Datasets:
J0nasW
/
paperswithcode

Dataset card Data Studio Files
xet
Community
1
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
6b23687b-6041-4dfd-b505-ef53ff342935
cross-lingual-learning-to-rank-with-shared
null
null
https://aclanthology.org/N18-2073
https://aclanthology.org/N18-2073.pdf
Cross-Lingual Learning-to-Rank with Shared Representations
Cross-lingual information retrieval (CLIR) is a document retrieval task where the documents are written in a language different from that of the user{'}s query. This is a challenging problem for data-driven approaches due to the general lack of labeled training data. We introduce a large-scale dataset derived from Wikipedia to support CLIR research in 25 languages. Further, we present a simple yet effective neural learning-to-rank model that shares representations across languages and reduces the data requirement. This model can exploit training data in, for example, Japanese-English CLIR to improve the results of Swahili-English CLIR.
['Shota Sasaki', 'Shigehiko Schamoni', 'Kevin Duh', 'Kentaro Inui', 'Shuo Sun']
2018-06-01
null
null
null
naacl-2018-6
['cross-lingual-information-retrieval']
['natural-language-processing']
[-3.59444559e-01 -6.19033098e-01 -7.19714463e-01 -3.64551663e-01 -1.38394094e+00 -7.28936493e-01 6.99157476e-01 2.16343284e-01 -9.00473535e-01 6.45786762e-01 6.28204942e-01 -3.58196080e-01 -4.95910108e-01 -3.18029404e-01 -2.18886107e-01 -1.91296548e-01 1.87051937e-01 7.69442976e-01 -4.53249179e-02 -6.62591815e-01 3.68925363e-01 4.74688023e-01 -1.28869820e+00 8.29482555e-01 8.27372789e-01 6.96356833e-01 4.98882264e-01 2.90112048e-01 -5.62711418e-01 9.26759422e-01 -5.02846360e-01 -6.32456988e-02 1.18051514e-01 3.43294255e-02 -9.97618437e-01 -4.59245294e-01 5.18325448e-01 -2.09984645e-01 -5.16956806e-01 7.95872390e-01 6.52106345e-01 3.59091759e-01 8.95432651e-01 -8.13947856e-01 -1.62026811e+00 8.79914582e-01 -5.25816381e-01 1.83179468e-01 9.94841382e-02 -1.01442361e+00 1.35137105e+00 -1.28091729e+00 8.79610598e-01 1.36540759e+00 3.29904348e-01 6.97480142e-01 -7.02214837e-01 -6.01372480e-01 1.48703400e-02 2.56834596e-01 -1.71949124e+00 -2.24206775e-01 6.02535486e-01 -4.47774053e-01 9.92908478e-01 1.67831108e-01 -2.49608740e-01 8.07443619e-01 -1.80988044e-01 9.64513719e-01 8.63914132e-01 -8.60834062e-01 -2.44376555e-01 3.77255559e-01 7.55116224e-01 3.08995366e-01 2.66319394e-01 -1.41546056e-01 -3.67724746e-01 -2.05796242e-01 2.90816605e-01 8.85573551e-02 -1.38082698e-01 -4.37452607e-02 -1.00672567e+00 1.10865796e+00 5.10637701e-01 7.25838840e-01 -1.30801082e-01 -1.57451943e-01 7.85527468e-01 5.76568961e-01 6.90182626e-01 7.25378633e-01 -8.86163831e-01 5.96852362e-01 -8.77558410e-01 9.57458466e-02 5.28553665e-01 9.59177136e-01 6.89074337e-01 -2.48995334e-01 -3.63966018e-01 1.54752684e+00 4.10824478e-01 5.07880807e-01 7.52402246e-01 -6.10065401e-01 4.18118656e-01 6.21380627e-01 7.05005601e-02 -7.75320768e-01 -2.01846182e-01 -2.34715000e-01 -6.75168216e-01 -4.44413036e-01 1.32677723e-02 1.54667795e-01 -7.56849289e-01 1.35152483e+00 -2.23678067e-01 -6.90104485e-01 4.70147312e-01 8.14003408e-01 1.42446530e+00 9.11168337e-01 -5.25875911e-02 -1.92831740e-01 1.27691269e+00 -1.15852356e+00 -9.29170191e-01 -3.35577339e-01 1.06480455e+00 -7.93634653e-01 1.17025459e+00 2.06081077e-01 -6.76604092e-01 -5.00588655e-01 -7.62129366e-01 -7.16304302e-01 -1.11217773e+00 6.00139141e-01 6.00911677e-01 2.10005283e-01 -1.29023767e+00 -1.76937152e-02 -3.34390402e-02 -3.64016175e-01 -1.80615380e-01 1.38580874e-01 -5.30470848e-01 -6.83422804e-01 -1.67563522e+00 1.09546757e+00 8.53598952e-01 1.00634523e-01 -4.61040318e-01 -4.77690279e-01 -9.03386652e-01 -9.96377543e-02 2.94390887e-01 -1.26424907e-02 1.07162583e+00 -7.58257031e-01 -7.99386382e-01 1.09030735e+00 -2.27497086e-01 -2.77798180e-03 7.67635405e-02 -3.44506830e-01 -6.87226653e-01 -1.91433832e-01 3.56233269e-01 5.37144840e-01 1.96203679e-01 -1.44745064e+00 -4.67673242e-01 -4.95816410e-01 -1.89578861e-01 4.39750075e-01 -6.56961203e-01 8.88689220e-01 -1.03612006e+00 -5.33558309e-01 -2.23499119e-01 -8.18481266e-01 -1.02130584e-01 -4.56082433e-01 -1.13059096e-01 -9.06115592e-01 5.79977989e-01 -1.05039108e+00 1.56792915e+00 -2.06908178e+00 -1.32966787e-01 6.55832365e-02 -3.30041684e-02 3.50850910e-01 -5.60822785e-01 8.35029185e-01 3.52113619e-02 4.76260155e-01 4.65396047e-01 -8.06405321e-02 -1.42059237e-01 1.58255696e-01 -3.79079252e-01 7.07115680e-02 -1.32677466e-01 9.48281467e-01 -8.86506736e-01 -6.41771376e-01 -2.30944246e-01 5.83290398e-01 -1.72956392e-01 1.43048361e-01 4.00508791e-02 6.53080642e-02 -6.85622931e-01 5.18795311e-01 2.61709183e-01 -2.80660212e-01 1.51520386e-01 -9.99120772e-02 -3.29384327e-01 5.05504906e-01 -6.43835485e-01 1.86265612e+00 -7.22561300e-01 7.15569377e-01 -2.50171751e-01 -7.58847415e-01 9.68175709e-01 4.45056468e-01 3.16444218e-01 -1.24990773e+00 -6.01290427e-02 5.87576210e-01 -1.54280961e-01 -4.67685282e-01 9.74004328e-01 3.13774794e-01 -3.21206301e-01 6.57331765e-01 -3.10398072e-01 1.55381501e-01 6.54780626e-01 1.96404696e-01 6.99287653e-01 -2.09738597e-01 2.98278004e-01 -6.28687501e-01 6.80770516e-01 2.85604835e-01 3.56432259e-01 8.86957765e-01 2.86851674e-01 3.98158371e-01 -3.79226923e-01 -6.30833030e-01 -1.03903055e+00 -7.13085115e-01 -3.66907805e-01 1.83535278e+00 -1.58069089e-01 -3.62575501e-01 -1.05852231e-01 -4.87827390e-01 -8.81298929e-02 6.78115368e-01 -5.16110420e-01 -5.73148998e-03 -5.89825451e-01 -4.07397658e-01 4.18036073e-01 3.22360545e-01 2.73340821e-01 -1.36357200e+00 2.15612069e-01 1.88433360e-02 -3.84963423e-01 -9.87172127e-01 -7.93563902e-01 2.78183103e-01 -5.51471531e-01 -6.63390517e-01 -8.74746740e-01 -1.35766160e+00 5.35215497e-01 6.47264779e-01 1.42065310e+00 1.68646216e-01 -3.36216509e-01 3.13733459e-01 -6.74033105e-01 -5.03482401e-01 -3.71681780e-01 2.87599862e-01 1.15481861e-01 -6.49214745e-01 1.04482687e+00 5.44872820e-01 -1.64617598e-01 1.79596126e-01 -1.00571430e+00 -4.49533015e-01 4.46267635e-01 9.43369627e-01 5.75882554e-01 3.08485419e-01 7.47671723e-01 -9.69183803e-01 1.22396255e+00 -5.24743676e-01 -4.66348410e-01 9.61971998e-01 -7.09332049e-01 2.67048389e-01 4.33597714e-01 -4.64526474e-01 -9.53154683e-01 -4.57404166e-01 3.30386788e-01 -2.30385050e-01 2.27821633e-01 1.30229318e+00 2.58229405e-01 4.24790233e-01 8.08862090e-01 1.82218775e-01 -4.30812865e-01 -8.78241360e-01 3.21022838e-01 1.41965234e+00 2.20522314e-01 -7.67741919e-01 3.24304491e-01 4.19119932e-02 -5.62060654e-01 -8.41865122e-01 -1.28372407e+00 -1.03186345e+00 -7.43903816e-01 2.85653956e-02 7.45985210e-01 -1.33041453e+00 -2.15724707e-01 1.55396909e-01 -1.37520123e+00 -8.09880644e-02 2.02789038e-01 5.65113425e-01 1.27549231e-01 1.99265659e-01 -8.76711130e-01 -6.64705336e-01 -6.83926404e-01 -8.86442900e-01 1.08865654e+00 -8.73161107e-02 -1.05880983e-01 -1.14678025e+00 4.16577876e-01 5.12143612e-01 4.05128837e-01 -6.99534893e-01 1.30268157e+00 -1.15384901e+00 8.31146240e-02 -4.35294360e-01 -6.10210955e-01 5.92011571e-01 4.15254146e-01 -2.09787134e-02 -7.19310582e-01 -6.06637836e-01 -4.77561831e-01 -7.92128503e-01 8.80198002e-01 6.63609356e-02 1.10223949e+00 -3.14028621e-01 -1.63022708e-02 -8.72943401e-02 1.46180940e+00 4.61519033e-01 3.09222460e-01 6.25862658e-01 7.61932015e-01 8.93261433e-01 5.51901996e-01 -3.43037806e-02 4.98744458e-01 7.41454303e-01 -5.04491985e-01 -1.51839823e-01 -8.43930021e-02 -2.71618098e-01 2.57746816e-01 1.37942994e+00 1.44392416e-01 -2.92361766e-01 -1.07393253e+00 7.37638116e-01 -1.78261673e+00 -7.27592647e-01 -7.38451108e-02 2.16177797e+00 1.14139032e+00 -5.42543650e-01 -3.96178454e-01 -2.78466612e-01 6.75121725e-01 -1.18407190e-01 -3.04796815e-01 -5.57086527e-01 -2.83098131e-01 3.81302200e-02 4.15562332e-01 6.60938203e-01 -1.05836546e+00 1.35153759e+00 6.74808216e+00 1.05318654e+00 -1.04955626e+00 5.53826801e-02 2.50289679e-01 2.87785381e-01 -4.98346806e-01 -4.12950456e-01 -1.06480277e+00 1.16538152e-01 8.22745621e-01 -4.60586488e-01 4.86595422e-01 9.11880374e-01 1.20405525e-01 3.23497921e-01 -1.16524613e+00 1.03012085e+00 5.80761254e-01 -1.06566584e+00 5.01928151e-01 -1.16940871e-01 9.09990668e-01 4.11735892e-01 3.51067334e-02 6.37612164e-01 6.79039896e-01 -1.08182251e+00 1.86830476e-01 2.01664463e-01 1.02489495e+00 -8.23091745e-01 8.67713749e-01 4.42167401e-01 -8.80233824e-01 5.96768782e-02 -7.06323147e-01 3.91333252e-01 -2.95193911e-01 3.05771023e-01 -6.36310995e-01 5.03646076e-01 1.00682569e+00 7.61892080e-01 -7.68335819e-01 7.10070670e-01 -5.83962910e-02 1.29725561e-01 8.91041458e-02 -1.58654720e-01 4.36945885e-01 -1.29503012e-01 -8.59772507e-03 1.49910426e+00 4.02488768e-01 -3.27650160e-01 4.89645004e-01 4.08708543e-01 -3.52262527e-01 6.88624799e-01 -1.08945870e+00 -3.53143364e-01 4.84698743e-01 1.04373515e+00 -5.55866174e-02 -1.64999619e-01 -7.07288384e-01 8.29981565e-01 7.06849992e-01 6.23010576e-01 1.04039773e-01 -5.22789001e-01 3.65324497e-01 -1.99713871e-01 -1.75713375e-01 -2.12225825e-01 1.82386443e-01 -1.24295104e+00 -9.96897891e-02 -8.91569734e-01 8.75485897e-01 -8.73503804e-01 -1.88599551e+00 9.60779846e-01 1.31672323e-01 -1.16381705e+00 -7.04341292e-01 -9.63194251e-01 1.42652526e-01 1.20613801e+00 -1.90222776e+00 -1.29897070e+00 2.81182736e-01 8.52085829e-01 7.45661438e-01 -8.42695177e-01 9.96518314e-01 6.16141975e-01 -2.93198228e-01 5.90628684e-01 9.12769794e-01 6.53155088e-01 1.08430052e+00 -9.31894481e-01 -1.98753506e-01 5.54035664e-01 3.62399668e-01 1.05106151e+00 1.89385310e-01 -6.14573479e-01 -1.34490991e+00 -9.76120234e-01 1.82376051e+00 -4.93462473e-01 9.25119996e-01 -1.91312537e-01 -9.74252224e-01 6.55190766e-01 4.16834027e-01 -1.75724640e-01 1.03922856e+00 5.99909186e-01 -8.22896242e-01 -1.14074066e-01 -7.63210535e-01 6.78075612e-01 5.67468464e-01 -1.05085719e+00 -8.26840162e-01 5.07069588e-01 7.01862931e-01 1.21182755e-01 -7.87262678e-01 1.99395478e-01 4.86466676e-01 -1.60408899e-01 1.16731393e+00 -1.11414146e+00 4.36621487e-01 -2.57202685e-01 -5.44310153e-01 -1.31670976e+00 -5.84131360e-01 1.09478936e-01 3.10169071e-01 1.20463192e+00 6.26263142e-01 -2.12039515e-01 1.61381081e-01 5.59878290e-01 8.55108500e-02 -9.70009714e-02 -7.71976411e-01 -8.94626439e-01 6.54367566e-01 -3.73570859e-01 2.28219971e-01 1.29924941e+00 7.37370402e-02 6.73501194e-01 -4.54134017e-01 -1.03812620e-01 3.24263483e-01 -1.35765728e-02 2.06091523e-01 -1.44342399e+00 9.32590589e-02 -4.28818911e-01 2.33673200e-01 -8.97138655e-01 7.34218121e-01 -1.50070941e+00 2.04684854e-01 -1.84664941e+00 6.26925528e-01 -5.99540293e-01 -8.51715565e-01 5.83761990e-01 -9.96927470e-02 1.13682292e-01 2.32118033e-02 5.09310901e-01 -7.37125516e-01 3.59940261e-01 1.09185171e+00 -5.22378683e-01 -2.14123726e-01 -4.50264245e-01 -8.95329475e-01 4.01568025e-01 4.86695141e-01 -5.00963569e-01 -5.32280147e-01 -1.16505241e+00 5.15646636e-01 7.71757495e-03 -3.98922384e-01 -1.34122863e-01 3.73668879e-01 -2.12134421e-01 2.75593996e-01 -7.31212437e-01 -7.73434639e-02 -8.54200780e-01 -3.07602525e-01 6.83366880e-02 -1.07789361e+00 4.68108892e-01 1.25646546e-01 1.42030597e-01 -6.39818966e-01 -4.84481037e-01 4.68732506e-01 -3.39377493e-01 -1.09350491e+00 2.77986765e-01 -4.13064599e-01 1.65108502e-01 3.27273339e-01 4.03075576e-01 -4.89997327e-01 -2.82902390e-01 -2.30473012e-01 4.40070063e-01 4.69678752e-02 1.04808342e+00 6.05521381e-01 -1.68434787e+00 -1.19218481e+00 -1.29406154e-01 9.46056426e-01 -3.40716809e-01 -1.27885798e-02 1.22796670e-02 -4.70981091e-01 1.08648539e+00 4.15116102e-02 -9.65134427e-02 -1.25075877e+00 8.73525023e-01 -2.33111717e-02 -6.75667167e-01 -3.11346680e-01 4.96623427e-01 2.35366851e-01 -1.12903273e+00 3.30493867e-01 4.22737360e-01 -8.84627819e-01 3.49604815e-01 9.76839662e-01 1.88657135e-01 3.02510530e-01 -7.23344028e-01 -3.12460035e-01 4.13341135e-01 -6.69589639e-01 -2.07923323e-01 1.28897047e+00 -4.05975103e-01 -5.97806990e-01 8.36893320e-01 1.64949417e+00 -6.13467470e-02 -1.98083874e-02 -8.73603523e-01 5.91888070e-01 -4.20296878e-01 4.25103754e-01 -1.00140345e+00 -9.66470301e-01 8.08353901e-01 5.45020759e-01 -1.75482348e-01 8.28699589e-01 2.37018779e-01 3.92995119e-01 1.26702547e+00 4.73814517e-01 -1.61024487e+00 -5.47601692e-02 1.03340375e+00 1.10159230e+00 -1.55714536e+00 -1.46377996e-01 1.96501657e-01 -6.40550792e-01 1.14831567e+00 4.47307378e-01 7.12949857e-02 8.51095021e-01 -9.03659686e-02 5.33347428e-01 -1.15780160e-01 -7.75073469e-01 -3.06533903e-01 8.81079674e-01 4.49943304e-01 1.16291785e+00 -8.69911686e-02 -7.02669501e-01 5.74916124e-01 1.45314023e-01 -2.46609747e-02 2.38450781e-01 8.74104321e-01 -2.62461275e-01 -1.32324147e+00 -8.90665203e-02 3.64238441e-01 -6.61246598e-01 -6.82509243e-01 -7.12867081e-01 7.07593143e-01 -3.54284674e-01 1.09909379e+00 6.60448372e-02 -3.67529355e-02 -1.99129526e-02 8.62602592e-02 -6.65870383e-02 -8.95088017e-01 -4.52582836e-01 2.10355401e-01 1.97124079e-01 -8.95763561e-02 -5.20834327e-01 -3.52198005e-01 -9.11281168e-01 -4.36171144e-02 -1.37575373e-01 5.12232482e-01 8.05745125e-01 9.12540138e-01 2.41742015e-01 9.00921822e-02 7.78767705e-01 -2.72844434e-01 -2.32029036e-01 -1.05561924e+00 -7.94465721e-01 4.40768570e-01 2.14570865e-01 -2.82604694e-01 -3.17016780e-01 -2.26026669e-01]
[11.358192443847656, 9.81503677368164]
5bf58c07-f327-46d0-bcab-44ad0cebb4c6
hotpotqa-a-dataset-for-diverse-explainable
1809.09600
null
http://arxiv.org/abs/1809.09600v1
http://arxiv.org/pdf/1809.09600v1.pdf
HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answering
Existing question answering (QA) datasets fail to train QA systems to perform complex reasoning and provide explanations for answers. We introduce HotpotQA, a new dataset with 113k Wikipedia-based question-answer pairs with four key features: (1) the questions require finding and reasoning over multiple supporting documents to answer; (2) the questions are diverse and not constrained to any pre-existing knowledge bases or knowledge schemas; (3) we provide sentence-level supporting facts required for reasoning, allowing QA systems to reason with strong supervision and explain the predictions; (4) we offer a new type of factoid comparison questions to test QA systems' ability to extract relevant facts and perform necessary comparison. We show that HotpotQA is challenging for the latest QA systems, and the supporting facts enable models to improve performance and make explainable predictions.
['Peng Qi', 'Saizheng Zhang', 'Yoshua Bengio', 'Ruslan Salakhutdinov', 'Zhilin Yang', 'William W. Cohen', 'Christopher D. Manning']
2018-09-25
hotpotqa-a-dataset-for-diverse-explainable-1
https://aclanthology.org/D18-1259
https://aclanthology.org/D18-1259.pdf
emnlp-2018-10
['multi-hop-question-answering']
['knowledge-base']
[-1.21766217e-01 9.29441035e-01 -1.66580096e-01 -6.90689564e-01 -1.45605314e+00 -7.81450450e-01 4.68076795e-01 4.23322767e-01 1.25836655e-01 1.20023429e+00 6.03837967e-01 -7.72815764e-01 -5.48138916e-01 -1.02968037e+00 -9.82934654e-01 3.55302334e-01 7.81844463e-03 1.24241352e+00 8.87918711e-01 -1.05257666e+00 9.13679749e-02 -6.50743544e-02 -1.40057373e+00 1.10408998e+00 1.69797945e+00 9.49458838e-01 -1.23823360e-01 9.50716197e-01 -5.01587808e-01 1.61596358e+00 -5.78578413e-01 -1.11265147e+00 1.81567967e-01 -6.50592089e-01 -1.82860386e+00 -5.69909155e-01 9.99117076e-01 -2.14178771e-01 -2.76149422e-01 5.38839161e-01 1.95656661e-02 4.44875620e-02 5.99334538e-01 -1.33319128e+00 -1.19110906e+00 6.01860523e-01 3.63725662e-01 3.53396654e-01 1.14721656e+00 3.14981133e-01 1.36104703e+00 -5.51533818e-01 8.44916880e-01 1.18067205e+00 6.96376920e-01 8.56973529e-01 -9.19808626e-01 1.56305730e-02 -2.55603939e-01 8.95146132e-01 -7.22305715e-01 -3.15748274e-01 1.22657776e-01 -1.46957943e-02 1.41411626e+00 7.02320099e-01 3.85104120e-01 7.23718703e-01 2.01453373e-01 5.53298831e-01 9.58342671e-01 -4.33465451e-01 1.39007479e-01 4.23605740e-02 5.19313514e-01 9.58160937e-01 1.10779166e-01 -4.27493751e-01 -4.52033281e-01 -2.45642900e-01 6.90393075e-02 -5.67795336e-01 -3.93811107e-01 -2.33356562e-03 -1.24714530e+00 9.56889570e-01 7.06350625e-01 4.66625206e-03 -2.88748056e-01 -3.00382048e-01 2.41238222e-01 7.80317247e-01 -2.33662978e-01 1.41441321e+00 -9.77452934e-01 -8.16702768e-02 -4.48921144e-01 1.02473438e+00 1.46612716e+00 9.22764897e-01 8.89243245e-01 -5.19349694e-01 -4.89911854e-01 5.11652946e-01 -5.41715175e-02 7.12112963e-01 3.07339102e-01 -1.68349493e+00 1.21139205e+00 1.09603012e+00 3.75061691e-01 -8.71016085e-01 -4.20149416e-01 9.16728228e-02 -1.52356565e-01 -4.39437062e-01 7.64202476e-01 -1.26049906e-01 -6.00742400e-01 1.46177495e+00 1.94729462e-01 -3.94724011e-01 7.11518764e-01 6.88749313e-01 1.56350851e+00 5.51693797e-01 1.03566229e-01 9.12527218e-02 1.57467556e+00 -1.16504073e+00 -7.29914486e-01 -4.41463679e-01 8.79764736e-01 -5.01154006e-01 1.40095246e+00 3.72844748e-02 -1.41831195e+00 -3.75634491e-01 -6.89231992e-01 -7.16057479e-01 -4.96587187e-01 -4.32569295e-01 8.28469813e-01 2.83488244e-01 -1.03822696e+00 1.36114419e-01 -6.92488104e-02 -1.30707338e-01 2.62823284e-01 8.40050951e-02 -4.12653983e-01 -6.74890697e-01 -1.87359869e+00 1.53966522e+00 4.86766428e-01 -4.62246686e-01 -6.32872701e-01 -1.08778322e+00 -1.12642837e+00 3.71714056e-01 6.57348812e-01 -1.33412445e+00 1.60358548e+00 -4.44931924e-01 -1.10647237e+00 5.60009956e-01 -5.70196390e-01 -7.26127326e-01 3.31159890e-01 -1.28718644e-01 -6.12459600e-01 7.33378470e-01 5.81648529e-01 8.43157530e-01 -4.15177196e-02 -9.90220904e-01 -5.68743765e-01 -2.08476767e-01 8.92978311e-01 2.22815096e-01 1.13308892e-01 -1.24613799e-01 -2.05243081e-01 5.28383069e-02 1.83824062e-01 -3.99825960e-01 -7.07853884e-02 -3.66002887e-01 -6.73711717e-01 -5.55427134e-01 3.90214413e-01 -1.00757718e+00 8.68007123e-01 -1.19646251e+00 -1.31238014e-01 -1.94805592e-01 3.43791157e-01 1.83872208e-01 -3.68150890e-01 6.86509430e-01 5.17840907e-02 1.64786577e-01 -1.88544765e-01 3.19553792e-01 2.56406724e-01 5.24031222e-01 -7.93770134e-01 -7.30116785e-01 6.49559438e-01 1.42116690e+00 -9.81002152e-01 -8.21663558e-01 -2.29294538e-01 -3.33717078e-01 -7.98587859e-01 4.15560514e-01 -1.00104678e+00 2.18834758e-01 -4.96703058e-01 5.99037170e-01 4.66472834e-01 -4.16118681e-01 -1.77555054e-01 -1.31433830e-01 5.06864130e-01 1.23768914e+00 -8.30521166e-01 1.38451564e+00 -4.15796578e-01 4.38111335e-01 -4.62464184e-01 -6.24098182e-01 6.89654768e-01 2.86390960e-01 -1.19944997e-01 -1.07003438e+00 -4.29837286e-01 3.73009354e-01 1.56146672e-03 -1.10134053e+00 6.05836689e-01 -1.04559563e-01 -1.22089915e-01 4.45889533e-01 8.64446387e-02 -8.25388551e-01 6.93199635e-01 8.04870486e-01 1.47215629e+00 -3.46331984e-01 -5.06463349e-02 -1.58040926e-01 6.11866355e-01 9.97332513e-01 4.72127885e-01 8.73822272e-01 -9.33352299e-03 4.03851390e-01 6.30860806e-01 -6.16560876e-01 -7.48243868e-01 -1.21001554e+00 5.92239983e-02 9.49326456e-01 8.88435990e-02 -5.17654419e-01 -6.22473359e-01 -1.11346936e+00 4.07403074e-02 1.34985709e+00 -4.70280498e-01 8.30208212e-02 -6.39665484e-01 -1.82154905e-02 8.31965625e-01 5.86115003e-01 6.20169938e-01 -1.06702125e+00 -4.23650444e-01 8.96250904e-02 -1.15892243e+00 -1.22721541e+00 -7.98788220e-02 5.89579418e-02 -9.40626502e-01 -1.77092242e+00 -1.17293343e-01 -6.92774057e-01 4.86671865e-01 1.81844026e-01 1.96010065e+00 5.32984555e-01 1.80441976e-01 6.64816082e-01 -5.04672468e-01 -4.71584082e-01 -7.46707201e-01 3.85985188e-02 -3.99167627e-01 -1.02454686e+00 5.38269401e-01 -1.87216982e-01 -3.55246842e-01 4.37585860e-01 -7.32778668e-01 8.07130113e-02 1.97149336e-01 7.20854998e-01 3.68495464e-01 4.50824909e-02 8.88798296e-01 -1.06464827e+00 9.33399081e-01 -6.30913436e-01 -1.23301916e-01 1.05116987e+00 -3.56446207e-01 3.21588010e-01 7.96800613e-01 2.58612990e-01 -1.38052404e+00 -4.36897427e-01 -3.31417382e-01 4.46157068e-01 -1.37615725e-01 6.91627085e-01 -2.32426390e-01 2.30180979e-01 1.32499373e+00 -3.87361944e-02 -2.18386948e-01 -8.20312724e-02 7.89474607e-01 2.68252820e-01 8.96091461e-01 -9.82100010e-01 9.20375407e-01 1.17154635e-01 -1.34472385e-01 -4.99077290e-02 -1.53189254e+00 -1.08875714e-01 -4.99050081e-01 2.30742007e-01 8.10146451e-01 -7.43526697e-01 -1.09273338e+00 -5.40166736e-01 -1.28955984e+00 -3.64088774e-01 -4.53718364e-01 2.63433754e-01 -8.74188304e-01 4.24705863e-01 -8.82061660e-01 -4.77453917e-01 -5.03959656e-01 -7.31792986e-01 5.39164543e-01 4.67893571e-01 -5.04699826e-01 -1.01275456e+00 7.63069913e-02 1.34228122e+00 4.26816493e-01 -2.18088124e-02 1.60134792e+00 -9.62037742e-01 -6.91172600e-01 -2.64786258e-02 -2.95444995e-01 1.00692749e-01 -1.58802822e-01 -8.69718120e-02 -6.83617651e-01 3.95438015e-01 -5.38960733e-02 -1.09339249e+00 5.66314042e-01 -1.52143672e-01 9.53835130e-01 -6.48443103e-01 8.35874453e-02 -9.09928158e-02 9.71771240e-01 -1.94360480e-01 8.26084137e-01 4.63085771e-01 2.59360850e-01 1.07178044e+00 7.35621810e-01 -3.54646772e-01 1.29340541e+00 3.79614115e-01 5.14730275e-01 2.73928165e-01 -1.74936354e-01 -5.50716162e-01 5.79287298e-04 7.51924336e-01 1.24761991e-01 -1.35332361e-01 -1.18315077e+00 8.53695869e-01 -1.87624395e+00 -1.15113664e+00 -7.36742795e-01 1.61238050e+00 1.21706688e+00 1.22450721e-02 -1.95954517e-01 -1.36347398e-01 1.36064634e-01 -2.90190756e-01 -6.25178218e-01 -5.84372282e-01 -3.98004651e-01 4.39898759e-01 -9.99166071e-02 7.62150168e-01 -5.34912109e-01 9.15994942e-01 6.82987309e+00 4.88953799e-01 -3.33457701e-02 -4.66534644e-02 2.57210374e-01 2.50081450e-01 -9.89362538e-01 1.87420085e-01 -6.18696868e-01 1.27438102e-02 1.11201847e+00 -1.84409738e-01 3.51154655e-01 6.43363833e-01 -4.43268448e-01 -1.74370989e-01 -1.24557340e+00 1.62923798e-01 6.95486218e-02 -1.84339929e+00 4.95963544e-01 -5.43887079e-01 7.42638409e-01 -1.63567364e-01 -2.18148768e-01 9.17196989e-01 8.55416536e-01 -1.33267975e+00 2.65194386e-01 6.44748867e-01 2.58979917e-01 -6.02464676e-01 9.51312006e-01 6.06901705e-01 -7.08683074e-01 -3.85199904e-01 -7.04206765e-01 -2.21442506e-01 3.16143841e-01 3.11774433e-01 -1.08671308e+00 1.13433301e+00 7.52986372e-01 2.18949035e-01 -8.93422663e-01 9.36615169e-01 -9.15743470e-01 5.70326149e-01 -1.59267202e-01 -1.19818941e-01 2.73999333e-01 3.23945224e-01 2.34257221e-01 7.05045581e-01 1.55020759e-01 5.85594475e-01 -1.40873313e-01 9.69569027e-01 -2.09830970e-01 -7.64042735e-02 -2.26690888e-01 1.77967936e-01 5.85866272e-01 8.25948834e-01 -8.68040174e-02 -5.15932918e-01 -4.01860565e-01 6.35657310e-01 6.44618034e-01 2.29148507e-01 -5.15371025e-01 -3.86645257e-01 4.03150469e-01 -9.61987581e-03 5.29328287e-02 1.96023345e-01 -3.57817262e-01 -1.38976693e+00 3.27646434e-01 -1.35376883e+00 1.13264799e+00 -1.47491884e+00 -1.71548629e+00 7.43297338e-01 -1.61693737e-01 -7.62310386e-01 -6.42580092e-01 -6.15628600e-01 -6.88670218e-01 1.07150578e+00 -1.95023215e+00 -1.14011836e+00 -3.95165622e-01 9.41581249e-01 4.11448926e-01 -9.85151250e-03 1.16956985e+00 -2.79248711e-02 6.53580874e-02 3.72413367e-01 -5.69008112e-01 9.31960419e-02 8.55126739e-01 -1.69170201e+00 5.21921635e-01 4.99014914e-01 2.68874258e-01 8.22931945e-01 9.32602942e-01 -7.69230127e-01 -1.43495309e+00 -7.08292484e-01 1.58550644e+00 -1.49422383e+00 6.37369454e-01 1.26587391e-01 -1.39360607e+00 8.65710020e-01 5.09500444e-01 -1.94377109e-01 8.79910529e-01 6.09381616e-01 -9.05237138e-01 3.04427799e-02 -1.30850458e+00 4.48757201e-01 8.11849177e-01 -7.00728536e-01 -1.81468034e+00 7.79172182e-01 1.25980139e+00 -8.43764067e-01 -1.00015628e+00 4.66402948e-01 1.28117055e-01 -1.02149844e+00 1.02109933e+00 -1.47144711e+00 9.81407106e-01 -4.42432016e-01 -1.67228252e-01 -1.14953065e+00 -1.25663728e-01 -1.20276995e-01 -5.05575061e-01 9.87683177e-01 1.12617302e+00 -4.68055487e-01 7.47920334e-01 1.40103948e+00 -3.76678139e-01 -8.72370601e-01 -9.23908830e-01 -5.10980368e-01 2.68249691e-01 -3.59672278e-01 9.97096479e-01 1.07363129e+00 4.77656662e-01 5.98476350e-01 8.40369537e-02 3.95425111e-01 4.50542271e-02 5.40666342e-01 8.06657612e-01 -1.09326744e+00 -3.74005914e-01 -1.45672327e-02 8.06874931e-02 -1.13356745e+00 1.72359318e-01 -9.09412622e-01 -2.45126486e-02 -2.41797113e+00 1.61538616e-01 -4.08126503e-01 4.03523743e-01 7.57146955e-01 -5.93245208e-01 -1.58243820e-01 7.72131532e-02 2.25319099e-02 -1.12440884e+00 4.61963177e-01 1.44426799e+00 -1.78566396e-01 3.01557809e-01 -1.51775191e-02 -1.08217359e+00 4.49535847e-01 4.57698971e-01 -1.57408193e-01 -7.61985838e-01 -7.40718901e-01 8.05454791e-01 5.03156662e-01 4.41058040e-01 -8.29078436e-01 4.88842994e-01 -3.38033974e-01 3.91284108e-01 -4.38631445e-01 2.52131164e-01 -6.32652700e-01 -4.64937657e-01 4.38201666e-01 -5.51738679e-01 3.66178304e-01 3.27337980e-01 2.21284091e-01 -5.14280081e-01 -4.72610474e-01 1.36689320e-01 -2.50921845e-01 -6.81551874e-01 -1.14564344e-01 -1.22205906e-01 9.65212643e-01 6.70328021e-01 6.06471077e-02 -1.29453802e+00 -7.90180385e-01 -4.89909708e-01 1.07261491e+00 1.68837458e-01 2.85618126e-01 5.45991361e-01 -1.07777607e+00 -9.09950197e-01 -4.94800091e-01 4.46788132e-01 1.56863555e-01 5.89086175e-01 3.63196969e-01 -6.73155487e-01 8.16190362e-01 -2.02601388e-01 -1.68058455e-01 -8.85836899e-01 5.19851327e-01 5.06768644e-01 -6.24160171e-01 -3.97477776e-01 8.96635890e-01 -3.10372174e-01 -1.22126746e+00 -1.22497499e-01 -2.81526119e-01 -4.23174709e-01 -2.64069110e-01 7.65511334e-01 1.92413688e-01 2.84220815e-01 -1.31115586e-01 -2.12267041e-01 5.16494326e-02 -6.07723929e-02 1.69437826e-01 1.27632499e+00 -5.03377616e-02 -2.11665109e-01 -4.94212992e-02 5.32180190e-01 -3.44158225e-02 -6.95353329e-01 -3.74002159e-01 3.20090562e-01 -4.64278668e-01 -7.39058435e-01 -1.58094954e+00 -3.45148832e-01 9.63827848e-01 -2.16716111e-01 5.67638516e-01 7.84598708e-01 3.64150196e-01 1.23618877e+00 1.14117646e+00 3.25661302e-01 -8.55196416e-01 2.58196712e-01 1.02052641e+00 1.17233455e+00 -1.30032408e+00 -2.75741428e-01 -6.90441310e-01 -9.28324163e-01 1.15236723e+00 1.17054939e+00 4.51288044e-01 -1.40694499e-01 -3.12189788e-01 4.38067198e-01 -6.76117778e-01 -1.24601007e+00 -2.31928736e-01 6.67339504e-01 7.12258875e-01 2.75274664e-01 -1.79229990e-01 -1.93793457e-02 8.76992285e-01 -9.37626481e-01 -2.40870014e-01 4.81774151e-01 5.13740063e-01 -5.94460845e-01 -9.09779727e-01 -2.08142430e-01 7.07483053e-01 -1.74054056e-01 -2.28737742e-01 -7.16433883e-01 8.07350934e-01 -1.64069921e-01 1.47682953e+00 -3.96467477e-01 -5.43673150e-02 7.20842063e-01 3.67844522e-01 4.25489068e-01 -7.90977240e-01 -7.59056330e-01 -1.44084370e+00 7.67051756e-01 -6.23892367e-01 -2.48883087e-02 -1.92160606e-01 -1.61932087e+00 -4.70036805e-01 -1.32672325e-01 9.76240695e-01 1.21171847e-01 1.42708731e+00 6.21855319e-01 4.15124565e-01 4.13162820e-02 3.08831990e-01 -6.18162513e-01 -8.33367169e-01 8.84004459e-02 6.68550909e-01 6.72123507e-02 -2.14017943e-01 -2.76879996e-01 -7.78935999e-02]
[11.033415794372559, 7.937178611755371]
d72ea368-e295-4833-89ef-8a3dd45e878b
quality-aware-generative-adversarial-networks
1911.03149
null
https://arxiv.org/abs/1911.03149v1
https://arxiv.org/pdf/1911.03149v1.pdf
Quality Aware Generative Adversarial Networks
Generative Adversarial Networks (GANs) have become a very popular tool for implicitly learning high-dimensional probability distributions. Several improvements have been made to the original GAN formulation to address some of its shortcomings like mode collapse, convergence issues, entanglement, poor visual quality etc. While a significant effort has been directed towards improving the visual quality of images generated by GANs, it is rather surprising that objective image quality metrics have neither been employed as cost functions nor as regularizers in GAN objective functions. In this work, we show how a distance metric that is a variant of the Structural SIMilarity (SSIM) index (a popular full-reference image quality assessment algorithm), and a novel quality aware discriminator gradient penalty function that is inspired by the Natural Image Quality Evaluator (NIQE, a popular no-reference image quality assessment algorithm) can each be used as excellent regularizers for GAN objective functions. Specifically, we demonstrate state-of-the-art performance using the Wasserstein GAN gradient penalty (WGAN-GP) framework over CIFAR-10, STL10 and CelebA datasets.
['Sumohana S. Channappayya', 'Parimala Kancharla']
2019-11-08
quality-aware-generative-adversarial-networks-1
http://papers.nips.cc/paper/8560-quality-aware-generative-adversarial-networks
http://papers.nips.cc/paper/8560-quality-aware-generative-adversarial-networks.pdf
neurips-2019-12
['no-reference-image-quality-assessment']
['computer-vision']
[ 1.14591703e-01 -2.78029665e-02 3.19213808e-01 -3.76805902e-01 -1.11395943e+00 -3.58590990e-01 7.84946084e-01 -3.00851017e-01 -4.60811913e-01 1.11490309e+00 2.49387190e-01 1.84779465e-01 -1.56999856e-01 -8.10474217e-01 -4.78339791e-01 -1.03549302e+00 8.06837156e-02 5.60587168e-01 -1.72327295e-01 -2.23296851e-01 2.12219507e-01 5.49013674e-01 -1.29378939e+00 -1.13686949e-01 1.09013236e+00 9.05632377e-01 -1.35649949e-01 8.26553345e-01 1.49641499e-01 8.85173380e-01 -7.65907288e-01 -8.59971523e-01 3.07567686e-01 -9.70114231e-01 -8.21242988e-01 -3.27743679e-01 6.05174005e-01 -6.21690229e-02 -3.92801374e-01 1.27828372e+00 8.01374137e-01 2.13525727e-01 1.09363651e+00 -1.38970256e+00 -1.06413054e+00 2.33004898e-01 -3.78245592e-01 3.24019760e-01 9.51550305e-02 3.96636903e-01 1.01418829e+00 -6.50045693e-01 4.80466276e-01 1.30232573e+00 6.33109689e-01 7.06004679e-01 -1.39796793e+00 -4.08513844e-01 -4.48317081e-01 2.77929842e-01 -1.26700330e+00 -6.24472573e-02 7.66882658e-01 -2.31745034e-01 5.53640783e-01 2.33545035e-01 4.09922391e-01 1.25976348e+00 3.83990824e-01 6.07982457e-01 1.37478638e+00 -3.77695352e-01 3.97188991e-01 1.61203131e-01 -3.27918410e-01 7.44623840e-01 -5.69990203e-02 3.36515099e-01 -3.61221522e-01 -9.53144059e-02 8.93368959e-01 -2.82519817e-01 -3.64899218e-01 -4.26771909e-01 -1.15778184e+00 1.02979243e+00 9.01754677e-01 3.54402125e-01 -3.82452279e-01 2.96382993e-01 9.06905904e-02 5.63834310e-01 5.11600852e-01 5.87341726e-01 1.51142150e-01 -2.97214061e-01 -9.54948366e-01 2.33043298e-01 5.12980878e-01 4.24483269e-01 7.27466226e-01 4.43207145e-01 -6.25579834e-01 6.63716197e-01 3.31455171e-01 6.90141439e-01 5.35190642e-01 -1.04369950e+00 6.20191917e-02 2.54610330e-01 9.33085382e-02 -9.48423862e-01 -1.62284955e-01 -6.04794145e-01 -1.29231334e+00 8.44810903e-01 4.33122367e-01 1.27203360e-01 -9.32914138e-01 1.89348173e+00 1.23284072e-01 9.99726281e-02 1.18604280e-01 1.00580895e+00 7.53220022e-01 5.83485186e-01 1.25785977e-01 -6.52179793e-02 9.58351195e-01 -9.50847924e-01 -5.96736491e-01 3.06616634e-01 1.09444968e-01 -7.92502940e-01 1.33898461e+00 6.50569141e-01 -1.38344336e+00 -8.06265593e-01 -1.23879623e+00 5.78570254e-02 -4.29803669e-01 -2.83418387e-01 5.49168646e-01 9.61820304e-01 -1.20607209e+00 8.36673439e-01 -7.58742452e-01 -8.94788280e-02 6.36526704e-01 1.39936328e-01 -2.71548808e-01 -3.42494100e-02 -1.07100034e+00 1.13380468e+00 1.31241292e-01 -2.16057137e-01 -1.23854542e+00 -6.87073052e-01 -4.76687193e-01 -7.10362270e-02 -1.55636802e-01 -9.01783943e-01 5.95863938e-01 -1.15171731e+00 -1.97645831e+00 9.95162189e-01 5.09180367e-01 -5.33559084e-01 6.89724624e-01 -1.03264935e-01 -5.63122690e-01 2.36047000e-01 -1.57826856e-01 9.60690200e-01 1.17482162e+00 -1.24642026e+00 -1.44061759e-01 -2.77266383e-01 1.28394868e-02 1.68759242e-01 -1.91714745e-02 -1.14447415e-01 1.75007731e-01 -9.69054222e-01 -2.47435153e-01 -8.66054773e-01 3.87491137e-02 8.17055628e-02 -3.10499907e-01 -3.06614935e-02 5.75851977e-01 -6.13716185e-01 7.92191446e-01 -1.86018550e+00 5.28143048e-01 1.82027325e-01 1.81498170e-01 4.82181460e-01 -3.94321084e-01 2.32562780e-01 -9.41834822e-02 2.28523076e-01 -6.31192744e-01 -3.60251486e-01 1.53770074e-01 2.09025964e-01 -1.52730823e-01 5.28463185e-01 2.80168980e-01 1.18641639e+00 -9.21341002e-01 -3.63674611e-01 3.30789655e-01 9.30929482e-01 -3.58064741e-01 4.51874286e-01 5.03296852e-02 7.90772319e-01 1.96332596e-02 3.73642355e-01 8.80791903e-01 -1.16344839e-01 -5.32011628e-01 -4.03294504e-01 3.53330731e-01 -1.19723842e-01 -8.86277020e-01 1.85688126e+00 -4.00437713e-01 5.77893078e-01 -2.87156850e-01 -9.38629210e-01 8.72475684e-01 2.65360415e-01 2.86248714e-01 -9.47465003e-01 8.53023008e-02 2.42201146e-02 2.71640662e-02 -1.62900150e-01 2.86806196e-01 -4.67595309e-01 1.86875015e-01 2.87592202e-01 4.28771079e-01 -3.42574656e-01 7.51497597e-02 1.10168286e-01 1.04664636e+00 3.15729439e-01 1.48341879e-01 -2.82060683e-01 7.46580958e-01 -3.56950700e-01 1.84465259e-01 5.60935259e-01 -2.82413155e-01 1.22520900e+00 4.08358574e-01 -2.73057431e-01 -1.50585747e+00 -1.51763856e+00 -6.15785010e-02 6.71383440e-01 -7.99232721e-02 -1.55227005e-01 -8.62588942e-01 -8.04133296e-01 -2.91713417e-01 7.30732560e-01 -8.01648438e-01 -2.82892883e-01 -4.03652221e-01 -1.04218626e+00 8.82607102e-01 4.50526983e-01 8.78105044e-01 -1.17023778e+00 -3.89250338e-01 1.30188763e-01 -5.59879690e-02 -7.23325729e-01 -2.96304852e-01 5.57543449e-02 -6.94315195e-01 -9.16364193e-01 -1.17514932e+00 -3.00837606e-01 4.49545532e-01 -2.93402821e-01 1.46215725e+00 -1.87358305e-01 -3.50788563e-01 4.52646464e-01 -3.81132960e-01 -6.36885762e-02 -7.55545199e-01 -2.45338112e-01 -1.24204770e-01 2.41215602e-02 -4.48514223e-02 -7.69030035e-01 -8.66657853e-01 4.12738979e-01 -1.21174431e+00 -1.30969539e-01 6.22693241e-01 1.06033444e+00 7.28055120e-01 -1.30491838e-01 5.56686521e-01 -8.34887803e-01 7.24887013e-01 -1.42094165e-01 -4.57309633e-01 2.40913391e-01 -8.27968180e-01 4.51615781e-01 7.19982028e-01 -2.85701096e-01 -1.05301511e+00 -5.61130464e-01 -5.50956905e-01 -5.11269152e-01 -2.04060927e-01 8.18010420e-02 -1.42043918e-01 -4.93284941e-01 8.71061087e-01 2.33811200e-01 1.47352666e-02 -3.12229663e-01 5.87017596e-01 2.68412143e-01 7.14478135e-01 -4.20099199e-01 1.05407667e+00 5.04076064e-01 2.91346788e-01 -4.54192907e-01 -7.16725826e-01 -7.89692774e-02 -4.62080121e-01 -4.78469171e-02 1.03332448e+00 -8.05947423e-01 -6.10784173e-01 5.69470704e-01 -8.34377944e-01 -2.62017727e-01 -6.01224482e-01 4.84865606e-01 -7.93044567e-01 3.99403244e-01 -3.87563139e-01 -6.17739439e-01 -5.66971362e-01 -1.07625568e+00 8.91243696e-01 3.82973760e-01 1.37235790e-01 -1.26674807e+00 3.19811583e-01 2.40287468e-01 9.76001203e-01 8.28598559e-01 9.64479446e-01 -8.93808156e-02 -3.53050977e-01 2.81559024e-02 -3.28152955e-01 9.10460353e-01 -7.96793029e-02 -1.69214457e-01 -1.04580045e+00 -4.02859628e-01 2.03003958e-01 -6.20008051e-01 7.87263811e-01 3.94197941e-01 1.07821155e+00 -2.59955734e-01 3.26016396e-01 9.92306769e-01 1.75535071e+00 1.07907459e-01 1.28611457e+00 1.49673909e-01 6.66804135e-01 5.03401794e-02 2.11925939e-01 1.85037822e-01 -2.81110313e-02 7.25677788e-01 5.81034780e-01 -1.74884275e-01 -5.65282226e-01 -1.54700100e-01 3.49904090e-01 7.64253378e-01 -4.89865959e-01 -4.56308335e-01 -5.27343214e-01 8.97624195e-02 -1.57860088e+00 -1.11205399e+00 8.65835845e-02 2.26698899e+00 7.36148894e-01 9.44145471e-02 2.19522625e-01 1.85731843e-01 4.41221982e-01 2.67587781e-01 -6.12477779e-01 -4.57633972e-01 -4.74144459e-01 5.24393260e-01 3.38247716e-01 3.93847585e-01 -9.75882053e-01 5.72172344e-01 6.37771845e+00 1.06272757e+00 -1.03117406e+00 4.10815746e-01 8.93103778e-01 1.98591799e-01 -3.63548428e-01 -3.01607281e-01 -2.02725247e-01 5.93363225e-01 8.32466304e-01 6.23714402e-02 6.78398609e-01 8.06291759e-01 -2.20762938e-01 -2.70427633e-02 -1.08831167e+00 1.42317104e+00 2.99978375e-01 -1.06936848e+00 1.32848933e-01 1.48128659e-01 9.17132556e-01 -3.33549790e-02 6.59055829e-01 2.51423866e-01 3.20688069e-01 -1.56412899e+00 4.66884106e-01 8.17835331e-01 1.01516831e+00 -9.85974014e-01 8.66200268e-01 -3.24545652e-02 -8.21370304e-01 3.92673105e-01 -6.18293583e-01 3.97397280e-01 1.11931138e-01 6.38262510e-01 -4.00660723e-01 5.33312559e-01 7.00806379e-01 4.92536873e-01 -7.35510349e-01 1.03152585e+00 -3.04667622e-01 5.65800130e-01 1.04796991e-01 2.12157205e-01 5.09759426e-01 -3.88417214e-01 7.60618329e-01 9.51736867e-01 2.77319342e-01 -2.24972859e-01 -4.23302740e-01 1.18130779e+00 -2.79204279e-01 -1.31122828e-01 -4.49196607e-01 5.00541925e-02 -2.14159787e-01 1.35203373e+00 -5.22624731e-01 -1.57802731e-01 -1.13093130e-01 1.29920888e+00 2.65461922e-01 3.33276629e-01 -1.03021586e+00 -3.58012110e-01 6.42558455e-01 -1.29521772e-01 2.54118681e-01 -5.27676754e-02 -1.25876173e-01 -1.26336765e+00 -2.70811826e-01 -1.08452523e+00 1.20238796e-01 -1.10000873e+00 -1.66346204e+00 8.99976015e-01 -2.87823081e-01 -1.39718926e+00 -1.53692991e-01 -5.05364895e-01 -7.88909614e-01 1.02969897e+00 -1.54155874e+00 -1.02889323e+00 -5.30353665e-01 8.20206523e-01 2.41334334e-01 -4.67334926e-01 8.82260799e-01 2.96616137e-01 -1.23287201e-01 8.10794473e-01 3.23006749e-01 -3.70817631e-02 6.41863286e-01 -1.68038666e+00 2.06466883e-01 8.48752797e-01 3.58548969e-01 1.91886425e-01 8.48264217e-01 -2.41736099e-01 -1.11054063e+00 -1.01473200e+00 1.74096003e-01 -5.47701716e-01 3.38091373e-01 -2.92658806e-02 -8.94310892e-01 2.43722290e-01 6.14941061e-01 6.89690607e-03 5.70938826e-01 -4.62986797e-01 -3.21540475e-01 -1.63723797e-01 -1.57364547e+00 3.73900205e-01 9.01909471e-01 -4.80150372e-01 -4.87844288e-01 2.41941437e-01 4.83262390e-01 -1.94824487e-01 -1.06925702e+00 2.97792047e-01 3.70371580e-01 -1.35910153e+00 1.16641653e+00 -4.29963112e-01 4.34386849e-01 -3.28654319e-01 -1.67057306e-01 -1.72016799e+00 -4.85667259e-01 -6.55772746e-01 -1.01076782e-01 1.22869897e+00 -6.74988925e-02 -5.55783510e-01 6.43584430e-01 2.27700815e-01 -2.10307520e-02 -7.33691514e-01 -1.03687441e+00 -8.19293261e-01 3.02466035e-01 -1.70020983e-01 5.84953904e-01 5.69665372e-01 -6.28222764e-01 3.13332707e-01 -7.71556497e-01 -1.38582736e-01 9.90507245e-01 -2.14550838e-01 7.17038929e-01 -1.13662958e+00 -4.46480632e-01 -6.70986295e-01 -7.65623927e-01 -7.28883982e-01 -8.21604431e-02 -8.94977868e-01 -1.75380602e-01 -1.33629036e+00 8.18653479e-02 -3.99802208e-01 -6.10452414e-01 4.08852659e-03 -6.45300150e-02 7.27698207e-01 2.39579171e-01 2.10250646e-01 -5.89015365e-01 9.82650042e-01 1.55702674e+00 -2.44090542e-01 2.21682787e-01 -1.08915428e-02 -5.68309665e-01 4.95925814e-01 6.79700434e-01 -6.29587412e-01 -4.06684816e-01 -2.78937042e-01 5.80331206e-01 -1.07746184e-01 5.58886766e-01 -1.52251458e+00 -7.82568678e-02 1.04680277e-01 5.78446269e-01 -2.48438805e-01 4.06461060e-01 -6.81177914e-01 3.59452039e-01 4.51983064e-01 -3.79720509e-01 1.44345090e-01 -2.16946408e-01 5.40887773e-01 -5.61641872e-01 -2.31575280e-01 1.30173934e+00 -2.24662989e-01 -4.63203222e-01 4.48430955e-01 1.81662887e-01 3.92177731e-01 8.48151088e-01 1.36850439e-02 -3.40982527e-01 -4.27731395e-01 -5.05823731e-01 -2.93776870e-01 6.20510221e-01 1.35144219e-01 7.23733664e-01 -1.64361191e+00 -1.00845861e+00 3.48529071e-02 -2.44014766e-02 -2.62595564e-01 1.39372185e-01 7.60386467e-01 -8.21708202e-01 1.25735074e-01 -7.29035676e-01 -4.59259182e-01 -9.50426102e-01 5.17942667e-01 4.94676650e-01 -6.08568907e-01 -4.37404186e-01 1.08724582e+00 1.87054485e-01 -1.09260432e-01 9.39112827e-02 1.30190551e-02 -8.43299478e-02 -2.72765905e-01 5.38911641e-01 6.75051689e-01 7.90989697e-02 -6.59096122e-01 -2.30057284e-01 6.51235104e-01 3.46729755e-01 -1.80698648e-01 1.39164388e+00 -1.26928806e-01 -3.08114290e-03 3.01342458e-01 1.36447680e+00 -2.20750511e-01 -1.31887019e+00 -1.19610511e-01 -3.71444941e-01 -4.95306939e-01 2.05292806e-01 -1.09076881e+00 -1.29981279e+00 1.13238382e+00 1.27821469e+00 4.96962696e-01 1.33055651e+00 -1.80146933e-01 7.93471754e-01 2.64072195e-02 3.65844429e-01 -9.49214756e-01 5.13410330e-01 2.31832311e-01 1.16477847e+00 -1.49074244e+00 -2.01011762e-01 1.55377850e-01 -8.68536592e-01 8.82454813e-01 3.87514979e-01 -5.73392630e-01 4.76799965e-01 -2.34727599e-02 1.20009538e-02 5.38770668e-02 -3.08317930e-01 -2.04010561e-01 5.32331228e-01 9.39916790e-01 4.50122386e-01 -1.03441477e-02 -2.57794321e-01 -5.72592579e-02 -2.70962358e-01 -2.02700064e-01 3.00651371e-01 2.77582020e-01 -1.55462489e-01 -1.19555008e+00 -1.97045535e-01 4.31177914e-01 -5.69387138e-01 -1.42350778e-01 1.31244957e-01 5.59001803e-01 3.43849391e-01 7.03100801e-01 -1.30644128e-01 -3.31344068e-01 2.16124672e-02 -2.70777252e-02 8.04458141e-01 -2.13680238e-01 -6.58975124e-01 -3.63517106e-01 -4.28438962e-01 -6.39138579e-01 -7.90125787e-01 -8.33033249e-02 -7.22065032e-01 -3.99228215e-01 -7.95733854e-02 1.93419755e-01 8.18304777e-01 6.76092744e-01 1.64724231e-01 5.50229609e-01 6.98950946e-01 -8.65031779e-01 -4.93338853e-01 -1.12655282e+00 -6.74417794e-01 8.80444765e-01 1.69133499e-01 -7.25168467e-01 -4.81220782e-01 4.76930551e-02]
[11.676581382751465, -0.5857764482498169]
f6fe793b-8aed-4254-b8ec-7b481ba19c4b
fdnerf-semantics-driven-face-reconstruction
2306.00783
null
https://arxiv.org/abs/2306.00783v1
https://arxiv.org/pdf/2306.00783v1.pdf
FDNeRF: Semantics-Driven Face Reconstruction, Prompt Editing and Relighting with Diffusion Models
The ability to create high-quality 3D faces from a single image has become increasingly important with wide applications in video conferencing, AR/VR, and advanced video editing in movie industries. In this paper, we propose Face Diffusion NeRF (FDNeRF), a new generative method to reconstruct high-quality Face NeRFs from single images, complete with semantic editing and relighting capabilities. FDNeRF utilizes high-resolution 3D GAN inversion and expertly trained 2D latent-diffusion model, allowing users to manipulate and construct Face NeRFs in zero-shot learning without the need for explicit 3D data. With carefully designed illumination and identity preserving loss, as well as multi-modal pre-training, FD-NeRF offers users unparalleled control over the editing process enabling them to create and edit face NeRFs using just single-view images, text prompts, and explicit target lighting. The advanced features of FDNeRF have been designed to produce more impressive results than existing 2D editing approaches that rely on 2D segmentation maps for editable attributes. Experiments show that our FDNeRF achieves exceptionally realistic results and unprecedented flexibility in editing compared with state-of-the-art 3D face reconstruction and editing methods. Our code will be available at https://github.com/BillyXYB/FDNeRF.
['Tai Chi-Keung Tang', 'Yu-Wing', 'Tianyuan Dai', 'Yanbo Xu', 'Hao Zhang']
2023-06-01
null
null
null
null
['video-editing', '3d-face-reconstruction', 'face-reconstruction']
['computer-vision', 'computer-vision', 'computer-vision']
[ 2.43977115e-01 3.46617401e-01 3.40513617e-01 -4.92244124e-01 -7.08084166e-01 -5.99817216e-01 6.32635236e-01 -7.55424678e-01 6.39802888e-02 4.59403396e-01 1.87588573e-01 9.26891863e-02 1.65142149e-01 -6.59294307e-01 -7.25646675e-01 -2.51748413e-01 2.97371238e-01 7.10758030e-01 -3.28887969e-01 -2.84184694e-01 -1.02659188e-01 8.26968491e-01 -1.56896973e+00 5.37841115e-03 8.41768920e-01 8.45549643e-01 2.75381267e-01 8.15236509e-01 -2.06191540e-01 3.45476329e-01 -3.07361752e-01 -7.89906859e-01 4.94612157e-01 -4.38551486e-01 -3.57483804e-01 3.53021055e-01 7.93045282e-01 -7.81872928e-01 -3.22681099e-01 6.71622753e-01 7.91556418e-01 1.56363621e-01 6.53301001e-01 -1.14399695e+00 -9.45631802e-01 -3.99384648e-02 -6.32811308e-01 -3.91990811e-01 7.43437767e-01 1.55112624e-01 5.11489928e-01 -1.19181252e+00 1.00512338e+00 1.43586648e+00 7.38088846e-01 1.15180755e+00 -1.41602671e+00 -7.64335215e-01 -1.29471868e-01 -2.90168881e-01 -1.36688697e+00 -1.05630183e+00 7.61970460e-01 -4.46768820e-01 7.42214918e-01 2.56607413e-01 7.87991107e-01 1.30556262e+00 -2.37578489e-02 4.97103930e-01 1.02304769e+00 -4.48520273e-01 -1.66846476e-02 1.09307811e-01 -7.00622261e-01 1.05681109e+00 -3.31975192e-01 1.83872983e-01 -6.41689599e-01 -1.95109695e-02 1.41752434e+00 1.35722989e-02 -4.23590094e-01 -4.24161851e-01 -8.29264224e-01 6.17320657e-01 -3.85603197e-02 -2.49623388e-01 -3.31053436e-01 2.63343215e-01 -1.59844384e-02 3.65122795e-01 8.59987915e-01 1.96214154e-01 -1.11323908e-01 -3.28569978e-01 -1.01478958e+00 5.53175099e-02 5.82530499e-01 1.27468383e+00 5.96027076e-01 3.62572670e-01 -6.21983875e-03 1.08875573e+00 1.52825028e-01 7.03635752e-01 -1.61795959e-01 -1.59917331e+00 -1.32253226e-02 2.19302066e-02 9.61397886e-02 -7.17926085e-01 1.83468714e-01 1.34114861e-01 -6.79015279e-01 6.70058429e-01 -5.19325994e-02 -9.11695361e-02 -1.33594930e+00 1.73577487e+00 6.39795482e-01 2.70935476e-01 -2.32510999e-01 7.37199426e-01 8.55626166e-01 5.83856761e-01 -4.19164509e-01 -1.49225861e-01 9.82458353e-01 -7.09630787e-01 -7.94451714e-01 6.92317262e-02 -4.78091463e-02 -9.31296885e-01 1.38707948e+00 4.07828182e-01 -1.42238140e+00 -3.56138319e-01 -6.38607442e-01 -4.74895805e-01 9.05806944e-02 -1.26584873e-01 6.71732903e-01 8.17212880e-01 -1.47932053e+00 5.29236794e-01 -7.86225617e-01 -3.75081599e-01 8.83426070e-01 3.46822441e-01 -7.27846801e-01 -4.26524907e-01 -6.94574177e-01 7.77475238e-01 -3.84060919e-01 8.68336856e-02 -1.19506061e+00 -1.04434454e+00 -9.92038548e-01 -1.46643326e-01 3.34483892e-01 -9.45501149e-01 1.26527107e+00 -8.28718245e-01 -2.03752160e+00 1.12927806e+00 -2.78971523e-01 2.69612134e-01 7.80371428e-01 -3.23023379e-01 -2.16352239e-01 2.51871049e-01 9.95825529e-02 9.96862233e-01 1.40154147e+00 -1.49480867e+00 6.38254583e-02 -3.27717930e-01 -1.14707602e-02 3.69367570e-01 -1.78685397e-01 -4.42150533e-02 -7.64491439e-01 -7.92551219e-01 -2.51999974e-01 -6.94360077e-01 1.10823385e-01 7.08767593e-01 -2.57479131e-01 2.33798593e-01 1.20786428e+00 -8.93528879e-01 5.07491529e-01 -2.32917500e+00 3.84059370e-01 2.22659230e-01 4.34933662e-01 1.72504514e-01 -3.28160673e-01 2.39495650e-01 9.50326100e-02 1.05278067e-01 -3.26321810e-01 -1.06039011e+00 9.18939263e-02 2.77809203e-01 -1.52039841e-01 2.43886262e-01 1.24257624e-01 9.15487170e-01 -6.63158715e-01 -4.35778558e-01 4.27131206e-01 1.19389164e+00 -7.21626401e-01 3.91454339e-01 -1.93643928e-01 8.32190931e-01 -1.15583926e-01 8.56907427e-01 7.41525054e-01 -2.32479684e-02 -2.13067327e-02 -1.31697580e-01 1.25979766e-01 -3.88749719e-01 -9.92272198e-01 2.21975899e+00 -7.86026359e-01 5.34429610e-01 5.28243780e-01 -2.59613991e-01 8.06746662e-01 3.34530413e-01 3.93007815e-01 -6.27796233e-01 1.14674978e-01 1.23666100e-01 -7.65539169e-01 -3.15249741e-01 4.81995672e-01 -3.45809549e-01 1.61622807e-01 5.55508137e-01 2.65453786e-01 -7.79204130e-01 -1.62825629e-01 4.83366936e-01 7.51705766e-01 5.59008300e-01 -2.79145718e-01 5.00141382e-02 -9.02810786e-03 -5.93913496e-01 4.80123997e-01 4.02788967e-01 3.77130419e-01 1.09496844e+00 1.38802260e-01 -3.44363339e-02 -1.29004371e+00 -1.44589126e+00 -4.19293158e-02 6.94079876e-01 -6.22167587e-02 -2.94503301e-01 -9.25350547e-01 -5.07345676e-01 -4.56910999e-03 7.92926550e-01 -6.01124763e-01 1.60483271e-02 -3.39636862e-01 -5.34393042e-02 4.83428836e-01 3.76700461e-01 5.28410673e-01 -7.96705306e-01 -2.40718618e-01 -1.09493837e-01 -1.26064003e-01 -1.03007555e+00 -1.11120212e+00 -3.66924256e-01 -6.77267075e-01 -7.09562242e-01 -1.00766218e+00 -7.53078997e-01 9.37700033e-01 1.41072512e-01 1.09024644e+00 3.00418884e-02 -6.18420780e-01 1.00159824e+00 -2.82420665e-01 -2.08166197e-01 -5.43987989e-01 -3.27378750e-01 1.93082318e-01 -1.13016553e-02 -2.78211832e-01 -1.07065511e+00 -5.40264010e-01 2.58290440e-01 -8.09606671e-01 2.75030851e-01 2.34206796e-01 7.25203454e-01 6.69720054e-01 -3.22950780e-01 4.44090575e-01 -1.06092143e+00 4.84663248e-01 -7.22952536e-05 -5.33722878e-01 1.22587055e-01 -4.87168968e-01 -1.61803007e-01 4.32116866e-01 -4.87435490e-01 -1.56827927e+00 -5.00577427e-02 -4.13741887e-01 -1.01088810e+00 -1.02550544e-01 -1.92787826e-01 -4.31475610e-01 -3.75795454e-01 5.97653270e-01 2.03133374e-02 3.38256329e-01 -5.58453262e-01 7.57064521e-01 5.62858224e-01 6.80257618e-01 -6.25440955e-01 8.76747191e-01 5.31906188e-01 -2.39484876e-01 -9.18058813e-01 -4.29419816e-01 2.38917202e-01 -5.72997391e-01 -4.29022074e-01 9.03838992e-01 -9.66231167e-01 -5.17298996e-01 6.56522155e-01 -1.00322592e+00 -6.89808846e-01 -6.92837477e-01 1.21978827e-01 -7.52728105e-01 1.85361758e-01 -7.10900486e-01 -6.45410001e-01 -3.57919663e-01 -1.10053980e+00 1.34870374e+00 2.11994290e-01 -3.08413476e-01 -8.67977858e-01 -2.09189013e-01 6.01258755e-01 4.87623572e-01 4.84856486e-01 7.62162149e-01 2.85912424e-01 -7.31488347e-01 8.45991671e-02 4.65840893e-03 3.97628754e-01 3.52553964e-01 1.17453367e-01 -1.09431684e+00 -3.64827186e-01 -1.74453869e-01 -4.14997041e-01 4.71434325e-01 3.48753154e-01 1.12990308e+00 -1.49446487e-01 -1.54590130e-01 1.13523173e+00 1.16097093e+00 1.83034629e-01 8.13412905e-01 -4.20215726e-01 1.03656960e+00 2.84687608e-01 3.09310406e-01 5.07022262e-01 4.08603609e-01 6.69065416e-01 1.99677929e-01 -1.76259965e-01 -4.62677419e-01 -5.19632697e-01 3.50347370e-01 7.18707085e-01 -3.75458360e-01 -2.06344128e-01 -5.15514731e-01 2.19117865e-01 -1.23022068e+00 -9.83038843e-01 3.47586989e-01 2.08813620e+00 7.98377752e-01 -3.18115175e-01 -1.77162662e-01 -2.68502682e-01 7.52688408e-01 4.21169735e-02 -7.70088613e-01 -5.43978274e-01 7.87836872e-03 6.97336197e-01 1.00664675e-01 8.29601467e-01 -6.55599594e-01 1.06318390e+00 6.01996994e+00 8.18789959e-01 -1.07621384e+00 4.57582355e-01 6.17157340e-01 -4.89787012e-01 -8.29931080e-01 -2.65843987e-01 -4.54038084e-01 2.36186296e-01 4.12963361e-01 1.08030260e-01 8.67528677e-01 4.55547154e-01 1.80161163e-01 4.59277406e-02 -1.00919938e+00 1.37794054e+00 3.89779866e-01 -1.49181271e+00 1.15983769e-01 2.85938859e-01 8.77536714e-01 -3.69981229e-01 2.71283478e-01 -2.15679742e-02 3.80275607e-01 -1.20806313e+00 8.51962209e-01 8.24096024e-01 1.84514630e+00 -8.43344092e-01 -3.04211471e-02 -2.49046199e-02 -9.11643744e-01 1.95615187e-01 -1.22856192e-01 3.48942518e-01 5.82907736e-01 6.67881310e-01 -5.39842248e-01 4.97127593e-01 7.52285600e-01 6.38103426e-01 -3.83059531e-02 3.84470165e-01 -2.26653576e-01 1.06352702e-01 -4.95976448e-01 5.84143579e-01 -4.43918854e-01 -4.13114458e-01 6.38487220e-01 6.25371933e-01 7.30179191e-01 5.43043971e-01 -1.33562118e-01 9.33697879e-01 -4.68676239e-01 -1.84432894e-01 -7.05591500e-01 -1.90907136e-01 5.67595422e-01 1.22758055e+00 -5.30813873e-01 -7.06742480e-02 -1.18720651e-01 1.77118635e+00 2.37747326e-01 3.22283030e-01 -8.84704351e-01 -2.95987546e-01 8.68083537e-01 4.61979806e-01 3.95934641e-01 -4.37494636e-01 -1.04712406e-02 -1.26214659e+00 5.93005605e-02 -7.03512430e-01 -1.77625746e-01 -1.13404298e+00 -1.28884685e+00 7.92128265e-01 -1.02054134e-01 -8.86138916e-01 -2.27147684e-01 -2.96657383e-01 -5.75110495e-01 7.30488122e-01 -1.23159254e+00 -1.59712863e+00 -4.79510307e-01 8.72844040e-01 5.31738281e-01 -2.30366021e-01 9.82600093e-01 5.25280476e-01 -2.49768078e-01 8.88052642e-01 -7.57420063e-02 -1.63663328e-01 8.96664023e-01 -9.12310779e-01 5.79845130e-01 6.06766641e-01 1.36339322e-01 3.93730611e-01 4.28231627e-01 -7.62687027e-01 -1.57484198e+00 -9.27373707e-01 3.60822976e-01 -7.00039864e-01 -1.85961485e-01 -6.91461921e-01 -5.62958956e-01 8.56093228e-01 2.97926992e-01 6.24797530e-02 7.05259144e-01 -1.38880640e-01 -2.50735670e-01 -1.29633129e-01 -1.68479145e+00 6.32070899e-01 1.72976410e+00 -7.36269832e-01 -1.62602007e-01 1.53654888e-01 6.93091035e-01 -7.82862604e-01 -9.40747559e-01 2.46227756e-01 6.33857489e-01 -9.79146779e-01 9.54436362e-01 -5.58028147e-02 4.34896618e-01 -1.26921818e-01 -9.50636044e-02 -1.43573952e+00 -2.00904429e-01 -1.11172593e+00 -6.15638942e-02 1.50594401e+00 9.60175619e-02 -5.45227170e-01 6.82768941e-01 8.36293340e-01 -3.71795475e-01 -6.80110991e-01 -7.47931898e-01 -4.20605659e-01 -2.98796862e-01 -2.53824770e-01 7.19055235e-01 1.08485341e+00 -6.83607221e-01 1.11220643e-01 -6.96347296e-01 1.01270385e-01 8.90501142e-01 -4.18033227e-02 9.61543560e-01 -1.21381450e+00 -2.45531499e-01 4.94182968e-05 -1.78983361e-01 -1.03513110e+00 2.63240665e-01 -8.96237552e-01 -1.30734235e-01 -1.59244275e+00 -1.49633855e-01 -5.66428900e-01 5.71215093e-01 5.22360563e-01 2.37283096e-01 6.50721610e-01 2.43652835e-01 2.55101342e-02 -3.30339931e-02 9.19985175e-01 1.54172969e+00 1.42285734e-01 -1.83645621e-01 -3.00578564e-01 -7.18831897e-01 6.10819936e-01 5.16245067e-01 -2.13623106e-01 -7.80589700e-01 -8.23272586e-01 -6.56705350e-02 1.35977209e-01 4.21979457e-01 -6.69150889e-01 2.93817893e-02 -7.86423013e-02 8.08167219e-01 -1.53133050e-01 9.59998071e-01 -7.10663676e-01 7.82598078e-01 -2.08441451e-01 2.97843516e-02 2.97435131e-02 1.22950211e-01 4.51014578e-01 1.19192414e-01 1.38917178e-01 9.16273117e-01 -3.48215282e-01 -6.76782489e-01 8.13743234e-01 -6.68035671e-02 9.89264771e-02 1.07396376e+00 -5.09132624e-01 3.43500488e-02 -7.66805828e-01 -1.07148445e+00 2.48152837e-02 1.07721007e+00 4.64648038e-01 1.10575879e+00 -1.44158173e+00 -6.48015976e-01 7.03402519e-01 -2.98801780e-01 3.11476767e-01 7.08289444e-01 3.70526612e-01 -6.61384165e-01 -4.37153906e-01 -3.00386667e-01 -5.48034132e-01 -1.37368047e+00 1.05479211e-01 3.05857062e-01 3.13686371e-01 -9.24033821e-01 1.17546761e+00 9.56154540e-02 -8.10857415e-01 1.31706176e-02 3.62626642e-01 4.90468889e-01 -1.31433636e-01 3.88252854e-01 1.47787467e-01 -6.01825453e-02 -4.63818550e-01 -8.12894553e-02 7.59394586e-01 -1.20129794e-01 -4.24890280e-01 1.52886009e+00 -2.94083148e-01 2.04532351e-02 9.94826555e-02 1.07119823e+00 3.22008997e-01 -1.72984266e+00 7.50182420e-02 -9.78568673e-01 -1.06426620e+00 7.33123645e-02 -7.71094859e-01 -1.36731720e+00 8.25415373e-01 5.40706217e-01 -4.65201944e-01 1.25560462e+00 7.33801052e-02 1.01481390e+00 -1.28182963e-01 6.14152908e-01 -9.70956564e-01 3.25450182e-01 2.69750357e-01 1.05063069e+00 -8.95166397e-01 -8.11226591e-02 -6.21417642e-01 -8.19859326e-01 8.44050646e-01 6.17203832e-01 5.59757352e-02 6.48355484e-01 5.49755454e-01 4.05355953e-02 -1.91796809e-01 -4.02452737e-01 2.58449852e-01 1.04521671e-02 8.74261975e-01 9.13604945e-02 -9.64561328e-02 4.52696472e-01 1.78850174e-01 -3.30461770e-01 1.38223484e-01 4.80106533e-01 7.87830949e-01 8.13497156e-02 -1.31561446e+00 -2.53676385e-01 5.18071949e-01 -1.40552878e-01 -6.31115288e-02 -1.67600498e-01 5.01463830e-01 6.65332526e-02 5.20262778e-01 2.30207726e-01 -1.70249820e-01 3.42321247e-01 1.04436375e-01 1.06156445e+00 -7.43634343e-01 -1.49597749e-01 7.63509050e-02 1.33591935e-01 -7.72617519e-01 -2.78699756e-01 -6.19831979e-01 -1.16768277e+00 -7.02041864e-01 -2.01374814e-01 -1.66206315e-01 6.27021790e-01 5.90415835e-01 9.22938824e-01 2.71305472e-01 7.84462869e-01 -1.33547616e+00 -5.40828379e-03 -5.63744009e-01 -6.81287348e-01 3.06865931e-01 2.76274085e-01 -8.02303672e-01 -1.74018115e-01 3.23435426e-01]
[12.695623397827148, -0.3709513247013092]
9d3b4342-66db-43a0-b8a9-fef812b9349b
high-fidelity-3d-face-generation-from-natural
2305.03302
null
https://arxiv.org/abs/2305.03302v1
https://arxiv.org/pdf/2305.03302v1.pdf
High-Fidelity 3D Face Generation from Natural Language Descriptions
Synthesizing high-quality 3D face models from natural language descriptions is very valuable for many applications, including avatar creation, virtual reality, and telepresence. However, little research ever tapped into this task. We argue the major obstacle lies in 1) the lack of high-quality 3D face data with descriptive text annotation, and 2) the complex mapping relationship between descriptive language space and shape/appearance space. To solve these problems, we build Describe3D dataset, the first large-scale dataset with fine-grained text descriptions for text-to-3D face generation task. Then we propose a two-stage framework to first generate a 3D face that matches the concrete descriptions, then optimize the parameters in the 3D shape and texture space with abstract description to refine the 3D face model. Extensive experimental results show that our method can produce a faithful 3D face that conforms to the input descriptions with higher accuracy and quality than previous methods. The code and Describe3D dataset are released at https://github.com/zhuhao-nju/describe3d .
['Xun Cao', 'Yuanxun Lu', 'Yiyu Zhuang', 'Linjia Huang', 'Hao Zhu', 'Menghua Wu']
2023-05-05
null
http://openaccess.thecvf.com//content/CVPR2023/html/Wu_High-Fidelity_3D_Face_Generation_From_Natural_Language_Descriptions_CVPR_2023_paper.html
http://openaccess.thecvf.com//content/CVPR2023/papers/Wu_High-Fidelity_3D_Face_Generation_From_Natural_Language_Descriptions_CVPR_2023_paper.pdf
cvpr-2023-1
['face-model', 'face-generation', 'text-to-3d', 'text-annotation']
['computer-vision', 'computer-vision', 'computer-vision', 'natural-language-processing']
[-1.00828484e-01 1.37423009e-01 4.41102907e-02 -6.91479146e-01 -4.29697484e-01 -3.74551117e-01 7.17046797e-01 -5.57884932e-01 4.38826948e-01 4.36770320e-01 3.93572658e-01 3.93188335e-02 2.71257579e-01 -7.30868399e-01 -4.31399912e-01 -2.31809631e-01 5.06087363e-01 8.54030728e-01 -4.52157073e-02 -2.47274593e-01 9.93317738e-02 7.97949135e-01 -1.83029485e+00 1.20640300e-01 5.83648205e-01 1.06324983e+00 3.66259031e-02 2.37554297e-01 -4.82185930e-01 1.26653895e-01 -3.49441797e-01 -6.25369966e-01 2.32707947e-01 -6.05712235e-01 -6.00075424e-01 5.75226605e-01 3.32463473e-01 -3.34697396e-01 -6.33706078e-02 8.93424451e-01 5.39565027e-01 -5.01908101e-02 8.56366456e-01 -1.45625067e+00 -9.04584110e-01 5.30996248e-02 -5.97088993e-01 -7.15930343e-01 6.70693934e-01 1.42225586e-02 5.69548368e-01 -1.24930656e+00 8.91598344e-01 1.70722771e+00 4.89673913e-01 1.18811166e+00 -1.05380929e+00 -7.44325042e-01 -3.89745124e-02 -3.24283898e-01 -1.72914588e+00 -8.64419878e-01 9.92758930e-01 -4.78841543e-01 4.88842010e-01 1.90293461e-01 8.33261967e-01 1.24893105e+00 -1.74369529e-01 5.91701508e-01 9.25006747e-01 -4.02800977e-01 -1.07490784e-02 3.36425781e-01 -5.61001003e-01 1.07826757e+00 -4.64614891e-02 -3.70369628e-02 -6.41021132e-01 -2.49947175e-01 1.01973689e+00 -1.31886154e-01 -5.84863126e-02 -3.96001995e-01 -9.10569012e-01 6.80608630e-01 -4.23287079e-02 5.13892174e-02 -1.85950503e-01 -8.48251507e-02 2.45704278e-01 1.89149622e-02 8.46970022e-01 6.17958456e-02 -3.61418843e-01 -1.75464034e-01 -7.28405178e-01 5.66032946e-01 7.42444217e-01 1.48792732e+00 7.00446010e-01 6.75889403e-02 1.54391229e-01 1.07872486e+00 6.45863712e-01 8.08530569e-01 2.17255071e-01 -9.01588619e-01 2.01207533e-01 6.39561713e-01 -8.54557753e-02 -9.69970226e-01 -2.02215046e-01 2.94976681e-01 -8.74581575e-01 1.03785254e-01 5.34779988e-02 1.26256511e-01 -7.78401434e-01 1.53819060e+00 6.84438527e-01 -4.50227112e-02 -1.05323605e-01 9.75776315e-01 1.38918877e+00 5.73939562e-01 -3.78934652e-01 -5.60683534e-02 1.44319677e+00 -6.48842096e-01 -7.71322966e-01 -5.52791506e-02 2.87888169e-01 -9.59579349e-01 1.38149381e+00 -8.85728225e-02 -1.11167812e+00 -5.13529778e-01 -7.83925653e-01 -3.18032026e-01 -9.25818831e-02 2.70335734e-01 4.91944015e-01 7.95422077e-01 -1.01214194e+00 2.47219533e-01 -4.81866717e-01 -4.91895407e-01 6.72598600e-01 2.27011457e-01 -7.09962964e-01 -8.69859883e-04 -8.75548303e-01 6.50783479e-01 1.00837030e-01 -1.88807852e-03 -6.40106022e-01 -6.60572469e-01 -1.11684227e+00 -3.92095119e-01 4.06979889e-01 -8.00669730e-01 1.20469463e+00 -6.94858789e-01 -1.80752432e+00 1.42168355e+00 -4.17910516e-01 4.02526051e-01 5.67592025e-01 1.09124519e-01 -1.99901760e-01 -2.56404370e-01 1.92128360e-01 8.06698859e-01 9.19551075e-01 -1.62350035e+00 -1.45716891e-01 -6.58942163e-01 -1.95978940e-01 1.39812961e-01 -7.58165196e-02 3.90123092e-02 -1.05331981e+00 -5.61932027e-01 2.46520057e-01 -8.55745554e-01 1.48373216e-01 5.47263682e-01 -5.50393999e-01 -4.22726780e-01 8.61259222e-01 -3.59974712e-01 7.37893164e-01 -2.15673852e+00 8.95148963e-02 1.67598560e-01 2.29508713e-01 1.03300922e-01 -2.13079721e-01 3.46390158e-01 1.80126861e-01 4.32726651e-01 -1.38475984e-01 -8.43329072e-01 1.94914401e-01 5.63838705e-02 -2.63136595e-01 2.75821626e-01 3.45211387e-01 7.84877419e-01 -6.52425408e-01 -7.66106725e-01 1.44355103e-01 8.30688596e-01 -6.65372074e-01 4.42133784e-01 -3.95441979e-01 5.86040020e-01 -7.38982081e-01 8.88747692e-01 8.37193668e-01 -4.30752002e-02 -3.84933129e-02 -2.77672350e-01 2.91179735e-02 3.02401967e-02 -1.08690035e+00 1.86021888e+00 -3.68196875e-01 3.92630488e-01 1.59740210e-01 -3.63356501e-01 1.46448803e+00 3.11453581e-01 4.67007220e-01 -5.92546880e-01 2.87236601e-01 3.44765306e-01 -6.15563333e-01 -6.03421509e-01 4.49990332e-01 -5.31974077e-01 -9.91535112e-02 5.65938294e-01 2.21939329e-02 -9.57217097e-01 -1.35963127e-01 -4.96625155e-03 2.88163304e-01 6.07211411e-01 1.35206163e-01 -1.68355405e-01 5.72372675e-01 -2.26913810e-01 6.83038414e-01 1.02376007e-01 -3.23003083e-02 9.74696100e-01 5.87809801e-01 -4.69860464e-01 -1.26443529e+00 -8.66593182e-01 -1.93094388e-01 4.80376184e-01 1.48164719e-01 -7.48820901e-01 -8.99442017e-01 -5.79040527e-01 4.04825248e-03 4.31675553e-01 -6.68556809e-01 3.59108001e-02 -3.25011820e-01 -2.12244228e-01 6.09669805e-01 2.00834408e-01 5.72266936e-01 -9.48556602e-01 -8.17473903e-02 -2.67748505e-01 -1.76600605e-01 -1.24278402e+00 -8.22216988e-01 -6.26205802e-01 -6.67555988e-01 -9.77377236e-01 -5.95216513e-01 -9.22217250e-01 1.05882609e+00 -3.86564136e-02 1.13310349e+00 4.17164147e-01 -3.88491571e-01 1.94731310e-01 -4.20273751e-01 -5.11175871e-01 -7.16923892e-01 -2.24221453e-01 3.43742281e-01 4.39050868e-02 2.74572343e-01 -4.46463913e-01 -1.31802917e-01 4.75571990e-01 -8.66795182e-01 6.09839737e-01 8.13733786e-02 6.97181463e-01 8.22235882e-01 -4.36222516e-02 4.36311394e-01 -9.33713555e-01 5.39318979e-01 -1.34644747e-01 -6.33515179e-01 6.83007613e-02 -3.46775711e-01 7.29769170e-02 4.97260779e-01 -3.68489951e-01 -1.11931694e+00 1.93996146e-01 -6.01988196e-01 -5.33971190e-01 -3.68832827e-01 1.16307139e-01 -6.86313391e-01 1.62873641e-01 5.19531965e-01 2.94667721e-01 4.11585361e-01 -5.99325061e-01 3.45048696e-01 9.09883857e-01 2.80257285e-01 -9.00167525e-01 9.94241834e-01 5.91202676e-01 2.26792786e-02 -9.78851974e-01 -6.80499971e-01 1.37363568e-01 -9.79866147e-01 -2.58875668e-01 7.27307498e-01 -7.45176792e-01 -7.39365220e-01 4.88744646e-01 -1.34064174e+00 -2.86276996e-01 -1.27161667e-01 3.46382149e-02 -9.16402340e-01 1.73460305e-01 -2.72318095e-01 -8.34732831e-01 -3.29999834e-01 -1.19200647e+00 1.60349822e+00 1.84108973e-01 -2.47345686e-01 -7.67663121e-01 -9.24287736e-02 4.77291375e-01 1.73988461e-01 3.52112889e-01 1.03191626e+00 -1.76435150e-02 -2.82851934e-01 -9.24268663e-02 -1.89388081e-01 1.15885578e-01 4.38782454e-01 3.40358406e-01 -9.37681258e-01 3.81618701e-02 -1.74925297e-01 -4.22136784e-01 5.79194278e-02 1.06520802e-01 1.21405172e+00 -3.60006779e-01 -2.16433689e-01 8.12114298e-01 1.10548282e+00 1.00499608e-01 5.17244995e-01 -1.99045151e-01 7.80518353e-01 8.95442307e-01 5.71044385e-01 6.29232466e-01 6.37941420e-01 9.95104313e-01 2.15173349e-01 1.58310626e-02 -3.40108246e-01 -8.62169027e-01 1.30179822e-01 7.72558391e-01 2.18899120e-02 -6.21576458e-02 -9.53523695e-01 3.36472392e-01 -1.55754042e+00 -7.51463592e-01 6.19447455e-02 1.99758959e+00 8.39435518e-01 -1.37706146e-01 1.93106338e-01 -1.07979037e-01 6.35052860e-01 -8.48066583e-02 -4.87714291e-01 -2.44559169e-01 -2.47603077e-02 -1.42375886e-01 -2.84526706e-01 5.46221137e-01 -7.59789765e-01 1.26759028e+00 5.74412441e+00 7.54556894e-01 -1.11457181e+00 -6.06115237e-02 6.10247016e-01 -3.28685157e-02 -6.72421396e-01 -1.14254050e-01 -9.76793647e-01 3.10384661e-01 3.96505654e-01 -2.53610253e-01 3.62504601e-01 7.05609143e-01 4.25695509e-01 3.00502479e-01 -1.19944024e+00 1.50537539e+00 4.19202834e-01 -1.43959689e+00 4.63973939e-01 1.64382264e-01 7.43056655e-01 -5.19105136e-01 -9.05156285e-02 5.31101488e-02 -9.01188254e-02 -1.21323287e+00 1.14336312e+00 6.19132042e-01 1.39078748e+00 -6.93779945e-01 3.90674829e-01 4.79092926e-01 -1.24506593e+00 5.22577584e-01 -3.66035908e-01 3.07300806e-01 2.07621560e-01 3.82321239e-01 -8.59592497e-01 2.76411057e-01 6.50899053e-01 7.33988762e-01 -3.34927648e-01 5.50785005e-01 -1.24276869e-01 5.11560217e-02 -1.90124124e-01 -1.67980626e-01 -2.66295999e-01 -4.02646422e-01 5.63666284e-01 6.96108878e-01 5.25596440e-01 4.36632782e-01 3.62613760e-02 1.35336971e+00 -3.56334329e-01 2.95276642e-01 -9.78764594e-01 -2.16332957e-01 6.57747388e-01 1.04831362e+00 -5.64484417e-01 -6.34484291e-02 -3.22765827e-01 1.02997839e+00 1.42169595e-01 8.22789073e-02 -6.17965817e-01 -1.51545241e-01 7.57863343e-01 5.43142200e-01 -1.29595995e-01 -2.94725865e-01 -2.92750061e-01 -1.18308842e+00 2.58223504e-01 -8.79817903e-01 -1.23557009e-01 -1.05711484e+00 -1.37564480e+00 1.00325096e+00 5.20473011e-02 -1.25550866e+00 -2.98664272e-01 -5.63700914e-01 -2.61981934e-01 9.31477129e-01 -1.22209382e+00 -1.56906140e+00 -4.78656441e-01 5.85559785e-01 6.59725130e-01 -3.83642107e-01 1.06954682e+00 1.94961444e-01 -4.66363072e-01 6.50899410e-01 -3.51654798e-01 9.32692513e-02 6.49501622e-01 -6.78344607e-01 7.38203764e-01 2.64415950e-01 -5.40096387e-02 4.79909003e-01 5.60876608e-01 -7.23075092e-01 -1.63512456e+00 -9.81813669e-01 9.90864813e-01 -6.51564240e-01 2.14208499e-01 -8.08742702e-01 -6.47099137e-01 3.97346377e-01 -2.24148259e-01 5.29537797e-02 6.12675607e-01 -1.38512850e-01 -3.92550141e-01 2.90851183e-02 -1.50828600e+00 8.44652951e-01 1.37029386e+00 -6.55784011e-01 -2.30508685e-01 2.20514938e-01 5.60558736e-01 -7.31351495e-01 -9.46440160e-01 3.32098365e-01 7.07427740e-01 -9.49374974e-01 7.03057349e-01 -3.46063286e-01 4.67387944e-01 -3.28848392e-01 -2.23974466e-01 -1.13777995e+00 9.16325524e-02 -7.24238455e-01 1.54515624e-01 1.49611485e+00 3.05171490e-01 -3.71653736e-01 9.19765413e-01 7.40282714e-01 -8.27359632e-02 -8.87328565e-01 -8.05456996e-01 -5.09136796e-01 2.94941347e-02 -4.11923438e-01 1.31536126e+00 7.95008898e-01 -3.02301109e-01 3.82706344e-01 -4.37899083e-01 -1.42883942e-01 6.87680602e-01 2.48366266e-01 1.24062717e+00 -1.38611901e+00 2.97793627e-01 -4.21525210e-01 -4.66214389e-01 -1.23134303e+00 5.77835441e-01 -9.00613844e-01 1.64494962e-01 -1.50836694e+00 3.16153705e-01 -8.33239853e-01 8.98490667e-01 5.67121446e-01 1.40285671e-01 3.86703372e-01 7.78814703e-02 1.91929698e-01 -1.51250243e-01 9.89187658e-01 1.64737833e+00 6.23570643e-02 -1.82244360e-01 -1.18485868e-01 -9.09802556e-01 7.61991739e-01 6.16495073e-01 -2.30324298e-01 -4.76714551e-01 -5.03657758e-01 1.85508709e-02 2.70031124e-01 3.33798915e-01 -5.25998235e-01 -1.58219650e-01 -5.00412047e-01 3.64274204e-01 -6.49930239e-01 6.99150503e-01 -8.02989721e-01 3.08692336e-01 -1.34710163e-01 -1.38372838e-01 -8.99595544e-02 2.33441219e-01 5.02564572e-02 -5.08178324e-02 -7.94052053e-03 8.98057222e-01 -1.44789338e-01 -4.86314654e-01 8.17622900e-01 4.42729034e-02 7.55821094e-02 8.16080570e-01 -4.52692956e-01 -7.41218850e-02 -5.35493970e-01 -6.03754520e-01 8.10854062e-02 9.56618488e-01 9.44076478e-01 1.03823555e+00 -1.80621028e+00 -1.03205967e+00 8.27435672e-01 3.61865938e-01 2.47592151e-01 1.98457822e-01 2.28804395e-01 -5.69302678e-01 2.64389575e-01 -2.23808721e-01 -6.53714240e-01 -1.41347837e+00 8.67170393e-02 4.35233414e-01 5.38514495e-01 -7.35435247e-01 8.32626402e-01 2.16068462e-01 -1.01661813e+00 6.82661161e-02 1.04651459e-01 5.46617098e-02 -2.69191235e-01 6.31313682e-01 -6.78727850e-02 -1.00546837e-01 -1.26487255e+00 -3.92724752e-01 9.65277195e-01 2.36014411e-01 -1.98568612e-01 1.34874225e+00 -2.79504567e-01 -2.20658079e-01 2.42575824e-01 1.28974962e+00 4.02230844e-02 -1.19530559e+00 -2.01873675e-01 -3.42225492e-01 -9.24152613e-01 -1.90708071e-01 -3.58528316e-01 -1.13920069e+00 7.89675474e-01 2.96098411e-01 -7.27246627e-02 9.47860658e-01 3.53027880e-01 8.04601967e-01 6.94551915e-02 5.66644073e-01 -8.48845482e-01 2.62629062e-01 5.93723118e-01 1.35655940e+00 -1.12893915e+00 -2.86458209e-02 -7.65852630e-01 -8.26021135e-01 1.03975391e+00 8.35222363e-01 2.66744226e-01 8.23036432e-01 3.75189573e-01 2.52069712e-01 -4.61267680e-01 -6.71362936e-01 4.80673183e-03 4.33740854e-01 8.62797499e-01 5.59333444e-01 2.95047183e-03 9.21250507e-02 7.40321577e-01 -5.40441513e-01 -3.26638259e-02 3.23844850e-01 6.41161680e-01 -1.12442583e-01 -1.33441162e+00 -4.19652849e-01 2.53654987e-01 -8.78307000e-02 1.89282194e-01 -6.97234333e-01 6.00116134e-01 1.72150522e-01 8.76144290e-01 7.47131743e-03 -5.10765553e-01 6.42900229e-01 1.00099787e-01 6.27626657e-01 -8.82603943e-01 -2.42361221e-02 2.16519892e-01 -1.82159990e-02 -4.38371927e-01 -2.64752388e-01 -5.83078206e-01 -1.42168772e+00 -7.25080669e-01 -2.13671803e-01 1.00371949e-01 8.78195286e-01 7.65652478e-01 5.34788787e-01 -5.09406179e-02 8.69846523e-01 -1.11251581e+00 -1.30281940e-01 -7.14336574e-01 -7.63297975e-01 5.94209552e-01 2.07959756e-01 -7.67645597e-01 -1.46113425e-01 3.29823285e-01]
[12.799188613891602, -0.19566722214221954]
ce587792-ed27-4249-b944-87559c63fc7a
cross-domain-named-entity-recognition-via
null
null
https://openreview.net/forum?id=pfjbxxqih3x
https://openreview.net/pdf?id=pfjbxxqih3x
Cross-domain Named Entity Recognition via Graph Matching
Cross-domain NER is a practical yet challenging problem since the data scarcity in the real-world scenario. A common practice is first to learn a NER model in a rich-resource general domain and then adapt the model to specific domains. Due to the mismatch problem between entity types across domains, the wide knowledge in the general domain can not effectively transfer to the target domain NER model. To this end, we model the label relationship as a probability distribution and construct label graphs in both source and target label spaces. To enhance the contextual representation with label structures, we fuse the label graph into the word embedding output by BERT. By representing label relationships as graphs, we formulate cross-domain NER as a graph matching problem. Furthermore, the proposed method has good applicability with pre-training methods and is potentially capable of other cross-domain prediction tasks. Empirical results on four datasets show that our method outperforms a series of transfer learning, multi-task learning, and few-shot learning methods.
['Anonymous']
2021-11-16
null
null
null
acl-arr-november-2021-11
['cross-domain-named-entity-recognition']
['natural-language-processing']
[ 2.20939949e-01 6.40711561e-02 -3.99724901e-01 -3.52442592e-01 -8.22765887e-01 -7.43555367e-01 5.72671533e-01 3.33631247e-01 -5.53844988e-01 8.23287189e-01 2.60871738e-01 1.72422752e-02 -1.20060958e-01 -1.17627585e+00 -3.90022397e-01 -4.24837232e-01 3.38027149e-01 7.51448214e-01 4.54378515e-01 -3.42685401e-01 -9.54328030e-02 1.56576559e-01 -1.15240920e+00 1.00112431e-01 1.07025754e+00 7.32436061e-01 2.94354439e-01 -4.02116105e-02 -7.12188661e-01 5.10728300e-01 -4.33471501e-01 -8.35667670e-01 1.08401790e-01 -3.17209035e-01 -9.20048296e-01 -2.18187094e-01 2.58275092e-01 2.69726604e-01 -3.98301482e-01 1.33888793e+00 5.95854521e-01 5.12470484e-01 9.08104300e-01 -1.26871729e+00 -1.08593798e+00 4.89249766e-01 -5.03220558e-01 4.41187210e-02 3.12067688e-01 -4.52388197e-01 1.23763919e+00 -7.36080706e-01 8.42619359e-01 1.18185413e+00 7.59188890e-01 6.35152996e-01 -1.16613734e+00 -8.93659234e-01 2.14438275e-01 2.96749353e-01 -1.14861572e+00 3.62872961e-03 8.36424232e-01 -4.77008879e-01 8.91710758e-01 -3.61368179e-01 2.18582287e-01 1.28014791e+00 -1.01956479e-01 4.37972009e-01 1.05962825e+00 -5.63442349e-01 1.46658450e-01 1.47626206e-01 2.07930699e-01 5.60831308e-01 2.06310213e-01 -5.70533536e-02 -2.36721903e-01 -3.50037783e-01 4.58660424e-01 1.77938387e-01 -1.76044211e-01 -5.92574537e-01 -9.50565338e-01 8.63097966e-01 4.66677606e-01 5.77421725e-01 -1.31481692e-01 -2.95869708e-01 5.65959275e-01 3.01800519e-01 7.05978632e-01 4.61849242e-01 -6.20172620e-01 3.07782024e-01 -5.97417831e-01 -1.24823526e-02 9.31524217e-01 1.27198017e+00 1.14188409e+00 -3.11560988e-01 -6.74937889e-02 1.31802976e+00 2.95278281e-01 3.16667557e-01 6.76828086e-01 -3.40274751e-01 8.62074077e-01 7.05955982e-01 -6.19042069e-02 -9.91594791e-01 -3.10162038e-01 -1.99871629e-01 -6.84727311e-01 -2.04280451e-01 4.51672703e-01 -2.85670638e-01 -8.31348300e-01 1.92593217e+00 4.53310132e-01 5.70351720e-01 3.33455056e-01 4.76188481e-01 8.49509060e-01 7.48095989e-01 8.04008186e-01 4.97667007e-02 1.56199682e+00 -1.10086775e+00 -7.54813075e-01 -6.95510626e-01 1.02492917e+00 -5.64735830e-01 9.55644786e-01 -1.41538411e-01 -2.05285549e-01 -6.44130468e-01 -8.82112324e-01 -9.14006308e-02 -9.55924511e-01 -1.47788405e-01 3.83715898e-01 3.76749039e-01 -5.27971208e-01 6.25339746e-01 -2.25872099e-01 -7.94218063e-01 2.90889561e-01 3.19993459e-02 -5.38114190e-01 -7.67844796e-01 -1.81325471e+00 1.31073308e+00 1.00570941e+00 -5.54198921e-01 -4.11479264e-01 -7.76391804e-01 -1.08358717e+00 2.19497055e-01 4.36677277e-01 -6.04687095e-01 8.30733240e-01 -7.22084343e-01 -1.11817884e+00 9.11907315e-01 7.22004026e-02 -3.33198793e-02 -3.13604884e-02 4.27218787e-02 -1.03546834e+00 -2.48649657e-01 3.36832494e-01 4.15652692e-01 4.28749591e-01 -1.31883407e+00 -8.16144288e-01 -3.12048584e-01 1.54399917e-01 3.73989165e-01 -7.01854706e-01 -8.86707231e-02 -3.20479751e-01 -6.49205863e-01 -3.57025832e-01 -8.47919106e-01 -2.26558477e-01 -4.20694888e-01 -4.52528037e-02 -5.52492023e-01 7.09966362e-01 -6.25236690e-01 1.30255222e+00 -2.14198565e+00 5.27065769e-02 8.60303491e-02 2.23203935e-02 3.05143327e-01 -5.37835062e-01 6.54170930e-01 -2.76256114e-01 9.47074816e-02 -2.67668992e-01 -3.26920524e-02 1.76198304e-01 4.40430462e-01 -8.13738406e-02 1.62559569e-01 2.88095698e-02 7.63365984e-01 -1.40177166e+00 -7.19766200e-01 3.88809741e-02 3.38710815e-01 -2.22444430e-01 4.41641718e-01 -3.43660712e-02 1.21745966e-01 -6.85891867e-01 5.18757641e-01 6.72742903e-01 -3.60432655e-01 6.82810843e-01 -4.43128586e-01 3.08666080e-01 1.00869156e-01 -1.10847628e+00 1.97520268e+00 -7.90670812e-01 2.95459837e-01 -4.26108032e-01 -1.27625799e+00 1.22814643e+00 3.84709477e-01 4.35245037e-01 -6.06895089e-01 8.47402364e-02 1.81954741e-01 -2.54304707e-01 -4.56811041e-01 3.50947589e-01 -6.40653014e-01 -4.12331522e-01 3.34182441e-01 7.42311597e-01 6.20078333e-02 2.69996554e-01 -4.74275351e-02 1.02259016e+00 1.66681185e-01 7.35887408e-01 -1.48732483e-01 3.42909217e-01 8.22967291e-02 8.19914043e-01 2.66040355e-01 -1.66723028e-01 1.75147280e-01 2.22938776e-01 -1.76361978e-01 -8.87698591e-01 -1.04694653e+00 -1.07379079e-01 1.45104706e+00 3.33160877e-01 -4.45796788e-01 -5.07966876e-01 -1.36281431e+00 2.32534111e-02 9.04381275e-01 -6.44076109e-01 -3.41665715e-01 -4.10405427e-01 -5.82950294e-01 4.55792487e-01 6.10388041e-01 3.83514047e-01 -1.04837942e+00 1.97919115e-01 4.47741717e-01 -3.15278560e-01 -1.27457488e+00 -6.12110376e-01 3.02522123e-01 -7.08158553e-01 -1.16776609e+00 -6.52309179e-01 -1.30495954e+00 5.09240448e-01 2.66124427e-01 1.31508553e+00 -3.22477520e-01 2.37453282e-02 3.30440462e-01 -5.72088838e-01 -1.02271609e-01 -3.44390899e-01 1.66104212e-01 -1.27447277e-01 -1.81115746e-01 9.82669175e-01 -5.72136998e-01 -1.97155133e-01 2.94908464e-01 -9.30597603e-01 -3.94894242e-01 4.15513933e-01 1.15916717e+00 5.33909500e-01 1.99646667e-01 7.88575709e-01 -1.38412547e+00 7.45449841e-01 -1.05433536e+00 -4.49648857e-01 8.21561337e-01 -7.09731460e-01 1.47269607e-01 7.19186366e-01 -6.35074854e-01 -1.42902696e+00 8.15075934e-02 8.87841433e-02 -1.18304074e-01 -3.67291987e-01 7.29215026e-01 -5.72229564e-01 1.33734941e-01 6.77309930e-01 -1.03304915e-01 -5.37720203e-01 -7.09025145e-01 6.92738950e-01 6.01890504e-01 2.63435662e-01 -8.27031076e-01 8.02458704e-01 3.34993489e-02 -1.26136899e-01 -5.09393334e-01 -1.30469835e+00 -8.16495001e-01 -9.46988642e-01 -2.71135606e-02 9.78214025e-01 -1.04907632e+00 1.68500558e-01 -2.35338602e-02 -1.28545856e+00 -1.13061257e-01 -2.00058609e-01 4.29201066e-01 -2.55128771e-01 5.41068673e-01 -4.14037406e-01 -2.67448694e-01 -7.91227520e-02 -7.06207573e-01 8.49636018e-01 4.56745118e-01 -3.72202955e-02 -1.82012272e+00 6.46127522e-01 1.45251140e-01 6.86120614e-02 7.69204348e-02 1.31501389e+00 -1.32872701e+00 1.17341518e-01 -1.49752468e-01 -4.90896583e-01 2.44819432e-01 4.41254795e-01 -5.60430169e-01 -7.94106007e-01 -2.64321953e-01 -3.41513038e-01 -4.76813972e-01 5.68358421e-01 -1.91637337e-01 7.28539348e-01 6.01031296e-02 -6.82555854e-01 2.86497951e-01 1.80445516e+00 1.56788528e-01 4.09450620e-01 4.57327753e-01 8.67390573e-01 8.21895480e-01 8.60095203e-01 2.92311430e-01 6.52048051e-01 6.74842536e-01 -1.61534883e-02 -1.92693826e-02 -3.40880603e-01 -6.87905550e-01 1.42250046e-01 1.11513889e+00 2.49674723e-01 -5.27956784e-01 -1.14110827e+00 7.83411920e-01 -1.88803506e+00 -8.17582250e-01 1.20135553e-01 1.82372081e+00 9.37049448e-01 -1.98825568e-01 -1.38686702e-01 -5.87817907e-01 1.20860815e+00 1.20079547e-01 -5.27242661e-01 -1.25052199e-01 -6.26511276e-02 3.08826208e-01 5.68798721e-01 8.76248181e-02 -1.26298201e+00 1.14454663e+00 5.64940166e+00 1.09875238e+00 -5.54562032e-01 4.28799272e-01 4.45574895e-02 7.03780532e-01 -3.00719172e-01 1.61218598e-01 -9.40569103e-01 4.57492352e-01 9.74428594e-01 -4.54153627e-01 1.76896587e-01 8.91094089e-01 -5.87413549e-01 4.74599928e-01 -1.13347721e+00 9.57545936e-01 1.51250467e-01 -8.48092794e-01 1.03059120e-01 5.66709153e-02 8.45831454e-01 5.81184141e-02 -3.55325818e-01 9.04193223e-01 9.35389638e-01 -7.81394064e-01 -2.77658291e-02 2.78565973e-01 9.17209506e-01 -7.36075759e-01 8.40558052e-01 3.37655425e-01 -1.58851349e+00 -4.54446673e-02 -8.39069366e-01 3.62977386e-01 2.89283365e-01 4.39157814e-01 -6.80170655e-01 1.02427423e+00 3.58273804e-01 9.90756273e-01 -4.39927310e-01 1.01320148e+00 -4.76799697e-01 2.59631366e-01 7.29257688e-02 1.67685151e-01 2.25178793e-01 -4.73672301e-01 1.06015153e-01 1.51184618e+00 6.93116009e-01 8.65091011e-02 4.18619335e-01 4.83043224e-01 -4.77701247e-01 4.71577525e-01 -9.79180813e-01 -1.72951058e-01 7.57956564e-01 1.40791559e+00 -3.80317181e-01 -3.41774791e-01 -1.03333676e+00 1.06601095e+00 7.71134198e-01 4.02928144e-01 -7.72076964e-01 -6.73920810e-01 6.13657713e-01 -3.35676789e-01 2.77318299e-01 3.08317151e-02 1.34322587e-02 -1.45150089e+00 -2.91179478e-01 -5.30199766e-01 1.03871977e+00 -5.45688748e-01 -2.15365744e+00 6.98505998e-01 -2.19402742e-02 -1.41391003e+00 -9.53948125e-02 -7.15656340e-01 -5.67621708e-01 8.83149981e-01 -1.99220490e+00 -1.37186205e+00 -1.89031914e-01 6.46406889e-01 4.85533535e-01 -3.68758470e-01 1.25321019e+00 6.74983740e-01 -3.84628683e-01 5.80773413e-01 4.04542536e-01 4.15765822e-01 1.24246371e+00 -1.29225433e+00 3.33434284e-01 6.27200425e-01 2.68278748e-01 4.06739563e-01 3.87839258e-01 -8.25225353e-01 -8.01756263e-01 -1.42240727e+00 1.16130745e+00 -3.13896686e-01 1.09427261e+00 -2.68483937e-01 -1.24258053e+00 7.47508168e-01 5.13614655e-01 2.65937150e-01 1.24906945e+00 4.50421482e-01 -8.03979993e-01 4.94357087e-02 -1.16333067e+00 4.01671976e-01 1.13985002e+00 -7.28133798e-01 -1.19174361e+00 4.19956952e-01 6.02896273e-01 8.27259477e-03 -1.32242203e+00 2.53519297e-01 1.81337073e-01 -2.54741341e-01 9.54563260e-01 -1.05420482e+00 2.87578523e-01 -6.34613112e-02 -3.19685698e-01 -2.10210729e+00 -5.32274842e-01 2.12793928e-02 1.94723643e-02 1.87883651e+00 6.03307724e-01 -6.08837664e-01 5.74712396e-01 6.63295746e-01 6.54328242e-02 -1.47561789e-01 -8.59957159e-01 -1.11014295e+00 4.65053350e-01 7.85085931e-02 6.71708465e-01 1.58752739e+00 3.36015135e-01 7.22860336e-01 -2.80489117e-01 2.17867583e-01 7.00723886e-01 2.04118460e-01 2.81822830e-01 -1.68871069e+00 -1.46924973e-01 -3.29663604e-02 -2.17951819e-01 -8.89495015e-01 8.80972862e-01 -1.33892548e+00 2.00733989e-02 -1.67356539e+00 2.51516193e-01 -7.91570961e-01 -8.31033647e-01 5.43082535e-01 -3.63265246e-01 -1.57681271e-01 1.57454774e-01 6.28962368e-02 -8.05967391e-01 6.44837737e-01 1.13660777e+00 -2.28324383e-01 1.13919735e-01 -3.17447066e-01 -6.54300213e-01 6.46257579e-01 7.67192483e-01 -9.65531945e-01 -5.54230034e-01 -4.57748652e-01 1.51474461e-01 -1.30193317e-02 -1.43626928e-01 -7.52965808e-01 4.28515375e-01 -3.43540400e-01 9.74543020e-02 -8.90137628e-02 2.16672286e-01 -1.18934798e+00 3.59389782e-02 -1.00992136e-02 -2.78671503e-01 -5.21320291e-02 -3.40899304e-02 9.30290878e-01 -4.44566846e-01 -5.57937086e-01 9.54748571e-01 -2.15648666e-01 -1.30196249e+00 5.55983126e-01 1.02069817e-01 7.12272346e-01 1.07264674e+00 -7.77210370e-02 -5.19092321e-01 7.43990839e-02 -7.89360523e-01 3.38913113e-01 3.75410080e-01 7.06901193e-01 3.13918144e-01 -1.83917391e+00 -4.43831265e-01 -2.29400054e-01 7.29672968e-01 -2.83358663e-01 3.86384368e-01 2.42563680e-01 1.77808687e-01 1.30726680e-01 -3.79523814e-01 -3.15975361e-02 -1.04384530e+00 8.33516061e-01 9.91394594e-02 -6.94540739e-01 -5.16955674e-01 4.62836713e-01 3.46482754e-01 -9.70260262e-01 -2.04200260e-02 3.18876475e-01 -6.73440635e-01 4.53323126e-01 3.07272047e-01 2.85170525e-01 -4.05913889e-02 -7.02147424e-01 -3.01342100e-01 7.85574615e-01 7.08697140e-02 2.16757879e-01 1.48688054e+00 -2.79979110e-01 1.34679126e-02 4.12991673e-01 1.50289178e+00 -1.92624629e-01 -9.04636204e-01 -7.93472230e-01 5.74622452e-01 -3.52946132e-01 -2.55542606e-01 -7.92071164e-01 -7.96254814e-01 8.41794550e-01 4.76860225e-01 1.39851868e-01 8.76688063e-01 2.19485760e-01 8.52897525e-01 3.55586052e-01 4.10508454e-01 -1.42510819e+00 -5.18461801e-02 7.16023922e-01 3.86190742e-01 -1.30044103e+00 -2.50825793e-01 -5.33886135e-01 -8.25110674e-01 1.19661236e+00 9.48908627e-01 -3.01093906e-02 9.02636230e-01 -3.86490859e-02 1.08885631e-01 -1.40769973e-01 -5.63009322e-01 -4.84753072e-01 4.54611927e-01 1.07121813e+00 5.66616416e-01 6.59786761e-02 -2.65187651e-01 7.73969233e-01 3.17768335e-01 -1.48556188e-01 2.39460662e-01 5.98325133e-01 -3.81872356e-01 -1.81200945e+00 1.25993460e-01 3.09762686e-01 -2.64677525e-01 -2.02464804e-01 -2.08253354e-01 8.34725857e-01 4.08572584e-01 9.46408331e-01 -1.57204181e-01 -3.79654408e-01 6.78207457e-01 5.54886162e-01 2.37448096e-01 -1.02589428e+00 -4.17759091e-01 -2.78440863e-01 3.94039899e-01 -1.64993525e-01 -5.80021262e-01 -4.60412413e-01 -1.09712422e+00 3.96171473e-02 -3.77075106e-01 3.96550685e-01 4.13922161e-01 1.08623064e+00 3.64175290e-01 5.03455162e-01 3.71101111e-01 -2.52835453e-01 -5.84887564e-01 -9.72801387e-01 -9.30304110e-01 8.27310264e-01 -2.56885946e-01 -1.05852401e+00 -2.52310127e-01 2.99632903e-02]
[9.746838569641113, 9.613967895507812]
8f3f742e-029a-4a07-b829-fbb773e00a4c
runtime-analysis-of-competitive-co
2206.15238
null
https://arxiv.org/abs/2206.15238v1
https://arxiv.org/pdf/2206.15238v1.pdf
Runtime Analysis of Competitive co-Evolutionary Algorithms for Maximin Optimisation of a Bilinear Function
Co-evolutionary algorithms have a wide range of applications, such as in hardware design, evolution of strategies for board games, and patching software bugs. However, these algorithms are poorly understood and applications are often limited by pathological behaviour, such as loss of gradient, relative over-generalisation, and mediocre objective stasis. It is an open challenge to develop a theory that can predict when co-evolutionary algorithms find solutions efficiently and reliable. This paper provides a first step in developing runtime analysis for population-based competitive co-evolutionary algorithms. We provide a mathematical framework for describing and reasoning about the performance of co-evolutionary processes. An example application of the framework shows a scenario where a simple co-evolutionary algorithm obtains a solution in polynomial expected time. Finally, we describe settings where the co-evolutionary algorithm needs exponential time with overwhelmingly high probability to obtain a solution.
['Per Kristian Lehre']
2022-06-30
null
null
null
null
['board-games']
['playing-games']
[ 1.76090047e-01 -1.90864071e-01 1.93222135e-01 1.49887905e-01 -4.85995620e-01 -6.46967888e-01 2.27358952e-01 1.59113690e-01 -3.53377402e-01 7.74907768e-01 -6.46201253e-01 -5.98202407e-01 -3.54090512e-01 -8.05092275e-01 -8.43512952e-01 -8.64285231e-01 -4.75638449e-01 5.59168458e-01 3.24658930e-01 -3.73141110e-01 6.27898037e-01 2.60028481e-01 -1.85219252e+00 -3.35661680e-01 9.19890404e-01 5.68938553e-01 -1.23042308e-01 1.41386402e+00 1.66826352e-01 1.47712141e-01 -8.96240234e-01 -5.22242427e-01 3.59234661e-01 -1.03036463e+00 -5.25895298e-01 -1.27263471e-01 -1.74816355e-01 1.97488174e-01 6.88478723e-02 1.39662504e+00 6.48855865e-01 -1.29837707e-01 3.14217865e-01 -1.61011672e+00 -2.62569010e-01 6.84195161e-01 -8.49649131e-01 2.27081433e-01 2.02120200e-01 1.30085528e-01 9.52136874e-01 -3.24398011e-01 4.89025444e-01 1.10794592e+00 8.39214861e-01 5.89499235e-01 -1.22953606e+00 -5.43070495e-01 -2.18980283e-01 -8.74219686e-02 -1.59443557e+00 -1.48418605e-01 2.46766791e-01 2.33944505e-02 9.15258646e-01 8.45884085e-01 9.42367971e-01 4.86343861e-01 4.24873561e-01 7.71650851e-01 9.74733531e-01 -7.06209600e-01 5.40579677e-01 1.23569243e-01 -2.42655173e-01 6.84189439e-01 8.10039282e-01 4.22949940e-01 -4.84623164e-01 -7.56947398e-01 3.72679114e-01 -2.32693121e-01 -1.49949893e-01 -5.84688723e-01 -6.39384151e-01 8.05354059e-01 -7.07296804e-02 2.09765226e-01 -1.01653129e-01 5.13848543e-01 5.03315590e-02 7.54503071e-01 9.25252810e-02 7.97328293e-01 -5.96603036e-01 -6.57976627e-01 -8.75094891e-01 5.56003094e-01 1.04666483e+00 6.90795839e-01 3.50022644e-01 -1.38557434e-01 4.55129892e-01 3.85789245e-01 1.39817178e-01 2.25954652e-01 4.83212650e-01 -1.02664995e+00 -1.78080663e-01 3.09739411e-01 1.52342826e-01 -8.86143565e-01 -2.66323805e-01 -5.53740561e-01 -4.28709984e-01 5.42213857e-01 3.55193228e-01 -3.37848425e-01 -2.20942244e-01 1.85852826e+00 4.63926584e-01 3.27879429e-01 -3.52151250e-03 4.81714457e-01 -3.76260877e-01 4.53256756e-01 -4.78023887e-01 -6.49944186e-01 1.05508959e+00 -9.01289403e-01 -2.52452403e-01 -1.15217283e-01 7.28536904e-01 -6.84985518e-01 7.95517206e-01 4.87045377e-01 -1.44943285e+00 1.19192697e-01 -1.27222776e+00 6.18475080e-01 4.22409317e-03 -5.31021714e-01 6.49917364e-01 1.35513449e+00 -1.10674632e+00 8.12402606e-01 -1.07562649e+00 -4.25643116e-01 2.90810287e-01 5.55045784e-01 1.14186056e-01 7.56957680e-02 -5.24883986e-01 6.15834534e-01 1.02289170e-01 -1.10715486e-01 -4.34175253e-01 -6.20854676e-01 -3.23321253e-01 1.42134801e-01 6.42025352e-01 -9.03421879e-01 1.25221598e+00 -1.21993363e+00 -1.42861700e+00 7.69174635e-01 -1.57186761e-01 -6.04422987e-01 7.26227820e-01 1.50925517e-01 -1.92532167e-01 -4.72702503e-01 -2.04566389e-01 4.98347618e-02 7.74158716e-01 -8.37007999e-01 -9.97072577e-01 -4.95169073e-01 -3.76653150e-02 1.71119392e-01 -3.10402513e-01 1.97641537e-01 -1.21129513e-01 -2.85267353e-01 -2.80166678e-02 -1.26986146e+00 -6.38807118e-01 -2.06446379e-01 -1.14734173e-01 8.02310854e-02 4.19000924e-01 6.73598945e-02 1.34297836e+00 -2.12003350e+00 3.22134823e-01 3.44822049e-01 3.27239454e-01 -6.17576204e-02 9.50567201e-02 5.79195499e-01 1.58548966e-01 4.92247552e-01 -3.18472385e-01 1.13165908e-01 2.60116994e-01 -2.31329203e-02 -6.18497767e-02 6.69224739e-01 5.97794205e-02 7.47222185e-01 -9.76461291e-01 -1.50024265e-01 -4.29296076e-01 2.90774647e-02 -1.01174903e+00 -1.39309019e-01 -1.36727840e-01 -9.39166695e-02 -2.05672741e-01 5.79837322e-01 4.06155258e-01 -3.08733463e-01 4.74552423e-01 7.02992380e-01 -3.08409810e-01 -1.04687005e-01 -1.19110584e+00 1.14127314e+00 -3.66685957e-01 7.03687012e-01 2.16227666e-01 -8.31989884e-01 5.49006045e-01 -1.47909343e-01 3.32713187e-01 -3.42462391e-01 5.50730526e-01 6.65023029e-01 6.27458215e-01 2.45725475e-02 7.75920212e-01 -1.47756645e-02 -2.24437505e-01 9.54333782e-01 -4.58664328e-01 -2.61506766e-01 4.01920050e-01 -7.51516037e-03 1.55443907e+00 -3.14699590e-01 1.74867526e-01 -2.06405684e-01 2.77527899e-01 1.79251090e-01 4.66094464e-01 9.86295700e-01 -1.09776318e-01 4.66417551e-01 7.04004765e-01 -2.86195040e-01 -1.19519055e+00 -6.61096275e-01 6.54690713e-02 9.16552007e-01 4.78465617e-01 -6.76540911e-01 -6.88239217e-01 -1.41620740e-01 -6.74902499e-02 7.05937445e-01 -6.31566107e-01 -7.12446392e-01 -3.64153147e-01 -1.17870629e+00 4.90938663e-01 1.04256175e-01 3.11272666e-02 -5.92642426e-01 -1.39806974e+00 2.59501129e-01 4.64782447e-01 -4.65125918e-01 -4.45529789e-01 3.59771758e-01 -9.27761436e-01 -9.93190765e-01 -5.16525745e-01 -6.35985494e-01 6.58433557e-01 2.07388863e-01 1.23143637e+00 6.45927906e-01 -7.89825618e-01 1.21047109e-01 -1.72134154e-02 -3.86579692e-01 -6.95987761e-01 1.50830382e-02 7.12069049e-02 -3.87743860e-01 1.70483753e-01 -6.08701885e-01 -3.86433840e-01 3.81408811e-01 -5.92850745e-01 -4.06400532e-01 4.01998788e-01 1.10344112e+00 4.91551042e-01 8.62770319e-01 6.72541186e-02 -5.82308769e-01 1.10982084e+00 -6.50101364e-01 -1.10082579e+00 5.20006299e-01 -8.71710539e-01 2.43921295e-01 5.54382861e-01 -6.87396824e-01 -4.75739032e-01 -9.56289247e-02 5.13968803e-02 -2.68991709e-01 4.98083085e-01 4.25733745e-01 6.16050437e-02 -3.46461684e-01 6.82555556e-01 3.22612792e-01 1.83126181e-01 -8.37471560e-02 7.02971593e-02 4.23399806e-01 5.26669145e-01 -8.74403358e-01 7.20024824e-01 2.91564822e-01 2.04174280e-01 -7.39662826e-01 -2.61061192e-01 -4.46357578e-03 5.80601767e-02 -1.20196603e-01 1.23400345e-01 -3.07301611e-01 -9.35530484e-01 4.33884948e-01 -8.02455604e-01 -2.36990079e-01 -3.78015280e-01 -1.97003469e-01 -6.58176005e-01 2.37275481e-01 -1.79285944e-01 -1.17635059e+00 -1.57747835e-01 -1.27365530e+00 6.92924440e-01 6.77905202e-01 -3.11338514e-01 -7.06534505e-01 2.91709930e-01 -1.64361387e-01 4.54583973e-01 1.45268396e-01 9.11801875e-01 -5.45634151e-01 -5.01382589e-01 -2.72093356e-01 5.00853717e-01 -3.38812530e-01 -2.46477842e-01 6.39299870e-01 -3.09923977e-01 -4.36946541e-01 -8.94780159e-02 -1.29622117e-01 2.69986421e-01 4.12624627e-01 9.81333077e-01 -2.11473450e-01 -5.32446980e-01 7.07526684e-01 1.40779209e+00 5.01026154e-01 5.81206739e-01 5.96238196e-01 -5.40132001e-02 4.71103042e-01 7.44653225e-01 7.38774180e-01 2.79882830e-02 5.86724937e-01 2.60226160e-01 5.80025554e-01 5.62620282e-01 3.25001194e-03 3.18368524e-01 5.87933302e-01 9.95552391e-02 -3.83672595e-01 -9.70121384e-01 5.25242686e-01 -1.99354339e+00 -1.03358746e+00 -3.13557312e-02 2.56981254e+00 8.56019437e-01 3.78022283e-01 6.52117610e-01 3.88393641e-01 9.33620453e-01 -7.52432287e-01 -8.54450881e-01 -7.45182514e-01 -1.85222521e-01 4.29121882e-01 7.86749363e-01 3.35883737e-01 -5.88393331e-01 7.73656368e-01 7.28309536e+00 8.09208393e-01 -9.64124978e-01 4.73653190e-02 6.63680553e-01 -2.64548510e-01 -3.42385352e-01 1.72728777e-01 -4.38941061e-01 5.68386853e-01 1.16873789e+00 -1.06197095e+00 6.05392635e-01 9.14382875e-01 -2.96276182e-01 -2.05694601e-01 -9.41026092e-01 8.46477509e-01 -1.79079458e-01 -1.29488337e+00 -8.48875403e-01 3.52170140e-01 1.02392876e+00 -2.55216122e-01 3.69381845e-01 7.59861693e-02 7.55372643e-01 -1.01294196e+00 7.77889013e-01 -9.79764909e-02 4.47216958e-01 -1.49778759e+00 4.80890512e-01 4.42521393e-01 -7.04930484e-01 -4.17214483e-01 -4.54286218e-01 -1.51980355e-01 1.33646771e-01 1.72397614e-01 -5.17462909e-01 2.16581404e-01 8.25608432e-01 -2.14282330e-02 -3.87431771e-01 1.76364219e+00 2.07742870e-01 3.35819691e-01 -7.02638984e-01 -5.81097603e-01 2.45378576e-02 -3.16204339e-01 8.45829964e-01 5.59358895e-01 7.24087894e-01 -9.07077640e-03 -2.35864148e-01 7.82237709e-01 1.95503712e-01 2.47016326e-02 -3.37801456e-01 -4.07348156e-01 8.04459333e-01 8.87410760e-01 -1.26677489e+00 -3.32461521e-02 2.31555793e-02 9.79476094e-01 -2.60262266e-02 -1.58761233e-01 -1.08983302e+00 -4.79694605e-01 9.55400646e-01 -8.06428194e-02 3.01656932e-01 -2.29170159e-01 -5.67440093e-01 -9.03503656e-01 -2.13514984e-01 -1.16177070e+00 3.08352500e-01 -4.00625616e-01 -9.66046631e-01 5.02131045e-01 -3.09170008e-01 -8.31323981e-01 -4.27641451e-01 -3.70401919e-01 -6.32508695e-01 3.94306064e-01 -6.64969921e-01 -1.19758584e-01 5.19874468e-02 7.02414885e-02 3.64322543e-01 -2.56749779e-01 4.37112838e-01 1.29427388e-01 -7.19056785e-01 8.99959385e-01 3.58767182e-01 -9.25721586e-01 3.01745176e-01 -1.15092707e+00 5.19255757e-01 1.05885303e+00 1.54647246e-01 6.87488079e-01 1.36828148e+00 -4.31079835e-01 -2.01365256e+00 -4.75079209e-01 5.04354298e-01 -4.09604549e-01 1.01358891e+00 -1.22535080e-01 -5.27831554e-01 2.07901523e-01 -1.72427222e-01 -2.20761925e-01 4.70568389e-01 2.67060131e-01 -7.62564465e-02 1.49948493e-01 -1.17183805e+00 1.15067029e+00 1.22028661e+00 -1.38671875e-01 -6.85319258e-03 2.76324928e-01 4.52670902e-01 -5.39821029e-01 -3.73935461e-01 9.64482054e-02 8.03680599e-01 -1.25047719e+00 8.70201528e-01 -4.96046335e-01 3.57232034e-01 -2.93643385e-01 5.31088412e-02 -1.18560803e+00 -1.49276182e-01 -1.35661030e+00 -2.99578637e-01 7.98299074e-01 3.80316943e-01 -7.90876567e-01 9.52730834e-01 4.39290464e-01 2.09773824e-01 -9.02888298e-01 -7.40112662e-01 -1.13125539e+00 2.58072555e-01 -3.07600111e-01 6.57223344e-01 6.90214396e-01 1.63971027e-03 3.85242738e-02 -3.16166520e-01 6.10378273e-02 5.60832918e-01 2.29483530e-01 7.64674485e-01 -1.12280655e+00 -8.81289840e-01 -1.12955225e+00 -5.79343319e-01 -7.79660523e-01 4.28211391e-02 -4.76183057e-01 1.50859103e-01 -6.06840074e-01 2.88739383e-01 -6.94926322e-01 -5.17333522e-02 -1.42529532e-01 -2.28927433e-01 2.27058992e-01 9.62599926e-03 -9.71085206e-02 -6.42698467e-01 2.47279793e-01 6.93158746e-01 1.85090788e-02 -2.98064053e-01 3.39011729e-01 -9.57204759e-01 4.93117332e-01 8.41158688e-01 -8.65585923e-01 -3.33035111e-01 -1.08534433e-01 1.01114237e+00 1.53923541e-01 -3.36368196e-03 -8.81214917e-01 4.32700604e-01 -2.46725991e-01 -3.29911560e-01 -1.48164228e-01 -9.31020603e-02 -7.51906633e-01 6.62940264e-01 1.12473011e+00 -2.25228131e-01 7.56947398e-01 1.19490348e-01 6.30998313e-01 1.85066268e-01 -5.91081142e-01 8.09280634e-01 1.07862398e-01 -3.39169532e-01 6.61606267e-02 -5.97570777e-01 1.22093998e-01 1.32913280e+00 -6.08887255e-01 -2.11852655e-01 -3.68112713e-01 -3.56495708e-01 3.24194580e-01 1.06631458e+00 2.95864254e-01 3.07107925e-01 -8.65984559e-01 -5.36659539e-01 1.67222336e-01 -8.99228156e-02 -4.11863327e-01 1.70572773e-02 1.00511456e+00 -8.34587336e-01 -1.89181700e-01 -2.72172391e-01 -4.76316214e-01 -1.83272028e+00 4.63396847e-01 5.04364073e-01 -7.81476870e-02 -2.06717879e-01 1.29339242e+00 -1.67149767e-01 -2.31875107e-01 -5.88989176e-04 2.66550452e-01 8.04916561e-01 -4.31002945e-01 5.76941609e-01 4.59025443e-01 -5.83041459e-02 -2.53634870e-01 -5.86491168e-01 4.13138688e-01 7.02029690e-02 -4.01452750e-01 1.30715930e+00 -2.58415155e-02 -4.95105118e-01 3.59261215e-01 7.81282783e-01 1.29948691e-01 -8.12925160e-01 3.77738059e-01 1.91856638e-01 -6.42544806e-01 -1.32088095e-01 -5.55092931e-01 -8.67891312e-01 6.30348742e-01 5.11796772e-01 4.59881872e-01 1.34963250e+00 -2.50988841e-01 4.94109303e-01 4.64823931e-01 8.22724581e-01 -9.46614921e-01 3.55387921e-03 3.75422508e-01 3.70852917e-01 -6.64416075e-01 5.31111285e-02 -1.44603968e-01 -2.49357671e-01 1.08648324e+00 5.53532481e-01 -3.48554164e-01 5.29841542e-01 8.56875122e-01 -5.12910426e-01 -6.91562891e-02 -1.38890409e+00 3.65989581e-02 -2.20626503e-01 3.46061200e-01 1.94355056e-01 7.33513087e-02 -6.64880812e-01 4.76745009e-01 -6.63372993e-01 -3.28668803e-01 8.50115478e-01 1.25496399e+00 -4.57432806e-01 -1.35139692e+00 -4.82007653e-01 1.62398249e-01 -7.28274047e-01 -8.35435316e-02 -5.26576042e-01 6.90681756e-01 -1.47508264e-01 7.04603791e-01 6.19348548e-02 -5.00713825e-01 -1.87240601e-01 -1.77357316e-01 1.05853105e+00 -2.45536625e-01 -6.96737945e-01 7.34814778e-02 -8.39940310e-02 -4.06762928e-01 1.45316646e-01 -9.46338117e-01 -9.95117545e-01 -8.64691019e-01 -6.94392204e-01 3.32249105e-01 6.70149446e-01 7.35085368e-01 5.44409871e-01 4.22017664e-01 6.66295409e-01 -3.59373242e-01 -6.98428035e-01 -3.25762257e-02 -5.63456893e-01 -2.24669293e-01 4.82316092e-02 -4.85324055e-01 -4.54896301e-01 -3.83072257e-01]
[5.773055076599121, 4.060076713562012]
64b6c39a-3612-4849-a735-987891bc2550
automatic-evaluation-of-herding-behavior-in
2303.12016
null
https://arxiv.org/abs/2303.12016v1
https://arxiv.org/pdf/2303.12016v1.pdf
Automatic evaluation of herding behavior in towed fishing gear using end-to-end training of CNN and attention-based networks
This paper considers the automatic classification of herding behavior in the cluttered low-visibility environment that typically surrounds towed fishing gear. The paper compares three convolutional and attention-based deep action recognition network architectures trained end-to-end on a small set of video sequences captured by a remotely controlled camera and classified by an expert in fishing technology. The sequences depict a scene in front of a fishing trawl where the conventional herding mechanism has been replaced by directed laser light. The goal is to detect the presence of a fish in the sequence and classify whether or not the fish reacts to the lasers. A two-stream CNN model, a CNN-transformer hybrid, and a pure transformer model were trained end-to-end to achieve 63%, 54%, and 60% 10-fold classification accuracy on the three-class task when compared to the human expert. Inspection of the activation maps learned by the three networks raises questions about the attributes of the sequences the models may be learning, specifically whether changes in viewpoint introduced by human camera operators that affect the position of laser lines in the video frames may interfere with the classification. This underlines the importance of careful experimental design when capturing scientific data for automatic end-to-end evaluation and the usefulness of inspecting the trained models.
['Torfi Thorhallsson', 'Martin Eineborg', 'Týr Vilhjálmsson', 'Orri Steinn Guðfinnsson']
2023-03-21
null
null
null
null
['experimental-design']
['methodology']
[ 4.03550923e-01 -1.89758003e-01 6.74295127e-01 -4.82656389e-01 -6.46479577e-02 -6.97439551e-01 4.37179893e-01 -2.12541908e-01 -9.46075857e-01 8.75979289e-02 -9.32244733e-02 -7.99196959e-03 4.64854278e-02 -4.94030625e-01 -9.32538986e-01 -8.37282538e-01 -4.59570646e-01 1.49081558e-01 3.47993761e-01 -7.83059001e-02 4.68450963e-01 6.83825910e-01 -1.80132115e+00 3.99611831e-01 -2.93525875e-01 9.54358101e-01 1.42213032e-01 1.11041951e+00 4.27011222e-01 1.03915215e+00 -7.93636441e-01 -2.87190557e-01 5.27623236e-01 -4.46518689e-01 -3.75997722e-01 3.32438320e-01 9.41276789e-01 -6.45759523e-01 -2.82357514e-01 1.01860368e+00 2.24378034e-01 1.02306038e-01 4.23611730e-01 -9.27248955e-01 2.07584314e-02 1.11949570e-01 -3.64429384e-01 7.55287528e-01 9.36193839e-02 7.43956089e-01 5.81572890e-01 -4.94909316e-01 3.44232559e-01 1.03854585e+00 7.03479528e-01 4.28218782e-01 -1.02706897e+00 -5.50102949e-01 -1.79768413e-01 1.65451378e-01 -1.02077484e+00 -6.23906851e-01 3.58624250e-01 -8.08923483e-01 8.35538387e-01 -2.77061369e-02 8.63885820e-01 1.05375350e+00 5.79392016e-01 1.84796885e-01 7.61451721e-01 -9.58975852e-02 2.03592062e-01 -2.22852662e-01 -1.07361399e-01 8.69280398e-01 1.33949801e-01 4.47887570e-01 -7.28629649e-01 -1.16403952e-01 9.06262159e-01 -3.01683228e-02 -3.95170212e-01 -3.04106534e-01 -1.24673438e+00 5.19128799e-01 3.81479293e-01 7.25867078e-02 -4.58399385e-01 1.87355638e-01 5.43530166e-01 4.13760036e-01 3.76726329e-01 7.11837709e-01 -4.18506086e-01 1.19134588e-02 -7.80533731e-01 1.54796988e-01 6.78757012e-01 5.18871307e-01 4.81569886e-01 2.94567376e-01 1.96246922e-01 1.86671495e-01 1.89165369e-01 4.54694390e-01 3.33898515e-01 -1.25573587e+00 3.48119251e-02 4.25922483e-01 2.78996646e-01 -8.72450352e-01 -4.73565638e-01 -4.53158945e-01 -2.46585861e-01 8.89540493e-01 6.42767072e-01 -3.72198820e-01 -6.47580445e-01 1.15094888e+00 4.43445332e-02 2.37309709e-01 9.44804698e-02 1.33399045e+00 5.80233216e-01 4.34406251e-01 6.72205389e-02 1.32499933e-01 1.43834734e+00 -5.40330052e-01 -4.02886480e-01 -4.09725934e-01 3.56103420e-01 -5.22701383e-01 7.24549890e-01 4.01058882e-01 -8.84035349e-01 -6.80862308e-01 -1.14620221e+00 3.22105438e-01 -1.51395455e-01 2.17120215e-01 1.10858001e-01 2.32393190e-01 -8.24095428e-01 7.69465029e-01 -1.13652122e+00 -3.81107092e-01 3.69619727e-01 4.01757658e-01 -5.45920193e-01 1.18708730e-01 -7.49312639e-01 9.90711391e-01 -1.91224575e-01 6.55251443e-01 -1.67408323e+00 -2.82281876e-01 -7.88381338e-01 2.52625465e-01 7.60063976e-02 -1.20249815e-01 1.73866117e+00 -1.46457350e+00 -1.27745736e+00 9.79955435e-01 1.56335399e-01 -5.64483166e-01 6.40803576e-01 -3.90515357e-01 -4.08708155e-02 4.77952898e-01 2.18635455e-01 4.51330572e-01 8.34003329e-01 -8.00308049e-01 -1.00085473e+00 -5.42356431e-01 2.17683956e-01 3.40739518e-01 2.26918042e-01 1.78898796e-01 1.43114656e-01 -4.56153721e-01 -1.90527990e-01 -7.58003414e-01 2.33798809e-02 5.87943852e-01 2.50474036e-01 1.10843040e-01 1.09597337e+00 -2.54604638e-01 1.02073774e-01 -2.35493684e+00 -1.04349405e-01 -2.05164120e-01 -2.13199645e-01 3.48733962e-01 -3.06791246e-01 4.45081919e-01 -1.72646791e-01 -2.89305627e-01 2.85679065e-02 1.54286074e-02 -5.76363742e-01 1.71053424e-01 -2.43193358e-01 1.01867664e+00 1.24138266e-01 1.86167732e-01 -1.06468093e+00 -1.03222162e-01 1.36252493e-01 3.85772228e-01 -2.05284595e-01 5.72685540e-01 -1.04240432e-01 5.13669014e-01 -7.51998276e-02 5.49752474e-01 1.06398486e-01 3.16561520e-01 6.04704693e-02 -2.42978379e-01 -5.16494274e-01 2.04912517e-02 -6.35973155e-01 1.23817384e+00 -9.57185179e-02 1.42445469e+00 2.34616920e-01 -1.00851059e+00 8.05007875e-01 4.47204620e-01 8.61746296e-02 -5.42433619e-01 2.46421084e-01 -1.01482205e-01 3.45930785e-01 -1.14902532e+00 3.64928663e-01 -3.40083331e-01 2.28783742e-01 2.93176919e-01 7.88506344e-02 3.29230279e-01 1.13654323e-01 -3.15270483e-01 1.13494492e+00 2.80727804e-01 -1.84283182e-01 -3.99395436e-01 1.78791136e-02 1.88245282e-01 3.68861258e-01 8.10367346e-01 -3.77780199e-01 4.41993475e-01 5.71163297e-01 -1.04832685e+00 -6.70838118e-01 -5.69221079e-01 1.47385955e-01 1.47672844e+00 2.88403511e-01 2.88751781e-01 -6.49182439e-01 -4.07677233e-01 -2.63309270e-01 5.59471071e-01 -1.09605825e+00 -3.35477501e-01 -5.29960930e-01 -3.74968290e-01 7.41883039e-01 5.34845233e-01 7.25336611e-01 -1.17501533e+00 -1.77027535e+00 1.01299323e-01 -3.45438831e-02 -9.72426772e-01 -1.61556199e-01 3.60015750e-01 -8.25935841e-01 -1.49919760e+00 -3.20503354e-01 -7.04091370e-01 5.68703353e-01 3.67908686e-01 7.03616440e-01 1.16736153e-02 -2.10095838e-01 4.55468953e-01 -4.25759166e-01 -5.14105618e-01 -2.55458176e-01 -4.84140664e-01 -4.78312559e-02 2.72184670e-01 5.42451560e-01 -2.33584940e-01 -7.36575246e-01 4.61464167e-01 -6.94811344e-01 -3.43650877e-01 5.07194102e-01 4.94535446e-01 -7.05890805e-02 1.22155815e-01 -2.52140015e-01 -2.83009827e-01 5.35490178e-02 -2.98130661e-01 -8.03970695e-01 -9.44109187e-02 2.23893240e-01 -9.72905159e-02 4.22437102e-01 -4.49388236e-01 -6.88639104e-01 4.06957597e-01 3.14056605e-01 -6.55250132e-01 -4.63328511e-01 3.09482843e-01 2.81381428e-01 -1.16728954e-01 7.22944856e-01 2.58347929e-01 3.35804850e-01 -2.47227892e-01 -5.46785653e-01 5.24080753e-01 7.47138202e-01 2.77726173e-01 4.95633364e-01 6.67148113e-01 -1.04443319e-01 -1.11435437e+00 -7.15261996e-01 -1.87057436e-01 -5.08937776e-01 -7.18626559e-01 1.06246376e+00 -8.63817096e-01 -1.01645839e+00 7.77635038e-01 -1.20295918e+00 -5.69369674e-01 -2.88508609e-02 6.21083796e-01 -2.85328358e-01 6.12951107e-02 -3.55592728e-01 -9.11641419e-01 -2.18505278e-01 -1.16054213e+00 1.04354608e+00 1.89264357e-01 3.50860581e-02 -5.50395727e-01 -5.11927605e-02 2.32780352e-01 4.05814737e-01 9.88335758e-02 4.63640511e-01 -7.11032033e-01 -2.19879568e-01 -5.14767468e-01 2.30157495e-01 3.82149696e-01 1.84194501e-02 3.14268351e-01 -1.13780308e+00 -2.09109455e-01 1.81270063e-01 -1.53863266e-01 6.47012949e-01 3.40761989e-01 6.02551520e-01 -1.54728055e-01 -7.50657618e-02 3.17829996e-01 1.09199965e+00 5.13737798e-01 5.44940114e-01 3.67861420e-01 1.97691247e-01 9.30122316e-01 4.98552829e-01 2.76477307e-01 -1.36531457e-01 5.80654800e-01 9.54821110e-01 2.79073529e-02 3.20905030e-01 1.38113514e-01 8.03681076e-01 -4.39728588e-01 -2.36019403e-01 -3.22311789e-01 -8.01177025e-01 5.48134089e-01 -1.44467962e+00 -1.18810940e+00 -1.84190333e-01 2.20168257e+00 1.03191599e-01 1.35937244e-01 2.23821446e-01 5.93954884e-02 8.03312480e-01 -4.43780310e-02 -4.48493600e-01 -2.43815750e-01 2.67406195e-01 -2.56560892e-01 8.41761410e-01 3.40931028e-01 -1.33305585e+00 5.31676292e-01 6.23996925e+00 -1.55614868e-01 -1.42362142e+00 -3.39590013e-01 3.41394007e-01 -2.74612159e-01 8.75596642e-01 -2.76200175e-01 -4.71166551e-01 3.98830503e-01 8.97250891e-01 4.77544010e-01 1.61268279e-01 8.28364670e-01 4.84032005e-01 -4.14500922e-01 -1.29370809e+00 4.94687945e-01 2.22677574e-01 -1.04062021e+00 -2.11140797e-01 1.03272386e-01 -7.50447586e-02 2.32690513e-01 -1.96429253e-01 -1.50218057e-02 6.37091547e-02 -8.15163195e-01 1.04902852e+00 6.62163317e-01 4.61009234e-01 -3.11363041e-01 1.07880652e+00 4.29807156e-01 -8.38001788e-01 -1.72394797e-01 -2.13485986e-01 -6.42441988e-01 -1.85898572e-01 -2.61050165e-01 -7.65880823e-01 -2.56813675e-01 1.09498012e+00 5.14867365e-01 -5.53560197e-01 1.14568079e+00 -1.26991272e-01 6.99543834e-01 -2.63465166e-01 -2.01377094e-01 5.36571980e-01 3.97498980e-02 7.14628994e-01 1.23283875e+00 1.24828726e-01 2.69407123e-01 -3.20280731e-01 4.46756780e-01 1.91056624e-01 -6.63452208e-01 -7.93873906e-01 3.00827101e-02 -4.70495149e-02 1.06226182e+00 -6.61186934e-01 -1.31923854e-01 -3.78251255e-01 7.43015766e-01 -1.25860617e-01 2.69693553e-01 -5.41900635e-01 -4.87236083e-01 8.09319317e-01 2.41265669e-01 6.98323667e-01 -1.11769401e-01 2.32933939e-01 -7.34567881e-01 -1.31470680e-01 -6.99340463e-01 2.62469143e-01 -1.11670148e+00 -1.03065765e+00 6.73211634e-01 -8.52346234e-03 -1.53102243e+00 -1.64064139e-01 -7.69562006e-01 -8.28695834e-01 5.40942788e-01 -1.16277695e+00 -6.59756422e-01 -6.68330491e-01 1.31826550e-01 8.13316047e-01 -3.05868864e-01 6.42682076e-01 -1.92983949e-03 -3.82754952e-01 7.89018646e-02 -1.30738050e-01 7.03688025e-01 2.52970010e-01 -9.51299071e-01 1.32394806e-01 9.58925247e-01 -1.48734292e-02 1.02522530e-01 1.11817896e+00 -3.26967925e-01 -1.29744971e+00 -1.12030983e+00 5.36697149e-01 -4.31027800e-01 4.84708250e-01 -1.64811566e-01 -9.07702208e-01 8.55573893e-01 3.75391930e-01 1.58286065e-01 6.03313923e-01 -6.12071335e-01 1.26655875e-02 -1.93128452e-01 -9.52007234e-01 2.63508469e-01 5.77200115e-01 -2.54306585e-01 -7.74962783e-01 3.18141848e-01 -2.07637563e-01 -3.31186622e-01 -2.29745641e-01 1.56505466e-01 9.27939713e-01 -1.14280868e+00 4.90363449e-01 -9.06010807e-01 6.93999290e-01 -4.23451215e-01 -2.54848562e-02 -1.16437995e+00 -2.35580772e-01 -1.57313287e-01 7.19297171e-01 4.35808182e-01 2.99421512e-02 -3.09952110e-01 7.31982291e-01 3.23683143e-01 -1.12098530e-01 -2.71042347e-01 -1.00227404e+00 -5.58584630e-01 -3.56420130e-01 -1.43072486e-01 -1.14708878e-01 6.21726096e-01 -1.89251915e-01 4.90410000e-01 -4.48007792e-01 6.76150322e-01 5.54631948e-01 -1.91959664e-02 7.72357643e-01 -1.15488875e+00 -1.64542764e-01 -2.36446083e-01 -9.83353317e-01 -8.56034279e-01 -1.08859099e-01 -2.05287740e-01 4.67915118e-01 -1.04619932e+00 -2.30899751e-01 4.79357868e-01 2.15291128e-01 6.53488934e-01 3.10749352e-01 4.87250805e-01 -7.73331597e-02 2.06415042e-01 -4.11337584e-01 3.63553435e-01 7.12698281e-01 -2.44267900e-02 1.41661540e-01 2.22297341e-01 -1.42542258e-01 9.40379143e-01 5.93789935e-01 -6.39074862e-01 1.21028926e-02 -9.10896719e-01 1.81999281e-02 2.58174211e-01 8.90114248e-01 -1.27458811e+00 4.38966483e-01 1.32045805e-01 4.45887923e-01 -3.58745307e-02 4.32596356e-01 -1.21929467e+00 1.16749048e-01 7.53185630e-01 -6.86604381e-01 1.03840813e-01 3.46071512e-01 7.21695602e-01 -2.01897696e-01 -3.25156212e-01 1.00408590e+00 -5.46332777e-01 -8.50705385e-01 -3.14932793e-01 -1.26911390e+00 -2.51710832e-01 9.81654644e-01 -4.29479569e-01 -3.84782851e-01 -3.89031589e-01 -6.85604692e-01 1.77922949e-01 1.51772425e-01 3.17901015e-01 5.69700241e-01 -4.96903777e-01 -8.68409753e-01 5.48833013e-01 6.72095269e-02 -1.65007889e-01 8.35291669e-02 8.92349005e-01 -1.11096966e+00 -2.87079029e-02 -4.98800695e-01 -6.67954803e-01 -1.54900205e+00 -1.76347494e-02 1.13457704e+00 4.86116976e-01 -6.01215780e-01 9.62667525e-01 1.47280395e-01 7.17141330e-02 5.00551641e-01 -2.75495410e-01 -2.40588158e-01 7.02929497e-02 6.70529246e-01 4.22573149e-01 6.82529807e-02 -5.39929926e-01 -4.03752595e-01 4.73897696e-01 7.83221796e-02 -6.39940649e-02 1.40885532e+00 2.19130784e-01 2.49353066e-01 6.68749213e-01 9.67252851e-01 -5.65352619e-01 -1.74211764e+00 -6.60474300e-02 -3.43896836e-01 -5.50155759e-01 2.97942549e-01 -7.88172662e-01 -1.28729212e+00 1.27295113e+00 9.53639925e-01 2.70758361e-01 8.93084407e-01 -2.36259282e-01 1.22074589e-01 8.66580606e-01 1.47863731e-01 -7.33065486e-01 -4.13321704e-02 5.05988896e-01 9.31148827e-01 -1.15626872e+00 -1.61522612e-01 4.12675261e-01 -9.20426250e-01 1.42940617e+00 6.81713343e-01 -3.87740135e-01 3.11748832e-01 3.12395096e-01 5.68153799e-01 -6.39719844e-01 -7.76550412e-01 -1.24521151e-01 -2.12777391e-01 4.18409437e-01 7.32297972e-02 -4.24385011e-01 4.88800526e-01 -8.44623372e-02 7.44257960e-03 -1.52217746e-02 8.48307371e-01 1.16671789e+00 -7.47348785e-01 3.51887569e-02 -2.05578715e-01 3.13229591e-01 -4.78495687e-01 1.44399062e-01 -5.70955336e-01 8.20366859e-01 2.91508287e-01 8.74071777e-01 6.85026765e-01 -4.47833508e-01 7.57883787e-01 8.25552456e-03 1.61312774e-01 -6.38008535e-01 -8.87496650e-01 6.72354177e-02 -8.19403678e-03 -3.84673297e-01 -7.76304305e-01 -8.83072436e-01 -7.74204373e-01 1.32290050e-01 -1.49509877e-01 3.19037378e-01 5.00076950e-01 9.80842769e-01 7.74198696e-02 2.46359050e-01 6.03546143e-01 -1.42118669e+00 -4.67428565e-01 -1.23965359e+00 -5.83597124e-01 5.89343905e-02 7.80502558e-01 -5.54811895e-01 -8.10725808e-01 4.66662675e-01]
[8.474173545837402, -1.0724670886993408]
14e6a65b-1ace-4c63-a174-06a7f7310a63
rl-dwa-omnidirectional-motion-planning-for
2211.04993
null
https://arxiv.org/abs/2211.04993v2
https://arxiv.org/pdf/2211.04993v2.pdf
RL-DWA Omnidirectional Motion Planning for Person Following in Domestic Assistance and Monitoring
Robot assistants are emerging as high-tech solutions to support people in everyday life. Following and assisting the user in the domestic environment requires flexible mobility to safely move in cluttered spaces. We introduce a new approach to person following for assistance and monitoring. Our methodology exploits an omnidirectional robotic platform to detach the computation of linear and angular velocities and navigate within the domestic environment without losing track of the assisted person. While linear velocities are managed by a conventional Dynamic Window Approach (DWA) local planner, we trained a Deep Reinforcement Learning (DRL) agent to predict optimized angular velocities commands and maintain the orientation of the robot towards the user. We evaluate our navigation system on a real omnidirectional platform in various indoor scenarios, demonstrating the competitive advantage of our solution compared to a standard differential steering following.
['Marcello Chiaberge', 'Mauro Martini', 'Andrea Eirale']
2022-11-09
null
null
null
null
['motion-planning']
['robots']
[-1.81640685e-01 4.95956987e-01 9.92682427e-02 -3.22314352e-01 8.14993829e-02 -5.34604251e-01 7.51629651e-01 -4.67078835e-01 -1.24119329e+00 1.07914937e+00 3.68199736e-01 -4.69176203e-01 -3.54995310e-01 -6.26910567e-01 -2.54490674e-01 -5.60614049e-01 -2.70662189e-01 9.16688144e-01 2.12614089e-01 -8.53105605e-01 6.28805012e-02 8.63049448e-01 -1.42878556e+00 -4.51256752e-01 9.99288797e-01 1.52384460e-01 6.65103495e-01 1.03657985e+00 6.94399416e-01 6.11741006e-01 -2.88430065e-01 4.88609016e-01 3.29744667e-01 2.17373103e-01 -7.98679352e-01 8.07303116e-02 -2.44088337e-01 -6.93140268e-01 -7.07032204e-01 3.88760448e-01 8.96675646e-01 7.23839819e-01 5.48032403e-01 -1.54357779e+00 6.69990480e-02 7.80588984e-02 1.33172438e-01 -1.32695794e-01 1.08125722e+00 2.74713159e-01 3.56692284e-01 -4.51275468e-01 9.29121315e-01 1.11045277e+00 7.90839911e-01 6.52564228e-01 -6.86865211e-01 1.68823287e-01 2.51501322e-01 4.48045582e-01 -1.18513346e+00 -5.03422439e-01 1.54537439e-01 -1.67597994e-01 1.16055894e+00 1.00559682e-01 7.19315827e-01 1.24259603e+00 3.70339245e-01 5.00836551e-01 2.35774949e-01 -4.12760466e-01 3.71269077e-01 1.69826910e-01 9.23330523e-03 9.25263941e-01 3.00418735e-01 9.59821567e-02 -1.71278313e-01 3.95893194e-02 7.30684578e-01 8.61333087e-02 -5.36260366e-01 -1.14715374e+00 -1.18484426e+00 5.91274977e-01 5.88081479e-01 -5.96994348e-03 -7.47214854e-01 1.04945235e-01 1.12771481e-01 2.54923999e-01 -5.38029194e-01 3.88612986e-01 -5.31892538e-01 -4.77569729e-01 -7.52966898e-03 8.18543077e-01 8.93360972e-01 1.42382002e+00 1.01360820e-01 -2.16475755e-01 -9.93440449e-02 6.15647018e-01 3.20859700e-01 5.61430275e-01 3.50944102e-01 -1.42753303e+00 6.56081021e-01 5.06751001e-01 1.37966716e+00 -6.86532915e-01 -1.36884058e+00 -9.94320959e-02 -5.89397430e-01 6.73725069e-01 9.11161840e-01 -8.04014862e-01 -2.72029698e-01 1.31244695e+00 6.62456036e-01 -7.79231846e-01 1.52053058e-01 1.13673151e+00 9.44667831e-02 3.67340267e-01 -2.53946006e-01 1.61233187e-01 1.17892706e+00 -1.27781224e+00 -6.01831019e-01 -6.76254153e-01 8.88267636e-01 -1.00354970e-01 7.85388768e-01 6.02178037e-01 -5.89989305e-01 -3.86120170e-01 -1.08236527e+00 -1.07983805e-01 -2.88274974e-01 2.48398542e-01 3.95021915e-01 5.48539460e-01 -1.20597517e+00 7.76705682e-01 -9.04974163e-01 -1.08768415e+00 -2.42873505e-01 6.86227560e-01 -5.23698390e-01 -1.56286493e-01 -1.04858077e+00 1.29725718e+00 -9.30121616e-02 2.68414229e-01 -4.54734325e-01 -1.29795834e-01 -1.02754545e+00 -3.85261059e-01 1.17888980e-01 -9.92221415e-01 1.62997687e+00 2.12833688e-01 -1.92428267e+00 3.82828981e-01 -7.05980211e-02 -4.38945800e-01 1.07351649e+00 -8.51025701e-01 -2.48176921e-02 -1.27788514e-01 3.88151944e-01 4.99379128e-01 3.04519802e-01 -1.01660991e+00 -1.16258216e+00 -4.93956774e-01 2.02826560e-01 9.57870185e-01 -8.09528455e-02 -4.85125244e-01 -1.93950504e-01 5.77957332e-02 1.36889443e-01 -1.39612544e+00 -7.49602675e-01 2.66205072e-01 -1.70891702e-01 -1.33422375e-01 6.82638645e-01 -4.73467201e-01 5.19686282e-01 -1.56793749e+00 2.84868926e-01 1.53934762e-01 -3.17909807e-01 -1.44144133e-01 2.22332329e-01 3.73032779e-01 6.36660576e-01 -8.92397106e-01 1.34980842e-01 -3.75246704e-01 2.14797929e-01 5.10424137e-01 -8.27998072e-02 7.53245831e-01 -6.74499989e-01 3.87645870e-01 -1.30650759e+00 -7.93664008e-02 4.45667297e-01 6.36792600e-01 -5.30223370e-01 3.98952425e-01 4.22857732e-01 6.37888253e-01 -5.77965140e-01 3.19154739e-01 3.60771447e-01 4.87358749e-01 5.37589967e-01 4.27682489e-01 -6.00217998e-01 2.51068175e-01 -1.52510703e+00 1.50648248e+00 -7.44295537e-01 2.50792801e-01 7.97897220e-01 -7.15007186e-01 7.46374667e-01 2.11004004e-01 4.44451660e-01 -8.25028300e-01 1.42792940e-01 2.18060672e-01 -5.16879439e-01 -5.54224014e-01 8.98405254e-01 5.60743451e-01 -3.26043516e-01 2.40371332e-01 -3.02980453e-01 8.46023578e-03 -1.13857433e-01 -2.18070209e-01 1.43586254e+00 7.97688663e-01 5.16746640e-01 -4.21329319e-01 6.42919183e-01 3.77947897e-01 2.73714215e-01 1.23980296e+00 -8.29069555e-01 1.32091165e-01 -2.56743491e-01 -7.11583972e-01 -1.05706215e+00 -1.20535481e+00 2.34084636e-01 1.56363487e+00 5.01863122e-01 -8.36894140e-02 -6.33423805e-01 -3.83390933e-01 1.70832783e-01 7.29252815e-01 -8.17828998e-02 1.87800705e-01 -1.20564508e+00 -4.14442182e-01 3.34243357e-01 5.86187661e-01 5.98853469e-01 -1.06639087e+00 -1.66139305e+00 4.91952926e-01 -4.74820435e-01 -1.27426016e+00 -5.03354728e-01 3.15223873e-01 -4.32682633e-01 -9.28201735e-01 -6.94695830e-01 -1.19611406e+00 6.00292087e-01 5.87692678e-01 4.12913412e-01 -3.51398826e-01 -4.27590907e-02 9.56310213e-01 -2.40324929e-01 -2.65272260e-01 -7.75042325e-02 4.54648063e-02 8.68378639e-01 -5.03020406e-01 -3.25924419e-02 -5.81691861e-01 -8.26740623e-01 4.51613039e-01 4.61198360e-01 4.36184295e-02 1.47815481e-01 5.87827384e-01 -1.01449922e-01 -1.54753506e-01 1.80263877e-01 -9.61450636e-02 8.97722304e-01 -2.49423981e-01 -4.59656149e-01 -1.49293646e-01 -5.25347769e-01 7.30285272e-02 7.86403656e-01 -1.93084404e-01 -1.19172442e+00 5.71346462e-01 4.90728840e-02 8.25181305e-01 -3.25209081e-01 -1.76412836e-01 -2.52429247e-01 7.02354982e-02 8.53353381e-01 8.63728002e-02 1.63691603e-02 -2.94760317e-01 5.73643446e-01 8.95495892e-01 1.14661956e+00 -3.69302452e-01 6.82059169e-01 6.35824561e-01 4.33099233e-02 -1.03770101e+00 8.13431367e-02 -7.06396520e-01 -1.12805378e+00 -4.16401207e-01 6.55532539e-01 -7.70114303e-01 -1.33390760e+00 4.90500510e-01 -1.29503465e+00 -7.84766138e-01 -3.10963932e-02 4.12661821e-01 -1.10759211e+00 3.64540279e-01 -3.42479050e-01 -1.04133534e+00 -3.62666130e-01 -9.18338180e-01 8.85294497e-01 4.58367616e-01 -7.64604032e-01 -7.52746999e-01 2.20206618e-01 2.75708605e-02 4.68209296e-01 2.69522637e-01 1.72259644e-01 -1.51654676e-01 1.90605726e-02 -2.60349602e-01 1.71779305e-01 -7.79107869e-01 1.21838219e-01 -9.34216440e-01 -4.44981545e-01 -5.03069043e-01 -2.72217542e-01 1.31563039e-03 -7.98039362e-02 2.94844419e-01 1.81714773e-01 -4.83184338e-01 -8.89664471e-01 3.71680707e-01 7.67527997e-01 4.42185193e-01 5.10563612e-01 1.39876926e+00 5.95861077e-01 8.54802191e-01 1.07102203e+00 8.81698310e-01 9.20122623e-01 1.25299394e+00 4.05406922e-01 4.96312022e-01 5.52225232e-01 -1.38581127e-01 5.45856953e-01 -1.96096867e-01 -5.00364959e-01 -6.19010441e-02 -8.67072642e-01 5.51214516e-01 -2.51228786e+00 -8.80436599e-01 -4.54589315e-02 2.10199857e+00 3.50541443e-01 6.69735074e-02 2.37056658e-01 2.04800531e-01 4.32792157e-01 -3.90673161e-01 -5.37285447e-01 -6.25708580e-01 4.41930443e-01 -3.36005598e-01 9.33388770e-01 1.15442669e+00 -1.16183770e+00 8.22469413e-01 6.01194143e+00 -4.64215487e-01 -5.39930284e-01 -2.11176381e-01 -5.28526723e-01 5.04983263e-03 5.73169708e-01 -4.17262286e-01 -8.56242776e-01 8.72832388e-02 7.14822471e-01 1.84351265e-01 7.16224670e-01 1.48270679e+00 7.24371791e-01 -3.50319415e-01 -1.08479774e+00 9.14176106e-01 -4.66519564e-01 -6.23590112e-01 -7.39973187e-01 4.69093397e-02 2.51789868e-01 -3.20275761e-02 -4.06482399e-01 6.02819800e-01 7.02903330e-01 -7.61022985e-01 8.34814250e-01 6.20817542e-01 2.93102562e-01 -8.63982201e-01 5.97848237e-01 1.00958574e+00 -1.29371881e+00 -7.62183964e-01 -2.15730757e-01 -6.96690619e-01 6.81343496e-01 -2.59323001e-01 -1.42866480e+00 2.79033393e-01 8.75079215e-01 1.11646816e-01 -9.87364799e-02 8.45834315e-01 -2.76166558e-01 -4.05149043e-01 -3.79658401e-01 -5.22055030e-01 3.32296848e-01 -4.76657927e-01 8.71158004e-01 1.20301151e+00 3.69912446e-01 -9.64336377e-03 1.40327528e-01 8.57293531e-02 7.33538449e-01 -2.96479613e-01 -9.39262033e-01 9.70292389e-01 3.33023131e-01 9.84682441e-01 -2.98228920e-01 1.04762554e-01 1.50803834e-01 1.65208614e+00 3.21357965e-01 2.72420675e-01 -6.30097687e-01 -8.68065119e-01 9.54774678e-01 2.00485468e-01 1.89653575e-01 -8.80879164e-01 -1.70742914e-01 -7.28048027e-01 2.53172874e-01 -4.30642933e-01 1.94318220e-01 -8.37416649e-01 -2.62852967e-01 3.87787640e-01 -9.35462862e-02 -1.28142822e+00 -9.54688609e-01 -9.02119398e-01 -4.26406711e-01 6.37198508e-01 -1.19823647e+00 -7.25876153e-01 -8.56420279e-01 7.31803179e-01 6.54011965e-01 -3.25534940e-01 1.09476221e+00 -1.53905138e-01 7.43884919e-03 3.61414284e-01 3.11152309e-01 -2.74318814e-01 7.29824841e-01 -1.45627403e+00 4.42665666e-01 6.63204074e-01 -8.90076458e-01 6.99623466e-01 1.18226254e+00 -4.37037110e-01 -1.54527080e+00 -8.76411796e-01 1.12709248e+00 -6.08334184e-01 1.45745084e-01 -3.23817343e-01 -1.21646740e-01 8.19109023e-01 -1.61167830e-01 -2.13993564e-01 -9.01539698e-02 -1.06729031e-01 3.76387179e-01 -7.10197017e-02 -1.46355629e+00 1.35000801e+00 1.62938809e+00 7.25302845e-02 -5.02901137e-01 2.34844774e-01 4.19877440e-01 -5.86330056e-01 -1.83646560e-01 8.64959359e-02 1.07202375e+00 -7.90523827e-01 1.04064643e+00 -3.01932484e-01 -5.61933815e-01 -4.28341925e-01 -2.39540473e-01 -1.50867283e+00 -5.84550261e-01 -1.06782806e+00 -1.53535411e-01 5.67570210e-01 4.06486504e-02 -6.26098275e-01 9.72439229e-01 9.71081674e-01 -2.07510203e-01 -3.34133834e-01 -1.08460760e+00 -5.47309279e-01 -4.93873537e-01 -4.04873967e-01 4.05503780e-01 1.28417119e-01 7.87017465e-01 2.93969959e-01 -6.72326505e-01 6.47626877e-01 5.96183062e-01 -6.98103070e-01 1.10251188e+00 -1.11832750e+00 -2.05648076e-02 -8.35289955e-02 -4.09089237e-01 -1.36635792e+00 2.99103148e-02 -4.23936367e-01 6.00725770e-01 -2.13523269e+00 -5.52935123e-01 -3.65050644e-01 4.61392790e-01 2.37360537e-01 4.05374378e-01 -3.74517053e-01 -1.10803828e-01 -1.76949240e-02 -6.61733985e-01 5.02479017e-01 8.95397604e-01 -2.23487522e-02 -8.00627351e-01 5.43865204e-01 -5.12385547e-01 9.10359919e-01 9.34758365e-01 3.40803862e-02 -4.85112906e-01 -3.31794560e-01 -1.62766427e-02 2.33791903e-01 4.25775558e-01 -1.35893869e+00 7.79151142e-01 -2.39034742e-01 5.28459132e-01 -4.69163120e-01 4.65136051e-01 -9.27376688e-01 -2.56869674e-01 1.14662921e+00 1.22684434e-01 1.89312398e-01 -1.78723022e-01 4.47078526e-01 8.70597899e-01 -8.56543928e-02 6.29851282e-01 -2.44617220e-02 -1.04412580e+00 -2.34573245e-01 -1.20361781e+00 -5.15397847e-01 9.32742655e-01 -3.43273431e-01 -2.05959603e-01 -8.30847144e-01 -8.52918923e-01 7.97058880e-01 4.70143944e-01 6.43527508e-01 6.29370093e-01 -1.04398799e+00 -1.43233880e-01 6.79136366e-02 -9.50678065e-03 4.96827178e-02 -8.22402313e-02 4.24750090e-01 -9.25176203e-01 8.97961617e-01 -5.68688333e-01 -3.84703726e-01 -1.13823521e+00 1.91489980e-01 6.60210907e-01 -1.95520714e-01 -1.07829881e+00 2.31274366e-01 -4.08077449e-01 -1.41802692e+00 6.18627489e-01 -2.01114416e-01 -6.95446253e-01 -5.01843750e-01 5.76799393e-01 1.06905556e+00 -2.01227978e-01 -5.37567496e-01 -7.28598952e-01 2.90449172e-01 2.65257657e-01 -6.50123417e-01 1.37644470e+00 -8.40939939e-01 4.12282109e-01 -2.90382095e-02 4.64447409e-01 -9.75937694e-02 -1.57560039e+00 8.85557663e-03 2.71503001e-01 -7.12096021e-02 -6.06509209e-01 -5.68958044e-01 2.35325634e-01 8.05395097e-03 8.34096968e-01 -1.73323750e-01 4.03134555e-01 -4.06849176e-01 5.08699000e-01 1.40914881e+00 1.06909466e+00 -1.52993298e+00 -9.02790874e-02 9.16053236e-01 1.14689338e+00 -1.08804953e+00 -6.74468651e-02 -1.15550131e-01 -8.28138947e-01 1.23473728e+00 8.11545908e-01 -1.73232302e-01 4.31762397e-01 3.38684410e-01 2.65503347e-01 4.61168289e-01 -1.39243826e-01 4.53654267e-02 -3.89901370e-01 1.55557942e+00 -4.87094931e-02 1.28095418e-01 -1.50658488e-01 6.88310921e-01 -7.17559814e-01 -6.37836531e-02 8.49338889e-01 1.47450912e+00 -9.58883524e-01 -7.19579279e-01 -6.85169756e-01 -2.52756029e-01 4.17893887e-01 8.36023092e-01 2.27750782e-02 9.57484365e-01 -2.86889989e-02 1.35069168e+00 4.57450897e-02 -3.06494623e-01 9.35063124e-01 1.43254427e-02 5.55136859e-01 -2.74026394e-01 -6.17646873e-02 -5.35966158e-01 3.58195037e-01 -9.22069311e-01 1.10879011e-01 -9.14113700e-01 -1.69840515e+00 -1.05657354e-01 4.60485697e-01 -4.18753102e-02 6.05910003e-01 9.76189673e-01 1.43816143e-01 3.45529586e-01 5.83133578e-01 -1.63942444e+00 -7.93830156e-01 -8.09035599e-01 -4.97505844e-01 -4.26742770e-02 7.13348985e-01 -7.44961262e-01 2.82816421e-02 -4.81374711e-01]
[4.800148963928223, 0.9056942462921143]
8b1809a3-4416-4009-8003-f0517269182a
mapping-probability-word-problems-to
null
null
https://aclanthology.org/2021.emnlp-main.294
https://aclanthology.org/2021.emnlp-main.294.pdf
Mapping probability word problems to executable representations
While solving math word problems automatically has received considerable attention in the NLP community, few works have addressed probability word problems specifically. In this paper, we employ and analyse various neural models for answering such word problems. In a two-step approach, the problem text is first mapped to a formal representation in a declarative language using a sequence-to-sequence model, and then the resulting representation is executed using a probabilistic programming system to provide the answer. Our best performing model incorporates general-domain contextualised word representations that were finetuned using transfer learning on another in-domain dataset. We also apply end-to-end models to this task, which bring out the importance of the two-step approach in obtaining correct solutions to probability problems.
['Walter Daelemans', 'Luc De Raedt', 'Jesse Davis', 'Angelika Kimmig', 'Pietro Totis', 'Pieter Fivez', 'Simon Suster']
null
null
null
null
emnlp-2021-11
['contextualised-word-representations']
['natural-language-processing']
[ 4.35402840e-01 4.16152447e-01 1.27351806e-01 -4.98841554e-01 -1.03615963e+00 -6.21393800e-01 6.00269675e-01 3.80054563e-01 -5.72251141e-01 8.90322685e-01 1.69464648e-01 -6.33793533e-01 -4.68918145e-01 -1.29272604e+00 -8.42276454e-01 -1.75811812e-01 3.26395482e-01 9.52432036e-01 3.76197547e-01 -6.70494437e-02 5.68014681e-01 1.93729863e-01 -1.37770462e+00 2.87511230e-01 9.06383157e-01 7.31245220e-01 3.10741693e-01 8.39168429e-01 -8.31998885e-01 8.21879089e-01 -6.51164532e-01 -5.80318213e-01 -1.71073005e-01 -1.98779404e-01 -1.30549300e+00 -3.88259560e-01 2.52602607e-01 5.50253429e-02 -8.09540153e-02 1.03850400e+00 2.62709469e-01 4.18617815e-01 1.04658055e+00 -7.57557631e-01 -8.23062122e-01 8.14666748e-01 -8.01070184e-02 3.33744496e-01 7.86528647e-01 -6.80060089e-02 1.36868227e+00 -7.59215891e-01 3.72294575e-01 1.44383585e+00 5.37761450e-01 5.48515856e-01 -1.53161073e+00 -3.09899181e-01 2.51812097e-02 3.81146997e-01 -1.27455735e+00 1.06387794e-01 5.70723116e-01 -4.37590241e-01 1.38884878e+00 -7.54332468e-02 4.73916829e-01 8.17391634e-01 1.18545249e-01 7.38277614e-01 8.42518449e-01 -7.84179986e-01 4.49398458e-01 1.33787662e-01 4.23281670e-01 5.79954147e-01 1.51661962e-01 -1.10836945e-01 -4.23921496e-01 -2.41788730e-01 5.83075345e-01 -4.39059258e-01 -3.55390497e-02 -2.04034954e-01 -9.35442090e-01 1.09181368e+00 1.37086332e-01 4.48887646e-01 -5.33181846e-01 2.71117687e-01 1.60894901e-01 1.61951110e-01 5.56263030e-01 9.29696977e-01 -7.84674883e-01 -3.35670561e-01 -1.26025379e+00 7.52946496e-01 1.31072915e+00 5.84973156e-01 7.32023001e-01 -2.44259447e-01 -2.98266232e-01 6.99753642e-01 4.81645465e-01 2.61478841e-01 5.90463102e-01 -7.16221511e-01 5.65301359e-01 3.98839355e-01 7.10563287e-02 -1.07418132e+00 -1.67032108e-01 -1.16728162e-02 -6.41507357e-02 -1.44076720e-01 6.53659284e-01 -2.90219691e-02 -8.75499487e-01 1.77304101e+00 1.74148813e-01 2.49034375e-01 2.79547811e-01 6.60426319e-01 7.84210026e-01 1.14187324e+00 4.97565210e-01 7.30739310e-02 1.46515560e+00 -5.75418174e-01 -4.92839307e-01 -4.10821319e-01 4.49087828e-01 -5.06914258e-01 1.02354968e+00 6.27421558e-01 -1.17519140e+00 -4.33099270e-01 -9.15201187e-01 -2.34688342e-01 -5.93985498e-01 -2.00715616e-01 2.59364098e-01 6.29640341e-01 -1.05289507e+00 7.30018854e-01 -4.56420690e-01 -3.58296424e-01 3.23851138e-01 2.19802603e-01 -8.45447835e-03 -3.99718106e-01 -1.55896938e+00 1.23881054e+00 8.34632277e-01 -2.28268713e-01 -6.19539678e-01 -8.32316101e-01 -1.04516852e+00 3.02276343e-01 4.41357821e-01 -8.87157798e-01 1.41894209e+00 -7.69890845e-01 -1.65940619e+00 8.32269073e-01 -2.97554940e-01 -5.44522822e-01 1.32228225e-01 -2.66172498e-01 -1.05446666e-01 1.03748791e-01 8.91413018e-02 5.82974255e-01 7.96178281e-01 -9.67666447e-01 -4.53602403e-01 -2.07560062e-01 2.51391351e-01 -8.57364163e-02 -1.11932270e-01 1.72146708e-01 -3.35531652e-01 -6.20153427e-01 -3.83749567e-02 -5.01310110e-01 -2.63583571e-01 -3.97153735e-01 -1.32303625e-01 -9.52774942e-01 1.10425949e-01 -6.71131790e-01 1.22488987e+00 -1.83932948e+00 4.48053122e-01 3.47497970e-01 -2.52869260e-02 4.06474054e-01 -2.84176677e-01 3.71939421e-01 -2.98659354e-01 1.49237067e-01 -3.85227412e-01 -1.94714963e-01 4.09783036e-01 3.90296787e-01 -7.85369694e-01 -1.15439412e-03 6.19005203e-01 9.70138729e-01 -1.00661623e+00 -5.88932812e-01 2.66347468e-01 3.48028660e-01 -7.24391043e-01 6.52273893e-01 -9.23538029e-01 -2.18065187e-01 -5.12185991e-01 1.96250111e-01 4.01122451e-01 -2.43704803e-02 2.38547161e-01 2.29595944e-01 1.58950612e-01 7.54645050e-01 -1.30635822e+00 1.68441212e+00 -7.04680145e-01 4.17991489e-01 -5.53212821e-01 -1.29074502e+00 9.51910794e-01 2.22887725e-01 -2.98021417e-02 -4.10860270e-01 8.42724890e-02 1.42106548e-01 -1.11269534e-01 -6.81830525e-01 6.55414045e-01 -4.99437094e-01 -2.62419194e-01 5.69421768e-01 3.99844497e-01 -7.39713728e-01 4.60974216e-01 2.63638556e-01 8.84221792e-01 6.32170975e-01 4.49912608e-01 -1.96526781e-01 6.77799404e-01 2.70985156e-01 1.00044489e-01 8.27777743e-01 9.67473462e-02 7.04550505e-01 7.51585841e-01 -4.89274710e-01 -7.97773540e-01 -1.08876979e+00 -3.27130444e-02 1.13169408e+00 -3.72602552e-01 -5.39584458e-01 -8.29844534e-01 -5.88481128e-01 -3.03940643e-02 1.40734065e+00 -5.59372127e-01 -2.40315676e-01 -4.93559003e-01 -4.44617093e-01 4.42964733e-01 4.62704718e-01 2.06574276e-02 -1.44877505e+00 -6.19039834e-01 4.99359488e-01 -8.30866024e-02 -9.08172429e-01 -1.64395217e-02 4.27744001e-01 -7.18598604e-01 -1.02017176e+00 -6.82045519e-01 -7.92198360e-01 3.54006559e-01 -3.77122760e-01 1.45837057e+00 -1.71783641e-01 -3.41279775e-01 5.32280803e-01 -2.43974969e-01 -5.66653788e-01 -3.65766764e-01 2.19720542e-01 -1.02008171e-01 -2.96771556e-01 9.04127836e-01 -4.78299052e-01 1.95771526e-03 -3.00518781e-01 -1.06631637e+00 -3.73397142e-01 4.96569723e-01 5.39066195e-01 3.78910393e-01 8.88175443e-02 4.66645449e-01 -8.89847398e-01 1.02503192e+00 -6.39328003e-01 -7.52832830e-01 2.76760161e-01 -4.01971638e-01 5.50390303e-01 5.51187873e-01 -4.98646647e-01 -9.28042591e-01 1.03729099e-01 -3.42921764e-01 -1.52361050e-01 -4.73369449e-01 9.32979345e-01 -1.53813884e-01 3.88520330e-01 7.43602276e-01 3.55168521e-01 -1.74156561e-01 -2.19001040e-01 6.63924992e-01 3.05368513e-01 4.02523726e-01 -1.34313536e+00 7.90227890e-01 -4.00607914e-01 -3.32959741e-01 -7.06566393e-01 -1.20141590e+00 -3.01384211e-01 -4.48408008e-01 1.10457689e-01 9.75581467e-01 -4.52062517e-01 -5.25022566e-01 6.53396100e-02 -1.46188450e+00 -4.58941281e-01 -4.14769351e-01 3.05133402e-01 -6.70729041e-01 4.11669254e-01 -3.18531215e-01 -9.83052194e-01 -2.08433226e-01 -9.20790076e-01 8.90508652e-01 3.03454220e-01 -7.37755001e-01 -1.12230885e+00 4.01620388e-01 1.82079747e-01 5.26523471e-01 -3.15515041e-01 1.54814684e+00 -1.01982296e+00 -3.62752318e-01 -2.70546675e-01 -3.35800439e-01 3.34233403e-01 -3.78025591e-01 -2.10176915e-01 -8.12844872e-01 5.03752172e-01 3.27838324e-02 -6.80241168e-01 8.20003331e-01 2.90327758e-01 1.44146609e+00 -1.20435029e-01 -1.39143035e-01 1.18502542e-01 1.37990391e+00 -2.58212119e-01 8.32800388e-01 3.52540523e-01 3.06161046e-01 7.32324362e-01 4.08552438e-01 1.61623523e-01 6.75501943e-01 3.54605317e-01 1.31947085e-01 5.07448554e-01 1.06850863e-01 -3.64796489e-01 2.22043529e-01 3.78338814e-01 2.06739038e-01 -2.98047543e-01 -1.22156107e+00 8.27446163e-01 -1.82720006e+00 -8.97553325e-01 6.15716167e-02 1.74668050e+00 1.30043089e+00 1.63041592e-01 -1.29637361e-01 1.65860057e-01 4.64114010e-01 8.92351493e-02 -1.88448268e-03 -7.11071253e-01 3.81330103e-01 9.27364111e-01 -4.77564819e-02 6.55629516e-01 -9.00484025e-01 1.28294408e+00 6.70091820e+00 1.04155087e+00 -8.13283384e-01 -1.81066826e-01 4.19174731e-01 2.85611570e-01 -4.00914311e-01 -1.88273545e-02 -6.93228126e-01 1.68138728e-01 1.29276168e+00 -1.85575694e-01 5.22379041e-01 1.01044893e+00 -2.07151920e-01 -4.80220646e-01 -1.41678584e+00 7.44707763e-01 2.07356364e-01 -1.26663625e+00 2.63563961e-01 -3.14401448e-01 4.06958759e-01 -4.13206875e-01 -1.05932996e-01 7.12162971e-01 5.43306887e-01 -1.35971534e+00 6.83462143e-01 7.90916979e-01 3.76466841e-01 -9.20282483e-01 7.83264518e-01 5.73066771e-01 -8.40374589e-01 2.75503784e-01 -5.50343454e-01 -3.79670382e-01 1.76000372e-01 6.86603904e-01 -9.80211794e-01 5.41388929e-01 4.23292339e-01 3.02098125e-01 -3.99874777e-01 8.22051287e-01 -7.26708055e-01 8.05171847e-01 -3.08239430e-01 -6.36717796e-01 2.68176168e-01 -1.28850415e-01 3.13496590e-01 1.34499514e+00 2.32599765e-01 2.66525477e-01 5.70661984e-02 1.38338459e+00 1.76226482e-01 7.53295422e-03 -4.21122253e-01 -3.02786708e-01 1.06549278e-01 1.04934859e+00 -4.21591043e-01 -4.96371567e-01 -2.75352418e-01 8.10696185e-01 7.84510374e-01 2.81459540e-01 -6.79357946e-01 -6.49608076e-01 3.11333627e-01 -1.33370921e-01 4.85697240e-01 -2.08680242e-01 -4.63947564e-01 -1.02591598e+00 -1.25246257e-01 -7.55951226e-01 4.98173922e-01 -1.01063681e+00 -1.58443701e+00 3.47282320e-01 3.12116086e-01 -2.75666058e-01 -5.54617345e-01 -8.88003111e-01 -6.80676401e-01 1.30569673e+00 -1.58245683e+00 -7.45836616e-01 2.63358563e-01 3.87201816e-01 4.80087191e-01 -9.02222171e-02 1.18773413e+00 -1.51444199e-02 -2.30533347e-01 2.42127940e-01 -4.87749070e-01 4.78693433e-02 4.84191239e-01 -1.50301874e+00 3.87357622e-01 7.85016000e-01 2.97497064e-01 1.03687787e+00 9.38775241e-01 -6.08682871e-01 -1.39594483e+00 -1.02864802e+00 1.59672630e+00 -7.88208723e-01 9.36851621e-01 -2.55670518e-01 -1.10048091e+00 5.68356216e-01 3.56842428e-01 -1.67418703e-01 7.53333449e-01 3.13166112e-01 -5.33610106e-01 3.40906888e-01 -1.15603292e+00 4.05764967e-01 3.80378723e-01 -5.55348396e-01 -1.68942785e+00 2.68967569e-01 7.75383413e-01 -3.96947473e-01 -5.65039396e-01 -1.13332890e-01 2.35522464e-01 -3.85876507e-01 9.07915771e-01 -9.23917413e-01 1.00447309e+00 -2.10153714e-01 -3.24814707e-01 -1.50183582e+00 -2.27382958e-01 -2.74275929e-01 -8.44041035e-02 1.13576365e+00 6.77416801e-01 -5.39931953e-01 6.55418575e-01 7.80395925e-01 1.29174024e-01 -7.16855288e-01 -8.87607396e-01 -3.50570381e-01 6.11565769e-01 -1.06576765e+00 5.25558770e-01 6.36959195e-01 2.36126706e-01 4.40379828e-01 -1.55198577e-04 1.46431699e-01 3.13978553e-01 2.02973589e-01 6.08184993e-01 -1.24315143e+00 -4.25230354e-01 -3.96834284e-01 -1.85749516e-01 -1.07359695e+00 6.44848228e-01 -1.04114258e+00 5.16429424e-01 -1.78529191e+00 7.97305405e-02 -2.47457683e-01 -9.43086445e-02 4.72336471e-01 -4.97170091e-01 2.01095100e-02 -4.92454227e-03 -4.30943191e-01 -5.58901191e-01 4.62187707e-01 5.86563826e-01 -1.49283782e-01 1.55763656e-01 -3.82768922e-02 -7.38915920e-01 6.42329872e-01 7.73591101e-01 -6.88818574e-01 -1.82971194e-01 -5.30170262e-01 6.44746721e-01 -3.29261906e-02 3.17514896e-01 -8.63341093e-01 2.78182298e-01 -3.87290418e-01 3.16646129e-01 -4.60389405e-01 2.66270995e-01 -7.27067828e-01 -5.83275437e-01 2.23956674e-01 -4.92571920e-01 -5.50788492e-02 4.36486751e-01 5.76503217e-01 -2.19920963e-01 -8.28400850e-01 5.79273582e-01 -3.14741045e-01 -6.52323961e-01 -3.84212546e-02 -6.68573201e-01 3.74036193e-01 8.67960393e-01 2.83469576e-02 2.05924273e-01 -1.75789386e-01 -7.13235915e-01 2.51455218e-01 -5.91955557e-02 3.35187018e-01 6.43706560e-01 -9.42195237e-01 -7.70392418e-01 -6.60561547e-02 4.16694023e-02 3.11216533e-01 -9.98481512e-02 2.38425240e-01 -5.90605497e-01 6.14788592e-01 1.29099622e-01 -4.19046313e-01 -7.55466342e-01 6.95711315e-01 2.83503592e-01 -4.24948871e-01 -3.05846453e-01 1.05478644e+00 -1.95914760e-01 -7.98577130e-01 3.13310117e-01 -5.08564055e-01 -3.17174941e-01 -1.60364192e-02 7.11865127e-01 1.52632901e-02 -8.60400796e-02 -9.58106220e-02 -3.63464773e-01 5.17307460e-01 8.02861899e-03 -4.70484734e-01 1.50145555e+00 2.94958502e-01 -2.17992455e-01 4.45127934e-01 8.03894043e-01 -2.01912254e-01 -6.03836358e-01 -4.51762497e-01 4.24429476e-01 -2.70605952e-01 -3.60233262e-02 -8.71578753e-01 -4.77335870e-01 9.45979416e-01 -3.01225446e-02 2.73458898e-01 6.91744268e-01 3.21823269e-01 6.02348626e-01 6.74072266e-01 1.82735875e-01 -9.31530058e-01 1.07984237e-01 1.21022391e+00 7.22639084e-01 -1.05437827e+00 -1.35795414e-01 -3.40791643e-01 -4.82819915e-01 1.45835590e+00 4.77478832e-01 -3.64291698e-01 4.37225491e-01 8.43209103e-02 -2.45143846e-01 -2.62195855e-01 -6.93027854e-01 -2.29787350e-01 3.94760728e-01 6.88824356e-01 5.43697715e-01 -7.67276064e-02 -2.71914840e-01 1.02872419e+00 -5.00785291e-01 2.50258803e-01 3.64662319e-01 7.68264472e-01 -5.82930863e-01 -1.21731532e+00 -4.58355665e-01 2.94035882e-01 -3.44286352e-01 -4.83728230e-01 -4.22669202e-01 5.11469364e-01 -3.85355391e-02 8.55166435e-01 1.85376462e-02 7.18055442e-02 4.06159759e-01 7.34556913e-01 6.94534719e-01 -1.06303906e+00 -6.52121842e-01 -4.65089887e-01 1.81788206e-01 -3.54943871e-01 -1.31087601e-01 -5.12630522e-01 -1.28491485e+00 -1.91757791e-02 8.34524911e-03 3.60243618e-01 6.27438962e-01 1.45796216e+00 -3.34300436e-02 7.59255707e-01 -8.06020722e-02 -7.69102871e-01 -7.45020866e-01 -9.99708891e-01 -2.05800697e-01 9.08425525e-02 -5.65632172e-02 -3.59283119e-01 -2.08489895e-01 -4.15817983e-02]
[9.555259704589844, 7.5344133377075195]
3122e6e4-fee6-47fa-97d3-702fc5faff17
generating-videos-with-scene-dynamics
1609.02612
null
http://arxiv.org/abs/1609.02612v3
http://arxiv.org/pdf/1609.02612v3.pdf
Generating Videos with Scene Dynamics
We capitalize on large amounts of unlabeled video in order to learn a model of scene dynamics for both video recognition tasks (e.g. action classification) and video generation tasks (e.g. future prediction). We propose a generative adversarial network for video with a spatio-temporal convolutional architecture that untangles the scene's foreground from the background. Experiments suggest this model can generate tiny videos up to a second at full frame rate better than simple baselines, and we show its utility at predicting plausible futures of static images. Moreover, experiments and visualizations show the model internally learns useful features for recognizing actions with minimal supervision, suggesting scene dynamics are a promising signal for representation learning. We believe generative video models can impact many applications in video understanding and simulation.
['Antonio Torralba', 'Hamed Pirsiavash', 'Carl Vondrick']
2016-09-08
generating-videos-with-scene-dynamics-1
http://papers.nips.cc/paper/6194-generating-videos-with-scene-dynamics
http://papers.nips.cc/paper/6194-generating-videos-with-scene-dynamics.pdf
neurips-2016-12
['self-supervised-action-recognition']
['computer-vision']
[ 3.17979306e-01 4.01180416e-01 -1.17600001e-01 -2.97610939e-01 -4.34997082e-01 -6.25614643e-01 9.36740398e-01 -6.95094049e-01 4.29141782e-02 6.46272421e-01 6.94498956e-01 -4.77908254e-01 6.65628314e-01 -8.82359207e-01 -1.44051516e+00 -5.89306533e-01 -4.88650113e-01 7.86582939e-03 3.41790259e-01 -1.60452202e-01 -8.29293281e-02 4.10344481e-01 -1.42743802e+00 7.26922870e-01 3.17067474e-01 6.80085778e-01 7.81692117e-02 1.25649273e+00 3.74373704e-01 1.83122659e+00 -7.11265147e-01 -1.23968944e-01 4.07931387e-01 -6.17751122e-01 -6.85215116e-01 4.04333740e-01 5.33256531e-01 -9.76274252e-01 -1.24901974e+00 5.35454571e-01 -6.36195624e-03 3.21190268e-01 6.21720076e-01 -1.48296690e+00 -7.61817157e-01 6.22673094e-01 -2.72278607e-01 4.54770118e-01 5.24715483e-01 8.53978276e-01 4.08972144e-01 -4.50986028e-01 9.36415255e-01 1.40030921e+00 4.27509904e-01 9.14957881e-01 -1.12158978e+00 -6.38650298e-01 5.25203407e-01 2.83843070e-01 -1.02528131e+00 -6.42872930e-01 5.62927246e-01 -5.89876890e-01 9.07447219e-01 2.44054228e-01 9.00966227e-01 1.89586854e+00 3.19107443e-01 1.08827281e+00 4.94718313e-01 2.34950315e-02 1.72395349e-01 -3.11438203e-01 -4.54929262e-01 7.08154559e-01 -1.17506318e-01 3.40049684e-01 -3.96875620e-01 1.77932158e-01 1.42490292e+00 9.48597342e-02 -3.12532514e-01 -2.29794458e-01 -1.31036401e+00 6.98731124e-01 4.21264321e-01 -1.53484613e-01 -3.61454725e-01 9.20575440e-01 1.18281648e-01 1.77327663e-01 2.52804369e-01 3.93669456e-01 -2.51194146e-02 -3.72333080e-01 -8.65306973e-01 4.21795964e-01 6.25125587e-01 1.14064467e+00 4.14793998e-01 9.29873168e-01 -4.33465123e-01 3.75222005e-02 -6.51890039e-02 6.82773948e-01 2.76619017e-01 -1.58345115e+00 2.42361575e-01 1.89925358e-01 2.39266366e-01 -9.19868052e-01 1.39468804e-01 2.76788473e-01 -7.01465964e-01 3.65925193e-01 4.52860802e-01 -6.10269010e-01 -1.23836923e+00 1.71056545e+00 -1.06364889e-02 1.10093439e+00 1.03685632e-01 9.20588553e-01 6.77779377e-01 1.17966211e+00 1.81832999e-01 -3.15574259e-02 8.16067219e-01 -1.00386727e+00 -2.97945082e-01 -2.97621489e-01 1.61294088e-01 -3.00791711e-01 7.62016058e-01 2.38250345e-01 -1.31751442e+00 -7.40143776e-01 -6.03765786e-01 3.85132581e-02 7.57499561e-02 -2.02460900e-01 9.47085142e-01 3.63061279e-01 -1.28403378e+00 5.88813961e-01 -1.39216483e+00 -2.50561595e-01 6.75066173e-01 1.67291701e-01 -3.23007792e-01 -5.22858314e-02 -9.02458787e-01 3.57676029e-01 2.94108033e-01 -1.59323111e-01 -1.84845555e+00 -6.17276728e-01 -1.02022409e+00 8.29405636e-02 3.15511227e-01 -1.03032768e+00 1.27373707e+00 -1.51727509e+00 -1.58022439e+00 4.11277235e-01 -2.24897161e-01 -1.05909419e+00 6.68302774e-01 -3.07419688e-01 -3.70769709e-01 5.67936599e-01 -1.68461159e-01 1.11748815e+00 1.15401375e+00 -1.09991372e+00 -4.54777658e-01 1.70926973e-01 5.75703502e-01 1.10725693e-01 -3.47360969e-02 -1.16324715e-01 -3.58569533e-01 -9.71589923e-01 -3.93140882e-01 -9.29487944e-01 -4.87944096e-01 1.00468859e-01 -2.06013367e-01 2.83142954e-01 1.14187038e+00 -7.35067725e-01 8.05704534e-01 -1.84313846e+00 1.30145565e-01 -2.66042352e-01 2.11411402e-01 6.37011230e-02 -3.30891550e-01 4.38548028e-01 -1.60851970e-01 2.69729823e-01 2.82776773e-01 5.45574948e-02 -4.18897629e-01 1.94672510e-01 -8.95511150e-01 2.15117335e-01 6.01530492e-01 1.36403739e+00 -1.02177608e+00 -1.40598357e-01 4.50397491e-01 6.03689194e-01 -7.94926226e-01 5.23931623e-01 -7.93786764e-01 8.91466677e-01 -5.13956487e-01 3.96652848e-01 1.51299208e-01 -4.59498882e-01 2.62364328e-01 3.04897651e-02 2.48283342e-01 1.85973883e-01 -7.99937189e-01 1.45500517e+00 -3.00950319e-01 1.05108738e+00 -3.10134202e-01 -8.38506043e-01 4.48378831e-01 1.95311889e-01 5.29822528e-01 -4.69698280e-01 -1.78926900e-01 -6.86180353e-01 -1.24348447e-01 -5.73582113e-01 5.06909907e-01 8.13710988e-02 1.15736127e-01 5.53394139e-01 6.82318658e-02 -7.99134299e-02 7.53484592e-02 6.98884726e-01 1.40006065e+00 6.57423794e-01 -1.37688994e-01 1.09273419e-01 -1.19284533e-01 1.09893687e-01 4.55402911e-01 7.45532870e-01 2.71307435e-02 7.65842795e-01 6.33254051e-01 -8.11057925e-01 -1.30772996e+00 -1.13214743e+00 5.45666516e-01 9.14456308e-01 2.53301859e-01 -3.79778117e-01 -8.26734424e-01 -6.26299322e-01 -1.41657338e-01 7.83651114e-01 -7.78059483e-01 -3.71868134e-01 -8.58556330e-01 -3.31909388e-01 5.38604081e-01 1.01279199e+00 2.64871508e-01 -1.19158876e+00 -6.97294116e-01 1.60133809e-01 -1.71262994e-01 -1.31773424e+00 -3.91307056e-01 -3.74398559e-01 -7.28935003e-01 -1.13609397e+00 -6.31125987e-01 -5.27691007e-01 5.79600036e-01 5.01177251e-01 1.32717061e+00 -9.23372582e-02 -2.35922053e-01 9.47324395e-01 -3.05615574e-01 -9.05753747e-02 -9.01388526e-01 -5.34841001e-01 8.31942782e-02 -5.23440056e-02 -1.69604775e-02 -7.03669250e-01 -8.39326322e-01 1.65489405e-01 -9.93873417e-01 5.40225565e-01 1.97931260e-01 6.77923501e-01 3.09258789e-01 -1.16651930e-01 2.49511912e-01 -8.31473827e-01 6.06573895e-02 -7.65342057e-01 -5.55928409e-01 1.40432179e-01 2.67571826e-02 -4.63820435e-03 8.66912067e-01 -7.13352740e-01 -1.26557207e+00 -6.83655124e-03 3.12302947e-01 -1.05379164e+00 -3.66439521e-01 -1.40169665e-01 1.26627870e-02 2.14952767e-01 7.13685453e-01 4.50829238e-01 -1.39267475e-03 1.38161927e-01 6.54361069e-01 -1.32999634e-02 7.33308852e-01 -5.47536910e-01 9.42966580e-01 7.11937428e-01 -6.61978573e-02 -8.04197848e-01 -4.30431277e-01 2.42231414e-01 -3.65208775e-01 -4.61108476e-01 9.94433582e-01 -1.45204043e+00 -7.45802343e-01 4.09283698e-01 -1.16744018e+00 -9.46473539e-01 -4.15671855e-01 3.49114716e-01 -8.59529734e-01 1.55150205e-01 -9.61842477e-01 -6.97193563e-01 2.24224135e-01 -1.00766563e+00 1.20130193e+00 3.32164109e-01 -2.35343277e-01 -1.04448700e+00 -1.98626220e-01 2.96174616e-01 1.87374592e-01 5.96966982e-01 4.64324623e-01 -1.39474347e-01 -1.58454621e+00 2.11248741e-01 5.89658357e-02 2.88727403e-01 -3.45577411e-02 4.72765028e-01 -9.37250376e-01 -2.38729179e-01 -3.06047678e-01 -4.58318174e-01 9.89022851e-01 6.17837965e-01 1.51635921e+00 -7.59441257e-01 -3.60270023e-01 9.07438636e-01 9.67748582e-01 4.62253660e-01 1.09787726e+00 -6.18577749e-02 8.67204726e-01 1.83612108e-01 4.43691105e-01 5.63400924e-01 2.48362586e-01 4.04849023e-01 4.98445421e-01 -1.43604860e-01 -3.80955815e-01 -7.12692380e-01 8.52280080e-01 2.62589127e-01 -3.32511783e-01 -7.06672966e-01 -7.99084127e-01 3.27576160e-01 -1.87493682e+00 -1.74778163e+00 1.53484464e-01 1.82428503e+00 3.26253355e-01 1.57717437e-01 3.00681055e-01 -4.37657148e-01 6.55994713e-01 4.51770872e-01 -7.93691993e-01 -1.35000572e-01 -1.46164358e-01 2.16505542e-01 4.78859186e-01 3.13832939e-01 -1.32710874e+00 1.23451996e+00 7.34623766e+00 4.34517235e-01 -1.22840989e+00 -3.22611272e-01 1.16483736e+00 -3.12549919e-01 -4.46150899e-01 4.62474301e-02 -4.41389024e-01 6.45564556e-01 1.25035334e+00 -3.82215291e-01 6.00944519e-01 9.28086698e-01 5.27797818e-01 5.72322570e-02 -1.25273871e+00 7.99964726e-01 1.06852114e-01 -2.00196576e+00 6.40207171e-01 8.26499537e-02 1.06965506e+00 -1.86667349e-02 1.08903885e-01 3.78657043e-01 9.63570416e-01 -1.40087032e+00 8.97352934e-01 6.99496567e-01 8.62726867e-01 -4.07400280e-01 1.40637428e-01 2.36452937e-01 -1.07754612e+00 -1.22725047e-01 -2.13462979e-01 -2.75072098e-01 4.54457700e-01 -5.37065901e-02 -8.11021984e-01 5.06510250e-02 5.04802048e-01 1.13812554e+00 -5.51585853e-01 5.64521968e-01 -2.08463058e-01 9.93684649e-01 -1.28507167e-01 2.62426138e-01 3.19098413e-01 -8.57845247e-02 4.81156617e-01 1.20360267e+00 4.49013978e-01 5.09560406e-01 3.17948222e-01 1.03129768e+00 -1.48785919e-01 -6.32037938e-01 -1.10921299e+00 -4.18664545e-01 2.40904838e-01 7.42760658e-01 -7.57569611e-01 -6.21941864e-01 -2.90818840e-01 1.20092893e+00 -2.72124112e-02 7.13067055e-01 -1.35415244e+00 3.53347570e-01 9.15698707e-01 2.88904399e-01 5.02966583e-01 -3.51628929e-01 2.12795749e-01 -1.56803632e+00 -2.42747039e-01 -1.00184715e+00 4.57671992e-02 -1.19551194e+00 -6.90976560e-01 4.95964885e-01 9.19872988e-03 -1.38885105e+00 -1.02985561e+00 -5.99039912e-01 -8.86414945e-01 2.95550764e-01 -9.04658794e-01 -1.31473339e+00 -3.42707008e-01 8.12383413e-01 9.52980518e-01 -1.33132204e-01 5.59198499e-01 -1.63763732e-01 -4.66879725e-01 1.63304701e-01 -1.45965651e-01 3.96855384e-01 3.40343833e-01 -1.09366679e+00 1.03009677e+00 1.23030698e+00 5.22958755e-01 3.60042810e-01 7.42030323e-01 -7.03209221e-01 -1.76806283e+00 -1.42880404e+00 -3.96562666e-02 -8.35213661e-01 7.15914249e-01 -2.87796706e-01 -6.28935397e-01 1.14696479e+00 2.32175291e-01 1.89298660e-01 3.70539606e-01 -4.47526485e-01 -3.95647079e-01 1.61190018e-01 -6.40227616e-01 9.07875955e-01 1.33187258e+00 -4.98327196e-01 -1.69981197e-01 4.46503371e-01 9.48826253e-01 -5.42605162e-01 -6.28167152e-01 2.10296333e-01 5.90816617e-01 -1.11323059e+00 1.23446810e+00 -1.13955986e+00 1.03745413e+00 -9.94052738e-02 3.79373766e-02 -1.06627572e+00 -3.96781445e-01 -9.21774030e-01 -6.36546135e-01 8.61530781e-01 -4.14302945e-02 -2.74352301e-02 9.81243014e-01 9.46681857e-01 -2.81517357e-02 -3.45770031e-01 -4.22803640e-01 -7.13927805e-01 -3.58310789e-02 -4.52873021e-01 3.67528707e-01 7.16101348e-01 -4.13690329e-01 2.13433981e-01 -7.00439155e-01 1.99516892e-01 3.02395433e-01 1.51444316e-01 1.24353445e+00 -4.81253207e-01 -6.03952050e-01 -2.16747314e-01 -6.84212148e-01 -1.37814426e+00 3.63871545e-01 -5.11306763e-01 -1.98008001e-01 -1.23141837e+00 2.48151898e-01 1.47556448e-02 -4.39707674e-02 3.33107889e-01 -2.08092049e-01 2.61962712e-01 4.43279028e-01 -2.77939234e-02 -7.00636864e-01 5.81890523e-01 1.35127640e+00 -3.06421727e-01 1.79734454e-02 -1.13494240e-01 -5.49945176e-01 8.04811478e-01 5.80010355e-01 -7.83675537e-02 -7.39679158e-01 -5.89310110e-01 -3.44193816e-01 5.46290874e-01 8.39832425e-01 -1.19793129e+00 -1.31958574e-01 -7.60712862e-01 8.00220311e-01 -5.72011471e-02 5.93766212e-01 -4.47939843e-01 4.10739899e-01 3.11307847e-01 -3.76484871e-01 -6.94049597e-02 3.00882339e-01 8.76586437e-01 -8.18087533e-02 4.71335441e-01 5.16639829e-01 -4.57989633e-01 -1.14839303e+00 5.71875036e-01 -6.89218402e-01 1.74151316e-01 1.31527460e+00 -2.93325722e-01 -5.55350125e-01 -1.11838222e+00 -8.17344069e-01 9.42799151e-02 8.10777664e-01 6.72735631e-01 7.98116565e-01 -1.22846079e+00 -6.15490556e-01 1.07843839e-01 -3.05138499e-01 -2.00835794e-01 4.19485837e-01 1.06887229e-01 -8.33823562e-01 2.46963888e-01 -3.87778938e-01 -6.83152676e-01 -1.10194945e+00 7.76376188e-01 1.86050653e-01 7.34406384e-03 -8.08577299e-01 7.86800385e-01 8.12330782e-01 3.45322281e-01 5.37134446e-02 -3.93704087e-01 1.04185961e-01 -5.90399802e-01 7.74036884e-01 2.57705033e-01 -7.49708295e-01 -5.97050488e-01 9.62776132e-03 4.74986397e-02 1.87826842e-01 4.13350984e-02 1.34747398e+00 5.60361966e-02 3.54404390e-01 3.33668470e-01 8.82024109e-01 -2.90366292e-01 -2.21953750e+00 3.59475404e-01 -5.46686113e-01 -5.05009174e-01 -2.30255574e-01 -5.63790858e-01 -1.13922822e+00 1.00809383e+00 1.69027105e-01 1.91608742e-02 1.07618296e+00 -1.95231177e-02 7.99508393e-01 5.25503814e-01 6.32171214e-01 -5.02844870e-01 4.62103218e-01 3.47392440e-01 8.43473792e-01 -1.15420258e+00 -1.70087785e-01 -2.35322610e-01 -9.98494565e-01 1.27270913e+00 9.37553406e-01 -4.54614967e-01 3.74884635e-01 6.45642340e-01 -6.17349073e-02 8.41852054e-02 -1.26689935e+00 -4.71058302e-02 2.04871371e-01 7.55210102e-01 2.29775354e-01 2.47104988e-02 5.63213229e-01 1.80515438e-01 -2.86832266e-03 1.23545572e-01 9.35420513e-01 7.06549168e-01 -2.50786304e-01 -7.82768428e-01 -1.25524744e-01 3.48356247e-01 -3.38756800e-01 -2.42492687e-02 -2.17176542e-01 7.40465045e-01 -1.08733103e-01 7.47909725e-01 3.65697265e-01 -5.25558352e-01 -2.91840613e-01 -6.92077503e-02 7.04908431e-01 -5.80063224e-01 -3.13446522e-01 4.54984494e-02 5.10325320e-02 -1.15209162e+00 -4.65276867e-01 -6.50969267e-01 -1.06727910e+00 -4.79394674e-01 2.76358157e-01 -2.84284651e-01 8.53987560e-02 5.25998950e-01 6.40730321e-01 6.21411145e-01 6.31168902e-01 -1.17923546e+00 -3.08713876e-02 -6.53368950e-01 -2.38188237e-01 7.60897517e-01 3.02889705e-01 -3.76156479e-01 -1.43062755e-01 1.00879395e+00]
[10.730000495910645, -0.6026548743247986]
2f3f2763-4432-46f7-a671-a3b4eca50c88
multimodal-manoeuvre-and-trajectory
2303.16109
null
https://arxiv.org/abs/2303.16109v1
https://arxiv.org/pdf/2303.16109v1.pdf
Multimodal Manoeuvre and Trajectory Prediction for Autonomous Vehicles Using Transformer Networks
Predicting the behaviour (i.e. manoeuvre/trajectory) of other road users, including vehicles, is critical for the safe and efficient operation of autonomous vehicles (AVs), a.k.a. automated driving systems (ADSs). Due to the uncertain future behaviour of vehicles, multiple future behaviour modes are often plausible for a vehicle in a given driving scene. Therefore, multimodal prediction can provide richer information than single-mode prediction enabling AVs to perform a better risk assessment. To this end, we propose a novel multimodal prediction framework that can predict multiple plausible behaviour modes and their likelihoods. The proposed framework includes a bespoke problem formulation for manoeuvre prediction, a novel transformer-based prediction model, and a tailored training method for multimodal manoeuvre and trajectory prediction. The performance of the framework is evaluated using two public benchmark highway driving datasets, namely NGSIM and highD. The results show that the proposed framework outperforms the state-of-the-art multimodal methods in the literature in terms of prediction error and is capable of predicting plausible manoeuvre and trajectory modes.
['Mehrdad Dianati', 'Konstantinos Koufos', 'Sajjad Mozaffari']
2023-03-28
null
null
null
null
['trajectory-prediction']
['computer-vision']
[-6.93619847e-02 1.83617603e-02 -3.64919484e-01 -7.01457262e-01 -9.28779185e-01 -1.76924959e-01 1.11164105e+00 1.20714724e-01 -1.80470139e-01 5.25437593e-01 7.48740584e-02 -8.29562545e-01 -2.46688709e-01 -8.75088215e-01 -7.40044713e-01 -7.88302779e-01 -4.30449955e-02 3.82397979e-01 6.02826416e-01 -4.78533894e-01 2.94391751e-01 5.83406031e-01 -2.17298913e+00 1.28986254e-01 8.63262653e-01 1.14826834e+00 3.04293156e-01 6.40669525e-01 1.50872633e-01 8.41468811e-01 1.75509900e-01 -6.28867686e-01 -1.21848933e-01 2.38469213e-01 -4.43865091e-01 -1.39566377e-01 1.89553350e-01 -3.09639454e-01 -7.18149543e-01 6.51121020e-01 4.50637564e-02 2.85183340e-01 1.09120584e+00 -1.82330346e+00 2.16065556e-01 1.26268595e-01 -3.11559349e-01 9.37138796e-02 1.58146828e-01 3.02890480e-01 6.26110733e-01 -8.33849669e-01 2.90251553e-01 1.29902303e+00 3.37662280e-01 4.96120751e-01 -9.11218822e-01 -6.79118156e-01 -9.86106470e-02 1.01334298e+00 -1.49127054e+00 -5.19212902e-01 7.18042731e-01 -6.52748764e-01 5.76379538e-01 2.68471569e-01 3.19406748e-01 1.00619757e+00 8.80020380e-01 1.05145466e+00 7.10290611e-01 3.24897885e-01 1.81568548e-01 4.71484929e-01 -2.23930068e-02 4.83238608e-01 -3.33902240e-01 4.42556947e-01 -3.65062773e-01 -9.78719220e-02 -4.21631604e-01 -4.06056717e-02 3.97346735e-01 -3.29518288e-01 -1.13447976e+00 7.68277407e-01 -1.45866470e-02 -5.95616937e-01 -5.43615222e-01 1.81005672e-01 4.82803464e-01 4.78571914e-02 1.12940438e-01 -4.89827126e-01 -1.87809438e-01 -4.74842191e-01 -6.66435242e-01 7.68352747e-01 4.04689729e-01 1.01286709e+00 9.26578641e-01 -1.44731954e-01 -3.54518145e-01 5.88048160e-01 7.00219512e-01 8.22769701e-01 -1.80640504e-01 -7.58341432e-01 7.40035415e-01 5.43765128e-01 2.52980262e-01 -1.13249397e+00 -3.85669947e-01 4.41697128e-02 -6.85015380e-01 1.44578740e-01 1.59097731e-01 2.76198909e-02 -6.87272906e-01 1.40037739e+00 4.38831210e-01 6.51317775e-01 3.47903222e-01 6.27970755e-01 5.99948704e-01 1.21398902e+00 5.28837025e-01 4.73212264e-02 1.14378226e+00 -7.13442147e-01 -6.88012242e-01 -2.64902502e-01 7.88704872e-01 -6.50159776e-01 4.95981187e-01 3.12969089e-01 -7.47629583e-01 -6.39158547e-01 -9.56320643e-01 3.01990062e-01 -4.62437719e-01 5.91560490e-02 2.64830709e-01 5.00919282e-01 -5.48796713e-01 2.98600376e-01 -6.18064642e-01 -1.63079709e-01 3.52119118e-01 1.69330001e-01 -3.90681118e-01 -3.30686212e-01 -1.51623464e+00 1.32922101e+00 5.82296669e-01 2.06064612e-01 -1.28588498e+00 -7.62763202e-01 -9.05217409e-01 -2.49866784e-01 3.42736065e-01 -2.40883335e-01 1.07132256e+00 -2.01951399e-01 -1.01070535e+00 3.81988704e-01 -4.75245655e-01 -5.84641278e-01 6.29574955e-01 1.36545852e-01 -9.32694137e-01 -1.45095140e-01 -1.40970424e-01 7.41305053e-01 8.07831764e-01 -1.42708051e+00 -1.28216267e+00 -1.44349515e-01 -2.73690790e-01 3.02208215e-01 5.37310317e-02 -1.21316627e-01 -3.19919258e-01 7.02183321e-02 -3.55248779e-01 -1.02460742e+00 -2.65861541e-01 -3.34996462e-01 -4.35574889e-01 -6.81796432e-01 1.26402724e+00 -6.92109406e-01 1.36400592e+00 -2.16565084e+00 2.15632897e-02 4.61339086e-01 5.00809774e-02 2.73328036e-01 -1.55298665e-01 6.25582099e-01 3.92140180e-01 -2.82652408e-01 -4.48535085e-02 -4.45679158e-01 2.35438347e-01 4.61906374e-01 -2.88407773e-01 4.95746732e-01 1.94096044e-01 6.70204401e-01 -8.00093353e-01 -5.11811793e-01 8.16853285e-01 5.08090615e-01 -3.66830663e-03 4.46329713e-01 -1.70747399e-01 4.73845273e-01 -5.20047247e-01 5.10651886e-01 9.42180276e-01 6.08086407e-01 -2.45726198e-01 -1.47868142e-01 -4.43324238e-01 3.15995663e-02 -9.01980042e-01 9.04549122e-01 -5.91965377e-01 7.94577360e-01 -5.02719045e-01 -6.94895566e-01 1.14896429e+00 2.21686721e-01 3.24989110e-01 -7.90710986e-01 2.01173022e-01 3.43017906e-01 -8.25802088e-02 -7.32640028e-01 7.90999174e-01 1.64984971e-01 -1.76403075e-01 -3.78566049e-02 -3.99122864e-01 1.40130281e-01 1.28171116e-01 3.00123394e-01 7.89364576e-01 -5.51132783e-02 -1.19915135e-01 -1.38789520e-01 9.20602441e-01 -1.08781718e-01 5.00357032e-01 4.08009082e-01 -3.48818988e-01 -7.76874274e-02 5.72763383e-01 -4.94371861e-01 -1.18736672e+00 -1.09663916e+00 -4.00200516e-01 6.80394471e-01 6.73543155e-01 -3.71502310e-01 -4.80206162e-01 -5.46198308e-01 1.00738160e-01 1.39185369e+00 -4.43436325e-01 -5.36064267e-01 -2.72314519e-01 -3.60826880e-01 6.01022601e-01 4.29770917e-01 2.40755916e-01 -6.26487553e-01 -4.57369626e-01 2.45055750e-01 -3.19631666e-01 -1.31470811e+00 -1.11050107e-01 -4.28311408e-01 -3.55081201e-01 -8.79252791e-01 -1.71621561e-01 -4.71973479e-01 2.54657924e-01 3.48438501e-01 6.89060450e-01 -1.01270229e-01 3.01345617e-01 1.96199149e-01 -1.51582658e-01 -5.35070121e-01 -9.98448968e-01 -4.24296148e-02 1.14362314e-01 5.29220998e-01 6.35867715e-01 -8.70958790e-02 -7.02134252e-01 8.89261365e-01 -5.78984022e-01 4.31535095e-01 6.44700050e-01 5.83755136e-01 6.84957922e-01 4.40528393e-01 8.27101350e-01 -4.13903445e-01 2.59303421e-01 -1.37645519e+00 -6.66202366e-01 2.29306728e-01 -7.19837070e-01 -1.44697785e-01 6.40052676e-01 -2.75651902e-01 -1.36601877e+00 2.47023940e-01 -3.68183076e-01 -1.31185114e-01 -5.90846300e-01 4.84727502e-01 -3.42287332e-01 5.73713146e-02 2.33412907e-01 5.38189709e-01 1.18091829e-01 -1.89940855e-01 2.56894141e-01 9.27644253e-01 5.11836946e-01 -1.69620916e-01 9.44960356e-01 2.59506851e-01 4.23388869e-01 -9.62420702e-01 -2.68623799e-01 -7.15428412e-01 -4.04708326e-01 -1.08372939e+00 7.43905723e-01 -1.03615046e+00 -9.57166553e-01 6.48352921e-01 -1.11074042e+00 8.03638995e-02 5.47822118e-01 5.03399432e-01 -6.49550080e-01 2.32382819e-01 -2.08363198e-02 -1.24460018e+00 2.53547490e-01 -1.57317448e+00 1.12133026e+00 1.51099607e-01 3.74107100e-02 -9.60636854e-01 -2.09584951e-01 6.93797052e-01 3.97828043e-01 2.66804665e-01 8.66969883e-01 -4.24674600e-01 -7.53659904e-01 -5.44076920e-01 -3.31732839e-01 2.13444471e-01 -3.78145814e-01 8.94910917e-02 -8.19967330e-01 2.00190190e-02 -6.61255896e-01 5.05841821e-02 7.62216330e-01 3.05229574e-02 1.05575383e+00 -3.86830956e-01 -5.79688251e-01 1.12136610e-01 1.12695467e+00 4.21794236e-01 1.07099903e+00 2.04884544e-01 7.50801980e-01 1.05656779e+00 1.34680629e+00 5.47788203e-01 1.11813271e+00 9.35962558e-01 8.75537157e-01 3.06571364e-01 2.31049314e-01 -5.38705349e-01 6.24697924e-01 4.29619223e-01 1.38888955e-01 -4.03953671e-01 -1.16465199e+00 7.88098633e-01 -2.17553616e+00 -1.09633589e+00 -7.51048684e-01 2.04065943e+00 1.01662114e-01 2.24336684e-01 3.21773201e-01 2.99401104e-01 5.80499291e-01 7.06258044e-02 -5.77261388e-01 -6.50341451e-01 2.05409497e-01 -7.67336249e-01 8.53104413e-01 4.24162894e-01 -1.06242061e+00 6.80876076e-01 5.14086342e+00 1.35923398e+00 -7.32726395e-01 -6.78159893e-02 6.38869762e-01 3.90641689e-01 -6.03505671e-01 -5.61734959e-02 -1.20689893e+00 7.63401449e-01 1.53252661e+00 -2.22314283e-01 2.02883169e-01 7.78494537e-01 7.57500768e-01 -2.67130405e-01 -9.82912481e-01 7.73537636e-01 -7.71430507e-02 -1.36397207e+00 6.24132529e-02 7.49902204e-02 4.69806880e-01 1.45798057e-01 2.15663627e-01 3.05523396e-01 -3.17671120e-01 -9.45579529e-01 9.31762815e-01 1.00569046e+00 3.64516884e-01 -1.27259302e+00 1.08500326e+00 7.57936418e-01 -1.22928548e+00 -3.30971599e-01 -8.32212493e-02 3.72838140e-01 7.22672045e-01 1.91667527e-01 -8.81013215e-01 7.40954459e-01 4.19511437e-01 9.62620258e-01 -6.16272926e-01 1.08490205e+00 1.56912953e-01 6.90326452e-01 -1.27235711e-01 -2.66720742e-01 5.18741369e-01 -1.13153547e-01 7.92512655e-01 1.11797583e+00 5.81709206e-01 -1.45841196e-01 -1.13659002e-01 4.17717844e-01 3.98808300e-01 8.80794525e-02 -8.33001792e-01 2.89331317e-01 6.91765189e-01 1.09373116e+00 1.22137323e-01 -5.09449355e-02 -5.52452266e-01 1.63915530e-01 -2.25365385e-01 2.15646207e-01 -1.19018602e+00 -2.05639392e-01 8.57972264e-01 3.99786562e-01 6.44278377e-02 -1.56324953e-01 -2.14982361e-01 -5.05324304e-01 -9.64499265e-02 -3.21121812e-01 1.19842008e-01 -6.12114310e-01 -1.04284787e+00 5.54081857e-01 3.77232730e-01 -1.65941906e+00 -3.57299268e-01 -4.43968982e-01 -7.88183630e-01 7.23529458e-01 -1.87644553e+00 -1.54416144e+00 -1.67303890e-01 4.95018870e-01 5.64790964e-01 -6.34191453e-01 1.68954715e-01 6.10991359e-01 -5.73732972e-01 5.23100853e-01 1.67102516e-01 -6.11012101e-01 2.25470841e-01 -7.93583870e-01 3.20271313e-01 7.65566647e-01 -6.58078790e-01 -9.94452313e-02 1.08858836e+00 -5.32158315e-01 -1.78354418e+00 -1.70493329e+00 1.15362847e+00 -3.79002333e-01 7.14681029e-01 -9.27051976e-02 -7.94722438e-01 3.20678204e-01 -7.94585943e-02 -3.87514859e-01 4.76403296e-01 -2.62887299e-01 1.44018963e-01 -5.14404774e-01 -9.75038350e-01 7.07736492e-01 5.11383057e-01 -3.81164163e-01 -2.08233252e-01 1.16997221e-02 2.28249282e-01 -9.89744440e-02 -8.25773895e-01 6.01840913e-01 8.29837024e-01 -7.98023045e-01 8.60142350e-01 -3.91361982e-01 4.63380575e-01 -5.10731816e-01 -4.01505798e-01 -1.11362934e+00 -7.05876872e-02 -2.75651306e-01 -4.10416037e-01 1.14237094e+00 6.08460546e-01 -5.34026682e-01 5.42472363e-01 9.40377474e-01 -4.92057085e-01 -9.39557731e-01 -1.31918442e+00 -6.72279119e-01 3.83429346e-04 -1.23762989e+00 8.67086589e-01 8.01637918e-02 -1.08115293e-01 4.56359461e-02 -8.82476270e-01 6.16643012e-01 9.09602404e-01 -4.85536456e-01 1.04880559e+00 -9.73651290e-01 2.92599767e-01 -3.40141267e-01 -1.04194951e+00 -1.03846228e+00 4.66054946e-01 -7.07780123e-01 2.40577072e-01 -1.39769840e+00 4.99963909e-02 -4.54422385e-01 -1.52667344e-01 1.53434619e-01 -7.02397674e-02 8.51218104e-02 -5.86044863e-02 -8.08544084e-02 -6.58547580e-01 8.23683679e-01 9.05417919e-01 -3.17678750e-01 6.79294244e-05 5.19604683e-01 -1.30034700e-01 5.54048300e-01 9.79185283e-01 -3.86057079e-01 -6.26416504e-01 6.03057444e-02 1.97333723e-01 5.11197448e-01 8.22018564e-01 -8.19415271e-01 3.41204792e-01 -5.55556059e-01 -3.03886682e-01 -1.39090180e+00 4.66682434e-01 -1.00888133e+00 4.51816440e-01 2.94417232e-01 -3.12984943e-01 -1.18834630e-01 2.93617368e-01 9.38342273e-01 -2.09529087e-01 6.17679395e-02 6.57655239e-01 8.26204181e-01 -1.32314980e+00 6.72411561e-01 -1.11656082e+00 -5.43347239e-01 1.66589773e+00 -3.93968135e-01 -2.56987065e-01 -3.90327930e-01 -3.22300613e-01 8.68620515e-01 7.01773316e-02 1.04463172e+00 1.05286121e+00 -1.63032746e+00 -8.61478567e-01 1.22639358e-01 7.93894589e-01 -4.11190689e-01 8.78836632e-01 9.34753060e-01 -2.33187854e-01 3.43036145e-01 -1.07792601e-01 -6.67931259e-01 -1.25211775e+00 3.51118058e-01 1.74120814e-01 2.52831042e-01 -3.81684721e-01 2.55084392e-02 5.50898910e-02 -4.85787839e-01 1.12211481e-01 2.77337849e-01 -6.92791879e-01 -1.69648811e-01 5.91547132e-01 9.15409446e-01 -7.08205774e-02 -1.57959139e+00 -4.11724627e-01 2.22647011e-01 3.95114571e-02 1.58923313e-01 8.69190276e-01 -5.89482665e-01 3.65021497e-01 5.99684477e-01 1.23331749e+00 -5.11255085e-01 -1.36664975e+00 4.91468273e-02 -9.98433456e-02 -5.71751714e-01 3.12180221e-01 -6.84003592e-01 -7.86088288e-01 1.03053677e+00 6.77150548e-01 8.69203210e-02 7.78965473e-01 3.07046901e-02 1.28429079e+00 1.37642682e-01 3.97685140e-01 -1.15328872e+00 -5.21620393e-01 4.25192922e-01 7.09588587e-01 -1.52656901e+00 -4.94822055e-01 -4.16090041e-01 -1.08536339e+00 9.69167233e-01 5.48793137e-01 3.37253034e-01 8.91376317e-01 -2.38589063e-01 -9.31783989e-02 4.71985601e-02 -1.13828576e+00 -7.31401592e-02 8.13101590e-01 6.08125269e-01 -2.07295835e-01 4.67501491e-01 -7.37015978e-02 4.73801970e-01 4.15577330e-02 -1.55426398e-01 3.87595922e-01 3.60686243e-01 -5.04146576e-01 -9.07474041e-01 -2.43133798e-01 7.06923246e-01 1.66826218e-01 4.09899473e-01 3.62239569e-01 4.36497509e-01 3.05430830e-01 1.43710434e+00 -6.85693771e-02 -1.03086531e+00 5.13174355e-01 -1.70682639e-01 -1.48791492e-01 1.90777406e-01 1.17896199e-01 -4.99564856e-01 4.79191869e-01 -6.23558462e-01 -2.44272292e-01 -9.29493904e-01 -1.14279413e+00 -9.54295456e-01 2.71825995e-02 2.48757545e-02 8.46729398e-01 1.26957810e+00 3.86559516e-01 3.05977255e-01 1.10511005e+00 -8.16276312e-01 -3.95055324e-01 -4.54338878e-01 -5.39130509e-01 3.87162596e-01 3.64807039e-01 -1.00662494e+00 -1.77086800e-01 -2.11432502e-01]
[5.826858043670654, 1.1155730485916138]
f5f897a4-1d6f-40f0-9de8-f216dba91a49
question-type-identification-for-academic
2211.13727
null
https://arxiv.org/abs/2211.13727v1
https://arxiv.org/pdf/2211.13727v1.pdf
Question-type Identification for Academic Questions in Online Learning Platform
Online learning platforms provide learning materials and answers to students' academic questions by experts, peers, or systems. This paper explores question-type identification as a step in content understanding for an online learning platform. The aim of the question-type identifier is to categorize question types based on their structure and complexity, using the question text, subject, and structural features. We have defined twelve question-type classes, including Multiple-Choice Question (MCQ), essay, and others. We have compiled an internal dataset of students' questions and used a combination of weak-supervision techniques and manual annotation. We then trained a BERT-based ensemble model on this dataset and evaluated this model on a separate human-labeled test set. Our experiments yielded an F1-score of 0.94 for MCQ binary classification and promising results for 12-class multilabel classification. We deployed the model in our online learning platform as a crucial enabler for content understanding to enhance the student learning experience.
['Saurabh Khanwalkar', "Johnson D'Souza", 'Alok Goel', 'Azam Rabiee']
2022-11-24
null
null
null
null
['type']
['speech']
[-1.42720891e-02 4.55845386e-01 -3.92328113e-01 -4.74431515e-01 -1.10236418e+00 -1.07448578e+00 4.54054028e-01 9.13980186e-01 -4.07626063e-01 4.99232054e-01 -9.41420533e-03 -1.02172792e+00 -4.36419159e-01 -7.77952552e-01 -3.94973785e-01 3.53371985e-02 4.35699880e-01 1.88298464e-01 6.24720812e-01 -4.32183474e-01 6.16319537e-01 3.21431637e-01 -1.84025097e+00 5.72189808e-01 1.29267395e+00 1.19492197e+00 2.22694114e-01 7.60145307e-01 -7.28058040e-01 1.38433015e+00 -7.30541646e-01 -6.20308995e-01 -2.20189020e-01 -6.00783825e-01 -1.41377938e+00 -7.54765272e-02 1.03409016e+00 1.34134769e-01 4.30611409e-02 8.57838035e-01 1.38724014e-01 2.87455827e-01 5.38369894e-01 -1.04940259e+00 -6.30744338e-01 5.72135746e-01 1.36094600e-01 2.35196546e-01 9.81702030e-01 -4.34273422e-01 1.36767888e+00 -7.74261236e-01 8.07781637e-01 7.12180734e-01 5.51178694e-01 4.63176459e-01 -8.28142643e-01 -4.50476944e-01 -7.64134526e-02 6.36163294e-01 -8.04150820e-01 -1.46686763e-01 3.90019804e-01 -7.85542786e-01 2.81463057e-01 4.55550522e-01 4.04224515e-01 5.17231762e-01 3.26575607e-01 8.62017214e-01 1.61487126e+00 -8.65005970e-01 3.39192182e-01 7.06567466e-01 8.91328514e-01 9.19704318e-01 -1.93555176e-01 -5.74661791e-01 -4.01857555e-01 -1.48004983e-02 -8.33059475e-02 -7.85660669e-02 -3.77177924e-01 7.00088143e-02 -7.35879004e-01 8.77689540e-01 2.04468429e-01 3.59315574e-01 2.60275602e-01 -6.70359612e-01 -5.03546745e-02 8.34801674e-01 1.53570816e-01 9.03573692e-01 -6.45731926e-01 -1.00391559e-01 -7.88938105e-01 8.68349075e-02 1.37137699e+00 8.84632170e-01 6.54809594e-01 -5.97998977e-01 -3.20468336e-01 9.14237440e-01 4.50557202e-01 8.52715448e-02 6.60602689e-01 -9.34968948e-01 3.77628505e-01 1.24126232e+00 -4.25057888e-01 -6.44207060e-01 -4.10125256e-01 -5.22140205e-01 -1.03438087e-01 1.24102950e-01 7.65288055e-01 -1.06715135e-01 -6.45499825e-01 1.19296801e+00 4.87953484e-01 -3.04418206e-01 1.27358604e-02 4.83975500e-01 1.70746803e+00 2.28082910e-01 1.76310107e-01 1.84582859e-01 1.77072549e+00 -1.11272371e+00 -8.92743945e-01 4.47274327e-01 1.19772923e+00 -1.00410903e+00 8.71721148e-01 3.86133283e-01 -8.64136517e-01 -6.56451881e-01 -7.51964927e-01 -1.72072917e-01 -7.26661205e-01 1.31812662e-01 6.78896680e-02 9.62029517e-01 -8.83557737e-01 3.24366868e-01 6.50278013e-03 -1.84551477e-01 2.07823738e-01 4.59246151e-02 -3.40773731e-01 -1.89835932e-02 -1.12332451e+00 7.16072261e-01 -1.19412281e-01 -6.81674063e-01 -6.15650654e-01 -8.73071373e-01 -7.53350317e-01 2.85585672e-01 1.91876203e-01 -2.29229242e-01 1.45368016e+00 -6.44530237e-01 -1.62848330e+00 1.28515494e+00 1.72715694e-01 -3.38557698e-02 2.42987216e-01 8.36890414e-02 -3.84563148e-01 4.91195589e-01 1.91884711e-01 2.68269628e-01 3.31241190e-01 -9.95780408e-01 -9.38018501e-01 -3.18811476e-01 3.23954642e-01 1.46153420e-01 -4.83380556e-01 1.54151589e-01 1.11470101e-02 -3.44275147e-01 3.39278787e-01 -7.31402755e-01 2.11980581e-01 -2.11824089e-01 1.14522483e-02 -9.21838939e-01 7.18570113e-01 -9.36125576e-01 1.33565450e+00 -1.57806087e+00 -4.61684883e-01 2.82939404e-01 4.18437541e-01 2.51542807e-01 -1.63756758e-02 3.54943722e-01 -1.63622186e-01 2.67561197e-01 2.78504640e-01 4.17286344e-02 1.38897717e-01 -2.41464168e-01 -2.63105750e-01 -3.27653512e-02 -1.03056885e-01 6.54674947e-01 -1.01874745e+00 -7.74271250e-01 -2.86961589e-02 -2.82305777e-01 -5.21465003e-01 8.74228716e-01 -3.24158788e-01 8.85865018e-02 -4.26013052e-01 8.20595145e-01 1.74896836e-01 -2.34226540e-01 5.57200387e-02 2.20601112e-01 -1.16929449e-01 5.84910333e-01 -1.02551806e+00 1.40531409e+00 -6.42787874e-01 7.85355568e-01 2.35975400e-01 -7.78628707e-01 1.28427684e+00 4.40980703e-01 2.46211171e-01 -6.57371819e-01 2.38216907e-01 2.88377315e-01 -7.60472640e-02 -8.34519029e-01 5.86642861e-01 2.49376103e-01 -9.24143717e-02 6.21932685e-01 5.75657904e-01 -1.31779820e-01 3.16789061e-01 4.15643215e-01 1.22037601e+00 1.91440850e-01 3.85531597e-02 -5.49995184e-01 8.85352790e-01 2.36059949e-02 -2.92296465e-02 7.89430022e-01 -5.03711820e-01 4.63816792e-01 4.94863093e-01 -8.29173401e-02 -2.53345907e-01 -7.51496315e-01 -4.40199047e-01 1.85313737e+00 -1.82678342e-01 -3.17880839e-01 -8.65309298e-01 -1.16823459e+00 -6.50475547e-02 6.62871361e-01 -4.07031506e-01 7.47061223e-02 -2.20392674e-01 -7.86691159e-02 2.41668686e-01 2.69040093e-02 2.81771213e-01 -8.01426411e-01 -2.22033337e-01 1.83686748e-01 -2.09012434e-01 -1.17204094e+00 -3.24801832e-01 4.00155783e-01 -4.31852758e-01 -1.38489962e+00 -3.53725553e-01 -1.18070102e+00 6.08595848e-01 2.46300578e-01 1.34903419e+00 4.32966918e-01 -7.46109486e-02 1.00121176e+00 -7.24261045e-01 -4.40347612e-01 -4.54208106e-01 4.39089507e-01 -2.80184686e-01 -9.98101160e-02 4.05310422e-01 -8.12788010e-02 -5.08641303e-01 4.98477131e-01 -7.71107018e-01 -4.84963536e-01 1.83623925e-01 5.90134978e-01 2.97077268e-01 -4.17691618e-01 1.00272667e+00 -9.54062641e-01 8.17577660e-01 -6.95258617e-01 -5.16212165e-01 6.47071004e-01 -7.55496323e-01 -1.09713830e-01 5.58706164e-01 -3.11744183e-01 -1.10920715e+00 -2.69490361e-01 -5.12591422e-01 2.82800999e-02 -3.91351342e-01 7.04198360e-01 -3.76296073e-01 -6.59981906e-01 7.23271370e-01 -9.69207063e-02 -5.77514768e-02 -6.64756536e-01 1.57379135e-01 1.12966061e+00 3.40329260e-01 -8.96898389e-01 4.14215624e-01 -3.40157062e-01 -1.17650352e-01 -7.02256501e-01 -1.52780247e+00 -8.92309427e-01 -6.50677443e-01 -7.25514889e-01 7.96293020e-01 -9.15050805e-01 -1.09760165e+00 1.15543000e-01 -7.59502709e-01 -3.11803758e-01 -1.79003477e-01 3.44993442e-01 9.11322385e-02 2.76454508e-01 -5.54306984e-01 -6.12810910e-01 -1.21888302e-01 -1.00624835e+00 3.86580497e-01 6.83097005e-01 -2.30711624e-01 -1.37447262e+00 2.36543909e-01 1.50905538e+00 2.30503008e-01 -1.09837770e-01 9.54067945e-01 -1.44026113e+00 -2.98289448e-01 -2.45207205e-01 1.10317878e-02 3.16544533e-01 -2.28179544e-01 -8.14847946e-02 -1.01036561e+00 9.88122597e-02 1.26351893e-01 -7.92110026e-01 6.92644596e-01 -4.36823517e-01 1.40770435e+00 -2.43809983e-01 8.50639716e-02 1.15607195e-01 1.20131290e+00 -2.05853999e-01 5.98303080e-02 7.13192880e-01 7.09481776e-01 1.11217749e+00 5.61175525e-01 -7.73404837e-02 9.67029989e-01 3.78365666e-01 4.38200682e-01 4.03891355e-01 -1.84594691e-01 -2.97948331e-01 3.68896633e-01 1.29395723e+00 4.35726702e-01 -5.26750274e-02 -1.33418381e+00 6.70201540e-01 -1.30394030e+00 -7.35619426e-01 -7.15726972e-01 1.87485206e+00 1.06171656e+00 -5.01105189e-02 -9.38180462e-02 6.81774914e-02 4.92252380e-01 -2.22772777e-01 9.83766317e-02 -7.43011475e-01 1.77921832e-01 5.05364656e-01 3.00908566e-01 5.29539943e-01 -9.31631744e-01 5.23067653e-01 6.27944851e+00 9.13156390e-01 -6.64456904e-01 1.01316333e-01 4.75462139e-01 6.05324030e-01 -5.51655233e-01 1.87617149e-02 -1.33666372e+00 2.98203379e-01 1.41387999e+00 1.15947761e-01 -1.77859187e-01 7.88896203e-01 -2.90400356e-01 -2.90455997e-01 -8.60447943e-01 3.00739586e-01 4.05326873e-01 -1.24237490e+00 -1.25828728e-01 -5.95873334e-02 9.54572499e-01 -1.78403437e-01 -2.06677794e-01 7.29367971e-01 3.96181583e-01 -9.14996982e-01 5.36525309e-01 5.09981573e-01 2.80330688e-01 -3.56017977e-01 7.26440191e-01 5.89498520e-01 -7.68380821e-01 -2.81797260e-01 -1.63571797e-02 -1.86767250e-01 -6.06624961e-01 2.89540082e-01 -9.44764018e-01 5.31582057e-01 6.63542449e-01 2.73131877e-01 -1.06437004e+00 1.09170234e+00 -3.66984814e-01 1.12169003e+00 1.08067714e-01 -4.43626553e-01 7.28313476e-02 -2.16789395e-01 1.52319390e-02 1.12014961e+00 1.52180091e-01 3.98778617e-01 3.90029281e-01 4.42215204e-01 -3.19957495e-01 5.91560662e-01 1.06391078e-02 1.98037967e-01 5.61976135e-01 1.65232265e+00 -6.54794335e-01 -2.68436521e-01 -6.17594182e-01 1.79183453e-01 3.65105420e-01 3.14254127e-02 -2.22934768e-01 -7.61832952e-01 9.16678682e-02 1.67935252e-01 -5.74155301e-02 1.72633201e-01 -2.67005414e-01 -1.28617918e+00 -1.63456872e-01 -1.24367356e+00 8.33412945e-01 -5.28189898e-01 -1.24804068e+00 2.71304935e-01 -3.82891178e-01 -1.20502865e+00 1.19780123e-01 -9.69733059e-01 -7.67013371e-01 7.50294328e-01 -1.83122456e+00 -8.61719906e-01 -5.86791039e-01 4.66225833e-01 2.70538628e-01 -3.65893751e-01 9.25750911e-01 2.17286140e-01 -7.03398705e-01 8.73213649e-01 4.00920600e-01 3.55178297e-01 1.02556801e+00 -1.72650874e+00 -6.06371045e-01 3.24908465e-01 1.70835912e-01 3.48022878e-01 4.66387808e-01 -2.43203267e-01 -1.34787834e+00 -1.02511954e+00 1.25821567e+00 -9.31649029e-01 9.99540091e-01 -8.08523875e-03 -1.24704957e+00 4.46648031e-01 5.94171822e-01 -1.99798733e-01 1.70428312e+00 2.49443635e-01 -5.00491500e-01 1.81487501e-02 -1.18677497e+00 8.16831961e-02 2.15687126e-01 -7.36588538e-01 -1.25803661e+00 4.73273575e-01 4.32603687e-01 -3.51056218e-01 -1.78656483e+00 1.26087308e-01 3.67326230e-01 -8.64966214e-01 5.24729788e-01 -7.08311081e-01 7.59026289e-01 7.86668286e-02 4.59032655e-02 -9.41484869e-01 2.90510338e-02 -2.35610873e-01 -1.08178079e-01 1.53231966e+00 3.88818800e-01 -3.44844699e-01 9.09596443e-01 8.68533492e-01 -4.10743028e-01 -8.64210963e-01 -6.55377626e-01 -3.58355701e-01 5.78318596e-01 -1.67662337e-01 2.99974680e-01 1.36717975e+00 5.23945808e-01 6.04521751e-01 6.25284314e-01 -1.58436850e-01 5.99664152e-01 5.87014914e-01 6.14985466e-01 -1.45948958e+00 8.45416114e-02 -6.44123375e-01 -2.83512741e-01 -8.99506748e-01 4.90584403e-01 -1.37972534e+00 -1.87390730e-01 -1.37074435e+00 9.81357470e-02 -4.38881397e-01 -2.89902151e-01 4.27193910e-01 -6.23662770e-01 5.19595593e-02 3.70935537e-02 1.14487693e-01 -1.12124383e+00 1.73270762e-01 1.35438764e+00 6.88827336e-02 1.64992020e-01 8.63324404e-02 -7.10769653e-01 6.51203990e-01 7.71493495e-01 -4.67479080e-01 -9.51084960e-03 -2.72502210e-02 3.42171133e-01 1.35831445e-01 4.16639969e-02 -8.48980665e-01 4.34314728e-01 -2.66036332e-01 1.94356978e-01 -4.81347889e-01 -2.45489299e-01 -7.13940620e-01 -8.36564779e-01 3.32527786e-01 -9.37545180e-01 -3.40560943e-01 -8.31119865e-02 1.64333284e-01 -4.38074291e-01 -9.88032341e-01 6.37599289e-01 -4.80548441e-02 -5.68637371e-01 -1.19074270e-01 -5.63196063e-01 4.61779118e-01 9.57527697e-01 7.91453104e-03 -5.94776094e-01 -3.07791591e-01 -7.32385039e-01 6.48343146e-01 1.14634641e-01 5.91938198e-01 4.07045305e-01 -1.07624662e+00 -8.14811170e-01 -9.68661383e-02 3.35342765e-01 -2.86514431e-01 3.95275354e-02 4.67454880e-01 -4.87466365e-01 6.65874302e-01 -2.81152576e-01 -4.96115029e-01 -1.61313760e+00 4.41738367e-02 2.73285896e-01 -6.01598084e-01 1.71549171e-01 9.13392663e-01 -4.88326192e-01 -1.16575730e+00 3.77521157e-01 -1.96639240e-01 -1.21525824e+00 5.94237447e-01 5.94824493e-01 4.00208592e-01 2.87501633e-01 -4.19102132e-01 9.05255526e-02 3.33956629e-01 -1.10745832e-01 1.74031168e-01 9.58391368e-01 -1.45479381e-01 -1.55271441e-01 4.41073179e-01 1.37185979e+00 1.35299563e-01 -3.46002400e-01 -4.17605728e-01 5.77603400e-01 -1.19611435e-01 1.53533146e-01 -9.45497751e-01 -5.69802821e-01 8.50797117e-01 6.64080083e-01 7.44345605e-01 7.40083814e-01 -2.85093347e-03 5.04933119e-01 5.85244179e-01 -4.20300639e-04 -1.13856947e+00 3.18268657e-01 1.03302336e+00 4.79221523e-01 -1.58255541e+00 -1.03532195e-01 -4.62103248e-01 -2.69253515e-02 1.34571052e+00 9.24015701e-01 3.05924535e-01 7.78755665e-01 -3.39627415e-01 4.45512682e-01 -1.60709247e-01 -8.45512033e-01 -1.81011349e-01 8.75271261e-01 7.43962778e-03 1.03253865e+00 -2.89995987e-02 -5.24487197e-01 8.53147388e-01 -4.71811652e-01 -2.04042420e-01 8.18909645e-01 1.16984832e+00 -9.54816639e-01 -1.20558774e+00 -5.84227383e-01 6.69324577e-01 -9.61768568e-01 7.68710375e-02 -4.88912255e-01 2.95589954e-01 9.33458954e-02 1.29584777e+00 -2.31209844e-01 -3.74159187e-01 2.84113765e-01 4.23660427e-01 4.14667845e-01 -1.00098789e+00 -1.47692132e+00 -6.83035672e-01 2.44545385e-01 -9.95939821e-02 -3.12731236e-01 -5.14320374e-01 -9.65023220e-01 1.45371303e-01 -6.87307477e-01 8.75391185e-01 8.73316467e-01 1.17266440e+00 1.24559902e-01 6.23668730e-01 8.87012184e-01 1.36641875e-01 -8.81386042e-01 -9.49873924e-01 -1.59192070e-01 4.66230452e-01 2.04718471e-01 -1.81952193e-01 -6.30597830e-01 1.55035287e-01]
[11.313522338867188, 8.3779935836792]
5be7d787-0860-4577-b12d-1dbf829d81e1
telling-stories-through-multi-user-dialogue
2105.15054
null
https://arxiv.org/abs/2105.15054v1
https://arxiv.org/pdf/2105.15054v1.pdf
Telling Stories through Multi-User Dialogue by Modeling Character Relations
This paper explores character-driven story continuation, in which the story emerges through characters' first- and second-person narration as well as dialogue -- requiring models to select language that is consistent with a character's persona and their relationships with other characters while following and advancing the story. We hypothesize that a multi-task model that trains on character dialogue plus character relationship information improves transformer-based story continuation. To this end, we extend the Critical Role Dungeons and Dragons Dataset (Rameshkumar and Bailey, 2020) -- consisting of dialogue transcripts of people collaboratively telling a story while playing the role-playing game Dungeons and Dragons -- with automatically extracted relationships between each pair of interacting characters as well as their personas. A series of ablations lend evidence to our hypothesis, showing that our multi-task model using character relationships improves story continuation accuracy over strong baselines.
['Mark O. Riedl', 'Prithviraj Ammanabrolu', 'Wai Man Si']
2021-05-31
null
https://aclanthology.org/2021.sigdial-1.30
https://aclanthology.org/2021.sigdial-1.30.pdf
sigdial-acl-2021-7
['story-continuation']
['computer-vision']
[ 1.74219698e-01 2.98315138e-01 -4.99861389e-02 -4.54964459e-01 -9.12558794e-01 -8.93887758e-01 1.44481063e+00 3.31861645e-01 -3.40400159e-01 9.81161892e-01 1.29506695e+00 6.30840436e-02 1.38137892e-01 -8.16004395e-01 -1.98603570e-01 -1.64810613e-01 -1.76970020e-01 1.01160407e+00 7.26295188e-02 -8.87633145e-01 3.34735066e-01 -3.15583587e-01 -9.93673921e-01 8.92303765e-01 3.27218205e-01 -3.89368013e-02 -3.43437269e-02 1.26727045e+00 -8.35843850e-04 1.56032479e+00 -8.18483710e-01 -7.86132634e-01 -2.18473047e-01 -9.59628880e-01 -1.30442393e+00 -1.59843266e-01 3.91203225e-01 -7.09309578e-01 -6.41228855e-01 5.34601919e-02 6.15913272e-01 3.92564744e-01 7.55433083e-01 -1.24229777e+00 -3.36152256e-01 1.39124620e+00 -3.83872628e-01 1.49597168e-01 1.26952088e+00 1.83071375e-01 1.52403748e+00 -4.51902986e-01 1.00618839e+00 1.32870376e+00 1.11935771e+00 8.33524168e-01 -1.27033854e+00 -4.42409277e-01 7.56307393e-02 -2.13161297e-03 -8.66719842e-01 -6.59589291e-01 8.28494608e-01 -6.27161145e-01 1.26895678e+00 3.83439898e-01 9.96957183e-01 1.78462577e+00 -1.07140616e-01 9.94978249e-01 5.63470721e-01 -2.41252571e-01 -4.13185358e-01 -8.85625109e-02 2.49538213e-01 3.39284688e-01 -5.41899025e-01 -3.07473272e-01 -1.07673275e+00 -5.13346434e-01 6.95160389e-01 -7.93005109e-01 4.90708873e-02 2.32776642e-01 -1.65879691e+00 9.22594786e-01 -3.04923445e-01 5.15508652e-01 -1.94239631e-01 5.03466606e-01 8.49240720e-01 4.83603209e-01 5.51931441e-01 8.75760019e-01 5.87996021e-02 -9.58682299e-01 -6.95415080e-01 1.08839118e+00 1.27931738e+00 1.00449038e+00 5.89122474e-02 -2.89771616e-01 -6.61663353e-01 1.08878958e+00 -5.70023321e-02 -2.66848564e-01 2.06695557e-01 -1.00867510e+00 6.35176778e-01 1.99799716e-01 2.76709497e-01 -8.75904262e-01 -4.58957583e-01 1.74079061e-01 -5.62801361e-01 -2.29263306e-01 7.61520207e-01 -4.96654183e-01 2.07505271e-01 2.14343286e+00 1.89550221e-01 7.85195362e-03 2.53216267e-01 5.05649805e-01 1.26927853e+00 7.11796284e-01 5.04300408e-02 -2.37151355e-01 1.53881967e+00 -8.00432920e-01 -6.16031587e-01 -1.63218781e-01 7.56398797e-01 -5.46273649e-01 1.17064202e+00 2.91490883e-01 -1.42017221e+00 -4.97235328e-01 -8.67098153e-01 -4.51079667e-01 3.17387998e-01 -1.33852467e-01 8.77971053e-01 2.06291810e-01 -1.05965841e+00 6.94459617e-01 -3.25077951e-01 -7.40901053e-01 1.55008510e-01 -1.07189104e-01 -3.16338688e-01 4.94108081e-01 -1.40020561e+00 1.06719911e+00 2.79606462e-01 -3.54330808e-01 -9.44858551e-01 -6.09705091e-01 -8.56873631e-01 -3.30270737e-01 2.55539984e-01 -7.66407847e-01 1.70666504e+00 -5.69246531e-01 -1.61341012e+00 1.14741039e+00 2.05133893e-02 -4.88089770e-01 9.71274674e-01 -6.06148362e-01 4.80666123e-02 9.69209895e-02 3.89230192e-01 6.77725017e-01 2.97523737e-01 -1.04924285e+00 -7.40673125e-01 1.44703478e-01 4.82399791e-01 8.26870799e-01 -1.86160237e-01 6.37020826e-01 -4.45504785e-02 -6.16468847e-01 -3.60871166e-01 -6.37358248e-01 5.06398119e-02 -3.62064928e-01 -7.11165488e-01 -7.27546096e-01 4.04401183e-01 -6.62262142e-01 1.10463166e+00 -1.60158050e+00 2.62122095e-01 -4.31217670e-01 6.35362506e-01 -3.75617832e-01 -2.73191154e-01 1.18897331e+00 7.24128187e-02 2.67808733e-04 -8.87904689e-03 -8.03278387e-01 -7.20917434e-02 -1.64973021e-01 -2.81825423e-01 4.20976281e-01 1.00416854e-01 1.16124225e+00 -1.12499452e+00 -6.10230088e-01 -3.88046727e-02 2.62092888e-01 -3.15166771e-01 3.93753529e-01 -5.41756570e-01 5.69976091e-01 -2.96289027e-01 -1.54961944e-02 -3.14234123e-02 -1.35336161e-01 2.12568730e-01 4.49715763e-01 -2.55869061e-01 1.06850076e+00 -4.70808774e-01 1.77067554e+00 -3.71532440e-01 1.18788505e+00 -1.05644921e-02 -3.58803362e-01 9.03382659e-01 6.78493321e-01 4.33371067e-01 3.82041596e-02 8.96024927e-02 -2.88744062e-01 1.83945611e-01 -6.60896719e-01 7.65315950e-01 -4.40895677e-01 -7.31648207e-01 1.30094409e+00 -3.57225984e-02 -8.70751560e-01 3.60891491e-01 8.40577304e-01 1.26538002e+00 4.56296772e-01 4.98938948e-01 -8.65105391e-02 2.38998011e-01 1.49162218e-01 2.09186628e-01 1.07518458e+00 3.89990062e-02 6.79777920e-01 1.05551779e+00 -4.28662568e-01 -1.32549083e+00 -7.86449492e-01 3.36275786e-01 1.50699735e+00 -1.81655169e-01 -7.56458104e-01 -6.70476735e-01 -4.07391965e-01 -3.01001102e-01 1.24068820e+00 -8.13935518e-01 2.41289422e-01 -1.14311361e+00 -3.64102274e-01 1.16576171e+00 2.52187878e-01 4.91242677e-01 -1.19370735e+00 -8.47040892e-01 5.87756217e-01 -8.19212794e-01 -1.08092332e+00 -4.94767904e-01 -2.68505394e-01 -2.32882991e-01 -1.07472289e+00 -6.86349452e-01 -6.69179082e-01 -1.19670980e-01 -1.12072080e-01 1.35779834e+00 7.71454573e-02 2.56557882e-01 4.43359047e-01 -6.86256409e-01 8.99319944e-04 -9.95713770e-01 2.58491099e-01 -1.48598775e-01 -1.88367054e-01 -7.61752576e-02 -1.10479057e+00 1.77965894e-01 7.96270650e-03 -5.40802777e-02 8.86890590e-01 -7.21013993e-02 7.24475861e-01 -4.13753211e-01 -5.39969563e-01 5.70763290e-01 -1.22655189e+00 1.25726748e+00 -7.06988692e-01 3.09880614e-01 1.29097030e-01 -1.09824231e-02 -4.70244557e-01 7.51573980e-01 -6.50984466e-01 -1.14682925e+00 -3.07972282e-01 -7.42845088e-02 4.30783898e-01 -4.11736853e-02 2.93503016e-01 -9.02317092e-03 7.19464660e-01 9.16055322e-01 3.22776109e-01 -6.99493960e-02 -1.26022607e-01 5.87055802e-01 4.40461010e-01 1.09758604e+00 -1.01148462e+00 7.28559196e-01 1.69539601e-01 -3.91260505e-01 -8.48604798e-01 -7.88101137e-01 -4.30004925e-01 -6.52657926e-01 -7.74344742e-01 7.65571892e-01 -1.03690219e+00 -9.25742149e-01 7.58186340e-01 -1.91166019e+00 -7.67797530e-01 -2.02716336e-01 2.21117198e-01 -9.75083530e-01 2.73903698e-01 -1.04240632e+00 -9.11666870e-01 -4.07979846e-01 -3.91199142e-01 6.80451751e-01 1.55372256e-02 -1.43662930e+00 -9.92970407e-01 3.71782839e-01 6.39773726e-01 3.72729115e-02 6.94819033e-01 9.78339195e-01 -1.06559217e+00 5.30194119e-02 -1.65966511e-01 6.86767772e-02 -4.59816724e-01 2.86704320e-02 -2.61427388e-02 -6.91393912e-01 7.46292099e-02 -1.20764256e-01 -9.48995292e-01 4.54865068e-01 1.28038079e-02 1.57439783e-01 -4.97019053e-01 -2.84450471e-01 6.73929453e-02 4.84963596e-01 -1.28214210e-01 4.84021276e-01 1.02312051e-01 6.91554308e-01 1.11348367e+00 4.56175655e-01 8.39173019e-01 9.79373336e-01 6.44239485e-01 -1.26323104e-01 2.98175663e-01 -1.05915606e-01 -8.16551864e-01 4.94096845e-01 4.82910901e-01 -3.07952315e-01 -6.58681929e-01 -8.94172966e-01 7.17447698e-01 -2.12045765e+00 -1.50230050e+00 -6.80070758e-01 1.53987539e+00 1.32452846e+00 2.38279670e-01 7.65113354e-01 -1.67824626e-01 8.44740033e-01 7.61982858e-01 -4.08734351e-01 -3.01938802e-01 -4.59717542e-01 -3.08654368e-01 -3.57588500e-01 9.05713916e-01 -9.33188319e-01 1.26855099e+00 6.66806221e+00 4.89683628e-01 -3.52878660e-01 2.45453596e-01 7.22760797e-01 -5.96899331e-01 -5.94627678e-01 9.50850621e-02 -6.59576654e-01 1.73896745e-01 6.75330222e-01 -5.18777430e-01 4.66096103e-01 2.99768418e-01 3.82977992e-01 -3.46226484e-01 -1.76140738e+00 7.32393444e-01 5.17108500e-01 -1.61836159e+00 -1.86279297e-01 -2.60105163e-01 5.36249876e-01 -4.23452079e-01 -3.82982463e-01 3.45941633e-01 9.31861162e-01 -1.16269362e+00 1.22791529e+00 2.60616183e-01 9.18523371e-01 -3.78892779e-01 3.67451847e-01 6.97496295e-01 -1.03298891e+00 2.09712923e-01 3.41934174e-01 -7.24647224e-01 6.85056746e-01 -2.99972951e-01 -1.12902009e+00 2.29993872e-02 5.90028167e-02 7.48308599e-01 -9.17972028e-02 3.07674736e-01 -5.47520339e-01 7.36693263e-01 -2.14284614e-01 -6.08511269e-01 2.66353309e-01 -5.97154833e-02 1.03084779e+00 1.53125882e+00 -2.55501717e-01 7.05303609e-01 8.74069110e-02 1.04670823e+00 -7.44766444e-02 9.41930935e-02 -6.50997877e-01 -2.66108096e-01 6.31489456e-01 1.09144235e+00 -5.78403831e-01 -3.63923669e-01 -2.28402510e-01 1.20745659e+00 3.48946750e-01 1.60554182e-02 -6.54723406e-01 -1.48180887e-01 5.72951198e-01 4.13123995e-01 -4.19732928e-01 -3.55334759e-01 -4.91120636e-01 -8.38319957e-01 -3.57577235e-01 -9.98454154e-01 3.13171923e-01 -9.46568489e-01 -1.26174080e+00 6.29767537e-01 4.25559670e-01 -8.31248164e-01 -9.04438615e-01 3.27583730e-01 -1.28754354e+00 7.78679490e-01 -4.77827609e-01 -1.62597752e+00 -5.72041869e-02 3.61919045e-01 8.82132292e-01 -3.22077125e-01 7.74411321e-01 -3.70762289e-01 -2.28457227e-01 6.91948771e-01 -5.07957697e-01 5.98895490e-01 6.28045738e-01 -1.15168118e+00 1.17639327e+00 5.39135039e-01 3.04393172e-01 5.28320551e-01 1.07287288e+00 -9.60007668e-01 -8.21496904e-01 -4.63085473e-01 1.46340859e+00 -8.12779367e-01 8.41281414e-01 -8.74028325e-01 -6.64017260e-01 8.17201674e-01 6.76407754e-01 -8.64876866e-01 8.37448955e-01 4.99777704e-01 -3.88710827e-01 6.66599691e-01 -9.52641129e-01 1.00340855e+00 1.49019384e+00 -8.37293565e-01 -1.01308215e+00 5.95476687e-01 7.34617174e-01 -4.25467879e-01 -5.12008607e-01 -1.93384543e-01 7.32888758e-01 -1.04148281e+00 6.77353561e-01 -9.35853362e-01 1.20607126e+00 3.49862278e-01 5.66385239e-02 -1.20661807e+00 -1.43042997e-01 -1.39365208e+00 1.53458253e-01 1.61554623e+00 5.17552435e-01 -6.83073625e-02 7.45030105e-01 8.11912239e-01 -1.44282639e-01 -4.19722766e-01 -8.08261156e-01 -2.12208971e-01 2.75458574e-01 -5.61223209e-01 2.76603162e-01 1.18938410e+00 7.48316407e-01 1.20219326e+00 -8.01475585e-01 -5.93019783e-01 1.60005748e-01 -1.83549076e-01 1.13385785e+00 -1.28031898e+00 -6.90583110e-01 -6.71005368e-01 1.79834470e-01 -1.21865821e+00 4.05566365e-01 -7.42905796e-01 2.28181988e-01 -1.72046089e+00 5.60546815e-01 -2.89653093e-01 8.12565029e-01 4.54178661e-01 -1.81878030e-01 -9.38299671e-02 3.20607126e-01 3.01406711e-01 -8.08276296e-01 6.28094852e-01 1.15247047e+00 1.48040184e-04 -5.33264220e-01 2.14632317e-01 -8.18945706e-01 8.44310164e-01 5.02994359e-01 -4.06345278e-01 -5.36144853e-01 -2.54721940e-01 5.71680129e-01 7.50084579e-01 3.37619871e-01 -5.92046440e-01 4.85783041e-01 -2.28856876e-01 4.39264020e-03 -5.15257835e-01 7.81080544e-01 3.10014904e-01 1.10222526e-01 1.53904587e-01 -1.16318130e+00 -2.19545215e-02 -5.15789799e-02 3.11291754e-01 9.98468399e-02 -2.55627543e-01 3.99919540e-01 -4.51006085e-01 -3.10601473e-01 -1.36740625e-01 -1.07253885e+00 4.41775620e-01 9.57432985e-01 -5.45535743e-01 -3.54361117e-01 -1.48370230e+00 -5.23169100e-01 5.16718090e-01 2.86236465e-01 3.78792018e-01 6.34159982e-01 -1.16530859e+00 -1.70912516e+00 -4.15982664e-01 9.54600144e-03 -1.98794305e-02 -1.44728562e-02 3.64807665e-01 -3.83616656e-01 -2.15758439e-02 -6.17002323e-02 -9.03160721e-02 -1.60354745e+00 -3.48538429e-01 2.75041163e-01 -2.70139605e-01 -8.40629578e-01 1.56031501e+00 -1.95499539e-01 -3.75979006e-01 7.10584074e-02 1.90259263e-01 -5.57007432e-01 4.64715928e-01 6.78643584e-01 2.92749316e-01 -8.45877409e-01 -7.38960683e-01 -1.23833962e-01 1.05018817e-01 -2.05228195e-01 -1.05976689e+00 1.36587119e+00 -1.30680665e-01 -2.67064691e-01 1.07771206e+00 8.95156562e-01 2.43926212e-01 -1.24610722e+00 -1.51493996e-01 7.35447630e-02 -2.25848272e-01 -6.81129754e-01 -9.40985024e-01 1.27891392e-01 5.33481419e-01 -8.39630604e-01 4.06937629e-01 2.25838035e-01 4.55197841e-01 1.02510977e+00 3.49199891e-01 3.22357446e-01 -9.17287767e-01 6.74550891e-01 1.01295424e+00 1.42452466e+00 -1.01295030e+00 1.06272520e-02 -2.43251085e-01 -1.28926587e+00 1.07675695e+00 8.16839993e-01 -1.41786665e-01 9.15714055e-02 3.26913923e-01 -1.15046903e-01 -1.49191558e-01 -1.35555327e+00 3.87837328e-02 -2.38665879e-01 7.43799388e-01 8.86257350e-01 1.06030613e-01 -2.98057526e-01 9.51685131e-01 -9.48358834e-01 -4.56755221e-01 8.99237216e-01 4.53736156e-01 -1.15413502e-01 -9.44155991e-01 -1.35199845e-01 4.35703307e-01 -2.29444169e-02 -2.88691819e-01 -1.32008994e+00 7.18300700e-01 5.00837639e-02 1.37567008e+00 2.86771297e-01 -6.77994430e-01 8.56261626e-02 1.59392685e-01 6.33548498e-01 -1.05495322e+00 -1.43260407e+00 -1.85679108e-01 1.27356029e+00 -1.24344639e-01 -1.12382263e-01 -1.30795431e+00 -1.28083241e+00 -7.76660740e-01 -5.65801486e-02 2.11939871e-01 -5.84273934e-02 1.11212933e+00 -3.22826833e-01 1.59663111e-01 4.77942497e-01 -6.72582626e-01 -3.03759426e-01 -1.46504962e+00 -2.55989552e-01 4.17187870e-01 1.92039490e-01 -1.05974890e-01 -9.85185057e-03 -3.29941027e-02]
[12.32728385925293, 8.394868850708008]
a7f444cd-617f-4f11-9a91-dcf691240317
aggregated-text-transformer-for-scene-text
2211.13984
null
https://arxiv.org/abs/2211.13984v1
https://arxiv.org/pdf/2211.13984v1.pdf
Aggregated Text Transformer for Scene Text Detection
This paper explores the multi-scale aggregation strategy for scene text detection in natural images. We present the Aggregated Text TRansformer(ATTR), which is designed to represent texts in scene images with a multi-scale self-attention mechanism. Starting from the image pyramid with multiple resolutions, the features are first extracted at different scales with shared weight and then fed into an encoder-decoder architecture of Transformer. The multi-scale image representations are robust and contain rich information on text contents of various sizes. The text Transformer aggregates these features to learn the interaction across different scales and improve text representation. The proposed method detects scene texts by representing each text instance as an individual binary mask, which is tolerant of curve texts and regions with dense instances. Extensive experiments on public scene text detection datasets demonstrate the effectiveness of the proposed framework.
['Cheng Jin', 'Yingbin Zheng', 'Xiangcheng Du', 'Zhao Zhou']
2022-11-25
null
null
null
null
['scene-text-detection']
['computer-vision']
[ 7.18973041e-01 -3.95271212e-01 7.93459415e-02 -3.22576165e-01 -8.11668038e-01 -1.53857186e-01 7.38000631e-01 1.89020298e-02 -1.78165391e-01 1.38422385e-01 5.77037752e-01 2.28946269e-01 3.63735944e-01 -8.67772818e-01 -7.80826211e-01 -6.68148518e-01 5.87274909e-01 1.72623545e-01 7.35441804e-01 2.57517546e-02 4.32091981e-01 9.44003016e-02 -1.55451155e+00 1.16505539e+00 9.08124030e-01 1.03621972e+00 7.17276752e-01 1.05389023e+00 -6.31368279e-01 1.53596890e+00 -7.75525868e-01 -5.13239384e-01 2.24076167e-01 -2.66756296e-01 -5.28238297e-01 7.34433591e-01 8.76502037e-01 -7.25707173e-01 -8.65745544e-01 1.22687316e+00 5.35869658e-01 -1.49916679e-01 6.24233007e-01 -8.28887820e-01 -1.00907838e+00 7.87107706e-01 -1.02384305e+00 6.16379499e-01 2.10651442e-01 2.44432651e-02 9.83508527e-01 -1.28431523e+00 1.51613429e-01 1.64595997e+00 4.91630375e-01 8.69269073e-02 -7.79774070e-01 -5.36468506e-01 3.67425352e-01 2.35924587e-01 -1.42443812e+00 -3.40326428e-01 5.13399959e-01 -4.02483910e-01 1.08981299e+00 3.39724004e-01 3.78067583e-01 9.01597977e-01 5.26894748e-01 1.55071378e+00 7.93729782e-01 -2.75141656e-01 -3.32311451e-01 2.53802121e-01 9.66902748e-02 9.52324927e-01 3.13528359e-01 -6.38038754e-01 -8.59402597e-01 1.89595029e-01 7.44951844e-01 3.81481349e-01 -2.22105440e-02 4.46072035e-02 -1.29507864e+00 8.16489995e-01 4.38338280e-01 4.02327299e-01 -2.14063272e-01 1.22545242e-01 6.95103228e-01 5.47844507e-02 5.99324644e-01 -3.56098890e-01 -1.66352212e-01 3.49076867e-01 -1.01095808e+00 -1.73810512e-01 1.75074100e-01 1.15855598e+00 5.95849574e-01 2.34127745e-01 -5.68690538e-01 1.10163593e+00 2.27576599e-01 9.15864646e-01 8.58074546e-01 -4.14396375e-02 1.05709291e+00 1.06850278e+00 -3.83730114e-01 -1.13145626e+00 -2.41034478e-01 -1.48509428e-01 -1.11225522e+00 -1.22351773e-01 -1.56662345e-01 1.52721092e-01 -1.00583172e+00 8.86090100e-01 3.39837253e-01 1.60017118e-01 6.51564598e-02 6.43032312e-01 1.10522139e+00 1.07854438e+00 5.45437075e-02 1.61940157e-01 1.64271295e+00 -1.29316449e+00 -8.87919188e-01 -5.39909780e-01 3.66586745e-01 -9.93033409e-01 1.07414436e+00 2.22762942e-01 -1.08659887e+00 -7.83628941e-01 -9.09513950e-01 -5.17635405e-01 -6.25402272e-01 7.30863571e-01 2.42028236e-02 5.95519245e-01 -9.46848750e-01 3.55130583e-02 -5.19683182e-01 -6.72502995e-01 5.73088884e-01 1.45011753e-01 8.85992274e-02 -1.23046853e-01 -8.73842478e-01 6.06681585e-01 4.77156788e-01 -4.64964658e-02 -7.83141971e-01 -1.37714416e-01 -9.53933179e-01 3.36915851e-01 2.53649861e-01 -4.20964569e-01 9.54027176e-01 -1.29005480e+00 -1.25323200e+00 8.65532100e-01 -4.39975679e-01 -4.32196289e-01 4.42199647e-01 -3.26682985e-01 -4.85677361e-01 4.82194066e-01 6.08331561e-01 5.07799447e-01 1.59966040e+00 -9.35924768e-01 -1.12841117e+00 -3.32150072e-01 -3.28313649e-01 7.48968899e-01 -7.54272819e-01 4.90489900e-01 -7.79172957e-01 -8.32181215e-01 8.06634407e-03 -3.01358134e-01 -3.02853901e-02 -5.40230535e-02 -6.27911150e-01 -9.14655626e-02 1.37127531e+00 -8.11095238e-01 1.19273627e+00 -2.08329701e+00 7.51222000e-02 -4.05485213e-01 2.82763034e-01 2.28675269e-02 -2.34264746e-01 4.00368065e-01 2.40239412e-01 -7.41101950e-02 -1.02120815e-02 -3.92417461e-01 -3.16377059e-02 -1.11691311e-01 -5.40793657e-01 4.51415867e-01 3.32435995e-01 1.06424057e+00 -3.59346449e-01 -1.22527456e+00 8.98641706e-01 3.52804452e-01 -2.79239029e-01 1.32917270e-01 -1.37661174e-01 -2.47441426e-01 -8.22160721e-01 7.42057443e-01 8.55628133e-01 -6.24648750e-01 -3.41040581e-01 -4.28416818e-01 -1.78575397e-01 -6.65079430e-02 -1.16244662e+00 1.27526891e+00 -2.37704292e-01 9.22921062e-01 1.18973166e-01 -8.96360219e-01 7.61426568e-01 6.12953901e-02 4.14876491e-01 -8.09884548e-01 4.57415819e-01 -3.58916730e-01 -5.88196158e-01 -7.93889344e-01 9.24753070e-01 4.56003726e-01 -1.97623074e-01 2.79898763e-01 -6.37920499e-02 -2.84859985e-01 3.15548390e-01 4.78041559e-01 1.02802753e+00 -2.79670268e-01 3.01349312e-01 -1.64523304e-01 9.14713144e-01 -2.61200070e-02 7.06185624e-02 9.40302134e-01 -1.16867468e-01 4.99784499e-01 1.72488347e-01 -6.07467830e-01 -1.24621677e+00 -9.50432003e-01 -2.88425714e-01 1.61281574e+00 3.35886419e-01 -4.46851343e-01 -6.44725442e-01 -7.34819412e-01 -1.01922238e-02 2.18443558e-01 -7.89490223e-01 1.29267514e-01 -4.40039843e-01 -8.94178629e-01 4.47880864e-01 6.24004185e-01 1.22623861e+00 -1.02342200e+00 -6.70828521e-01 -7.50853419e-02 -3.48330885e-01 -1.58019197e+00 -8.31905901e-01 2.35680774e-01 -5.95935225e-01 -9.18512821e-01 -5.97123682e-01 -1.19114351e+00 6.56377137e-01 7.15458035e-01 7.53075182e-01 4.95250598e-02 -7.56185114e-01 4.70821619e-01 -5.97668111e-01 -3.07268649e-01 -3.93814653e-01 -2.00921595e-01 -4.82950091e-01 5.21874964e-01 3.94937962e-01 1.64805442e-01 -5.06915927e-01 7.15146139e-02 -1.13609338e+00 3.95485967e-01 8.47542644e-01 9.16028500e-01 3.14672858e-01 5.79232395e-01 1.82639033e-01 -7.27223456e-01 6.33143067e-01 -1.95153490e-01 -4.79802370e-01 3.42340410e-01 3.47371176e-02 -3.22186381e-01 8.77692997e-01 -3.71810526e-01 -1.29376650e+00 3.47724795e-01 2.36260355e-01 -3.26940238e-01 -1.76354885e-01 4.33245040e-02 -4.20582071e-02 -2.22632531e-02 4.37717438e-01 1.04681218e+00 -5.89781463e-01 -1.65744945e-01 4.06001329e-01 1.23857844e+00 3.43890488e-01 -2.94563621e-01 8.15246701e-01 6.48594618e-01 -4.84349012e-01 -1.35921371e+00 -9.89239931e-01 -6.68822885e-01 -8.83402109e-01 -1.12795308e-01 1.23839319e+00 -1.36836827e+00 -2.43931100e-01 7.23452151e-01 -1.03273714e+00 -2.58715451e-01 -1.47456959e-01 3.72757427e-02 -4.47957158e-01 7.77327836e-01 -1.00448668e+00 -7.81129539e-01 -6.72738135e-01 -1.11797225e+00 2.05622005e+00 2.43903786e-01 4.79540586e-01 -8.32983792e-01 -5.69365442e-01 5.14211297e-01 1.85653031e-01 -1.79342076e-01 7.06812143e-01 -4.44407225e-01 -7.59727597e-01 -2.18578532e-01 -8.42969239e-01 1.47836581e-01 1.84567481e-01 -3.01084127e-02 -1.02738571e+00 -2.90987521e-01 -1.02192469e-01 -4.26621437e-01 1.05109286e+00 3.72482687e-01 1.40120697e+00 -2.56336033e-01 -2.85320073e-01 4.20863479e-01 1.60264456e+00 1.98197067e-02 5.22476792e-01 3.97895336e-01 9.79821920e-01 1.67896748e-01 3.39359045e-01 6.55227125e-01 4.18705791e-01 2.11709857e-01 2.63279021e-01 -4.73124832e-01 -2.88683772e-01 -1.00770690e-01 5.12291729e-01 7.88186550e-01 7.25116849e-01 -5.12843966e-01 -7.63658226e-01 4.86871332e-01 -1.78383720e+00 -1.03245831e+00 -2.56417513e-01 1.45170796e+00 4.98793751e-01 1.16799086e-01 -1.05804987e-01 4.38869596e-02 1.17141974e+00 4.28912491e-01 -7.62502670e-01 -2.35755384e-01 -6.21825695e-01 -8.47854838e-02 6.32583439e-01 1.53431818e-01 -1.48432183e+00 1.38698697e+00 6.62417650e+00 9.99214828e-01 -9.42529559e-01 4.54283925e-03 6.14249885e-01 4.99883592e-02 3.93388867e-01 -4.37369317e-01 -9.09367681e-01 5.35686910e-01 3.48965764e-01 -4.40493494e-01 2.45879650e-01 9.29554403e-01 1.25308037e-02 -1.46406740e-01 -5.22075772e-01 1.08427203e+00 6.57908797e-01 -1.26350331e+00 6.35623574e-01 -4.56523001e-01 9.60304141e-01 3.01713437e-01 2.02930927e-01 2.45838121e-01 5.10218501e-01 -7.30542004e-01 7.40149140e-01 2.08575845e-01 8.61247122e-01 -4.54641372e-01 4.13679898e-01 2.83768505e-01 -1.88147044e+00 -5.24706423e-01 -8.52913916e-01 2.96178311e-01 -1.58482641e-01 3.53395730e-01 -6.94255590e-01 3.87308717e-01 1.02920902e+00 1.24344206e+00 -1.06312692e+00 7.09894657e-01 1.52376875e-01 3.69308442e-01 -2.24829033e-01 -3.27134162e-01 3.00029516e-01 2.89197266e-02 2.46611506e-01 1.69624937e+00 2.66929895e-01 -2.64713094e-02 4.56701010e-01 5.78682303e-01 -8.59366059e-02 5.41510463e-01 -5.97561419e-01 -1.07012335e-02 2.24704087e-01 1.29490042e+00 -1.00073576e+00 -8.96409094e-01 -8.91658187e-01 1.44933224e+00 1.16603732e-01 3.08353692e-01 -8.73294890e-01 -5.21787882e-01 3.62927765e-02 -2.14789823e-01 7.08530068e-01 2.23328680e-01 -4.02774304e-01 -1.47111225e+00 6.80715218e-02 -9.75648642e-01 7.01772213e-01 -1.28301430e+00 -1.14778149e+00 6.24842763e-01 -3.99653673e-01 -1.19463491e+00 5.21017373e-01 -6.92066312e-01 -6.09934986e-01 5.52419007e-01 -1.41139543e+00 -1.58808053e+00 -5.05239248e-01 1.15936148e+00 1.62762976e+00 -2.77160704e-01 3.91662121e-01 1.18664555e-01 -9.33576047e-01 4.86169040e-01 5.19709527e-01 4.23593223e-01 4.71051663e-01 -1.23639345e+00 5.38704038e-01 1.13855624e+00 6.79036751e-02 5.81850372e-02 4.26225454e-01 -8.95080030e-01 -1.43423009e+00 -1.51248276e+00 3.96621078e-01 -4.47964609e-01 8.40126336e-01 -5.13970315e-01 -7.82805622e-01 8.05162787e-01 6.86557055e-01 -9.79468971e-02 6.49584904e-02 -5.39560080e-01 -3.21467400e-01 -8.07645097e-02 -1.03750920e+00 6.78494751e-01 7.01643765e-01 -6.33882880e-01 -7.15360522e-01 8.06150377e-01 7.46886849e-01 -4.19161707e-01 -5.15572608e-01 1.39547482e-01 1.30812213e-01 -9.36350763e-01 9.17299330e-01 -1.00767322e-01 5.38967013e-01 -9.76326913e-02 -3.51915568e-01 -7.54841924e-01 -5.00756741e-01 -1.50817350e-01 3.99399027e-02 1.18005335e+00 -7.74598792e-02 -4.03892428e-01 6.87736154e-01 -5.10798804e-02 -4.73898090e-02 -2.79683143e-01 -9.30651605e-01 -2.32811496e-01 4.78614680e-03 -8.33690912e-02 5.41325867e-01 7.74463773e-01 -1.86794437e-02 6.65294826e-01 -6.40921712e-01 3.95103931e-01 6.91491604e-01 2.56633133e-01 7.62722790e-01 -8.49643350e-01 1.31893560e-01 -5.67692220e-01 -4.78448182e-01 -1.49611282e+00 8.89364444e-03 -8.03502023e-01 1.66151300e-01 -1.72358668e+00 7.71839082e-01 2.66908467e-01 -6.22481368e-02 2.29268432e-01 -4.95933533e-01 3.05567861e-01 2.41645202e-01 7.48514608e-02 -1.10461140e+00 6.89257383e-01 1.15739059e+00 -6.44811571e-01 3.12421203e-01 -3.99141818e-01 -3.94705892e-01 8.84049594e-01 7.03432739e-01 -2.48625696e-01 -1.78270295e-01 -7.88821518e-01 -1.30037844e-01 -7.95153305e-02 2.18174711e-01 -1.12099922e+00 3.91865671e-01 -6.45613372e-02 9.87821698e-01 -1.39320827e+00 3.16922307e-01 -9.68889952e-01 -6.21451259e-01 4.56010431e-01 -5.48515022e-01 1.18969253e-03 3.26118201e-01 7.04833508e-01 -4.24394421e-02 1.95408855e-02 1.00124276e+00 -4.88744676e-02 -7.70288765e-01 2.28313327e-01 -6.66109085e-01 -3.77705768e-02 1.04194844e+00 -2.63418049e-01 -5.49594700e-01 -1.61281720e-01 -2.67689377e-01 3.81551087e-01 2.60591120e-01 6.29613042e-01 9.43022728e-01 -1.09634626e+00 -8.95781696e-01 3.85798454e-01 2.30742201e-01 -1.12210803e-01 5.08880079e-01 4.48636562e-01 -4.58718330e-01 5.01509190e-01 2.07103929e-03 -9.16271091e-01 -1.55071306e+00 7.78527677e-01 4.88638759e-01 -3.34862471e-01 -1.12320173e+00 7.55281925e-01 7.91437328e-01 -6.63897721e-03 2.58074820e-01 -4.11048621e-01 -3.47943872e-01 -1.38271078e-01 8.85944605e-01 6.92884177e-02 -2.08323538e-01 -1.09745944e+00 -8.93878564e-02 1.10295081e+00 -3.47087592e-01 2.87124008e-01 1.02656579e+00 -7.57121027e-01 -1.12186678e-01 4.30231065e-01 1.05457520e+00 1.47884269e-03 -1.20926583e+00 -7.35663533e-01 -3.36219400e-01 -7.65658975e-01 1.37340918e-01 -5.24519861e-01 -1.04411709e+00 9.06732082e-01 6.55240357e-01 4.13302541e-01 1.40064096e+00 -7.94439465e-02 8.49654794e-01 5.17884254e-01 -1.49563432e-01 -1.33276045e+00 7.79719293e-01 5.07909954e-01 8.10041428e-01 -1.41476011e+00 1.00110807e-01 -3.72281730e-01 -9.69185412e-01 1.21501505e+00 7.90035665e-01 -2.64339954e-01 2.21534401e-01 6.85951471e-01 3.93400975e-02 -2.46940330e-01 -8.37088764e-01 -5.22433639e-01 9.59460810e-02 3.91668707e-01 2.39370659e-01 -2.61671722e-01 2.75494307e-01 1.66785911e-01 7.97384977e-02 -3.44174504e-01 4.93750781e-01 8.25067580e-01 -1.02311909e+00 -4.17216271e-01 -7.37787902e-01 9.57299113e-01 -5.63953280e-01 -3.88596505e-01 -6.00359201e-01 4.88439173e-01 -5.02371602e-02 1.05741239e+00 8.49601105e-02 -3.83831978e-01 3.28542113e-01 7.06805885e-02 2.15627983e-01 -6.11464500e-01 -6.76299036e-01 4.75551933e-01 -3.81402582e-01 -3.18349600e-01 -2.64544487e-01 -7.08067596e-01 -1.12892604e+00 -7.36935213e-02 -4.75537419e-01 -1.51139751e-01 3.59076738e-01 6.35399044e-01 1.27511889e-01 8.97140503e-01 9.62772787e-01 -6.08369470e-01 -3.70116204e-01 -1.02464747e+00 -6.77145898e-01 5.03694236e-01 4.92946953e-01 -1.37412772e-01 -3.10957342e-01 5.94221711e-01]
[12.034929275512695, 2.2582921981811523]
dcb497c6-0ec7-4c42-a161-6776cbbd3d9a
semantic-role-labeling-as-dependency-parsing-1
null
null
https://openreview.net/forum?id=bmF2qC-CUG
https://openreview.net/pdf?id=bmF2qC-CUG
Semantic Role Labeling as Dependency Parsing: Exploring Latent Tree Structures Inside Arguments
Semantic role labeling (SRL) is a fundamental yet challenging task in the NLP community. Recent works of SRL mainly fall into two lines: 1) BIO-based; 2) span-based. Despite ubiquity, they share some intrinsic drawbacks of not explicitly considering internal argument structures, which may potentially hinder the model's expressiveness. To remedy this, we propose to reduce SRL to a dependency parsing task and regard the flat argument spans as latent subtrees. In particular, we equip our formulation with a novel span-constrained TreeCRF to make tree structures span-aware, and further extend it to the second-order case. Experiments on CoNLL05 and CoNLL12 benchmarks reveal that the results of our methods outperform all previous works and achieve the state-of-the-art.
['Anonymous']
2021-11-16
null
null
null
acl-arr-november-2021-11
['semantic-role-labeling']
['natural-language-processing']
[ 4.71214980e-01 5.57036757e-01 -8.17731082e-01 -3.45518529e-01 -6.44787669e-01 -9.10172701e-01 5.52700222e-01 2.20942959e-01 -1.74771503e-01 9.29217637e-01 6.20470524e-01 -5.98241806e-01 -2.98581362e-01 -6.47944808e-01 -5.25269747e-01 -6.01106882e-01 2.77647525e-01 3.37656677e-01 6.45519555e-01 -2.62823731e-01 2.13699341e-01 2.09212378e-01 -1.35344803e+00 3.32638562e-01 1.02110648e+00 7.51976907e-01 -1.28545731e-01 -3.58048305e-02 -3.82679105e-01 1.12048733e+00 -4.64942425e-01 -4.08100903e-01 -1.04691908e-02 -4.12168384e-01 -1.27798486e+00 -8.43467563e-02 1.28889196e-02 1.48304373e-01 -1.18845575e-01 9.19110119e-01 1.64483383e-01 3.08419652e-02 2.35983878e-01 -1.33264959e+00 -5.67141652e-01 1.08776891e+00 -7.54112124e-01 2.69432753e-01 4.15015757e-01 -4.88956839e-01 1.61979091e+00 -6.03033245e-01 7.83827007e-01 1.61803162e+00 5.56727111e-01 5.21134079e-01 -1.18158865e+00 -3.83556098e-01 8.40354383e-01 3.86137128e-01 -8.31787109e-01 -2.91466027e-01 9.68393564e-01 -6.90216869e-02 8.21612358e-01 2.38593355e-01 3.72394592e-01 8.44658852e-01 -3.86403203e-01 1.03288627e+00 1.44078803e+00 -6.02560341e-01 2.06705451e-01 -5.82408309e-01 6.68040395e-01 6.32805586e-01 3.80797416e-01 -4.15369570e-01 -6.11975074e-01 -5.25575876e-01 6.05066359e-01 -3.69600385e-01 -1.75750345e-01 -2.39094317e-01 -1.10319114e+00 1.00888658e+00 4.69648689e-02 4.01745141e-01 7.69515783e-02 2.00871021e-01 4.82519239e-01 9.05653611e-02 5.93168497e-01 3.87316823e-01 -6.39791667e-01 -3.49895768e-02 -5.65064609e-01 4.82298791e-01 8.58107865e-01 9.46110606e-01 3.53123277e-01 -4.31496173e-01 -1.25463158e-01 9.63419437e-01 3.21521610e-01 -1.81954384e-01 3.22373718e-01 -1.14658201e+00 7.07901120e-01 8.55807483e-01 7.38091692e-02 -6.39358938e-01 -5.05501509e-01 -3.07360888e-01 -2.38044411e-01 -3.97978634e-01 5.69829762e-01 2.72423387e-01 -5.80846429e-01 1.91960573e+00 6.40834808e-01 1.75965309e-01 4.19454090e-02 7.04850972e-01 7.72500515e-01 5.02402425e-01 5.20819485e-01 -4.39024210e-01 1.66696966e+00 -1.41857040e+00 -9.16551232e-01 -4.65317249e-01 6.94643378e-01 -3.82742524e-01 1.29271579e+00 2.00387150e-01 -9.38492298e-01 -1.72816634e-01 -9.85253155e-01 -3.01636845e-01 -2.24701017e-01 -3.02931182e-02 1.00002146e+00 6.15794420e-01 -7.34284878e-01 6.12739801e-01 -5.67209721e-01 -6.69232458e-02 3.60882878e-01 -2.10709013e-02 -2.21990913e-01 -1.77149981e-01 -1.55469680e+00 6.85133517e-01 6.53142750e-01 -5.89387827e-02 -4.64117259e-01 -7.37909317e-01 -1.04587555e+00 2.98539475e-02 1.13090587e+00 -5.80689847e-01 1.43722332e+00 -4.47068930e-01 -1.32536232e+00 9.43847358e-01 -4.62934375e-01 -4.14134264e-01 4.62256700e-01 -5.02314746e-01 -1.86169982e-01 1.83479130e-01 5.10802448e-01 2.43084624e-01 4.25633609e-01 -1.39545274e+00 -7.69684792e-01 -3.71591419e-01 7.41924703e-01 4.27336991e-01 -2.26755545e-01 3.88309330e-01 -4.78529125e-01 -7.80095696e-01 3.97693992e-01 -9.12101328e-01 -3.43454748e-01 -2.23511577e-01 -4.62150425e-01 -9.83411193e-01 5.92218459e-01 -4.20529187e-01 1.64335942e+00 -1.97292519e+00 8.43510404e-02 -3.23210031e-01 1.84650496e-01 1.46478176e-01 6.14889152e-02 5.38655162e-01 -2.29230106e-01 5.21965325e-01 -5.84668040e-01 -4.11341727e-01 1.22572534e-01 6.93329692e-01 -5.11337698e-01 3.55296224e-01 2.06196666e-01 8.93696904e-01 -1.15258503e+00 -9.04187560e-01 -6.08186610e-02 -8.76588970e-02 -4.50240493e-01 1.00949556e-01 -5.20052433e-01 4.19295698e-01 -6.65160835e-01 7.57623911e-01 6.25988007e-01 -3.15879464e-01 7.77891874e-01 -3.79575901e-02 -1.69950590e-01 1.04548693e+00 -1.04344690e+00 1.65517545e+00 -2.84818143e-01 -6.79660290e-02 1.43227890e-01 -1.12470412e+00 5.31279445e-01 1.95701882e-01 5.46466827e-01 -3.72177899e-01 -1.90607101e-01 2.50339866e-01 1.45604551e-01 -2.63217211e-01 3.25925201e-01 -1.97768286e-01 -4.96340305e-01 3.71251136e-01 -3.46818328e-01 1.66079313e-01 4.38403815e-01 6.82350770e-02 1.04528737e+00 6.09431207e-01 7.13097990e-01 -5.64945042e-01 8.20568025e-01 2.96004526e-02 1.18430877e+00 5.80451608e-01 -3.18624228e-01 3.91332835e-01 1.24590194e+00 -2.43534848e-01 -7.25324631e-01 -9.04014528e-01 -2.31805369e-01 1.37780297e+00 3.70247811e-01 -6.94434941e-01 -8.03168654e-01 -1.41032052e+00 -8.95086899e-02 7.87069261e-01 -5.21169424e-01 2.16408476e-01 -1.29919171e+00 -9.80111420e-01 7.85384655e-01 8.71989489e-01 4.57668424e-01 -9.91533518e-01 -6.79322422e-01 5.25524020e-01 -5.90194881e-01 -1.61853588e+00 -2.07685962e-01 2.86321849e-01 -8.37182224e-01 -1.41958487e+00 -3.39207679e-01 -6.89090431e-01 5.31619012e-01 5.53796589e-02 1.20178926e+00 2.45086804e-01 1.18172273e-01 -1.08682252e-01 -8.08455467e-01 -2.40029395e-01 -7.83519149e-02 2.01999336e-01 -2.69493192e-01 -2.29909301e-01 1.03865176e-01 -6.66750073e-01 -4.98046219e-01 2.88618058e-01 -7.25919664e-01 -1.46593647e-02 3.47244352e-01 9.08930302e-01 7.69970477e-01 1.20262675e-01 7.10027218e-01 -1.60965800e+00 5.85951805e-01 -5.74731886e-01 -3.05175126e-01 4.28062379e-01 -8.04937959e-01 1.39063910e-01 6.61945224e-01 -1.37589365e-01 -1.34747744e+00 -2.84899399e-02 -2.51441121e-01 1.87035456e-01 5.04408069e-02 5.36322236e-01 -6.53914452e-01 3.40856761e-01 7.32656643e-02 -2.17324756e-02 -4.86753970e-01 -8.01841021e-01 4.52275217e-01 1.40750811e-01 3.69958401e-01 -1.17073727e+00 6.47247434e-01 5.95526993e-01 1.90325767e-01 -5.26204407e-01 -1.55052471e+00 -4.39126134e-01 -5.41194081e-01 2.91695863e-01 7.62969375e-01 -7.24356055e-01 -5.36912978e-01 1.97078288e-01 -1.35994375e+00 -2.79426485e-01 -2.51448900e-01 -1.93468690e-01 -3.92969579e-01 8.83415818e-01 -8.32386613e-01 -7.38178313e-01 -1.69584483e-01 -8.70469868e-01 1.08543956e+00 1.46667793e-01 -3.09754133e-01 -1.00393534e+00 -9.55418199e-02 6.24983668e-01 -5.72851487e-03 4.01461899e-01 1.54671979e+00 -8.44809890e-01 -2.07285568e-01 3.29224795e-01 -3.76838744e-01 1.56257287e-01 2.12569851e-02 -4.80388016e-01 -7.61112392e-01 1.57690316e-01 1.19040333e-01 -2.65466869e-01 8.52822959e-01 3.71608622e-02 1.44220281e+00 -3.65023911e-01 -5.15563309e-01 2.83950657e-01 1.19069505e+00 1.28381222e-01 5.57239592e-01 5.22371352e-01 6.55458927e-01 9.74075139e-01 1.15700746e+00 1.56397343e-01 7.02940464e-01 7.64045179e-01 5.63692689e-01 1.67441443e-01 -1.79496333e-01 -6.89628184e-01 1.16079733e-01 6.41278565e-01 -1.26738727e-01 -3.70136857e-01 -7.11760998e-01 6.51804328e-01 -2.32587361e+00 -5.42899132e-01 -4.93689626e-01 1.68423152e+00 9.67959523e-01 3.27655733e-01 1.37777016e-01 6.11333787e-01 7.10820913e-01 7.32350528e-01 -4.12548959e-01 -2.75641322e-01 -2.23238945e-01 2.41117716e-01 2.92930394e-01 3.92368972e-01 -1.27858758e+00 1.39620626e+00 6.28473949e+00 8.60234022e-01 -4.32916909e-01 4.63730842e-01 5.10694206e-01 3.58404249e-01 -5.48682272e-01 5.47703326e-01 -1.16605008e+00 4.24249351e-01 5.79626203e-01 -8.40612724e-02 1.33997634e-01 1.01467097e+00 -1.45579964e-01 -4.14055511e-02 -1.00719106e+00 6.01234198e-01 -1.04412250e-01 -1.12067831e+00 -4.62882146e-02 -1.85588580e-02 4.14855123e-01 -4.04836893e-01 -3.71801585e-01 3.46981704e-01 4.49654698e-01 -1.03210163e+00 9.09974992e-01 -2.16261312e-01 6.49622142e-01 -5.06453514e-01 5.21497130e-01 3.83827269e-01 -1.44472790e+00 -2.60926276e-01 -2.72582024e-01 -4.01445895e-01 4.98320967e-01 7.15406179e-01 -1.99750081e-01 8.37431848e-01 6.13095164e-01 7.83717752e-01 -4.20969278e-01 4.43627924e-01 -1.05198860e+00 1.04166353e+00 -1.00616559e-01 8.06933790e-02 2.36304075e-01 -7.84743354e-02 5.33813536e-01 9.96396780e-01 -1.29857421e-01 2.67023444e-01 4.37111229e-01 7.52346873e-01 -1.32635400e-01 6.74627796e-02 -2.27375627e-01 -6.22780435e-02 8.61920536e-01 1.10768723e+00 -1.00612271e+00 -1.90930620e-01 -3.47581863e-01 8.43835056e-01 4.34602022e-01 4.16787416e-02 -9.23642218e-01 -8.63315538e-02 6.25944436e-01 1.97172180e-01 1.72548994e-01 -2.24546075e-01 -4.99364018e-01 -1.11747575e+00 3.25886875e-01 -8.93076479e-01 9.31628585e-01 -3.28917533e-01 -1.23856640e+00 4.54410702e-01 4.67046708e-01 -7.82757223e-01 -1.54499143e-01 -7.35939085e-01 -2.71203488e-01 4.36325073e-01 -2.08468366e+00 -1.43214500e+00 1.52699694e-01 2.23918676e-01 8.22780013e-01 4.90376800e-01 7.07950771e-01 2.09273115e-01 -4.76265132e-01 4.79602337e-01 -4.56217080e-01 -3.48173380e-02 3.81528765e-01 -1.41828406e+00 5.19436896e-01 1.00002837e+00 4.96889129e-02 8.20497811e-01 5.02791226e-01 -6.35112882e-01 -1.12151241e+00 -9.64857638e-01 1.36318576e+00 -6.87659979e-01 8.31166208e-01 -3.54018748e-01 -9.18818831e-01 8.51146519e-01 -8.68115202e-02 2.44754493e-01 6.12391829e-01 3.52758765e-01 -7.24421620e-01 2.96180904e-01 -1.03155088e+00 5.38765907e-01 1.92520487e+00 -3.97913814e-01 -1.16415060e+00 1.60768107e-01 1.24664783e+00 -4.29440439e-01 -7.91614890e-01 6.73659563e-01 3.57694119e-01 -8.49180758e-01 1.05439138e+00 -5.69648445e-01 5.74728429e-01 -3.94006580e-01 -1.07440829e-01 -9.23914850e-01 -2.98086137e-01 -7.74788618e-01 -3.29640955e-01 1.54777002e+00 2.97328234e-01 -8.64485562e-01 7.22845316e-01 5.11192024e-01 -3.17349285e-01 -9.51941311e-01 -1.12424207e+00 -9.51307654e-01 2.50136226e-01 -4.43504006e-01 7.46144772e-01 9.63256538e-01 1.75693899e-01 6.58907175e-01 -2.34335884e-01 3.34886163e-02 6.67333961e-01 5.70969641e-01 1.92330554e-01 -1.46331227e+00 -4.56426263e-01 -4.37489837e-01 4.02533740e-01 -1.30524278e+00 6.96001649e-01 -8.97648275e-01 -3.18434015e-02 -1.73858905e+00 1.95796713e-01 -8.68888974e-01 -2.36813426e-01 9.98947501e-01 -4.71740365e-01 -6.12049028e-02 1.78504765e-01 3.94122213e-01 -8.04724932e-01 3.59293908e-01 1.16930342e+00 2.17467457e-01 4.61284928e-02 -8.45704749e-02 -1.17641902e+00 1.11463404e+00 6.33507073e-01 -8.19944978e-01 -5.74934065e-01 -4.20778692e-01 4.32175517e-01 -9.99107398e-03 1.11345120e-01 -4.20138538e-01 -1.35508806e-01 -3.81726831e-01 -3.45584095e-01 -3.82290244e-01 6.43054470e-02 -5.62567532e-01 -3.24367970e-01 3.69902402e-01 -4.83879149e-01 -2.05265358e-02 -1.10712782e-01 6.61238074e-01 -3.52219284e-01 -6.41487658e-01 4.89262611e-01 -2.37536758e-01 -8.99861276e-01 1.52649563e-02 -3.70571017e-02 4.81570899e-01 8.07195842e-01 8.32472593e-02 -6.25256717e-01 4.67682295e-02 -4.44501370e-01 2.89138138e-01 3.61304998e-01 4.57845896e-01 1.41218424e-01 -1.01636767e+00 -4.99650061e-01 -3.84835809e-01 1.79088429e-01 4.41785604e-01 4.84578423e-02 6.73284292e-01 -1.35148406e-01 5.17719865e-01 4.19405192e-01 4.49631223e-03 -1.12590575e+00 5.28996408e-01 -1.07887290e-01 -1.09354913e+00 -6.81729794e-01 7.13681221e-01 3.78975958e-01 -4.89343822e-01 6.31783828e-02 -2.46752396e-01 -6.48433506e-01 7.79025257e-02 2.80111581e-01 3.27024221e-01 -1.51801303e-01 -5.69941640e-01 -7.38039494e-01 4.63484049e-01 1.20644607e-01 4.83731702e-02 1.37404335e+00 -8.67396370e-02 -6.04783952e-01 2.30014265e-01 7.44847476e-01 3.30753922e-01 -1.02436483e+00 -3.89589816e-01 8.48722816e-01 -1.91694155e-01 -3.55122328e-01 -8.30632627e-01 -6.27577424e-01 7.08837509e-01 -2.81590402e-01 2.91910559e-01 1.00328588e+00 4.49672163e-01 9.67844844e-01 6.86014444e-02 4.89670455e-01 -1.11811531e+00 -5.75248301e-02 7.06804752e-01 7.89514780e-01 -6.78258419e-01 1.41905010e-01 -1.40898561e+00 -5.36101103e-01 9.21657979e-01 6.03098869e-01 -8.43568817e-02 4.06323105e-01 2.47888118e-01 -2.16740415e-01 -1.24001101e-01 -7.21930206e-01 -4.51713622e-01 -5.65878451e-02 4.42726403e-01 6.70125127e-01 -3.91802229e-02 -1.21508765e+00 1.15744162e+00 -7.62767494e-02 -2.34157309e-01 3.04712206e-01 1.22124481e+00 -2.09084049e-01 -1.71747267e+00 -1.27111793e-01 -6.90641925e-02 -9.62556660e-01 -3.44019026e-01 -4.62711602e-01 8.89645100e-01 2.81691641e-01 1.10271168e+00 -3.26224715e-01 8.48768577e-02 4.43192273e-01 3.65215659e-01 5.16532958e-01 -9.35702443e-01 -5.16175508e-01 1.58805549e-02 7.77896583e-01 -5.77332497e-01 -6.96824491e-01 -7.70268023e-01 -1.58732057e+00 3.34623545e-01 -1.77808359e-01 2.76762426e-01 2.91078269e-01 1.17994225e+00 1.53812274e-01 4.89155263e-01 3.96415025e-01 -2.58593380e-01 -8.03545237e-01 -7.45817780e-01 -3.37443769e-01 5.28681636e-01 -1.62273198e-01 -9.58059788e-01 -3.21362019e-01 3.94532047e-02]
[10.216400146484375, 9.264005661010742]
4c75d41e-d769-4255-a94d-9a889b4b6813
automatic-icd-coding-exploiting-discourse
null
null
https://aclanthology.org/2022.coling-1.254
https://aclanthology.org/2022.coling-1.254.pdf
Automatic ICD Coding Exploiting Discourse Structure and Reconciled Code Embeddings
The International Classification of Diseases (ICD) is the foundation of global health statistics and epidemiology. The ICD is designed to translate health conditions into alphanumeric codes. A number of approaches have been proposed for automatic ICD coding, since manual coding is labor-intensive and there is a global shortage of healthcare workers. However, existing studies did not exploit the discourse structure of clinical notes, which provides rich contextual information for code assignment. In this paper, we exploit the discourse structure by leveraging section type classification and section type embeddings. We also focus on the class-imbalanced problem and the heterogeneous writing style between clinical notes and ICD code definitions. The proposed reconciled embedding approach is able to tackle them simultaneously. Experimental results on the MIMIC dataset show that our model outperforms all previous state-of-the-art models by a large margin. The source code is available at https://github.com/discnet2022/discnet
['Wanchun Yang', 'Bo Sang', 'Fuxin Zhang', 'Bozheng Zhang', 'Shurui Zhang']
null
null
null
null
coling-2022-10
['medical-code-prediction', 'epidemiology']
['medical', 'medical']
[-5.79313701e-03 5.66281855e-01 -8.62481713e-01 -1.68320119e-01 -6.67649567e-01 -5.69007337e-01 2.25391746e-01 7.93357670e-01 -8.82774666e-02 5.91040373e-01 9.35311258e-01 -5.04394889e-01 -1.13282003e-01 -7.27376699e-01 -9.20944207e-04 -3.38632762e-01 9.66127515e-02 5.51527917e-01 -4.24978405e-01 1.63089439e-01 4.51639630e-02 -6.80654570e-02 -9.91128564e-01 6.03437722e-01 1.08239079e+00 8.82032692e-01 2.98089371e-03 4.47043538e-01 -3.92232955e-01 1.13793707e+00 -3.74196112e-01 -6.93353772e-01 -9.46630314e-02 -7.26159155e-01 -1.03537750e+00 -2.47049674e-01 -1.27674296e-01 -2.64099449e-01 -3.85915339e-01 1.05810595e+00 6.20866716e-01 -4.94591832e-01 9.21634197e-01 -1.12105894e+00 -1.14562440e+00 6.96008801e-01 -1.12473384e-01 3.80230658e-02 5.89741826e-01 -2.50774831e-01 1.20044613e+00 -9.18359697e-01 8.38523924e-01 8.96840990e-01 8.93150270e-01 7.04062760e-01 -8.81637812e-01 -5.55998862e-01 -1.17079571e-01 3.19518507e-01 -1.33121121e+00 -2.82700330e-01 8.62418950e-01 -1.05046797e+00 9.76960003e-01 3.75223994e-01 8.51177990e-01 1.28038299e+00 2.62377888e-01 4.25203085e-01 8.48774850e-01 -4.01311338e-01 1.50245652e-01 2.65669078e-01 2.80940562e-01 7.38839209e-01 6.01214051e-01 -3.06599170e-01 3.70455161e-03 -6.25008404e-01 5.92009902e-01 4.87610251e-01 -4.93513167e-01 -1.39644206e-01 -1.49746788e+00 9.90932822e-01 3.54561538e-01 4.73051459e-01 -4.33498681e-01 -2.48954996e-01 8.19336593e-01 1.93789780e-01 5.71933031e-01 5.71619153e-01 -4.41611081e-01 -4.06255662e-01 -7.16391206e-01 3.94930840e-02 1.04087400e+00 8.85829270e-01 2.21113101e-01 -3.83179456e-01 -2.29239717e-01 1.09755290e+00 4.33856577e-01 3.02619576e-01 6.52717769e-01 -8.80345821e-01 7.42414057e-01 1.20186543e+00 -1.61604211e-01 -1.38899875e+00 -4.88460988e-01 -1.94499835e-01 -1.36384511e+00 -5.75076938e-01 1.43685520e-01 -2.99299210e-01 -5.44923723e-01 1.46703815e+00 2.07523048e-01 -8.22809190e-02 7.25750476e-02 5.85241318e-01 1.15925038e+00 3.29138100e-01 2.27600168e-02 -3.22483033e-01 1.70261431e+00 -7.52961338e-01 -1.23803675e+00 9.24067572e-02 1.02358973e+00 -7.70333648e-01 6.04484439e-01 -1.99058000e-02 -8.25410128e-01 -2.12729350e-01 -7.17711687e-01 -2.08550945e-01 -2.90632814e-01 2.87880659e-01 3.58322471e-01 5.13084590e-01 -6.10207975e-01 5.74677661e-02 -7.63825595e-01 -4.26808834e-01 7.73323238e-01 -9.17367712e-02 -4.17959213e-01 -1.54531943e-02 -1.26345694e+00 8.07379901e-01 3.27726901e-01 -1.55477449e-01 -6.16926625e-02 -9.97633994e-01 -9.82908309e-01 3.02707106e-02 1.11819930e-01 -8.43184471e-01 8.52208495e-01 -7.14708805e-01 -1.00706053e+00 1.16518033e+00 -1.26817480e-01 -1.59895048e-01 4.50504333e-01 -2.35724915e-02 -5.21994114e-01 2.76966780e-01 5.46144098e-02 3.20774436e-01 1.74115703e-01 -9.32722390e-01 -2.19092742e-01 -2.44636834e-01 1.38244495e-01 -2.78445035e-02 -8.16639364e-01 1.31399408e-01 -4.60776202e-02 -1.10462379e+00 -4.18517292e-02 -9.08735991e-01 4.21718284e-02 2.58995891e-01 -4.63396639e-01 -3.92382264e-01 3.54371890e-02 -9.13028181e-01 2.07702065e+00 -2.12331605e+00 2.03030348e-01 -1.71780929e-01 5.93622029e-01 2.19814807e-01 3.76633763e-01 8.76683116e-01 -2.23245472e-01 5.28453588e-01 -5.29779255e-01 -2.71027923e-01 1.37032330e-01 3.56588870e-01 -3.46884519e-01 4.79245961e-01 -2.61154529e-02 8.71095836e-01 -1.01145756e+00 -8.49126995e-01 -4.42434996e-02 5.43524325e-01 -9.31583047e-01 3.33706528e-01 1.80490211e-01 2.49764413e-01 -3.87102544e-01 6.64413989e-01 6.43920481e-01 -6.71119034e-01 6.61951602e-01 -1.31423935e-01 3.14034894e-02 6.30419135e-01 -8.57750356e-01 1.64495814e+00 -2.88775325e-01 4.38871533e-01 -3.34677011e-01 -1.13591492e+00 5.93873024e-01 7.65827894e-01 8.27334344e-01 -1.84396878e-01 1.35085970e-01 1.99170753e-01 1.09169453e-01 -1.02530241e+00 4.80055399e-02 -4.12403159e-02 -2.76600063e-01 3.69323581e-01 -1.54382095e-01 1.51843846e-01 3.29720229e-03 1.25691414e-01 1.30500495e+00 -5.04960835e-01 9.39272523e-01 -3.01899225e-01 3.64187598e-01 5.40908091e-02 9.98348832e-01 4.13605899e-01 -3.06131780e-01 9.21570599e-01 7.84299850e-01 -6.69707179e-01 -1.07794082e+00 -9.53404129e-01 -6.50981724e-01 3.79926115e-01 -1.33772880e-01 -7.10474372e-01 -7.12658167e-01 -7.40934014e-01 1.60581052e-01 2.79774904e-01 -9.46082413e-01 -7.53331790e-03 -4.87906575e-01 -7.11992443e-01 7.74471879e-01 5.60714781e-01 3.22626919e-01 -1.05157709e+00 -6.29529834e-01 2.99734890e-01 -7.04407871e-01 -8.74041200e-01 -4.13280070e-01 -2.17722803e-01 -7.24087954e-01 -1.49443316e+00 -7.21377373e-01 -9.80456591e-01 8.03366601e-01 -3.03595304e-01 1.25683320e+00 3.44039828e-01 -3.62123460e-01 2.93693632e-01 -7.96728432e-01 -4.71185386e-01 -6.59935355e-01 3.69863302e-01 -6.62614685e-03 -9.86105353e-02 5.44264019e-01 -3.49237800e-01 -8.69887590e-01 -1.44688532e-01 -1.01117277e+00 3.65399182e-01 4.12785321e-01 9.67755139e-01 2.42333025e-01 -4.03978437e-01 7.46485233e-01 -1.17379797e+00 6.71888649e-01 -9.11380291e-01 6.54092580e-02 1.72006860e-01 -7.32844830e-01 -2.18593776e-01 6.13133430e-01 -1.72083452e-01 -7.65447259e-01 -3.78117979e-01 -4.17993903e-01 1.92809012e-02 -2.17855766e-01 9.00796592e-01 1.53443575e-01 6.14460289e-01 5.41081965e-01 -3.47832814e-02 -8.52072462e-02 -5.67098260e-01 -8.66810083e-02 1.25489044e+00 1.07903508e-02 -1.47567496e-01 3.51810783e-01 4.29757476e-01 -4.92727906e-01 -5.02553523e-01 -8.40157330e-01 -5.38803399e-01 -5.31317592e-01 1.11002341e-01 1.09983385e+00 -9.29572165e-01 -3.60787809e-01 1.26767367e-01 -1.52117133e+00 -7.09809363e-02 -1.97756588e-01 5.02139211e-01 -2.81491935e-01 3.36571068e-01 -7.27555811e-01 -3.88624728e-01 -3.48891497e-01 -8.40512633e-01 6.88531518e-01 -3.03911030e-01 -8.20533335e-01 -1.31483626e+00 5.38933098e-01 3.33752036e-01 2.68441021e-01 5.80282629e-01 1.37168622e+00 -7.61336625e-01 9.38773826e-02 -2.72568110e-02 -3.03883165e-01 1.88679934e-01 7.49839485e-01 -1.92204922e-01 -6.97350144e-01 -2.68390309e-02 6.48020431e-02 -1.28711268e-01 8.25458229e-01 2.49815941e-01 1.37240505e+00 -9.00049686e-01 -4.12334114e-01 6.42280638e-01 1.41663826e+00 2.34241351e-01 5.35042048e-01 1.83997795e-01 9.30468678e-01 4.96715695e-01 1.86943218e-01 8.98524106e-01 8.73180628e-01 6.83526218e-01 1.79094061e-01 -6.66874275e-02 -8.66614431e-02 3.71528082e-02 -5.29092401e-02 1.45416498e+00 2.41357535e-02 -2.16466501e-01 -1.65692592e+00 9.43483531e-01 -1.98525500e+00 -9.77430582e-01 -1.18602522e-01 1.62975049e+00 1.40553808e+00 -2.59162217e-01 -1.53503045e-01 3.30877334e-01 4.12836999e-01 6.60748109e-02 -2.48914629e-01 -5.06719828e-01 1.98004991e-01 -1.99951027e-02 1.72012970e-01 3.08104634e-01 -1.05152738e+00 1.91094145e-01 5.85278749e+00 4.23612863e-01 -7.58163929e-01 3.65592510e-01 4.86269176e-01 1.77207991e-01 -4.83737320e-01 -4.26126689e-01 -3.53604108e-01 8.78485084e-01 7.37647593e-01 -2.84926385e-01 -1.87354069e-02 5.74943483e-01 -8.47278535e-02 3.96895230e-01 -1.10789216e+00 1.13352132e+00 1.84346139e-01 -1.57927012e+00 -3.24022919e-02 1.44370705e-01 8.76028836e-01 -1.22010395e-01 -6.66296284e-04 9.46276709e-02 7.46855810e-02 -9.16300535e-01 3.30018967e-01 4.71974790e-01 1.00162697e+00 -3.62670332e-01 1.08635068e+00 3.74505669e-02 -1.11934125e+00 -3.34986210e-01 -1.82693437e-01 -1.54849485e-01 -1.71576962e-02 9.58502650e-01 -6.13071501e-01 7.02834845e-01 6.36266470e-01 1.12220466e+00 -4.55164611e-01 9.22467470e-01 -9.62585956e-02 7.11483955e-01 2.64417320e-01 2.22595915e-01 -1.66831285e-01 -3.10717225e-02 3.06912571e-01 1.54994142e+00 4.84848827e-01 2.45748952e-01 2.16931105e-01 7.62504458e-01 -3.51196051e-01 2.79484242e-01 -8.35720778e-01 -2.34373197e-01 6.20085776e-01 9.01261568e-01 -5.16507268e-01 -5.86551785e-01 -6.25305235e-01 8.02755773e-01 2.21605897e-01 1.24104984e-01 -9.55401659e-01 -4.29321200e-01 8.72082770e-01 1.68896094e-01 1.44580960e-01 1.16110817e-01 -5.66966176e-01 -1.57671380e+00 1.67349666e-01 -9.52091217e-01 6.99081838e-01 -2.57399261e-01 -1.33442628e+00 5.88176131e-01 -5.35051897e-02 -1.57561302e+00 -2.83589512e-01 -4.72200334e-01 -2.53349170e-02 3.80622655e-01 -1.45600438e+00 -9.74257648e-01 -2.74133861e-01 2.43726298e-01 4.29050148e-01 -1.79590195e-01 1.34586227e+00 7.01885104e-01 -5.13840199e-01 8.03259194e-01 3.55228275e-01 8.37343216e-01 7.45200455e-01 -1.28626084e+00 3.34242061e-02 3.40466678e-01 -2.41707534e-01 8.12106609e-01 3.51029634e-01 -7.07278490e-01 -1.11535943e+00 -1.33037150e+00 1.58043504e+00 -6.37006521e-01 6.15674794e-01 -3.59181702e-01 -1.05719650e+00 5.98312438e-01 4.10165846e-01 1.90887805e-02 1.43472302e+00 1.25925452e-03 -5.23972690e-01 1.35950834e-01 -1.13914049e+00 4.34352726e-01 1.13582683e+00 -6.42873108e-01 -9.70366657e-01 4.64559436e-01 6.45183146e-01 -3.83820504e-01 -1.33278942e+00 2.50347942e-01 7.19527006e-01 -6.74036324e-01 9.39858615e-01 -5.02430260e-01 1.08544600e+00 -1.61070079e-02 -1.10667303e-01 -1.12837696e+00 -5.27150095e-01 -2.36159876e-01 -2.35077977e-01 1.25698650e+00 3.25067610e-01 -8.56286585e-01 2.77293146e-01 6.15301073e-01 -7.90306926e-02 -1.06292415e+00 -1.05458248e+00 -5.80470324e-01 2.49687240e-01 -1.54469877e-01 5.62871397e-01 1.66988385e+00 6.08460844e-01 -1.65941432e-01 -2.35215873e-01 -6.74636588e-02 3.02110344e-01 2.40179077e-01 2.11339086e-01 -1.33776069e+00 -2.54358232e-01 -5.70960402e-01 -4.62628394e-01 -4.70047325e-01 1.52947426e-01 -1.28436625e+00 -3.87020290e-01 -1.97883391e+00 5.28321028e-01 -4.53554124e-01 -5.24039626e-01 6.89304531e-01 -2.92669058e-01 1.93565249e-01 2.00582847e-01 3.86097878e-01 -4.81432736e-01 2.34444231e-01 8.32058728e-01 -3.66302103e-01 -1.35598198e-01 -5.22714555e-01 -9.81272101e-01 7.81165183e-01 1.13155878e+00 -9.55949903e-01 -2.14207828e-01 -5.78074038e-01 6.75100446e-01 2.67928056e-02 1.89080358e-01 -7.68188715e-01 1.49433360e-01 -2.04941198e-01 3.50844413e-02 -4.86905612e-02 -6.50362670e-02 -8.94635439e-01 3.75151068e-01 7.37380087e-01 -6.87931538e-01 3.39812428e-01 4.19793017e-02 3.35796177e-01 -3.52822036e-01 -1.83933496e-01 5.40865958e-01 -2.89794654e-02 -1.24224983e-02 1.76028639e-01 -6.39233351e-01 5.85596442e-01 7.85975754e-01 2.02946886e-02 -3.63894343e-01 -1.18721034e-02 -5.20794749e-01 7.13566989e-02 4.70311821e-01 5.62229037e-01 6.78020716e-01 -1.74734604e+00 -8.66086483e-01 2.06285521e-01 5.19590318e-01 -1.82665646e-01 2.26773545e-01 1.14829051e+00 -7.09189534e-01 5.05630314e-01 -9.89397168e-02 -2.89812088e-01 -1.27288973e+00 5.90756893e-01 9.68342349e-02 -5.57369113e-01 -7.54240990e-01 5.03028512e-01 1.92808181e-01 -5.70396721e-01 1.59877300e-01 -7.27360964e-01 -4.81021821e-01 5.31735778e-01 6.32463813e-01 3.74841064e-01 -1.53161228e-01 -5.70803463e-01 -6.14861190e-01 6.62154675e-01 -2.09049159e-03 2.80121416e-01 1.47150028e+00 -7.76935071e-02 -4.15227443e-01 5.17093837e-01 1.59851277e+00 -9.41354558e-02 -2.99047947e-01 -1.93202198e-01 -4.63855155e-02 -3.99642944e-01 -1.85150102e-01 -7.04244733e-01 -8.78868759e-01 8.71061206e-01 4.59740758e-01 4.38546628e-01 1.05006814e+00 -2.55251899e-02 7.91976213e-01 1.44477516e-01 -5.23956753e-02 -1.03021514e+00 -8.19232538e-02 3.83662432e-01 1.02403462e+00 -1.34173107e+00 -1.41184658e-01 -3.70479316e-01 -6.44666016e-01 1.07098353e+00 1.17813513e-01 6.96379468e-02 9.95194137e-01 3.02798569e-01 3.42977077e-01 -1.84759811e-01 -7.76083589e-01 1.85584679e-01 2.67259359e-01 5.03179431e-01 8.62803519e-01 3.20824295e-01 -6.66205049e-01 9.67950284e-01 -1.52917668e-01 2.22942337e-01 5.59440434e-01 8.91932786e-01 3.34595516e-02 -1.24758041e+00 -2.52467811e-01 7.76654124e-01 -8.16985786e-01 -2.63182968e-01 -3.70984614e-01 4.39119399e-01 4.41822916e-01 8.66089821e-01 6.76460639e-02 -2.59324282e-01 1.67266294e-01 2.28458628e-01 1.12127475e-01 -7.87511945e-01 -7.35529184e-01 -4.12513822e-01 4.69232425e-02 -3.23707193e-01 -5.29674053e-01 -6.57800019e-01 -1.27799404e+00 -3.42442930e-01 1.31110981e-01 3.95652279e-02 1.44270733e-01 5.09785056e-01 8.65292549e-01 7.79216826e-01 3.45622331e-01 -1.76699817e-01 -2.37495288e-01 -8.95640492e-01 -2.45458812e-01 6.88444138e-01 6.75082207e-01 -6.05058610e-01 -3.54637712e-01 3.88120681e-01]
[8.014861106872559, 6.927388668060303]
8ff8381e-7423-491c-9e26-f09a23e775e1
prediction-of-the-motion-of-chest-internal
2207.05951
null
https://arxiv.org/abs/2207.05951v1
https://arxiv.org/pdf/2207.05951v1.pdf
Prediction of the motion of chest internal points using a recurrent neural network trained with real-time recurrent learning for latency compensation in lung cancer radiotherapy
During the radiotherapy treatment of patients with lung cancer, the radiation delivered to healthy tissue around the tumor needs to be minimized, which is difficult because of respiratory motion and the latency of linear accelerator systems. In the proposed study, we first use the Lucas-Kanade pyramidal optical flow algorithm to perform deformable image registration of chest computed tomography scan images of four patients with lung cancer. We then track three internal points close to the lung tumor based on the previously computed deformation field and predict their position with a recurrent neural network (RNN) trained using real-time recurrent learning (RTRL) and gradient clipping. The breathing data is quite regular, sampled at approximately 2.5Hz, and includes artificial drift in the spine direction. The amplitude of the motion of the tracked points ranged from 12.0mm to 22.7mm. Finally, we propose a simple method for recovering and predicting 3D tumor images from the tracked points and the initial tumor image based on a linear correspondence model and Nadaraya-Watson non-linear regression. The root-mean-square error, maximum error, and jitter corresponding to the RNN prediction on the test set were smaller than the same performance measures obtained with linear prediction and least mean squares (LMS). In particular, the maximum prediction error associated with the RNN, equal to 1.51mm, is respectively 16.1% and 5.0% lower than the maximum error associated with linear prediction and LMS. The average prediction time per time step with RTRL is equal to 119ms, which is less than the 400ms marker position sampling time. The tumor position in the predicted images appears visually correct, which is confirmed by the high mean cross-correlation between the original and predicted images, equal to 0.955.
['Ritu Bhusal Chhatkuli', 'Kazuyuki Demachi', 'Mitsuru Uesaka', 'Michel Pohl']
2022-07-13
null
null
null
null
['respiratory-motion-forecasting', 'time-series-prediction']
['medical', 'time-series']
[ 7.85085037e-02 2.90742129e-01 -2.25951940e-01 2.50981927e-01 -8.10188174e-01 -2.68267483e-01 2.45150924e-01 -1.15814403e-01 -6.71382308e-01 7.09615648e-01 8.95542055e-02 -1.04054391e-01 -3.10389131e-01 -3.78762960e-01 -4.53189671e-01 -1.20591247e+00 -3.65414955e-02 5.02450705e-01 5.42549253e-01 1.74828485e-01 1.22468702e-01 8.35528433e-01 -7.79430628e-01 5.01226485e-02 6.07631505e-01 8.59216452e-01 5.55837214e-01 8.75419259e-01 1.35185540e-01 9.46635425e-01 -3.51686686e-01 1.35886073e-01 -1.40627235e-01 -5.07212102e-01 -7.09262967e-01 -7.16811121e-02 4.25724871e-02 -2.66996354e-01 -5.84592998e-01 7.00850904e-01 4.96766835e-01 6.95282742e-02 5.99088848e-01 -7.48292983e-01 6.03889525e-02 1.16454028e-02 -5.59516191e-01 2.40108684e-01 7.71055147e-02 1.48036266e-02 1.65302292e-01 -7.37753570e-01 8.59834909e-01 3.20593268e-01 1.11289394e+00 8.32879186e-01 -7.64223337e-01 -2.29487196e-01 -5.27444482e-01 2.73887534e-02 -1.31753409e+00 -1.53393865e-01 4.69996631e-01 -5.55798292e-01 6.69997334e-01 5.90319693e-01 8.22174132e-01 7.27901280e-01 1.20822167e+00 -6.39746562e-02 6.23369157e-01 -4.08585817e-01 -1.39322318e-02 2.21875921e-01 -1.56457663e-01 9.22310770e-01 1.49112046e-01 2.56775230e-01 7.79730603e-02 -1.51656657e-01 1.09506202e+00 4.05158371e-01 -8.25543880e-01 -1.66241631e-01 -1.43887246e+00 5.57638228e-01 8.09957504e-01 8.75177979e-01 -4.39729661e-01 3.18643093e-01 2.95043975e-01 -5.02270997e-01 1.13726720e-01 -7.19526932e-02 -4.90357541e-02 -2.04941228e-01 -8.17462683e-01 -3.99677336e-01 5.88186204e-01 6.69183969e-01 8.83703977e-02 6.52169883e-02 -2.16021582e-01 1.91474959e-01 6.08307540e-01 5.82465827e-01 1.27005613e+00 -7.89547205e-01 9.07188952e-02 2.76792556e-01 7.39258304e-02 -8.28811646e-01 -8.46760869e-01 -5.21758437e-01 -1.22476387e+00 4.33854461e-02 6.33348346e-01 -2.66189757e-03 -6.60646915e-01 1.18453908e+00 4.42996413e-01 4.05573636e-01 1.45312563e-01 9.41710234e-01 8.43947828e-01 4.66415316e-01 -2.94165999e-01 -9.65059400e-01 1.05999768e+00 -1.03684533e+00 -9.45056438e-01 1.31201610e-01 1.17481816e+00 -8.14111292e-01 7.27653801e-01 -1.55905753e-01 -1.20225835e+00 -4.33288842e-01 -5.82464278e-01 3.90186071e-01 5.94663918e-01 2.10301206e-01 -8.81087407e-02 3.54246110e-01 -1.02213621e+00 9.93411124e-01 -1.35429931e+00 -2.16291517e-01 3.38958018e-02 4.55622107e-01 -3.22259784e-01 2.04452112e-01 -7.88726807e-01 9.39052820e-01 2.08303239e-03 4.41348195e-01 -5.09933412e-01 -9.10417676e-01 -2.86585391e-01 -2.50053734e-01 -5.15716709e-02 -6.61590159e-01 1.24262059e+00 -7.29351938e-01 -1.73074508e+00 6.15760148e-01 -4.00894135e-01 -3.08456630e-01 7.23073184e-01 1.24812439e-01 -2.26351559e-01 3.13774586e-01 -6.18901812e-02 1.56442329e-01 4.57089514e-01 -9.40798461e-01 -1.67806104e-01 -1.94507807e-01 -9.20726418e-01 3.03581208e-01 -1.13541767e-01 -2.18277708e-01 -4.12582725e-01 -3.99380207e-01 6.06649756e-01 -1.41583395e+00 -5.36818326e-01 2.98998475e-01 -2.49980390e-01 2.68697441e-01 8.46357286e-01 -7.17906654e-01 1.11412144e+00 -2.10253811e+00 -1.34208173e-01 2.06914425e-01 6.48855567e-02 -9.86273750e-04 4.80148584e-01 -9.57714468e-02 -3.34821790e-01 -2.24265665e-01 -2.89753735e-01 -1.81146748e-02 -8.61114740e-01 2.10957319e-01 1.11619510e-01 8.97316515e-01 -7.31255710e-01 1.09937060e+00 -8.31150353e-01 -6.17011666e-01 5.81063144e-02 6.62128270e-01 -9.82833952e-02 1.19856641e-01 3.74941707e-01 8.70931447e-01 -3.80739689e-01 2.07553118e-01 4.61104572e-01 -2.73488581e-01 3.20748636e-03 -3.63723576e-01 -3.74562442e-01 -3.99455316e-02 -8.58255923e-01 1.62202501e+00 -3.52922708e-01 6.01204693e-01 -2.64399558e-01 -4.25107121e-01 7.75541604e-01 8.05393338e-01 1.34075761e+00 -6.14828348e-01 1.93928018e-01 3.78378153e-01 2.29322147e-02 -6.88338459e-01 1.07119791e-01 -2.77947247e-01 4.70137864e-01 4.29722488e-01 -5.41251302e-01 -2.87396997e-01 -3.87397885e-01 -7.06253052e-02 1.16232181e+00 -3.85324880e-02 3.00413549e-01 -1.50681928e-01 8.36812437e-01 1.07898943e-01 5.59008479e-01 3.67698759e-01 -3.13918740e-01 8.26481998e-01 1.56101137e-01 -3.29483032e-01 -8.66031468e-01 -7.55362034e-01 -3.97466302e-01 -1.00505508e-01 2.75781900e-01 -1.39327556e-01 -6.37933969e-01 -6.64400637e-01 -3.83952916e-01 4.27527875e-01 -2.92815596e-01 -1.53015941e-01 -1.07837856e+00 -5.85936964e-01 2.87076950e-01 5.01419604e-01 2.64950871e-01 -1.10108423e+00 -9.01718497e-01 3.54356468e-01 -3.78291547e-01 -1.23852885e+00 -5.62854469e-01 -5.78016276e-03 -1.42292416e+00 -1.12697363e+00 -8.42839599e-01 -7.74849474e-01 9.93007064e-01 2.63818830e-01 7.89531231e-01 2.24649206e-01 -4.87599552e-01 4.06786919e-01 4.17430438e-02 1.33571789e-01 -7.46338904e-01 -2.68004268e-01 -4.11522724e-02 -4.31157023e-01 -3.73504758e-01 -1.79628134e-01 -6.07743144e-01 6.49754286e-01 -5.98340154e-01 3.48447673e-02 4.32243645e-01 9.69757438e-01 1.04154825e+00 1.80882454e-01 -1.57620385e-01 -4.90554035e-01 1.17100552e-01 -1.59583807e-01 -6.24457240e-01 4.81389016e-02 -4.52919573e-01 -1.50749907e-01 6.88666940e-01 -5.97299933e-01 -8.62231731e-01 4.77326512e-01 1.19975423e-02 -6.23430729e-01 7.91848004e-02 1.53239295e-01 6.47659838e-01 -5.52706599e-01 8.20060194e-01 1.84669897e-01 2.90405810e-01 2.80793101e-01 -1.44344836e-01 3.42041969e-01 6.28165960e-01 1.83305621e-01 8.48517895e-01 6.53432131e-01 5.19267261e-01 -7.34955788e-01 -5.75512528e-01 -3.42681974e-01 -8.63377571e-01 -5.61922908e-01 9.85910833e-01 -5.93104005e-01 -1.07555485e+00 4.58020896e-01 -1.23927641e+00 -4.40989822e-01 -6.79588556e-01 1.25895047e+00 -7.14228213e-01 5.83133459e-01 -8.59940410e-01 -5.03634512e-01 -7.95352519e-01 -1.23040378e+00 6.76225185e-01 3.09248418e-01 -1.85297593e-01 -1.20078373e+00 1.46968529e-01 3.81568894e-02 4.90194947e-01 3.69135082e-01 5.28150737e-01 -2.40498781e-01 -5.52940130e-01 -4.49701160e-01 1.49678811e-01 -1.42088115e-01 3.58996928e-01 1.69339523e-01 -7.24944234e-01 -5.28396249e-01 8.75063241e-01 3.62233162e-01 1.74750686e-01 9.71712232e-01 1.11531556e+00 -4.50456664e-02 -7.95576453e-01 5.59538186e-01 1.47499466e+00 5.08976161e-01 5.83750546e-01 2.16376036e-01 8.34891200e-01 2.51961172e-01 6.64497077e-01 3.43104154e-01 -1.20445807e-02 8.61530662e-01 6.53969347e-01 1.10699628e-02 -1.88400477e-01 1.95299804e-01 3.06298822e-01 1.14415777e+00 -2.01077342e-01 -2.89086811e-02 -1.02032030e+00 7.05649704e-02 -1.68151379e+00 -8.89169216e-01 -8.21197450e-01 2.52822113e+00 5.40410459e-01 2.40114897e-01 -3.90418291e-01 6.56719282e-02 6.53543293e-01 -1.81191340e-01 -3.06565136e-01 -2.43262514e-01 4.50829953e-01 -8.82580504e-02 8.64340067e-01 6.99921727e-01 -7.13144243e-01 1.86829060e-01 5.79325247e+00 5.24677038e-01 -1.43792927e+00 1.58455268e-01 2.87758648e-01 -1.25561915e-02 1.89146008e-02 -8.53684396e-02 -7.99873233e-01 6.27526045e-01 1.26099944e+00 -2.16209501e-01 1.99540675e-01 5.23569643e-01 5.23971796e-01 -4.16610122e-01 -8.45482945e-01 9.31938708e-01 8.22514147e-02 -1.18017590e+00 -6.84425831e-01 3.07755291e-01 6.72854364e-01 1.81418583e-01 -1.66202977e-01 -3.63530517e-01 -3.30319673e-01 -8.24833870e-01 1.76330537e-01 1.07259965e+00 8.13790858e-01 -4.94599849e-01 9.24314022e-01 8.95075977e-01 -1.16517031e+00 9.25960988e-02 -3.48643929e-01 6.26205444e-01 2.30159178e-01 6.13353193e-01 -1.29709208e+00 6.83591783e-01 4.86637175e-01 6.34588718e-01 -1.25706300e-01 1.11944568e+00 -1.40820622e-01 4.02851820e-01 -5.26907802e-01 -1.57535938e-03 -6.89542443e-02 -2.80491829e-01 5.73006630e-01 7.18496799e-01 7.60957658e-01 9.47306082e-02 -9.96646956e-02 2.96449900e-01 1.80677801e-01 3.68287303e-02 -4.35271829e-01 5.42170465e-01 2.03507021e-01 1.40916669e+00 -7.65634954e-01 -1.27709110e-03 -2.01046497e-01 9.70044851e-01 -2.01093599e-01 -6.24743402e-02 -1.27953935e+00 -6.95567951e-02 -1.74311712e-01 4.06538635e-01 -1.64233875e-02 -2.29601264e-01 -1.39070317e-01 -8.01984072e-01 9.87121090e-02 -8.57128724e-02 2.21976891e-01 -8.70606005e-01 -6.42473221e-01 9.20849144e-01 -3.43088657e-01 -1.65605187e+00 -5.96378326e-01 -2.44525805e-01 -9.23439324e-01 7.73822725e-01 -1.15806115e+00 -4.79819179e-01 -8.23109746e-01 6.42209470e-01 2.06344455e-01 1.81453764e-01 1.17656016e+00 5.76847009e-02 -5.11074841e-01 4.34075475e-01 4.05492306e-01 2.19146479e-02 6.03962362e-01 -8.70158434e-01 -8.72664750e-02 5.21775603e-01 -2.24982068e-01 2.37079393e-02 3.81773710e-01 -6.97121143e-01 -1.39361012e+00 -1.03934836e+00 9.25281882e-01 -4.30505216e-01 5.37863135e-01 2.50204295e-01 -1.13760328e+00 6.39586389e-01 -7.04207364e-03 8.45727682e-01 2.25431055e-01 -1.09843099e+00 7.19412327e-01 -3.61666456e-02 -1.34217334e+00 3.58804703e-01 5.28571546e-01 -3.25071782e-01 -7.21174181e-02 6.91573441e-01 4.77240652e-01 -1.09237313e+00 -1.16717184e+00 5.87786555e-01 4.18385565e-01 -7.44364738e-01 8.88492525e-01 1.48504227e-01 3.68339219e-03 -3.37838084e-01 3.34994823e-01 -1.02831423e+00 -5.10903478e-01 -4.71034139e-01 3.38289887e-02 6.42117560e-01 2.17786402e-01 -6.10752881e-01 1.32098508e+00 6.10380054e-01 -2.64688164e-01 -1.06151259e+00 -1.40437853e+00 -5.49122036e-01 -6.05593435e-02 -1.33817941e-01 -3.55965108e-01 6.53031945e-01 -1.67503759e-01 -2.12715492e-01 -1.96511596e-02 3.25566381e-01 4.86026883e-01 -1.72335178e-01 2.66684085e-01 -9.55106914e-01 -2.43082106e-01 -7.97821060e-02 -4.21178550e-01 -9.50185776e-01 -7.79453665e-02 -1.01336896e+00 1.47816539e-01 -1.55193055e+00 5.38977347e-02 -3.84046793e-01 -2.11088642e-01 1.23327479e-01 1.60411343e-01 2.82046217e-02 -2.93143988e-02 6.22283995e-01 1.05025396e-01 4.15085703e-01 1.78899908e+00 1.78528190e-01 -3.41619045e-01 6.10079587e-01 2.71371692e-01 9.04253721e-01 6.78595960e-01 -8.60286772e-01 -2.34885409e-01 -9.54818279e-02 -2.57500499e-01 9.00531948e-01 2.40874648e-01 -1.25016129e+00 6.91975415e-01 1.88373448e-03 5.73688865e-01 -8.07951927e-01 3.00791949e-01 -1.35712790e+00 6.50991619e-01 1.27410698e+00 3.04226000e-02 2.18815893e-01 1.12934284e-01 4.23653662e-01 -1.20707296e-01 -6.67025089e-01 9.52065945e-01 -1.07684195e-01 -6.62762001e-02 4.99222994e-01 -5.27548492e-01 -2.33575612e-01 1.43181336e+00 -4.63141084e-01 -8.30553547e-02 -1.52059987e-01 -8.94371867e-01 -3.23416740e-01 1.89644203e-01 -2.23056346e-01 8.34223211e-01 -1.48759592e+00 -3.96478504e-01 2.82995254e-01 -3.98233116e-01 2.29690596e-01 3.92181218e-01 1.64367568e+00 -8.20300400e-01 4.74936545e-01 2.69156247e-01 -1.07099605e+00 -1.40852273e+00 3.67293149e-01 1.06496382e+00 -6.34029627e-01 -9.44109857e-01 5.84239364e-01 -7.15278760e-02 -5.47935553e-02 2.91810594e-02 -4.98272270e-01 -3.20736945e-01 -3.72009069e-01 1.85869440e-01 4.82465655e-01 2.98742741e-01 -8.34776819e-01 -4.50141013e-01 1.22096789e+00 5.43410443e-02 1.60270870e-01 9.54738140e-01 2.58935094e-02 -4.57648374e-02 2.28591025e-01 1.30284917e+00 1.15922228e-01 -1.22030210e+00 -1.08154647e-01 -7.25204067e-04 -3.51855189e-01 1.08331233e-01 -4.25418198e-01 -1.12486279e+00 5.92146993e-01 1.03007162e+00 1.03638865e-01 1.04219306e+00 -1.75173625e-01 7.94326246e-01 -3.32220010e-02 2.45926175e-02 -5.69040895e-01 2.27753833e-01 5.02353609e-01 7.85804808e-01 -8.81547213e-01 1.87642828e-01 -4.88454938e-01 -5.99804997e-01 1.65454054e+00 5.67394555e-01 -1.46231115e-01 6.83426440e-01 4.12783414e-01 1.20244600e-01 1.14050157e-01 -6.45868838e-01 5.25341809e-01 2.96110541e-01 3.31576988e-02 5.82146168e-01 -2.24282950e-01 -2.80898333e-01 8.09013695e-02 4.03355695e-02 1.98358521e-01 6.86440170e-01 9.64038849e-01 -4.53898340e-01 -6.50600493e-01 -6.34480298e-01 6.53705671e-02 -4.17816341e-01 3.30956280e-01 4.70457107e-01 8.68454814e-01 -2.74899840e-01 4.52213526e-01 2.20641404e-01 -3.37910578e-02 5.68720281e-01 -2.55756557e-01 4.94897604e-01 -1.97109967e-01 -6.57441437e-01 4.33253527e-01 -5.25631726e-01 -7.14939535e-01 -4.99640256e-01 -5.25700092e-01 -2.00437689e+00 -6.64654449e-02 -4.79331583e-01 2.47722685e-01 7.82323062e-01 7.33308256e-01 -7.76785836e-02 7.44952261e-01 9.43393409e-01 -6.88302517e-01 -3.14136177e-01 -6.34683013e-01 -3.96046847e-01 2.91662421e-02 2.99991667e-01 -1.59357652e-01 -7.48220444e-01 -1.93706572e-01]
[13.802556991577148, -2.603996992111206]
d3fe67ab-23d8-4ea1-b373-8728e4f9bbd0
fingers-angle-calculation-using-level-set
1406.3418
null
http://arxiv.org/abs/1406.3418v1
http://arxiv.org/pdf/1406.3418v1.pdf
Fingers' Angle Calculation using Level-Set Method
In the current age, use of natural communication in human computer interaction is a known and well installed thought. Hand gesture recognition and gesture based applications has gained a significant amount of popularity amongst people all over the world. It has a number of applications ranging from security to entertainment. These applications generally are real time applications and need fast, accurate communication with machines. On the other end, gesture based communications have few limitations also like bent finger information is not provided in vision based techniques. In this paper, a novel method for fingertip detection and for angle calculation of both hands bent fingers is discussed. Angle calculation has been done before with sensor based gloves/devices. This study has been conducted in the context of natural computing for calculating angles without using any wired equipment, colors, marker or any device. The pre-processing and segmentation of the region of interest is performed in a HSV color space and a binary format respectively. Fingertips are detected using level-set method and angles were calculated using geometrical analysis. This technique requires no training for system to perform the task.
['J. L. Raheja', 'Ankit Chaudhary', 'K. Das', 'S. Raheja']
2014-06-13
null
null
null
null
['fingertip-detection']
['computer-vision']
[ 3.34531128e-01 -5.27748406e-01 2.27678671e-01 -2.87188321e-01 1.23397909e-01 -9.75464582e-01 4.14169729e-01 1.70336202e-01 -8.36338460e-01 6.05257928e-01 -2.40559697e-01 -3.97087216e-01 -3.80132794e-01 -6.94506228e-01 1.53896138e-01 -4.90611553e-01 2.23382652e-01 4.25356299e-01 5.89137197e-01 -1.66122913e-01 7.55567193e-01 9.90199387e-01 -1.31006074e+00 -2.82352775e-01 1.57333583e-01 6.05410635e-01 1.64920166e-01 1.24812984e+00 -5.90652972e-02 3.21464315e-02 -4.42827642e-01 -2.52086972e-03 6.60872757e-01 -2.51226842e-01 -3.87234449e-01 -2.39742827e-02 -2.07069963e-01 -7.52953529e-01 7.48738945e-02 5.28254330e-01 8.59384596e-01 3.75180811e-01 7.54198492e-01 -1.06374049e+00 1.37785226e-01 -7.30895400e-02 -9.37152922e-01 1.72271375e-02 4.83375967e-01 1.32647827e-01 1.13910779e-01 -5.44442534e-01 5.18780231e-01 1.11445892e+00 3.44817936e-01 4.51229692e-01 -4.11849409e-01 -6.45238280e-01 -4.20869499e-01 1.31336108e-01 -1.41241729e+00 1.95498943e-01 7.49206424e-01 -3.94567698e-01 1.08995330e+00 4.51918691e-01 5.20022929e-01 3.74352068e-01 -7.81328753e-02 2.21212104e-01 1.36756623e+00 -9.02671218e-01 3.25012840e-02 1.13116845e-01 3.22693467e-01 6.41000926e-01 1.76275805e-01 -1.05368653e-02 -7.15464354e-02 1.74325883e-01 1.21347201e+00 4.44836378e-01 -1.29920244e-01 7.26480410e-02 -1.15920758e+00 2.42012590e-01 1.94840446e-01 6.62298322e-01 -4.99000132e-01 5.72455302e-02 2.26933330e-01 -6.10190891e-02 -3.48925143e-01 9.55858529e-02 -4.64257777e-01 -7.37992883e-01 -7.53327131e-01 -7.08835423e-02 6.67770088e-01 7.49368846e-01 -2.96210796e-02 -2.56439835e-01 3.09707642e-01 3.31095964e-01 5.76178789e-01 5.71096539e-01 2.82319516e-01 -2.92642474e-01 4.81443316e-01 8.82275581e-01 3.78884852e-01 -9.34711635e-01 -4.64213282e-01 4.55778241e-01 -5.78575790e-01 9.04027522e-01 9.91112232e-01 -4.03455138e-01 -1.20968652e+00 6.92189157e-01 3.99069399e-01 -7.02088177e-02 -2.26991892e-01 9.36886668e-01 4.66341048e-01 6.40568614e-01 8.39284807e-02 3.72649916e-02 1.49761033e+00 -2.08950892e-01 -4.84426528e-01 6.00604355e-01 -1.11964397e-01 -1.43958378e+00 1.12943769e+00 9.51625049e-01 -5.83707869e-01 -2.96870887e-01 -1.01226568e+00 2.29010507e-02 -7.75544763e-01 3.24396729e-01 3.98079842e-01 1.21363819e+00 -5.82961679e-01 4.15018141e-01 -1.04199016e+00 -1.07108152e+00 -6.65845489e-03 7.94117272e-01 -8.42847750e-02 2.98177779e-01 -3.31785977e-01 9.09621775e-01 1.09986864e-01 2.71724164e-01 2.67777026e-01 1.59791753e-01 -9.87657532e-02 -3.38622719e-01 8.25322345e-02 1.05616279e-01 8.61888766e-01 -5.01779616e-01 -1.67184019e+00 6.95384920e-01 -1.13096476e-01 1.76317453e-01 8.51104856e-01 -2.80834228e-01 -2.91831583e-01 1.08308397e-01 -5.47808588e-01 1.80946812e-01 7.37097919e-01 -8.08986902e-01 -5.86020648e-01 -8.47009659e-01 -5.99794388e-02 3.39467973e-01 -1.37883916e-01 4.45685506e-01 -1.78241655e-01 -4.86954808e-01 3.48806828e-01 -8.52059245e-01 2.27120608e-01 1.99484885e-01 -5.32901466e-01 -2.76410699e-01 1.42973089e+00 -1.06023681e+00 1.02043283e+00 -1.60688698e+00 -2.86146283e-01 9.64362144e-01 -1.53215677e-01 6.11872733e-01 5.22015572e-01 6.01128280e-01 4.18023676e-01 -1.14768557e-01 -7.60217309e-02 3.60097378e-01 -8.80845040e-02 4.77068983e-02 1.50304690e-01 2.23243698e-01 -1.97157204e-01 2.27610096e-01 -5.42831242e-01 -6.34815812e-01 8.46376836e-01 9.60666955e-01 -2.23366972e-02 1.59910589e-01 3.70541155e-01 6.81395650e-01 -5.21667421e-01 9.40355659e-01 5.89989722e-01 5.44013917e-01 1.14493690e-01 -4.26286936e-01 -4.70365644e-01 -4.08296078e-01 -1.60065329e+00 1.03527308e+00 -5.77347636e-01 7.67234862e-01 -4.22652904e-03 -7.13696361e-01 9.50335324e-01 6.07551515e-01 3.04722190e-01 -2.44685575e-01 6.17044747e-01 3.79555970e-01 1.47253245e-01 -8.66395593e-01 3.11321765e-01 2.44563952e-01 5.21987140e-01 7.31937468e-01 -4.09582466e-01 -3.55126560e-02 2.84644049e-02 -3.42785627e-01 5.71083009e-01 4.02780682e-01 5.51270247e-01 1.68326333e-01 5.41776597e-01 1.73185449e-02 -1.62247971e-01 3.51556987e-01 -1.62891775e-01 4.43304390e-01 -1.58473507e-01 -5.53942360e-02 -1.09709764e+00 -9.32801962e-01 -2.83388868e-02 7.92481422e-01 7.55050182e-02 2.55702913e-01 -7.03434587e-01 -1.18856467e-01 -1.59617350e-01 7.00965673e-02 -1.59243718e-01 7.81765223e-01 -8.19604039e-01 -1.55441463e-01 2.95154482e-01 6.45473301e-01 8.33054900e-01 -1.27912366e+00 -1.57974482e+00 1.98314488e-01 6.81935668e-01 -8.10836852e-01 -5.91383055e-02 -2.66921699e-01 -1.16369939e+00 -1.33334339e+00 -1.30856586e+00 -8.22479367e-01 9.27889824e-01 3.99600938e-02 3.20539236e-01 2.68055111e-01 -8.97338033e-01 6.24269545e-01 -7.86942244e-01 -1.06544518e+00 2.98750907e-01 -3.54390331e-02 -1.70691013e-01 -3.79192233e-01 8.44458163e-01 -4.66663390e-01 -8.39133024e-01 3.29842269e-01 -5.44596612e-01 -3.33010584e-01 5.82871079e-01 2.87212312e-01 5.08598052e-02 5.35651855e-02 4.97038476e-02 -4.39951509e-01 9.69004273e-01 9.15405974e-02 -6.27035141e-01 3.02879989e-01 -3.34219784e-01 -4.91863154e-02 2.31114909e-01 -3.31358582e-01 -8.65324199e-01 2.73132980e-01 1.19296029e-01 3.99498403e-01 -5.38887799e-01 9.62663665e-02 -1.78283546e-02 -3.24927181e-01 4.58945304e-01 3.92752327e-02 1.12810545e-01 -7.44455755e-01 -1.10450849e-01 1.39645028e+00 3.66579503e-01 -3.18475693e-01 6.95877373e-01 2.72096068e-01 2.61094958e-01 -1.33613014e+00 5.01569152e-01 -7.21177459e-01 -1.03485501e+00 -6.36630595e-01 9.10735428e-01 2.53780305e-01 -1.18645489e+00 7.76970625e-01 -1.27882397e+00 7.74317086e-02 5.59767663e-01 7.85931826e-01 1.69758275e-01 4.12195772e-01 -8.35807323e-02 -1.65747929e+00 -6.63887262e-01 -8.42831612e-01 5.01659214e-01 7.81845450e-01 -3.21832895e-01 -5.97130418e-01 -4.00976568e-01 1.20127806e-02 5.59055805e-01 6.78778231e-01 6.25439465e-01 -1.19916312e-01 -3.37772191e-01 -7.57912159e-01 -3.65806729e-01 -4.03416976e-02 7.57206917e-01 3.63926977e-01 -6.61618233e-01 1.06509000e-01 -3.12271476e-01 1.53770879e-01 1.99015781e-01 3.47493827e-01 8.40791166e-01 -5.03522307e-02 -2.52993315e-01 -1.48663461e-01 1.79015756e+00 9.85936582e-01 4.91794288e-01 2.59857923e-01 7.06673801e-01 5.49778402e-01 6.04500592e-01 6.94621742e-01 -1.03662349e-01 4.66814339e-01 1.29220948e-01 8.35854784e-02 1.15182288e-01 2.95999765e-01 -2.18895636e-02 2.37493098e-01 -1.08714879e+00 -1.65950134e-01 -1.11585963e+00 2.91520923e-01 -1.36444378e+00 -7.36746788e-01 -4.39854532e-01 2.61689258e+00 5.73343635e-01 -1.57265678e-01 4.73911494e-01 1.03417575e+00 8.91571164e-01 -6.66809797e-01 -3.99068415e-01 -7.48460054e-01 5.94311178e-01 9.89184201e-01 7.81868756e-01 6.56132758e-01 -8.22918594e-01 6.81448102e-01 4.67166185e+00 -6.87462091e-02 -1.62050295e+00 -1.66690394e-01 2.11529508e-02 8.01061019e-02 5.56304216e-01 -3.22316110e-01 -5.05706191e-01 5.12620568e-01 8.03981349e-02 4.26043779e-01 5.65245271e-01 4.99731332e-01 5.57693124e-01 -7.36593246e-01 -7.08599567e-01 1.19893074e+00 -3.88096929e-01 -5.70159078e-01 -1.30713329e-01 -3.81655209e-02 1.79466084e-01 -6.05460167e-01 -1.77559331e-01 -4.29439336e-01 -3.63576382e-01 -1.08268952e+00 2.03725874e-01 7.06603587e-01 7.65725613e-01 -7.97461510e-01 8.39459300e-01 1.83806315e-01 -1.31295180e+00 1.18721403e-01 -1.86590776e-02 -3.34681481e-01 2.60869950e-01 -1.00711696e-01 -1.22259974e+00 -1.22834258e-01 4.76134330e-01 -2.95579672e-01 -1.19683400e-01 1.34462202e+00 1.21870033e-01 5.71648836e-01 -9.95490670e-01 -8.27225864e-01 1.11822397e-01 -2.52110660e-01 4.15552348e-01 1.46118367e+00 4.86319363e-01 5.00802100e-01 -3.03911954e-01 2.67096043e-01 5.77442229e-01 4.11118895e-01 -2.90854365e-01 -2.61470556e-01 5.72697103e-01 1.07230127e+00 -1.28102529e+00 -1.00589395e-01 -1.59493074e-01 1.11134183e+00 -8.33583415e-01 2.15512857e-01 -2.84483194e-01 -1.34221900e+00 3.59954797e-02 6.20590329e-01 -1.49971455e-01 -9.48982894e-01 -4.59498674e-01 -3.84786904e-01 1.41979590e-01 -3.60082120e-01 3.76745500e-02 -4.69907254e-01 -5.53394735e-01 1.73488826e-01 1.30491138e-01 -1.16803646e+00 -2.96971291e-01 -1.26118958e+00 -8.15308750e-01 1.14371097e+00 -7.64678121e-01 -1.13906562e+00 -9.79843557e-01 7.63778031e-01 6.81935668e-01 -2.08147764e-01 8.69018316e-01 1.66222185e-01 -2.05775738e-01 4.76157367e-01 -6.31948262e-02 5.24447441e-01 5.74110031e-01 -1.24383843e+00 -1.47808105e-01 9.36965227e-01 -1.75823763e-01 9.59145248e-01 7.67618179e-01 -6.43006921e-01 -1.43608665e+00 -5.64659573e-02 5.23724198e-01 -2.07746811e-02 2.90267449e-02 -6.23787381e-02 -2.83679396e-01 2.70584136e-01 1.79973274e-01 -2.70169914e-01 5.57796001e-01 -3.35427970e-01 2.13498458e-01 2.46581733e-02 -1.77749801e+00 3.90494734e-01 5.78997970e-01 -3.09006304e-01 -3.73653114e-01 5.82631938e-02 -5.43486297e-01 -4.17685360e-01 -5.72627246e-01 -1.25894085e-01 1.31653368e+00 -9.27462876e-01 7.30371058e-01 -3.21603090e-01 -2.74610549e-01 -4.16184902e-01 1.17865521e-02 -5.18707693e-01 3.74318153e-01 -8.16327453e-01 2.69691765e-01 1.31436396e+00 1.46129623e-01 -7.69183278e-01 1.04595649e+00 1.07965040e+00 6.57155037e-01 -4.56695199e-01 -5.68851531e-01 -4.07904685e-01 -6.12660766e-01 -2.60762781e-01 3.03307652e-01 5.40914178e-01 1.45977423e-01 -1.06404640e-01 1.01853916e-02 4.07751761e-02 6.42656147e-01 -2.96040893e-01 6.24862075e-01 -1.51543248e+00 -3.59313725e-03 -1.52644873e-01 -8.12051535e-01 -4.67824548e-01 -1.04186416e+00 -8.21332484e-02 -2.73834914e-01 -1.64644670e+00 -3.20262015e-01 -4.89064693e-01 1.13124013e-01 3.65934044e-01 3.41906011e-01 4.42323923e-01 2.31555685e-01 1.47570491e-01 3.44335169e-01 -6.18319511e-01 1.02848721e+00 3.12541723e-01 -6.66048110e-01 3.72839630e-01 1.42368257e-01 5.91545045e-01 1.07934034e+00 -1.82064213e-02 -5.29636621e-01 -1.55678362e-01 6.99658915e-02 -2.17175886e-01 4.84006017e-01 -1.00133300e+00 2.22974166e-01 -1.94375336e-01 6.06810927e-01 -5.89652956e-01 2.64973998e-01 -1.38730597e+00 -1.46313116e-01 5.63546181e-01 1.31940231e-01 2.85551220e-01 -1.70592457e-01 1.88512072e-01 3.51873547e-01 -4.21669781e-01 6.16989374e-01 -3.97931397e-01 -8.55632305e-01 -2.75394857e-01 -4.34191436e-01 -6.76901400e-01 1.28088117e+00 -1.23418844e+00 2.31756017e-01 -3.72708768e-01 -7.23764002e-01 -2.80155420e-01 2.27334604e-01 8.94115269e-02 6.51268840e-01 -7.62576640e-01 -2.41836891e-01 1.79076623e-02 -3.20698291e-01 -2.05598939e-02 -3.15263778e-01 7.00326204e-01 -1.45956218e+00 5.22056937e-01 -8.77130449e-01 -1.71227574e-01 -2.03271818e+00 -1.52280182e-01 -5.18531315e-02 3.59393537e-01 -4.18432146e-01 5.70645750e-01 -1.17696929e+00 1.55095994e-01 3.68539125e-01 -7.14447260e-01 -5.82221925e-01 -9.58353728e-02 4.96719062e-01 1.03731132e+00 -1.40407264e-01 -5.63579261e-01 -5.47078490e-01 1.23834205e+00 2.64072806e-01 -7.73904264e-01 1.20810854e+00 2.51622535e-02 5.16296886e-02 2.83231735e-01 8.76464128e-01 1.80373177e-01 -8.14032316e-01 1.73904479e-01 1.26327783e-01 -7.58775294e-01 -4.27790731e-02 -1.20341349e+00 -6.79409683e-01 1.10984373e+00 1.32960606e+00 8.77711326e-02 1.05166352e+00 -5.34439445e-01 7.48186886e-01 5.22764981e-01 5.04509628e-01 -1.50590491e+00 -3.41394395e-01 2.21511379e-01 8.66391838e-01 -1.20956051e+00 1.38256088e-01 -3.42406869e-01 -4.96275365e-01 1.76203775e+00 3.88153940e-01 -2.83351541e-01 9.76663172e-01 4.99354780e-01 1.88543558e-01 2.44798791e-02 4.88120526e-01 -2.98003644e-01 1.84921354e-01 1.07834542e+00 9.72069263e-01 2.14327872e-01 -1.04661179e+00 1.82942808e-01 -2.26233944e-01 4.91241813e-01 3.32651019e-01 1.74817407e+00 -5.79600871e-01 -1.23623645e+00 -8.47842634e-01 6.49794757e-01 -7.27198601e-01 2.18974486e-01 -5.67674220e-01 9.47783947e-01 1.74080774e-01 1.08976734e+00 -4.18605357e-02 -4.84499305e-01 1.37146801e-01 2.68042624e-01 1.01487827e+00 -1.78966418e-01 -5.81285655e-01 -1.76639766e-01 -2.53836811e-01 -1.43944815e-01 -3.83170009e-01 -5.15975237e-01 -1.62188447e+00 -3.54994863e-01 -1.67775825e-01 -2.19276905e-01 1.52349031e+00 8.63301456e-01 -1.05337679e-01 -8.04614499e-02 1.32066280e-01 -1.08811867e+00 -3.58553529e-01 -1.18045819e+00 -5.06865740e-01 1.15143590e-01 4.91356254e-02 -5.91306508e-01 2.54342347e-01 1.93977877e-01]
[6.484547138214111, -0.28094691038131714]
8a20808a-c999-4e61-aee5-50233f1ea4a0
chatgpt-chemistry-assistant-for-text-mining
2306.11296
null
https://arxiv.org/abs/2306.11296v1
https://arxiv.org/pdf/2306.11296v1.pdf
ChatGPT Chemistry Assistant for Text Mining and Prediction of MOF Synthesis
We use prompt engineering to guide ChatGPT in the automation of text mining of metal-organic frameworks (MOFs) synthesis conditions from diverse formats and styles of the scientific literature. This effectively mitigates ChatGPT's tendency to hallucinate information -- an issue that previously made the use of Large Language Models (LLMs) in scientific fields challenging. Our approach involves the development of a workflow implementing three different processes for text mining, programmed by ChatGPT itself. All of them enable parsing, searching, filtering, classification, summarization, and data unification with different tradeoffs between labor, speed, and accuracy. We deploy this system to extract 26,257 distinct synthesis parameters pertaining to approximately 800 MOFs sourced from peer-reviewed research articles. This process incorporates our ChemPrompt Engineering strategy to instruct ChatGPT in text mining, resulting in impressive precision, recall, and F1 scores of 90-99%. Furthermore, with the dataset built by text mining, we constructed a machine-learning model with over 86% accuracy in predicting MOF experimental crystallization outcomes and preliminarily identifying important factors in MOF crystallization. We also developed a reliable data-grounded MOF chatbot to answer questions on chemical reactions and synthesis procedures. Given that the process of using ChatGPT reliably mines and tabulates diverse MOF synthesis information in a unified format, while using only narrative language requiring no coding expertise, we anticipate that our ChatGPT Chemistry Assistant will be very useful across various other chemistry sub-disciplines.
['Omar M. Yaghi', 'Jennifer T. Chayes', 'Christian Borgs', 'Oufan Zhang', 'Zhiling Zheng']
2023-06-20
null
null
null
null
['chatbot', 'prompt-engineering', 'chatbot']
['methodology', 'natural-language-processing', 'natural-language-processing']
[ 1.20036483e-01 1.54034555e-01 -4.23086524e-01 -1.11054271e-01 -8.77173781e-01 -8.25081170e-01 4.45131123e-01 6.64999664e-01 -8.95268470e-02 8.42029572e-01 1.01081245e-01 -1.12816370e+00 -2.38752842e-01 -6.99538171e-01 -6.48501933e-01 -3.21427613e-01 3.22170734e-01 5.51801145e-01 -2.90127218e-01 -6.92150518e-02 1.02810144e+00 5.73572934e-01 -1.23157704e+00 6.42064869e-01 1.22758663e+00 4.00182217e-01 6.48023367e-01 5.38760602e-01 -6.49924636e-01 1.12809896e+00 -6.94400132e-01 -4.60701108e-01 -2.30616897e-01 -2.35461488e-01 -1.06247103e+00 -3.84661764e-01 2.43391749e-02 2.50274211e-01 1.74941532e-02 5.37487805e-01 3.25355262e-01 -1.28997102e-01 6.36882842e-01 -6.42533183e-01 -6.81022584e-01 7.25616038e-01 -1.01947457e-01 1.75421804e-01 9.08978820e-01 6.28583670e-01 9.68853295e-01 -8.83685172e-01 1.25384390e+00 1.25969040e+00 2.95037061e-01 4.34088349e-01 -1.26645160e+00 -1.04680574e+00 -2.72455495e-02 6.62574247e-02 -1.06589925e+00 -6.60461009e-01 1.39144003e-01 -8.52976382e-01 1.65625155e+00 7.31404349e-02 7.27950811e-01 1.21773434e+00 7.31764734e-01 7.31479898e-02 8.99347484e-01 -4.85311627e-01 4.31662738e-01 3.72160822e-01 1.58966586e-01 5.87216973e-01 3.12307507e-01 -2.97534585e-01 -9.65266526e-01 -4.54424709e-01 3.37507159e-01 7.53125083e-03 1.02625795e-01 2.40289137e-01 -8.91061544e-01 8.38622153e-01 -2.94476718e-01 1.70817703e-01 -1.52830765e-01 -2.86613107e-01 2.08658725e-01 4.82275754e-01 3.59299749e-01 1.19128180e+00 -6.85969353e-01 -3.74772340e-01 -8.45432937e-01 2.39707798e-01 1.22217476e+00 1.32254028e+00 5.83208025e-01 -1.99029908e-01 2.32035443e-01 6.10452950e-01 5.56512594e-01 1.21821456e-01 1.11191757e-01 -9.02433991e-01 4.82344240e-01 7.43582547e-01 5.26734255e-02 -6.01914883e-01 -4.28363830e-01 -3.19551490e-02 9.66613889e-02 1.11569479e-01 4.26948428e-01 -2.22227469e-01 -7.66987622e-01 1.24524212e+00 1.75666690e-01 -8.10013950e-01 2.14650229e-01 3.28354567e-01 1.22587931e+00 7.75518358e-01 6.95052803e-01 -3.93414408e-01 1.31647098e+00 -3.60562176e-01 -6.05169475e-01 2.60747522e-02 8.55216980e-01 -1.19058275e+00 1.05739522e+00 8.81632507e-01 -1.16121864e+00 -6.48066923e-02 -1.19093633e+00 -4.76257324e-01 -6.91613853e-01 -5.96847087e-02 1.11497676e+00 6.15349948e-01 -5.05693376e-01 1.16008568e+00 -6.77762389e-01 -5.16914964e-01 6.71288371e-01 5.18930018e-01 -3.52215707e-01 -1.33926436e-01 -6.25204563e-01 1.13339555e+00 4.98430371e-01 -3.70863497e-01 -9.35985863e-01 -1.28318083e+00 -5.41721821e-01 -9.50228646e-02 8.18672061e-01 -5.95915496e-01 1.23482215e+00 -6.63075522e-02 -1.52364683e+00 5.57442009e-01 -6.07390344e-01 8.06698129e-02 2.08619252e-01 -6.98944647e-03 -5.01748502e-01 -8.60704780e-02 4.05773044e-01 3.36583465e-01 3.50689381e-01 -7.10576594e-01 -4.93499964e-01 -3.54824096e-01 -2.12499902e-01 -1.58193395e-01 1.32170855e-03 2.80521780e-01 -1.48407102e-01 -2.05843329e-01 9.92379617e-03 -3.38763595e-01 -4.09663439e-01 -2.94946671e-01 -4.24751103e-01 -3.91149133e-01 5.52733600e-01 -5.59861541e-01 1.23873234e+00 -1.63725448e+00 7.36348331e-02 2.59395748e-01 6.58403337e-01 -2.24445432e-01 5.22592962e-02 9.02527750e-01 -1.23845570e-01 5.89015365e-01 2.10538581e-01 2.76100934e-01 -8.44540745e-02 -2.10326202e-02 -2.60593712e-01 1.66979760e-01 3.54936123e-01 9.26560700e-01 -9.46210444e-01 -3.32444519e-01 2.11890683e-01 1.13867268e-01 -6.18179977e-01 -7.95444660e-03 -7.80183673e-01 4.84470785e-01 -6.34134412e-01 1.12005758e+00 2.40932822e-01 -5.62599480e-01 6.42111003e-01 7.51389340e-02 -8.96298587e-01 1.02144289e+00 -8.65651846e-01 1.94308245e+00 -3.64370525e-01 3.98706079e-01 -4.82570864e-02 -5.29208183e-01 1.11031568e+00 2.80667156e-01 4.79357660e-01 -6.50488198e-01 1.36928588e-01 3.04125696e-01 6.36093095e-02 -7.19873130e-01 3.47991645e-01 -9.31927413e-02 1.89931661e-01 6.69016719e-01 6.04898810e-01 -3.78337532e-01 8.47566366e-01 4.37988669e-01 1.20233929e+00 3.15642685e-01 2.44124532e-01 -5.20325840e-01 2.14135081e-01 6.50994539e-01 3.61625046e-01 7.42440879e-01 2.80440629e-01 -1.33763468e-02 7.57948339e-01 -5.57926655e-01 -1.14290118e+00 -5.44636905e-01 -4.11675304e-01 1.08420932e+00 -5.03992736e-01 -1.42594540e+00 -4.65578109e-01 -3.62316549e-01 -1.50271401e-01 8.05048764e-01 -2.27459431e-01 1.90574184e-01 -3.54698412e-02 -9.22760129e-01 2.13088647e-01 -8.88848156e-02 -1.70889184e-01 -9.69605267e-01 -2.39256769e-01 6.57677174e-01 2.34257430e-01 -9.42709506e-01 1.01500042e-01 8.26423049e-01 -6.75134361e-01 -1.25704026e+00 -1.08164713e-01 -5.74231267e-01 1.64610058e-01 5.28498217e-02 1.04168427e+00 -2.18594059e-01 -6.21518254e-01 -1.31740406e-01 -9.01400223e-02 -1.04959333e+00 -9.77574408e-01 3.63891453e-01 -2.29669929e-01 -1.05106294e+00 7.40379333e-01 -6.70908570e-01 -1.50493726e-01 -8.38904700e-04 -6.31957293e-01 2.03543872e-01 5.02583861e-01 2.90162206e-01 4.52105820e-01 -3.10388714e-01 7.01212764e-01 -1.14078915e+00 7.27235675e-01 -7.83315897e-01 -6.06837392e-01 4.95217532e-01 -9.87538874e-01 2.01201066e-01 6.50886834e-01 -4.40628409e-01 -8.02481472e-01 -4.28280570e-02 -2.04068273e-01 2.26923823e-01 -2.27608263e-01 7.66455173e-01 -3.75519425e-01 -4.57518287e-02 7.66438246e-01 -2.41395161e-01 1.90820564e-02 -4.39908415e-01 2.94648081e-01 8.02960873e-01 1.17796592e-01 -8.85416746e-01 3.85594755e-01 -2.56283619e-02 -6.79692179e-02 -1.02053487e+00 -4.15582269e-01 -3.34134459e-01 -2.84430146e-01 5.23547307e-02 7.51037478e-01 -9.10806715e-01 -1.37124169e+00 -2.41673350e-01 -1.25575566e+00 -1.23828977e-01 8.17017928e-02 4.38171089e-01 -9.49410349e-02 2.13628918e-01 -6.38256311e-01 -6.36771858e-01 -5.35824537e-01 -1.31450057e+00 6.42072797e-01 2.64523059e-01 -9.83017981e-01 -6.20683312e-01 3.12324171e-03 5.75075209e-01 9.79783237e-02 2.86159575e-01 1.46491444e+00 -7.83631742e-01 -7.03850091e-01 3.23484093e-01 -2.88721696e-02 -4.99726474e-01 2.18141571e-01 4.43929344e-01 -8.81148636e-01 2.27411196e-01 -8.37210342e-02 -3.85835022e-01 4.79088008e-01 7.25209638e-02 9.70623910e-01 -2.65230924e-01 -3.56302023e-01 4.90273893e-01 1.19656706e+00 6.98056996e-01 4.76425171e-01 5.67848504e-01 7.09507644e-01 7.19298124e-01 3.82631272e-01 4.82788295e-01 -7.35690892e-02 1.08442113e-01 3.05652749e-02 3.02066237e-01 5.36379516e-01 -2.96666443e-01 5.46251833e-01 5.09471416e-01 -2.77496427e-01 -1.76808983e-01 -9.78713334e-01 -3.91833149e-02 -1.53387880e+00 -7.89365828e-01 -3.98482889e-01 1.78814328e+00 1.03599429e+00 2.26972386e-01 -2.65575331e-02 -2.02186808e-01 8.47325101e-02 -3.23745281e-01 -7.71141052e-01 -8.69135320e-01 3.14441919e-02 8.44335437e-01 4.89004940e-01 3.88944179e-01 -6.04316592e-01 1.24808383e+00 6.91094398e+00 9.13526595e-01 -1.13302636e+00 -2.04669923e-01 4.63606119e-01 -3.74930233e-01 -5.22231698e-01 2.96355516e-01 -9.39900696e-01 3.04957390e-01 1.35592103e+00 -4.93334234e-01 7.03588545e-01 8.34415257e-01 6.72293365e-01 -1.21428847e-01 -1.36277223e+00 7.34606445e-01 -5.30979931e-01 -2.29260755e+00 2.39073396e-01 2.25575566e-01 3.15796405e-01 -4.50034030e-02 -3.58940870e-01 1.25838667e-01 7.24469543e-01 -1.50826216e+00 7.98366547e-01 3.47717315e-01 8.20483327e-01 -6.88916683e-01 4.45601866e-02 1.73203230e-01 -7.64671564e-01 -8.86757523e-02 -5.02985477e-01 -2.74252772e-01 -8.04164335e-02 6.68020129e-01 -1.23507261e+00 9.02739584e-01 5.70764244e-01 9.46156323e-01 -3.64689022e-01 3.60352039e-01 -5.58134122e-03 4.14617926e-01 -2.25681558e-01 -3.49709332e-01 5.94464410e-03 -4.07411784e-01 2.48480245e-01 1.37739444e+00 1.55127019e-01 2.04829127e-01 5.04125878e-02 1.48334467e+00 -3.50209959e-02 3.62976164e-01 -5.25146604e-01 -9.23723817e-01 6.55555844e-01 1.21835971e+00 -7.91969657e-01 -1.57078326e-01 -4.07916725e-01 2.34248832e-01 2.92057320e-02 3.00713390e-01 -3.71326238e-01 -3.17397863e-01 7.63459027e-01 1.74875766e-01 1.01481043e-01 -4.08208996e-01 -8.16751420e-01 -8.28308821e-01 -2.38260776e-01 -1.19119477e+00 2.78424799e-01 -6.84660494e-01 -1.09744155e+00 2.70719945e-01 -2.69069821e-01 -4.47638512e-01 1.49567142e-01 -8.28320026e-01 -6.36423111e-01 1.04673135e+00 -9.34363127e-01 -7.87912488e-01 1.93931296e-01 1.66596994e-01 6.07895613e-01 -3.07955593e-01 9.03799713e-01 6.26426637e-02 -8.29732060e-01 1.50868341e-01 1.01438105e-01 -5.21207094e-01 8.27963531e-01 -1.11339796e+00 3.18371087e-01 1.62239477e-01 -1.08567961e-01 1.42904556e+00 7.42502153e-01 -1.09307313e+00 -1.96156585e+00 -8.45158160e-01 9.21387136e-01 -9.17760193e-01 9.14367974e-01 -6.74819887e-01 -8.61029148e-01 2.13845506e-01 1.53218195e-01 -9.38390732e-01 1.18986535e+00 2.67979532e-01 -2.37671047e-01 4.13599849e-01 -9.77275550e-01 6.75470352e-01 9.78443146e-01 -6.01305246e-01 -4.59706515e-01 6.45100594e-01 7.56195664e-01 -4.07669574e-01 -1.37035990e+00 -1.22827597e-01 5.57505965e-01 -3.69253069e-01 7.58525908e-01 -1.08784831e+00 9.48434591e-01 -2.31510594e-01 1.51450217e-01 -7.19705760e-01 -1.50850907e-01 -1.14225888e+00 1.51799470e-01 1.08792758e+00 1.09018791e+00 -4.00616556e-01 6.21616721e-01 1.08798802e+00 -3.22258800e-01 -7.40306675e-01 -4.09144431e-01 -2.20122606e-01 3.51957172e-01 -6.35510564e-01 4.63608146e-01 9.11002040e-01 6.31284773e-01 7.66857147e-01 1.82758972e-01 -2.81502187e-01 8.63420591e-02 -1.02840532e-02 7.32534885e-01 -1.38105381e+00 -1.89225793e-01 -2.83192843e-01 1.98875472e-01 -6.42010987e-01 -1.09286830e-01 -1.30822301e+00 -4.80377644e-01 -1.54529381e+00 3.80011797e-01 -2.22782046e-01 1.53073326e-01 6.60776794e-01 3.10484737e-01 -5.88251829e-01 -6.25047684e-02 3.76876086e-01 -5.49870729e-01 -1.23229176e-02 1.18640757e+00 -8.61512721e-02 -5.88408291e-01 -7.05167115e-01 -1.39499998e+00 4.20539141e-01 5.98963201e-01 -7.77479172e-01 -2.25175336e-01 1.20098941e-01 4.43600267e-01 -1.51570335e-01 -3.59647349e-02 -6.49101198e-01 3.70745361e-01 -6.56426787e-01 6.21903062e-01 -4.83461946e-01 -3.04642022e-01 -3.08676302e-01 6.23675227e-01 3.30572426e-01 -3.51745278e-01 -3.06574911e-01 4.61109400e-01 1.79248869e-01 2.93057203e-01 -2.97728896e-01 3.64421934e-01 -6.84869766e-01 -2.29734600e-01 1.22296862e-01 -1.11537802e+00 -1.52792618e-01 6.97167635e-01 -2.38896489e-01 -6.63985074e-01 1.00685716e-01 -6.79128766e-01 2.79061556e-01 6.82406366e-01 3.20508987e-01 2.09058523e-01 -4.36657906e-01 -2.54255503e-01 1.28980145e-01 7.54051358e-02 4.89378497e-02 -4.67927419e-02 5.63474834e-01 -7.69421101e-01 7.51275599e-01 -2.48694718e-02 -3.50469351e-01 -1.08116388e+00 5.10897815e-01 -1.16778180e-01 -2.16624960e-01 -4.65034842e-01 3.15932781e-01 1.15913283e-02 -2.97656745e-01 1.68627441e-01 -4.66545135e-01 -5.72745204e-02 1.11747369e-01 6.71049178e-01 3.21693540e-01 4.63531554e-01 1.75629273e-01 -3.40231001e-01 2.14471564e-01 -5.75277150e-01 -1.21179633e-01 1.75491834e+00 3.01150233e-01 -8.35045353e-02 4.62783158e-01 8.07497263e-01 2.04687893e-01 -8.43873143e-01 2.69893169e-01 5.14922500e-01 5.81000783e-02 -1.98039874e-01 -1.10958529e+00 -1.74164683e-01 5.71561933e-01 -5.36551550e-02 -1.60501033e-01 2.85952628e-01 1.89056769e-01 4.28285062e-01 8.14647973e-01 2.67985880e-01 -1.21623218e+00 -1.87539190e-01 7.02631891e-01 7.66255021e-01 -1.03909767e+00 3.64886582e-01 -5.02025843e-01 -4.13896233e-01 1.62262106e+00 3.76986772e-01 7.30935335e-01 2.46902272e-01 5.09383261e-01 -6.46288097e-02 -7.92228997e-01 -1.10719824e+00 2.59620011e-01 -2.07325861e-01 4.51161563e-01 8.52272093e-01 -2.55733907e-01 -4.23894495e-01 9.57793891e-01 -4.35876846e-01 2.92890131e-01 5.79144835e-01 1.36787033e+00 -6.19264364e-01 -1.61141348e+00 -3.91141713e-01 4.33899224e-01 -5.54756820e-01 -3.78173620e-01 -1.06278014e+00 6.90731108e-01 1.43306151e-01 1.35034144e+00 -4.61800098e-01 -3.31121802e-01 1.95806831e-01 4.42794502e-01 4.36364323e-01 -9.13392425e-01 -7.95701385e-01 2.78424710e-01 5.03699720e-01 -4.38364953e-01 -1.08163312e-01 -5.47396243e-01 -1.49054861e+00 -7.47800946e-01 -2.75848150e-01 6.26579285e-01 8.04611802e-01 1.12881970e+00 6.54792368e-01 4.49778348e-01 7.88547918e-02 -4.41643715e-01 2.20299233e-02 -8.63983035e-01 -2.61130214e-01 -1.71974286e-01 -1.97481439e-01 -4.99846578e-01 6.12864494e-02 6.22126125e-02]
[4.795236587524414, 5.891251564025879]
6bdf4638-50e5-4f88-8a97-949d155df4f0
fedcut-a-spectral-analysis-framework-for
2211.13389
null
https://arxiv.org/abs/2211.13389v1
https://arxiv.org/pdf/2211.13389v1.pdf
FedCut: A Spectral Analysis Framework for Reliable Detection of Byzantine Colluders
This paper proposes a general spectral analysis framework that thwarts a security risk in federated Learning caused by groups of malicious Byzantine attackers or colluders, who conspire to upload vicious model updates to severely debase global model performances. The proposed framework delineates the strong consistency and temporal coherence between Byzantine colluders' model updates from a spectral analysis lens, and, formulates the detection of Byzantine misbehaviours as a community detection problem in weighted graphs. The modified normalized graph cut is then utilized to discern attackers from benign participants. Moreover, the Spectral heuristics is adopted to make the detection robust against various attacks. The proposed Byzantine colluder resilient method, i.e., FedCut, is guaranteed to converge with bounded errors. Extensive experimental results under a variety of settings justify the superiority of FedCut, which demonstrates extremely robust model performance (MP) under various attacks. It was shown that FedCut's averaged MP is 2.1% to 16.5% better than that of the state of the art Byzantine-resilient methods. In terms of the worst-case model performance (MP), FedCut is 17.6% to 69.5% better than these methods.
['Qiang Yang', 'Xingxing Tang', 'Lixin Fan', 'Hanlin Gu']
2022-11-24
null
null
null
null
['community-detection']
['graphs']
[-1.12670757e-01 2.69058570e-02 -1.03813879e-01 3.78389090e-01 -5.91896892e-01 -9.96339023e-01 6.67643726e-01 1.01075247e-01 -2.35838555e-02 6.47348106e-01 -3.05414200e-01 -4.34842199e-01 -4.76219565e-01 -8.11303198e-01 -5.90787232e-01 -9.78016913e-01 -6.87313259e-01 2.46283382e-01 2.14771599e-01 -1.38733715e-01 2.53933847e-01 4.10349458e-01 -9.06846166e-01 -1.52650341e-01 8.61960888e-01 6.03575826e-01 -4.95552033e-01 6.49995208e-01 9.47034732e-02 9.55718577e-01 -7.81055868e-01 -8.33916962e-01 6.15751326e-01 -3.17609698e-01 -7.04991341e-01 1.60414234e-01 1.88905030e-01 -7.06040412e-02 -5.29813766e-01 1.70386970e+00 1.14391856e-01 -1.93849325e-01 2.47081503e-01 -2.08942652e+00 -3.00977111e-01 9.80931044e-01 -9.16169405e-01 4.17814583e-01 2.29307726e-01 3.12314272e-01 8.93432617e-01 -3.56612235e-01 7.73142517e-01 1.34599829e+00 6.86585784e-01 3.93591732e-01 -1.18760836e+00 -1.27583492e+00 2.06062958e-01 3.65063101e-01 -1.44471371e+00 -2.04372987e-01 8.09830129e-01 -2.96005070e-01 3.41781825e-01 6.35668516e-01 4.40631419e-01 1.08869898e+00 3.66559803e-01 1.75372362e-01 1.26438856e+00 -2.19138980e-01 2.85372585e-01 5.49333952e-02 2.85321712e-01 9.09719944e-01 9.59777534e-01 4.87175763e-01 -4.97086197e-01 -9.95447874e-01 4.00663078e-01 -3.87809239e-02 -4.47923213e-01 -4.55763429e-01 -6.60908282e-01 9.46274817e-01 3.42787147e-01 3.29719223e-02 -1.19806834e-01 2.13440612e-01 5.11125207e-01 5.42708576e-01 3.91179681e-01 6.36973083e-02 -1.64404348e-01 1.95080459e-01 -5.85587919e-01 -2.33242307e-02 1.08504963e+00 6.97182775e-01 4.67205107e-01 1.23224162e-01 5.42480707e-01 1.50072649e-01 5.04530072e-01 3.09027970e-01 4.14035358e-02 -8.50140333e-01 3.54638010e-01 6.68742478e-01 4.00341960e-04 -1.59616482e+00 -2.34834224e-01 -8.26857090e-01 -8.45170081e-01 1.25802070e-01 3.32926691e-01 -1.56198218e-01 -2.93778867e-01 1.73639345e+00 8.58741879e-01 7.66102791e-01 8.67275521e-02 8.93734753e-01 1.58872828e-01 4.35759544e-01 1.33135235e-02 -7.93455184e-01 1.09017837e+00 -8.11487615e-01 -7.01894522e-01 2.93306649e-01 5.68390071e-01 -6.28748238e-01 3.02701473e-01 5.37541568e-01 -8.52469921e-01 2.71902561e-01 -1.09841514e+00 1.05711937e+00 1.51640107e-03 -8.74744236e-01 5.97892046e-01 1.29775119e+00 -8.52550685e-01 4.24065590e-01 -7.71720171e-01 -2.97097266e-01 3.70576859e-01 3.90366167e-01 -3.69432569e-01 5.01037948e-02 -1.22208285e+00 5.68251133e-01 4.21005815e-01 -3.97618294e-01 -1.58623266e+00 -7.53599346e-01 -5.35403192e-01 -1.92414261e-02 8.64581227e-01 -5.07367611e-01 1.01448286e+00 -7.55443513e-01 -8.70425165e-01 4.12994504e-01 4.57085490e-01 -8.39328885e-01 1.09709489e+00 3.75921935e-01 -6.60126507e-01 5.83281994e-01 -2.19111398e-01 -6.29169822e-01 8.05058241e-01 -1.60050964e+00 -5.52048028e-01 -4.51609284e-01 1.63159683e-01 4.63705743e-03 -5.40753484e-01 3.27545166e-01 2.27855772e-01 -4.97960895e-01 7.30087385e-02 -8.89143944e-01 -4.04769063e-01 -2.40629435e-01 -5.31513631e-01 -1.02703735e-01 1.26965773e+00 -3.61960351e-01 1.26790142e+00 -1.75661707e+00 -3.23428899e-01 6.62498772e-01 7.51437604e-01 3.61757100e-01 -1.30489394e-01 9.14000332e-01 8.23581442e-02 4.95971799e-01 -1.17492184e-01 -8.50059092e-02 -8.68428722e-02 -6.39430583e-02 -2.09034771e-01 1.20508087e+00 -6.42056406e-01 2.24286601e-01 -1.10021162e+00 -2.58979142e-01 -1.34171426e-01 2.09200203e-01 -1.80125400e-01 8.07429180e-02 1.28431231e-01 1.17620789e-01 -4.19383824e-01 6.02557540e-01 1.05483234e+00 -3.57304096e-01 7.71673858e-01 1.81633249e-01 2.10430369e-01 -1.89286724e-01 -1.48081648e+00 8.59757543e-01 -4.13365513e-02 1.10847518e-01 7.37187862e-01 -5.83515942e-01 9.00171041e-01 4.36965674e-01 5.99992096e-01 -9.00911018e-02 3.62578630e-01 2.44915113e-01 2.53043864e-02 -2.35300094e-01 1.03156224e-01 2.80426703e-02 1.40430480e-01 9.86377537e-01 -6.42551929e-02 5.00649571e-01 -3.21174040e-02 9.01955605e-01 1.50164235e+00 -6.54970109e-01 3.31342578e-01 -6.37673140e-01 7.07416713e-01 -2.00350881e-01 7.76333630e-01 9.22178805e-01 -6.85823977e-01 -1.85541391e-01 6.26030445e-01 -3.01832318e-01 -6.38355672e-01 -1.08618224e+00 4.21267480e-01 9.08406198e-01 5.67249000e-01 -7.08469093e-01 -9.43258762e-01 -1.16909635e+00 1.11936286e-01 4.19317216e-01 -3.46466690e-01 -3.27991277e-01 -1.76487654e-01 -9.25587893e-01 9.70568299e-01 -1.98065951e-01 7.93122947e-01 -3.42035204e-01 -2.28620067e-01 1.54537335e-01 -4.40752134e-02 -9.16369796e-01 -5.77816784e-01 -3.25283796e-01 -5.44502497e-01 -1.97082245e+00 6.88776374e-02 -3.99103075e-01 6.40828550e-01 7.80316114e-01 6.07159257e-01 6.05359674e-01 -2.10153192e-01 6.15537167e-01 -3.06260973e-01 -1.78009301e-01 -7.64214158e-01 -3.09573501e-01 5.40214598e-01 5.46744704e-01 1.04155570e-01 -8.43895376e-01 -6.62475467e-01 3.93150151e-01 -9.30319250e-01 -4.27924603e-01 6.55639395e-02 7.06637800e-01 -7.32610300e-02 5.71037173e-01 9.09962535e-01 -1.11338341e+00 8.54903162e-01 -9.22444403e-01 -8.42402458e-01 3.82277697e-01 -1.17149377e+00 -3.58070672e-01 8.07140946e-01 -5.79844773e-01 -8.55257809e-01 -2.91361004e-01 7.23667979e-01 -6.79774046e-01 2.91700125e-01 2.97461033e-01 -6.50969222e-02 -7.36992180e-01 9.22102690e-01 2.89571553e-01 3.05249333e-01 -1.68145657e-01 3.70125204e-01 5.38488567e-01 2.99109697e-01 -4.54940915e-01 1.28866792e+00 4.99178588e-01 8.29941854e-02 -5.04723847e-01 -1.57717407e-01 -4.46084619e-01 7.66429901e-02 -6.53283596e-01 2.74549425e-02 -7.82857001e-01 -1.10309601e+00 6.20164633e-01 -8.98892939e-01 9.20202285e-02 4.20769364e-01 1.10995539e-01 -1.68416128e-01 1.02092350e+00 -9.59843755e-01 -1.36165845e+00 -6.72002554e-01 -7.07524061e-01 -4.31001261e-02 2.57711709e-01 2.38442004e-01 -1.11979854e+00 2.49468852e-02 5.69847703e-01 1.80652931e-01 6.88051522e-01 6.12333000e-01 -1.24838209e+00 -5.19743383e-01 -3.38248074e-01 -8.31296444e-02 -7.50687346e-02 9.32273418e-02 1.63691506e-01 -6.95119619e-01 -8.86244833e-01 1.28294796e-01 3.09499763e-02 3.83271903e-01 -3.53770435e-01 7.72376001e-01 -1.16561115e+00 -3.63793790e-01 3.37937623e-01 1.46442533e+00 1.30902633e-01 1.41037971e-01 2.84324616e-01 4.92093176e-01 4.68521208e-01 3.99595380e-01 7.56469131e-01 2.37489954e-01 1.09092139e-01 1.17181385e+00 2.69275934e-01 1.34244263e-01 -3.89157712e-01 3.21515530e-01 7.69678414e-01 1.20762005e-01 -1.22845322e-01 -8.76388788e-01 4.84822214e-01 -1.83412552e+00 -9.47575986e-01 -4.19009179e-01 2.09212661e+00 5.65791428e-01 9.19524133e-02 3.68960559e-01 4.29788172e-01 1.34434521e+00 2.83021748e-01 -3.94518316e-01 -3.48765105e-01 1.71228349e-02 -2.47448772e-01 8.68344426e-01 5.06963074e-01 -7.06521630e-01 6.41852081e-01 5.40421820e+00 9.47014689e-01 -6.44207716e-01 4.76300627e-01 3.43179345e-01 2.29421072e-02 -1.69452310e-01 4.17977333e-01 -2.37249956e-01 5.45419455e-01 1.00217450e+00 -1.01758111e+00 9.22490776e-01 8.62699926e-01 1.21128395e-01 2.73537397e-01 -5.81953228e-01 7.61373639e-01 6.15985729e-02 -1.20347023e+00 -1.57630771e-01 2.80744940e-01 9.72564042e-01 -1.72658890e-01 -1.02316849e-01 1.22598307e-02 9.55918610e-01 -7.33784258e-01 4.52059627e-01 7.79499784e-02 3.54729384e-01 -1.35034132e+00 4.28170115e-01 6.42060041e-01 -1.35012662e+00 -5.08635342e-01 -1.98388621e-01 -7.89482892e-02 -1.19519308e-01 4.15097654e-01 -9.99205470e-01 1.00965142e+00 3.21789980e-01 2.83493519e-01 -4.46783841e-01 1.18324816e+00 -1.71952054e-01 8.70160162e-01 -1.92509085e-01 8.95197853e-04 2.21240908e-01 -1.57786027e-01 1.07946396e+00 1.05139375e+00 5.27946912e-02 9.49990600e-02 4.48496610e-01 5.61812460e-01 -3.77831280e-01 1.78033188e-01 -6.50391698e-01 8.36270005e-02 1.19160402e+00 1.48587334e+00 -5.91993570e-01 -2.40407705e-01 1.09753564e-01 4.20066327e-01 2.12533012e-01 4.49118078e-01 -8.94388735e-01 -4.06058133e-01 6.35724664e-01 1.02033518e-01 -1.88467234e-01 -7.30701610e-02 -2.03275353e-01 -1.06142724e+00 -3.26355964e-01 -1.34837425e+00 1.17213082e+00 -1.60425544e-01 -1.42674875e+00 5.55490136e-01 -2.82360286e-01 -1.07098734e+00 1.46752998e-01 9.54912156e-02 -9.98567641e-01 4.53871280e-01 -1.10855639e+00 -1.03278697e+00 -1.18501239e-01 9.41047013e-01 7.94167519e-02 -2.86482155e-01 5.03282726e-01 -1.22296605e-02 -7.43368030e-01 9.04546618e-01 6.08574552e-03 2.32230942e-03 3.76586378e-01 -8.21286142e-01 6.42269850e-03 1.40408957e+00 -2.97141522e-02 8.13042343e-01 9.17116761e-01 -8.67643058e-01 -1.63340378e+00 -1.23339176e+00 3.92282665e-01 9.85663980e-02 1.25954044e+00 -2.65736580e-01 -8.40688050e-01 6.41162813e-01 1.88723221e-01 2.60621756e-01 5.20613849e-01 -2.19522581e-01 -9.56919611e-01 -8.90045092e-02 -1.82518387e+00 7.03904510e-01 1.10432827e+00 -5.09272695e-01 -1.63722292e-01 5.56016445e-01 8.01619887e-01 5.39811775e-02 -7.01386988e-01 2.26191446e-01 1.53649136e-01 -1.07473052e+00 6.59995317e-01 -8.60069752e-01 -4.53889281e-01 -2.45819375e-01 -1.19528651e-01 -1.03614628e+00 -3.19200397e-01 -1.39703310e+00 -8.79383266e-01 1.18442416e+00 -1.11227944e-01 -1.27695739e+00 6.45677745e-01 2.45429501e-01 5.23725867e-01 -4.19116378e-01 -1.31435072e+00 -1.20740557e+00 1.32727837e-02 -1.59462035e-01 6.56732738e-01 1.48176563e+00 2.70061672e-01 -2.70701170e-01 -3.94872695e-01 7.34671950e-01 1.56946051e+00 -1.98161021e-01 7.31252432e-01 -1.27034378e+00 -3.66011336e-02 -4.01896328e-01 -4.81486112e-01 -7.64581934e-02 3.17089289e-01 -1.03257465e+00 -5.05466700e-01 -9.23173428e-01 4.45161462e-01 -2.46357709e-01 -6.38210654e-01 3.48133266e-01 -6.65862709e-02 1.52434930e-01 5.59006691e-01 3.10302645e-01 -7.11988688e-01 1.42486349e-01 8.14545095e-01 -2.01484680e-01 -5.94270155e-02 3.78368832e-02 -7.27245271e-01 4.61696476e-01 9.32463765e-01 -8.47604394e-01 -5.08468211e-01 4.75795954e-01 -5.71665633e-03 4.25414503e-01 5.98523736e-01 -9.09219623e-01 5.01104772e-01 -5.02787411e-01 -3.56902331e-01 -8.28748941e-02 -2.32301429e-01 -8.56881857e-01 6.43759370e-01 1.12594974e+00 -1.82771340e-01 2.27210112e-02 -2.88987726e-01 1.11708403e+00 4.15820718e-01 -8.67561549e-02 8.68567765e-01 -2.34345570e-02 -8.75551030e-02 5.70873201e-01 -2.27825478e-01 7.20103160e-02 1.53160965e+00 5.69983199e-02 -8.21460724e-01 -6.14183187e-01 -5.27791679e-01 4.13692027e-01 4.86236125e-01 2.02914625e-01 4.31344390e-01 -1.18141782e+00 -8.09334219e-01 -1.32202014e-01 -2.33982503e-01 -7.31478572e-01 4.03708279e-01 9.67185020e-01 -2.70312995e-01 -9.92985293e-02 -1.83520280e-02 -3.06367636e-01 -1.77432287e+00 8.71955872e-01 5.21858931e-01 -3.75145495e-01 -5.50200343e-01 5.78054190e-01 -2.62096047e-01 -2.12515742e-01 3.76329690e-01 6.51533484e-01 2.39090860e-01 -2.34864205e-02 5.49561381e-01 1.06930256e+00 -1.09595448e-01 -6.02049172e-01 -4.65180486e-01 4.60026003e-02 -1.90714702e-01 4.11004759e-02 9.64207530e-01 -3.15603018e-01 -3.70442867e-01 -2.42329806e-01 1.06915796e+00 8.86057317e-02 -1.14436579e+00 -3.61966550e-01 1.31511360e-01 -7.33242989e-01 -8.26653372e-03 -8.54973257e-01 -1.30703282e+00 1.90234199e-01 3.36953998e-01 8.15721810e-01 9.77052271e-01 -4.30958807e-01 7.83962011e-01 3.27078439e-02 1.01805592e+00 -7.64821231e-01 -1.76872138e-03 1.32902563e-01 3.76462013e-01 -6.26760900e-01 5.30802272e-02 -6.87662005e-01 -3.50671113e-01 9.98804569e-01 5.86316824e-01 -2.99775213e-01 6.06879771e-01 1.20756559e-01 1.22608822e-02 -2.17600077e-01 -1.06470799e+00 3.37191731e-01 -3.03011864e-01 6.67895317e-01 -5.70519090e-01 2.70534664e-01 -6.03773296e-01 7.95336008e-01 -2.36618742e-02 -4.35858369e-01 1.10704243e+00 8.40618074e-01 -5.79301715e-01 -7.71726668e-01 -7.10115433e-01 4.59772870e-02 -6.89535618e-01 4.41621363e-01 -6.54812217e-01 6.52463734e-01 -2.58045316e-01 1.60310900e+00 -7.30376601e-01 -8.01282167e-01 -1.35537386e-01 -7.76662752e-02 6.13537477e-03 3.16980034e-02 -1.00161314e+00 -7.02441484e-02 4.25730161e-02 -6.62297130e-01 -1.25728771e-01 -2.23211512e-01 -1.13946533e+00 -9.48987603e-01 -5.80703735e-01 6.65616572e-01 6.10206068e-01 5.32671213e-01 2.25694522e-01 1.52056307e-01 1.53120100e+00 -9.44351777e-02 -1.21029365e+00 -5.77800333e-01 -8.19100499e-01 3.19882482e-01 2.61795223e-01 -6.40389919e-01 -1.15611935e+00 -3.79628628e-01]
[5.739081859588623, 7.110145092010498]
90348fb7-4c19-4e93-b04a-c8864fa587f3
diffusion-models-for-video-prediction-and
2206.07696
null
https://arxiv.org/abs/2206.07696v3
https://arxiv.org/pdf/2206.07696v3.pdf
Diffusion Models for Video Prediction and Infilling
Predicting and anticipating future outcomes or reasoning about missing information in a sequence are critical skills for agents to be able to make intelligent decisions. This requires strong, temporally coherent generative capabilities. Diffusion models have shown remarkable success in several generative tasks, but have not been extensively explored in the video domain. We present Random-Mask Video Diffusion (RaMViD), which extends image diffusion models to videos using 3D convolutions, and introduces a new conditioning technique during training. By varying the mask we condition on, the model is able to perform video prediction, infilling, and upsampling. Due to our simple conditioning scheme, we can utilize the same architecture as used for unconditional training, which allows us to train the model in a conditional and unconditional fashion at the same time. We evaluate RaMViD on two benchmark datasets for video prediction, on which we achieve state-of-the-art results, and one for video generation. High-resolution videos are provided at https://sites.google.com/view/video-diffusion-prediction.
['Andrea Dittadi', 'Didrik Nielsen', 'Stefan Bauer', 'Arash Mehrjou', 'Tobias Höppe']
2022-06-15
null
null
null
null
['video-prediction']
['computer-vision']
[ 1.52877554e-01 -8.21189955e-02 -8.71584862e-02 -2.78472573e-01 -3.16440970e-01 -3.68421793e-01 1.10139930e+00 -3.78208965e-01 -1.97295457e-01 7.84281254e-01 4.40877050e-01 -3.30565691e-01 3.36018473e-01 -8.90803635e-01 -1.00788903e+00 -7.95896709e-01 -2.70744890e-01 3.78261149e-01 3.81367594e-01 4.88022640e-02 -2.20342912e-02 1.48936853e-01 -1.52330601e+00 7.03099191e-01 5.89165568e-01 8.70874941e-01 5.36852837e-01 1.00792098e+00 1.19716190e-01 1.41394174e+00 -4.26729739e-01 -4.13219780e-01 1.53942436e-01 -6.70327961e-01 -5.35023153e-01 2.52874732e-01 1.42798200e-01 -6.36144340e-01 -8.25383008e-01 7.03038812e-01 3.30685139e-01 3.35864812e-01 7.61789382e-01 -1.14457393e+00 -8.28188777e-01 7.10438192e-01 -2.89676279e-01 3.12299967e-01 5.08086860e-01 5.97815454e-01 6.68640733e-01 -8.43574345e-01 8.60120237e-01 1.20848286e+00 3.74038815e-01 1.01812017e+00 -1.28133452e+00 -4.66794550e-01 4.29435134e-01 3.75019908e-01 -1.06326962e+00 -4.37587976e-01 6.31535649e-01 -5.82766473e-01 1.07938468e+00 1.14507161e-01 1.03705025e+00 1.76316845e+00 3.94277364e-01 1.02272129e+00 9.05277431e-01 -5.91567904e-02 2.92308271e-01 -2.56691009e-01 -6.13758624e-01 7.49163270e-01 -3.15901309e-01 3.54725569e-01 -6.40056908e-01 1.77170619e-01 1.00192094e+00 3.15525323e-01 -4.93518442e-01 -1.95170581e-01 -1.47217572e+00 7.69227386e-01 4.79355991e-01 1.93499610e-01 -6.69154406e-01 5.67816854e-01 -7.97637459e-03 3.42419058e-01 6.69408858e-01 7.53315240e-02 -1.38241053e-01 -4.05632615e-01 -1.04776084e+00 4.47365016e-01 7.48124361e-01 8.04893076e-01 4.93505985e-01 3.03047568e-01 -6.14898801e-01 4.46070850e-01 2.53662527e-01 4.82407749e-01 5.83656788e-01 -1.17240918e+00 2.89126068e-01 1.42479405e-01 8.49063843e-02 -5.72492063e-01 -1.49756521e-01 -3.02298903e-01 -9.69040930e-01 3.88250768e-01 1.72334895e-01 -5.08468330e-01 -1.23761153e+00 1.82855725e+00 2.10665166e-01 8.29957902e-01 1.68368787e-01 9.11986947e-01 6.79589570e-01 1.13794446e+00 8.99828896e-02 -9.35087353e-02 7.75280476e-01 -1.30267274e+00 -4.68544543e-01 -2.29307428e-01 4.72437978e-01 -4.68785614e-01 8.53760362e-01 4.05678004e-01 -1.12704074e+00 -6.29781544e-01 -7.51523018e-01 3.83540690e-02 -1.23912059e-02 -2.14184411e-02 7.85615325e-01 2.59901047e-01 -1.31758857e+00 8.18548560e-01 -1.31876910e+00 -1.94113821e-01 5.57834983e-01 2.36190874e-02 -2.80951172e-01 -2.50058353e-01 -1.08263946e+00 6.06264591e-01 1.61132455e-01 -1.13072947e-01 -1.66951072e+00 -7.04020023e-01 -8.58024359e-01 6.10547215e-02 2.56401986e-01 -1.08224845e+00 1.44988728e+00 -1.03114486e+00 -1.66070974e+00 3.82375211e-01 -2.60402471e-01 -1.06914723e+00 9.19560254e-01 -2.83586442e-01 -1.98112741e-01 1.02771610e-01 -9.56078991e-02 1.22102082e+00 1.19880641e+00 -9.96933699e-01 -5.44479787e-01 4.20173667e-02 1.82062924e-01 2.56096423e-01 -1.48199946e-01 -3.99320960e-01 -6.47568166e-01 -1.04944587e+00 -3.35048825e-01 -1.03113437e+00 -4.93634373e-01 -5.82266115e-02 -1.56050399e-01 -4.91486080e-02 8.96302819e-01 -7.16317773e-01 1.20536780e+00 -2.00955749e+00 4.89761680e-01 -1.71667680e-01 2.09264502e-01 8.57458711e-02 -1.32063553e-01 3.18956941e-01 7.53509253e-02 8.06651786e-02 -3.30635577e-01 -9.01567757e-01 7.21335690e-03 1.98089674e-01 -4.06017065e-01 4.73170131e-02 3.27057660e-01 1.02610648e+00 -8.14749420e-01 -1.70681551e-01 3.36428374e-01 8.34172666e-01 -9.48364973e-01 4.25091803e-01 -7.95020759e-01 7.28376091e-01 -2.86929697e-01 2.80852199e-01 2.78047055e-01 -4.74852949e-01 5.39849065e-02 1.85717732e-01 1.07277766e-01 2.93690413e-01 -8.20217133e-01 1.91113222e+00 -3.93285602e-01 8.31997812e-01 -3.78141463e-01 -7.08465874e-01 6.15736365e-01 5.33974230e-01 3.90128940e-01 -6.47905111e-01 -4.36021611e-02 -1.14386633e-01 -1.29634157e-01 -4.66948032e-01 4.08624053e-01 -1.30219888e-02 2.28287458e-01 5.64298689e-01 1.93595156e-01 1.06143523e-02 4.80281174e-01 5.16888857e-01 1.17174911e+00 5.70913017e-01 -1.97215587e-01 1.68518052e-01 2.01393858e-01 -1.73294082e-01 4.30231780e-01 7.65123963e-01 2.23999709e-01 7.11743474e-01 5.08829534e-01 -4.96058136e-01 -1.08261263e+00 -1.00889897e+00 4.70162630e-01 7.58304060e-01 -8.98038894e-02 -4.71718401e-01 -8.11573923e-01 -6.19121850e-01 -1.98097169e-01 9.79560077e-01 -9.15801167e-01 -1.43047452e-01 -4.92920488e-01 -6.16487682e-01 1.56253994e-01 6.80630565e-01 4.22573894e-01 -1.27775121e+00 -7.02933311e-01 3.32876533e-01 -6.17915392e-02 -1.04733098e+00 -4.33893770e-01 -1.17681205e-01 -7.68865764e-01 -6.28753901e-01 -1.08720636e+00 -5.32578588e-01 4.97943848e-01 2.44388461e-01 1.22728276e+00 6.73843026e-02 -1.65090768e-03 5.30896604e-01 -3.44153255e-01 -4.48973812e-02 -6.50056720e-01 -6.58654124e-02 -7.66059682e-02 3.82289514e-02 -6.55781478e-02 -7.68127680e-01 -8.54423285e-01 8.52493346e-02 -9.04406786e-01 6.07837498e-01 5.83335936e-01 8.32263112e-01 6.29028022e-01 -5.22637926e-02 2.71583289e-01 -7.61874676e-01 5.12664795e-01 -7.03790843e-01 -5.43362319e-01 5.10088820e-03 -3.06976169e-01 8.34831372e-02 5.15207231e-01 -6.30421817e-01 -1.27618551e+00 1.45926885e-02 -3.68188798e-01 -9.53216195e-01 -2.93245018e-01 4.57862973e-01 2.05117092e-01 2.33426854e-01 4.42319125e-01 3.89017522e-01 -5.03659584e-02 -3.09122235e-01 4.98857677e-01 8.35720226e-02 3.67554426e-01 -3.79243404e-01 4.05245602e-01 5.07985234e-01 -1.93391383e-01 -5.02666771e-01 -5.56714356e-01 1.06909111e-01 -3.45865786e-01 -4.47811961e-01 1.11180067e+00 -1.07425618e+00 -5.05706847e-01 5.76162159e-01 -1.07555664e+00 -1.06906891e+00 -4.23047036e-01 5.29638648e-01 -7.72711039e-01 5.51201887e-02 -7.24101484e-01 -6.35059655e-01 -7.41640031e-02 -1.18591213e+00 7.69033611e-01 2.02245250e-01 -1.06516071e-01 -1.14458776e+00 7.92063475e-02 5.61548322e-02 6.06717408e-01 1.91376656e-01 4.69570667e-01 -2.82387465e-01 -1.00027013e+00 1.23379968e-01 2.55569100e-01 2.96859145e-01 -1.40388921e-01 1.31578818e-01 -7.71766186e-01 -2.31035262e-01 -9.42362323e-02 -2.97686100e-01 1.46636677e+00 5.58810294e-01 1.25481439e+00 -4.05327946e-01 -4.63428199e-01 6.84265077e-01 9.92323041e-01 2.20306650e-01 8.04873884e-01 1.78075228e-02 5.86806118e-01 2.58211344e-01 3.36251944e-01 6.14995241e-01 5.58531404e-01 6.17529333e-01 5.16288102e-01 3.23011875e-02 -4.98370111e-01 -4.97557312e-01 6.56348705e-01 5.35232484e-01 -2.73281008e-01 -7.39132166e-01 -7.04029083e-01 5.49393833e-01 -2.00268173e+00 -1.39211416e+00 2.39735022e-01 1.90488732e+00 6.10049129e-01 3.23727489e-01 1.34735614e-01 -2.03093842e-01 4.44377333e-01 4.14256364e-01 -6.66567504e-01 -1.19132679e-02 -5.99966198e-02 8.32202360e-02 1.45163685e-01 6.70618713e-01 -1.15420437e+00 1.11595547e+00 5.96822977e+00 8.10584903e-01 -1.20544875e+00 1.72294110e-01 1.01073623e+00 -4.20995146e-01 -5.25382578e-01 -2.80302428e-02 -6.67424083e-01 5.44252396e-01 9.45149720e-01 4.30403799e-02 5.75978756e-01 6.88394964e-01 3.71429831e-01 -1.01697175e-02 -1.12120950e+00 7.75043130e-01 2.14432050e-02 -1.75467539e+00 3.27931792e-01 -7.74977356e-02 9.97196853e-01 2.27905005e-01 3.99788201e-01 3.16837788e-01 5.85723758e-01 -1.06942415e+00 8.83549094e-01 8.64905059e-01 5.86201191e-01 -4.64941889e-01 3.29462767e-01 4.77315485e-01 -9.67025518e-01 -1.01344399e-01 -2.02765197e-01 -1.50090739e-01 6.07799530e-01 4.68515038e-01 -7.48193562e-01 1.86497793e-01 7.40193069e-01 1.09958088e+00 -3.46473902e-01 9.99200642e-01 -5.50339818e-01 8.37071657e-01 -2.14686900e-01 -5.88208623e-02 2.40485996e-01 -1.35714993e-01 5.89979887e-01 1.10090363e+00 7.54280329e-01 7.43407905e-02 8.19799155e-02 1.02741647e+00 -1.51242211e-01 -3.97797197e-01 -5.02567112e-01 -1.31068498e-01 1.59538686e-01 9.04964685e-01 -5.38410366e-01 -5.13723791e-01 -5.07286727e-01 1.35774410e+00 3.12268674e-01 5.55858314e-01 -1.18157339e+00 3.29474330e-01 6.86549664e-01 1.43654346e-01 6.82037055e-01 -4.68951732e-01 1.38744548e-01 -1.39495015e+00 -1.23936310e-01 -7.11537838e-01 1.46463856e-01 -9.43208814e-01 -1.02732646e+00 7.59607732e-01 -1.14478029e-01 -1.08409500e+00 -7.14601398e-01 -3.44704032e-01 -7.64054537e-01 6.30770385e-01 -1.29146206e+00 -1.11184537e+00 -3.94034863e-01 7.79596150e-01 8.65745366e-01 -1.07798725e-01 6.91116810e-01 2.61622876e-01 -4.80918139e-01 1.36547357e-01 -1.85385972e-01 -8.23591724e-02 4.76479173e-01 -1.13022935e+00 8.20103109e-01 1.05880284e+00 4.81312931e-01 2.51901567e-01 7.96582520e-01 -7.75828898e-01 -1.24413121e+00 -1.27229333e+00 5.94068944e-01 -3.60209823e-01 5.74363053e-01 -2.89931744e-01 -7.76027501e-01 9.94486749e-01 4.77829635e-01 -6.48723468e-02 4.53285366e-01 -3.53091091e-01 -4.20374945e-02 2.32367501e-01 -6.91359282e-01 8.62851441e-01 1.23536348e+00 -1.67464182e-01 -1.96497813e-01 4.05520737e-01 1.03101075e+00 -6.65555000e-01 -6.13319516e-01 1.83846474e-01 4.58286524e-01 -1.23026597e+00 8.29975903e-01 -5.43432057e-01 8.15966249e-01 -2.86907881e-01 -3.70831750e-02 -1.40051353e+00 -3.95175129e-01 -8.56672168e-01 -6.84719622e-01 9.78218377e-01 6.96323037e-01 -4.70638186e-01 9.00785089e-01 5.28972328e-01 -2.21642330e-01 -9.86889124e-01 -5.88727295e-01 -5.53467512e-01 -1.59520879e-01 -4.86380041e-01 6.15272224e-01 4.61541235e-01 -3.59724879e-01 1.25046819e-01 -8.41915071e-01 7.15631992e-02 3.61939520e-01 1.83012024e-01 8.12679529e-01 -6.63620651e-01 -7.82786787e-01 -5.07442415e-01 -2.51422673e-01 -1.72262299e+00 6.45289049e-02 -5.98980844e-01 -1.47196606e-01 -1.78044069e+00 7.43187591e-02 -2.70396799e-01 -1.98086612e-02 3.90433848e-01 -1.37310147e-01 3.78561795e-01 3.98259133e-01 2.95054644e-01 -6.60721064e-01 9.43683088e-01 1.34188259e+00 -1.59776479e-01 -1.66062325e-01 1.62326902e-01 -3.38694543e-01 6.26942456e-01 9.37449634e-01 -3.04803312e-01 -6.42983735e-01 -6.68698728e-01 1.50028110e-01 2.78437555e-01 5.25372624e-01 -1.23576725e+00 2.25507602e-01 -1.21735506e-01 6.55478179e-01 -3.00525784e-01 7.99406946e-01 -5.36142588e-01 4.94314164e-01 5.28424382e-01 -4.72326189e-01 1.81630775e-01 6.30122572e-02 7.61124074e-01 -2.08178133e-01 6.79206550e-02 5.04790246e-01 -2.11962447e-01 -1.02087390e+00 8.26620758e-01 -5.77999711e-01 -1.92481562e-01 1.22081923e+00 7.36066997e-02 -2.76039541e-01 -8.90560806e-01 -1.04729211e+00 2.15758160e-01 5.82663000e-01 5.42981207e-01 9.03632343e-01 -1.32189369e+00 -8.19512904e-01 2.19929367e-01 -2.71943539e-01 -7.53348544e-02 5.60620248e-01 7.72743464e-01 -5.40205419e-01 2.66830325e-01 -1.39033675e-01 -5.02858460e-01 -9.65349078e-01 7.79503763e-01 2.66324252e-01 -2.88988531e-01 -7.60144532e-01 1.03346622e+00 3.35521370e-01 1.49558663e-01 1.70493692e-01 -3.02003205e-01 -1.76062986e-01 -2.40894347e-01 6.20947897e-01 6.17598854e-02 -3.65324020e-01 -4.60820884e-01 7.30134845e-02 1.33265704e-01 -1.11038819e-01 -4.08420146e-01 1.39677036e+00 -2.19051749e-01 3.14900368e-01 3.33540142e-01 7.33849168e-01 -2.63023794e-01 -2.12189245e+00 -4.65487391e-02 -6.23600781e-01 -4.64463860e-01 -1.10467318e-02 -6.16711617e-01 -1.34820294e+00 1.13765550e+00 3.93692881e-01 3.94346595e-01 1.02745223e+00 -1.07945818e-02 8.30919087e-01 -4.22618305e-03 2.36299559e-01 -5.57577014e-01 3.60401183e-01 4.79671419e-01 9.51138198e-01 -1.03227723e+00 -3.16095680e-01 -1.27532125e-01 -9.77479160e-01 8.52874637e-01 5.75095117e-01 -2.54620641e-01 6.69385612e-01 4.14637655e-01 -2.02813119e-01 1.04325257e-01 -1.51258385e+00 -1.83270559e-01 2.38773331e-01 6.12479329e-01 4.52000350e-01 6.62267432e-02 8.19832683e-02 2.24232778e-01 -1.26154900e-01 1.53973266e-01 4.13724095e-01 7.50354469e-01 -3.14017862e-01 -1.00887418e+00 -1.79642830e-02 4.12550449e-01 -3.63501459e-01 -2.76299298e-01 -3.00366003e-02 3.81929994e-01 -2.73386966e-02 7.73274720e-01 2.91058362e-01 -5.08535445e-01 -1.91392154e-01 -4.10729535e-02 5.88882089e-01 -3.45501393e-01 -1.61040902e-01 1.41002059e-01 1.32325627e-02 -7.39990294e-01 -5.52788556e-01 -8.15552473e-01 -1.07052648e+00 -4.80548620e-01 1.22145496e-01 3.58273014e-02 3.42939019e-01 6.73077822e-01 6.30422413e-01 8.57072711e-01 3.14177752e-01 -1.33730149e+00 -1.64746329e-01 -9.63100612e-01 -1.60525292e-01 3.27915102e-01 3.45471174e-01 -6.73385382e-01 -1.49405167e-01 5.08791566e-01]
[10.69731616973877, -0.5737518668174744]
bd17cafc-cd95-4a28-ad47-66db4a33e890
probing-inter-modality-visual-parsing-with
2106.13488
null
https://arxiv.org/abs/2106.13488v4
https://arxiv.org/pdf/2106.13488v4.pdf
Probing Inter-modality: Visual Parsing with Self-Attention for Vision-Language Pre-training
Vision-Language Pre-training (VLP) aims to learn multi-modal representations from image-text pairs and serves for downstream vision-language tasks in a fine-tuning fashion. The dominant VLP models adopt a CNN-Transformer architecture, which embeds images with a CNN, and then aligns images and text with a Transformer. Visual relationship between visual contents plays an important role in image understanding and is the basic for inter-modal alignment learning. However, CNNs have limitations in visual relation learning due to local receptive field's weakness in modeling long-range dependencies. Thus the two objectives of learning visual relation and inter-modal alignment are encapsulated in the same Transformer network. Such design might restrict the inter-modal alignment learning in the Transformer by ignoring the specialized characteristic of each objective. To tackle this, we propose a fully Transformer visual embedding for VLP to better learn visual relation and further promote inter-modal alignment. Specifically, we propose a metric named Inter-Modality Flow (IMF) to measure the interaction between vision and language modalities (i.e., inter-modality). We also design a novel masking optimization mechanism named Masked Feature Regression (MFR) in Transformer to further promote the inter-modality learning. To the best of our knowledge, this is the first study to explore the benefit of Transformer for visual feature learning in VLP. We verify our method on a wide range of vision-language tasks, including Image-Text Retrieval, Visual Question Answering (VQA), Visual Entailment and Visual Reasoning. Our approach not only outperforms the state-of-the-art VLP performance, but also shows benefits on the IMF metric.
['Jiebo Luo', 'Houqiang Li', 'Jianlong Fu', 'Houwen Peng', 'Bei Liu', 'Yupan Huang', 'Hongwei Xue']
2021-06-25
probing-inter-modality-visual-parsing-with-1
http://proceedings.neurips.cc/paper/2021/hash/23fa71cc32babb7b91130824466d25a5-Abstract.html
http://proceedings.neurips.cc/paper/2021/file/23fa71cc32babb7b91130824466d25a5-Paper.pdf
neurips-2021-12
['visual-entailment']
['reasoning']
[ 6.15310743e-02 -9.19046253e-02 -3.55610102e-01 -4.07048643e-01 -5.28337181e-01 -5.09715915e-01 8.79988730e-01 -1.42671674e-01 -4.09486324e-01 1.36417150e-01 3.58389348e-01 -3.76193941e-01 3.89092416e-02 -7.48448491e-01 -8.95768285e-01 -6.53671563e-01 5.11681557e-01 -3.27949412e-02 2.01020718e-01 -2.02738330e-01 2.04072729e-01 2.08484918e-01 -1.56831849e+00 8.41781616e-01 7.14444518e-01 1.12054694e+00 5.23686528e-01 4.83084172e-01 -4.56874847e-01 1.19218087e+00 -1.22877121e-01 -4.90920514e-01 3.55270021e-02 -4.59763587e-01 -9.45755541e-01 1.39801607e-01 4.86944258e-01 -1.77932963e-01 -4.32260275e-01 1.00538123e+00 4.08435732e-01 -7.60087967e-02 8.22233319e-01 -1.54839385e+00 -1.36341202e+00 4.62056488e-01 -7.64220119e-01 2.14965299e-01 2.40458041e-01 4.40005988e-01 1.39019942e+00 -1.07810080e+00 3.17815632e-01 1.39111352e+00 5.04464924e-01 5.67234397e-01 -1.07252967e+00 -4.72843498e-01 3.12706709e-01 4.81518418e-01 -1.23216987e+00 -3.77931923e-01 9.62859571e-01 -5.11765003e-01 1.04330814e+00 2.00540885e-01 6.06903911e-01 1.04334342e+00 1.04614004e-01 1.04750621e+00 1.24280226e+00 -3.44286501e-01 -3.18434715e-01 3.25799942e-01 3.15270312e-02 9.16221797e-01 -3.55160475e-01 2.06351280e-01 -7.19797850e-01 4.89059955e-01 7.10427761e-01 2.49703955e-02 -4.85914409e-01 -2.44366646e-01 -1.42591369e+00 8.07107627e-01 6.58200145e-01 3.12566549e-01 -1.81282222e-01 1.46173745e-01 3.01060230e-01 3.23396415e-01 8.14834908e-02 2.03747135e-02 -3.69954616e-01 1.68668091e-01 -7.01114774e-01 -2.89492011e-01 4.68807846e-01 7.71781266e-01 8.57388794e-01 -1.12029130e-03 -6.03006065e-01 8.64639759e-01 7.30230093e-01 5.83144605e-01 5.24776459e-01 -7.51905024e-01 6.78559005e-01 7.77841926e-01 -4.71956760e-01 -8.93223643e-01 -2.98822552e-01 -1.64023951e-01 -1.02766526e+00 2.12623894e-01 3.88553619e-01 4.05160725e-01 -9.14590001e-01 1.92351782e+00 2.00776488e-01 -4.61810753e-02 2.73311883e-01 1.06678879e+00 1.43730962e+00 7.50859499e-01 1.80507004e-01 -1.18781224e-01 1.72028518e+00 -1.24887395e+00 -5.10175288e-01 -3.68339866e-01 2.97624737e-01 -9.81444597e-01 1.58688831e+00 -7.92634860e-02 -1.07163668e+00 -9.09016371e-01 -7.81941891e-01 -6.96964383e-01 -3.43509823e-01 1.92873865e-01 4.98119354e-01 2.81456590e-01 -1.02531838e+00 -2.58573499e-02 -5.14955223e-01 -3.12276602e-01 4.25993145e-01 2.09100634e-01 -4.47144747e-01 -1.84966803e-01 -1.32525563e+00 9.61437404e-01 2.21671537e-01 2.02818140e-01 -9.52860415e-01 -7.13002324e-01 -1.06358361e+00 -1.39820352e-02 2.75604784e-01 -1.08192074e+00 8.67923915e-01 -1.08161569e+00 -1.36683667e+00 1.24425232e+00 -3.11141253e-01 -2.17968240e-01 3.27796340e-01 8.77834931e-02 -3.17741483e-01 2.21785903e-01 7.82543719e-02 9.90278900e-01 1.17857087e+00 -1.38173437e+00 -4.89103675e-01 -2.58035928e-01 4.59668785e-01 4.43516821e-01 -4.31643426e-01 -1.75963268e-01 -8.65610600e-01 -5.73859990e-01 -5.74263260e-02 -5.64201653e-01 2.83494473e-01 1.65103987e-01 -3.54072571e-01 -3.83489758e-01 7.77983725e-01 -5.61637223e-01 8.93038690e-01 -2.11058736e+00 2.71263927e-01 -1.41075879e-01 3.32514703e-01 1.64809495e-01 -5.10183454e-01 2.36394837e-01 -7.58786872e-02 2.39484268e-03 -6.99183494e-02 -5.37669063e-01 2.49612890e-03 3.88118893e-01 -4.00945127e-01 3.66892099e-01 4.49379146e-01 1.50442338e+00 -7.29418397e-01 -8.21023464e-01 4.29417521e-01 7.25323677e-01 -5.04749060e-01 3.58108580e-01 -2.80249774e-01 5.25973439e-01 -3.27400148e-01 8.67775619e-01 5.60524642e-01 -5.91075778e-01 -2.23933086e-01 -8.87922168e-01 -1.53668985e-01 -2.44641844e-02 -6.22364700e-01 1.80269206e+00 -7.25033104e-01 6.89037919e-01 -8.83673802e-02 -1.29016030e+00 7.54540384e-01 1.37576923e-01 2.36829713e-01 -1.18349624e+00 1.85979158e-01 -2.04472855e-01 -9.73091573e-02 -8.96412373e-01 2.59359986e-01 -1.59566373e-01 2.34740391e-01 3.55381161e-01 2.37852484e-01 7.71835074e-02 -2.97406525e-03 1.40385598e-01 5.26391864e-01 1.57643229e-01 2.69529670e-01 4.77239192e-02 1.02923024e+00 -2.57455289e-01 2.32795969e-01 6.17130995e-01 -3.50518793e-01 6.13242507e-01 5.70547104e-01 -1.48906589e-01 -8.50221157e-01 -1.25176609e+00 -4.67173615e-03 1.27149260e+00 3.93061817e-01 -3.14114153e-01 -3.22671533e-01 -8.12181830e-01 -5.04771173e-02 4.82397318e-01 -7.54059315e-01 -2.68005133e-01 -3.55648935e-01 -5.20714164e-01 5.91415048e-01 6.20629251e-01 7.90826261e-01 -1.25585115e+00 -2.48863220e-01 -3.54349285e-01 -5.38215518e-01 -1.51950073e+00 -7.96469271e-01 3.05593703e-02 -4.79705751e-01 -1.14580369e+00 -6.27660334e-01 -1.17649293e+00 5.49897492e-01 3.98475736e-01 1.13957286e+00 6.73520491e-02 -4.48087566e-02 8.07921827e-01 -3.07511926e-01 -5.12284264e-02 -2.41537571e-01 -1.35820389e-01 -3.13319445e-01 2.59793550e-01 4.00950104e-01 -5.47786057e-01 -7.15051711e-01 3.34229261e-01 -9.55031097e-01 3.40156376e-01 9.30786848e-01 1.08930063e+00 7.52409637e-01 -1.88047782e-01 1.94317222e-01 -3.85236293e-01 5.28321147e-01 -2.39283517e-01 -4.14690286e-01 7.41428256e-01 -4.60818350e-01 2.82424390e-01 6.45279408e-01 -6.25414968e-01 -1.01911521e+00 -1.07717216e-01 -9.76179242e-02 -7.43107855e-01 -7.30270222e-02 4.86913502e-01 -4.66540664e-01 -2.20494956e-01 2.91758776e-01 5.99822640e-01 5.23960926e-02 -9.47766006e-02 8.79575610e-01 4.61056471e-01 6.63748860e-01 -4.28198487e-01 8.23905587e-01 6.36864185e-01 1.19593583e-01 -7.49490619e-01 -1.04041815e+00 -4.60599899e-01 -5.11918843e-01 -1.57857150e-01 1.28126621e+00 -9.91953194e-01 -1.07873642e+00 4.24413770e-01 -1.33886063e+00 -2.79849231e-01 -1.41551957e-01 4.36008781e-01 -5.43157697e-01 5.97267866e-01 -4.69987690e-01 -5.06317675e-01 -4.16167408e-01 -1.26688540e+00 1.08159196e+00 2.06287935e-01 3.31002563e-01 -1.09998643e+00 -1.56027108e-01 7.90216804e-01 2.64242709e-01 -2.40032017e-01 9.41681087e-01 -2.71091551e-01 -6.75264478e-01 4.89746541e-01 -8.48019004e-01 6.17050886e-01 -5.06798774e-02 -1.90505579e-01 -1.18251431e+00 -1.32805243e-01 7.27397501e-02 -5.58217347e-01 1.25113094e+00 3.73200178e-01 1.20113480e+00 -1.47870928e-01 -3.93340457e-03 8.75746071e-01 1.40382993e+00 -9.01741162e-02 7.28734136e-01 3.74473423e-01 1.19086325e+00 7.70140707e-01 5.16079962e-01 5.17325737e-02 8.52706194e-01 6.26023233e-01 6.19542181e-01 -4.08099592e-01 -5.72047591e-01 -2.74706125e-01 5.83810508e-01 8.68902028e-01 -1.47989253e-02 -1.88539147e-01 -8.99531424e-01 4.58687454e-01 -1.91001034e+00 -9.34048057e-01 -4.36382219e-02 1.84434068e+00 8.53304088e-01 -1.79787785e-01 -1.08645007e-01 -1.09709762e-01 5.50440967e-01 3.80896419e-01 -4.03629839e-01 -1.94748759e-01 -3.58619303e-01 1.77283362e-02 2.19528288e-01 5.60439944e-01 -1.15373170e+00 1.04927301e+00 4.94449282e+00 8.30218256e-01 -1.32842886e+00 3.05052012e-01 4.99256641e-01 1.21261463e-01 -4.63197887e-01 2.42852606e-02 -6.92789555e-01 2.72111297e-01 3.01376015e-01 2.62082607e-01 5.25420070e-01 5.47600687e-01 4.05588448e-02 1.55735597e-01 -1.22233820e+00 1.41855407e+00 4.58784461e-01 -1.31397593e+00 3.99203807e-01 -5.96315302e-02 4.43563104e-01 7.33645409e-02 3.68567765e-01 2.95693815e-01 -2.39087850e-01 -1.26707709e+00 8.07829440e-01 5.94963074e-01 8.53817284e-01 -5.30954003e-01 5.15948236e-01 7.92289376e-02 -1.55492842e+00 -5.52319586e-02 -2.73818254e-01 2.68187702e-01 2.51780987e-01 3.41248691e-01 -2.80834436e-01 5.90940595e-01 8.23814034e-01 1.04843628e+00 -7.63812125e-01 5.40635765e-01 -3.80983084e-01 2.82071590e-01 1.29606158e-01 2.44384080e-01 3.24868828e-01 -1.60557702e-01 4.55154687e-01 1.12202740e+00 -5.16454540e-02 -2.26162612e-01 1.04729019e-01 1.18704581e+00 -3.83427553e-02 5.42643629e-02 -6.40470147e-01 -1.71502054e-01 1.85472921e-01 1.30848944e+00 -3.45807791e-01 -1.30258128e-01 -9.01873648e-01 1.08218908e+00 3.33377212e-01 5.66980720e-01 -1.00172019e+00 -7.97739029e-02 5.78865230e-01 -2.01589227e-01 5.08789539e-01 8.98084487e-04 -2.60805964e-01 -1.35564590e+00 2.46372312e-01 -8.98137510e-01 3.89681220e-01 -1.02503932e+00 -1.63279545e+00 6.32222295e-01 -1.44518971e-01 -1.31354344e+00 5.22145443e-02 -8.71240735e-01 -5.33776522e-01 9.11916256e-01 -2.03326702e+00 -1.97868443e+00 -4.65504378e-01 1.35884392e+00 4.58861768e-01 -2.41963595e-01 5.10843158e-01 2.47564316e-01 -3.69146198e-01 7.01164901e-01 -3.72451842e-01 2.56149471e-01 7.34118044e-01 -1.07861376e+00 -1.29371524e-01 8.58844221e-01 4.22246575e-01 5.97778857e-01 2.53914893e-01 -2.56140023e-01 -1.54635370e+00 -1.15466666e+00 8.95206630e-01 -4.20310706e-01 8.11011970e-01 -2.65469819e-01 -9.10602391e-01 5.56937695e-01 4.69256550e-01 1.95328310e-01 5.57878673e-01 -4.23573554e-02 -8.34796011e-01 -2.40431488e-01 -6.58087313e-01 6.82789385e-01 9.53345001e-01 -1.24089777e+00 -8.09690654e-01 1.76154211e-01 9.31799948e-01 -2.16065317e-01 -8.00044119e-01 4.71326202e-01 5.82450151e-01 -1.05814075e+00 1.37419033e+00 -4.43485439e-01 6.69762254e-01 -4.90352213e-01 -4.04579878e-01 -8.98317039e-01 -2.29946777e-01 -1.85016334e-01 -1.46553174e-01 1.53430879e+00 2.68806726e-01 -5.72000682e-01 2.60998279e-01 1.50733158e-01 -6.16233535e-02 -8.31523776e-01 -7.21057057e-01 -5.10451376e-01 9.83976647e-02 -5.24604440e-01 2.70517439e-01 1.07019866e+00 -3.22984636e-01 6.61719024e-01 -5.24325967e-01 2.90767938e-01 5.81354141e-01 3.14662397e-01 5.14051318e-01 -7.21500218e-01 -4.73950535e-01 -7.09446251e-01 -2.61857569e-01 -1.18306375e+00 4.63200241e-01 -1.13111567e+00 -2.32055590e-01 -1.64263773e+00 4.84040082e-01 -8.04358125e-02 -4.26341742e-01 6.31829381e-01 -1.69011340e-01 3.84523749e-01 3.71784657e-01 2.90469974e-01 -7.35857606e-01 7.42134988e-01 1.71263337e+00 -4.30930942e-01 -3.54819968e-02 -3.86316568e-01 -7.03799129e-01 6.35366738e-01 6.32385373e-01 -6.83213025e-02 -7.12592959e-01 -6.52785063e-01 2.20830575e-01 -4.00766693e-02 8.58193517e-01 -5.65316319e-01 3.72585773e-01 -2.00595409e-01 4.07911837e-01 -6.94398522e-01 3.30699980e-01 -7.42071092e-01 -2.95366645e-01 1.79580048e-01 -4.34421360e-01 7.65680373e-02 9.14742351e-02 4.87852424e-01 -5.89126289e-01 1.15947179e-01 6.96548462e-01 -1.26294613e-01 -1.04649603e+00 4.87848461e-01 -3.90052311e-02 2.98413426e-01 8.09289813e-01 -2.09319085e-01 -7.04544365e-01 -2.64736295e-01 -5.24466813e-01 4.20252800e-01 3.89617920e-01 6.23769403e-01 9.48391140e-01 -1.38946629e+00 -5.31714499e-01 2.46073589e-01 4.81503904e-01 -1.12756707e-01 4.23022598e-01 1.29990828e+00 -2.08284467e-01 4.26948905e-01 -2.85431802e-01 -9.25256550e-01 -1.35094655e+00 7.69513786e-01 4.91237015e-01 -2.51218021e-01 -5.88653207e-01 8.84071648e-01 8.74616742e-01 -2.71652490e-01 3.08149964e-01 -2.14261144e-01 -4.96427000e-01 3.65396328e-02 5.30855894e-01 -1.99410841e-01 -2.60402858e-01 -9.91464734e-01 -4.53689933e-01 1.06839466e+00 2.19205450e-02 -5.79619855e-02 8.78989637e-01 -4.87700641e-01 -4.20453399e-01 4.72662747e-01 1.51984942e+00 -2.24137530e-01 -1.21149516e+00 -4.70133454e-01 -4.44931358e-01 -2.27246284e-01 8.02479833e-02 -6.65375054e-01 -1.37001550e+00 1.30690682e+00 5.52657247e-01 -4.87403758e-02 1.36991143e+00 3.61489713e-01 6.94414675e-01 1.98003799e-01 -2.26451576e-01 -7.51523316e-01 6.86983764e-01 6.16613925e-01 1.04442036e+00 -1.57213569e+00 -2.41043597e-01 -2.52144486e-01 -1.01388443e+00 1.10960567e+00 8.41668725e-01 2.78555542e-01 6.51159286e-01 -1.20254584e-01 2.31219172e-01 -3.33137661e-01 -6.78718626e-01 -7.48251379e-01 9.00272548e-01 7.60495663e-01 3.85628551e-01 -2.39011079e-01 9.11369221e-04 4.15646791e-01 -5.38452901e-03 -2.19669476e-01 -1.47294119e-01 5.24861395e-01 -1.18946038e-01 -1.08073831e+00 -2.23855972e-01 9.45160538e-02 -2.35887274e-01 -4.87786055e-01 -2.91826218e-01 8.41494262e-01 4.27781641e-01 9.59399819e-01 8.77967328e-02 -4.46165681e-01 1.59643814e-01 -9.45805907e-02 6.96615398e-01 -1.79020807e-01 -5.46676397e-01 9.60307941e-02 -2.36440703e-01 -7.59153485e-01 -8.90275061e-01 -1.86680153e-01 -1.20125723e+00 -1.75303966e-01 -1.57301091e-02 -3.26754719e-01 3.66581768e-01 1.08504367e+00 1.04083717e-01 6.46182358e-01 5.10655582e-01 -6.03156567e-01 -2.61800706e-01 -6.31119311e-01 -3.76854032e-01 6.93043590e-01 5.25426984e-01 -6.68661237e-01 -3.63450557e-01 1.27262950e-01]
[10.751249313354492, 1.4415854215621948]
23ce90fd-9e7c-427c-9f91-b7f7c5a4d6b2
neural-network-compression-via-effective
2206.03596
null
https://arxiv.org/abs/2206.03596v1
https://arxiv.org/pdf/2206.03596v1.pdf
Neural Network Compression via Effective Filter Analysis and Hierarchical Pruning
Network compression is crucial to making the deep networks to be more efficient, faster, and generalizable to low-end hardware. Current network compression methods have two open problems: first, there lacks a theoretical framework to estimate the maximum compression rate; second, some layers may get over-prunned, resulting in significant network performance drop. To solve these two problems, this study propose a gradient-matrix singularity analysis-based method to estimate the maximum network redundancy. Guided by that maximum rate, a novel and efficient hierarchical network pruning algorithm is developed to maximally condense the neuronal network structure without sacrificing network performance. Substantial experiments are performed to demonstrate the efficacy of the new method for pruning several advanced convolutional neural network (CNN) architectures. Compared to existing pruning methods, the proposed pruning algorithm achieved state-of-the-art performance. At the same or similar compression ratio, the new method provided the highest network prediction accuracy as compared to other methods.
['Ze Wang', 'Yilong Yin', 'Li Lian', 'Ziqi Zhou']
2022-06-07
null
null
null
null
['neural-network-compression', 'neural-network-compression']
['methodology', 'miscellaneous']
[ 3.44082892e-01 8.11791718e-02 -1.70675665e-01 -1.84859395e-01 1.59461156e-01 8.55072290e-02 1.88424569e-02 1.74180016e-01 -5.76527059e-01 6.30789876e-01 -2.20463574e-01 -3.24392736e-01 -4.61132109e-01 -8.47589791e-01 -4.54489112e-01 -5.48588753e-01 -2.81342119e-02 -5.56101725e-02 4.87952173e-01 6.11395948e-02 4.77243394e-01 6.61599934e-01 -1.63868845e+00 2.10364833e-01 8.21017623e-01 1.45371044e+00 5.70928276e-01 2.22045138e-01 -6.38530531e-04 8.33791018e-01 -5.65334618e-01 -5.20974457e-01 4.58060950e-01 -2.90048093e-01 -5.86787760e-01 -2.68165618e-01 2.37628981e-01 -6.24674082e-01 -7.48369455e-01 1.28343582e+00 2.45069101e-01 1.86625272e-01 3.86329263e-01 -1.04903460e+00 -1.28532186e-01 9.37950730e-01 -3.95946115e-01 4.23938960e-01 -3.50860119e-01 -4.09187764e-01 8.85358214e-01 -6.11807108e-01 4.32436883e-01 1.07313979e+00 6.72788441e-01 5.52061439e-01 -1.11597192e+00 -1.06813228e+00 2.23036632e-02 2.28141353e-01 -1.47081900e+00 -5.35857141e-01 8.23685229e-01 4.77519147e-02 1.03121340e+00 -2.11571017e-03 1.16700697e+00 6.53551817e-01 2.77236611e-01 4.43887055e-01 5.81022978e-01 -3.35771561e-01 3.99450958e-01 -2.53545284e-01 1.25658751e-01 9.29146409e-01 8.12946677e-01 -1.73514888e-01 -5.78716040e-01 1.37908906e-01 9.39692080e-01 1.91676229e-01 -1.85465112e-01 -4.37125079e-02 -7.67779768e-01 6.52456224e-01 5.86540878e-01 3.42821926e-01 -2.04234451e-01 3.31587166e-01 6.39729142e-01 1.47607297e-01 2.59362638e-01 3.75833094e-01 -3.21084827e-01 -2.03531340e-01 -1.33936751e+00 2.79950649e-01 8.69016707e-01 8.14378679e-01 4.55266833e-01 3.71462911e-01 1.69420332e-01 9.94120359e-01 3.23742598e-01 -8.61050040e-02 5.36153555e-01 -1.28697264e+00 6.85525954e-01 8.37511957e-01 -7.21208513e-01 -1.37343514e+00 -3.39693874e-01 -7.92539775e-01 -1.31637120e+00 2.90045172e-01 1.02586947e-01 2.03429718e-05 -5.84393978e-01 1.69171429e+00 -3.88969891e-02 1.38219208e-01 4.36646715e-02 6.05551779e-01 6.85841799e-01 4.63887244e-01 -8.64037797e-02 -2.79944211e-01 1.00318134e+00 -9.79782522e-01 -6.84261680e-01 -2.24322647e-01 5.32074392e-01 -4.66074109e-01 5.73280454e-01 4.96182710e-01 -1.22907186e+00 -4.46292579e-01 -1.59114206e+00 3.42797674e-02 -7.93005154e-02 2.17751250e-01 7.96100259e-01 6.34619653e-01 -1.04868221e+00 1.23125577e+00 -1.02326488e+00 -1.62921101e-01 9.39720273e-01 6.95126534e-01 -7.77543038e-02 -1.08236568e-02 -7.97286689e-01 6.68514431e-01 8.65705788e-01 1.01442218e-01 -6.08758867e-01 -7.20120907e-01 -5.14236391e-01 6.25911832e-01 -8.18751007e-02 -6.16108298e-01 1.14586878e+00 -8.01443696e-01 -1.48243546e+00 3.15736055e-01 -4.70559224e-02 -1.08941031e+00 1.52068287e-01 -2.48495251e-01 -1.15734607e-01 6.03315592e-01 -3.86951059e-01 9.61299062e-01 5.66111803e-01 -8.21815908e-01 -5.17686665e-01 -3.48958075e-01 -1.49250284e-01 2.25208819e-01 -1.00255263e+00 -2.16343641e-01 -4.52442199e-01 -7.81232119e-01 6.18477881e-01 -6.73006713e-01 -2.87691087e-01 4.08225328e-01 -2.14871451e-01 -2.54718252e-02 1.02594173e+00 -5.07392645e-01 1.45931029e+00 -1.97728217e+00 3.09252664e-02 2.43041813e-01 6.41391635e-01 7.14805543e-01 2.28786040e-02 7.98444822e-02 2.15248764e-02 2.62486309e-01 -1.06129512e-01 -3.20471764e-01 -5.84779501e-01 2.39748865e-01 -2.60763586e-01 2.18370602e-01 3.48675847e-02 5.17394722e-01 -6.10346913e-01 -6.16629243e-01 1.15702681e-01 6.38492942e-01 -8.17435443e-01 -1.28110692e-01 1.83071434e-01 -1.50108680e-01 -2.75542140e-01 5.60422957e-01 6.42111838e-01 -4.47891392e-02 2.67946154e-01 -4.43663120e-01 1.17019616e-01 2.64684111e-01 -9.17420566e-01 1.55609930e+00 2.91245896e-03 9.35459733e-01 7.05454424e-02 -1.32888401e+00 1.17519248e+00 1.42785072e-01 6.12841845e-01 -5.70721745e-01 5.47721326e-01 2.62936622e-01 3.23376030e-01 -8.00987631e-02 3.67862999e-01 1.52101994e-01 5.72745264e-01 1.48789287e-01 2.40565151e-01 1.91520467e-01 3.91467035e-01 1.26077756e-01 1.31284475e+00 -4.24594194e-01 9.41596404e-02 -3.79811555e-01 2.87228405e-01 -3.18511397e-01 7.67152131e-01 5.84468007e-01 -3.07252705e-01 3.10419470e-01 6.67387068e-01 -4.20567483e-01 -1.17116511e+00 -7.39702165e-01 -1.69460803e-01 5.86703062e-01 6.46945536e-02 -7.54840374e-01 -1.12674379e+00 -1.41139001e-01 -3.95659149e-01 2.02254966e-01 -8.48565921e-02 -4.66176212e-01 -4.44327027e-01 -5.73166788e-01 9.75259244e-01 3.73149693e-01 1.23128557e+00 -8.81343424e-01 -8.35177183e-01 2.64549881e-01 -1.26468882e-01 -1.28720379e+00 1.41588356e-02 2.63019621e-01 -1.91257429e+00 -9.96943772e-01 -4.87750620e-01 -9.47495997e-01 8.75583589e-01 4.17336673e-01 6.03780389e-01 5.13416648e-01 -2.19618589e-01 -4.22210693e-01 -2.31184229e-01 -2.72333443e-01 -1.46831393e-01 3.17503363e-01 6.76281601e-02 -4.30564135e-01 2.31710508e-01 -1.02747786e+00 -7.34798253e-01 -1.18861072e-01 -9.46557581e-01 3.50788891e-01 8.95466089e-01 7.61851370e-01 6.93476200e-01 7.38815069e-01 5.99798441e-01 -6.21727288e-01 8.65137160e-01 -7.52880722e-02 -5.90199590e-01 -1.02681033e-01 -1.02386606e+00 1.88941196e-01 8.62984598e-01 -5.57073891e-01 -7.17123449e-01 1.45547971e-01 -2.00205237e-01 -5.93004167e-01 2.15787798e-01 7.02997506e-01 -5.16304793e-03 -4.71028984e-01 4.71288919e-01 1.57085255e-01 1.73495308e-01 -4.90333825e-01 -1.81462467e-01 2.29867727e-01 5.80630898e-01 -2.74719507e-01 4.33783472e-01 2.36835986e-01 4.60481107e-01 -9.01188076e-01 -5.35973728e-01 -9.14998651e-02 -6.01157308e-01 -1.79613039e-01 4.06167805e-01 -8.61518264e-01 -6.49942517e-01 5.77524066e-01 -1.06310463e+00 -1.98064998e-01 -2.75725294e-02 5.69419801e-01 -3.15137148e-01 5.06411433e-01 -9.59236205e-01 -6.58935964e-01 -8.04622173e-01 -1.01404500e+00 3.02565157e-01 2.89326161e-01 1.22411445e-01 -5.81770182e-01 -3.32169443e-01 1.99449092e-01 5.54666340e-01 7.25180805e-02 1.15347207e+00 -4.07149702e-01 -5.24360716e-01 -2.74173230e-01 -5.32312036e-01 6.23575985e-01 -3.50482881e-01 1.66264623e-01 -4.74306643e-01 -3.04871976e-01 2.48186007e-01 -2.41232872e-01 1.00856316e+00 3.62573713e-01 1.70405221e+00 -6.06404483e-01 -3.06191832e-01 8.92384768e-01 1.57023323e+00 1.51470721e-01 8.33536983e-01 3.84677917e-01 3.68015438e-01 1.58820510e-01 9.68548059e-02 4.70799565e-01 -1.27647564e-01 3.24717671e-01 6.46110833e-01 2.34402329e-01 -2.99710929e-01 -2.81304568e-01 1.16266310e-01 1.39416790e+00 -1.45075828e-01 -1.09173162e-02 -6.71183348e-01 2.92732358e-01 -1.72237790e+00 -9.23960209e-01 2.28181332e-01 2.04730105e+00 8.31985414e-01 6.08693004e-01 -2.45380342e-01 6.40828371e-01 6.93947911e-01 -6.02836125e-02 -6.94321394e-01 -3.71733367e-01 -2.38529276e-02 3.16779494e-01 6.38628125e-01 -6.13363236e-02 -7.95078158e-01 7.85580695e-01 6.90464401e+00 1.07701004e+00 -1.04925072e+00 -4.96361554e-02 6.96564376e-01 -4.32929158e-01 1.98563978e-01 -3.35495546e-02 -9.86820877e-01 2.83619970e-01 1.14650273e+00 -7.60670111e-04 6.34780228e-01 1.03660369e+00 5.81249222e-02 4.19078954e-02 -7.41390228e-01 1.16028786e+00 -1.62566066e-01 -1.66577065e+00 3.53576779e-01 2.39109725e-01 5.93793869e-01 1.58609539e-01 -2.03946784e-01 1.95987120e-01 -1.18131891e-01 -8.30023944e-01 6.80139780e-01 2.88078815e-01 7.10249186e-01 -1.16043603e+00 7.56517231e-01 4.77815598e-01 -1.17230284e+00 -4.16512251e-01 -1.02920294e+00 -3.60802293e-01 -1.62904397e-01 8.97765219e-01 -4.75735337e-01 1.04439799e-02 6.91569686e-01 6.84904099e-01 -6.31025195e-01 1.44173789e+00 2.11209003e-02 6.62533879e-01 -4.67221171e-01 -1.81839153e-01 1.68770533e-02 -2.04941630e-01 4.11715806e-01 9.99851882e-01 5.58192432e-01 5.90798147e-02 -3.04550618e-01 8.74662936e-01 -5.66115141e-01 -1.16999224e-02 -5.53037941e-01 -2.52183408e-01 9.08301055e-01 1.28341544e+00 -1.13629770e+00 -1.23878479e-01 -2.42618263e-01 6.47767723e-01 7.42895782e-01 2.40923315e-01 -6.36901677e-01 -7.25083768e-01 4.13092613e-01 1.48622230e-01 2.98129469e-01 -3.77227932e-01 -6.81302667e-01 -9.47342455e-01 6.83451071e-03 -5.74102700e-01 -4.34663370e-02 -4.59328264e-01 -6.70717835e-01 6.53893352e-01 -6.55779094e-02 -1.07406020e+00 1.15536392e-01 -7.24262357e-01 -5.25454640e-01 3.00327718e-01 -1.21358395e+00 -5.14182866e-01 -3.16811889e-01 4.37736362e-01 5.08424997e-01 -5.47085941e-01 8.20158780e-01 6.05581880e-01 -8.89945149e-01 8.77891839e-01 1.60528973e-01 6.25174642e-02 -3.66360508e-02 -4.49587286e-01 -7.99418315e-02 1.01439631e+00 8.11758414e-02 7.97520578e-01 3.63773078e-01 -4.29699868e-01 -1.24519920e+00 -1.11108410e+00 6.56429112e-01 5.87024331e-01 4.12344158e-01 -2.79464334e-01 -8.52341354e-01 3.67802233e-01 -5.43167740e-02 -1.37580216e-01 5.65940678e-01 -1.32678792e-01 -3.52769375e-01 -5.90862274e-01 -1.31230140e+00 7.50160873e-01 1.03008533e+00 -1.87038109e-01 -4.07730490e-01 3.38834561e-02 9.09189105e-01 -3.08938026e-01 -8.31567407e-01 5.79431474e-01 6.99697196e-01 -1.00590920e+00 9.46072638e-01 -7.83819929e-02 8.49157214e-01 1.06427997e-01 -2.45011657e-01 -7.47662783e-01 -3.41124326e-01 -4.33374047e-01 -8.98471355e-01 1.13326943e+00 3.27321202e-01 -4.97568011e-01 1.11623681e+00 5.48012316e-01 -1.87697858e-01 -1.35675335e+00 -1.08246768e+00 -8.63235533e-01 -1.70304030e-01 -3.66848767e-01 3.21490169e-01 4.37314421e-01 1.21893279e-01 2.96832174e-01 -1.35229155e-01 -1.64712563e-01 6.77578509e-01 -4.22942787e-01 2.80118167e-01 -1.39182687e+00 -6.00945838e-02 -9.62231696e-01 -7.14037538e-01 -1.14375710e+00 6.62174821e-02 -9.02116060e-01 -2.33057693e-01 -1.45391595e+00 2.51255065e-01 -3.91973048e-01 -4.03239548e-01 4.03403044e-01 3.48819077e-01 2.93520123e-01 2.74909258e-01 4.07858074e-01 -4.46083277e-01 6.60625517e-01 1.10209286e+00 5.66724353e-02 -1.62328824e-01 9.68503114e-03 -8.11360002e-01 7.55721569e-01 1.08121145e+00 -5.93956590e-01 -8.01719308e-01 -5.92939138e-01 1.03903629e-01 -6.90664947e-02 1.86387032e-01 -1.84432006e+00 6.49717391e-01 2.64609724e-01 4.61092472e-01 -7.55444884e-01 2.98399121e-01 -8.73917341e-01 -1.08631372e-01 8.37108612e-01 -4.38824087e-01 -1.57780293e-03 3.47463638e-01 5.36154509e-01 -2.59938478e-01 -7.24840641e-01 9.77160215e-01 3.33520696e-02 -3.54963452e-01 4.37318027e-01 -5.12164056e-01 -2.99618274e-01 7.29657054e-01 -4.37617034e-01 -3.35618943e-01 -7.68304467e-02 -2.79320627e-01 -1.28225014e-01 8.59082714e-02 1.26997381e-01 1.07816398e+00 -1.27244020e+00 -2.11236611e-01 2.58375466e-01 -5.19470870e-01 1.91267788e-01 1.36465475e-01 6.04340553e-01 -1.01214290e+00 5.70658684e-01 -5.19160807e-01 -2.29140759e-01 -1.29520786e+00 3.76848340e-01 1.51410371e-01 -4.95172858e-01 -8.25855613e-01 8.44673812e-01 -3.16870332e-01 -8.52132365e-02 6.98561132e-01 -3.90252084e-01 -2.09622607e-01 -3.55072379e-01 6.49844527e-01 5.46377957e-01 2.00222433e-01 -2.91147798e-01 -8.36561713e-03 2.55467087e-01 -2.01742038e-01 1.13971010e-01 1.45070755e+00 3.47216576e-02 -3.39437425e-01 -6.81021810e-02 1.28714883e+00 -7.42842913e-01 -1.17902827e+00 -1.75642326e-01 -8.02334473e-02 -2.74267137e-01 5.02477646e-01 -2.95213282e-01 -1.73564219e+00 8.48938167e-01 5.87163568e-01 2.60733590e-02 1.48510838e+00 -6.07858241e-01 1.06501496e+00 8.83669257e-01 3.47182184e-01 -1.16622961e+00 1.58543944e-01 6.03111207e-01 5.54308772e-01 -7.05905914e-01 4.84939933e-01 -4.65733081e-01 -8.60508457e-02 1.46363235e+00 8.78894091e-01 -3.93039972e-01 9.15131807e-01 3.36699963e-01 -6.68380439e-01 -3.16091120e-01 -8.14497471e-01 2.90353239e-01 1.02920309e-02 4.67197835e-01 3.12689245e-01 -3.10461879e-01 -5.90461791e-01 6.36358500e-01 -4.37982678e-01 1.79782555e-01 4.32237089e-01 8.36178422e-01 -8.51111770e-01 -8.72036517e-01 2.08015680e-01 9.70647752e-01 -8.29414427e-01 -2.17007220e-01 -6.74658269e-02 6.49198115e-01 -6.96375668e-02 8.98970664e-01 6.21783882e-02 -7.47690916e-01 1.01663068e-01 -1.74887791e-01 4.80550051e-01 -3.69004726e-01 -3.84843200e-01 8.71695462e-04 -1.85241267e-01 -5.46955109e-01 -3.49755496e-01 -1.42821342e-01 -1.36835802e+00 -7.28612602e-01 -5.98106444e-01 -1.29888579e-01 8.73338461e-01 8.87042165e-01 4.60910380e-01 6.92158818e-01 2.13743880e-01 -9.10113692e-01 -4.02885556e-01 -8.82307470e-01 -5.36557734e-01 5.09087443e-02 -1.04090206e-01 -5.79741240e-01 -2.06784472e-01 -1.62083283e-01]
[8.532817840576172, 3.08404541015625]
40e71700-b99c-4768-8a68-a143dcfc0a9c
stip-a-spatiotemporal-information-preserving
2206.04381
null
https://arxiv.org/abs/2206.04381v1
https://arxiv.org/pdf/2206.04381v1.pdf
STIP: A SpatioTemporal Information-Preserving and Perception-Augmented Model for High-Resolution Video Prediction
Although significant achievements have been achieved by recurrent neural network (RNN) based video prediction methods, their performance in datasets with high resolutions is still far from satisfactory because of the information loss problem and the perception-insensitive mean square error (MSE) based loss functions. In this paper, we propose a Spatiotemporal Information-Preserving and Perception-Augmented Model (STIP) to solve the above two problems. To solve the information loss problem, the proposed model aims to preserve the spatiotemporal information for videos during the feature extraction and the state transitions, respectively. Firstly, a Multi-Grained Spatiotemporal Auto-Encoder (MGST-AE) is designed based on the X-Net structure. The proposed MGST-AE can help the decoders recall multi-grained information from the encoders in both the temporal and spatial domains. In this way, more spatiotemporal information can be preserved during the feature extraction for high-resolution videos. Secondly, a Spatiotemporal Gated Recurrent Unit (STGRU) is designed based on the standard Gated Recurrent Unit (GRU) structure, which can efficiently preserve spatiotemporal information during the state transitions. The proposed STGRU can achieve more satisfactory performance with a much lower computation load compared with the popular Long Short-Term (LSTM) based predictive memories. Furthermore, to improve the traditional MSE loss functions, a Learned Perceptual Loss (LP-loss) is further designed based on the Generative Adversarial Networks (GANs), which can help obtain a satisfactory trade-off between the objective quality and the perceptual quality. Experimental results show that the proposed STIP can predict videos with more satisfactory visual quality compared with a variety of state-of-the-art methods. Source code has been available at \url{https://github.com/ZhengChang467/STIPHR}.
['Wen Gao', 'Siwei Ma', 'Shanshe Wang', 'Xinfeng Zhang', 'Zheng Chang']
2022-06-09
null
null
null
null
['video-prediction']
['computer-vision']
[ 2.17259690e-01 -3.73169541e-01 -1.05729528e-01 -8.33389908e-02 -7.65215337e-01 2.25453839e-01 2.99583197e-01 -4.26781505e-01 -1.99461922e-01 7.50988126e-01 1.97570190e-01 6.86640367e-02 -8.58021900e-02 -8.84772480e-01 -9.66802537e-01 -9.87752914e-01 2.47425511e-01 -3.83256733e-01 4.19100404e-01 -1.04742616e-01 5.49595021e-02 1.54626355e-01 -1.50788724e+00 2.66222298e-01 9.89514887e-01 1.35028100e+00 7.10348964e-01 3.39827210e-01 1.56281352e-01 1.10626459e+00 -2.51747102e-01 -2.26336062e-01 -2.48315465e-02 -6.51041806e-01 -2.29394734e-01 -1.40684769e-01 3.04632308e-03 -3.59153688e-01 -8.97963405e-01 1.15872598e+00 5.80275118e-01 4.10275549e-01 3.10968488e-01 -1.04974914e+00 -7.96508789e-01 2.64190704e-01 -4.71249610e-01 1.31312668e-01 1.09860033e-01 2.33998299e-01 7.08101630e-01 -7.97767103e-01 4.19356495e-01 1.06080520e+00 4.23628807e-01 5.24500608e-01 -7.96491206e-01 -9.21353638e-01 2.39879698e-01 7.14389920e-01 -1.55919993e+00 -4.27313447e-01 9.29451525e-01 -1.96074143e-01 7.14019835e-01 2.47924596e-01 5.93378365e-01 1.18344581e+00 4.70134079e-01 8.65112960e-01 8.89702380e-01 -4.63644974e-03 -6.62136376e-02 -4.54321466e-02 -4.73826945e-01 7.61545002e-01 -3.47430676e-01 2.99980491e-01 -4.36537057e-01 5.23275733e-01 1.24802577e+00 4.39825863e-01 -6.54503345e-01 -9.00805518e-02 -1.18693328e+00 5.37361622e-01 7.38756061e-01 5.71400821e-01 -4.87193614e-01 8.64045843e-02 4.68450844e-01 2.07736671e-01 4.13128495e-01 -7.96385333e-02 -2.10534289e-01 -1.85203537e-01 -9.23484981e-01 -1.14632763e-01 7.24162757e-02 9.45348442e-01 6.00365460e-01 5.32578707e-01 -5.35147429e-01 8.63818705e-01 2.89532185e-01 4.48823243e-01 7.70044625e-01 -8.99170101e-01 6.75173163e-01 2.63273507e-01 6.51650429e-02 -1.38172722e+00 1.96733028e-02 -6.09789312e-01 -1.32037854e+00 9.66689661e-02 -1.98072836e-01 -1.51251424e-02 -8.46891224e-01 1.82747185e+00 -8.23020488e-02 5.53655982e-01 1.47254154e-01 1.02655661e+00 8.64640594e-01 1.34463179e+00 9.77705345e-02 -6.02021813e-01 9.60131347e-01 -1.00806868e+00 -9.13857639e-01 -6.02534525e-02 4.05207016e-02 -5.71915388e-01 1.07452118e+00 2.34587833e-01 -1.18272007e+00 -1.07123017e+00 -1.11041570e+00 -3.77559625e-02 -6.68978542e-02 1.61206096e-01 1.47218376e-01 1.00018688e-01 -9.58380044e-01 5.72054148e-01 -9.17942107e-01 1.57741737e-02 3.73292506e-01 2.15192154e-01 -1.49937272e-01 -2.27185681e-01 -1.38611758e+00 6.49106681e-01 4.20363247e-01 3.14037949e-01 -9.68235791e-01 -4.90208238e-01 -9.42500353e-01 1.70496866e-01 4.93383288e-01 -5.31611264e-01 7.89326966e-01 -9.97841835e-01 -1.55642939e+00 1.67680532e-01 -1.91685095e-01 -3.59544724e-01 3.87277514e-01 -9.67651159e-02 -6.43019080e-01 1.98158696e-01 -2.50110216e-02 5.48644960e-01 9.40552950e-01 -1.15066433e+00 -6.31128192e-01 -1.83794603e-01 -2.62471110e-01 4.29635197e-01 -5.15375316e-01 -7.90426582e-02 -6.61665082e-01 -1.16274512e+00 -2.22003683e-02 -7.01416016e-01 -3.94332074e-02 -1.89984888e-01 -2.53932089e-01 8.72784927e-02 9.68504965e-01 -1.10418773e+00 1.39333701e+00 -2.24774218e+00 3.46890748e-01 -1.86194286e-01 -5.25851436e-02 5.81705749e-01 -1.62971094e-01 1.18199907e-01 1.04823979e-02 -1.77595727e-02 -1.57182127e-01 -2.13411406e-01 -2.08921194e-01 1.49615437e-01 -3.99983436e-01 1.10375278e-01 -3.75498906e-02 9.88740742e-01 -7.37354040e-01 -4.22797978e-01 5.56220472e-01 9.00394499e-01 -2.21112192e-01 4.53023881e-01 -2.34360486e-01 7.02704191e-01 -5.34740567e-01 3.95429730e-01 5.65826356e-01 -1.28075108e-01 -2.49259442e-01 -4.21048403e-01 -2.02215537e-01 -3.71374562e-02 -9.00264263e-01 1.80763972e+00 -6.89945161e-01 4.22063887e-01 -1.59539148e-01 -9.87223744e-01 9.77403343e-01 5.56152284e-01 4.08092588e-01 -1.24190176e+00 7.86999613e-02 1.08858041e-01 -3.48028213e-01 -4.75250006e-01 2.80419886e-01 -1.97010636e-01 1.32082760e-01 -1.32642061e-01 2.38573886e-02 5.20151198e-01 -1.71887398e-01 -7.36424774e-02 8.18838656e-01 4.64895546e-01 -9.79365483e-02 2.27104127e-01 8.43953192e-01 -5.25141120e-01 1.04375553e+00 2.21938252e-01 1.86379801e-03 6.34012878e-01 1.83143076e-02 -2.90577590e-01 -1.05489004e+00 -9.32598233e-01 1.05314389e-01 6.66793942e-01 5.55676579e-01 -2.32618213e-01 -5.01128793e-01 -3.60686749e-01 -5.42141497e-01 7.04803050e-01 -3.49221438e-01 -5.18801689e-01 -5.99765241e-01 -3.56460690e-01 3.36855710e-01 6.64097250e-01 1.18939638e+00 -1.37529719e+00 -4.30084735e-01 4.09189731e-01 -4.74031270e-01 -1.05874789e+00 -6.01035416e-01 -3.92931223e-01 -7.37673461e-01 -5.36748767e-01 -1.06441748e+00 -7.74353027e-01 3.45435977e-01 1.63207188e-01 5.41895449e-01 -1.57323733e-01 1.07349969e-01 -1.61495693e-02 -5.32482266e-01 1.54524505e-01 -1.48336828e-01 -2.54550844e-01 -6.27615005e-02 2.73277104e-01 7.26354346e-02 -8.23157012e-01 -7.75487006e-01 4.34751868e-01 -1.11405385e+00 4.23703492e-01 8.04397941e-01 9.28531826e-01 9.57720041e-01 4.67209250e-01 6.01953983e-01 -3.89670342e-01 1.93344906e-01 -4.59565043e-01 -5.13859749e-01 2.56937802e-01 -6.01699889e-01 -8.04546177e-02 1.06271029e+00 -4.76381898e-01 -1.34637332e+00 -2.74808526e-01 -4.05357182e-01 -1.08373392e+00 3.29549611e-02 4.53098506e-01 -5.00765264e-01 -2.25640368e-02 -1.00769401e-01 9.85433817e-01 -1.20526060e-01 -5.06604314e-01 1.09929539e-01 4.82994467e-01 6.08219624e-01 -2.18691707e-01 5.44128239e-01 8.82011130e-02 -4.96429093e-02 -5.38509011e-01 -6.79768682e-01 -2.04577181e-03 -1.27600148e-01 -2.49105960e-01 1.02476728e+00 -1.15256751e+00 -6.17528617e-01 6.83557570e-01 -9.39714968e-01 -2.89141715e-01 -6.49487153e-02 7.06594825e-01 -7.93522775e-01 3.65341365e-01 -9.21762824e-01 -7.04973519e-01 -3.85093659e-01 -1.24781132e+00 6.90249741e-01 5.50352395e-01 5.12888134e-01 -8.00828099e-01 -3.32039535e-01 3.79043132e-01 4.59347576e-01 3.99695337e-01 7.33213067e-01 -1.15256138e-01 -9.49161410e-01 -7.53728766e-03 -2.68499702e-01 6.21816814e-01 2.26297323e-02 -3.40167612e-01 -6.02368891e-01 -3.92195016e-01 2.66880691e-01 -1.60879701e-01 1.00715709e+00 4.89626825e-01 1.44444442e+00 -6.09444976e-01 -1.46291599e-01 8.54385138e-01 1.65444791e+00 6.77663028e-01 1.14120293e+00 3.73420656e-01 1.02570248e+00 1.56749636e-01 9.28732336e-01 3.81067306e-01 3.41860175e-01 8.10644805e-01 4.61424619e-01 -9.01899580e-03 -1.97909594e-01 -6.78158462e-01 6.22068584e-01 1.07475984e+00 -2.61912346e-01 -3.08459073e-01 -4.17531848e-01 4.13650334e-01 -2.08473349e+00 -1.10525274e+00 2.96764433e-01 2.21721053e+00 7.46318340e-01 1.12516388e-01 -2.22631767e-01 2.20121995e-01 7.81369209e-01 4.65828836e-01 -7.00420320e-01 -1.88284740e-01 -1.57527268e-01 -2.28804182e-02 2.35450521e-01 2.79136986e-01 -9.75371480e-01 7.54436731e-01 4.05932665e+00 1.44451773e+00 -1.25801659e+00 2.33760148e-01 7.47192562e-01 -7.67183080e-02 -2.20837742e-01 -1.75032392e-01 -5.53259134e-01 1.04214478e+00 8.83370578e-01 1.17326630e-02 5.80486953e-01 6.37368143e-01 4.96322721e-01 2.08607197e-01 -6.21762633e-01 1.19082820e+00 8.89379606e-02 -1.20343089e+00 2.40973115e-01 -6.85576946e-02 6.42198801e-01 -2.96642452e-01 4.52585876e-01 3.43406230e-01 -2.87998259e-01 -1.05005515e+00 5.87573767e-01 1.13541985e+00 1.27538300e+00 -1.19991696e+00 8.21286619e-01 4.34662819e-01 -1.53774452e+00 -3.04680228e-01 -4.88627046e-01 2.71830052e-01 2.23239094e-01 5.01727760e-01 9.94258076e-02 8.21840942e-01 8.24390709e-01 1.04214430e+00 -2.49312729e-01 8.39422882e-01 -3.05662215e-01 4.45021361e-01 9.61892083e-02 2.34975994e-01 2.67261714e-01 -4.03996468e-01 7.63566852e-01 8.56242418e-01 5.54078341e-01 3.77196401e-01 2.38061715e-02 8.11138690e-01 -1.11615673e-01 5.15848659e-02 -4.11253452e-01 1.64493799e-01 3.73865157e-01 1.03561890e+00 -2.00714156e-01 -2.90919721e-01 -4.24497366e-01 1.24101019e+00 9.34798568e-02 5.31433582e-01 -1.06745684e+00 -5.58273077e-01 4.37639803e-01 1.42325684e-01 6.73459351e-01 -1.19079417e-02 1.44539207e-01 -1.28867555e+00 4.33739573e-01 -8.13805044e-01 2.40446955e-01 -1.08167470e+00 -9.60956395e-01 6.95119917e-01 -2.60199636e-01 -1.63629615e+00 -1.30384609e-01 -1.70201492e-02 -6.07090771e-01 9.47004855e-01 -1.59714460e+00 -1.28386378e+00 -3.46523851e-01 8.95524621e-01 6.57664180e-01 -2.19079301e-01 5.27476311e-01 5.32402337e-01 -7.60092139e-01 6.64074898e-01 2.87259609e-01 9.06208232e-02 5.15216708e-01 -6.84527397e-01 -1.81519523e-01 1.10535872e+00 -1.42915621e-01 3.45081538e-01 3.82455081e-01 -5.66014826e-01 -1.21237528e+00 -1.53095055e+00 5.05334020e-01 2.58975357e-01 2.99104363e-01 4.32022251e-02 -1.04397905e+00 6.11972153e-01 1.78340942e-01 3.74999456e-02 2.38154247e-01 -5.26227236e-01 -1.78753793e-01 -4.61274832e-01 -1.00107241e+00 4.96073276e-01 8.89747024e-01 -5.29042721e-01 -2.65009612e-01 -4.57359590e-02 9.54288363e-01 -3.44860256e-01 -1.05622053e+00 8.14683139e-01 4.80556011e-01 -1.08661664e+00 9.15049493e-01 -7.19507560e-02 7.68553674e-01 -5.13123453e-01 -2.66128391e-01 -1.17906213e+00 -4.57467228e-01 -3.89037639e-01 -3.46513599e-01 1.38469470e+00 1.60223816e-03 -5.85902870e-01 4.48338509e-01 3.98483604e-01 -2.10044175e-01 -1.14950049e+00 -9.10155594e-01 -7.82895625e-01 -2.54498333e-01 -2.10594490e-01 3.80649149e-01 6.05368018e-01 -3.33582610e-01 2.54556984e-01 -9.15410042e-01 1.65997475e-01 5.60566247e-01 1.98802024e-01 2.78031260e-01 -6.03944600e-01 -3.38115692e-01 -2.62408704e-01 -5.13720870e-01 -1.26889086e+00 4.97757923e-04 -4.75232005e-01 6.13015592e-02 -1.54393458e+00 2.26673111e-01 -2.83585548e-01 -7.79128313e-01 3.58699024e-01 -2.08941311e-01 2.02750936e-01 3.47237915e-01 2.31986940e-01 -7.54042506e-01 1.20450628e+00 1.53747761e+00 -7.50514641e-02 -1.38970047e-01 4.76111472e-02 -4.30994332e-01 5.27289510e-01 7.02061653e-01 -2.16363698e-01 -5.33051968e-01 -3.88564020e-01 -1.49701372e-01 6.81712925e-01 4.90775168e-01 -1.33540201e+00 3.02692533e-01 -8.96886364e-02 6.37599647e-01 -7.18339503e-01 6.27862453e-01 -8.32010329e-01 3.68367523e-01 4.30671096e-01 -2.61277288e-01 -3.87172550e-02 8.25666338e-02 7.46479511e-01 -5.83705723e-01 3.31724167e-01 9.15711761e-01 -7.83527642e-02 -9.17067409e-01 7.75324762e-01 -7.01812655e-02 -2.70806730e-01 1.05262375e+00 -2.37550169e-01 -1.35774985e-01 -4.95195627e-01 -7.16125071e-01 1.72631606e-01 4.01940584e-01 5.45304000e-01 1.12680697e+00 -1.63362050e+00 -6.76222205e-01 1.65384769e-01 -6.82268143e-02 -9.25678164e-02 8.75855505e-01 8.26955318e-01 -2.27099717e-01 3.88849884e-01 -4.67624187e-01 -3.52048695e-01 -1.02608097e+00 7.61310399e-01 2.27585614e-01 -3.69173408e-01 -7.32077599e-01 7.03072965e-01 3.67908388e-01 7.24367127e-02 1.34478450e-01 5.75239435e-02 -4.16205168e-01 -3.19389135e-01 6.06023133e-01 3.40485275e-01 -2.90523440e-01 -8.34132016e-01 -1.18046612e-01 6.29901588e-01 6.41714334e-02 6.76706284e-02 1.39270389e+00 -3.94185662e-01 5.03467731e-02 4.17625189e-01 1.39160848e+00 -1.88461527e-01 -1.58036375e+00 -3.97614062e-01 -6.82331443e-01 -6.26965821e-01 2.09925070e-01 -6.21555030e-01 -1.44772351e+00 9.10512567e-01 8.16203475e-01 -1.41772166e-01 1.69749808e+00 -3.73680651e-01 1.42301595e+00 -1.65289432e-01 3.67893845e-01 -8.51020515e-01 1.83418751e-01 2.97921002e-01 9.30996478e-01 -9.69577551e-01 -2.44311169e-01 -1.59595266e-01 -8.24931085e-01 8.35259736e-01 7.28670776e-01 -2.30103537e-01 5.02602577e-01 -3.72655578e-02 -2.99943388e-01 2.53442585e-01 -7.14214027e-01 -8.55175927e-02 3.81279111e-01 4.30150211e-01 1.17458753e-01 -1.71512626e-02 -2.53193021e-01 8.24797511e-01 1.57140195e-01 4.07918207e-02 1.08569607e-01 4.87965167e-01 -3.29108596e-01 -8.95302296e-01 -8.96154121e-02 4.79681909e-01 -4.43889886e-01 -2.16352984e-01 4.08074498e-01 4.12164509e-01 3.92046988e-01 8.68553698e-01 -1.38756931e-01 -8.11021388e-01 1.91772372e-01 -3.21905702e-01 2.98882514e-01 -2.41992608e-01 -7.18714446e-02 2.16280296e-01 -2.73228616e-01 -8.59899759e-01 -5.21088660e-01 -4.31890190e-01 -1.18028343e+00 -2.15372086e-01 -1.38757467e-01 1.05685465e-01 2.51470625e-01 8.77125680e-01 4.08851266e-01 8.63852501e-01 8.89674842e-01 -6.82650387e-01 -1.30698755e-01 -8.70946944e-01 -6.12125754e-01 3.19070548e-01 2.64170915e-01 -6.17584467e-01 -1.23361401e-01 1.09400563e-01]
[11.085329055786133, -1.746423602104187]
0e0c0dfe-6ea2-4acd-ab42-f00382730a0b
clickbait-detection-with-style-aware-title
null
null
https://aclanthology.org/2020.ccl-1.106
https://aclanthology.org/2020.ccl-1.106.pdf
Clickbait Detection with Style-aware Title Modeling and Co-attention
Clickbait is a form of web content designed to attract attention and entice users to click on specific hyperlinks. The detection of clickbaits is an important task for online platforms to improve the quality of web content and the satisfaction of users. Clickbait detection is typically formed as a binary classification task based on the title and body of a webpage, and existing methods are mainly based on the content of title and the relevance between title and body. However, these methods ignore the stylistic patterns of titles, which can provide important clues on identifying clickbaits. In addition, they do not consider the interactions between the contexts within title and body, which are very important for measuring their relevance for clickbait detection. In this paper, we propose a clickbait detection approach with style-aware title modeling and co-attention. Specifically, we use Transformers to learn content representations of title and body, and respectively compute two content-based clickbait scores for title and body based on their representations. In addition, we propose to use a character-level Transformer to learn a style-aware title representation by capturing the stylistic patterns of title, and we compute a title stylistic score based on this representation. Besides, we propose to use a co-attention network to model the relatedness between the contexts within title and body, and further enhance their representations by encoding the interaction information. We compute a title-body matching score based on the representations of title and body enhanced by their interactions. The final clickbait score is predicted by a weighted summation of the aforementioned four kinds of scores. Extensive experiments on two benchmark datasets show that our approach can effectively improve the performance of clickbait detection and consistently outperform many baseline methods.
['Yongfeng Huang', 'Tao Qi', 'Fangzhao Wu', 'Chuhan Wu']
null
null
null
null
ccl-2020-10
['clickbait-detection']
['natural-language-processing']
[-2.93327756e-02 -6.28750682e-01 -5.93000948e-01 -4.70175415e-01 -6.25028729e-01 -4.88964409e-01 6.75601840e-01 3.16174120e-01 -2.45310083e-01 2.40669683e-01 3.42802078e-01 -2.72379577e-01 -1.71301156e-01 -8.54106367e-01 -6.65365458e-01 -2.71469206e-01 2.61488527e-01 1.15575239e-01 6.06272340e-01 -2.50281513e-01 4.97670084e-01 -1.15674168e-01 -1.56020463e+00 2.87910789e-01 1.22540069e+00 1.67315710e+00 3.04713696e-01 4.93812025e-01 -6.47393167e-01 5.89165092e-01 -5.67971885e-01 -2.52960384e-01 -5.65515310e-02 -4.79696900e-01 -4.64129776e-01 1.07390530e-01 7.75466859e-01 -4.12933439e-01 -6.53439283e-01 1.12089980e+00 1.17197642e-02 -9.76446643e-02 6.11814678e-01 -9.92759883e-01 -6.49857283e-01 5.35709083e-01 -6.62323296e-01 4.60078895e-01 3.41000617e-01 5.21939881e-02 1.75012279e+00 -7.30506420e-01 1.55307412e-01 1.18642724e+00 3.17915976e-01 7.76880160e-02 -8.83195162e-01 -6.34055257e-01 5.81748962e-01 3.92734557e-01 -1.14260054e+00 8.81068408e-02 8.76226366e-01 -4.85271513e-01 2.40599886e-01 5.91991782e-01 5.99322379e-01 9.29597497e-01 -1.46882370e-01 1.39804995e+00 7.08357692e-01 -2.06258297e-01 -1.05986409e-01 7.84367695e-02 6.56467617e-01 6.22459948e-01 1.34926468e-01 -3.15916330e-01 -4.07692105e-01 -3.26291323e-02 6.32824540e-01 4.17881787e-01 -2.43401513e-01 -1.21037066e-01 -1.03597546e+00 9.17432010e-01 8.08533609e-01 2.27025598e-01 -2.31258795e-01 2.09549785e-01 3.48023474e-01 -1.06374234e-01 4.35398132e-01 2.64502048e-01 -1.25516087e-01 -1.05214633e-01 -9.16133702e-01 4.16093618e-01 5.15978515e-01 9.31412816e-01 9.03899550e-01 -3.72392714e-01 -6.77910388e-01 1.15669167e+00 4.14618343e-01 5.88128030e-01 6.94126666e-01 -3.73604178e-01 8.30591619e-01 1.04649448e+00 1.73310727e-01 -1.24432361e+00 4.39459179e-03 -6.41996145e-01 -5.22127867e-01 -3.47473234e-01 3.09402585e-01 2.31639475e-01 -9.48806703e-01 1.39584959e+00 3.61096598e-02 -1.00092433e-01 -7.24527240e-01 9.49722409e-01 7.47037709e-01 6.33732378e-01 9.05784518e-02 1.30789220e-01 1.62786615e+00 -1.02441072e+00 -9.04573619e-01 -3.76684755e-01 5.76453209e-01 -8.76916230e-01 1.57058990e+00 2.62205135e-02 -9.71295834e-01 -5.16559780e-01 -1.01219809e+00 -1.88179299e-01 -4.83850896e-01 4.74343151e-01 5.19744813e-01 3.98983419e-01 -1.79269299e-01 5.87106943e-01 -3.95093739e-01 -1.13121249e-01 3.66901875e-01 2.05364563e-02 4.90228355e-01 1.11148350e-01 -1.42929697e+00 3.54576230e-01 2.70800442e-01 6.45779725e-03 -4.33454096e-01 -7.48243868e-01 -6.42219543e-01 3.30644697e-01 5.51680982e-01 -3.05482388e-01 1.33026671e+00 -1.00836670e+00 -1.02008700e+00 4.40881997e-01 -2.90917039e-01 -1.59965858e-01 6.70149326e-01 -4.24356192e-01 -5.51316381e-01 3.19501758e-02 2.72618085e-01 2.13627815e-01 7.61733413e-01 -8.75928402e-01 -1.17597735e+00 -2.24552035e-01 1.30733758e-01 1.48766115e-01 -7.29786277e-01 -6.96907118e-02 -1.17573261e+00 -7.53248513e-01 2.47920677e-01 -5.56723714e-01 1.68861002e-01 5.25800586e-02 -6.57336056e-01 -5.58392048e-01 8.01396787e-01 -9.16435719e-01 1.94241214e+00 -2.22044992e+00 -2.62945175e-01 3.91415745e-01 5.85561574e-01 4.07660276e-01 -5.07308170e-02 1.88248381e-01 3.12589288e-01 5.35384834e-01 1.94923058e-01 1.29354849e-01 2.92919517e-01 -1.81021690e-01 -4.96351004e-01 -1.89375103e-01 1.48456916e-01 9.02707517e-01 -8.14200044e-01 -5.33152223e-01 -6.95884377e-02 5.47861680e-02 -4.47748125e-01 4.23608780e-01 -5.71231067e-01 -3.00510406e-01 -1.01383996e+00 4.26808000e-01 6.16739392e-01 -5.53885818e-01 -1.49881691e-01 -3.51255327e-01 -1.67028472e-01 7.73840725e-01 -9.04118419e-01 8.68244886e-01 -4.84645009e-01 6.01702034e-01 -2.73773938e-01 -5.49860537e-01 9.40105915e-01 -1.99403554e-01 1.10405371e-01 -1.08641529e+00 6.18726667e-03 3.96127909e-01 -7.11847842e-02 -4.75289196e-01 4.65321720e-01 3.87617826e-01 -9.90584269e-02 5.07480323e-01 -4.49634790e-01 3.15517426e-01 5.70119500e-01 6.17477357e-01 9.98124361e-01 -8.84950981e-02 -2.61166282e-02 1.07052311e-01 6.99165642e-01 -3.48315626e-01 3.73414129e-01 8.55817139e-01 -6.00133650e-02 6.27180755e-01 8.13258290e-01 -3.16685706e-01 -1.00794017e+00 -8.42546761e-01 6.90840334e-02 1.31769252e+00 3.98291647e-01 -5.56998789e-01 -6.57059848e-01 -1.05373049e+00 1.71355933e-01 5.40555418e-01 -7.55999565e-01 -2.76204318e-01 -6.50018632e-01 -5.04240453e-01 2.57420391e-01 7.13222265e-01 8.21728110e-01 -6.88708723e-01 9.27517191e-02 2.43983254e-01 -4.31960583e-01 -8.43029320e-01 -1.09773219e+00 -1.11164257e-01 -5.40765643e-01 -1.03030884e+00 -9.58209455e-01 -8.12308133e-01 4.42614943e-01 4.57920849e-01 7.93943405e-01 3.56230676e-01 -1.50286466e-01 8.02690461e-02 -2.92089462e-01 -4.81886752e-02 6.21627122e-02 3.80091488e-01 -4.08645928e-01 4.65031236e-01 3.26718599e-01 -2.95756757e-01 -9.05583918e-01 7.52895772e-01 -9.92429018e-01 -1.34938806e-01 6.42908990e-01 8.10704291e-01 2.70698220e-01 -8.73056129e-02 2.99728006e-01 -9.06768978e-01 9.45271075e-01 -5.71059346e-01 -4.63534564e-01 3.78123730e-01 -6.49309456e-01 1.80428997e-01 5.50322771e-01 -6.49453759e-01 -1.01784885e+00 -4.54581320e-01 -1.09653667e-01 -2.61742830e-01 2.43931875e-01 8.59756887e-01 -3.37543875e-01 1.46450847e-01 2.13516623e-01 5.08675396e-01 -3.48009288e-01 -8.07744265e-01 7.68446028e-02 8.34632933e-01 1.42835066e-01 -2.93842942e-01 7.67772615e-01 8.92660320e-02 -4.31826055e-01 -4.65386152e-01 -1.24096131e+00 -8.09188366e-01 -1.99693978e-01 -1.41090736e-01 5.18294632e-01 -5.59760571e-01 -8.97966027e-01 3.87802780e-01 -9.58614647e-01 1.84790313e-01 1.67370424e-01 3.06415677e-01 -1.24001212e-01 6.75942779e-01 -6.26861632e-01 -5.90734065e-01 -1.33840397e-01 -1.00018227e+00 9.48707461e-01 4.38855708e-01 -6.23826422e-02 -9.76000607e-01 -2.40388885e-01 3.81643921e-01 4.32550728e-01 -3.42333853e-01 1.33612788e+00 -8.96904826e-01 -7.75604248e-01 -6.21264517e-01 -8.59658301e-01 2.31227532e-01 2.03751817e-01 -4.47295606e-03 -5.86351275e-01 1.49126813e-01 -4.34739262e-01 -8.09726641e-02 1.10231888e+00 1.53017566e-01 1.45802736e+00 -5.79831064e-01 -4.13088173e-01 3.07451457e-01 1.09192860e+00 8.20025504e-02 5.49333096e-01 4.76469368e-01 8.52509916e-01 4.07238811e-01 7.17538238e-01 5.62070429e-01 3.35150510e-01 8.89420390e-01 4.90068972e-01 2.93483492e-02 -1.01607554e-01 -7.40713716e-01 1.86084256e-01 8.49603951e-01 2.55134404e-01 -1.45741925e-01 -7.07312167e-01 3.82659823e-01 -1.77499330e+00 -7.65948176e-01 -3.37055027e-01 2.23952985e+00 9.66960728e-01 5.75635552e-01 3.42999309e-01 -2.73831049e-03 9.99197721e-01 2.89661914e-01 -6.04814649e-01 4.10711467e-02 1.43678054e-01 -2.03328386e-01 3.97856355e-01 2.06704557e-01 -1.06402600e+00 7.17696309e-01 5.43467665e+00 1.23185110e+00 -1.10672832e+00 -2.26805285e-01 6.18465424e-01 8.27630237e-02 -5.89830399e-01 -1.52663156e-01 -1.25100100e+00 1.15712368e+00 4.14912194e-01 -2.95161784e-01 4.01079774e-01 1.07411242e+00 3.42138499e-01 1.71993244e-02 -9.97917116e-01 7.77607501e-01 2.98196580e-02 -1.21697211e+00 3.18968862e-01 5.42737097e-02 4.36426073e-01 -2.68893927e-01 3.33765805e-01 5.95165789e-01 1.02888376e-01 -5.13838291e-01 7.42741883e-01 4.68349248e-01 4.28161085e-01 -3.65516424e-01 5.69003820e-01 2.29860410e-01 -1.45363414e+00 -2.13222533e-01 -1.79107368e-01 2.18327820e-01 -1.70160338e-01 6.30595028e-01 -5.62640846e-01 2.00781092e-01 5.14796317e-01 8.17787349e-01 -9.96512413e-01 1.48250461e+00 -2.40881011e-01 8.74375045e-01 -2.09906325e-01 -7.04632878e-01 4.63629335e-01 -8.69365111e-02 3.26642156e-01 1.11565912e+00 1.05007730e-01 -2.87741631e-01 2.81769067e-01 1.03057611e+00 -3.65009218e-01 3.15570861e-01 1.44751608e-01 -3.83356452e-01 5.26662230e-01 1.20525002e+00 -4.50744629e-01 -6.36004686e-01 -5.27745426e-01 7.25726008e-01 3.38290066e-01 4.06100571e-01 -9.92887914e-01 -9.81562674e-01 5.68975568e-01 4.30898935e-01 5.44609308e-01 4.04738598e-02 -2.53802925e-01 -1.26290810e+00 4.92193520e-01 -6.70113981e-01 4.16165859e-01 -6.91605091e-01 -1.44720721e+00 3.22935134e-01 -2.27596387e-01 -1.31013453e+00 2.93254882e-01 -4.27657634e-01 -9.48629797e-01 9.58654821e-01 -1.61629212e+00 -1.03788912e+00 -6.95326865e-01 3.05987090e-01 7.03864872e-01 2.54805475e-01 2.11995952e-02 5.23009837e-01 -7.55981028e-01 7.52506316e-01 2.48339608e-01 3.92969787e-01 5.63966215e-01 -1.27664542e+00 5.02581894e-01 3.90516609e-01 -1.69861943e-01 9.38574493e-01 4.83473122e-01 -8.49404156e-01 -1.03126061e+00 -1.11724579e+00 1.03266978e+00 -5.56329973e-02 1.06838381e+00 -3.75596136e-01 -1.14751184e+00 4.18765098e-01 -1.81491137e-01 -2.53266156e-01 5.11580288e-01 4.88206446e-01 -6.74261153e-01 -3.51083368e-01 -4.33254987e-01 7.03823030e-01 9.20435011e-01 -5.33559501e-01 -6.77634597e-01 3.99252474e-01 6.80871308e-01 7.95749947e-02 -4.75662500e-01 9.89857092e-02 6.38733685e-01 -6.36369526e-01 8.59686673e-01 -5.24371743e-01 6.56205118e-01 -1.24490730e-01 2.49677032e-01 -1.08131194e+00 -5.94939590e-01 -1.82675675e-01 -1.11045502e-01 1.60150707e+00 3.99865806e-01 -4.44428355e-01 8.34266663e-01 4.73233581e-01 -1.92355569e-02 -6.91441596e-01 -4.94583398e-01 -7.15423405e-01 -3.02546144e-01 -1.71491086e-01 7.25786328e-01 5.00746608e-01 8.96157348e-04 6.05900168e-01 -3.72847199e-01 -2.17044815e-01 4.32206422e-01 3.35373759e-01 6.08668447e-01 -1.25055444e+00 -1.87977314e-01 -9.85262215e-01 -2.17938274e-01 -1.78735101e+00 -2.42019743e-01 -8.19578528e-01 5.11559099e-02 -1.42776227e+00 5.03190994e-01 -5.10517299e-01 -5.18392980e-01 1.78917453e-01 -6.69447124e-01 5.98901138e-02 1.48603078e-02 6.84190571e-01 -1.02940321e+00 6.24933481e-01 1.25433862e+00 -4.33803618e-01 -5.29344156e-02 3.10009181e-01 -6.82761550e-01 7.68519878e-01 5.51859617e-01 -2.38698483e-01 -9.55464169e-02 -2.74253041e-01 4.29083347e-01 -2.04367742e-01 3.04962903e-01 -8.35818648e-01 1.97532326e-01 -2.93480903e-01 3.39988828e-01 -8.27540636e-01 1.60711333e-01 -5.43392420e-01 -6.74485683e-01 2.76639760e-01 -9.26828563e-01 -3.08958203e-01 -2.34466538e-01 9.42173541e-01 -3.15170676e-01 -4.08337206e-01 4.10848707e-01 -6.42415881e-02 -5.95726371e-01 3.96407902e-01 -4.89947468e-01 2.66421527e-01 6.00976348e-01 -6.15140684e-02 -2.89096922e-01 -4.56392795e-01 -4.09952581e-01 4.47909325e-01 7.60349929e-02 5.87751031e-01 5.47939241e-01 -1.54079318e+00 -3.36666942e-01 2.64875025e-01 4.47580814e-01 -5.34868121e-01 9.99237224e-02 5.66523314e-01 -3.58763963e-01 6.17582023e-01 7.41984099e-02 -4.01041955e-01 -1.24037600e+00 4.17950034e-01 1.91659972e-01 -4.81232375e-01 -1.02833457e-01 7.35696614e-01 6.34193242e-01 -2.12594979e-02 4.82900351e-01 -4.38195616e-01 -3.89877707e-01 1.61450848e-01 7.53298938e-01 2.86849678e-01 -5.11475746e-03 -2.11712480e-01 -1.02046631e-01 5.34105003e-01 -6.33766174e-01 7.89878517e-02 9.03189123e-01 -2.96100587e-01 1.17799953e-01 4.14641768e-01 1.39023173e+00 4.81249839e-02 -1.16376340e+00 -6.01229131e-01 3.15102786e-01 -7.50167191e-01 2.39426360e-01 -7.64352858e-01 -1.05951905e+00 1.18211985e+00 2.70733625e-01 4.35335100e-01 9.32958901e-01 1.58710498e-02 1.19920671e+00 2.70870179e-01 -2.74561614e-01 -1.14045501e+00 6.20459259e-01 4.47907537e-01 6.89663529e-01 -1.14460826e+00 -1.06966272e-01 -6.21056795e-01 -4.65929359e-01 1.11583984e+00 7.87296653e-01 2.49716993e-02 4.61313337e-01 -3.96184981e-01 -2.02547088e-01 -2.25164536e-02 -5.27469039e-01 -4.15987939e-01 8.05383027e-01 1.02487989e-02 4.37118143e-01 -8.51034895e-02 -7.19520271e-01 8.46557260e-01 2.31419653e-01 -2.60631800e-01 2.57913843e-02 7.15777993e-01 -8.71438265e-01 -1.03598857e+00 -1.47494182e-01 9.09549773e-01 -5.49509883e-01 -4.01401728e-01 -4.74115014e-01 3.60268056e-01 -2.02142179e-01 7.47753501e-01 6.28701895e-02 -5.56511939e-01 3.13172430e-01 -1.06403545e-01 -1.60336301e-01 -4.59062815e-01 -6.33338928e-01 3.75482500e-01 5.68042360e-02 -3.01912427e-01 2.54813075e-01 -4.59432006e-01 -8.08147371e-01 -2.79342592e-01 -6.53034568e-01 4.98307794e-01 7.11194217e-01 8.64812374e-01 3.29540163e-01 5.33872426e-01 9.92258251e-01 -2.84698963e-01 -8.05181742e-01 -9.67992902e-01 -6.29012764e-01 7.70490825e-01 1.41219556e-01 -7.01073766e-01 -4.37523037e-01 -1.78029478e-01]
[7.7728190422058105, 9.650147438049316]
1d6ad714-c37f-4016-b76d-5b8fa409d990
chatcad-interactive-computer-aided-diagnosis
2302.07257
null
https://arxiv.org/abs/2302.07257v1
https://arxiv.org/pdf/2302.07257v1.pdf
ChatCAD: Interactive Computer-Aided Diagnosis on Medical Image using Large Language Models
Large language models (LLMs) have recently demonstrated their potential in clinical applications, providing valuable medical knowledge and advice. For example, a large dialog LLM like ChatGPT has successfully passed part of the US medical licensing exam. However, LLMs currently have difficulty processing images, making it challenging to interpret information from medical images, which are rich in information that supports clinical decisions. On the other hand, computer-aided diagnosis (CAD) networks for medical images have seen significant success in the medical field by using advanced deep-learning algorithms to support clinical decision-making. This paper presents a method for integrating LLMs into medical-image CAD networks. The proposed framework uses LLMs to enhance the output of multiple CAD networks, such as diagnosis networks, lesion segmentation networks, and report generation networks, by summarizing and reorganizing the information presented in natural language text format. The goal is to merge the strengths of LLMs' medical domain knowledge and logical reasoning with the vision understanding capability of existing medical-image CAD models to create a more user-friendly and understandable system for patients compared to conventional CAD systems. In the future, LLM's medical knowledge can be also used to improve the performance of vision-based medical-image CAD models.
['Dinggang Shen', 'Qian Wang', 'Xi Ouyang', 'Zihao Zhao', 'Sheng Wang']
2023-02-14
null
null
null
null
['logical-reasoning']
['reasoning']
[ 1.37426853e-01 7.16556847e-01 -3.54566962e-01 -7.02027798e-01 -5.64250708e-01 -1.02913462e-01 2.13141859e-01 6.67803586e-01 -1.51297122e-01 5.63735485e-01 2.95558453e-01 -6.84780419e-01 -1.45427629e-01 -9.79663134e-01 -1.45412579e-01 -3.29648882e-01 1.38976440e-01 7.31477141e-01 1.40685722e-01 -7.99995963e-04 9.66123044e-02 5.93776286e-01 -1.14873981e+00 9.97057259e-01 1.09344888e+00 9.05830264e-01 5.56330502e-01 6.64388955e-01 -4.98821139e-01 1.50904369e+00 -5.28850913e-01 -3.42865437e-01 -3.76466185e-01 -5.35615683e-01 -9.70631540e-01 1.44441351e-01 1.95148990e-01 -4.64174092e-01 -2.22369477e-01 1.09508741e+00 4.12033677e-01 -2.89451689e-01 4.60538447e-01 -9.43974555e-01 -7.85468578e-01 6.48418307e-01 -2.52699167e-01 9.48985219e-02 2.64744163e-01 1.27152786e-01 4.70894009e-01 -4.77768064e-01 8.07885468e-01 1.37322140e+00 4.72082734e-01 6.48622274e-01 -5.03710091e-01 -5.43965995e-01 -8.59336406e-02 5.13028622e-01 -9.81835365e-01 -1.48377091e-01 5.00007331e-01 -4.86489832e-01 1.04501748e+00 1.70104370e-01 6.54144645e-01 5.91894865e-01 6.12235844e-01 1.06938612e+00 7.09248245e-01 -5.20538449e-01 2.03771442e-01 2.45136306e-01 3.80756199e-01 1.09640324e+00 3.75802726e-01 -3.10702890e-01 -2.51511335e-01 -4.44479287e-03 9.03216958e-01 2.78732151e-01 -4.99970792e-03 1.73738211e-01 -1.22277462e+00 9.11270320e-01 8.65998507e-01 5.66149652e-01 -5.00722945e-01 -2.19939873e-01 3.81874949e-01 -4.76461835e-02 2.97973007e-01 4.36101913e-01 -1.29529759e-01 1.32379577e-01 -8.54039848e-01 7.09524006e-03 6.97973430e-01 6.76209688e-01 1.72816813e-01 -2.34155625e-01 -3.08506817e-01 8.74580026e-01 5.67805886e-01 3.78672391e-01 8.39687526e-01 -1.01221359e+00 3.77544194e-01 1.02573490e+00 -4.66325849e-01 -1.11991608e+00 -6.33694708e-01 -2.86543638e-01 -1.26378214e+00 6.11999035e-02 8.91198497e-03 -1.11451186e-01 -1.18597174e+00 1.13547289e+00 6.39280826e-02 -3.52562196e-03 2.99088061e-01 7.75518417e-01 1.49491060e+00 4.62865949e-01 4.54642564e-01 6.54807836e-02 1.58205557e+00 -9.64551389e-01 -9.85911250e-01 -3.72097671e-01 9.14081216e-01 -5.72954237e-01 6.95084691e-01 3.67906719e-01 -1.15627527e+00 -6.05693579e-01 -9.03189182e-01 -2.69912928e-01 -3.56104136e-01 4.67826635e-01 8.85334730e-01 2.25704730e-01 -1.15624321e+00 2.47404128e-01 -9.57203746e-01 -4.72460687e-01 1.00590396e+00 3.62292260e-01 -1.16922930e-01 -4.61326867e-01 -1.21951282e+00 1.29729223e+00 4.91945297e-01 2.56038308e-01 -5.89636922e-01 -7.33366072e-01 -1.04511547e+00 -5.24931066e-02 2.51732975e-01 -1.03144884e+00 1.61620235e+00 -7.37482488e-01 -1.20160115e+00 1.14368641e+00 -2.68134266e-01 -6.44168556e-01 3.82863939e-01 8.68099928e-02 -4.17658389e-01 6.14336848e-01 1.57227382e-01 1.19831491e+00 2.67001897e-01 -8.49842548e-01 -7.26661444e-01 -3.89793664e-01 1.55288965e-01 1.72334895e-01 -1.37090102e-01 -1.86356425e-01 -5.03227293e-01 -4.18250948e-01 4.51776572e-02 -5.05490303e-01 -6.96256638e-01 3.98048341e-01 -4.29764748e-01 -2.55415052e-01 7.08507061e-01 -8.18033755e-01 1.18900228e+00 -2.04638863e+00 -2.43835345e-01 2.46603321e-02 6.61094904e-01 7.22181916e-01 5.75858690e-02 5.27425148e-02 -1.41223241e-02 2.26221368e-01 1.99441072e-02 -1.95527095e-02 -5.67476273e-01 3.73004198e-01 1.56913429e-01 -1.34793743e-01 3.80508274e-01 1.15757346e+00 -7.71242678e-01 -1.12855995e+00 6.54162049e-01 3.43748450e-01 -5.51631987e-01 1.98462442e-01 -3.69036525e-01 5.57362974e-01 -6.79629982e-01 6.49782062e-01 3.72792900e-01 -8.48801076e-01 1.81297556e-01 -2.67422467e-01 2.69362301e-01 1.12735234e-01 -6.56423569e-01 1.59662318e+00 -5.24184644e-01 6.76157951e-01 9.04041901e-02 -1.16881323e+00 6.34634316e-01 5.78818202e-01 5.21269023e-01 -5.70645988e-01 1.93965077e-01 -4.29958627e-02 2.70686418e-01 -1.14582229e+00 8.38910341e-02 -2.17426807e-01 3.55585456e-01 2.83071935e-01 -7.86160007e-02 -3.87668431e-01 2.46069744e-01 5.18603802e-01 8.49592984e-01 -3.52624744e-01 5.35197318e-01 3.19755703e-01 7.94467270e-01 4.89950180e-01 3.21839601e-01 6.02860987e-01 -1.09535143e-01 4.20550585e-01 3.39338988e-01 -4.96458560e-01 -6.78007126e-01 -9.29908812e-01 -1.69981658e-01 3.47602338e-01 -4.04698439e-02 -2.79113889e-01 -8.02654028e-01 -5.79130471e-01 -2.24413380e-01 7.87573040e-01 -2.58213907e-01 -1.82182133e-01 -3.86745065e-01 -6.39716446e-01 5.28582573e-01 7.74466991e-01 9.33629811e-01 -1.51837337e+00 -7.00670838e-01 2.68997550e-01 -4.08947170e-01 -1.22159624e+00 -5.11019342e-02 -2.28520215e-01 -1.16627097e+00 -1.15399122e+00 -7.60184467e-01 -1.18910849e+00 1.05385876e+00 4.73537296e-02 9.96452272e-01 2.96390295e-01 -8.38440776e-01 3.44168752e-01 -2.58690953e-01 -8.09335947e-01 -9.65358019e-01 -1.03048682e-01 -4.90254372e-01 -4.69623268e-01 4.92947578e-01 -1.31532820e-02 -5.57945311e-01 -2.32468963e-01 -1.17651320e+00 6.81665063e-01 1.01909733e+00 8.31656575e-01 5.80941856e-01 1.02978297e-01 6.06768847e-01 -1.12534285e+00 8.48756492e-01 -4.49695051e-01 -1.13820851e-01 4.73552167e-01 -5.61952651e-01 -1.75406307e-01 3.96469772e-01 -2.21717000e-01 -1.44376421e+00 -1.11643218e-01 -4.22581464e-01 -9.33188647e-02 -4.99169111e-01 1.03923762e+00 1.73677728e-01 1.13723889e-01 5.98971486e-01 -9.99536067e-02 3.91888201e-01 -1.83318570e-01 3.53494495e-01 1.00912368e+00 6.72000170e-01 -9.25605446e-02 1.46532685e-01 2.41007164e-01 3.94290760e-02 -8.42525661e-01 -1.03866625e+00 -3.43552053e-01 -5.13214231e-01 -1.73914969e-01 1.16309524e+00 -8.04422081e-01 -7.38495767e-01 3.97920012e-01 -1.27881420e+00 -2.40953406e-03 -2.84087032e-01 6.09570503e-01 -2.19493091e-01 4.62507755e-01 -1.07046783e+00 -4.39674050e-01 -5.90965390e-01 -1.24375069e+00 7.56173849e-01 5.55131316e-01 -5.67781270e-01 -1.31401694e+00 -4.03280765e-01 7.63940275e-01 3.76797616e-01 1.22727811e-01 1.52588344e+00 -6.38852477e-01 -4.69280571e-01 -5.34746647e-01 -3.91200304e-01 4.89882112e-01 3.80400151e-01 -8.65224525e-02 -6.37871742e-01 2.61397719e-01 1.22391276e-01 -2.32918620e-01 6.29777908e-01 9.41130519e-01 1.43542671e+00 -8.85018110e-02 -7.12201953e-01 1.60501122e-01 1.11643445e+00 5.90855718e-01 5.38507342e-01 -2.54346114e-02 6.59133494e-01 6.48870468e-01 5.75884640e-01 8.60331056e-04 5.78822434e-01 3.61136235e-02 2.49954551e-01 -6.47759497e-01 -3.51089776e-01 -5.37834466e-02 -1.72463339e-02 9.26775992e-01 1.82488486e-01 -5.79044968e-02 -1.24221063e+00 3.82725865e-01 -1.91097522e+00 -5.59697509e-01 -1.98706806e-01 1.49571419e+00 1.06477237e+00 1.20336220e-01 -6.37075484e-01 -2.24345073e-01 5.71346343e-01 -3.74880254e-01 -7.02930093e-01 -4.06733125e-01 1.86860129e-01 3.08298945e-01 1.68897748e-01 4.06888932e-01 -9.37056363e-01 9.07711625e-01 6.67465782e+00 7.57092893e-01 -1.20467842e+00 6.67275488e-02 9.47743416e-01 2.33264267e-01 -4.44282889e-02 -4.89664078e-01 -5.76554954e-01 -2.20806152e-03 8.65778625e-01 -1.40948266e-01 -2.66840667e-01 9.33850288e-01 5.47720134e-01 -4.60666180e-01 -1.23142445e+00 1.08543682e+00 1.29224822e-01 -2.02769208e+00 4.65257227e-01 -1.10057816e-01 5.15701354e-01 -2.38916948e-01 -8.89734402e-02 2.29287684e-01 4.98010725e-01 -1.25898528e+00 2.43827105e-02 6.33749127e-01 8.51441801e-01 -3.30249608e-01 1.29341400e+00 5.21770000e-01 -7.97982693e-01 -6.45320443e-03 -1.38564453e-01 -3.93785872e-02 2.08347023e-01 6.49820268e-01 -1.49855387e+00 5.93958080e-01 4.81128544e-01 8.61476839e-01 -6.94387496e-01 1.16684949e+00 -2.25656986e-01 4.74454403e-01 1.32495552e-01 6.59750868e-03 2.84009725e-01 8.21715221e-02 7.56771788e-02 1.19662118e+00 1.02145970e-01 4.21438724e-01 4.35380250e-01 9.04449463e-01 4.19009849e-02 1.24852642e-01 -6.50208116e-01 -4.52851385e-01 6.60964921e-02 1.28299201e+00 -1.03330016e+00 -8.92682850e-01 -5.10125637e-01 4.98254091e-01 -2.60115176e-01 9.80418995e-02 -5.28518260e-01 -2.02233046e-01 2.71794617e-01 8.53476226e-02 -3.11884403e-01 1.28324956e-01 -6.35662675e-01 -9.94037569e-01 -3.59882057e-01 -8.90780330e-01 4.74646032e-01 -1.02198100e+00 -1.43479085e+00 6.63372040e-01 -6.56901523e-02 -1.05428088e+00 -5.53703725e-01 -9.06157076e-01 -4.44666564e-01 8.15715492e-01 -1.39227569e+00 -1.38076913e+00 -4.94211882e-01 5.49652576e-01 8.58044028e-01 -3.60528022e-01 1.09657216e+00 2.26034373e-01 -2.75197059e-01 1.27616644e-01 -4.21410471e-01 4.89775479e-01 6.67516708e-01 -1.02550185e+00 1.55376736e-02 4.54340726e-01 -2.49018058e-01 6.28387392e-01 1.23612233e-01 -8.53254914e-01 -8.06924760e-01 -1.19127989e+00 9.02840972e-01 -1.73918396e-01 3.02420676e-01 3.48814517e-01 -8.58371198e-01 6.26291275e-01 2.17451572e-01 -1.95090532e-01 9.44962084e-01 -3.65539938e-01 2.45056942e-01 -1.19372737e-02 -1.42915130e+00 5.75173795e-01 5.00633776e-01 -3.51452827e-01 -9.04665530e-01 5.75030148e-01 7.77233064e-01 -5.71520686e-01 -7.62399256e-01 5.33430099e-01 2.15085492e-01 -4.45800364e-01 8.89860332e-01 -8.62147391e-01 8.31819594e-01 -1.56047106e-01 2.62982368e-01 -1.12489378e+00 -1.54663816e-01 6.06060028e-02 1.21210076e-01 8.17259312e-01 4.23864961e-01 -5.17458379e-01 6.43653989e-01 7.88418710e-01 -2.83310473e-01 -8.74217272e-01 -4.46305484e-01 3.80708687e-02 3.10681341e-03 -5.78493476e-01 1.70394540e-01 8.68634105e-01 2.46803969e-01 3.57495427e-01 1.97014675e-01 1.69451550e-01 2.92250842e-01 -8.51346478e-02 1.82405546e-01 -1.41310203e+00 -1.82647873e-02 -5.17703533e-01 -4.75004882e-01 -7.22608089e-01 -4.38917130e-02 -1.19299257e+00 -5.88634610e-02 -2.38421893e+00 3.41828793e-01 -3.84610027e-01 -2.51942594e-02 7.66159952e-01 -3.92957479e-02 4.17994708e-02 1.97804272e-01 4.87352759e-02 -4.94760484e-01 -1.41895413e-01 1.75065315e+00 -4.75935489e-01 -2.28831902e-01 6.18774407e-02 -9.49931204e-01 1.09693789e+00 7.09207475e-01 -2.24080265e-01 -4.65953022e-01 -4.25597221e-01 1.72589663e-02 4.71970975e-01 3.43217045e-01 -9.95032310e-01 6.60889685e-01 -1.04402706e-01 6.54255033e-01 -7.87342310e-01 1.64651908e-02 -6.62008405e-01 -1.64564520e-01 8.66166353e-01 -6.29279196e-01 -3.45759302e-01 3.34189326e-01 1.40780509e-01 -7.46413052e-01 -2.91850030e-01 5.89845061e-01 -7.24391580e-01 -8.31609011e-01 1.84230641e-01 -8.14410448e-01 -2.22191751e-01 1.15021849e+00 -1.21844850e-01 -3.94933879e-01 -4.58020270e-01 -1.12979186e+00 5.55624843e-01 -2.14102030e-01 4.83689904e-01 1.05754101e+00 -9.44398880e-01 -6.28447413e-01 1.10969484e-01 -7.24359751e-02 5.21933496e-01 5.77522039e-01 8.35318446e-01 -1.07365143e+00 7.96577275e-01 -3.10249209e-01 -7.91133225e-01 -1.44174123e+00 2.58512616e-01 2.79887289e-01 -6.31792665e-01 -7.12620795e-01 5.69882512e-01 2.78250158e-01 -3.24134707e-01 3.98968697e-01 -7.59036005e-01 -4.83872741e-01 8.83456727e-04 8.26703191e-01 -2.25980356e-02 6.03054464e-02 -2.16137409e-01 -2.70416558e-01 3.87252867e-01 -4.48696852e-01 3.17074060e-02 1.39543045e+00 -8.30602124e-02 -4.01882738e-01 2.80699521e-01 6.39080048e-01 -5.13043404e-01 -4.07729506e-01 -2.98445493e-01 4.02012430e-02 1.72588140e-01 2.06444129e-01 -1.31445265e+00 -1.14873922e+00 1.32902980e+00 6.76436484e-01 -7.10432380e-02 1.10355234e+00 1.99847162e-01 8.27248096e-01 6.37641609e-01 1.76120907e-01 -9.24615264e-01 1.98982909e-01 2.17224523e-01 7.38885760e-01 -1.42715013e+00 -2.64185648e-02 -6.08688533e-01 -1.02404702e+00 1.55480564e+00 7.06637502e-01 2.17881709e-01 7.64266908e-01 4.10581261e-01 4.33461279e-01 -4.00821328e-01 -5.89030027e-01 2.01641675e-02 3.13559175e-01 6.59370840e-01 5.96271098e-01 1.76904887e-01 -1.43248811e-01 7.43233979e-01 4.04836563e-03 8.08988214e-01 5.00295997e-01 9.30627108e-01 -4.48833138e-01 -1.00445175e+00 -4.31036264e-01 7.77943552e-01 -6.04924202e-01 -2.24343419e-01 -3.25293154e-01 6.16054237e-01 4.21159238e-01 9.67300951e-01 1.40178837e-02 2.02556537e-03 4.53477167e-02 1.34372145e-01 2.52151519e-01 -1.07191372e+00 -3.97012413e-01 7.38851773e-03 7.33313560e-02 -2.92609334e-01 -6.01056576e-01 -1.38734490e-01 -1.77430093e+00 3.30845662e-03 1.07983477e-01 -6.93084821e-02 5.04888594e-01 1.15209329e+00 3.83781344e-01 1.07518458e+00 -1.77912012e-01 -3.57267767e-01 -8.18484053e-02 -1.07000792e+00 -1.54257730e-01 1.21494137e-01 2.79060692e-01 -1.53271988e-01 4.15084779e-01 4.74322826e-01]
[15.021306037902832, -1.5832418203353882]
2da38865-9017-4454-9359-b3b1649b66d2
call-attention-to-rumors-deep-attention-based
1704.05973
null
http://arxiv.org/abs/1704.05973v1
http://arxiv.org/pdf/1704.05973v1.pdf
Call Attention to Rumors: Deep Attention Based Recurrent Neural Networks for Early Rumor Detection
The proliferation of social media in communication and information dissemination has made it an ideal platform for spreading rumors. Automatically debunking rumors at their stage of diffusion is known as \textit{early rumor detection}, which refers to dealing with sequential posts regarding disputed factual claims with certain variations and highly textual duplication over time. Thus, identifying trending rumors demands an efficient yet flexible model that is able to capture long-range dependencies among postings and produce distinct representations for the accurate early detection. However, it is a challenging task to apply conventional classification algorithms to rumor detection in earliness since they rely on hand-crafted features which require intensive manual efforts in the case of large amount of posts. This paper presents a deep attention model on the basis of recurrent neural networks (RNN) to learn \textit{selectively} temporal hidden representations of sequential posts for identifying rumors. The proposed model delves soft-attention into the recurrence to simultaneously pool out distinct features with particular focus and produce hidden representations that capture contextual variations of relevant posts over time. Extensive experiments on real datasets collected from social media websites demonstrate that (1) the deep attention based RNN model outperforms state-of-the-arts that rely on hand-crafted features; (2) the introduction of soft attention mechanism can effectively distill relevant parts to rumors from original posts in advance; (3) the proposed method detects rumors more quickly and accurately than competitors.
['Xue Li', 'Jun Zhang', 'Lin Wu', 'Hongzhi Yin', 'Yang Wang', 'Tong Chen']
2017-04-20
null
null
null
null
['deep-attention', 'deep-attention']
['computer-vision', 'natural-language-processing']
[ 6.46631643e-02 -3.17949653e-01 -2.81258434e-01 -1.58293828e-01 -4.62528497e-01 -1.53852329e-01 9.26634490e-01 1.86361447e-01 -1.95370004e-01 4.51937973e-01 5.62231481e-01 -5.21856070e-01 -6.67851567e-02 -5.10534227e-01 -3.73630852e-01 -3.91617477e-01 -2.04509348e-01 4.78315383e-01 1.82575315e-01 -6.69986188e-01 7.31525123e-01 3.94813091e-01 -1.12712419e+00 5.85024357e-01 5.23786843e-01 9.87300217e-01 2.02751189e-01 4.01014000e-01 -2.99436212e-01 1.58080757e+00 -8.25374246e-01 -9.33354273e-02 -1.31722420e-01 -6.89501166e-01 -9.51238811e-01 2.50635445e-01 -7.48429373e-02 -5.55545747e-01 -6.88285589e-01 9.73581791e-01 2.43268013e-01 1.97069272e-01 5.21893024e-01 -8.16911280e-01 -1.19857824e+00 8.59487295e-01 -1.01842225e+00 1.20009208e+00 1.75685048e-01 -6.93838075e-02 9.70757306e-01 -9.83622193e-01 4.20426548e-01 1.13012826e+00 7.18127489e-01 2.79609323e-01 -6.73951209e-01 -7.43619740e-01 2.07216680e-01 6.78702176e-01 -1.06959355e+00 -3.72511804e-01 1.09473789e+00 -5.94993234e-01 8.42097342e-01 2.35576689e-01 2.79002964e-01 1.57448435e+00 4.04949903e-01 9.32232380e-01 1.13620460e+00 7.88568258e-02 -1.94392025e-01 3.05801351e-02 6.50611877e-01 5.87054491e-01 9.12805088e-04 1.04205171e-02 -6.54354513e-01 -4.67587888e-01 5.38942039e-01 6.69495761e-01 -3.77645284e-01 5.23617327e-01 -9.59736586e-01 1.27443123e+00 6.25805676e-01 5.54489911e-01 -1.01708233e+00 -3.63552094e-01 7.31366336e-01 6.93843067e-01 8.34156275e-01 3.38713318e-01 -2.56345272e-02 1.60219088e-01 -1.39454973e+00 1.95036650e-01 6.13366365e-01 4.10033822e-01 5.78093469e-01 1.90537140e-01 -1.73915014e-01 9.59994912e-01 -1.05690554e-01 7.86295235e-02 1.02118528e+00 -7.18079880e-02 4.19770509e-01 6.09855115e-01 2.15059981e-01 -1.52357185e+00 -5.76329052e-01 -8.23629141e-01 -1.11457992e+00 -2.93264836e-01 -8.16788822e-02 -1.15169883e-01 -8.96430075e-01 1.14563787e+00 -1.54516418e-02 1.50952086e-01 -2.08444223e-01 1.01917934e+00 6.74203813e-01 1.04238784e+00 -3.99060190e-01 -6.54112756e-01 1.30220115e+00 -1.05886602e+00 -9.98550594e-01 -2.27000788e-01 -3.71278822e-02 -7.29352534e-01 5.70434928e-01 2.05043346e-01 -8.42610121e-01 -1.50588587e-01 -1.03395212e+00 -4.82168533e-02 -2.49894291e-01 -3.66690814e-01 3.70822251e-01 8.73058960e-02 -6.56183541e-01 6.99171007e-01 -4.94826615e-01 -1.53377444e-01 6.02641463e-01 8.26737732e-02 2.82796562e-01 2.17462361e-01 -1.62406635e+00 9.27348673e-01 1.41942799e-01 5.55274367e-01 -1.54302549e+00 -1.30250603e-01 -4.33242589e-01 4.70497236e-02 5.09044051e-01 -4.01429594e-01 1.24135566e+00 -1.25168025e+00 -1.56752479e+00 5.99861205e-01 -1.62425578e-01 -8.40164840e-01 6.53345406e-01 -3.56526047e-01 -6.35294676e-01 1.08671464e-01 2.01172054e-01 -3.81181508e-01 1.55997920e+00 -8.01718712e-01 -4.05990809e-01 -4.63482410e-01 -3.32135499e-01 -6.31439015e-02 -3.03147167e-01 6.14807248e-01 1.70769513e-01 -7.66270816e-01 2.11917639e-01 -6.46433771e-01 -6.10710718e-02 -7.64940739e-01 -6.08950078e-01 -3.01217943e-01 1.19038725e+00 -1.00227511e+00 1.31226981e+00 -1.61844838e+00 -4.63181250e-02 -8.86337757e-02 6.04870319e-01 4.14929152e-01 2.89243221e-01 6.57513678e-01 8.38603452e-02 1.72580946e-02 2.54010353e-02 -1.88084587e-01 -3.36753130e-01 -2.38592342e-01 -9.06111598e-01 7.35529125e-01 8.26208573e-03 8.43977451e-01 -8.93845797e-01 -1.09741144e-01 -2.29100972e-01 3.92063856e-01 -1.12406395e-01 3.63507748e-01 -2.12473616e-01 4.02798891e-01 -6.16949916e-01 4.69348967e-01 5.78928053e-01 -6.93764389e-01 -1.73415858e-02 2.45180309e-01 -1.32748500e-01 8.01980734e-01 -4.43261385e-01 8.73634160e-01 -3.47058743e-01 8.59071076e-01 1.18629493e-01 -9.91680145e-01 1.16919959e+00 4.67441678e-01 1.74024016e-01 -5.69947362e-01 5.82407475e-01 1.73996449e-01 -3.03517818e-01 -5.83802402e-01 8.00866306e-01 -4.58005935e-01 -1.25944465e-01 1.04064584e+00 -2.98931301e-01 5.60475230e-01 -4.40271884e-01 3.35141540e-01 1.03696346e+00 -7.32280433e-01 4.42859501e-01 -1.07499599e-01 5.49229980e-01 -2.22079739e-01 4.80296254e-01 8.92401218e-01 -2.54982978e-01 3.89937729e-01 5.11815965e-01 -6.98393643e-01 -8.38659167e-01 -4.29392695e-01 1.20182715e-01 1.32956398e+00 -1.04243830e-02 1.19295329e-01 -2.31409609e-01 -7.99285889e-01 -5.10645211e-02 6.91956162e-01 -9.76616502e-01 1.62525531e-02 -5.87678134e-01 -9.95641708e-01 3.95747602e-01 2.61770397e-01 5.32409370e-01 -1.46483696e+00 -4.31996822e-01 4.09877181e-01 -2.58245468e-01 -7.37058938e-01 -6.67734087e-01 -2.08171960e-02 -7.25882053e-01 -9.31076705e-01 -8.94656897e-01 -6.55692697e-01 4.09293950e-01 6.93471074e-01 6.06624663e-01 2.43276358e-01 2.17236474e-01 -3.24769139e-01 -4.04617310e-01 -2.42233813e-01 -4.98812884e-01 1.22414097e-01 -3.86339426e-02 5.12333512e-01 2.27168158e-01 -7.53661692e-01 -4.25108135e-01 3.18581372e-01 -8.55630040e-01 -6.70968369e-02 5.63943446e-01 1.12899339e+00 -1.24911100e-01 9.85572860e-02 1.07097924e+00 -1.21854138e+00 1.36638451e+00 -1.28862751e+00 -2.91302919e-01 -1.61820546e-01 -6.68878615e-01 1.62808634e-02 6.34874642e-01 -5.55829704e-01 -1.07628453e+00 -9.30360615e-01 2.17096791e-01 -6.51934743e-01 2.68769115e-01 8.67348075e-01 6.87057853e-01 3.47209573e-01 7.81074703e-01 7.68690228e-01 1.23472363e-01 -5.92828751e-01 6.88003153e-02 1.19280267e+00 4.34327334e-01 -1.46987941e-02 5.27006209e-01 4.65574652e-01 -5.40750146e-01 -8.10715020e-01 -1.17857945e+00 -7.39158630e-01 -2.20435277e-01 -7.13452697e-03 3.34900707e-01 -5.87661207e-01 -7.68334448e-01 6.41726673e-01 -1.33061147e+00 1.53568611e-01 4.20296311e-01 1.33175716e-01 -1.73485786e-01 5.93920708e-01 -1.36998200e+00 -1.13977909e+00 -9.46711421e-01 -7.09366143e-01 4.16449338e-01 5.08606508e-02 -1.96927354e-01 -9.37423050e-01 1.07443690e-01 6.47806227e-01 8.64979029e-01 7.33007193e-02 7.07740605e-01 -1.22652400e+00 -3.44922364e-01 -4.58704412e-01 -4.81209636e-01 2.16373309e-01 2.31947854e-01 -4.31845993e-01 -8.97320271e-01 -3.56332213e-01 5.74198365e-01 -3.33803833e-01 1.30725610e+00 1.00803375e-01 9.11598206e-01 -1.11208773e+00 -1.08737484e-01 2.01237551e-03 8.45378697e-01 -5.24101816e-02 3.49797010e-01 4.14248109e-01 6.06791794e-01 4.95859206e-01 1.20216332e-01 7.56384611e-01 3.96490842e-01 4.32049900e-01 5.70278823e-01 2.11246237e-01 3.17915469e-01 -2.29023501e-01 6.51652157e-01 1.07293868e+00 -3.83441336e-02 -2.21614569e-01 -7.42614031e-01 9.14845347e-01 -1.85052407e+00 -1.49591720e+00 -8.52238461e-02 1.98476648e+00 8.30092847e-01 2.05206379e-01 3.05560887e-01 3.89157683e-02 1.12045038e+00 7.38520086e-01 -3.99098724e-01 -4.95995402e-01 -7.21969232e-02 -2.41863161e-01 2.07171082e-01 5.91328621e-01 -8.23620439e-01 6.73241317e-01 5.30360413e+00 4.29401368e-01 -1.53299522e+00 4.43587393e-01 4.90856469e-01 -2.56007761e-01 -1.81743324e-01 -1.95613205e-01 -6.99011624e-01 8.04320693e-01 1.05005944e+00 -3.10580760e-01 5.24487853e-01 6.30165279e-01 4.83005792e-01 5.01791537e-01 -5.54889619e-01 6.62223697e-01 3.93741608e-01 -1.57842422e+00 3.13371509e-01 -5.68413734e-02 9.22524333e-01 3.71784776e-01 2.53065854e-01 3.77690971e-01 1.84447810e-01 -1.05581188e+00 7.09268391e-01 4.74031895e-01 5.06382771e-02 -7.41073251e-01 1.10935843e+00 7.75260210e-01 -4.30651456e-01 -3.91212374e-01 -4.36850518e-01 -1.91243917e-01 3.48046035e-01 6.60487652e-01 -1.55832839e+00 3.51953089e-01 3.73859227e-01 8.80185544e-01 -1.98792472e-01 7.03263223e-01 -2.48363629e-01 8.40685189e-01 1.22549601e-01 -2.35942975e-01 7.24913657e-01 3.09176296e-01 8.33144605e-01 1.30698466e+00 1.45378202e-01 2.56404560e-02 1.42965823e-01 9.00858164e-01 -3.54099631e-01 1.24792665e-01 -2.40624472e-01 -2.88579077e-01 3.56741846e-01 1.01332533e+00 -3.99955779e-01 -3.08641404e-01 -1.08593278e-01 1.04415929e+00 4.91239727e-01 3.00878376e-01 -7.13019788e-01 -5.09131551e-02 2.06337243e-01 5.75380743e-01 5.31445742e-01 -7.11843371e-02 -4.73782457e-02 -1.21837282e+00 -2.45210826e-01 -8.87450159e-01 6.83439076e-01 -4.81852919e-01 -1.69918180e+00 1.14881265e+00 -4.42377716e-01 -7.30125487e-01 -3.66691500e-01 -1.07550500e-02 -1.00961852e+00 8.74127328e-01 -1.74309468e+00 -1.11142159e+00 -2.32080102e-01 8.28271747e-01 9.71117616e-01 -8.21044594e-02 3.75880390e-01 9.91373882e-02 -8.94562364e-01 2.77725130e-01 4.31996956e-02 2.42728144e-01 4.49744374e-01 -7.13697016e-01 2.95989573e-01 7.25453556e-01 -1.35305598e-01 9.38731909e-01 1.01432800e+00 -8.88838053e-01 -9.81399775e-01 -1.08485484e+00 1.16002035e+00 -2.15581685e-01 1.05993164e+00 8.72013867e-02 -1.45597470e+00 1.01273048e+00 4.92060333e-01 -2.09905177e-01 2.69490689e-01 2.33042330e-01 -4.69798714e-01 1.56672269e-01 -9.41574991e-01 2.01688722e-01 4.78411913e-01 -8.37565064e-01 -1.03215837e+00 5.82287848e-01 3.14989597e-01 -1.04698718e-01 -2.58403420e-01 7.18636438e-02 2.14770615e-01 -9.58268404e-01 4.72544551e-01 -9.23322082e-01 5.62437534e-01 4.11415361e-02 3.35163653e-01 -1.04055405e+00 -5.62967896e-01 -1.24279296e+00 -3.94914150e-01 9.48088884e-01 2.61264056e-01 -7.22382128e-01 4.32951957e-01 -1.87207162e-01 4.06254977e-02 -7.89063394e-01 -8.15593660e-01 -4.48086530e-01 -1.77065641e-01 -8.95449221e-02 3.65845710e-01 1.39783180e+00 2.94981331e-01 7.48458207e-01 -1.10451829e+00 1.95428729e-01 4.90834832e-01 5.53186417e-01 4.35747743e-01 -1.18968189e+00 -8.65135491e-02 -6.52989507e-01 -1.99980527e-01 -1.35534275e+00 1.99736908e-01 -7.97578514e-01 -3.30320522e-02 -1.12907612e+00 4.61691350e-01 -1.39791638e-01 -6.97398722e-01 2.53073752e-01 -2.37537622e-01 -1.20063461e-02 -1.12478763e-01 8.44229698e-01 -5.71324050e-01 7.29713798e-01 1.00420606e+00 -2.91741043e-01 -2.59695441e-01 4.42251235e-01 -8.37595761e-01 6.41370535e-01 7.06359804e-01 -7.50759900e-01 -1.34667203e-01 -1.59935609e-01 3.92845452e-01 5.75010598e-01 4.31050062e-01 -2.55589068e-01 4.76257831e-01 -1.84598699e-01 -8.88050534e-03 -8.29767942e-01 2.02449769e-01 -1.90781325e-01 -2.63523310e-01 2.16793224e-01 -5.49632072e-01 2.46454045e-01 -5.28142601e-02 1.03835356e+00 -1.45247251e-01 -1.91558421e-01 7.72717535e-01 -3.76261860e-01 -5.56404948e-01 2.36744955e-01 -8.01443577e-01 -1.76016067e-03 6.71662390e-01 2.55565226e-01 -4.34876621e-01 -9.25960302e-01 -6.41006231e-01 -4.48643602e-03 1.43405527e-01 6.52977288e-01 8.08500171e-01 -9.56255555e-01 -1.09972143e+00 1.35223508e-01 -2.26715297e-01 -4.22758311e-01 4.59338844e-01 1.23165286e+00 -9.89641398e-02 4.71107215e-01 -1.83660805e-01 -2.10882276e-01 -9.61117566e-01 8.33068252e-01 2.67989665e-01 -5.00824928e-01 -8.42694461e-01 8.07687283e-01 5.32410778e-02 9.91014689e-02 -1.87622502e-01 2.72121243e-02 -4.37903106e-01 3.34355980e-01 9.87714171e-01 5.37459671e-01 2.09274013e-02 -8.02878380e-01 -2.34218761e-01 -2.70359993e-01 -9.59366500e-01 2.14938805e-01 1.55320513e+00 -3.88236016e-01 -3.79627764e-01 4.69794035e-01 1.09201670e+00 9.50146765e-02 -9.26834583e-01 -8.19524646e-01 1.77035347e-01 -2.86769807e-01 5.22471011e-01 -6.57859981e-01 -1.07623804e+00 8.37984800e-01 -1.82597354e-01 5.66712976e-01 7.38931298e-01 -3.56288888e-02 1.24949312e+00 1.60380676e-01 6.78927898e-02 -7.29507923e-01 3.50732088e-01 7.36328900e-01 1.20505655e+00 -9.98311520e-01 1.06236242e-01 1.12170525e-01 -8.87484968e-01 1.20162261e+00 6.32790551e-02 -3.34682137e-01 5.76625049e-01 -3.16567004e-01 2.17647050e-02 -7.16714263e-01 -9.05176878e-01 2.72340566e-01 2.38140345e-01 -9.28224549e-02 1.47913054e-01 -8.18737298e-02 -2.77771652e-01 8.25762510e-01 -5.66385575e-02 -3.94378342e-02 8.73833418e-01 6.63330972e-01 -7.90384531e-01 -2.32082918e-01 -4.78240877e-01 5.05126417e-01 -9.94826198e-01 -1.94188267e-01 -4.81661797e-01 3.40733320e-01 -4.42785859e-01 9.76382375e-01 -1.99283570e-01 -4.20881689e-01 -2.79499620e-01 2.31371503e-02 -1.85015917e-01 -5.07800043e-01 -8.08062792e-01 -8.78697634e-02 1.56032115e-01 -1.24606766e-01 -1.87567398e-02 -7.17294216e-01 -8.09469461e-01 -7.06487477e-01 -6.23988628e-01 3.86134326e-01 3.03414375e-01 1.24907124e+00 5.55482924e-01 4.55131203e-01 1.03825176e+00 -8.14198673e-01 -1.22230721e+00 -1.48678553e+00 -5.29223621e-01 4.46670175e-01 9.72530127e-01 -6.33934438e-01 -6.03074372e-01 -4.21206892e-01]
[8.162402153015137, 10.134195327758789]
4d54cc7e-47f0-4be4-b273-89e47ff99086
two-baselines-for-unsupervised-dependency
null
null
https://aclanthology.org/W12-1910
https://aclanthology.org/W12-1910.pdf
Two baselines for unsupervised dependency parsing
null
['Anders S{\\o}gaard']
2012-06-01
null
null
null
ws-2012-6
['unsupervised-dependency-parsing']
['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]
[-7.315014839172363, 3.6626956462860107]
74ecfa65-83c6-4d29-8d61-9da797cb5c5a
dsdp-a-blind-docking-strategy-accelerated-by
2303.09916
null
https://arxiv.org/abs/2303.09916v1
https://arxiv.org/pdf/2303.09916v1.pdf
DSDP: A Blind Docking Strategy Accelerated by GPUs
Virtual screening, including molecular docking, plays an essential role in drug discovery. Many traditional and machine-learning based methods are available to fulfil the docking task. The traditional docking methods are normally extensively time-consuming, and their performance in blind docking remains to be improved. Although the runtime of docking based on machine learning is significantly decreased, their accuracy is still limited. In this study, we take the advantage of both traditional and machine-learning based methods, and present a method Deep Site and Docking Pose (DSDP) to improve the performance of blind docking. For the traditional blind docking, the entire protein is covered by a cube, and the initial positions of ligands are randomly generated in the cube. In contract, DSDP can predict the binding site of proteins and provide an accurate searching space and initial positions for the further conformational sampling. The docking task of DSDP makes use of the score function and a similar but modified searching strategy of AutoDock Vina, accelerated by implementation in GPUs. We systematically compare its performance with the state-of-the-art methods, including Autodock Vina, GNINA, QuickVina, SMINA, and DiffDock. DSDP reaches a 29.8% top-1 success rate (RMSD < 2 {\AA}) on an unbiased and challenging test dataset with 1.2 s wall-clock computational time per system. Its performances on DUD-E dataset and the time-split PDBBind dataset used in EquiBind, TankBind, and DiffDock are also effective, presenting a 57.2% and 41.8% top-1 success rate with 0.8 s and 1.0 s per system, respectively.
['Yi Qin Gao', 'Jun Zhang', 'Xiaohan Lin', 'Dajiong Yue', 'Siyuan Jiang', 'Hong Zhang', 'Yupeng Huang']
2023-03-16
null
null
null
null
['drug-discovery', 'blind-docking', 'molecular-docking']
['medical', 'medical', 'medical']
[-2.98567444e-01 -5.97724617e-01 4.55416031e-02 -1.37986869e-01 -6.61863863e-01 -8.61465514e-01 1.41804606e-01 2.17446789e-01 -7.19022036e-01 1.25701141e+00 -4.06813443e-01 -6.16656184e-01 2.62511939e-01 -4.19100106e-01 -7.56302297e-01 -1.22474647e+00 -3.68110016e-02 5.98512232e-01 4.34459537e-01 -2.74116069e-01 4.68958408e-01 6.60941660e-01 -1.23009157e+00 5.35570756e-02 1.13504863e+00 4.71440881e-01 5.25630414e-01 2.59902775e-01 -2.64588781e-02 3.56148034e-02 -5.55787981e-01 -4.83451873e-01 4.23613749e-02 -5.00964403e-01 -5.50877988e-01 -7.22768188e-01 -3.85312065e-02 -3.25742476e-02 6.36062324e-02 9.82702911e-01 1.47543216e+00 1.29595445e-03 6.88758254e-01 -3.87171805e-01 -1.53488144e-01 -3.28833222e-01 -3.44299018e-01 7.09089190e-02 8.06839406e-01 2.45475158e-01 4.58470345e-01 -1.14370334e+00 5.66419244e-01 7.89295018e-01 7.02418327e-01 6.03990436e-01 -1.29636455e+00 -1.06102991e+00 -1.81434736e-01 2.67668694e-01 -1.48458815e+00 -1.03760868e-01 1.95480101e-02 -5.99270642e-01 1.13148987e+00 4.00552779e-01 6.59163654e-01 8.90151083e-01 5.65471470e-01 7.98793659e-02 1.08316457e+00 -2.58539855e-01 7.09815979e-01 -2.05608681e-01 -2.11422324e-01 4.27486807e-01 2.52824366e-01 1.27582386e-01 -4.85980690e-01 -7.95041144e-01 7.16160119e-01 4.27986026e-01 -5.22333860e-01 -3.10546756e-01 -9.98392999e-01 9.72769737e-01 6.38003588e-01 -2.63559014e-01 -3.33306462e-01 -5.94718277e-01 3.33422631e-01 -9.76102874e-02 2.53974795e-02 6.37952447e-01 -7.76503623e-01 -6.13865592e-02 -5.86414754e-01 5.87657273e-01 8.26372802e-01 5.63295901e-01 5.13800740e-01 -4.86938655e-01 -4.65893932e-02 4.16963667e-01 3.06412041e-01 3.39910775e-01 5.52347422e-01 -2.52294123e-01 1.12746753e-01 5.24957061e-01 5.82795382e-01 -6.39239907e-01 -8.32964063e-01 -3.20251584e-02 -8.28230739e-01 5.87824225e-01 4.50620532e-01 -9.83087495e-02 -1.05558205e+00 1.35340273e+00 7.10952401e-01 -1.09900258e-01 2.83184528e-01 1.20641768e+00 9.71629083e-01 5.71035624e-01 2.48667747e-01 -7.81871736e-01 1.35797405e+00 -8.26120734e-01 -4.39603686e-01 4.23889518e-01 6.79032683e-01 -1.18009806e+00 7.33111560e-01 7.96756566e-01 -7.62176156e-01 -3.83314818e-01 -1.04586840e+00 1.95436090e-01 -2.21288979e-01 1.89988002e-01 7.44782925e-01 6.20218158e-01 -5.86664855e-01 8.97868574e-01 -8.19810987e-01 -1.38436660e-01 2.57859766e-01 1.06295514e+00 -6.99875474e-01 -4.40584356e-03 -9.85350668e-01 7.72027433e-01 5.49170852e-01 -4.54907604e-02 -8.39491129e-01 -5.90847671e-01 -3.12427670e-01 -4.56218034e-01 1.64089963e-01 -5.52667856e-01 1.10678697e+00 -3.69482845e-01 -1.77943921e+00 4.90208000e-01 -4.46347445e-01 -7.28643611e-02 3.96837562e-01 -8.81583244e-02 -1.05913348e-01 -1.22642890e-01 1.36144832e-01 2.67712712e-01 1.23107724e-01 -7.96224236e-01 -2.74866521e-01 -6.38392985e-01 -1.58823326e-01 4.86642927e-01 3.12053204e-01 2.75944710e-01 -4.77752805e-01 -3.94916147e-01 2.73270786e-01 -8.53848934e-01 -5.90483308e-01 -3.77322674e-01 -4.53376114e-01 -9.71232206e-02 2.07133412e-01 -5.41568518e-01 1.19560885e+00 -1.74529183e+00 3.76956135e-01 4.09425408e-01 2.85118818e-01 7.09142387e-01 3.00918102e-01 7.47976601e-01 -4.02456254e-01 -1.33015573e-01 9.16242599e-02 2.57508397e-01 -5.08941948e-01 -9.70065817e-02 8.24429560e-03 7.36910939e-01 -4.01336789e-01 4.68085438e-01 -7.63041914e-01 -1.40509769e-01 3.55692059e-02 6.12333238e-01 -6.92205012e-01 3.58178943e-01 -1.56846628e-01 9.44630027e-01 -6.02374196e-01 8.52958202e-01 1.04128063e+00 -3.50286096e-01 6.21487141e-01 -2.48723045e-01 -5.22061169e-01 2.28870288e-01 -1.24467707e+00 1.68080997e+00 2.82406330e-01 -4.70963359e-01 -3.35413903e-01 -3.97931457e-01 9.31360662e-01 3.54799598e-01 1.83644190e-01 -9.38340724e-01 2.70488746e-02 6.59222841e-01 2.60793000e-01 -3.04464698e-01 -1.85898751e-01 -1.31403655e-01 4.09609497e-01 -3.57543044e-02 -1.39417484e-01 5.29598594e-01 6.87466636e-02 -5.06520458e-02 8.91345143e-01 3.70546073e-01 5.71784794e-01 -2.41869926e-01 6.28851950e-01 3.38356227e-01 6.24902248e-01 2.38898486e-01 1.43557757e-01 5.22961855e-01 4.04200643e-01 -8.30330193e-01 -9.66055214e-01 -4.79075700e-01 -6.27439380e-01 9.58919525e-01 3.83538097e-01 -8.08775187e-01 -9.32374299e-01 -5.05244136e-01 4.72691096e-02 -8.23104009e-02 -3.70565504e-01 1.83830876e-02 -3.69986504e-01 -1.41310108e+00 2.65234977e-01 2.87858024e-02 6.17083386e-02 -8.45926821e-01 -1.62959874e-01 6.50034666e-01 2.42127120e-01 -4.21020538e-01 -2.89940685e-01 5.86632013e-01 -7.69808114e-01 -1.52861118e+00 -8.65357220e-01 -6.41434848e-01 5.94218135e-01 2.26634234e-01 6.99754298e-01 -1.20733820e-01 -3.79121453e-01 -8.41626048e-01 -3.74328911e-01 -3.37556601e-01 1.05641358e-01 -1.94719121e-01 5.50626874e-01 -4.72505242e-01 6.33921385e-01 -5.86035252e-01 -1.04200757e+00 8.37400854e-01 -3.47851425e-01 -8.91867056e-02 6.02787495e-01 1.17743790e+00 1.30854714e+00 -3.54421347e-01 1.64547876e-01 -6.82328641e-01 3.02163064e-01 -3.13320100e-01 -9.18776155e-01 -7.77374357e-02 -6.19879425e-01 2.74797101e-02 7.90361106e-01 -4.67540860e-01 -4.95590210e-01 5.93251407e-01 -4.66384172e-01 -2.60681421e-01 2.77817957e-02 5.23694873e-01 -5.01182139e-01 -7.42721319e-01 8.45353067e-01 3.01016003e-01 2.24231631e-01 -1.04878044e+00 -3.56407017e-01 7.05478907e-01 1.49360150e-01 -3.98228854e-01 3.57291192e-01 1.90744460e-01 1.31011382e-01 -3.17454398e-01 -2.82779485e-01 -6.19808733e-01 -3.90017241e-01 4.45358366e-01 8.20709050e-01 -1.07495689e+00 -1.57903171e+00 4.57797527e-01 -1.20945215e+00 -1.24441773e-01 7.12924600e-01 8.21309447e-01 -2.52289355e-01 7.92029381e-01 -2.39213362e-01 -4.97610539e-01 -6.33615851e-01 -1.82098758e+00 8.69202733e-01 2.84811199e-01 -2.52587833e-02 -3.53516519e-01 5.06993353e-01 3.11271042e-01 5.69023415e-02 4.02973413e-01 6.78264558e-01 -9.98800635e-01 -3.92720044e-01 -2.98422784e-01 -1.63139850e-02 -1.09741464e-01 -2.26493590e-02 -1.65127128e-01 -7.66602635e-01 -5.19593358e-01 -3.64059031e-01 -3.06198329e-01 4.82220501e-01 3.25228125e-01 9.00340021e-01 -2.09999084e-01 -7.19513595e-01 9.37296510e-01 1.54340029e+00 8.00967872e-01 7.86510289e-01 6.59247100e-01 5.97293019e-01 2.23783050e-02 9.35128450e-01 6.37910664e-01 9.51825827e-02 1.20135951e+00 8.57262909e-01 -2.77286440e-01 5.79788983e-01 6.36274219e-02 2.24347115e-01 3.05776715e-01 -8.63680124e-01 -3.03426776e-02 -9.46729183e-01 -3.09849024e-01 -1.84551513e+00 -9.07135010e-01 -6.12175047e-01 2.70212865e+00 1.40339422e+00 -1.96099579e-01 4.31222558e-01 -1.39511809e-01 5.45416355e-01 -6.79200292e-01 -8.24307323e-01 -1.33998320e-01 -6.76842257e-02 5.38062155e-01 5.09157419e-01 3.35687459e-01 -1.15642881e+00 9.01783347e-01 5.84039068e+00 1.04387498e+00 -1.21865475e+00 -1.92134351e-01 2.23625451e-01 -2.35285447e-03 3.65042388e-01 5.78595027e-02 -1.14266074e+00 7.57474542e-01 8.99073780e-01 1.31471604e-01 3.21967453e-01 9.64222729e-01 3.35781246e-01 -2.36872986e-01 -9.25997317e-01 1.21452940e+00 -4.09654826e-01 -1.55954242e+00 -2.99335620e-03 4.56762761e-01 3.60460967e-01 1.00692980e-01 -2.75835961e-01 -1.19432462e-02 -8.18111748e-03 -1.37469292e+00 1.55082271e-01 3.69562179e-01 9.87951994e-01 -1.04258537e+00 1.12756062e+00 2.63707787e-01 -9.99505162e-01 3.92310619e-01 -7.70538867e-01 -5.69781587e-02 -2.60756999e-01 4.52047408e-01 -7.33760178e-01 8.35532129e-01 7.26616085e-01 3.83017033e-01 -3.07022661e-01 1.53302109e+00 1.79769117e-02 1.99513480e-01 -3.25920582e-01 -3.15885216e-01 1.39059961e-01 -5.51034033e-01 1.54148236e-01 9.80172753e-01 -1.32378489e-02 4.28257942e-01 4.34761703e-01 3.23908269e-01 1.38836443e-01 5.19732535e-01 9.59664509e-02 2.93673933e-01 5.34415185e-01 1.02080417e+00 -3.93181950e-01 -5.33821695e-02 -2.61916608e-01 1.13186240e+00 1.20211355e-01 2.33360887e-01 -9.70522642e-01 -6.55500412e-01 1.20111573e+00 2.53751576e-02 3.24616104e-01 5.25272191e-02 3.37968796e-01 -1.02448654e+00 2.55728046e-05 -1.17355120e+00 2.25182027e-01 -5.09764671e-01 -1.00063896e+00 5.62230349e-01 -5.18470824e-01 -1.31815398e+00 2.76964188e-01 -1.00232148e+00 -5.32634676e-01 1.52251458e+00 -1.34920943e+00 -4.22413379e-01 -3.37866932e-01 6.64943874e-01 2.27200419e-01 -2.88268298e-01 1.32173491e+00 4.37073648e-01 -8.61507952e-01 6.47310019e-01 8.88868570e-01 -2.55413204e-01 1.13385463e+00 -1.15871429e+00 3.05227995e-01 1.51175395e-01 -5.39631903e-01 1.13647127e+00 7.26704359e-01 -9.01716828e-01 -1.69805026e+00 -8.97830963e-01 4.80898559e-01 -4.32206124e-01 2.10540280e-01 -2.55231559e-01 -1.00154269e+00 -2.72926651e-02 -3.82297128e-01 -1.28536280e-02 1.21086442e+00 -1.33051574e-01 -2.08184212e-01 3.40610653e-01 -1.09667575e+00 4.13478792e-01 8.20410907e-01 -1.07102290e-01 -1.56723976e-01 7.96506166e-01 4.79070842e-01 -9.30510759e-01 -9.37714696e-01 3.57285053e-01 6.25797451e-01 -1.08868253e+00 1.13306832e+00 -8.70374918e-01 -3.54574889e-01 -8.79145086e-01 -1.02180876e-01 -8.99292707e-01 -6.41586065e-01 -8.96409929e-01 1.53498262e-01 5.41219831e-01 4.78660375e-01 -6.95689917e-01 9.15458977e-01 2.60773450e-01 -2.28056535e-01 -1.08159971e+00 -1.15486658e+00 -7.52424538e-01 -2.54984237e-02 2.52795547e-01 5.33877432e-01 7.31318414e-01 1.76029354e-01 4.66473609e-01 -2.15622410e-01 2.84465492e-01 4.66817915e-01 5.91273373e-03 9.02148128e-01 -1.48395979e+00 -3.16228211e-01 9.45428610e-02 -6.50823951e-01 -8.34950030e-01 -4.37020749e-01 -6.89089894e-01 -5.16573310e-01 -1.39739418e+00 4.68449056e-01 -1.62963122e-01 -1.39252739e-02 7.11216509e-01 -2.73320466e-01 8.27202350e-02 -3.54528397e-01 5.98314643e-01 -5.29671967e-01 4.51363713e-01 1.21196115e+00 2.24593848e-01 -5.19807518e-01 1.26902908e-01 -3.79718006e-01 5.35245419e-01 6.75879598e-01 -5.11644304e-01 4.26916964e-02 3.65397334e-01 3.49645585e-01 -2.75554985e-01 1.66495696e-01 -7.64940917e-01 2.22409666e-01 -2.35035673e-01 7.21700132e-01 -6.42932951e-01 2.23961696e-01 -5.44204712e-01 6.45836055e-01 8.15319598e-01 7.70422339e-01 -3.48697938e-02 3.75861228e-01 6.21294498e-01 -6.74943402e-02 2.99232826e-02 8.19837511e-01 -1.26265600e-01 -3.35341364e-01 3.49920183e-01 -1.27429217e-01 -4.05939370e-01 1.05236697e+00 -6.65944159e-01 -1.13543943e-01 3.53825271e-01 -9.42500055e-01 4.29502539e-02 7.53572404e-01 -3.52550983e-01 5.43759763e-01 -1.01110721e+00 -4.03505147e-01 2.61665881e-01 2.20670894e-01 5.38600646e-02 2.29257837e-01 1.02988052e+00 -1.09917724e+00 6.92454576e-01 -2.25645646e-01 -4.36502397e-01 -1.55143213e+00 6.28888369e-01 5.91325045e-01 7.77629316e-02 -2.79476166e-01 8.19498777e-01 3.24177295e-01 -3.25448841e-01 2.78295010e-01 6.37243167e-02 -2.66052753e-01 1.28546171e-02 9.32356656e-01 2.21936360e-01 6.31446362e-01 -4.31484848e-01 -8.16877306e-01 7.58764267e-01 -4.83379781e-01 6.19296551e-01 1.26939273e+00 5.16351163e-01 -2.24904940e-01 -4.30967480e-01 8.87084424e-01 4.23918478e-02 -1.08846784e+00 2.27017701e-01 -3.37044507e-01 -3.26576412e-01 -1.14771441e-01 -1.19278347e+00 -2.61515141e-01 6.61280215e-01 9.44635212e-01 -4.66765255e-01 8.10153604e-01 -1.44766763e-01 5.01017153e-01 5.22477508e-01 7.84696043e-01 -5.06944835e-01 -2.63851285e-01 4.43072200e-01 7.02440202e-01 -1.27290201e+00 2.54445612e-01 -3.90477866e-01 -5.42290926e-01 1.30126357e+00 5.70785940e-01 2.00111300e-01 5.67576475e-02 -1.00185022e-01 -1.02326475e-01 -2.93177754e-01 -3.37677896e-01 -1.73819195e-02 1.35313779e-01 5.65940559e-01 6.16621614e-01 -8.69490653e-02 -6.66858375e-01 1.01787317e+00 1.48779482e-01 -3.04011721e-02 5.16516529e-02 1.06063521e+00 -6.25867188e-01 -1.56069136e+00 -6.78190529e-01 -1.43131480e-01 -6.63201153e-01 -3.24660182e-01 -7.23125756e-01 6.82405949e-01 1.50523484e-01 5.91809452e-01 -5.34290671e-01 -3.82776171e-01 5.84530950e-01 -5.73367476e-02 4.67385918e-01 -4.93589103e-01 -7.23241746e-01 3.76526415e-01 -2.57972538e-01 -7.06478477e-01 -1.68906301e-01 -2.93288827e-01 -1.57467878e+00 -4.90360498e-01 -7.61481047e-01 7.80532598e-01 8.29578996e-01 5.71039021e-01 8.14037621e-01 2.76272036e-02 6.03258371e-01 -1.11243045e+00 -4.92570996e-01 -7.14013219e-01 -5.73760092e-01 -2.43250161e-01 1.65352091e-01 -9.14555013e-01 -2.24339217e-01 -2.60684580e-01]
[4.877192497253418, 5.533206939697266]
7a54e472-fd56-4fea-bfec-b2304dd2901a
eye-in-the-sky-drone-based-object-tracking
1910.08259
null
https://arxiv.org/abs/1910.08259v1
https://arxiv.org/pdf/1910.08259v1.pdf
Eye in the Sky: Drone-Based Object Tracking and 3D Localization
Drones, or general UAVs, equipped with a single camera have been widely deployed to a broad range of applications, such as aerial photography, fast goods delivery and most importantly, surveillance. Despite the great progress achieved in computer vision algorithms, these algorithms are not usually optimized for dealing with images or video sequences acquired by drones, due to various challenges such as occlusion, fast camera motion and pose variation. In this paper, a drone-based multi-object tracking and 3D localization scheme is proposed based on the deep learning based object detection. We first combine a multi-object tracking method called TrackletNet Tracker (TNT) which utilizes temporal and appearance information to track detected objects located on the ground for UAV applications. Then, we are also able to localize the tracked ground objects based on the group plane estimated from the Multi-View Stereo technique. The system deployed on the drone can not only detect and track the objects in a scene, but can also localize their 3D coordinates in meters with respect to the drone camera. The experiments have proved our tracker can reliably handle most of the detected objects captured by drones and achieve favorable 3D localization performance when compared with the state-of-the-art methods.
['Jenq-Neng Hwang', 'Gaoang Wang', 'Zhichao Lei', 'Haotian Zhang']
2019-10-18
null
null
null
null
['drone-based-object-tracking']
['computer-vision']
[-2.51236320e-01 -8.66204977e-01 3.39615613e-01 2.88691908e-01 -1.90601349e-01 -7.96569347e-01 4.01285082e-01 3.76776569e-02 -5.28988004e-01 4.85936344e-01 -7.13437378e-01 2.80656010e-01 -2.21190080e-01 -4.54793483e-01 -5.74817121e-01 -8.32716346e-01 -2.11844221e-01 3.48747671e-01 7.91572750e-01 -2.20785215e-01 -5.80633767e-02 8.19852352e-01 -1.69252419e+00 -4.73307133e-01 3.92958075e-01 1.14345467e+00 5.28413773e-01 6.03862286e-01 5.55019438e-01 2.24710450e-01 -5.69990575e-01 -1.79528505e-01 5.58064103e-01 8.00497159e-02 2.04786688e-01 2.18578771e-01 9.18748915e-01 -6.92686081e-01 -3.25351983e-01 1.23395920e+00 3.70900244e-01 1.69619888e-01 2.08462358e-01 -1.53059602e+00 -1.94237549e-02 -2.05719709e-01 -9.26376939e-01 3.81338984e-01 3.02758604e-01 -2.52132677e-02 5.35617590e-01 -6.40993178e-01 4.40180749e-01 1.12238896e+00 8.98329377e-01 2.91857034e-01 -5.18998384e-01 -7.65318096e-01 1.70132771e-01 1.09040096e-01 -1.60366583e+00 -1.29604757e-01 4.34230834e-01 -6.35742426e-01 3.50664020e-01 -6.06713369e-02 8.90846670e-01 6.01871192e-01 5.23496628e-01 4.25724536e-01 6.28154099e-01 -1.50672212e-01 -2.97345519e-01 3.35836411e-02 -2.17626959e-01 9.56041992e-01 8.83049309e-01 4.29711610e-01 -2.32429326e-01 -1.17860168e-01 9.57911253e-01 6.06216073e-01 -6.22199655e-01 -7.24823594e-01 -1.59456539e+00 5.59760928e-01 4.27712351e-01 1.67416602e-01 -5.24010837e-01 2.19782338e-01 2.45323613e-01 -1.37319922e-01 3.45540643e-01 1.02169789e-01 -3.96605402e-01 1.43501163e-01 -8.50085914e-01 4.19152975e-01 3.65097612e-01 1.30432940e+00 5.22688270e-01 1.84823975e-01 1.16952337e-01 1.39535829e-01 5.90243578e-01 1.14012063e+00 2.33836681e-01 -3.16310108e-01 3.41835171e-01 6.94161892e-01 6.44782841e-01 -1.48772264e+00 -5.25638998e-01 -5.88017583e-01 -7.86818743e-01 3.77453417e-01 1.33708492e-01 -3.72241408e-01 -4.47674245e-01 1.30355608e+00 7.61698723e-01 3.95477474e-01 3.41000296e-02 1.17933011e+00 6.81461215e-01 6.58585310e-01 -2.00799555e-01 -2.28624895e-01 1.42927170e+00 -6.78520083e-01 -6.97875142e-01 -3.01614940e-01 4.46141064e-01 -8.68788004e-01 4.64347973e-02 4.86938000e-01 -7.16892183e-01 -8.45042229e-01 -1.11985803e+00 5.61534166e-01 -2.04832867e-01 7.89107621e-01 2.54949182e-01 4.68293756e-01 -8.50410283e-01 1.53005958e-01 -9.57097173e-01 -5.28740525e-01 1.74350619e-01 4.84902114e-01 -4.36947882e-01 4.41596471e-03 -7.78573990e-01 9.91438508e-01 6.86135888e-01 3.99735570e-01 -1.14326692e+00 -3.59393239e-01 -9.24008965e-01 -1.28747031e-01 5.78358710e-01 -6.96829677e-01 8.59286964e-01 -6.82855308e-01 -9.51472104e-01 6.49332106e-01 1.46328166e-01 -4.72704560e-01 4.74077940e-01 -5.60866892e-01 -5.41120231e-01 1.86698228e-01 1.42073065e-01 4.00167078e-01 8.93717170e-01 -1.07081199e+00 -1.39750934e+00 -4.99486029e-01 8.78590345e-02 2.99112409e-01 -2.87263691e-01 1.19740568e-01 -2.50600845e-01 -3.84082496e-01 1.26322061e-01 -1.19575059e+00 3.69588844e-03 4.05173093e-01 -2.14618072e-02 -5.63660711e-02 1.47274137e+00 -2.65937686e-01 8.24301839e-01 -2.06966591e+00 2.30336353e-01 -3.20828825e-01 1.27319813e-01 7.00457394e-01 5.05466126e-02 3.30732852e-01 2.88847506e-01 -6.13829195e-01 2.52981126e-01 -2.75695443e-01 -3.60284746e-01 -3.63103569e-01 -3.05334121e-01 9.96554911e-01 -2.83689886e-01 3.65455270e-01 -1.08882105e+00 -3.57804269e-01 5.89159787e-01 5.49706042e-01 -1.23160481e-01 2.44697466e-01 -4.38011587e-02 4.62349683e-01 -5.62591672e-01 8.20212662e-01 9.19517577e-01 5.63455895e-02 -3.75609159e-01 -4.29969609e-01 -6.47780001e-01 -6.45087659e-01 -1.50263798e+00 1.42942512e+00 -2.16441199e-01 5.93349874e-01 3.99405301e-01 -4.58927393e-01 1.02505219e+00 4.28935647e-01 5.32258928e-01 1.07390061e-01 1.76404878e-01 1.23494692e-01 -1.01189196e-01 -4.25900280e-01 7.11136758e-01 1.18300512e-01 4.46077846e-02 -1.30767360e-01 5.72525412e-02 -1.45591637e-02 2.39817753e-01 -1.47267476e-01 8.97258282e-01 2.15430051e-01 4.19564575e-01 -1.24058530e-01 7.88016498e-01 3.48613083e-01 6.75937951e-01 3.85109991e-01 -1.87494367e-01 1.96224704e-01 -4.34883296e-01 -6.70340776e-01 -6.84078813e-01 -7.31073797e-01 -1.37622850e-02 6.21619225e-01 8.86196554e-01 -2.61164844e-01 -3.71571660e-01 -4.02311951e-01 1.11686938e-01 -1.17193364e-01 -8.21205527e-02 -4.55889069e-02 -4.56912726e-01 -6.36373878e-01 3.57100427e-01 2.63963908e-01 9.08703208e-01 -4.89243746e-01 -1.41234350e+00 3.43913227e-01 1.35793194e-01 -1.51706910e+00 -4.11750913e-01 -3.35214853e-01 -7.14784086e-01 -1.39312232e+00 -7.41382480e-01 -7.21101880e-01 5.53637743e-01 1.10337853e+00 4.40924674e-01 1.63287893e-01 -3.89943361e-01 7.57131219e-01 -3.78488302e-01 -5.21990240e-01 5.68715781e-02 -3.08737516e-01 6.64943397e-01 2.94361025e-01 1.45061210e-01 -1.08667620e-01 -6.11391485e-01 6.72592640e-01 -7.72471547e-01 -2.25850016e-01 6.04408741e-01 5.18683851e-01 4.17681396e-01 4.54847991e-01 5.10379486e-02 -1.16959780e-01 -2.29816940e-02 -2.25158900e-01 -1.60464847e+00 2.29682863e-01 3.59924249e-02 -7.21276283e-01 5.87181449e-01 -5.52862167e-01 -4.51993197e-01 4.95522469e-01 3.59702855e-01 -9.36964214e-01 -3.50007862e-01 2.05181062e-01 -8.71347487e-02 -6.38164818e-01 1.84048697e-01 2.27031350e-01 -1.55845016e-01 -1.17564633e-01 -9.93950367e-02 5.99064469e-01 4.96067613e-01 6.39659241e-02 1.27503705e+00 6.55734479e-01 4.20940757e-01 -1.07030082e+00 -8.02021503e-01 -7.77960956e-01 -8.56227756e-01 -6.10721827e-01 9.71246362e-01 -1.45598292e+00 -9.88612115e-01 6.67040884e-01 -1.49038804e+00 3.05415362e-01 3.39949518e-01 1.02854419e+00 -2.89046049e-01 4.16722596e-01 -1.90510169e-01 -1.00519860e+00 -3.26251239e-01 -1.16530979e+00 1.36859727e+00 5.50206780e-01 4.29204822e-01 -1.16272688e+00 4.89079766e-02 4.87252232e-03 1.80368498e-01 5.92619777e-01 3.25790718e-02 -3.06015253e-01 -1.13449335e+00 -6.48398817e-01 -9.90150962e-03 2.77797468e-02 2.46777132e-01 1.24165162e-01 -6.33322358e-01 -9.67845023e-01 -8.73793513e-02 5.64176999e-02 3.98954898e-01 6.21975243e-01 3.20102096e-01 1.89237162e-01 -9.73335803e-01 7.95927227e-01 1.69606054e+00 4.24975425e-01 -8.80924985e-02 2.69547850e-01 7.99634397e-01 2.11802557e-01 1.36966050e+00 6.12091780e-01 1.75521240e-01 9.89730716e-01 9.78347540e-01 6.93685515e-03 2.69309133e-01 -3.59668583e-02 5.01002133e-01 3.91428202e-01 -2.15741456e-01 -5.75886309e-01 -7.20120907e-01 4.23168778e-01 -1.94657564e+00 -8.43170166e-01 -4.06162292e-01 2.39182806e+00 -2.57060856e-01 -2.52593812e-02 6.63254634e-02 -2.76725024e-01 1.26250410e+00 1.05789602e-01 -3.83449316e-01 5.63277900e-01 3.66058275e-02 -4.79451656e-01 1.04258323e+00 1.12113692e-01 -1.54989183e+00 8.99145544e-01 5.25956917e+00 4.89030272e-01 -1.14471531e+00 2.54711304e-02 -4.82819766e-01 3.57759371e-02 7.22442150e-01 -1.03534885e-01 -1.60477102e+00 5.16196668e-01 3.28844666e-01 -8.75541717e-02 1.17653824e-01 9.24994648e-01 1.02391563e-01 -1.47831947e-01 -8.02968681e-01 1.38014555e+00 4.27191645e-01 -1.13212132e+00 -5.00956438e-02 2.89861947e-01 7.15177596e-01 -7.32019097e-02 1.12041213e-01 -1.17214501e-01 -3.73706967e-02 -2.65351444e-01 7.63700664e-01 4.18480039e-01 4.94899422e-01 -6.85924590e-01 1.00757205e+00 6.99119866e-01 -1.72865748e+00 -2.15345576e-01 -7.55406499e-01 -9.78004783e-02 3.33007932e-01 3.02684009e-01 -7.00129509e-01 7.47854829e-01 8.32716167e-01 1.03182745e+00 -3.78819168e-01 1.70578969e+00 -1.58501551e-01 -5.37544899e-02 -3.71728152e-01 -1.17281884e-01 4.36193347e-01 -4.47405666e-01 9.46356833e-01 5.88712394e-01 1.08050036e+00 3.25221062e-01 6.74165606e-01 5.40285408e-01 2.64151990e-01 -1.86246902e-01 -9.80282068e-01 2.37898111e-01 4.40443546e-01 1.61472297e+00 -8.24357688e-01 -3.00499946e-01 -4.66822445e-01 7.65060425e-01 -2.84666181e-01 -3.94353177e-03 -9.96849418e-01 -4.03320730e-01 6.62811637e-01 7.64089450e-02 6.58331633e-01 -6.89889491e-01 8.71576190e-01 -1.22114468e+00 -2.96926796e-02 -4.91006494e-01 1.04278333e-01 -9.44060504e-01 -7.38863409e-01 8.48120332e-01 1.62778497e-01 -1.95251048e+00 2.63989996e-02 -8.78236651e-01 -5.17634749e-01 6.12794757e-01 -1.37871253e+00 -1.40241718e+00 -8.27602029e-01 5.86896598e-01 4.91177797e-01 -4.32870120e-01 5.30994356e-01 4.80638146e-01 -5.91347694e-01 -6.63644809e-04 1.24179132e-01 1.39991745e-01 7.36900032e-01 -7.95643926e-01 -2.35965233e-02 1.21700132e+00 1.68233201e-01 4.27597493e-01 5.82817256e-01 -7.84995079e-01 -1.67285836e+00 -1.31735206e+00 2.27289811e-01 -5.62341094e-01 5.89793801e-01 -2.80236691e-01 -3.42207223e-01 9.45036113e-01 6.53529465e-02 4.51079637e-01 2.73114353e-01 -4.90929931e-01 4.81056213e-01 -4.32894081e-01 -1.04287350e+00 8.47746730e-02 8.23920131e-01 1.28507391e-01 -4.81278300e-01 5.08004367e-01 4.82743621e-01 -9.22168970e-01 -7.15622663e-01 5.19776344e-01 4.71002251e-01 -9.04158175e-01 9.69286740e-01 -1.01887129e-01 -4.28835601e-01 -9.91286755e-01 -1.32399842e-01 -1.39873827e+00 -4.03349876e-01 -5.49054325e-01 4.87379823e-03 1.12040329e+00 -4.43383396e-01 -6.29653215e-01 5.53777754e-01 -2.00693294e-01 -3.04943353e-01 -6.93562925e-02 -9.86346304e-01 -1.01152408e+00 -8.83587122e-01 2.81132519e-01 5.29246867e-01 4.75703508e-01 -6.44190550e-01 1.70324400e-01 -6.49639010e-01 1.13456106e+00 9.46676016e-01 3.87845814e-01 1.06883621e+00 -1.75743330e+00 1.30568057e-01 1.12882935e-01 -9.70904768e-01 -1.28505278e+00 -3.88099663e-02 -4.07282650e-01 8.38360414e-02 -1.25541747e+00 -1.49659514e-01 -3.69140744e-01 3.25609744e-02 9.51343682e-03 1.32721826e-01 3.77627999e-01 2.71017104e-01 2.02088505e-01 -9.57285643e-01 4.75397944e-01 9.93436158e-01 -2.91955676e-02 9.44030434e-02 3.73934686e-01 7.26746768e-03 9.01891351e-01 2.99512446e-01 -4.33905035e-01 -1.50612757e-01 -6.02207541e-01 -2.67926021e-03 3.14573556e-01 8.15138042e-01 -1.62251723e+00 6.34700656e-01 1.43290922e-01 5.22524059e-01 -1.20671630e+00 6.98846638e-01 -1.43486381e+00 3.20021123e-01 8.21505010e-01 4.85103935e-01 5.48747957e-01 5.39106548e-01 7.14156389e-01 -3.37428033e-01 -2.35074490e-01 8.00420284e-01 -2.34119937e-01 -9.58223760e-01 6.32369399e-01 -3.23243052e-01 -3.59721571e-01 1.64775157e+00 -1.86731532e-01 -2.88185805e-01 2.20304951e-02 -2.63202339e-01 3.43044013e-01 5.42974114e-01 4.09000546e-01 7.42472053e-01 -1.36071253e+00 -5.56002438e-01 3.82492214e-01 1.76225796e-01 1.06766112e-01 3.01420659e-01 9.29418921e-01 -8.53731453e-01 7.28996813e-01 -4.17421132e-01 -9.88117099e-01 -1.58757496e+00 6.82422936e-01 3.42585802e-01 1.68773025e-01 -5.08154213e-01 6.72117651e-01 1.24196902e-01 7.33029023e-02 1.80737033e-01 -5.42458951e-01 -9.59204808e-02 7.09867999e-02 7.00872719e-01 4.88880038e-01 -1.06769413e-01 -9.28520799e-01 -5.38780153e-01 1.45731711e+00 1.36205956e-01 3.91682625e-01 1.14668727e+00 -2.85533190e-01 -1.40125370e-02 2.36512810e-01 5.87882459e-01 -6.08998314e-02 -1.48180449e+00 -6.16577007e-02 -3.32690895e-01 -7.68871844e-01 3.39255691e-01 -1.97516903e-01 -1.05950391e+00 8.54579329e-01 1.02101970e+00 3.24350834e-01 8.75849843e-01 -4.12041873e-01 8.16592395e-01 5.50746679e-01 1.02642751e+00 -5.12417734e-01 -1.17685594e-01 2.99430430e-01 4.16628242e-01 -1.35295331e+00 2.08295733e-01 -2.29064822e-01 -3.50632519e-01 1.24601972e+00 7.17191398e-01 -3.73839051e-01 3.80912542e-01 1.61044732e-01 1.12344205e-01 -1.80846438e-01 -4.63492036e-01 -4.49039608e-01 2.14398831e-01 7.41672575e-01 -1.16709873e-01 -2.83634633e-01 3.14902842e-01 -9.25207138e-02 2.76766658e-01 -2.26721808e-01 5.34201145e-01 8.54665637e-01 -7.39170551e-01 -5.74038863e-01 -8.85642231e-01 -3.72551084e-02 -1.89819977e-01 2.18830764e-01 7.33298436e-02 1.21726394e+00 5.58772922e-01 8.33236098e-01 2.11551428e-01 -3.12511444e-01 5.35359740e-01 -5.25578380e-01 6.20575488e-01 -6.30710363e-01 -3.57783109e-01 1.68055147e-01 -4.53783214e-01 -2.72303194e-01 -7.19448686e-01 -7.80602515e-01 -7.71850228e-01 -6.71477523e-03 -8.85757089e-01 1.80495813e-01 6.78343892e-01 7.68456519e-01 1.83018029e-01 3.47103536e-01 5.60140014e-01 -1.22022140e+00 -3.86329234e-01 -6.71657205e-01 -9.11966562e-01 -1.24990344e-01 6.66366100e-01 -1.23021460e+00 -3.66576970e-01 -3.61909531e-03]
[6.898179531097412, -1.8734655380249023]
d769dc41-b508-4d4b-b997-6ed7e470ffac
a-neural-network-transliteration-model-in-low
null
null
https://aclanthology.org/2017.mtsummit-papers.26
https://aclanthology.org/2017.mtsummit-papers.26.pdf
A Neural Network Transliteration Model in Low Resource Settings
null
['Fatiha Sadat', 'Tan Le']
null
null
null
null
mtsummit-2017-9
['transliteration']
['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]
[-7.2482805252075195, 3.801905632019043]
59716fd1-0e27-4c11-ab27-d58486513d1c
pyramid-stereo-matching-network
1803.08669
null
http://arxiv.org/abs/1803.08669v1
http://arxiv.org/pdf/1803.08669v1.pdf
Pyramid Stereo Matching Network
Recent work has shown that depth estimation from a stereo pair of images can be formulated as a supervised learning task to be resolved with convolutional neural networks (CNNs). However, current architectures rely on patch-based Siamese networks, lacking the means to exploit context information for finding correspondence in illposed regions. To tackle this problem, we propose PSMNet, a pyramid stereo matching network consisting of two main modules: spatial pyramid pooling and 3D CNN. The spatial pyramid pooling module takes advantage of the capacity of global context information by aggregating context in different scales and locations to form a cost volume. The 3D CNN learns to regularize cost volume using stacked multiple hourglass networks in conjunction with intermediate supervision. The proposed approach was evaluated on several benchmark datasets. Our method ranked first in the KITTI 2012 and 2015 leaderboards before March 18, 2018. The codes of PSMNet are available at: https://github.com/JiaRenChang/PSMNet.
['Yong-Sheng Chen', 'Jia-Ren Chang']
2018-03-23
pyramid-stereo-matching-network-1
http://openaccess.thecvf.com/content_cvpr_2018/html/Chang_Pyramid_Stereo_Matching_CVPR_2018_paper.html
http://openaccess.thecvf.com/content_cvpr_2018/papers/Chang_Pyramid_Stereo_Matching_CVPR_2018_paper.pdf
cvpr-2018-6
['stereo-depth-estimation', 'stereo-lidar-fusion']
['computer-vision', 'computer-vision']
[ 9.23226997e-02 1.68880939e-01 -1.23163581e-01 -3.24909925e-01 -7.94839382e-01 -4.03138787e-01 5.93870997e-01 1.17566422e-01 -4.77604479e-01 5.32517374e-01 1.45037621e-01 8.58798623e-02 -1.34461462e-01 -8.33576083e-01 -1.06485534e+00 -3.94736558e-01 -7.40546063e-02 3.74154359e-01 5.12948394e-01 -1.92253307e-01 4.29843515e-01 6.54962957e-01 -1.60959983e+00 4.47324455e-01 8.69585752e-01 1.36700535e+00 3.46728683e-01 3.99562418e-01 5.98041601e-02 8.22313666e-01 -2.16265738e-01 -1.32434160e-01 8.26872885e-01 -1.75626297e-02 -8.21122110e-01 3.30733247e-02 1.02918422e+00 -4.49832171e-01 -6.00222766e-01 8.69169354e-01 2.23550960e-01 1.35727867e-01 2.55470037e-01 -1.10017133e+00 -3.51957202e-01 1.92514300e-01 -7.19052732e-01 2.95310766e-01 2.82590568e-01 1.61216944e-01 1.23550010e+00 -1.17142642e+00 7.24849343e-01 1.15518796e+00 5.98496377e-01 2.47697473e-01 -1.04237270e+00 -5.72978497e-01 1.32414490e-01 2.62790442e-01 -1.28780377e+00 -9.50155184e-02 8.41189086e-01 -4.78928596e-01 1.11276472e+00 -9.69833508e-02 8.29490364e-01 7.49043703e-01 1.35552883e-01 9.44381952e-01 1.15155494e+00 -1.95717424e-01 1.38801215e-02 -1.90652072e-01 7.60276467e-02 9.83522773e-01 4.87386361e-02 2.47018486e-01 -9.31442320e-01 1.54589891e-01 1.07211006e+00 2.70670474e-01 -2.94029802e-01 -4.64325428e-01 -1.06954062e+00 8.00006628e-01 1.15730584e+00 8.21877345e-02 -3.05953711e-01 1.15065128e-01 1.35491163e-01 1.43561453e-01 5.60410202e-01 4.18200880e-01 -4.85463977e-01 3.25728327e-01 -1.02229893e+00 2.82678246e-01 4.94082302e-01 8.77403080e-01 1.42169690e+00 -4.51333672e-01 -3.99529561e-02 6.70111299e-01 -1.41280433e-02 1.74403518e-01 1.59390479e-01 -1.22957516e+00 7.88187444e-01 8.72864723e-01 -1.31310239e-01 -9.32912469e-01 -4.04025704e-01 -3.82020444e-01 -8.73520732e-01 3.99930745e-01 5.02767146e-01 1.22492090e-01 -9.44587350e-01 1.35888660e+00 3.34504753e-01 4.59897697e-01 -2.15325043e-01 1.00392079e+00 8.62479031e-01 5.68955243e-01 -4.15104002e-01 5.27604282e-01 1.08962417e+00 -1.46937573e+00 -1.21583126e-01 -3.64481091e-01 3.66018236e-01 -6.94564641e-01 8.59805942e-01 3.37231547e-01 -1.25632322e+00 -5.59070170e-01 -1.31993508e+00 -7.07062423e-01 -5.60543418e-01 1.18510770e-02 5.08371055e-01 -7.51883090e-02 -1.32255101e+00 8.84802878e-01 -8.16806436e-01 -2.76925117e-01 7.80293822e-01 4.18840319e-01 -6.23008490e-01 -3.82088006e-01 -1.00652659e+00 5.75211942e-01 3.01186502e-01 3.52805495e-01 -8.11664522e-01 -8.01650643e-01 -1.14666283e+00 -1.24002313e-02 1.30247280e-01 -7.61687577e-01 9.66549993e-01 -9.41207945e-01 -1.22807264e+00 1.17551720e+00 -1.45772949e-01 -5.35500526e-01 5.93556404e-01 -4.04134274e-01 3.32160562e-01 4.21241790e-01 3.71122301e-01 1.06936669e+00 7.30174422e-01 -9.83444035e-01 -8.98882449e-01 -5.12982607e-01 3.87369215e-01 2.80997694e-01 -4.19129506e-02 -3.76758009e-01 -7.41889954e-01 -5.92369139e-01 3.62861395e-01 -7.55848646e-01 -2.53313750e-01 1.66866034e-01 -4.82674032e-01 -1.33739233e-01 8.04838717e-01 -6.42364502e-01 7.26199210e-01 -2.03043938e+00 3.78488004e-01 2.34545618e-01 3.60377312e-01 -8.80038291e-02 -1.35034114e-01 2.72128791e-01 -1.24747329e-03 -2.23854631e-01 -4.09437388e-01 -7.65426338e-01 -1.41158164e-01 1.88622922e-01 -1.10096827e-01 4.83592868e-01 4.29110229e-01 9.55477715e-01 -8.18769395e-01 -3.10140997e-01 3.70557278e-01 4.99245524e-01 -7.84041762e-01 2.70325929e-01 -1.97911069e-01 6.64977133e-01 -1.67749569e-01 8.08669567e-01 8.72676373e-01 -4.25101817e-01 -1.90898046e-01 -2.00552255e-01 -2.68198878e-01 3.59277397e-01 -1.00686920e+00 2.19666052e+00 -4.20796454e-01 7.35211074e-01 1.42147809e-01 -9.87541139e-01 7.84605503e-01 -1.43928677e-01 4.68563765e-01 -8.26175332e-01 1.20812031e-02 3.49034458e-01 -4.03341234e-01 -1.45466760e-01 5.77831507e-01 3.07779402e-01 -1.35720875e-02 5.14963493e-02 2.19892636e-01 -3.63689542e-01 2.13925093e-01 9.63206887e-02 1.14463091e+00 1.88953891e-01 3.80185172e-02 -2.80013859e-01 5.39999723e-01 -1.17047422e-01 7.20103621e-01 4.64946032e-01 -1.59766108e-01 1.01571095e+00 4.86173004e-01 -7.61479914e-01 -9.65474367e-01 -1.11183798e+00 -1.02241591e-01 7.41366565e-01 3.95356059e-01 -3.27016085e-01 -4.77783948e-01 -5.66000462e-01 9.73710418e-02 -1.32236958e-01 -8.60180318e-01 2.45034724e-01 -7.11699843e-01 -1.66344181e-01 1.35271892e-01 6.31837964e-01 9.36500907e-01 -1.11270702e+00 -6.72389746e-01 -3.81184854e-02 -1.44877270e-01 -1.32783973e+00 -6.55174851e-01 4.61930275e-01 -9.07157540e-01 -1.37015224e+00 -8.10014904e-01 -9.50410306e-01 6.75989866e-01 3.06133926e-01 1.23894715e+00 1.63081408e-01 -3.52151662e-01 1.78149462e-01 -2.00759828e-01 -1.45424172e-01 2.56438464e-01 4.24073547e-01 -3.67565721e-01 -3.07971649e-02 1.19654723e-01 -9.33960259e-01 -1.01507497e+00 2.89858490e-01 -8.99723828e-01 3.29133511e-01 7.28960335e-01 9.35277939e-01 8.24195504e-01 -2.99571395e-01 -1.10801263e-03 -7.25678086e-01 -3.60091552e-02 -3.52474034e-01 -9.35594738e-01 -3.23798740e-03 -2.79990524e-01 -6.81333197e-03 3.60075682e-01 1.90728873e-01 -7.45049477e-01 2.76995122e-01 -1.05337933e-01 -6.15901709e-01 -1.31415993e-01 2.61476189e-01 -1.88764587e-01 -3.43650997e-01 3.36596727e-01 1.33742288e-01 -1.36775553e-01 -3.75655234e-01 1.48870155e-01 6.44420683e-02 5.14078557e-01 -5.58777153e-01 8.35975707e-01 7.66465068e-01 1.77572727e-01 -5.55684447e-01 -1.24197066e+00 -6.39528394e-01 -9.91247654e-01 -9.31208283e-02 1.04672730e+00 -1.20379961e+00 -3.41637403e-01 6.32417262e-01 -1.05185378e+00 -6.44742787e-01 -2.95334280e-01 2.45165139e-01 -5.99985123e-01 2.27527767e-01 -7.65069008e-01 -1.45043999e-01 -3.02917033e-01 -1.05980086e+00 1.38238490e+00 3.93309116e-01 1.27548799e-01 -9.74787056e-01 8.89729410e-02 6.57255113e-01 1.57274365e-01 3.86920542e-01 5.28566420e-01 -1.44175649e-01 -1.30590880e+00 -7.40294755e-02 -4.89070714e-01 4.36682284e-01 -1.00428602e-02 -1.35971755e-01 -1.10457873e+00 -2.93990850e-01 -2.66319633e-01 -5.12597442e-01 1.22966278e+00 5.10064065e-01 1.27878428e+00 -1.49474606e-01 -3.57128642e-02 1.13552129e+00 1.54145515e+00 -2.76930779e-01 7.41261482e-01 5.88636696e-01 8.87498438e-01 7.02667415e-01 3.77320319e-01 3.30791146e-01 5.66546082e-01 6.70697033e-01 7.97587752e-01 -2.67224401e-01 -2.83729523e-01 -4.02655274e-01 8.67449865e-02 5.31294167e-01 -5.12588955e-02 1.90503374e-01 -9.37429130e-01 6.34477139e-01 -1.81301618e+00 -7.15941250e-01 3.22231352e-02 2.04048514e+00 4.82264072e-01 3.36515456e-01 -7.16762841e-02 -1.27173513e-02 4.71957505e-01 3.66219550e-01 -5.39280951e-01 -1.03251025e-01 -4.62569535e-01 3.78390700e-01 7.27458358e-01 6.82759345e-01 -1.40965164e+00 9.73310113e-01 4.62873411e+00 7.40608692e-01 -1.08189917e+00 1.28631994e-01 7.72475719e-01 -2.14790508e-01 -1.14515208e-01 8.24598670e-02 -6.45503819e-01 1.78697392e-01 2.02403247e-01 1.92698032e-01 2.74366170e-01 5.35495222e-01 1.17050409e-02 -3.76980245e-01 -1.11069405e+00 1.06305599e+00 8.27926490e-03 -1.63848138e+00 -1.22920237e-03 1.07711680e-01 1.26280653e+00 5.72014630e-01 3.07323724e-01 9.97881964e-02 1.26220852e-01 -1.06352782e+00 6.61564469e-01 4.34621751e-01 4.75095689e-01 -6.75788403e-01 6.22010887e-01 1.99480414e-01 -1.40738070e+00 -1.83614299e-01 -2.47623712e-01 -2.13713065e-01 -1.00175709e-01 6.50248110e-01 -4.53099549e-01 8.19374859e-01 1.08600628e+00 1.26647317e+00 -5.81611872e-01 1.23351705e+00 -3.94784957e-01 1.14496490e-02 -4.01840895e-01 4.73811537e-01 5.86881757e-01 -1.44676968e-01 3.75035495e-01 8.88607442e-01 3.04303825e-01 -1.25071913e-01 2.59332538e-01 7.72858620e-01 -2.50739962e-01 -2.31698900e-02 -6.53507471e-01 4.36799377e-01 1.70418799e-01 1.24766660e+00 -6.47954226e-01 -1.83769628e-01 -5.05434513e-01 1.12850618e+00 8.26926231e-01 1.91950470e-01 -5.24107933e-01 -2.26508528e-01 6.74908936e-01 2.64344722e-01 4.62219983e-01 -3.55096430e-01 -5.47848344e-01 -1.27679026e+00 2.09313080e-01 -5.14945924e-01 4.65486646e-01 -7.67219186e-01 -1.21834135e+00 6.91743076e-01 -1.97954729e-01 -1.29387164e+00 2.26351559e-01 -7.28936374e-01 -7.53038168e-01 9.06063259e-01 -1.92862260e+00 -1.20218444e+00 -7.04542160e-01 7.95850039e-01 6.62854970e-01 1.83965508e-02 4.70209241e-01 3.45062405e-01 -5.51871002e-01 4.80897278e-01 -6.78183138e-02 2.06393689e-01 6.52189434e-01 -1.35917592e+00 3.81578654e-01 7.45540500e-01 -3.29971053e-02 2.49420792e-01 2.74540126e-01 -4.25164849e-01 -9.50982273e-01 -1.30869865e+00 8.59302461e-01 -3.86725634e-01 5.74190140e-01 -4.92456406e-01 -8.40176642e-01 6.02270603e-01 3.42078477e-01 4.37651455e-01 1.71323732e-01 -6.24871030e-02 -5.75407863e-01 -3.07658911e-01 -9.87688899e-01 4.14879590e-01 1.14411581e+00 -7.19062328e-01 -3.44572753e-01 4.28531975e-01 6.97973728e-01 -7.56347418e-01 -7.92808592e-01 4.59645927e-01 3.75256747e-01 -1.38601112e+00 9.38147187e-01 -1.64447241e-02 1.00259590e+00 -3.00386399e-01 -2.64324039e-01 -1.04894221e+00 -1.21587543e-02 -4.29095119e-01 7.31929168e-02 6.97150826e-01 4.40267384e-01 -6.65346980e-01 1.17664921e+00 3.57529372e-01 -3.36022824e-01 -1.02408171e+00 -1.09270334e+00 -7.82308817e-01 1.70505568e-01 -1.84816211e-01 3.33220959e-01 8.53343904e-01 -2.36533135e-01 1.31491825e-01 -2.34591424e-01 3.49344015e-01 8.12407553e-01 2.76204705e-01 7.78508008e-01 -1.07307220e+00 -2.39078328e-01 -5.56934118e-01 -6.13667130e-01 -1.36438560e+00 2.09969819e-01 -1.06350350e+00 -6.83271035e-04 -1.51594555e+00 1.86101258e-01 -4.14032698e-01 -2.41433248e-01 5.72528064e-01 -1.15965076e-01 5.44488668e-01 1.82315558e-01 2.85652995e-01 -5.91049016e-01 6.84348464e-01 1.28157675e+00 -2.57024556e-01 -9.64924917e-02 -1.88638702e-01 -2.20400959e-01 6.49195015e-01 9.40958679e-01 -2.22035527e-01 -3.68525207e-01 -6.95884705e-01 1.04011796e-01 -1.36001152e-03 7.92537749e-01 -1.32868624e+00 4.96858805e-01 1.26743972e-01 4.16208476e-01 -8.83173287e-01 5.42435527e-01 -7.68402874e-01 -6.07242025e-02 3.76862675e-01 -2.27279291e-01 2.32948586e-01 3.41778368e-01 4.58233625e-01 -5.76414108e-01 1.74655795e-01 7.29183018e-01 -1.91555023e-01 -9.86371100e-01 8.05602551e-01 1.35027915e-01 1.50275722e-01 8.48428130e-01 -2.35980943e-01 -2.26249903e-01 -8.53489786e-02 -6.15904570e-01 4.83776659e-01 6.35833681e-01 3.28344464e-01 8.33036542e-01 -1.24076664e+00 -4.62325007e-01 2.85958111e-01 1.83418691e-01 6.15985811e-01 5.51425993e-01 8.41927707e-01 -8.97586167e-01 4.35721636e-01 -5.64322531e-01 -7.87977755e-01 -9.18573380e-01 1.84427530e-01 5.96560836e-01 -5.27455330e-01 -6.79958105e-01 1.14546919e+00 5.53143561e-01 -5.00218749e-01 4.14043307e-01 -5.04740119e-01 -1.23809926e-01 -1.32380977e-01 2.91651636e-01 1.31677920e-02 2.09468514e-01 -5.36776006e-01 -2.50982791e-01 8.83349061e-01 -1.01323940e-01 4.78261262e-02 1.44335771e+00 -1.06865153e-01 -2.35586300e-01 1.80708855e-01 1.54318333e+00 -4.20453697e-01 -1.74911654e+00 -5.10704219e-01 2.09868364e-02 -5.81135035e-01 6.88219145e-02 -4.65971857e-01 -1.34579360e+00 9.51849401e-01 4.28974986e-01 -2.45456845e-01 1.25398505e+00 1.98606569e-02 9.81584311e-01 2.96690464e-01 2.95130283e-01 -1.04992008e+00 3.26028496e-01 7.84400880e-01 1.10663164e+00 -1.64454210e+00 -1.74081072e-01 -4.07283962e-01 -2.36569107e-01 1.07506347e+00 9.11302388e-01 -6.18085682e-01 9.73150551e-01 8.95767286e-02 -8.15975200e-03 -4.35506761e-01 -6.79117501e-01 -3.06734025e-01 5.18552423e-01 3.52011353e-01 1.33314922e-01 -1.79201171e-01 1.48038596e-01 3.46065760e-01 -1.56081498e-01 -1.88078851e-01 3.24588865e-01 1.06400681e+00 -3.36311936e-01 -9.26331639e-01 -1.78149447e-01 2.23620757e-01 -2.29049951e-01 -3.31063598e-01 -2.89438009e-01 7.51723766e-01 3.68137896e-01 6.33244991e-01 3.23774636e-01 -3.43362033e-01 4.91116285e-01 -3.38872939e-01 5.12709141e-01 -6.46177232e-01 -5.57333708e-01 -2.28930742e-01 -2.40032136e-01 -1.05560434e+00 -5.51237643e-01 -5.82198858e-01 -1.04124117e+00 -2.22335681e-01 3.28735083e-01 -1.92940801e-01 3.04454058e-01 8.27780306e-01 2.99530059e-01 3.53912264e-01 6.65966749e-01 -1.35480571e+00 7.49011487e-02 -6.42125368e-01 -6.18853509e-01 2.39393398e-01 6.23390377e-01 -6.16224587e-01 -3.99787366e-01 -1.09978825e-01]
[8.754830360412598, -2.376689910888672]
466147c3-703d-41ab-8185-85c6a4c948c5
pdanet-polarity-consistent-deep-attention
1909.05693
null
https://arxiv.org/abs/1909.05693v1
https://arxiv.org/pdf/1909.05693v1.pdf
PDANet: Polarity-consistent Deep Attention Network for Fine-grained Visual Emotion Regression
Existing methods on visual emotion analysis mainly focus on coarse-grained emotion classification, i.e. assigning an image with a dominant discrete emotion category. However, these methods cannot well reflect the complexity and subtlety of emotions. In this paper, we study the fine-grained regression problem of visual emotions based on convolutional neural networks (CNNs). Specifically, we develop a Polarity-consistent Deep Attention Network (PDANet), a novel network architecture that integrates attention into a CNN with an emotion polarity constraint. First, we propose to incorporate both spatial and channel-wise attentions into a CNN for visual emotion regression, which jointly considers the local spatial connectivity patterns along each channel and the interdependency between different channels. Second, we design a novel regression loss, i.e. polarity-consistent regression (PCR) loss, based on the weakly supervised emotion polarity to guide the attention generation. By optimizing the PCR loss, PDANet can generate a polarity preserved attention map and thus improve the emotion regression performance. Extensive experiments are conducted on the IAPS, NAPS, and EMOTIC datasets, and the results demonstrate that the proposed PDANet outperforms the state-of-the-art approaches by a large margin for fine-grained visual emotion regression. Our source code is released at: https://github.com/ZizhouJia/PDANet.
['Guiguang Ding', 'Leida Li', 'Kurt Keutzer', 'Zizhou Jia', 'Sicheng Zhao', 'Hui Chen']
2019-09-11
null
null
null
null
['deep-attention', 'deep-attention']
['computer-vision', 'natural-language-processing']
[-2.40510508e-01 -4.12379980e-01 -1.11641608e-01 -4.78808343e-01 -2.28503883e-01 -3.47900271e-01 1.50993213e-01 -5.06991446e-02 -1.39356166e-01 5.50580561e-01 2.45347708e-01 6.63755760e-02 1.04011618e-01 -5.54337502e-01 -6.93640053e-01 -6.98227882e-01 1.98153049e-01 -1.49869844e-01 -2.46459469e-01 -2.62525439e-01 1.86202586e-01 4.86439794e-01 -1.22142923e+00 4.02566880e-01 1.06359017e+00 1.59422290e+00 -1.38007939e-01 4.80071753e-01 -1.76178500e-01 1.17461002e+00 -5.77387929e-01 -3.16381037e-01 -1.46826148e-01 -4.95364934e-01 -5.34040570e-01 -8.19110349e-02 1.42968073e-01 1.13464795e-01 -2.09054932e-01 1.13107324e+00 5.58380485e-01 9.25153866e-02 6.23008132e-01 -1.69545197e+00 -9.64148760e-01 2.16165066e-01 -1.03911781e+00 1.75539508e-01 -9.43210125e-02 -6.28590658e-02 1.07704973e+00 -8.35722625e-01 4.16369051e-01 1.26976883e+00 6.92016661e-01 3.93051445e-01 -8.88480365e-01 -9.36304450e-01 7.31614292e-01 3.96180063e-01 -1.59415948e+00 1.07469270e-03 1.29521239e+00 -3.67999494e-01 7.06424534e-01 1.13878608e-01 7.41832316e-01 9.75459933e-01 1.01932436e-01 7.84722984e-01 1.18045783e+00 -1.93213046e-01 2.45744102e-02 2.56764233e-01 1.03482148e-02 8.05215418e-01 -2.35480011e-01 -2.56312072e-01 -4.20857102e-01 1.87022433e-01 7.10390568e-01 -6.34913100e-03 -3.55396569e-01 -3.86364520e-01 -8.07349384e-01 8.76552641e-01 1.04575181e+00 1.72533095e-01 -4.24416065e-01 1.97355285e-01 6.64730191e-01 2.28319660e-01 8.58041108e-01 2.10466042e-01 -5.43224275e-01 -5.69994785e-02 -4.58109260e-01 -4.07215990e-02 4.74171966e-01 6.93811655e-01 8.46978366e-01 8.25899765e-02 -2.72994131e-01 1.07950711e+00 2.31538594e-01 2.18485266e-01 2.34474152e-01 -6.24431014e-01 3.88571411e-01 8.29667509e-01 -1.90701708e-01 -1.66387248e+00 -6.55245662e-01 -4.06301677e-01 -1.27707684e+00 3.04520547e-01 5.50307333e-02 -5.31564355e-01 -8.08307052e-01 1.89087248e+00 3.12627584e-01 6.69599175e-02 -1.47154525e-01 1.32408357e+00 1.14496398e+00 9.12200511e-01 3.10905308e-01 -1.13670118e-01 1.23229253e+00 -1.31757390e+00 -7.96227157e-01 -2.30541602e-01 2.53956169e-01 -4.45943177e-01 1.37223721e+00 3.93012106e-01 -9.05053139e-01 -6.02446735e-01 -9.48864996e-01 -1.39369324e-01 -5.83404005e-01 4.35462654e-01 6.76564872e-01 2.78208107e-01 -9.16562855e-01 2.73726374e-01 -4.33939129e-01 -1.00762799e-01 7.14173913e-01 1.86067507e-01 -3.89963686e-01 1.53063431e-01 -1.29412818e+00 6.28325641e-01 -1.25628039e-01 6.80619836e-01 -4.38133091e-01 -5.96134663e-01 -1.00558650e+00 2.57151186e-01 1.16757177e-01 -4.02880847e-01 8.27778280e-01 -1.68424618e+00 -1.46006417e+00 6.62861049e-01 -3.83039981e-01 2.02752203e-01 3.38605642e-01 -2.71139387e-02 -4.54826266e-01 3.23100716e-01 -1.30894464e-02 8.28034759e-01 7.88095057e-01 -1.53558791e+00 -4.92705911e-01 -3.51091713e-01 -4.19409983e-02 4.02894944e-01 -6.06633663e-01 1.64297372e-01 -9.37092006e-01 -7.47834444e-01 -2.26704642e-01 -7.41042137e-01 -2.65577376e-01 1.15782574e-01 -6.54682815e-01 -1.77866012e-01 8.70985866e-01 -5.72628617e-01 1.33344436e+00 -2.19831657e+00 4.53448921e-01 4.14157391e-01 4.59018618e-01 2.03527007e-02 -3.12673926e-01 -1.04304165e-01 -3.55323732e-01 3.50985169e-01 -1.72574013e-01 -3.23319018e-01 5.35729602e-02 -8.21199045e-02 -1.06385380e-01 4.28961903e-01 6.24522030e-01 1.23370349e+00 -5.44080257e-01 -5.48935056e-01 1.72090262e-01 7.28965223e-01 -3.47677439e-01 2.03808904e-01 4.54607718e-02 5.69094598e-01 -5.06695569e-01 9.23596919e-01 7.70695865e-01 -4.42709059e-01 4.14771065e-02 -6.13665581e-01 -1.41405717e-01 -2.62484431e-01 -8.08810115e-01 1.38294852e+00 -6.19851947e-01 8.62032890e-01 2.57896304e-01 -1.03021216e+00 1.22587383e+00 9.35385674e-02 4.72285002e-01 -1.02629435e+00 4.52295631e-01 -1.68337211e-01 -1.95071086e-01 -4.09898847e-01 4.29611385e-01 -1.49816334e-01 -3.92959982e-01 2.35711351e-01 -1.11124784e-01 2.37599477e-01 -1.70035079e-01 6.01128079e-02 6.11793756e-01 -3.06176692e-02 2.04243854e-01 -1.33059561e-01 3.92975628e-01 -3.35110985e-02 7.63993204e-01 3.85611892e-01 -4.74618018e-01 6.83877289e-01 9.95198548e-01 -5.21763384e-01 -7.66710579e-01 -5.85184097e-01 2.41834279e-02 1.20471513e+00 5.79454839e-01 -2.37923548e-01 -5.80799639e-01 -8.54144573e-01 -9.68873873e-02 -5.39303906e-02 -1.09836829e+00 -2.79686153e-01 -3.58049452e-01 -8.39835644e-01 3.14297229e-01 7.07853913e-01 7.06769705e-01 -1.56363010e+00 -1.40781149e-01 -8.33137929e-02 -3.42949748e-01 -9.34541225e-01 -3.70853126e-01 2.18208611e-01 -2.44773775e-01 -9.37534332e-01 -7.81393707e-01 -9.38997567e-01 7.40192115e-01 7.33707696e-02 1.15198743e+00 1.73090860e-01 -2.01129630e-01 1.97849035e-01 -6.16452038e-01 -4.30815339e-01 5.63890398e-01 1.98457450e-01 -4.98980999e-01 3.05967540e-01 3.87316555e-01 -4.91911650e-01 -7.39963591e-01 2.37698182e-01 -6.01726890e-01 9.22983140e-03 5.61478615e-01 9.18345332e-01 8.65543127e-01 2.69525856e-01 6.66370571e-01 -8.89194369e-01 9.37152147e-01 -6.63985372e-01 -2.25118354e-01 2.81155705e-01 -2.54525125e-01 -2.79802203e-01 8.96751642e-01 -5.82707226e-01 -1.12665796e+00 5.78541756e-02 -2.63346195e-01 -8.38811517e-01 -3.03058088e-01 5.77645004e-01 -3.96466881e-01 -4.20857817e-01 2.30651274e-01 4.99008447e-02 -3.67351353e-01 -4.92062382e-02 4.80638176e-01 3.67425889e-01 2.13836744e-01 -4.97111708e-01 3.63981783e-01 4.75553185e-01 -2.63857007e-01 -5.23643255e-01 -7.65729785e-01 -2.47961387e-01 -2.95568407e-01 -4.15964752e-01 9.75384235e-01 -8.72545779e-01 -1.05845225e+00 6.50681019e-01 -9.79040086e-01 -6.70135319e-01 8.81135017e-02 1.92325130e-01 -3.37234110e-01 6.37874678e-02 -7.42737710e-01 -5.82550466e-01 -3.29459846e-01 -1.08459818e+00 1.03261554e+00 5.60294628e-01 -1.07476242e-01 -1.11142266e+00 -1.10983849e-01 -3.37763429e-02 3.25984508e-01 6.06378376e-01 7.90061295e-01 -5.78290112e-02 -1.04930989e-01 -6.71133995e-02 -9.17177022e-01 2.66453087e-01 -7.40699619e-02 2.23661870e-01 -9.01057839e-01 5.90494350e-02 -3.67629141e-01 -6.92548633e-01 1.07862973e+00 5.94472110e-01 1.74664092e+00 -2.48878419e-01 -1.49673209e-01 9.75933790e-01 1.35917139e+00 1.48236006e-01 7.91435242e-01 3.83462459e-01 1.26382637e+00 7.56678402e-01 7.36203611e-01 6.15744233e-01 7.23334968e-01 5.55770278e-01 6.37747407e-01 -7.82014608e-01 1.24194529e-02 -8.09683055e-02 -1.22705415e-01 7.08450377e-01 -8.85189697e-02 -3.57982099e-01 -6.36210024e-01 4.60027456e-01 -2.04396462e+00 -7.02245653e-01 -9.28357691e-02 1.39981854e+00 6.16039455e-01 -8.04398358e-02 -3.65701877e-02 -1.16097756e-01 8.27472270e-01 4.43635195e-01 -7.21340775e-01 -7.68650055e-01 -2.51364499e-01 1.97088867e-01 2.00916514e-01 3.87517720e-01 -1.30937040e+00 1.09743619e+00 4.83538342e+00 8.16871464e-01 -1.69601214e+00 1.69429984e-02 1.21941245e+00 -1.60710156e-01 -3.41526806e-01 -4.89978850e-01 -4.76747125e-01 5.89380324e-01 2.46199086e-01 1.53113246e-01 4.54137832e-01 9.17608976e-01 2.28143662e-01 2.13589132e-01 -4.03297007e-01 1.14263105e+00 2.90849924e-01 -1.12941289e+00 -6.87396452e-02 -3.17941576e-01 7.94392765e-01 -2.74283409e-01 2.51252592e-01 4.18205321e-01 1.28862292e-01 -1.22273004e+00 7.69503772e-01 7.22023010e-01 8.32287014e-01 -1.20985281e+00 7.25488722e-01 -2.12684348e-01 -1.62359285e+00 -1.24178648e-01 -2.91163325e-01 2.81908526e-03 2.00703908e-02 4.77540374e-01 8.54723230e-02 3.42417479e-01 1.13479364e+00 1.12190318e+00 -4.67564076e-01 7.46811748e-01 -4.37274575e-01 3.97308439e-01 7.62958750e-02 -2.45968118e-01 4.07151520e-01 -1.74088597e-01 3.32368575e-02 1.40135622e+00 6.76149353e-02 2.63722837e-01 2.38843635e-02 8.15874875e-01 -3.95172238e-01 2.93510139e-01 -1.94605455e-01 1.77581817e-01 3.20521533e-01 1.85011208e+00 -7.17622519e-01 -1.71169430e-01 -3.57695460e-01 1.11967862e+00 9.22422111e-01 6.70978189e-01 -1.09956491e+00 -8.19547355e-01 9.11307871e-01 -4.91797656e-01 5.00511229e-01 2.56694347e-01 -6.03112578e-01 -1.07082570e+00 1.59841061e-01 -8.58840525e-01 3.06429684e-01 -1.15311527e+00 -1.54476011e+00 1.04071856e+00 -5.42022169e-01 -1.06577027e+00 3.48631591e-01 -7.13709772e-01 -1.04273689e+00 9.36124623e-01 -1.88291121e+00 -1.25643492e+00 -8.15031469e-01 9.42198277e-01 1.82811037e-01 2.55949467e-01 5.71163952e-01 3.97182167e-01 -9.34202135e-01 8.10114264e-01 -2.36936048e-01 1.74463183e-01 7.43467033e-01 -1.19874620e+00 -1.70001671e-01 5.33280611e-01 -2.96484828e-01 2.86913186e-01 3.42587441e-01 -3.69995236e-01 -1.02931762e+00 -1.34605932e+00 5.96108913e-01 3.09237689e-02 6.33568525e-01 -4.71763700e-01 -1.00817657e+00 4.23326671e-01 3.61219615e-01 4.16844726e-01 5.00241280e-01 3.15895319e-01 -5.62603712e-01 -2.79222727e-01 -9.97042179e-01 6.23001456e-01 7.03290403e-01 -5.40620983e-01 -1.32134737e-04 4.77082767e-02 7.02997983e-01 -3.35072517e-01 -7.92718530e-01 5.21912932e-01 6.75633073e-01 -8.21754336e-01 8.51520658e-01 -5.10209382e-01 8.60981703e-01 -2.72595257e-01 1.68712705e-01 -1.59559894e+00 -6.20171487e-01 -1.02373004e-01 1.24614254e-01 1.23607159e+00 4.58852798e-01 -3.93570632e-01 6.36259377e-01 4.20111448e-01 -2.11635306e-01 -1.24391210e+00 -5.31982660e-01 -1.05901718e-01 1.08736932e-01 -4.16970342e-01 5.40124357e-01 1.35418355e+00 8.72905478e-02 3.40187520e-01 -6.51372671e-01 1.69927418e-01 9.71793458e-02 3.15261632e-01 6.18362963e-01 -9.87765968e-01 1.36659652e-01 -9.20708776e-01 -2.82596320e-01 -7.99551964e-01 4.28868532e-01 -4.18029606e-01 2.36874655e-01 -1.67443705e+00 2.12873340e-01 -5.75188100e-01 -5.88483334e-01 6.99905694e-01 -4.24210310e-01 7.43957758e-01 9.58060175e-02 -1.29355758e-01 -9.97274518e-01 8.78133476e-01 1.45565736e+00 -1.33658871e-01 -2.52746075e-01 -4.97731358e-01 -1.12641084e+00 7.47493207e-01 1.22553313e+00 -2.59070009e-01 -3.12698811e-01 -2.69905239e-01 4.87381905e-01 -1.57346576e-01 4.72793847e-01 -5.89684665e-01 2.01232255e-01 -2.60770440e-01 8.23661327e-01 -5.10554373e-01 3.42162430e-01 -8.38165164e-01 -2.74833649e-01 -3.74495089e-02 -4.28092271e-01 1.10356644e-01 4.83629227e-01 4.20030534e-01 -6.31497741e-01 4.15646523e-01 7.87223756e-01 -7.58085726e-03 -1.02635372e+00 6.66755021e-01 -2.08787650e-01 1.72272459e-01 1.04645395e+00 3.96145619e-02 -5.71407616e-01 -4.18864101e-01 -7.34195828e-01 4.66395706e-01 2.45554969e-01 5.29455066e-01 7.34420300e-01 -1.61396468e+00 -5.07286727e-01 9.75737348e-02 2.90687472e-01 -1.54975012e-01 7.80697405e-01 9.49335158e-01 -2.92111367e-01 2.44334284e-02 -5.79645932e-01 -3.39932919e-01 -1.27729142e+00 4.71758544e-01 4.85554785e-01 -3.16878706e-01 -2.90309668e-01 1.10017908e+00 6.78218186e-01 -6.49396956e-01 2.36237422e-01 -1.73149928e-01 -6.53330982e-01 1.05929285e-01 5.07304192e-01 -1.71076898e-02 -1.71298087e-01 -8.03555131e-01 -4.71383363e-01 8.67732227e-01 3.92242409e-02 2.88065493e-01 1.32527208e+00 -3.30491751e-01 -3.44599694e-01 3.70532632e-01 1.39580452e+00 4.49253544e-02 -1.51990998e+00 -3.62587124e-02 -5.90478301e-01 -3.84489268e-01 1.49339542e-01 -9.79234457e-01 -1.78805017e+00 1.16316032e+00 4.52453434e-01 2.11633965e-01 1.75576687e+00 7.68442601e-02 5.07006764e-01 -6.02460057e-02 -3.49151731e-01 -1.04667854e+00 3.20925415e-01 6.01484716e-01 1.05602741e+00 -1.33862233e+00 -2.38427743e-01 -3.79142284e-01 -1.35758662e+00 1.03300846e+00 1.13074541e+00 -2.34208107e-01 7.60055482e-01 2.24932134e-01 3.80061030e-01 -4.90402400e-01 -6.73939168e-01 -2.59702682e-01 4.65044260e-01 5.23320258e-01 4.77575600e-01 1.49371460e-01 -1.04051962e-01 1.05687773e+00 9.62017849e-02 -1.25549331e-01 4.99396324e-02 4.55337077e-01 -1.81497082e-01 -6.41978443e-01 -1.58122465e-01 3.12767774e-01 -4.09792781e-01 -3.56225848e-01 -7.36620367e-01 8.00006390e-01 3.44795793e-01 8.37501526e-01 3.39942247e-01 -6.29515171e-01 4.28581923e-01 -3.79419774e-01 1.48902729e-01 -1.11917548e-01 -6.59020841e-01 8.26349705e-02 2.97071505e-02 -6.50700092e-01 -4.21952933e-01 -1.84350178e-01 -1.28235650e+00 -3.51276755e-01 -6.28712252e-02 1.00014552e-01 4.95407373e-01 6.79116786e-01 5.27342677e-01 9.14574325e-01 9.44163144e-01 -9.21647251e-01 4.41543400e-01 -8.11362088e-01 -6.95023179e-01 4.26886439e-01 3.27519953e-01 -7.10628748e-01 -3.01941246e-01 -1.82738110e-01]
[13.536092758178711, 1.7609022855758667]
d66e1393-8026-4a38-a7ea-f77b2d24558e
moving-objects-detection-with-a-moving-camera
2001.05238
null
https://arxiv.org/abs/2001.05238v1
https://arxiv.org/pdf/2001.05238v1.pdf
Moving Objects Detection with a Moving Camera: A Comprehensive Review
During about 30 years, a lot of research teams have worked on the big challenge of detection of moving objects in various challenging environments. First applications concern static cameras but with the rise of the mobile sensors studies on moving cameras have emerged over time. In this survey, we propose to identify and categorize the different existing methods found in the literature. For this purpose, we propose to classify these methods according to the choose of the scene representation: one plane or several parts. Inside these two categories, the methods are grouped according to eight different approaches: panoramic background subtraction, dual cameras, motion compensation, subspace segmentation, motion segmentation, plane+parallax, multi planes and split image in blocks. A reminder of methods for static cameras is provided as well as the challenges with both static and moving cameras. Publicly available datasets and evaluation metrics are also surveyed in this paper.
['Marie-Neige Chapel', 'Thierry Bouwmans']
2020-01-15
null
null
null
null
['motion-segmentation']
['computer-vision']
[ 5.65219700e-01 -6.63867474e-01 -1.14859931e-01 1.84417367e-02 -2.92809039e-01 -1.00948942e+00 7.06589460e-01 -4.40393150e-01 -4.67152566e-01 4.32602525e-01 2.24191379e-02 -1.19701317e-02 3.12193017e-02 -2.36582488e-01 -3.03044349e-01 -8.80429864e-01 3.01362216e-01 -1.05081506e-01 9.28252459e-01 8.82139057e-02 4.42133069e-01 5.07322192e-01 -1.84859145e+00 4.23862398e-01 5.07957697e-01 7.23076165e-01 2.48383567e-01 1.03801143e+00 1.92784846e-01 8.12929988e-01 -5.56944013e-01 -5.17110646e-01 6.29168928e-01 -5.03808916e-01 -6.78835750e-01 5.94193637e-01 8.71444464e-01 -4.59606528e-01 -3.38415653e-01 1.31461167e+00 4.61522967e-01 1.18530691e-01 2.97205299e-01 -1.39743173e+00 1.13894552e-01 1.20302700e-01 -7.87216544e-01 8.79724085e-01 7.28926003e-01 2.76592523e-02 2.50848800e-01 -9.32041168e-01 7.08995581e-01 1.24352705e+00 6.60893440e-01 5.07606566e-01 -7.01736510e-01 -1.69018596e-01 4.28183705e-01 6.05649233e-01 -1.21715438e+00 -6.40201628e-01 6.14445984e-01 -7.10357845e-01 7.76158571e-01 5.36435723e-01 5.57914317e-01 1.01541793e+00 6.20725155e-02 7.48617768e-01 1.21673191e+00 -6.08234763e-01 -5.39892279e-02 2.00027645e-01 4.56038356e-01 3.47256750e-01 5.85580289e-01 -1.24494135e-01 -3.58832538e-01 -1.09443523e-01 5.65180302e-01 1.22032195e-01 -5.67635655e-01 -6.99684978e-01 -1.42773271e+00 2.55825460e-01 -5.10749519e-01 3.26528549e-01 4.21282873e-02 -3.75631988e-01 3.68528217e-01 -8.96733031e-02 6.13957755e-02 -2.21376285e-01 -3.75066370e-01 -1.08398154e-01 -1.11557543e+00 2.80869037e-01 6.49316788e-01 1.05475700e+00 4.28916454e-01 5.98178804e-02 7.87118003e-02 7.02855885e-01 1.38272777e-01 6.34536505e-01 5.93309760e-01 -1.03665328e+00 6.42859578e-01 2.09764346e-01 2.67654419e-01 -1.16213131e+00 -4.13143784e-01 1.30920246e-01 -8.65066171e-01 1.92122921e-01 6.24541283e-01 -3.62178832e-01 -6.36819541e-01 9.83312905e-01 5.30395091e-01 4.14934188e-01 7.81540573e-03 8.21833134e-01 9.20090735e-01 5.80904543e-01 -2.78448701e-01 -4.46828753e-01 1.44741321e+00 -1.25990117e+00 -8.03585231e-01 -2.54664838e-01 2.35977788e-02 -1.04648125e+00 2.49631867e-01 8.70624185e-01 -1.20783329e+00 -7.25268722e-01 -9.79778647e-01 1.93937629e-01 -5.88868678e-01 2.89313167e-01 4.84409451e-01 1.18736482e+00 -1.10626781e+00 3.44594747e-01 -7.98032761e-01 -8.11099648e-01 1.17270023e-01 3.91463280e-01 -2.03105345e-01 -1.34098426e-01 -6.54460788e-01 8.91540587e-01 2.55927593e-01 7.05991760e-02 -8.37428987e-01 -2.18447417e-01 -6.14394426e-01 -3.75698030e-01 4.53451902e-01 -8.58278453e-01 1.13874519e+00 -1.16303957e+00 -1.50466907e+00 1.16732478e+00 -3.31909984e-01 -2.86665410e-01 6.89584792e-01 -4.36485320e-01 -7.97292948e-01 3.82236749e-01 -1.19090192e-01 3.53822470e-01 9.16985631e-01 -1.13688099e+00 -1.39495838e+00 -4.66433257e-01 1.82307601e-01 4.98524457e-01 -9.40840691e-02 6.45150900e-01 -9.82051969e-01 -5.17447591e-01 -7.63945356e-02 -9.48253214e-01 -2.58767724e-01 -5.37250698e-01 -3.24673712e-01 2.02356935e-01 1.20400047e+00 -7.01114058e-01 1.41576445e+00 -1.97135663e+00 2.55371422e-01 -3.84269595e-01 6.77459314e-02 4.37455177e-01 2.56774932e-01 1.87138915e-01 -1.30468860e-01 -3.40763271e-01 -2.39696264e-01 -1.55962154e-01 -4.07065481e-01 -1.34808272e-01 -2.78617978e-01 8.88071895e-01 -3.53704035e-01 4.12106574e-01 -6.68761790e-01 -5.39765418e-01 7.66013265e-01 2.52377868e-01 -1.17233448e-01 -1.06954262e-01 4.43996102e-01 3.81521791e-01 -6.40044734e-02 9.13356781e-01 1.21383691e+00 1.79570824e-01 1.00158177e-01 -3.11289430e-01 -6.20417714e-01 -3.85918647e-01 -1.86893058e+00 1.23654497e+00 1.77133039e-01 8.19777131e-01 4.68139142e-01 -7.85392225e-01 2.92552799e-01 2.89704621e-01 6.81444526e-01 -1.36285380e-01 8.81752595e-02 1.33361489e-01 -1.07045226e-01 -5.55061698e-01 8.23501587e-01 2.63525993e-01 1.67434782e-01 -2.37552255e-01 1.02971541e-02 5.47441989e-02 6.02070510e-01 -1.03102267e-01 8.56998086e-01 2.46709615e-01 6.47175312e-01 -5.36906235e-02 1.20426261e+00 1.96471363e-01 6.36281908e-01 7.42846131e-01 -6.90230608e-01 9.23142850e-01 -1.74540713e-01 -4.81467813e-01 -6.88083410e-01 -9.05249655e-01 -6.61911666e-02 7.62688220e-01 7.56633520e-01 -3.27380419e-01 -1.00704205e+00 -4.43293154e-01 -4.55010414e-01 6.95254281e-02 -2.63194323e-01 3.99396360e-01 -7.96984911e-01 -1.17955899e+00 3.58288646e-01 3.44392300e-01 7.63309300e-01 -5.62695444e-01 -1.04266846e+00 -1.89224720e-01 -5.02413452e-01 -1.63581300e+00 -3.30982804e-01 -1.80427656e-01 -1.08565545e+00 -1.26125133e+00 -1.07482111e+00 -7.17395723e-01 4.05352622e-01 1.27434671e+00 1.07003593e+00 -4.03399289e-01 -3.04493189e-01 1.09295619e+00 -3.89667392e-01 -2.78989911e-01 1.12572148e-01 -3.26104552e-01 2.09091365e-01 3.63554865e-01 6.63148701e-01 -2.94835031e-01 -7.92374969e-01 7.18462825e-01 -1.00974977e+00 -4.06514406e-02 4.03411835e-01 1.06486760e-01 3.28824729e-01 1.35765644e-02 -4.15153533e-01 -6.39018714e-01 -3.40023115e-02 -3.92605871e-01 -9.49889004e-01 3.12308967e-01 -1.31699771e-01 -7.93095291e-01 2.35836625e-01 -3.82694691e-01 -1.16740334e+00 5.07201195e-01 3.34367394e-01 -3.19300354e-01 -6.81567609e-01 -3.24612170e-01 -3.54850113e-01 -3.99623245e-01 4.32643831e-01 3.85877401e-01 -5.65333366e-01 -4.11443323e-01 4.16158289e-01 7.00182080e-01 6.97983623e-01 -7.21430629e-02 8.16468239e-01 1.21211851e+00 -1.46810621e-01 -1.31018734e+00 -4.38390404e-01 -1.27861238e+00 -1.08850956e+00 -6.29899561e-01 1.19044232e+00 -9.15170074e-01 -2.95404524e-01 1.05281067e+00 -1.24933875e+00 -1.11500556e-02 7.91511983e-02 6.45050585e-01 -5.28542638e-01 1.03304183e+00 -3.75681043e-01 -9.00829136e-01 -2.35193707e-02 -1.32142925e+00 1.08905995e+00 5.10051310e-01 1.43099278e-01 -8.82232606e-01 2.16352925e-01 4.90689367e-01 1.66122183e-01 3.94207984e-01 -4.93004359e-02 -1.21958837e-01 -8.97900164e-01 -1.54893756e-01 1.20162763e-01 3.93902928e-01 7.05984607e-02 2.40210712e-01 -1.01520348e+00 -1.33614659e-01 3.15720350e-01 4.66324657e-01 9.20508444e-01 8.52233112e-01 6.41969800e-01 9.13090035e-02 -7.56054044e-01 8.88729274e-01 1.66121924e+00 7.56639540e-01 8.42876613e-01 7.14126408e-01 6.68628573e-01 6.17599308e-01 7.51097620e-01 4.10034776e-01 1.96880579e-01 8.97132933e-01 3.26685339e-01 1.85409680e-01 -3.08609754e-02 5.00082493e-01 7.46360660e-01 6.79869413e-01 -5.73127866e-01 -3.27862829e-01 -9.21453595e-01 4.37643707e-01 -1.66787362e+00 -1.32396221e+00 -6.98163509e-01 2.25538874e+00 -4.79416810e-02 -1.19715050e-01 4.83597696e-01 1.77282408e-01 1.22687769e+00 2.88786769e-01 -1.29392654e-01 4.75225821e-02 -7.28483975e-01 -4.56513911e-01 8.85559320e-01 5.32322884e-01 -1.70553827e+00 8.10678780e-01 6.70890284e+00 6.31632209e-01 -9.72033978e-01 1.79015711e-01 5.07670283e-01 7.54122138e-02 6.36815190e-01 1.55472279e-01 -1.45237982e+00 5.91881752e-01 5.85886955e-01 -9.89827607e-03 2.19326466e-01 8.00853133e-01 1.25118628e-01 -6.82353616e-01 -7.89331555e-01 1.55601132e+00 7.53272295e-01 -9.94364262e-01 -5.27333096e-02 -1.06044695e-01 9.51628149e-01 4.19168323e-02 -7.40064215e-03 -9.63197276e-02 -1.25085369e-01 -5.14084637e-01 8.34557593e-01 5.83473146e-01 1.63336501e-01 -4.85923737e-01 5.81112146e-01 3.75539660e-01 -1.37036860e+00 -2.76574999e-01 -4.22410309e-01 -1.80069342e-01 4.51811671e-01 3.92240793e-01 -1.09511092e-01 9.12520409e-01 1.07077134e+00 7.71922052e-01 -9.29694414e-01 1.60465801e+00 3.32524061e-01 3.26587647e-01 -1.91097975e-01 3.32485884e-01 -3.60418484e-02 -6.93291247e-01 7.99163103e-01 1.59328294e+00 3.81709397e-01 1.40822768e-01 1.64226294e-01 3.99942659e-02 5.92848420e-01 -4.24514525e-02 -5.62460124e-01 4.74745274e-01 5.39473072e-02 1.48986554e+00 -1.17400146e+00 -5.80335259e-01 -9.41339254e-01 1.29383934e+00 -5.86275280e-01 4.02409226e-01 -9.40346599e-01 -2.37113133e-01 5.53037941e-01 -3.01195937e-03 3.90786886e-01 -6.06249928e-01 2.09224541e-02 -1.43804669e+00 2.42565069e-02 -1.09073329e+00 6.61578536e-01 -7.01488793e-01 -7.42756963e-01 4.32121575e-01 5.02580225e-01 -1.53088057e+00 1.47368923e-01 -1.01224482e+00 -4.84100699e-01 5.11087775e-01 -1.30040121e+00 -8.00017774e-01 -8.68289232e-01 8.24697554e-01 9.80879128e-01 -1.78566143e-01 2.08120078e-01 6.47807300e-01 -8.62373471e-01 -1.40179947e-01 4.22785401e-01 1.24661028e-01 7.75496244e-01 -1.08875239e+00 2.14905262e-01 1.44997633e+00 1.22025356e-01 5.38854897e-01 6.44561410e-01 -3.60437870e-01 -1.24229455e+00 -8.82361948e-01 3.65026265e-01 -9.83310997e-01 2.08952278e-01 -2.32837543e-01 -4.20138121e-01 6.30614400e-01 5.14607847e-01 3.29137361e-03 4.84728932e-01 -6.80262029e-01 2.20833659e-01 -1.78907692e-01 -7.97247946e-01 5.28316021e-01 9.41848040e-01 1.24822348e-01 -5.44459522e-01 2.50440985e-01 1.51932472e-02 -5.47003746e-01 -2.63584346e-01 4.65030730e-01 4.47437078e-01 -1.53636718e+00 1.31690609e+00 -2.48389810e-01 -8.90538692e-02 -8.08811843e-01 -3.28199238e-01 -6.33576751e-01 -1.59373403e-01 -8.14414501e-01 -2.17455432e-01 1.21717751e+00 -2.44708344e-01 -3.76061648e-01 7.44277537e-01 2.95140564e-01 8.40193313e-03 -1.42883748e-01 -8.08251858e-01 -6.80269599e-01 -2.80735373e-01 -2.64615893e-01 -1.10908382e-01 9.61017787e-01 -4.23580796e-01 2.55416721e-01 -5.31436563e-01 3.76507074e-01 7.24602222e-01 1.16590701e-01 1.09957707e+00 -1.14690757e+00 -2.80919820e-01 -5.89750528e-01 -8.29002321e-01 -1.41012311e+00 -4.11337435e-01 -2.28201747e-01 -2.70137191e-01 -1.43560874e+00 5.53602934e-01 4.41269815e-01 1.39214203e-01 -4.31672275e-01 -2.30299592e-01 4.06812519e-01 3.55175614e-01 3.56393188e-01 -1.14804316e+00 -2.05923736e-01 6.77465320e-01 -1.86758637e-01 6.11131974e-02 4.00349587e-01 -2.75137216e-01 1.34762967e+00 6.29817963e-01 -9.58133191e-02 -1.87633842e-01 -3.31849784e-01 4.99764495e-02 -1.17095076e-01 4.49552894e-01 -1.48026717e+00 5.37046671e-01 -2.27976844e-01 5.37305295e-01 -1.00049329e+00 4.49517637e-01 -9.58711624e-01 3.42071623e-01 4.52682912e-01 3.75643075e-01 4.21961844e-01 1.40525341e-01 6.53634429e-01 -5.64957500e-01 -3.85799944e-01 1.03859627e+00 -5.11648595e-01 -1.29760432e+00 -2.25906707e-02 -9.45522666e-01 -1.00190215e-01 1.47392821e+00 -8.61371934e-01 -1.69584468e-01 -2.77908832e-01 -7.18600690e-01 8.51149857e-03 4.74216223e-01 4.53950018e-01 4.00819689e-01 -9.95053411e-01 -3.34368855e-01 -3.05698991e-01 -2.79672772e-01 -2.09847361e-01 6.81751251e-01 1.27774787e+00 -9.18534100e-01 8.99963379e-01 -3.36962759e-01 -8.76729012e-01 -1.98585153e+00 8.36565793e-01 4.68758792e-01 -5.07064909e-02 -2.57332802e-01 5.83256304e-01 4.19894040e-01 1.89386234e-01 2.69414812e-01 -4.09829438e-01 -6.22649133e-01 1.18302964e-01 7.43485868e-01 1.13657975e+00 5.77829331e-02 -1.12518585e+00 -6.69390559e-01 1.18346834e+00 3.43297273e-01 -1.34351715e-01 8.81948888e-01 -7.33479500e-01 -6.87699392e-02 5.82660377e-01 7.85030067e-01 4.22569692e-01 -1.14035380e+00 1.20128565e-01 6.77103028e-02 -7.60531902e-01 -4.28924978e-01 -2.84534901e-01 -8.15843046e-01 7.96959817e-01 9.96505499e-01 3.06054622e-01 1.52418303e+00 -3.21535200e-01 2.62896478e-01 2.40315303e-01 4.25857335e-01 -1.06894016e+00 -6.79362640e-02 6.10175669e-01 3.37123603e-01 -1.12283230e+00 1.69478938e-01 -1.02751613e+00 -6.20655239e-01 1.36869514e+00 6.43286347e-01 1.71597255e-03 5.74226737e-01 2.92625248e-01 1.63891390e-01 1.78132445e-01 -1.20980777e-01 -3.58041108e-01 1.80019632e-01 9.58473086e-01 3.61041307e-01 -4.36903179e-01 -2.42431983e-01 2.33043656e-01 1.53572693e-01 -2.72049546e-01 8.12166810e-01 9.04386044e-01 -5.77066064e-01 -9.89309728e-01 -1.12062442e+00 8.71023221e-04 -7.27704763e-01 1.68999776e-01 -6.25935614e-01 9.39737797e-01 5.28770864e-01 1.22749007e+00 -1.75626203e-01 -3.13236900e-02 5.04219949e-01 -1.73064008e-01 6.93233609e-01 -3.57488811e-01 -3.97589117e-01 -7.17861578e-03 -1.02902040e-01 -6.08809948e-01 -1.24702013e+00 -1.15135336e+00 -3.24882716e-01 -2.38104351e-02 -3.91282469e-01 -1.01616919e-01 4.85321820e-01 6.32084489e-01 2.03162730e-02 2.96215177e-01 2.49806166e-01 -1.32887316e+00 -1.03279665e-01 -7.11571753e-01 -4.75490808e-01 3.22583467e-01 3.39737147e-01 -6.12267554e-01 -4.84443665e-01 6.55521512e-01]
[8.873080253601074, -0.8660454154014587]
e54fff29-058d-4c32-8572-a6ec1c28fbb3
modified-parametric-multichannel-wiener
2306.17317
null
https://arxiv.org/abs/2306.17317v1
https://arxiv.org/pdf/2306.17317v1.pdf
Modified Parametric Multichannel Wiener Filter \\for Low-latency Enhancement of Speech Mixtures with Unknown Number of Speakers
This paper introduces a novel low-latency online beamforming (BF) algorithm, named Modified Parametric Multichannel Wiener Filter (Mod-PMWF), for enhancing speech mixtures with unknown and varying number of speakers. Although conventional BFs such as linearly constrained minimum variance BF (LCMV BF) can enhance a speech mixture, they typically require such attributes of the speech mixture as the number of speakers and the acoustic transfer functions (ATFs) from the speakers to the microphones. When the mixture attributes are unavailable, estimating them by low-latency processing is challenging, hindering the application of the BFs to the problem. In this paper, we overcome this problem by modifying a conventional Parametric Multichannel Wiener Filter (PMWF). The proposed Mod-PMWF can adaptively form a directivity pattern that enhances all the speakers in the mixture without explicitly estimating these attributes. Our experiments will show the proposed BF's effectiveness in interference reduction ratios and subjective listening tests.
['Takehiro Moriya', 'Shoko Araki', 'Tomohiro Nakatani', 'Ning Guo']
2023-06-29
null
null
null
null
['low-latency-processing']
['robots']
[ 2.32475027e-01 -3.05397898e-01 3.65349919e-01 -2.00983062e-01 -1.00661588e+00 -6.63646996e-01 3.20515126e-01 -5.34216642e-01 -3.92244279e-01 6.12390399e-01 6.14985406e-01 -6.14309847e-01 -4.17627037e-01 -2.32622698e-01 -4.46015418e-01 -9.55100715e-01 -2.07004875e-01 -1.02183260e-01 2.55766094e-01 -4.05280627e-02 -3.58166993e-02 5.23260713e-01 -1.45877910e+00 2.35059813e-01 7.68395066e-01 1.04056668e+00 7.43751466e-01 1.15486026e+00 -9.99069437e-02 1.84820145e-01 -8.66653502e-01 1.47457290e-02 4.92385477e-01 -2.53919750e-01 4.51625623e-02 7.52678066e-02 6.18361942e-02 -4.83670264e-01 -3.45708847e-01 9.26450312e-01 9.96451497e-01 6.24500275e-01 6.56038105e-01 -1.14229465e+00 6.28921315e-02 6.46725774e-01 -5.22014797e-01 3.60409766e-01 1.62239969e-01 -1.68487862e-01 4.27181691e-01 -1.09508646e+00 -1.03729069e-01 1.42095232e+00 4.68099862e-01 5.20181060e-01 -8.27316999e-01 -1.09293342e+00 2.09573388e-01 1.96407840e-01 -1.53834093e+00 -1.19424951e+00 7.88130224e-01 -1.56791449e-01 6.19706511e-01 6.96612835e-01 2.81795293e-01 5.48774302e-01 -3.74701560e-01 5.60428858e-01 1.04156685e+00 -4.62173820e-01 3.40820223e-01 1.08015791e-01 -1.17565095e-01 1.28446802e-01 -5.42559959e-02 7.28358179e-02 -7.28199422e-01 -2.75331080e-01 8.97637069e-01 -4.99020845e-01 -6.05714560e-01 1.47342011e-01 -1.14429760e+00 2.27764621e-01 -1.88592017e-01 3.73295397e-01 -3.24692994e-01 -1.66646868e-01 -1.74102679e-01 2.94989675e-01 4.08160686e-01 1.10206328e-01 -3.37790072e-01 -1.19550869e-01 -1.14453888e+00 3.81349698e-02 8.37016463e-01 1.09182572e+00 3.79961252e-01 4.13494676e-01 -1.30627245e-01 9.09016907e-01 7.36471176e-01 9.02522266e-01 2.60027051e-01 -9.38636899e-01 6.78420305e-01 -3.13018650e-01 5.18999636e-01 -7.15941548e-01 -2.05477029e-01 -7.38375247e-01 -9.21791911e-01 8.69330019e-02 3.74226063e-01 -6.05315506e-01 -7.77757168e-01 1.68827176e+00 6.28645539e-01 5.44819832e-01 1.57157972e-01 9.95016694e-01 5.82443655e-01 9.30151999e-01 -3.33339155e-01 -7.11545348e-01 8.22851300e-01 -9.13723528e-01 -1.12632942e+00 -1.91778570e-01 -1.45402968e-01 -1.37599802e+00 7.28622377e-01 5.40217102e-01 -1.32097006e+00 -6.41686678e-01 -1.12142146e+00 6.06232643e-01 1.37132630e-01 1.76087916e-02 3.29472393e-01 1.34471262e+00 -1.19912481e+00 -1.04054973e-01 -6.10854387e-01 3.26658010e-01 -1.59201145e-01 6.98925674e-01 6.13776967e-02 5.41690737e-02 -9.92909729e-01 4.98905867e-01 -1.85544610e-01 5.38895607e-01 -9.40761805e-01 -5.97634912e-01 -4.49197143e-01 2.47948989e-01 2.08424345e-01 -3.97481918e-01 1.38958490e+00 -8.76883745e-01 -1.90434599e+00 -3.62980366e-01 -6.08904779e-01 7.80328119e-04 3.66851956e-01 -2.45256141e-01 -9.63516176e-01 -5.95672987e-02 -4.03870165e-01 3.15574378e-01 1.29865456e+00 -1.47368467e+00 -6.90619111e-01 -4.26142029e-02 -7.93139488e-02 4.75724429e-01 -5.03421962e-01 1.32867321e-01 -3.09968054e-01 -9.31039870e-01 4.34488147e-01 -4.68392193e-01 -3.38994652e-01 -1.63879201e-01 -4.45188522e-01 2.70053953e-01 8.73577476e-01 -8.06788385e-01 1.44252801e+00 -2.52966595e+00 -1.15286723e-01 5.52253783e-01 -1.52614832e-01 3.74965340e-01 -1.68713942e-01 2.78365701e-01 1.34099215e-01 -2.20666438e-01 -2.09763311e-02 -4.09505129e-01 5.13192602e-02 -8.23677480e-02 -2.37869531e-01 4.27394062e-01 -3.34817022e-01 9.73128229e-02 -6.43027663e-01 -2.52619356e-01 3.80366623e-01 6.55065238e-01 -5.12500226e-01 5.65684795e-01 3.29530418e-01 5.77890813e-01 -1.49710938e-01 4.31792736e-01 1.14433384e+00 5.26543379e-01 -1.84506308e-02 -4.27113146e-01 -2.59876996e-01 1.01873137e-01 -2.00264359e+00 1.11881912e+00 -7.29113400e-01 4.13695067e-01 1.08039713e+00 -5.62626660e-01 8.69702041e-01 8.17554176e-01 2.97153443e-01 -1.47957891e-01 -9.82751474e-02 3.63263458e-01 3.65547061e-01 -4.04129922e-01 3.15163761e-01 -2.13946372e-01 4.07082021e-01 3.55295986e-01 -3.49545442e-02 1.63712695e-01 1.31207347e-01 3.96758951e-02 8.15055549e-01 -4.05818909e-01 2.32991055e-01 -3.29067558e-01 8.53557169e-01 -8.77730906e-01 5.21238983e-01 8.60754251e-01 -1.45210326e-01 2.51173884e-01 -4.93427843e-01 5.44951737e-01 -6.55601978e-01 -1.45068228e+00 4.56956588e-02 1.12625957e+00 -3.20630819e-02 -1.03058919e-01 -7.12160408e-01 -8.12875759e-03 -3.61537516e-01 6.02209926e-01 2.33131722e-01 2.17190832e-01 -4.78265822e-01 -5.46338439e-01 4.53892320e-01 2.57520437e-01 6.24457657e-01 -2.53252655e-01 -6.22366779e-02 5.06172836e-01 -3.34271938e-01 -1.02947211e+00 -1.03409147e+00 9.28210542e-02 -6.30918682e-01 -4.57984030e-01 -7.00181961e-01 -7.59104490e-01 5.81161082e-01 6.83930695e-01 3.31111342e-01 -3.96793544e-01 2.47436821e-01 4.13733840e-01 -1.53698608e-01 -4.53132749e-01 -3.72156233e-01 -4.54786062e-01 4.64403212e-01 4.94248301e-01 -1.02221474e-01 -8.88043523e-01 -6.97214544e-01 6.55068576e-01 -7.59783924e-01 -6.52687624e-02 6.70419991e-01 4.58215803e-01 1.99873745e-01 7.01239884e-01 8.31802189e-01 -2.73300558e-01 8.45965087e-01 -2.73453891e-01 -4.67628121e-01 1.16893418e-01 -6.22245595e-02 -2.31913999e-01 6.49984062e-01 -8.59520674e-01 -1.55052054e+00 -4.85727238e-03 -3.30859989e-01 -2.07903013e-01 -2.08634466e-01 1.88672721e-01 -9.17413950e-01 -2.09325463e-01 2.75503963e-01 1.66208088e-01 -3.39432091e-01 -6.36703908e-01 4.42646086e-01 1.29827893e+00 6.96455419e-01 -1.80887386e-01 9.62745070e-01 1.59915254e-01 -1.35055380e-02 -1.20876467e+00 -2.74636656e-01 -9.60175514e-01 -3.05616617e-01 -4.49075013e-01 3.17343831e-01 -9.96326447e-01 -5.10166287e-01 5.25699973e-01 -9.45151687e-01 -1.35835230e-01 1.58989072e-01 1.09533954e+00 -2.82115668e-01 3.59603256e-01 -4.23682928e-01 -1.62585354e+00 -2.61111796e-01 -1.10250354e+00 6.83083117e-01 -2.73272563e-02 -2.73776464e-02 -7.71412373e-01 -2.33718365e-01 4.45329010e-01 8.22563350e-01 -3.65242273e-01 5.74519217e-01 -1.95017010e-01 -5.10216296e-01 5.82828224e-02 2.27683395e-01 4.25623894e-01 4.91166145e-01 -3.21985990e-01 -1.19238317e+00 -3.50319952e-01 5.19526660e-01 5.04045248e-01 4.42689717e-01 8.69129062e-01 8.28956723e-01 -4.62349653e-01 -2.50697881e-01 5.31133354e-01 1.04657590e+00 6.57820940e-01 3.79700989e-01 -1.52636766e-01 3.71734470e-01 4.91825283e-01 4.58380669e-01 6.58842742e-01 4.59446525e-03 5.03685176e-01 1.07686810e-01 -1.36901930e-01 -2.23312795e-01 1.97583571e-01 7.19489634e-01 1.46492922e+00 -3.76409627e-02 -6.44091904e-01 -2.82570988e-01 2.77484596e-01 -1.39213848e+00 -7.68638253e-01 -9.47687216e-03 2.37217855e+00 7.96755850e-01 -3.88666289e-03 7.36205205e-02 6.38274610e-01 1.11080623e+00 9.55040976e-02 -4.35400665e-01 -3.61341178e-01 -4.00111228e-02 1.78249866e-01 4.99863148e-01 1.07522035e+00 -9.95481968e-01 3.33794892e-01 5.93813848e+00 1.06229758e+00 -8.14610898e-01 2.84736127e-01 2.51800537e-01 -3.40414435e-01 -2.88987219e-01 -4.94537503e-01 -7.73078620e-01 3.95077109e-01 1.02236295e+00 -2.77826730e-02 9.04128075e-01 2.84017980e-01 7.53089190e-01 -1.09910153e-01 -7.84711242e-01 1.07876039e+00 -1.57061830e-01 -5.69302619e-01 -1.88755020e-01 -1.16445623e-01 3.92653078e-01 -4.93327618e-01 2.81365484e-01 5.83951958e-02 4.42583710e-02 -7.56594956e-01 6.12839162e-01 3.82371485e-01 5.82020223e-01 -9.22166944e-01 3.64161909e-01 5.07279754e-01 -1.51823819e+00 -3.15116316e-01 -2.79292732e-01 -4.70649712e-02 5.07084131e-01 9.89644289e-01 -7.92135835e-01 4.74459201e-01 3.99695456e-01 -2.94231296e-01 2.83861104e-02 1.31384671e+00 -3.53726111e-02 9.65855658e-01 -5.65634251e-01 7.61942938e-02 -1.62404515e-02 -7.42448494e-02 9.00783598e-01 1.28334475e+00 7.85138905e-01 3.80326539e-01 6.30006045e-02 2.35008284e-01 1.09524049e-01 1.62287265e-01 3.25097479e-02 1.45249158e-01 7.64914513e-01 1.20957541e+00 -4.85892832e-01 -1.93326786e-01 -3.20918202e-01 5.39733589e-01 -4.63415295e-01 7.94883549e-01 -6.82937086e-01 -2.74085701e-01 7.86386192e-01 2.36842763e-02 3.40480238e-01 -5.39330959e-01 -7.88792521e-02 -7.36299157e-01 -1.06684878e-01 -9.24826443e-01 -4.42098863e-02 -6.97478592e-01 -1.12262404e+00 4.77380753e-01 -1.20132543e-01 -1.32193196e+00 -3.98468599e-02 -3.21032763e-01 -4.66006786e-01 1.18076146e+00 -1.30360472e+00 -7.63047576e-01 8.12594518e-02 9.07403231e-01 7.73349226e-01 -1.49603218e-01 6.55061305e-01 6.98858261e-01 -2.53925651e-01 6.48514867e-01 4.41278696e-01 -4.40711081e-01 7.78672755e-01 -1.09319937e+00 1.25855908e-01 1.04776990e+00 -8.93601391e-04 6.24747515e-01 1.04153883e+00 -4.32526648e-01 -1.42800307e+00 -9.45433140e-01 6.66489720e-01 2.85869151e-01 3.69068116e-01 -6.19137347e-01 -5.59603095e-01 9.55596939e-02 3.45829576e-01 -4.36117977e-01 1.04223990e+00 -2.29646415e-01 1.17915727e-01 -3.38739872e-01 -1.09715068e+00 7.43986130e-01 1.07662416e+00 -2.53529757e-01 -2.58964986e-01 1.54822052e-01 5.37502110e-01 -3.16929638e-01 -6.37841403e-01 2.95283139e-01 6.74328744e-01 -6.74210131e-01 1.13576257e+00 8.33482593e-02 -4.17531163e-01 -6.23935878e-01 -5.71323335e-01 -1.52593708e+00 -4.29760039e-01 -1.03930354e+00 -2.26047277e-01 1.56912220e+00 4.77615833e-01 -6.48086965e-01 3.80360067e-01 4.09425557e-01 -2.65801460e-01 -3.82219285e-01 -1.09817612e+00 -8.10027301e-01 -4.94244546e-01 -7.40729988e-01 7.54495025e-01 5.11103213e-01 -1.94485977e-01 1.73115209e-01 -6.48648143e-01 9.11432981e-01 6.98919654e-01 -3.66690367e-01 6.73034370e-01 -7.23111451e-01 -8.46066058e-01 -2.23821983e-01 3.07269469e-02 -1.65896213e+00 -2.37032697e-01 -3.88565481e-01 3.99210334e-01 -1.34924006e+00 -3.92434895e-01 -4.24355894e-01 -3.41515630e-01 -2.52542228e-01 -3.00174296e-01 -1.18444808e-01 -6.09672256e-03 -1.52665287e-01 -2.54449546e-02 3.85595858e-01 1.23574400e+00 -1.33438006e-01 -3.75417322e-01 6.22694492e-01 -3.96073133e-01 6.92691982e-01 7.16950119e-01 -2.41360694e-01 -8.13267350e-01 -4.37124938e-01 -4.21993196e-01 4.26411867e-01 -1.44088134e-01 -1.10312414e+00 6.53790891e-01 -1.07604519e-01 4.48211908e-01 -7.46271849e-01 8.39386642e-01 -1.03825068e+00 4.25875634e-01 1.93610579e-01 -1.63260445e-01 -1.72767520e-01 2.39768624e-01 6.57052398e-01 -1.14040881e-01 -2.02427596e-01 5.11040151e-01 1.20281495e-01 -4.54798192e-01 8.54514912e-02 -9.69740033e-01 -4.40643549e-01 5.29910266e-01 -1.80181980e-01 -8.15688893e-02 -9.52183366e-01 -8.95437002e-01 2.82884873e-02 -3.06104690e-01 9.52695608e-02 7.68312573e-01 -1.09758854e+00 -5.83613217e-01 4.87596720e-01 -6.93022132e-01 -2.17802435e-01 4.72186863e-01 8.19530487e-01 6.53105741e-03 4.00382578e-01 2.48012051e-01 -3.09815288e-01 -1.60006511e+00 3.92458320e-01 2.93414414e-01 1.14700235e-01 -1.26373013e-02 1.08139503e+00 4.05154884e-01 -2.34631851e-01 4.48359638e-01 -7.47589469e-02 -2.21838534e-01 -3.41634750e-01 6.65723264e-01 7.34960079e-01 4.82694730e-02 -6.43485546e-01 -1.80933043e-01 3.27902168e-01 2.74510235e-01 -8.15549850e-01 8.83676171e-01 -8.25157523e-01 -5.61346626e-03 3.48432422e-01 8.01569045e-01 6.21269524e-01 -1.25723326e+00 -1.55645505e-01 -2.65408486e-01 -7.92290628e-01 5.09087741e-01 -7.00426936e-01 -8.76409590e-01 8.59052062e-01 9.89261746e-01 2.11556396e-03 1.52979720e+00 -4.14882094e-01 8.44751298e-01 3.24112624e-01 5.23185968e-01 -9.34451699e-01 -1.90281987e-01 2.48206526e-01 6.45648062e-01 -6.45682454e-01 -3.02688479e-01 -7.86204696e-01 -1.90217435e-01 9.90199327e-01 4.13980931e-01 4.27866995e-01 1.01950336e+00 7.88614511e-01 4.81246233e-01 5.71887374e-01 -5.58679461e-01 -1.89957678e-01 3.41064304e-01 6.72098160e-01 3.69166911e-01 2.02807963e-01 -2.28798226e-01 8.31975996e-01 -3.78006607e-01 -4.45237428e-01 3.91928971e-01 7.37679660e-01 -8.98057044e-01 -1.20451725e+00 -9.80304360e-01 4.93683755e-01 -4.57377553e-01 -2.05792084e-01 8.12640041e-02 9.38052908e-02 2.15788439e-01 1.82479692e+00 -5.46446331e-02 -4.39457506e-01 3.89567494e-01 -8.63628685e-02 4.74859685e-01 -4.17721570e-01 -4.10265207e-01 1.05015683e+00 2.39233553e-01 -8.32404271e-02 -6.06082857e-01 -5.61428845e-01 -7.70030499e-01 -2.62846947e-01 -7.26330280e-01 4.41274017e-01 8.48529279e-01 9.84362483e-01 1.02144824e-02 7.15087235e-01 1.01751268e+00 -9.64359701e-01 -6.51384950e-01 -1.23830271e+00 -8.87754142e-01 -2.80923426e-01 4.97387230e-01 -5.27880013e-01 -5.59037328e-01 4.10834048e-03]
[15.070100784301758, 5.829063415527344]
584f5cf1-a1f0-4a84-875c-ee620e363755
a-novel-context-aware-multimodal-framework
2103.02636
null
https://arxiv.org/abs/2103.02636v1
https://arxiv.org/pdf/2103.02636v1.pdf
A Novel Context-Aware Multimodal Framework for Persian Sentiment Analysis
Most recent works on sentiment analysis have exploited the text modality. However, millions of hours of video recordings posted on social media platforms everyday hold vital unstructured information that can be exploited to more effectively gauge public perception. Multimodal sentiment analysis offers an innovative solution to computationally understand and harvest sentiments from videos by contextually exploiting audio, visual and textual cues. In this paper, we, firstly, present a first of its kind Persian multimodal dataset comprising more than 800 utterances, as a benchmark resource for researchers to evaluate multimodal sentiment analysis approaches in Persian language. Secondly, we present a novel context-aware multimodal sentiment analysis framework, that simultaneously exploits acoustic, visual and textual cues to more accurately determine the expressed sentiment. We employ both decision-level (late) and feature-level (early) fusion methods to integrate affective cross-modal information. Experimental results demonstrate that the contextual integration of multimodal features such as textual, acoustic and visual features deliver better performance (91.39%) compared to unimodal features (89.24%).
['Amir Hussain', 'Erik Cambria', 'Mandar Gogate', 'Kia Dashtipour']
2021-03-03
null
null
null
null
['persian-sentiment-anlysis']
['natural-language-processing']
[ 3.54004949e-01 -3.87471437e-01 1.36949956e-01 -5.83256304e-01 -1.37476516e+00 -8.05056930e-01 7.72457242e-01 4.78145003e-01 -6.71328247e-01 5.25588214e-01 4.23887461e-01 2.60263920e-01 2.70997465e-01 -2.59891242e-01 -1.92835897e-01 -9.93751287e-01 1.23692274e-01 -2.89861560e-01 5.23567200e-03 -5.42093635e-01 3.76069039e-01 1.16160095e-01 -2.05382562e+00 8.45480204e-01 3.70398492e-01 1.43550968e+00 2.46571023e-02 8.54228854e-01 -8.27317163e-02 9.25843954e-01 -6.19505465e-01 -8.61833036e-01 -2.99879849e-01 -2.29520932e-01 -5.49252152e-01 1.77862793e-02 1.62252143e-01 -5.65107353e-02 1.83842644e-01 1.09000647e+00 7.19790697e-01 1.86174929e-01 5.40274739e-01 -1.21539640e+00 -2.51711786e-01 5.03532290e-01 -6.18626535e-01 -2.19219811e-02 1.04006958e+00 -8.77931491e-02 1.16760492e+00 -1.02656329e+00 5.66837311e-01 1.10570049e+00 4.35271174e-01 8.82936865e-02 -7.89042771e-01 -5.40880680e-01 1.01654008e-01 4.02848363e-01 -1.24767923e+00 -6.56882048e-01 1.17072570e+00 -3.65426540e-01 7.80169964e-01 4.63470638e-01 5.76075733e-01 1.38389373e+00 -4.33604326e-03 9.97218788e-01 1.36348855e+00 -5.68707228e-01 -8.48723110e-03 6.09845221e-01 -1.02093574e-02 4.25143003e-01 -5.07447779e-01 -4.16165859e-01 -1.20361602e+00 -2.45637029e-01 -2.88098067e-01 -1.06162243e-01 -1.06419750e-01 1.21632658e-01 -1.21605825e+00 8.02086294e-01 -2.47464199e-02 3.75184000e-01 -6.55752599e-01 -9.54574719e-02 6.66537821e-01 2.43622810e-01 4.69922781e-01 1.42066866e-01 -3.56623560e-01 -6.56098366e-01 -9.42176461e-01 -5.84008507e-02 6.46852374e-01 5.49448907e-01 8.31547439e-01 4.04886007e-02 1.86261125e-02 1.08337533e+00 5.98412633e-01 1.03629661e+00 5.91592371e-01 -8.05298984e-01 5.01266897e-01 5.34192681e-01 -4.61121416e-03 -1.55808485e+00 -6.05108440e-01 4.44482416e-01 -4.30888265e-01 -3.37856710e-01 5.57442894e-03 -3.01512450e-01 -3.11432034e-01 1.39189255e+00 4.29728478e-01 -2.96044052e-01 2.72884548e-01 9.89893794e-01 1.24852645e+00 7.11525798e-01 1.64719805e-01 -2.37624168e-01 1.87660909e+00 -4.95345175e-01 -7.98095047e-01 6.62660226e-02 4.30929661e-01 -1.31030810e+00 9.60080922e-01 7.24264860e-01 -9.23564553e-01 -3.84390652e-01 -1.00857091e+00 1.16957150e-01 -6.61095977e-01 3.04465979e-01 4.02616858e-01 1.04260409e+00 -7.77122915e-01 -1.05341941e-01 -5.38435340e-01 -3.45077872e-01 1.50046468e-01 2.67251194e-01 -6.66877031e-01 4.28232327e-02 -1.23052680e+00 5.12323499e-01 1.33581519e-01 2.32055202e-01 -4.68126059e-01 -9.38608497e-02 -1.14867592e+00 -3.05384129e-01 3.42027962e-01 -1.01731747e-01 1.13757932e+00 -1.29280448e+00 -1.71307111e+00 8.58030438e-01 -4.81961966e-01 -4.13376130e-02 -2.88733970e-02 -2.10455999e-01 -6.80859625e-01 8.31606507e-01 -9.75633711e-02 7.44095683e-01 9.79218364e-01 -1.23039508e+00 -8.31066608e-01 -5.32767594e-01 7.09675178e-02 3.64300996e-01 -7.58623481e-01 5.01411021e-01 -5.16397536e-01 -4.98362184e-01 -6.38069287e-02 -9.51748490e-01 1.95019782e-01 -6.15726352e-01 -4.69528168e-01 -1.55789688e-01 1.01868105e+00 -6.51064873e-01 1.10310471e+00 -2.24353480e+00 1.83189392e-01 2.71719754e-01 2.62133572e-02 1.50209554e-02 -6.57439679e-02 8.55957270e-01 2.43297562e-01 -1.56237647e-01 2.71123294e-02 -7.03390598e-01 2.96679497e-01 -5.83256967e-02 -2.95130461e-01 3.81180078e-01 2.76222140e-01 7.09049225e-01 -6.58146381e-01 -7.38510966e-01 4.61399823e-01 8.99340868e-01 -4.35898721e-01 -6.75123259e-02 1.20477341e-01 5.18947959e-01 -5.87781549e-01 1.20652664e+00 3.98552507e-01 2.50946671e-01 1.34945303e-01 -5.48937678e-01 -2.67737150e-01 -2.86151201e-01 -1.04382372e+00 1.69695556e+00 -5.18382490e-01 9.28693831e-01 2.42465109e-01 -9.11622107e-01 9.19882655e-01 4.64226604e-01 6.12954497e-01 -7.85157859e-01 6.24391854e-01 5.26604950e-02 -6.44293010e-01 -8.68505061e-01 1.12463772e+00 -2.27205694e-01 -8.33992183e-01 1.38792947e-01 2.22850263e-01 -1.02260798e-01 2.67678082e-01 1.30496606e-01 6.55879200e-01 -1.85777824e-02 6.79817349e-02 3.31104934e-01 9.70811486e-01 -1.29802689e-01 1.10435151e-01 4.33680475e-01 -3.35192770e-01 5.80121934e-01 6.19118750e-01 -5.95623143e-02 -4.22829658e-01 -5.01993656e-01 6.88847825e-02 1.55101180e+00 1.07284032e-01 -6.00790441e-01 -6.39032185e-01 -4.05066878e-01 -4.05627459e-01 3.83334041e-01 -4.13302511e-01 1.76932052e-01 -1.25058681e-01 -8.70451927e-01 5.82393050e-01 3.16149205e-01 3.09156418e-01 -1.08148789e+00 -7.05536485e-01 -7.85067454e-02 -6.35668039e-01 -1.49059618e+00 -3.67775597e-02 -4.31799032e-02 -2.03308448e-01 -8.85131836e-01 -6.66298985e-01 -5.38086236e-01 2.33474478e-01 1.50526151e-01 7.73040831e-01 -3.55073720e-01 -2.45133623e-01 1.08859563e+00 -9.10182118e-01 -4.85824287e-01 -6.21581227e-02 -1.14428587e-01 -2.01370995e-02 7.06191778e-01 7.30017900e-01 -2.31979296e-01 -3.80012244e-01 2.47614175e-01 -9.30020154e-01 -3.37306112e-01 5.95772862e-01 6.17164373e-01 5.56228518e-01 -9.28185508e-02 6.67082727e-01 -5.11830509e-01 7.21499145e-01 -4.56694961e-01 -2.98968852e-01 -2.81766225e-02 1.79728698e-02 -4.73350197e-01 3.70649844e-01 -2.55270183e-01 -1.24711835e+00 2.07515255e-01 -2.55447268e-01 -1.14539430e-01 -4.58608091e-01 9.37273860e-01 -1.19519517e-01 1.17371166e-02 2.32161075e-01 1.16832398e-01 -2.58343190e-01 -1.78020760e-01 3.92538399e-01 1.15006244e+00 5.62849700e-01 -5.49449027e-01 2.85833448e-01 7.30920553e-01 -1.42218009e-01 -1.39576364e+00 -6.83011532e-01 -9.71394122e-01 -7.06138372e-01 -8.10649931e-01 1.14494550e+00 -1.15643191e+00 -1.24207973e+00 5.49007058e-01 -8.97317111e-01 3.66654813e-01 1.77578837e-01 6.25469983e-01 -5.50016522e-01 5.03795505e-01 -4.64149892e-01 -1.18620896e+00 -2.07040176e-01 -1.32564378e+00 1.72286427e+00 1.95319086e-01 -3.14172566e-01 -7.93033302e-01 2.34339442e-02 9.20042932e-01 3.77234705e-02 3.20681006e-01 2.68236876e-01 -6.96488321e-01 -9.32032429e-03 -5.44107199e-01 -2.88943976e-01 3.24863553e-01 -2.17170164e-01 1.18219770e-01 -1.56961572e+00 1.31011633e-02 -1.06768742e-01 -7.20959663e-01 7.85994649e-01 2.22265288e-01 7.85336554e-01 -8.53677839e-02 1.87968593e-02 -4.25506663e-03 1.23643613e+00 2.72359401e-01 6.94782197e-01 3.15815389e-01 6.01326525e-01 9.75584328e-01 9.12982047e-01 8.93307865e-01 7.98606694e-01 5.30927956e-01 4.60450143e-01 1.07649021e-01 4.33599025e-01 1.94900662e-01 7.23791420e-01 1.21141410e+00 -2.23161787e-01 -3.87368679e-01 -8.12601566e-01 4.14223045e-01 -1.74544418e+00 -1.11254895e+00 -5.68835065e-02 1.88279331e+00 5.76860189e-01 -1.11878954e-01 2.26780698e-01 4.52215552e-01 5.92636228e-01 2.28551567e-01 9.03067142e-02 -5.81440866e-01 -4.90669131e-01 -1.28879234e-01 6.96836710e-02 2.50671178e-01 -1.48555219e+00 5.97467959e-01 5.53621197e+00 8.59860122e-01 -1.23426616e+00 5.14575206e-02 3.39570552e-01 -2.52961367e-01 -2.08502769e-01 -4.40275490e-01 -6.71370864e-01 3.91269714e-01 1.23794663e+00 3.83022726e-01 5.96578754e-02 4.73193526e-01 3.40862185e-01 -5.48344135e-01 -7.26977229e-01 1.27671552e+00 6.60874724e-01 -8.32250237e-01 -2.99821109e-01 -9.33135971e-02 5.80392897e-01 -7.27087259e-02 4.13177669e-01 2.50225067e-01 -3.87428075e-01 -7.22549975e-01 9.08157945e-01 7.51507163e-01 4.98319328e-01 -1.14791679e+00 1.10530043e+00 3.16029899e-02 -1.30195487e+00 -5.60635701e-02 8.25399533e-02 1.15119345e-01 3.22881699e-01 3.46747130e-01 -6.25088036e-01 8.30198109e-01 9.66042757e-01 9.30594265e-01 -6.19089305e-01 4.88889277e-01 9.70148817e-02 3.62460434e-01 -1.53503194e-01 -3.80838871e-01 3.80679965e-01 -1.09754458e-01 5.77971101e-01 1.62685931e+00 2.36917555e-01 -6.90764338e-02 1.52413473e-01 9.05937329e-02 1.72864899e-01 5.60033441e-01 -6.03906214e-01 -4.27340239e-01 6.92678690e-02 1.62675881e+00 -7.69874334e-01 -9.22468081e-02 -8.32769930e-01 7.59706199e-01 -1.78661764e-01 2.73929387e-01 -9.02904153e-01 -4.86229271e-01 4.07107353e-01 -6.43526971e-01 3.45119029e-01 -2.34706536e-01 1.28565207e-01 -1.16323280e+00 1.96054019e-02 -9.05440331e-01 3.07706267e-01 -1.08516586e+00 -1.13039541e+00 7.67666876e-01 -1.62712082e-01 -1.44798362e+00 -3.84432226e-01 -7.64260232e-01 -2.14375034e-01 5.19373000e-01 -1.58171964e+00 -1.47397852e+00 -2.44913012e-01 7.80695975e-01 4.81888175e-01 -3.69740069e-01 9.80940878e-01 5.16289651e-01 -5.76819062e-01 4.81175214e-01 7.18854135e-03 2.15897150e-02 9.68532026e-01 -9.37859893e-01 -8.98927391e-01 3.73470128e-01 1.09153248e-01 3.16547334e-01 7.54101872e-01 -1.79124147e-01 -1.91947830e+00 -6.93988144e-01 9.27568376e-01 -5.66671014e-01 9.12541509e-01 -1.28207251e-01 -3.41988653e-01 2.29593396e-01 4.60910231e-01 -4.82677519e-01 1.32107317e+00 7.03985766e-02 -2.42445976e-01 -1.25748470e-01 -1.03212523e+00 5.21557927e-01 1.37326464e-01 -9.35894728e-01 -4.75299031e-01 -1.07688218e-01 4.65528190e-01 -1.12724356e-01 -1.02319169e+00 4.05338466e-01 9.11175072e-01 -1.24811053e+00 8.68011892e-01 -1.99034680e-02 5.62316954e-01 -2.98654675e-01 -9.30956781e-01 -1.01730156e+00 6.60972655e-01 -5.39100289e-01 3.17464411e-01 1.60789061e+00 4.75430608e-01 -4.12508279e-01 4.53030288e-01 3.53462517e-01 -4.58866172e-02 -4.59178180e-01 -7.89686739e-01 -5.96032441e-02 -5.99046648e-01 -1.05691159e+00 2.44464561e-01 7.84226418e-01 5.13420582e-01 4.23799306e-01 -5.65838754e-01 1.07309617e-01 1.92996368e-01 1.80216119e-01 6.84863865e-01 -1.01995337e+00 2.28642076e-01 -4.35524970e-01 -7.91362822e-01 -5.09920657e-01 2.29404613e-01 -4.98958945e-01 -8.62497743e-03 -9.53019857e-01 2.07607985e-01 1.76069617e-01 -4.27596629e-01 2.02424914e-01 1.70571327e-01 9.98461008e-01 3.20791274e-01 -8.28874186e-02 -1.11497331e+00 5.69680929e-01 9.03178215e-01 -2.73642242e-01 -4.05028053e-02 -3.11754048e-01 -7.16815472e-01 9.78181601e-01 5.46754837e-01 -1.90429702e-01 -1.29225463e-01 4.69044223e-03 7.21594274e-01 1.60737962e-01 3.77202123e-01 -7.48408616e-01 3.16469193e-01 6.89724926e-03 4.20482993e-01 -1.05388784e+00 1.08753538e+00 -8.81007791e-01 -3.31766427e-01 -1.29083052e-01 -2.69984156e-01 4.70750593e-02 4.01403666e-01 6.53087974e-01 -9.33676481e-01 -1.64073601e-01 2.61311799e-01 2.30711415e-01 -9.84159589e-01 -2.88235933e-01 -8.77750397e-01 -2.40585193e-01 9.87679899e-01 -7.73593783e-02 -2.53049821e-01 -6.68714881e-01 -9.17142093e-01 1.92919075e-01 -2.12857649e-02 4.81479377e-01 7.62638569e-01 -1.30096185e+00 -5.04236341e-01 1.01078220e-01 4.64109123e-01 -7.45306492e-01 6.97513759e-01 1.23884380e+00 -1.82772681e-01 5.25831699e-01 -4.97258008e-02 -8.83402228e-01 -1.69992244e+00 1.82043597e-01 -2.71845102e-01 2.28552192e-01 1.75073594e-01 8.25410426e-01 -1.81915045e-01 -3.23581934e-01 1.71392649e-01 5.53260408e-02 -7.90019631e-01 1.12504482e+00 7.71560311e-01 2.88935661e-01 1.89863712e-01 -1.42406619e+00 -5.26332855e-01 8.05530727e-01 3.12520802e-01 -5.51852226e-01 1.23249578e+00 -6.68700457e-01 -1.69943020e-01 8.99414420e-01 1.33953941e+00 3.00585836e-01 -5.78076780e-01 -1.32149354e-01 -9.50508565e-02 -3.17632616e-01 2.33242661e-02 -7.45149612e-01 -7.30650365e-01 1.09121025e+00 5.04142225e-01 6.70216501e-01 1.48626041e+00 1.29015883e-02 7.02331245e-01 5.14140546e-01 2.34823421e-01 -1.41949439e+00 2.56807268e-01 3.60237122e-01 7.62225926e-01 -1.55670345e+00 -1.38512835e-01 -1.65438741e-01 -1.26495683e+00 1.20624638e+00 4.15999480e-02 3.23527724e-01 6.68891430e-01 1.44379646e-01 4.04867500e-01 -2.16581389e-01 -6.50902510e-01 -4.12512451e-01 6.19260132e-01 4.63691026e-01 4.57380801e-01 2.42212676e-02 -1.26754539e-02 1.01556778e+00 -2.89444864e-01 -4.24749553e-01 5.12204349e-01 1.04132605e+00 -5.67338049e-01 -8.59170198e-01 -7.42885590e-01 3.08551099e-02 -9.20138180e-01 -1.23139217e-01 -4.36329335e-01 4.59665477e-01 -1.85322780e-02 1.59804571e+00 -1.84893712e-01 -5.53989053e-01 2.87256449e-01 1.65784702e-01 3.26740295e-01 -1.72528863e-01 -5.97537875e-01 5.59054852e-01 3.57623875e-01 -6.39182925e-01 -1.31333411e+00 -8.75707924e-01 -9.72810626e-01 -1.04237005e-01 -2.03452885e-01 6.74455836e-02 1.17647088e+00 1.01839399e+00 1.19260408e-01 3.96435201e-01 8.83770645e-01 -1.32512069e+00 1.70690358e-01 -7.91695952e-01 -4.69243824e-01 2.91397691e-01 5.15670478e-01 -5.90991676e-01 -4.11284655e-01 4.95983005e-01]
[13.111824989318848, 5.219996452331543]
526a8be4-eab3-4281-b2c1-3e489d6ee2a2
recognizing-and-extracting-cybersecurtity
2208.01693
null
https://arxiv.org/abs/2208.01693v1
https://arxiv.org/pdf/2208.01693v1.pdf
Recognizing and Extracting Cybersecurtity-relevant Entities from Text
Cyber Threat Intelligence (CTI) is information describing threat vectors, vulnerabilities, and attacks and is often used as training data for AI-based cyber defense systems such as Cybersecurity Knowledge Graphs (CKG). There is a strong need to develop community-accessible datasets to train existing AI-based cybersecurity pipelines to efficiently and accurately extract meaningful insights from CTI. We have created an initial unstructured CTI corpus from a variety of open sources that we are using to train and test cybersecurity entity models using the spaCy framework and exploring self-learning methods to automatically recognize cybersecurity entities. We also describe methods to apply cybersecurity domain entity linking with existing world knowledge from Wikidata. Our future work will survey and test spaCy NLP tools and create methods for continuous integration of new information extracted from text.
['Anupam Joshi', 'Tim Finin', 'Priyanka Ranade', 'Michael Maiden', 'Casey Hanks']
2022-08-02
null
null
null
null
['self-learning']
['natural-language-processing']
[ 1.01623125e-01 3.66221815e-01 -6.09709024e-01 1.89103618e-01 -4.02287632e-01 -1.35482943e+00 1.03792000e+00 7.09964693e-01 -1.48395553e-01 7.44284570e-01 4.32928443e-01 -9.67755616e-01 -6.33408308e-01 -1.23575902e+00 -5.13304532e-01 4.15089965e-01 -5.14983177e-01 7.86451578e-01 3.95688146e-01 -3.94116014e-01 4.44833934e-01 7.21528232e-01 -5.73536992e-01 1.97976902e-01 8.13483000e-01 4.19429570e-01 -8.04801822e-01 5.17134547e-01 3.20043764e-03 9.04969871e-01 -1.04648173e+00 -9.00268734e-01 4.59211439e-01 3.69248271e-01 -1.36596167e+00 -9.80702639e-01 4.78662670e-01 4.24213633e-02 -4.19046044e-01 1.25024533e+00 -1.02355592e-01 -1.92240551e-01 4.88645136e-01 -1.82319260e+00 -5.11147201e-01 8.78865719e-01 -3.36868428e-02 5.38383245e-01 5.50736248e-01 1.70633763e-01 8.81809056e-01 -2.30795547e-01 1.58293748e+00 1.12441325e+00 7.89081573e-01 3.81747156e-01 -5.35670698e-01 -1.04801917e+00 1.96732339e-02 4.74220008e-01 -8.92680049e-01 1.69051260e-01 6.57470584e-01 -4.57613051e-01 1.73306525e+00 2.59708792e-01 3.04367155e-01 1.50590599e+00 4.77974951e-01 3.14543247e-01 7.77801871e-01 -5.10088503e-01 -5.11485375e-02 2.95687824e-01 7.77681589e-01 8.08025956e-01 1.05803537e+00 3.91581357e-01 -2.67710268e-01 -7.09779680e-01 4.68888313e-01 -1.43985614e-01 5.90848885e-02 -8.28868300e-02 -1.29553556e+00 7.84740150e-01 4.91034806e-01 7.74795473e-01 -2.29222074e-01 -8.20505768e-02 6.41819596e-01 4.23211426e-01 4.75209385e-01 1.44000494e+00 -8.64588797e-01 -2.21620262e-01 -2.24155411e-01 1.67232051e-01 1.51769483e+00 8.31343114e-01 4.47261065e-01 6.31778985e-02 6.85904384e-01 4.69591916e-02 9.75063071e-02 3.85894328e-01 9.06353910e-03 -5.06154478e-01 9.24314559e-01 8.42799485e-01 1.53896689e-01 -1.36720824e+00 -4.52273399e-01 -1.46854326e-01 1.81857347e-02 1.52092338e-01 2.47781456e-01 -4.80306894e-01 -9.78479385e-01 1.29795444e+00 3.69781464e-01 6.68530285e-01 4.43399787e-01 4.00429219e-01 8.13623011e-01 5.71099222e-01 6.87013209e-01 4.06196803e-01 1.28220761e+00 -3.18406343e-01 -3.47340941e-01 -2.34889567e-01 1.00843489e+00 -3.83754760e-01 1.51494727e-01 5.35696030e-01 -4.63249952e-01 2.71967322e-01 -1.27964878e+00 4.97004360e-01 -1.63945818e+00 -7.44349241e-01 1.03738809e+00 9.03329611e-01 -4.92372304e-01 5.88021815e-01 -5.59348285e-01 -5.25171638e-01 4.58560169e-01 9.44075584e-02 -6.87598050e-01 -1.54374346e-01 -1.91272342e+00 1.66394937e+00 1.33488321e+00 -7.59645343e-01 -1.15209854e+00 -1.13982499e+00 -1.02739155e+00 -3.72264646e-02 9.78214383e-01 -2.41747543e-01 6.29645288e-01 -2.61929333e-01 -5.37716150e-01 4.70265359e-01 7.99470961e-01 -6.13315940e-01 -4.05863434e-01 -8.18493143e-02 -1.31262326e+00 3.15304726e-01 8.14915895e-02 1.62641898e-01 4.48955834e-01 -1.13967943e+00 -5.48677683e-01 -4.87758964e-01 4.90046144e-01 -2.55843371e-01 -6.73840880e-01 6.83456242e-01 1.53591782e-01 -5.39410233e-01 -7.38784611e-01 -9.04400408e-01 -5.87951429e-02 -8.06873798e-01 -6.21189892e-01 -7.90977255e-02 1.21084869e+00 -1.18827653e+00 1.27133441e+00 -1.45699620e+00 -1.20133907e-01 7.27488816e-01 5.22761166e-01 7.66724527e-01 -4.29367334e-01 1.00615561e+00 -4.76127803e-01 7.65735388e-01 1.41429722e-01 8.51968706e-01 1.61894873e-01 7.82194436e-02 -7.89151013e-01 -2.00881958e-01 3.72284144e-01 9.49768305e-01 -1.32545054e+00 -3.71908724e-01 8.73542503e-02 4.84210216e-02 -3.76729697e-01 -2.21330538e-01 -5.85791945e-01 3.52976248e-02 -7.39426553e-01 9.45878386e-01 4.84959394e-01 -9.61004645e-02 4.78541940e-01 -1.67729467e-01 -1.01587504e-01 5.21092296e-01 -6.54592812e-01 1.32329297e+00 -3.04860204e-01 5.08088112e-01 -2.57589445e-02 -6.36113346e-01 6.79723680e-01 4.91135091e-01 4.85638887e-01 -3.57387275e-01 5.78180049e-03 -1.42427683e-01 -1.91922501e-01 -1.75874010e-01 3.47806752e-01 1.40058428e-01 -5.45517743e-01 6.38326526e-01 7.13725686e-01 -5.75910546e-02 1.72422022e-01 7.31585026e-01 1.66344249e+00 -2.03993484e-01 3.54202241e-01 -7.17498735e-02 2.95471132e-01 9.97600794e-01 6.39819562e-01 4.35550094e-01 -1.11501096e-02 -7.23081291e-01 5.63942790e-01 -9.38092709e-01 -9.06453431e-01 -1.03027534e+00 -6.54352978e-02 6.18508220e-01 -4.98784930e-02 -9.57053661e-01 -6.17072523e-01 -1.56531727e+00 -1.58277899e-02 1.21107674e+00 -3.34544212e-01 -1.59424230e-01 -4.45555657e-01 -5.44165730e-01 1.11371648e+00 4.99746531e-01 2.52425969e-01 -1.18984556e+00 -2.56732702e-01 2.45554879e-01 -1.50510579e-01 -1.41150999e+00 2.70079166e-01 -8.83632712e-03 -1.87828615e-01 -1.70343518e+00 5.34151137e-01 -3.42135161e-01 4.00807619e-01 6.37102723e-02 1.04621553e+00 4.68428805e-02 -7.44489372e-01 7.43545473e-01 -4.53797996e-01 -1.20798492e+00 -7.09467709e-01 2.80467689e-01 2.67784417e-01 -9.36897218e-01 8.09748054e-01 -3.11771184e-01 1.14247285e-01 2.07192495e-01 -1.36736143e+00 -4.54165280e-01 1.67366356e-01 1.43426880e-01 -2.37569243e-01 6.11048818e-01 6.23132706e-01 -1.05508578e+00 8.10913026e-01 -1.07433462e+00 -8.43275726e-01 9.22301531e-01 -5.91876984e-01 -1.43607393e-01 5.98286033e-01 -2.41431251e-01 -1.17973137e+00 -6.08575225e-01 1.94481835e-01 -1.55190021e-01 -3.09581369e-01 9.89962935e-01 -2.39478171e-01 -5.15445769e-01 9.39330459e-01 -5.84502995e-01 -4.84240979e-01 -4.49962243e-02 7.59213090e-01 4.25165772e-01 3.83250624e-01 -9.15002465e-01 1.57955718e+00 -4.42512706e-03 2.45951070e-03 -7.07676768e-01 -7.34755635e-01 -3.05940449e-01 -7.28110790e-01 -1.41126528e-01 8.72875452e-01 -4.28103089e-01 -7.87287593e-01 1.65743213e-02 -1.13736939e+00 -2.71765310e-02 -1.81171903e-03 3.60057116e-01 6.13794811e-02 6.05158687e-01 -5.79836905e-01 -3.63896847e-01 -7.86361098e-01 -5.04094779e-01 4.23402876e-01 1.52585238e-01 -3.46640438e-01 -1.67409837e+00 7.40203857e-01 5.04175127e-01 2.75927097e-01 7.72868931e-01 1.35595083e+00 -1.52739012e+00 -5.76855779e-01 -6.08491302e-01 -3.54305267e-01 -5.20278849e-02 1.31932795e-01 -6.26832154e-03 -4.95778948e-01 -1.52298510e-01 -6.30730569e-01 -5.16576529e-01 1.66250676e-01 -6.52990103e-01 4.12947983e-01 -7.03459084e-01 -7.42925286e-01 2.79936373e-01 1.56218398e+00 4.82263982e-01 6.69131875e-01 8.76186967e-01 7.17930138e-01 8.07065368e-01 5.59289455e-01 7.72981122e-02 3.65360856e-01 8.93655047e-02 3.13514948e-01 3.44071358e-01 3.47804040e-01 -6.86710000e-01 9.01722386e-02 3.60951066e-01 -1.29831791e-01 -1.38953045e-01 -1.75295329e+00 6.46752000e-01 -1.57332766e+00 -1.48528337e+00 2.32245058e-01 1.72311735e+00 7.27566361e-01 3.89060318e-01 -3.38713676e-01 -2.34952271e-01 2.62030989e-01 -1.36894986e-01 -3.41263741e-01 -5.18020391e-01 2.03791276e-01 4.84802425e-01 9.00094032e-01 4.76571411e-01 -1.32996583e+00 1.67396379e+00 6.57280016e+00 8.13923955e-01 -7.84491837e-01 -9.69568193e-02 -1.57148600e-01 2.94191003e-01 -4.35365021e-01 5.73468685e-01 -9.90916133e-01 -2.92231012e-02 1.66691101e+00 -7.14861989e-01 5.29048026e-01 9.06165957e-01 -5.96627295e-01 4.37992752e-01 -5.52147150e-01 1.62536860e-01 1.95122570e-01 -1.86859953e+00 1.51733607e-01 2.78045863e-01 4.99113828e-01 1.42511040e-01 -2.94712543e-01 5.88035941e-01 1.43675470e+00 -7.76275098e-01 -5.00784861e-03 4.42917764e-01 6.06498301e-01 -7.69133568e-01 5.22935808e-01 6.99226111e-02 -1.01940262e+00 -5.18344164e-01 -2.64607608e-01 5.07852316e-01 1.03226881e-02 2.25119472e-01 -1.57460403e+00 1.38988972e+00 4.87876236e-01 9.22853708e-01 -8.74389112e-01 7.27732241e-01 -3.81110072e-01 6.51757956e-01 -2.52717227e-01 1.14289984e-01 4.26176101e-01 -4.84312735e-02 9.22040641e-01 1.25692284e+00 7.10526034e-02 4.40562099e-01 3.39083076e-01 6.50312006e-01 -8.50995854e-02 -4.04611319e-01 -1.38720012e+00 -9.37816262e-01 7.43723929e-01 1.51901901e+00 -5.79439163e-01 -5.01467109e-01 -4.47042823e-01 2.86092877e-01 5.53103276e-02 2.48319387e-01 -6.67509675e-01 -8.36672962e-01 9.44549024e-01 -1.04802765e-01 -1.04254492e-01 -7.18280613e-01 -1.21889092e-01 -1.50879884e+00 -5.98430991e-01 -1.24938369e+00 1.13985765e+00 -8.94245446e-01 -1.48844075e+00 8.96386862e-01 6.25322580e-01 -1.07552552e+00 -4.06179875e-01 -1.00477266e+00 -8.84450734e-01 6.11601651e-01 -1.31369805e+00 -1.78157592e+00 -1.05128437e-01 7.47550130e-01 -1.39284506e-01 -6.24643266e-01 1.08908689e+00 -9.47673619e-02 -6.20923579e-01 2.07450330e-01 -4.56996143e-01 6.39000237e-01 7.32455134e-01 -9.91621971e-01 1.19267273e+00 1.00921822e+00 3.53445321e-01 1.16212308e+00 3.63447458e-01 -1.75854886e+00 -1.49320853e+00 -1.12819290e+00 6.44021332e-01 -1.57358944e+00 1.81669784e+00 -3.22865367e-01 -9.41971719e-01 1.10333312e+00 4.85017926e-01 -7.75426388e-01 1.11423159e+00 3.54687035e-01 -1.07805133e+00 4.03061837e-01 -1.34977281e+00 5.84939957e-01 1.09412527e+00 -5.88139355e-01 -1.31384397e+00 5.77329636e-01 1.12373066e+00 2.01374993e-01 -1.51241553e+00 3.94378781e-01 -1.02482900e-01 1.53817743e-01 1.23828936e+00 -1.53229225e+00 -5.62479906e-02 -2.03843594e-01 3.64576519e-01 -1.44232965e+00 -1.74216583e-01 -6.33231699e-01 -1.02887653e-01 1.17164266e+00 7.36863792e-01 -1.09467244e+00 5.30156732e-01 1.06442726e+00 1.42133340e-01 2.50401556e-01 -7.28156447e-01 -1.11999691e+00 2.40631774e-01 -4.07236189e-01 6.72219872e-01 1.91652501e+00 9.36374784e-01 2.03831628e-01 9.95638445e-02 5.29700577e-01 9.94978011e-01 -1.92699164e-01 5.70232034e-01 -1.73876119e+00 2.34335870e-01 -1.56392321e-01 -6.38976276e-01 4.82424825e-01 4.80260313e-01 -1.10726058e+00 -8.73960078e-01 -1.51981854e+00 -7.61094466e-02 -4.47745025e-01 -3.12767029e-01 1.19116223e+00 1.36256889e-01 -3.13561946e-01 1.36454776e-01 5.99375702e-02 -5.04253566e-01 -1.27052203e-01 5.92390478e-01 -3.46040159e-01 -4.93732244e-02 -6.79542601e-01 -7.70304203e-01 7.91395962e-01 9.34457898e-01 -5.31285882e-01 -4.83534843e-01 -2.02047646e-01 3.71689469e-01 -1.09032355e-02 3.37083429e-01 -1.03014803e+00 6.01025581e-01 -6.13686979e-01 4.22949314e-01 -3.79126847e-01 1.80775560e-02 -8.44389260e-01 1.70497000e-02 2.90048361e-01 -4.92623597e-02 1.34715125e-01 7.95009851e-01 4.38823014e-01 -7.13495836e-02 -3.57683897e-01 1.72632158e-01 -1.62951440e-01 -1.11042595e+00 4.33364332e-01 -3.84093314e-01 1.61115795e-01 1.33969343e+00 6.36774972e-02 -1.24606740e+00 -6.24532551e-02 -4.60041612e-01 2.56840497e-01 4.84827369e-01 9.21279311e-01 6.57345414e-01 -9.59649980e-01 -4.58839387e-01 -9.32960436e-02 3.06979567e-01 -1.02445185e+00 7.70539641e-02 -4.14809957e-02 -5.89032471e-01 9.35652971e-01 -8.46220315e-01 6.57867610e-01 -1.09571505e+00 1.12088847e+00 -5.33844009e-02 -8.57637823e-01 -5.70760071e-01 1.05378427e-01 -3.75190943e-01 -8.93932879e-01 -4.60507115e-03 2.33858258e-01 -5.27380586e-01 -3.93328846e-01 8.22795928e-01 1.86733752e-01 -1.31853759e-01 -2.25808263e-01 -7.92161644e-01 1.55031770e-01 -3.55693430e-01 -2.37432286e-01 1.65635383e+00 8.86986375e-01 -1.93047553e-01 -6.67112589e-01 7.69625306e-01 1.58454850e-01 -2.95746773e-01 -1.64273337e-01 9.47969496e-01 -3.79166126e-01 -2.97208726e-01 -1.65139651e+00 -3.45632881e-01 4.81779993e-01 -7.31849968e-02 3.66862088e-01 8.49484980e-01 -8.94352198e-02 8.03295135e-01 9.49984968e-01 9.39039469e-01 -1.13893580e+00 6.03627339e-02 9.10454988e-01 8.99592698e-01 -7.45142519e-01 6.75056353e-02 -6.84420943e-01 -8.51966560e-01 1.14514041e+00 8.66452694e-01 -7.32548833e-02 9.80298162e-01 5.91921628e-01 -5.40222265e-02 -7.60374010e-01 -7.99857140e-01 3.65142524e-03 3.52871388e-01 1.06863654e+00 -1.40560180e-01 -1.40388692e-02 7.06553236e-02 6.06049299e-01 4.50643003e-02 -2.98842311e-01 4.73040313e-01 1.20175505e+00 -2.93496549e-01 -1.61946344e+00 -4.96933609e-01 2.13652164e-01 -5.06971419e-01 -3.93438518e-01 -1.43512404e+00 1.13468778e+00 -1.32571325e-01 9.66370165e-01 -5.02429426e-01 -8.27138901e-01 2.92897671e-01 2.30718821e-01 6.44059554e-02 -6.94519639e-01 -1.05311859e+00 -7.65413046e-01 9.49052751e-01 -5.53788662e-01 1.16526999e-01 -2.65425891e-01 -1.23917162e+00 -4.45423633e-01 -8.11791867e-02 4.77120906e-01 9.42148566e-01 7.17661202e-01 7.27963626e-01 1.76594406e-01 9.19651762e-02 -8.47840905e-02 9.13921837e-03 -7.30062604e-01 7.69493431e-02 3.74446452e-01 -5.62856197e-01 -5.45223296e-01 -5.45939198e-03 -1.27229661e-01]
[6.516456604003906, 7.452075481414795]
12991382-dbed-4e02-8acc-b35dda6c3fd9
prise-demystifying-deep-lucas-kanade-with
2303.11526
null
https://arxiv.org/abs/2303.11526v1
https://arxiv.org/pdf/2303.11526v1.pdf
PRISE: Demystifying Deep Lucas-Kanade with Strongly Star-Convex Constraints for Multimodel Image Alignment
The Lucas-Kanade (LK) method is a classic iterative homography estimation algorithm for image alignment, but often suffers from poor local optimality especially when image pairs have large distortions. To address this challenge, in this paper we propose a novel Deep Star-Convexified Lucas-Kanade (PRISE) method for multimodel image alignment by introducing strongly star-convex constraints into the optimization problem. Our basic idea is to enforce the neural network to approximately learn a star-convex loss landscape around the ground truth give any data to facilitate the convergence of the LK method to the ground truth through the high dimensional space defined by the network. This leads to a minimax learning problem, with contrastive (hinge) losses due to the definition of strong star-convexity that are appended to the original loss for training. We also provide an efficient sampling based algorithm to leverage the training cost, as well as some analysis on the quality of the solutions from PRISE. We further evaluate our approach on benchmark datasets such as MSCOCO, GoogleEarth, and GoogleMap, and demonstrate state-of-the-art results, especially for small pixel errors. Code can be downloaded from https://github.com/Zhang-VISLab.
['Ziming Zhang', 'Xinming Huang', 'Yiqing Zhang']
2023-03-21
null
http://openaccess.thecvf.com//content/CVPR2023/html/Zhang_PRISE_Demystifying_Deep_Lucas-Kanade_With_Strongly_Star-Convex_Constraints_for_Multimodel_CVPR_2023_paper.html
http://openaccess.thecvf.com//content/CVPR2023/papers/Zhang_PRISE_Demystifying_Deep_Lucas-Kanade_With_Strongly_Star-Convex_Constraints_for_Multimodel_CVPR_2023_paper.pdf
cvpr-2023-1
['homography-estimation']
['computer-vision']
[-3.04149166e-02 2.31784657e-02 -1.05203539e-01 -2.12880686e-01 -1.02714455e+00 -3.98171902e-01 3.94167900e-01 -2.20232621e-01 -3.81930262e-01 5.64021647e-01 8.46966878e-02 -1.10179275e-01 -6.62537143e-02 -4.42021102e-01 -1.09849191e+00 -8.12270641e-01 1.65182412e-01 5.09229481e-01 -2.00821593e-01 -1.16830923e-01 2.16872558e-01 2.87194848e-01 -9.07347023e-01 -2.56848693e-01 1.09294295e+00 9.16630626e-01 2.03320757e-01 6.19462729e-01 5.14310062e-01 6.58966780e-01 -8.59955251e-02 -6.40870094e-01 7.61847973e-01 -3.65557998e-01 -7.62154758e-01 2.22011790e-01 1.06248915e+00 -4.16498691e-01 -2.75575906e-01 1.27271259e+00 5.62763572e-01 1.11961536e-01 3.18749100e-01 -1.35397220e+00 -5.30462503e-01 2.15557322e-01 -7.92747200e-01 -2.24195048e-01 -7.30952993e-03 1.80947632e-01 1.14725018e+00 -1.03345156e+00 5.96555889e-01 9.04404581e-01 9.27088261e-01 1.32736877e-01 -1.19877446e+00 -4.23066258e-01 -3.01224776e-02 1.90100670e-01 -1.44347942e+00 -4.59049493e-01 7.81161427e-01 -4.08273757e-01 6.94141030e-01 3.22977714e-02 6.00704610e-01 7.73039639e-01 1.23307794e-01 6.84363604e-01 9.30273235e-01 -3.63303304e-01 -1.55380934e-01 -1.00520596e-01 -1.87359706e-01 9.15248275e-01 1.06430739e-01 2.67776668e-01 -4.75412369e-01 4.46431525e-02 1.02044308e+00 -2.11144924e-01 -5.24474263e-01 -1.02614880e+00 -1.28388691e+00 6.79424047e-01 9.76550758e-01 -1.61177099e-01 -3.73457342e-01 4.42990303e-01 9.46822912e-02 6.16480932e-02 3.61753404e-01 4.24518406e-01 -2.01750129e-01 -2.59132162e-02 -7.63937294e-01 2.39544019e-01 5.53084195e-01 9.23800468e-01 1.05051279e+00 3.84405069e-02 1.12429671e-01 7.63799667e-01 3.08624655e-01 6.02360308e-01 1.55601963e-01 -1.24984109e+00 7.91845918e-01 2.62091964e-01 1.64899305e-01 -1.42199242e+00 -2.01663241e-01 -8.53095114e-01 -1.07379186e+00 2.73099363e-01 5.09923339e-01 -2.73612887e-02 -6.74997211e-01 1.75532949e+00 5.24424791e-01 4.17397052e-01 -1.68325230e-01 1.15947402e+00 2.88618416e-01 5.89741945e-01 -5.59157193e-01 -1.88195240e-02 8.17336679e-01 -1.39326191e+00 -4.04166102e-01 -4.54236388e-01 7.27804959e-01 -7.31911302e-01 1.21245217e+00 2.75493354e-01 -1.21050060e+00 -3.48164171e-01 -1.06619227e+00 -4.28257942e-01 1.08106114e-01 1.76552013e-01 2.25764811e-01 2.26786718e-01 -1.08015943e+00 5.85142672e-01 -9.13018107e-01 -8.99787620e-02 5.04297733e-01 2.54638284e-01 -4.04483140e-01 -2.30368320e-02 -7.04952776e-01 9.21794236e-01 1.74942717e-01 2.96979249e-01 -7.47261941e-01 -8.57387483e-01 -8.46704781e-01 1.62497610e-02 5.84256649e-01 -9.19608414e-01 1.04406357e+00 -1.13623047e+00 -1.51401663e+00 9.65381622e-01 -1.03810258e-01 -5.15845180e-01 9.23542440e-01 -4.84587818e-01 3.14768940e-01 -5.21072047e-03 1.45526111e-01 7.90266275e-01 6.77742660e-01 -1.42507970e+00 -4.36879992e-01 -2.71875918e-01 7.22222179e-02 4.24547613e-01 -1.27564386e-01 -3.00009131e-01 -6.79708660e-01 -7.12193727e-01 2.53829420e-01 -1.14493251e+00 -2.47897446e-01 3.16592723e-01 -5.87957084e-01 2.42578670e-01 3.73703212e-01 -9.51573133e-01 1.02031016e+00 -1.99590504e+00 4.41602916e-01 4.03689563e-01 2.43753225e-01 5.51779047e-02 -2.12814286e-01 1.66532502e-01 -5.81149980e-02 -3.32898021e-01 -5.54458320e-01 -6.75996184e-01 -1.25506977e-02 1.16502225e-01 -3.16619575e-01 8.75859261e-01 -1.27144203e-01 1.02175844e+00 -8.49163711e-01 -2.34594703e-01 2.90144205e-01 4.08256322e-01 -6.65010273e-01 1.46571591e-01 1.92116015e-02 4.97058749e-01 8.24846607e-03 4.56853688e-01 7.74823546e-01 -4.56413060e-01 -1.30476221e-01 -4.35425341e-01 -1.40861049e-01 1.10346369e-01 -1.16704154e+00 1.84278393e+00 -5.65285146e-01 7.26303399e-01 1.95267156e-01 -1.09234548e+00 6.72655225e-01 -1.04432248e-01 5.28835237e-01 -7.13216245e-01 1.69501990e-01 4.21526998e-01 -2.07370415e-01 -7.61926994e-02 3.54868591e-01 2.26752371e-01 3.23603362e-01 2.14589730e-01 -1.99667171e-01 -2.24728167e-01 -4.43662181e-02 3.54479067e-02 5.74231148e-01 2.00388327e-01 2.60306478e-01 -3.44205052e-01 5.69835663e-01 7.84407929e-02 5.82552731e-01 6.08099341e-01 8.68454389e-03 8.21811438e-01 2.84697145e-01 -5.07756293e-01 -1.37143135e+00 -9.20662463e-01 -9.04451218e-03 4.60074812e-01 4.21855360e-01 -1.76655889e-01 -8.61760557e-01 -5.27533174e-01 -1.21858172e-01 4.07241911e-01 -3.87867004e-01 -1.64172482e-02 -7.50560641e-01 -6.07200384e-01 2.39925891e-01 3.36435616e-01 8.58506322e-01 -5.33161283e-01 -4.82785523e-01 9.71422158e-03 -3.83615613e-01 -1.27220798e+00 -9.05685365e-01 -1.44354820e-01 -7.01512933e-01 -1.04210532e+00 -8.58039260e-01 -7.67205715e-01 9.79370296e-01 3.51332933e-01 1.10397243e+00 2.00727522e-01 -4.10204828e-02 2.81093359e-01 -2.89837047e-02 -5.98030388e-02 -1.82199627e-01 1.57572314e-01 1.22655055e-03 6.45914003e-02 -2.16887876e-01 -5.27658463e-01 -7.48120666e-01 6.50033176e-01 -5.34493268e-01 3.51345122e-01 4.36995357e-01 9.95538354e-01 9.38522100e-01 -2.45840877e-01 -1.18699215e-01 -3.89909118e-01 1.81646496e-01 -1.79527611e-01 -1.21907997e+00 1.90497816e-01 -7.17717648e-01 7.59112239e-02 6.19108319e-01 -2.33533666e-01 -5.75151026e-01 1.96805224e-01 -5.77029102e-02 -7.12292433e-01 3.88387978e-01 4.21109170e-01 -1.48612514e-01 -6.58755720e-01 5.43343425e-01 2.39048630e-01 2.14827850e-01 -3.11440170e-01 4.25290644e-01 3.08376491e-01 1.03876925e+00 -4.55219448e-01 1.13334239e+00 7.28946745e-01 2.26768851e-01 -5.75955451e-01 -9.61163402e-01 -4.42017078e-01 -5.38973987e-01 -2.05359250e-01 7.40537226e-01 -9.83079493e-01 -5.73575497e-01 7.32610524e-01 -1.05960643e+00 -7.34808922e-01 -1.86800614e-01 5.41915953e-01 -8.67841363e-01 6.39836192e-01 -2.17485473e-01 -3.89197975e-01 -3.66602570e-01 -1.23420477e+00 1.00861549e+00 1.09854368e-02 1.25105545e-01 -1.10952091e+00 1.44514740e-01 4.30088490e-01 2.09421933e-01 3.11024457e-01 4.63353395e-01 1.90710500e-02 -9.45403576e-01 -8.67213234e-02 -3.04883301e-01 5.70550799e-01 -1.21284142e-01 -8.24228600e-02 -5.11371791e-01 -5.54750741e-01 8.66761506e-02 -3.48003775e-01 7.70739377e-01 6.25262201e-01 1.08381033e+00 -5.86840808e-01 -6.87446743e-02 1.45418644e+00 1.58283269e+00 -2.88773209e-01 6.77127361e-01 7.24979937e-01 1.11719692e+00 4.13580328e-01 6.70822442e-01 2.76370108e-01 7.31521368e-01 1.09718597e+00 7.00750113e-01 -3.06218058e-01 -9.76355076e-02 -3.36832613e-01 2.87590444e-01 7.55412698e-01 8.83750618e-02 -1.60495013e-01 -9.03701365e-01 5.07958293e-01 -2.04423428e+00 -7.24784076e-01 -2.43647426e-01 2.47014093e+00 8.06507409e-01 -1.45870715e-01 2.53350697e-02 -1.64274395e-01 5.65792143e-01 2.32509449e-01 -5.51390946e-01 1.54626798e-02 -2.78310090e-01 -2.73723990e-01 8.32060933e-01 9.75808322e-01 -1.19435787e+00 1.03602147e+00 5.27018499e+00 7.52608418e-01 -1.34523058e+00 6.56247661e-02 6.41476452e-01 -9.04137790e-02 -7.18867853e-02 8.06191415e-02 -5.88787854e-01 4.35413063e-01 3.45969766e-01 -1.97408363e-01 7.21972942e-01 7.63296783e-01 3.00491959e-01 -1.08323328e-01 -9.35829282e-01 1.23051465e+00 1.64609462e-01 -1.50567174e+00 -2.85159081e-01 2.03328684e-01 1.12752402e+00 3.69768232e-01 2.52786815e-01 -3.87975872e-01 3.73401582e-01 -8.27090204e-01 7.47850657e-01 3.41460556e-01 7.19330907e-01 -6.52750075e-01 5.61393201e-01 2.46280804e-01 -9.30343449e-01 2.81630475e-02 -4.41740483e-01 1.26038894e-01 3.57736945e-01 6.05025887e-01 -5.62145889e-01 6.12624347e-01 6.35977626e-01 9.20058548e-01 -3.09033036e-01 1.23663914e+00 -3.51053923e-01 1.87685609e-01 -5.69029927e-01 4.15969163e-01 3.64454061e-01 -6.14218891e-01 7.02581048e-01 7.48730421e-01 4.43949312e-01 -1.62941352e-01 2.20781177e-01 9.06426668e-01 -2.18006268e-01 7.68566206e-02 -4.30438459e-01 3.31260830e-01 3.70321870e-01 1.22185028e+00 -3.84998262e-01 -9.10732374e-02 -2.86291271e-01 1.17809331e+00 5.35233557e-01 3.76562119e-01 -8.61399889e-01 -1.54494300e-01 6.91191196e-01 -2.38579675e-03 2.31038660e-01 -4.46972162e-01 -5.19435942e-01 -1.25295925e+00 4.06487346e-01 -9.38858569e-01 1.62753895e-01 -8.82662177e-01 -1.03303230e+00 4.18400437e-01 -1.53914377e-01 -1.40452790e+00 -1.25133917e-01 -4.43110347e-01 -6.11029685e-01 9.08873320e-01 -1.65637374e+00 -1.18388319e+00 -6.77260101e-01 4.34461832e-01 2.23092645e-01 -5.10496534e-02 1.84219599e-01 4.27163064e-01 -5.98069906e-01 7.74249315e-01 3.40385765e-01 1.06701210e-01 7.29695976e-01 -1.21273816e+00 4.92131829e-01 1.08653009e+00 2.63374537e-01 4.07218874e-01 6.29181147e-01 -3.53550553e-01 -1.23679090e+00 -1.01787806e+00 7.84928381e-01 -4.97924000e-01 6.60871744e-01 -1.51165485e-01 -7.77156770e-01 7.78438032e-01 1.45767942e-01 5.02965748e-02 7.66944587e-02 -3.32264632e-01 -2.98374534e-01 -2.80878961e-01 -8.59574676e-01 7.39485800e-01 1.06073153e+00 -3.54634523e-01 1.27997464e-02 5.16953349e-01 5.27052343e-01 -9.38400030e-01 -6.75912976e-01 2.72362083e-01 4.99915659e-01 -1.07120883e+00 1.14343381e+00 -2.38084093e-01 6.78002119e-01 -5.14309943e-01 -9.25631002e-02 -1.43705893e+00 -2.56848514e-01 -8.31737399e-01 -8.68977308e-02 8.09529305e-01 2.96404988e-01 -7.33228147e-01 8.46025050e-01 3.04410666e-01 -2.43398070e-01 -9.45431054e-01 -9.90500391e-01 -8.78956437e-01 1.12032875e-01 -2.15771317e-01 3.86478364e-01 9.53430653e-01 -4.48544055e-01 1.01630084e-01 -6.59402847e-01 5.20125806e-01 1.03731608e+00 -2.83781346e-02 1.20221794e+00 -7.06002593e-01 -4.92375493e-01 -5.45858860e-01 -4.37068373e-01 -1.58547091e+00 1.41740263e-01 -9.51223314e-01 2.18747199e-01 -1.32053602e+00 8.87891874e-02 -5.11983216e-01 1.80378303e-01 3.64327669e-01 -2.18160436e-01 3.29223990e-01 3.20398539e-01 5.76557755e-01 -3.64040017e-01 6.94317222e-01 1.35422099e+00 -7.26889521e-02 -1.83103770e-01 -9.06903148e-02 -4.66277540e-01 6.31654143e-01 6.80451393e-01 -3.47725332e-01 -3.48928779e-01 -9.53697443e-01 4.45997804e-01 -1.55023023e-01 7.90445387e-01 -8.96349549e-01 3.75710517e-01 -4.43005748e-02 2.73197219e-02 -4.99155879e-01 2.65533358e-01 -8.43803167e-01 2.42312372e-01 4.10566598e-01 -2.87384361e-01 2.39590988e-01 -5.74912876e-02 3.00707817e-01 -1.66851372e-01 -7.56385329e-04 1.05078411e+00 2.02704534e-01 -3.92023087e-01 6.35405064e-01 4.57112074e-01 2.83587575e-01 7.83300698e-01 -3.16808701e-01 -2.65016854e-01 -6.07667744e-01 -2.93852687e-01 5.81189871e-01 9.10059452e-01 1.56574205e-01 4.97008443e-01 -1.39352167e+00 -7.98284173e-01 1.75788850e-01 -1.00815529e-03 4.76332903e-01 7.61181414e-02 1.35558259e+00 -9.49341893e-01 2.42022678e-01 -1.66434303e-01 -8.40732813e-01 -1.16078675e+00 3.59955221e-01 7.94503510e-01 -2.18345240e-01 -7.06695735e-01 9.84147072e-01 3.18451166e-01 -5.95375717e-01 3.18563730e-01 -2.23468795e-01 3.38324338e-01 -3.82054716e-01 1.56792596e-01 5.41483700e-01 -5.03257066e-02 -8.02565575e-01 -1.68881655e-01 9.21639025e-01 6.86251149e-02 -1.74452186e-01 1.18150425e+00 -3.16551417e-01 -1.87785834e-01 -8.24562982e-02 1.53598642e+00 -9.78205577e-02 -1.64842403e+00 -4.32566643e-01 -2.31486440e-01 -7.03883410e-01 1.15108065e-01 -5.01611054e-01 -1.32443535e+00 7.83090889e-01 5.07839382e-01 -2.17119545e-01 1.04158247e+00 -1.37645036e-01 9.15873289e-01 3.81102055e-01 1.73221022e-01 -1.06838524e+00 1.80937201e-01 5.26320457e-01 1.10568249e+00 -1.33393359e+00 7.83395618e-02 -3.82872075e-01 -5.49991786e-01 1.00069427e+00 5.39379716e-01 -4.71728444e-01 3.93617600e-01 6.94126263e-02 2.29814321e-01 -2.69460864e-02 -2.40057498e-01 7.36920089e-02 3.04648668e-01 4.15481806e-01 -4.90423851e-02 -2.40916803e-01 -3.34743671e-02 -1.12281144e-01 -4.28308189e-01 -1.57891870e-01 5.05499423e-01 5.63631475e-01 -1.07444182e-01 -9.01757360e-01 -2.89237827e-01 7.53093213e-02 -1.08158655e-01 -1.42248139e-01 -2.85694003e-01 6.42854393e-01 -1.69879749e-01 4.82455671e-01 1.36918724e-01 -9.32661593e-02 2.50814736e-01 -5.87207377e-01 5.17936885e-01 -1.00771643e-01 -1.39675513e-01 4.72894795e-02 -1.32836893e-01 -9.08748567e-01 -3.86205226e-01 -6.27172351e-01 -1.05294037e+00 -4.85459954e-01 -2.54760146e-01 -1.19966574e-01 6.40923440e-01 8.38517725e-01 3.97249132e-01 1.02110751e-01 9.23178375e-01 -9.95590627e-01 -8.31137717e-01 -5.00236392e-01 -1.20037004e-01 3.60002875e-01 6.09123945e-01 -4.34638470e-01 -5.26642442e-01 -2.07342971e-02]
[8.64970588684082, -2.2700560092926025]
d59e79b3-a6c3-4132-8e88-b1f0fc5292d3
teaching-arithmetic-to-small-transformers
2307.03381
null
https://arxiv.org/abs/2307.03381v1
https://arxiv.org/pdf/2307.03381v1.pdf
Teaching Arithmetic to Small Transformers
Large language models like GPT-4 exhibit emergent capabilities across general-purpose tasks, such as basic arithmetic, when trained on extensive text data, even though these tasks are not explicitly encoded by the unsupervised, next-token prediction objective. This study investigates how small transformers, trained from random initialization, can efficiently learn arithmetic operations such as addition, multiplication, and elementary functions like square root, using the next-token prediction objective. We first demonstrate that conventional training data is not the most effective for arithmetic learning, and simple formatting changes can significantly improve accuracy. This leads to sharp phase transitions as a function of training data scale, which, in some cases, can be explained through connections to low-rank matrix completion. Building on prior work, we then train on chain-of-thought style data that includes intermediate step results. Even in the complete absence of pretraining, this approach significantly and simultaneously improves accuracy, sample complexity, and convergence speed. We also study the interplay between arithmetic and text data during training and examine the effects of few-shot prompting, pretraining, and model scale. Additionally, we discuss length generalization challenges. Our work highlights the importance of high-quality, instructive data that considers the particular characteristics of the next-word prediction objective for rapidly eliciting arithmetic capabilities.
['Dimitris Papailiopoulos', 'Kangwook Lee', 'Jason D. Lee', 'Kartik Sreenivasan', 'Nayoung Lee']
2023-07-07
null
null
null
null
['low-rank-matrix-completion', 'matrix-completion']
['methodology', 'methodology']
[ 3.66795629e-01 -1.50402039e-02 -7.61879459e-02 -1.53111830e-01 -5.43285608e-01 -4.57108289e-01 6.94722354e-01 5.53333461e-01 -4.77586418e-01 3.48358512e-01 4.68487889e-01 -5.58202147e-01 -2.00444162e-01 -1.00585008e+00 -8.11620295e-01 -3.51527989e-01 -3.72443199e-01 6.08215690e-01 -1.87225491e-01 -5.95462918e-01 5.04393399e-01 1.13770731e-01 -1.49606705e+00 4.47135597e-01 1.18676901e+00 7.41446793e-01 2.69889057e-01 6.63706899e-01 -1.56225622e-01 1.24029613e+00 -5.23357391e-01 -4.11772490e-01 3.21799725e-01 -1.82921931e-01 -7.73667276e-01 -2.27736399e-01 5.10654509e-01 -4.52507406e-01 -2.76401401e-01 7.33855844e-01 3.65923017e-01 4.54848319e-01 7.26440430e-01 -8.26466382e-01 -8.74971032e-01 1.11509693e+00 -2.27361042e-02 1.68975785e-01 2.27559641e-01 6.88152075e-01 1.30933046e+00 -9.55620944e-01 2.64479011e-01 1.12882972e+00 9.17503119e-01 1.36190042e-01 -1.59423590e+00 -5.56605756e-01 4.59308289e-02 1.28202856e-01 -1.33992410e+00 -6.29309535e-01 3.77140492e-01 -4.35284793e-01 1.39735794e+00 -8.62517506e-02 7.42241085e-01 9.01374876e-01 3.25690329e-01 8.22705090e-01 1.00920045e+00 -4.75536019e-01 1.99075237e-01 -3.34317386e-01 3.31981897e-01 8.89113486e-01 1.82128966e-01 -4.02788147e-02 -7.94800878e-01 8.43207911e-02 5.48532188e-01 -1.95427030e-01 -1.70446083e-01 -9.65379849e-02 -1.31856751e+00 6.52690411e-01 1.27156720e-01 2.47260377e-01 -3.03670228e-01 3.04893315e-01 4.53470051e-01 7.33147025e-01 3.41561317e-01 1.17493916e+00 -7.22363591e-01 -6.23950005e-01 -9.62220192e-01 1.53938532e-01 9.91527319e-01 7.38963723e-01 7.56127298e-01 3.96415472e-01 -2.99698651e-01 5.45684338e-01 -1.51867658e-01 4.78931308e-01 6.20622635e-01 -9.73020494e-01 6.41972125e-01 5.27672887e-01 -1.33224070e-01 -8.90992761e-01 -6.31593764e-01 -7.08208740e-01 -6.06629610e-01 -1.27045959e-01 6.69857442e-01 -2.34395966e-01 -6.91171944e-01 1.86757195e+00 -1.55232117e-01 -6.29172772e-02 2.61333555e-01 4.98213261e-01 4.94327575e-01 8.16448629e-01 2.33839527e-01 -1.18162654e-01 1.23631740e+00 -9.78599906e-01 -4.78639454e-01 -4.99051720e-01 1.42285466e+00 -6.38556361e-01 1.72103000e+00 6.22810364e-01 -1.34048140e+00 -6.05914950e-01 -9.78778481e-01 -3.53110373e-01 -3.22752863e-01 7.05759216e-04 9.86718655e-01 7.06916094e-01 -1.20745051e+00 1.00747991e+00 -9.23925102e-01 -2.68271208e-01 2.69571424e-01 4.03925776e-01 -2.90706363e-02 -2.84475744e-01 -1.19474757e+00 1.13090551e+00 4.97416288e-01 -2.72565067e-01 -7.76716888e-01 -1.16078568e+00 -8.34906638e-01 5.40954113e-01 3.38271767e-01 -7.41485834e-01 1.23016310e+00 -7.90559471e-01 -1.75549924e+00 4.99734700e-01 1.36500979e-02 -6.53693080e-01 6.92955628e-02 -3.13121706e-01 2.08887998e-02 -6.08130321e-02 -3.30784708e-01 8.00400436e-01 6.50758684e-01 -5.67852795e-01 -3.93658653e-02 -2.15335459e-01 2.79319674e-01 4.47292447e-01 -7.39319086e-01 -2.30274871e-01 -5.78527339e-02 -6.14765644e-01 -1.10319763e-01 -6.65403605e-01 -1.22417934e-01 -3.74457687e-01 -1.29574612e-01 -3.45197946e-01 1.10076189e-01 -6.07004106e-01 1.21976781e+00 -2.25088763e+00 1.83442459e-01 1.13890558e-01 2.01565892e-01 -3.54185067e-02 -4.68340516e-01 5.93722522e-01 8.31531808e-02 1.93421930e-01 -8.10328722e-02 -4.25785094e-01 2.71180779e-01 2.50436775e-02 -4.42355812e-01 2.65670359e-01 3.46504331e-01 1.22970176e+00 -8.57863128e-01 -2.54690915e-01 1.57368332e-01 9.94263142e-02 -1.11581790e+00 -4.41972725e-02 -5.85109472e-01 2.10881740e-01 6.94110841e-02 3.09002548e-01 6.40163198e-02 -6.68763220e-01 1.09864026e-01 -1.72695771e-01 -1.73927248e-02 7.71909893e-01 -8.29710901e-01 1.91913855e+00 -7.40336776e-01 6.58591926e-01 -2.82281697e-01 -1.14280403e+00 6.12723410e-01 -2.87058633e-02 2.59061456e-01 -8.94403517e-01 4.40974347e-02 9.03313383e-02 6.00538254e-01 -2.75786847e-01 7.83882380e-01 -2.02279299e-01 -1.94594059e-02 6.78176939e-01 1.91605181e-01 -4.66042459e-01 5.76605797e-01 4.36986744e-01 1.31930089e+00 -6.62019253e-02 2.19547555e-01 -4.81197566e-01 7.96795171e-03 1.06004164e-01 1.53207509e-02 9.86927271e-01 2.28537455e-01 9.46931243e-02 6.02394283e-01 -3.05546731e-01 -1.12078106e+00 -7.79815853e-01 -1.02467872e-01 1.72144949e+00 -3.05585504e-01 -1.03643870e+00 -4.42634821e-01 -4.50989157e-02 -8.90516117e-02 9.95062828e-01 -4.83490288e-01 -4.61995333e-01 -5.39330006e-01 -7.67783999e-01 6.62943304e-01 4.45724070e-01 4.86996621e-01 -8.05960357e-01 -6.38376176e-01 2.40963265e-01 2.53914706e-02 -1.17303336e+00 -2.86770791e-01 4.75233823e-01 -1.15186393e+00 -6.73934996e-01 -2.96468645e-01 -5.37091613e-01 5.80884457e-01 -8.64479914e-02 1.41013229e+00 5.55336893e-01 2.56951209e-02 4.93563354e-01 -1.41737849e-01 -3.99270318e-02 -4.65876490e-01 4.66308415e-01 7.37473965e-02 -5.43226063e-01 1.27869710e-01 -9.42134023e-01 -2.72461593e-01 -1.15282811e-01 -5.82562089e-01 5.48242569e-01 8.00953269e-01 9.34225559e-01 1.76419914e-02 1.12910539e-01 2.74576128e-01 -8.55311513e-01 7.49041140e-01 -3.52927953e-01 -3.92979443e-01 2.23569900e-01 -5.80279350e-01 5.96354306e-01 1.00126469e+00 -6.96540236e-01 -7.08408475e-01 -2.93781370e-01 8.81287158e-02 -1.71417207e-01 1.30658448e-01 9.55451727e-01 3.05374324e-01 -4.16846648e-02 1.09344411e+00 5.27728617e-01 -3.32263410e-02 -1.83065608e-01 4.50767428e-01 1.17160633e-01 2.71291614e-01 -1.16123283e+00 6.97902143e-01 -2.28819791e-02 -5.43142715e-03 -9.33488607e-01 -8.95795107e-01 1.04860924e-01 -4.92868006e-01 2.13347927e-01 6.16296768e-01 -1.02685082e+00 -7.41770327e-01 4.90282953e-01 -9.64505851e-01 -1.18764889e+00 -4.35248554e-01 4.92561430e-01 -4.55787659e-01 2.05127954e-01 -1.06315374e+00 -6.16626441e-01 -3.78549844e-01 -1.14314520e+00 8.46875489e-01 -1.96443379e-01 -4.61548537e-01 -1.23490512e+00 -3.49157155e-01 2.28152394e-01 7.62937009e-01 -3.46131653e-01 1.62459433e+00 -8.48426700e-01 -7.67174363e-01 1.79544136e-01 -3.18338983e-02 1.99024141e-01 -2.61617362e-01 -1.54213771e-01 -7.33682871e-01 -2.47598812e-01 1.50796408e-02 -9.06720638e-01 8.04399908e-01 9.59211290e-02 1.21526122e+00 -4.40081179e-01 1.71265393e-01 7.30051219e-01 1.05409324e+00 -2.70090491e-01 4.48755533e-01 1.18831605e-01 6.33056402e-01 2.63785452e-01 1.16549671e-01 5.04047215e-01 5.36514342e-01 2.72902638e-01 -4.76641320e-02 3.00345242e-01 6.71889111e-02 -3.40291142e-01 6.00225091e-01 1.31024241e+00 -3.59757878e-02 1.71833541e-02 -1.36899912e+00 2.25049779e-01 -1.32669711e+00 -8.43876958e-01 2.06841916e-01 2.17936778e+00 1.28129041e+00 4.59019512e-01 -1.89066246e-01 2.14265764e-01 -4.29673046e-02 1.79760844e-01 -4.26386565e-01 -3.78257513e-01 -1.72840897e-02 8.44707131e-01 4.86728132e-01 5.87469578e-01 -6.86895430e-01 1.20554352e+00 6.94265699e+00 9.73295510e-01 -1.19842196e+00 9.76572931e-02 7.15126693e-01 -1.89200744e-01 -4.38335538e-01 -5.62883960e-03 -5.65664649e-01 2.41824746e-01 1.41492534e+00 -9.46085826e-02 9.17459369e-01 6.83692276e-01 -1.27611484e-03 -5.48477694e-02 -1.56405103e+00 6.66420519e-01 8.77839327e-03 -1.33510005e+00 1.89505607e-01 -1.42398700e-01 7.71617174e-01 2.64391750e-02 1.96449548e-01 1.03491426e+00 5.49595475e-01 -1.41489661e+00 5.96918702e-01 2.49884158e-01 7.46409237e-01 -4.70697403e-01 1.83642268e-01 6.27434671e-01 -1.02868164e+00 -2.60757804e-01 -3.94752234e-01 -8.05425406e-01 -2.59825617e-01 4.54061687e-01 -7.09481835e-01 8.30363929e-02 1.45709679e-01 7.05214679e-01 -8.38109612e-01 6.81749821e-01 -2.68157542e-01 8.21770310e-01 -4.26126122e-01 -2.05362633e-01 1.78554982e-01 3.45729887e-02 5.09586632e-02 1.06352317e+00 2.82037050e-01 4.70780462e-01 3.21245581e-01 8.34068716e-01 -8.28236714e-02 2.34693363e-01 -3.91147375e-01 -4.64286417e-01 6.01483822e-01 8.23354840e-01 -6.45683050e-01 -4.94101912e-01 -3.91300738e-01 6.10452592e-01 6.94082797e-01 3.86634827e-01 -4.87459898e-01 -1.26901641e-01 4.01369274e-01 8.06041211e-02 2.80281901e-01 -7.72219896e-01 -7.57327974e-01 -1.36968219e+00 -2.88642079e-01 -1.34036481e+00 1.01071939e-01 -7.92931557e-01 -1.04558480e+00 -7.18390346e-02 1.61610532e-03 -4.87137645e-01 -2.41442427e-01 -8.08000207e-01 -7.09814668e-01 5.79109728e-01 -9.61629689e-01 -7.81680405e-01 -1.01246201e-01 4.50715363e-01 6.64049625e-01 4.74268105e-03 9.18386877e-01 5.76867089e-02 -5.66005468e-01 7.50003874e-01 1.16131328e-01 5.78152016e-02 4.82138902e-01 -1.18739140e+00 6.26259148e-01 7.76284635e-01 3.00183773e-01 1.11971068e+00 7.82112837e-01 -5.39416373e-01 -1.90892136e+00 -7.11942613e-01 7.37680435e-01 -5.49578488e-01 1.00609636e+00 -6.23781443e-01 -8.86772335e-01 8.61841857e-01 1.22991718e-01 -2.79041111e-01 5.94529748e-01 7.33740270e-01 -4.73966271e-01 2.86461357e-02 -5.54573715e-01 9.07112122e-01 1.21832037e+00 -7.30262935e-01 -7.47043192e-01 5.11968791e-01 9.54371452e-01 -5.73445141e-01 -1.08344424e+00 3.14074457e-01 3.44870776e-01 -6.21106863e-01 1.10078776e+00 -7.09271848e-01 9.91358757e-01 3.50574791e-01 -2.17246085e-01 -1.41476786e+00 -3.15414071e-01 -6.62052214e-01 -2.69636422e-01 8.09308648e-01 6.12525940e-01 -4.55223858e-01 6.48570061e-01 7.33673275e-01 -2.69136906e-01 -9.00204957e-01 -4.31944311e-01 -7.45056689e-01 5.23386419e-01 -7.71724105e-01 4.27338123e-01 1.05272007e+00 4.62226361e-01 7.42687345e-01 -1.52782664e-01 -2.06921667e-01 2.25185126e-01 3.49419713e-02 9.87299323e-01 -9.85572994e-01 -7.51586795e-01 -6.12203598e-01 -2.04140291e-01 -1.46091652e+00 1.60291374e-01 -1.13850570e+00 -7.59246796e-02 -1.20661604e+00 -3.31314765e-02 -6.62582159e-01 -2.27918960e-02 5.68552315e-01 -2.49262288e-01 -1.49412155e-01 3.93890947e-01 1.63563550e-01 -6.61543608e-01 5.98878741e-01 1.19803953e+00 -2.76359655e-02 -1.28101960e-01 -3.44450623e-01 -7.65694320e-01 5.25290072e-01 8.64229620e-01 -1.67352870e-01 -5.98291814e-01 -7.68735170e-01 6.23311400e-01 1.17831245e-01 5.37985973e-02 -1.36076033e+00 4.91912663e-01 -1.82855338e-01 3.43536556e-01 -2.18869060e-01 4.76473808e-01 -3.35425377e-01 -3.57653975e-01 5.15871644e-01 -7.29381442e-01 3.93999547e-01 5.01630664e-01 2.03923538e-01 2.71061093e-01 -2.76195645e-01 5.42141795e-01 -2.72627383e-01 -5.52194834e-01 3.14002596e-02 -5.36045730e-01 6.16970003e-01 4.46776420e-01 1.92238446e-02 -5.28046250e-01 -3.84574145e-01 -4.40475672e-01 1.59877062e-01 5.04996002e-01 4.51048277e-02 2.77980596e-01 -9.89230573e-01 -7.65821815e-01 1.99608520e-01 -1.51613921e-01 1.42046320e-03 1.98279601e-02 1.18715060e+00 -3.93698364e-01 4.38888639e-01 -7.08830506e-02 -5.96445620e-01 -7.38349080e-01 6.46026552e-01 9.84758288e-02 -6.45386755e-01 -5.01646936e-01 8.59472632e-01 1.71069950e-01 -4.55927491e-01 2.68605947e-01 -8.84140432e-01 1.17413163e-01 -9.29988548e-02 3.85718793e-01 1.01152077e-01 6.66736886e-02 1.05151720e-01 5.58805130e-02 3.32334697e-01 -9.66203734e-02 -2.46916320e-02 1.20807827e+00 2.59642571e-01 -1.47418246e-01 6.89933002e-01 9.51508284e-01 -4.91219126e-02 -1.01840079e+00 -3.92672867e-01 5.87429702e-02 -3.24517898e-02 -1.96327612e-01 -6.99010670e-01 -7.02154696e-01 1.11507070e+00 5.89315407e-02 -3.28873582e-02 8.95024240e-01 -4.24785078e-01 5.38699448e-01 1.10318267e+00 3.68329883e-01 -9.49830890e-01 6.11631513e-01 1.21737421e+00 4.89312857e-01 -1.16033638e+00 1.47567391e-01 -7.28572831e-02 -3.78241122e-01 1.14614666e+00 7.97130823e-01 -1.21338755e-01 4.12887186e-01 4.29470599e-01 -5.47843635e-01 -1.07432611e-01 -1.28979015e+00 -1.45887313e-02 8.83825719e-02 2.46314004e-01 6.55547976e-01 -1.89910620e-01 -4.24018204e-02 6.23635173e-01 -9.81847882e-01 -1.33247197e-01 3.31468821e-01 9.58333910e-01 -5.95142484e-01 -8.15426290e-01 -1.78275287e-01 8.61929595e-01 -3.30100626e-01 -9.98576224e-01 6.23854846e-02 8.06645513e-01 -1.89605713e-01 6.27726912e-01 3.14573169e-01 -5.61458588e-01 -3.12932134e-02 4.89464104e-01 6.97329819e-01 -8.57395232e-01 -8.78213704e-01 -6.37750566e-01 2.36188278e-01 -3.82717788e-01 1.57768071e-01 -4.83799547e-01 -1.19393253e+00 -8.20873022e-01 -2.43544281e-01 1.23612024e-01 2.47038797e-01 1.25663388e+00 3.71976584e-01 4.69752997e-01 3.26348953e-02 -8.84205341e-01 -1.02081537e+00 -1.17082596e+00 -4.27381806e-02 3.99109006e-01 8.78922716e-02 -3.45916420e-01 -3.94946396e-01 1.08604617e-02]
[9.664321899414062, 7.377455711364746]
4244b51a-7577-4c5c-98b6-b49423473ad9
transforma-at-semeval-2019-task-6-offensive
1903.05280
null
http://arxiv.org/abs/1903.05280v3
http://arxiv.org/pdf/1903.05280v3.pdf
Offensive Language Analysis using Deep Learning Architecture
SemEval-2019 Task 6 (Zampieri et al., 2019b) requires us to identify and categorise offensive language in social media. In this paper we will describe the process we took to tackle this challenge. Our process is heavily inspired by Sosa (2017) where he proposed CNN-LSTM and LSTM-CNN models to conduct twitter sentiment analysis. We decided to follow his approach as well as further his work by testing out different variations of RNN models with CNN. Specifically, we have divided the challenge into two parts: data processing and sampling and choosing the optimal deep learning architecture. In preprocessing, we experimented with two techniques, SMOTE and Class Weights to counter the imbalance between classes. Once we are happy with the quality of our input data, we proceed to choosing the optimal deep learning architecture for this task. Given the quality and quantity of data we have been given, we found that the addition of CNN layer provides very little to no additional improvement to our model's performance and sometimes even lead to a decrease in our F1-score. In the end, the deep learning architecture that gives us the highest macro F1-score is a simple BiLSTM-CNN.
['Ryan Ong']
2019-03-12
null
null
null
null
['twitter-sentiment-analysis', 'abuse-detection']
['natural-language-processing', 'natural-language-processing']
[-2.24308167e-02 1.37805670e-01 6.11551255e-02 -5.23798645e-01 -5.03651679e-01 -5.17954409e-01 7.61412084e-01 2.74017423e-01 -8.65916491e-01 4.09663141e-01 3.32609743e-01 -5.18085659e-01 1.85990453e-01 -7.17445850e-01 -5.04481196e-01 -3.02149266e-01 1.56345546e-01 2.73163646e-01 5.65760536e-03 -7.53517330e-01 4.44796115e-01 3.04187506e-01 -1.32416916e+00 5.70133388e-01 5.00433445e-01 1.04304636e+00 -3.49544942e-01 6.13410592e-01 -6.44907132e-02 1.22623444e+00 -6.78336322e-01 -7.13153243e-01 1.94811508e-01 -3.17524761e-01 -1.09854937e+00 -3.43482733e-01 1.82381526e-01 -1.55641407e-01 5.90810850e-02 9.47598219e-01 4.88782018e-01 2.64444292e-01 5.46419442e-01 -8.38368177e-01 -3.15448970e-01 9.64870334e-01 -4.03605342e-01 3.84013742e-01 7.14904629e-03 1.53935984e-01 1.01526952e+00 -9.08961475e-01 5.09051144e-01 1.03776646e+00 8.76459777e-01 6.69479132e-01 -1.09812629e+00 -6.51913285e-01 1.42026946e-01 9.28933546e-02 -1.13373685e+00 -3.96942198e-01 6.25868738e-01 -4.09627199e-01 1.01980472e+00 1.14023976e-01 6.65530145e-01 1.41936707e+00 6.11347985e-03 8.84419501e-01 1.11638904e+00 -5.00054598e-01 3.43153514e-02 3.05624604e-01 3.29397470e-01 2.58221537e-01 1.45090939e-02 -1.12965763e-01 -2.96840429e-01 6.77296668e-02 2.09590629e-01 -1.04789503e-01 1.10383339e-01 2.59182036e-01 -9.47860360e-01 1.04520273e+00 7.34309316e-01 7.07935333e-01 -3.65340889e-01 8.73010978e-02 7.79125214e-01 5.12181640e-01 7.25813687e-01 8.69453669e-01 -6.39448643e-01 -1.44289270e-01 -1.10135972e+00 2.59325027e-01 8.60651016e-01 1.71917781e-01 2.95092046e-01 1.90452784e-01 -6.53654858e-02 9.63328183e-01 -1.34752274e-01 2.24392023e-02 7.94466674e-01 -5.44029355e-01 3.73883098e-01 4.76519525e-01 -6.24525696e-02 -1.07632303e+00 -5.12996852e-01 -5.62136233e-01 -6.33831739e-01 1.54091492e-01 4.40675497e-01 -5.67175329e-01 -8.74043763e-01 1.67474997e+00 -1.52238980e-01 -1.31386012e-01 -1.11081889e-02 5.56740105e-01 7.39432275e-01 5.01818240e-01 1.97055876e-01 1.74234301e-01 1.20479059e+00 -8.28024447e-01 -5.07782996e-01 -2.38213882e-01 9.42894578e-01 -8.83188725e-01 1.06272769e+00 5.73216081e-01 -8.27893674e-01 -5.74788690e-01 -1.05399144e+00 2.72648269e-03 -7.85544276e-01 -1.10831365e-01 4.41072702e-01 6.52910590e-01 -1.05480492e+00 9.88948584e-01 -6.98447943e-01 -3.79378229e-01 4.54278588e-01 5.04443169e-01 -1.52272046e-01 3.57879281e-01 -1.46616435e+00 1.27829957e+00 4.92443949e-01 3.16250950e-01 -5.00109911e-01 -3.59113693e-01 -6.02027535e-01 -5.34666851e-02 2.93842971e-01 -2.81394273e-01 1.62042725e+00 -1.47052598e+00 -1.41725552e+00 9.06176031e-01 8.98422897e-02 -7.88839221e-01 4.26875889e-01 -3.41313720e-01 -2.59495169e-01 -4.46250916e-01 -4.88677584e-02 6.05898082e-01 4.75468129e-01 -9.94983196e-01 -6.68183506e-01 -2.80239433e-01 1.56718984e-01 3.70613672e-02 -5.21664679e-01 4.37297821e-01 9.66087431e-02 -5.96485257e-01 -3.29217225e-01 -9.07162547e-01 -2.61542380e-01 -7.75816202e-01 -5.16779661e-01 -3.80998433e-01 5.45993209e-01 -5.43459594e-01 1.06173933e+00 -2.08725834e+00 -1.70470715e-01 2.43144810e-01 2.99697697e-01 5.94227850e-01 -1.34060577e-01 3.98291528e-01 -3.26260239e-01 3.89802665e-01 3.74145582e-02 -5.64055264e-01 -3.38216759e-02 -5.74962646e-02 -4.82022077e-01 3.23829114e-01 2.02630088e-01 8.97210777e-01 -8.07995558e-01 -1.38216794e-01 4.49794307e-02 6.99200392e-01 -4.68127519e-01 6.61540870e-03 -1.33154362e-01 1.56649128e-01 -2.41091520e-01 4.20180738e-01 3.45084518e-01 -2.71796230e-02 3.29128951e-02 -2.32767444e-02 -2.32604891e-01 6.69577062e-01 -8.31500113e-01 1.07404923e+00 -6.11204386e-01 1.05538619e+00 -7.46466592e-02 -8.31939399e-01 9.26557362e-01 2.80050397e-01 3.25445503e-01 -8.72408450e-01 8.40734601e-01 4.72514600e-01 4.19115633e-01 -3.16354156e-01 7.46909201e-01 -4.18986857e-01 -4.87989299e-02 4.94563967e-01 6.87501878e-02 7.03283027e-02 2.17465341e-01 -6.89672455e-02 8.69161665e-01 -9.40767601e-02 1.78470179e-01 -4.06426698e-01 4.48597193e-01 -4.94092368e-02 1.95322290e-01 6.56494617e-01 -1.96599409e-01 4.91927832e-01 7.19453275e-01 -7.06984937e-01 -1.00131762e+00 -4.49791014e-01 3.09821721e-02 1.38927460e+00 -5.93309641e-01 -2.77114898e-01 -8.80168319e-01 -6.43813372e-01 -3.61763567e-01 9.02674079e-01 -1.06862020e+00 -1.50049895e-01 -6.98376298e-01 -9.13179576e-01 6.62482142e-01 3.69207263e-01 2.82126486e-01 -1.62213969e+00 -7.25005627e-01 2.17950538e-01 -9.96198058e-02 -7.86903620e-01 -1.67482913e-01 7.50660598e-01 -6.53441548e-01 -9.08109546e-01 -4.99010473e-01 -5.78378022e-01 2.20781118e-01 -7.82590806e-02 1.07806551e+00 3.01868450e-02 1.74041718e-01 -3.26635003e-01 -6.10316515e-01 -8.56489062e-01 -3.93895775e-01 7.14836240e-01 -9.19962227e-02 7.60656297e-02 7.03371227e-01 -3.23724031e-01 -3.98090512e-01 -1.86916336e-01 -9.76810932e-01 -2.48188540e-01 2.74442583e-01 5.78001440e-01 6.11472595e-03 -9.43385214e-02 4.56459373e-01 -1.27032721e+00 1.08025527e+00 -5.48174202e-01 -1.92012608e-01 -3.13652873e-01 -5.61577678e-01 -5.62016442e-02 1.00572312e+00 -4.33447987e-01 -4.49539453e-01 -1.74738526e-01 -6.61623418e-01 -1.15184717e-01 -6.83227926e-02 6.97560012e-01 2.77045399e-01 1.10769615e-01 9.05297458e-01 -1.49372607e-01 -1.57534108e-01 -3.99254054e-01 1.64498359e-01 7.69415736e-01 1.32205084e-01 -1.42422736e-01 4.45721775e-01 1.76365003e-01 -4.21719640e-01 -5.76776683e-01 -1.32523036e+00 -2.14266866e-01 -6.68114185e-01 -1.90038875e-01 9.04359519e-01 -6.98937714e-01 -9.61668551e-01 6.02681816e-01 -9.67768013e-01 -6.16577029e-01 -2.99489677e-01 2.39349917e-01 -1.84174642e-01 -6.56193346e-02 -7.45191216e-01 -7.15766728e-01 -5.64354181e-01 -1.13578939e+00 5.24023712e-01 3.18413042e-02 -5.17619610e-01 -1.18203568e+00 3.00510019e-01 2.04096928e-01 9.39865947e-01 4.35181409e-01 6.68289542e-01 -1.23878253e+00 2.84896225e-01 -4.63960886e-01 -8.53089392e-02 7.03505158e-01 -1.24641262e-01 5.58271445e-02 -1.32611716e+00 -1.42271101e-01 1.55541003e-01 -5.85607886e-01 1.17245436e+00 2.90984869e-01 1.23455524e+00 -3.28250468e-01 2.13783324e-01 4.62216407e-01 1.27730024e+00 8.01649541e-02 6.29941523e-01 9.47781205e-01 8.10165644e-01 8.45261931e-01 2.88510412e-01 3.69431041e-02 3.80635172e-01 4.40589607e-01 4.30236071e-01 -3.12156737e-01 2.94514984e-01 -2.75848269e-01 6.41475856e-01 7.53404498e-01 -7.16418475e-02 -2.63813823e-01 -1.03948283e+00 5.14871836e-01 -1.73583019e+00 -1.04622018e+00 -1.82582662e-01 1.90676677e+00 7.90968716e-01 5.92239916e-01 5.87646127e-01 2.89823174e-01 4.60855126e-01 3.04341435e-01 -1.63259909e-01 -1.10654998e+00 -1.41682345e-02 4.22196120e-01 7.00421989e-01 6.08141720e-01 -1.24657643e+00 1.20610321e+00 6.10493040e+00 8.10653448e-01 -1.59671021e+00 -6.25552312e-02 9.98023927e-01 -3.10905963e-01 -1.09361187e-01 -2.55783826e-01 -7.72353113e-01 4.51073557e-01 1.50529981e+00 1.43623382e-01 3.10495406e-01 7.17599809e-01 1.52782932e-01 1.04582541e-01 -8.55552673e-01 5.15253246e-01 1.16639860e-01 -1.03444421e+00 -5.62447347e-02 5.37942760e-02 4.96745259e-01 6.10023737e-01 1.03592165e-01 7.25994587e-01 6.96581423e-01 -1.44029307e+00 8.28327656e-01 2.76247799e-01 2.04399645e-01 -8.97759140e-01 1.20248282e+00 3.91657591e-01 -4.56654280e-01 -2.56025881e-01 -7.56307989e-02 -4.43895638e-01 -1.28919274e-01 5.29533565e-01 -9.11862850e-01 1.05011590e-01 6.96614683e-01 6.68287814e-01 -6.02072299e-01 6.52604640e-01 -2.40972564e-01 8.23968887e-01 -1.97494552e-01 -4.09912884e-01 6.37953162e-01 6.85403123e-02 1.20041788e-01 1.50449026e+00 2.67694928e-02 -3.00078243e-01 5.04171029e-02 4.31297690e-01 -2.88763881e-01 3.69068444e-01 -5.07081509e-01 8.18862617e-02 4.49046828e-02 1.33763587e+00 -8.04041028e-01 -4.21820194e-01 -2.21561000e-01 5.46941996e-01 4.98590738e-01 4.36114743e-02 -5.90350866e-01 -4.89733040e-01 4.61824715e-01 2.40098417e-01 3.02290529e-01 7.44589195e-02 -5.41560829e-01 -1.07053030e+00 -2.77795166e-01 -1.14091456e+00 3.47613245e-01 -4.95114177e-01 -1.19383490e+00 1.06508613e+00 -2.23338723e-01 -8.01638067e-01 -3.87423307e-01 -7.77114749e-01 -7.74216294e-01 7.60882974e-01 -1.52056515e+00 -1.06359839e+00 4.39992845e-02 3.70467693e-01 5.35249293e-01 -1.25440478e-01 6.76026165e-01 3.63004088e-01 -7.26153135e-01 5.66799223e-01 -5.32600507e-02 5.05323172e-01 7.21675634e-01 -1.26387489e+00 5.37794292e-01 6.59764707e-01 3.39011997e-02 7.80351758e-01 9.13891971e-01 -3.01542044e-01 -6.44072115e-01 -1.04844296e+00 1.33488739e+00 -6.00686908e-01 1.13594282e+00 -4.66501892e-01 -8.02538812e-01 8.08831751e-01 5.53567946e-01 -3.83953303e-01 7.84974277e-01 3.53794962e-01 -3.28009844e-01 -4.12852913e-02 -8.84157777e-01 6.70415103e-01 4.20860469e-01 -4.30806488e-01 -6.20051205e-01 1.35711744e-01 4.58986849e-01 -2.03018621e-01 -6.48779452e-01 4.15176392e-01 4.81516510e-01 -1.02335918e+00 5.45865476e-01 -8.22658002e-01 8.90040338e-01 7.49249458e-02 -1.13160655e-01 -1.48139167e+00 -2.07735479e-01 -2.58207738e-01 3.51573706e-01 1.08561957e+00 8.37689281e-01 -5.26096463e-01 7.29424894e-01 3.72063816e-01 -3.97346951e-02 -9.36726630e-01 -4.99334931e-01 -4.13254321e-01 5.36924005e-01 -7.07956135e-01 3.67793024e-01 1.14464033e+00 -9.32383537e-02 5.83408237e-01 -4.38402325e-01 -4.58129674e-01 2.37323418e-01 -1.45017564e-01 7.61775792e-01 -1.22916949e+00 -9.51409787e-02 -9.69778776e-01 -2.23173741e-02 -5.07021785e-01 8.78163427e-02 -9.40932870e-01 -2.91785263e-02 -1.35992432e+00 1.28797427e-01 -4.04061191e-02 -6.68923020e-01 8.76340806e-01 -9.96039882e-02 6.60295427e-01 3.34742248e-01 8.67382437e-02 -5.49883187e-01 1.61499634e-01 8.86771023e-01 3.04795895e-02 -2.86577195e-01 1.25657260e-01 -1.21559978e+00 7.49324977e-01 1.19043875e+00 -5.62776625e-01 -2.87604202e-02 -3.56847584e-01 6.34395480e-01 -5.18560588e-01 2.74681747e-01 -8.16196203e-01 7.80617148e-02 1.31574631e-01 4.91326064e-01 -4.13607001e-01 2.29007229e-01 -5.73811829e-01 -4.10435706e-01 6.76012099e-01 -6.49928033e-01 1.61631480e-01 3.93808424e-01 -1.74144015e-01 -4.11961794e-01 -4.17690933e-01 7.81530917e-01 -3.09015900e-01 -4.04742360e-01 2.58933958e-02 -6.94261432e-01 -5.52188009e-02 5.30535102e-01 -9.19246376e-02 -1.97206527e-01 -4.51604813e-01 -8.37731600e-01 2.58150160e-01 2.24113315e-01 6.15349889e-01 1.40273437e-01 -9.33643579e-01 -8.34883809e-01 2.14775708e-02 -2.56027821e-02 -3.22327524e-01 -7.61043355e-02 8.06310356e-01 -5.17496765e-01 4.78825837e-01 -2.69641608e-01 -1.36606410e-01 -1.17223406e+00 3.64555001e-01 4.58477825e-01 -5.33388972e-01 -1.86766982e-01 8.85438025e-01 -3.19408625e-01 -6.74130619e-01 2.28384554e-01 -2.77399749e-01 -7.48791575e-01 6.67605937e-01 4.76386994e-01 1.75361753e-01 3.94310355e-01 -6.49494171e-01 -2.44819328e-01 1.99682146e-01 -4.45929796e-01 -9.97326300e-02 1.75681841e+00 3.14566761e-01 -1.33794909e-02 8.03050280e-01 1.37114239e+00 1.90738123e-02 -7.63147116e-01 2.50782780e-02 1.44177094e-01 -1.04173198e-01 2.28735358e-01 -8.06496084e-01 -1.08978748e+00 1.03470933e+00 3.36531729e-01 8.04300785e-01 9.09843385e-01 -3.66287053e-01 8.66093397e-01 3.32441568e-01 -1.10679291e-01 -1.25528765e+00 -1.45542398e-01 1.15004969e+00 5.79393923e-01 -1.40785456e+00 -1.11661248e-01 2.51364052e-01 -7.49966502e-01 1.19803059e+00 4.83558059e-01 -4.34509575e-01 7.20645547e-01 6.80447072e-02 4.06045586e-01 -4.45840299e-01 -8.67874384e-01 -3.02609533e-01 2.74304241e-01 9.45951715e-02 8.91702354e-01 -2.61180918e-03 -2.83785045e-01 4.40557837e-01 -8.76023948e-01 1.02378726e-02 6.22054458e-01 6.00918293e-01 -5.84244490e-01 -1.03060555e+00 -9.53852460e-02 5.52223563e-01 -1.02175009e+00 -2.57726222e-01 -7.56675303e-01 1.07671726e+00 1.65941551e-01 8.99709046e-01 4.36155424e-02 -7.67662406e-01 5.70393085e-01 1.13730200e-01 -1.46420756e-02 -6.98144138e-01 -1.40053308e+00 2.57530306e-02 2.35669911e-01 -3.42739165e-01 -2.32129022e-01 -6.11837983e-01 -8.85138452e-01 -6.00476682e-01 -2.24517047e-01 1.81620911e-01 8.69882107e-01 1.08784914e+00 -1.28357140e-02 6.31175160e-01 7.05451131e-01 -8.39881539e-01 -5.77192962e-01 -1.35534835e+00 -2.58193910e-01 2.78787374e-01 5.28715193e-01 -1.86377227e-01 -5.21783888e-01 -2.69984037e-01]
[10.884710311889648, 7.503410339355469]
7d57fc68-ec63-44d1-b24e-0ab789abd64b
class-specific-channel-attention-for-few-shot
2209.01332
null
https://arxiv.org/abs/2209.01332v2
https://arxiv.org/pdf/2209.01332v2.pdf
Class-Specific Channel Attention for Few-Shot Learning
Few-Shot Learning (FSL) has attracted growing attention in computer vision due to its capability in model training without the need for excessive data. FSL is challenging because the training and testing categories (the base vs. novel sets) can be largely diversified. Conventional transfer-based solutions that aim to transfer knowledge learned from large labeled training sets to target testing sets are limited, as critical adverse impacts of the shift in task distribution are not adequately addressed. In this paper, we extend the solution of transfer-based methods by incorporating the concept of metric-learning and channel attention. To better exploit the feature representations extracted by the feature backbone, we propose Class-Specific Channel Attention (CSCA) module, which learns to highlight the discriminative channels in each class by assigning each class one CSCA weight vector. Unlike general attention modules designed to learn global-class features, the CSCA module aims to learn local and class-specific features with very effective computation. We evaluated the performance of the CSCA module on standard benchmarks including miniImagenet, Tiered-ImageNet, CIFAR-FS, and CUB-200-2011. Experiments are performed in inductive and in/cross-domain settings. We achieve new state-of-the-art results.
['Ming-Ching Chang', 'Jun-Wei Hsieh', 'Ying-Yu Chen']
2022-09-03
null
null
null
null
['few-shot-image-classification']
['computer-vision']
[ 3.90870780e-01 -8.11802670e-02 -1.38159797e-01 -5.29128730e-01 -9.11224246e-01 -1.45605341e-01 5.81729829e-01 -1.59576043e-01 -5.58352530e-01 8.27170014e-01 7.22846836e-02 1.18086219e-01 -2.16695175e-01 -7.74548590e-01 -7.74701238e-01 -8.06815803e-01 -2.10072786e-01 2.57443130e-01 4.18867826e-01 -1.19295627e-01 1.10715389e-01 1.78321868e-01 -1.44945180e+00 5.53952575e-01 6.96765602e-01 1.31150258e+00 4.27292615e-01 3.44557196e-01 -1.56923801e-01 7.70079136e-01 -4.67389166e-01 -1.27826825e-01 1.94194734e-01 -7.20465660e-01 -7.48306215e-01 -8.01573973e-03 4.60173011e-01 -1.20833084e-01 -3.35107505e-01 1.01280248e+00 8.00987124e-01 2.69798487e-01 8.26878846e-01 -1.33468187e+00 -9.52895224e-01 3.77551943e-01 -5.56392610e-01 4.62719232e-01 -6.58484772e-02 1.40901268e-01 1.02433491e+00 -1.24564183e+00 6.23002470e-01 1.11223269e+00 5.59423506e-01 8.95172715e-01 -1.21326339e+00 -7.95457184e-01 2.55200565e-01 8.74616325e-01 -1.37506640e+00 -2.26610392e-01 8.32396746e-01 -5.71117282e-01 8.61654282e-01 -1.13504574e-01 3.99103224e-01 1.25021076e+00 -7.45704249e-02 9.66817796e-01 1.13571513e+00 -4.80363071e-01 4.12607729e-01 2.75577903e-01 1.31655678e-01 4.25156474e-01 -7.33139366e-02 1.45814358e-03 -7.13477433e-01 3.03765591e-02 3.93403620e-01 1.85907990e-01 -3.83866340e-01 -6.81186378e-01 -1.04777896e+00 1.05204010e+00 8.15435052e-01 3.07739466e-01 -3.16340744e-01 2.26084813e-02 4.84992266e-01 3.49536568e-01 5.47484279e-01 3.51850063e-01 -6.22244716e-01 8.51713344e-02 -6.10015333e-01 3.11610922e-02 5.54995596e-01 9.04083848e-01 1.06468511e+00 -8.11106935e-02 -6.50176346e-01 1.05442250e+00 3.14932317e-02 3.06264341e-01 8.40548158e-01 -5.02204180e-01 2.74926037e-01 4.29905623e-01 -3.67715716e-01 -5.54821074e-01 -2.37485826e-01 -6.84512913e-01 -6.98509097e-01 1.09717585e-01 1.01649590e-01 -2.63522059e-01 -1.17850399e+00 1.71703649e+00 1.19325913e-01 4.47774082e-01 2.25704759e-02 8.14263999e-01 7.84172952e-01 5.82996607e-01 2.29917601e-01 8.93049613e-02 9.26363349e-01 -1.16191089e+00 -4.06453401e-01 -2.84945518e-01 7.06972241e-01 -4.03645873e-01 1.16159141e+00 -2.95445360e-02 -6.43602610e-01 -6.55528903e-01 -1.05872858e+00 1.32219285e-01 -6.12278581e-01 -2.40769655e-01 4.85704660e-01 3.16855878e-01 -8.45299959e-01 5.87659299e-01 -3.72156501e-01 -3.75100672e-01 1.11342072e+00 1.65290147e-01 -2.83654571e-01 -3.88574779e-01 -1.30333745e+00 7.96694040e-01 3.79278094e-01 -2.71642387e-01 -1.19512892e+00 -9.49245036e-01 -8.79111350e-01 3.78567666e-01 2.07744673e-01 -2.93111831e-01 1.03848207e+00 -1.25253534e+00 -1.36282384e+00 7.35924780e-01 2.61159629e-01 -5.42804420e-01 2.45768785e-01 3.16757639e-03 -1.74196869e-01 1.41514689e-01 1.27044708e-01 9.33816075e-01 1.04122639e+00 -1.05218148e+00 -7.62198508e-01 -3.98079932e-01 -1.42751947e-01 8.41164738e-02 -7.65356243e-01 -1.62316903e-01 -2.40543261e-01 -6.28663063e-01 -4.36274618e-01 -7.59323418e-01 -1.54784009e-01 7.87198469e-02 -6.05557114e-02 -2.65080512e-01 9.85418499e-01 -3.03673595e-01 9.62767243e-01 -2.30426884e+00 9.07307193e-02 4.98429462e-02 2.83456296e-02 6.24192178e-01 -3.33307564e-01 2.22024307e-01 -6.93918467e-02 -3.26231837e-01 -3.50919843e-01 -4.44871485e-02 -1.03802301e-01 2.15806812e-01 -2.64348179e-01 3.26412469e-01 4.66000944e-01 9.86521482e-01 -1.00116336e+00 -3.60846907e-01 2.20007017e-01 4.90976751e-01 -6.27040982e-01 1.31150767e-01 2.71897437e-03 3.73743564e-01 -4.11911309e-01 6.40783727e-01 4.80008066e-01 -2.20774025e-01 -1.43739685e-01 -2.90045366e-02 2.34568641e-01 -1.14212044e-01 -7.30972469e-01 1.89578962e+00 -5.13775349e-01 6.86375439e-01 -3.88647437e-01 -1.37998223e+00 8.95061851e-01 1.75935015e-01 3.70082527e-01 -9.94875729e-01 9.73895192e-02 3.57165039e-02 3.43332916e-01 -3.88250679e-01 -1.90105215e-01 -1.62007138e-01 2.13205926e-02 1.21307284e-01 7.63021767e-01 2.49470413e-01 7.85369575e-02 9.37092751e-02 1.07005227e+00 3.24689411e-02 2.55548537e-01 -3.76214623e-01 3.44225675e-01 -1.51331306e-01 6.52702093e-01 6.83547437e-01 -4.85994607e-01 6.90031946e-01 2.85315335e-01 -3.20350885e-01 -7.04462588e-01 -1.08325970e+00 -3.92385989e-01 1.47219813e+00 5.38734719e-02 -1.14320464e-01 -7.20628202e-01 -8.48039806e-01 7.12211132e-02 5.57420909e-01 -1.04377711e+00 -6.68596089e-01 -1.81054488e-01 -7.51124203e-01 2.52760470e-01 7.73468435e-01 6.57307148e-01 -1.27173090e+00 -8.60393524e-01 2.00703502e-01 2.09136978e-01 -9.48477626e-01 -5.60768843e-01 4.08610404e-01 -7.18861759e-01 -1.23634231e+00 -1.03808343e+00 -1.12726331e+00 7.16401458e-01 3.63599986e-01 8.40418816e-01 -2.84931332e-01 -6.97236300e-01 4.73780066e-01 -5.82005560e-01 -7.67808020e-01 3.23362678e-01 9.44037959e-02 -2.60487467e-01 5.47936261e-01 6.09467506e-01 -4.73263055e-01 -7.37460673e-01 3.52092594e-01 -6.62264645e-01 -2.28205353e-01 7.22237885e-01 1.26343501e+00 4.23830032e-01 -4.49812382e-01 1.01065600e+00 -1.06184125e+00 4.15054440e-01 -7.12699413e-01 -2.29124740e-01 2.16363028e-01 -5.78878760e-01 -7.18983933e-02 5.07303119e-01 -5.11038959e-01 -1.10435569e+00 4.75202985e-02 2.34432071e-01 -6.44388080e-01 -7.83353671e-02 4.46584970e-01 -2.71262467e-01 -2.13670969e-01 9.09846961e-01 2.06228673e-01 -9.96675417e-02 -4.80353832e-01 3.78157139e-01 6.50465846e-01 4.12791431e-01 -3.41880083e-01 6.62780762e-01 5.13512671e-01 -3.34471107e-01 -7.77746856e-01 -9.23493564e-01 -6.77778661e-01 -6.77075386e-01 -2.00045824e-01 9.16256130e-01 -9.70867097e-01 -1.61698297e-01 4.65944558e-01 -8.77062678e-01 -4.90383059e-01 -7.00231612e-01 6.27920628e-01 -5.51294208e-01 -1.63761124e-01 -3.06424856e-01 -3.68399769e-01 -2.93954045e-01 -1.05388749e+00 7.99299240e-01 3.07804465e-01 2.55793761e-02 -1.00695479e+00 1.22861922e-01 -3.47272009e-02 6.46722078e-01 6.66721761e-02 8.92200291e-01 -9.03655589e-01 -3.03925335e-01 -1.76810756e-01 -3.92392755e-01 5.39148092e-01 5.36406413e-02 -5.97534537e-01 -1.37072706e+00 -4.61466372e-01 -2.72442997e-01 -8.18668783e-01 1.35904288e+00 4.45899040e-01 1.45323384e+00 1.63891777e-01 -2.96747386e-01 7.84711599e-01 1.34270012e+00 2.51193315e-01 5.83698571e-01 3.34505022e-01 6.33820176e-01 5.54811954e-01 5.50602138e-01 4.77526635e-01 8.82948413e-02 5.35501361e-01 3.91954213e-01 -6.95635825e-02 -4.04897392e-01 -1.21244706e-01 1.83901072e-01 4.87946570e-01 -6.35305867e-02 1.40818268e-01 -8.20324779e-01 7.98355460e-01 -1.86150372e+00 -8.60757351e-01 5.00658214e-01 2.08587694e+00 7.81359911e-01 1.00497358e-01 -1.21712074e-01 5.93467429e-03 7.42018878e-01 1.69517741e-01 -7.81297147e-01 -1.88375846e-01 2.98398212e-02 4.67325866e-01 5.19467056e-01 1.10030346e-01 -1.23559499e+00 8.91380906e-01 5.41727781e+00 1.12946522e+00 -1.21395779e+00 4.36434180e-01 6.38106287e-01 -2.83094466e-01 2.32432727e-02 -2.03904644e-01 -7.99649358e-01 4.14980054e-01 8.47427309e-01 -1.80089116e-01 2.57232547e-01 1.06584847e+00 -3.80096585e-01 1.03929386e-01 -1.19074249e+00 1.19857383e+00 3.41440976e-01 -1.30832958e+00 1.80107057e-02 -6.00921027e-02 9.53593373e-01 4.29460019e-01 2.93340743e-01 8.77334774e-01 -6.11427985e-02 -9.42581832e-01 4.69563097e-01 4.65196341e-01 9.54279363e-01 -5.99861085e-01 6.52699232e-01 1.70932934e-01 -9.17878389e-01 -3.70644689e-01 -7.50692248e-01 -1.15009703e-01 -2.73431718e-01 2.36485645e-01 -7.00009108e-01 1.45614713e-01 1.00578976e+00 9.56751347e-01 -5.61491132e-01 1.31743681e+00 -1.18738681e-01 6.94370687e-01 1.97305813e-01 -5.45608066e-02 4.43927675e-01 6.48335069e-02 1.97055727e-01 1.17824900e+00 1.48233801e-01 -1.30179062e-01 1.95950881e-01 6.52366161e-01 -3.55073839e-01 4.77999337e-02 -6.96679771e-01 1.88595012e-01 1.49951547e-01 1.23073030e+00 -5.24293959e-01 -3.98273677e-01 -8.24561536e-01 9.57211077e-01 5.20514309e-01 4.74140644e-01 -7.35022604e-01 -6.72926247e-01 6.08387113e-01 4.84931543e-02 7.96696424e-01 2.44864732e-01 -1.81456991e-02 -1.03148329e+00 -1.95184857e-01 -5.56206048e-01 5.05482137e-01 -4.00266737e-01 -1.52582085e+00 6.10599697e-01 -1.80589352e-02 -1.28558350e+00 1.85961835e-02 -7.51574159e-01 -8.50954413e-01 8.75731707e-01 -1.92304838e+00 -1.18134511e+00 -3.71176034e-01 9.92555797e-01 8.87762785e-01 -6.01625502e-01 9.00664926e-01 5.13941944e-01 -3.22087049e-01 7.87019134e-01 3.73797923e-01 3.35850060e-01 1.01031995e+00 -1.12887847e+00 5.98550774e-03 3.89825046e-01 2.11679757e-01 1.49098679e-01 2.87850767e-01 -2.36925498e-01 -1.00509858e+00 -1.46864235e+00 6.02246940e-01 -1.82913885e-01 5.97889900e-01 -4.60032374e-01 -1.04178393e+00 6.20177627e-01 1.51076332e-01 7.84800410e-01 8.96364391e-01 7.59599432e-02 -6.30477786e-01 -3.26361060e-01 -8.77176583e-01 7.17211589e-02 1.06011772e+00 -6.12572730e-01 -5.07328928e-01 2.50353545e-01 5.93898237e-01 5.57837896e-02 -4.85467076e-01 3.06906849e-01 3.99118781e-01 -7.34203756e-01 8.62040222e-01 -9.47877526e-01 4.22006071e-01 7.34993741e-02 -3.86918932e-01 -1.71167839e+00 -6.00622237e-01 -1.08271815e-01 -8.20382871e-03 9.77007031e-01 4.71771300e-01 -4.99529749e-01 6.03564620e-01 2.80055076e-01 -3.70540887e-01 -9.30395842e-01 -9.90716100e-01 -8.40262711e-01 7.65040591e-02 -2.09434718e-01 3.13591480e-01 1.05334044e+00 -3.84891825e-03 5.83579421e-01 -3.46698672e-01 -2.45794326e-01 6.20498240e-01 1.06079832e-01 4.53367889e-01 -1.24825168e+00 -4.55985099e-01 -3.50428700e-01 -6.56146288e-01 -4.11318630e-01 2.29908600e-01 -1.15426886e+00 1.58949271e-01 -1.23470795e+00 4.68604952e-01 -3.27714086e-01 -9.46044803e-01 5.91418624e-01 -1.35571092e-01 4.28886920e-01 4.21380699e-01 4.46069874e-02 -9.04397368e-01 9.10516918e-01 1.21205592e+00 -4.30821806e-01 -2.57371590e-02 5.25602326e-03 -5.54968417e-01 6.89221203e-01 7.41465747e-01 -4.38384980e-01 -6.52137220e-01 -3.89121801e-01 -2.52436399e-01 -5.05642891e-01 3.75823379e-01 -1.16065943e+00 1.72038004e-01 8.44810307e-02 5.99384069e-01 -2.00462878e-01 2.77170122e-01 -7.05897868e-01 -6.29238069e-01 5.09523630e-01 -5.71981549e-01 -6.31992340e-01 6.50878102e-02 7.77161002e-01 -3.54406059e-01 -1.95521966e-01 1.06748700e+00 -1.02415964e-01 -1.22379613e+00 7.24081457e-01 -4.88081425e-02 2.99844474e-01 1.38205087e+00 -1.23637125e-01 -2.68923432e-01 -8.07147697e-02 -8.05059254e-01 3.05091709e-01 -4.01367731e-02 6.66767418e-01 7.93875456e-01 -1.44746268e+00 -6.11140847e-01 3.29807580e-01 6.92965686e-01 -1.33005947e-01 4.75983441e-01 6.69065714e-01 -8.79883692e-02 4.08577770e-01 -6.38749897e-01 -5.30580521e-01 -8.04794371e-01 7.40821183e-01 2.89701611e-01 -2.37393137e-02 -4.61607546e-01 1.11540270e+00 6.82582438e-01 -3.62077445e-01 3.62465203e-01 -8.29063822e-03 -3.34997237e-01 2.50744730e-01 8.27434123e-01 3.68766129e-01 8.55657980e-02 -4.32576358e-01 -3.20640802e-01 4.27664608e-01 -2.67490178e-01 2.48605311e-01 1.50297558e+00 5.07472306e-02 3.82406116e-01 4.51187491e-01 1.55048954e+00 -6.86031640e-01 -1.64475822e+00 -7.34714210e-01 3.11577227e-02 -3.59388560e-01 1.84555382e-01 -7.84801543e-01 -1.17135799e+00 1.33756530e+00 1.00029755e+00 -2.17088297e-01 1.09426844e+00 2.17753381e-01 7.18087316e-01 2.63840258e-01 3.89314175e-01 -1.34602606e+00 2.69383669e-01 5.39322078e-01 8.07566524e-01 -1.52168524e+00 -5.06446004e-01 4.55416180e-03 -7.41291881e-01 8.40146542e-01 8.83009136e-01 -1.97100505e-01 9.61247563e-01 -1.02111869e-01 -1.34844229e-01 -5.68578504e-02 -7.24988341e-01 -4.34135765e-01 4.27409917e-01 9.11999106e-01 2.20040977e-01 7.63107017e-02 -4.67277542e-02 7.57247388e-01 4.05848444e-01 1.72465425e-02 1.15687415e-01 1.00828362e+00 -6.19247854e-01 -8.45956981e-01 -6.56464770e-02 7.14420795e-01 -2.74976462e-01 -1.85244501e-01 -1.45052761e-01 6.53371334e-01 2.40319267e-01 5.58414161e-01 2.46697277e-01 -3.85553151e-01 2.94213682e-01 2.03043446e-01 4.22888726e-01 -8.95313382e-01 -2.40436569e-01 -8.05277303e-02 -3.96870553e-01 -5.71139753e-01 -3.56416255e-01 -4.81321871e-01 -1.11907911e+00 3.18971395e-01 -1.85144752e-01 1.03426650e-01 5.23745835e-01 9.17924821e-01 4.94057447e-01 8.45490098e-01 6.50977910e-01 -9.24880803e-01 -7.20835686e-01 -1.18391943e+00 -8.02793920e-01 5.72221220e-01 2.58230448e-01 -9.52577651e-01 -1.87011734e-01 1.11184716e-02]
[9.9443941116333, 2.7743747234344482]
729c100d-c9c4-4645-9e99-e44203205516
towards-a-robust-detection-of-language-model
2306.05871
null
https://arxiv.org/abs/2306.05871v1
https://arxiv.org/pdf/2306.05871v1.pdf
Towards a Robust Detection of Language Model Generated Text: Is ChatGPT that Easy to Detect?
Recent advances in natural language processing (NLP) have led to the development of large language models (LLMs) such as ChatGPT. This paper proposes a methodology for developing and evaluating ChatGPT detectors for French text, with a focus on investigating their robustness on out-of-domain data and against common attack schemes. The proposed method involves translating an English dataset into French and training a classifier on the translated data. Results show that the detectors can effectively detect ChatGPT-generated text, with a degree of robustness against basic attack techniques in in-domain settings. However, vulnerabilities are evident in out-of-domain contexts, highlighting the challenge of detecting adversarial text. The study emphasizes caution when applying in-domain testing results to a wider variety of content. We provide our translated datasets and models as open-source resources. https://gitlab.inria.fr/wantoun/robust-chatgpt-detection
['Djamé Seddah', 'Benoît Sagot', 'Virginie Mouilleron', 'Wissam Antoun']
2023-06-09
null
null
null
null
['adversarial-text']
['adversarial']
[ 1.10141234e-02 -6.61815777e-02 1.21171132e-01 -1.02873534e-01 -1.33988166e+00 -1.15980101e+00 9.92405653e-01 4.32918280e-01 -3.41930568e-01 5.66929221e-01 2.35543381e-02 -6.92784190e-01 2.52727956e-01 -8.40424061e-01 -6.50466859e-01 -3.25907588e-01 4.30493690e-02 5.43981016e-01 3.75295371e-01 -3.62105578e-01 3.11950237e-01 3.65146130e-01 -6.09533310e-01 7.44680226e-01 8.57588232e-01 2.89633363e-01 -1.08401105e-01 1.07082856e+00 -1.28270030e-01 6.86346710e-01 -1.16924477e+00 -9.83579636e-01 4.82492954e-01 -3.87207299e-01 -9.81983185e-01 -3.13482791e-01 3.91849250e-01 -8.75139162e-02 -4.10947859e-01 1.07738829e+00 8.05898726e-01 -1.91779613e-01 6.14284217e-01 -1.31614792e+00 -6.76731408e-01 9.06405747e-01 -1.98036268e-01 4.53759551e-01 9.18908954e-01 4.32917297e-01 7.80895889e-01 -6.49349153e-01 7.30265737e-01 1.37263560e+00 6.62117541e-01 4.64157552e-01 -1.03890431e+00 -6.18319154e-01 -4.19491351e-01 -6.88814521e-02 -1.12979400e+00 -2.96994507e-01 3.71348798e-01 -3.47559631e-01 1.10924029e+00 2.61642665e-01 -2.81330049e-01 1.82792044e+00 4.08951789e-01 8.37677956e-01 1.21764851e+00 -7.17825770e-01 5.85959628e-02 5.80146253e-01 -6.78419322e-02 4.51430708e-01 -4.26895507e-02 9.63060483e-02 -2.63811320e-01 -6.64523900e-01 2.27915078e-01 -5.58710933e-01 7.24329501e-02 2.65130341e-01 -8.90485704e-01 1.23100686e+00 -1.94352061e-01 7.17434168e-01 -2.66653523e-02 -4.84120339e-01 9.45361197e-01 7.32949138e-01 6.85635746e-01 5.29930651e-01 -5.32922447e-01 -3.41623187e-01 -9.23051775e-01 2.20658630e-01 1.27994239e+00 1.18950140e+00 2.13466108e-01 -1.51994377e-01 -2.84975469e-02 6.90862477e-01 9.16746184e-02 5.49510956e-01 5.14066339e-01 -3.28892946e-01 1.23886001e+00 3.00341696e-01 -2.37242997e-01 -8.98470461e-01 -1.08042665e-01 -5.78875747e-03 -4.18315351e-01 -1.31642655e-01 7.53232837e-01 -5.61380684e-01 -3.80141109e-01 1.38765693e+00 2.07713276e-01 -2.08745211e-01 2.54545629e-01 4.63977426e-01 2.89542526e-01 7.17369139e-01 2.43197307e-01 1.88809961e-01 1.43807995e+00 -4.35106188e-01 -3.98953915e-01 -2.44729862e-01 1.13585615e+00 -1.21019828e+00 1.15951908e+00 4.37279254e-01 -6.89151764e-01 -2.47571632e-01 -7.56913245e-01 1.21021092e-01 -6.73188508e-01 -7.08595440e-02 4.93289530e-02 1.40741694e+00 -7.26103485e-01 3.86429846e-01 -3.84433240e-01 -7.20731616e-01 1.65477470e-01 8.05440024e-02 -4.99253035e-01 2.99987998e-02 -1.72896969e+00 1.18724084e+00 3.76987845e-01 -3.07055324e-01 -1.01319420e+00 -4.16049451e-01 -6.66883230e-01 -2.33693376e-01 4.37601209e-01 1.23076029e-01 1.52722490e+00 -6.94301307e-01 -1.26067042e+00 8.63296270e-01 1.05229728e-01 -6.98444664e-01 1.05143034e+00 -4.49084282e-01 -8.64977717e-01 2.84049660e-01 2.92562902e-01 -6.34922087e-02 8.66541088e-01 -7.02196836e-01 -3.90205771e-01 -6.85870424e-02 1.18137047e-01 -2.59306699e-01 -4.67145652e-01 9.44318950e-01 -7.76256993e-02 -7.72222638e-01 -7.01128483e-01 -8.31429183e-01 -2.51664035e-02 -4.75946844e-01 -7.54698575e-01 -1.92370161e-01 1.09940410e+00 -1.00350451e+00 1.16422558e+00 -1.99392259e+00 -5.21512032e-01 2.51109153e-01 -2.21515998e-01 7.31088758e-01 -4.16566879e-01 1.13154054e+00 -3.07891648e-02 5.27883410e-01 -1.38160571e-01 -1.77047327e-01 6.47653565e-02 -3.58749405e-02 -5.78705728e-01 4.99949753e-01 5.50780594e-01 7.85857499e-01 -7.59809017e-01 -5.24364233e-01 7.16063082e-02 1.18427508e-01 -4.30254132e-01 1.55555546e-01 -3.93883109e-01 3.30199987e-01 -6.75156891e-01 5.50609350e-01 6.32397771e-01 3.41883212e-01 2.66648024e-01 5.33174574e-01 -2.87449714e-02 6.34831071e-01 -1.01824486e+00 1.14470041e+00 -4.05720562e-01 7.86579013e-01 1.35034010e-01 -7.85037875e-01 1.06473422e+00 5.52177370e-01 -7.55702108e-02 -6.06406510e-01 4.20524389e-01 2.30998456e-01 5.61748482e-02 -6.82287633e-01 4.04077172e-01 1.28989667e-01 -5.30045986e-01 8.69748771e-01 3.00159514e-01 2.26668129e-03 4.74276721e-01 4.98625368e-01 1.29477930e+00 -2.29399174e-01 1.45838127e-01 -2.17890844e-01 8.35649073e-01 4.73437039e-03 -2.72561647e-02 9.79813159e-01 -3.58243883e-01 2.23332822e-01 8.46298218e-01 -2.44689360e-02 -1.17717874e+00 -8.03645790e-01 -4.14015949e-02 1.08130550e+00 -5.29180050e-01 -5.32403111e-01 -1.02924967e+00 -1.23776734e+00 -8.97482410e-02 8.91992927e-01 -3.31059963e-01 -1.34560347e-01 -7.41847217e-01 -6.23300076e-01 1.58843279e+00 1.32175863e-01 4.43938404e-01 -1.15526915e+00 -7.22914189e-02 2.85156101e-01 -3.18909317e-01 -1.50253260e+00 -5.02159238e-01 7.06614926e-02 -4.45276618e-01 -1.24247169e+00 -5.07745981e-01 -8.22141349e-01 3.12999159e-01 -2.94959515e-01 9.65748310e-01 -1.06576987e-01 -4.58631992e-01 3.89877468e-01 -7.26932645e-01 -3.84943396e-01 -1.54080749e+00 3.59274477e-01 1.59330573e-02 -1.96999058e-01 7.18096614e-01 -7.06595331e-02 1.41861945e-01 4.98197168e-01 -9.84584689e-01 -8.41376245e-01 4.47305620e-01 4.46330488e-01 -2.85665631e-01 -7.29452120e-04 5.46445549e-01 -1.27816176e+00 1.28878498e+00 -6.96998894e-01 -5.28143287e-01 3.05492342e-01 -4.60475385e-02 -3.76742445e-02 9.58214343e-01 -8.00497055e-01 -1.07378829e+00 -3.35072458e-01 -6.24415398e-01 2.18808316e-02 -6.02297664e-01 3.79242867e-01 -4.10087347e-01 -1.41636938e-01 1.15866482e+00 2.42850423e-01 -1.92715943e-01 -5.49561799e-01 3.44549865e-01 9.94976044e-01 5.33678979e-02 -7.47050524e-01 1.14016974e+00 -1.51472256e-01 -6.19179070e-01 -1.09883916e+00 -2.82976210e-01 -4.46614921e-01 -6.54525220e-01 -2.88788266e-02 4.88128811e-01 -7.39266694e-01 -3.63847882e-01 7.32213557e-01 -1.06556499e+00 -4.25373942e-01 2.46040940e-01 2.49084353e-01 -3.05696517e-01 9.10576701e-01 -1.16386759e+00 -7.24465966e-01 -5.08618355e-01 -1.00663483e+00 7.77316809e-01 -2.55367547e-01 -4.59887803e-01 -1.07099831e+00 3.32247943e-01 5.52972615e-01 2.07010061e-01 -3.62933427e-02 8.01261604e-01 -1.47549200e+00 -1.19071655e-01 -7.02986002e-01 9.04173106e-02 5.00261664e-01 -1.56622455e-01 1.41519859e-01 -9.98331785e-01 -3.59022468e-01 1.90344423e-01 -4.92720991e-01 1.85444161e-01 -3.42335522e-01 3.97844940e-01 -6.33272707e-01 -4.70870510e-02 3.32147479e-02 1.28136933e+00 4.33698706e-02 6.22697592e-01 5.47373295e-01 2.25795016e-01 5.73042154e-01 6.94939494e-01 3.17348480e-01 -1.44501582e-01 4.69586372e-01 1.11185312e-01 2.43775085e-01 3.78897071e-01 -3.93996000e-01 8.79872143e-01 2.35750273e-01 4.68119204e-01 -7.36158907e-01 -1.13627708e+00 4.15821612e-01 -1.32484615e+00 -1.02059889e+00 -3.50308955e-01 1.69658518e+00 8.47050190e-01 4.35703933e-01 5.03248990e-01 2.19281808e-01 8.41515899e-01 -9.49173234e-03 -9.00058821e-03 -1.11573720e+00 -2.24918783e-01 2.89595455e-01 5.37332177e-01 5.40121734e-01 -1.30071115e+00 1.12411606e+00 6.12934875e+00 9.92199659e-01 -1.09400618e+00 2.45252535e-01 4.65822369e-01 2.84308523e-01 2.46800870e-01 -1.08813785e-01 -9.95150089e-01 5.57907045e-01 1.63812065e+00 -3.23804587e-01 9.50743407e-02 8.75567794e-01 2.62656510e-01 2.12226689e-01 -7.61832118e-01 2.00656936e-01 2.00458750e-01 -9.71058011e-01 -1.00016363e-01 2.69385371e-02 2.21474171e-01 3.38738412e-01 -4.85374639e-03 5.98622739e-01 6.09987259e-01 -6.94248497e-01 6.66562200e-01 -3.18401992e-01 7.10200191e-01 -7.43534565e-01 8.97047937e-01 7.80228376e-01 -7.42377579e-01 -1.07508764e-01 -3.78477901e-01 1.63944691e-01 5.51661253e-02 2.94376165e-01 -1.39864624e+00 6.16339862e-01 4.74955857e-01 3.21060956e-01 -8.16326499e-01 7.77510107e-01 -4.16407377e-01 9.40632880e-01 -5.20167768e-01 -3.90464991e-01 3.58153820e-01 1.20900637e-02 6.34130776e-01 1.86957073e+00 2.90841818e-01 -3.66123110e-01 1.55681312e-01 6.81461573e-01 -8.65781158e-02 4.37965423e-01 -7.19970524e-01 -4.79913980e-01 3.87620091e-01 1.11927617e+00 -5.83582759e-01 -1.75593436e-01 -5.06503582e-01 9.44047928e-01 1.66157171e-01 7.95332268e-02 -8.37459445e-01 -5.04680157e-01 4.54703182e-01 2.22217262e-01 1.20548181e-01 -2.58510202e-01 -4.57619354e-02 -1.17061865e+00 5.28047197e-02 -1.59474611e+00 6.23738527e-01 -2.98261702e-01 -1.50869262e+00 7.43962765e-01 5.12490161e-02 -1.08495736e+00 -3.52757543e-01 -8.22106779e-01 -8.96895170e-01 8.93533051e-01 -9.46287036e-01 -1.22983110e+00 4.79613066e-01 8.79444003e-01 5.68784773e-01 -3.95499080e-01 9.39943433e-01 4.79207456e-01 -5.14368951e-01 1.07036507e+00 1.60887927e-01 1.00440383e+00 8.79065156e-01 -9.69378531e-01 9.94293034e-01 1.28195167e+00 2.26372838e-01 6.15099013e-01 7.95045078e-01 -7.59065688e-01 -1.09278643e+00 -1.02001357e+00 1.21518350e+00 -8.22108924e-01 1.28272104e+00 -1.13763094e+00 -9.64102328e-01 7.16288745e-01 3.86900872e-01 -5.94155550e-01 5.95409095e-01 -1.38743207e-01 -7.40919650e-01 5.61778307e-01 -1.48380327e+00 4.72811759e-01 4.81618583e-01 -1.01069224e+00 -6.97537422e-01 7.12537646e-01 4.42835510e-01 -2.30227575e-01 -7.77873456e-01 -2.55848676e-01 1.11917667e-01 -7.07344234e-01 6.88394785e-01 -7.22925186e-01 2.02890977e-01 1.29316881e-01 1.09386131e-01 -1.09067428e+00 1.26964718e-01 -1.02659631e+00 1.37733519e-01 1.55230319e+00 6.32103503e-01 -9.77557898e-01 6.63697958e-01 4.34823424e-01 2.08631188e-01 5.40070869e-02 -1.04742002e+00 -9.53738034e-01 6.08795702e-01 -6.66742563e-01 1.89561799e-01 1.00488722e+00 4.04733777e-01 2.40251139e-01 -2.95577645e-01 4.42476183e-01 3.36511731e-01 -5.15555739e-01 8.90388310e-01 -8.60541224e-01 -2.85185009e-01 2.79823951e-02 -4.24680889e-01 -5.43043077e-01 2.06205517e-01 -8.28663111e-01 -1.80546701e-01 -8.17777514e-01 -2.09143966e-01 -1.78905115e-01 2.94844836e-01 5.39273977e-01 1.39030609e-02 2.63748795e-01 4.57180709e-01 -8.45768973e-02 -2.60795832e-01 -3.66357751e-02 6.35715127e-01 -1.17440477e-01 2.34539866e-01 2.91023016e-01 -4.91730988e-01 4.99345124e-01 1.34039593e+00 -1.01362872e+00 -9.09857452e-02 -1.47681653e-01 2.20674053e-02 -1.85619816e-01 1.79481909e-01 -9.96077240e-01 3.23320031e-02 5.61220646e-02 7.33161792e-02 -2.74426073e-01 4.65611964e-02 -7.18043566e-01 -4.03415710e-01 5.82508624e-01 -4.10129249e-01 3.70456189e-01 4.89948243e-01 3.87250245e-01 -8.81661400e-02 -5.17571092e-01 7.91835904e-01 -1.87239528e-01 -4.45199609e-01 6.90286085e-02 -9.73680735e-01 4.95213807e-01 1.14183700e+00 6.55467063e-02 -6.71285987e-01 -4.40258831e-01 -6.24039710e-01 7.47843087e-02 4.13934261e-01 7.49307096e-01 2.54654348e-01 -5.80565691e-01 -1.06914377e+00 4.82866764e-01 2.54764464e-02 -8.36089313e-01 7.55390227e-02 5.00964761e-01 -7.85283685e-01 4.73846823e-01 -9.78114307e-02 -3.26768517e-01 -1.71392190e+00 7.03267157e-01 2.27091298e-01 -5.89380741e-01 -5.64260840e-01 6.31923020e-01 -4.63653088e-01 -8.18176270e-01 4.02780920e-02 3.65518183e-02 9.37180594e-02 -8.21426883e-02 6.12006843e-01 2.19461307e-01 2.10363239e-01 -7.38461673e-01 -3.12061638e-01 -5.46522439e-02 -3.21396142e-01 -2.42295340e-01 9.53727663e-01 7.28692338e-02 2.55519114e-02 -5.77873215e-02 1.32621682e+00 5.44435203e-01 -4.86671537e-01 -1.59785882e-01 4.42546934e-01 -3.19890171e-01 -5.94989181e-01 -9.35059071e-01 -4.83731747e-01 9.42277431e-01 3.31816912e-01 5.40822923e-01 6.27762496e-01 -5.13232686e-03 8.46372008e-01 5.98271012e-01 4.52139109e-01 -1.03511155e+00 1.01162186e-02 1.02816296e+00 7.22549140e-01 -9.53161001e-01 -3.16493064e-01 -4.74842936e-01 -8.48568678e-01 1.15086210e+00 4.13112670e-01 -2.12341934e-01 6.04315221e-01 5.50072372e-01 5.23063004e-01 2.18110532e-01 -6.09281182e-01 3.08712095e-01 -2.38409668e-01 8.68169785e-01 3.75364393e-01 -1.21933006e-01 -3.19997311e-01 3.20078462e-01 -3.93447489e-01 -2.79810876e-01 8.06767344e-01 1.12535024e+00 -9.01869833e-02 -1.57686758e+00 -7.43225276e-01 2.60486871e-01 -1.05907834e+00 -4.04981852e-01 -9.51841772e-01 1.06642318e+00 -2.47123644e-01 1.36186278e+00 -4.10750002e-01 -4.42143738e-01 3.31454158e-01 4.76032406e-01 3.25026847e-02 -7.07752705e-01 -1.26862323e+00 6.46597892e-02 6.30348980e-01 -2.03229025e-01 7.41826743e-02 -7.38722503e-01 -7.69140244e-01 -6.69074118e-01 -3.41441542e-01 5.29651463e-01 4.52542305e-01 8.25489819e-01 3.70347559e-01 -1.11096375e-01 4.60167915e-01 -3.41539979e-01 -8.44902635e-01 -1.30471265e+00 -4.85330135e-01 5.73292017e-01 -1.39630288e-02 5.83221503e-02 -6.18108273e-01 -1.32331967e-01]
[6.0574493408203125, 8.075825691223145]
e933abff-c959-4eca-b135-19848ae45dd4
sar-to-optical-image-synthesis-for-cloud
null
null
https://www.isprs-ann-photogramm-remote-sens-spatial-inf-sci.net/IV-1/5/2018/
https://www.isprs-ann-photogramm-remote-sens-spatial-inf-sci.net/IV-1/5/2018/
SAR TO OPTICAL IMAGE SYNTHESIS FOR CLOUD REMOVAL WITH GENERATIVE ADVERSARIAL NETWORKS
Optical imagery is often affected by the presence of clouds. Aiming to reduce their effects, different reconstruction techniques have been proposed in the last years. A common alternative is to extract data from active sensors, like Synthetic Aperture Radar (SAR), because they are almost independent on the atmospheric conditions and solar illumination. On the other hand, SAR images are more complex to interpret than optical images requiring particular handling. Recently, Conditional Generative Adversarial Networks (cGANs) have been widely used in different image generation tasks presenting state-of-the-art results. One application of cGANs is learning a nonlinear mapping function from two images of different domains. In this work, we combine the fact that SAR images are hardly affected by clouds with the ability of cGANS for image translation in order to map optical images from SAR ones so as to recover regions that are covered by clouds. Experimental results indicate that the proposed solution achieves better classification accuracy than SAR based classification.
['R. Q. Feitosa', 'D. A. B. Oliveira', 'P. N. Happ', 'J. D. Bermudez']
2018-09-26
null
null
null
isprs-annals-of-the-photogrammetry-remote-2
['cloud-removal']
['computer-vision']
[ 7.34571218e-01 -1.34831756e-01 2.41058677e-01 -3.32846820e-01 -5.21372080e-01 -5.40075660e-01 7.79678583e-01 -1.74382716e-01 -3.78596187e-01 1.13489342e+00 -3.35251510e-01 1.11480594e-01 -1.60573319e-01 -1.04601467e+00 -6.81836665e-01 -1.09536135e+00 2.48268589e-01 4.17716742e-01 -1.86636373e-02 -2.52784073e-01 -8.35612603e-03 9.75418746e-01 -1.63821101e+00 7.00984383e-03 1.34345078e+00 8.10124099e-01 4.85149980e-01 2.77621657e-01 -5.41921705e-02 5.78377962e-01 -5.16357660e-01 5.70318215e-02 5.35805523e-01 -4.32998002e-01 -1.87221706e-01 4.36704844e-01 2.68267453e-01 -2.23876625e-01 -1.18378170e-01 1.36486173e+00 2.11223021e-01 -2.71670464e-02 8.08356047e-01 -8.22057247e-01 -4.49666440e-01 3.03234249e-01 -2.80629575e-01 -1.29759774e-01 -2.14325003e-02 1.64334565e-01 3.67712051e-01 -5.61438859e-01 4.66853827e-01 9.73233163e-01 1.89199030e-01 2.14780584e-01 -1.37202692e+00 -4.24142450e-01 -1.31442174e-01 1.07452109e-01 -1.26042533e+00 -2.54491568e-01 9.39223349e-01 -5.26980758e-01 1.94048196e-01 4.67625439e-01 4.52046752e-01 1.03776634e+00 1.86596602e-01 3.06600392e-01 1.80093074e+00 -4.33477014e-01 2.64062911e-01 5.11688650e-01 -2.85552472e-01 2.02523604e-01 5.89965880e-01 1.52062982e-01 5.03909811e-02 3.78551036e-02 4.88906026e-01 1.57267883e-01 -7.51393795e-01 -3.75986189e-01 -9.94830191e-01 9.17802989e-01 7.79078364e-01 5.83968341e-01 -6.05295360e-01 -3.02989632e-01 -2.77287871e-01 1.82960495e-01 3.63564998e-01 5.54121017e-01 -1.85672578e-03 6.48212135e-01 -9.52136219e-01 2.81169176e-01 4.80640501e-01 3.92289340e-01 9.21903908e-01 5.65869570e-01 1.20233223e-01 4.78791416e-01 6.75892457e-02 9.81746852e-01 5.07572472e-01 -3.82611632e-01 2.16116890e-01 4.70661283e-01 2.52676278e-01 -1.11961734e+00 -8.91276523e-02 -8.53346586e-01 -1.31685925e+00 5.90587616e-01 1.76777482e-01 -9.29763988e-02 -9.90161538e-01 1.30898833e+00 7.42799230e-03 1.19354054e-01 4.95365262e-01 1.04994285e+00 4.17503893e-01 8.00678134e-01 -1.41797915e-01 -3.16850036e-01 9.06248868e-01 -2.92691678e-01 -6.50470138e-01 -6.27327383e-01 7.01173991e-02 -8.68069291e-01 6.15218103e-01 5.25899172e-01 -5.08745074e-01 -7.20269799e-01 -1.20288491e+00 5.30688524e-01 -5.31943262e-01 3.27509731e-01 4.71454412e-01 5.53739667e-01 -6.63812220e-01 5.33675730e-01 -6.59229100e-01 -2.04067335e-01 2.51048356e-01 -2.18981202e-03 -3.46130520e-01 -1.41169384e-01 -1.00542021e+00 1.03159857e+00 5.68294644e-01 3.85793656e-01 -7.32761979e-01 -3.11690897e-01 -7.09727824e-01 -7.29584768e-02 2.68278956e-01 -2.88987249e-01 3.97490442e-01 -1.38845885e+00 -1.09510958e+00 7.79132485e-01 2.54502296e-01 -8.01593840e-01 5.57920754e-01 -1.95168942e-01 -6.07490063e-01 2.76637226e-01 -1.29641145e-01 2.63626069e-01 1.19929743e+00 -1.54602063e+00 -2.00874776e-01 -5.01392186e-01 -1.46787643e-01 1.47628970e-02 -1.04721412e-01 -3.25349748e-01 3.10946375e-01 -6.49972379e-01 2.25824922e-01 -1.07028997e+00 -3.11170369e-01 -1.71070620e-01 -5.47357261e-01 6.51835084e-01 1.05911446e+00 -5.53056359e-01 4.15130407e-01 -1.87214804e+00 3.65324646e-01 1.24793828e-01 -3.85583937e-01 6.03678465e-01 -1.21213876e-01 2.84673780e-01 -1.04818307e-01 -1.01987936e-01 -7.59878755e-01 -4.60916907e-02 -4.73990500e-01 2.83829123e-01 -6.74800992e-01 6.08699620e-01 4.79098737e-01 6.33525252e-01 -3.37484568e-01 -2.57552236e-01 4.03298467e-01 6.42365575e-01 -5.29650450e-02 4.00566459e-01 -5.43457448e-01 1.14808023e+00 -5.56867421e-01 4.44268078e-01 9.99461055e-01 1.00583345e-01 1.91780731e-01 -2.66766101e-01 -1.39864385e-01 -4.11050409e-01 -9.63106096e-01 1.19122171e+00 -3.86091650e-01 4.68009591e-01 7.02883229e-02 -1.30896652e+00 1.31138742e+00 2.29678020e-01 2.68294334e-01 -8.03666770e-01 1.53668568e-01 3.20517510e-01 3.59238237e-02 -4.73403901e-01 3.94375443e-01 -3.31961274e-01 2.25320190e-01 3.77462246e-02 -3.22716475e-01 -6.90499783e-01 -5.93319237e-02 -3.66995305e-01 3.96117210e-01 2.44188175e-01 3.34715486e-01 5.33260033e-02 7.36689985e-01 7.34830722e-02 5.29936850e-01 3.93489808e-01 4.51831222e-01 7.24303961e-01 -9.44298357e-02 -4.47883517e-01 -8.87058198e-01 -9.41274166e-01 -2.84394771e-01 -2.29465201e-01 -6.97633624e-02 6.57888770e-01 -5.97490013e-01 -3.68790329e-01 -2.48349965e-01 7.23884404e-01 -3.89417976e-01 5.22666723e-02 -5.51300108e-01 -1.00496781e+00 2.73040682e-01 -9.41905677e-02 8.88559759e-01 -1.00049937e+00 -7.38073587e-01 2.61282951e-01 -3.02476674e-01 -1.50442564e+00 4.33133096e-01 -7.56006166e-02 -9.89337325e-01 -1.00494552e+00 -9.35473800e-01 -2.77268469e-01 7.86519527e-01 3.51865441e-01 9.04911339e-01 -2.16233641e-01 -5.35526037e-01 1.09131262e-01 -5.65832317e-01 -5.46939492e-01 -6.94309294e-01 -1.78936988e-01 -7.02922344e-02 7.47972846e-01 -8.72347206e-02 -7.64979661e-01 -3.36855710e-01 2.11027667e-01 -1.42284179e+00 -3.21272798e-02 9.88311231e-01 6.30113482e-01 6.93436861e-01 1.99380249e-01 3.12265843e-01 -9.50086176e-01 1.13434158e-01 -3.17163408e-01 -1.12914097e+00 1.40287906e-01 -4.04312104e-01 7.24259838e-02 8.30735922e-01 -3.14100325e-01 -1.19569218e+00 3.66769522e-01 6.07390292e-02 -3.47152740e-01 -5.12652338e-01 4.48715061e-01 -1.84127539e-01 -3.94220442e-01 6.93709135e-01 6.58240616e-01 1.76225360e-02 -4.50417131e-01 -7.60203786e-03 5.72083831e-01 4.95583951e-01 -5.72500415e-02 1.43771589e+00 7.84295321e-01 3.83729100e-01 -1.29748070e+00 -8.39640081e-01 -5.43774553e-02 -4.80907232e-01 -1.62417382e-01 1.10070395e+00 -9.83023703e-01 1.52051672e-02 5.64735293e-01 -9.83624816e-01 -2.26186682e-02 -1.50530338e-01 8.05714309e-01 -3.42621297e-01 4.78713661e-01 9.66216158e-03 -9.76381063e-01 -4.53019947e-01 -1.10628748e+00 5.67644060e-01 5.27067900e-01 4.48547810e-01 -6.65105700e-01 -1.10307574e-01 4.02885199e-01 4.60882366e-01 7.32781649e-01 7.79466033e-01 -2.14650080e-01 -1.05370486e+00 -1.52319998e-01 -1.15474142e-01 8.27082276e-01 3.78111035e-01 -1.82673141e-01 -1.01401341e+00 -5.09922624e-01 3.53752404e-01 -8.29997435e-02 7.95066595e-01 3.97414118e-01 8.45234215e-01 -3.47596824e-01 -1.51372522e-01 6.77873373e-01 1.96015251e+00 3.33664089e-01 1.01095712e+00 2.45072320e-01 4.15930808e-01 6.27855420e-01 9.27211225e-01 1.65178135e-01 -3.50231618e-01 8.65580320e-01 9.40372825e-01 -2.96137393e-01 1.10445030e-01 1.05246060e-01 2.97048301e-01 2.52807945e-01 -2.06475228e-01 -4.59818780e-01 -6.10523880e-01 4.41503733e-01 -1.46506763e+00 -9.78938639e-01 -3.76818568e-01 2.35476422e+00 3.06150109e-01 1.02745697e-01 -5.04410803e-01 2.97528774e-01 5.71778893e-01 3.73441160e-01 -1.90371022e-01 3.15503627e-02 -5.77041090e-01 5.71290433e-01 6.03609443e-01 3.60508621e-01 -1.04062986e+00 5.79754293e-01 4.56666374e+00 4.87027287e-01 -1.49741948e+00 -6.46446571e-02 4.03003246e-01 4.10533577e-01 -2.74852693e-01 1.77859291e-01 -4.47003007e-01 6.05642319e-01 4.74626333e-01 2.61721551e-01 2.70027697e-01 4.88798290e-01 2.25442067e-01 -3.04296196e-01 -3.83401334e-01 7.78197467e-01 2.35237002e-01 -9.47281063e-01 3.16619545e-01 1.37567624e-01 8.08483958e-01 -1.81119129e-01 1.67960256e-01 -2.04332769e-01 -2.70598922e-02 -1.09299588e+00 3.91049743e-01 9.35092330e-01 5.56919277e-01 -7.11579144e-01 9.82545853e-01 5.82756341e-01 -6.80235505e-01 1.54852290e-02 -5.18324137e-01 -2.73440871e-02 1.98881850e-01 1.01229453e+00 -8.56141031e-01 1.00617766e+00 4.15829480e-01 3.06096017e-01 -5.67207813e-01 9.46920753e-01 -3.94206285e-01 5.27378201e-01 -1.54268444e-01 1.87131777e-01 2.60492980e-01 -8.28203678e-01 8.99061561e-01 7.24584937e-01 7.55036354e-01 3.14100161e-02 4.45582047e-02 1.15720201e+00 3.17286849e-01 1.58418715e-02 -1.07613659e+00 -2.15167761e-01 1.59965560e-03 1.29871857e+00 -4.87010837e-01 -4.53118272e-02 -9.78178531e-02 8.93348694e-01 -2.70539790e-01 3.77163291e-01 -7.29408741e-01 -9.96157974e-02 5.07995486e-01 2.90482074e-01 4.66195911e-01 -4.07187223e-01 1.56167880e-01 -1.04510868e+00 -2.00640298e-02 -8.56782377e-01 -2.17633516e-01 -1.13094473e+00 -9.86236334e-01 9.79701817e-01 -8.21557567e-02 -1.75748277e+00 -2.69638062e-01 -5.43891013e-01 -3.49541962e-01 1.00962627e+00 -1.92700052e+00 -1.32353592e+00 -7.78784931e-01 5.36458254e-01 3.52784485e-01 -3.11261743e-01 9.06459153e-01 1.21250413e-01 -9.63973626e-02 -2.58218557e-01 1.58024028e-01 1.46257028e-01 4.94111478e-01 -8.94847512e-01 -1.04238041e-01 1.28385055e+00 2.63812453e-01 9.47140828e-02 9.96098399e-01 -5.43511748e-01 -1.13555324e+00 -1.23056304e+00 4.87161815e-01 2.54199117e-01 1.96752548e-01 -4.11980972e-02 -8.93884540e-01 5.63338935e-01 2.39614189e-01 1.78887099e-01 2.06067562e-01 -7.28598416e-01 -8.53634700e-02 -5.07365584e-01 -1.23900068e+00 3.44311535e-01 2.03326672e-01 -2.33798057e-01 -5.11652470e-01 3.18909824e-01 3.14272285e-01 -1.53429598e-01 -4.64296997e-01 7.01486349e-01 -6.45270990e-03 -1.31882870e+00 9.21010077e-01 -1.47359312e-01 3.35095823e-01 -5.85215211e-01 -2.11671263e-01 -1.54995918e+00 3.69197093e-02 -8.42526257e-02 5.75505137e-01 1.03839862e+00 1.26884922e-01 -9.88645196e-01 4.16586637e-01 -1.44305915e-01 4.53163944e-02 3.35470587e-02 -7.31390357e-01 -7.74235785e-01 -2.18208149e-01 1.49314761e-01 4.39133227e-01 1.09859860e+00 -1.03419530e+00 2.23625913e-01 -5.47766924e-01 6.30430818e-01 9.88901377e-01 6.25110745e-01 8.91392946e-01 -1.59620976e+00 -4.75406528e-01 1.12651512e-01 -4.71081674e-01 -3.79318386e-01 2.43994202e-02 -6.61037922e-01 -7.53319785e-02 -1.32439995e+00 -1.57950357e-01 -5.33910990e-01 6.69795275e-02 2.39311442e-01 1.31987959e-01 6.88253522e-01 2.79271811e-01 1.70173526e-01 4.63358104e-01 5.75994909e-01 1.30424607e+00 -3.67632240e-01 5.84366880e-02 3.23663056e-01 -1.49234444e-01 6.40024543e-01 9.77731168e-01 -5.09061456e-01 -3.20057452e-01 -3.25541198e-01 1.55241430e-01 1.59991279e-01 6.35408342e-01 -1.46577179e+00 -1.95642084e-01 -2.01144010e-01 4.95332181e-01 -5.83291292e-01 5.47292709e-01 -1.14605892e+00 7.07210958e-01 5.05686164e-01 1.40114009e-01 -2.24743649e-01 1.22687630e-01 5.24406612e-01 -5.75131059e-01 -4.02652085e-01 1.15970600e+00 -4.30179000e-01 -4.83058244e-01 1.42765671e-01 -4.11202401e-01 -4.47075903e-01 9.84077156e-01 -7.44831264e-02 -1.69427395e-01 -5.20493984e-01 -5.62301219e-01 -4.30495560e-01 4.02017087e-01 2.26050049e-01 5.32276750e-01 -8.12711835e-01 -9.35348213e-01 3.15423101e-01 1.39550596e-01 3.57922278e-02 3.88603300e-01 7.67572165e-01 -7.74254382e-01 4.09822911e-01 -5.77819943e-01 -5.36417186e-01 -1.28463650e+00 4.95139837e-01 4.00229216e-01 -3.08437943e-01 -4.18755859e-01 1.28324747e-01 3.91838402e-02 -2.70076483e-01 -3.86222929e-01 -1.95720255e-01 -3.12497139e-01 4.33204845e-02 2.98154265e-01 -5.91571704e-02 2.35819861e-01 -7.13230908e-01 2.50575487e-02 6.40899956e-01 2.77390689e-01 -7.58464485e-02 1.47830045e+00 1.58499748e-01 -1.54163808e-01 2.08846107e-01 7.27237105e-01 3.62127602e-01 -9.58445013e-01 -3.03844362e-01 -2.73243308e-01 -5.69396734e-01 2.41052121e-01 -8.40678990e-01 -1.23131692e+00 8.93590987e-01 8.97476494e-01 3.85591120e-01 1.37339926e+00 -4.99487430e-01 3.33557904e-01 2.45937183e-01 6.25141740e-01 -7.33926296e-01 -1.83645636e-01 -3.66327888e-03 9.95219111e-01 -1.25867176e+00 9.58474427e-02 -5.50070941e-01 -5.43969750e-01 1.10108078e+00 1.30508110e-01 -1.95255786e-01 2.58560508e-01 2.06077740e-01 2.22705618e-01 1.30360499e-01 4.78430353e-02 -6.22694075e-01 1.41839534e-01 6.40375078e-01 -2.25718748e-02 2.20129803e-01 -3.71759504e-01 -2.15721205e-02 5.17875254e-02 -2.59257276e-02 6.31901085e-01 6.83909118e-01 -2.87213326e-01 -1.22198164e+00 -8.44957292e-01 1.94228396e-01 -2.90914148e-01 6.54409379e-02 -3.79626393e-01 9.37632203e-01 1.21743135e-01 9.02406454e-01 -1.43753765e-02 -6.95336461e-02 2.90206037e-02 -6.77287132e-02 5.03754973e-01 -3.90223086e-01 -1.51787266e-01 -1.02342308e-01 -1.97594777e-01 -1.57616839e-01 -8.59812260e-01 -5.19303977e-01 -7.87377775e-01 1.87928587e-01 -3.80265892e-01 2.18666241e-01 1.04629278e+00 8.83432508e-01 2.20359303e-02 5.15478253e-01 8.63622069e-01 -9.41596210e-01 -4.70924288e-01 -1.04202139e+00 -8.34772289e-01 3.84925634e-01 3.51483434e-01 -5.23777366e-01 -4.77435827e-01 1.73459798e-02]
[10.04766845703125, -1.9665217399597168]
04de5d55-72bd-483b-992e-0fe61dbc986e
automatic-fine-grained-glomerular-lesion
2203.05847
null
https://arxiv.org/abs/2203.05847v1
https://arxiv.org/pdf/2203.05847v1.pdf
Automatic Fine-grained Glomerular Lesion Recognition in Kidney Pathology
Recognition of glomeruli lesions is the key for diagnosis and treatment planning in kidney pathology; however, the coexisting glomerular structures such as mesangial regions exacerbate the difficulties of this task. In this paper, we introduce a scheme to recognize fine-grained glomeruli lesions from whole slide images. First, a focal instance structural similarity loss is proposed to drive the model to locate all types of glomeruli precisely. Then an Uncertainty Aided Apportionment Network is designed to carry out the fine-grained visual classification without bounding-box annotations. This double branch-shaped structure extracts common features of the child class from the parent class and produces the uncertainty factor for reconstituting the training dataset. Results of slide-wise evaluation illustrate the effectiveness of the entire scheme, with an 8-22% improvement of the mean Average Precision compared with remarkable detection methods. The comprehensive results clearly demonstrate the effectiveness of the proposed method.
['Guang Yang', 'Zhihong Liu', 'Guotong Xie', 'Caihong Zeng', 'Guyue Zhang', 'Peng Tang', 'Fengyi Li', 'Yang Nan']
2022-03-11
null
null
null
null
['fine-grained-image-classification']
['computer-vision']
[ 2.51665115e-01 2.24857852e-01 9.73689370e-03 -5.27082503e-01 -8.40931535e-01 -4.65444833e-01 4.14341837e-01 3.32584441e-01 -2.27560401e-01 7.40129173e-01 -1.07732534e-01 -6.18555620e-02 -3.87482852e-01 -7.12650001e-01 -5.23762584e-01 -9.89448488e-01 -3.66951860e-02 7.01772392e-01 1.05871350e-01 4.82926250e-01 4.33512449e-01 9.67490256e-01 -1.17058837e+00 2.41503209e-01 1.23439884e+00 9.23016489e-01 3.23533177e-01 5.29297292e-01 -5.56379378e-01 3.20982367e-01 -3.51171970e-01 -2.94099987e-01 1.89484105e-01 1.71093810e-02 -5.73647738e-01 2.86992699e-01 6.01062059e-01 -2.51234889e-01 -8.60697776e-02 1.20840585e+00 5.49710751e-01 -3.39551121e-01 1.14634931e+00 -6.59176290e-01 -4.55406398e-01 5.12786925e-01 -5.76680064e-01 3.78949702e-01 -2.56936610e-01 9.97221991e-02 7.67213762e-01 -9.58540201e-01 5.95703661e-01 1.08444011e+00 3.99507076e-01 2.95946419e-01 -1.08539498e+00 -4.06882852e-01 -1.50643867e-02 -4.50746194e-02 -1.59702027e+00 -2.48901546e-02 4.88694251e-01 -5.68526328e-01 3.66739422e-01 4.15067673e-01 3.47973883e-01 2.21098453e-01 3.56567293e-01 6.67794287e-01 1.11064947e+00 -3.47516328e-01 9.60110202e-02 2.55243212e-01 1.65791333e-01 9.67306316e-01 6.39935195e-01 1.57059561e-02 1.55907452e-01 -2.30676960e-02 1.07595968e+00 3.27477366e-01 -1.08100727e-01 -4.97789472e-01 -9.74593699e-01 2.85093248e-01 8.05073738e-01 4.44986641e-01 -2.30652153e-01 1.32063314e-01 1.90710306e-01 -3.16056848e-01 2.02603206e-01 3.37251186e-01 1.31100342e-01 4.97607619e-01 -8.91063571e-01 3.44209783e-02 2.65432417e-01 7.92619407e-01 5.47437549e-01 -3.06098491e-01 -7.64359772e-01 6.66329741e-01 3.75878960e-01 5.56614459e-01 6.87009543e-02 -4.14622068e-01 2.10450619e-01 1.28347743e+00 -3.61231677e-02 -6.31424487e-01 -4.22846735e-01 -6.61038280e-01 -1.11836898e+00 2.76621163e-01 5.72641432e-01 2.77308971e-01 -1.68262458e+00 1.03128314e+00 3.51045460e-01 1.91050455e-01 -3.87269616e-01 9.08261299e-01 6.91947043e-01 3.44784915e-01 3.32190573e-01 3.52218375e-02 1.52431798e+00 -6.46600664e-01 -4.16812986e-01 7.15453699e-02 2.63262063e-01 -5.01216352e-01 6.74224138e-01 -1.47253126e-01 -7.56799400e-01 -2.32778564e-01 -1.06587875e+00 6.95630237e-02 -1.28406748e-01 6.40340865e-01 4.38533008e-01 4.85958010e-01 -6.69173300e-01 5.47608852e-01 -7.65083194e-01 -2.97475159e-01 8.91005158e-01 3.16928893e-01 -4.07413930e-01 -4.60885577e-02 -5.97985029e-01 8.39638293e-01 2.99471796e-01 4.55535471e-01 -7.62375951e-01 -1.04205525e+00 -4.73733217e-01 3.57623547e-01 -1.37218118e-01 -5.39718032e-01 7.51628101e-01 -2.76410997e-01 -8.14965606e-01 9.58348989e-01 -1.11131728e-01 -2.59996057e-01 8.84625554e-01 1.42019838e-01 -4.38244408e-03 3.24140877e-01 -2.66580060e-02 3.93370152e-01 6.73179984e-01 -1.23219478e+00 -8.46727371e-01 -5.14344335e-01 -2.99010694e-01 1.93697318e-01 -4.95176539e-02 -4.33095634e-01 -5.79214573e-01 -4.57263559e-01 3.73664558e-01 -6.12676799e-01 -5.06209075e-01 4.47229683e-01 -6.93946481e-01 -1.34895533e-01 4.11098123e-01 -4.30655211e-01 1.27819335e+00 -2.12117267e+00 -2.85537899e-01 7.70035565e-01 4.84666616e-01 3.46879333e-01 1.63336441e-01 1.97308175e-02 1.18057586e-01 2.58641690e-01 -1.44959927e-01 -7.19362870e-02 4.45402712e-02 6.14830852e-03 -2.02298522e-01 5.02356052e-01 3.59117776e-01 9.20286238e-01 -6.20466113e-01 -8.58919263e-01 4.86043513e-01 3.68761241e-01 -1.67564093e-03 2.05700904e-01 6.82287291e-02 1.56693906e-01 -7.52601981e-01 9.61766303e-01 9.09435511e-01 -6.15849972e-01 3.31535965e-01 -4.65167433e-01 -1.29105067e-02 -2.30555609e-01 -1.24196959e+00 1.10091531e+00 1.60116889e-02 4.87253517e-02 -3.25772278e-02 -5.41160941e-01 8.81931484e-01 5.77219203e-02 9.78658572e-02 -4.24962789e-01 9.67015103e-02 2.65345007e-01 -8.47667754e-02 -4.98047113e-01 -9.22175050e-02 -2.98680902e-01 1.71468616e-01 3.91726661e-03 -1.14126697e-01 1.18343607e-02 2.81702548e-01 6.70916587e-02 9.73202288e-01 -3.89166355e-01 5.39451778e-01 -4.57654864e-01 6.42621934e-01 -1.01183824e-01 4.34290767e-01 8.48443687e-01 -4.87854838e-01 7.73292243e-01 3.44871610e-01 -5.06183922e-01 -1.01317060e+00 -1.43394077e+00 -5.85766137e-01 3.23850840e-01 4.10276055e-01 1.82016596e-01 -5.62730074e-01 -1.01059008e+00 4.95775521e-01 1.15072414e-01 -7.32669175e-01 1.52497262e-01 -4.32607502e-01 -8.43825519e-01 5.75125635e-01 5.73005199e-01 4.51960593e-01 -6.24323428e-01 -3.22694004e-01 -6.21818602e-02 5.93816899e-02 -4.82161969e-01 -3.32066685e-01 1.04101434e-01 -1.01039088e+00 -1.18186748e+00 -7.95802712e-01 -1.03965771e+00 1.28918552e+00 -4.29749861e-02 6.46940231e-01 3.63516062e-01 -1.09088218e+00 -2.66251832e-01 2.65842527e-01 -1.28361285e-01 -1.70394182e-01 -2.40214482e-01 -2.60779917e-01 -2.11166795e-02 2.93401092e-01 -3.57030094e-01 -9.41773474e-01 2.10715160e-01 -6.25271738e-01 -2.21676394e-01 1.10704887e+00 9.36607242e-01 9.10585821e-01 1.49333449e-02 5.90513647e-01 -1.02921557e+00 2.70657718e-01 -1.02370866e-02 -6.91236258e-01 7.96241820e-01 -8.66060495e-01 2.90566206e-01 2.15825081e-01 -7.98097253e-02 -1.21663940e+00 2.48414487e-01 2.06012771e-01 -3.59300047e-01 -1.70895025e-01 1.90633968e-01 -5.77172171e-03 -3.93142283e-01 6.51165426e-01 1.63209245e-01 -9.67324227e-02 -4.59403038e-01 2.87895739e-01 6.43454909e-01 5.17881691e-01 -5.01651704e-01 7.74669886e-01 6.48372531e-01 1.55951068e-01 -5.03154337e-01 -4.33410883e-01 -6.11819923e-01 -3.32320213e-01 -1.97937578e-01 5.73955059e-01 -6.41589761e-01 -7.02181101e-01 2.48570383e-01 -7.66054809e-01 3.63210052e-01 -2.15259194e-01 5.06595910e-01 -2.83989925e-02 5.61643541e-01 -8.90164077e-01 -6.53337002e-01 -6.71012580e-01 -9.65077579e-01 9.73292768e-01 6.98000133e-01 2.79871136e-01 -7.46851683e-01 5.56096807e-02 3.81544769e-01 3.26558918e-01 7.06544816e-02 1.29554570e+00 -6.17682815e-01 -1.12220407e+00 -4.42622632e-01 -7.18049586e-01 3.75589207e-02 1.00294210e-01 3.31042379e-01 -9.81372595e-01 -7.78168514e-02 -3.69849712e-01 -1.28717627e-02 1.06086969e+00 6.68306768e-01 1.01896894e+00 8.61383155e-02 -7.91895568e-01 5.64660490e-01 1.78489041e+00 1.58655643e-01 7.52133489e-01 1.04610778e-01 4.13088024e-01 3.73599529e-01 6.28147900e-01 3.16912472e-01 -8.31835717e-02 -8.25398043e-02 4.77144659e-01 -2.94512987e-01 -2.07157955e-01 -1.92336723e-01 -4.60545391e-01 1.01294585e-01 -6.04944900e-02 -1.28347069e-01 -9.02320743e-01 7.08367109e-01 -1.61401153e+00 -7.78070092e-01 -5.44328988e-02 2.12366581e+00 6.80556476e-01 3.82001586e-02 -5.01042962e-01 -3.51898909e-01 1.24860799e+00 -1.76985294e-01 -5.17405748e-01 1.75355509e-01 1.06972836e-01 1.47765428e-01 4.69971418e-01 4.65022057e-01 -1.24612164e+00 5.11805296e-01 6.39414215e+00 1.13814354e+00 -1.15746081e+00 -5.54887950e-01 9.18987155e-01 9.94481370e-02 -2.66002417e-01 -2.12439984e-01 -1.00698054e+00 2.51970470e-01 7.14723906e-03 -3.06798875e-01 1.48310345e-02 5.83866477e-01 -6.46477193e-02 -2.12413356e-01 -1.09011757e+00 6.67271912e-01 -2.49601975e-01 -1.53678870e+00 3.44643980e-01 -7.19229504e-02 6.52938187e-01 -2.12635055e-01 5.60992919e-02 2.94209644e-02 -4.96724155e-03 -1.21478844e+00 4.12024930e-02 8.52585733e-01 9.75238740e-01 -5.42548418e-01 9.44993079e-01 1.20319635e-01 -1.20041609e+00 1.95987090e-01 -5.01816511e-01 3.38966608e-01 3.66231203e-02 1.06458902e+00 -1.45699036e+00 5.42047799e-01 3.92790914e-01 7.07441270e-02 -7.27156401e-01 1.66461182e+00 -1.88310221e-01 2.28230163e-01 -4.27788526e-01 -6.10568076e-02 -2.04458609e-02 -1.63700014e-01 4.56881344e-01 1.28482568e+00 4.34590995e-01 -1.97555013e-02 -4.91430461e-02 1.07775104e+00 -1.78220898e-01 1.54797420e-01 -1.42424345e-01 -2.81329393e-01 6.24291778e-01 1.62164152e+00 -9.86220658e-01 -3.21337402e-01 -7.36792758e-03 5.65406442e-01 5.58846295e-01 2.78029054e-01 -6.77150905e-01 -5.09784579e-01 2.95639604e-01 6.81355968e-02 3.47220749e-01 4.82955426e-01 -7.04010427e-01 -8.61346126e-01 2.07117260e-01 -1.99205622e-01 5.64077854e-01 -3.75447839e-01 -1.49075902e+00 3.96478534e-01 -4.99892473e-01 -1.05125988e+00 1.50783822e-01 -6.29219115e-01 -8.27590883e-01 1.03079593e+00 -1.45587981e+00 -1.25205958e+00 -3.67032498e-01 1.25332817e-01 -2.13879757e-02 1.88667312e-01 7.65511453e-01 3.21620256e-01 -4.89858270e-01 5.24316967e-01 4.00953859e-01 3.23912978e-01 8.06186199e-01 -1.53172266e+00 -1.13782026e-01 8.14291894e-01 -2.97426730e-01 7.98639536e-01 4.65663135e-01 -8.39029133e-01 -8.25299203e-01 -1.33222163e+00 1.08294523e+00 -1.82903975e-01 3.27763438e-01 -3.51260928e-03 -7.74112225e-01 5.25349200e-01 -3.36203009e-01 4.30645853e-01 4.06250238e-01 -9.12677348e-02 -2.25385740e-01 -2.54431814e-01 -1.56892252e+00 4.66152549e-01 6.26423717e-01 -3.07629764e-01 -7.27963328e-01 1.59219563e-01 2.47830704e-01 -5.64303875e-01 -8.60499382e-01 7.34570503e-01 6.93434954e-01 -6.96887076e-01 1.02847624e+00 -4.53877181e-01 3.76459509e-01 -7.90267050e-01 1.01216905e-01 -8.57536793e-01 -5.52397907e-01 1.68803915e-01 1.85879901e-01 1.27067447e+00 3.59361708e-01 -3.28534186e-01 1.14119887e+00 5.66115975e-01 -1.66308925e-01 -7.98182607e-01 -9.28771615e-01 -4.65702444e-01 -1.33743554e-01 2.51071960e-01 2.78167397e-01 5.66110134e-01 5.14463484e-02 -5.05476519e-02 7.80396089e-02 5.55488944e-01 1.05380750e+00 6.05336845e-01 7.96217173e-02 -1.21753943e+00 9.92273390e-02 -3.08488697e-01 -7.55354941e-01 -6.76854312e-01 -4.69105065e-01 -1.00426590e+00 1.88124031e-01 -1.78929865e+00 6.33293629e-01 -7.34015882e-01 -6.36079192e-01 2.47685790e-01 -4.26954091e-01 2.36338884e-01 -1.00901611e-01 8.23852271e-02 -5.11364400e-01 1.67527899e-01 1.50865138e+00 -3.91650915e-01 7.23070279e-02 1.91112161e-01 -7.05673277e-01 5.37112236e-01 4.97730643e-01 -2.94753522e-01 -3.11441272e-01 -6.07501483e-03 -2.38932788e-01 -1.36731222e-01 4.60437953e-01 -9.68320668e-01 4.04850185e-01 -1.95009485e-01 9.19267178e-01 -7.08407640e-01 5.05943336e-02 -6.91909432e-01 1.84868425e-01 7.38281906e-01 -3.70989770e-01 -7.11200178e-01 1.16242962e-02 9.08054769e-01 -1.66003197e-01 -3.41776043e-01 9.90109026e-01 -1.16692193e-01 -5.24330199e-01 3.65843683e-01 -1.79210827e-01 -1.95275396e-01 1.21552515e+00 -1.84887186e-01 -7.34695494e-01 2.74874657e-01 -9.55700457e-01 5.04851818e-01 2.16686577e-01 -1.51637718e-01 6.08805418e-01 -1.18430352e+00 -6.32444561e-01 2.38468125e-01 3.84065390e-01 2.95395970e-01 7.77477503e-01 7.70534813e-01 -8.25059712e-01 6.65151536e-01 -2.23257914e-01 -9.04654682e-01 -1.34510696e+00 4.03466702e-01 5.63296795e-01 -5.73635280e-01 -5.92459202e-01 1.09982193e+00 7.58741677e-01 -2.49537528e-01 2.71238685e-01 -6.45877719e-01 -2.26228416e-01 -2.57787734e-01 5.28575063e-01 3.57122034e-01 1.20332375e-01 1.01718307e-01 -5.54248750e-01 4.08578187e-01 -6.42538607e-01 4.66905981e-01 1.08349204e+00 -1.57496586e-01 -1.65452108e-01 -2.89426651e-02 7.48536170e-01 3.87349315e-02 -1.22200251e+00 -2.26756066e-01 -3.31507102e-02 -5.47888458e-01 -9.22676250e-02 -1.19198263e+00 -7.50165999e-01 8.75176609e-01 9.60853219e-01 -2.51526684e-01 8.32141221e-01 7.44521469e-02 2.08335340e-01 2.42312089e-01 2.49527648e-01 -8.18394721e-01 -2.98853755e-01 6.40173331e-02 5.82550764e-01 -1.05170000e+00 1.66287929e-01 -8.09542537e-01 -3.70802283e-01 1.26941657e+00 7.41581321e-01 -2.63372809e-01 4.55166250e-01 3.53258312e-01 1.37311220e-01 -3.76097053e-01 -4.74498689e-01 -1.45476729e-01 4.30569053e-01 3.28277409e-01 1.99702218e-01 3.01876575e-01 -4.49373186e-01 6.26304686e-01 6.76147461e-01 1.00861870e-01 2.14460522e-01 7.91416347e-01 -9.15501595e-01 -6.87589586e-01 -2.12321624e-01 8.82949591e-01 -4.13372487e-01 2.19425131e-02 -2.16594398e-01 7.40938187e-01 7.76069909e-02 2.54141539e-01 9.46141630e-02 -2.47707572e-02 1.24707676e-01 6.22221269e-02 6.17909908e-01 -5.10054469e-01 -4.90514547e-01 2.94253051e-01 -9.91396457e-02 -3.16897243e-01 -3.21371742e-02 -2.63149917e-01 -1.61877811e+00 -2.63911579e-02 -4.43341851e-01 1.52730107e-01 4.05329645e-01 9.49701548e-01 2.76744246e-01 5.25389791e-01 4.79252249e-01 -3.70799661e-01 -8.35289776e-01 -7.42909610e-01 -8.85715365e-01 2.06592828e-01 3.86721879e-01 -4.35529858e-01 -4.53372657e-01 2.64186077e-02]
[15.132658958435059, -2.8118937015533447]
56cf346e-b31f-429a-94a8-de00054d5f33
instance-shadow-detection-with-a-single-stage
2207.04614
null
https://arxiv.org/abs/2207.04614v1
https://arxiv.org/pdf/2207.04614v1.pdf
Instance Shadow Detection with A Single-Stage Detector
This paper formulates a new problem, instance shadow detection, which aims to detect shadow instance and the associated object instance that cast each shadow in the input image. To approach this task, we first compile a new dataset with the masks for shadow instances, object instances, and shadow-object associations. We then design an evaluation metric for quantitative evaluation of the performance of instance shadow detection. Further, we design a single-stage detector to perform instance shadow detection in an end-to-end manner, where the bidirectional relation learning module and the deformable maskIoU head are proposed in the detector to directly learn the relation between shadow instances and object instances and to improve the accuracy of the predicted masks. Finally, we quantitatively and qualitatively evaluate our method on the benchmark dataset of instance shadow detection and show the applicability of our method on light direction estimation and photo editing.
['Chi-Wing Fu', 'Pheng-Ann Heng', 'Xiaowei Hu', 'Tianyu Wang']
2022-07-11
null
null
null
null
['shadow-detection']
['computer-vision']
[ 8.67864311e-01 4.99705791e-01 1.56005129e-01 -7.46355653e-01 -5.18725336e-01 -3.17778349e-01 6.76809907e-01 -3.94940317e-01 -1.55110639e-02 4.97297108e-01 -6.41740188e-02 -1.24712393e-01 3.98244321e-01 -6.26724899e-01 -7.95363426e-01 -6.92716360e-01 3.27186882e-01 7.21765935e-01 8.60687435e-01 3.28372389e-01 2.01560438e-01 5.66456437e-01 -1.39574146e+00 4.85127032e-01 8.32801819e-01 1.10799873e+00 6.18079007e-01 7.63648033e-01 1.42986357e-01 7.07723379e-01 -7.05896080e-01 -3.70924741e-01 3.66101682e-01 -2.67195672e-01 -4.10279483e-01 4.66442406e-01 1.03255868e+00 -5.04316092e-01 -6.82311729e-02 6.39984310e-01 5.42690516e-01 1.30283773e-01 6.93049252e-01 -1.31218636e+00 -2.11573929e-01 8.43231529e-02 -7.29607224e-01 -4.71233651e-02 2.19990239e-01 3.94441396e-01 8.40506434e-01 -1.37406075e+00 6.81846261e-01 1.22123873e+00 4.25893486e-01 3.78950596e-01 -1.12145305e+00 -6.78013086e-01 4.23557490e-01 1.61148399e-01 -1.04588366e+00 -5.25421560e-01 7.79828608e-01 -2.84059405e-01 5.23826003e-01 4.47574705e-01 7.34738171e-01 6.66463614e-01 3.61839943e-02 1.15518260e+00 1.40341544e+00 -3.01197290e-01 1.03752106e-01 4.08641815e-01 1.02122687e-01 1.08813238e+00 -1.49854347e-01 2.92653978e-01 -5.92034817e-01 -1.08141892e-01 4.13877845e-01 -7.31086805e-02 -5.02630115e-01 -4.38112259e-01 -1.07510626e+00 2.21643224e-01 3.95389199e-01 -3.81122231e-01 -2.10197344e-01 2.34249890e-01 1.81356017e-02 -2.95330018e-01 6.10557854e-01 1.26575287e-02 -5.22216558e-01 5.42310238e-01 -9.28890228e-01 1.46345451e-01 7.99046993e-01 1.12889028e+00 8.31356823e-01 -3.60314876e-01 -8.99260759e-01 6.02192163e-01 1.22579753e-01 9.09884453e-01 -4.41619873e-01 -9.04552877e-01 3.87037069e-01 8.00704598e-01 2.38488838e-01 -7.32791662e-01 -2.35595435e-01 -3.50671768e-01 -3.59286785e-01 3.70229751e-01 1.02430291e-01 7.40566552e-02 -1.10009611e+00 1.30274844e+00 8.40313315e-01 7.40406156e-01 -1.58760205e-01 9.74526882e-01 9.22124386e-01 5.90232074e-01 -1.52726725e-01 -1.92028582e-01 1.20401263e+00 -1.35895360e+00 -5.48630178e-01 -4.13696766e-01 2.04186261e-01 -9.08344209e-01 1.44202149e+00 1.36729423e-02 -1.08690453e+00 -4.84320343e-01 -7.58901358e-01 -2.26055637e-01 -7.99538568e-02 7.17318237e-01 5.12695909e-01 2.71873027e-01 -4.84706521e-01 1.52643472e-01 -6.26025319e-01 -9.66387019e-02 6.74617827e-01 2.42403433e-01 1.82052135e-01 -1.21546246e-01 -4.40160781e-01 7.56578743e-01 1.04103364e-01 7.46070370e-02 -1.23114407e+00 -9.99439061e-01 -6.40477955e-01 1.14650071e-01 6.68392956e-01 -7.05253422e-01 1.15206993e+00 -1.06340599e+00 -1.27958095e+00 1.01937783e+00 -5.43917120e-01 -1.92209199e-01 5.56770265e-01 -1.66879803e-01 -1.21889345e-01 -1.59561709e-01 5.88048883e-02 5.49800575e-01 1.09014559e+00 -2.00483012e+00 -1.25743580e+00 -2.45976850e-01 3.59642327e-01 3.47513944e-01 -5.78156747e-02 -1.26637071e-01 -9.66720343e-01 -3.94809365e-01 -9.23450515e-02 -1.05168641e+00 -1.95525456e-02 1.70858324e-01 -8.01765442e-01 -1.24609150e-01 1.32978010e+00 -7.10915148e-01 1.02680647e+00 -2.23833370e+00 -2.34224014e-02 2.83878028e-01 1.98798284e-01 1.90879285e-01 -2.44795606e-02 -5.33520505e-02 3.94346982e-01 -6.80131376e-01 -7.19650984e-01 -7.03505337e-01 -1.61482841e-02 4.83159184e-01 -5.57021320e-01 3.42186779e-01 1.06861860e-01 7.99238980e-01 -9.01699781e-01 -5.03159165e-01 4.04892147e-01 3.86305124e-01 -2.49826387e-01 6.17445052e-01 -5.55767298e-01 3.67229849e-01 -2.54848868e-01 8.28858912e-01 8.87752771e-01 -2.17200086e-01 2.38074318e-01 -6.72298074e-01 -4.56540026e-02 -6.92310631e-02 -1.26429272e+00 1.13560987e+00 -7.84714699e-01 9.01276231e-01 6.02424927e-02 -1.13961510e-01 5.54362655e-01 -1.92042992e-01 2.89647996e-01 -5.08936584e-01 -1.03519060e-01 -1.21844776e-01 -2.80230850e-01 -4.93415564e-01 3.29797924e-01 2.23790124e-01 4.18096811e-01 3.53394002e-01 -4.55031633e-01 -3.55073452e-01 -9.37314481e-02 3.09924871e-01 9.56576765e-01 2.77438432e-01 1.19690098e-01 -1.11176580e-01 6.24189615e-01 -1.38158441e-01 4.08829093e-01 5.31674504e-01 2.00757995e-01 7.16400564e-01 1.31857827e-01 -1.62409738e-01 -5.47426701e-01 -1.10005045e+00 -1.79865196e-01 1.32531202e+00 5.86273551e-01 -1.12012751e-01 -6.91366851e-01 -1.12050545e+00 2.31478974e-01 1.00872540e+00 -6.89211130e-01 1.67374820e-01 -6.78736985e-01 -6.50942683e-01 -5.32762110e-02 5.80384433e-01 4.77854609e-01 -1.13201737e+00 -1.05681193e+00 -2.36972436e-01 -1.03429765e-01 -1.44186735e+00 -7.08499551e-01 8.05246681e-02 -2.56425768e-01 -1.43034375e+00 -3.38306546e-01 -7.81953692e-01 1.01536977e+00 4.21442926e-01 1.30416703e+00 2.56258726e-01 -7.03919351e-01 5.28697729e-01 1.27143532e-01 -7.06704199e-01 -3.11842978e-01 -3.36594760e-01 -4.04312015e-01 4.13658410e-01 -3.74883085e-01 -2.10985959e-01 -9.41110671e-01 4.79893893e-01 -8.33260655e-01 5.39619803e-01 7.32386053e-01 5.29674113e-01 7.66256273e-01 -2.17894047e-01 -1.66478172e-01 -1.28700948e+00 1.68165788e-01 1.33617550e-01 -1.07914114e+00 6.37914777e-01 -6.93177402e-01 -1.97831094e-01 1.32173896e-01 -8.25429037e-02 -1.50721383e+00 7.10101485e-01 4.02285457e-01 -3.50330830e-01 1.64031913e-03 -2.39642829e-01 -5.23125052e-01 -2.51432806e-01 4.94745642e-01 2.29137108e-01 -5.69612741e-01 -2.58720726e-01 6.35273159e-01 4.75508034e-01 6.25309944e-01 -5.17235696e-01 1.12276351e+00 9.62677360e-01 1.22598961e-01 -6.88292205e-01 -1.35326958e+00 -6.21845186e-01 -5.98158658e-01 -4.94037896e-01 7.07944214e-01 -6.96269572e-01 -7.15416014e-01 2.37745345e-01 -1.44461036e+00 -8.32284510e-01 -2.95205265e-01 -2.38509893e-01 -3.91607821e-01 1.75163940e-01 -6.45657480e-02 -8.89040053e-01 -2.82805502e-01 -1.07989609e+00 1.81476259e+00 6.74906075e-02 3.77712071e-01 -7.23145545e-01 -4.51570675e-02 3.53734970e-01 -6.76097199e-02 2.23194674e-01 9.28870261e-01 2.12472435e-02 -1.13546944e+00 1.01354346e-01 -7.45960772e-01 5.66456616e-01 6.60917461e-02 8.61150119e-03 -1.33935249e+00 -1.17988989e-01 -4.45382327e-01 5.13230599e-02 9.50857997e-01 2.14119762e-01 1.45833290e+00 -2.72077501e-01 -7.32842982e-01 5.85331857e-01 1.53279865e+00 9.20553971e-03 4.46639359e-01 -2.01584861e-01 1.06858718e+00 6.44736767e-01 1.07458377e+00 4.40759450e-01 3.91193390e-01 9.21514809e-01 4.56293911e-01 -6.69839144e-01 -7.86479712e-01 1.12161167e-01 2.81632304e-01 3.61980908e-02 -1.62890211e-01 -3.53588700e-01 -8.30804706e-01 2.17511907e-01 -1.92370963e+00 -6.71621263e-01 -3.87275994e-01 2.13527989e+00 6.98736012e-01 4.20963280e-02 -1.50113612e-01 -2.16030568e-01 5.61654270e-01 1.06635936e-01 -6.32151425e-01 3.11005443e-01 -2.43919902e-02 1.25556812e-01 6.34435236e-01 9.32928383e-01 -9.33290184e-01 1.28295565e+00 6.12553501e+00 5.20862877e-01 -8.78785133e-01 1.51662022e-01 5.31250358e-01 -2.30255686e-02 -2.42947713e-01 1.53132662e-01 -9.82953906e-01 3.58809292e-01 1.05737165e-01 2.82062829e-01 5.87192297e-01 6.97039068e-01 4.39474940e-01 -4.67462420e-01 -1.41331828e+00 6.83737636e-01 2.40616605e-01 -1.24237323e+00 -2.43084505e-03 9.87600256e-03 9.40108359e-01 -2.45512098e-01 -4.72499505e-02 -2.55857538e-02 5.54265603e-02 -5.81104219e-01 8.03767025e-01 7.01454163e-01 8.59539747e-01 -2.86539644e-01 2.64495373e-01 2.14678213e-01 -1.35591447e+00 -1.35108948e-01 -2.66930819e-01 2.72577554e-01 2.55472243e-01 8.38145971e-01 -1.44155240e+00 2.63127863e-01 3.03014398e-01 4.54427183e-01 -6.34721458e-01 1.08450961e+00 -6.34562135e-01 5.67545831e-01 -1.87209725e-01 3.33383858e-01 -2.87212759e-01 -7.67420754e-02 5.80180764e-01 1.44019544e+00 -2.46897608e-01 2.41288036e-01 4.34841156e-01 8.45002651e-01 -1.75285786e-01 -2.47006968e-01 -2.25689381e-01 6.01683080e-01 5.36780655e-01 1.57446587e+00 -8.44802797e-01 -5.98245203e-01 -3.75245214e-01 1.32798851e+00 1.84052840e-01 5.40485620e-01 -1.37736428e+00 2.95589808e-02 4.78469312e-01 4.48093712e-01 2.19613269e-01 1.35048449e-01 -4.26894218e-01 -7.12713480e-01 2.70379543e-01 -5.45218349e-01 1.01053551e-01 -1.09356463e+00 -1.04746902e+00 2.73273259e-01 -1.93476379e-02 -8.77657533e-01 2.91225076e-01 -5.37811041e-01 -6.59687281e-01 6.16194785e-01 -1.65194416e+00 -1.47800362e+00 -9.88058150e-01 6.22283101e-01 7.85849094e-01 2.87939310e-01 5.42224646e-01 2.03640983e-01 -4.62103456e-01 3.42408419e-01 -2.45996825e-02 -2.77102999e-02 6.64960086e-01 -1.41626954e+00 3.48474503e-01 8.17626417e-01 1.55434281e-01 4.41293344e-02 6.80373967e-01 -7.74918854e-01 -1.20631051e+00 -1.52418590e+00 5.19405961e-01 -8.11138868e-01 2.20895365e-01 -6.88275099e-01 -5.37759960e-01 5.61579764e-01 -4.63252105e-02 3.48758131e-01 -1.66600347e-02 -9.93171036e-02 -1.89427868e-01 -4.64393616e-01 -1.00303209e+00 6.16459668e-01 1.48529732e+00 -4.61414039e-01 -3.13406229e-01 9.71416414e-01 7.04663813e-01 -8.89609694e-01 -3.00803065e-01 6.13747180e-01 5.48231006e-01 -9.70424592e-01 1.20765030e+00 -2.19794378e-01 4.26855266e-01 -5.34063101e-01 -1.95975810e-01 -9.62314069e-01 1.12566259e-02 -3.04710180e-01 -3.96038592e-01 1.27578974e+00 5.09848535e-01 -2.38061413e-01 7.99881220e-01 5.05091071e-01 -3.80217075e-01 -1.16020083e+00 -5.57872057e-01 -4.59349781e-01 -9.15994585e-01 -3.22002858e-01 5.36295056e-01 4.22640502e-01 -8.18683088e-01 3.58756125e-01 -2.57512659e-01 6.44358635e-01 8.33798230e-01 8.29096138e-01 1.19328988e+00 -8.92167270e-01 -5.58441937e-01 6.71031401e-02 -9.45567712e-03 -1.01309192e+00 3.02033126e-01 -6.92064703e-01 4.95923609e-01 -1.82073331e+00 6.84510410e-01 -6.21118963e-01 -6.84161112e-02 5.01232028e-01 -5.76701164e-01 3.72537971e-01 2.89178491e-01 2.59727657e-01 -7.42444336e-01 4.08794969e-01 1.36145914e+00 -2.99708933e-01 -3.11270744e-01 3.73078287e-01 -1.43259406e-01 7.92659760e-01 2.27454856e-01 -4.58160758e-01 -4.35884476e-01 -3.71921808e-01 -2.63952792e-01 -8.36928859e-02 9.10869539e-01 -9.31819141e-01 -5.37440218e-02 -4.01121438e-01 4.34974670e-01 -7.33921826e-01 6.19134605e-01 -1.06764638e+00 -9.64426473e-02 3.46869349e-01 -2.25446552e-01 -4.92904395e-01 -1.34753734e-01 7.19806254e-01 3.72507572e-01 3.00498575e-01 1.00963438e+00 1.36789707e-02 -6.89149976e-01 5.79131007e-01 3.28083426e-01 3.71625274e-02 1.18867934e+00 -1.43003985e-01 -2.66798556e-01 3.90621275e-02 -6.25151217e-01 1.50778756e-01 5.20380616e-01 2.74158359e-01 6.79545999e-01 -8.84806454e-01 -5.22642195e-01 7.06561804e-02 3.74268115e-01 2.60082453e-01 -1.42023429e-01 8.04293096e-01 -1.70192018e-01 4.69296910e-02 2.30833918e-01 -7.09999084e-01 -1.71994555e+00 2.73713142e-01 5.11200786e-01 1.53023973e-01 -7.59473205e-01 9.17928278e-01 1.00088501e+00 -1.46141693e-01 6.08623862e-01 -5.82041919e-01 2.52012879e-01 -3.57861727e-01 3.73516560e-01 4.66798812e-01 7.35712573e-02 -3.44794780e-01 -4.25378501e-01 4.01111424e-01 2.42680192e-01 1.90857779e-02 1.04627144e+00 -5.58831654e-02 -1.78296193e-01 1.92322046e-01 8.46302271e-01 3.35782468e-01 -1.65854526e+00 -2.77135402e-01 -2.53210783e-01 -8.13979387e-01 4.90457118e-02 -1.26539874e+00 -1.11451364e+00 5.22541761e-01 9.18173671e-01 -1.81598112e-01 1.13846302e+00 2.79717922e-01 8.05102706e-01 3.64666015e-01 1.50017157e-01 -1.10940099e+00 1.89506263e-01 2.79032528e-01 9.30624843e-01 -1.22027922e+00 1.58246234e-01 -1.01703560e+00 -6.39058709e-01 7.56501853e-01 8.33887398e-01 1.61648065e-01 6.73820317e-01 6.04602396e-01 8.59922022e-02 -4.30445790e-01 -5.07525265e-01 -3.80371541e-01 6.51955068e-01 5.62898874e-01 6.41042888e-02 3.10100734e-01 -3.37704308e-02 1.07479371e-01 1.35205686e-01 -1.22128196e-01 1.86412498e-01 7.05984950e-01 -6.36143148e-01 -9.06330645e-01 -4.38846856e-01 4.98810440e-01 2.13621646e-01 -2.47469082e-01 -8.83285701e-01 5.94136417e-01 4.64887708e-01 6.65844619e-01 -1.16079152e-01 -1.53367355e-01 5.96728921e-01 -2.35679582e-01 6.75972879e-01 -7.78474391e-01 -4.12910908e-01 -1.49616569e-01 1.10807642e-01 -8.17337751e-01 -3.46995592e-01 -4.90466446e-01 -1.42131341e+00 1.69285089e-01 -4.88305420e-01 -3.46772492e-01 6.50484622e-01 9.89742339e-01 2.90239483e-01 6.73736930e-01 7.37272978e-01 -1.06514311e+00 -1.94338784e-01 -6.53132975e-01 -3.39542419e-01 3.98928583e-01 2.65887231e-01 -7.62007713e-01 -7.48315081e-02 3.44586670e-01]
[10.851092338562012, -4.111710548400879]
75f9d925-14c1-48d9-bf54-be0e8bb2bfcc
dynamic-size-message-scheduling-for-multi
2306.10134
null
https://arxiv.org/abs/2306.10134v1
https://arxiv.org/pdf/2306.10134v1.pdf
Dynamic Size Message Scheduling for Multi-Agent Communication under Limited Bandwidth
Communication plays a vital role in multi-agent systems, fostering collaboration and coordination. However, in real-world scenarios where communication is bandwidth-limited, existing multi-agent reinforcement learning (MARL) algorithms often provide agents with a binary choice: either transmitting a fixed number of bytes or no information at all. This limitation hinders the ability to effectively utilize the available bandwidth. To overcome this challenge, we present the Dynamic Size Message Scheduling (DSMS) method, which introduces a finer-grained approach to scheduling by considering the actual size of the information to be exchanged. Our contribution lies in adaptively adjusting message sizes using Fourier transform-based compression techniques, enabling agents to tailor their messages to match the allocated bandwidth while striking a balance between information loss and transmission efficiency. Receiving agents can reliably decompress the messages using the inverse Fourier transform. Experimental results demonstrate that DSMS significantly improves performance in multi-agent cooperative tasks by optimizing the utilization of bandwidth and effectively balancing information value.
['Raphaël Avalos', 'Ann Nowé', 'Yuan YAO', 'Denis Steckelmacher', 'Qingshuang Sun']
2023-06-16
null
null
null
null
['multi-agent-reinforcement-learning']
['methodology']
[ 2.07899101e-02 1.45690516e-02 -4.46303278e-01 -1.28426626e-01 -5.97718596e-01 -3.23350221e-01 5.62919796e-01 7.53881156e-01 -9.53060091e-01 1.15108788e+00 6.02585124e-03 -2.08292812e-01 -3.79096210e-01 -9.66531336e-01 -2.38361433e-01 -7.72941232e-01 -7.40831614e-01 8.22224379e-01 2.09979281e-01 -3.51242095e-01 5.13918459e-01 2.40861818e-01 -1.44330430e+00 -2.10634366e-01 6.16212845e-01 8.95712256e-01 5.16503632e-01 9.84515488e-01 -1.87101483e-01 9.34614241e-01 -1.23818016e+00 2.11704910e-01 3.24577719e-01 -4.20521289e-01 -8.23845446e-01 2.69632280e-01 -5.05045593e-01 -9.25001442e-01 -1.92197457e-01 5.56544006e-01 4.12515283e-01 2.71050930e-01 3.20547819e-01 -1.79056633e+00 -6.97196871e-02 9.37699199e-01 -7.92576790e-01 4.66848046e-01 4.08403188e-01 -3.09194885e-02 6.46002114e-01 1.06865585e-01 5.27919471e-01 1.26959872e+00 2.73033887e-01 3.70289117e-01 -1.14339936e+00 -5.78855276e-01 1.14132226e-01 2.73277372e-01 -1.06890750e+00 -5.71978748e-01 4.09071863e-01 -2.94688884e-02 8.49894464e-01 4.03401703e-01 7.42441297e-01 2.00568646e-01 5.03899753e-01 4.65334237e-01 1.03826392e+00 -6.10629022e-01 6.65856719e-01 -2.08601132e-01 -4.85767096e-01 4.54085767e-01 1.07168756e-01 -8.44274387e-02 -1.06138730e+00 -4.97793645e-01 8.56631279e-01 -2.77413815e-01 -3.20055842e-01 -2.37277448e-01 -1.59222007e+00 9.80351150e-01 3.34627628e-02 1.51607469e-01 -7.95741558e-01 4.87708509e-01 6.39714301e-01 9.29574192e-01 4.82618898e-01 4.20016348e-01 -2.81125396e-01 -6.13087118e-01 -6.15888596e-01 4.30306792e-01 1.11548388e+00 8.32499325e-01 8.03731740e-01 -2.12559514e-02 1.90153375e-01 6.63888335e-01 3.01548362e-01 4.39062357e-01 5.36117792e-01 -1.48789239e+00 4.76876229e-01 1.87532976e-01 2.82396108e-01 -5.66106856e-01 -5.75362444e-01 -4.55954000e-02 -6.52192950e-01 5.71402490e-01 4.77641016e-01 -6.16964102e-01 -5.14243729e-02 1.76916349e+00 6.82264209e-01 -1.34308532e-01 3.48150402e-01 9.81830239e-01 -1.23846054e-01 7.28362083e-01 -6.38628602e-02 -6.65617764e-01 1.21591485e+00 -9.08036649e-01 -6.67573392e-01 -3.28227356e-02 6.56074762e-01 -9.04312491e-01 5.44309735e-01 2.53623221e-02 -1.40081537e+00 8.01579729e-02 -1.02579558e+00 5.51710784e-01 1.31435350e-01 -6.20520413e-01 6.91624165e-01 5.88577449e-01 -1.11189079e+00 4.45079625e-01 -9.75172579e-01 -4.89422157e-02 -1.31252930e-01 5.57866573e-01 -1.43436760e-01 3.52475792e-01 -9.63595390e-01 5.99806547e-01 2.20550954e-01 -4.85432953e-01 -7.99475789e-01 -8.01416516e-01 -5.45510709e-01 1.58291966e-01 6.49385989e-01 -7.23690450e-01 1.66127658e+00 -9.16935325e-01 -1.79526544e+00 1.28639210e-02 4.32795525e-01 -8.68595600e-01 5.36456764e-01 3.59617829e-01 1.61104620e-01 4.72211659e-01 2.46818550e-02 7.15188086e-01 7.86713779e-01 -1.08444071e+00 -1.01458406e+00 -5.31406403e-02 5.80292583e-01 8.13911140e-01 -6.54607475e-01 -2.41964474e-01 1.65408254e-01 -3.89807045e-01 -2.59335369e-01 -9.36790287e-01 -3.45377952e-01 4.29799184e-02 5.36940508e-02 -3.44241083e-01 8.53248537e-01 -1.47772292e-02 9.22385395e-01 -1.93000460e+00 5.18355854e-02 1.85753062e-01 4.59404111e-01 -3.63553613e-02 -3.86645079e-01 1.05749369e+00 7.42479086e-01 -1.79181680e-01 1.29371613e-01 -3.54222924e-01 2.56692648e-01 4.86094475e-01 4.21674997e-02 5.37626088e-01 -4.34984684e-01 1.67857483e-01 -9.28131878e-01 -5.35732746e-01 -4.91446145e-02 2.82814592e-01 -6.75208211e-01 4.84577984e-01 -2.92447180e-01 3.09447616e-01 -7.04221487e-01 2.31583953e-01 4.78613734e-01 -3.66302401e-01 5.55699050e-01 3.62594515e-01 -2.29525983e-01 2.76393503e-01 -1.32092297e+00 1.32398605e+00 -6.85913682e-01 5.80526114e-01 8.37359667e-01 -9.02115822e-01 5.58251381e-01 2.50217140e-01 1.15227520e+00 -9.12741005e-01 -3.74141932e-02 4.33321819e-02 2.50463150e-02 -2.14752659e-01 8.03867221e-01 6.45304248e-02 1.70240384e-02 1.33005166e+00 -6.39308274e-01 -2.00339854e-01 6.43330455e-01 3.80495429e-01 1.33159363e+00 -3.71988893e-01 2.53846288e-01 -1.88873902e-01 2.06427500e-01 1.98823154e-01 3.59753311e-01 8.55987012e-01 -3.20155948e-01 -4.25256133e-01 4.54597771e-01 -4.11154091e-01 -1.22052002e+00 -7.03005493e-01 3.07712555e-01 1.49647999e+00 3.07591885e-01 -3.92984748e-01 -9.30876911e-01 -2.81008363e-01 1.61612779e-01 3.75987589e-01 -2.51220196e-01 1.22428238e-01 -6.33044243e-01 -6.21507406e-01 1.36879385e-01 -2.00435370e-02 6.11094177e-01 -1.04467702e+00 -1.66718328e+00 6.57924175e-01 -3.68489116e-01 -8.74479592e-01 -8.32340539e-01 6.68888167e-02 -4.70713496e-01 -1.13693714e+00 -5.52209139e-01 -3.85313720e-01 5.96395671e-01 7.21442044e-01 9.14233029e-01 4.90287542e-01 -2.51085848e-01 9.32149172e-01 -4.43543047e-01 -2.98509002e-01 -7.80970573e-01 1.38393775e-01 2.61681229e-02 -3.87048364e-01 -1.43673927e-01 -5.52815497e-01 -7.57718861e-01 2.26842016e-01 -1.23922360e+00 1.35938942e-01 4.73200560e-01 7.83301175e-01 1.25457436e-01 2.87851632e-01 9.31489885e-01 -3.92090410e-01 1.19760191e+00 -4.69369590e-01 -7.43985593e-01 -4.35731038e-02 -6.60186172e-01 6.84043691e-02 7.42991269e-01 -5.78395844e-01 -6.58271551e-01 -4.32355523e-01 1.89972505e-01 1.91450670e-01 2.66757727e-01 4.08717066e-01 6.34990513e-01 -3.32215369e-01 2.19139218e-01 2.54105449e-01 7.55855203e-01 1.96809098e-01 1.46666825e-01 7.48892605e-01 -3.68066542e-02 -6.22123897e-01 2.33137935e-01 3.89333755e-01 9.07648280e-02 -9.34060156e-01 -5.36170490e-02 -1.87453225e-01 1.27271172e-02 -3.78572285e-01 1.14450417e-01 -6.40738189e-01 -1.54130805e+00 3.29345733e-01 -8.47790182e-01 -8.63845348e-01 -1.46328568e-01 6.97133541e-01 -9.93534029e-01 5.12953877e-01 -8.55584025e-01 -8.85452092e-01 -3.24916303e-01 -1.27260935e+00 7.13547051e-01 6.04862627e-03 -9.92436856e-02 -8.65470946e-01 1.18672110e-01 1.17591687e-01 9.73075032e-01 -8.24710280e-02 7.07764626e-01 -4.99737442e-01 -8.27695727e-01 3.53230327e-01 -3.95134650e-02 -3.80199343e-01 6.91054314e-02 -5.54220736e-01 -3.58571522e-02 -8.87970924e-01 -1.60554081e-01 -6.75585866e-01 1.80119067e-01 2.86122233e-01 7.88951457e-01 -6.29123569e-01 -1.59520909e-01 1.31383896e-01 1.31450641e+00 1.71543762e-01 1.20937489e-02 7.91617393e-01 -2.07313821e-01 5.32759845e-01 7.16728747e-01 1.48483086e+00 9.54157293e-01 7.44231999e-01 8.09653699e-01 2.11200595e-01 -3.67855579e-02 2.42272928e-01 3.84678811e-01 7.49354959e-01 2.08043054e-01 -3.99258703e-01 -6.28915966e-01 3.32230031e-01 -1.95432878e+00 -8.02035093e-01 5.26544034e-01 2.32638097e+00 1.12614536e+00 1.11105643e-01 3.14762235e-01 1.95559978e-01 5.19818604e-01 7.46461153e-02 -7.32252955e-01 -4.70198661e-01 4.16595846e-01 -4.12135154e-01 7.19292402e-01 6.98311210e-01 -6.00718975e-01 4.38639700e-01 6.30952358e+00 7.50723541e-01 -9.05783236e-01 1.27868829e-02 3.09571981e-01 -3.59359175e-01 -1.63110003e-01 -2.79482454e-01 -4.21066135e-01 7.21193314e-01 1.37101424e+00 -7.81089783e-01 1.08256435e+00 4.74206239e-01 6.44668937e-01 -6.59003496e-01 -8.85209620e-01 7.11321890e-01 -1.12979524e-01 -1.51836145e+00 -1.72901869e-01 4.11793292e-01 3.26292932e-01 -1.24213919e-01 -2.49909088e-01 -2.31947340e-02 5.30544043e-01 -5.52615583e-01 7.39261806e-01 1.80889666e-01 4.03051138e-01 -9.86511827e-01 4.42976147e-01 5.53364754e-01 -1.26619089e+00 -1.91124424e-01 -3.97196978e-01 -4.30638313e-01 3.50955188e-01 3.88157398e-01 -1.10275304e+00 1.70922592e-01 3.65939677e-01 -6.62794337e-02 1.32226557e-01 9.25542355e-01 5.72566450e-01 1.74816638e-01 -5.28658330e-01 -3.21422368e-01 3.03949267e-01 -2.16919258e-01 6.44581974e-01 7.84554839e-01 2.05000862e-01 2.25300834e-01 8.57776821e-01 2.17638686e-01 -1.17779821e-01 6.61008805e-02 -1.69292271e-01 4.70832996e-02 1.19877410e+00 9.80182707e-01 -8.42239201e-01 -3.73370230e-01 -3.81723255e-01 6.54615641e-01 4.41702485e-01 1.85735956e-01 -6.43549085e-01 -4.79337364e-01 8.97554159e-01 -1.48344442e-01 1.40993431e-01 -7.09733129e-01 3.31210077e-01 -5.12462318e-01 -1.19181678e-01 -8.39802146e-01 3.10567528e-01 -2.90313154e-01 -1.00582969e+00 3.55266094e-01 -8.65300521e-02 -9.57664847e-01 -6.81236625e-01 2.69606471e-01 -1.19628519e-01 2.44764537e-01 -1.93124378e+00 -7.61824429e-01 -7.71588013e-02 5.67596257e-01 8.12132835e-01 -2.55355537e-01 7.59935081e-01 2.78351218e-01 -7.23164007e-02 3.74590605e-01 2.62952238e-01 -4.46172476e-01 6.93776965e-01 -1.10479665e+00 -1.82878554e-01 1.57186195e-01 -3.11148584e-01 2.51059562e-01 9.05142128e-01 -2.78793067e-01 -2.08104825e+00 -6.01322889e-01 4.35534477e-01 5.67930281e-01 7.85355330e-01 6.69986978e-02 -5.47639966e-01 3.46127898e-01 5.87469459e-01 -2.42011905e-01 8.02623868e-01 -2.37440258e-01 1.17818214e-01 -2.42966160e-01 -1.16260827e+00 5.02293289e-01 4.37287778e-01 -6.14225008e-02 -3.07861920e-02 5.55584073e-01 6.56156659e-01 -4.55655187e-01 -1.12794399e+00 -1.84326366e-01 4.24864054e-01 -9.28518295e-01 6.51774347e-01 -2.47921914e-01 1.51288673e-01 -9.06542316e-02 -6.81010708e-02 -1.78350925e+00 1.24490611e-01 -1.03769743e+00 9.09759700e-02 7.94020653e-01 7.99019709e-02 -7.12198138e-01 9.13890243e-01 6.05103791e-01 1.19766869e-01 -6.75074041e-01 -1.11672449e+00 -6.37106478e-01 -1.97429657e-01 4.97468971e-02 7.79004812e-01 8.03206444e-01 6.39769197e-01 1.20050954e-02 -3.08141917e-01 7.16396868e-02 9.39290464e-01 1.95560247e-01 1.02368200e+00 -8.24705660e-01 -4.02563244e-01 -3.71817976e-01 3.23283002e-02 -1.32953322e+00 2.38338530e-01 -3.63266289e-01 -1.13784168e-02 -1.33628988e+00 1.04762994e-01 -1.03686929e+00 2.47021526e-01 2.68114805e-01 3.50349009e-01 5.94345592e-02 5.99339366e-01 4.70927685e-01 -1.09075499e+00 5.29522419e-01 1.22039199e+00 5.21337800e-02 -3.20469558e-01 -7.31141567e-02 -2.52748877e-01 3.06846052e-01 1.09136522e+00 -1.84209883e-01 -6.31985486e-01 -5.33513486e-01 1.43381834e-01 7.67635405e-01 2.39654109e-02 -7.68693089e-01 7.78361022e-01 -6.56958997e-01 -1.36252448e-01 -1.15814678e-01 3.94365251e-01 -8.83886039e-01 3.45133469e-02 8.75626981e-01 -8.08198094e-01 5.08961022e-01 -1.22542292e-01 7.72413135e-01 -3.18600573e-02 -1.79939076e-01 5.71478546e-01 -2.06702873e-01 -4.67973620e-01 1.29851058e-01 -1.02339745e+00 2.34862100e-02 1.37723708e+00 8.38277191e-02 -4.48466033e-01 -8.10124099e-01 -3.62038165e-01 7.34161317e-01 5.01411676e-01 1.73740789e-01 6.59225821e-01 -1.05228937e+00 -8.15855682e-01 2.42894202e-01 -2.97796160e-01 -2.90337861e-01 2.40828544e-01 6.53059006e-01 -5.27830541e-01 2.31437489e-01 -4.57471877e-01 -3.29299271e-01 -1.40595901e+00 2.83933908e-01 1.24736398e-01 -3.84073436e-01 -6.41305029e-01 3.23763758e-01 -3.85495543e-01 1.13832965e-01 5.29380560e-01 2.38220155e-01 2.37736758e-02 -6.56584874e-02 8.76023412e-01 6.54989243e-01 -9.91529524e-02 -1.24905154e-01 -2.35401653e-02 -7.59084756e-03 -1.54662430e-01 -5.03716469e-01 1.34481537e+00 -7.98076987e-01 -3.25889111e-01 -2.91157681e-02 7.47171879e-01 -1.07273847e-01 -1.50856435e+00 -3.16642135e-01 -2.37557113e-01 -8.25737357e-01 3.72947037e-01 -4.17125344e-01 -1.06548500e+00 8.62446576e-02 1.60564482e-01 1.01833797e+00 1.03176415e+00 -2.59119868e-01 1.17974234e+00 4.50702965e-01 8.16283405e-01 -1.39944828e+00 4.84639019e-01 2.66966462e-01 4.88201022e-01 -8.33289862e-01 1.42386675e-01 -2.02352315e-01 -5.75664401e-01 1.33482313e+00 5.04711747e-01 1.42451718e-01 5.18344820e-01 8.12628448e-01 4.09373902e-02 -1.50148320e-04 -1.26233029e+00 -3.87368016e-02 -9.08448935e-01 7.44902790e-01 1.75451651e-01 1.90144092e-01 -7.01613784e-01 -2.67185748e-01 -2.27041334e-01 -2.82476932e-01 1.21454895e+00 1.41008401e+00 -1.04573715e+00 -1.48336089e+00 -5.78584850e-01 4.77549851e-01 -3.55620652e-01 2.12506726e-01 2.95781732e-01 6.37086511e-01 -5.52645206e-01 1.43597138e+00 5.00285983e-01 3.42886865e-01 -6.37422577e-02 -5.17379761e-01 3.16854388e-01 -2.43613452e-01 -5.19250751e-01 1.56654343e-01 2.67335176e-01 -4.59912091e-01 -6.71057522e-01 -3.53079557e-01 -1.47722304e+00 -1.09542942e+00 -1.51207373e-01 5.24472952e-01 9.60401416e-01 7.43811548e-01 4.26532865e-01 6.40799046e-01 9.08764601e-01 -9.85635698e-01 -1.03711903e+00 -5.82776785e-01 -5.75813591e-01 3.52220125e-02 8.27195883e-01 -6.83429897e-01 -2.56518930e-01 -1.84402391e-01]
[3.94704008102417, 2.1835954189300537]
04797f33-00e9-487c-b389-bed6a135e076
attending-to-characters-in-neural-sequence
1611.04361
null
http://arxiv.org/abs/1611.04361v1
http://arxiv.org/pdf/1611.04361v1.pdf
Attending to Characters in Neural Sequence Labeling Models
Sequence labeling architectures use word embeddings for capturing similarity, but suffer when handling previously unseen or rare words. We investigate character-level extensions to such models and propose a novel architecture for combining alternative word representations. By using an attention mechanism, the model is able to dynamically decide how much information to use from a word- or character-level component. We evaluated different architectures on a range of sequence labeling datasets, and character-level extensions were found to improve performance on every benchmark. In addition, the proposed attention-based architecture delivered the best results even with a smaller number of trainable parameters.
['Gamal K. O. Crichton', 'Sampo Pyysalo', 'Marek Rei']
2016-11-14
attending-to-characters-in-neural-sequence-1
https://aclanthology.org/C16-1030
https://aclanthology.org/C16-1030.pdf
coling-2016-12
['grammatical-error-detection']
['natural-language-processing']
[ 3.78106892e-01 -2.07239494e-01 -3.15815866e-01 -3.21323335e-01 -7.22612321e-01 -6.71053290e-01 5.49720407e-01 4.99041736e-01 -1.02010477e+00 6.82066917e-01 2.92931855e-01 -4.62044984e-01 2.82673657e-01 -7.64505088e-01 -3.13022166e-01 -5.50309479e-01 2.96848221e-03 5.47613859e-01 4.23415154e-01 -2.96360016e-01 5.63115478e-01 5.66548288e-01 -1.39048576e+00 3.01701546e-01 6.52664244e-01 5.78160882e-01 4.92849529e-01 8.65561068e-01 -6.51247919e-01 3.56808752e-01 -6.42049193e-01 -3.40266734e-01 3.40260379e-02 -2.70648092e-01 -9.15877342e-01 -2.10262373e-01 3.27327281e-01 -1.00889221e-01 -2.35502020e-01 8.85523736e-01 6.31653070e-01 3.75978380e-01 5.00892460e-01 -7.20883787e-01 -1.06855774e+00 5.83038926e-01 -1.57185286e-01 6.68049812e-01 4.32363003e-01 4.16057050e-01 1.39950871e+00 -9.66821671e-01 6.41329587e-01 1.15936792e+00 5.51646948e-01 8.40565503e-01 -1.27531362e+00 -2.83591926e-01 4.73027557e-01 4.67482299e-01 -1.14676392e+00 -2.39903823e-01 4.01668280e-01 -2.77652860e-01 1.84401703e+00 1.18461363e-01 3.33735734e-01 1.28190458e+00 1.55426055e-01 6.62353456e-01 8.37614059e-01 -5.69982648e-01 2.72764534e-01 -1.17597975e-01 9.12890017e-01 5.97004950e-01 3.69438499e-01 -2.18602195e-01 -1.35531932e-01 -5.07016301e-01 2.55916208e-01 -5.70688434e-02 -3.07278782e-01 -4.58156690e-02 -1.05950999e+00 1.02681315e+00 2.71728516e-01 7.38089561e-01 -3.82067740e-01 1.61623791e-01 6.38225317e-01 3.36107314e-01 4.68919873e-01 7.83927023e-01 -7.24175632e-01 -2.13609830e-01 -4.63063926e-01 6.97580981e-04 6.59188688e-01 7.57590473e-01 6.48748696e-01 1.36221215e-01 -5.67694366e-01 9.85376418e-01 1.19387999e-01 -9.66007784e-02 1.17628944e+00 -2.87915140e-01 4.09490466e-01 3.94916147e-01 1.67794595e-03 -5.91984451e-01 -5.55302024e-01 -4.68477368e-01 -3.44617218e-01 -1.06193781e-01 1.73165545e-01 -9.69447345e-02 -1.37248683e+00 1.87442505e+00 -1.41019344e-01 3.85749787e-01 9.45831016e-02 7.45649338e-01 7.00550973e-01 7.93541074e-01 4.83208120e-01 -3.18052858e-04 1.60379231e+00 -1.08041000e+00 -6.92922413e-01 -5.63743532e-01 8.89843643e-01 -5.44188559e-01 1.11717367e+00 1.47159174e-01 -9.80590463e-01 -6.00782931e-01 -1.27104747e+00 -2.29267180e-01 -6.19548857e-01 -2.90873408e-01 2.70108879e-01 7.40265191e-01 -1.20777893e+00 6.84109271e-01 -5.93986571e-01 -6.12239003e-01 1.87481597e-01 1.64542824e-01 -2.74802774e-01 -2.51965314e-01 -1.47010601e+00 1.33523893e+00 8.43303978e-01 -6.43925443e-02 -7.46112168e-01 -5.75419486e-01 -9.63785052e-01 4.18897688e-01 1.08843341e-01 -5.68303168e-01 1.09847569e+00 -5.41989386e-01 -1.42340350e+00 8.10820282e-01 -2.97740817e-01 -6.38516366e-01 2.39345413e-02 -2.86041290e-01 -5.62889218e-01 2.04933852e-01 -3.00660402e-01 7.76091576e-01 6.06661618e-01 -8.33894610e-01 -2.74892420e-01 -1.30061001e-01 1.42880725e-02 -3.77700478e-02 -8.12794507e-01 1.31032228e-01 -2.63384759e-01 -8.27166200e-01 -3.30143541e-01 -8.73979628e-01 -5.08791566e-01 -1.88402504e-01 3.38592604e-02 -5.36557078e-01 5.68406880e-01 -4.54981297e-01 1.24650931e+00 -1.96519804e+00 1.51502803e-01 -2.09697589e-01 2.13168655e-02 8.69891107e-01 -8.99676621e-01 7.04403818e-01 -1.61421388e-01 4.91109967e-01 -3.11523348e-01 -3.04342210e-01 -6.06410354e-02 4.63079333e-01 -1.51256874e-01 2.06398010e-01 7.03645170e-01 9.83669102e-01 -1.05115724e+00 -2.13111877e-01 1.09377958e-01 5.07771790e-01 -5.31820714e-01 3.12072217e-01 -5.29524982e-01 -2.14194074e-01 -3.82816851e-01 3.12159359e-01 4.36324626e-01 -3.47200394e-01 4.58867848e-01 2.11351112e-01 2.72082150e-01 6.34756863e-01 -6.53056502e-01 1.80071485e+00 -7.22607434e-01 5.54976285e-01 -4.19958711e-01 -9.96614575e-01 1.15004599e+00 4.25396502e-01 -1.64314192e-02 -4.51962024e-01 1.73895404e-01 1.12092026e-01 4.69847172e-01 -4.38648403e-01 7.28720486e-01 -1.26407966e-01 -1.04902364e-01 6.05094373e-01 3.16737086e-01 3.66570473e-01 2.02780813e-01 4.44243476e-02 1.52928519e+00 1.12049282e-03 5.48641324e-01 -3.12955290e-01 7.25416422e-01 -3.07366773e-02 4.81963515e-01 6.24672294e-01 -3.08557659e-01 6.04481995e-01 4.64994699e-01 -7.74723232e-01 -1.04955661e+00 -7.74881184e-01 8.37934949e-03 1.59921265e+00 -5.61645702e-02 -5.13761997e-01 -4.77507204e-01 -8.36038828e-01 -9.25098732e-02 1.03175020e+00 -6.94052398e-01 -3.42135042e-01 -8.06691289e-01 -7.67430246e-01 5.69586337e-01 7.78080523e-01 -2.06689775e-01 -1.23439765e+00 -5.95406234e-01 6.45138085e-01 4.20684107e-02 -1.09853673e+00 -6.07368231e-01 5.99297225e-01 -8.62525165e-01 -7.43174970e-01 -8.57826233e-01 -9.96660352e-01 3.74664038e-01 1.66447923e-01 1.25125504e+00 3.58776420e-01 -3.74243408e-01 3.05383909e-03 -7.31376290e-01 -9.63077396e-02 -3.95242393e-01 6.23437226e-01 -9.17434245e-02 -1.86455071e-01 7.41047144e-01 -3.30892384e-01 -3.79330128e-01 1.31131515e-01 -9.49699283e-01 -6.54364765e-01 4.55491692e-01 1.04352546e+00 1.69763431e-01 -6.95870519e-01 8.57587039e-01 -8.26125085e-01 9.76782084e-01 -6.57293975e-01 -2.63031542e-01 2.74369985e-01 -5.37007987e-01 6.57578766e-01 8.55631530e-01 -6.36827648e-01 -7.43043005e-01 -1.15191832e-01 -6.98271155e-01 -3.24709624e-01 -2.59117514e-01 3.15679669e-01 3.46334949e-02 2.29442611e-01 5.93723536e-01 3.81301254e-01 -1.18399307e-01 -8.24040949e-01 6.16369486e-01 5.90024352e-01 1.59019098e-01 -5.64061105e-01 2.76734620e-01 -3.58730219e-02 -5.10250330e-01 -7.32236862e-01 -6.05765104e-01 -5.32680094e-01 -6.32922351e-01 5.11589468e-01 1.00464082e+00 -5.33186972e-01 -2.90850967e-01 1.15056597e-01 -1.54794061e+00 -1.01796672e-01 -1.19765408e-01 3.77157390e-01 -2.48740926e-01 5.55390894e-01 -8.20741773e-01 -4.86874849e-01 -3.56514990e-01 -1.21901298e+00 6.47967756e-01 -4.41865474e-02 -6.17234945e-01 -1.03815913e+00 4.71919566e-01 -7.63750225e-02 5.23413122e-01 -1.61737889e-01 1.31555712e+00 -1.45550025e+00 -8.26216564e-02 -3.31513017e-01 -8.55989661e-03 3.50176454e-01 8.68020207e-02 -8.33957195e-02 -1.09516096e+00 -3.04085106e-01 -1.77546978e-01 -4.48885828e-01 1.26655626e+00 -1.69934019e-01 1.22165930e+00 -1.12657316e-01 -3.38023752e-01 3.92015517e-01 1.39804387e+00 2.76293248e-01 6.80303454e-01 4.29607838e-01 5.03856122e-01 3.49738419e-01 1.68445677e-01 2.62214929e-01 2.36498192e-02 6.56018615e-01 4.15426522e-01 3.04120451e-01 -1.89506188e-01 -1.32910565e-01 2.91349024e-01 8.56027603e-01 2.50247240e-01 -8.27397943e-01 -8.78722370e-01 7.27724969e-01 -1.63672614e+00 -8.67882729e-01 8.47204402e-02 1.84193754e+00 8.69336724e-01 2.80229956e-01 3.12704295e-02 7.47267157e-03 7.49166548e-01 4.89220679e-01 -6.04962885e-01 -1.11045265e+00 -7.27716181e-03 5.99239469e-01 4.74654078e-01 4.43533689e-01 -7.43902981e-01 1.10796916e+00 7.52703047e+00 8.97666276e-01 -9.73720312e-01 2.05053374e-01 3.89982045e-01 4.13292088e-02 -6.51450515e-01 -3.26457918e-01 -9.16461051e-01 5.79958737e-01 1.28756130e+00 -1.13688625e-01 -1.42724618e-01 7.06621766e-01 -3.12328428e-01 2.61252314e-01 -1.08422673e+00 4.71582144e-01 2.61555523e-01 -1.38308382e+00 3.17086101e-01 -5.43710962e-02 3.99915427e-01 2.93391168e-01 -2.14935005e-01 4.28923488e-01 5.02004623e-01 -9.56130922e-01 3.16846669e-01 2.84631789e-01 5.14180124e-01 -7.67641306e-01 8.94011736e-01 1.94780618e-01 -1.00628495e+00 1.73150171e-02 -6.19423330e-01 -2.74233431e-01 3.00770879e-01 2.40806758e-01 -8.49963009e-01 3.84420395e-01 1.56224534e-01 5.42474866e-01 -5.47138810e-01 8.98941219e-01 -3.31898183e-01 5.96322894e-01 -7.03065172e-02 -4.71441358e-01 7.36770511e-01 6.62718788e-02 3.28537345e-01 1.78245342e+00 2.27101937e-01 9.23430026e-02 2.16742814e-01 4.37237084e-01 -1.31557584e-01 -7.14744721e-03 -5.32449484e-01 -2.23382756e-01 5.08324742e-01 1.27671504e+00 -7.38381267e-01 -5.37679136e-01 -5.29628038e-01 1.16245127e+00 9.52027738e-01 2.19913453e-01 -6.94340169e-01 -5.41394055e-01 1.17522502e+00 -1.88877881e-01 7.00903833e-01 -4.44809526e-01 -1.12124048e-01 -1.01607704e+00 -2.95432866e-01 -6.58231497e-01 7.36382902e-01 -4.74306166e-01 -1.42691934e+00 1.11470997e+00 -2.09751546e-01 -7.42257059e-01 -4.05088007e-01 -9.72586513e-01 -7.32211828e-01 9.69749451e-01 -1.65267205e+00 -5.32759607e-01 2.41240218e-01 3.69365327e-02 9.07739997e-01 -2.28407085e-01 1.24323809e+00 1.53797761e-01 -6.38211548e-01 7.69999743e-01 7.61409625e-02 1.41634360e-01 5.04709423e-01 -1.14257884e+00 1.02316177e+00 6.61651850e-01 4.48938459e-01 6.12837613e-01 5.59006989e-01 -3.93542618e-01 -1.23501348e+00 -9.26515460e-01 1.11711502e+00 -3.67327720e-01 6.92114532e-01 -5.41817069e-01 -1.48256254e+00 6.71729386e-01 6.13027036e-01 7.80305043e-02 1.02397859e+00 1.66961521e-01 -7.48747647e-01 4.88695145e-01 -1.06008363e+00 4.96866554e-01 1.18948984e+00 -5.11942327e-01 -1.13173425e+00 6.45662919e-02 1.16473138e+00 2.86051277e-02 -6.12909436e-01 5.84085472e-02 4.47746962e-01 -5.83048105e-01 8.69580865e-01 -1.12058556e+00 2.73936719e-01 -1.01715058e-01 -6.79200590e-02 -1.60575521e+00 -8.24052095e-01 -4.10732388e-01 -1.60615787e-01 1.00030828e+00 7.32706726e-01 -8.41190159e-01 6.63401425e-01 2.57504016e-01 -3.43394905e-01 -8.54602158e-01 -6.55837238e-01 -1.13371861e+00 4.21664506e-01 -3.42463672e-01 6.93892896e-01 8.85178983e-01 1.07973784e-01 5.82848430e-01 -2.94849843e-01 -1.06872447e-01 1.01464354e-01 -9.92379338e-02 4.49733771e-02 -8.71136606e-01 -2.06424147e-01 -6.77173734e-01 -6.37425721e-01 -1.12518489e+00 5.63318789e-01 -1.13483787e+00 2.60682195e-01 -1.52291334e+00 -1.81685872e-02 -1.36533096e-01 -8.41314614e-01 6.02716267e-01 -7.06123590e-01 3.46945375e-01 2.50345677e-01 -1.47912309e-01 -3.88759673e-01 5.23427427e-01 6.93889737e-01 -2.22562984e-01 1.26557246e-01 -5.01452744e-01 -6.06391549e-01 5.21300077e-01 1.16822433e+00 -5.77697098e-01 -3.32608372e-01 -9.15317416e-01 -3.19331177e-02 -4.21812057e-01 -1.33856386e-01 -7.73057759e-01 4.41723363e-03 -7.51875639e-02 2.42570013e-01 -4.89637285e-01 3.61923963e-01 -5.13344646e-01 -3.05358768e-01 7.46586800e-01 -8.11054766e-01 4.31965888e-01 3.31470937e-01 5.84418356e-01 -1.93704180e-02 -7.04526246e-01 7.84663260e-01 -1.79112583e-01 -1.04975295e+00 1.81411207e-01 -8.54749918e-01 1.75184965e-01 1.05952537e+00 -2.06973359e-01 -1.60306394e-01 6.16735294e-02 -6.10392988e-01 2.10163057e-01 4.01327521e-01 7.35534728e-01 6.03019059e-01 -1.27816010e+00 -7.35289931e-01 2.77366906e-01 5.59260190e-01 -7.18633533e-01 -1.07057534e-01 5.26343193e-03 -6.26611412e-01 4.87873018e-01 -3.08421254e-01 -2.67710149e-01 -1.03892624e+00 1.01443791e+00 1.67068839e-03 -4.08499330e-01 -5.07845402e-01 9.39750016e-01 -1.63520902e-01 -3.22603106e-01 2.56268114e-01 -4.95405585e-01 -3.73179108e-01 2.91238159e-01 8.95197272e-01 3.79214019e-01 2.53477246e-01 -4.82733935e-01 -6.74211621e-01 5.56774378e-01 -5.10120571e-01 5.76421656e-02 1.24534631e+00 7.67470375e-02 3.37559253e-01 4.14470404e-01 1.44751287e+00 -5.04228055e-01 -9.83171165e-01 -4.42359090e-01 4.83412236e-01 -3.89612764e-01 -1.67212173e-01 -6.77275419e-01 -7.74623990e-01 1.16002703e+00 3.46411735e-01 1.29674792e-01 7.60356724e-01 -1.57540038e-01 1.06729102e+00 4.74135011e-01 1.98234841e-01 -8.34461153e-01 8.73921323e-04 9.57699955e-01 4.92900312e-01 -1.19914854e+00 -2.17160389e-01 -2.54856739e-02 -5.93450010e-01 1.25444031e+00 6.75927520e-01 -3.23536098e-01 3.91865432e-01 3.54225427e-01 1.03039414e-01 -4.55754576e-03 -1.26251554e+00 -4.82370704e-01 2.07173020e-01 7.22683191e-01 7.69028783e-01 -1.75105378e-01 -7.07069993e-01 2.49465331e-01 3.36373419e-01 -2.81816870e-01 4.08792883e-01 1.00227606e+00 -8.27512860e-01 -1.49525464e+00 -7.76230777e-03 3.94769371e-01 -5.51735938e-01 -4.12803173e-01 -4.06336337e-01 4.51221019e-01 -1.33959934e-01 8.58487070e-01 4.21446919e-01 -1.96863785e-01 2.38621071e-01 5.71769476e-01 3.21866661e-01 -1.12318289e+00 -9.57824945e-01 -3.25813234e-01 2.71930903e-01 -4.81441826e-01 -2.15496808e-01 -2.85660565e-01 -1.21553779e+00 -3.69317271e-02 -2.38269106e-01 2.28626028e-01 3.44347179e-01 9.33730841e-01 4.42689389e-01 5.36110461e-01 4.22103554e-01 -5.54445028e-01 -7.98338592e-01 -1.14591300e+00 -1.28174737e-01 4.41478550e-01 2.93707341e-01 -3.87603700e-01 -6.36805519e-02 -2.76895136e-01]
[10.578577041625977, 8.68929386138916]
8c9d1ee0-d7b7-4d70-92eb-118d9ba2189f
a-new-network-based-algorithm-for-human
1502.06075
null
http://arxiv.org/abs/1502.06075v1
http://arxiv.org/pdf/1502.06075v1.pdf
A new network-based algorithm for human activity recognition in video
In this paper, a new network-transmission-based (NTB) algorithm is proposed for human activity recognition in videos. The proposed NTB algorithm models the entire scene as an error-free network. In this network, each node corresponds to a patch of the scene and each edge represents the activity correlation between the corresponding patches. Based on this network, we further model people in the scene as packages while human activities can be modeled as the process of package transmission in the network. By analyzing these specific "package transmission" processes, various activities can be effectively detected. The implementation of our NTB algorithm into abnormal activity detection and group activity recognition are described in detail in the paper. Experimental results demonstrate the effectiveness of our proposed algorithm.
['Bin Sheng', 'Weiyao Lin', 'Hongxiang Li', 'Jianxin Wu', 'Yuanzhe Chen', 'Hanli Wang']
2015-02-21
null
null
null
null
['activity-recognition-in-videos', 'group-activity-recognition']
['computer-vision', 'computer-vision']
[ 2.63947576e-01 -1.75139844e-01 -1.80542737e-01 4.19127569e-02 6.05757654e-01 -1.34618044e-01 2.46970281e-01 -4.06549051e-02 8.48073736e-02 2.24701285e-01 8.93732607e-02 1.13843277e-01 -2.52269953e-01 -1.04484355e+00 -4.59989548e-01 -7.25833476e-01 -4.63659376e-01 9.61585268e-02 5.96149862e-01 3.14011693e-01 -7.72401169e-02 5.36837578e-01 -9.86967981e-01 1.56629622e-01 5.19076884e-01 9.62686896e-01 2.56080121e-01 8.98067772e-01 2.51061078e-02 1.49870729e+00 -1.09448445e+00 -1.04496948e-01 -5.17301299e-02 -6.84561074e-01 -5.26029050e-01 7.95976281e-01 -2.76612192e-01 -3.81495059e-01 -8.27348650e-01 9.57399964e-01 4.74608876e-02 9.90007445e-02 4.30460274e-01 -1.71705317e+00 3.93060874e-03 3.03122371e-01 -6.83526754e-01 4.67887819e-01 7.88843989e-01 -2.44389504e-01 2.98110068e-01 -5.53309798e-01 4.87052917e-01 1.26706362e+00 6.62014306e-01 -1.91878027e-03 -9.25713420e-01 -4.76698339e-01 1.42338455e-01 4.81773794e-01 -1.34878039e+00 -6.82502910e-02 9.05825257e-01 -4.80281830e-01 8.95600379e-01 3.78344864e-01 1.47477043e+00 8.19301009e-01 4.31246668e-01 1.10044670e+00 4.59663242e-01 -3.04674506e-01 3.49474028e-02 -2.98258394e-01 3.72885942e-01 9.41760898e-01 5.88159621e-01 -4.34820026e-01 -6.71508968e-01 -2.41506293e-01 1.07829618e+00 4.51716334e-01 -3.31019431e-01 -3.32079381e-01 -1.13309062e+00 2.15590298e-01 -2.22577658e-02 2.39440605e-01 -4.10183519e-01 2.53501952e-01 4.06041354e-01 1.85690358e-01 4.49440658e-01 -2.98190773e-01 1.51997477e-01 -3.36985111e-01 -8.01622152e-01 -1.59673765e-01 1.04081464e+00 1.02277422e+00 5.31204104e-01 -8.06293339e-02 -8.42378363e-02 8.27982306e-01 5.64801872e-01 3.96754682e-01 9.12292395e-03 -9.74820435e-01 2.51478672e-01 9.27534163e-01 -4.12098914e-02 -1.77265728e+00 -2.40775913e-01 -2.70207852e-01 -9.76052344e-01 -1.30635157e-01 1.94876954e-01 -8.33554864e-02 -3.15240324e-01 1.12599146e+00 3.61054331e-01 7.47076631e-01 -2.50793159e-01 3.84356678e-01 4.93044376e-01 1.17597544e+00 -4.56648059e-02 -7.03850746e-01 1.28110135e+00 -1.38565767e+00 -1.17595780e+00 1.73378199e-01 5.33537090e-01 -5.05250871e-01 2.32352048e-01 7.26560831e-01 -8.69384348e-01 -5.82904935e-01 -1.05496395e+00 6.76910818e-01 -4.45981622e-02 9.49245542e-02 4.66884106e-01 7.43452847e-01 -1.16417277e+00 1.84500247e-01 -1.01394069e+00 -7.75487304e-01 4.18862998e-01 1.17909364e-01 -1.08176492e-01 -2.29063079e-01 -6.14642143e-01 2.01205388e-01 9.33981240e-02 4.41042036e-01 -1.19035709e+00 1.03402860e-01 -7.85534501e-01 9.93779525e-02 7.41042376e-01 -4.99894053e-01 9.97687459e-01 -1.15258276e+00 -8.16264749e-01 5.37501454e-01 -4.22175705e-01 -3.28637540e-01 4.34631974e-01 -7.88406655e-02 -8.13672125e-01 6.41128719e-01 -1.19967163e-01 -1.84765622e-01 6.27509236e-01 -1.07235909e+00 -6.45321906e-01 -9.16536525e-02 -5.33694495e-03 1.57436416e-01 -4.20973331e-01 5.31653523e-01 -1.06658709e+00 -8.51187587e-01 1.39326528e-01 -6.89611077e-01 -2.20942777e-02 2.83700138e-01 -3.71395260e-01 -1.89696833e-01 1.34284770e+00 -8.24950993e-01 1.70234966e+00 -2.43542266e+00 -3.08249086e-01 6.46961451e-01 2.88192064e-01 -4.43965420e-02 1.25875011e-01 6.53484285e-01 9.95036438e-02 2.11330201e-03 1.69474080e-01 -2.10411593e-01 -4.38876688e-01 4.86261636e-01 2.71459609e-01 7.49561906e-01 -4.08120394e-01 4.31133896e-01 -9.10442531e-01 -8.53736341e-01 2.59943873e-01 2.88496375e-01 -2.62492418e-01 4.00467575e-01 3.67956221e-01 2.20445499e-01 -6.23887837e-01 8.27087820e-01 8.12946320e-01 -2.13488385e-01 4.43916619e-01 -1.34952292e-01 -6.33972604e-03 -2.50932217e-01 -1.31626666e+00 1.04002583e+00 1.55294240e-01 8.99747074e-01 1.65861100e-01 -1.32935452e+00 7.04408824e-01 4.70039725e-01 8.24134648e-01 -2.27266416e-01 -3.53263132e-02 -3.58861655e-01 -1.14494920e-01 -8.26518536e-01 2.66106218e-01 7.57956684e-01 1.02674045e-01 5.13197601e-01 -2.32988913e-02 7.99702108e-01 4.11431968e-01 3.09387684e-01 1.69766402e+00 -2.88098305e-02 4.76370126e-01 1.97351794e-03 5.66118360e-01 -3.58453877e-02 8.01196456e-01 1.09407139e+00 -5.34205377e-01 -1.07016630e-01 6.17352188e-01 -2.62652606e-01 -5.90941787e-01 -1.13219011e+00 3.58433664e-01 7.39645600e-01 5.89492381e-01 -7.68091083e-01 -1.01204681e+00 -6.01004124e-01 -5.25441766e-01 5.61651681e-03 -4.42049176e-01 -1.36568872e-02 -6.09468937e-01 -6.86360717e-01 4.45432752e-01 4.34004754e-01 1.15550375e+00 -1.12715447e+00 -3.38365525e-01 3.35822403e-01 -4.77365017e-01 -1.27512562e+00 -4.23146725e-01 -5.30588746e-01 -1.00376165e+00 -1.45767331e+00 -2.76967436e-01 -1.17972457e+00 1.13009524e+00 7.72760391e-01 8.73478234e-01 5.95854819e-01 -4.20818508e-01 9.43561971e-01 -6.76262379e-01 -7.89049489e-04 -3.73230249e-01 -7.59364903e-01 -7.17754662e-02 6.05742991e-01 2.71018684e-01 -4.28888142e-01 -7.06256270e-01 7.00438678e-01 -5.88614166e-01 -2.40476690e-02 2.80570716e-01 3.98783922e-01 3.89252067e-01 1.07603121e+00 -1.07644118e-01 -5.29364109e-01 4.88167942e-01 -5.96534789e-01 -4.57235634e-01 4.83714789e-01 -4.10938412e-02 -8.79647970e-01 2.59766608e-01 -5.72484791e-01 -1.24173975e+00 5.34054860e-02 4.26479548e-01 -2.95236170e-01 -2.58627027e-01 4.60294724e-01 -5.39870441e-01 -6.82348609e-02 -1.34390011e-01 3.65674824e-01 6.98705157e-03 -3.86180103e-01 -3.86310577e-01 6.36337399e-01 3.98833454e-01 -2.53131092e-01 5.16749203e-01 6.40306413e-01 1.18707478e-01 -1.24160600e+00 -3.22148472e-01 -7.07755089e-01 -4.60635632e-01 -1.11551070e+00 1.03860211e+00 -9.18819308e-01 -9.22138631e-01 1.18196392e+00 -1.19096518e+00 -2.87741005e-01 1.53487697e-01 5.14169455e-01 -3.04293543e-01 7.71680593e-01 -8.56639028e-01 -1.29942465e+00 1.71817377e-01 -7.99381971e-01 6.32873654e-01 9.72163454e-02 -1.69300079e-01 -1.31544149e+00 -5.23258299e-02 3.69581521e-01 -1.52091876e-01 2.27553695e-01 4.28399295e-01 -4.59039211e-01 -9.74280298e-01 -4.43987966e-01 -4.85430704e-04 4.19522941e-01 3.87438774e-01 3.13322574e-01 -4.17859614e-01 -1.19150974e-01 4.41124171e-01 5.73425889e-01 2.95784116e-01 5.56806326e-01 1.52436471e+00 -3.68665069e-01 -9.10183132e-01 3.90477508e-01 1.15545297e+00 7.38413155e-01 9.28030252e-01 1.89197809e-01 7.87449360e-01 6.14920735e-01 8.11732948e-01 6.56093240e-01 -7.43388981e-02 6.51695669e-01 3.13748538e-01 -1.71231374e-01 -4.64384444e-02 -1.53236836e-01 6.36586785e-01 9.09947395e-01 -3.21315855e-01 -8.88587534e-01 -7.05023289e-01 3.96117389e-01 -2.04337287e+00 -1.48582518e+00 -6.15612984e-01 1.86029506e+00 -1.04511464e-02 2.34275535e-01 2.73373932e-01 4.58450317e-01 1.22252178e+00 1.66220143e-01 -6.56223446e-02 -2.16600634e-02 6.40117750e-02 -3.85066271e-01 3.52382660e-01 4.89880145e-02 -1.03500104e+00 3.42086434e-01 7.14553165e+00 9.73537982e-01 -2.95086384e-01 1.32303476e-01 4.75926071e-01 7.41742328e-02 5.36239803e-01 -1.44697502e-01 -5.81133068e-01 8.59064639e-01 4.31462884e-01 -1.62784860e-01 2.72981316e-01 7.77006567e-01 8.68591309e-01 -7.31685221e-01 -7.88063645e-01 1.25081360e+00 3.73145401e-01 -8.46039951e-01 2.59980410e-02 1.25779003e-01 2.51336962e-01 -6.39975190e-01 -4.70150411e-01 -2.34821737e-01 1.55587733e-01 -6.04184210e-01 4.77802485e-01 6.28925681e-01 1.63440958e-01 -6.53149605e-01 4.73426878e-01 5.42639136e-01 -1.72784686e+00 -3.89996618e-01 -1.45690307e-01 -2.24883184e-01 5.64599752e-01 7.23033190e-01 -5.19849837e-01 4.63449389e-01 1.01273227e+00 1.22248423e+00 -5.38623273e-01 1.37923539e+00 -1.85305178e-01 9.48452711e-01 -2.13145867e-01 -1.14243470e-01 -1.89482510e-01 -6.61900401e-01 7.62869477e-01 1.30831754e+00 6.40223742e-01 1.47544578e-01 4.64153707e-01 4.36781406e-01 1.03651598e-01 -2.22514898e-01 -8.07917237e-01 1.22157328e-01 4.54563230e-01 1.07506621e+00 -1.08972752e+00 -6.66448414e-01 -7.40835011e-01 1.16797030e+00 -2.54260242e-01 4.04868275e-01 -1.07188725e+00 -2.68495798e-01 4.04074639e-01 4.37433660e-01 1.95013225e-01 -2.44424298e-01 3.05420786e-01 -1.02800405e+00 -1.57839265e-02 -5.60246646e-01 4.98542279e-01 -1.19321597e+00 -9.87346470e-01 1.89549670e-01 2.06690371e-01 -1.54109836e+00 4.84772295e-01 -3.89850110e-01 -6.61601841e-01 2.01596305e-01 -5.60083389e-01 -7.55682170e-01 -7.78186381e-01 1.03203702e+00 6.35587573e-01 -1.67601988e-01 3.95165503e-01 3.83841842e-01 -9.61606801e-01 1.37046397e-01 2.06266847e-02 6.01589561e-01 1.08408310e-01 -7.10555732e-01 3.70970331e-02 1.35123956e+00 -2.10377891e-02 4.84920919e-01 5.81924021e-01 -1.14629173e+00 -1.01747787e+00 -1.23050344e+00 6.42978370e-01 -6.12392351e-02 5.93984783e-01 -4.87268180e-01 -6.13434911e-01 8.71997535e-01 3.82023782e-01 -1.55576766e-01 8.25772882e-01 -2.36298099e-01 1.80313542e-01 -3.25771034e-01 -1.01152921e+00 4.92772698e-01 1.55864012e+00 -3.12723368e-01 -5.34496903e-01 4.85150725e-01 3.76288801e-01 7.43419752e-02 -6.94462836e-01 2.16841102e-01 4.22405034e-01 -9.28348660e-01 8.99115503e-01 1.01927325e-01 -1.17157146e-01 -5.47208726e-01 2.77225047e-01 -1.00226712e+00 -4.57807451e-01 -5.43605864e-01 -5.63004851e-01 1.03566432e+00 -2.51377404e-01 -4.27152723e-01 7.34158695e-01 1.85595651e-03 -4.48718108e-02 -3.13986331e-01 -7.26124644e-01 -1.08142078e+00 -1.35697567e+00 -4.93963242e-01 3.33534598e-01 7.72647500e-01 1.91843569e-01 -2.79133189e-02 -6.84843242e-01 3.40459734e-01 8.63094926e-01 -3.88705194e-01 7.42790043e-01 -1.18234372e+00 -6.69499934e-01 -4.61179279e-02 -8.10842991e-01 -1.44541156e+00 -2.42075384e-01 -2.62135834e-01 -8.10893700e-02 -1.73157609e+00 7.49892235e-01 -6.74859732e-02 8.29338096e-03 2.57356435e-01 3.02770704e-01 2.03573272e-01 -8.14477354e-02 3.89537841e-01 -1.09686100e+00 7.76153151e-03 1.02147698e+00 -1.06429957e-01 -6.95787892e-02 1.54608980e-01 9.31377038e-02 9.29617703e-01 6.94777846e-01 -3.87684882e-01 -5.27507842e-01 -1.73972756e-01 1.31749853e-01 3.44889760e-01 4.79736805e-01 -1.21463966e+00 4.19779837e-01 1.26523618e-02 4.66147542e-01 -9.26772892e-01 4.05137300e-01 -1.22784925e+00 6.45825446e-01 9.81469452e-01 1.95301518e-01 5.38792498e-02 -2.68243343e-01 1.10545611e+00 -2.72426754e-01 -2.32401222e-01 2.55087256e-01 -2.42000688e-02 -7.89087951e-01 3.31365645e-01 -1.16522801e+00 -7.25464880e-01 1.44623590e+00 -8.79925787e-01 -4.13712382e-01 -5.35624921e-01 -1.07175410e+00 1.97862461e-01 4.38421100e-01 1.20919727e-01 7.14140713e-01 -1.28771091e+00 -5.51714981e-03 2.74422407e-01 4.05961610e-02 -2.74948508e-01 4.27625149e-01 1.15396857e+00 -1.03551435e+00 -9.66515765e-02 -9.86635387e-02 -6.70567989e-01 -1.98813093e+00 5.97463369e-01 4.45784599e-01 -3.17148685e-01 -3.99925202e-01 3.80015284e-01 3.43245029e-01 3.21041018e-01 4.11678404e-01 -7.03071728e-02 -5.55026352e-01 -3.58173966e-01 6.34619355e-01 1.01778495e+00 -4.33365941e-01 -7.76735127e-01 -2.57334799e-01 4.48422343e-01 3.56436968e-01 2.11092293e-01 1.05291641e+00 -4.09414202e-01 -6.41971767e-01 3.17806333e-01 9.31770623e-01 -1.24887392e-01 -9.34999108e-01 1.34343892e-01 -8.89573768e-02 -7.55129397e-01 -2.62339830e-01 -3.76552790e-01 -1.23072696e+00 3.61487865e-01 5.02195179e-01 6.96031630e-01 1.46978092e+00 2.84094419e-02 5.42063653e-01 2.26720944e-01 5.62811494e-01 -1.22306967e+00 6.47435367e-01 1.40844271e-01 3.72971028e-01 -3.37145209e-01 -6.97450489e-02 -1.14344275e+00 -1.63424388e-01 9.25067484e-01 6.33140147e-01 -1.91226259e-01 9.12192166e-01 3.81314576e-01 -5.69716096e-01 -4.03399885e-01 -5.32202840e-01 1.67964086e-01 -3.24828595e-01 8.69458914e-01 -1.43223375e-01 -1.17429093e-01 -3.44948292e-01 3.65189373e-01 4.89992291e-01 2.03889698e-01 8.61229062e-01 1.18935144e+00 -6.56147599e-01 -8.32015157e-01 -8.25411141e-01 5.03449440e-01 -3.89809847e-01 4.46523309e-01 -2.61718839e-01 6.07212365e-01 2.00324059e-01 1.30615711e+00 5.56810141e-01 -4.31672275e-01 4.60238427e-01 -3.17120463e-01 5.25265694e-01 -5.06338537e-01 -1.87586144e-01 5.57260990e-01 2.40567029e-01 -9.04435337e-01 -9.82173622e-01 -8.59362841e-01 -8.13486695e-01 -5.18184841e-01 -1.91775173e-01 4.78647575e-02 1.46685779e-01 6.21973813e-01 9.42585766e-02 7.15120494e-01 7.57565141e-01 -2.59937793e-01 3.99992704e-01 -8.48491490e-01 -1.14065242e+00 3.54551822e-01 -8.59894231e-02 -7.43550837e-01 -4.49542999e-01 4.79709834e-01]
[8.359685897827148, 0.4949393570423126]
bb12f7f7-352b-4f1e-9581-abeeefb25e8b
rethinking-privacy-preserving-deep-learning
2006.11601
null
https://arxiv.org/abs/2006.11601v2
https://arxiv.org/pdf/2006.11601v2.pdf
Rethinking Privacy Preserving Deep Learning: How to Evaluate and Thwart Privacy Attacks
This paper investigates capabilities of Privacy-Preserving Deep Learning (PPDL) mechanisms against various forms of privacy attacks. First, we propose to quantitatively measure the trade-off between model accuracy and privacy losses incurred by reconstruction, tracing and membership attacks. Second, we formulate reconstruction attacks as solving a noisy system of linear equations, and prove that attacks are guaranteed to be defeated if condition (2) is unfulfilled. Third, based on theoretical analysis, a novel Secret Polarization Network (SPN) is proposed to thwart privacy attacks, which pose serious challenges to existing PPDL methods. Extensive experiments showed that model accuracies are improved on average by 5-20% compared with baseline mechanisms, in regimes where data privacy are satisfactorily protected.
['Kam Woh Ng', 'Tianyu Zhang', 'Chee Seng Chan', 'Chang Liu', 'Ce Ju', 'Qiang Yang', 'Lixin Fan']
2020-06-20
null
null
null
null
['privacy-preserving-deep-learning', 'privacy-preserving-deep-learning']
['methodology', 'natural-language-processing']
[ 2.78270274e-01 3.61478239e-01 -8.46193824e-03 -4.50600743e-01 -9.50766027e-01 -1.10611463e+00 5.61731577e-01 1.73931852e-01 -5.71063817e-01 9.47846472e-01 -1.14782825e-01 -6.25303388e-01 1.42798834e-02 -7.15814114e-01 -1.09905636e+00 -9.90696192e-01 -1.61698923e-01 -2.24261478e-01 -2.07075492e-01 2.06133947e-01 2.89958209e-01 9.31650102e-01 -9.59980726e-01 1.36456445e-01 5.86694181e-01 1.23929965e+00 -8.19362938e-01 3.54461253e-01 2.66477019e-01 5.12989044e-01 -5.09616494e-01 -9.36898053e-01 8.79159093e-01 1.08991098e-02 -7.22726405e-01 -4.04354513e-01 7.49558806e-01 -8.01710188e-01 -6.79611206e-01 1.67223597e+00 2.62083560e-01 -3.32482338e-01 2.25244135e-01 -1.73187625e+00 -3.08808446e-01 6.36579871e-01 -4.17459249e-01 -1.73951760e-01 -2.01031685e-01 6.01956025e-02 9.72534180e-01 -3.46139401e-01 4.45151120e-01 9.58972275e-01 8.66334617e-01 8.38171959e-01 -1.70157933e+00 -1.38379216e+00 -6.73836991e-02 -1.95832744e-01 -1.22063696e+00 -7.86660314e-01 5.04872203e-01 -9.14974436e-02 4.55315292e-01 5.91369808e-01 2.91346341e-01 1.37471199e+00 5.33088505e-01 4.80723202e-01 1.25431800e+00 3.87609727e-03 2.39985287e-01 5.53300261e-01 3.78365308e-01 3.58154058e-01 9.99533117e-01 3.66781056e-01 -6.67478144e-01 -8.81893516e-01 6.18172050e-01 -4.45466399e-01 -6.04456186e-01 -7.25984752e-01 -3.16017687e-01 9.03710902e-01 1.02236401e-02 -1.90180108e-01 8.30773450e-03 1.79276049e-01 4.59022373e-01 7.44330943e-01 3.11926812e-01 4.13880050e-01 -5.25123775e-01 3.10810775e-01 -6.36134982e-01 5.48657477e-01 1.02681887e+00 9.59803820e-01 5.68287313e-01 -1.53675526e-01 1.47993546e-02 1.88962668e-01 3.73994648e-01 3.49886656e-01 -1.54390574e-01 -1.33422554e+00 6.22751057e-01 3.26163247e-02 3.82917464e-01 -1.25778425e+00 7.82635137e-02 -5.38535357e-01 -9.99673307e-01 1.82314560e-01 4.49158460e-01 -4.95812833e-01 -2.40139589e-01 2.14111495e+00 1.00078799e-01 -2.58673890e-03 4.57838565e-01 4.56199139e-01 3.59672129e-01 4.73081976e-01 1.16174646e-01 -4.56488907e-01 1.18523407e+00 -1.93177789e-01 -6.50698602e-01 -6.14481680e-02 6.75737321e-01 -1.94893152e-01 5.92344880e-01 3.33738267e-01 -1.08870614e+00 1.26290366e-01 -1.17557597e+00 -1.46515682e-01 4.92142737e-02 -5.53032100e-01 7.14467525e-01 1.40404105e+00 -1.16129386e+00 6.10753536e-01 -8.24496627e-01 5.50290905e-02 1.02751338e+00 7.09025383e-01 -7.70343900e-01 9.80136469e-02 -1.26403511e+00 1.65497482e-01 2.40908235e-01 -1.17671061e-02 -1.01972210e+00 -1.02073050e+00 -5.67652702e-01 3.36909622e-01 -1.84893552e-02 -4.32244658e-01 1.08226204e+00 -5.39463282e-01 -1.08602262e+00 1.20873618e+00 1.52902409e-01 -1.24079418e+00 1.12413907e+00 -2.34816015e-01 -3.07286143e-01 1.56754673e-01 -4.04348046e-01 3.72626394e-01 5.22253275e-01 -1.62175202e+00 -6.04982436e-01 -6.22391462e-01 2.09666744e-01 -1.82205051e-01 -7.67147660e-01 2.90830750e-02 -7.35083893e-02 -3.33755642e-01 6.57782555e-02 -1.04151738e+00 -3.00062865e-01 4.30194080e-01 -6.81109369e-01 3.87648165e-01 1.15553355e+00 -5.38077354e-01 7.70036340e-01 -2.31680298e+00 -6.92542434e-01 5.09607911e-01 5.15313327e-01 5.34956574e-01 -1.39992032e-02 2.37820804e-01 2.63735414e-01 4.93816853e-01 -4.27782416e-01 -7.09382415e-01 2.60802925e-01 2.00615540e-01 -7.96023965e-01 9.19034064e-01 -3.80320162e-01 6.16352439e-01 -2.93324679e-01 -1.83823586e-01 -3.29925299e-01 5.59816301e-01 -7.08659708e-01 -2.52408138e-03 -6.00679293e-02 4.61336255e-01 -4.45049852e-01 4.03281748e-01 1.47093153e+00 -4.87690270e-02 5.74104786e-01 1.72345042e-02 1.86310440e-01 2.87304491e-01 -9.36766863e-01 9.33582008e-01 5.27554899e-02 7.62820899e-01 5.88128865e-01 -7.42132127e-01 7.46454537e-01 2.63240546e-01 3.40555400e-01 -5.04935086e-01 3.99494767e-01 6.48517981e-02 -3.61224681e-01 -5.03805354e-02 2.45034128e-01 -7.67108798e-03 -1.20914601e-01 4.72352296e-01 -2.61754513e-01 3.60703081e-01 -7.89231479e-01 7.28468075e-02 1.08307767e+00 -6.47001863e-01 -1.54070854e-01 -6.82523906e-01 3.79201621e-01 -3.65427971e-01 1.06375432e+00 1.25043833e+00 -6.21661484e-01 1.89213961e-01 1.13641298e+00 -3.15838426e-01 -8.41798663e-01 -1.07941794e+00 -4.67215627e-01 6.06634676e-01 1.50717065e-01 -3.24962229e-01 -1.05962992e+00 -9.23909783e-01 4.30595815e-01 6.11556411e-01 -4.12187785e-01 -2.47666717e-01 -3.66147280e-01 -9.68849838e-01 1.25666142e+00 1.91464096e-01 1.16675496e+00 -2.92495221e-01 -2.18626469e-01 -2.11610034e-01 -9.27935466e-02 -1.36375499e+00 -2.03990117e-01 5.65269217e-02 -9.95541930e-01 -9.70506191e-01 -4.08931077e-02 -4.79777962e-01 6.57691479e-01 2.91173518e-01 6.25599623e-01 9.00389403e-02 2.54620254e-01 2.82728344e-01 3.30407351e-01 -3.73080939e-01 -5.00195920e-01 1.35152027e-01 1.79322243e-01 1.55680493e-01 5.25063396e-01 -8.66231978e-01 -4.97157574e-01 3.49215716e-02 -8.56520593e-01 -3.86147141e-01 2.63428152e-01 5.57924509e-01 5.70688307e-01 2.34499946e-01 2.59510517e-01 -1.31826389e+00 6.54852867e-01 -1.79067343e-01 -1.18441212e+00 2.89769292e-01 -9.42902625e-01 9.42398757e-02 6.33678794e-01 -2.31135592e-01 -1.02805388e+00 1.26651913e-01 -2.36395121e-01 -5.45325696e-01 -7.00645968e-02 -3.67434509e-02 -7.30493248e-01 -8.96664858e-01 4.38199371e-01 3.54572922e-01 1.74083680e-01 -4.70832258e-01 1.27088070e-01 4.97316867e-01 6.01811171e-01 -8.58674109e-01 8.99563313e-01 6.76000357e-01 4.67925817e-01 -6.39706314e-01 -8.00292969e-01 3.27720314e-01 -2.62638688e-01 2.57994145e-01 4.91672248e-01 -9.30928230e-01 -1.41798663e+00 8.18200469e-01 -1.04973340e+00 -4.09032404e-02 1.12627901e-01 2.37015396e-01 -3.09778720e-01 7.47259021e-01 -9.07191217e-01 -9.90262449e-01 -5.03922880e-01 -8.94948006e-01 4.14575249e-01 1.13925040e-01 2.40783811e-01 -9.66820538e-01 -3.90163749e-01 5.29993713e-01 3.45114022e-01 6.81401074e-01 7.95367181e-01 -9.87433255e-01 -7.71313071e-01 -4.00221229e-01 -3.80886316e-01 5.32195091e-01 -4.05752957e-01 -3.02156955e-01 -1.33988595e+00 -7.80128181e-01 5.97975671e-01 -4.04049605e-01 7.99552917e-01 1.07098952e-01 1.70477331e+00 -1.16532063e+00 -2.96436220e-01 1.19069123e+00 1.51506400e+00 5.78873977e-02 7.24626541e-01 2.69790560e-01 4.85915929e-01 6.48093581e-01 -1.78891618e-03 6.17365837e-01 3.27802241e-01 2.62421876e-01 6.25393987e-01 2.26131558e-01 6.44427657e-01 -4.73573089e-01 1.28383279e-01 1.16088122e-01 5.67680895e-01 -1.93195522e-01 -6.43159926e-01 2.30397403e-01 -1.65664113e+00 -8.30659091e-01 -3.78337592e-01 2.42733455e+00 9.35901999e-01 3.56687933e-01 -1.25002056e-01 -1.34996958e-02 3.66524011e-01 3.72055560e-01 -6.83461070e-01 -3.99118632e-01 -4.57236081e-01 -2.41179820e-02 1.33791459e+00 5.56496918e-01 -1.26012552e+00 8.90691996e-01 7.05953074e+00 5.26849627e-01 -1.04414737e+00 1.29517019e-01 1.05268526e+00 -1.09753050e-01 -4.48983997e-01 7.72514716e-02 -9.75352108e-01 4.24077570e-01 9.15080488e-01 -5.06538332e-01 3.55504423e-01 7.81876028e-01 -2.71922320e-01 4.32335854e-01 -1.14601243e+00 7.79423058e-01 -2.72476107e-01 -1.44276845e+00 -1.07126739e-02 7.87399173e-01 5.91560364e-01 -1.35526851e-01 5.15217543e-01 -2.57355813e-02 8.30184281e-01 -9.37894881e-01 3.74519914e-01 3.18043649e-01 6.61084414e-01 -1.21718776e+00 4.81272817e-01 4.08173203e-01 -4.21247572e-01 -3.32959294e-01 -6.48242593e-01 1.35760099e-01 -3.07572722e-01 6.39371157e-01 -2.86643535e-01 5.31642258e-01 6.86232090e-01 1.48804545e-01 -1.92960560e-01 6.61316931e-01 -2.63884276e-01 8.05906892e-01 -4.31496233e-01 4.53700483e-01 1.61275193e-01 -1.13746464e-01 6.12226903e-01 9.57015514e-01 1.61256209e-01 1.39699280e-01 -4.19248998e-01 1.07411325e+00 -7.26279318e-01 -3.39881837e-01 -6.27337754e-01 3.49638308e-03 1.02196825e+00 9.24601972e-01 -2.81326845e-02 1.58564791e-01 -1.50036681e-02 8.91558945e-01 1.59820989e-01 2.54241645e-01 -5.50715744e-01 -2.29652181e-01 1.40877974e+00 -1.14250835e-02 1.64612532e-02 -1.84780121e-01 -7.45684206e-01 -1.26010501e+00 8.12641829e-02 -9.14285600e-01 4.76288646e-01 6.66668490e-02 -1.51848066e+00 2.21485093e-01 -4.03050005e-01 -7.67769575e-01 4.05047596e-01 -3.15124482e-01 -3.33086997e-01 8.48915637e-01 -1.51337826e+00 -8.37322950e-01 2.54126430e-01 7.01449752e-01 -6.10236466e-01 -1.72707006e-01 1.03443027e+00 2.80099332e-01 -6.88351870e-01 1.48923314e+00 5.62403798e-01 5.01030624e-01 3.28445941e-01 -8.32767606e-01 5.86678863e-01 1.08611548e+00 8.41811448e-02 8.60524654e-01 7.53028989e-01 -4.22337145e-01 -1.60877228e+00 -1.28779471e+00 1.03643429e+00 -2.62396455e-01 3.21194351e-01 -9.39172804e-01 -1.15392971e+00 9.04703200e-01 -1.37270629e-01 3.97147030e-01 9.23984885e-01 -1.02181792e-01 -8.90489519e-01 -3.70186299e-01 -1.93561482e+00 4.24784184e-01 9.19664979e-01 -8.00797343e-01 6.92476425e-03 2.38379732e-01 8.84479642e-01 -4.40436244e-01 -7.87993371e-01 3.53317827e-01 7.66311705e-01 -1.13925028e+00 9.99629319e-01 -8.57417047e-01 -3.75798643e-02 1.92913786e-01 -6.51556134e-01 -5.92654049e-01 -2.78659537e-03 -1.07098019e+00 -3.19105715e-01 1.29607916e+00 2.63438612e-01 -1.09848773e+00 1.34070110e+00 1.17314887e+00 5.67842424e-01 -2.62764931e-01 -1.22259486e+00 -1.01860678e+00 5.41474938e-01 -3.57589573e-01 8.71519566e-01 1.37353420e+00 -4.25453752e-01 -3.87023479e-01 -6.22764766e-01 7.95279264e-01 1.34963298e+00 -2.96866447e-01 7.82556117e-01 -1.35245359e+00 -1.83327585e-01 -1.16895132e-01 -2.57004648e-01 -8.29806805e-01 5.54923177e-01 -7.59275854e-01 -4.46425140e-01 -8.05471838e-01 3.59387062e-02 -4.59638476e-01 -6.95975780e-01 5.82211256e-01 3.39625746e-01 1.22840300e-01 3.61763358e-01 3.15759271e-01 -1.72634289e-01 4.40026194e-01 7.19295084e-01 5.82943782e-02 2.85731498e-02 3.81515443e-01 -1.29656613e+00 4.09166396e-01 8.66417408e-01 -8.27310503e-01 -4.64160949e-01 -3.55357200e-01 1.37617692e-01 2.60333121e-01 6.62665248e-01 -9.69776869e-01 2.56703764e-01 -2.08531141e-01 1.53936177e-01 -3.05136889e-01 3.42578083e-01 -1.12257493e+00 3.56433123e-01 6.23862386e-01 -6.32999480e-01 -4.56421703e-01 1.80381775e-01 6.65796041e-01 2.22630277e-01 2.07589611e-01 1.15562212e+00 2.41213813e-01 2.24835109e-02 5.13139725e-01 -1.61541253e-01 3.53071396e-03 1.04033267e+00 5.37940301e-03 -6.57164693e-01 -3.81389618e-01 -5.46103716e-01 3.01898479e-01 5.93237340e-01 -3.54055054e-02 2.42987469e-01 -1.16135144e+00 -5.97914815e-01 4.83234644e-01 -1.51233658e-01 -2.76255786e-01 2.79346704e-01 4.48466003e-01 -3.67103726e-01 3.87850314e-01 -2.08066821e-01 -1.63836092e-01 -1.56573689e+00 3.73944581e-01 6.43944919e-01 -2.64465153e-01 -7.72853851e-01 1.14716673e+00 1.40088692e-01 -3.08838964e-01 8.35728943e-01 1.10584512e-01 4.67849672e-01 -2.08514899e-01 5.41531920e-01 8.09558183e-02 5.33746555e-02 -3.90079737e-01 -3.21068257e-01 -4.91814837e-02 -3.48479688e-01 -1.14893578e-01 1.27065349e+00 -1.81318194e-01 -8.28210562e-02 -2.93164283e-01 1.57813334e+00 9.78062674e-02 -1.56589842e+00 -4.69934285e-01 -1.38395727e-01 -8.54996443e-01 1.17239289e-01 -6.93216801e-01 -1.28056610e+00 8.38536859e-01 5.60714602e-01 3.79899472e-01 1.02838695e+00 -4.51233327e-01 1.05420732e+00 6.26313686e-01 5.64428329e-01 -7.53362060e-01 -7.09136426e-01 3.24737459e-01 4.34149116e-01 -9.90615964e-01 -1.26285031e-01 -3.95793587e-01 -3.94403845e-01 6.74975038e-01 4.68791336e-01 -3.95492911e-02 8.76723528e-01 5.57071865e-01 -8.77179652e-02 2.02611789e-01 -8.10879648e-01 8.58876526e-01 -2.00534314e-01 6.54264629e-01 -2.38641277e-01 8.98367241e-02 -2.44639292e-01 1.03072381e+00 -4.68152791e-01 -1.18472651e-01 5.35024285e-01 8.88149440e-01 -2.75855690e-01 -1.24379992e+00 -2.48983771e-01 2.31018171e-01 -1.02951217e+00 2.53337026e-01 -6.08035684e-01 5.95973432e-01 -8.14289302e-02 7.99938500e-01 -2.87403911e-02 -5.35207510e-01 1.31520983e-02 -1.82131276e-01 1.60009935e-01 5.81519268e-02 -6.65681779e-01 -2.89753795e-01 9.89768207e-02 -6.51759446e-01 4.84705493e-02 -7.02054143e-01 -8.61885905e-01 -9.11199331e-01 -9.08285845e-03 1.08113647e-01 5.52634299e-01 6.21387482e-01 6.60821974e-01 -2.35228121e-01 9.52703536e-01 5.69488965e-02 -1.16477597e+00 -9.04743969e-02 -9.36152756e-01 8.98238644e-02 5.96783757e-01 -1.27312392e-01 -8.08138311e-01 -3.83862078e-01]
[5.919155597686768, 6.968176364898682]
43aa4f84-92ae-4cd4-b399-19f570c22789
one-system-to-rule-them-all-a-universal
2112.08261
null
https://arxiv.org/abs/2112.08261v1
https://arxiv.org/pdf/2112.08261v1.pdf
One System to Rule them All: a Universal Intent Recognition System for Customer Service Chatbots
Customer service chatbots are conversational systems designed to provide information to customers about products/services offered by different companies. Particularly, intent recognition is one of the core components in the natural language understating capabilities of a chatbot system. Among the different intents that a chatbot is trained to recognize, there is a set of them that is universal to any customer service chatbot. Universal intents may include salutation, switch the conversation to a human agent, farewells, among others. A system to recognize those universal intents will be very helpful to optimize the training process of specific customer service chatbots. We propose the development of a universal intent recognition system, which is trained to recognize a selected group of 11 intents that are common in 28 different chatbots. The proposed system is trained considering state-of-the-art word-embedding models such as word2vec and BERT, and deep classifiers based on convolutional and recurrent neural networks. The proposed model is able to discriminate between those universal intents with a balanced accuracy up to 80.4\%. In addition, the proposed system is equally accurate to recognize intents expressed both in short and long text requests. At the same time, misclassification errors often occurs between intents with very similar semantic fields such as farewells and positive comments. The proposed system will be very helpful to optimize the training process of a customer service chatbot because some of the intents will be already available and detected by our system. At the same time, the proposed approach will be a suitable base model to train more specific chatbots by applying transfer learning strategies.
['Andres Felipe Tejada-Castro', 'Juan Esteban Jaramillo', 'Jose Luis Pemberty-Tamayo', 'Juan Carlos Guerrero-Sierra', 'Juan Camilo Vasquez-Correa']
2021-12-15
null
null
null
null
['intent-recognition']
['natural-language-processing']
[-1.79376096e-01 3.68926935e-02 1.11861512e-01 -5.34053028e-01 -3.22234184e-01 -5.16341627e-01 5.70731044e-01 -6.63357154e-02 -3.37580323e-01 4.06446666e-01 6.47971258e-02 -3.85060072e-01 1.86650723e-01 -7.72397101e-01 1.25339895e-01 -6.34755254e-01 4.02704060e-01 8.39582562e-01 1.23981439e-01 -8.29063654e-01 5.39311171e-01 3.55720967e-01 -1.34933400e+00 7.58220911e-01 7.16955483e-01 9.16061163e-01 7.07446814e-01 6.82941556e-01 -8.95855844e-01 1.35995913e+00 -9.10805166e-01 -6.05461955e-01 -1.59876794e-01 -3.33960593e-01 -1.20171309e+00 -9.43964794e-02 -3.05365384e-01 -1.49360240e-01 -7.30104893e-02 7.29565918e-01 3.32890093e-01 1.65661499e-01 6.35432184e-01 -1.49145365e+00 -6.57993972e-01 7.26526737e-01 2.39412755e-01 9.62808356e-02 5.59494138e-01 2.26122156e-01 1.30909109e+00 -7.56236196e-01 4.55185384e-01 1.11763752e+00 5.18522024e-01 6.12601101e-01 -6.49739087e-01 -6.13754988e-01 -4.64392193e-02 6.16325796e-01 -1.07612658e+00 -9.18804780e-02 9.79741096e-01 -3.86127979e-01 1.09995675e+00 4.20802236e-01 2.67241985e-01 1.18004656e+00 4.33634818e-01 8.60070825e-01 7.98332810e-01 -1.98207840e-01 1.68684855e-01 7.95538127e-01 5.86034417e-01 3.90278161e-01 -2.78796971e-01 -6.25938237e-01 -2.38318637e-01 -1.36409104e-01 2.65250593e-01 3.96049023e-01 3.19829918e-02 2.54376590e-01 -5.29632688e-01 1.18275213e+00 4.68182057e-01 8.79242301e-01 -3.09609056e-01 -1.74214855e-01 7.88219094e-01 5.43761730e-01 3.65914673e-01 5.50590098e-01 -4.63176727e-01 -5.46653152e-01 -2.48602390e-01 3.61781120e-02 1.38877165e+00 1.11231673e+00 9.59324658e-01 -1.55358255e-01 -2.00562403e-01 1.13397193e+00 2.40714550e-01 5.50239533e-02 8.21182847e-01 -4.33219612e-01 4.19405937e-01 1.16105700e+00 8.80428553e-02 -9.88684893e-01 -4.83601511e-01 -1.29091009e-01 -5.97295642e-01 -1.05516650e-01 1.81603834e-01 -1.81607619e-01 -4.89243746e-01 1.03146768e+00 -3.89856249e-02 -8.96984264e-02 1.85046673e-01 8.08143377e-01 1.22414994e+00 7.84759820e-01 -3.13647017e-02 5.15522212e-02 1.68598568e+00 -1.11620760e+00 -1.00073540e+00 -3.74585837e-01 9.54022646e-01 -1.17820978e+00 1.33848262e+00 8.12863037e-02 -3.64325553e-01 -4.39778775e-01 -6.37651145e-01 -2.21322794e-02 -8.07039022e-01 1.21895939e-01 8.20146561e-01 7.81885624e-01 -6.60552442e-01 3.10662478e-01 -3.45866024e-01 -7.06655204e-01 -1.21579096e-01 3.75571251e-01 -1.79171368e-01 -1.44226700e-01 -1.34747183e+00 1.22908139e+00 1.96706623e-01 1.96978062e-01 -6.72883809e-01 -4.31534231e-01 -8.07102382e-01 3.09917569e-01 1.19105726e-01 -1.68074086e-01 1.42891192e+00 -9.97669160e-01 -1.64158392e+00 6.50179803e-01 -1.92225143e-01 -3.72033119e-01 1.47035733e-01 1.77388906e-01 -5.76155007e-01 -1.63697854e-01 1.35818660e-01 8.60059708e-02 5.76112032e-01 -8.61887574e-01 -7.06461489e-01 -1.16263479e-01 4.24446911e-01 2.42137045e-01 -5.80749393e-01 3.74745011e-01 -5.96906543e-02 -5.55307083e-02 -2.56930709e-01 -8.44681859e-01 1.08653437e-02 -5.01634657e-01 -2.31440768e-01 -8.07084858e-01 1.25667095e+00 -5.55474341e-01 8.83041322e-01 -1.95663190e+00 -4.56865013e-01 -2.65703589e-01 -4.70624715e-02 4.99607921e-01 -2.53917813e-01 1.04171968e+00 1.49537534e-01 -4.91960254e-03 2.74052560e-01 -4.27325070e-01 8.50415006e-02 3.18263292e-01 -1.51748568e-01 3.90168317e-02 3.10399354e-01 6.89137995e-01 -6.82560146e-01 -3.04445386e-01 4.41192985e-01 3.24559271e-01 -3.44820201e-01 6.94083154e-01 -1.42628670e-01 2.77406573e-01 -5.40380836e-01 4.76044148e-01 5.03076851e-01 -1.08492851e-01 2.22098082e-01 -8.49196166e-02 -1.18780069e-01 8.38877976e-01 -6.86044812e-01 1.21269608e+00 -1.17896664e+00 5.83395720e-01 1.19050674e-01 -1.26988876e+00 1.43863177e+00 6.51039898e-01 3.89413744e-01 -2.65538365e-01 4.02585030e-01 1.97297543e-01 2.15211898e-01 -7.39152074e-01 6.37515366e-01 -2.19778195e-01 -2.43048444e-01 6.84392810e-01 3.91442299e-01 -1.33842900e-01 2.81181037e-01 -1.15238335e-02 1.26396334e+00 -5.50516605e-01 2.67396420e-01 -2.23384406e-02 9.31836188e-01 -6.26707589e-03 4.18432921e-01 7.34575272e-01 -4.05225992e-01 4.15933281e-01 5.48829019e-01 -6.98942542e-01 -5.67847192e-01 -5.01174092e-01 9.27614421e-02 1.36514938e+00 1.30627871e-01 -3.35049987e-01 -2.93289334e-01 -1.00355995e+00 -3.99572730e-01 8.64649355e-01 -3.79851103e-01 -2.48259023e-01 -3.80567431e-01 -3.33247423e-01 4.84885514e-01 1.88175306e-01 6.25298679e-01 -1.64845514e+00 -2.39937261e-01 4.89825696e-01 -4.48205173e-01 -1.13040972e+00 -4.02731866e-01 5.16038477e-01 -4.41395193e-01 -1.12833130e+00 -4.18386817e-01 -1.13393235e+00 2.47603953e-01 4.84274656e-01 9.84225273e-01 3.01817805e-01 -7.18653202e-02 9.57638398e-03 -9.60279107e-01 -3.96924347e-01 -6.88460052e-01 4.23918307e-01 -1.97520226e-01 2.82085180e-01 8.37716281e-01 -5.23921907e-01 -3.45089674e-01 6.07431173e-01 -6.78909004e-01 -3.52659851e-01 4.08945650e-01 9.46574628e-01 -4.73178059e-01 -6.59448728e-02 7.22687483e-01 -1.08408952e+00 1.04606712e+00 -6.81031406e-01 -1.13414340e-01 3.05979729e-01 -4.48707879e-01 -2.36463636e-01 1.11930072e+00 -4.63429749e-01 -1.19537473e+00 -1.11305378e-01 -7.35273063e-01 2.47572586e-02 -4.84979093e-01 4.71633911e-01 -1.20614506e-01 -7.40599409e-02 4.76171494e-01 2.32499972e-01 4.07770164e-02 -6.17848337e-01 1.19029708e-01 1.47725916e+00 -3.21916848e-01 -1.33810982e-01 1.72276706e-01 8.06941316e-02 -7.16119945e-01 -9.07902181e-01 -3.56039554e-01 -1.21361256e+00 -2.65555412e-01 -2.79455513e-01 7.26767659e-01 -4.33256179e-01 -1.20107830e+00 4.06026065e-01 -1.62749541e+00 6.85190931e-02 4.44347784e-02 2.80880719e-01 -3.29635262e-01 2.21271336e-01 -9.40202773e-01 -1.02383578e+00 -5.79543173e-01 -1.41756654e+00 7.60068059e-01 2.85118252e-01 -5.29783845e-01 -1.05906093e+00 -1.49577245e-01 7.09343791e-01 6.21703207e-01 -7.53971159e-01 9.31725085e-01 -1.52624345e+00 -1.25441954e-01 -5.95699787e-01 -5.82543574e-02 5.35321951e-01 5.23426354e-01 -2.43689135e-01 -9.69330311e-01 3.36916670e-02 3.43029320e-01 -3.50987315e-01 4.66065824e-01 -6.23943992e-02 5.96353650e-01 -5.32174766e-01 -1.49355516e-01 6.66341186e-02 1.10977054e+00 8.48264456e-01 7.75826335e-01 2.01403886e-01 3.79954010e-01 7.59834886e-01 7.29380012e-01 3.27939689e-01 3.27176303e-01 8.62832248e-01 6.49484575e-01 9.56386924e-02 3.20102781e-01 -2.26204507e-02 5.36680460e-01 1.01693153e+00 1.05643384e-01 -2.07377821e-01 -7.25739360e-01 5.81880331e-01 -1.80715632e+00 -1.04530656e+00 -3.12979251e-01 1.53814888e+00 6.23259604e-01 -3.03266440e-02 4.60640974e-02 8.49244818e-02 7.71429598e-01 2.62931257e-01 -1.94220960e-01 -1.26354873e+00 4.59919900e-01 2.22841263e-01 1.08382232e-01 4.09228146e-01 -8.97913992e-01 1.21085763e+00 5.46360779e+00 8.36444020e-01 -1.24509406e+00 3.01476806e-01 3.58568251e-01 4.65771139e-01 -1.77995324e-01 1.46431625e-02 -8.65728021e-01 4.10343468e-01 8.28222096e-01 -4.73651960e-02 3.57778698e-01 1.36659074e+00 2.62714356e-01 1.05342910e-01 -8.84763420e-01 8.18904459e-01 3.05215150e-01 -1.32656956e+00 -1.50971055e-01 -4.12573636e-01 2.65001118e-01 -1.56452760e-01 -2.55166262e-01 8.42344284e-01 4.68343496e-01 -8.46397936e-01 -2.94294092e-03 6.72623590e-02 1.43334061e-01 -6.96294010e-01 1.62316895e+00 6.55568302e-01 -1.14120495e+00 -1.71122044e-01 -6.61951244e-01 -4.27564174e-01 2.22890854e-01 3.72584999e-01 -1.56297588e+00 4.96812105e-01 5.98683298e-01 6.69221461e-01 -2.58938551e-01 6.89462900e-01 -1.20683216e-01 5.38618267e-01 -3.88512090e-02 -9.84484553e-01 3.34794015e-01 -3.92577112e-01 2.60392606e-01 1.30141628e+00 1.94187984e-01 -1.31972134e-01 1.70537707e-04 8.79989386e-01 9.83357877e-02 4.93474633e-01 -9.24792171e-01 -3.06705594e-01 3.69655937e-01 1.52192640e+00 -5.47107697e-01 -1.72601193e-01 -4.63877499e-01 1.11818802e+00 3.45867932e-01 1.59604087e-01 -7.93738067e-01 -7.92485714e-01 1.03635025e+00 -2.59979486e-01 2.70981938e-01 1.32990167e-01 -3.10258627e-01 -9.69325304e-01 -1.71338722e-01 -7.98328340e-01 2.11581826e-01 -7.58804142e-01 -1.52738428e+00 9.67259169e-01 -6.09699190e-01 -1.25135350e+00 -3.82152706e-01 -7.22726524e-01 -1.21271420e+00 9.51073766e-01 -1.18826544e+00 -1.18096638e+00 -2.85229534e-01 6.76490963e-01 9.97783542e-01 -4.88876402e-01 1.10898054e+00 3.89453262e-01 -4.35777098e-01 5.18411517e-01 -1.82750732e-01 3.81927580e-01 7.19560981e-01 -8.99714053e-01 1.74051583e-01 3.26297551e-01 1.41169712e-01 8.82197380e-01 5.96008182e-01 -2.34213084e-01 -1.13563263e+00 -1.02083480e+00 1.49724102e+00 -2.08700985e-01 7.30845630e-01 -5.80360770e-01 -8.17547679e-01 7.82330394e-01 4.59968209e-01 -4.90509599e-01 7.57695436e-01 5.14071465e-01 -2.59223908e-01 -2.47728005e-01 -1.12920833e+00 5.45837522e-01 3.73408675e-01 -7.46235192e-01 -7.68983841e-01 7.17955232e-01 7.61932313e-01 -6.64027482e-02 -6.43018842e-01 -7.06618279e-02 3.96045923e-01 -1.08246124e+00 6.17448866e-01 -7.82094181e-01 4.08413976e-01 1.13625035e-01 1.30652070e-01 -1.25828600e+00 -1.66337669e-01 -4.63590920e-01 5.10740578e-01 1.54196453e+00 3.97084445e-01 -8.96085978e-01 8.39041650e-01 6.08151972e-01 -2.25928679e-01 -5.32944620e-01 -9.13375974e-01 -6.62346363e-01 -1.80235356e-01 -5.02322495e-01 5.50815225e-01 1.05087125e+00 5.89016438e-01 7.85312235e-01 -4.78457183e-01 -3.02512407e-01 -2.43569732e-01 2.64828920e-01 8.24918926e-01 -1.07535481e+00 3.10427360e-02 -2.57898599e-01 -6.51932538e-01 -1.19280291e+00 2.97288358e-01 -9.79519844e-01 3.70333195e-01 -1.53047872e+00 -9.68971252e-02 -6.52699471e-01 -9.87833738e-02 6.16218328e-01 4.90894243e-02 5.11143431e-02 2.29139671e-01 9.79844555e-02 -3.40449750e-01 5.41461468e-01 1.03031528e+00 -3.88757437e-01 -3.01632732e-01 6.32743359e-01 -6.91043019e-01 6.96586967e-01 1.07048416e+00 -5.69644034e-01 -3.04324239e-01 -1.93669405e-02 -8.69863257e-02 1.44276768e-01 5.07179275e-02 -6.23789907e-01 2.51134157e-01 -2.22130790e-01 -5.88856220e-01 -3.77030283e-01 5.01870513e-01 -1.17229247e+00 -1.59343109e-01 5.36749601e-01 -4.72743779e-01 -5.13151102e-02 -1.99845925e-01 1.94552898e-01 -4.93599415e-01 -8.63756061e-01 4.92267698e-01 -2.93339372e-01 -7.56700516e-01 -1.17221937e-01 -1.02474594e+00 -1.44494772e-01 9.02757883e-01 -7.31549710e-02 -3.71301740e-01 -9.37852740e-01 -4.11792397e-01 5.10008298e-02 -7.48512894e-02 7.90083766e-01 4.90488321e-01 -1.06485784e+00 -3.79185438e-01 1.05229141e-02 3.96808028e-01 -5.10177553e-01 2.63045907e-01 7.42170691e-01 -5.88939905e-01 6.54388547e-01 -2.62830317e-01 -3.79757464e-01 -1.40224874e+00 3.12410146e-01 2.48901039e-01 -6.11836672e-01 -3.09518993e-01 8.38885605e-01 -2.47250404e-02 -8.63425791e-01 2.48018175e-01 -2.99780637e-01 -6.72106445e-01 1.39025584e-01 5.31808078e-01 6.74435198e-02 3.75431389e-01 -6.07524216e-01 -5.36063313e-01 8.46351758e-02 -3.17642361e-01 2.59984583e-01 1.19450545e+00 1.56840026e-01 -2.37241387e-01 5.46957314e-01 1.46265960e+00 -2.39362001e-01 -3.65527898e-01 5.68290502e-02 7.24383965e-02 -3.35094124e-01 -4.06726390e-01 -7.16479421e-01 -9.06093061e-01 1.02403557e+00 3.77324104e-01 6.51551306e-01 7.82080531e-01 2.51519270e-02 9.41747725e-01 7.03634262e-01 4.24835622e-01 -1.07322395e+00 2.94821188e-02 9.67388809e-01 9.57091808e-01 -1.39968348e+00 -5.67277908e-01 -4.56525534e-01 -1.07806706e+00 1.27140713e+00 7.21622884e-01 -1.13793574e-01 4.97068018e-01 1.81484237e-01 5.32395661e-01 -2.54994363e-01 -9.09281015e-01 -2.52900034e-01 -5.69936112e-02 7.40665555e-01 7.12913573e-01 -2.27876622e-02 -8.43718767e-01 8.46077979e-01 -8.88188183e-02 -1.20959878e-01 6.75250173e-01 6.70494914e-01 -3.64466190e-01 -1.29188168e+00 -5.35868444e-02 4.62050378e-01 -1.84870124e-01 -2.07688168e-01 -6.06374383e-01 6.83582187e-01 -3.48982438e-02 1.58785260e+00 -5.65277524e-02 -7.68188357e-01 3.08323145e-01 4.92720902e-01 -3.97833973e-01 -9.65557814e-01 -1.35860670e+00 -5.01608193e-01 4.60454673e-01 -3.65684718e-01 -3.49362433e-01 -3.72525126e-01 -1.08081019e+00 -4.51024622e-01 -5.92266262e-01 5.38393974e-01 7.30199814e-01 1.22879004e+00 6.08220510e-02 4.48830187e-01 9.61463571e-01 -5.01728177e-01 -4.79499757e-01 -1.41063797e+00 -6.74135268e-01 6.52106583e-01 -1.09966047e-01 -5.00121057e-01 -5.46949685e-01 -1.94891870e-01]
[12.712361335754395, 7.712646484375]
cb01a4b3-7332-462b-8b16-810485180364
transformers-in-medical-imaging-a-survey
2201.09873
null
https://arxiv.org/abs/2201.09873v1
https://arxiv.org/pdf/2201.09873v1.pdf
Transformers in Medical Imaging: A Survey
Following unprecedented success on the natural language tasks, Transformers have been successfully applied to several computer vision problems, achieving state-of-the-art results and prompting researchers to reconsider the supremacy of convolutional neural networks (CNNs) as {de facto} operators. Capitalizing on these advances in computer vision, the medical imaging field has also witnessed growing interest for Transformers that can capture global context compared to CNNs with local receptive fields. Inspired from this transition, in this survey, we attempt to provide a comprehensive review of the applications of Transformers in medical imaging covering various aspects, ranging from recently proposed architectural designs to unsolved issues. Specifically, we survey the use of Transformers in medical image segmentation, detection, classification, reconstruction, synthesis, registration, clinical report generation, and other tasks. In particular, for each of these applications, we develop taxonomy, identify application-specific challenges as well as provide insights to solve them, and highlight recent trends. Further, we provide a critical discussion of the field's current state as a whole, including the identification of key challenges, open problems, and outlining promising future directions. We hope this survey will ignite further interest in the community and provide researchers with an up-to-date reference regarding applications of Transformer models in medical imaging. Finally, to cope with the rapid development in this field, we intend to regularly update the relevant latest papers and their open-source implementations at \url{https://github.com/fahadshamshad/awesome-transformers-in-medical-imaging}.
['Huazhu Fu', 'Fahad Shahbaz Khan', 'Munawar Hayat', 'Muhammad Haris Khan', 'Syed Waqas Zamir', 'Salman Khan', 'Fahad Shamshad']
2022-01-24
null
null
null
null
['medical-image-denoising', 'medical-object-detection', 'medical-report-generation']
['computer-vision', 'computer-vision', 'medical']
[ 6.18365884e-01 1.75764561e-01 -1.31554082e-01 -3.71471405e-01 -6.99585438e-01 -4.55380887e-01 1.37288481e-01 6.07274137e-02 -3.64726305e-01 3.14497799e-01 1.02669127e-01 -5.62648714e-01 -2.14855015e-01 -6.14890456e-01 -3.07391882e-01 -8.41584980e-01 -2.52053708e-01 1.26407504e-01 1.19265497e-01 -3.06899995e-02 -8.81429389e-02 6.71977937e-01 -1.07350719e+00 5.40213048e-01 6.32317960e-01 1.24559820e+00 3.49632293e-01 4.16854620e-01 -5.39864972e-02 9.26779687e-01 -4.78003204e-01 -5.10309339e-01 -4.00725678e-02 -2.80800283e-01 -1.02338517e+00 1.24922514e-01 2.41309330e-01 -1.00779548e-01 -5.75907707e-01 9.87149656e-01 7.51049936e-01 -1.89498737e-01 3.83601665e-01 -8.56557190e-01 -7.24751294e-01 5.43710887e-01 -3.23956251e-01 6.12018466e-01 -7.45731145e-02 5.74604683e-02 8.61119270e-01 -8.64170134e-01 6.22236967e-01 5.66486716e-01 8.79038692e-01 7.80359209e-01 -7.68869817e-01 -5.52066922e-01 1.72091946e-01 4.05939013e-01 -1.23052561e+00 -4.11770314e-01 7.36046791e-01 -3.45128268e-01 1.06282294e+00 5.84551454e-01 7.54062057e-01 1.05816674e+00 5.56623399e-01 1.02720690e+00 9.78297889e-01 -2.55025506e-01 -6.98608309e-02 -1.43115252e-01 -1.10185556e-02 9.92789507e-01 -5.80122061e-02 -5.25910221e-02 -4.07836735e-01 -1.03711158e-01 9.89046574e-01 4.47259955e-02 -5.15921295e-01 -1.71609461e-01 -1.47290719e+00 7.18289196e-01 6.90722704e-01 7.43795395e-01 -4.51211959e-01 9.60345417e-02 6.00933254e-01 3.58064175e-02 3.29189122e-01 2.82837719e-01 -3.03899497e-01 2.30627000e-01 -9.10922527e-01 -5.31918518e-02 3.50470483e-01 7.67083168e-01 1.07627906e-01 2.05839351e-01 -1.87640518e-01 9.80186403e-01 2.29265410e-02 2.14982808e-01 5.06405294e-01 -7.77857363e-01 -9.44887623e-02 2.25881159e-01 -5.30764580e-01 -7.68839777e-01 -7.06898630e-01 -9.90002990e-01 -1.44205427e+00 -6.02270104e-02 1.48990363e-01 2.04905853e-01 -9.66861606e-01 1.41721785e+00 1.95127260e-02 1.66897446e-01 -2.85587788e-01 7.06904888e-01 1.26977670e+00 2.76034176e-01 7.48896822e-02 -3.26797932e-01 1.82420433e+00 -9.06997323e-01 -9.29011047e-01 -8.43168497e-02 2.83904880e-01 -9.28076982e-01 6.47933006e-01 5.59966683e-01 -1.15517020e+00 -3.53017032e-01 -9.12387908e-01 -1.11335754e-01 -2.44050786e-01 2.93716669e-01 7.73523748e-01 5.56827664e-01 -1.23376179e+00 5.73564827e-01 -1.28024709e+00 -5.10205030e-01 8.21530342e-01 5.73797941e-01 -3.78235653e-02 -2.85654273e-02 -1.10337698e+00 1.06123662e+00 -7.79995397e-02 3.89961779e-01 -8.80835295e-01 -8.62138152e-01 -6.84381485e-01 -3.49811822e-01 2.68171579e-01 -1.00909793e+00 1.55177522e+00 -7.46068299e-01 -1.26290202e+00 1.19584680e+00 -1.18910983e-01 -5.90328574e-01 3.99618655e-01 2.92982906e-01 -5.22058308e-01 3.66629899e-01 1.25305697e-01 7.12557375e-01 4.24726635e-01 -7.55653501e-01 -6.42275810e-01 -1.91653743e-01 2.62778625e-02 -2.42243260e-02 -4.58834559e-01 3.40786070e-01 -5.28729677e-01 -9.74084973e-01 2.13775113e-01 -8.29273641e-01 -6.04390383e-01 3.77846599e-01 -4.41478312e-01 -1.96383685e-01 4.33923125e-01 -4.28968906e-01 1.26908684e+00 -2.08537459e+00 -8.98589566e-02 -8.10618624e-02 7.05051363e-01 2.27062836e-01 4.44873981e-02 2.11001262e-01 -4.38324153e-01 -1.44767817e-02 -4.87641037e-01 -2.57009774e-01 -3.82424980e-01 1.90593988e-01 -2.28599742e-01 7.23087788e-01 8.42575505e-02 1.29189742e+00 -9.13944423e-01 -5.13727605e-01 5.43298244e-01 7.96413839e-01 -6.16267957e-02 -2.74341822e-01 2.79469103e-01 7.78996050e-01 -6.04099989e-01 8.29433143e-01 4.89859313e-01 -6.60904646e-01 1.76100194e-01 -5.77247858e-01 -1.96056917e-01 3.25094491e-01 -6.96981430e-01 1.58008206e+00 -5.01901686e-01 7.07679749e-01 3.16452533e-01 -1.37208939e+00 5.91794670e-01 7.38593161e-01 1.11273336e+00 -5.87239921e-01 2.50728399e-01 5.06482065e-01 2.90997587e-02 -5.48115969e-01 4.53962088e-02 -4.78367686e-01 2.10076973e-01 1.06454223e-01 4.73695546e-02 -3.61121930e-02 -4.61378098e-02 -2.06136443e-02 1.11126435e+00 -2.91504800e-01 6.07533574e-01 -2.90004015e-01 6.85756624e-01 -6.84556440e-02 3.25612217e-01 7.41662502e-01 -5.22444189e-01 7.34667838e-01 2.36467451e-01 -8.52832079e-01 -6.21596754e-01 -1.07922304e+00 -5.34771681e-01 5.53605974e-01 -1.24812119e-01 -4.37764108e-01 -6.87855482e-01 -7.16867626e-01 -6.31963611e-01 9.96409208e-02 -8.70869875e-01 2.26012990e-01 -8.40928495e-01 -1.20406413e+00 7.30670035e-01 6.62657976e-01 5.42131960e-01 -1.20240450e+00 -1.04290521e+00 2.74904162e-01 -6.19844079e-01 -1.36909056e+00 -2.31774807e-01 2.86886543e-01 -1.26151645e+00 -1.09724379e+00 -1.19315290e+00 -1.12531006e+00 6.20362878e-01 1.61469668e-01 1.21539378e+00 2.37197071e-01 -6.16288543e-01 4.64350522e-01 -2.36906037e-01 -3.54912341e-01 -2.04761207e-01 1.06294647e-01 -2.59590328e-01 -2.30196580e-01 2.24174783e-02 -5.13397872e-01 -8.15883279e-01 1.39196858e-01 -1.08731043e+00 3.77735049e-02 5.28387785e-01 9.70049202e-01 7.29994535e-01 -5.17648198e-02 2.70907015e-01 -9.11670804e-01 4.65356857e-01 -1.78242758e-01 -2.01393560e-01 3.29643488e-01 -4.80209827e-01 -3.21822226e-01 4.46287155e-01 -1.11533776e-01 -7.28330553e-01 9.39088762e-02 -6.67931795e-01 -1.98348999e-01 -1.86783314e-01 6.01087391e-01 3.40466380e-01 -3.35491806e-01 7.37156272e-01 4.03656334e-01 3.14234532e-02 -2.79031485e-01 7.50040188e-02 4.77450848e-01 5.94329596e-01 -3.70596737e-01 3.05206597e-01 1.02332342e+00 1.33208886e-01 -9.35476303e-01 -9.24408615e-01 -4.98604089e-01 -3.66250813e-01 -2.80171990e-01 9.69774306e-01 -4.84954268e-01 -5.97215772e-01 6.12260938e-01 -1.27928114e+00 -8.72848257e-02 -3.44011813e-01 3.64128143e-01 -5.95069885e-01 4.77823943e-01 -1.06189907e+00 -3.66513461e-01 -6.64100230e-01 -1.78033841e+00 8.27482820e-01 1.10335067e-01 -1.19386703e-01 -1.28451550e+00 -3.44202310e-01 3.96002442e-01 5.79492569e-01 3.50005418e-01 8.29238653e-01 -3.78574222e-01 -6.40776932e-01 -6.29743338e-02 -3.32502127e-02 2.65944064e-01 2.45603159e-01 -2.36096859e-01 -1.12110758e+00 -1.87398806e-01 3.83714110e-01 1.81708504e-02 8.38018596e-01 1.02408850e+00 1.53608561e+00 1.69510216e-01 -6.46231592e-01 7.95890272e-01 1.25613046e+00 2.70955414e-01 6.80094659e-01 2.59020358e-01 3.29754919e-01 5.58207393e-01 2.61197776e-01 2.30357468e-01 1.85352907e-01 6.09824896e-01 6.42145276e-01 -5.59457183e-01 -4.78503644e-01 3.22529703e-01 -2.17581972e-01 1.02106404e+00 -2.91068852e-01 -1.45121530e-01 -1.01827979e+00 4.69089299e-01 -1.49639881e+00 -6.05355978e-01 -1.91827208e-01 1.89644027e+00 5.74223220e-01 -4.00157012e-02 -1.77413076e-01 4.76685278e-02 7.16012776e-01 2.94584095e-01 -4.23286140e-01 6.26705140e-02 -1.00576095e-01 6.23131692e-01 5.22022307e-01 3.65720361e-01 -1.29672503e+00 7.66898870e-01 6.68402290e+00 9.69382524e-01 -1.67353582e+00 3.63472164e-01 9.68073666e-01 1.79049298e-01 -1.89158067e-01 -4.00742859e-01 -2.87932575e-01 -1.02519868e-02 5.60940921e-01 1.08558938e-01 2.62845784e-01 4.60608542e-01 1.72221348e-01 1.00143269e-01 -7.85512030e-01 1.19912553e+00 1.19877815e-01 -1.87673330e+00 -1.03094473e-01 8.23989213e-02 3.49184752e-01 4.72903043e-01 4.79303867e-01 -2.53966063e-01 -1.86909050e-01 -1.16538250e+00 5.81466734e-01 8.74625742e-02 1.18864477e+00 -2.13879138e-01 7.00735331e-01 6.29662499e-02 -1.44687808e+00 4.99416552e-02 -2.04408646e-01 1.73076540e-01 3.07521701e-01 7.55163908e-01 -6.56766117e-01 7.70316541e-01 9.02636588e-01 8.76787245e-01 -2.37422332e-01 1.25722682e+00 -1.09530263e-01 5.24345696e-01 -8.48365948e-02 1.57580778e-01 3.72947663e-01 1.85893551e-01 5.13418615e-01 1.35003304e+00 1.72804356e-01 2.91446745e-01 -6.24108082e-03 7.36281931e-01 1.14807390e-01 8.98126811e-02 -4.32143182e-01 1.13415346e-01 7.56905749e-02 1.43587923e+00 -1.25314617e+00 -3.91319335e-01 -5.92763543e-01 8.49453628e-01 -9.45385471e-02 3.41252536e-01 -9.30064559e-01 -2.92102933e-01 5.30097067e-01 1.98243499e-01 -2.50154659e-02 -1.70806915e-01 -5.49529672e-01 -1.00047600e+00 1.09468652e-02 -8.87647212e-01 5.62486529e-01 -6.77838743e-01 -1.02175462e+00 1.07986689e+00 2.70681176e-02 -1.38827586e+00 3.73504497e-02 -9.14576352e-01 -2.95299679e-01 5.89614749e-01 -1.66439605e+00 -1.10283148e+00 -1.91928834e-01 5.62756896e-01 6.88389838e-01 5.59054874e-02 9.94943798e-01 5.96414745e-01 -2.70335317e-01 4.99752730e-01 -1.12108558e-01 4.07687932e-01 5.74638009e-01 -8.42070460e-01 4.48099256e-01 5.62810004e-01 2.63722181e-01 8.03065419e-01 5.12929499e-01 -2.03184709e-01 -1.03486300e+00 -9.37692225e-01 9.38005745e-01 -2.57593125e-01 6.17712736e-01 -6.46644309e-02 -5.86526573e-01 7.78988540e-01 3.81127059e-01 3.10440361e-01 7.26027608e-01 -2.84827530e-01 -3.87669839e-02 -2.26300955e-01 -1.20982039e+00 5.45345724e-01 1.00387263e+00 -3.21735144e-01 -1.78111821e-01 4.96246159e-01 3.06332469e-01 -8.10828447e-01 -8.58001113e-01 6.98651254e-01 5.97232342e-01 -1.09420800e+00 1.10476804e+00 -2.01677322e-01 3.19646209e-01 -5.13061695e-02 1.63738444e-01 -1.00604928e+00 -4.99346405e-01 -5.34733713e-01 3.68720859e-01 4.94722039e-01 1.82173535e-01 -7.18602538e-01 8.09722245e-01 1.58113927e-01 -6.55299366e-01 -1.40929031e+00 -1.21266365e+00 -3.63546491e-01 3.32026243e-01 -7.73679912e-01 2.93122977e-01 9.21241045e-01 7.47981668e-02 5.93610993e-03 -1.66932181e-01 3.01955044e-02 5.76498806e-01 -7.13744313e-02 -6.37114644e-02 -9.39044893e-01 -1.08350202e-01 -6.92604005e-01 -4.64863241e-01 -1.22673929e+00 -1.58407688e-01 -1.10217118e+00 -1.67186901e-01 -1.74672234e+00 2.54307151e-01 -3.25667441e-01 -3.74114215e-01 5.94134629e-01 2.51321465e-01 8.32229018e-01 9.24844444e-02 1.08108945e-01 -3.64375025e-01 -2.95301303e-02 1.65426660e+00 -4.20060724e-01 2.64582902e-01 2.48426914e-01 -8.60570729e-01 7.33454823e-01 9.39842224e-01 -2.86278367e-01 -3.12615871e-01 -6.46746635e-01 1.28683701e-01 9.26437825e-02 4.85594571e-01 -1.00758183e+00 4.41022635e-01 5.26012927e-02 2.44069040e-01 -5.33896387e-01 2.54695296e-01 -7.67855406e-01 9.77559164e-02 7.94096947e-01 -2.55273253e-01 2.74876446e-01 2.35083222e-01 5.33599444e-02 -4.31659222e-01 -2.77791142e-01 1.08219492e+00 -7.08470762e-01 -7.00465679e-01 5.93023479e-01 -7.36828923e-01 -1.50035888e-01 8.80719602e-01 -3.09299350e-01 -9.73161757e-02 -3.21560264e-01 -1.07093656e+00 5.37109971e-02 -4.28774245e-02 2.40078226e-01 8.46297979e-01 -9.32536364e-01 -7.02748537e-01 1.36888996e-01 1.08863354e-01 1.60994418e-02 6.81377172e-01 1.39286971e+00 -6.03560567e-01 6.96934104e-01 -3.69862355e-02 -7.66025543e-01 -1.28103876e+00 4.89531338e-01 8.80159080e-01 -4.89369601e-01 -8.28309238e-01 9.61415708e-01 4.57298577e-01 -2.66019583e-01 3.22264493e-01 -6.39247000e-01 -2.78773338e-01 -2.60629803e-01 4.12869275e-01 8.91175568e-02 6.14033997e-01 -4.86758590e-01 -7.12198257e-01 6.79974318e-01 -5.85169084e-02 1.47375893e-02 1.42633426e+00 3.64980325e-02 -3.51747394e-01 1.85079515e-01 1.04071391e+00 -2.23823473e-01 -6.28185213e-01 -2.13068455e-01 -3.54330003e-01 5.55607192e-02 1.48876891e-01 -8.70368481e-01 -1.58906794e+00 1.22341514e+00 7.12665319e-01 1.95423350e-01 1.60713661e+00 3.12737703e-01 8.18292439e-01 -3.79699990e-02 3.95563543e-01 -6.61648214e-01 1.31362006e-01 5.19400537e-01 9.94433522e-01 -8.62263441e-01 7.10653439e-02 -8.34466577e-01 -4.82987076e-01 1.41502893e+00 9.40116867e-02 2.72816513e-02 1.08627057e+00 6.19014561e-01 4.00172353e-01 -4.74289328e-01 -3.98613960e-01 -4.21030037e-02 3.19601148e-01 8.84058416e-01 9.60767329e-01 7.77502358e-02 -3.09652656e-01 3.27806711e-01 -1.79549977e-01 3.08035821e-01 3.53739411e-01 8.77742767e-01 -4.08862717e-02 -1.34418869e+00 -3.19317788e-01 6.67394876e-01 -8.49068820e-01 -4.41294461e-01 2.79319212e-02 4.62343395e-01 2.73166180e-01 8.10462654e-01 -1.17031343e-01 -1.39137357e-01 3.26566547e-01 -3.44445467e-01 7.03995049e-01 -5.67701221e-01 -8.67198348e-01 3.83943796e-01 -1.47805214e-01 -5.16041338e-01 -6.62326217e-01 -6.25453591e-01 -1.17296636e+00 -8.24321732e-02 -1.54930681e-01 -7.83491209e-02 4.77783442e-01 7.78450072e-01 -7.78831774e-03 9.62955654e-01 1.76032051e-01 -5.15532136e-01 -4.12144125e-01 -7.01349556e-01 -4.48024392e-01 -1.71653897e-01 4.19895768e-01 -4.59671825e-01 4.27394882e-02 8.35749283e-02]
[14.525227546691895, -2.595310926437378]
32ab03cd-57fd-4c33-8113-57df6a46c9d1
psla-improving-audio-event-classification
2102.01243
null
https://arxiv.org/abs/2102.01243v3
https://arxiv.org/pdf/2102.01243v3.pdf
PSLA: Improving Audio Tagging with Pretraining, Sampling, Labeling, and Aggregation
Audio tagging is an active research area and has a wide range of applications. Since the release of AudioSet, great progress has been made in advancing model performance, which mostly comes from the development of novel model architectures and attention modules. However, we find that appropriate training techniques are equally important for building audio tagging models with AudioSet, but have not received the attention they deserve. To fill the gap, in this work, we present PSLA, a collection of training techniques that can noticeably boost the model accuracy including ImageNet pretraining, balanced sampling, data augmentation, label enhancement, model aggregation and their design choices. By training an EfficientNet with these techniques, we obtain a single model (with 13.6M parameters) and an ensemble model that achieve mean average precision (mAP) scores of 0.444 and 0.474 on AudioSet, respectively, outperforming the previous best system of 0.439 with 81M parameters. In addition, our model also achieves a new state-of-the-art mAP of 0.567 on FSD50K.
['James Glass', 'Yu-An Chung', 'Yuan Gong']
2021-02-02
null
null
null
null
['audio-tagging']
['audio']
[ 2.63091952e-01 -6.68849945e-02 -1.58997610e-01 -4.51564968e-01 -1.10005343e+00 -4.48181301e-01 4.16984946e-01 1.09571166e-01 -6.91407919e-01 3.37751031e-01 3.44512671e-01 6.68633878e-02 2.93976087e-02 -2.68801212e-01 -5.45113444e-01 -4.33602214e-01 -3.07315737e-01 5.49135745e-01 4.43015009e-01 -7.08697066e-02 -2.71990150e-02 -9.41625051e-03 -1.73400462e+00 6.14699781e-01 4.38135713e-01 1.42623508e+00 7.66325742e-02 8.53831172e-01 8.08889568e-02 8.64743233e-01 -7.12068737e-01 -4.46985871e-01 -3.61069404e-02 -1.25263065e-01 -1.02517200e+00 -2.87969321e-01 7.74122477e-01 -2.57829189e-01 -3.34819168e-01 7.14431405e-01 9.85511959e-01 1.18122600e-01 4.04716462e-01 -1.15479970e+00 -4.95744824e-01 1.05525732e+00 -5.14882147e-01 3.68720591e-01 7.19224140e-02 -2.29367092e-01 1.19063699e+00 -9.33435202e-01 6.71866238e-02 1.22318637e+00 8.57215703e-01 6.87370658e-01 -9.51218367e-01 -1.19214034e+00 1.11977816e-01 4.78536397e-01 -1.60976076e+00 -7.93542624e-01 3.66769403e-01 -2.53051281e-01 1.16572332e+00 2.87713319e-01 5.61288476e-01 8.44847322e-01 -1.06153592e-01 8.93825650e-01 9.87053931e-01 -7.75562048e-01 1.08117707e-01 2.83471316e-01 2.05166325e-01 3.54803056e-01 -2.15425447e-01 -3.00556123e-02 -9.10518885e-01 -1.32399589e-01 4.08033490e-01 -4.09306884e-01 -1.11659504e-01 9.33010653e-02 -9.44479644e-01 5.59081018e-01 4.08082366e-01 2.99747705e-01 5.35663739e-02 4.82854754e-01 6.60976648e-01 3.41782093e-01 6.33619368e-01 6.39989793e-01 -5.68131268e-01 -4.37032312e-01 -1.00694585e+00 2.20486447e-01 4.79964703e-01 1.00654435e+00 2.43244722e-01 1.11836836e-01 -1.43795088e-01 1.20822155e+00 1.25208050e-01 4.43580121e-01 5.56649268e-01 -1.04297960e+00 4.64459062e-01 3.31174642e-01 -9.01060179e-02 -5.13660073e-01 -3.87182385e-01 -7.03046739e-01 -8.71233165e-01 -2.57380426e-01 -5.81577308e-02 -2.20431294e-03 -1.10199749e+00 1.75257361e+00 2.99086049e-02 5.96706331e-01 -1.39059275e-01 7.18798935e-01 9.75156307e-01 5.47340035e-01 5.59008360e-01 2.42207557e-01 1.35278332e+00 -1.20375931e+00 -7.74930000e-01 -3.61332893e-01 6.54756010e-01 -1.04330170e+00 1.05976474e+00 6.07436240e-01 -1.11310351e+00 -8.85270059e-01 -1.15219414e+00 8.78070593e-02 -3.42998952e-01 1.27582118e-01 6.45759583e-01 6.35393798e-01 -1.25545192e+00 4.75565940e-01 -6.86303794e-01 -4.33077782e-01 3.92103463e-01 6.46098256e-01 -1.56585336e-01 1.35032877e-01 -1.29190254e+00 6.53814137e-01 5.20535648e-01 -2.33932942e-01 -9.55772460e-01 -8.50249946e-01 -4.45197105e-01 1.31661102e-01 4.28429246e-01 -5.26675582e-01 1.69986832e+00 -7.10917890e-01 -1.27196479e+00 6.97453380e-01 2.02528477e-01 -8.36795866e-01 6.02281420e-03 -5.55351377e-01 -7.48228312e-01 1.97888035e-02 -1.12892315e-01 1.43795490e+00 5.90802073e-01 -8.37635338e-01 -1.07406557e+00 -9.09951180e-02 1.86330602e-01 2.90849626e-01 -8.96129847e-01 4.33156610e-01 -7.00263143e-01 -6.78689480e-01 -2.24957719e-01 -1.15392816e+00 -6.00055493e-02 -3.97756636e-01 -7.18320832e-02 -2.81417102e-01 5.67086697e-01 -6.26323640e-01 1.68041611e+00 -2.44452667e+00 -2.09842205e-01 -1.09079525e-01 2.81165950e-02 6.24521136e-01 -2.81763256e-01 2.20289901e-01 1.47069665e-02 2.78138936e-01 -4.31221351e-02 -5.25601983e-01 4.44610268e-02 1.11655675e-01 -3.05687010e-01 -5.57014979e-02 2.22913086e-01 6.70827925e-01 -7.49102235e-01 -5.70766091e-01 8.31131637e-02 5.09283662e-01 -8.96764815e-01 1.09818466e-01 -1.88817337e-01 1.69180661e-01 -2.53483914e-02 6.45078361e-01 4.61001068e-01 -2.39459276e-01 2.20896788e-02 -5.72049897e-03 1.05551042e-01 6.12118959e-01 -1.22804141e+00 1.94165313e+00 -4.84526187e-01 7.07015455e-01 3.48841175e-02 -5.52623868e-01 7.11238980e-01 6.49739027e-01 5.93079567e-01 -5.47235191e-01 1.84584692e-01 1.19638316e-01 1.31329358e-01 -2.49223456e-01 7.14370728e-01 1.35005012e-01 -5.11347018e-02 3.13728392e-01 4.94787067e-01 3.12968731e-01 2.14199185e-01 1.51557550e-01 1.23267770e+00 -2.58329719e-01 -3.08989417e-02 -1.37065455e-01 3.42927277e-01 -2.62684375e-01 3.51419777e-01 7.48697579e-01 -2.33197853e-01 6.28698170e-01 -1.26638010e-01 -3.46246094e-01 -9.81679618e-01 -6.06235385e-01 -3.47719759e-01 1.56221545e+00 -1.77589491e-01 -8.16933155e-01 -8.33701491e-01 -6.13882065e-01 -1.15082353e-01 3.43323141e-01 -5.00020921e-01 -3.14673007e-01 -5.15813828e-01 -8.44840944e-01 1.10208476e+00 9.28856194e-01 6.37741566e-01 -1.16401768e+00 -2.49899507e-01 4.13942307e-01 -4.28507477e-01 -1.16735947e+00 -4.64459211e-01 5.84307432e-01 -7.40110099e-01 -5.55098534e-01 -5.56587279e-01 -8.44320595e-01 -1.00394972e-02 2.51324117e-01 1.42785907e+00 -1.88518432e-03 -2.40206681e-02 1.78237334e-01 -6.16109252e-01 -7.92776108e-01 -2.38897964e-01 6.98734224e-01 1.05707727e-01 -1.48567349e-01 5.60208201e-01 -5.35993814e-01 -3.90950173e-01 3.95347804e-01 -8.76446307e-01 -8.31872374e-02 6.34986103e-01 6.47898495e-01 6.29024744e-01 -3.03344559e-02 8.25112402e-01 -7.23871112e-01 4.15192813e-01 -3.04824531e-01 -2.22426310e-01 1.17988117e-01 -7.61935115e-01 -2.36965150e-01 2.04455540e-01 -5.94702423e-01 -8.82596374e-01 9.49044600e-02 -5.83337605e-01 -3.69120687e-01 -4.57862467e-01 3.89978617e-01 -8.29370916e-02 -9.78101045e-03 7.53040195e-01 -1.40109435e-01 -3.67768288e-01 -9.33739126e-01 2.84711331e-01 1.10727096e+00 5.79591453e-01 -4.06914890e-01 2.51588553e-01 2.05817357e-01 -3.99321079e-01 -8.03289890e-01 -1.31111002e+00 -9.25189912e-01 -4.67159301e-01 -8.58595744e-02 5.92991292e-01 -1.41084194e+00 -3.62444252e-01 4.06923085e-01 -8.49774480e-01 -2.90836573e-01 -2.87382543e-01 4.98219252e-01 -3.09323847e-01 -6.35561273e-02 -6.42820179e-01 -7.98733413e-01 -5.11457920e-01 -8.36483657e-01 1.12354386e+00 -8.28121305e-02 -5.32336891e-01 -6.79827034e-01 2.19152197e-01 4.76766616e-01 4.80761349e-01 -4.37762499e-01 7.08311617e-01 -9.02285218e-01 -3.53517920e-01 -1.82416603e-01 -1.42854199e-01 5.91039598e-01 -2.02724487e-01 -3.42071682e-01 -1.71803272e+00 -3.42639685e-01 -4.40314084e-01 -5.47289193e-01 1.08666563e+00 4.11065698e-01 1.45497108e+00 1.25773577e-02 -3.59492183e-01 5.04992664e-01 1.10542810e+00 4.67271537e-01 5.40625811e-01 4.44959134e-01 4.63574797e-01 1.37380064e-01 6.49786532e-01 3.19722354e-01 3.02539647e-01 1.08837116e+00 3.48752350e-01 -1.60832584e-01 -5.55358052e-01 -2.77923644e-01 3.43027204e-01 1.08579791e+00 -5.21776229e-02 -3.18014294e-01 -9.24531877e-01 4.89766806e-01 -1.70678329e+00 -7.96372652e-01 1.10720709e-01 2.13216138e+00 8.86106610e-01 3.14666599e-01 3.50766093e-01 5.74639201e-01 6.09797716e-01 1.14918528e-02 -1.78100109e-01 -2.00418845e-01 2.69995425e-02 4.70464975e-01 3.29106688e-01 3.48783523e-01 -1.47472751e+00 1.08059990e+00 6.89106512e+00 1.05979168e+00 -9.65462744e-01 2.29463547e-01 5.18909633e-01 -1.91208690e-01 2.79004782e-01 -2.63399869e-01 -1.30974269e+00 4.38412696e-01 1.51220715e+00 1.51854560e-01 3.12842488e-01 9.57652211e-01 -2.59013921e-01 2.47300252e-01 -9.07473505e-01 1.07129908e+00 2.12961212e-01 -9.84827995e-01 2.61754207e-02 1.01382263e-01 7.25623488e-01 1.84913412e-01 1.84566423e-01 7.22087443e-01 2.01577052e-01 -8.23761582e-01 7.91187346e-01 1.25828981e-01 8.37449610e-01 -8.80418122e-01 1.09305358e+00 -8.22849795e-02 -1.24888253e+00 -2.25689754e-01 -4.76080894e-01 -2.04148009e-01 2.27328092e-02 5.02825260e-01 -1.22257721e+00 2.64887065e-01 1.17724991e+00 5.61615944e-01 -6.64838731e-01 1.45585704e+00 -5.32068079e-03 1.12373865e+00 -4.28861797e-01 9.21180658e-03 1.86887890e-01 6.00647390e-01 1.19231902e-01 1.54829347e+00 2.51803428e-01 -2.73650438e-01 2.43377328e-01 1.01039261e-01 -3.72450262e-01 1.79375947e-01 -2.11869955e-01 7.45659769e-02 7.14348614e-01 1.30299234e+00 -5.52992523e-01 -5.13737977e-01 -2.99647659e-01 4.59353805e-01 2.20673725e-01 -5.71181346e-03 -1.06572044e+00 -3.86177987e-01 7.02247679e-01 1.54666051e-01 2.68655300e-01 6.87548891e-02 -2.01800644e-01 -8.01247954e-01 -1.42424151e-01 -9.07081008e-01 5.74072003e-01 -6.58748031e-01 -9.57598805e-01 8.71622980e-01 -2.14359611e-02 -1.18873429e+00 -1.34976178e-01 -2.50460953e-01 -2.17300519e-01 4.52039301e-01 -1.49547744e+00 -1.08594489e+00 -2.78446287e-01 2.66547501e-01 5.88944972e-01 -2.62541503e-01 1.11039901e+00 1.11288118e+00 -4.10976589e-01 1.01593029e+00 -1.00968458e-01 7.91999251e-02 1.06007254e+00 -1.22372985e+00 6.18799627e-01 4.49933022e-01 7.01041460e-01 3.09100837e-01 5.06815374e-01 -1.46734223e-01 -5.89037299e-01 -1.32740819e+00 1.08567083e+00 -4.52370435e-01 6.72685385e-01 -4.30331737e-01 -8.30132902e-01 6.31275833e-01 3.66992235e-01 -2.62775719e-01 9.55068648e-01 6.18004739e-01 -3.32387626e-01 -4.43377495e-01 -6.68694317e-01 3.42890501e-01 1.14295828e+00 -4.80506539e-01 -2.74263799e-01 2.11883530e-01 8.51718664e-01 -3.97492379e-01 -1.05139160e+00 7.01228797e-01 5.94911993e-01 -5.92069864e-01 1.01294672e+00 -5.79777956e-01 6.34059832e-02 -2.21575037e-01 -3.05290610e-01 -1.19079828e+00 -6.27859294e-01 -4.44953501e-01 -2.98977554e-01 1.67076433e+00 5.74363112e-01 -7.72267058e-02 9.02822673e-01 2.40418166e-01 -6.14367723e-01 -8.11554551e-01 -9.09840047e-01 -8.73396337e-01 -1.43890023e-01 -8.15933108e-01 7.46755123e-01 7.79103458e-01 -6.66566566e-02 6.69911146e-01 -6.24077737e-01 -4.89170998e-02 2.41415828e-01 -3.46256196e-01 7.49556124e-01 -1.29331493e+00 -2.31329769e-01 -2.68950045e-01 -6.00310981e-01 -1.18122518e+00 4.82979268e-02 -7.77462304e-01 -4.56109783e-03 -1.26430893e+00 3.40015501e-01 -7.29999304e-01 -9.11561012e-01 8.84513438e-01 -2.72986647e-02 7.03192592e-01 2.19035551e-01 1.50098324e-01 -1.05759180e+00 3.83648902e-01 8.00333500e-01 -1.74909025e-01 -2.31175154e-01 7.05698356e-02 -8.79147828e-01 7.75518715e-01 1.13675773e+00 -5.23029268e-01 -4.65890080e-01 -6.30552411e-01 9.68848765e-02 -4.68632936e-01 1.39735729e-01 -1.64610946e+00 1.51317075e-01 4.31621045e-01 3.63461673e-01 -5.95532596e-01 7.43495107e-01 -8.28066051e-01 1.34665385e-01 2.33330578e-01 -6.98003531e-01 1.36404738e-01 5.06994963e-01 3.92157137e-01 -4.79001731e-01 -1.95273504e-01 7.15448737e-01 9.04829353e-02 -8.85278225e-01 2.80228287e-01 -2.26185292e-01 1.50854602e-01 5.20475745e-01 1.79261729e-01 -2.81793803e-01 -2.89358526e-01 -9.13673580e-01 6.93567619e-02 -1.16887046e-02 7.49518156e-01 2.92251736e-01 -1.57726955e+00 -6.77070916e-01 1.01711266e-01 3.10068339e-01 -2.21248806e-01 2.60758042e-01 6.21519208e-01 -2.54849587e-02 7.16397524e-01 -8.62712692e-03 -7.24501073e-01 -1.59946692e+00 2.80759066e-01 3.47168408e-02 -4.14279848e-01 -3.66946012e-01 1.06776953e+00 -6.25644922e-02 -2.88169503e-01 8.13353956e-01 -3.22097361e-01 -3.00658971e-01 1.69416532e-01 8.37944806e-01 4.09340113e-01 5.34910500e-01 -4.84978884e-01 -3.29302669e-01 3.65987420e-01 -2.77744055e-01 -1.70580134e-01 1.34465957e+00 9.30279791e-02 3.57910007e-01 3.87510091e-01 9.51547980e-01 -2.43095160e-01 -9.40894604e-01 -2.94897020e-01 -2.08494104e-02 -3.43511969e-01 3.11212838e-01 -1.18574810e+00 -1.04059076e+00 1.07892549e+00 1.07117450e+00 3.78299445e-01 1.22271335e+00 1.88371405e-01 8.46527398e-01 1.89050451e-01 2.50279456e-01 -1.17526627e+00 1.04232185e-01 7.31226206e-01 6.84677660e-01 -1.07395673e+00 -1.35724604e-01 -3.08396101e-01 -6.10635281e-01 4.91349697e-01 6.99365795e-01 1.85500383e-01 7.56853938e-01 5.30921578e-01 1.84683725e-01 2.81396322e-02 -1.09949958e+00 -3.08605820e-01 5.88778377e-01 6.12579703e-01 7.79010713e-01 9.68992934e-02 1.68275252e-01 6.94183409e-01 -3.95649701e-01 5.71234748e-02 -2.72408426e-02 8.72265995e-01 -6.84069395e-01 -1.22575486e+00 -1.88228786e-01 4.79729772e-01 -8.46332908e-01 -3.19409460e-01 -4.77666706e-01 7.47460306e-01 2.83787161e-01 9.48024213e-01 1.42853588e-01 -8.23874235e-01 4.53056723e-01 3.03477377e-01 3.24451357e-01 -6.61193669e-01 -7.98972487e-01 2.82499015e-01 2.78299719e-01 -1.65496975e-01 -6.11275554e-01 -3.71706218e-01 -9.12192523e-01 -1.96212783e-01 -7.17786610e-01 3.26209396e-01 6.58545852e-01 6.50347710e-01 6.05712295e-01 8.41553450e-01 2.62923688e-01 -7.03959286e-01 -3.77469808e-01 -1.39473200e+00 -5.80447018e-01 8.14066678e-02 3.79255489e-02 -6.93560064e-01 -1.97619364e-01 2.23540708e-01]
[15.213525772094727, 5.124805927276611]
21d342c4-ed63-465c-aa96-b7c389e20648
motion-magnification-in-robotic-sonography
2307.03698
null
https://arxiv.org/abs/2307.03698v1
https://arxiv.org/pdf/2307.03698v1.pdf
Motion Magnification in Robotic Sonography: Enabling Pulsation-Aware Artery Segmentation
Ultrasound (US) imaging is widely used for diagnosing and monitoring arterial diseases, mainly due to the advantages of being non-invasive, radiation-free, and real-time. In order to provide additional information to assist clinicians in diagnosis, the tubular structures are often segmented from US images. To improve the artery segmentation accuracy and stability during scans, this work presents a novel pulsation-assisted segmentation neural network (PAS-NN) by explicitly taking advantage of the cardiac-induced motions. Motion magnification techniques are employed to amplify the subtle motion within the frequency band of interest to extract the pulsation signals from sequential US images. The extracted real-time pulsation information can help to locate the arteries on cross-section US images; therefore, we explicitly integrated the pulsation into the proposed PAS-NN as attention guidance. Notably, a robotic arm is necessary to provide stable movement during US imaging since magnifying the target motions from the US images captured along a scan path is not manually feasible due to the hand tremor. To validate the proposed robotic US system for imaging arteries, experiments are carried out on volunteers' carotid and radial arteries. The results demonstrated that the PAS-NN could achieve comparable results as state-of-the-art on carotid and can effectively improve the segmentation performance for small vessels (radial artery).
['Zhongliang Jiang', 'Nassir Navab', 'Yuan Bi', 'Dianye Huang']
2023-07-07
null
null
null
null
['motion-magnification']
['computer-vision']
[ 4.78144735e-03 9.84445959e-02 -1.97414249e-01 -2.56499708e-01 -5.03777564e-01 -5.50625384e-01 -1.55266792e-01 -4.81115341e-01 -1.39561966e-01 4.70940739e-01 3.66915427e-02 -4.33745533e-01 -2.23460466e-01 -4.46961671e-01 -4.40794677e-01 -8.90200913e-01 -3.98270249e-01 2.07083479e-01 4.26704735e-01 -1.20275110e-01 1.21668912e-01 5.97010732e-01 -9.01561797e-01 -3.95186573e-01 1.23461723e+00 8.40851426e-01 4.09188151e-01 5.28269112e-01 -1.67848244e-01 3.77059996e-01 -3.74147028e-01 2.59152949e-02 2.22958893e-01 -6.35987997e-01 -6.98675215e-01 4.98099253e-02 7.78683722e-02 -6.62704706e-01 -5.56763768e-01 1.10031784e+00 9.56319392e-01 2.53047645e-01 3.83080512e-01 -4.83795971e-01 -6.59432828e-01 8.47061455e-01 -9.71999824e-01 8.16750467e-01 -1.48333013e-01 3.90932620e-01 4.10407811e-01 -2.93213755e-01 7.84430146e-01 9.86830473e-01 4.55656886e-01 3.43812823e-01 -8.78888607e-01 -5.13927877e-01 -1.41056359e-01 4.09368962e-01 -8.37768972e-01 -3.57620865e-02 1.29841733e+00 -3.63542676e-01 1.96754009e-01 3.77496868e-01 8.43025208e-01 6.65425718e-01 3.06374371e-01 7.96922982e-01 7.74598122e-01 -2.02810094e-01 -5.91840921e-03 -1.25898615e-01 3.85058850e-01 6.68046713e-01 1.70534134e-01 2.22372606e-01 2.77653515e-01 2.30517223e-01 1.56864762e+00 -1.04908787e-01 -7.42299795e-01 -8.72644335e-02 -1.22891057e+00 5.51478803e-01 8.56554866e-01 7.43946612e-01 -5.82888007e-01 4.28433381e-02 4.74387586e-01 -2.10486844e-01 1.21506536e-02 5.20474732e-01 -1.23430520e-01 -4.55159128e-01 -5.32820880e-01 -4.47718471e-01 2.84755588e-01 3.46536368e-01 -2.13322118e-02 5.43217063e-01 -2.59349465e-01 6.68869793e-01 2.27594763e-01 5.23111641e-01 1.03215027e+00 -1.27287948e+00 3.89849022e-02 3.34503651e-01 -1.55096566e-02 -1.08588266e+00 -7.63178825e-01 -7.19466090e-01 -1.04889214e+00 -7.82402083e-02 6.47836745e-01 -3.26720953e-01 -9.87381637e-01 1.38494110e+00 6.88652635e-01 5.24099946e-01 -1.81363523e-01 1.81900680e+00 1.06192470e+00 3.84460866e-01 5.32285273e-02 -4.44665760e-01 1.50878739e+00 -6.92368150e-01 -9.11862195e-01 3.43130864e-02 5.69706321e-01 -5.64535201e-01 1.20049655e+00 4.15106006e-02 -1.00235415e+00 -8.35752249e-01 -9.06542659e-01 4.68743384e-01 4.77181166e-01 3.79052818e-01 7.03910291e-01 6.70343876e-01 -4.94854122e-01 7.73898721e-01 -1.47007620e+00 1.82752371e-01 4.57951963e-01 7.78456032e-02 7.11197928e-02 3.07913899e-01 -1.34495902e+00 6.50516868e-01 6.10178225e-02 5.83228052e-01 -1.51574105e-01 -9.28761184e-01 -6.51547253e-01 -1.41046390e-01 3.19335639e-01 -2.97986507e-01 1.14878738e+00 -6.27961636e-01 -1.89401960e+00 2.77455479e-01 -5.00354916e-02 -1.84781447e-01 6.33679628e-01 -8.06856826e-02 -4.51086700e-01 1.00616384e+00 -7.83996582e-02 2.03147471e-01 7.56741583e-01 -9.77125406e-01 -9.77468863e-02 -3.84703904e-01 -2.89588898e-01 -1.98738858e-01 -1.41456217e-01 -1.47119015e-01 -3.51048410e-01 -9.49981511e-01 6.78596318e-01 -1.17404306e+00 -1.79089770e-01 1.70410335e-01 -2.86904037e-01 2.28786934e-02 8.02191019e-01 -1.27008498e+00 1.23027885e+00 -1.95005143e+00 6.60107508e-02 3.78648549e-01 2.22956061e-01 6.85751498e-01 -4.01837714e-02 -6.39941752e-01 -1.32109553e-01 -3.33399791e-03 -4.01943445e-01 6.26480043e-01 -5.92730522e-01 2.00203821e-01 -1.23186400e-02 5.62012434e-01 1.29371285e-02 1.10485744e+00 -8.01763237e-01 -6.96780562e-01 2.99000293e-01 4.08050001e-01 -1.12524077e-01 1.45647809e-01 1.14573300e-01 1.31504703e+00 -8.30633581e-01 5.73197961e-01 8.07858527e-01 -1.47422820e-01 -9.32005607e-03 -7.06520617e-01 -9.14351195e-02 -1.91578045e-01 -1.26883757e+00 1.61448145e+00 -4.88913774e-01 4.52264577e-01 1.90576151e-01 -1.17953277e+00 7.40341604e-01 7.09008455e-01 7.24219143e-01 -9.11590934e-01 6.79069221e-01 3.13307434e-01 6.63379610e-01 -1.40271664e+00 -4.54978287e-01 1.79216415e-01 4.74175304e-01 6.31878197e-01 -3.43916148e-01 4.92960960e-03 2.71604538e-01 -1.02070875e-01 5.23284674e-01 2.57254064e-01 -1.00712195e-01 -2.26813689e-01 7.54652858e-01 -6.37404472e-02 5.87445974e-01 4.90934998e-01 -5.92280030e-01 5.06298482e-01 2.14932024e-01 -6.87383354e-01 -6.39881730e-01 -9.34166253e-01 -5.29530764e-01 3.42891723e-01 6.69975281e-01 4.86155808e-01 -5.57729721e-01 -5.37851095e-01 -1.50594711e-01 3.89890283e-01 -4.57685828e-01 -9.38360393e-02 -1.26449740e+00 -7.02116132e-01 4.18464929e-01 7.96693027e-01 6.64710999e-01 -1.08858669e+00 -1.15875709e+00 5.57762027e-01 -5.73184192e-01 -9.97022390e-01 -5.18971860e-01 -3.95562410e-01 -1.10258269e+00 -1.03407383e+00 -1.27957582e+00 -8.05268884e-01 6.66884959e-01 1.77020162e-01 4.85312194e-01 -6.99274391e-02 -7.58104146e-01 -7.72299469e-02 -1.98753819e-01 -1.53109767e-02 -5.29259861e-01 -2.49078810e-01 -1.90681979e-01 -9.10168365e-02 -1.02254078e-01 -7.70349741e-01 -1.05340183e+00 4.00333881e-01 -5.74678600e-01 -1.54643819e-01 7.60496855e-01 6.60232008e-01 4.34775323e-01 1.04446396e-01 7.29404449e-01 -4.68088299e-01 4.98898357e-01 -1.05777390e-01 -4.86311495e-01 -7.33127771e-03 3.02021261e-02 2.35813055e-02 5.43876529e-01 -9.65264142e-01 -1.29965889e+00 -8.46113339e-02 -4.05742735e-01 -5.01074553e-01 -2.30777040e-01 6.29559100e-01 4.62073386e-01 -2.87584722e-01 8.05824697e-01 4.92663831e-01 6.04227185e-01 -2.64865130e-01 1.99489772e-01 6.69178069e-01 8.14706326e-01 -5.20962894e-01 3.72350007e-01 3.96401703e-01 2.00647682e-01 -9.57467377e-01 -3.06453943e-01 -3.46206129e-01 -7.25917935e-01 -5.54766536e-01 9.30311084e-01 -2.85212129e-01 -9.64269102e-01 3.56709540e-01 -1.04285276e+00 3.85461748e-03 -8.83779526e-02 9.55494106e-01 -1.63985461e-01 1.05350554e+00 -1.23881114e+00 -6.31002188e-01 -7.63500035e-01 -1.60054326e+00 5.56788921e-01 6.79580748e-01 -6.73821643e-02 -7.12829947e-01 -3.18191290e-01 1.99245349e-01 7.79927671e-01 4.71956253e-01 8.97137582e-01 -1.40070736e-01 -4.88796979e-01 -2.66705543e-01 -3.03854048e-01 -2.16010585e-02 4.61311430e-01 1.41056692e-02 -6.18123114e-01 1.12440594e-01 8.69752318e-02 6.29966706e-02 4.58424568e-01 1.19588101e+00 1.24914658e+00 -3.66376489e-02 -2.61497349e-01 5.78897774e-01 8.98796141e-01 6.02353215e-01 4.22828525e-01 1.12711146e-01 6.47835433e-01 6.41568363e-01 5.22264957e-01 2.85015851e-01 -1.87337294e-01 4.12164330e-01 2.52268106e-01 -2.28361994e-01 -2.80149728e-01 4.60585624e-01 -2.55776703e-01 7.21741259e-01 -5.81948638e-01 4.78733897e-01 -4.89826918e-01 3.73235226e-01 -1.49955654e+00 -6.88085079e-01 -5.47393441e-01 1.90373015e+00 1.02577829e+00 1.44969076e-01 1.01181440e-01 5.33486484e-03 7.24481642e-01 -1.64799675e-01 -8.39493394e-01 1.27166316e-01 2.05985069e-01 2.28042290e-01 3.08904141e-01 2.40561485e-01 -1.19411445e+00 1.97165772e-01 5.30944109e+00 4.83001530e-01 -1.86637938e+00 4.36969288e-02 5.63856959e-01 3.21711540e-01 5.42462841e-02 -4.80067730e-01 -1.61227793e-01 6.81787014e-01 6.86968863e-01 2.42012948e-01 3.67208980e-02 6.41732812e-01 6.70537412e-01 2.81259604e-02 -4.64761257e-01 7.62676239e-01 -5.70170820e-01 -1.23947692e+00 -3.85466993e-01 -3.11795473e-01 3.86907190e-01 -2.95629233e-01 -1.75364409e-02 -1.22020781e-01 -2.42873803e-01 -4.69276696e-01 2.36970350e-01 4.87791061e-01 8.43367636e-01 -5.61007082e-01 8.64339054e-01 2.34582499e-01 -1.18348181e+00 1.39029562e-01 -2.20253199e-01 3.86455178e-01 7.34519780e-01 5.82066298e-01 -5.58104038e-01 6.31110013e-01 5.17010450e-01 6.24454021e-01 -3.61149430e-01 1.08119667e+00 -2.16174141e-01 9.69690084e-01 -4.20985073e-01 1.16317943e-01 1.94528818e-01 -5.38561821e-01 8.58621240e-01 9.11095619e-01 5.69493055e-01 3.54696542e-01 7.50339031e-02 1.05856442e+00 4.28152382e-01 1.28557347e-02 1.28198965e-02 1.15552224e-01 3.88695598e-01 1.32864761e+00 -1.10674334e+00 -3.84414822e-01 -2.46118471e-01 5.22732973e-01 -4.74915653e-01 5.27702987e-01 -9.29567516e-01 -5.63660145e-01 -1.26800044e-02 2.36594096e-01 9.79806036e-02 -2.79504746e-01 -4.88310367e-01 -9.10673440e-01 -1.05307950e-03 -3.80206794e-01 3.18250597e-01 -6.98916256e-01 -8.87893617e-01 6.86621249e-01 -1.06652506e-01 -1.40380764e+00 -1.96684375e-01 -2.67677844e-01 -8.38489532e-01 8.56146932e-01 -1.52379787e+00 -7.42926776e-01 -6.26661658e-01 1.56376883e-01 4.91190642e-01 2.09249392e-01 3.84845406e-01 3.01139861e-01 -6.49227262e-01 3.68559867e-01 -3.31166476e-01 3.67042094e-01 6.63876534e-01 -1.11020577e+00 -2.56253220e-02 9.33433592e-01 -4.79842931e-01 7.99588740e-01 6.69651628e-01 -8.00642788e-01 -1.32565558e+00 -8.26747894e-01 -2.02508748e-01 3.49138200e-01 5.11417449e-01 6.65801346e-01 -1.15697873e+00 3.78994912e-01 -1.85783897e-02 5.08691132e-01 3.24538171e-01 -8.51669610e-01 3.49897355e-01 1.20341770e-01 -1.10342216e+00 5.44996977e-01 6.95918918e-01 1.45962551e-01 -5.35839081e-01 1.16842091e-01 7.83867896e-01 -9.69261169e-01 -1.06595623e+00 5.12933910e-01 6.84684098e-01 -3.31811577e-01 1.16883206e+00 -3.82280648e-01 6.69996858e-01 -2.82997668e-01 6.82040513e-01 -1.22432327e+00 -4.28920031e-01 -5.43522835e-01 -3.13404292e-01 7.81718910e-01 -4.11664248e-02 -9.28371966e-01 8.17950010e-01 5.41936815e-01 -3.52319330e-01 -6.40348494e-01 -9.65620041e-01 -5.21118402e-01 -4.85208221e-02 -1.41694456e-01 2.09402338e-01 8.84017885e-01 -3.14309373e-02 -9.36507583e-02 -3.86789478e-02 4.14914578e-01 7.75602281e-01 4.03486580e-01 2.66250402e-01 -1.21609259e+00 -3.67465228e-01 -6.83228493e-01 -2.98707455e-01 -1.22576535e+00 -2.25746959e-01 -6.59389138e-01 -1.41231805e-01 -1.47033894e+00 -3.21350396e-01 -5.99646628e-01 8.00872371e-02 1.30899817e-01 -6.07850492e-01 2.84254491e-01 -2.26537317e-01 4.15522307e-01 3.05754930e-01 2.93102890e-01 2.43284702e+00 -1.22165695e-01 -7.13965356e-01 5.38400471e-01 -3.30288947e-01 5.39588869e-01 5.94108403e-01 8.07959139e-02 -4.07948196e-01 -4.89038527e-02 -5.55277169e-01 7.04331458e-01 1.05205521e-01 -7.08893776e-01 1.60393029e-01 2.97523979e-02 3.62454891e-01 -5.39066970e-01 -4.23741713e-02 -8.76252592e-01 3.77388224e-02 1.00933862e+00 -2.21003503e-01 -4.54063863e-01 1.17440321e-01 2.30641752e-01 -2.28055954e-01 -3.52681726e-01 1.04454911e+00 -4.21827495e-01 -6.67647004e-01 1.10101797e-01 -5.36753893e-01 -7.59535059e-02 1.09563768e+00 -1.74481705e-01 -1.78570420e-01 -2.64071792e-01 -1.09737313e+00 9.07578543e-02 -4.09457773e-01 7.59702027e-02 6.67610109e-01 -1.03839421e+00 -6.13645077e-01 4.98975396e-01 -2.92771161e-01 1.86282724e-01 9.24858272e-01 1.56691396e+00 -9.73459482e-01 3.00691366e-01 -1.78039178e-01 -1.02547812e+00 -9.33766782e-01 4.01701480e-01 6.70798779e-01 4.22667935e-02 -1.19922388e+00 6.27922595e-01 -2.31065527e-01 -3.77727710e-02 2.16216311e-01 -6.69636786e-01 -5.47262192e-01 -3.00756723e-01 5.21990001e-01 5.91334581e-01 5.57497703e-02 -5.49519539e-01 -5.61210152e-04 1.04964709e+00 1.81826577e-01 2.40524262e-01 1.11794209e+00 -4.01997715e-01 9.58649665e-02 -2.18640640e-01 8.64149570e-01 -1.91278994e-01 -1.37764919e+00 -3.17164093e-01 -3.77346247e-01 -4.52233970e-01 3.13244551e-01 -7.57775366e-01 -1.70320094e+00 9.90925968e-01 9.13112998e-01 1.74808055e-01 1.08003235e+00 -1.01914078e-01 1.37943017e+00 -1.86402336e-01 1.75141409e-01 -6.20736539e-01 7.22254366e-02 -2.93965161e-01 8.01659405e-01 -1.20755363e+00 -2.10301936e-01 -8.84812057e-01 -7.29463935e-01 1.69861829e+00 6.78664505e-01 -1.43417984e-01 6.86865687e-01 2.48940095e-01 4.12111163e-01 2.08041240e-02 3.53816390e-01 1.44339222e-02 3.62079620e-01 4.55684543e-01 3.63560855e-01 -1.07861482e-01 -5.62312722e-01 5.32636225e-01 1.11585490e-01 2.55980134e-01 6.09840870e-01 1.11106217e+00 -5.29999793e-01 -4.99592900e-01 -5.98123193e-01 5.96399903e-01 -6.24988616e-01 5.14858365e-01 5.37710488e-01 6.11941278e-01 -2.86526501e-01 6.73092365e-01 -1.53661534e-01 1.23585194e-01 4.43527788e-01 -1.95272580e-01 5.07978320e-01 1.31265536e-01 -9.21266302e-02 6.71709239e-01 -2.04092279e-01 -4.35665935e-01 -4.26768810e-01 -4.77574944e-01 -1.84570336e+00 2.60755599e-01 -3.21204394e-01 -5.02894574e-04 6.11244917e-01 9.76616085e-01 2.07606807e-01 1.02932239e+00 7.03261554e-01 -1.09401011e+00 -5.86181998e-01 -1.14653683e+00 -5.66306710e-01 5.28729141e-01 3.03512812e-01 -7.74777114e-01 -4.68238801e-01 3.25632662e-01]
[14.331058502197266, -2.4940061569213867]
5b80a0c1-a827-4a76-94df-3984eee01429
enabling-joint-radar-communication-operation
2305.15069
null
https://arxiv.org/abs/2305.15069v1
https://arxiv.org/pdf/2305.15069v1.pdf
Enabling Joint Radar-Communication Operation in Shift Register-Based PMCW Radars
This article introduces adaptations to the conventional frame structure in binary phase-modulated continuous wave (PMCW) radars with sequence generation via linear-feedbck shift registers and additional processing steps to enable joint radar-communication (RadCom) operation. In this context, a preamble structure based on pseudorandom binary sequences (PRBSs) that is compatible with existing synchronization algorithms is outlined, and the allocation of pilot PRBS blocks is discussed. Finally, results from proof-of-concept measurements are presented to illustrate the effects of the choice of system and signal parameters and validate the investigated PMCW-based RadCom system and synchronization strategy.
['Thomas Zwick', 'Akanksha Bhutani', 'Yueheng Li', 'Theresa Antes', 'Benjamin Nuss', 'Axel Diewald', 'Elizabeth Bekker', 'Lucas Giroto de Oliveira']
2023-05-24
null
null
null
null
['joint-radar-communication']
['robots']
[ 6.29338801e-01 -1.47765195e-02 -3.63410823e-02 -3.62738401e-01 -5.33547997e-01 -3.48133028e-01 1.11482334e+00 -6.01555444e-02 -7.10650146e-01 1.30239534e+00 -3.77196521e-02 -7.58598447e-01 -9.05602753e-01 -4.64957267e-01 1.97235093e-01 -1.00044751e+00 -4.86622036e-01 2.42749110e-01 1.23231136e-03 -4.18050408e-01 3.90735924e-01 9.62511361e-01 -9.82797861e-01 -5.80588877e-01 2.47905612e-01 8.21482778e-01 1.68920487e-01 1.01312840e+00 3.10316354e-01 5.52888632e-01 -1.43274534e+00 1.17689101e-02 4.66233462e-01 -5.70725083e-01 1.30549267e-01 -3.00464034e-01 -8.35494697e-02 -1.19222130e-03 -5.47069609e-01 7.46180415e-01 6.02345109e-01 -2.39403516e-01 7.96795249e-01 -8.36133778e-01 1.71090782e-01 7.22493827e-01 -1.92706481e-01 3.92698556e-01 2.68608034e-01 1.94937065e-01 2.15002760e-01 -1.64423287e-01 5.35190225e-01 5.00672221e-01 6.21055841e-01 -1.51164439e-02 -1.02713811e+00 -1.09143865e+00 -7.80712485e-01 2.57372171e-01 -1.47949946e+00 -5.20456314e-01 6.62337303e-01 -1.99724272e-01 7.40362644e-01 4.35517758e-01 9.56980288e-01 5.97344041e-01 8.72988522e-01 -4.63795930e-01 1.30654514e+00 -9.70337033e-01 4.12858695e-01 -4.20316249e-01 6.94448709e-01 2.52472218e-02 1.15885150e+00 9.38334644e-01 -3.07225317e-01 -2.39080086e-01 4.03016776e-01 -6.38972461e-01 -7.39816248e-01 -5.03798604e-01 -1.03894269e+00 8.35966527e-01 -3.68567407e-01 6.96678400e-01 -4.33838546e-01 3.91986012e-01 -2.40115076e-01 5.75091064e-01 -1.52791794e-02 5.40364921e-01 1.20296903e-01 -1.48799926e-01 -1.22253466e+00 2.34618500e-01 1.04707098e+00 1.04139900e+00 4.22260523e-01 8.46513271e-01 -1.94571093e-01 -9.85707901e-03 5.28107107e-01 1.28449678e+00 1.00407422e-01 -3.11662465e-01 -1.19258724e-01 -5.24030089e-01 2.38398954e-01 -4.27340955e-01 -6.03671789e-01 -1.11466014e+00 -6.13963306e-01 3.13073665e-01 6.25228137e-02 -7.29814291e-01 -9.77105737e-01 1.10835552e+00 -2.66444795e-02 2.49171272e-01 6.82276845e-01 5.99417329e-01 4.99891847e-01 5.57848036e-01 -3.69923234e-01 -5.95304906e-01 1.46013582e+00 3.58656831e-02 -1.35750902e+00 -9.52384248e-02 1.28250942e-01 -1.02408588e+00 -6.27140164e-01 2.84338027e-01 -7.22788751e-01 -3.79658490e-01 -1.91050065e+00 1.19422209e+00 1.06023714e-01 5.31610064e-02 3.24373990e-01 1.57474518e+00 -7.95207381e-01 6.05303943e-02 -2.75381237e-01 1.12424381e-01 -2.24417806e-01 3.53586644e-01 2.51321346e-01 2.16009989e-01 -1.38813448e+00 1.13346338e+00 3.29412252e-01 3.92182052e-01 -4.48931605e-01 -6.15963757e-01 -8.44280601e-01 -3.21714491e-01 -2.05843136e-01 -2.91176558e-01 1.30271375e+00 -8.95842537e-02 -1.76435757e+00 2.36945495e-01 2.18076780e-01 -1.49675071e+00 6.84418902e-02 4.59380075e-02 -1.10061979e+00 6.36393607e-01 -4.01052684e-01 -1.16255596e-01 1.34193134e+00 -8.08981061e-01 -5.13686538e-01 4.25878793e-01 -5.79975367e-01 -3.64642978e-01 8.63992751e-01 -1.23228682e-02 7.39185631e-01 -6.12059355e-01 2.81850189e-01 -5.26024282e-01 -3.58582348e-01 -9.24313486e-01 1.84654281e-01 5.65214992e-01 7.77007103e-01 -2.10803404e-01 1.16188645e+00 -1.87701499e+00 -3.85013759e-01 7.16782987e-01 -1.91562429e-01 4.56779480e-01 2.36443654e-01 8.97560120e-01 -3.04068565e-01 -8.02913368e-01 -1.39455795e-01 5.94509661e-01 -7.31000751e-02 1.90236382e-02 -6.59726918e-01 9.53956723e-01 -7.74165466e-02 5.59483826e-01 -3.34184140e-01 1.48297131e-01 3.11082929e-01 1.55999348e-01 2.00575009e-01 -2.17211023e-01 1.30722791e-01 2.51688391e-01 -2.79345185e-01 3.83827448e-01 1.03204012e+00 5.30680239e-01 3.43821228e-01 -3.00794870e-01 -6.91451073e-01 2.29320288e-01 -1.07290268e+00 8.26467693e-01 -1.00515068e-01 9.08295572e-01 4.62165833e-01 -1.05171812e+00 1.49212456e+00 4.81009662e-01 3.59038979e-01 -7.57220447e-01 4.72251415e-01 1.20993294e-01 5.95899105e-01 -4.09811847e-02 7.36544549e-01 -3.84653181e-01 -1.23887368e-01 3.58133525e-01 2.53065854e-01 -5.40106654e-01 3.16973001e-01 1.24832898e-01 1.16266668e+00 5.48336050e-03 7.46031404e-01 -5.78421235e-01 9.11254168e-01 1.50664076e-01 2.15721399e-01 7.91660488e-01 1.87142417e-02 -3.16356830e-02 -1.42805139e-02 4.68160361e-02 -7.25463510e-01 -8.66415024e-01 -5.46022534e-01 -3.57976824e-01 3.95752341e-01 -3.46648484e-01 -3.00369486e-02 2.72329718e-01 -1.67465396e-03 9.27457929e-01 2.95460850e-01 -3.84136401e-02 -8.07368815e-01 -8.92717004e-01 8.45163047e-01 -3.93260121e-01 3.56878489e-01 -4.17951077e-01 -1.12063241e+00 5.97788155e-01 4.73057538e-01 -1.35721719e+00 4.38728839e-01 6.74262345e-01 -5.90868831e-01 -1.08756435e+00 -5.87068439e-01 -4.86092895e-01 4.40109253e-01 5.18111587e-01 6.09271824e-01 -4.36189353e-01 -4.38207239e-01 8.45575869e-01 -7.60790169e-01 -4.11228567e-01 -6.47880375e-01 -3.73642147e-01 -6.61149919e-02 -2.79823184e-01 3.52584004e-01 -4.06526506e-01 -1.96748719e-01 1.58882588e-01 -5.63814938e-01 -2.96500802e-01 8.61264884e-01 7.24587619e-01 -3.63929570e-01 3.94296646e-02 5.60476661e-01 -1.97589144e-01 6.23187602e-01 7.87662789e-02 -1.41360986e+00 6.73643425e-02 -5.12309015e-01 6.68251216e-02 1.31601691e-01 -7.71536157e-02 -7.96569109e-01 -2.63097167e-01 -1.48959622e-01 5.08548975e-01 -3.97406295e-02 4.20664459e-01 1.68171570e-01 -8.64682138e-01 5.15972674e-01 5.59514880e-01 3.77301365e-01 1.09837323e-01 2.42056921e-01 8.03896487e-01 7.22916603e-01 -2.34400421e-01 1.61989725e+00 1.93040550e-01 4.27300215e-01 -1.48678434e+00 -1.97010249e-01 -2.94457167e-01 -1.15226910e-01 -4.71624166e-01 5.10381997e-01 -9.29600656e-01 -5.36692679e-01 4.33404028e-01 -1.09733236e+00 7.17509985e-02 -4.05762047e-01 1.45231605e+00 -3.65721047e-01 5.64656198e-01 -1.98283792e-01 -1.28353357e+00 -2.80304998e-01 -5.42874873e-01 4.69887018e-01 1.28874972e-01 -2.46934220e-01 -4.59127039e-01 1.65571243e-01 -8.40377659e-02 7.81561911e-01 1.75587818e-01 5.20384371e-01 -4.20357525e-01 -1.07375383e+00 -5.78493118e-01 2.37682983e-01 -8.79800245e-02 -1.55597419e-01 -6.01086915e-01 -1.94179878e-01 -6.76327050e-01 5.73040605e-01 2.84089178e-01 4.99852568e-01 7.08008528e-01 -3.70818555e-01 1.54958904e-01 -4.39890474e-01 6.40396893e-01 1.59487283e+00 8.36431742e-01 8.99966657e-01 3.12831491e-01 -5.45711339e-01 3.18814511e-03 9.34117973e-01 4.94277418e-01 -2.78751254e-01 5.65817893e-01 -4.09404412e-02 4.61870521e-01 -1.32488266e-01 5.24414659e-01 3.77132356e-01 5.21365762e-01 7.98194855e-02 -1.65574208e-01 -3.82004023e-01 -1.10102212e-02 -9.44606960e-01 -1.16252983e+00 -5.99444509e-01 1.94623291e+00 5.30937970e-01 4.43579137e-01 -5.56195617e-01 3.84872377e-01 7.29964852e-01 5.87875605e-01 3.90526533e-01 -2.66750216e-01 -3.02935094e-01 1.25310361e+00 1.46164906e+00 8.63631725e-01 -7.65507519e-01 2.57536471e-01 6.44909191e+00 9.09617841e-01 -9.83067930e-01 2.46595275e-02 -6.09446168e-01 3.16502482e-01 -4.53802198e-02 3.89865041e-01 -1.15467930e+00 3.64567280e-01 1.35594571e+00 -5.83434522e-01 -2.07528785e-01 -4.43430319e-02 3.51740420e-02 -7.76016116e-01 -3.45391333e-01 8.38079929e-01 1.87077045e-01 -1.28975832e+00 -4.40272868e-01 1.44481972e-01 2.07192644e-01 -3.31494272e-01 -2.41585046e-01 3.22540998e-01 1.41219914e-01 -4.47928458e-01 8.05697620e-01 5.95541537e-01 6.34429574e-01 -5.28126597e-01 9.74045932e-01 1.98326409e-01 -1.08584249e+00 6.63042814e-02 -2.91127771e-01 -2.20491022e-01 8.96850586e-01 9.04580951e-01 -9.47729051e-01 1.30723941e+00 -2.16942519e-01 -1.80318981e-01 -2.13458240e-01 1.35595560e+00 -3.95542800e-01 9.78305936e-01 -5.32855392e-01 -5.50632179e-01 2.35956162e-01 -3.64229023e-01 9.99236703e-01 1.10262942e+00 1.01376545e+00 4.52308297e-01 -4.16885018e-01 2.08056286e-01 9.73354936e-01 -5.80107927e-01 -4.82350886e-01 1.03174902e-01 7.48956919e-01 1.06476188e+00 -2.94638306e-01 -1.54191673e-01 -1.67878553e-01 -9.11630914e-02 -1.00529861e+00 4.22824383e-01 -8.42590272e-01 -1.05016029e+00 5.31548440e-01 1.77406549e-01 7.67780960e-01 -1.16791058e+00 4.27919477e-02 -4.95130748e-01 -5.70961833e-01 -4.43052143e-01 -1.26429519e-03 -5.57649016e-01 -5.83853543e-01 6.25904799e-01 4.97577012e-01 -1.79979479e+00 -6.60824418e-01 -3.98916483e-01 -2.15118989e-01 9.82679069e-01 -1.80304682e+00 -7.90737212e-01 -1.82664394e-01 1.12268977e-01 -2.83046186e-01 -7.04523504e-01 7.45620072e-01 -1.11154251e-01 3.42419624e-01 1.00976966e-01 2.43351787e-01 -1.14739887e-01 3.68376464e-01 -5.28245926e-01 3.58940549e-02 1.17844522e+00 -7.93044418e-02 5.75462043e-01 1.23543417e+00 -9.43450093e-01 -1.62713218e+00 -5.32769918e-01 6.53279424e-01 4.51519102e-01 7.91917503e-01 -2.63039023e-01 3.55820581e-02 1.67096764e-01 9.05509233e-01 -6.53962076e-01 6.57364190e-01 -6.41191244e-01 2.94004411e-01 -1.94002673e-01 -1.01056266e+00 5.84986866e-01 2.46722847e-01 1.20995507e-01 -1.10913169e+00 5.15766442e-02 2.52201557e-01 -6.76435471e-01 -5.11120021e-01 7.11378813e-01 6.41509116e-01 -7.19481409e-01 1.01755321e+00 2.63068080e-01 -6.78580880e-01 -9.32744801e-01 -2.67379582e-01 -1.16538990e+00 -3.98049831e-01 -1.37459707e+00 8.39076415e-02 6.72775328e-01 -4.84230220e-02 -1.13660157e+00 6.80449247e-01 -6.49904668e-01 -1.74370736e-01 1.66620612e-02 -1.59518015e+00 -1.09079289e+00 -5.57319880e-01 -4.51556265e-01 3.21196735e-01 3.35415244e-01 9.30872858e-02 5.56482375e-01 -5.55690944e-01 5.76089740e-01 1.00444520e+00 1.93821877e-01 8.87895584e-01 -1.24065959e+00 -5.81091225e-01 -1.13683626e-01 -9.13473487e-01 -1.12520826e+00 -2.12481841e-01 -6.48983240e-01 1.52625097e-02 -1.13983667e+00 -1.02621150e+00 -3.46608192e-01 2.42718920e-01 -4.87069935e-01 5.53089857e-01 2.75860727e-01 2.24404782e-01 -7.74113983e-02 1.14678711e-01 5.77820539e-01 4.47520673e-01 -2.64781620e-02 1.67180717e-01 5.10313272e-01 -1.78201441e-02 1.35710537e-01 7.79348671e-01 -4.63686764e-01 -1.73610806e-01 5.36589026e-01 -1.22579888e-01 6.39216959e-01 3.49664420e-01 -1.72101319e+00 6.50803626e-01 1.96717560e-01 2.84939677e-01 -1.40465915e+00 6.20835423e-01 -1.18510175e+00 8.17168713e-01 1.42807245e+00 3.68654847e-01 -1.70813307e-01 1.40379206e-03 6.26630723e-01 -1.84387445e-01 -9.36792731e-01 8.31362247e-01 6.03433788e-01 -5.71535408e-01 -3.96602929e-01 -1.26661098e+00 -4.26816851e-01 1.52063632e+00 -6.96403444e-01 -8.36950541e-01 -5.13359189e-01 -4.42800999e-01 1.68416128e-01 -1.37428403e-01 -2.45191216e-01 5.83716094e-01 -9.95194554e-01 -8.50749731e-01 7.38603592e-01 -1.01095706e-01 -8.65448058e-01 1.83893055e-01 9.01853025e-01 -9.66850936e-01 1.16523790e+00 -3.84265423e-01 -6.23731196e-01 -1.45495367e+00 -1.73474729e-01 3.22265297e-01 -1.54137984e-01 -6.74975216e-01 2.20206529e-01 -8.84244919e-01 3.56603533e-01 2.84988843e-02 -2.13230744e-01 -2.53033459e-01 7.18389302e-02 6.09623134e-01 -5.35303820e-03 1.78084508e-01 -2.54584610e-01 -3.64712387e-01 7.66227484e-01 2.93042481e-01 -7.38862693e-01 1.02064300e+00 1.06182611e-02 -1.11795060e-01 -5.17813675e-02 2.41910085e-01 4.89168286e-01 -6.66058958e-01 -9.56677794e-02 2.35450879e-01 -2.83282697e-01 -1.59821808e-01 -4.29092169e-01 -3.87146413e-01 1.85007274e-01 6.95861936e-01 3.33462179e-01 9.02753711e-01 -5.81436872e-01 3.69234800e-01 7.61221409e-01 1.12610257e+00 -7.72449315e-01 -5.88677883e-01 6.79998338e-01 3.51252228e-01 1.01126581e-01 8.03069592e-01 -3.86249006e-01 -1.41948506e-01 1.56602931e+00 -3.73276114e-01 -2.58485496e-01 1.08982038e+00 1.07049477e+00 1.89594507e-01 -2.91455299e-01 -5.57946801e-01 -4.99679089e-01 -3.73374552e-01 1.10749304e+00 1.33678894e-02 -1.29524693e-01 -1.40717983e+00 3.53187233e-01 -6.26104653e-01 -1.21171556e-01 1.13487017e+00 1.44957256e+00 -1.08613384e+00 -1.43986118e+00 -1.20464158e+00 3.56090367e-01 -1.79409146e-01 -8.29215646e-02 1.61288217e-01 1.15359569e+00 -1.93979576e-01 9.29728687e-01 8.81499276e-02 -2.02214718e-01 2.75824368e-01 1.14571616e-01 8.18408370e-01 -2.17581764e-01 -3.74093324e-01 2.03811765e-01 4.96076256e-01 2.48740032e-01 -8.27527225e-01 -6.81168318e-01 -9.54016268e-01 6.58238605e-02 -6.12608790e-01 9.25154090e-01 7.99084246e-01 7.91000724e-01 -2.38557681e-01 7.84637928e-01 6.77228034e-01 -4.46409255e-01 -7.62717783e-01 -9.99981463e-01 -1.01914322e+00 -8.86345744e-01 3.53460461e-01 -4.37845379e-01 -3.07433605e-01 -2.94066966e-01]
[6.376824855804443, 1.2535187005996704]
0b8431ec-88b9-4f04-b94a-93e5f7145818
content-based-table-retrieval-for-web-queries
1706.02427
null
http://arxiv.org/abs/1706.02427v1
http://arxiv.org/pdf/1706.02427v1.pdf
Content-Based Table Retrieval for Web Queries
Understanding the connections between unstructured text and semi-structured table is an important yet neglected problem in natural language processing. In this work, we focus on content-based table retrieval. Given a query, the task is to find the most relevant table from a collection of tables. Further progress towards improving this area requires powerful models of semantic matching and richer training and evaluation resources. To remedy this, we present a ranking based approach, and implement both carefully designed features and neural network architectures to measure the relevance between a query and the content of a table. Furthermore, we release an open-domain dataset that includes 21,113 web queries for 273,816 tables. We conduct comprehensive experiments on both real world and synthetic datasets. Results verify the effectiveness of our approach and present the challenges for this task.
['Junwei Bao', 'Nan Duan', 'Zhao Yan', 'Yuanhua Lv', 'Zhoujun Li', 'Ming Zhou', 'Duyu Tang']
2017-06-08
null
null
null
null
['table-retrieval']
['natural-language-processing']
[ 1.64966583e-01 -2.85063572e-02 -4.23691183e-01 -4.48463678e-01 -1.31874251e+00 -7.72713363e-01 4.83565986e-01 9.24647331e-01 -4.27300692e-01 7.51294971e-01 6.02748811e-01 -7.92284161e-02 -2.00511098e-01 -1.08937705e+00 -7.63350070e-01 1.35362118e-01 2.72633508e-03 8.93595695e-01 3.68437320e-01 -5.46392620e-01 3.89312416e-01 1.18295237e-01 -1.33910596e+00 7.87774622e-01 7.24644363e-01 1.37884986e+00 -9.06881690e-02 3.37464035e-01 -4.92702633e-01 1.14760375e+00 -6.33175492e-01 -8.12619984e-01 1.84306651e-01 -1.85314968e-01 -1.24563849e+00 -1.81819946e-01 4.66937691e-01 -1.27388209e-01 -4.10070032e-01 1.02826691e+00 3.75653654e-01 1.56102657e-01 5.11626244e-01 -1.00760913e+00 -6.75442278e-01 1.03664720e+00 -2.00702429e-01 3.47875297e-01 5.83229244e-01 -5.14800429e-01 1.48885286e+00 -9.10539448e-01 8.24932277e-01 1.28477859e+00 3.86136115e-01 2.10774943e-01 -8.86517704e-01 -3.49143028e-01 1.75378434e-02 1.99209929e-01 -1.40967965e+00 -5.95990300e-01 6.53209984e-01 -9.77992415e-02 1.04674661e+00 3.26940417e-01 1.20782517e-01 6.58526301e-01 8.53939652e-02 7.36860335e-01 5.44580817e-01 -4.77144182e-01 1.44736797e-01 4.36880499e-01 1.67266950e-01 5.39481521e-01 6.47811592e-01 -6.42013609e-01 -6.29458010e-01 -3.22196782e-01 3.16597164e-01 -1.95152789e-01 1.02171764e-01 -5.66338062e-01 -1.32834268e+00 7.32291877e-01 5.57627499e-01 3.38443846e-01 -3.47186565e-01 -8.89041200e-02 5.58242142e-01 4.09765244e-01 3.12196404e-01 8.49615455e-01 -5.75022221e-01 7.21883029e-02 -6.15540683e-01 4.21403795e-01 1.19757998e+00 1.23990512e+00 7.37960696e-01 -6.16181195e-01 -4.49172318e-01 8.44055772e-01 5.25339618e-02 4.04887974e-01 1.66429073e-01 -9.45983112e-01 1.10405290e+00 1.13223386e+00 1.78299069e-01 -1.35213685e+00 -3.39416474e-01 -1.95034027e-01 -6.64086640e-01 -5.77163696e-01 5.20741224e-01 3.13321650e-01 -4.67365652e-01 1.41592717e+00 1.62697524e-01 -9.11165357e-01 2.73916572e-01 6.20841682e-01 1.03467691e+00 6.11501694e-01 -4.21559587e-02 -6.44740323e-03 1.50612664e+00 -7.86017358e-01 -8.27601433e-01 -2.65542448e-01 6.45535111e-01 -6.98901236e-01 1.09508741e+00 4.15498614e-02 -1.30541778e+00 -3.60977560e-01 -8.47993255e-01 -5.01607299e-01 -9.90705311e-01 1.98500305e-01 5.04067183e-01 3.35201919e-01 -8.74400496e-01 2.95209110e-01 -3.64985943e-01 -6.13544762e-01 3.07007581e-01 3.12454700e-01 -3.26747775e-01 -2.63420969e-01 -1.45013082e+00 7.36483693e-01 6.56638324e-01 -1.33829072e-01 -2.12375551e-01 -5.78670025e-01 -7.74503112e-01 4.35287654e-01 9.04435754e-01 -5.67750216e-01 1.27201116e+00 -3.10551882e-01 -7.36763835e-01 1.06677985e+00 -3.35638940e-01 -5.52758694e-01 1.45838290e-01 -2.91688740e-01 -4.44259465e-01 2.80224979e-01 3.87170732e-01 6.30000472e-01 1.46055192e-01 -1.11689520e+00 -6.23636246e-01 -4.41431344e-01 1.81156591e-01 2.71340191e-01 -5.94590545e-01 3.52988422e-01 -9.65886295e-01 -6.55429184e-01 1.43693119e-01 -5.21718323e-01 2.88478434e-02 -2.01448314e-02 -5.87563396e-01 -4.19069111e-01 2.76823699e-01 -6.87685132e-01 1.56872976e+00 -1.83639240e+00 -2.38618091e-01 3.38518620e-01 3.02691787e-01 -7.33940452e-02 4.44668494e-02 8.94526482e-01 2.91800946e-01 3.14017147e-01 -2.61957765e-01 1.73054487e-01 1.27619714e-01 -1.58250332e-02 -6.31968975e-01 -2.68276632e-01 2.53827244e-01 1.09968531e+00 -6.70721650e-01 -7.51644790e-01 -2.94543266e-01 1.14874527e-01 -6.17526472e-01 9.45467129e-02 -5.80702543e-01 -2.53143281e-01 -6.57873571e-01 9.38920975e-01 2.19519228e-01 -5.87322831e-01 3.38910788e-01 -2.84121811e-01 3.91749173e-01 8.40661645e-01 -1.14160502e+00 1.67778468e+00 -1.85643151e-01 4.28301334e-01 -2.82173187e-01 -7.48058736e-01 1.02509761e+00 3.80132124e-02 5.69051623e-01 -1.18652081e+00 9.10939947e-02 3.35137039e-01 -3.04915577e-01 -3.57790411e-01 9.12855625e-01 3.68189752e-01 -4.92982537e-01 5.22767186e-01 -2.84749657e-01 5.34519069e-02 7.50773311e-01 5.84171832e-01 1.22432864e+00 -2.10763529e-01 4.47972864e-01 -3.68102401e-01 7.83758283e-01 5.05345821e-01 1.51050642e-01 9.36501384e-01 -1.73595697e-02 3.93925279e-01 6.50283992e-01 -6.48450375e-01 -8.99993598e-01 -9.01443481e-01 7.10239783e-02 1.31097317e+00 2.87361681e-01 -7.50325441e-01 -5.33290267e-01 -8.12875271e-01 4.17927742e-01 4.94455040e-01 -6.92927182e-01 -2.35482067e-01 -7.34873116e-01 -5.80391407e-01 4.01695967e-01 6.64545655e-01 5.70678830e-01 -1.18114984e+00 -3.11060905e-01 2.89488465e-01 -7.06527412e-01 -1.29169357e+00 -5.65159142e-01 2.26648405e-01 -7.26611137e-01 -1.21963906e+00 -1.47098064e-01 -8.62986863e-01 4.00660694e-01 2.92455018e-01 1.93079185e+00 1.28929973e-01 -1.58519536e-01 2.44679645e-01 -3.04109275e-01 -4.88475978e-01 -3.82362276e-01 5.23198605e-01 -3.19785684e-01 -3.76112252e-01 8.56305480e-01 8.81048851e-03 -3.41736734e-01 4.11256075e-01 -1.00215435e+00 -2.70714134e-01 5.83122611e-01 5.47608256e-01 6.65257454e-01 1.68197781e-01 4.44133103e-01 -1.04388046e+00 9.20392632e-01 -5.35507560e-01 -7.57435501e-01 7.57557690e-01 -8.32999408e-01 4.91423309e-01 5.95900953e-01 5.29385284e-02 -8.75543773e-01 -1.18333839e-01 1.26802698e-01 7.85766840e-02 1.31029755e-01 8.38180542e-01 -3.07603210e-01 4.37090099e-01 7.77138531e-01 1.60398036e-01 -3.12038958e-01 -5.89214146e-01 3.35129321e-01 6.49632335e-01 5.64533770e-01 -8.21185291e-01 8.16226959e-01 5.33471584e-01 -1.68269902e-01 -3.26829851e-01 -1.19430959e+00 -5.75811386e-01 -6.35701895e-01 2.11906999e-01 5.24982452e-01 -9.82250333e-01 -7.58385003e-01 -1.79419845e-01 -9.34407711e-01 -3.23319919e-02 -2.71280348e-01 9.86945480e-02 -3.68099034e-01 2.23533548e-02 -5.66366076e-01 -4.37804759e-01 -4.39534962e-01 -7.64017045e-01 1.15927756e+00 -2.48725377e-02 -2.95962155e-01 -8.73911381e-01 5.50707094e-02 5.41964173e-01 6.02668941e-01 -9.00195390e-02 1.31340885e+00 -1.20303798e+00 -8.40716004e-01 -4.24323857e-01 -5.91599643e-01 -2.02829570e-01 2.91898668e-01 -1.89123437e-01 -6.68350577e-01 -3.05254571e-02 -2.68484503e-01 -7.64341950e-01 1.02030587e+00 -9.10792053e-02 1.23850954e+00 -4.50511545e-01 -3.15025628e-01 3.36568534e-01 1.54224741e+00 4.37737375e-01 4.83678043e-01 7.20773697e-01 5.91656804e-01 9.86064315e-01 8.38040709e-01 4.52847093e-01 8.49225819e-01 5.71592808e-01 1.49430469e-01 2.12372154e-01 9.71557647e-02 -6.29940450e-01 -1.60755545e-01 7.13061273e-01 5.10432541e-01 -6.68412089e-01 -1.09872854e+00 4.94035631e-01 -1.59376323e+00 -8.63821805e-01 2.75024027e-01 2.14356613e+00 1.07263470e+00 4.68785763e-01 3.39322127e-02 2.19305754e-01 5.54050446e-01 8.02873895e-02 -5.53703904e-01 -2.50735819e-01 -2.76295334e-01 9.42816958e-02 3.96622598e-01 2.40284055e-01 -1.31206119e+00 1.08765209e+00 6.60756397e+00 5.63880265e-01 -6.20361447e-01 -5.49385905e-01 8.49654317e-01 4.79094312e-02 -4.66905922e-01 -2.20559165e-01 -1.03454173e+00 8.80738944e-02 1.00230277e+00 -5.89821815e-01 4.86681759e-01 7.45646060e-01 -2.38609269e-01 -1.78328797e-01 -1.28083241e+00 8.43355775e-01 2.62274653e-01 -1.46257591e+00 4.29007918e-01 -2.05792934e-01 3.62846076e-01 -2.82715917e-01 -5.10925874e-02 5.28522849e-01 2.58847564e-01 -9.42236423e-01 5.05533040e-01 3.17932904e-01 7.00017810e-01 -6.86088562e-01 7.13193715e-01 -4.63522039e-02 -1.32725072e+00 2.99638771e-02 -5.19611776e-01 4.87845354e-02 -5.55856526e-01 2.95366675e-01 -7.31794417e-01 5.26036382e-01 1.00844967e+00 7.30497718e-01 -1.04495466e+00 6.87815130e-01 1.14926413e-01 1.97814316e-01 -2.60664850e-01 -3.36737454e-01 8.50554928e-02 1.65309250e-01 4.55090255e-02 1.13298106e+00 -2.15656847e-01 -1.59930456e-02 1.03219591e-01 8.54880691e-01 -6.95115209e-01 4.38011557e-01 -7.05830753e-01 -2.38404900e-01 6.36661947e-01 1.14077413e+00 -9.52551842e-01 -7.05916047e-01 -4.18297440e-01 5.43424129e-01 5.65135837e-01 2.24204063e-01 -4.32211339e-01 -6.15546405e-01 5.19442201e-01 5.81513485e-03 2.70255208e-01 8.96488428e-02 -4.61810589e-01 -1.28262401e+00 6.02339149e-01 -1.11468863e+00 1.00213599e+00 -7.51161695e-01 -1.38312745e+00 7.83074081e-01 1.04466416e-01 -1.00494480e+00 -4.75198358e-01 -4.57462490e-01 -5.00143692e-02 7.74344146e-01 -1.63948631e+00 -5.66987991e-01 -2.98575073e-01 5.53891361e-01 4.75355804e-01 -1.09154798e-01 7.89323568e-01 4.25147623e-01 -3.24021637e-01 5.86540103e-01 2.48562127e-01 6.20768666e-01 8.92847657e-01 -1.37120557e+00 7.07214832e-01 6.10004127e-01 3.69563431e-01 8.12496781e-01 5.24018288e-01 -7.30959535e-01 -1.73965037e+00 -1.00433707e+00 1.54312634e+00 -9.31246877e-01 7.22788513e-01 -5.89535415e-01 -1.07768941e+00 5.61672747e-01 2.53981411e-01 -9.54065174e-02 6.05460584e-01 2.00574398e-01 -6.61826134e-01 -3.99005353e-01 -1.11796439e+00 4.72018301e-01 1.02887988e+00 -7.16785729e-01 -7.38281906e-01 4.48441088e-01 9.14706409e-01 -4.08306748e-01 -9.77028072e-01 3.28631312e-01 5.65888107e-01 -7.52602935e-01 1.07479644e+00 -7.81906128e-01 7.13937700e-01 -5.38596734e-02 -4.67209339e-01 -8.74850273e-01 -1.36709973e-01 -3.49574387e-01 -1.84808031e-01 1.14897668e+00 6.95279777e-01 -2.83417761e-01 1.07381392e+00 8.60143065e-01 3.30724776e-01 -7.41330445e-01 -3.43099952e-01 -6.42226756e-01 1.68257594e-01 -2.10123867e-01 7.29884744e-01 7.20960438e-01 1.26199007e-01 5.37134290e-01 -1.70763768e-02 -6.25795648e-02 4.24771965e-01 4.62928295e-01 6.53024077e-01 -1.39019084e+00 2.08614618e-01 -3.26275259e-01 1.23717787e-03 -8.63761723e-01 -9.45702381e-03 -9.51834083e-01 -9.96498466e-02 -1.83389258e+00 3.86980742e-01 -3.22214752e-01 -4.99221563e-01 3.66030693e-01 -3.59583825e-01 8.81333575e-02 7.93003365e-02 4.16421592e-01 -1.17197001e+00 2.90123224e-01 8.86177242e-01 -4.04293597e-01 -2.88048144e-02 -2.21091956e-01 -1.14777052e+00 2.73835868e-01 8.33899677e-01 -7.27262616e-01 -5.08478045e-01 -5.87217748e-01 7.09595680e-01 2.96443045e-01 -1.02988958e-01 -9.38489914e-01 5.45006275e-01 -1.05284832e-01 4.76010412e-01 -8.87333751e-01 1.38881117e-01 -9.57920849e-01 -5.06196976e-01 2.63082415e-01 -1.02577245e+00 6.29921734e-01 2.52005786e-01 5.00016153e-01 -6.13940597e-01 1.69074535e-01 2.23259047e-01 -3.52696031e-01 -6.28116727e-01 2.59729862e-01 -1.84123199e-02 7.81181514e-01 4.64560419e-01 2.83395588e-01 -5.33149779e-01 -4.29981649e-01 -2.79854298e-01 4.98776376e-01 2.89956003e-01 6.81209266e-01 6.60610676e-01 -1.24061847e+00 -6.48153424e-01 1.23793267e-01 7.12983906e-01 -1.81845680e-01 -3.66553396e-01 2.14603543e-01 -5.20489037e-01 1.01752222e+00 -6.14143126e-02 -1.61828622e-01 -9.49502110e-01 9.39454734e-01 -1.21693257e-02 -7.07869947e-01 -2.83982575e-01 5.62865019e-01 -1.99480146e-01 -5.54427981e-01 6.34237647e-01 -4.12775129e-01 -4.08691704e-01 1.03842497e-01 6.56439602e-01 1.29058287e-01 5.44134617e-01 -3.12849939e-01 -4.62030500e-01 1.32090122e-01 -3.45757395e-01 3.39550562e-02 1.02851832e+00 -2.49639273e-01 -3.19705665e-01 4.98099267e-01 1.29532170e+00 4.22054715e-02 -4.35843945e-01 -8.69391263e-01 6.91622913e-01 -4.13947761e-01 -4.24067497e-01 -1.02743268e+00 -7.78021932e-01 5.58606803e-01 9.55381840e-02 3.10603201e-01 1.08067763e+00 1.33232236e-01 9.08593357e-01 1.30970418e+00 2.39749432e-01 -1.00816762e+00 9.78764966e-02 6.03008807e-01 7.73746669e-01 -1.59552777e+00 1.26161696e-02 -5.23064494e-01 -5.05108714e-01 1.20505869e+00 7.04016864e-01 1.45764515e-01 5.71203411e-01 6.41107112e-02 2.35174939e-01 -4.85655785e-01 -1.25045049e+00 -3.27704966e-01 5.33084750e-01 1.63070261e-01 7.31611848e-01 -3.73767912e-01 -1.77284673e-01 3.53990734e-01 -2.96201080e-01 -5.75634912e-02 3.00174385e-01 1.33005786e+00 -4.79341894e-01 -1.09881067e+00 -2.92105436e-01 7.58208752e-01 -6.57575250e-01 -3.26604664e-01 -7.92580307e-01 7.17947304e-01 -5.59177518e-01 1.09351671e+00 1.56719089e-01 -2.77246803e-01 5.11677861e-01 2.72517323e-01 1.36309918e-02 -5.67396343e-01 -8.33457589e-01 -3.01735014e-01 1.88960478e-01 -6.00459278e-01 -2.91907191e-01 -3.81984472e-01 -1.07254755e+00 -3.62856150e-01 2.05662653e-01 4.69883054e-01 4.59952980e-01 5.65257728e-01 5.10549068e-01 3.64428192e-01 4.96030986e-01 2.33222455e-01 -6.72834098e-01 -7.52687275e-01 -3.36533546e-01 5.48846364e-01 9.15898085e-02 -3.78285944e-01 -5.62504306e-02 2.50450727e-02]
[9.752492904663086, 7.9173903465271]
d39aeab5-746b-4354-b41c-83b7592d0b77
deepsolo-let-transformer-decoder-with-1
2305.19957
null
https://arxiv.org/abs/2305.19957v1
https://arxiv.org/pdf/2305.19957v1.pdf
DeepSolo++: Let Transformer Decoder with Explicit Points Solo for Text Spotting
End-to-end text spotting aims to integrate scene text detection and recognition into a unified framework. Dealing with the relationship between the two sub-tasks plays a pivotal role in designing effective spotters. Although Transformer-based methods eliminate the heuristic post-processing, they still suffer from the synergy issue between the sub-tasks and low training efficiency. In this paper, we present DeepSolo, a simple DETR-like baseline that lets a single decoder with explicit points solo for text detection and recognition simultaneously and efficiently. Technically, for each text instance, we represent the character sequence as ordered points and model them with learnable explicit point queries. After passing a single decoder, the point queries have encoded requisite text semantics and locations. Furthermore, we show the surprisingly good extensibility of our method, in terms of character class, language type, and task. On the one hand, DeepSolo not only performs well in English scenes but also masters the Chinese transcription with complex font structure and a thousand-level character classes. On the other hand, based on the extensibility of DeepSolo, we launch DeepSolo++ for multilingual text spotting, making a further step to let Transformer decoder with explicit points solo for multilingual text detection, recognition, and script identification all at once. Extensive experiments on public benchmarks demonstrate that our simple approach achieves better training efficiency compared with Transformer-based models and outperforms the previous state-of-the-art. In addition, DeepSolo and DeepSolo++ are also compatible with line annotations, which require much less annotation cost than polygons. The code is available at \url{https://github.com/ViTAE-Transformer/DeepSolo}.
['DaCheng Tao', 'Bo Du', 'Tongliang Liu', 'Juhua Liu', 'Shanshan Zhao', 'Jing Zhang', 'Maoyuan Ye']
2023-05-31
null
null
null
null
['text-spotting', 'scene-text-detection']
['computer-vision', 'computer-vision']
[ 1.35037050e-01 -4.82492030e-01 -1.51861861e-01 -2.74220258e-01 -1.18514121e+00 -8.97235990e-01 4.32428837e-01 -6.29524840e-03 -4.44178909e-01 2.62594163e-01 -9.56227165e-03 -5.64119875e-01 3.60170424e-01 -5.82702994e-01 -7.50997365e-01 -5.20874679e-01 6.38206303e-01 7.62800097e-01 4.67953712e-01 -1.81989789e-01 4.31604862e-01 1.25189140e-01 -1.10785091e+00 7.05966771e-01 9.60024893e-01 8.27513695e-01 5.80314159e-01 8.40425730e-01 -3.90031964e-01 7.27955759e-01 -5.43844700e-01 -7.81773865e-01 1.38770878e-01 -3.04458410e-01 -7.38007545e-01 2.04490989e-01 6.47601426e-01 -4.83386785e-01 -4.64298934e-01 9.83568966e-01 8.17673802e-01 -3.67862999e-01 4.57683563e-01 -8.58497202e-01 -4.56878394e-01 9.24663186e-01 -5.32791913e-01 -5.79740815e-02 3.35451156e-01 3.42664689e-01 1.31554413e+00 -1.33646452e+00 4.36362118e-01 8.67636919e-01 9.70379472e-01 3.86180341e-01 -1.01261783e+00 -3.50713581e-01 9.97662172e-02 1.41126871e-01 -1.38521397e+00 -5.90931594e-01 3.50431561e-01 -3.99903506e-01 1.10369205e+00 6.12187922e-01 2.27191791e-01 1.23722482e+00 -1.96268544e-01 1.35812104e+00 6.67212784e-01 -6.44596219e-01 -6.23649172e-02 1.18486002e-01 1.77846417e-01 9.28528547e-01 -7.08391666e-02 -5.79105794e-01 -6.33902431e-01 1.80083230e-01 4.71159548e-01 -9.51566249e-02 -2.71050930e-01 1.27044888e-02 -1.43863213e+00 4.90095586e-01 -1.41837239e-01 3.45308602e-01 2.06233889e-01 1.06899165e-01 7.01095104e-01 2.83677310e-01 2.02590957e-01 2.74151176e-01 -4.35523331e-01 -4.62526262e-01 -1.21977389e+00 -3.06807458e-02 7.71064878e-01 1.31460357e+00 6.66225910e-01 -3.49295847e-02 -4.58191007e-01 1.26248443e+00 1.11021334e-02 6.69777691e-01 5.79933882e-01 -2.37728402e-01 1.18064570e+00 6.16863787e-01 -2.24716827e-01 -6.40908659e-01 -1.45834818e-01 -4.16454732e-01 -6.68119073e-01 -1.92702040e-01 5.29040992e-01 -9.36666783e-03 -8.54427516e-01 1.21219599e+00 -3.64986882e-02 -1.62805885e-01 -3.78475130e-01 8.44383895e-01 5.05151212e-01 8.23922157e-01 -4.30306107e-01 3.03650469e-01 1.60442209e+00 -1.37781656e+00 -5.38344622e-01 -5.43284893e-01 1.04579115e+00 -1.25695205e+00 1.63538802e+00 4.67158705e-01 -8.24298382e-01 -3.46544892e-01 -1.01884007e+00 -4.82663423e-01 -3.24062288e-01 8.39524150e-01 3.53418320e-01 8.39865327e-01 -9.47114229e-01 3.39205474e-01 -8.07590783e-01 -5.19397855e-01 1.81999922e-01 2.25651637e-01 -7.46779665e-02 1.68755725e-02 -8.63052249e-01 6.21340871e-01 2.73795336e-01 1.32151842e-01 -4.93747205e-01 -4.57703382e-01 -5.57010353e-01 9.74408463e-02 6.07663095e-01 -3.52256119e-01 1.40253127e+00 -7.75424600e-01 -1.62278175e+00 9.17577982e-01 -3.27149719e-01 -1.63045347e-01 9.12752569e-01 -3.27530473e-01 -2.21116170e-01 8.63147825e-02 2.80680686e-01 4.43932772e-01 8.03086638e-01 -7.95709372e-01 -8.40269208e-01 -2.18199804e-01 -3.29728067e-01 3.96602243e-01 -6.92998469e-01 2.69366831e-01 -1.36659360e+00 -8.73314321e-01 -1.66455124e-04 -8.00267935e-01 2.43946239e-01 4.21221815e-02 -9.09941316e-01 -2.75082320e-01 8.04407597e-01 -7.97639906e-01 1.50461388e+00 -2.35180974e+00 4.08892743e-02 -1.71690725e-03 1.28516570e-01 2.07835361e-01 -1.42119259e-01 7.11235225e-01 1.85229853e-01 2.25694850e-01 -1.58966810e-01 -7.59494603e-01 4.65906709e-01 -4.70121838e-02 -5.11891305e-01 4.09428298e-01 -1.12028327e-02 9.08162057e-01 -6.07495129e-01 -6.18801653e-01 2.67183512e-01 1.24451987e-01 -5.12207389e-01 3.81915793e-02 -4.12069172e-01 -8.35612863e-02 -3.34373385e-01 8.57766867e-01 5.05431771e-01 -2.95051605e-01 3.74130338e-01 -2.08529029e-02 -2.87318259e-01 6.05768085e-01 -1.27253270e+00 1.68499875e+00 -2.87726164e-01 1.01691103e+00 -4.98601384e-02 -7.94042528e-01 6.93899512e-01 2.24043980e-01 1.49238989e-01 -8.23548853e-01 9.23500583e-03 6.15778863e-01 -4.92068768e-01 -4.79988128e-01 8.89105499e-01 3.99780005e-01 -2.11353213e-01 4.70289201e-01 -3.39133404e-02 -5.00106178e-02 3.80259037e-01 2.38151222e-01 1.03164959e+00 3.36307853e-01 -1.01750391e-02 -2.32124468e-03 4.45538104e-01 6.79100007e-02 3.45606834e-01 1.03920233e+00 1.35143891e-01 8.99973691e-01 7.04141855e-01 -2.06068426e-01 -1.37479842e+00 -7.27019727e-01 -1.07092768e-01 1.51538432e+00 -5.87389022e-02 -7.14897275e-01 -9.95381117e-01 -7.36613929e-01 -1.56170428e-01 5.79854786e-01 -3.94101650e-01 3.55646282e-01 -8.72658372e-01 -7.85066664e-01 1.21203017e+00 5.94797850e-01 5.01613677e-01 -6.96483016e-01 -2.06783742e-01 1.01834022e-01 -5.76524138e-01 -1.32236433e+00 -1.09324956e+00 3.79252851e-01 -5.52835643e-01 -7.74224997e-01 -9.00232792e-01 -9.93844032e-01 4.73686367e-01 3.45795572e-01 9.84684110e-01 2.20945939e-01 -3.11312437e-01 9.97878611e-02 -5.41533053e-01 1.25059132e-02 -3.64662737e-01 5.37429273e-01 -4.37701404e-01 -4.64857119e-04 2.89499938e-01 -8.18498954e-02 -4.44880038e-01 5.95179200e-01 -8.64204526e-01 3.85018975e-01 5.38614094e-01 9.43491936e-01 3.67982149e-01 -1.52889580e-01 1.05553744e-02 -8.67290854e-01 3.26053262e-01 -5.64612038e-02 -6.90920353e-01 5.77683508e-01 -4.27076817e-01 -3.12737413e-02 8.26943338e-01 -2.10044041e-01 -7.71472514e-01 2.27686942e-01 -3.53133202e-01 -1.29458830e-01 -4.93128262e-02 4.65477914e-01 -3.52186859e-01 2.41771176e-01 6.91955030e-01 7.63906300e-01 -3.60777855e-01 -7.56896019e-01 1.77714005e-01 1.05419266e+00 5.14321506e-01 -7.06166089e-01 8.09930503e-01 2.52133727e-01 -6.62800431e-01 -8.78202856e-01 -6.71209276e-01 -5.66475451e-01 -7.38778651e-01 -1.83974095e-02 6.96732581e-01 -8.87680709e-01 -6.20997429e-01 9.32885230e-01 -1.09331644e+00 -7.56364822e-01 4.77244221e-02 9.09142271e-02 -5.31560779e-01 9.44860578e-01 -7.86617458e-01 -5.20564198e-01 -4.13471848e-01 -1.34373641e+00 1.52819073e+00 -2.18732700e-01 5.75147718e-02 -8.14620912e-01 -3.05060983e-01 5.57902098e-01 1.39137954e-01 -4.79231626e-01 8.78967226e-01 -6.55541241e-01 -7.67527640e-01 -3.04221749e-01 -4.41386908e-01 9.90011171e-02 -1.79580361e-01 9.51010361e-03 -9.44153786e-01 -2.83635706e-01 -3.30379516e-01 -2.47903243e-01 9.32978928e-01 -1.46796510e-01 1.23795903e+00 -1.73685238e-01 -1.54406071e-01 9.11297321e-01 1.45128608e+00 -7.83488750e-02 7.53064632e-01 6.49086773e-01 9.78335083e-01 3.20470840e-01 4.67116445e-01 6.24327123e-01 4.85257775e-01 1.00551248e+00 2.34356821e-01 -1.14441454e-01 -2.71235973e-01 -4.23374236e-01 6.54565096e-01 9.92211401e-01 3.61812323e-01 -6.45491660e-01 -1.03925490e+00 2.87445962e-01 -1.94640696e+00 -9.30948675e-01 -5.17547071e-01 2.25166154e+00 1.04009390e+00 7.49711245e-02 2.39151433e-01 2.54063040e-01 7.54142821e-01 1.82249129e-01 -3.51478994e-01 -1.82867602e-01 -4.86375153e-01 -3.01237521e-03 7.85401106e-01 5.18061399e-01 -1.15376914e+00 1.41490746e+00 5.46095181e+00 1.37347579e+00 -1.33038998e+00 8.80962014e-02 5.25632977e-01 -3.24230269e-02 -1.44680709e-01 -1.54983820e-02 -1.29431534e+00 7.06991792e-01 6.47925973e-01 2.18230724e-01 6.00623548e-01 7.09638119e-01 9.61402133e-02 2.93217991e-02 -1.17285895e+00 1.10924196e+00 2.22220749e-01 -1.34988046e+00 -8.15310923e-04 -1.09469034e-01 3.23639005e-01 3.02311927e-01 8.26575309e-02 2.88375109e-01 2.89684534e-02 -8.60067725e-01 1.36925805e+00 9.72449183e-02 1.14706111e+00 -2.47796401e-01 4.54813451e-01 4.20809388e-01 -1.24345458e+00 1.45597328e-02 -3.43056917e-01 3.11474800e-01 -2.25019827e-02 5.31129122e-01 -9.44688559e-01 5.27386189e-01 5.35509109e-01 7.41534412e-01 -9.05575871e-01 1.03295386e+00 -2.64613330e-01 8.60760689e-01 -4.12364393e-01 -2.74236083e-01 3.74064088e-01 -7.33801946e-02 3.84630144e-01 1.89234126e+00 4.65908766e-01 -4.70760167e-01 2.48674661e-01 6.60984159e-01 -9.24358815e-02 4.08639222e-01 -2.83475280e-01 -1.28885195e-01 4.41573441e-01 1.06079221e+00 -9.47883010e-01 -5.15423477e-01 -5.29844105e-01 1.53486705e+00 3.85656148e-01 2.25826204e-01 -1.01816797e+00 -6.58418834e-01 1.79230139e-01 -1.13769369e-02 6.22463346e-01 -3.37866783e-01 -6.78580046e-01 -1.65317440e+00 4.07264739e-01 -1.33419907e+00 3.13643068e-01 -6.96482778e-01 -9.53331828e-01 5.20112395e-01 -5.46951771e-01 -1.31257439e+00 1.40313944e-02 -9.52650964e-01 -3.88342112e-01 9.07548428e-01 -1.29358351e+00 -1.29596829e+00 -1.21580690e-01 6.55235112e-01 1.11213255e+00 -1.11751288e-01 6.65962756e-01 6.28179073e-01 -1.02381372e+00 1.09047699e+00 5.17786562e-01 7.17424512e-01 1.04786217e+00 -1.50465822e+00 8.71443450e-01 1.12136686e+00 2.72863865e-01 4.18466181e-01 4.10097033e-01 -5.47981620e-01 -1.71348894e+00 -9.65035439e-01 1.14843833e+00 -6.20514691e-01 8.12728167e-01 -9.66487765e-01 -7.90250361e-01 6.83401287e-01 2.87812859e-01 -4.03355002e-01 4.90074515e-01 1.76097184e-01 -5.48850894e-01 -1.83963217e-02 -4.80038404e-01 8.80762577e-01 8.96423519e-01 -7.43669689e-01 -2.61206329e-01 6.77974403e-01 3.68646055e-01 -6.77331567e-01 -4.29937005e-01 -1.70616254e-01 4.89060938e-01 -1.01352870e+00 6.79261625e-01 -7.96382651e-02 4.58562762e-01 -3.05318892e-01 -2.42907181e-01 -8.29772592e-01 -1.94906235e-01 -8.00474524e-01 1.27667218e-01 1.32760358e+00 7.05991268e-01 -5.52690089e-01 8.13639283e-01 1.20890014e-01 -4.75838035e-01 -5.56434095e-01 -7.90792286e-01 -7.96588540e-01 8.42796937e-02 -7.34573662e-01 4.88892376e-01 9.26287174e-01 3.77510339e-02 4.45262909e-01 -7.24374712e-01 8.74624774e-02 3.25544029e-01 7.19292834e-02 8.98745358e-01 -7.94060707e-01 -5.84839642e-01 -7.51141250e-01 -4.61712256e-02 -1.73472869e+00 -2.02472702e-01 -9.86224711e-01 2.43359715e-01 -1.39203107e+00 2.73760170e-01 -4.52113092e-01 2.31915861e-01 8.31871450e-01 -2.86876529e-01 3.45969737e-01 3.99843186e-01 4.40439880e-01 -8.19428444e-01 2.33902618e-01 8.50167096e-01 -1.59742221e-01 -1.07201897e-02 -9.65866968e-02 -5.33444285e-01 4.81652647e-01 7.60034502e-01 -3.41563791e-01 -1.51020726e-02 -1.18658483e+00 4.92607921e-01 4.15697768e-02 3.17492276e-01 -8.68591726e-01 5.69715917e-01 1.03403859e-01 1.56961635e-01 -8.09973001e-01 2.04333067e-01 -5.69629431e-01 -3.45603019e-01 2.05504656e-01 -2.77169198e-01 1.23904042e-01 1.65637787e-02 1.59451082e-01 -7.39673227e-02 -6.04496598e-01 5.97987115e-01 5.07886633e-02 -7.16054082e-01 1.86239406e-01 -6.23013198e-01 8.05577487e-02 5.09421587e-01 -4.83091831e-01 -5.60922623e-01 -1.57312959e-01 -3.17157656e-01 1.91739842e-01 6.80251181e-01 3.80498886e-01 3.34674567e-01 -8.35893929e-01 -6.54658198e-01 2.26157695e-01 2.05539107e-01 -1.50879711e-01 8.58654827e-02 1.06144035e+00 -9.00243461e-01 6.17591560e-01 2.30508208e-01 -7.39392579e-01 -1.26370382e+00 3.34080964e-01 4.11371708e-01 -4.24313098e-01 -7.66799986e-01 7.08148837e-01 9.40934494e-02 -6.77718818e-01 5.35646737e-01 -3.03963631e-01 3.03261608e-01 -6.06757635e-03 4.91897851e-01 2.60935187e-01 3.84530932e-01 -3.79018962e-01 -2.57218838e-01 6.81011260e-01 -2.56215572e-01 -2.30799034e-01 1.05748641e+00 -2.00097576e-01 -1.28568619e-01 3.25862825e-01 1.00282943e+00 4.63032395e-01 -1.12612391e+00 -3.29970270e-01 2.94535756e-01 -4.05964911e-01 -1.55976251e-01 -1.04873729e+00 -8.09176922e-01 1.10852313e+00 2.47471362e-01 5.74562401e-02 1.01612365e+00 -2.99423605e-01 9.35594082e-01 5.75771272e-01 1.31196290e-01 -1.30160046e+00 1.33509800e-01 9.88731086e-01 5.11573672e-01 -1.10620046e+00 -2.52028316e-01 -5.54588795e-01 -6.48140252e-01 1.34011006e+00 5.78950763e-01 3.86895359e-01 -2.32723396e-04 5.52804708e-01 1.57308996e-01 1.90985411e-01 -7.46608317e-01 -2.29428083e-01 1.92992479e-01 2.26669654e-01 6.34096563e-01 -6.58937171e-02 -1.44064380e-02 3.99603665e-01 -1.91229418e-01 -2.20517144e-01 3.80329221e-01 8.18309665e-01 -5.40979564e-01 -1.32310927e+00 -5.38789570e-01 3.22189510e-01 -5.38277328e-01 -6.42609358e-01 -5.00932276e-01 6.16567492e-01 -1.89918384e-01 6.86279893e-01 -1.37344018e-01 -4.01682198e-01 2.82835245e-01 2.11792931e-01 1.99871525e-01 -6.45000041e-01 -8.15204442e-01 4.61353391e-01 1.22757927e-01 -2.36419231e-01 4.67296302e-01 -6.62207365e-01 -1.05608130e+00 -5.71945727e-01 -4.13361102e-01 -2.54396945e-02 6.27395570e-01 8.49856198e-01 3.98165196e-01 3.05924118e-01 6.25583768e-01 -5.57156742e-01 -8.39262128e-01 -9.38901722e-01 -3.39453429e-01 1.49774095e-02 7.30853528e-02 3.51477377e-02 -9.27951410e-02 1.86343431e-01]
[11.978839874267578, 2.255518674850464]
239ee1de-183d-435f-ae98-4ef209cb3b30
dfr-tsd-a-deep-learning-based-framework-for
2006.02578
null
https://arxiv.org/abs/2006.02578v1
https://arxiv.org/pdf/2006.02578v1.pdf
DFR-TSD: A Deep Learning Based Framework for Robust Traffic Sign Detection Under Challenging Weather Conditions
Robust traffic sign detection and recognition (TSDR) is of paramount importance for the successful realization of autonomous vehicle technology. The importance of this task has led to a vast amount of research efforts and many promising methods have been proposed in the existing literature. However, the SOTA (SOTA) methods have been evaluated on clean and challenge-free datasets and overlooked the performance deterioration associated with different challenging conditions (CCs) that obscure the traffic images captured in the wild. In this paper, we look at the TSDR problem under CCs and focus on the performance degradation associated with them. To overcome this, we propose a Convolutional Neural Network (CNN) based TSDR framework with prior enhancement. Our modular approach consists of a CNN-based challenge classifier, Enhance-Net, an encoder-decoder CNN architecture for image enhancement, and two separate CNN architectures for sign-detection and classification. We propose a novel training pipeline for Enhance-Net that focuses on the enhancement of the traffic sign regions (instead of the whole image) in the challenging images subject to their accurate detection. We used CURE-TSD dataset consisting of traffic videos captured under different CCs to evaluate the efficacy of our approach. We experimentally show that our method obtains an overall precision and recall of 91.1% and 70.71% that is 7.58% and 35.90% improvement in precision and recall, respectively, compared to the current benchmark. Furthermore, we compare our approach with SOTA object detection networks, Faster-RCNN and R-FCN, and show that our approach outperforms them by a large margin.
['Md. Kamrul Hasan', 'Uday Kamal', 'Sabbir Ahmed']
2020-06-03
null
null
null
null
['traffic-sign-detection']
['computer-vision']
[ 2.93284774e-01 -2.60703683e-01 1.37227699e-01 -3.17782551e-01 -7.52411246e-01 -1.94927827e-01 7.36489773e-01 -8.35322738e-01 -6.72728896e-01 3.79162818e-01 -2.00472355e-01 -3.82950634e-01 2.91578293e-01 -4.58448350e-01 -7.24368691e-01 -7.62306154e-01 6.07146695e-02 -1.41160995e-01 7.90608048e-01 -2.47868419e-01 1.77959412e-01 4.00939018e-01 -2.04936075e+00 2.73073435e-01 9.49861526e-01 1.35059857e+00 7.10003003e-02 7.59180248e-01 3.84345762e-02 9.18069541e-01 -7.83419132e-01 -5.49642682e-01 5.09465992e-01 -1.32357225e-01 -1.33244574e-01 -6.00412488e-02 1.05184817e+00 -6.14697993e-01 -6.10974669e-01 9.88505900e-01 6.68253541e-01 -1.53909460e-01 3.83868098e-01 -1.63333297e+00 -3.57279062e-01 3.77355479e-02 -7.87108541e-01 2.28803486e-01 -2.69483119e-01 6.31104887e-01 5.52876115e-01 -9.41590726e-01 6.02558374e-01 1.09020030e+00 7.43050337e-01 6.05103672e-01 -6.66646063e-01 -1.02843451e+00 -2.10542325e-02 6.30662560e-01 -1.14040232e+00 -5.92100203e-01 6.02214158e-01 -3.14318359e-01 8.54172587e-01 -6.75952584e-02 3.46346736e-01 1.11023009e+00 6.48388639e-02 1.11305535e+00 1.19293773e+00 -1.99302122e-01 1.06697604e-02 4.61308509e-02 1.36890203e-01 6.38119400e-01 5.48863471e-01 4.64592785e-01 -1.79524124e-01 5.12607098e-01 2.61067748e-01 -1.54009163e-01 -6.84846118e-02 -2.69286305e-01 -9.06470895e-01 2.59842306e-01 4.91090029e-01 1.30659938e-01 -2.54979491e-01 4.73798841e-01 6.12916291e-01 1.79558441e-01 -4.02491726e-02 -1.61194399e-01 -3.84350747e-01 -2.64662743e-01 -1.00516033e+00 2.28382856e-01 4.24522400e-01 9.57054019e-01 3.10052216e-01 4.09193128e-01 -3.39665323e-01 7.67054260e-01 2.48619825e-01 1.03299165e+00 1.31851047e-01 -6.84941411e-01 7.34167218e-01 4.99881476e-01 -1.87646877e-02 -9.62360144e-01 -2.15076476e-01 -7.68956900e-01 -7.05048680e-01 7.14772463e-01 4.68379796e-01 -1.52777910e-01 -1.29540825e+00 1.59971499e+00 1.57002971e-01 4.79867429e-01 2.27744594e-01 1.13241804e+00 8.27142298e-01 4.31148320e-01 2.99941927e-01 2.79986054e-01 1.40817487e+00 -1.11488867e+00 -6.46277368e-01 -2.34903514e-01 5.67556977e-01 -8.06238353e-01 7.31847823e-01 3.32161993e-01 -7.55936801e-01 -7.99892902e-01 -1.25513411e+00 -2.97701377e-02 -5.82953215e-01 6.33077860e-01 3.44656467e-01 9.37722445e-01 -1.04580998e+00 2.15575337e-01 -6.38457716e-01 -4.56591636e-01 7.19500065e-01 2.56436199e-01 -4.07658517e-01 -3.10398281e-01 -8.95908177e-01 1.11348426e+00 1.05097249e-01 5.15338659e-01 -1.03829896e+00 -4.99669403e-01 -5.25341690e-01 -1.09107852e-01 5.15021622e-01 -3.23284894e-01 1.06236196e+00 -8.12833011e-01 -1.31842101e+00 8.12527835e-01 -2.41100907e-01 -6.78070903e-01 8.29140902e-01 -2.29007930e-01 -7.81155109e-01 6.08963147e-02 -1.33971991e-02 7.10021496e-01 9.24651504e-01 -1.22212052e+00 -1.09309399e+00 -7.61462003e-02 -3.68775949e-02 -2.12548926e-01 2.43302565e-02 3.14762473e-01 -7.14295447e-01 -3.57282698e-01 -1.95586264e-01 -8.93629611e-01 1.68159693e-01 2.90015310e-01 -1.07396625e-01 -6.78985044e-02 1.48963737e+00 -8.05658281e-01 1.00411367e+00 -2.17433047e+00 -6.19741201e-01 -9.77619737e-02 1.86712161e-01 1.07020140e+00 -4.01711434e-01 -1.74077204e-03 1.31848022e-01 -1.23248741e-01 -3.60743374e-01 -3.64461124e-01 1.78032935e-01 1.94481835e-01 -1.52773008e-01 3.90060425e-01 5.95666349e-01 1.11159432e+00 -6.02847099e-01 -4.50202763e-01 4.94355798e-01 6.81111753e-01 -3.66558820e-01 3.35288718e-02 7.11586922e-02 1.28646806e-01 -2.58117706e-01 1.00430274e+00 1.29339826e+00 1.24877751e-01 -2.18952060e-01 -4.86524373e-01 -3.23717564e-01 -1.18932100e-02 -1.06094205e+00 9.56337214e-01 -2.65642583e-01 1.12667394e+00 1.38067931e-01 -9.20724273e-01 8.87422740e-01 9.56264734e-02 1.88074604e-01 -1.36098266e+00 4.35734212e-01 5.96525192e-01 2.18222082e-01 -7.83624709e-01 5.25806189e-01 1.44735157e-01 1.95584089e-01 1.06018856e-01 -7.67723192e-03 3.58640671e-01 3.69658798e-01 1.27003670e-01 1.14414620e+00 1.25904307e-01 -2.95265377e-01 2.19409674e-01 7.53250003e-01 -5.41199744e-02 5.03549516e-01 6.81715488e-01 -7.66879201e-01 5.18776476e-01 3.31648380e-01 -2.67941177e-01 -1.07016468e+00 -7.91007936e-01 -1.88164443e-01 7.33482122e-01 3.94824177e-01 8.45127553e-02 -5.24428546e-01 -8.95219743e-01 9.93074998e-02 5.08203268e-01 -6.54918909e-01 -1.58003747e-01 -9.55556035e-01 -6.29337013e-01 1.01527607e+00 7.57300496e-01 1.16923153e+00 -7.81121552e-01 -7.08039939e-01 2.74185333e-02 8.61514453e-03 -1.71917343e+00 -3.98544729e-01 -9.05884430e-02 -5.15988111e-01 -1.32921529e+00 -7.55195737e-01 -7.42501557e-01 4.20514047e-01 6.80036545e-01 7.41265833e-01 1.56251371e-01 -3.69826168e-01 8.03006813e-02 -3.25292468e-01 -4.48038936e-01 -3.46242219e-01 -2.13912055e-01 -2.89271951e-01 3.99408430e-01 5.29151976e-01 -1.97773889e-01 -8.51248503e-01 7.36997187e-01 -8.58691931e-01 -8.77425447e-02 1.05319238e+00 7.90060699e-01 -2.51038722e-03 -3.72935086e-01 5.25526822e-01 -3.05711895e-01 2.35542655e-01 -1.95166960e-01 -8.98751199e-01 1.68430761e-01 -5.46010733e-01 7.15543423e-03 1.51735708e-01 -2.46238977e-01 -1.18344080e+00 8.46769884e-02 -3.67150933e-01 -4.86372352e-01 -3.22221875e-01 -3.68232913e-02 -1.91815004e-01 -4.08655703e-01 3.58267069e-01 3.81248593e-01 1.21119425e-01 -3.84885550e-01 2.48139724e-01 9.67094958e-01 7.13855028e-01 -6.66067973e-02 9.31907892e-01 6.46134675e-01 4.13049944e-02 -7.42379785e-01 -5.97215950e-01 -4.88075495e-01 -2.96278507e-01 -6.69249237e-01 7.84950614e-01 -1.02352369e+00 -7.62374759e-01 8.43077958e-01 -1.01564312e+00 -1.04387730e-01 -9.75878350e-03 5.18201411e-01 -1.49315774e-01 4.60706323e-01 -2.96073973e-01 -8.69720638e-01 -4.60635960e-01 -1.35560322e+00 1.17435288e+00 3.98564935e-01 4.75136995e-01 -4.67832178e-01 -2.42616564e-01 6.24901712e-01 1.07041073e+00 2.81997114e-01 3.03842545e-01 -4.09145653e-01 -9.89397645e-01 -4.66512471e-01 -9.30639088e-01 6.88930273e-01 -2.05064088e-01 5.26289977e-02 -1.38503110e+00 -8.87883753e-02 -5.54339290e-01 -2.12172985e-01 1.27330983e+00 3.32596064e-01 6.37302756e-01 1.60095677e-01 -3.29449981e-01 5.53608239e-01 1.42788625e+00 2.42231399e-01 9.69541371e-01 3.70791584e-01 5.83698213e-01 2.88015366e-01 7.01701760e-01 -2.88831927e-02 4.42774743e-01 8.74218941e-01 6.87858641e-01 -2.48146966e-01 -8.70481014e-01 -1.58135012e-01 7.23654509e-01 3.58801484e-01 -9.57273319e-02 -2.79638082e-01 -6.05461895e-01 6.89056754e-01 -1.71939623e+00 -1.03914785e+00 -5.35563409e-01 1.91918254e+00 2.60485381e-01 2.00896084e-01 1.62786677e-01 3.46211761e-01 7.36981213e-01 4.75760326e-02 -4.44920748e-01 -3.40064198e-01 -4.70533878e-01 1.30914852e-01 8.25288117e-01 1.12123966e-01 -1.20080042e+00 1.01090193e+00 5.74277639e+00 8.57760370e-01 -1.48784363e+00 1.68512926e-01 2.97985643e-01 1.55603871e-01 4.37657714e-01 -3.06938469e-01 -9.30026650e-01 5.42565525e-01 8.21003079e-01 3.65317732e-01 -1.24759383e-01 9.68115091e-01 3.68078142e-01 -7.49466196e-02 -5.40617287e-01 8.48955095e-01 3.06433588e-01 -1.03292620e+00 -2.19266117e-01 5.77957928e-02 7.88554013e-01 3.79909247e-01 1.28324971e-01 5.78160405e-01 4.61297296e-02 -8.01580667e-01 8.96412790e-01 3.90946865e-01 9.02161002e-01 -4.27388698e-01 1.21777093e+00 4.85275872e-02 -1.37636936e+00 -2.01137617e-01 -1.66977227e-01 1.92815140e-01 3.45687866e-01 4.25147206e-01 -7.41779804e-01 4.07407075e-01 5.92311502e-01 7.50114381e-01 -7.80372381e-01 1.54362261e+00 -2.95093030e-01 7.17076838e-01 -3.74524385e-01 -5.70980273e-02 3.44388783e-01 1.90870821e-01 6.24587357e-01 1.46486664e+00 3.08277071e-01 -3.51350933e-01 -2.48226911e-01 8.26436341e-01 -1.75727997e-02 -2.40767911e-01 -3.93252105e-01 2.77152240e-01 1.54243186e-01 1.22639394e+00 -4.41401690e-01 -5.16925275e-01 -4.68164116e-01 7.68307447e-01 -1.37878463e-01 3.32280606e-01 -1.24487066e+00 -4.57589000e-01 7.87795246e-01 2.32701972e-02 8.26861322e-01 -1.03311002e-01 -1.05742231e-01 -1.06490803e+00 4.48472500e-01 -7.25557506e-01 1.59936368e-01 -8.01574349e-01 -9.46399152e-01 6.13402724e-01 -3.02974880e-01 -1.61048424e+00 2.06911191e-01 -1.07097399e+00 -6.55602336e-01 5.45931160e-01 -2.23722506e+00 -1.36861515e+00 -7.71105766e-01 4.04091656e-01 4.49791670e-01 -1.16017133e-01 2.15328466e-02 9.77358639e-01 -7.81285107e-01 9.74496424e-01 2.90262625e-02 3.26107591e-01 7.50976205e-01 -8.35729718e-01 4.86904204e-01 1.25769210e+00 -4.28422451e-01 5.70272282e-02 4.72420275e-01 -4.39422697e-01 -1.53480816e+00 -1.39069581e+00 9.40719724e-01 -3.31854224e-01 4.66676593e-01 -1.72815412e-01 -7.07791507e-01 2.56741881e-01 1.31998658e-01 3.30199540e-01 -1.70066819e-01 -5.84526658e-01 -5.83018243e-01 -4.43555444e-01 -1.11875129e+00 4.30051893e-01 1.16893244e+00 -2.38272667e-01 -3.14146787e-01 -3.12280536e-01 2.91869670e-01 -2.88939297e-01 -2.56413400e-01 6.63868368e-01 9.47282732e-01 -9.29529369e-01 9.67656910e-01 -2.86381811e-01 2.93983668e-01 -9.62191880e-01 -1.75955832e-01 -8.13802660e-01 1.92722410e-01 -1.45290583e-01 -1.13776557e-01 1.12481070e+00 3.64316732e-01 -6.67104602e-01 6.79828823e-01 3.89988124e-01 -4.48040366e-01 -3.98487389e-01 -9.95096385e-01 -9.11657572e-01 -2.41002887e-01 -8.07654321e-01 4.35307443e-01 4.12705809e-01 -5.61883152e-01 9.74196941e-02 -7.05426276e-01 2.75741667e-01 8.60887229e-01 -8.35419297e-02 1.14505982e+00 -9.34718490e-01 1.13946989e-01 -6.23999536e-01 -8.01360786e-01 -1.16531348e+00 -3.62953335e-01 -5.31765878e-01 3.04380029e-01 -1.39802635e+00 4.09347378e-02 -2.97083825e-01 -2.74369568e-01 4.56632257e-01 -2.05683768e-01 6.27033710e-01 3.65387142e-01 -3.27569544e-02 -9.56424534e-01 6.49202645e-01 1.30866098e+00 -2.44089693e-01 1.63449854e-01 -1.34401724e-01 -4.56821084e-01 3.97796810e-01 7.00697958e-01 -4.10397977e-01 9.63676944e-02 -4.34517622e-01 -2.35394120e-01 -4.90841478e-01 8.74314964e-01 -1.34907162e+00 4.36819077e-01 3.12299192e-01 2.10975379e-01 -9.56566095e-01 2.55042225e-01 -9.05063868e-01 -1.21748619e-01 6.72865748e-01 -4.84541468e-02 -2.03939930e-01 4.18135881e-01 5.40164471e-01 -4.12170559e-01 2.05408439e-01 9.87752497e-01 4.69227999e-01 -1.12283266e+00 1.10159583e-01 -2.92976141e-01 -9.45562720e-02 9.98458505e-01 -5.64834118e-01 -6.67404830e-01 -2.10287243e-01 -8.14565495e-02 3.86957169e-01 6.19629920e-02 7.29156375e-01 6.91513240e-01 -1.22761869e+00 -8.78772855e-01 2.80113757e-01 4.17047560e-01 -4.82960820e-01 3.88803929e-01 1.31323862e+00 -5.63793838e-01 5.84623814e-01 -3.91308427e-01 -7.78251171e-01 -1.42285824e+00 2.01696530e-01 5.18867135e-01 -4.96216342e-02 -5.54634035e-01 5.55951178e-01 -7.04395771e-02 -2.31863245e-01 5.25577545e-01 -5.63636184e-01 -2.49127492e-01 -1.90987781e-01 6.68506205e-01 5.74062824e-01 2.85450965e-01 -9.37368929e-01 -4.43957686e-01 8.44365835e-01 2.07156092e-02 2.88622290e-01 1.19405496e+00 -5.53192645e-02 4.56313699e-01 -3.82744133e-01 1.32960999e+00 -1.83539689e-01 -1.39088273e+00 -2.38709450e-01 -5.20399213e-02 -5.19461691e-01 1.61747903e-01 -1.06223428e+00 -1.47941220e+00 1.05084348e+00 1.16561639e+00 -1.89154863e-01 1.18059945e+00 -2.48939231e-01 9.52404857e-01 2.37425417e-01 1.24047875e-01 -1.02964652e+00 -9.85163078e-02 5.21638870e-01 7.12609529e-01 -1.62663376e+00 -4.14511055e-01 -3.39974523e-01 -6.38007462e-01 9.97322023e-01 6.51685119e-01 -1.33332089e-01 3.76539499e-01 3.29685539e-01 3.53649259e-01 -1.66079074e-01 -6.71630859e-01 -8.62373710e-01 4.11311209e-01 6.12052798e-01 6.42912984e-02 -1.74718231e-01 -2.72792310e-01 3.36762965e-01 2.19303772e-01 3.85218263e-01 1.99550599e-01 9.03118312e-01 -5.09528935e-01 -7.97359228e-01 -3.30766231e-01 2.05629796e-01 -2.03338534e-01 7.57248029e-02 -3.31674695e-01 9.85310674e-01 5.22709191e-01 1.08793342e+00 -1.50993302e-01 -6.50750339e-01 6.64337039e-01 -1.19672948e-02 1.29822329e-01 1.91979244e-01 -4.79832530e-01 -1.78512052e-01 3.79017413e-01 -6.65753663e-01 -5.43414712e-01 -6.14992082e-01 -9.70869899e-01 -3.14596832e-01 -5.50626755e-01 -3.75348657e-01 8.88796449e-01 9.01842296e-01 5.62330067e-01 6.26959562e-01 5.39448380e-01 -7.78389692e-01 -3.41573507e-01 -9.01437044e-01 -2.34952956e-01 4.38490540e-01 5.42582691e-01 -8.49158525e-01 -4.02958661e-01 -3.19347568e-02]
[7.967831611633301, -0.853804886341095]
80f0fbf1-968e-4c48-bb4a-f5f1cbcad596
proceedings-of-the-workshop-on-computational-3
null
null
https://aclanthology.org/W12-0400
https://aclanthology.org/W12-0400.pdf
Proceedings of the Workshop on Computational Approaches to Deception Detection
null
['']
2012-04-01
null
null
null
ws-2012-4
['deception-detection']
['miscellaneous']
[-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]
[-7.315954685211182, 3.736192464828491]
f5c7d0bb-38ec-4301-b213-802d31ec4fbf
robotic-navigation-autonomy-for-subretinal
2301.07204
null
https://arxiv.org/abs/2301.07204v1
https://arxiv.org/pdf/2301.07204v1.pdf
Robotic Navigation Autonomy for Subretinal Injection via Intelligent Real-Time Virtual iOCT Volume Slicing
In the last decade, various robotic platforms have been introduced that could support delicate retinal surgeries. Concurrently, to provide semantic understanding of the surgical area, recent advances have enabled microscope-integrated intraoperative Optical Coherent Tomography (iOCT) with high-resolution 3D imaging at near video rate. The combination of robotics and semantic understanding enables task autonomy in robotic retinal surgery, such as for subretinal injection. This procedure requires precise needle insertion for best treatment outcomes. However, merging robotic systems with iOCT introduces new challenges. These include, but are not limited to high demands on data processing rates and dynamic registration of these systems during the procedure. In this work, we propose a framework for autonomous robotic navigation for subretinal injection, based on intelligent real-time processing of iOCT volumes. Our method consists of an instrument pose estimation method, an online registration between the robotic and the iOCT system, and trajectory planning tailored for navigation to an injection target. We also introduce intelligent virtual B-scans, a volume slicing approach for rapid instrument pose estimation, which is enabled by Convolutional Neural Networks (CNNs). Our experiments on ex-vivo porcine eyes demonstrate the precision and repeatability of the method. Finally, we discuss identified challenges in this work and suggest potential solutions to further the development of such systems.
['Iulian Iordachita', 'M. Ali Nasseri', 'Nassir Navab', 'Peter Gehlbach', 'Benjamin Busam', 'Alejandro Martin-Gomez', 'Peiyao Zhang', 'Michael Sommersperger', 'Shervin Dehghani']
2023-01-17
null
null
null
null
['trajectory-planning']
['robots']
[-6.09281845e-02 1.28281251e-01 1.29544586e-01 -3.37074548e-02 -1.41932607e-01 -7.96599388e-01 -2.30783728e-04 -3.67693067e-01 -5.13838470e-01 2.10489810e-01 -7.69878700e-02 -3.49173576e-01 -3.05948883e-01 -3.14823568e-01 -5.60148180e-01 -4.75642413e-01 -1.22161293e-02 5.39261699e-01 2.55149752e-01 3.03846058e-02 5.50109982e-01 1.02680767e+00 -1.61600137e+00 -1.06443539e-01 1.04889190e+00 1.21130681e+00 6.93388402e-01 5.34383893e-01 6.46640500e-03 5.02313256e-01 -1.85256839e-01 -1.25496089e-01 6.75722480e-01 -5.83545305e-03 -5.96173048e-01 -2.07215119e-02 6.60498559e-01 -7.16342330e-01 -6.79113790e-02 8.85473609e-01 6.42056644e-01 -3.54804218e-01 4.43007380e-01 -5.43812394e-01 -1.65834188e-01 8.29493180e-02 -2.68286437e-01 -1.41013980e-01 9.07028168e-02 4.70626116e-01 1.00948736e-01 -8.32553148e-01 1.05937302e+00 6.18030667e-01 6.04128420e-01 5.61649144e-01 -6.62606418e-01 -4.78641272e-01 -4.71688747e-01 -6.32567108e-02 -7.62992680e-01 -4.38360423e-01 3.04270923e-01 -1.06602597e+00 8.02878559e-01 -1.43365189e-01 1.28842092e+00 1.85869694e-01 1.11817956e+00 -3.40686948e-03 1.05390108e+00 -2.52112806e-01 2.75772791e-02 -3.41748036e-02 -3.86049569e-01 8.84232461e-01 3.61075312e-01 5.18945456e-01 -1.27207547e-01 2.21245110e-01 1.48239088e+00 2.06989586e-01 -6.25286758e-01 -7.82257915e-01 -1.43650007e+00 2.49980494e-01 7.73797274e-01 8.71994905e-03 -6.53680623e-01 2.95409590e-01 2.05038667e-01 -1.03699170e-01 1.03951141e-01 1.06217349e+00 -2.53924310e-01 -1.33864015e-01 -3.46115112e-01 -2.12945938e-01 6.46600366e-01 1.14765322e+00 3.57041627e-01 -2.94379592e-01 -2.04999700e-01 4.48011875e-01 6.05865300e-01 3.99036616e-01 3.70992869e-01 -1.42826486e+00 2.01928139e-01 5.75121701e-01 4.35961545e-01 -5.22162139e-01 -8.50129426e-01 -4.41723168e-01 -5.94265103e-01 6.81050301e-01 3.35570961e-01 -1.38228789e-01 -1.09258521e+00 7.74164259e-01 4.67316061e-01 6.10163473e-02 -1.80747032e-01 1.21525478e+00 7.31249452e-01 -4.77950945e-02 -3.86736214e-01 -4.37034309e-01 1.38980484e+00 -9.12926733e-01 -7.88410664e-01 -9.87020656e-02 7.07938135e-01 -9.36744869e-01 8.11009765e-01 3.54046792e-01 -1.06750715e+00 -1.83637708e-01 -8.83293867e-01 -1.20290011e-01 -1.05469869e-02 4.63331848e-01 7.83325434e-01 2.56741405e-01 -1.03207397e+00 4.77163792e-01 -1.05953050e+00 -4.15308416e-01 6.41893506e-01 9.33062971e-01 -3.99986625e-01 4.45628911e-02 -2.46534511e-01 1.21685278e+00 3.04111596e-02 5.03314078e-01 -4.18427557e-01 -9.13330674e-01 -5.50094485e-01 -4.37980235e-01 1.69283301e-01 -1.08645511e+00 1.29372358e+00 -3.85624230e-01 -2.27717924e+00 1.01810169e+00 5.46133257e-02 -4.03963894e-01 3.13963085e-01 -4.77515817e-01 9.56000015e-02 4.98778552e-01 -1.52700126e-01 7.18171418e-01 7.28452027e-01 -9.18907106e-01 -3.89267147e-01 -7.29946911e-01 7.41801932e-02 8.11454132e-02 1.27475768e-01 1.72737673e-01 -4.95686650e-01 -1.34951502e-01 6.25894666e-01 -1.31306958e+00 -5.90793848e-01 7.81536877e-01 -1.06401108e-01 1.81751087e-01 4.46850777e-01 -3.79405767e-01 5.76859772e-01 -2.01641893e+00 1.13794968e-01 -1.45212889e-01 6.63947463e-01 4.39762503e-01 2.99312379e-02 -3.37502100e-02 1.20007254e-01 -2.36934379e-01 3.10061365e-01 1.12727284e-02 -6.73458219e-01 -2.23863468e-01 -1.56469926e-01 6.40012801e-01 -3.53931218e-01 1.15165329e+00 -6.78434968e-01 -2.87117124e-01 4.61577654e-01 3.80602241e-01 -5.87366521e-01 2.64720082e-01 -2.33496815e-01 1.08487475e+00 -2.62693912e-01 7.80670345e-01 7.16675758e-01 -1.77610770e-01 -6.12104423e-02 -6.63139045e-01 -6.25796258e-01 1.69141330e-02 -6.03409767e-01 1.98310256e+00 -5.25626481e-01 4.53839093e-01 2.17014611e-01 -3.63033339e-02 7.91999519e-01 1.75579235e-01 6.20908439e-01 -5.42443275e-01 6.59241974e-01 6.74693823e-01 8.82833973e-02 -8.81445527e-01 4.31504339e-01 -2.65170336e-01 5.46919584e-01 2.00321794e-01 -6.19204938e-02 -6.91056848e-01 -3.78039390e-01 -3.00149590e-01 6.28152251e-01 4.32931930e-01 1.56973571e-01 9.92036387e-02 2.03386813e-01 2.83287674e-01 2.95618623e-01 4.48934853e-01 -3.21716428e-01 7.54695117e-01 2.34341130e-01 -5.58122516e-01 -8.72915924e-01 -6.59834743e-01 -2.40772560e-01 -2.34357327e-01 6.90472782e-01 9.70645323e-02 -5.98155379e-01 -4.63903427e-01 4.25028093e-02 -1.89779043e-01 -2.52677053e-01 1.98584780e-01 -5.69680035e-01 -1.61907569e-01 -1.58675030e-01 2.70819992e-01 2.36530080e-01 -7.93392658e-01 -1.14669693e+00 2.73112953e-01 1.52666569e-01 -1.45466375e+00 -2.80438036e-01 -3.55821729e-01 -1.34456325e+00 -1.28767407e+00 -6.99189305e-01 -8.42847466e-01 7.24226177e-01 4.52452362e-01 4.77475703e-01 -2.96876818e-01 -7.66562521e-01 6.86695278e-01 -5.66583760e-02 -6.70978546e-01 -2.09980801e-01 -2.88587511e-01 2.17216223e-01 -2.37038150e-01 -8.57202634e-02 -3.87775809e-01 -1.13479376e+00 4.26911026e-01 -3.73697996e-01 1.63216874e-01 1.01274562e+00 4.94729429e-01 8.38351190e-01 -6.40918672e-01 1.44140437e-01 -5.37818968e-01 3.95079523e-01 4.68273163e-02 -1.26365769e+00 1.51341528e-01 -4.05111969e-01 -3.27871710e-01 2.98364133e-01 -3.45170170e-01 -8.34585249e-01 3.37980032e-01 2.20741391e-01 -9.49422300e-01 4.26936112e-02 4.77172077e-01 4.19724196e-01 -9.23772693e-01 8.67720366e-01 -1.33993402e-01 1.13650048e+00 5.41261658e-02 3.78609300e-01 9.17129993e-01 5.43673158e-01 1.11913376e-01 3.91281337e-01 8.86279404e-01 1.85274929e-01 -4.29119736e-01 -6.26530647e-01 -4.40959573e-01 -9.36252773e-01 -6.29304707e-01 8.18847656e-01 -8.21693122e-01 -1.15658522e+00 6.79827750e-01 -1.38952672e+00 -3.23285222e-01 -2.45462954e-01 1.20192516e+00 -7.84824848e-01 1.99816033e-01 -4.57546085e-01 -4.32396621e-01 -6.36448443e-01 -1.61292160e+00 1.09640789e+00 4.49192792e-01 9.46622416e-02 -5.84793568e-01 -1.28911063e-01 6.43108249e-01 5.79811096e-01 1.11342207e-01 4.72084165e-01 2.03876466e-01 -1.50153637e+00 -4.81533259e-01 -2.75924057e-01 2.42905885e-01 2.46598661e-01 6.64938167e-02 -7.66134441e-01 -1.88494861e-01 2.01694146e-01 -3.01249139e-02 2.85544157e-01 1.03293312e+00 1.10275614e+00 -1.07363880e-01 -6.63383126e-01 1.07017255e+00 1.44068921e+00 3.28251898e-01 8.97396266e-01 2.49337003e-01 7.14561164e-01 7.29510725e-01 1.11469293e+00 2.52711594e-01 2.67082691e-01 9.41793561e-01 8.02592635e-01 3.45173031e-02 -3.33884716e-01 2.90433884e-01 -8.61733593e-03 7.96912491e-01 -5.26174486e-01 4.33140457e-01 -9.22158301e-01 3.55981588e-01 -1.36446893e+00 -3.20665896e-01 -1.97291687e-01 2.50249052e+00 5.54615855e-01 -4.67330307e-01 -3.45756739e-01 -7.61270285e-01 3.88548106e-01 -5.81042171e-01 -7.59754539e-01 -5.24850264e-02 5.33252954e-01 2.29979217e-01 8.42858911e-01 4.07011032e-01 -9.17529464e-01 1.13409007e+00 6.14987898e+00 1.36552706e-01 -1.75711977e+00 3.90634798e-02 -1.53157517e-01 -2.06898317e-01 8.43838081e-02 -2.14436930e-02 -7.42089808e-01 2.04688251e-01 4.08980399e-01 1.85396507e-01 3.79168510e-01 4.43841308e-01 2.92735130e-01 -3.35815430e-01 -8.34801733e-01 1.14462411e+00 7.37784654e-02 -1.86450827e+00 -2.93508880e-02 3.75779301e-01 3.81879151e-01 5.23013413e-01 5.17617613e-02 -4.29670066e-01 -3.87254536e-01 -8.36196899e-01 2.27572605e-01 1.02955639e+00 1.11153197e+00 -3.06044251e-01 6.77906513e-01 1.48272917e-01 -7.26335108e-01 -2.93272585e-01 -4.41620201e-01 1.37063354e-01 8.39530006e-02 6.00731730e-01 -1.19682574e+00 4.45363641e-01 7.20447123e-01 8.92721057e-01 -2.58721232e-01 1.70549834e+00 -4.03682105e-02 -3.14259320e-01 -8.28953162e-02 3.72126251e-02 -2.74268925e-01 -5.64019680e-01 7.90964425e-01 2.72498846e-01 5.00525653e-01 2.32835263e-01 -2.75653124e-01 8.86197031e-01 2.17775375e-01 1.69811606e-01 -9.37596798e-01 6.43827394e-02 3.83679867e-01 1.52348411e+00 -6.56448245e-01 2.98577547e-01 -2.18944237e-01 6.98034167e-01 5.95808104e-02 2.30835944e-01 -5.82686543e-01 -5.47482073e-01 8.54690194e-01 2.59789467e-01 -9.23028141e-02 -6.27659857e-01 -4.74955261e-01 -1.08515561e+00 3.50167811e-01 -1.97396979e-01 -4.51389879e-01 -1.19290674e+00 -6.59226239e-01 5.72537363e-01 -6.35898590e-01 -2.00225902e+00 3.44023146e-02 -1.03206694e+00 -2.05887482e-01 8.40547800e-01 -1.65782619e+00 -1.12887180e+00 -6.61405146e-01 2.97183603e-01 1.95278779e-01 -2.78224736e-01 7.14975178e-01 -1.54885799e-01 -3.14739227e-01 -2.82226782e-02 -8.03455636e-02 -1.15547583e-01 9.84548450e-01 -6.90591514e-01 -1.76464453e-01 5.52829027e-01 -5.00866532e-01 7.99868107e-01 1.29952654e-01 -6.48381114e-01 -1.88452041e+00 -1.14160573e+00 2.90856063e-01 -6.91651583e-01 5.27826011e-01 -1.22411689e-02 -4.33038443e-01 8.49654198e-01 -9.63073447e-02 3.87198627e-01 4.65023547e-01 -2.65831321e-01 6.44697472e-02 -2.68181175e-01 -1.16283786e+00 7.26962984e-01 9.92856085e-01 -2.73651987e-01 -3.07446420e-01 5.99464953e-01 7.64184356e-01 -1.25687909e+00 -1.20200908e+00 7.02051759e-01 9.95357811e-01 -8.80079985e-01 8.00072372e-01 -7.18142390e-02 4.95047837e-01 -5.46762288e-01 5.67738891e-01 -1.05766916e+00 -7.66728669e-02 -8.31915200e-01 1.09060071e-01 2.53031284e-01 9.01164263e-02 -1.23423564e+00 7.83816755e-01 6.35510981e-01 -6.44221365e-01 -1.01937759e+00 -1.06416166e+00 -6.05916977e-01 -2.82185853e-01 -1.50619149e-01 1.84466615e-01 3.48681957e-01 1.47289366e-01 -2.70471126e-01 1.52040556e-01 6.23049259e-01 5.14984071e-01 3.12757820e-01 7.93375492e-01 -1.33471179e+00 1.46636903e-01 -3.60169917e-01 -9.79753435e-01 -1.10685551e+00 -7.45469704e-02 -9.15219665e-01 6.70581236e-02 -1.66769826e+00 -1.66809991e-01 -7.25948989e-01 2.99150884e-01 1.48221716e-01 4.71623898e-01 2.36543030e-01 -1.08379032e-02 7.96240270e-01 -2.19655067e-01 4.02487278e-01 2.01336432e+00 2.53966153e-01 -6.88834429e-01 1.36669308e-01 -3.81799310e-01 5.62954307e-01 7.35484838e-01 -2.01758116e-01 -2.59371042e-01 -5.74103117e-01 1.33230597e-01 4.54482675e-01 5.98063290e-01 -1.20029390e+00 9.11539674e-01 1.56167194e-01 1.55062601e-02 -4.00631040e-01 3.52983981e-01 -1.04764080e+00 9.35022756e-02 6.27892077e-01 2.05251247e-01 -7.33193457e-01 3.50732028e-01 5.36384702e-01 -1.45431459e-01 9.42534730e-02 9.38685298e-01 -6.73272088e-02 -3.23529989e-01 6.11999035e-01 -2.73660511e-01 -4.22291756e-01 1.44909084e+00 -6.97817922e-01 -4.74202931e-01 2.11713597e-01 -9.75686312e-01 1.84701890e-01 6.60586834e-01 2.94315994e-01 1.17579365e+00 -6.66142106e-01 -3.45710933e-01 4.28649336e-01 3.01858157e-01 4.32026267e-01 4.49967027e-01 1.64648151e+00 -1.27327001e+00 7.20097065e-01 -4.16079491e-01 -1.13136351e+00 -1.24833083e+00 2.87657499e-01 1.19941127e+00 4.29847270e-01 -7.81482697e-01 6.21923089e-01 1.63411021e-01 -7.64782071e-01 8.23447630e-02 -6.66231811e-01 -4.78596002e-01 -2.78995484e-01 2.92795360e-01 -1.91369094e-02 2.85050929e-01 -1.57324001e-01 -1.30069152e-01 1.37455106e+00 -1.91638768e-01 1.56100258e-01 1.22404659e+00 -2.63221174e-01 -6.76936209e-01 8.62356722e-02 5.70102692e-01 -8.88905823e-02 -1.29480839e+00 3.70224826e-02 -4.26853031e-01 -6.54693902e-01 2.83725530e-01 -9.65054274e-01 -1.34287715e+00 7.35439897e-01 9.13876355e-01 -5.16612291e-01 9.85362530e-01 -3.66112636e-03 6.06132209e-01 2.11705297e-01 1.09590876e+00 -6.92903817e-01 -2.07321152e-01 4.22831535e-01 1.24904215e+00 -1.12913692e+00 -6.21856377e-02 -1.06518960e+00 -4.35323417e-01 1.52448356e+00 7.69186080e-01 2.68609319e-02 9.21454608e-01 2.19483241e-01 3.27808499e-01 -5.87415040e-01 -2.46434912e-01 -5.98189496e-02 3.70718569e-01 6.18348181e-01 2.62771547e-01 -2.95245647e-03 -2.56937146e-01 1.79010913e-01 -2.20272597e-02 7.55174696e-01 1.01039708e+00 8.56728852e-01 -3.30042809e-01 -6.99653327e-01 -7.33101740e-02 6.52160406e-01 -1.34544522e-01 1.61902867e-02 5.79874590e-02 5.08440793e-01 1.16507716e-01 6.82381690e-01 2.07285315e-01 -2.95794994e-01 5.20142019e-01 -4.34935361e-01 9.78354871e-01 -8.98075938e-01 -5.86411119e-01 1.27762645e-01 -2.30724424e-01 -1.15245044e+00 -3.36009592e-01 -4.21597421e-01 -1.32237899e+00 3.43849808e-01 -4.08795744e-01 -3.25934947e-01 1.25277734e+00 6.46345496e-01 9.90103900e-01 3.92296523e-01 5.50357580e-01 -1.21620214e+00 -1.17322370e-01 -7.56596088e-01 -7.22637236e-01 -6.55882835e-01 3.82169187e-01 -6.94956064e-01 -2.02046975e-01 -1.87848955e-01]
[13.798877716064453, -3.0485522747039795]
69c03413-7218-48c1-af0f-205053cc35e0
adaptive-fine-grained-predicates-learning-for
2207.04602
null
https://arxiv.org/abs/2207.04602v1
https://arxiv.org/pdf/2207.04602v1.pdf
Adaptive Fine-Grained Predicates Learning for Scene Graph Generation
The performance of current Scene Graph Generation (SGG) models is severely hampered by hard-to-distinguish predicates, e.g., woman-on/standing on/walking on-beach. As general SGG models tend to predict head predicates and re-balancing strategies prefer tail categories, none of them can appropriately handle hard-to-distinguish predicates. To tackle this issue, inspired by fine-grained image classification, which focuses on differentiating hard-to-distinguish objects, we propose an Adaptive Fine-Grained Predicates Learning (FGPL-A) which aims at differentiating hard-to-distinguish predicates for SGG. First, we introduce an Adaptive Predicate Lattice (PL-A) to figure out hard-to-distinguish predicates, which adaptively explores predicate correlations in keeping with model's dynamic learning pace. Practically, PL-A is initialized from SGG dataset, and gets refined by exploring model's predictions of current mini-batch. Utilizing PL-A, we propose an Adaptive Category Discriminating Loss (CDL-A) and an Adaptive Entity Discriminating Loss (EDL-A), which progressively regularize model's discriminating process with fine-grained supervision concerning model's dynamic learning status, ensuring balanced and efficient learning process. Extensive experimental results show that our proposed model-agnostic strategy significantly boosts performance of benchmark models on VG-SGG and GQA-SGG datasets by up to 175% and 76% on Mean Recall@100, achieving new state-of-the-art performance. Moreover, experiments on Sentence-to-Graph Retrieval and Image Captioning tasks further demonstrate practicability of our method.
['Jingkuan Song', 'Heng Tao Shen', 'Pengpeng Zeng', 'Lianli Gao', 'Xinyu Lyu']
2022-07-11
null
null
null
null
['scene-graph-generation', 'fine-grained-image-classification']
['computer-vision', 'computer-vision']
[ 3.34862918e-01 4.17231858e-01 -4.05877501e-01 -3.81183743e-01 -8.25398982e-01 -4.99459922e-01 6.79004073e-01 2.13285357e-01 -5.69545738e-02 6.89441621e-01 4.64519113e-02 -2.40235776e-01 -2.36611158e-01 -1.16518211e+00 -8.38874817e-01 -6.35710418e-01 -1.70693919e-01 8.42181623e-01 6.24476790e-01 -6.02739155e-02 -5.56800701e-02 2.31564328e-01 -1.78801572e+00 4.97027844e-01 1.12983525e+00 1.32057083e+00 2.83999085e-01 3.73138875e-01 -2.26403296e-01 8.48818839e-01 -3.60250622e-01 -6.29088521e-01 1.90628141e-01 -2.02292219e-01 -9.44990456e-01 2.75983721e-01 6.32268250e-01 3.95135283e-02 -6.55926317e-02 1.05247271e+00 3.01962197e-01 -1.11828335e-01 7.17513442e-01 -1.39115572e+00 -7.96567380e-01 7.04471886e-01 -5.21295011e-01 1.16888888e-01 3.90295684e-01 2.79343188e-01 1.50220978e+00 -7.44518220e-01 6.14837050e-01 1.50407851e+00 4.31220621e-01 5.25414288e-01 -1.13746142e+00 -7.06973672e-01 7.45534897e-01 4.05251890e-01 -1.49390578e+00 5.10101505e-02 8.56553793e-01 -4.31811154e-01 6.52512372e-01 4.40322191e-01 5.88010550e-01 1.10721517e+00 1.40974954e-01 1.04720497e+00 1.05101943e+00 -1.06624752e-01 1.02472469e-01 -1.10445738e-01 4.97628599e-02 8.35170925e-01 3.10380399e-01 -1.30362496e-01 -4.91364390e-01 1.07858190e-02 4.84436810e-01 -4.17853296e-01 -3.55156839e-01 -6.52438879e-01 -1.10796630e+00 6.94636226e-01 4.07093048e-01 -1.90634340e-01 -4.09699678e-01 -1.41904995e-01 3.97694200e-01 9.93021056e-02 3.01278710e-01 2.95649379e-01 -6.21978879e-01 8.53803158e-02 -5.97061276e-01 3.68938267e-01 7.06089616e-01 1.40232587e+00 9.13731992e-01 -3.27454120e-01 -9.18529451e-01 8.51282775e-01 3.25813681e-01 3.22521299e-01 3.76214355e-01 -4.98171300e-01 6.69893801e-01 8.93263757e-01 -2.12515682e-01 -1.03088355e+00 -3.58977407e-01 -5.51639616e-01 -9.41807985e-01 -4.33153361e-01 2.76585937e-01 1.76286161e-01 -1.17654955e+00 1.75762904e+00 4.15396392e-01 2.47809529e-01 -7.25843757e-02 5.91381907e-01 1.06822634e+00 4.81046408e-01 7.93618381e-01 -1.19112462e-01 1.49499619e+00 -1.18522871e+00 -1.36725605e-01 -4.35169131e-01 5.94460785e-01 -1.64458394e-01 1.50788116e+00 2.01151952e-01 -8.85511100e-01 -8.30430925e-01 -7.01550782e-01 8.79683197e-02 -4.18145776e-01 6.62965998e-02 6.74231529e-01 5.54941177e-01 -1.01171100e+00 4.92565423e-01 -5.60873687e-01 -3.10030073e-01 7.85621464e-01 2.62814492e-01 -4.14739788e-01 -1.19070105e-01 -1.34492290e+00 3.57654065e-01 1.00254083e+00 -1.78008288e-01 -9.48239267e-01 -8.51632416e-01 -1.08592606e+00 2.83060014e-01 6.61246061e-01 -9.95577931e-01 9.31911111e-01 -6.16637826e-01 -1.16469491e+00 1.21746409e+00 6.35231733e-02 -5.28591275e-01 7.04492927e-01 -2.16157082e-02 -5.88132024e-01 2.49094635e-01 4.14035618e-01 9.12506402e-01 9.20279205e-01 -1.26095736e+00 -9.51145172e-01 -2.61362106e-01 2.91895509e-01 4.31562334e-01 -2.28603005e-01 -2.05054149e-01 -9.20261204e-01 -5.94532490e-01 2.63183683e-01 -8.20941508e-01 -1.92128479e-01 -1.85403571e-01 -7.26904869e-01 -7.01933622e-01 5.96825659e-01 -3.17569137e-01 1.35805392e+00 -2.09589124e+00 -3.38993818e-01 1.64552554e-01 3.18606704e-01 2.90553123e-01 -3.71691793e-01 2.43765533e-01 -6.53033331e-02 6.34171590e-02 -1.03951588e-01 -3.61761898e-01 1.86902061e-01 4.48815942e-01 -3.19693625e-01 1.31673515e-01 6.37646437e-01 1.09054255e+00 -1.17086053e+00 -9.19969141e-01 1.38912141e-01 1.09879561e-01 -6.04867756e-01 2.26839960e-01 -5.84395707e-01 2.05521807e-01 -8.49475145e-01 9.60212350e-01 7.59272158e-01 -7.49781609e-01 2.19913557e-01 -2.79110014e-01 3.06249529e-01 2.53761709e-01 -1.14662993e+00 1.21499765e+00 -1.40721634e-01 -1.00505523e-01 -3.49641889e-01 -9.76342559e-01 1.00909364e+00 -1.77737594e-01 1.41066015e-01 -8.20728421e-01 -1.73528761e-01 6.37819916e-02 -1.13343477e-01 -3.15876037e-01 4.86116946e-01 9.43356678e-02 -3.23169529e-01 -3.04497749e-01 -1.51073588e-02 8.92240778e-02 4.93577778e-01 2.01785713e-01 1.09267354e+00 3.24051619e-01 3.09335351e-01 -4.42244530e-01 1.00121617e+00 1.97466044e-03 8.30987394e-01 1.00467253e+00 -3.70864481e-01 6.61039710e-01 7.22948074e-01 -3.22752595e-01 -4.62036043e-01 -1.03589571e+00 -8.53447840e-02 1.34157920e+00 5.41046441e-01 -5.22725344e-01 -5.98260343e-01 -1.20300448e+00 7.10990727e-02 6.74889863e-01 -7.70059168e-01 -1.50332496e-01 -5.30308127e-01 -7.29849994e-01 4.20994192e-01 6.29815936e-01 7.97871292e-01 -1.39497638e+00 -2.26013452e-01 2.53605157e-01 -2.17225686e-01 -1.28092313e+00 -2.24573970e-01 5.00475615e-02 -4.72708225e-01 -1.30070901e+00 -3.76427561e-01 -9.93031144e-01 7.49404609e-01 -3.44248414e-02 1.62864876e+00 -7.20762238e-02 1.08827159e-01 2.48367906e-01 -4.79413360e-01 -2.04037398e-01 -2.16026515e-01 1.06581330e-01 -3.85101050e-01 1.98512495e-01 2.60692567e-01 -3.65215242e-01 -6.77700520e-01 4.63941187e-01 -6.58057272e-01 2.55543053e-01 7.08053887e-01 8.03694427e-01 1.18240166e+00 2.18160257e-01 4.71519947e-01 -1.39736533e+00 4.42813307e-01 -5.71884394e-01 -5.68916321e-01 5.68819702e-01 -7.41591930e-01 -7.67040402e-02 8.95264745e-01 -4.88338023e-01 -9.24377441e-01 -1.31985247e-01 -1.08702481e-01 -5.93887627e-01 -3.24503511e-01 3.14279735e-01 -7.60939121e-01 8.18057135e-02 3.71436149e-01 4.55502301e-01 -6.37323380e-01 -1.40842110e-01 4.68503088e-01 2.64290899e-01 6.92649424e-01 -9.91870522e-01 8.33494842e-01 2.56854236e-01 -7.37477886e-03 -4.39767003e-01 -1.04884422e+00 -5.44593811e-01 -1.98202297e-01 3.16302516e-02 8.55446279e-01 -1.16369951e+00 -7.65112281e-01 6.31524146e-01 -6.05805695e-01 -6.23404860e-01 -2.69691616e-01 -1.32791698e-01 -5.67696929e-01 4.60578263e-01 -4.71536577e-01 -4.81127828e-01 -3.16202760e-01 -9.46980000e-01 1.49993515e+00 3.45157206e-01 7.27652684e-02 -9.65370953e-01 -2.50516176e-01 5.02637982e-01 7.75583312e-02 4.12948698e-01 1.06501210e+00 -8.57509196e-01 -7.05758035e-01 1.73524290e-01 -5.73731363e-01 2.89731860e-01 2.51916610e-02 -3.31563115e-01 -8.00492942e-01 -3.74754131e-01 -5.45396686e-01 -4.08612698e-01 8.55224788e-01 1.32678851e-01 1.83588970e+00 -5.26055098e-01 -6.23394907e-01 6.39665902e-01 1.34317064e+00 -2.32910626e-02 6.55156434e-01 4.62635994e-01 1.04683340e+00 4.63821650e-01 1.09500599e+00 4.04184401e-01 9.35799062e-01 4.76888508e-01 6.31070733e-01 -2.86431704e-02 -4.25059706e-01 -7.61455238e-01 -1.34272933e-01 2.80013412e-01 1.62106797e-01 -5.32004058e-01 -8.20106387e-01 7.56405950e-01 -1.90655899e+00 -6.51034772e-01 -6.19187839e-02 1.95911205e+00 9.40315962e-01 4.95973140e-01 1.12528484e-02 -7.40423650e-02 6.69716656e-01 5.16368389e-01 -5.10563254e-01 -2.46209651e-02 -2.96915829e-01 2.53281534e-01 5.37523150e-01 2.02336252e-01 -1.54587376e+00 1.40132844e+00 4.70323181e+00 1.33115399e+00 -9.71130550e-01 -1.55392647e-01 8.95994186e-01 5.30447006e-01 -4.36716765e-01 6.58787116e-02 -1.15539908e+00 7.38062203e-01 5.44990301e-01 -7.11753592e-02 1.40317921e-02 1.06154394e+00 -1.62079245e-01 1.99677959e-01 -1.17986047e+00 9.90169168e-01 -2.65102219e-02 -1.21652687e+00 4.99071896e-01 -5.78553341e-02 7.42954433e-01 -5.91971576e-02 -1.66395187e-01 8.78318429e-01 4.25534397e-01 -7.21996069e-01 8.62352788e-01 1.74548566e-01 8.18594217e-01 -4.36087459e-01 5.11074126e-01 4.10355479e-01 -1.56582975e+00 -1.10304505e-01 -3.74492437e-01 1.84917092e-01 -3.38516757e-02 4.66692239e-01 -1.03158855e+00 8.55123639e-01 8.20246875e-01 6.88298047e-01 -9.29962456e-01 9.38844204e-01 -6.49576187e-01 8.09576988e-01 -1.66732818e-01 -1.49801001e-02 4.13043886e-01 -1.14403069e-02 3.75870705e-01 1.33734679e+00 -5.47755323e-02 1.65727019e-01 6.49797022e-01 5.85811377e-01 -2.29115427e-01 6.89075068e-02 -1.75247878e-01 9.78439525e-02 5.86845398e-01 1.22762549e+00 -7.00578749e-01 -3.93872678e-01 -2.64606833e-01 8.61483693e-01 4.90430623e-01 2.90698081e-01 -9.61962581e-01 -1.68967053e-01 5.90356648e-01 4.51966047e-01 6.31954491e-01 2.76437551e-01 4.27869223e-02 -9.04127359e-01 1.97415948e-01 -8.31999481e-01 9.96953070e-01 -7.32761025e-01 -1.48965085e+00 7.20322669e-01 1.77653939e-01 -1.25040364e+00 -1.15611196e-01 -5.76551557e-01 -3.72732133e-01 6.04096174e-01 -1.93591487e+00 -1.72127855e+00 -5.01335144e-01 8.19044530e-01 5.49732089e-01 1.69841707e-01 5.97117007e-01 2.36637354e-01 -4.06649798e-01 9.08997834e-01 -5.30194342e-01 1.62024766e-01 5.36647439e-01 -1.45506036e+00 5.10221064e-01 9.43608046e-01 9.25693959e-02 2.11602405e-01 4.46336240e-01 -8.24207723e-01 -1.11642337e+00 -1.72922015e+00 1.04421699e+00 -3.73371601e-01 8.00663352e-01 -3.26095670e-01 -9.67675626e-01 7.51267612e-01 -2.94746071e-01 4.27763909e-01 4.44740921e-01 2.54913241e-01 -6.27309442e-01 -2.62682587e-01 -1.13653123e+00 5.41468799e-01 1.47524524e+00 -3.39100003e-01 -5.58347464e-01 6.20111644e-01 8.04210782e-01 -6.63538694e-01 -8.10815990e-01 8.94659102e-01 1.22508854e-01 -7.80507684e-01 9.87990379e-01 -6.65956855e-01 4.15694237e-01 -4.40694392e-01 -1.36875823e-01 -8.92606914e-01 -8.13435197e-01 -5.69094419e-01 -2.80276150e-01 1.38665986e+00 3.01399201e-01 -8.04059744e-01 1.01852357e+00 4.25097495e-01 -4.70034361e-01 -1.07851517e+00 -6.25855625e-01 -9.25424635e-01 -1.50176466e-01 -3.28821242e-01 8.25188816e-01 9.34194684e-01 -6.24352455e-01 1.41330600e-01 -1.03185341e-01 5.50203562e-01 7.07301378e-01 4.46113437e-01 8.07608485e-01 -1.27510977e+00 -5.33382773e-01 -5.03227174e-01 -5.94757974e-01 -1.16424978e+00 2.07110390e-01 -7.99087822e-01 3.88176069e-02 -1.76894081e+00 3.22153836e-01 -7.88093746e-01 -5.06441176e-01 9.25821304e-01 -6.44729793e-01 3.39311153e-01 2.33854458e-01 8.65767822e-02 -1.18953061e+00 5.71850240e-01 1.49362159e+00 -3.40713322e-01 -9.48950369e-03 2.54916754e-02 -1.10112560e+00 7.12704599e-01 4.80562121e-01 -2.70859540e-01 -7.53783047e-01 -1.65645093e-01 1.08938485e-01 -1.35198802e-01 4.44790989e-01 -9.27017331e-01 2.73515321e-02 -2.41458118e-01 2.20307484e-01 -7.60498703e-01 1.20225966e-01 -4.02963340e-01 1.65895194e-01 1.00309901e-01 -4.86601554e-02 -3.16512793e-01 2.19066724e-01 7.14630961e-01 -4.01638538e-01 2.25315675e-01 5.40821135e-01 -9.52633023e-02 -1.37018442e+00 7.33302772e-01 3.47039998e-01 5.79741776e-01 1.09515738e+00 -4.67324644e-01 -5.47589898e-01 1.01084463e-01 -7.01993585e-01 7.71799862e-01 3.70753884e-01 5.81245959e-01 4.22944188e-01 -1.20989895e+00 -7.95406342e-01 2.77655303e-01 7.01254845e-01 4.36883181e-01 4.36216623e-01 4.50350612e-01 -3.53144914e-01 3.01629752e-01 1.13443531e-01 -8.39003921e-01 -1.18618608e+00 6.25822544e-01 -6.75175386e-03 -9.11346316e-01 -6.00748658e-01 1.28791654e+00 7.79439211e-01 -3.02783579e-01 1.75632134e-01 -4.46344435e-01 -4.11818922e-01 -2.79300325e-02 1.66141868e-01 -4.32281569e-02 9.32269394e-02 -5.87902844e-01 -5.77293038e-01 4.88336861e-01 -3.58109325e-01 7.32146382e-01 1.03753817e+00 -6.59160241e-02 1.04414873e-01 -6.27774522e-02 8.87322962e-01 -4.32577766e-02 -1.58214498e+00 -1.87677518e-01 -1.87734421e-02 -4.49468672e-01 -4.21468526e-01 -9.72632706e-01 -8.74605894e-01 5.09296358e-01 2.73890227e-01 4.27143186e-01 1.55284083e+00 4.76178259e-01 6.86401725e-01 1.64633602e-01 6.37505591e-01 -8.77375126e-01 -1.03866979e-02 5.17485917e-01 6.75694704e-01 -1.23174417e+00 -1.83843687e-01 -9.30839181e-01 -7.77652383e-01 4.54836190e-01 9.79787230e-01 4.29186560e-02 4.39649910e-01 -2.71901786e-01 -2.56318629e-01 -2.36091360e-01 -7.21353590e-01 -3.79820645e-01 6.15790367e-01 6.29736483e-01 -4.44024354e-02 4.96330470e-01 -2.34624192e-01 7.33005464e-01 -3.49508405e-01 -2.68098384e-01 -5.89049943e-02 5.97949982e-01 -2.10833400e-01 -1.10817957e+00 7.41095692e-02 5.63884079e-01 -2.47353941e-01 -2.77841687e-01 -1.86723351e-01 1.01601231e+00 4.45281386e-01 7.71601915e-01 -1.05131557e-02 -1.95589304e-01 5.20034850e-01 -2.01004341e-01 4.34889346e-01 -7.04940319e-01 -4.00128812e-01 -1.30554959e-01 3.77635747e-01 -8.11084330e-01 -2.30585933e-01 -5.90805709e-01 -1.19429088e+00 2.45226517e-01 -1.52821690e-01 1.62605733e-01 -1.27984453e-02 7.94061065e-01 4.77260292e-01 5.96561968e-01 5.00256777e-01 -4.32838112e-01 -5.28591573e-01 -6.85663581e-01 -6.72585189e-01 7.38727808e-01 -1.41305029e-02 -8.09054792e-01 -1.84267834e-01 -9.91282170e-04]
[10.300887107849121, 1.7562081813812256]
da007d13-97dd-435b-a5e9-4b5d6e0e5f70
a-comparative-attention-framework-for-better
2210.13923
null
https://arxiv.org/abs/2210.13923v1
https://arxiv.org/pdf/2210.13923v1.pdf
A Comparative Attention Framework for Better Few-Shot Object Detection on Aerial Images
Few-Shot Object Detection (FSOD) methods are mainly designed and evaluated on natural image datasets such as Pascal VOC and MS COCO. However, it is not clear whether the best methods for natural images are also the best for aerial images. Furthermore, direct comparison of performance between FSOD methods is difficult due to the wide variety of detection frameworks and training strategies. Therefore, we propose a benchmarking framework that provides a flexible environment to implement and compare attention-based FSOD methods. The proposed framework focuses on attention mechanisms and is divided into three modules: spatial alignment, global attention, and fusion layer. To remain competitive with existing methods, which often leverage complex training, we propose new augmentation techniques designed for object detection. Using this framework, several FSOD methods are reimplemented and compared. This comparison highlights two distinct performance regimes on aerial and natural images: FSOD performs worse on aerial images. Our experiments suggest that small objects, which are harder to detect in the few-shot setting, account for the poor performance. Finally, we develop a novel multiscale alignment method, Cross-Scales Query-Support Alignment (XQSA) for FSOD, to improve the detection of small objects. XQSA outperforms the state-of-the-art significantly on DOTA and DIOR.
['Anissa Mokraoui', 'Pierre Le Jeune']
2022-10-25
null
null
null
null
['few-shot-object-detection']
['computer-vision']
[ 2.90585101e-01 -4.33284581e-01 -6.44361824e-02 -5.12584522e-02 -5.84003806e-01 -3.81641775e-01 6.70071959e-01 4.43959124e-02 -5.99892080e-01 2.80194879e-01 -2.06687689e-01 5.98306917e-02 -5.05821919e-03 -6.13358498e-01 -5.31645179e-01 -6.22111082e-01 9.11258236e-02 -4.39219996e-02 9.75050330e-01 -3.94857228e-01 1.73575744e-01 5.39610863e-01 -1.96720099e+00 1.75137714e-01 7.89621890e-01 1.16379523e+00 5.86854994e-01 7.61059642e-01 1.91906691e-02 7.37819731e-01 -7.15979040e-01 -3.48039418e-01 5.57638228e-01 -4.43314165e-01 -6.94334924e-01 7.17192292e-02 9.72883880e-01 -6.40239835e-01 -2.57280737e-01 1.18069184e+00 7.18277156e-01 3.01024795e-01 5.81254184e-01 -1.29979599e+00 -6.84827149e-01 3.55787158e-01 -6.65661216e-01 8.17811131e-01 1.71004459e-01 3.77416223e-01 1.18656087e+00 -1.27445865e+00 4.74704355e-01 1.17471230e+00 6.89985693e-01 3.60670537e-01 -1.05270410e+00 -3.89644593e-01 2.62504399e-01 3.85237008e-01 -1.38189936e+00 -4.35815603e-01 3.89903516e-01 -4.83822882e-01 1.11839914e+00 4.17830616e-01 4.46792603e-01 1.05762935e+00 8.03313032e-02 1.01895916e+00 1.00700665e+00 -5.58431327e-01 1.40364632e-01 7.33686537e-02 1.78080037e-01 6.32816315e-01 3.10002893e-01 1.72187582e-01 -3.88076127e-01 1.11009240e-01 3.90099466e-01 1.70307815e-01 -3.38271677e-01 -5.60338616e-01 -1.11799967e+00 7.22151935e-01 6.50536478e-01 2.33767554e-01 -3.61551791e-01 -1.19673617e-01 4.41042632e-01 4.39308956e-03 1.36541620e-01 5.64206660e-01 -1.34913296e-01 1.62686720e-01 -1.08585358e+00 4.02091056e-01 4.08939660e-01 9.55925107e-01 5.28939426e-01 3.40933949e-02 -7.25721717e-01 8.35703075e-01 1.21479640e-02 5.49647868e-01 4.44040060e-01 -7.66778469e-01 2.31882140e-01 6.55018628e-01 3.30219299e-01 -1.18830502e+00 -3.91665518e-01 -4.53694701e-01 -5.27861893e-01 4.13311452e-01 2.97750175e-01 1.05662763e-01 -9.62669909e-01 1.37675333e+00 3.73550475e-01 2.06406917e-02 -1.34162745e-02 1.33413804e+00 1.04553366e+00 5.51280379e-01 1.37303635e-01 9.11467448e-02 1.36479211e+00 -1.30377877e+00 -5.76555967e-01 -4.99356687e-01 4.45447266e-01 -7.91606426e-01 1.29721272e+00 5.82296737e-02 -1.07329333e+00 -8.54460359e-01 -1.25976121e+00 -1.97190959e-02 -6.28490448e-01 5.35940826e-01 4.59908098e-01 5.19039690e-01 -7.16526747e-01 5.73815703e-01 -7.28284240e-01 -8.32258701e-01 4.89291728e-01 5.53577319e-02 -4.52527776e-02 -3.99863832e-02 -1.01016641e+00 9.42149103e-01 4.84234989e-01 -8.81672837e-03 -1.17276478e+00 -4.71493036e-01 -8.49695027e-01 1.68869317e-01 6.70245409e-01 -5.03373504e-01 1.30961478e+00 -7.26572096e-01 -1.27048063e+00 9.66540098e-01 2.92700887e-01 -6.83281422e-01 5.03327906e-01 -5.03455877e-01 -2.57801980e-01 1.93336755e-01 2.99331307e-01 8.58867347e-01 9.64226484e-01 -1.01621020e+00 -8.80893528e-01 -1.72367185e-01 4.14763421e-01 1.88419685e-01 -4.02533054e-01 4.10091758e-01 -6.16036713e-01 -6.53294921e-01 -3.09053749e-01 -6.99410021e-01 -1.36064738e-01 3.18923861e-01 -4.19771612e-01 -1.08544581e-01 1.02432668e+00 -2.18075782e-01 1.35462415e+00 -2.22777581e+00 9.84618738e-02 -4.23640519e-01 9.92206857e-03 8.01910281e-01 -3.34474534e-01 3.58056009e-01 9.53070000e-02 -1.94237411e-01 -3.55528206e-01 -3.81949663e-01 9.21007618e-02 4.15289514e-02 -3.12472463e-01 3.94390821e-01 6.22315168e-01 8.10708344e-01 -8.98029327e-01 -5.57213962e-01 4.21750426e-01 1.06172509e-01 -4.11746442e-01 3.33609551e-01 -1.27338186e-01 -1.46895517e-02 -1.52484581e-01 1.07340860e+00 7.16613650e-01 -3.53761643e-01 -2.69327611e-01 -2.12389767e-01 -4.32187676e-01 -1.45528361e-01 -1.32663369e+00 1.60532582e+00 -2.70653144e-02 5.48562109e-01 4.14585881e-02 -8.35219920e-01 7.97342718e-01 -1.41045764e-01 1.76857367e-01 -6.14132106e-01 3.84860992e-01 6.53183460e-02 1.11684725e-01 -6.28883481e-01 8.54876876e-01 2.28081673e-01 1.49328709e-01 8.17421973e-02 4.06705886e-01 -6.98799733e-03 6.95803940e-01 2.85045862e-01 1.16775739e+00 1.08806118e-01 6.28222346e-01 -2.66616762e-01 4.74638790e-01 1.39827490e-01 5.15540838e-01 1.02909267e+00 -5.95813990e-01 7.99996138e-01 3.40853333e-01 -3.30537587e-01 -8.84512007e-01 -8.39321375e-01 -2.72930115e-01 1.49845278e+00 5.83546817e-01 -5.62145472e-01 -8.35653603e-01 -7.77037263e-01 -6.85178190e-02 4.05076772e-01 -7.58194387e-01 1.01496875e-02 -1.84096217e-01 -9.50990498e-01 4.80279505e-01 7.28785872e-01 6.28752470e-01 -1.05067945e+00 -1.26018417e+00 1.93631679e-01 4.73932773e-02 -1.39194870e+00 -4.68819976e-01 1.43985957e-01 -4.59994197e-01 -1.24864423e+00 -9.05050755e-01 -5.26992381e-01 2.45809451e-01 9.78400886e-01 9.88435388e-01 6.48168251e-02 -7.08087325e-01 3.51579785e-01 -6.41421854e-01 -7.10942268e-01 -9.73669142e-02 -1.34374080e-02 8.71384293e-02 8.09690505e-02 3.89824778e-01 1.04689216e-02 -8.56246293e-01 6.15254939e-01 -9.73088145e-01 -2.04873100e-01 8.17523956e-01 9.99481559e-01 5.10395050e-01 -4.10323292e-01 1.63435385e-01 -6.56273305e-01 2.70389467e-01 -3.44516158e-01 -7.93922186e-01 3.25871974e-01 -4.56292629e-01 -2.95311064e-01 5.56734860e-01 -4.90756452e-01 -7.35426545e-01 9.68139023e-02 1.81808975e-02 -7.62547612e-01 -3.07581216e-01 1.21071316e-01 6.93583339e-02 -3.13445508e-01 1.04641473e+00 1.11614816e-01 -1.62020981e-01 -4.80572045e-01 1.55140519e-01 6.95632219e-01 7.11818159e-01 -1.32246763e-01 5.96773922e-01 5.32485187e-01 -3.74392837e-01 -1.02539575e+00 -1.07067239e+00 -8.47061336e-01 -6.65183544e-01 -1.46681383e-01 9.37286198e-01 -8.54808450e-01 -5.48918284e-02 5.09837508e-01 -1.14018488e+00 -2.95582175e-01 -4.26597863e-01 4.16228473e-01 -4.33949023e-01 3.07588965e-01 -4.44210947e-01 -7.74411440e-01 -4.96908724e-01 -1.25158834e+00 1.41274261e+00 3.01236540e-01 -6.09724820e-02 -4.10542399e-01 -4.91750278e-02 1.14787892e-01 4.69691366e-01 2.19354838e-01 4.45127219e-01 -6.27381206e-01 -6.12514436e-01 -1.91417471e-01 -5.05262792e-01 3.38364571e-01 -1.63644418e-01 3.24199528e-01 -1.15706384e+00 -4.29592371e-01 -4.02243078e-01 -4.96602386e-01 1.02014446e+00 4.50204223e-01 9.76267636e-01 9.33092907e-02 -3.79682332e-01 6.11104369e-01 1.39198434e+00 1.14626214e-02 6.26452029e-01 4.72023040e-01 3.51143420e-01 4.28499371e-01 1.12528646e+00 4.66174126e-01 8.37829709e-02 9.49209809e-01 7.11015821e-01 -2.84209341e-01 -3.16379666e-01 1.35531545e-01 4.62249249e-01 3.07434082e-01 -1.56319812e-01 -3.65907282e-01 -8.10790002e-01 5.88729918e-01 -1.88940215e+00 -1.01411259e+00 -1.18370719e-01 2.14268899e+00 2.31037036e-01 2.82956690e-01 4.05258477e-01 1.35595605e-01 7.06788838e-01 4.02792484e-01 -4.41160738e-01 -6.01148941e-02 -2.18885675e-01 5.37082627e-02 4.55735177e-01 -5.55049954e-03 -1.54724443e+00 9.09117222e-01 6.63263607e+00 8.90684724e-01 -1.07837903e+00 2.33761400e-01 2.65219092e-01 -2.12904755e-02 5.60165644e-01 -2.44417310e-01 -1.13507795e+00 4.35556412e-01 5.08754551e-01 -2.72851791e-02 1.18678190e-01 1.14046931e+00 -1.31574972e-02 -1.29476577e-01 -9.85214531e-01 9.99668896e-01 4.01085556e-01 -1.19664478e+00 -1.05042711e-01 -4.39818174e-01 7.71510661e-01 3.69414568e-01 -7.81979878e-03 5.64559758e-01 1.74133070e-02 -7.06697226e-01 8.78088415e-01 1.58279181e-01 5.34889221e-01 -2.93778986e-01 8.41379106e-01 2.44486153e-01 -1.27893424e+00 -4.74685252e-01 -7.63190269e-01 4.46116664e-02 -1.70771442e-02 1.29646733e-01 -4.22472417e-01 5.29230237e-01 1.14309537e+00 7.15938389e-01 -9.45102394e-01 1.47594988e+00 -5.79974055e-02 2.24695340e-01 -1.27552465e-01 -7.25794733e-02 5.48606873e-01 1.14002228e-01 7.94378519e-01 1.41565621e+00 3.32462221e-01 2.96867080e-02 3.06884497e-01 9.48863745e-01 6.37835339e-02 1.61285773e-01 -6.65580451e-01 7.27189984e-03 3.09046239e-01 1.58639193e+00 -8.22631717e-01 -3.79030436e-01 -7.31677294e-01 9.87280130e-01 2.64376640e-01 1.64701387e-01 -1.06783426e+00 -4.81319278e-01 4.96913970e-01 1.43230677e-01 6.74099386e-01 -6.48997948e-02 3.15379471e-01 -1.05117369e+00 4.23386600e-03 -1.06569958e+00 6.50016129e-01 -7.87914038e-01 -1.18682325e+00 7.53026009e-01 1.24437898e-01 -1.41926932e+00 1.89048186e-01 -8.76026928e-01 -8.93398106e-01 3.65607202e-01 -1.61115229e+00 -1.11315000e+00 -8.64721715e-01 4.23654646e-01 9.34675753e-01 -1.78777292e-01 6.46950305e-01 4.06462878e-01 -8.88041914e-01 5.55625021e-01 1.02806715e-02 6.97591752e-02 6.67661667e-01 -1.27311671e+00 4.76124316e-01 1.21325338e+00 3.26165646e-01 2.77191937e-01 5.88737726e-01 -2.57091612e-01 -1.21498418e+00 -1.20780766e+00 1.56151220e-01 -4.24912959e-01 6.06006026e-01 -2.79470265e-01 -1.02738607e+00 3.19135338e-01 1.71909079e-01 4.95231479e-01 3.70295525e-01 -2.11299270e-01 -1.47276580e-01 -1.16936795e-01 -8.35509062e-01 4.14520264e-01 1.00431669e+00 -1.68088600e-01 -5.82425535e-01 3.05582970e-01 6.59067810e-01 -4.05149370e-01 -5.20261824e-01 6.37608111e-01 3.56566042e-01 -1.27850091e+00 1.06551468e+00 -5.48143148e-01 3.76100063e-01 -5.65996528e-01 -3.49974543e-01 -1.06133485e+00 -4.00703937e-01 -4.88578051e-01 -3.26767206e-01 1.13409567e+00 1.51297599e-01 -2.40835339e-01 4.83006716e-01 4.86843511e-02 -2.10486308e-01 -7.53149867e-01 -6.34276271e-01 -1.15922415e+00 -3.54307860e-01 -8.24453235e-02 4.48817641e-01 7.24828184e-01 -4.08409268e-01 3.71044546e-01 -4.28302705e-01 2.56755739e-01 4.97781336e-01 3.87208492e-01 1.03588223e+00 -9.89611149e-01 -3.74823749e-01 -4.83496845e-01 -7.56761849e-01 -9.76095855e-01 -3.70206088e-01 -5.13490558e-01 2.74170846e-01 -1.32953644e+00 3.55430216e-01 -8.35189223e-02 -5.13968050e-01 3.84893596e-01 -4.40404385e-01 5.59390664e-01 4.59652662e-01 2.29580134e-01 -1.04735816e+00 7.02850282e-01 1.00282252e+00 -2.82589972e-01 -1.40631020e-01 -1.37727737e-01 -3.48390013e-01 8.49664569e-01 5.92495501e-01 -3.72821987e-01 -8.00829008e-02 -3.66364807e-01 -2.92850882e-01 -3.89649540e-01 5.79085648e-01 -1.49345386e+00 2.43381545e-01 -1.78352296e-02 2.95557022e-01 -7.33524799e-01 3.22217375e-01 -6.16307497e-01 -3.89593452e-01 7.02107072e-01 -1.50444582e-01 1.00887708e-01 2.88279891e-01 6.54478490e-01 -2.75441498e-01 -3.99375558e-01 1.15913427e+00 -4.58155610e-02 -1.22077978e+00 2.57677406e-01 -1.91707090e-01 1.35631099e-01 1.37785816e+00 -2.90284365e-01 -3.54944050e-01 1.47880977e-02 -4.45528448e-01 2.88719982e-01 3.16604018e-01 5.98878205e-01 6.24293506e-01 -1.21461785e+00 -7.67371774e-01 1.73948586e-01 6.18988276e-01 -9.02716070e-02 2.72538871e-01 1.13805234e+00 -5.89207828e-01 3.82032216e-01 -4.02169168e-01 -8.01246941e-01 -1.42042398e+00 8.25409949e-01 3.92415494e-01 -1.61064342e-01 -6.30810976e-01 9.68433142e-01 4.43657190e-01 -1.79888010e-01 3.67393494e-01 -5.01437247e-01 -2.25839630e-01 2.35471725e-01 8.72256339e-01 5.46616018e-01 1.71866357e-01 -6.28436685e-01 -3.39929223e-01 5.52962542e-01 -1.77719250e-01 4.43082273e-01 1.20310462e+00 -1.38582960e-01 3.17839682e-01 2.86454350e-01 6.99374139e-01 -3.83978665e-01 -1.35894585e+00 -3.27703178e-01 -9.23994258e-02 -7.71775067e-01 8.99110287e-02 -5.43408751e-01 -8.67364168e-01 1.03695798e+00 9.71679211e-01 3.25212806e-01 1.16777337e+00 -1.72626287e-01 5.35831571e-01 4.00954962e-01 1.97781533e-01 -1.15290511e+00 3.75257581e-01 3.98420423e-01 9.49647844e-01 -1.51659429e+00 1.62177235e-01 -3.85189623e-01 -7.20390916e-01 9.40632761e-01 9.10613060e-01 -1.17195994e-01 1.68218732e-01 1.68793842e-01 -1.92286521e-01 -2.55205363e-01 -4.49614495e-01 -7.11648524e-01 5.04158139e-01 4.37378645e-01 1.30075052e-01 -3.39310080e-01 -1.14928588e-01 4.97480184e-01 3.37848991e-01 -1.59278482e-01 3.96032155e-01 9.60613370e-01 -6.47587895e-01 -5.77203453e-01 -5.19639492e-01 4.36168164e-01 -5.23375988e-01 -5.89025877e-02 -4.89945531e-01 1.09690332e+00 2.39036888e-01 7.97862887e-01 4.64569516e-02 -2.80669570e-01 5.50956070e-01 -2.57530957e-01 3.63058209e-01 -7.52055287e-01 -5.11810720e-01 -6.21616319e-02 -1.46981508e-01 -7.68475652e-01 -6.32904530e-01 -5.11256754e-01 -7.41271496e-01 -4.87540439e-02 -8.69604707e-01 -2.29221627e-01 2.97131479e-01 6.77305639e-01 3.00543636e-01 6.64801359e-01 2.96808928e-01 -1.11622226e+00 -7.60910332e-01 -9.71785963e-01 -5.64829290e-01 3.83669555e-01 1.81383476e-01 -1.10002851e+00 -2.38119215e-01 -1.44916579e-01]
[8.880888938903809, -0.05310998857021332]
6fb5bb24-f00e-4a02-855e-873cb847748d
radformer-transformers-with-global-local
2211.04793
null
https://arxiv.org/abs/2211.04793v1
https://arxiv.org/pdf/2211.04793v1.pdf
RadFormer: Transformers with Global-Local Attention for Interpretable and Accurate Gallbladder Cancer Detection
We propose a novel deep neural network architecture to learn interpretable representation for medical image analysis. Our architecture generates a global attention for region of interest, and then learns bag of words style deep feature embeddings with local attention. The global, and local feature maps are combined using a contemporary transformer architecture for highly accurate Gallbladder Cancer (GBC) detection from Ultrasound (USG) images. Our experiments indicate that the detection accuracy of our model beats even human radiologists, and advocates its use as the second reader for GBC diagnosis. Bag of words embeddings allow our model to be probed for generating interpretable explanations for GBC detection consistent with the ones reported in medical literature. We show that the proposed model not only helps understand decisions of neural network models but also aids in discovery of new visual features relevant to the diagnosis of GBC. Source-code and model will be available at https://github.com/sbasu276/RadFormer
['Chetan Arora', 'Pankaj Gupta', 'Pratyaksha Rana', 'Mayank Gupta', 'Soumen Basu']
2022-11-09
null
null
null
null
['gallbladder-cancer-detection']
['computer-vision']
[-8.65302086e-02 6.36929750e-01 -8.79257098e-02 -3.71837765e-01 -9.48158324e-01 -2.69681841e-01 5.02075434e-01 4.17350054e-01 2.23273113e-02 3.89086306e-02 8.79011631e-01 -1.07521379e+00 -1.89558804e-01 -6.29796445e-01 -7.04847038e-01 -7.70631969e-01 -5.85102022e-01 4.32528377e-01 -3.44941527e-01 3.61565053e-02 1.17245279e-01 4.03792351e-01 -7.97038734e-01 5.95501006e-01 5.14703095e-01 9.38867807e-01 4.12092328e-01 1.10583365e+00 9.91274938e-02 9.95960891e-01 -2.60653496e-01 -2.57537872e-01 1.55381560e-01 -6.14256799e-01 -9.52777326e-01 -8.97544846e-02 1.01119317e-01 -4.71507430e-01 -4.39649552e-01 8.92304420e-01 5.43527067e-01 -3.04920346e-01 8.04858923e-01 -6.97548628e-01 -1.28692925e+00 6.05492651e-01 -2.75008887e-01 5.70352077e-01 1.03037871e-01 3.23747635e-01 8.55861664e-01 -9.87258673e-01 4.45414305e-01 1.12624252e+00 5.57521403e-01 8.37568283e-01 -7.33110011e-01 -4.83301520e-01 -9.83869806e-02 3.84248883e-01 -9.57165420e-01 -1.47484586e-01 4.09918785e-01 -3.27812701e-01 7.55909681e-01 4.54810351e-01 1.17456996e+00 1.26912606e+00 7.25552082e-01 7.64014542e-01 7.53134906e-01 -5.16322911e-01 -9.85894427e-02 -8.58016685e-02 7.34293982e-02 1.34619105e+00 6.04966640e-01 2.37942442e-01 -9.22021791e-02 -2.19829515e-01 1.14241588e+00 4.32161808e-01 -4.22379851e-01 -2.07280830e-01 -1.40234268e+00 1.20841086e+00 1.02444935e+00 2.32683539e-01 -5.16482174e-01 7.71319389e-01 2.34644949e-01 1.05571365e-02 1.94948897e-01 7.00830281e-01 -2.69080520e-01 2.86871582e-01 -1.73321128e-01 -2.16195703e-01 5.11245668e-01 7.31606185e-01 -2.87503004e-02 -5.81660122e-02 -7.57743418e-03 4.37290579e-01 6.85331821e-01 4.25697803e-01 8.68459940e-01 -8.72386456e-01 -2.17131376e-01 5.91025770e-01 -2.02040613e-01 -1.13916385e+00 -5.56690872e-01 -6.08479500e-01 -8.53220761e-01 1.95256341e-02 1.41279578e-01 2.14873448e-01 -1.35391605e+00 1.22653854e+00 -7.26295337e-02 2.25684524e-01 1.74020231e-02 1.15066767e+00 1.24058402e+00 2.57637471e-01 3.04340333e-01 4.39905286e-01 2.15099359e+00 -8.22111785e-01 -4.93919611e-01 -2.53085911e-01 7.85547912e-01 -4.98452812e-01 7.11806059e-01 1.28770575e-01 -9.01675045e-01 -9.88143757e-02 -8.72442424e-01 -5.97964466e-01 -2.14163780e-01 8.67322981e-02 9.19014215e-01 3.64396572e-01 -1.38180482e+00 4.92856950e-02 -1.28620863e+00 -5.35567343e-01 7.03608811e-01 2.71131009e-01 -3.23570132e-01 -2.10329577e-01 -1.09497583e+00 1.08497632e+00 3.97440605e-02 1.74132183e-01 -1.16708517e+00 -5.50978243e-01 -1.15881693e+00 2.59953707e-01 -2.69789323e-02 -1.07902968e+00 1.33010304e+00 -6.75125539e-01 -9.24487591e-01 1.13462472e+00 -1.36258498e-01 -8.56458664e-01 1.08447053e-01 9.10871848e-02 -1.95131823e-01 6.34636819e-01 7.08956569e-02 9.21073258e-01 7.61858404e-01 -1.25195920e+00 -5.82733214e-01 -3.07021648e-01 1.00505076e-01 -8.53538886e-02 1.01008348e-01 -3.72920513e-01 -3.42866063e-01 -9.16166067e-01 4.26174074e-01 -7.84086466e-01 -6.72086298e-01 3.77151847e-01 -2.70610720e-01 -2.30638400e-01 4.51183707e-01 -8.12702477e-01 6.62171304e-01 -2.13406801e+00 -1.74303815e-01 1.96544066e-01 8.98340285e-01 -1.20210303e-02 -2.14759171e-01 -2.00299099e-02 -4.18984383e-01 5.68603933e-01 1.29776970e-01 -6.28754795e-02 -3.27844232e-01 2.58248091e-01 -9.16607007e-02 6.34732008e-01 6.17522478e-01 1.36295879e+00 -1.17858803e+00 -5.12133121e-01 5.00782549e-01 6.32069170e-01 -5.33874929e-01 7.12526590e-02 1.41783819e-01 3.98971409e-01 -7.26807415e-01 8.75485122e-01 2.85987854e-01 -1.02901256e+00 3.43831033e-01 -3.34569365e-01 4.69418913e-01 3.52613449e-01 -3.24678063e-01 1.53499556e+00 -5.22785902e-01 7.37599432e-01 -1.61043391e-01 -1.06672215e+00 6.22185886e-01 5.70603669e-01 1.73612833e-01 -7.40359962e-01 4.03221607e-01 2.42332652e-01 3.37090462e-01 -7.59597719e-01 -3.73945385e-02 -2.13413239e-01 7.96027929e-02 5.49408257e-01 2.01345369e-01 7.93791786e-02 -2.66398877e-01 4.67390627e-01 1.10865319e+00 -3.53990465e-01 8.91711056e-01 -5.83588600e-01 2.08089277e-01 6.45959377e-02 2.13678762e-01 9.42418993e-01 -3.16840440e-01 6.66968524e-01 5.95726490e-01 -1.13050568e+00 -9.03865874e-01 -1.07179165e+00 -2.34239101e-01 5.36799431e-01 2.96937346e-01 -1.87084556e-01 -1.73946500e-01 -6.88965023e-01 5.63565679e-02 6.60878241e-01 -1.33239293e+00 -1.70389533e-01 -5.81852853e-01 -4.90993470e-01 3.42588037e-01 7.90178955e-01 -1.84272483e-01 -9.50136185e-01 -8.71413887e-01 1.32308155e-01 -1.79747343e-01 -5.17296374e-01 -4.02375937e-01 3.39659810e-01 -9.13479507e-01 -1.57749593e+00 -8.71159017e-01 -1.10266006e+00 1.12230504e+00 1.82436153e-01 1.23693335e+00 7.08400249e-01 -1.03754580e+00 7.46915996e-01 -4.20434028e-01 -5.97639441e-01 -6.98698997e-01 -3.09836686e-01 -2.16313422e-01 -4.86260027e-01 5.29014826e-01 -9.80267525e-02 -1.39797163e+00 -8.42073187e-02 -6.55925751e-01 4.26556677e-01 7.79209316e-01 1.14847362e+00 4.05519813e-01 -8.39126885e-01 2.72894949e-01 -7.07619011e-01 6.47885978e-01 -7.84551442e-01 -1.54575378e-01 5.40754711e-03 -5.32423258e-01 1.57635942e-01 2.79113740e-01 -1.48740605e-01 -6.29180491e-01 -2.44335249e-01 -3.43064457e-04 -4.48607057e-01 -9.71423183e-03 5.29562235e-01 7.93008089e-01 4.14556265e-03 8.00502777e-01 2.73084253e-01 3.00758362e-01 -2.64062643e-01 2.57197261e-01 4.52519655e-01 4.29729372e-01 -1.51056662e-01 2.80329466e-01 5.52670002e-01 -1.11084335e-01 -3.39452267e-01 -4.79724407e-01 -3.85149091e-01 8.93553067e-03 2.68786997e-02 8.76991928e-01 -9.95176494e-01 -7.71416843e-01 -2.82960802e-01 -1.05179489e+00 -3.61586027e-02 -2.42599085e-01 6.52937710e-01 -3.53878736e-01 1.01572849e-01 -9.72347975e-01 -4.69766021e-01 -5.19096315e-01 -1.47041082e+00 1.14278305e+00 1.64350092e-01 -3.97827893e-01 -1.30824363e+00 -2.85239339e-01 1.98855579e-01 4.25913900e-01 2.40924165e-01 1.21499979e+00 -8.27133358e-01 -9.59386826e-01 -2.47703806e-01 -5.41325271e-01 -2.01065853e-01 3.85355741e-01 -3.01989108e-01 -8.48757029e-01 -2.16419980e-01 3.22883762e-02 -5.01284786e-02 1.14165187e+00 9.30403709e-01 1.63722742e+00 -4.17057365e-01 -5.23680866e-01 1.04139769e+00 1.26382017e+00 3.64674002e-01 4.83206898e-01 2.59160846e-01 3.74243021e-01 6.46447316e-02 1.75360087e-02 3.35451871e-01 5.80491364e-01 -1.49256885e-01 9.27979648e-01 -5.44105113e-01 -2.59831756e-01 -1.33284792e-01 -2.72392601e-01 5.43503344e-01 -2.79885739e-01 -2.34938741e-01 -1.29608071e+00 1.03232622e+00 -1.41439760e+00 -5.30410945e-01 3.93357687e-02 1.59469831e+00 5.69241226e-01 -3.86372477e-01 -5.60568333e-01 -5.31475604e-01 3.20001036e-01 1.91396009e-02 -2.65167236e-01 -5.72451890e-01 3.35912466e-01 2.85129130e-01 6.89250469e-01 5.49555719e-01 -1.11252213e+00 3.85454327e-01 6.49810076e+00 2.51994550e-01 -1.28533685e+00 2.06924245e-01 1.22262001e+00 1.01001225e-01 -6.48998797e-01 -3.44754577e-01 -3.12407110e-02 2.12957188e-01 7.33839989e-01 -1.63687304e-01 1.03613753e-02 6.99224234e-01 1.63932636e-01 1.69085622e-01 -1.17171431e+00 1.08320200e+00 2.57990390e-01 -1.93959844e+00 3.07783186e-01 1.37196377e-01 3.67112279e-01 8.22928995e-02 3.77643079e-01 1.08657956e-01 4.84656125e-01 -1.47220898e+00 2.30873629e-01 5.18444955e-01 9.57995176e-01 -2.09668174e-01 1.03613007e+00 -3.20242435e-01 -8.67089570e-01 -7.01900423e-02 -8.55353102e-02 2.10136905e-01 -3.00656438e-01 1.20093105e-02 -1.61579323e+00 1.61776245e-01 6.77730739e-01 6.37747288e-01 -7.36450672e-01 9.43426788e-01 -2.63332337e-01 6.15178943e-01 -5.88775687e-02 -9.86733437e-02 4.84434724e-01 4.21666473e-01 4.67525482e-01 1.26919127e+00 5.73393822e-01 5.07177353e-01 -1.39604256e-01 9.82267141e-01 -2.33528391e-02 9.90921333e-02 -6.60405934e-01 -1.70889869e-01 2.62301713e-01 1.17206657e+00 -8.45023394e-01 -4.76368904e-01 -2.86439180e-01 9.16769683e-01 2.29702052e-02 3.13076973e-01 -5.85070431e-01 -1.11898176e-01 3.68751377e-01 8.60931575e-02 2.21634865e-01 2.11615369e-01 -4.41052407e-01 -1.11939967e+00 -2.64167368e-01 -6.98602140e-01 5.38701952e-01 -8.09652567e-01 -1.13767862e+00 5.93421102e-01 -4.01903033e-01 -1.45892429e+00 -4.99148935e-01 -1.06229365e+00 -7.80035675e-01 7.02434242e-01 -1.55297768e+00 -1.40275633e+00 -4.14205074e-01 2.39937320e-01 6.24718070e-01 -1.44628704e-01 1.37672102e+00 -2.74253458e-01 1.00938596e-01 4.08539057e-01 -1.15942292e-01 5.02302706e-01 4.59627509e-01 -1.49521196e+00 4.64269310e-01 4.96071130e-01 -1.31479368e-01 1.06731915e+00 5.73970139e-01 -4.11165804e-01 -1.42051995e+00 -1.06802547e+00 7.17952907e-01 -4.84186321e-01 6.48284256e-01 4.03723381e-02 -6.93035603e-01 9.42155600e-01 2.42195547e-01 5.21303415e-01 9.74283457e-01 -8.35747346e-02 -2.28234336e-01 3.05929869e-01 -1.05829203e+00 5.78461170e-01 5.42332709e-01 -2.72513777e-01 -7.62161732e-01 4.26794976e-01 6.57056987e-01 -6.47636712e-01 -8.37288439e-01 8.30011293e-02 7.01206028e-01 -6.05599284e-01 1.14576530e+00 -8.69381487e-01 7.48532772e-01 1.21478236e-03 8.29850286e-02 -1.31331718e+00 -6.75379455e-01 -1.49539307e-01 3.72799076e-02 -1.46976680e-01 5.49013615e-01 -7.19815075e-01 4.64680761e-01 3.92876863e-01 -3.19378793e-01 -1.06797409e+00 -7.10160017e-01 3.07579488e-02 1.22958839e-01 -2.58109242e-01 4.37555552e-01 1.11349642e+00 1.99939847e-01 -3.25245619e-01 5.98476343e-02 5.05245268e-01 6.48431838e-01 1.75702646e-01 1.04585268e-01 -9.51406181e-01 -1.83521330e-01 -3.54234457e-01 -7.44321585e-01 -7.85801649e-01 -5.77270031e-01 -1.22057927e+00 -1.89250469e-01 -2.00119948e+00 4.01198119e-01 -2.67707020e-01 -5.42700350e-01 7.04854488e-01 -2.11428985e-01 4.73946959e-01 1.83549106e-01 -3.77081782e-02 -2.04164460e-01 6.21475093e-02 1.48322773e+00 -3.38971466e-01 1.97603300e-01 -3.01492810e-01 -1.31722748e+00 7.91155994e-01 6.87865913e-01 -3.73689860e-01 -7.88520947e-02 -7.45797932e-01 1.84039906e-01 3.84657383e-01 9.65749502e-01 -3.52200657e-01 1.24010056e-01 -6.30960241e-02 8.92176330e-01 -2.05409810e-01 6.53800443e-02 -7.28800297e-01 -2.02904344e-01 9.65344429e-01 -3.44633788e-01 -2.29973271e-01 1.97408482e-01 6.14028752e-01 -9.90207419e-02 -4.03122008e-02 5.53054333e-01 -6.39824688e-01 -7.49643862e-01 2.00921953e-01 -5.91262698e-01 -2.90738225e-01 8.28706920e-01 -1.47012711e-01 -3.91647249e-01 -6.24256909e-01 -1.09229434e+00 3.69939893e-01 9.94783416e-02 4.57547903e-01 1.20381558e+00 -1.23259044e+00 -1.00349247e+00 4.90620732e-01 3.31198782e-01 1.80541113e-01 1.89752206e-01 1.03561556e+00 -1.09981906e+00 6.66531801e-01 4.82770018e-02 -6.76340818e-01 -1.11615360e+00 3.25388759e-01 6.63176060e-01 -1.43915251e-01 -1.07572854e+00 1.08267689e+00 7.07709551e-01 -9.83859375e-02 7.85815790e-02 -9.86451685e-01 -1.10161126e-01 -4.99087125e-01 6.57383025e-01 -3.45318556e-01 -1.97205022e-01 -1.32704765e-01 -4.35963035e-01 1.61346480e-01 -4.35186386e-01 5.17406166e-01 1.41697311e+00 -5.16549982e-02 5.12624718e-02 -4.38955933e-04 1.12590230e+00 -3.28878582e-01 -6.40820742e-01 -1.85869887e-01 -3.34978014e-01 -4.49578047e-01 1.77677199e-01 -1.00822914e+00 -1.17874873e+00 9.43768322e-01 1.01950455e+00 2.63888910e-02 1.05772221e+00 5.53998709e-01 4.89965916e-01 2.85606861e-01 9.25681591e-02 -1.71433330e-01 8.14534798e-02 1.38023868e-01 1.14380836e+00 -1.60668218e+00 -1.72005594e-02 1.21827237e-02 -5.56647718e-01 1.47316122e+00 2.88853616e-01 -5.03337860e-01 5.70014179e-01 2.40844563e-01 4.26188260e-01 -7.23025143e-01 -8.57463658e-01 8.53604749e-02 3.42591286e-01 5.96878469e-01 6.15328550e-01 5.20953119e-01 -1.25283882e-01 7.50585854e-01 3.31648928e-03 -1.26910082e-03 6.53342009e-01 5.79344034e-01 -3.05937290e-01 -6.18763924e-01 -3.59306306e-01 8.67344856e-01 -6.33940399e-01 -4.67278093e-01 4.54761088e-02 9.67911243e-01 1.57589074e-02 3.35557103e-01 4.77969259e-01 7.83093199e-02 -2.25990608e-01 -7.16689602e-02 3.74715656e-01 -5.93222380e-01 -2.08961606e-01 3.18963885e-01 -1.12585604e-01 -5.81628859e-01 -1.15168683e-01 -3.42464924e-01 -1.41676760e+00 2.06034020e-01 -8.87895599e-02 -9.78171360e-03 4.97722536e-01 5.90842962e-01 4.39026743e-01 8.12617779e-01 3.20079237e-01 -5.81582606e-01 -2.85983711e-01 -1.03414762e+00 -3.41200918e-01 5.31950653e-01 9.84650075e-01 -4.26875591e-01 -3.38537723e-01 4.28021431e-01]
[15.078195571899414, -2.4218318462371826]
4fcaede7-0ec4-4ed0-91c0-62462208a078
emergence-in-artificial-life
2105.03216
null
https://arxiv.org/abs/2105.03216v2
https://arxiv.org/pdf/2105.03216v2.pdf
Emergence in artificial life
Even when concepts similar to emergence have been used since antiquity, we lack an agreed definition. However, emergence has been identified as one of the main features of complex systems. Most would agree on the statement ``life is complex''. Thus, understanding emergence and complexity should benefit the study of living systems. It can be said that life emerges from the interactions of complex molecules. But how useful is this to understand living systems? Artificial life (ALife) has been developed in recent decades to study life using a synthetic approach: build it to understand it. ALife systems are not so complex, be them soft (simulations), hard (robots), or wet (protocells). Then, we can aim at first understanding emergence in ALife, for then using this knowledge in biology. I argue that to understand emergence and life, it becomes useful to use information as a framework. In a general sense, I define emergence as information that is not present at one scale but is present at another scale. This perspective avoids problems of studying emergence from a materialist framework, and can be also useful in the study of self-organization and complexity.
['Carlos Gershenson']
2021-04-30
null
null
null
null
['artificial-life']
['miscellaneous']
[-1.08553199e-02 1.42069399e-01 3.32297921e-01 3.07389498e-01 6.85271025e-01 -7.32564747e-01 9.05795336e-01 5.75058639e-01 -1.54323697e-01 6.90876126e-01 1.07143499e-01 -3.04170966e-01 -3.56782615e-01 -1.03816819e+00 -3.88173193e-01 -1.13139045e+00 -1.35613561e-01 2.33229563e-01 1.41390875e-01 -9.06174541e-01 1.55265138e-01 4.04797584e-01 -1.84445453e+00 -3.42331082e-01 9.94464219e-01 2.96294838e-01 4.18595731e-01 5.43325007e-01 -2.63271760e-02 6.00996971e-01 -5.63949645e-01 -7.14337900e-02 1.18535578e-01 -1.02164531e+00 -7.32400835e-01 -2.06456706e-01 -6.91301465e-01 4.68181133e-01 1.08678214e-01 6.75573587e-01 1.42272949e-01 -3.56485248e-01 5.76224446e-01 -9.19008434e-01 -7.01545775e-01 4.79862064e-01 2.35277843e-02 -1.21100813e-01 4.03541774e-01 1.81434304e-01 6.49937510e-01 -4.30264205e-01 8.36046457e-01 1.10678983e+00 4.06194359e-01 4.54545826e-01 -1.22786844e+00 -1.17103405e-01 -3.24050635e-02 -1.85782146e-02 -1.12474966e+00 -3.01422864e-01 3.96404326e-01 -7.54300177e-01 6.31102085e-01 2.74857283e-01 1.50786579e+00 7.86829650e-01 8.23691189e-01 2.04174519e-01 1.44025517e+00 -7.45345175e-01 4.74459738e-01 -3.26692276e-02 1.39755309e-01 3.68677139e-01 9.92841363e-01 3.07828486e-01 -6.36626959e-01 2.16332197e-01 6.95828259e-01 -1.19569145e-01 -3.49540353e-01 1.02553748e-01 -1.28014517e+00 5.13141572e-01 3.30631495e-01 1.11091256e+00 -3.24051023e-01 1.53598726e-01 -8.87587368e-02 5.38406968e-01 8.25114101e-02 1.07752025e+00 -4.16477114e-01 -4.51362461e-01 -3.34634334e-01 -8.06844682e-02 1.13193643e+00 2.65000165e-01 7.11238444e-01 -1.61095679e-01 7.47673929e-01 4.01986033e-01 2.57902682e-01 5.26019156e-01 4.99579042e-01 -8.30362380e-01 -5.26173651e-01 9.92859066e-01 -5.27544394e-02 -8.00406814e-01 -5.70544183e-01 -4.24153656e-01 -9.54242051e-01 5.16857147e-01 5.57789326e-01 -8.90925825e-02 -5.48698664e-01 1.86791968e+00 2.60519177e-01 -4.13423896e-01 3.42965961e-01 4.76280481e-01 6.58175349e-01 8.00922215e-01 -5.49390018e-01 -4.97903377e-01 1.16500533e+00 -3.85661691e-01 -5.21838903e-01 1.09447367e-01 3.50052178e-01 -5.15035450e-01 6.83421016e-01 6.34957314e-01 -8.28084350e-01 -1.86860651e-01 -1.14102781e+00 3.02979022e-01 -9.16496217e-01 -6.03288054e-01 9.39608872e-01 7.46396184e-01 -1.25447536e+00 6.39192343e-01 -9.65081573e-01 -1.25076199e+00 -2.03031749e-01 1.88486844e-01 -2.59545535e-01 4.24198031e-01 -8.92258883e-01 1.23001301e+00 2.47682303e-01 -1.20985135e-02 -5.06642401e-01 -1.78379178e-01 -4.11423177e-01 -6.94682300e-02 2.73543328e-01 -1.03001678e+00 5.97314000e-01 -1.19292486e+00 -1.50913620e+00 7.42144108e-01 -7.29839727e-02 -1.59212589e-01 1.35275111e-01 1.05948150e-01 1.05673661e-02 -3.80601846e-02 -2.83540357e-02 2.16554344e-01 2.53969193e-01 -1.59533346e+00 -2.06585899e-01 -4.39183712e-01 3.72977912e-01 5.24401627e-02 -8.16543773e-02 -1.07395895e-01 4.59359229e-01 -2.91762322e-01 4.96305913e-01 -1.05559814e+00 -2.60424942e-01 -1.88198105e-01 -5.81942573e-02 -3.23807687e-01 3.48293334e-01 -2.43087143e-01 9.43216264e-01 -1.95056403e+00 4.24127996e-01 4.93563414e-02 5.67833364e-01 5.87876029e-02 1.99699819e-01 1.25327790e+00 2.16178343e-01 4.88953829e-01 -4.04036283e-01 3.00559938e-01 -6.25656173e-02 3.80459070e-01 -3.57155763e-02 1.63986340e-01 -2.06401758e-02 8.34474504e-01 -1.00948274e+00 -7.52862841e-02 1.74446225e-01 6.14947140e-01 -2.93038011e-01 -2.07704157e-01 4.45807837e-02 7.31256902e-01 -2.13232100e-01 6.47376597e-01 3.27230811e-01 -3.73763770e-01 3.63600522e-01 5.04957438e-01 -7.92376578e-01 -3.36910016e-03 -6.96480870e-01 1.16905224e+00 -4.27096814e-01 8.29148293e-01 8.15016404e-03 -9.66907024e-01 8.64520311e-01 3.05887341e-01 5.94062686e-01 -5.04510701e-01 1.35432750e-01 3.79799366e-01 8.37975681e-01 -5.53660393e-01 -1.39878809e-01 -2.63974637e-01 3.47348824e-02 5.26027143e-01 -2.16007367e-01 -6.15424514e-01 3.59496474e-01 -4.81176935e-02 1.42261696e+00 8.11108723e-02 6.63784444e-01 -6.86945200e-01 4.42889720e-01 -7.44002163e-02 5.22553265e-01 5.15902460e-01 3.99316326e-02 1.38564110e-01 3.80758643e-01 -6.55860543e-01 -9.27807033e-01 -1.33565080e+00 -4.59077328e-01 5.63878417e-01 4.63611662e-01 -5.94025970e-01 -7.96301544e-01 1.70090050e-01 -2.81922758e-01 3.94017696e-01 -8.61110687e-01 -1.73376948e-01 -4.53534663e-01 -1.04688776e+00 6.56152144e-02 -2.98088044e-01 4.59402919e-01 -1.37452137e+00 -1.33072603e+00 2.59965658e-01 4.92596775e-02 -6.18089020e-01 6.57660067e-01 4.51075464e-01 -6.37956083e-01 -9.89763379e-01 -4.54387575e-01 -5.40560842e-01 6.91470265e-01 2.90049464e-01 1.09957862e+00 7.78268754e-01 -4.48105842e-01 4.61230069e-01 -8.21206987e-01 -5.38629949e-01 -8.18204641e-01 4.48920391e-02 4.42459971e-01 -3.51254523e-01 -1.28121814e-02 -1.04258251e+00 -6.16941631e-01 4.57162373e-02 -1.24178445e+00 2.80854702e-01 5.74668348e-01 6.66025817e-01 -4.57072333e-02 2.10600048e-01 8.84377182e-01 -3.01936090e-01 6.57521069e-01 -3.15515518e-01 -1.90304041e-01 4.93428528e-01 -6.93475485e-01 1.15460576e-02 8.08492482e-01 -3.66098434e-02 -4.86506552e-01 -2.70988941e-01 -5.69148436e-02 9.24394786e-01 -2.97233224e-01 6.06531799e-01 -1.25665143e-01 4.21980061e-02 4.92156059e-01 4.52813923e-01 9.38491523e-02 -7.64027685e-02 1.34541392e-01 6.07697189e-01 -1.28852591e-01 -7.01994181e-01 7.72916496e-01 6.71361506e-01 3.10281366e-01 -1.62185812e+00 -2.17723012e-01 2.59191453e-01 -7.61740446e-01 -4.57261682e-01 8.75975549e-01 -2.05906838e-01 -8.47776771e-01 4.86998290e-01 -9.77226019e-01 -4.20508206e-01 -6.70128644e-01 3.38165998e-01 -5.39009392e-01 2.87141260e-02 -2.40796909e-01 -9.96355176e-01 -3.29797976e-02 -8.58943462e-01 5.24201572e-01 5.73042810e-01 -4.41925734e-01 -1.17479742e+00 5.74237347e-01 -1.11632302e-01 7.20295727e-01 6.88989639e-01 9.57905114e-01 -1.55084625e-01 -6.09895647e-01 1.87569574e-01 3.68063360e-01 -1.00670405e-01 7.66839147e-01 6.62397206e-01 -6.00703359e-01 -1.60689786e-01 3.47685277e-01 1.15785964e-01 5.99276304e-01 2.73417056e-01 1.59006417e-01 -3.20836723e-01 -4.67939019e-01 2.91019529e-01 1.58506894e+00 6.89416528e-01 6.57395542e-01 5.57655931e-01 5.07455356e-02 1.30189812e+00 1.23026490e-01 3.17451328e-01 5.91537058e-01 3.63677531e-01 2.97090441e-01 7.75141120e-02 1.69826433e-01 1.59606412e-01 5.15569568e-01 1.34473050e+00 -4.58645046e-01 -1.55821621e-01 -1.25941479e+00 4.64172423e-01 -1.78035736e+00 -1.02083659e+00 -1.72724277e-01 2.14505529e+00 6.67298973e-01 -1.34050464e-02 2.28801876e-01 1.03232361e-01 4.84009981e-01 -3.84965748e-01 -3.67309988e-01 -5.93678296e-01 -6.38500452e-01 1.57635868e-01 -2.18278468e-01 6.77405715e-01 -4.64527607e-01 8.87620628e-01 6.74810457e+00 1.64120182e-01 -1.34414136e+00 -1.24254689e-01 3.55642170e-01 2.03059196e-01 -3.84068578e-01 6.55851722e-01 -3.31912488e-01 6.12415195e-01 7.11298764e-01 -4.89058614e-01 6.26918256e-01 1.64992437e-01 2.74229765e-01 -5.76129854e-01 -9.26093996e-01 7.70099521e-01 -1.96087942e-01 -1.06562901e+00 -8.27709436e-02 3.68410945e-01 5.80548882e-01 -2.69934356e-01 -2.20029250e-01 -1.96049720e-01 5.56726098e-01 -1.21987712e+00 5.68223894e-01 7.21761167e-01 1.88889965e-01 -4.09470916e-01 6.29636586e-01 7.35667109e-01 -1.16718125e+00 2.17810515e-02 -2.71766901e-01 -8.40503693e-01 2.88781673e-01 8.23799729e-01 -3.22130501e-01 4.64925021e-01 6.60288334e-01 7.18714833e-01 -7.26672649e-01 9.63181019e-01 -2.22533390e-01 3.34268749e-01 -5.80320418e-01 -4.12761897e-01 -1.73364565e-01 -7.71225274e-01 6.01716936e-01 5.99612176e-01 5.01472592e-01 2.20960647e-01 -3.53291094e-01 9.22837019e-01 7.41933584e-01 7.52325729e-02 -1.10126340e+00 -3.49477053e-01 3.20083231e-01 1.27909195e+00 -1.41390038e+00 -1.69146910e-01 -1.64136499e-01 6.28870249e-01 -2.41913915e-01 3.11594773e-02 -4.65442032e-01 -2.62678593e-01 5.95837712e-01 2.52275705e-01 -1.41828373e-01 -6.01707578e-01 -2.54836559e-01 -1.23467493e+00 -2.78992146e-01 -7.05441594e-01 -4.49354351e-01 -3.80220443e-01 -1.14262772e+00 6.13073349e-01 -1.30531445e-01 -8.79308701e-01 -8.91214237e-02 -4.55132931e-01 -5.75669706e-01 7.09994361e-02 -8.69450986e-01 -1.01803064e+00 -4.43937629e-01 -6.29875660e-02 1.30738527e-01 1.97743595e-01 8.12175870e-01 -4.15625215e-01 -5.62132061e-01 -1.81470841e-01 4.06288952e-01 -3.78273427e-01 2.15547532e-01 -1.33462930e+00 2.76309878e-01 7.56850481e-01 1.15697488e-01 9.01873887e-01 9.06145096e-01 -5.72855353e-01 -1.44069719e+00 -1.37611240e-01 7.89261281e-01 -5.74957013e-01 5.43000281e-01 -5.17735481e-01 -5.01511812e-01 1.79730654e-01 6.72789514e-01 -7.60009646e-01 8.86358500e-01 3.56782153e-02 1.22210473e-01 -1.66569222e-02 -8.79994988e-01 9.89150465e-01 1.19527817e+00 -3.13050374e-02 -7.05862343e-01 1.61422193e-01 7.15659440e-01 5.15985072e-01 -9.74057853e-01 4.05326992e-01 1.07542253e+00 -1.30441558e+00 6.63840950e-01 -8.23480785e-02 2.61083782e-01 -5.69003105e-01 3.43910009e-02 -1.31856191e+00 -3.41723233e-01 -8.65399063e-01 3.69577289e-01 1.25844610e+00 3.11897516e-01 -1.34967351e+00 1.09822072e-01 2.60232478e-01 1.16227292e-01 -7.38405287e-01 -8.43438029e-01 -1.17725360e+00 5.50172150e-01 2.95942515e-01 4.42868978e-01 1.10189915e+00 7.04711676e-01 4.58769530e-01 2.89503902e-01 -3.57023329e-01 3.15047383e-01 1.71852708e-01 8.39246690e-01 -1.81545877e+00 -2.48431921e-01 -6.36597693e-01 -4.41321015e-01 -5.32891572e-01 -2.35306725e-01 -6.14056408e-01 7.99386203e-02 -1.73935521e+00 1.11018859e-01 -4.22868639e-01 -5.73735647e-02 1.86622515e-02 5.22298515e-02 -7.09830970e-02 4.17659998e-01 4.91370082e-01 -2.00029150e-01 5.11131585e-01 1.14937663e+00 4.38304096e-01 -3.33782703e-01 -4.50090528e-01 -8.66536915e-01 8.31973553e-01 1.08486319e+00 -4.35122490e-01 -1.58132017e-01 1.52827337e-01 1.00379276e+00 -2.36197114e-01 2.01837555e-01 -1.34465647e+00 1.77434579e-01 -4.32321101e-01 -5.95819987e-02 -3.47386561e-02 2.56043881e-01 -7.58420885e-01 5.69407225e-01 1.15137839e+00 2.32619405e-01 1.83432922e-01 -2.08909959e-01 1.82913303e-01 4.06680591e-02 -2.85460442e-01 6.86422110e-01 -5.14512360e-01 -2.02050600e-02 -2.88516372e-01 -1.09119666e+00 -1.86897084e-01 1.29086149e+00 -4.59511966e-01 -7.09082663e-01 -3.65558296e-01 -6.84549034e-01 -2.29874421e-02 1.47987843e+00 1.58354416e-02 4.35755402e-01 -8.16629946e-01 -5.06844223e-01 -7.64140859e-02 -1.63705703e-02 -9.74682271e-02 -2.32963368e-01 7.81290412e-01 -1.17731738e+00 3.83887887e-01 -5.01559258e-01 -5.54868937e-01 -8.38313818e-01 5.91105998e-01 2.75755078e-01 2.18843371e-02 -4.82650727e-01 5.40408075e-01 6.19659424e-01 -2.06259429e-01 -5.99261045e-01 -3.61010522e-01 -1.77643672e-01 1.25217065e-01 3.45011234e-01 2.04996139e-01 -5.83547533e-01 -6.29924238e-01 -5.02953887e-01 1.13142574e+00 4.76203233e-01 -3.59838277e-01 1.60147250e+00 -3.63662988e-01 -9.88947868e-01 1.13756657e+00 3.93356979e-01 3.17203403e-01 -6.26565695e-01 5.01482546e-01 1.72549337e-01 -4.21017185e-02 -6.89149499e-01 -6.27975941e-01 -2.95461684e-01 7.23221660e-01 2.62882024e-01 1.22008741e+00 1.17828739e+00 2.83198714e-01 3.44414622e-01 4.81984109e-01 6.95896506e-01 -8.16033304e-01 2.20684886e-01 5.20960927e-01 9.73012030e-01 -9.32100832e-01 -1.50568306e-01 -4.28198814e-01 -5.01684397e-02 1.26558399e+00 3.79989564e-01 -2.49265730e-01 9.08652246e-01 3.60763103e-01 -1.25453398e-01 -2.32112944e-01 -1.10781324e+00 -6.79871976e-01 -3.07011038e-01 6.61343575e-01 7.17424095e-01 1.20204002e-01 -8.81747603e-01 2.09146768e-01 -6.00200772e-01 -1.84501648e-01 9.36940312e-01 1.05294633e+00 -9.59640801e-01 -1.52328694e+00 -4.00024295e-01 1.08250365e-01 -1.95807442e-01 -1.77143663e-02 -1.10864174e+00 7.58907259e-01 7.02291608e-01 1.15802610e+00 -2.15937465e-01 -5.02124965e-01 -3.37426394e-01 -3.51215377e-02 6.85344696e-01 -6.97449982e-01 -6.58506095e-01 -3.02537590e-01 -2.81669319e-01 -4.04124446e-02 -6.93785906e-01 -7.24532902e-01 -1.15105772e+00 -5.81615806e-01 -3.62315953e-01 5.61392784e-01 7.00798392e-01 9.94342089e-01 2.65777826e-01 4.48190153e-01 4.59166795e-01 -5.67752004e-01 3.52399468e-01 -7.34096587e-01 -5.72007775e-01 6.70729131e-02 2.52764046e-01 -5.69083035e-01 -6.82443619e-01 9.70000178e-02]
[5.603518962860107, 4.181462287902832]
72c6d11c-f6e2-4618-9d4c-fe34c334f1b9
bl-research-at-semeval-2022-task-1-deep
null
null
https://aclanthology.org/2022.semeval-1.11
https://aclanthology.org/2022.semeval-1.11.pdf
BL.Research at SemEval-2022 Task 1: Deep networks for Reverse Dictionary using embeddings and LSTM autoencoders
This paper describes our two deep learning systems that competed at SemEval-2022 Task 1 “CODWOE: Comparing Dictionaries and WOrd Embeddings”. We participated in the subtask for the reverse dictionary which consists in generating vectors from glosses. We use sequential models that integrate several neural networks, starting from Embeddings networks until the use of Dense networks, Bidirectional Long Short-Term Memory (BiLSTM) networks and LSTM networks. All glosses have been preprocessed in order to consider the best representation form of the meanings for all words that appears. We achieved very competitive results in reverse dictionary with a second position in English and French languages when using contextualized embeddings, and the same position for English, French and Spanish languages when using char embeddings.
['Youssef Miloudi', 'Christophe Bortolaso', 'Mokhtar Boumedyen Billami', 'Lina Nicolaieff', 'Julien Breton', 'Nihed Bendahman']
null
null
null
null
semeval-naacl-2022-7
['reverse-dictionary']
['natural-language-processing']
[-1.39757976e-01 1.11346776e-02 -1.46916181e-01 -2.39850208e-01 -3.65515381e-01 -4.90187436e-01 1.00118649e+00 1.02534458e-01 -1.31875718e+00 6.72214866e-01 5.54390252e-01 -6.70023620e-01 2.37623721e-01 -8.43583524e-01 -5.20493507e-01 -3.90689224e-01 9.42700207e-02 8.22674692e-01 -1.82624310e-01 -7.08934903e-01 4.28528376e-02 3.32783639e-01 -1.23247993e+00 6.33833468e-01 3.85322899e-01 7.80290782e-01 4.29986626e-01 5.61334848e-01 -3.75468761e-01 3.85314167e-01 -5.25900364e-01 -7.59437621e-01 2.67005801e-01 -1.76793665e-01 -6.52562618e-01 -6.11638963e-01 5.83353758e-01 -1.74354896e-01 -4.99940246e-01 1.01977181e+00 6.53295159e-01 1.91205934e-01 9.29129899e-01 -6.78604245e-01 -1.49473596e+00 1.24709034e+00 1.61962152e-01 4.21433240e-01 2.67914802e-01 1.48159951e-01 1.51991212e+00 -1.52168214e+00 7.80031204e-01 1.41578603e+00 7.71046281e-01 8.54595840e-01 -1.23668849e+00 -3.96246433e-01 2.37415269e-01 3.47652674e-01 -1.28019285e+00 -2.78093427e-01 1.61300153e-01 -3.11214745e-01 2.06139755e+00 -5.58013916e-02 7.59586334e-01 1.70939124e+00 4.17076021e-01 5.64691961e-01 1.00485420e+00 -7.90680528e-01 -1.84976608e-01 1.51737377e-01 4.64862436e-01 4.65495974e-01 3.77015501e-01 3.04904312e-01 -1.48450539e-01 2.62601525e-01 5.50919831e-01 9.40466747e-02 -5.53015135e-02 2.01922804e-01 -1.45589316e+00 9.90739763e-01 5.86440086e-01 1.00668621e+00 -6.35615945e-01 3.47597599e-01 6.62953198e-01 7.31406569e-01 1.37066424e-01 8.44780385e-01 -9.19156015e-01 1.17097780e-01 -6.59242451e-01 1.35187402e-01 6.31730318e-01 8.69867384e-01 4.89854276e-01 3.52505058e-01 -1.25657991e-01 1.16622186e+00 3.34845960e-01 6.66432977e-01 1.14472795e+00 -1.20634764e-01 4.80053186e-01 5.44740967e-02 -2.74768267e-02 -6.07159793e-01 -3.72406363e-01 -1.73468783e-01 -6.77297652e-01 -3.22200656e-02 1.84896290e-01 -4.17125404e-01 -1.36937797e+00 1.73734713e+00 -5.88776410e-01 -2.07388952e-01 3.19546580e-01 5.88633418e-01 1.05548978e+00 8.26892555e-01 3.83885860e-01 2.64763087e-01 1.50710297e+00 -1.14489651e+00 -8.18185031e-01 -5.05034983e-01 6.49155319e-01 -7.52652168e-01 1.24505556e+00 4.79833543e-01 -1.21386349e+00 -1.03507364e+00 -1.23860538e+00 -5.00718713e-01 -1.29858077e+00 1.05109466e-02 8.20936169e-03 3.61243278e-01 -1.47766638e+00 7.05242038e-01 -5.36271214e-01 -6.85196042e-01 -2.35123321e-01 3.69400650e-01 -4.14240897e-01 4.60535809e-02 -1.88734591e+00 1.64247870e+00 8.27221036e-01 2.51891613e-01 -9.95174289e-01 -3.04804862e-01 -1.15957344e+00 -1.94097888e-02 -3.13757151e-01 -6.47745550e-01 1.06404781e+00 -1.07894158e+00 -1.29041076e+00 1.18272984e+00 -2.14330815e-02 -8.89217854e-01 -9.20884386e-02 -3.98345917e-01 -9.15779591e-01 -4.63092715e-01 -4.28899489e-02 9.56963360e-01 6.07858419e-01 -9.43339884e-01 -5.45414984e-01 -3.28040197e-02 1.35861903e-01 5.51791079e-02 -4.89440441e-01 -6.89340755e-02 -1.80542514e-01 -9.99410212e-01 -5.57881594e-01 -8.57659161e-01 -3.31592053e-01 -5.45014679e-01 -1.43590346e-01 -5.42274237e-01 2.81239539e-01 -8.40236366e-01 1.46617258e+00 -2.09493208e+00 4.72375542e-01 3.52261737e-02 -1.97575614e-01 3.81133735e-01 -7.81853676e-01 8.05979729e-01 -3.84773046e-01 3.70755166e-01 1.69392124e-01 -6.19767129e-01 3.59828323e-01 8.46028566e-01 -6.15484595e-01 -2.81140562e-02 3.90722066e-01 1.25007081e+00 -9.63573158e-01 -1.46900475e-01 2.15744123e-01 5.99237204e-01 -5.21101177e-01 1.79414973e-01 -3.21077973e-01 -3.35174829e-01 4.03501749e-01 1.92595631e-01 2.49134302e-01 2.96790510e-01 3.38868082e-01 -3.46757829e-01 -1.93056896e-01 8.66376221e-01 -8.58315706e-01 1.93372834e+00 -1.15749013e+00 5.88596761e-01 -2.50901401e-01 -8.44938040e-01 9.41669166e-01 5.52082956e-01 -1.80960625e-01 -9.13616002e-01 1.56587496e-01 8.14886451e-01 1.14479624e-01 -3.76798004e-01 8.22199523e-01 -5.72771788e-01 -2.79043525e-01 1.44054145e-01 7.03711271e-01 -2.43996054e-01 4.68116909e-01 -2.31782332e-01 7.56381154e-01 1.93032548e-01 2.21827000e-01 -4.18525368e-01 2.96082020e-01 -1.35179520e-01 -3.04688513e-02 6.19720876e-01 1.47057801e-01 6.06539786e-01 1.56267017e-01 -8.58411849e-01 -1.14254618e+00 -1.19513571e+00 -4.92168777e-02 1.42096746e+00 -3.25794280e-01 -5.40362418e-01 -2.26853862e-01 -7.16760457e-01 -1.42746076e-01 1.36090696e+00 -8.89632702e-01 -2.38156661e-01 -8.61010015e-01 -3.89976442e-01 5.73673308e-01 6.65955365e-01 -3.17115150e-02 -1.81519723e+00 -2.86891401e-01 4.47106034e-01 1.21686541e-01 -9.68448937e-01 -5.77598870e-01 9.41450477e-01 -5.49278378e-01 -5.94785213e-01 -6.98762238e-01 -1.26662695e+00 9.49597284e-02 -3.78642440e-01 1.60788989e+00 4.32383195e-02 -1.61797851e-02 1.53783439e-02 -3.72997284e-01 -4.90137637e-01 -3.49640280e-01 4.85328615e-01 1.54495284e-01 -4.71158832e-01 7.00877488e-01 -5.38070023e-01 -3.49205375e-01 -1.50790244e-01 -9.37609673e-01 -4.86955881e-01 6.22523367e-01 1.03024220e+00 5.51820457e-01 -7.00021088e-01 4.59928989e-01 -8.65510404e-01 9.93799746e-01 -6.14535689e-01 -2.32110739e-01 1.21235833e-01 -5.50161242e-01 3.44205111e-01 9.82618392e-01 -4.37877208e-01 -4.23049718e-01 -4.65330034e-01 -8.48711431e-01 -2.08642229e-01 -1.41119063e-01 6.80622518e-01 4.75152917e-02 6.39245391e-01 8.27173114e-01 1.95149601e-01 -3.49724501e-01 -6.58124626e-01 9.19544399e-01 6.10346794e-01 3.46205711e-01 -5.12264729e-01 4.31073546e-01 -3.25980395e-01 -6.37242675e-01 -7.06822038e-01 -8.64089787e-01 -8.04389715e-02 -9.42596436e-01 3.67240787e-01 1.30385411e+00 -7.73745358e-01 1.80980265e-01 2.55336225e-01 -1.60236752e+00 -4.28187639e-01 -7.79817045e-01 4.95389491e-01 -2.04532713e-01 -8.58892649e-02 -8.07342231e-01 -3.92515548e-02 -5.65623641e-01 -1.14485919e+00 9.70228136e-01 -1.88441947e-01 -4.36947256e-01 -1.53345597e+00 4.71414506e-01 -4.55180138e-01 8.18731964e-01 1.11571904e-02 1.27510917e+00 -1.15579772e+00 2.13820726e-01 -3.89650203e-02 -3.74403000e-02 8.88874948e-01 4.39547077e-02 -6.22973926e-02 -8.10110271e-01 -3.32832545e-01 -4.53657359e-01 -5.54868758e-01 1.19844937e+00 9.01310742e-02 7.59271979e-01 -1.33974776e-01 -1.82520479e-01 5.92919290e-01 1.59721220e+00 3.49975348e-01 5.99872589e-01 3.67752671e-01 6.32300436e-01 2.32830510e-01 4.33690809e-02 -1.09531008e-01 3.46238405e-01 4.97083038e-01 3.66953790e-01 -4.59245108e-02 -5.92700362e-01 -3.01649958e-01 7.50000775e-01 1.47210574e+00 6.28782287e-02 -5.17221093e-01 -8.84229600e-01 1.13589430e+00 -1.48275232e+00 -7.43619084e-01 3.73505726e-02 1.92988682e+00 8.15240979e-01 3.02913934e-01 -1.88527212e-01 -2.25135908e-01 4.48619783e-01 4.79713827e-01 -1.18430853e-02 -1.37750244e+00 -2.59775639e-01 1.10953474e+00 2.19420999e-01 7.73034632e-01 -8.10426712e-01 1.31837928e+00 6.88078594e+00 7.42517591e-01 -1.15772653e+00 4.02573466e-01 7.49239698e-02 1.06419371e-02 -6.06850803e-01 -2.27906942e-01 -8.93151879e-01 3.82639259e-01 1.63962674e+00 2.67512828e-01 4.18071359e-01 6.47433281e-01 -3.70914310e-01 4.12728816e-01 -1.36022973e+00 7.36143947e-01 1.83021843e-01 -1.23519099e+00 7.36200571e-01 -2.19951287e-01 6.82776153e-01 8.43407452e-01 -6.31948784e-02 8.97610605e-01 6.50371134e-01 -1.32586277e+00 8.31845999e-01 5.85671663e-01 9.68963444e-01 -7.36935616e-01 8.97830963e-01 -1.42898738e-01 -1.10822153e+00 1.72775969e-01 -7.15378821e-01 -1.24920286e-01 2.22592518e-01 1.74179539e-01 -6.28543258e-01 3.88044983e-01 3.58562112e-01 8.76453400e-01 -4.22964245e-01 4.83688653e-01 -6.62178457e-01 4.52398777e-01 -2.07795560e-01 -2.59375125e-01 7.47518063e-01 -1.80568751e-02 2.07727551e-01 1.84206414e+00 4.32495207e-01 -5.38762093e-01 1.18651465e-01 5.00265181e-01 -2.67600864e-01 2.01293841e-01 -8.12135935e-01 -1.71341106e-01 4.11231756e-01 1.01225257e+00 -2.81476647e-01 -5.74824333e-01 -5.14938772e-01 1.27707064e+00 5.93486369e-01 2.87715465e-01 -8.03566396e-01 -7.15246558e-01 6.98596478e-01 -5.29896915e-02 6.04104459e-01 -5.92475772e-01 2.29566898e-02 -1.15267015e+00 -4.10315394e-01 -8.52236748e-01 3.96313280e-01 -7.51772285e-01 -1.59119987e+00 1.36476016e+00 -1.26429781e-01 -6.32398248e-01 -4.32272434e-01 -1.31084538e+00 -5.65980673e-01 1.11545289e+00 -1.74218440e+00 -1.12550390e+00 3.39349806e-01 6.26765609e-01 7.09445536e-01 -2.95079648e-01 1.48186791e+00 6.58594847e-01 -4.03959334e-01 6.42230630e-01 1.11282974e-01 2.51645565e-01 7.97534525e-01 -1.59380496e+00 9.05810833e-01 5.59051633e-01 8.57681870e-01 9.16316211e-01 4.27367717e-01 -2.65509099e-01 -9.36400712e-01 -1.09007084e+00 1.69609380e+00 -4.13136154e-01 1.11522043e+00 -5.70826948e-01 -5.92286348e-01 1.02807581e+00 1.18716788e+00 1.26186773e-01 5.96912205e-01 1.86053038e-01 -5.80100119e-01 1.04780175e-01 -5.18054485e-01 6.09575212e-01 1.02500117e+00 -8.60709727e-01 -1.21561730e+00 2.49535725e-01 9.21363652e-01 2.87572965e-02 -8.29276919e-01 1.05251506e-01 5.69723248e-01 -5.53967118e-01 7.66773224e-01 -9.27011728e-01 5.32023907e-01 4.48716953e-02 -4.48992461e-01 -1.95113647e+00 -4.12425756e-01 -4.49775398e-01 6.06911369e-02 8.67108166e-01 9.50751305e-01 -6.52754068e-01 -6.51004212e-03 -2.03210384e-01 -3.27665061e-01 -6.87846065e-01 -8.64401817e-01 -8.40895355e-01 6.12167001e-01 -5.53825200e-01 3.98730189e-01 8.26908410e-01 -3.38511556e-01 9.02431965e-01 -2.97454506e-01 -3.45177919e-01 -2.19130203e-01 -9.88288745e-02 2.38582119e-01 -8.89327228e-01 -1.51515931e-01 -6.28931105e-01 -2.44585082e-01 -1.15611196e+00 3.67924958e-01 -1.39351106e+00 -8.29869807e-02 -1.87138724e+00 -3.94535214e-01 -1.90989032e-01 -8.24244559e-01 6.84483290e-01 -1.97153509e-01 4.84847039e-01 2.51037180e-01 -2.10588217e-01 -3.19992542e-01 5.50409377e-01 7.39339411e-01 -2.26207092e-01 1.47208810e-01 -6.18730128e-01 -4.43633258e-01 5.27113676e-01 7.86992431e-01 -5.46612203e-01 -3.91115285e-02 -1.26467288e+00 5.10628343e-01 -6.24592423e-01 1.15874577e-02 -8.07806969e-01 -2.34862030e-01 3.06417376e-01 2.65475214e-01 -4.10265207e-01 2.90950179e-01 -9.36392844e-01 -9.57315639e-02 8.00266683e-01 -3.82444322e-01 8.40499103e-01 3.89337778e-01 1.60634574e-02 -3.75418454e-01 -4.99386460e-01 7.30896592e-01 -3.25826406e-01 -9.30967271e-01 8.62034932e-02 -6.25872552e-01 1.67786643e-01 5.78813910e-01 1.12591451e-02 1.18214255e-02 5.89936450e-02 -1.20331454e+00 4.32521030e-02 6.92730770e-02 8.40907156e-01 5.02325475e-01 -1.67357695e+00 -7.94458807e-01 4.32582855e-01 1.49494141e-01 -5.82605183e-01 -3.93167019e-01 3.86520445e-01 -5.79267442e-01 5.47073364e-01 -5.17687500e-01 7.01573193e-02 -7.35922039e-01 8.24753404e-01 4.53663856e-01 -7.15357780e-01 -2.67159849e-01 1.22515142e+00 -3.02787095e-01 -7.51929760e-01 2.52580196e-01 -9.26265240e-01 -4.34214085e-01 5.93451202e-01 3.57625365e-01 2.64929999e-02 2.63919443e-01 -5.78426182e-01 -4.22853529e-01 6.62246704e-01 -2.12133825e-01 -2.64669836e-01 1.47169352e+00 2.87170596e-02 -1.74962059e-01 7.71665871e-01 1.43825734e+00 1.78533927e-01 -3.72164279e-01 -2.15668797e-01 3.66070956e-01 3.17158371e-01 -8.30862299e-03 -9.98753667e-01 -1.02648151e+00 1.22635400e+00 5.56381881e-01 2.94108510e-01 6.92251146e-01 -3.24738681e-01 1.12352848e+00 6.15097404e-01 2.56325692e-01 -9.22427058e-01 4.75812815e-02 1.29472101e+00 1.19010365e+00 -9.78550851e-01 -4.41611856e-01 6.64934695e-01 -5.36741972e-01 1.49413526e+00 5.61338842e-01 -8.33325326e-01 8.07055295e-01 2.05217600e-01 2.06332237e-01 -2.78425843e-01 -8.93681943e-01 -4.04238820e-01 3.81362468e-01 4.61803824e-01 7.78260112e-01 1.09763682e-01 -7.35851765e-01 7.14909732e-01 -3.55214238e-01 -3.92529666e-01 2.02864423e-01 6.04319453e-01 -2.70698696e-01 -1.57776034e+00 8.51636678e-02 1.87259540e-01 -5.43152750e-01 -9.47760046e-01 -4.30005729e-01 1.13917482e+00 6.79623783e-01 5.60085595e-01 2.14812502e-01 -4.86472338e-01 7.27349222e-01 5.21559358e-01 4.29443836e-01 -1.12828290e+00 -1.15827155e+00 -2.14139909e-01 4.70994264e-01 -4.48088527e-01 -1.96492642e-01 -2.59241492e-01 -1.15217066e+00 1.56653404e-01 -1.97556883e-01 1.94606066e-01 6.59149826e-01 8.69735062e-01 1.68065950e-01 6.07778609e-01 2.31752887e-01 -8.49427819e-01 -4.65185612e-01 -1.42633796e+00 -3.44145328e-01 4.30884570e-01 3.06608498e-01 -3.15484017e-01 -2.10825473e-01 -2.76401713e-02]
[10.685410499572754, 9.12372875213623]
f65cb2ea-4ef7-4169-b555-fcb6d0cb3ee5
when-more-data-hurts-a-troubling-quirk-in-1
2205.12228
null
https://arxiv.org/abs/2205.12228v2
https://arxiv.org/pdf/2205.12228v2.pdf
When More Data Hurts: A Troubling Quirk in Developing Broad-Coverage Natural Language Understanding Systems
In natural language understanding (NLU) production systems, users' evolving needs necessitate the addition of new features over time, indexed by new symbols added to the meaning representation space. This requires additional training data and results in ever-growing datasets. We present the first systematic investigation of this incremental symbol learning scenario. Our analysis reveals a troubling quirk in building broad-coverage NLU systems: as the training dataset grows, performance on the new symbol often decreases if we do not accordingly increase its training data. This suggests that it becomes more difficult to learn new symbols with a larger training dataset. We show that this trend holds for multiple mainstream models on two common NLU tasks: intent recognition and semantic parsing. Rejecting class imbalance as the sole culprit, we reveal that the trend is closely associated with an effect we call source signal dilution, where strong lexical cues for the new symbol become diluted as the training dataset grows. Selectively dropping training examples to prevent dilution often reverses the trend, showing the over-reliance of mainstream neural NLU models on simple lexical cues. Code, models, and data are available at https://aka.ms/nlu-incremental-symbol-learning
['Yu Su', 'Jason Eisner', 'Benjamin Van Durme', 'Hao Fang', 'Sam Thomson', 'Adam Pauls', 'Emmanouil Antonios Platanios', 'Elias Stengel-Eskin']
2022-05-24
null
null
null
null
['intent-recognition']
['natural-language-processing']
[ 7.49638438e-01 3.01855594e-01 -4.91712332e-01 -3.89179677e-01 -4.96850550e-01 -7.71228552e-01 6.16054356e-01 4.17142242e-01 -3.39440644e-01 5.31551003e-01 5.35970509e-01 -6.39297068e-01 1.13203824e-01 -6.43015444e-01 -8.88545394e-01 -8.88405666e-02 2.48604104e-01 3.90603393e-01 2.39413772e-02 -3.39150697e-01 3.51723641e-01 -1.00061120e-02 -1.52775121e+00 3.01311612e-01 6.78769171e-01 6.77726030e-01 3.29343766e-01 5.59505224e-01 -6.49399221e-01 9.47024167e-01 -5.74558318e-01 -2.02731565e-01 2.67168373e-01 -6.25325501e-01 -8.70541990e-01 -2.58514553e-01 7.77612567e-01 -3.81087512e-01 -1.72103941e-01 8.14783871e-01 2.75462806e-01 3.26996520e-02 5.29283464e-01 -1.03693473e+00 -1.05168962e+00 1.10731053e+00 -2.29370162e-01 4.18052703e-01 3.46931666e-01 2.04788074e-01 1.53371727e+00 -7.08651304e-01 9.37780142e-01 1.23279142e+00 8.26521873e-01 6.34117484e-01 -1.29066050e+00 -6.41738534e-01 4.11086410e-01 -2.05181211e-01 -9.92932200e-01 -7.56698191e-01 6.73724174e-01 -4.93882537e-01 1.19737363e+00 2.20291927e-01 4.44758862e-01 1.28827429e+00 -1.63139179e-01 9.82240379e-01 8.58551502e-01 -7.83309400e-01 1.13659441e-01 1.13361023e-01 6.19788170e-01 6.07066572e-01 6.40331626e-01 -9.30529460e-02 -7.35986054e-01 -1.73369601e-01 7.25356877e-01 2.04595514e-02 -1.18409581e-01 -1.17082380e-01 -8.86390388e-01 7.50323653e-01 3.01415622e-01 5.95746219e-01 -5.74876219e-02 2.90726870e-01 4.85130340e-01 6.34544015e-01 2.93469459e-01 9.42901254e-01 -8.23191762e-01 -4.75275874e-01 -7.34209538e-01 2.40050212e-01 6.37274563e-01 8.88421953e-01 6.97295606e-01 -3.66865955e-02 1.22574354e-02 1.09139407e+00 1.10949025e-01 2.25006983e-01 8.23052764e-01 -8.73715758e-01 3.61827135e-01 7.10095942e-01 -2.06100434e-01 -5.89356542e-01 -4.18548793e-01 -7.57764280e-01 -1.24118574e-01 -8.11566114e-02 7.68034756e-01 -1.24978684e-01 -8.54154050e-01 2.18250847e+00 -7.57431611e-02 -6.20331094e-02 1.00120738e-01 4.06257808e-01 5.29393435e-01 4.46273416e-01 4.20267731e-01 -7.29389563e-02 1.33573508e+00 -5.37263572e-01 -6.32401407e-01 -8.22970271e-01 1.29302227e+00 -5.64962924e-01 1.68595529e+00 4.48860645e-01 -7.32617676e-01 -5.15780687e-01 -1.20382774e+00 -4.13322896e-01 -3.71425927e-01 -1.88081473e-01 9.00178909e-01 5.73368788e-01 -7.34582722e-01 5.66496789e-01 -2.59505659e-01 -4.32717860e-01 6.43355489e-01 -1.62118133e-02 -2.97657643e-02 -2.78389752e-01 -1.37120533e+00 8.31899464e-01 4.46685165e-01 -2.48649195e-01 -1.70663998e-01 -9.16665435e-01 -9.53503489e-01 1.32355899e-01 5.92057407e-01 -3.76073569e-01 1.59762955e+00 -1.32810831e+00 -1.09590530e+00 7.90861726e-01 -2.02214703e-01 -3.01118761e-01 3.68246019e-01 -4.93508905e-01 -2.36304641e-01 -4.42687184e-01 2.55066365e-01 6.78047478e-01 7.82751977e-01 -1.39075077e+00 -5.60110211e-01 -2.34312117e-01 -1.68489993e-01 1.45422995e-01 -2.99566448e-01 -2.97261924e-01 2.59270724e-02 -7.29928970e-01 2.25054234e-01 -6.53605044e-01 1.91120967e-01 -3.08415294e-01 -8.56264830e-02 -4.22085255e-01 7.47528911e-01 -6.09821260e-01 1.73445487e+00 -2.30961466e+00 -9.09023210e-02 8.74742046e-02 3.01999390e-01 2.05357090e-01 -2.43000686e-01 2.29840606e-01 -1.02523901e-01 3.94935250e-01 -3.44502598e-01 -2.74493515e-01 7.60407597e-02 3.13545018e-01 -5.78989804e-01 -1.31085888e-01 2.75355726e-01 1.18354893e+00 -1.24957657e+00 -2.30762601e-01 -1.17075764e-01 -1.56412944e-01 -6.39325678e-01 -2.67668396e-01 -5.25758028e-01 3.15047498e-03 -2.13530794e-01 7.13304818e-01 2.46817693e-01 -4.07005548e-01 2.30364099e-01 2.62007087e-01 -9.37564969e-02 7.15569317e-01 -9.31969047e-01 1.71559954e+00 -4.70798522e-01 7.41860688e-01 -2.76145577e-01 -7.34437287e-01 7.48587906e-01 1.25441283e-01 7.52448142e-02 -9.25568044e-01 1.44145444e-01 5.19755542e-01 5.23505390e-01 -4.88107771e-01 4.86103565e-01 -4.57641363e-01 -1.69543967e-01 6.10833168e-01 3.39178555e-02 -1.76751778e-01 2.01740801e-01 2.67933309e-01 1.27312398e+00 6.76895902e-02 4.22194928e-01 -2.98026115e-01 -1.48813566e-03 1.30414084e-01 5.11782706e-01 1.15112746e+00 -2.70867229e-01 2.62976348e-01 7.88037717e-01 -3.74600798e-01 -1.05444670e+00 -9.63318050e-01 -4.24143463e-01 1.47770143e+00 -4.28755552e-01 -5.62639475e-01 -5.02108514e-01 -8.60177636e-01 4.71394718e-01 1.20709312e+00 -6.46997392e-01 -3.37474674e-01 -4.90368068e-01 -4.47112471e-01 5.25370121e-01 5.53378105e-01 1.61889359e-01 -1.24245811e+00 -7.68082082e-01 2.92032570e-01 7.89106339e-02 -8.39141369e-01 -2.51872003e-01 4.18109626e-01 -1.10900831e+00 -8.70668054e-01 -3.01860899e-01 -5.95067739e-01 6.77404702e-01 1.14958808e-01 1.16168666e+00 4.13478464e-01 -2.27258176e-01 4.37928051e-01 -5.19799531e-01 -6.11291111e-01 -8.45408261e-01 3.18858624e-01 -4.13628332e-02 -4.98502910e-01 5.97780466e-01 -4.60063636e-01 -2.01812536e-01 -1.09012425e-01 -9.30990219e-01 6.72363341e-02 7.02432990e-01 8.56312215e-01 1.58437982e-01 -3.40307295e-01 1.05129862e+00 -1.32536364e+00 8.59314978e-01 -7.62636423e-01 -2.17790365e-01 1.45564377e-01 -7.17209160e-01 4.02821988e-01 5.98397613e-01 -5.29452562e-01 -7.85835922e-01 -2.94252485e-01 -5.94507940e-02 -1.38256401e-01 -1.61557823e-01 6.92623794e-01 9.14342329e-02 3.40934783e-01 1.11194015e+00 7.63580650e-02 -8.08516964e-02 -5.07972240e-01 4.72663343e-01 7.07126141e-01 2.24992976e-01 -6.40345097e-01 4.85276580e-01 -9.24891792e-03 -4.90540355e-01 -1.08112979e+00 -1.07104266e+00 -2.32539535e-01 -4.78748798e-01 -3.39721031e-02 4.93576437e-01 -5.83202779e-01 -1.67618364e-01 2.89910525e-01 -1.11438394e+00 -7.28250861e-01 -6.33296132e-01 1.17750153e-01 -1.87036738e-01 2.32044712e-01 -4.12797451e-01 -6.27165616e-01 -4.25461046e-02 -9.13170099e-01 7.29750931e-01 -1.83247980e-02 -1.00979948e+00 -9.63359952e-01 3.41903940e-02 3.33390653e-01 5.18146574e-01 -1.61830291e-01 1.49921679e+00 -1.07953656e+00 -3.65316957e-01 -1.63275376e-01 -2.04425603e-01 4.22936231e-01 5.14340818e-01 -2.72437900e-01 -1.09487581e+00 6.34745061e-02 2.96833012e-02 -5.46927392e-01 8.09616506e-01 1.27725005e-01 1.05837035e+00 -2.20108896e-01 -7.37953261e-02 3.37985337e-01 1.02058291e+00 2.96649635e-01 1.26675531e-01 1.97750106e-01 7.88051903e-01 6.44326746e-01 1.86038405e-01 7.47290701e-02 2.56689668e-01 3.33996266e-01 -1.89495504e-01 2.22101688e-01 -2.29915485e-01 -5.48064768e-01 2.68916577e-01 5.76089740e-01 3.62070352e-01 -6.05629310e-02 -1.22788405e+00 3.81880045e-01 -1.79728341e+00 -7.32351422e-01 7.38500580e-02 2.26036334e+00 1.09765577e+00 6.35016561e-01 -2.01388761e-01 4.85665053e-02 1.42697230e-01 3.12228620e-01 -8.18366289e-01 -4.76191819e-01 -8.83794427e-02 1.23081423e-01 4.92268264e-01 8.01423609e-01 -7.06276417e-01 1.26704490e+00 6.86181450e+00 5.37922382e-01 -1.30581629e+00 1.68418229e-01 9.06332612e-01 -9.91988629e-02 -7.76119113e-01 1.42066414e-02 -7.33808458e-01 4.66225147e-01 1.04721141e+00 -1.49104148e-01 4.32342142e-01 8.78428936e-01 -1.21597275e-01 -2.95475632e-01 -1.40667319e+00 8.37623537e-01 1.16238274e-01 -1.33905876e+00 1.61456481e-01 -6.86222538e-02 4.24002588e-01 1.84956118e-01 -3.08659300e-02 5.65784991e-01 2.90737271e-01 -9.74920928e-01 8.11576068e-01 2.37712190e-01 8.13302457e-01 -2.59364873e-01 3.98147374e-01 3.70573938e-01 -6.90352857e-01 -2.70675987e-01 -1.62069559e-01 -5.05077958e-01 6.93715587e-02 3.86168748e-01 -8.68661523e-01 -1.69841751e-01 2.40977868e-01 5.69649756e-01 -8.57522190e-01 3.45684141e-01 -3.54335576e-01 9.03726816e-01 -2.61952758e-01 1.80601943e-02 1.74892962e-01 8.53029937e-02 4.46914434e-01 1.19596457e+00 1.40070409e-01 7.80850574e-02 -9.99773815e-02 8.92550886e-01 -3.39238852e-01 3.08967173e-01 -1.11151981e+00 -4.60254550e-01 5.78786194e-01 6.52332544e-01 -7.87363946e-01 -6.92261994e-01 -6.69906497e-01 1.00027776e+00 5.10621428e-01 3.34109068e-01 -3.33356142e-01 3.09613738e-02 5.52847564e-01 1.56060666e-01 -1.23913042e-01 -2.57652640e-01 -8.86896133e-01 -1.12067986e+00 -5.95471077e-02 -9.17805076e-01 3.05878758e-01 -5.67597210e-01 -1.08555520e+00 1.51227996e-01 7.07637891e-02 -6.02659643e-01 -3.50119203e-01 -7.57353842e-01 -4.21488881e-01 7.83820510e-01 -1.34797931e+00 -7.07318068e-01 1.32987276e-01 -3.65183279e-02 8.99412453e-01 -5.44796027e-02 6.61291718e-01 1.01405457e-01 -4.24131304e-01 6.80027485e-01 -1.23513050e-01 3.61820981e-02 6.72291994e-01 -1.18831468e+00 8.49286675e-01 8.17793310e-01 4.62880135e-01 1.07119441e+00 6.75997376e-01 -1.02605951e+00 -1.19504702e+00 -4.61330801e-01 1.10888159e+00 -1.21007276e+00 8.39010298e-01 -7.87166834e-01 -1.24236715e+00 1.02133799e+00 -1.03462070e-01 -2.48112157e-01 9.45804536e-01 5.83078742e-01 -6.10684633e-01 3.07078093e-01 -7.92438745e-01 7.55279720e-01 1.50230932e+00 -6.50438845e-01 -7.92196214e-01 2.70594172e-02 1.09105062e+00 -1.60181209e-01 -3.99603248e-01 2.18048275e-01 7.05374420e-01 -6.63744032e-01 5.84122956e-01 -8.63020182e-01 4.26246047e-01 1.06967323e-01 -2.36282259e-01 -1.14052510e+00 -3.13399315e-01 -4.35251683e-01 -2.94518650e-01 9.17713583e-01 9.13582921e-01 -8.51107061e-01 6.48339868e-01 7.31126010e-01 -2.60327250e-01 -7.76666045e-01 -7.44916856e-01 -5.94778538e-01 1.77134514e-01 -8.81926239e-01 3.72408330e-01 1.36421430e+00 2.87723899e-01 7.23992765e-01 3.67274731e-02 -3.64338130e-01 3.66226703e-01 -3.33821297e-01 4.95667636e-01 -1.33446836e+00 -3.09528559e-01 -7.59577036e-01 1.01938965e-02 -1.13636470e+00 -6.21017627e-02 -1.11314702e+00 -1.24525487e-01 -1.40235770e+00 -8.16521570e-02 -6.78122878e-01 -3.37569058e-01 7.91673303e-01 -4.19528812e-01 -3.00629169e-01 3.59161228e-01 2.76448131e-01 -2.03087807e-01 2.06059709e-01 9.43942070e-01 8.49032179e-02 -4.73732352e-01 -1.94279701e-01 -1.14956892e+00 7.48717368e-01 8.25121701e-01 -3.01638693e-01 -7.79278576e-01 -6.27409935e-01 7.38154888e-01 -5.64301550e-01 1.03135139e-01 -8.52106929e-01 -9.38095599e-02 -1.50581719e-02 3.83179367e-01 -7.47495815e-02 3.70672755e-02 -7.02880025e-01 -3.26526999e-01 5.70238888e-01 -5.22866011e-01 1.48296744e-01 5.88312626e-01 1.59015790e-01 2.29424030e-01 -5.55267096e-01 5.91135502e-01 -2.28093401e-01 -9.35249984e-01 -2.77271360e-01 -2.99235582e-01 3.40026379e-01 5.74223161e-01 -4.26295727e-01 -3.64032179e-01 -2.01455846e-01 -5.92482328e-01 2.12426335e-01 4.47503120e-01 8.54762852e-01 3.88566613e-01 -1.05117989e+00 -4.03312802e-01 5.30046225e-01 2.75032640e-01 4.55281474e-02 1.47310421e-01 5.40943563e-01 -3.78287494e-01 4.15022939e-01 1.03315443e-01 -2.58718401e-01 -8.14913511e-01 2.13530302e-01 1.55309394e-01 -2.86732540e-02 -6.09980524e-01 1.11497045e+00 2.43917838e-01 -4.54836190e-01 2.09302962e-01 -8.91781986e-01 6.72092661e-03 4.77629185e-01 4.24210370e-01 8.77156034e-02 -1.20432712e-01 -1.75751865e-01 -4.39622886e-02 1.57364547e-01 -4.47960585e-01 -7.77908191e-02 1.19029510e+00 1.56611260e-02 1.87372431e-01 1.05681479e+00 9.83876050e-01 8.58837664e-02 -1.22669339e+00 -5.09862065e-01 3.42578053e-01 -3.57501954e-01 -7.11245760e-02 -9.43883419e-01 -3.43994647e-01 5.49232781e-01 3.22673351e-01 3.19719225e-01 5.50847948e-01 1.91707626e-01 7.60954857e-01 5.51753163e-01 2.10198954e-01 -1.25976348e+00 6.85134530e-02 9.15762544e-01 8.95322800e-01 -1.27653861e+00 -1.16497695e-01 -5.14315367e-02 -6.59650683e-01 8.28518629e-01 7.42563605e-01 1.51851654e-01 4.84540284e-01 1.68369681e-01 1.60758197e-01 -2.62065679e-01 -7.93992817e-01 -4.02237475e-02 -1.82532266e-01 2.38526046e-01 6.88539505e-01 4.32383940e-02 -3.24676454e-01 6.69791400e-01 -5.23181021e-01 -2.75691152e-02 4.49740261e-01 1.21237087e+00 -7.19524741e-01 -1.28823900e+00 -2.77511030e-01 9.73964453e-01 -1.11842327e-01 -6.08982861e-01 -5.68462253e-01 8.67956698e-01 3.04997891e-01 3.21785182e-01 2.50488520e-01 -5.52390814e-02 3.22917789e-01 1.03160441e+00 3.85087878e-01 -1.07732475e+00 -3.37120235e-01 -3.63240153e-01 1.76777810e-01 -3.75024557e-01 2.55411208e-01 -9.74263608e-01 -1.46600878e+00 -4.20700945e-02 -2.78255977e-02 -1.81046024e-01 5.20572484e-01 1.00700653e+00 4.06894505e-01 5.93519747e-01 2.50866693e-02 -3.92859399e-01 -5.07887900e-01 -1.09947062e+00 -1.12566225e-01 5.23689449e-01 5.08255839e-01 -6.20898962e-01 -4.79689687e-01 1.29525596e-02]
[10.672536849975586, 8.532660484313965]
bc9001dd-847a-430f-b3d9-0e5b64d34b33
rquge-reference-free-metric-for-evaluating
2211.01482
null
https://arxiv.org/abs/2211.01482v3
https://arxiv.org/pdf/2211.01482v3.pdf
RQUGE: Reference-Free Metric for Evaluating Question Generation by Answering the Question
Existing metrics for evaluating the quality of automatically generated questions such as BLEU, ROUGE, BERTScore, and BLEURT compare the reference and predicted questions, providing a high score when there is a considerable lexical overlap or semantic similarity between the candidate and the reference questions. This approach has two major shortcomings. First, we need expensive human-provided reference questions. Second, it penalises valid questions that may not have high lexical or semantic similarity to the reference questions. In this paper, we propose a new metric, RQUGE, based on the answerability of the candidate question given the context. The metric consists of a question-answering and a span scorer modules, using pre-trained models from existing literature, thus it can be used without any further training. We demonstrate that RQUGE has a higher correlation with human judgment without relying on the reference question. Additionally, RQUGE is shown to be more robust to several adversarial corruptions. Furthermore, we illustrate that we can significantly improve the performance of QA models on out-of-domain datasets by fine-tuning on synthetic data generated by a question generation model and re-ranked by RQUGE.
['Marzieh Saeidi', 'James Henderson', 'Angela Fan', 'Pouya Yanki', 'Majid Yazdani', 'Thomas Scialom', 'Alireza Mohammadshahi']
2022-11-02
null
null
null
null
['question-generation']
['natural-language-processing']
[ 6.20975904e-02 3.06233823e-01 3.13124180e-01 -3.13510150e-01 -1.61722779e+00 -8.37903738e-01 6.57177567e-01 2.83385187e-01 -4.68095183e-01 1.03700316e+00 3.92920107e-01 -2.27892146e-01 -1.71393588e-01 -7.77072906e-01 -6.20864570e-01 -4.33030501e-02 5.68841636e-01 5.61192572e-01 6.09979808e-01 -5.16803205e-01 4.42850202e-01 -3.10841594e-02 -1.44891703e+00 4.13657218e-01 1.49143481e+00 9.69065845e-01 1.73063993e-01 7.19337165e-01 -1.16077706e-01 9.62970257e-01 -1.13941371e+00 -8.77134025e-01 1.67831574e-02 -7.41574287e-01 -1.35264170e+00 -2.54166454e-01 7.50902116e-01 -1.33319214e-01 -1.98440012e-02 8.37632060e-01 4.97450501e-01 1.85603723e-01 6.91200793e-01 -1.02085221e+00 -8.30915987e-01 2.34659508e-01 2.55696982e-01 1.10679410e-01 9.53445017e-01 9.61964130e-02 1.35892522e+00 -8.89354765e-01 7.35357642e-01 1.21757555e+00 5.28503478e-01 6.02661252e-01 -1.04340136e+00 -2.94773489e-01 -4.02889848e-01 4.24706578e-01 -8.73951912e-01 -2.30004624e-01 4.85640883e-01 -4.32408005e-01 6.64126694e-01 4.86205399e-01 -4.59852517e-02 1.03515303e+00 1.44235969e-01 3.19750816e-01 1.31923139e+00 -6.31581306e-01 3.06535095e-01 3.09429526e-01 1.24046996e-01 5.01326799e-01 -6.18111752e-02 -8.10849220e-02 -1.81389675e-01 -4.09093589e-01 1.24495760e-01 -6.07074499e-01 -4.70755458e-01 -1.23590320e-01 -1.07423341e+00 1.13925195e+00 2.27457196e-01 4.53458399e-01 -2.29806110e-01 -1.72419935e-01 2.53694445e-01 5.45294583e-01 2.97751278e-01 1.15976357e+00 -5.48978567e-01 -1.57185972e-01 -1.03261840e+00 4.56557244e-01 1.07752228e+00 7.00928390e-01 8.26359630e-01 -3.30371141e-01 -6.26516461e-01 9.64825928e-01 2.34780312e-01 5.40052772e-01 8.02114189e-01 -1.28497493e+00 5.02115190e-01 6.92006230e-01 5.75588286e-01 -1.11453795e+00 -3.31064425e-02 -2.15619937e-01 -1.80462494e-01 -1.07977487e-01 6.38706446e-01 -2.92542460e-03 -5.47358871e-01 1.76340604e+00 2.34540358e-01 -2.35053614e-01 2.68769205e-01 7.66355515e-01 1.19870794e+00 6.30661130e-01 3.11411358e-02 -1.95110396e-01 1.48933518e+00 -1.06577313e+00 -7.71190524e-01 -1.83956370e-01 4.65948045e-01 -1.09081626e+00 1.66681433e+00 2.02418983e-01 -1.11527026e+00 -7.92188406e-01 -8.59519720e-01 -1.30506799e-01 -3.52535695e-01 1.19627662e-01 -1.68069333e-01 8.49539280e-01 -1.01034307e+00 4.61080402e-01 2.14657597e-02 -4.61850494e-01 -2.61745393e-01 -9.37227383e-02 -1.92814603e-01 -1.02759451e-01 -1.68229365e+00 1.24998772e+00 1.84603691e-01 -5.08077443e-01 -9.01650965e-01 -4.88084137e-01 -6.82977259e-01 1.23677075e-01 3.10201496e-01 -6.85434341e-01 1.58171129e+00 -8.29023778e-01 -1.48633003e+00 7.82467008e-01 -3.21685255e-01 -4.02931690e-01 4.26236600e-01 -2.34309852e-01 -6.59008741e-01 5.34492731e-01 3.11898082e-01 4.85323846e-01 7.79823124e-01 -1.25444460e+00 -4.68215376e-01 8.33611842e-03 4.85825151e-01 7.10323006e-02 -2.16111794e-01 6.68365881e-02 -1.01857834e-01 -6.35058165e-01 -2.76182711e-01 -6.83032274e-01 7.87766129e-02 -3.40850055e-01 -1.93916842e-01 -5.66680193e-01 5.59435844e-01 -9.71044600e-01 1.39406586e+00 -1.78818786e+00 -4.10408765e-01 -1.43102277e-02 -9.40367281e-02 5.76025188e-01 -5.06486714e-01 7.54984856e-01 1.08194940e-01 2.71644861e-01 -1.97080359e-01 3.86660516e-01 1.75775900e-01 -9.95258801e-03 -4.78430003e-01 -2.17599690e-01 2.57577032e-01 8.85820746e-01 -1.10197771e+00 -6.09354019e-01 -2.87801325e-01 1.39457220e-02 -4.90438342e-01 5.99703014e-01 -4.59094971e-01 2.12539390e-01 -1.59804419e-01 9.25331414e-02 3.43209922e-01 -2.84518093e-01 -8.05597678e-02 -1.59397200e-01 4.37262356e-01 8.32735300e-01 -8.89506876e-01 1.38869178e+00 -7.34685183e-01 5.25676548e-01 -3.94828767e-01 -6.76076949e-01 1.23613501e+00 5.41988254e-01 -9.61889103e-02 -9.70861077e-01 -1.68927208e-01 4.41100419e-01 -1.74190238e-01 -7.83790171e-01 7.94718027e-01 -1.31088942e-01 -2.04351634e-01 6.67503893e-01 3.05588007e-01 -4.94924188e-01 6.59132421e-01 3.80282760e-01 1.23212218e+00 9.96729452e-03 3.39663595e-01 -1.15481533e-01 9.05906498e-01 5.02805449e-02 2.04467669e-01 8.53783011e-01 -2.25312069e-01 6.77732885e-01 5.75724959e-01 1.48776814e-01 -8.89220476e-01 -1.23905611e+00 8.74341577e-02 1.12745953e+00 4.98554409e-02 -3.42077881e-01 -1.09074366e+00 -1.05905271e+00 -3.83356631e-01 1.24903214e+00 -5.08740366e-01 -1.04622640e-01 -4.68969852e-01 -1.74425513e-01 6.86601222e-01 1.77419677e-01 3.66581917e-01 -1.20354557e+00 -5.30177414e-01 2.80975074e-01 -9.26357567e-01 -1.05417514e+00 -6.69832289e-01 -3.42636108e-01 -6.83351219e-01 -1.29857218e+00 -6.25119627e-01 -8.17414284e-01 4.21394616e-01 1.40330762e-01 1.59253681e+00 1.34835288e-01 2.91230232e-01 7.06801236e-01 -7.46926248e-01 -1.48777053e-01 -9.19200480e-01 1.27028480e-01 -3.29452842e-01 -2.61151910e-01 4.58771348e-01 -1.85461804e-01 -6.59053385e-01 6.84081137e-01 -1.14213228e+00 -4.78098512e-01 3.55214626e-01 9.64762568e-01 3.17876786e-01 -4.21799332e-01 1.31541014e+00 -7.39572287e-01 1.20576966e+00 -4.63506907e-01 -1.80505797e-01 6.33332014e-01 -6.59790516e-01 1.68664694e-01 8.04835081e-01 -2.27722853e-01 -1.05335915e+00 -5.35001099e-01 -3.17963868e-01 2.87475973e-01 -2.56317437e-01 3.78977656e-01 -1.85418874e-01 2.99582947e-02 1.16392922e+00 6.03711568e-02 -1.69021830e-01 -3.78093600e-01 5.42912543e-01 8.22495282e-01 4.33112383e-01 -5.20959854e-01 8.87054026e-01 -7.77786504e-03 -4.02548134e-01 -4.64257836e-01 -1.18983316e+00 -5.21983027e-01 -3.56466740e-01 -2.87923843e-01 7.11147845e-01 -5.36093414e-01 -4.88420904e-01 -7.82745257e-02 -1.38237906e+00 9.09153372e-02 -4.04021561e-01 3.37432861e-01 -6.60381913e-01 7.71878600e-01 -4.39174473e-01 -6.05774999e-01 -5.55734754e-01 -9.79516923e-01 6.95012808e-01 2.69533813e-01 -6.48121178e-01 -1.04822612e+00 3.21594238e-01 8.59418631e-01 4.02836770e-01 1.40165210e-01 1.02570796e+00 -1.15116310e+00 -1.76734433e-01 -3.45203429e-01 -7.88030997e-02 7.00379312e-01 1.31843194e-01 -2.26670623e-01 -8.19165409e-01 -1.54211089e-01 3.02791476e-01 -8.39987516e-01 5.18800735e-01 -2.20688686e-01 7.43303418e-01 -7.16087461e-01 3.68119359e-01 -1.86951697e-01 1.33539796e+00 1.19481318e-01 8.26220095e-01 4.21949059e-01 1.37738958e-01 9.64076996e-01 9.99882400e-01 -2.20461134e-02 5.26291966e-01 5.41396856e-01 2.61639327e-01 3.91518623e-01 -1.93529576e-01 -5.41426241e-01 4.04502273e-01 9.58475411e-01 3.93334657e-01 -3.66263986e-01 -7.66542017e-01 7.68022895e-01 -1.44064629e+00 -1.05389619e+00 -3.50823015e-01 2.28411436e+00 1.13120294e+00 7.00578988e-02 5.75046279e-02 1.39850512e-01 6.47352934e-01 8.25614035e-02 -2.04960302e-01 -6.80822432e-01 -6.32734150e-02 4.98501539e-01 -1.62375182e-01 5.94443917e-01 -7.10369170e-01 6.83776438e-01 6.75796604e+00 9.61336851e-01 -5.73140144e-01 3.77558023e-01 4.70748782e-01 2.32957035e-01 -6.06840968e-01 1.15865409e-01 -5.36046326e-01 5.26795268e-01 1.28272271e+00 -3.39876324e-01 1.75837904e-01 6.77489281e-01 1.03290342e-01 -1.61966562e-01 -9.37052906e-01 4.52318192e-01 4.53961432e-01 -8.65257800e-01 1.31407440e-01 -4.14808184e-01 7.49014139e-01 -3.84246677e-01 -1.07706934e-01 4.08322901e-01 3.57135504e-01 -9.31874096e-01 5.35333157e-01 5.23969948e-01 6.76120102e-01 -5.80041230e-01 9.28523898e-01 3.69891763e-01 -6.19157791e-01 1.33142829e-01 -5.61144054e-01 1.51760831e-01 4.15575244e-02 4.95262295e-01 -1.06839144e+00 6.57641768e-01 3.80479664e-01 -2.21842453e-01 -9.67409313e-01 1.10272110e+00 -6.89127624e-01 7.84026027e-01 1.12994071e-02 -4.53027695e-01 2.33619481e-01 1.00034580e-01 4.01701331e-01 1.04211366e+00 4.75272715e-01 -1.40624136e-01 -1.83077216e-01 7.65859604e-01 -1.55784890e-01 5.62390089e-01 -2.65630543e-01 2.36605495e-01 5.59638023e-01 1.16473675e+00 -9.55118164e-02 -3.40562671e-01 -3.50560009e-01 9.25023019e-01 2.51610219e-01 2.59688646e-01 -8.11360359e-01 -7.65519083e-01 8.61971304e-02 1.00845337e-01 2.25566745e-01 2.05421105e-01 -1.35613784e-01 -1.07248759e+00 2.56291300e-01 -1.40666270e+00 4.41043437e-01 -1.04887974e+00 -1.38531673e+00 8.76133800e-01 -2.75443405e-01 -1.38880575e+00 -5.89855909e-01 -4.23777044e-01 -5.72245300e-01 9.86536622e-01 -1.54337239e+00 -6.78332746e-01 -4.21234906e-01 3.98590326e-01 4.44982648e-01 2.52154353e-03 9.13805306e-01 1.69803128e-01 -5.39926291e-02 8.09468269e-01 9.97425392e-02 3.96282375e-02 1.15099335e+00 -1.45690763e+00 3.23642284e-01 7.39240706e-01 1.58194512e-01 5.00930905e-01 8.50514531e-01 -4.92697269e-01 -5.77405572e-01 -1.07963598e+00 1.48729289e+00 -7.78093815e-01 6.95086837e-01 3.12567949e-02 -1.15562391e+00 1.48407608e-01 4.69273180e-01 -5.38324833e-01 9.77974176e-01 -1.17337994e-01 -5.76532841e-01 -4.40561101e-02 -1.32762778e+00 4.24184322e-01 5.22694826e-01 -6.85224354e-01 -1.29244506e+00 5.32678425e-01 9.06839550e-01 -7.29944780e-02 -9.55545425e-01 2.34637648e-01 2.71167129e-01 -1.11773729e+00 6.99396849e-01 -6.89270675e-01 7.05399215e-01 -2.89834797e-01 -1.39933139e-01 -1.47537220e+00 -2.09154442e-01 -4.55335647e-01 -1.53844245e-03 1.44941413e+00 8.15864503e-01 -4.38280225e-01 4.76957202e-01 5.53022027e-01 7.86038488e-02 -6.18682742e-01 -9.89172399e-01 -9.94180560e-01 3.29036772e-01 -4.69360501e-02 7.23647475e-01 6.79258287e-01 1.07975692e-01 6.90426171e-01 -2.55040109e-01 -1.17448404e-01 2.45634764e-01 5.69851547e-02 7.61691034e-01 -1.12831056e+00 -2.87474781e-01 -1.37234718e-01 -3.01051382e-02 -1.04933798e+00 -1.04397386e-02 -7.71206021e-01 2.70459235e-01 -1.72612214e+00 -2.30161920e-02 2.40459573e-02 -9.46671292e-02 7.19942898e-02 -7.35181928e-01 2.00014398e-01 1.39918998e-01 2.10439071e-01 -8.16766202e-01 7.10134208e-01 1.33343232e+00 2.46133469e-02 1.70720905e-01 -1.00366078e-01 -8.60391617e-01 5.49997509e-01 9.01471615e-01 -6.77197635e-01 -3.02561402e-01 -3.10765415e-01 3.26291949e-01 1.81950450e-01 3.33347559e-01 -1.13476825e+00 2.87233200e-02 -7.47865345e-03 -2.19252259e-02 -3.38829666e-01 8.82370248e-02 -6.28726602e-01 -3.03558469e-01 5.05565345e-01 -6.77640975e-01 1.92654595e-01 -1.75051332e-01 4.53089088e-01 -5.42499602e-01 -1.17104721e+00 8.37776303e-01 -2.06683666e-01 -3.90897244e-01 -2.01671720e-01 -3.49855870e-01 7.07814276e-01 8.17481697e-01 -1.09567419e-01 -4.95218635e-01 -8.33410323e-01 -3.83829832e-01 1.95625663e-01 3.79492760e-01 4.32859153e-01 5.61660945e-01 -1.33931625e+00 -9.99443293e-01 -4.04586077e-01 3.63826841e-01 -5.85719109e-01 1.57896355e-01 3.67056936e-01 -5.35489261e-01 7.11172342e-01 4.55099232e-02 -1.24285601e-01 -1.07832122e+00 5.36191642e-01 2.82157421e-01 -8.54321063e-01 2.48956412e-01 5.81947386e-01 -5.65664880e-02 -8.52132082e-01 5.45457080e-02 -1.01490319e-01 -5.22382259e-01 1.56770393e-01 6.51620328e-01 6.45417154e-01 1.80364579e-01 -6.05724931e-01 -8.46712962e-02 4.56334114e-01 1.19191378e-01 -4.02349919e-01 7.48542070e-01 -2.40409121e-01 -8.73632282e-02 2.12807983e-01 1.09450221e+00 3.05393398e-01 -7.11353719e-01 -4.04516608e-01 3.15557748e-01 -4.40845758e-01 -4.59153235e-01 -1.18464899e+00 -4.39133555e-01 8.76886845e-01 4.61352855e-01 6.82968855e-01 1.13771129e+00 -3.57582346e-02 1.10628033e+00 6.22448862e-01 1.83181241e-01 -1.25490713e+00 6.50244951e-01 7.95694649e-01 1.18387794e+00 -1.20851254e+00 -4.16621476e-01 -2.69370168e-01 -6.77785039e-01 1.01989114e+00 7.90778577e-01 -2.46083480e-03 2.22251564e-01 -5.70914388e-01 3.65929544e-01 -1.61769358e-03 -7.33910322e-01 -2.08580077e-01 6.15388811e-01 6.96629465e-01 5.27773380e-01 -2.10722312e-01 -9.04048741e-01 5.97598732e-01 -6.86155975e-01 -1.65010810e-01 6.45619929e-01 4.95068341e-01 -8.03096533e-01 -1.15825379e+00 -5.13676345e-01 6.14477694e-01 -8.04833710e-01 -2.16568425e-01 -5.62343717e-01 4.63620424e-01 -8.82499292e-02 1.61383104e+00 -2.86733449e-01 -3.26276302e-01 6.46530688e-01 3.16890210e-01 2.76379168e-01 -7.56922245e-01 -8.61510754e-01 -4.83976603e-01 3.97494256e-01 -3.58429998e-01 -3.69550854e-01 -3.94556731e-01 -7.41366565e-01 1.79870725e-02 -6.15929484e-01 8.22798610e-01 3.88218224e-01 9.58272696e-01 3.74681979e-01 2.42322057e-01 7.75540829e-01 1.42130673e-01 -9.86421645e-01 -1.33819556e+00 -1.24167763e-01 8.14090967e-01 2.47571729e-02 -3.16574305e-01 -6.27857149e-01 9.27599967e-02]
[11.513933181762695, 8.227394104003906]
44e88a45-0072-42c8-9ea9-4985027acb07
latentforensics-towards-lighter-deepfake
2303.17222
null
https://arxiv.org/abs/2303.17222v1
https://arxiv.org/pdf/2303.17222v1.pdf
LatentForensics: Towards lighter deepfake detection in the StyleGAN latent space
The classification of forged videos has been a challenge for the past few years. Deepfake classifiers can now reliably predict whether or not video frames have been tampered with. However, their performance is tied to both the dataset used for training and the analyst's computational power. We propose a deepfake classification method that operates in the latent space of a state-of-the-art generative adversarial network (GAN) trained on high-quality face images. The proposed method leverages the structure of the latent space of StyleGAN to learn a lightweight classification model. Experimental results on a standard dataset reveal that the proposed approach outperforms other state-of-the-art deepfake classification methods. To the best of our knowledge, this is the first study showing the interest of the latent space of StyleGAN for deepfake classification. Combined with other recent studies on the interpretation and manipulation of this latent space, we believe that the proposed approach can help in developing robust deepfake classification methods based on interpretable high-level properties of face images.
['Renaud Seguier', 'Simon Leglaive', 'Stephane Paquelet', 'Amine Kacete', 'Matthieu Delmas']
2023-03-30
null
null
null
null
['face-swapping']
['computer-vision']
[ 2.50873566e-01 1.29535347e-01 -2.81828344e-01 -4.13620561e-01 -5.05808949e-01 -6.83565319e-01 9.22104716e-01 -7.14584768e-01 8.08236524e-02 3.71655285e-01 9.09154862e-02 -3.11173201e-01 7.92477950e-02 -7.21452653e-01 -7.62938738e-01 -8.08566749e-01 -2.10853163e-02 4.04542089e-02 -3.35853636e-01 8.98064077e-02 3.00430864e-01 5.50117314e-01 -1.31995952e+00 5.02861083e-01 2.15317100e-01 1.43096101e+00 -3.83634686e-01 6.13347828e-01 2.20162556e-01 1.03409612e+00 -7.84867287e-01 -8.77276957e-01 3.11198443e-01 -6.40056551e-01 -8.25307429e-01 1.82210192e-01 6.44390941e-01 -7.39167809e-01 -5.75535297e-01 9.79138672e-01 1.81570962e-01 -4.11826879e-01 6.56951070e-01 -1.59683228e+00 -8.51241052e-01 1.73662439e-01 -1.23420469e-01 2.77860403e-01 1.03602901e-01 2.84045100e-01 6.81902587e-01 -9.59554791e-01 6.02332294e-01 1.37574160e+00 7.40674794e-01 7.01109827e-01 -1.09649217e+00 -1.17135811e+00 -2.57106781e-01 3.63964438e-01 -1.28462672e+00 -9.10996437e-01 1.14258945e+00 -7.39756584e-01 6.63658082e-01 1.02616429e-01 5.52140176e-01 1.66852343e+00 3.19865137e-01 6.56380892e-01 1.25757945e+00 -4.62569952e-01 2.50660479e-01 3.55628021e-02 -3.87180805e-01 1.12235200e+00 6.73085079e-03 3.06495130e-01 -7.97862113e-01 -2.24247605e-01 8.21837783e-01 -2.47119263e-01 -1.81782156e-01 -5.14472961e-01 -8.00015628e-01 1.19521427e+00 1.29015846e-02 2.71170616e-01 -2.65184827e-02 5.17940938e-01 4.59108591e-01 5.51029086e-01 7.88656235e-01 4.06333238e-01 -1.30256549e-01 -3.37400854e-01 -1.44077611e+00 -1.32875741e-01 6.83716953e-01 5.88596046e-01 4.77846384e-01 3.87769610e-01 1.18433677e-01 3.36206883e-01 4.24489170e-01 3.06585163e-01 4.25518423e-01 -9.81917024e-01 2.45108709e-01 3.12638313e-01 -1.72487691e-01 -1.34091341e+00 4.16376621e-01 -2.26051986e-01 -6.75884008e-01 5.29389024e-01 5.47374070e-01 1.46745071e-01 -8.09070706e-01 1.54513633e+00 2.10252255e-02 6.75153196e-01 1.49600744e-01 6.37895405e-01 4.96133387e-01 4.32010829e-01 -2.23587945e-01 9.17455703e-02 1.10992455e+00 -7.92278051e-01 -7.51607239e-01 -3.14433165e-02 4.06679302e-01 -8.54692340e-01 8.35395336e-01 5.40456295e-01 -9.78576303e-01 -4.82030004e-01 -1.34311688e+00 -3.64917815e-02 -2.31223717e-01 2.56219745e-01 7.71505594e-01 1.32380760e+00 -1.04269075e+00 6.74045444e-01 -8.94914687e-01 -1.49330616e-01 1.13503587e+00 2.79410034e-01 -5.57675421e-01 -1.22470722e-01 -1.06818855e+00 7.10052550e-01 2.09161147e-01 2.89153934e-01 -1.53450000e+00 -4.98232812e-01 -6.98807180e-01 -5.15979975e-02 3.74142587e-01 -4.03269261e-01 1.03774858e+00 -1.47029901e+00 -1.65343893e+00 1.19126403e+00 -9.27243531e-02 -4.85033780e-01 8.42132270e-01 -2.76324153e-01 -4.54167515e-01 5.47059476e-01 -1.51601538e-01 3.81026715e-01 1.90014648e+00 -1.04458153e+00 -9.72534791e-02 -3.16639990e-01 1.39066637e-01 -5.22089422e-01 -5.13958454e-01 1.88468501e-01 -2.63225026e-02 -1.08317769e+00 -2.39064366e-01 -9.27041590e-01 4.90568578e-01 4.42073762e-01 -1.99232742e-01 -1.16656594e-01 1.55801702e+00 -7.42304564e-01 1.02692926e+00 -2.27905846e+00 2.69053549e-01 9.05308500e-03 3.79947871e-01 4.89602059e-01 9.04301703e-02 2.96186298e-01 -2.63893545e-01 4.09193784e-01 1.57914739e-02 -5.84820271e-01 -1.13332994e-01 2.49021620e-01 -7.16073871e-01 9.95795727e-01 4.45830375e-01 1.10056043e+00 -7.58813441e-01 -1.79518864e-01 2.14005098e-01 5.15545011e-01 -4.80586141e-01 4.41233188e-01 -2.17867848e-02 5.15924811e-01 -2.42841527e-01 6.53542757e-01 6.99858189e-01 -2.60020554e-01 1.82103127e-01 -3.23570579e-01 2.26352811e-01 1.33680627e-01 -6.12057388e-01 1.64000976e+00 -3.19664776e-01 1.17073047e+00 -4.96584959e-02 -1.22693312e+00 6.56285524e-01 4.88551617e-01 2.35705048e-01 -2.35708326e-01 2.85994709e-01 1.35530859e-01 -2.05194354e-01 -4.42104697e-01 2.21847966e-01 -1.88579306e-01 7.23905414e-02 7.66136944e-01 4.69558775e-01 1.27327647e-02 -3.16028953e-01 2.27346703e-01 8.47567379e-01 2.58639395e-01 1.19348653e-01 -3.38317215e-01 4.28218722e-01 -6.14064276e-01 2.20881537e-01 8.37036431e-01 -1.89602897e-01 3.00178885e-01 7.73976624e-01 -7.02486396e-01 -1.31621063e+00 -7.46978462e-01 -1.13442771e-01 7.38133371e-01 -3.07704031e-01 -5.82525611e-01 -1.18037879e+00 -1.00989807e+00 -3.73523645e-02 4.46686864e-01 -9.95330751e-01 -4.87688094e-01 -5.29740036e-01 -4.13177073e-01 1.27732611e+00 3.36393744e-01 7.66665161e-01 -9.28513765e-01 -4.89273846e-01 -3.84862125e-02 -4.23346646e-02 -1.33611107e+00 -2.45266572e-01 -4.43732679e-01 -6.14368021e-01 -1.24584520e+00 -3.60285103e-01 -5.12196481e-01 6.68465734e-01 3.36450599e-02 1.02582407e+00 3.53260994e-01 -1.24055274e-01 6.13617539e-01 -5.02853155e-01 -1.55593619e-01 -9.68960762e-01 -2.67636687e-01 6.62653893e-02 4.87072408e-01 3.81338090e-01 -4.28692579e-01 -4.85632598e-01 4.81766611e-01 -1.04018176e+00 1.39433280e-01 2.29344085e-01 1.06603110e+00 -4.71909642e-02 3.89184684e-01 3.78524482e-01 -9.03753281e-01 2.24677920e-01 -5.31731427e-01 -6.00224137e-01 2.01818630e-01 -6.05170548e-01 1.86199635e-01 4.45853770e-01 -5.20199418e-01 -9.57970083e-01 -2.97049224e-01 -1.15443766e-02 -6.93504095e-01 -1.50599197e-01 2.60086358e-01 -2.97829986e-01 -4.76356119e-01 3.55589181e-01 2.68497586e-01 1.52583584e-01 -3.28789145e-01 3.13874930e-01 6.66524708e-01 5.96837282e-01 -4.90762025e-01 9.62036252e-01 8.01178694e-01 2.99040765e-01 -7.02278197e-01 -6.74347937e-01 8.55185986e-02 -6.52264178e-01 -3.53345990e-01 8.08739722e-01 -8.37541878e-01 -5.50723076e-01 1.22648013e+00 -1.07968879e+00 -3.42216641e-01 -2.28140652e-02 -6.60581216e-02 -7.04304516e-01 7.29654193e-01 -6.73179209e-01 -7.08817840e-01 -1.25711098e-01 -1.26291168e+00 1.09354830e+00 -4.27532881e-01 -1.52904660e-01 -1.19664454e+00 -4.68347251e-01 7.08898664e-01 3.17469716e-01 3.80504429e-01 7.62586653e-01 -3.18954825e-01 -9.00942206e-01 -1.98026106e-01 1.88300852e-02 8.13971877e-01 2.93606639e-01 3.81774977e-02 -1.39334631e+00 -5.34605086e-01 4.81691390e-01 -5.86336911e-01 9.49461758e-01 8.50935951e-02 1.54342878e+00 -8.15552056e-01 2.40094191e-03 1.01906490e+00 1.15803981e+00 -4.97566350e-02 1.03384268e+00 2.13102818e-01 7.31686532e-01 2.94415772e-01 3.47310990e-01 4.03652370e-01 -2.79692397e-03 7.24273682e-01 5.49808383e-01 4.00385037e-02 -2.37063378e-01 -4.42683578e-01 6.47470653e-01 4.33246225e-01 8.71154368e-02 -4.03484941e-01 -6.88719153e-01 2.10039705e-01 -1.51211083e+00 -1.17444348e+00 2.86784172e-01 1.80570710e+00 5.71058393e-01 1.69923946e-01 -2.16736451e-01 4.52548802e-01 4.14966345e-01 4.60263968e-01 -3.79321307e-01 -3.51429641e-01 -8.71738791e-02 3.39742064e-01 3.71780187e-01 3.13598573e-01 -1.26223385e+00 1.10254323e+00 6.49330568e+00 1.08635592e+00 -1.36877573e+00 4.06430334e-01 8.21550608e-01 3.51354629e-01 -4.40062257e-03 -6.38500834e-03 -6.49557114e-01 7.21159518e-01 9.95619357e-01 9.72976983e-02 6.26047850e-01 9.94295835e-01 -1.18655432e-02 2.14831054e-01 -1.35866749e+00 1.20382893e+00 5.96941590e-01 -1.54710925e+00 4.46190573e-02 4.34137464e-01 7.03005016e-01 -3.92968208e-01 5.63005209e-01 -6.06979281e-02 1.97610054e-02 -1.43596959e+00 9.71246779e-01 4.95809495e-01 1.17575884e+00 -6.09975219e-01 6.25282526e-01 1.84454769e-01 -7.95862198e-01 -1.00285202e-01 -1.33065253e-01 -5.59884310e-02 -2.55932599e-01 2.88356990e-01 -9.29292560e-01 1.66595846e-01 6.59872651e-01 9.63237584e-01 -6.57723963e-01 1.91651732e-01 -4.14602667e-01 1.06340408e+00 7.77051672e-02 6.68013811e-01 1.60994068e-01 1.25005782e-01 4.71409529e-01 1.04587030e+00 5.21994710e-01 -2.03293078e-02 -1.31395772e-01 9.32811916e-01 -4.10002083e-01 -4.87036079e-01 -8.13530743e-01 -6.47739649e-01 1.84339359e-01 1.05613148e+00 -5.84444106e-01 -3.67233127e-01 -3.87419939e-01 1.19948733e+00 1.08085766e-01 1.09262183e-01 -7.02551723e-01 2.36385539e-01 7.12029576e-01 6.44616261e-02 5.76294899e-01 -4.04452324e-01 -2.15670258e-01 -1.54469633e+00 -3.22149582e-02 -1.11657119e+00 3.46069485e-01 -7.35588729e-01 -1.37040710e+00 5.11395872e-01 -2.98241731e-02 -1.16806090e+00 -4.41279829e-01 -8.27403963e-01 -3.80111128e-01 5.72052240e-01 -1.53194547e+00 -1.56065214e+00 -2.66399354e-01 6.89192295e-01 5.01307905e-01 -7.87437916e-01 1.10998702e+00 -3.03479712e-02 -2.95423567e-01 9.49867368e-01 2.53626108e-01 4.97999400e-01 5.32908738e-01 -8.52640927e-01 5.30586362e-01 1.07326961e+00 1.84692800e-01 5.68232894e-01 5.26271284e-01 -5.61707437e-01 -1.43246973e+00 -8.49508286e-01 5.02379775e-01 -6.60565972e-01 6.74188137e-01 -6.90727472e-01 -7.22483039e-01 8.60943437e-01 1.95029676e-01 1.91343144e-01 9.07589734e-01 -2.30150715e-01 -7.65744209e-01 -5.64101078e-02 -1.39666879e+00 1.74567521e-01 8.63923848e-01 -1.04721415e+00 -3.97950500e-01 1.93067744e-01 1.74036726e-01 -2.57251322e-01 -7.66717076e-01 1.14281967e-01 7.80921459e-01 -1.13777876e+00 1.01040149e+00 -6.40016615e-01 7.16085255e-01 -9.47685819e-03 -1.51018262e-01 -1.26958656e+00 -1.35670260e-01 -7.65117168e-01 -5.31001508e-01 1.21120083e+00 -3.75136554e-01 -4.52434331e-01 9.86807704e-01 4.02846575e-01 1.40258059e-01 -3.93247485e-01 -1.09595942e+00 -7.24351645e-01 1.51290417e-01 -5.18653989e-01 6.48024082e-01 1.14750957e+00 -5.20673215e-01 -2.61644006e-01 -9.94236588e-01 1.52773663e-01 9.50592160e-01 -5.04075736e-02 9.45724189e-01 -1.08998954e+00 -3.95733833e-01 -1.37222469e-01 -9.33451593e-01 -7.50805855e-01 6.58811688e-01 -7.23106742e-01 -5.48312843e-01 -5.11027634e-01 6.44011796e-02 -2.60576487e-01 -1.74437121e-01 5.27121484e-01 3.09060186e-01 6.06910408e-01 1.69641942e-01 3.54558647e-01 -1.86311454e-01 5.01242280e-01 1.01814425e+00 -1.86866045e-01 5.32447040e-01 -3.40642124e-01 -5.74624479e-01 7.90412486e-01 8.04154098e-01 -6.17465913e-01 -3.74564290e-01 -3.67015153e-01 1.54669225e-01 -4.95978817e-02 9.32641745e-01 -8.74042988e-01 -6.92311227e-02 -1.29253834e-01 5.64011633e-01 -3.14129330e-02 5.20760298e-01 -8.63087893e-01 2.86111057e-01 4.04676169e-01 -3.54631215e-01 -2.34311089e-01 3.15837264e-02 4.65612084e-01 -2.31877401e-01 -9.79602933e-02 9.19870138e-01 -2.22853377e-01 -6.60993040e-01 2.68202692e-01 -4.65432197e-01 -3.54632437e-02 9.28032041e-01 -2.95283079e-01 -3.32346112e-01 -6.72449350e-01 -4.55594927e-01 -5.86904228e-01 7.04472601e-01 5.97184002e-01 7.57030308e-01 -1.43517554e+00 -5.72055459e-01 7.15528429e-01 -5.99078387e-02 -5.99888027e-01 1.39502645e-01 2.70572543e-01 -5.81694543e-01 3.40925425e-01 -5.14945388e-01 -4.76499885e-01 -1.43819511e+00 5.82136273e-01 4.33761239e-01 -1.31746411e-01 -4.59140211e-01 7.79325545e-01 2.23847687e-01 1.97894424e-01 -3.11385151e-02 7.16751441e-03 1.39984220e-01 -2.70321742e-02 6.04613423e-01 2.10972190e-01 5.15962355e-02 -1.09190702e+00 -2.53718704e-01 2.75599092e-01 -1.05733067e-01 2.26050499e-03 1.31712937e+00 -5.25120609e-02 -1.96007103e-01 1.07843041e-01 1.42824066e+00 3.99600491e-02 -1.58179510e+00 -9.35927853e-02 -2.77515829e-01 -1.07498789e+00 3.21280003e-01 -6.05769277e-01 -1.27658939e+00 9.77503121e-01 7.78226912e-01 -1.61447488e-02 1.13549137e+00 -1.23663813e-01 7.38063872e-01 1.00986645e-01 4.39891636e-01 -8.20749819e-01 6.57813191e-01 1.00658134e-01 9.20617938e-01 -1.12590158e+00 6.42474880e-03 -4.33574706e-01 -3.97481412e-01 1.37265146e+00 2.13993132e-01 -1.18444994e-01 8.00690293e-01 1.06130585e-01 -4.32793349e-02 -2.23334342e-01 -5.19397974e-01 5.24679422e-01 3.62738609e-01 4.21999991e-01 4.50298525e-02 -7.09545091e-02 2.25255072e-01 3.24265301e-01 -4.07563210e-01 3.28947186e-01 6.10041797e-01 8.92866910e-01 1.26596957e-01 -1.33772469e+00 -3.23900789e-01 1.68395177e-01 -9.65777218e-01 3.80974449e-02 -2.95574039e-01 4.64451164e-01 3.30339730e-01 1.12960792e+00 -1.85702994e-01 -4.29690897e-01 -3.42750341e-01 2.64170557e-01 6.87235773e-01 -2.72983700e-01 -3.30926508e-01 -2.01948404e-01 -2.19014641e-02 -6.35390937e-01 -6.98256433e-01 -6.90584064e-01 -4.97444302e-01 -6.26542449e-01 -2.32087210e-01 -1.34856971e-02 7.27165699e-01 1.13347888e+00 2.95331836e-01 -8.45903531e-02 7.14199960e-01 -9.59246993e-01 -5.65760195e-01 -7.14275122e-01 -5.98566234e-01 6.91486418e-01 5.71830571e-01 -8.36857677e-01 -4.86369699e-01 4.19380516e-01]
[12.569897651672363, 0.9930076003074646]
63793b2d-e611-417b-b44e-0dccee2b644e
large-scale-spectral-clustering-using
null
null
https://aclanthology.org/W18-1705
https://aclanthology.org/W18-1705.pdf
Large-scale spectral clustering using diffusion coordinates on landmark-based bipartite graphs
Spectral clustering has received a lot of attention due to its ability to separate nonconvex, non-intersecting manifolds, but its high computational complexity has significantly limited its applicability. Motivated by the document-term co-clustering framework by Dhillon (2001), we propose a landmark-based scalable spectral clustering approach in which we first use the selected landmark set and the given data to form a bipartite graph and then run a diffusion process on it to obtain a family of diffusion coordinates for clustering. We show that our proposed algorithm can be implemented based on very efficient operations on the affinity matrix between the given data and selected landmarks, thus capable of handling large data. Finally, we demonstrate the excellent performance of our method by comparing with the state-of-the-art scalable algorithms on several benchmark data sets.
['Khiem Pham', 'Guangliang Chen']
2018-06-01
null
null
null
ws-2018-6
['imagedocument-clustering']
['computer-vision']
[-2.03674957e-01 -3.58864546e-01 -5.83014823e-02 3.59019451e-02 -8.93806875e-01 -7.51149595e-01 4.54820246e-01 3.33680809e-01 -4.72281605e-01 3.37786555e-01 2.46446170e-02 -1.36079803e-01 -4.14497793e-01 -6.23412967e-01 -4.49494034e-01 -9.79695499e-01 -4.42968100e-01 7.42859185e-01 4.36412364e-01 2.47006357e-01 3.50616157e-01 7.03887463e-01 -1.35620558e+00 -2.58391619e-01 1.20148373e+00 5.64346671e-01 1.72899127e-01 4.05514061e-01 2.55170260e-02 -3.06492969e-02 -4.76687886e-02 -7.43359178e-02 2.44697720e-01 -3.93145502e-01 -9.72327173e-01 6.02473259e-01 2.58848816e-01 4.94446695e-01 -8.87636244e-02 1.15274894e+00 2.70054609e-01 5.44416785e-01 9.13042903e-01 -1.11947083e+00 -4.90583241e-01 2.69219726e-01 -1.09892404e+00 -8.18289518e-02 2.53174782e-01 -3.67079884e-01 1.19399679e+00 -1.07540226e+00 9.03494895e-01 9.29144621e-01 6.08374834e-01 2.27511868e-01 -1.57450104e+00 -3.10451955e-01 -6.08229032e-03 2.85985142e-01 -1.77357185e+00 -1.70798331e-01 9.75517869e-01 -4.25179720e-01 4.94503528e-01 3.71992320e-01 8.19930017e-01 3.03993195e-01 -3.56165349e-01 5.98898470e-01 1.10886610e+00 -3.67743999e-01 4.86896217e-01 -2.42995664e-01 2.17125773e-01 8.45739305e-01 1.56010404e-01 -3.95521730e-01 -4.66128327e-02 -4.93903399e-01 5.12943745e-01 -7.47959828e-03 -2.43121848e-01 -9.74572718e-01 -1.39025533e+00 1.04175663e+00 7.27441013e-01 5.67030489e-01 -3.37373972e-01 3.96140441e-02 2.24150762e-01 3.46405059e-02 5.27630329e-01 8.41925964e-02 -4.23624329e-02 6.55646920e-02 -1.06500614e+00 -7.54266679e-02 7.35812783e-01 8.52264225e-01 9.95520175e-01 -4.46030617e-01 4.43670481e-01 8.86553168e-01 3.31542253e-01 4.29635227e-01 1.71151578e-01 -8.85681152e-01 3.59252304e-01 8.14280808e-01 -7.77570978e-02 -1.37560129e+00 -6.77197754e-01 -1.16097160e-01 -1.12597466e+00 -2.12619528e-01 5.17597258e-01 7.86159933e-02 -7.31530488e-01 1.52697790e+00 6.34145141e-01 4.65445101e-01 5.87239712e-02 9.83846545e-01 3.67792815e-01 6.41615391e-01 -2.70832479e-01 -4.69994247e-01 9.12706077e-01 -9.37600255e-01 -3.15461099e-01 3.01881850e-01 8.33570600e-01 -7.69411504e-01 8.27424347e-01 3.23310673e-01 -9.50058639e-01 -2.09241241e-01 -8.22200596e-01 6.96027428e-02 -3.24832588e-01 7.44976848e-02 4.67286497e-01 5.63849092e-01 -1.29131126e+00 5.01830220e-01 -1.16378999e+00 -7.28818595e-01 2.59947449e-01 4.80032414e-01 -4.50936258e-01 -1.63744956e-01 -5.81841469e-01 4.49206799e-01 3.46956134e-01 7.10768998e-02 -5.03054857e-01 -2.74287701e-01 -7.10990608e-01 -1.44511178e-01 5.41085362e-01 -5.76996803e-01 2.58876264e-01 -6.42906010e-01 -1.26848686e+00 8.03541660e-01 -2.82388002e-01 -1.86726928e-01 4.20106411e-01 1.64827853e-01 -1.50537506e-01 5.25847554e-01 1.05441310e-01 4.49238926e-01 7.54539073e-01 -1.44379628e+00 -6.36559784e-01 -5.80516398e-01 -3.26209247e-01 3.55337471e-01 -5.97217023e-01 -1.58998668e-01 -9.24353361e-01 -4.82571512e-01 6.51689112e-01 -1.45250380e+00 -5.57969272e-01 -2.93314219e-01 -5.74030042e-01 -4.21512097e-01 1.06773984e+00 -3.25938106e-01 1.42774236e+00 -2.25004244e+00 8.79110992e-01 8.93895805e-01 3.72618675e-01 -2.28687331e-01 -1.87264681e-02 6.33220792e-01 4.23524994e-03 8.24450627e-02 -7.02969193e-01 -3.81432444e-01 -1.31702498e-01 1.37209594e-01 1.45638362e-01 1.15850711e+00 -3.37203830e-01 5.74517906e-01 -1.06897581e+00 -6.35874331e-01 2.49720290e-01 4.67927396e-01 -4.71876681e-01 -1.38337374e-01 1.90867990e-01 4.00614291e-01 -3.87843400e-01 4.03384656e-01 7.56162763e-01 -3.23025614e-01 3.46251011e-01 -1.92742750e-01 -1.75318882e-01 -3.53087485e-01 -1.69353902e+00 1.81754351e+00 -6.83243349e-02 3.66225988e-01 4.02756602e-01 -1.03166568e+00 8.53935838e-01 7.54007548e-02 1.04721105e+00 2.31807269e-02 -1.81495044e-02 4.53057379e-01 -1.68219045e-01 -1.23138145e-01 3.58431667e-01 8.97259042e-02 -1.54263573e-02 5.36188841e-01 -3.40872645e-01 -1.40123278e-01 5.06199896e-01 5.99819005e-01 1.08719635e+00 -2.85537571e-01 -2.18782909e-02 -7.58149445e-01 9.03510213e-01 1.91705942e-01 3.18811923e-01 1.30037457e-01 -3.03080603e-02 6.92111731e-01 2.04383016e-01 -1.00799747e-01 -9.43314075e-01 -1.09294796e+00 -2.08561301e-01 7.30661690e-01 3.74080360e-01 -4.88524437e-01 -1.05132759e+00 -7.66930461e-01 6.89483434e-02 1.57522172e-01 -7.03784525e-01 2.22199529e-01 -5.06602585e-01 -7.79321790e-01 1.10373393e-01 2.77557790e-01 3.51824582e-01 -5.09629309e-01 -1.85524836e-01 5.82994334e-02 -1.82569981e-01 -6.97565019e-01 -8.77692640e-01 -2.95106824e-02 -1.05309415e+00 -1.42206705e+00 -8.66681218e-01 -1.16805410e+00 1.20432985e+00 7.11958706e-01 7.11673319e-01 3.17092538e-01 -1.49050131e-01 6.69698238e-01 -3.27410698e-01 4.54339832e-01 -1.76607445e-01 3.11735123e-01 2.46993139e-01 4.60111618e-01 1.39655828e-01 -5.09619415e-01 -6.11666024e-01 5.21272659e-01 -1.02941477e+00 -2.45829046e-01 2.44387895e-01 6.60525322e-01 8.56208086e-01 4.31876332e-01 3.97971302e-01 -8.08244050e-01 6.01823092e-01 -6.33862734e-01 -7.67774284e-01 1.77308112e-01 -4.73984241e-01 -6.30653724e-02 5.75548351e-01 -1.96082920e-01 -5.27041674e-01 5.25968254e-01 2.98877716e-01 -5.28665602e-01 9.88998562e-02 6.20112002e-01 3.30093056e-02 -3.93954098e-01 3.03073466e-01 5.28679080e-02 6.40996620e-02 -5.52799106e-01 7.57601798e-01 5.84646106e-01 4.43420410e-01 -5.43996572e-01 1.09025824e+00 9.60152924e-01 3.13813090e-01 -9.63624358e-01 -4.32905197e-01 -1.15882039e+00 -1.23149109e+00 -1.34040728e-01 1.04830658e+00 -5.51217318e-01 -6.41015649e-01 1.78168684e-01 -8.00771952e-01 5.32189310e-02 -2.88437381e-02 4.54226375e-01 -6.36250556e-01 6.94825232e-01 -5.80007732e-01 -7.08543837e-01 -2.32166201e-02 -1.11844218e+00 8.18969250e-01 -1.40351683e-01 5.76203391e-02 -1.46954274e+00 3.38670611e-01 2.83457845e-01 1.03296209e-02 5.04790008e-01 7.89377213e-01 -5.37842393e-01 -5.05442202e-01 -2.22626746e-01 -1.10066891e-01 -2.52631847e-02 3.28714609e-01 1.56952709e-01 -2.62603253e-01 -7.13371634e-01 -1.23755470e-01 -4.85750996e-02 8.00895095e-01 3.53350937e-01 8.49260867e-01 -9.16889767e-05 -7.60128260e-01 6.10768318e-01 1.56515241e+00 -2.00429976e-01 3.72892261e-01 2.74702430e-01 1.03181577e+00 6.25799835e-01 5.30532658e-01 1.46582663e-01 4.83270586e-01 5.45311332e-01 2.29790732e-01 -3.20511162e-01 9.50512215e-02 7.66662136e-02 -1.98682398e-02 1.38164032e+00 -2.92402565e-01 1.17121369e-01 -1.11107218e+00 8.62167656e-01 -2.23327518e+00 -8.45647216e-01 -7.39097238e-01 2.31130838e+00 5.77949882e-01 -2.90851891e-01 5.74818313e-01 1.61079034e-01 9.31227803e-01 2.12745927e-02 -2.63622731e-01 -5.55909835e-02 -1.36997327e-01 -1.42525941e-01 6.19111359e-01 7.18590975e-01 -1.22800004e+00 9.36302304e-01 6.65186596e+00 9.25555289e-01 -7.76864171e-01 9.00545642e-02 4.39185679e-01 1.03400439e-01 -1.75529689e-01 1.04149636e-02 -3.31943184e-01 2.92288214e-01 7.63230979e-01 -4.24844682e-01 5.80071688e-01 6.02473676e-01 2.20981807e-01 -3.04062247e-01 -8.66742253e-01 9.73387837e-01 7.78719187e-02 -1.15348768e+00 -1.35142550e-01 3.27363104e-01 1.20392716e+00 5.20752929e-02 7.78701529e-02 -3.84948492e-01 4.05102134e-01 -7.76193619e-01 2.01416388e-01 1.71933919e-01 5.72698832e-01 -1.09422278e+00 2.92561442e-01 3.08733791e-01 -1.64859331e+00 5.99832833e-02 -4.52805072e-01 3.38992000e-01 1.28334835e-01 6.64263606e-01 -5.75923681e-01 8.54385197e-01 7.07919359e-01 8.88436258e-01 -4.67570633e-01 1.34225357e+00 8.54652300e-02 5.65873623e-01 -6.47192180e-01 1.63681246e-02 5.34878612e-01 -1.02960324e+00 5.56384444e-01 1.19939744e+00 4.99285102e-01 1.35913417e-01 5.62780976e-01 5.67325771e-01 -1.56805530e-01 6.92055106e-01 -5.87383449e-01 2.83345908e-01 3.43385309e-01 1.55485952e+00 -1.44770002e+00 -3.45822126e-01 -6.49817586e-01 1.02483213e+00 5.98820329e-01 4.67751265e-01 -6.17685735e-01 -3.40688139e-01 3.12034935e-01 7.86169767e-02 4.88438398e-01 -6.77670419e-01 -2.42236137e-01 -1.06764436e+00 -1.17087699e-01 -4.82371658e-01 7.45339274e-01 -3.01475972e-01 -1.29797482e+00 4.90404606e-01 -2.17358265e-02 -1.24248910e+00 1.45204484e-01 -2.10645124e-01 -4.11299109e-01 5.89082897e-01 -1.10778451e+00 -9.86263752e-01 -1.59067467e-01 9.72101450e-01 2.34761119e-01 1.19480319e-01 5.39501548e-01 3.93085480e-01 -4.85023350e-01 -3.25167887e-02 8.22628915e-01 8.47097412e-02 6.61207795e-01 -1.55574417e+00 4.64525223e-02 8.03420722e-01 3.46014231e-01 6.94816649e-01 4.45955068e-01 -7.86107957e-01 -1.69054401e+00 -9.74645972e-01 4.79026407e-01 -4.37535435e-01 1.08414876e+00 -3.74599457e-01 -1.02146065e+00 5.94602168e-01 1.81519076e-01 2.10394859e-02 7.18794763e-01 6.62116855e-02 -1.42546028e-01 -9.52114314e-02 -1.02896357e+00 6.84014678e-01 1.01640594e+00 -1.38689488e-01 -4.37463790e-01 5.66115737e-01 2.71819919e-01 -2.38774538e-01 -9.22676980e-01 1.21876024e-01 3.31159011e-02 -8.03931773e-01 9.02413309e-01 -1.93280742e-01 -1.63962901e-01 -6.37279451e-01 3.72431874e-02 -1.51234555e+00 -6.44013047e-01 -7.75089264e-01 1.20326169e-01 1.10288930e+00 3.90982807e-01 -6.08408511e-01 8.68253052e-01 4.22635764e-01 1.41154667e-02 -6.66327596e-01 -1.18413866e+00 -8.32340419e-01 1.53145686e-01 -1.59599975e-01 2.69646019e-01 1.06436777e+00 2.79615253e-01 3.03551346e-01 -2.06282094e-01 2.69039750e-01 9.72629130e-01 4.20049489e-01 7.10596144e-01 -1.45394254e+00 1.12118594e-01 -4.79906648e-01 -4.41503733e-01 -1.02519429e+00 2.68576652e-01 -1.25899303e+00 -5.09112999e-02 -1.66291797e+00 3.50275934e-01 -7.60724366e-01 -1.39032573e-01 1.47110686e-01 -1.67833313e-01 6.36939824e-01 1.23890266e-01 6.98036432e-01 -8.84839952e-01 5.04850566e-01 1.05288577e+00 -2.71982960e-02 -5.21812677e-01 -1.93806708e-01 -3.42644423e-01 6.68475211e-01 4.34112877e-01 -3.29576135e-01 -3.78503561e-01 -1.22585461e-01 -8.63637477e-02 -2.45291457e-01 -1.04859084e-01 -9.72901404e-01 5.18631458e-01 -1.05146796e-01 -9.22352914e-03 -5.57322323e-01 8.64968598e-02 -1.26266360e+00 2.16204435e-01 3.87105912e-01 1.38338720e-02 2.48317480e-01 -7.30055422e-02 9.44302678e-01 -2.56228536e-01 -3.22780311e-02 8.95690739e-01 1.52715996e-01 -4.10774380e-01 5.11412799e-01 -1.68552145e-01 -2.85012443e-02 1.55888009e+00 -1.50937587e-01 1.33357998e-02 -1.31728679e-01 -9.49128985e-01 4.34269637e-01 1.03604376e+00 5.12262657e-02 4.94474977e-01 -1.59736466e+00 -6.20263219e-01 -6.97880313e-02 -1.18477032e-01 1.10080436e-01 9.12022777e-03 1.31085503e+00 -6.63409770e-01 1.53850690e-01 2.74593979e-01 -9.27820265e-01 -1.43488908e+00 9.64922488e-01 -2.65901122e-04 -9.04698744e-02 -8.85829449e-01 3.96012217e-01 2.00335607e-02 -2.56147176e-01 1.19604789e-01 -4.04590443e-02 -5.70914410e-02 1.66975722e-01 1.71077967e-01 8.55461001e-01 -9.09365267e-02 -9.88875329e-01 -6.08346939e-01 1.12709606e+00 2.40581468e-01 -3.12146693e-01 1.36196363e+00 -3.69274884e-01 -5.10174692e-01 4.13221180e-01 1.51959741e+00 4.34873492e-01 -9.12180662e-01 -1.52232051e-01 4.34324086e-01 -5.22640526e-01 3.02732773e-02 -2.84400228e-02 -1.29902625e+00 4.82934266e-01 2.89130658e-01 7.30274796e-01 1.15365720e+00 3.42570484e-01 7.81145990e-01 2.75496155e-01 4.23426181e-01 -1.30373013e+00 -2.23413557e-02 2.24160895e-01 6.02987051e-01 -1.07877862e+00 1.63507253e-01 -8.61567616e-01 -4.96109307e-01 1.06800401e+00 9.73247290e-02 -4.39940929e-01 9.87294257e-01 -7.74312466e-02 6.51471224e-03 -3.40457588e-01 -2.99449027e-01 -4.12041754e-01 4.96169478e-01 4.97593552e-01 2.41478115e-01 9.09158066e-02 -6.35459304e-01 -1.25134245e-01 5.99136576e-02 -4.52268660e-01 3.89870346e-01 6.75940394e-01 -4.66931373e-01 -1.14179564e+00 -6.67402029e-01 3.98998767e-01 -1.10106312e-01 1.25207677e-01 -4.98856336e-01 6.96753383e-01 -2.60879159e-01 9.97600496e-01 3.55532765e-02 -1.87014818e-01 1.96266726e-01 -1.12826377e-01 2.74637341e-01 -5.81743121e-01 -3.46510142e-01 5.84534943e-01 -2.33149648e-01 -5.95124722e-01 -8.30002129e-01 -9.98908758e-01 -1.57752323e+00 -2.63660491e-01 -3.62540841e-01 6.59479320e-01 5.48458099e-01 7.14206219e-01 5.03977180e-01 -2.50337142e-02 9.50585008e-01 -9.55156684e-01 -1.43459797e-01 -6.21232092e-01 -8.88820291e-01 7.78353751e-01 8.30141380e-02 -7.92402565e-01 -6.89800262e-01 4.12576869e-02]
[7.5108323097229, 4.702158451080322]
d623c232-8f80-4e4d-a3e4-9ad484c38ea4
do-multi-document-summarization-models
2301.13844
null
https://arxiv.org/abs/2301.13844v1
https://arxiv.org/pdf/2301.13844v1.pdf
Do Multi-Document Summarization Models Synthesize?
Multi-document summarization entails producing concise synopses of collections of inputs. For some applications, the synopsis should accurately \emph{synthesize} inputs with respect to a key property or aspect. For example, a synopsis of film reviews all written about a particular movie should reflect the average critic consensus. As a more consequential example, consider narrative summaries that accompany biomedical \emph{systematic reviews} of clinical trial results. These narratives should fairly summarize the potentially conflicting results from individual trials. In this paper we ask: To what extent do modern multi-document summarization models implicitly perform this type of synthesis? To assess this we perform a suite of experiments that probe the degree to which conditional generation models trained for summarization using standard methods yield outputs that appropriately synthesize inputs. We find that existing models do partially perform synthesis, but do so imperfectly. In particular, they are over-sensitive to changes in input ordering and under-sensitive to changes in input compositions (e.g., the ratio of positive to negative movie reviews). We propose a simple, general method for improving model synthesis capabilities by generating an explicitly diverse set of candidate outputs, and then selecting from these the string best aligned with the expected aggregate measure for the inputs, or \emph{abstaining} when the model produces no good candidate. This approach improves model synthesis performance. We hope highlighting the need for synthesis (in some summarization settings), motivates further research into multi-document summarization methods and learning objectives that explicitly account for the need to synthesize.
['Byron C. Wallace', 'Iain J. Marshall', 'Stephanie C. Martinez', 'Jay DeYoung']
2023-01-31
null
null
null
null
['document-summarization']
['natural-language-processing']
[ 8.13222587e-01 4.01439786e-01 -5.23684323e-01 -4.40736145e-01 -1.22027194e+00 -9.79312778e-01 7.70741165e-01 4.65719074e-01 -1.85867473e-01 1.12185550e+00 8.41759026e-01 -3.46585602e-01 -1.02798402e-01 -5.34923136e-01 -7.57670164e-01 -3.72140050e-01 4.82711166e-01 3.79611611e-01 -1.91892505e-01 -1.19434431e-01 6.09252095e-01 2.05829516e-01 -1.12942147e+00 8.37596416e-01 1.03335929e+00 4.04355884e-01 -1.36042416e-01 9.86930072e-01 1.90947458e-01 8.19658041e-01 -1.13182771e+00 -7.38171399e-01 -5.20735160e-02 -1.07843542e+00 -6.03734434e-01 1.75712124e-01 5.48337162e-01 -3.27947199e-01 1.32423908e-01 1.08797979e+00 4.37019736e-01 -1.65169146e-02 1.06566465e+00 -9.56946373e-01 -5.93698919e-01 1.10107243e+00 -3.23864043e-01 4.50336844e-01 3.90946448e-01 3.88029784e-01 1.56148100e+00 -4.66420829e-01 7.20519304e-01 1.15235591e+00 4.78915006e-01 5.18757939e-01 -1.41136360e+00 -4.49546218e-01 3.78652185e-01 -4.42675561e-01 -7.06222177e-01 -6.02832377e-01 6.72900796e-01 -3.76217633e-01 1.05943513e+00 6.95787072e-01 4.61317509e-01 1.12857223e+00 6.83819354e-01 6.63811207e-01 9.55353379e-01 -1.47707686e-01 1.49348795e-01 3.56914699e-01 8.13992247e-02 1.71253175e-01 8.40020120e-01 -3.34429532e-01 -5.54014385e-01 -3.95283073e-01 2.99247116e-01 -4.61724013e-01 -1.90196350e-01 3.74237150e-01 -1.22718561e+00 8.65935504e-01 -1.29565895e-01 2.24627391e-01 -3.33738148e-01 1.13044493e-01 5.50406158e-01 2.70587325e-01 5.92914402e-01 1.32841456e+00 -3.87579322e-01 -9.64410696e-03 -1.40847468e+00 8.15430760e-01 9.60971355e-01 7.45653272e-01 3.58851641e-01 2.57464439e-01 -4.88933176e-01 6.74197257e-01 -9.59620327e-02 3.37009758e-01 4.13332999e-01 -1.19309688e+00 7.88174272e-01 3.89424503e-01 2.86513358e-01 -1.03697383e+00 -4.13834006e-01 -4.76719826e-01 -6.18961096e-01 -2.35199705e-01 2.51860589e-01 -5.41905642e-01 -5.69887519e-01 1.77399576e+00 -2.64760882e-01 -4.26875174e-01 3.17543924e-01 5.62730312e-01 8.47551584e-01 8.07459593e-01 8.50405842e-02 -8.39553237e-01 1.17226279e+00 -7.35330462e-01 -8.61170471e-01 -5.25221825e-01 5.90244174e-01 -8.94795358e-01 9.36189115e-01 4.96268988e-01 -1.60853791e+00 -1.96653560e-01 -1.33212149e+00 9.41900015e-02 1.13531686e-01 1.57700330e-01 3.29663575e-01 4.44543630e-01 -8.15703690e-01 8.80992115e-01 -5.13766348e-01 -9.87348780e-02 2.81323224e-01 3.10342997e-01 -7.80409276e-02 3.12687904e-01 -1.21037376e+00 1.14895785e+00 2.78812259e-01 -1.75729275e-01 -4.15261209e-01 -6.69277191e-01 -6.46325469e-01 1.58898428e-01 3.25902492e-01 -9.70544457e-01 1.56687617e+00 -1.27096951e+00 -1.23939276e+00 5.25902390e-01 -1.17099345e-01 -5.85334361e-01 4.65655535e-01 -8.13738778e-02 -3.00606102e-01 1.70998499e-01 1.53748021e-01 5.10313451e-01 6.72076225e-01 -1.22531855e+00 -4.75344032e-01 -1.43637553e-01 4.24168110e-02 4.24095929e-01 -1.40765414e-01 1.85489655e-01 7.40296254e-03 -8.63190472e-01 -3.85082901e-01 -7.67697096e-01 -3.75694811e-01 -5.64025223e-01 -7.78016806e-01 -1.71986163e-01 1.20447889e-01 -6.57990694e-01 1.73147118e+00 -1.63793349e+00 2.10778460e-01 8.78581703e-02 -6.98324144e-02 9.94801670e-02 -1.72628716e-01 7.02121496e-01 -2.86078658e-02 6.85943723e-01 -4.29610819e-01 -2.39707068e-01 -1.53963327e-01 -2.45804735e-03 -5.22563398e-01 2.25368053e-01 6.17764890e-01 7.75034964e-01 -8.64960313e-01 -5.29811800e-01 -3.66932899e-01 -8.37853029e-02 -8.39307725e-01 -7.56951123e-02 -6.13981187e-01 9.33370590e-02 -3.93438637e-01 2.61612326e-01 -8.84471368e-03 -4.40393358e-01 3.96476716e-01 -2.53484756e-01 -5.99927306e-02 7.12652147e-01 -1.04249847e+00 1.11995375e+00 -2.37324342e-01 6.63269520e-01 -2.51941204e-01 -8.40380728e-01 8.34499538e-01 3.07607710e-01 3.45798135e-01 -2.29889303e-01 1.04672238e-01 3.64984125e-01 3.37573230e-01 -3.05672854e-01 9.89324450e-01 -4.94762182e-01 -2.16061324e-01 8.41926396e-01 -1.56182215e-01 -7.36909211e-01 4.77512568e-01 5.50559700e-01 1.03582633e+00 -2.56937236e-01 7.06050098e-01 -1.13210715e-01 1.14003427e-01 3.24682474e-01 5.27297199e-01 9.66515362e-01 1.98855668e-01 1.04793465e+00 1.11205482e+00 4.79171015e-02 -1.36784804e+00 -7.97914863e-01 7.72505030e-02 6.53625131e-01 -3.52787018e-01 -7.36330986e-01 -7.58734763e-01 -5.89158297e-01 -1.74221575e-01 1.20578325e+00 -5.31973302e-01 -3.53896111e-01 -6.31847203e-01 -9.48261440e-01 6.26280665e-01 4.43963677e-01 -1.55421495e-01 -1.17005026e+00 -8.02356064e-01 4.20205861e-01 -2.91329324e-01 -7.82963395e-01 -5.58664083e-01 1.45270407e-01 -9.49238300e-01 -9.60816205e-01 -6.89183950e-01 -3.70311260e-01 6.72857165e-01 -1.69300556e-01 1.11831844e+00 -1.13653816e-01 3.22823554e-01 1.66573510e-01 -2.25016594e-01 -5.41777968e-01 -1.01897156e+00 2.22925603e-01 -2.04158388e-02 -3.24869394e-01 -4.14575227e-02 -4.34285045e-01 -3.50074470e-01 -1.49030581e-01 -1.24762237e+00 3.02314967e-01 7.29033470e-01 8.02301586e-01 4.06436980e-01 -2.28589818e-01 1.21647656e+00 -1.46609604e+00 1.41848683e+00 -6.03076160e-01 -1.16577707e-01 4.01374549e-01 -7.77604043e-01 2.04384431e-01 8.37210715e-01 -5.50196648e-01 -9.70526457e-01 -2.59491175e-01 1.36323757e-02 8.03942904e-02 2.38399118e-01 8.00180674e-01 7.54704103e-02 8.05958390e-01 1.06419575e+00 1.18456008e-02 2.72832047e-02 8.09109490e-03 4.19392228e-01 5.75451374e-01 4.15948570e-01 -5.98570228e-01 3.96386087e-01 7.18017444e-02 -2.99288392e-01 -5.90894580e-01 -8.31798077e-01 1.35750705e-02 -6.02955893e-02 -1.49689019e-01 5.77534735e-01 -7.51605034e-01 -1.59394443e-01 -5.59009761e-02 -1.43715990e+00 -3.47166210e-01 -4.64843690e-01 3.79625261e-01 -7.12711632e-01 2.27049112e-01 -4.51125503e-01 -6.95379496e-01 -6.24121308e-01 -1.13137734e+00 8.57660711e-01 3.55222374e-01 -1.08562970e+00 -7.85400271e-01 1.28592461e-01 1.40277222e-01 1.92961171e-01 3.47533584e-01 1.23209107e+00 -1.14907026e+00 -1.41119450e-01 -1.79766968e-01 1.77822620e-01 6.61162913e-01 3.65615904e-01 5.11652350e-01 -5.87719023e-01 -6.25957677e-04 -2.89236642e-02 -2.67194688e-01 8.76896024e-01 6.82319224e-01 8.53938937e-01 -9.35648024e-01 -1.56083420e-01 -1.01014405e-01 1.16174817e+00 2.48821497e-01 3.27900708e-01 1.63593646e-02 3.13073844e-01 8.66998553e-01 3.86029035e-01 5.27373075e-01 1.50828242e-01 3.94801319e-01 -9.58067849e-02 2.28595346e-01 1.52268773e-02 -2.35926464e-01 5.71612418e-01 8.31647158e-01 2.81542957e-01 -5.81899881e-01 -5.30876398e-01 7.58126080e-01 -1.60922420e+00 -1.19960475e+00 -1.30723137e-02 1.89127135e+00 1.27290213e+00 6.36503875e-01 3.30722064e-01 6.72383457e-02 5.41849256e-01 5.90365410e-01 -6.81525409e-01 -1.02179074e+00 -4.10125345e-01 2.58495927e-01 4.09410954e-01 5.70989192e-01 -5.50257146e-01 6.78048730e-01 6.64610338e+00 5.77183187e-01 -8.51241469e-01 -3.09507281e-01 8.65402222e-01 -4.87997532e-01 -1.07754815e+00 1.10734962e-01 -6.85942948e-01 3.29854310e-01 1.09960151e+00 -7.85112858e-01 4.30588499e-02 5.34043133e-01 4.75364476e-01 -3.31662387e-01 -1.31838608e+00 2.57985920e-01 2.62950003e-01 -1.74533117e+00 4.72547680e-01 -1.69896170e-01 1.28737724e+00 -4.78289396e-01 8.78090560e-02 3.39357369e-03 4.42486823e-01 -1.11405373e+00 8.04730356e-01 4.91573334e-01 6.35943115e-01 -6.73187196e-01 5.12057006e-01 3.84747237e-01 -5.19877732e-01 -6.32864684e-02 -8.03745463e-02 -1.91039741e-02 4.23166752e-01 5.48893869e-01 -9.52995896e-01 5.21419287e-01 1.66398752e-02 6.42426372e-01 -4.93550330e-01 7.34427154e-01 -2.59482682e-01 8.58435333e-01 -7.15186149e-02 -4.20709163e-01 2.00790301e-01 4.20293920e-02 7.64323294e-01 1.39743459e+00 2.04518571e-01 2.20531970e-01 -2.39983052e-01 7.74316430e-01 -1.88631222e-01 1.69939131e-01 -7.54656911e-01 -4.03448552e-01 4.33847636e-01 9.73085761e-01 -6.36555851e-01 -6.24471247e-01 -2.57158220e-01 5.07729471e-01 -8.46438259e-02 4.62906957e-01 -5.18103898e-01 -1.92338049e-01 3.53255510e-01 1.18844509e-01 4.18609641e-02 1.65420905e-01 -8.94283950e-01 -1.04295480e+00 3.44937630e-02 -1.45293355e+00 4.19820160e-01 -8.42431664e-01 -1.22960305e+00 5.14263153e-01 2.63281792e-01 -1.15944898e+00 -6.21578991e-01 -1.19547367e-01 -8.62930119e-01 7.20341563e-01 -1.04144645e+00 -5.66001296e-01 3.82972986e-01 -9.47292969e-02 9.25378859e-01 4.97379713e-02 4.91993040e-01 -2.16148332e-01 -3.70133132e-01 5.22400796e-01 -1.34966463e-01 -2.83967048e-01 9.90902543e-01 -1.25237715e+00 5.01088381e-01 8.81484210e-01 1.42357260e-01 7.69964814e-01 1.37828028e+00 -9.48556244e-01 -9.29809570e-01 -9.60907936e-01 1.11393774e+00 -5.64132571e-01 5.58965623e-01 2.06214115e-01 -7.56792128e-01 7.39858329e-01 4.18590307e-01 -8.97108018e-01 7.41025090e-01 -5.21764830e-02 -2.04517916e-01 -4.54627350e-02 -8.73155177e-01 9.89203990e-01 6.09384179e-01 -3.32872003e-01 -9.05120075e-01 4.57294226e-01 8.27412605e-01 -3.29682827e-01 -7.39265025e-01 4.47056502e-01 5.12902260e-01 -7.46145606e-01 5.47397435e-01 -1.02499902e+00 1.35379791e+00 -3.48179862e-02 -8.72401670e-02 -1.69022179e+00 -1.12347655e-01 -7.28219032e-01 3.40460688e-02 1.23322427e+00 1.02532160e+00 -2.39619225e-01 4.84730840e-01 6.91238701e-01 -3.19295049e-01 -9.93284345e-01 -5.31651795e-01 -4.27849501e-01 5.15711486e-01 -2.20316231e-01 2.86308080e-01 7.35271275e-01 3.26871634e-01 6.70111656e-01 -4.44381326e-01 -2.18997806e-01 1.74377561e-01 -4.51255515e-02 6.43181741e-01 -8.73624265e-01 -2.84745157e-01 -8.51439059e-01 2.22064286e-01 -7.79862404e-01 1.21717401e-01 -8.12752068e-01 3.60288145e-03 -1.96185398e+00 2.71647722e-01 -5.32312654e-02 9.31947082e-02 2.32769206e-01 -5.02402365e-01 -3.86063494e-02 4.00771618e-01 1.70990616e-01 -4.50763226e-01 1.59568161e-01 1.33856773e+00 -1.61436960e-01 -3.52275401e-01 2.64799416e-01 -1.48507059e+00 4.45820540e-01 9.21390653e-01 -4.99135822e-01 -5.66120923e-01 -2.05427513e-01 7.24910021e-01 3.42414230e-01 5.47561571e-02 -5.69623828e-01 2.21753627e-01 -4.80652601e-01 2.71880478e-01 -5.08956790e-01 9.63738114e-02 -2.06872821e-01 3.28914315e-01 2.92740077e-01 -9.90242183e-01 5.58812380e-01 2.68071443e-01 3.84360045e-01 -1.11282840e-01 -3.66372466e-01 5.85861862e-01 -3.57403874e-01 1.94808573e-01 -2.96508431e-01 -6.90906286e-01 4.50197667e-01 6.11505747e-01 -1.32587448e-01 -6.26500666e-01 -5.97878098e-01 -3.58369529e-01 2.28620380e-01 4.04018462e-01 3.34410399e-01 4.17162001e-01 -9.90598559e-01 -1.06251740e+00 -2.73760319e-01 -1.27659336e-01 -1.21124327e-01 6.64535537e-02 7.62585759e-01 -2.37450257e-01 4.78488296e-01 -5.21321893e-02 -8.63209963e-02 -1.03900206e+00 1.42002553e-01 2.08009571e-01 -6.10072255e-01 -3.12933922e-01 5.43025017e-01 4.53938469e-02 1.56239485e-02 -7.28716478e-02 -6.60899937e-01 -2.32609257e-01 5.18039465e-01 4.83487070e-01 2.65711248e-01 7.28017837e-02 -3.56727123e-01 -1.90531552e-01 1.47504359e-01 -3.29350352e-01 -6.17443323e-01 1.27398360e+00 1.83265999e-01 1.02223337e-01 6.83261395e-01 1.01939273e+00 2.90666819e-01 -1.16606534e+00 1.18088327e-01 -7.98239186e-02 2.29608733e-02 -4.67788756e-01 -1.02517307e+00 -5.96459508e-01 4.88097370e-01 -5.55430293e-01 3.36827844e-01 9.91106272e-01 -2.45998800e-02 4.76707250e-01 3.08394074e-01 -2.75930673e-01 -1.16504192e+00 2.31621161e-01 1.51198223e-01 1.13802671e+00 -8.32198679e-01 6.88038468e-01 7.03884363e-02 -1.19469178e+00 1.04599285e+00 5.14234066e-01 -2.92790115e-01 7.32168034e-02 1.58735618e-01 -7.83446804e-02 -2.49935389e-01 -1.26896417e+00 4.09584403e-01 4.02526468e-01 5.98607846e-02 5.48520982e-01 1.38599008e-01 -9.37734067e-01 8.23622167e-01 -5.37991345e-01 -3.18024576e-01 1.06186104e+00 6.07938826e-01 -4.14454401e-01 -9.14104939e-01 -2.25656807e-01 1.02276528e+00 -7.11693704e-01 -2.07032770e-01 -8.57881784e-01 6.99305594e-01 -4.56967771e-01 1.14982915e+00 -1.06970198e-01 -2.71573186e-01 4.44184363e-01 1.32063538e-01 2.67439663e-01 -9.16737199e-01 -1.08529472e+00 1.48502544e-01 6.60811305e-01 6.16061874e-02 -4.22545105e-01 -9.91602361e-01 -1.14412796e+00 -1.36341915e-01 -3.46459299e-01 2.19757423e-01 4.47750837e-01 1.05920470e+00 1.09694675e-01 6.87127709e-01 5.42008400e-01 -5.03692329e-01 -9.42905724e-01 -9.84756589e-01 -1.45529106e-01 3.03909630e-01 4.61053014e-01 -7.14390576e-02 -5.25930464e-01 2.11793259e-01]
[12.255932807922363, 9.38272476196289]
08860c5e-6199-408d-8498-f1073d9077c8
tbn-vit-temporal-bilateral-network-with
2112.01033
null
https://arxiv.org/abs/2112.01033v1
https://arxiv.org/pdf/2112.01033v1.pdf
TBN-ViT: Temporal Bilateral Network with Vision Transformer for Video Scene Parsing
Video scene parsing in the wild with diverse scenarios is a challenging and great significance task, especially with the rapid development of automatic driving technique. The dataset Video Scene Parsing in the Wild(VSPW) contains well-trimmed long-temporal, dense annotation and high resolution clips. Based on VSPW, we design a Temporal Bilateral Network with Vision Transformer. We first design a spatial path with convolutions to generate low level features which can preserve the spatial information. Meanwhile, a context path with vision transformer is employed to obtain sufficient context information. Furthermore, a temporal context module is designed to harness the inter-frames contextual information. Finally, the proposed method can achieve the mean intersection over union(mIoU) of 49.85\% for the VSPW2021 Challenge test dataset.
['Hongbin Wang', 'Leilei Cao', 'Bo Yan']
2021-12-02
null
null
null
null
['scene-parsing']
['computer-vision']
[ 1.58613473e-01 -1.41282022e-01 5.89223318e-02 -6.25602365e-01 -6.36609435e-01 -4.23760146e-01 5.05763948e-01 -4.80012089e-01 -4.99276072e-01 4.82872576e-01 3.82029057e-01 -2.06150785e-01 8.90311077e-02 -7.29133129e-01 -1.03263283e+00 -5.75678110e-01 1.52683705e-01 -4.66303617e-01 8.30434501e-01 -1.55217066e-01 -2.22728141e-02 1.37984648e-01 -1.53578913e+00 4.25017595e-01 7.75417387e-01 1.40335464e+00 7.22743273e-01 5.04511178e-01 -2.51554430e-01 8.85649741e-01 -1.97429821e-01 -4.48005825e-01 4.00258064e-01 -1.45438537e-01 -6.31182730e-01 -1.20842114e-01 7.93314993e-01 -3.73610616e-01 -6.55361056e-01 1.02779984e+00 4.06719267e-01 1.43716350e-01 -1.83229700e-01 -1.15495431e+00 -2.33963162e-01 4.59054351e-01 -7.36074209e-01 4.06757891e-01 2.76181132e-01 4.13345397e-01 7.35854030e-01 -8.17167819e-01 8.23693097e-01 1.10228968e+00 6.95568442e-01 2.88708806e-01 -7.57857382e-01 -9.88900006e-01 4.86405194e-01 5.68891287e-01 -9.13345158e-01 -3.90051752e-01 9.03583944e-01 -4.14603621e-01 8.04291368e-01 1.43513307e-01 7.77374148e-01 1.35581481e+00 2.65779078e-01 7.21991479e-01 6.69632852e-01 7.25405216e-02 2.64321212e-02 -3.81308347e-01 7.24610463e-02 6.19900882e-01 -9.54732746e-02 2.04805925e-01 -8.18606496e-01 5.81462204e-01 5.30977488e-01 9.07990113e-02 -1.64558008e-01 -5.14173135e-02 -1.27825522e+00 3.41947407e-01 5.48765719e-01 1.48729101e-01 -2.01420933e-01 3.12330812e-01 6.23707533e-01 -1.43425435e-01 1.79081932e-01 -1.02404229e-01 -4.34510857e-01 -3.02165300e-01 -9.79316056e-01 2.08062232e-01 2.28057861e-01 1.42778230e+00 5.75676262e-01 2.31134146e-01 -5.54418147e-01 5.90522110e-01 1.23908140e-01 4.26440030e-01 1.57781765e-01 -1.35264933e+00 9.35055137e-01 4.63031441e-01 -1.22890007e-02 -1.07876337e+00 -4.61636901e-01 -3.70976359e-01 -9.23361480e-01 2.06727069e-02 3.48504752e-01 -1.80603847e-01 -9.23929334e-01 1.98921978e+00 3.29066008e-01 6.42169833e-01 -4.62605916e-02 9.31190968e-01 9.11158323e-01 7.88989961e-01 2.80270875e-01 -1.80945069e-01 1.45561397e+00 -9.90618527e-01 -7.96144903e-01 -4.06582803e-01 2.84548163e-01 -6.22020543e-01 1.18947220e+00 3.04101199e-01 -9.64989483e-01 -8.75196218e-01 -1.13595855e+00 -4.35248464e-01 -9.72891673e-02 1.58972472e-01 6.07148886e-01 1.26277104e-01 -8.81300747e-01 3.45745951e-01 -6.37721658e-01 -2.68195331e-01 5.43570578e-01 -5.99149661e-03 -6.32859647e-01 -4.59128648e-01 -1.26118505e+00 4.40878868e-01 5.23585618e-01 6.07976019e-01 -8.44488025e-01 -7.56983519e-01 -1.09728968e+00 -8.66589099e-02 4.85779583e-01 -5.24395108e-01 9.69379306e-01 -7.60954142e-01 -1.22414219e+00 5.86120963e-01 -1.73185319e-01 -6.14720285e-01 8.38850379e-01 -4.48759019e-01 -4.06244218e-01 2.57883728e-01 3.98977846e-01 7.87954748e-01 6.35562479e-01 -8.65323603e-01 -1.04405606e+00 -7.45069981e-02 1.08915813e-01 7.95685649e-02 -2.34610468e-01 8.13942589e-03 -9.98653293e-01 -6.90580606e-01 9.48645771e-02 -5.96888304e-01 -1.37824342e-01 7.31682172e-03 -4.12424743e-01 3.27422693e-02 1.13870013e+00 -1.00644827e+00 1.22225654e+00 -2.49151349e+00 -1.04055554e-02 -2.49880880e-01 1.98210239e-01 2.81466246e-01 -1.39900297e-01 -3.44409756e-02 -1.97783895e-02 -1.61440760e-01 -3.67244154e-01 -4.15994078e-01 -8.59703049e-02 2.76220381e-01 -4.93786871e-01 2.43780091e-01 3.49980175e-01 9.09738839e-01 -8.63304257e-01 -6.64547741e-01 5.03893912e-01 3.85782599e-01 -7.96862185e-01 2.27319375e-01 -1.96265012e-01 4.88649398e-01 -5.02654254e-01 5.93418717e-01 9.04545844e-01 8.65235999e-02 -1.54274568e-01 -5.62787771e-01 -5.87373078e-01 -1.82535827e-01 -9.58348930e-01 2.21127272e+00 -2.25750864e-01 7.51767159e-01 2.68165141e-01 -8.06444883e-01 7.09600151e-01 -8.07798002e-03 5.59364974e-01 -1.20400906e+00 1.40381083e-01 -2.10379567e-02 -3.91433537e-01 -7.23117113e-01 6.01312876e-01 2.68370420e-01 -2.55029231e-01 -3.50062490e-01 6.46995082e-02 2.31333911e-01 2.94956923e-01 2.37555325e-01 1.36817086e+00 4.89652067e-01 -3.18994761e-01 -2.04914048e-01 4.97411311e-01 -1.66283525e-03 1.31739533e+00 3.23213220e-01 -4.44276989e-01 9.47757602e-01 5.70400953e-01 -5.34702659e-01 -9.42290664e-01 -9.84252155e-01 -8.69600475e-02 7.94163048e-01 5.67082286e-01 -3.64348978e-01 -7.57439792e-01 -5.08595645e-01 -3.66729558e-01 5.35816908e-01 -5.69855809e-01 -2.44147077e-01 -9.68746841e-01 -2.84363955e-01 4.42259490e-01 8.90748978e-01 1.28980148e+00 -1.12073898e+00 -8.52331817e-01 2.47541785e-01 -4.70616788e-01 -1.67832661e+00 -7.74775743e-01 1.20842464e-01 -4.46148694e-01 -1.20245385e+00 -4.17134732e-01 -9.26269531e-01 1.63139343e-01 3.54246199e-01 1.12291551e+00 -3.02495271e-01 -2.62782782e-01 4.96418029e-02 -4.12600935e-01 -8.05630535e-02 1.72414273e-01 -1.61301002e-01 -3.23115587e-01 7.04799220e-02 1.96904600e-01 -6.82498455e-01 -6.76476121e-01 3.30072910e-01 -8.33700836e-01 5.36024153e-01 4.69690144e-01 8.95810664e-01 8.11776638e-01 -3.44861336e-02 3.43976527e-01 -4.65346426e-01 -4.75636199e-02 -4.76938188e-01 -9.45499539e-01 1.90154523e-01 -3.09519414e-02 -2.21545011e-01 6.84945285e-01 -3.03485423e-01 -1.39470315e+00 2.42413774e-01 -1.88987508e-01 -8.01033378e-01 -1.11243524e-01 1.07762545e-01 -6.99070454e-01 2.65523463e-01 1.78443372e-01 1.62671924e-01 -4.06433344e-01 -4.01834965e-01 4.52262819e-01 2.69262969e-01 1.07887268e+00 -5.13812900e-01 6.69210911e-01 5.40077150e-01 -6.28629625e-02 -5.33987820e-01 -9.35858309e-01 -2.37304226e-01 -6.39778256e-01 -3.93488318e-01 1.50594306e+00 -1.17089224e+00 -6.91354394e-01 5.84962428e-01 -1.20515072e+00 -5.45968473e-01 -1.39110148e-01 3.65810007e-01 -4.28818613e-01 2.09421232e-01 -5.28838754e-01 -4.43033159e-01 -2.73486376e-01 -1.24881852e+00 1.08988571e+00 4.57039565e-01 2.41722971e-01 -4.14206505e-01 -3.30947936e-01 4.34395909e-01 2.75431812e-01 5.43204129e-01 6.12569928e-01 6.13591634e-03 -9.49210048e-01 1.66416213e-01 -6.90756083e-01 3.51355225e-01 -1.46126375e-01 -3.55049013e-03 -1.05973828e+00 8.69294107e-02 -1.71720609e-01 -2.00040210e-02 1.09940577e+00 5.72105348e-01 1.43503082e+00 7.68533200e-02 -2.61213303e-01 1.13886225e+00 1.28781509e+00 3.25419217e-01 7.74976254e-01 2.49899134e-01 1.10452068e+00 5.96865714e-01 9.37489271e-01 3.44841957e-01 5.71186423e-01 6.38985455e-01 5.60125053e-01 -1.44222394e-01 -4.02335107e-01 -5.42121470e-01 3.54462534e-01 6.06314182e-01 2.44958326e-01 -1.47786885e-01 -8.30105364e-01 5.97082138e-01 -2.01507354e+00 -1.04654109e+00 -2.28543997e-01 1.85344768e+00 5.05396426e-01 3.81912678e-01 -2.46840507e-01 -3.18137966e-02 8.34979594e-01 3.43258977e-01 -5.94861865e-01 -1.22917719e-01 -3.94151956e-01 -1.08515024e-01 6.09964371e-01 2.33280018e-01 -1.34605384e+00 1.03987789e+00 5.61453629e+00 8.75350952e-01 -1.09886730e+00 1.88117161e-01 6.59660578e-01 -1.23250142e-01 -2.34583512e-01 1.07787072e-03 -7.26446092e-01 8.20961475e-01 6.46975577e-01 1.71493724e-01 3.04791164e-02 7.67881036e-01 4.69004482e-01 -2.40547612e-01 -8.36016417e-01 1.11508250e+00 -2.10113332e-01 -1.44041717e+00 -3.21599066e-01 -1.09680757e-01 6.06971681e-01 1.06087610e-01 -1.26158576e-02 3.32811266e-01 1.63525969e-01 -7.50628531e-01 1.17615926e+00 6.43006444e-01 1.08589852e+00 -7.45122492e-01 6.64143980e-01 6.62263334e-02 -1.73106134e+00 -4.15238887e-01 -3.06030601e-01 1.31303012e-01 5.96070707e-01 6.78519428e-01 -5.49986139e-02 6.63527191e-01 1.23234129e+00 1.13415146e+00 -6.03855669e-01 1.06114018e+00 -4.63192500e-02 5.72799683e-01 -3.64046454e-01 5.35433829e-01 3.20210338e-01 -3.45600694e-01 4.37574744e-01 1.24748230e+00 3.05106461e-01 8.68195854e-03 2.24941835e-01 6.81633890e-01 -5.00043295e-02 -2.04739317e-01 -6.87446296e-01 4.41745818e-01 4.01965767e-01 1.25381088e+00 -6.03793740e-01 -2.31275201e-01 -5.18154800e-01 9.47930455e-01 1.35378495e-01 4.03584599e-01 -1.25933743e+00 -2.48042867e-01 9.04961288e-01 -7.61641040e-02 3.92392159e-01 -3.88324142e-01 -1.90302402e-01 -1.06332529e+00 4.77965444e-01 -4.17452991e-01 3.20618451e-01 -9.50195670e-01 -1.01197898e+00 6.75981879e-01 8.66961032e-02 -1.49633455e+00 1.83541492e-01 -5.14998674e-01 -9.06921744e-01 6.05613589e-01 -1.67893267e+00 -1.39997065e+00 -7.74933696e-01 7.07069397e-01 8.58100235e-01 -4.51892540e-02 1.76621899e-01 7.46008933e-01 -9.74785745e-01 5.27844965e-01 -3.50904524e-01 3.05670738e-01 6.68593705e-01 -1.00045764e+00 3.69874209e-01 1.38625312e+00 -2.73017138e-01 2.34920308e-01 5.40444195e-01 -6.61782205e-01 -1.46661603e+00 -1.59574568e+00 4.07706499e-01 -2.05443755e-01 4.23161775e-01 -4.62284148e-01 -7.75691628e-01 4.99722600e-01 2.04461187e-01 5.91077387e-01 -9.16869473e-03 -3.88606071e-01 -4.21354413e-01 -5.12165189e-01 -9.27305579e-01 6.00764871e-01 1.46909761e+00 -3.97270352e-01 -3.92632782e-01 6.54441640e-02 1.32466614e+00 -6.19856477e-01 -5.63110054e-01 6.00503862e-01 5.31326413e-01 -1.21943891e+00 9.79220629e-01 -1.62848383e-01 6.71086848e-01 -4.95679796e-01 -3.75403374e-01 -6.18257284e-01 -1.33248866e-01 -6.20039940e-01 1.82532996e-01 1.62361050e+00 5.22643328e-02 -2.59936422e-01 7.98294365e-01 6.79147243e-01 -6.20020211e-01 -6.22751772e-01 -1.10861170e+00 -5.63989878e-01 -1.58698350e-01 -8.15087259e-01 4.62945968e-01 6.01697624e-01 -4.10438716e-01 1.32632777e-01 -5.18971026e-01 7.48523474e-02 5.61299264e-01 1.47122547e-01 5.72497308e-01 -8.61785889e-01 -1.19388271e-02 -1.96126401e-01 -5.14475405e-01 -1.13525975e+00 2.28581250e-01 -6.29453659e-01 2.07020193e-01 -1.38860500e+00 7.96145424e-02 -3.61070275e-01 -3.74432564e-01 3.42755526e-01 -1.23383701e-01 1.41673774e-01 2.58200824e-01 -2.33706877e-01 -8.45719278e-01 7.22864032e-01 1.22273481e+00 -1.28439842e-02 1.38042837e-01 -3.32493722e-01 -2.95853198e-01 7.12639332e-01 6.91277564e-01 -1.53148279e-01 -5.33895493e-01 -7.52380729e-01 9.33940113e-02 2.27016792e-01 5.50980091e-01 -1.30512869e+00 2.57566124e-01 -2.79143572e-01 4.35243040e-01 -9.68185306e-01 3.76671314e-01 -9.45766807e-01 2.19648138e-01 1.59613505e-01 -9.61733311e-02 3.16371083e-01 3.44026178e-01 7.51340866e-01 -4.85977709e-01 3.76007050e-01 6.74639285e-01 1.13186851e-01 -1.35123932e+00 4.79992211e-01 4.49477732e-02 3.05376530e-01 1.18160081e+00 -3.61152112e-01 -2.92804807e-01 2.19668951e-02 -3.19154918e-01 7.03123033e-01 2.90745884e-01 6.95402026e-01 7.44600594e-01 -1.44128871e+00 -4.29515302e-01 3.57718468e-01 9.64258537e-02 4.20398384e-01 7.22456455e-01 7.73355305e-01 -6.12878084e-01 1.51556566e-01 -4.30783868e-01 -7.19707012e-01 -1.12151265e+00 5.06502926e-01 2.69272923e-01 6.87176809e-02 -1.15801370e+00 9.45637941e-01 5.39805830e-01 -3.56697813e-02 2.24657685e-01 -5.25543869e-01 -2.37596586e-01 9.03160498e-02 6.81105971e-01 1.29106686e-01 -1.14524238e-01 -7.46133626e-01 -3.34337980e-01 8.12434018e-01 3.45311999e-01 -2.66962573e-02 1.33341444e+00 -3.53338808e-01 8.38192701e-02 1.29246041e-01 1.21930432e+00 -3.38547304e-02 -1.95469034e+00 -1.92803130e-01 -9.71797481e-02 -5.58531463e-01 1.95723660e-02 -5.00350237e-01 -1.43002021e+00 1.10963380e+00 6.55906320e-01 -2.87708342e-01 1.31289339e+00 -2.29484126e-01 1.12158203e+00 3.20238583e-02 4.24754262e-01 -1.11645746e+00 1.68853924e-02 6.29513562e-01 9.43322361e-01 -1.26167166e+00 -4.35012728e-01 -5.97588241e-01 -6.68104291e-01 9.44226205e-01 1.04089177e+00 -2.99649518e-02 6.34026349e-01 4.89224106e-01 -8.82731900e-02 -1.00696199e-01 -7.09339917e-01 -2.78121501e-01 8.63349438e-02 5.44746876e-01 6.30970374e-02 -1.73362955e-01 -9.68202129e-02 9.71928656e-01 -2.05813512e-01 -6.36524484e-02 3.50041986e-01 7.24014938e-01 -2.66933262e-01 -6.48827732e-01 -4.73893695e-02 1.99827865e-01 -2.59614766e-01 -2.58672424e-02 2.55830526e-01 5.56657970e-01 5.81719518e-01 8.93083394e-01 1.42565474e-01 -7.92225659e-01 5.68940580e-01 -1.12473696e-01 5.14030382e-02 -2.49639004e-01 -4.70431179e-01 2.94701122e-02 1.26974031e-01 -1.29391646e+00 -2.62678653e-01 -5.50164163e-01 -1.30607164e+00 -1.39133990e-01 2.32570805e-02 -2.70609319e-01 4.98009413e-01 8.73472691e-01 4.24257636e-01 8.42886746e-01 3.50899696e-01 -8.50480378e-01 2.51371205e-01 -7.21886277e-01 -3.33199233e-01 4.18671727e-01 2.99738169e-01 -6.50562167e-01 -9.78744254e-02 2.56336927e-01]
[9.253369331359863, 0.02129863202571869]
eec49a7e-4b1e-48c4-8fdd-9dc9a45b1272
human-activity-behavioural-pattern
2306.13374
null
https://arxiv.org/abs/2306.13374v2
https://arxiv.org/pdf/2306.13374v2.pdf
Human Activity Behavioural Pattern Recognition in Smarthome with Long-hour Data Collection
The research on human activity recognition has provided novel solutions to many applications like healthcare, sports, and user profiling. Considering the complex nature of human activities, it is still challenging even after effective and efficient sensors are available. The existing works on human activity recognition using smartphone sensors focus on recognizing basic human activities like sitting, sleeping, standing, stair up and down and running. However, more than these basic activities is needed to analyze human behavioural pattern. The proposed framework recognizes basic human activities using deep learning models. Also, ambient sensors like PIR, pressure sensors, and smartphone-based sensors like accelerometers and gyroscopes are combined to make it hybrid-sensor-based human activity recognition. The hybrid approach helped derive more activities than the basic ones, which also helped derive human activity patterns or user profiling. User profiling provides sufficient information to identify daily living activity patterns and predict whether any anomaly exists. The framework provides the base for applications such as elderly monitoring when they are alone at home. The GRU model's accuracy of 95\% is observed to recognize the basic activities. Finally, Human activity patterns over time are recognized based on the duration and frequency of the activities. It is observed that human activity pattern, like, morning walking duration, varies depending on the day of the week.
['Geetha V', 'Ranjit Kolkar']
2023-06-23
null
null
null
null
['activity-recognition', 'human-activity-recognition', 'human-activity-recognition']
['computer-vision', 'computer-vision', 'time-series']
[ 1.56023145e-01 -2.49732003e-01 -4.62250262e-01 -2.20386878e-01 1.40647851e-02 4.34826650e-02 3.80595922e-01 1.92652881e-01 -4.89384055e-01 8.85014951e-01 5.75670004e-01 -1.89944059e-01 -8.82810205e-02 -9.68630195e-01 -1.40447840e-01 -7.55182266e-01 -2.50864267e-01 -1.31814957e-01 1.83082461e-01 -1.23605765e-01 3.55753973e-02 5.01740575e-01 -2.20044851e+00 1.64462194e-01 8.11604977e-01 1.02404833e+00 5.63652292e-02 7.26931870e-01 8.71737450e-02 8.15432191e-01 -8.34279895e-01 3.53087127e-01 -1.90789849e-01 -6.03315890e-01 -2.07928270e-01 1.08524896e-01 -3.30173165e-01 -3.30758572e-01 -2.71003097e-02 5.24864852e-01 5.57728469e-01 3.23903352e-01 5.86982012e-01 -1.23160338e+00 -1.23487599e-01 1.83183149e-01 -3.10812354e-01 3.94847870e-01 1.06489670e+00 -1.25157952e-01 2.99151003e-01 -4.07538176e-01 -2.74488717e-01 5.97453654e-01 7.67076552e-01 4.53345150e-01 -4.57866609e-01 -5.64141572e-01 -2.55440418e-02 6.13453627e-01 -1.28567100e+00 -3.40404809e-01 6.47848606e-01 -4.92575914e-01 1.16915047e+00 3.49151641e-01 1.31203711e+00 1.39133322e+00 3.77474576e-01 8.70155752e-01 8.17523360e-01 -5.23474813e-01 3.99461776e-01 -6.40977249e-02 4.96348023e-01 3.86361599e-01 6.14979148e-01 -2.66591519e-01 -4.29915845e-01 -1.92224070e-01 3.53348911e-01 6.97514892e-01 -3.02736133e-01 9.44218226e-03 -1.30016947e+00 2.60489136e-01 1.08241461e-01 8.38903606e-01 -5.57845056e-01 -4.08895761e-01 2.57962972e-01 -1.73996493e-01 1.54527023e-01 -4.25806455e-02 -4.29049462e-01 -9.08280790e-01 -9.62071240e-01 1.99898884e-01 1.04281700e+00 7.03939438e-01 6.55309916e-01 5.78600243e-02 -7.23737478e-02 7.86189020e-01 4.03911978e-01 6.74627721e-01 1.04442954e+00 -7.04331279e-01 3.16433161e-01 1.18309772e+00 4.18511540e-01 -1.10311639e+00 -9.87312019e-01 -1.34024099e-01 -1.11660254e+00 -9.67056826e-02 2.94979364e-01 -4.16670471e-01 -5.67222476e-01 1.33843803e+00 2.52405494e-01 2.16946334e-01 -4.63220589e-02 5.45575440e-01 6.28728211e-01 3.96917641e-01 2.14558139e-01 -3.99934441e-01 1.60987651e+00 -7.69875884e-01 -1.07872498e+00 -5.15186071e-01 8.45839739e-01 -2.67498866e-02 1.02927864e+00 3.26634705e-01 -5.70946872e-01 -6.42340124e-01 -1.41684711e+00 3.78859341e-01 -6.83307886e-01 1.48319155e-01 5.93400776e-01 1.33838356e+00 -6.37814939e-01 4.72522110e-01 -1.34007752e+00 -5.80983818e-01 3.17129016e-01 3.92853051e-01 -3.63018960e-01 3.23833853e-01 -1.20495212e+00 8.31335545e-01 4.52845544e-01 2.23679349e-01 -1.15597360e-01 -1.99562892e-01 -1.08050275e+00 -4.99913171e-02 -2.20526814e-01 -4.33692902e-01 8.62138450e-01 -6.45043492e-01 -1.54627025e+00 5.29025733e-01 -4.32233989e-01 -4.82051373e-01 5.39909661e-01 -4.80395615e-01 -9.56156671e-01 -2.56526113e-01 6.24397807e-02 -2.78070390e-01 2.31401101e-01 -2.37827495e-01 -8.00526381e-01 -8.44578803e-01 -4.58322406e-01 3.15484971e-01 -5.31799853e-01 -4.44706045e-02 1.23880498e-01 -2.21501753e-01 -1.26489799e-03 -7.29600966e-01 8.52684379e-02 -5.32088220e-01 -1.12900920e-01 -1.68257639e-01 9.12463605e-01 -9.65994358e-01 1.79984426e+00 -1.89915431e+00 -3.21725845e-01 2.80257761e-01 4.89756167e-02 2.54961580e-01 8.12653840e-01 3.02829593e-01 1.48005575e-01 -3.16356897e-01 -2.90174246e-01 -5.70505299e-02 -4.13521565e-02 4.49428141e-01 4.68897343e-01 4.28218633e-01 -3.94062757e-01 6.01689219e-01 -7.84495831e-01 -3.49523783e-01 6.61802828e-01 5.32248974e-01 -1.74225330e-01 2.14589298e-01 5.05287826e-01 7.34126925e-01 -3.33936751e-01 6.46201134e-01 1.27397895e-01 -2.40656943e-03 7.40641356e-02 7.90307075e-02 -1.35285810e-01 1.94347039e-01 -1.32479727e+00 1.34719753e+00 -4.13630068e-01 4.70701247e-01 -3.85202557e-01 -1.38868618e+00 1.04400051e+00 4.36719686e-01 8.64573002e-01 -8.00031066e-01 4.10231084e-01 1.64466888e-01 -3.88930738e-02 -8.78389060e-01 7.13914558e-02 3.80689830e-01 -3.08518857e-01 3.79600435e-01 -4.03982580e-01 9.40264583e-01 3.27220201e-01 -5.06520391e-01 1.39316094e+00 3.64887357e-01 1.05798554e+00 -1.80133939e-01 8.42921913e-01 -4.13495392e-01 5.40437341e-01 4.85272467e-01 -6.42101943e-01 1.81439474e-01 -1.21444941e-01 -5.45421422e-01 -5.33462048e-01 -8.75117719e-01 8.10103863e-02 1.17438900e+00 -5.78055307e-02 -2.90999264e-01 -9.04495955e-01 -2.07168326e-01 -2.97789186e-01 3.06050509e-01 -6.20129883e-01 -3.06216300e-01 -7.95360208e-01 -1.08786583e+00 7.19032884e-01 8.35969329e-01 1.32490361e+00 -1.38128519e+00 -1.24357009e+00 4.60279167e-01 -4.74077016e-01 -1.01596403e+00 1.97038408e-02 1.08023405e-01 -9.59746122e-01 -1.27413952e+00 -8.55850875e-01 -5.15786409e-01 2.57549793e-01 -6.34998083e-03 6.46829963e-01 -6.62965775e-02 -1.17270790e-01 4.22343910e-01 -4.26402301e-01 -6.61872327e-01 1.37030706e-01 2.39690885e-01 4.26750988e-01 1.36074558e-01 1.03644240e+00 -9.98938143e-01 -9.52978611e-01 4.91614163e-01 -2.82507122e-01 -2.91913122e-01 4.34402496e-01 2.64860272e-01 1.22286491e-01 1.05246753e-01 4.08669800e-01 -2.82481164e-01 6.35444462e-01 -6.57928526e-01 2.90030420e-01 3.13323438e-01 -5.45862496e-01 -2.43891388e-01 3.93343240e-01 -5.02529800e-01 -9.27249610e-01 -2.76165642e-02 -4.17817622e-01 5.91266155e-01 -7.02723324e-01 1.38923675e-01 -8.61069202e-01 4.46525365e-01 9.04839993e-01 4.33569223e-01 -1.16398484e-01 -5.81649482e-01 -4.26190734e-01 1.16144514e+00 4.72575486e-01 -1.92139372e-01 3.41335908e-02 4.28622812e-01 -2.93345988e-01 -1.26275468e+00 -5.45707047e-01 -7.63994634e-01 -7.98727691e-01 -4.06991869e-01 1.12709820e+00 -7.39775896e-01 -1.09742129e+00 1.00990486e+00 -6.77957654e-01 -3.38466287e-01 8.73383209e-02 5.73799610e-01 -5.16390443e-01 4.71135110e-01 -3.05971861e-01 -1.43758273e+00 -4.22125757e-01 -5.79326153e-01 8.31977129e-01 6.52166665e-01 -8.61335278e-01 -1.08322299e+00 9.75928754e-02 6.71254694e-01 3.47597808e-01 7.59859085e-01 2.14593962e-01 -4.47824717e-01 1.85119629e-01 -5.08319557e-01 3.05698216e-01 1.81773350e-01 7.52446830e-01 -5.23818612e-01 -6.27708077e-01 8.66771415e-02 1.42356873e-01 3.22095484e-01 1.45025671e-01 4.55739260e-01 9.74393606e-01 -4.27591205e-01 -5.46443582e-01 4.53703225e-01 7.21174181e-01 7.54660368e-01 1.20565236e+00 7.16342092e-01 8.23898494e-01 2.85635054e-01 3.71571958e-01 5.20217359e-01 5.87329566e-01 7.85461605e-01 -3.01978346e-02 9.83790904e-02 4.03945804e-01 -8.84304643e-02 6.58601701e-01 4.62511390e-01 -7.55616486e-01 1.32991105e-01 -9.67360973e-01 3.35137784e-01 -2.13789606e+00 -1.43551290e+00 -1.97050810e-01 2.24223018e+00 4.64771807e-01 1.44031212e-01 6.80562496e-01 1.14788604e+00 5.96422851e-01 1.85052529e-01 -5.81799746e-01 -3.58228117e-01 1.08123399e-01 1.03671335e-01 5.22005856e-01 1.57192782e-01 -1.08510017e+00 1.92512423e-01 6.20498037e+00 2.21459508e-01 -7.85833299e-01 1.64068505e-01 2.54081339e-01 -1.90022215e-02 5.01434267e-01 -6.55507326e-01 -9.60658848e-01 1.03526115e+00 1.18298435e+00 1.46632686e-01 1.11831143e-01 1.06515324e+00 7.41261363e-01 -6.20901346e-01 -8.39843273e-01 1.27499402e+00 2.07101807e-01 -4.72330272e-01 -4.18079376e-01 3.31807435e-01 2.75867581e-01 -2.81673223e-01 -6.77742660e-01 4.99795288e-01 -9.50550959e-02 -7.79673934e-01 1.53016910e-01 8.86466444e-01 2.43766487e-01 -6.36946261e-01 8.68603051e-01 9.03860271e-01 -1.44007528e+00 -3.46777946e-01 2.90890902e-01 -9.45853055e-01 2.95568883e-01 5.61042488e-01 -5.19977510e-01 2.41711244e-01 1.01200545e+00 8.29748690e-01 -6.03018165e-01 8.37135971e-01 -1.86396599e-01 5.77790141e-01 -5.81517398e-01 -4.01875526e-01 -2.43990690e-01 -3.04511219e-01 3.65911126e-02 1.21517432e+00 6.63565397e-01 1.93887711e-01 9.87202302e-02 1.50084212e-01 6.59377396e-01 6.37536049e-02 -7.63474166e-01 1.75614476e-01 2.36417010e-01 7.61793196e-01 -5.45947313e-01 -3.64071548e-01 -4.19498831e-01 9.53315437e-01 -1.98376104e-01 1.92059577e-01 -6.72768354e-01 -4.44563031e-01 7.19457328e-01 6.43143654e-01 -3.50155503e-01 -5.22716999e-01 -5.56853175e-01 -1.18527400e+00 3.38558227e-01 -6.68125212e-01 5.11636615e-01 -5.34864068e-01 -6.28994882e-01 -1.48217315e-02 2.04834923e-01 -1.39057076e+00 -5.65892816e-01 -6.25877976e-01 -8.32047403e-01 5.21103323e-01 -6.42470062e-01 -7.73059189e-01 -8.82521749e-01 9.79576111e-01 4.15751636e-01 -2.12575778e-01 9.48208034e-01 5.99577129e-01 -8.32351506e-01 5.54625213e-01 -2.92345375e-01 3.94632638e-01 -8.34245421e-03 -9.84763145e-01 1.52739855e-02 9.41226423e-01 -1.69217542e-01 7.79076993e-01 4.84588057e-01 -6.46222770e-01 -8.47395658e-01 -5.68610966e-01 1.09141088e+00 -4.65260863e-01 3.01350534e-01 -1.44123912e-01 -6.70099735e-01 5.53523660e-01 -2.77042866e-01 -2.19346777e-01 1.25414240e+00 1.90405890e-01 2.75675774e-01 -3.26029420e-01 -1.19889987e+00 6.68803096e-01 1.35508108e+00 -2.30714679e-01 -8.03493440e-01 -1.14804842e-02 -1.10924862e-01 -1.05094731e-01 -9.23129857e-01 4.64087397e-01 1.12938106e+00 -1.29096472e+00 9.41397548e-01 -1.65437147e-01 -2.00013071e-01 -3.06420624e-01 -1.15263745e-01 -8.81354272e-01 -3.05229455e-01 -1.89708635e-01 -8.89255702e-01 7.30524898e-01 1.10006355e-01 -8.59796345e-01 1.09122348e+00 8.96965027e-01 7.33959535e-03 -5.34660459e-01 -8.12907219e-01 -6.82213187e-01 -8.43470275e-01 -5.23036480e-01 8.33849788e-01 7.59777546e-01 5.87852657e-01 2.00571060e-01 -5.20589828e-01 -1.22952044e-01 4.85497147e-01 -5.39146841e-01 9.32524145e-01 -1.51951277e+00 -1.70428678e-01 -3.16675872e-01 -1.03782666e+00 -8.32363129e-01 -3.69519830e-01 -3.30968201e-02 -3.41402829e-01 -1.86018121e+00 -1.78676635e-01 1.62656143e-01 -4.31741834e-01 5.34432113e-01 -1.03439271e-01 5.79069369e-02 -4.39830005e-01 1.09726064e-01 -3.12411070e-01 2.57211655e-01 6.36905730e-01 -1.14137702e-01 -7.94907272e-01 6.82207167e-01 -5.65533519e-01 8.74656618e-01 1.21048629e+00 1.45592242e-01 -4.51903939e-01 -5.52463233e-02 1.62914142e-01 -7.83042908e-02 2.84810066e-01 -1.84177327e+00 1.64895326e-01 -4.69189174e-02 7.40298390e-01 -5.02404630e-01 2.74623930e-01 -9.70668733e-01 2.85391450e-01 8.40014338e-01 2.73455143e-01 -1.04409017e-01 -1.65643081e-01 4.62904930e-01 -7.00371638e-02 1.96316257e-01 4.30301964e-01 -1.77122429e-01 -8.45015943e-01 4.02147248e-02 -8.12254548e-01 -3.85984868e-01 1.10983622e+00 -1.09770548e+00 6.16811365e-02 -3.17410767e-01 -1.25416195e+00 -4.97140922e-02 1.10042177e-01 6.93511009e-01 2.26557180e-01 -1.45823395e+00 -7.58312817e-04 3.74200135e-01 2.81813920e-01 -2.54466176e-01 3.31835896e-01 1.07500887e+00 -4.90837485e-01 3.54646713e-01 -6.53120935e-01 -4.60881710e-01 -1.30406189e+00 1.75089777e-01 4.56075132e-01 -2.55894870e-01 -6.69658303e-01 -4.03983742e-02 -5.32637656e-01 -1.01340882e-01 4.90065455e-01 -5.93140602e-01 -9.62217450e-01 2.51833230e-01 1.08337188e+00 1.03901231e+00 1.04802221e-01 -7.76956439e-01 -7.15488911e-01 4.98946965e-01 5.56671619e-01 2.91366540e-02 9.93784308e-01 -2.33995467e-01 2.86037683e-01 8.21355522e-01 9.78613436e-01 -3.33269358e-01 -7.99439788e-01 1.44865319e-01 5.01694679e-01 -4.88861278e-02 -3.27330261e-01 -4.96197909e-01 -5.27552307e-01 8.16131651e-01 1.24178159e+00 3.81821603e-01 1.21731246e+00 -5.05682111e-01 9.90043938e-01 6.25789523e-01 8.44433069e-01 -1.41396928e+00 -2.44632512e-01 2.82518536e-01 4.16588157e-01 -1.09457445e+00 4.63543311e-02 -1.68581232e-02 -3.73441428e-01 7.34018326e-01 5.39089620e-01 -3.96181196e-02 9.56622720e-01 2.80462801e-01 -1.76521838e-01 1.31930396e-01 1.15654647e-01 -4.13179696e-01 1.22226477e-01 1.16003704e+00 5.94460011e-01 3.56910020e-01 -6.09679580e-01 9.41780329e-01 -5.21376550e-01 4.81321961e-01 2.71369219e-01 1.23548031e+00 -9.16229129e-01 -8.73286188e-01 -6.18222535e-01 7.12149382e-01 -5.64942539e-01 5.16900361e-01 -6.02279939e-02 5.79754829e-01 4.00533497e-01 1.23378909e+00 4.62663621e-02 -4.01706725e-01 6.66012585e-01 4.57710356e-01 3.55051994e-01 -4.49483365e-01 -3.53226602e-01 -5.02715409e-01 2.01592371e-01 -6.68432534e-01 -7.97059715e-01 -6.98466659e-01 -1.19468451e+00 -3.10977846e-01 3.27015340e-01 8.99059698e-02 4.75663692e-01 1.35373533e+00 1.66645154e-01 4.10808146e-01 3.87407690e-01 -1.03888583e+00 3.05482566e-01 -1.36420894e+00 -6.69868469e-01 3.32482070e-01 1.68977186e-01 -9.54416275e-01 -3.74210507e-01 1.43249512e-01]
[7.279477119445801, 0.659417450428009]
912a3b67-2a30-491f-807b-b25e4cb5675e
a-comparative-study-on-multichannel-speaker
2211.00511
null
https://arxiv.org/abs/2211.00511v3
https://arxiv.org/pdf/2211.00511v3.pdf
A Comparative Study on Multichannel Speaker-Attributed Automatic Speech Recognition in Multi-party Meetings
Speaker-attributed automatic speech recognition (SA-ASR) in multi-party meeting scenarios is one of the most valuable and challenging ASR task. It was shown that single-channel frame-level diarization with serialized output training (SC-FD-SOT), single-channel word-level diarization with SOT (SC-WD-SOT) and joint training of single-channel target-speaker separation and ASR (SC-TS-ASR) can be exploited to partially solve this problem. In this paper, we propose three corresponding multichannel (MC) SA-ASR approaches, namely MC-FD-SOT, MC-WD-SOT and MC-TS-ASR. For different tasks/models, different multichannel data fusion strategies are considered, including channel-level cross-channel attention for MC-FD-SOT, frame-level cross-channel attention for MC-WD-SOT and neural beamforming for MC-TS-ASR. Results on the AliMeeting corpus reveal that our proposed models can consistently outperform the corresponding single-channel counterparts in terms of the speaker-dependent character error rate.
['Li-Rong Dai', 'Shiliang Zhang', 'Qian Chen', 'Fan Yu', 'Zhihao Du', 'Jie Zhang', 'Mohan Shi']
2022-11-01
null
null
null
null
['speaker-separation']
['speech']
[ 4.22036380e-01 -1.82273313e-01 1.62961021e-01 -4.05528784e-01 -1.70689428e+00 -3.78497452e-01 5.95247984e-01 -1.29174232e-01 -3.00099403e-01 4.52863514e-01 6.03771865e-01 -5.92855036e-01 3.93698774e-02 3.10467035e-01 -5.51238716e-01 -7.99030602e-01 -6.78335801e-02 2.10918978e-01 -3.70641774e-03 -4.98735100e-01 -4.28208113e-02 2.20306054e-01 -1.37444615e+00 7.52916038e-01 6.77662194e-01 9.72086370e-01 4.54177678e-01 1.44534099e+00 -5.29702567e-02 1.89019039e-01 -8.37982357e-01 -1.42504051e-01 2.03721628e-01 -4.97530818e-01 -1.68932423e-01 1.17569141e-01 2.13178053e-01 6.08848222e-03 -6.34081423e-01 7.55450070e-01 1.12545741e+00 8.56599957e-02 3.68915826e-01 -1.13720870e+00 -3.24635178e-01 7.21570969e-01 -6.91820323e-01 3.34740847e-01 5.36104620e-01 -4.87882178e-03 7.75869131e-01 -1.23844314e+00 -5.96757978e-02 1.82223952e+00 3.89055163e-01 6.58086181e-01 -1.15573180e+00 -8.69365692e-01 1.58650756e-01 1.46702677e-01 -1.65747929e+00 -1.14874136e+00 4.75927889e-01 -1.25631675e-01 1.18922544e+00 6.21728063e-01 1.92369491e-01 1.20994067e+00 -6.96872594e-03 1.07045555e+00 1.11014414e+00 -5.59646428e-01 1.27446607e-01 -3.03540379e-02 1.90717548e-01 1.02409899e-01 -2.31540740e-01 2.18292356e-01 -1.27084231e+00 4.27522920e-02 5.72435200e-01 -4.89868850e-01 -4.73983765e-01 7.49386430e-01 -1.38190711e+00 5.52022219e-01 -1.74648970e-01 5.91466129e-01 -2.59711176e-01 -7.83024170e-03 3.04916888e-01 5.68830311e-01 6.64609134e-01 1.75466724e-02 -5.06271303e-01 -1.34359196e-01 -1.31958234e+00 -1.76395923e-02 5.12589276e-01 1.34609139e+00 3.10520053e-01 6.18024290e-01 -7.61264384e-01 1.10982549e+00 5.23046076e-01 9.66747999e-01 6.83335960e-01 -5.55603623e-01 7.02153265e-01 -5.17900705e-01 1.61291569e-01 -8.45771506e-02 -5.58871150e-01 -7.44479716e-01 -9.37122583e-01 -2.87147820e-01 4.06237170e-02 -3.80096883e-01 -1.11509788e+00 1.75226152e+00 1.08170755e-01 4.20929551e-01 3.95670474e-01 1.03114367e+00 8.03394258e-01 1.09529746e+00 -2.13693902e-01 -6.31894648e-01 1.50609159e+00 -8.74366403e-01 -1.31555307e+00 -2.99274623e-01 4.28099692e-01 -1.14828455e+00 5.26687562e-01 5.64226329e-01 -1.25181031e+00 -7.49804258e-01 -9.75278854e-01 3.07050347e-01 -8.34083930e-03 4.47616816e-01 -2.70543676e-02 1.10665572e+00 -1.30786216e+00 -3.51447135e-01 -4.33112800e-01 -2.06660539e-01 -8.91918782e-03 2.80812562e-01 -3.19962204e-01 -1.11823872e-01 -1.50150812e+00 6.11066699e-01 1.70368806e-01 3.81022245e-01 -1.14914811e+00 -4.13109481e-01 -7.44655967e-01 4.00662917e-04 2.65995920e-01 -2.06182137e-01 1.44378698e+00 -7.55757451e-01 -1.75477242e+00 5.60742617e-01 -1.00809073e+00 -5.07065833e-01 1.20581441e-01 -2.69688368e-01 -1.18237638e+00 6.54835254e-02 -2.05836982e-01 5.03308356e-01 1.14296961e+00 -1.42937136e+00 -7.31351137e-01 -5.60757101e-01 -5.62325358e-01 4.49665606e-01 -1.14566088e-01 4.57232714e-01 -2.57633597e-01 -9.09197867e-01 3.50489855e-01 -7.81144977e-01 8.55514556e-02 -8.74183774e-01 -8.31772566e-01 -3.08710665e-01 8.38464022e-01 -1.14051533e+00 1.65440786e+00 -2.50239563e+00 1.33850977e-01 7.27636591e-02 -6.32173568e-02 5.66651106e-01 -3.39180827e-01 3.62991929e-01 -1.43159837e-01 -1.98272005e-01 -1.43584758e-01 -7.77852118e-01 2.16798130e-02 -1.07591756e-01 2.74326727e-02 4.00709212e-01 2.61682570e-01 6.80603385e-01 -3.39087188e-01 -2.59839416e-01 3.39994520e-01 7.23268986e-01 -1.68175787e-01 3.86701226e-01 1.59171566e-01 5.61151445e-01 4.41953950e-02 6.10530138e-01 8.14723730e-01 3.94740701e-01 -6.38586730e-02 -1.23135544e-01 -2.79083759e-01 3.25403333e-01 -1.37156582e+00 1.73241532e+00 -6.27231658e-01 6.92045867e-01 8.09561789e-01 -9.00178254e-01 9.13398683e-01 1.06373417e+00 1.47735149e-01 -9.18568730e-01 -3.56688350e-02 6.27777100e-01 3.86824876e-01 -3.47241193e-01 5.61583757e-01 -3.64644736e-01 -4.64677028e-02 7.65672997e-02 2.58669853e-01 7.80814663e-02 -1.00354858e-01 3.52614343e-01 7.91265726e-01 -5.38182557e-01 8.86870101e-02 -2.78034627e-01 9.11037803e-01 -8.05616498e-01 4.30412143e-01 8.62805247e-01 -2.47128531e-01 9.01483119e-01 -7.63911977e-02 4.91360843e-01 -8.41968715e-01 -9.73474383e-01 -4.13769893e-02 1.19331777e+00 -4.43991572e-02 -1.89168409e-01 -8.69545817e-01 7.96940178e-03 -1.79762691e-01 1.15364349e+00 -3.67277376e-02 -4.39020293e-03 -6.14414692e-01 -7.21877217e-01 8.54367375e-01 3.12162280e-01 1.63023353e-01 -6.03398085e-01 1.66991092e-02 5.87513685e-01 -4.50990796e-01 -1.41335487e+00 -1.00438201e+00 6.05279148e-01 -2.91792125e-01 -2.37403303e-01 -1.31343043e+00 -5.32112837e-01 1.16398618e-01 7.22418845e-01 5.09478927e-01 -5.83697081e-01 -1.05194949e-01 5.54978192e-01 -5.60660899e-01 -4.52317685e-01 -6.48208261e-01 -4.09913957e-01 2.60201186e-01 7.68811405e-01 3.34290892e-01 -2.57760048e-01 -3.79399776e-01 4.21208382e-01 -7.15781808e-01 -1.54307097e-01 7.91234314e-01 8.16920340e-01 1.16986036e-01 -8.32071453e-02 9.79851246e-01 -3.78639042e-01 8.69079471e-01 -2.21233636e-01 -9.10067558e-02 3.06384176e-01 -2.60217845e-01 -1.09024487e-01 2.71450073e-01 -5.64207494e-01 -1.33180892e+00 -8.80171284e-02 -6.73480630e-01 -5.56759417e-01 -3.84752899e-01 3.20089430e-01 -5.23495436e-01 3.18781376e-01 4.53379452e-01 8.61609995e-01 -5.66122308e-02 -4.22523737e-01 3.85433644e-01 1.64549792e+00 5.34115374e-01 -1.18047386e-01 4.79588270e-01 8.52127224e-02 -6.18133664e-01 -1.53262365e+00 -3.64696085e-01 -1.02131641e+00 -3.42417091e-01 -1.08584948e-01 9.51114357e-01 -1.39507556e+00 -4.32893604e-01 8.83000791e-01 -1.34446883e+00 -1.96048945e-01 2.37280697e-01 8.49614501e-01 -3.37566763e-01 2.90709436e-01 -4.45054471e-01 -1.65246677e+00 -4.68932718e-01 -1.42433965e+00 1.48358583e+00 4.42766435e-02 -1.60299733e-01 -5.48146605e-01 -3.13202947e-01 5.80919683e-01 8.19802165e-01 -6.91905260e-01 4.15313721e-01 -9.47978556e-01 -1.25978485e-01 -6.32576421e-02 -4.99450639e-02 5.11936009e-01 2.84999231e-04 -6.06196165e-01 -1.32691503e+00 -5.24592459e-01 8.14752057e-02 9.40486863e-02 7.78804898e-01 1.06728148e+00 7.58149981e-01 -2.46286795e-01 -2.86143959e-01 2.24583730e-01 8.63432825e-01 6.53562009e-01 7.11690307e-01 -2.79735148e-01 4.47401196e-01 2.26852760e-01 4.44745719e-01 5.30379176e-01 4.05002475e-01 1.05655420e+00 -4.16966602e-02 -2.72735089e-01 -5.93174338e-01 1.62586033e-01 8.67050469e-01 1.14517069e+00 2.82202005e-01 -8.48321199e-01 -4.67424512e-01 4.53384072e-01 -1.53378308e+00 -8.60490322e-01 -5.28368771e-01 2.56364870e+00 5.71524799e-01 7.23057762e-02 3.13951164e-01 5.72109878e-01 1.27874303e+00 1.35104895e-01 -2.89492339e-01 -5.64728558e-01 -7.21161067e-01 1.64372370e-01 4.81322497e-01 8.94871473e-01 -8.51857185e-01 6.57287240e-01 5.04397011e+00 1.35467374e+00 -1.11419284e+00 6.74719095e-01 6.32939398e-01 -1.31851137e-01 -1.65583819e-01 -5.40233672e-01 -1.02712727e+00 5.01905382e-01 1.46287894e+00 7.39987642e-02 4.39630836e-01 6.76046088e-02 5.93866646e-01 -2.04698697e-01 -1.19046235e+00 1.40791714e+00 3.63437206e-01 -7.89333701e-01 -1.81968153e-01 -3.55770551e-02 2.71503061e-01 -4.23387103e-02 9.56193507e-02 2.34670997e-01 -1.57371074e-01 -9.22638118e-01 1.05140960e+00 7.92374387e-02 1.25379324e+00 -6.27123713e-01 5.68425059e-01 2.44335890e-01 -1.48112714e+00 -2.76183188e-01 1.52981550e-01 4.56229299e-01 4.93564755e-01 7.22720206e-01 -6.62493348e-01 1.01928759e+00 4.86511976e-01 2.45025188e-01 7.52382204e-02 8.44945312e-01 1.21684082e-01 9.10569727e-01 -4.25734401e-01 6.59559816e-02 1.09610468e-01 4.04910445e-01 1.13631582e+00 1.81619704e+00 6.82998598e-01 2.20263422e-01 4.28179651e-02 3.58347535e-01 2.53214955e-01 -2.32064724e-01 -1.57593384e-01 8.15849006e-02 7.45127797e-01 1.00191820e+00 -3.92120153e-01 -3.40679705e-01 -3.20890069e-01 1.04187202e+00 -5.42210639e-01 6.23187959e-01 -7.93912530e-01 -3.57776046e-01 6.46629214e-01 9.37892571e-02 3.95298421e-01 -4.55844730e-01 -4.13584001e-02 -9.73470807e-01 -3.23400736e-01 -1.04388416e+00 1.17994286e-01 -6.96724057e-01 -1.00548124e+00 7.21019387e-01 -3.29526842e-01 -1.06667125e+00 1.39429584e-01 -3.04380834e-01 -4.85324264e-01 1.04651856e+00 -1.73119509e+00 -1.09464169e+00 3.00156951e-01 1.09472704e+00 1.28663278e+00 -5.12094855e-01 8.00670981e-01 8.29049110e-01 -5.89570463e-01 1.12895143e+00 2.65477151e-01 -1.06393799e-01 6.42663360e-01 -1.02210760e+00 3.07194501e-01 1.05776644e+00 7.57839456e-02 2.68840313e-01 9.03319240e-01 -4.32146639e-01 -1.65059972e+00 -1.07744730e+00 9.57117081e-01 1.56619355e-01 2.13338450e-01 -8.86646569e-01 -6.89385355e-01 4.53185737e-01 2.94476897e-01 -3.79533201e-01 5.71127951e-01 -1.86327517e-01 -1.19951792e-01 -2.28201896e-01 -7.76302636e-01 3.17468226e-01 6.76139235e-01 -7.05513716e-01 -3.34728807e-01 2.50720114e-01 9.40295339e-01 -3.08182091e-01 -4.84460890e-01 5.09154201e-02 2.95474410e-01 -8.41495872e-01 9.03686047e-01 -5.89937046e-02 -2.57675767e-01 -3.89022827e-01 -7.78982818e-01 -1.41306353e+00 -2.02811182e-01 -1.35860312e+00 7.23782405e-02 1.49725270e+00 7.41024911e-01 -5.09297311e-01 2.06282958e-02 5.42672127e-02 -7.55454063e-01 3.06807347e-02 -1.54923069e+00 -8.66267323e-01 -3.09831232e-01 -8.51918757e-01 2.01720908e-01 4.18694705e-01 -1.94631875e-01 4.02546674e-01 -9.36109126e-01 6.17253721e-01 6.43291950e-01 -6.22395396e-01 4.63656276e-01 -9.10800934e-01 -2.60557026e-01 -3.05992484e-01 -3.97603866e-03 -1.46488118e+00 1.47442864e-02 -7.40031123e-01 4.24932152e-01 -1.41455412e+00 -2.98211336e-01 -1.18300714e-01 -3.92818600e-01 -9.26356167e-02 -1.02175340e-01 -1.77993208e-01 2.72902817e-01 -1.29145294e-01 -5.26607513e-01 5.24105430e-01 9.61255312e-01 -5.70007563e-02 -2.33585566e-01 2.81202465e-01 -5.28138459e-01 3.42958886e-03 3.06468546e-01 -3.19519728e-01 -1.32749289e-01 -1.77774251e-01 -5.32871723e-01 8.61350775e-01 1.16004329e-02 -1.00555682e+00 5.44524312e-01 3.66796285e-01 1.69582158e-01 -8.55430484e-01 1.00369346e+00 -5.17569304e-01 8.86432920e-03 3.79440159e-01 -6.76916599e-01 -3.83761138e-01 2.39416406e-01 7.14306056e-01 -3.80176783e-01 2.61858016e-01 7.72383571e-01 1.69530436e-02 -5.12627423e-01 -1.21693470e-01 -9.65789735e-01 -2.04234123e-01 7.67247856e-01 -1.92696095e-01 -1.42756864e-01 -7.53542125e-01 -8.34324002e-01 2.50349402e-01 -7.40529776e-01 5.73864162e-01 8.88780534e-01 -1.20724666e+00 -1.33392251e+00 4.71535802e-01 -2.89596058e-02 -2.17322484e-01 7.73349226e-01 1.15387893e+00 3.70421976e-01 1.04808557e+00 2.56777257e-01 -8.80387664e-01 -1.42101002e+00 2.42822453e-01 2.11869523e-01 1.07405148e-02 -2.70195812e-01 1.34916639e+00 4.80675966e-01 -1.26402587e-01 4.44369346e-01 -3.70511971e-02 5.20391874e-02 1.24519668e-03 5.09178996e-01 5.11505544e-01 3.73468310e-01 -1.08793271e+00 -4.24996793e-01 2.58317739e-01 7.36443046e-03 -6.79100811e-01 1.00218785e+00 -8.12262714e-01 2.37923518e-01 6.16837323e-01 1.10574555e+00 -3.19781038e-03 -8.87254417e-01 -3.62850517e-01 -2.51814961e-01 -1.44935951e-01 4.99137133e-01 -1.00827408e+00 -7.57557452e-01 1.25121498e+00 8.21673393e-01 1.82164460e-01 1.22449338e+00 -1.76775958e-02 9.26346421e-01 -1.09614819e-01 2.81818420e-01 -1.11048961e+00 1.31974384e-01 4.77294594e-01 1.08671021e+00 -1.10138869e+00 -7.32834756e-01 -2.86802799e-01 -8.47573817e-01 1.07994092e+00 2.84351528e-01 6.19739354e-01 7.39645302e-01 4.92619872e-01 2.08790883e-01 2.95947075e-01 -7.25946307e-01 -3.49635839e-01 3.24068367e-01 5.45559943e-01 7.08301842e-01 4.17099416e-01 4.47373725e-02 7.38994479e-01 -1.88647956e-01 -4.68725502e-01 4.23413157e-01 5.99862397e-01 -6.00103796e-01 -8.98677886e-01 -1.01418579e+00 2.79359996e-01 -4.80172753e-01 -3.46970797e-01 -1.68409143e-02 1.92714453e-01 -4.21749979e-01 1.76154029e+00 -1.02021560e-01 -5.82599699e-01 3.12543839e-01 4.46950316e-01 2.31063619e-01 -5.40918052e-01 -7.32444525e-01 1.32702637e+00 2.90576965e-01 -3.58921915e-01 -3.69073898e-01 -7.89907753e-01 -1.01035833e+00 3.38600092e-02 -7.42241800e-01 1.69453397e-01 1.07633746e+00 9.98742402e-01 4.90636289e-01 9.69153762e-01 6.99902177e-01 -9.90842521e-01 -5.02948344e-01 -1.41797888e+00 -9.71236110e-01 6.41407818e-02 8.54416132e-01 -3.89713049e-01 -4.85574067e-01 -1.22177802e-01]
[14.711441993713379, 5.828766345977783]
cd875d19-a141-436a-a005-1af6520461f0
high-resolution-image-reconstruction-with
null
null
http://openaccess.thecvf.com//content/CVPR2023/html/Takagi_High-Resolution_Image_Reconstruction_With_Latent_Diffusion_Models_From_Human_Brain_CVPR_2023_paper.html
http://openaccess.thecvf.com//content/CVPR2023/papers/Takagi_High-Resolution_Image_Reconstruction_With_Latent_Diffusion_Models_From_Human_Brain_CVPR_2023_paper.pdf
High-Resolution Image Reconstruction With Latent Diffusion Models From Human Brain Activity
Reconstructing visual experiences from human brain activity offers a unique way to understand how the brain represents the world, and to interpret the connection between computer vision models and our visual system. While deep generative models have recently been employed for this task, reconstructing realistic images with high semantic fidelity is still a challenging problem. Here, we propose a new method based on a diffusion model (DM) to reconstruct images from human brain activity obtained via functional magnetic resonance imaging (fMRI). More specifically, we rely on a latent diffusion model (LDM) termed Stable Diffusion. This model reduces the computational cost of DMs, while preserving their high generative performance. We also characterize the inner mechanisms of the LDM by studying how its different components (such as the latent vector Z, conditioning inputs C, and different elements of the denoising U-Net) relate to distinct brain functions. We show that our proposed method can reconstruct high-resolution images with high fidelity in straightforward fashion, without the need for any additional training and fine-tuning of complex deep-learning models. We also provide a quantitative interpretation of different LDM components from a neuroscientific perspective. Overall, our study proposes a promising method for reconstructing images from human brain activity, and provides a new framework for understanding DMs. Please check out our webpage at https://sites.google.com/view/stablediffusion-withbrain/.
['Shinji Nishimoto', 'Yu Takagi']
2023-01-01
null
null
null
cvpr-2023-1
['image-reconstruction']
['computer-vision']
[ 7.49342293e-02 -5.30625209e-02 2.44662672e-01 -2.82829791e-01 -2.39031062e-01 -4.91105556e-01 8.90040159e-01 -3.04201007e-01 -3.43982697e-01 5.87444663e-01 3.18764627e-01 5.07128201e-02 -2.33948991e-01 -8.20413589e-01 -7.23330140e-01 -1.06169617e+00 2.53127337e-01 3.39870691e-01 -4.36326228e-02 1.31561905e-01 2.62408983e-02 4.69834805e-01 -1.25033939e+00 6.82443157e-02 9.02839065e-01 8.30761194e-01 6.77903950e-01 3.23744267e-01 1.46916404e-01 7.74347246e-01 -3.49850267e-01 -3.38735074e-01 1.92926362e-01 -8.34528744e-01 -5.27979374e-01 1.73101604e-01 8.05030763e-02 -4.91936684e-01 -5.79095483e-01 1.20734692e+00 4.25927669e-01 2.03402400e-01 8.72298360e-01 -8.78309906e-01 -9.74118888e-01 3.62353563e-01 -3.99639487e-01 4.29611027e-01 1.54793849e-02 2.80686915e-01 6.42884195e-01 -9.00772750e-01 8.74945283e-01 1.02616000e+00 2.50651985e-01 6.53747916e-01 -1.67384899e+00 -4.96942103e-01 -3.02401166e-02 3.17734152e-01 -1.23643625e+00 -5.65937877e-01 8.37856889e-01 -8.62349749e-01 5.22419989e-01 -8.39426368e-02 9.13917899e-01 1.53347349e+00 4.03252125e-01 5.36335588e-01 1.50657082e+00 -6.66953027e-02 3.32828254e-01 3.96048576e-02 1.19049467e-01 7.79203475e-01 2.57872611e-01 6.30029142e-02 -5.30822217e-01 -4.33120653e-02 1.35892129e+00 1.83674753e-01 -5.63060164e-01 -4.09379214e-01 -1.22619033e+00 8.40610862e-01 4.79921192e-01 6.15466833e-01 -6.97014928e-01 2.68522412e-01 -1.07905157e-01 -7.54873455e-02 3.99389476e-01 1.26004130e-01 1.03638873e-01 2.51325011e-01 -1.03221822e+00 -5.72682172e-02 4.71244007e-01 3.87482315e-01 6.29880607e-01 2.68690199e-01 -7.60788769e-02 8.79323184e-01 4.07386303e-01 3.99338275e-01 5.77223063e-01 -1.26279664e+00 -7.66310841e-02 7.36105442e-02 -1.46874264e-01 -9.93498862e-01 -3.42906892e-01 -6.55183554e-01 -1.15814352e+00 1.52843520e-01 4.29953665e-01 -1.69926956e-02 -6.37879491e-01 2.08242679e+00 -8.86832539e-04 1.66743860e-01 -2.71409005e-01 1.06254268e+00 5.04727483e-01 5.92526138e-01 6.93289936e-02 -3.44620556e-01 1.27668560e+00 -5.00321209e-01 -6.67329431e-01 -2.70236969e-01 4.51317877e-02 -2.11977884e-01 8.64987075e-01 3.42086792e-01 -1.35196984e+00 -5.34698904e-01 -8.95543814e-01 -2.25700021e-01 3.50391492e-02 7.72771835e-02 6.39297307e-01 3.13699871e-01 -1.24097717e+00 5.84725320e-01 -1.22874272e+00 -3.20866734e-01 6.49846554e-01 -2.19082646e-02 -4.42223668e-01 -1.13499418e-01 -1.00946248e+00 1.02298856e+00 2.02157013e-02 2.55314380e-01 -1.28184378e+00 -6.11173689e-01 -6.45981193e-01 1.15383789e-01 3.53477560e-02 -1.23134911e+00 8.73465300e-01 -8.90303493e-01 -1.44263828e+00 8.52787375e-01 -3.24018538e-01 -4.42771286e-01 4.69266325e-01 8.55684876e-02 -4.93910201e-02 4.91571903e-01 5.73776104e-03 6.80294454e-01 1.01876712e+00 -1.37325275e+00 2.28178427e-01 -6.54621184e-01 -1.06507838e-01 6.11758151e-04 -1.69324890e-01 -2.57430643e-01 -2.20245495e-01 -7.75875449e-01 3.52310389e-01 -7.44947672e-01 -1.26621917e-01 2.13449046e-01 1.38395606e-02 1.57495171e-01 1.00542948e-01 -8.12543511e-01 8.34376395e-01 -1.99385583e+00 6.74796820e-01 9.69818309e-02 6.88997030e-01 1.23159653e-02 -1.23040833e-01 2.31760830e-01 -1.13080017e-01 -7.87698403e-02 -6.30700588e-01 -3.69927347e-01 2.14211922e-02 1.71552867e-01 -2.81886876e-01 7.59196579e-01 6.31000474e-02 1.16965604e+00 -7.99967349e-01 -1.56251058e-01 1.93851411e-01 9.68205035e-01 -6.16312087e-01 8.66221637e-02 1.47759030e-02 1.00522757e+00 -2.48691350e-01 2.01612905e-01 6.13983989e-01 -4.62292075e-01 4.79370117e-01 -3.32636178e-01 -2.43451484e-02 5.09422738e-03 -8.69312406e-01 1.91508675e+00 -4.13154662e-01 6.44407690e-01 2.25912333e-01 -1.45359528e+00 6.20366991e-01 2.18887076e-01 3.62255633e-01 -1.10267997e+00 2.46855080e-01 9.53102410e-02 2.27043152e-01 -5.05206168e-01 -1.79211438e-01 -5.70136845e-01 3.65061492e-01 6.12425864e-01 5.57538629e-01 2.07252249e-01 6.98465183e-02 2.55714238e-01 9.27712262e-01 1.47909760e-01 2.22303629e-01 -3.45950216e-01 2.62152016e-01 -6.44410253e-01 4.06734169e-01 6.63232863e-01 -1.43871352e-01 5.39776802e-01 6.03481770e-01 -1.28096044e-01 -9.34404492e-01 -1.36637592e+00 -2.77323455e-01 4.63189781e-01 -7.32018650e-02 5.72095392e-03 -9.54780579e-01 -1.43072428e-02 -3.01284701e-01 6.51457012e-01 -8.27461779e-01 -2.84225971e-01 -3.14164072e-01 -9.67818499e-01 3.69805604e-01 3.29474390e-01 5.37740946e-01 -1.01827145e+00 -5.98664880e-01 1.18286051e-01 -4.24044371e-01 -1.08275545e+00 -1.90411359e-01 -5.35759814e-02 -1.02871072e+00 -8.48582506e-01 -1.08761513e+00 -5.53560793e-01 8.33283365e-01 2.93487221e-01 1.00057471e+00 -2.11668581e-01 -4.03803736e-01 4.84357238e-01 1.44475728e-01 1.35319665e-01 -2.84671515e-01 -3.00854266e-01 7.17612207e-02 3.21460664e-01 -1.20912278e-02 -1.10541797e+00 -1.00536168e+00 1.38568953e-01 -1.21050417e+00 3.28006476e-01 5.77336848e-01 7.45580316e-01 7.43141532e-01 -1.56636029e-01 5.61653793e-01 -7.45688736e-01 6.72527313e-01 -7.88302422e-01 -6.12431347e-01 5.88736311e-03 -5.10586619e-01 2.55895853e-01 4.53688860e-01 -3.35673690e-01 -1.34937727e+00 -9.60961655e-02 -2.64323860e-01 -5.04485011e-01 -1.69091210e-01 4.43451226e-01 -2.03828171e-01 7.71905780e-02 6.30805016e-01 6.02471113e-01 1.10923275e-01 -5.71889460e-01 5.43873847e-01 6.31771535e-02 4.91473675e-01 -5.36110401e-01 6.24347687e-01 9.46258545e-01 -2.06468981e-02 -8.81369233e-01 -6.98087037e-01 -6.27883524e-02 -7.19538510e-01 -3.00079584e-01 1.06156433e+00 -8.36488426e-01 -6.52592063e-01 5.85670054e-01 -1.19340980e+00 -3.75252008e-01 -1.67615369e-01 6.11187935e-01 -7.36376047e-01 4.79612619e-01 -8.58749568e-01 -5.73680043e-01 -9.35656130e-02 -1.13238752e+00 6.95519805e-01 7.87859857e-02 -1.06039926e-01 -1.24884868e+00 1.46754250e-01 3.65001380e-01 4.51783806e-01 2.41970375e-01 9.72181857e-01 9.15301312e-03 -7.52220154e-01 3.87196362e-01 -1.70323819e-01 4.88250613e-01 -7.69706815e-02 -3.28269541e-01 -1.04898202e+00 -9.82821658e-02 6.85540557e-01 -8.59152153e-02 1.16032314e+00 8.41221869e-01 1.20658422e+00 -2.49227539e-01 -3.71054257e-03 8.13514054e-01 1.43616068e+00 9.50048566e-02 7.81887472e-01 -9.11654532e-02 5.58913589e-01 7.56034911e-01 -2.54265904e-01 3.57167721e-01 3.26207459e-01 4.83663678e-01 2.31787890e-01 -1.25211170e-02 -3.44409823e-01 -2.77432829e-01 4.09767061e-01 9.74990487e-01 -4.33394194e-01 -1.47607541e-02 -7.28102922e-01 4.25049752e-01 -1.69892585e+00 -1.22009742e+00 -2.01911535e-02 2.14756274e+00 7.87388384e-01 7.55955875e-02 -4.61383685e-02 -1.04603909e-01 4.55824375e-01 1.74301878e-01 -7.35653579e-01 6.45756498e-02 -3.70977372e-01 4.07321095e-01 1.70045942e-02 6.21548295e-01 -4.98595536e-01 6.47130311e-01 6.28651762e+00 5.73272765e-01 -1.16238999e+00 5.82499325e-01 4.95756298e-01 -2.19093114e-01 -6.09029710e-01 -7.94026181e-02 -3.55811536e-01 4.70009536e-01 9.52503026e-01 -2.00890332e-01 8.03264797e-01 3.84414166e-01 4.35125381e-01 -1.89370900e-01 -9.31449890e-01 1.18091202e+00 1.99055418e-01 -1.44981718e+00 2.87166268e-01 3.25271308e-01 6.50351346e-01 4.58860211e-02 3.07607412e-01 -1.89052135e-01 -3.50622013e-02 -9.42506313e-01 8.57037604e-01 1.09964490e+00 6.01577699e-01 -3.64300191e-01 1.91949353e-01 5.26644349e-01 -6.57089412e-01 5.46737202e-02 -4.32730615e-01 7.15281889e-02 2.14074418e-01 9.62244511e-01 -2.05376029e-01 1.35776609e-01 5.02077162e-01 8.22916806e-01 -2.93980956e-01 7.03875959e-01 -2.88104624e-01 4.38736141e-01 5.35117388e-02 4.19009924e-01 -1.58901230e-01 -5.95622301e-01 5.80071092e-01 9.38759446e-01 4.68654335e-01 2.54760146e-01 -3.82594168e-01 1.74192059e+00 -1.52512029e-01 -2.13334486e-01 -6.59519255e-01 -1.26881585e-01 7.33021423e-02 1.36231041e+00 -8.28964889e-01 -2.36599445e-01 -2.87253141e-01 1.20798707e+00 4.09872741e-01 7.06120610e-01 -9.22515094e-01 1.87883869e-01 6.35780334e-01 2.71322161e-01 1.54304877e-01 -6.59633934e-01 -1.97511077e-01 -1.59557903e+00 3.72902118e-02 -5.30091405e-01 -2.08858803e-01 -9.96505439e-01 -1.22206867e+00 5.84220707e-01 1.10700913e-03 -8.97924840e-01 -1.69006675e-01 -4.89756584e-01 -2.87438482e-01 1.01420987e+00 -1.53743505e+00 -7.32666373e-01 -2.91611910e-01 8.55891347e-01 3.66098434e-01 1.68501332e-01 7.68612087e-01 4.66923922e-01 -6.09426498e-01 -5.62287448e-03 1.70632482e-01 -6.16723597e-02 3.61936748e-01 -9.96352017e-01 2.53128767e-01 8.18686962e-01 4.17574406e-01 1.00070238e+00 5.32414734e-01 -4.36627448e-01 -1.29587412e+00 -7.19751596e-01 5.25039077e-01 -4.38643426e-01 6.44382954e-01 -5.39887488e-01 -1.00746882e+00 5.88292778e-01 3.43919098e-01 -6.75138459e-02 7.49101341e-01 -2.77388036e-01 -2.76306152e-01 3.20926122e-02 -9.77630258e-01 4.85158294e-01 1.14552891e+00 -8.30419421e-01 -5.25177181e-01 1.90381318e-01 2.45336697e-01 9.56053063e-02 -8.66830766e-01 -1.69471413e-01 5.51776707e-01 -1.14416790e+00 9.83024597e-01 -2.26504013e-01 5.42371213e-01 -1.86521545e-01 -4.09886539e-02 -1.53280342e+00 -6.70027673e-01 -2.53727019e-01 -3.70494395e-01 8.20411921e-01 5.16994670e-02 -7.84614146e-01 2.87550658e-01 5.65484285e-01 7.43590072e-02 -4.52691913e-01 -8.33056688e-01 -6.44372344e-01 1.35726795e-01 -3.78443539e-01 -7.31807109e-03 1.01048672e+00 -3.15730333e-01 3.30313891e-01 -3.61971557e-01 -1.32458815e-02 9.04302180e-01 -1.41404783e-02 6.64904863e-02 -1.29297531e+00 -4.58425432e-01 -5.94448090e-01 -1.31082758e-01 -9.67873633e-01 3.25973749e-01 -1.28158796e+00 -3.44955504e-01 -1.70787895e+00 4.42986310e-01 3.14220525e-02 -3.59220982e-01 2.06606805e-01 3.01206261e-01 3.53275687e-01 2.13353783e-01 5.59589207e-01 -1.60749272e-01 6.98084414e-01 1.49638855e+00 5.68511598e-02 2.02881604e-01 -4.12169635e-01 -7.60930121e-01 8.87765944e-01 7.36037850e-01 -4.79102582e-01 -5.52294850e-01 -6.79393828e-01 2.01970831e-01 2.18565479e-01 8.48465323e-01 -9.09888566e-01 2.09395126e-01 -1.68851446e-02 6.89783812e-01 1.39264435e-01 4.02091950e-01 -5.78280568e-01 4.12557960e-01 5.44945717e-01 -2.10025340e-01 -1.79136276e-01 -1.98576115e-02 5.89739501e-01 -1.75741553e-01 -1.16555847e-01 1.05293798e+00 -4.83720690e-01 -5.36856353e-01 4.07957464e-01 -6.28508210e-01 -7.57994037e-03 7.36179590e-01 -3.33302207e-02 -4.02145922e-01 -3.74181539e-01 -1.19599783e+00 -3.17577839e-01 2.79752702e-01 2.00406417e-01 7.52771676e-01 -1.34345531e+00 -5.27942896e-01 3.40762883e-01 -3.17736775e-01 -5.07647634e-01 6.70280933e-01 1.09091961e+00 -3.02984744e-01 3.08449358e-01 -6.34511948e-01 -4.07640368e-01 -5.46899199e-01 4.24408525e-01 6.15019739e-01 3.80473547e-02 -7.79724479e-01 6.56779647e-01 8.12884390e-01 2.93716416e-02 -9.72510874e-02 -2.57070124e-01 -1.69967547e-01 3.24458629e-01 5.12574673e-01 2.12359309e-01 -8.89132768e-02 -7.43986845e-01 -2.08857402e-01 4.55285370e-01 2.65997946e-01 -4.11894500e-01 1.52456796e+00 -3.19983929e-01 -2.72303998e-01 5.31305373e-01 1.14747572e+00 -1.86768055e-01 -1.43979645e+00 -2.37700656e-01 -3.49034578e-01 -3.02186966e-01 3.31516147e-01 -6.33983791e-01 -1.51363170e+00 1.18543780e+00 5.53626955e-01 6.83611184e-02 1.16528630e+00 1.22140646e-01 6.39221191e-01 4.32670824e-02 4.15237367e-01 -8.28868985e-01 2.63435602e-01 1.09998524e-01 1.15914321e+00 -9.54301834e-01 -2.12442890e-01 -2.12956849e-03 -5.94419062e-01 8.18088710e-01 3.14064443e-01 -2.83954531e-01 8.44793737e-01 -3.93914655e-02 -2.62224913e-01 -3.32409382e-01 -6.70204759e-01 -1.46929145e-01 2.09912926e-01 5.01304746e-01 3.79191190e-01 7.51827061e-02 -4.16989386e-01 7.21153736e-01 5.61096370e-02 9.68354717e-02 4.85465497e-01 5.11492014e-01 -2.85372168e-01 -9.66718316e-01 -1.50064409e-01 2.46121779e-01 -3.14082861e-01 -1.43597305e-01 -1.33114278e-01 3.49819511e-01 1.13367036e-01 5.62888324e-01 -1.72705203e-02 -1.91089846e-02 5.98462708e-02 2.74107039e-01 1.00171936e+00 -4.33631867e-01 -7.64394701e-02 2.98999608e-01 -3.96925032e-01 -5.92438519e-01 -6.92786336e-01 -7.98874974e-01 -1.07609773e+00 -2.73341656e-01 1.76414132e-01 -1.94046900e-01 7.04463601e-01 8.41496706e-01 4.41862196e-01 5.84241867e-01 3.67524773e-01 -8.84633839e-01 -2.43858591e-01 -8.20619702e-01 -9.19576228e-01 3.58551472e-01 2.42021397e-01 -8.79454613e-01 -3.83147866e-01 2.99603879e-01]
[10.7583646774292, 2.5105717182159424]
0cf7aeda-09af-439b-9479-f2f419db6f23
a-graph-similarity-for-deep-learning
null
null
http://proceedings.neurips.cc/paper/2020/hash/0004d0b59e19461ff126e3a08a814c33-Abstract.html
http://proceedings.neurips.cc/paper/2020/file/0004d0b59e19461ff126e3a08a814c33-Paper.pdf
A graph similarity for deep learning
Graph neural networks (GNNs) have been successful in learning representations from graphs. Many popular GNNs follow the pattern of aggregate-transform: they aggregate the neighbors' attributes and then transform the results of aggregation with a learnable function. Analyses of these GNNs explain which pairs of non-identical graphs have different representations. However, we still lack an understanding of how similar these representations will be. We adopt kernel distance and propose transform-sum-cat as an alternative to aggregate-transform to reflect the continuous similarity between the node neighborhoods in the neighborhood aggregation. The idea leads to a simple and efficient graph similarity, which we name Weisfeiler-Leman similarity (WLS). In contrast to existing graph kernels, WLS is easy to implement with common deep learning frameworks. In graph classification experiments, transform-sum-cat significantly outperforms other neighborhood aggregation methods from popular GNN models. We also develop a simple and fast GNN model based on transform-sum-cat, which obtains, in comparison with widely used GNN models, (1) a higher accuracy in node classification, (2) a lower absolute error in graph regression, and (3) greater stability in adversarial training of graph generation.
['Seongmin Ok']
2020-12-01
null
null
null
neurips-2020-12
['graph-similarity', 'graph-regression']
['graphs', 'graphs']
[-7.25251883e-02 2.28921145e-01 3.56012993e-02 -2.87689924e-01 -2.42104053e-01 -6.25258148e-01 7.42576420e-01 5.21792710e-01 -4.96125557e-02 6.22451305e-01 7.72470087e-02 -4.88399655e-01 -3.30814153e-01 -1.41449153e+00 -7.92448401e-01 -4.77007031e-01 -2.17036992e-01 3.08139771e-01 3.10484469e-01 -4.65479851e-01 -2.22181365e-01 6.73137367e-01 -1.03105783e+00 9.28630754e-02 1.01736403e+00 9.64662015e-01 -3.29402626e-01 6.98898077e-01 2.34022159e-02 8.19372892e-01 -5.07570386e-01 -5.98871469e-01 2.39759088e-01 -7.34234214e-01 -8.40803564e-01 -4.43640053e-01 6.44704938e-01 1.48457866e-02 -7.55041838e-01 1.21442616e+00 2.40785554e-01 2.98512459e-01 9.25901175e-01 -1.62235785e+00 -1.46299636e+00 8.98336947e-01 -2.67381102e-01 1.12931959e-01 5.34661949e-01 -5.03606014e-02 1.12450004e+00 -3.86217952e-01 5.19787192e-01 1.32561266e+00 1.25638878e+00 3.62617433e-01 -1.42897296e+00 -4.73691404e-01 6.78783506e-02 9.02570710e-02 -1.47795653e+00 1.27303198e-01 6.72041297e-01 1.86267011e-02 9.00003672e-01 2.64601231e-01 8.01796496e-01 1.02043200e+00 3.83275151e-01 5.51234961e-01 8.34614038e-01 -3.37371230e-01 9.57482830e-02 -1.86987817e-01 1.75321534e-01 9.99522150e-01 4.48734313e-01 -6.94161579e-02 -6.96903467e-02 -2.96776026e-01 9.10374939e-01 3.40951115e-01 -2.93992460e-01 -6.40691817e-01 -1.01712954e+00 1.08309770e+00 1.39933968e+00 5.30074835e-01 -3.06585550e-01 6.14132702e-01 2.63271958e-01 6.56009853e-01 5.84441125e-01 3.16556305e-01 6.23108409e-02 5.15268743e-01 -5.85758388e-01 2.52049476e-01 9.32681561e-01 7.62619734e-01 9.72440124e-01 3.70050520e-01 -2.52008419e-02 5.02179086e-01 2.30379909e-01 3.51888031e-01 4.51700836e-01 -4.13009584e-01 2.00163558e-01 8.83608818e-01 -6.39287293e-01 -1.42645431e+00 -4.55620855e-01 -4.89162296e-01 -1.22367883e+00 2.82472759e-01 3.04845870e-01 1.77241489e-01 -1.01959836e+00 1.90802264e+00 5.50511144e-02 5.56364059e-01 1.32758126e-01 4.73153919e-01 1.12642264e+00 3.09807926e-01 7.08736628e-02 4.96229708e-01 9.05153155e-01 -7.56017089e-01 -2.96348482e-01 -3.24937999e-02 9.31413233e-01 -2.66287208e-01 1.00582945e+00 -1.57743543e-01 -7.23841786e-01 -4.94164079e-01 -1.12388480e+00 1.43717322e-02 -9.87388194e-01 -3.39861184e-01 1.23969054e+00 8.46397340e-01 -1.64872706e+00 1.25032973e+00 -6.50628209e-01 -7.82387078e-01 5.56030095e-01 5.07130384e-01 -8.06177378e-01 -1.26480699e-01 -1.26809490e+00 7.78135419e-01 5.44389665e-01 -2.10290849e-01 -5.11959374e-01 -7.88654566e-01 -1.33904052e+00 2.81223774e-01 1.31019875e-01 -9.38257098e-01 6.64830923e-01 -9.71237183e-01 -1.14521909e+00 6.64146602e-01 2.89271861e-01 -7.60080338e-01 1.02905720e-01 1.67836294e-01 -5.83794057e-01 2.56542414e-02 9.76724550e-02 5.25194883e-01 6.78752840e-01 -9.33459282e-01 -8.65658075e-02 -1.26440093e-01 1.71921372e-01 8.83246660e-02 -4.54706520e-01 -4.12119538e-01 7.40010813e-02 -8.01003814e-01 1.43352196e-01 -8.68291140e-01 -2.28473350e-01 1.41360983e-01 -5.38881600e-01 -3.66742462e-01 8.97846162e-01 -3.83468747e-01 1.09799552e+00 -1.96106553e+00 1.54882194e-02 5.23207605e-01 8.59252274e-01 5.54098010e-01 -4.16011602e-01 6.18468583e-01 -4.58336890e-01 2.60786146e-01 -2.83286691e-01 7.51140863e-02 5.67589514e-02 2.56449759e-01 -1.47991508e-01 4.64789510e-01 2.50061572e-01 1.47884929e+00 -1.25339043e+00 -1.19029313e-01 3.06521535e-01 5.31074703e-01 -4.19690549e-01 5.03388345e-02 8.36999193e-02 -2.30385602e-01 -2.39846513e-01 5.94190300e-01 5.42188346e-01 -3.94248545e-01 1.79286003e-02 -3.06610912e-01 6.67795002e-01 1.33955210e-01 -8.42718244e-01 1.30847800e+00 -8.34003165e-02 6.68899119e-01 -3.11997622e-01 -1.26113498e+00 1.13420486e+00 4.67451289e-02 2.13889480e-01 -5.78191638e-01 5.72598167e-02 1.04679286e-01 9.65899974e-02 1.84145868e-01 3.19085389e-01 8.00776184e-02 4.45422344e-03 6.77446067e-01 3.73220235e-01 -1.46311790e-01 2.50182450e-01 6.44988656e-01 1.60875630e+00 -1.80009916e-01 5.64564526e-01 -4.09354627e-01 4.08893049e-01 -4.39968348e-01 7.57327974e-02 8.30564678e-01 -2.55211115e-01 5.53437293e-01 7.40161180e-01 -7.66262412e-01 -7.42725968e-01 -1.54684711e+00 4.84524280e-01 8.21477056e-01 8.58092755e-02 -5.86413980e-01 -7.23615170e-01 -1.04732156e+00 3.44676018e-01 6.18195593e-01 -9.81002927e-01 -6.95689499e-01 -4.10768420e-01 -5.03911376e-01 8.67005646e-01 7.40706801e-01 4.37173873e-01 -1.18981457e+00 4.30600911e-01 1.19254831e-02 2.54014015e-01 -9.14492786e-01 -7.08157897e-01 1.04828194e-01 -8.40240717e-01 -1.22557676e+00 -4.78401929e-01 -8.73564303e-01 8.68234217e-01 4.59164888e-01 1.28235734e+00 5.34807503e-01 -2.18440652e-01 5.38199365e-01 -4.09004629e-01 -1.56403556e-01 -6.67256296e-01 2.44575039e-01 8.06148350e-02 -2.08884571e-02 3.37418675e-01 -9.68411207e-01 -3.70334893e-01 7.56294429e-02 -9.71658945e-01 -4.24727798e-01 5.87586164e-01 7.06588030e-01 3.03974777e-01 4.04187925e-02 5.61086416e-01 -1.12289548e+00 1.25986516e+00 -6.94784284e-01 -2.39326656e-01 3.91661733e-01 -7.71675944e-01 2.55549937e-01 9.65592146e-01 -5.74945867e-01 -2.65343934e-01 -1.86669230e-01 -2.23906171e-02 -5.69305658e-01 3.33971977e-02 4.51512188e-01 1.25077799e-01 -6.85294390e-01 9.57754612e-01 1.92315891e-01 3.28124344e-01 9.36411880e-03 7.40623593e-01 -3.77678461e-02 5.72320223e-01 -4.24508065e-01 1.27045095e+00 3.10174495e-01 2.09717229e-01 -7.04113483e-01 -4.59546864e-01 -7.44994581e-02 -6.92242503e-01 -1.35602793e-02 7.23553538e-01 -4.88059759e-01 -6.47342145e-01 4.66119736e-01 -8.94804418e-01 -3.64785880e-01 -4.12519574e-01 3.66321027e-01 -4.21868026e-01 6.08791709e-01 -9.05145645e-01 -3.65598619e-01 -5.11705399e-01 -7.34864175e-01 6.71069682e-01 3.12842041e-01 -3.48813444e-01 -1.72982013e+00 2.07169931e-02 -2.67421693e-01 6.86997056e-01 7.26702034e-01 1.21689761e+00 -9.94896114e-01 -2.97343910e-01 -5.08659303e-01 -5.50027013e-01 4.05491620e-01 4.45478797e-01 2.08478631e-03 -5.51416636e-01 -4.41508591e-01 -5.68077743e-01 -1.82323068e-01 9.63432372e-01 4.04468864e-01 1.00129604e+00 -3.17063898e-01 -4.53792095e-01 9.10565853e-01 1.37621081e+00 -2.20566958e-01 7.54995108e-01 7.05600381e-02 1.14309549e+00 1.48036942e-01 -1.01123124e-01 -2.68705726e-01 3.31509620e-01 3.66509110e-01 5.92702627e-01 -5.12753308e-01 -4.16085839e-01 -5.61366737e-01 4.77869064e-01 8.29284251e-01 -1.12557031e-01 -3.80977422e-01 -7.37169445e-01 4.12073821e-01 -1.96231246e+00 -9.55009699e-01 -3.33399266e-01 2.00150752e+00 2.74861366e-01 1.68218911e-01 1.87815994e-01 -3.45407128e-02 8.43391478e-01 3.66510034e-01 -4.17782098e-01 -6.58487260e-01 -3.30740035e-01 5.32560587e-01 4.62170631e-01 4.94301379e-01 -1.17291069e+00 1.10021508e+00 6.75058365e+00 8.56038034e-01 -8.08496237e-01 -1.77772000e-01 4.99027222e-01 4.70705122e-01 -4.85504508e-01 9.62840170e-02 -1.93152055e-01 1.63075969e-01 9.42606866e-01 -4.79001850e-01 6.74394667e-01 9.41894114e-01 -4.91070241e-01 4.73688722e-01 -1.37832391e+00 7.53678918e-01 2.54021466e-01 -1.49794936e+00 4.97402191e-01 3.08266282e-02 6.93892539e-01 4.00155224e-02 -2.41290070e-02 4.68789428e-01 8.99658561e-01 -1.44637597e+00 3.26361358e-01 4.34388459e-01 5.93379796e-01 -7.21019387e-01 8.22766781e-01 -9.27342325e-02 -1.77458847e+00 1.29227445e-01 -6.09458089e-01 -2.61261240e-02 -2.12947085e-01 5.38990438e-01 -9.53604043e-01 9.45257545e-01 4.55529451e-01 9.64374125e-01 -8.79924297e-01 7.49184549e-01 -4.12393779e-01 4.70182210e-01 -1.84321135e-01 -1.85502425e-01 4.64785069e-01 -5.14067411e-01 2.77710408e-01 1.09504592e+00 3.93499464e-01 -2.50806302e-01 1.47515267e-01 1.24243212e+00 -3.64098907e-01 6.21128856e-05 -1.10158563e+00 -3.49429607e-01 5.20962298e-01 1.28337455e+00 -1.01897073e+00 -2.30557993e-01 -3.82456005e-01 1.20873582e+00 8.40293944e-01 4.30452913e-01 -8.40236902e-01 -7.69737303e-01 7.62099862e-01 1.14498451e-01 2.54072994e-01 -1.06846742e-01 1.88084081e-01 -7.83261180e-01 -1.03454344e-01 -6.69279039e-01 6.30985141e-01 -8.19797695e-01 -1.76365030e+00 9.10963058e-01 -1.89443439e-01 -9.28122044e-01 -1.30669266e-01 -8.66535962e-01 -1.07253206e+00 8.33158195e-01 -1.12425399e+00 -1.52570283e+00 -3.88459355e-01 9.57412720e-01 -1.90358892e-01 -2.29001597e-01 1.04962063e+00 1.26115054e-01 -3.50941539e-01 8.58963370e-01 -8.62113684e-02 5.71701467e-01 4.23218131e-01 -1.54727530e+00 1.03691351e+00 7.38223791e-01 5.00245035e-01 7.43587792e-01 2.94465810e-01 -7.22255588e-01 -1.11974049e+00 -1.25382662e+00 6.11368358e-01 -4.75514352e-01 9.91635144e-01 -3.61478597e-01 -1.10936916e+00 9.71960783e-01 5.61775714e-02 4.65416104e-01 4.97147352e-01 1.12549946e-01 -8.55453789e-01 -2.59020716e-01 -1.19734323e+00 7.04777300e-01 1.23938036e+00 -7.98395693e-01 -4.30993676e-01 4.60070640e-01 8.90175819e-01 -5.65781072e-03 -9.99775469e-01 2.89332092e-01 3.28886837e-01 -1.10824394e+00 1.19675255e+00 -8.37860584e-01 1.37982503e-01 -1.96167171e-01 2.62047574e-02 -1.68626571e+00 -8.22200239e-01 -6.19862974e-01 -1.32146627e-01 9.72550929e-01 3.89319599e-01 -1.14593065e+00 8.54919672e-01 3.15970361e-01 -1.04142986e-01 -7.43811548e-01 -6.54708147e-01 -1.10814166e+00 1.42514303e-01 -1.14250019e-01 8.54297519e-01 1.21390593e+00 -1.95732564e-02 2.72798419e-01 -2.77789552e-02 1.08439192e-01 6.87158585e-01 -7.35922083e-02 7.29604542e-01 -1.53819406e+00 1.89021812e-03 -7.40920007e-01 -1.18662906e+00 -5.72995722e-01 4.40621465e-01 -1.55149758e+00 -4.91539448e-01 -1.95459282e+00 8.08383152e-03 -3.06946099e-01 -5.45791805e-01 7.07299471e-01 -3.57458472e-01 3.46007735e-01 -2.91310754e-02 -1.51636496e-01 -6.21870100e-01 5.42306423e-01 9.72722411e-01 -2.21965104e-01 -4.57847863e-02 -5.45612350e-02 -8.96733701e-01 6.03668928e-01 8.11990917e-01 -4.54989940e-01 -6.11126661e-01 -1.27720699e-01 1.03937618e-01 -4.22673613e-01 4.33895648e-01 -1.22045600e+00 1.92119122e-01 1.76724315e-01 4.43910539e-01 -6.00011535e-02 5.93887419e-02 -5.04883707e-01 2.66031891e-01 7.59308040e-01 -1.03214391e-01 3.71036679e-01 -1.74652189e-02 7.67301619e-01 -8.00517797e-02 -7.19672069e-02 5.34606934e-01 -1.21461399e-01 -6.32729053e-01 6.17539108e-01 -1.16118424e-01 -2.27868333e-02 1.01459301e+00 -5.57022691e-01 -5.90422034e-01 -7.28324711e-01 -5.61788082e-01 -1.01276226e-01 5.08629203e-01 4.90088969e-01 7.65132666e-01 -1.77738869e+00 -8.24775219e-01 3.41838330e-01 7.82108158e-02 -3.39858711e-01 -8.27596486e-02 6.36518002e-01 -7.05705881e-01 1.14043050e-01 -2.75204867e-01 -3.72830451e-01 -1.10921264e+00 6.27408504e-01 4.09091115e-01 -4.35851216e-01 -7.90509999e-01 9.28638399e-01 2.21876562e-01 -8.10839474e-01 -1.92650884e-01 -4.15176153e-01 2.72971317e-02 -3.65627438e-01 3.20188582e-01 3.06888431e-01 4.40246537e-02 -6.18372202e-01 -3.55287671e-01 4.09911990e-01 -1.80164739e-01 5.80739260e-01 1.29694176e+00 4.05920208e-01 -4.45475310e-01 1.19784400e-01 1.32843268e+00 -1.11166187e-01 -8.43920588e-01 -2.77276337e-01 -5.73992617e-02 -2.18873844e-01 -2.99373955e-01 -3.59367430e-01 -1.29402268e+00 4.68900740e-01 3.09510976e-01 7.60238409e-01 9.31904972e-01 1.16951704e-01 8.06597054e-01 4.80064273e-01 2.41216019e-01 -5.30666053e-01 2.22069442e-01 5.34185410e-01 8.08703899e-01 -1.05121481e+00 2.36053929e-01 -6.25362754e-01 -3.44316006e-01 1.20947945e+00 5.88381767e-01 -6.07285559e-01 7.70396054e-01 3.59815396e-02 -6.83413371e-02 -4.61419582e-01 -4.14893448e-01 -3.47219229e-01 5.32608569e-01 1.25851190e+00 3.56888801e-01 1.84665203e-01 3.56877446e-02 3.58733326e-01 -3.92827183e-01 -5.17003357e-01 3.58972311e-01 5.57972193e-01 -2.37175271e-01 -1.08286250e+00 3.64931226e-02 7.92592168e-01 1.29536455e-02 -1.99232683e-01 -8.34513724e-01 1.13440764e+00 -3.23010504e-01 6.64396107e-01 -7.94807374e-02 -7.46161103e-01 3.00202459e-01 -1.36535957e-01 4.48751748e-01 -6.15532160e-01 -6.22773468e-01 -7.04044819e-01 -1.70813352e-02 -7.27140248e-01 -8.55846331e-02 -3.92051563e-02 -1.23671269e+00 -8.16355407e-01 -3.78586262e-01 1.60530269e-01 2.17132822e-01 7.35337317e-01 4.09839481e-01 7.15067863e-01 3.91562372e-01 -7.05925286e-01 -4.09231424e-01 -7.99953580e-01 -8.69684696e-01 9.37967122e-01 1.17334895e-01 -4.75819945e-01 -5.94590604e-01 -4.76382285e-01]
[6.885725021362305, 6.286672592163086]
fa6d6249-bfe0-439e-ba55-b80e28a878d8
lego-absa-a-prompt-based-task-assemblable
null
null
https://aclanthology.org/2022.coling-1.610
https://aclanthology.org/2022.coling-1.610.pdf
LEGO-ABSA: A Prompt-based Task Assemblable Unified Generative Framework for Multi-task Aspect-based Sentiment Analysis
Aspect-based sentiment analysis (ABSA) has received increasing attention recently. ABSA can be divided into multiple tasks according to the different extracted elements. Existing generative methods usually treat the output as a whole string rather than the combination of different elements and only focus on a single task at once. This paper proposes a unified generative multi-task framework that can solve multiple ABSA tasks by controlling the type of task prompts consisting of multiple element prompts. Further, the proposed approach can train on simple tasks and transfer to difficult tasks by assembling task prompts, like assembling Lego bricks. We conduct experiments on six ABSA tasks across multiple benchmarks. Our proposed multi-task approach achieves new state-of-the-art results in almost all tasks and competitive results in task transfer scenarios.
['Weipeng Yan', 'Yongjun Bao', 'Pengzhang Liu', 'Chao Liu', 'Zhiyuan Liu', 'Hanyu Liu', 'Jun Fang', 'Tianhao Gao']
null
null
null
null
coling-2022-10
['aspect-based-sentiment-analysis']
['natural-language-processing']
[ 3.68435830e-01 -2.14536667e-01 4.16626483e-01 -5.51023841e-01 -1.34127474e+00 -5.91686606e-01 7.17810094e-01 -1.11827753e-01 -4.05040413e-01 4.42575097e-01 6.42305240e-02 -1.47864655e-01 1.01508619e-02 -6.72807574e-01 -8.33106041e-01 -8.40874195e-01 5.59660554e-01 7.77732968e-01 1.14467271e-01 -4.26225930e-01 1.71167120e-01 -3.43639314e-01 -1.17139757e+00 8.12756121e-01 8.35853100e-01 7.43218064e-01 4.22760963e-01 6.05622411e-01 -3.83870006e-01 3.32574606e-01 -1.06970739e+00 -8.05663586e-01 1.64795332e-02 -5.40229142e-01 -7.64826655e-01 3.59471813e-02 2.27970809e-01 1.33595735e-01 5.10249972e-01 8.15427065e-01 7.44448841e-01 1.73534587e-01 7.55965412e-01 -1.45941186e+00 -7.83583164e-01 8.90917361e-01 -7.06286669e-01 1.18563576e-02 2.91902423e-01 2.63280515e-02 1.35115075e+00 -1.12865722e+00 1.55981421e-01 1.27394664e+00 5.30700624e-01 5.46568155e-01 -1.42185974e+00 -8.19772720e-01 6.19644523e-01 -6.31579058e-03 -9.78674531e-01 -1.99596565e-02 8.55058014e-01 -4.60709751e-01 1.31390285e+00 1.82035163e-01 5.78305960e-01 1.46082783e+00 3.63473862e-01 1.03464019e+00 1.20074046e+00 -2.43253112e-01 9.41695347e-02 -8.61245245e-02 3.83499205e-01 2.99668670e-01 3.19234282e-01 -6.37867212e-01 -7.22585797e-01 -1.40731364e-01 2.24891424e-01 -1.54011145e-01 1.75286427e-01 1.11197710e-01 -1.36617756e+00 1.01917374e+00 6.91014975e-02 1.53979883e-01 -2.70904362e-01 2.73338377e-01 4.33633506e-01 3.48886609e-01 6.11886501e-01 4.61414099e-01 -6.23454273e-01 -8.51564258e-02 -6.05024159e-01 6.86283529e-01 8.05369318e-01 1.14676821e+00 7.15518594e-01 2.03065023e-01 -8.26381087e-01 9.62515950e-01 1.99088454e-01 4.87998307e-01 5.03769338e-01 -2.88606137e-01 7.02776611e-01 5.62243044e-01 -3.74379940e-02 -7.21909702e-01 -3.92138004e-01 -4.97153968e-01 -7.50933886e-01 1.50471151e-01 2.36357555e-01 -5.77215672e-01 -1.09979308e+00 1.83334994e+00 2.24452928e-01 -2.12756589e-01 -2.43948866e-02 6.93236828e-01 1.05453646e+00 8.52701426e-01 2.28861660e-01 1.45698234e-01 1.89779365e+00 -1.43487060e+00 -8.36724341e-01 -6.56974912e-01 4.07645851e-01 -1.02426207e+00 1.44273639e+00 5.99029839e-01 -9.07841444e-01 -6.33473456e-01 -8.42852116e-01 -1.83230564e-01 -5.08816242e-01 4.41484243e-01 6.09722912e-01 5.42817831e-01 -1.00806022e+00 -2.80369837e-02 -5.60133398e-01 4.46724668e-02 2.85415947e-01 2.70676166e-01 4.59748581e-02 2.16503181e-02 -1.06085241e+00 5.93866825e-01 3.27654570e-01 2.10321054e-01 -1.02987647e+00 -7.50127792e-01 -8.51539493e-01 2.24695519e-01 5.59321463e-01 -1.19694138e+00 1.38592279e+00 -8.15739751e-01 -1.57731557e+00 8.51522744e-01 -2.01196998e-01 -6.52839020e-02 4.24639106e-01 -7.43747830e-01 -4.17596614e-03 -4.97690499e-01 1.51196510e-01 6.30882442e-01 1.08533502e+00 -1.27552772e+00 -6.36901975e-01 -1.51636824e-01 2.47464433e-01 4.09409761e-01 -2.78265357e-01 3.32355142e-01 -2.83348143e-01 -1.00868034e+00 -4.07440364e-01 -1.01680934e+00 -2.05591038e-01 -7.27445900e-01 -6.66475058e-01 -6.86631858e-01 6.74949467e-01 -3.31121057e-01 1.05780578e+00 -1.92890668e+00 5.15478253e-01 -4.08518374e-01 1.48161709e-01 7.03105778e-02 -3.31607670e-01 4.34954405e-01 -7.44009241e-02 9.99733955e-02 -2.29894191e-01 -8.39425385e-01 2.26030752e-01 -9.73929465e-03 -3.50963295e-01 -3.18793595e-01 3.16863656e-01 1.06482375e+00 -8.89154315e-01 -3.94450396e-01 -2.16606274e-01 3.75017434e-01 -4.17693675e-01 4.03197408e-01 -3.99264634e-01 2.92174548e-01 -6.03872240e-01 5.35417378e-01 5.36682904e-01 -3.92691642e-01 -7.35753775e-02 -1.51670322e-01 1.86196253e-01 3.65822077e-01 -1.00334191e+00 1.92283952e+00 -8.58988047e-01 4.84840453e-01 -1.46104589e-01 -9.11121726e-01 1.02362967e+00 3.96023631e-01 2.10537151e-01 -3.68085653e-01 2.06179440e-01 1.99737251e-02 2.18570873e-01 -2.96383828e-01 5.95358849e-01 -2.36422360e-01 -6.13619626e-01 7.67661095e-01 3.93381864e-01 -3.70624125e-01 4.02534217e-01 1.18607506e-01 9.32037055e-01 3.63449156e-01 3.39761674e-01 -3.02768201e-01 2.05598459e-01 -1.01401597e-01 4.49969590e-01 8.57292295e-01 2.97997296e-01 7.38984942e-01 6.58325732e-01 -4.90087062e-01 -9.37477767e-01 -8.52645099e-01 2.26913974e-01 1.69284391e+00 9.73681733e-03 -7.57901669e-01 -5.77914298e-01 -8.11446309e-01 -2.45284423e-01 8.66641104e-01 -8.85748029e-01 -9.86767486e-02 -5.04331350e-01 -1.38161504e+00 2.33650774e-01 6.14219308e-01 3.66107792e-01 -1.52420700e+00 -6.22109115e-01 2.60098219e-01 -3.81893396e-01 -1.17673421e+00 -5.77412069e-01 2.35982329e-01 -5.72100222e-01 -6.40165567e-01 -8.33532810e-01 -9.65713501e-01 5.62792301e-01 3.41240376e-01 1.50614130e+00 -3.63524288e-01 3.12965810e-02 6.84357360e-02 -5.47704458e-01 -7.63878226e-01 -1.93630829e-01 6.44115686e-01 -5.45921683e-01 3.59160870e-01 2.95814574e-01 -5.60663521e-01 -4.31370020e-01 2.70643234e-01 -9.15693760e-01 5.28689206e-01 8.71861517e-01 8.03219080e-01 6.19817495e-01 -3.01896721e-01 8.10016096e-01 -1.36049986e+00 1.23747730e+00 -5.62733412e-01 -2.21970916e-01 4.22370523e-01 -4.54243124e-01 8.48279893e-02 6.67805254e-01 -6.47697210e-01 -1.24765658e+00 -5.10192439e-02 -9.63605940e-03 -3.05702597e-01 -2.16542229e-01 5.91112971e-01 -2.26088822e-01 4.08023834e-01 3.97063255e-01 4.16155070e-01 -5.06734669e-01 -3.43556911e-01 2.74806589e-01 4.05853510e-01 3.35911997e-02 -9.88539815e-01 6.51832044e-01 1.02032796e-02 -1.37034565e-01 -3.40347916e-01 -1.07662714e+00 -2.88649023e-01 -2.83941776e-01 -3.61973792e-01 9.91934121e-01 -1.06857347e+00 -6.84567690e-01 9.11501467e-01 -1.39449489e+00 -4.08256531e-01 2.08505783e-02 2.94204708e-02 -3.56109440e-01 -1.94884658e-01 -5.48705041e-01 -4.95958954e-01 -8.81341934e-01 -1.55975878e+00 1.52932870e+00 3.03538859e-01 -4.10207212e-01 -8.13522637e-01 2.68568903e-01 4.19458270e-01 5.34083962e-01 1.99163243e-01 1.02853692e+00 -9.35412347e-01 -4.86956179e-01 -1.99079797e-01 3.19402814e-02 2.19649509e-01 2.96534508e-01 -1.10540427e-01 -9.50726271e-01 -2.69015759e-01 1.12672821e-01 -6.53846920e-01 9.76852596e-01 3.60068560e-01 1.08433151e+00 -2.58062541e-01 -2.36171186e-01 5.17749608e-01 1.21051025e+00 2.49559686e-01 4.73693281e-01 3.72951925e-01 9.98044610e-01 4.23322886e-01 5.68725944e-01 2.38033205e-01 4.60900873e-01 8.22260380e-01 2.05753565e-01 -7.11572543e-02 -2.87401062e-02 6.34892210e-02 4.60352898e-01 1.11754000e+00 -9.51107964e-02 -5.06521404e-01 -8.12906802e-01 6.79885387e-01 -1.93942976e+00 -5.54560125e-01 -2.54993916e-01 1.77791059e+00 1.01689410e+00 2.59542584e-01 2.65339732e-01 -1.43168345e-01 5.31622589e-01 4.62822795e-01 -4.14683640e-01 -6.74698412e-01 -2.40647774e-02 3.15675557e-01 -1.40733555e-01 3.19856644e-01 -1.12991738e+00 9.49631095e-01 6.08834982e+00 1.09943044e+00 -9.92338777e-01 3.80907565e-01 6.33329093e-01 -2.78756708e-01 -5.74318886e-01 -4.54985201e-02 -1.00599492e+00 6.59788430e-01 4.73008424e-01 -3.61851633e-01 1.86941415e-01 9.76253152e-01 -2.88779557e-01 -9.78854150e-02 -1.06652641e+00 8.52356136e-01 3.91876787e-01 -8.40235412e-01 5.17771125e-01 -2.62529641e-01 8.82945657e-01 -2.78422952e-01 2.92756259e-01 5.88761210e-01 5.70853710e-01 -9.72415507e-01 7.54258513e-01 1.63554728e-01 4.82960492e-01 -5.72532833e-01 7.38658309e-01 1.81894347e-01 -1.19856203e+00 1.33159474e-01 4.10047211e-02 -7.63497548e-03 3.21835637e-01 4.80520695e-01 -5.80643237e-01 8.37671936e-01 7.31628537e-01 4.94608134e-01 -6.50022507e-01 8.34908545e-01 -3.38559955e-01 6.53533399e-01 -6.86651841e-02 -3.62292230e-01 2.37123773e-01 -4.03863400e-01 6.26431525e-01 1.41764057e+00 4.05333817e-01 -2.68685400e-01 1.63725227e-01 8.13655734e-01 -1.07966233e-02 2.44033277e-01 -4.25734162e-01 1.36959672e-01 1.28371552e-01 1.64726424e+00 -7.75944114e-01 -2.55148113e-01 -6.36104167e-01 8.08391452e-01 4.21574622e-01 4.51854527e-01 -9.71572876e-01 -5.52906394e-01 4.46631819e-01 -1.89721644e-01 5.25471032e-01 -1.56796172e-01 -5.19656181e-01 -1.10363829e+00 2.29566783e-01 -1.14283395e+00 4.57379371e-01 -9.70083177e-01 -1.27437520e+00 1.01449192e+00 2.43174076e-01 -1.09951127e+00 -2.92960405e-01 -4.95052397e-01 -9.31982934e-01 9.62267756e-01 -1.18581998e+00 -1.56968665e+00 -4.16045666e-01 4.20405865e-01 1.12731016e+00 -9.91221070e-02 7.92533576e-01 1.59621269e-01 -6.39396191e-01 6.20877206e-01 -3.39171976e-01 5.91970654e-03 9.15373981e-01 -1.55069041e+00 6.78678989e-01 9.41596866e-01 2.43098035e-01 5.70563912e-01 7.06944585e-01 -5.63089550e-01 -1.11532402e+00 -9.24389958e-01 8.72909904e-01 -5.08734763e-01 5.25702417e-01 -8.92875314e-01 -8.95554423e-01 9.60362852e-01 7.39229083e-01 -4.31824714e-01 6.23775005e-01 5.84163487e-01 -3.90981585e-01 -2.89672017e-01 -4.57501739e-01 4.19504046e-01 7.87513196e-01 -2.61694580e-01 -5.91593742e-01 4.18908209e-01 8.90768528e-01 -6.91515446e-01 -3.52337956e-01 4.43385273e-01 3.58780861e-01 -6.17041767e-01 7.97370613e-01 -8.27307403e-01 8.10943067e-01 -1.47712141e-01 1.72382414e-01 -1.76313305e+00 -3.42204183e-01 -6.08791292e-01 9.23966989e-02 1.44767630e+00 6.94456100e-01 -5.70057273e-01 3.18938822e-01 2.84246325e-01 -3.86527121e-01 -9.51711595e-01 -6.44953430e-01 -5.00344932e-01 8.64550546e-02 -2.51193911e-01 5.71668446e-01 7.57937133e-01 -4.43725199e-01 1.01991665e+00 -5.44547021e-01 -1.40033334e-01 4.20832783e-01 5.89064896e-01 9.48787749e-01 -1.09790814e+00 -6.21946752e-01 -5.77721953e-01 1.77037686e-01 -6.48048103e-01 8.82711075e-03 -1.02748895e+00 3.30081731e-01 -1.68793273e+00 4.88275498e-01 -4.00733948e-01 -2.79572487e-01 8.00088108e-01 -9.53730524e-01 9.00410861e-02 2.82130539e-01 -9.79528483e-03 -7.40676582e-01 6.13209724e-01 1.29770744e+00 -2.46996462e-01 -1.72801375e-01 2.48657197e-01 -8.85620475e-01 4.90488082e-01 8.05952966e-01 -8.14017236e-01 -5.51220834e-01 -9.20373917e-01 7.99859345e-01 -1.34745285e-01 1.17463015e-01 -7.91463375e-01 8.94690007e-02 -1.28231302e-01 9.36322287e-02 -4.98593092e-01 3.79985660e-01 -4.01833057e-01 2.68677682e-01 1.10221319e-01 -3.91888052e-01 4.57300156e-01 3.63293082e-01 5.53142190e-01 -3.16911221e-01 -2.08133280e-01 2.44504392e-01 -4.60369773e-02 -3.00982594e-01 7.46170804e-02 -3.46593261e-01 2.69446254e-01 1.06091106e+00 1.89067706e-01 -4.10658479e-01 -2.37138420e-01 -6.34554386e-01 1.99651510e-01 -1.25316173e-01 7.36221969e-01 4.30787444e-01 -1.36621845e+00 -1.03205359e+00 5.60704283e-02 1.38286725e-01 5.16351700e-01 4.38642293e-01 5.30318856e-01 -6.21740334e-02 3.48940521e-01 -2.77473360e-01 -5.57902217e-01 -1.35405493e+00 4.20407325e-01 1.12122409e-01 -7.67178178e-01 -4.38062876e-01 1.04539299e+00 7.37489283e-01 -3.55425656e-01 -4.40276153e-02 -2.70301670e-01 -3.32912266e-01 4.32886481e-01 2.60743141e-01 3.33278114e-03 2.15141773e-01 -3.66554976e-01 -1.62525773e-01 7.08856583e-01 -3.90328974e-01 -1.31320804e-01 1.25871789e+00 1.52470648e-01 6.47460446e-02 7.67618835e-01 9.47041988e-01 -1.35484979e-01 -9.95284915e-01 -3.13510776e-01 -3.27016234e-01 -1.77702636e-01 -2.18725175e-01 -9.69074309e-01 -1.15093040e+00 8.58857512e-01 -1.31496519e-01 2.93996751e-01 1.23896182e+00 1.63991004e-01 7.55825698e-01 2.28024676e-01 1.82364196e-01 -8.29842269e-01 5.08171022e-01 4.29103285e-01 1.19377136e+00 -1.12865794e+00 -1.83149248e-01 -4.04889852e-01 -1.07579195e+00 7.84603894e-01 7.87809372e-01 -2.42687881e-01 3.69734287e-01 4.17826384e-01 1.76178783e-01 -2.34710038e-01 -1.10252309e+00 -9.59627926e-02 4.44225401e-01 4.24936116e-01 4.98983771e-01 1.89090282e-01 -5.68198800e-01 1.30122757e+00 -2.95288265e-01 -1.28441304e-01 2.35560611e-01 8.52723598e-01 -2.73281224e-02 -1.30114460e+00 -4.81383018e-02 5.79600871e-01 -6.57121539e-01 -3.02142292e-01 -2.80021071e-01 7.08776355e-01 -1.15178227e-01 7.35950768e-01 -7.15741962e-02 -3.73108655e-01 4.72899079e-01 3.73984635e-01 4.18518722e-01 -8.40470493e-01 -1.06268144e+00 2.08209664e-01 1.27762318e-01 -2.45158210e-01 -4.36627328e-01 -7.93648422e-01 -6.84644997e-01 1.18792146e-01 -1.55036926e-01 7.27094412e-02 5.88098228e-01 1.10512280e+00 4.06636477e-01 1.06338704e+00 5.16058624e-01 -7.05025733e-01 -4.33612436e-01 -1.33321142e+00 -1.80397168e-01 4.72398460e-01 3.63161340e-02 -5.45357108e-01 -2.12965626e-02 1.74996406e-01]
[11.48651123046875, 6.680897235870361]
cacc839d-b5ed-4355-9b2f-5f45cf3a372b
communication-efficient-tensor-factorization
2109.01718
null
https://arxiv.org/abs/2109.01718v2
https://arxiv.org/pdf/2109.01718v2.pdf
Communication Efficient Generalized Tensor Factorization for Decentralized Healthcare Networks
Tensor factorization has been proved as an efficient unsupervised learning approach for health data analysis, especially for computational phenotyping, where the high-dimensional Electronic Health Records (EHRs) with patients' history of medical procedures, medications, diagnosis, lab tests, etc., are converted to meaningful and interpretable medical concepts. Federated tensor factorization distributes the tensor computation to multiple workers under the coordination of a central server, which enables jointly learning the phenotypes across multiple hospitals while preserving the privacy of the patient information. However, existing federated tensor factorization algorithms encounter the single-point-failure issue with the involvement of the central server, which is not only easily exposed to external attacks but also limits the number of clients sharing information with the server under restricted uplink bandwidth. In this paper, we propose CiderTF, a communication-efficient decentralized generalized tensor factorization, which reduces the uplink communication cost by leveraging a four-level communication reduction strategy designed for a generalized tensor factorization, which has the flexibility of modeling different tensor distribution with multiple kinds of loss functions. Experiments on two real-world EHR datasets demonstrate that CiderTF achieves comparable convergence with a communication reduction up to 99.99%.
['Joyce C. Ho', 'Sivasubramanium Bhavani', 'Li Xiong', 'Jian Lou', 'Qiuchen Zhang', 'Jing Ma']
2021-09-03
null
null
null
null
['computational-phenotyping']
['medical']
[-3.40702653e-01 -1.20093383e-01 5.13484627e-02 -2.31225654e-01 -4.76954639e-01 -5.33173561e-01 -4.42385733e-01 5.24737656e-01 -1.97353631e-01 5.98364770e-01 3.01221371e-01 -5.16774178e-01 -4.65107083e-01 -6.59355462e-01 -4.47970837e-01 -9.20422375e-01 -4.32379454e-01 5.15273213e-01 -2.86923379e-01 3.39339077e-01 -4.88199919e-01 1.36030376e-01 -6.34834468e-01 4.95400101e-01 5.47935128e-01 1.20042789e+00 -3.14020813e-01 5.45446754e-01 9.36357379e-02 6.84238017e-01 -4.13066030e-01 -6.98497474e-01 2.80859381e-01 1.87866136e-01 -6.66755736e-01 1.16659574e-01 -3.41315329e-01 -6.00464523e-01 -5.98475575e-01 8.52481425e-01 6.51816845e-01 -3.36444825e-01 7.70334974e-02 -1.33916140e+00 -3.59819233e-01 8.38086128e-01 -5.28197587e-01 -1.97968595e-02 2.42738396e-01 -1.92982834e-02 7.20344424e-01 -3.27170074e-01 4.19964820e-01 1.02912009e+00 6.98751450e-01 2.79300988e-01 -1.23191714e+00 -6.10804260e-01 -8.98782164e-02 -1.01872303e-01 -1.43564367e+00 7.02950656e-02 3.62552822e-01 -5.37474155e-01 3.85356694e-01 7.39761353e-01 3.84459764e-01 6.43761158e-01 1.70835063e-01 7.02922583e-01 7.89113164e-01 1.33933648e-01 2.28044569e-01 1.12462923e-01 2.43182898e-01 5.41729569e-01 4.33548927e-01 -2.74456501e-01 -4.94082958e-01 -1.32649744e+00 5.01095355e-01 8.25323701e-01 -4.58380073e-01 -2.60122895e-01 -1.73342466e+00 3.68713319e-01 1.21236436e-01 2.57260576e-02 -6.59776151e-01 2.97043882e-02 8.04902136e-01 4.42512155e-01 4.66935635e-01 -1.95062011e-01 -7.38097787e-01 1.58827737e-01 -3.37394685e-01 -3.74317579e-02 8.05420280e-01 1.02123046e+00 6.17500365e-01 -2.15067923e-01 -5.12233496e-01 2.49067098e-01 2.13647157e-01 6.29893124e-01 2.92387426e-01 -6.61028266e-01 6.50626600e-01 8.04604888e-01 1.56265602e-01 -1.12845552e+00 -3.85619581e-01 -5.48580527e-01 -1.51667523e+00 -9.00318921e-01 9.54837948e-02 -7.57479250e-01 -1.61930010e-01 1.56952322e+00 1.01947963e+00 3.83823544e-01 4.21396866e-02 9.73775804e-01 3.92928958e-01 4.52330381e-01 -1.40980303e-01 -5.60853064e-01 1.78802156e+00 -4.45605695e-01 -8.15770209e-01 8.12721968e-01 9.97719526e-01 -7.00619876e-01 5.42755783e-01 6.10692739e-01 -5.87627292e-01 1.13796629e-01 -5.10758936e-01 9.86742526e-02 6.58713339e-04 3.58862191e-01 1.24445343e+00 8.64078164e-01 -6.39367640e-01 3.18968415e-01 -9.75404620e-01 -4.05145474e-02 5.43677986e-01 4.62861896e-01 -6.37878418e-01 -3.71157110e-01 -8.46209168e-01 -2.99110383e-01 8.01493675e-02 5.75225316e-02 -7.24019110e-01 -1.03151751e+00 -3.22168678e-01 2.45462790e-01 4.13439929e-01 -1.29965818e+00 9.15161014e-01 -1.79168791e-01 -1.14155948e+00 1.04649976e-01 7.84251466e-02 -2.20895484e-01 4.99940842e-01 -1.42499596e-01 -3.99298936e-01 1.49830550e-01 1.26271009e-01 -4.33583707e-01 5.44633389e-01 -6.56069875e-01 -5.02880096e-01 -9.07366037e-01 -1.37118608e-01 -2.00215057e-01 -7.92563021e-01 -1.10176198e-01 -2.69260198e-01 -6.92840338e-01 2.06989259e-01 -9.34463978e-01 -4.25164133e-01 5.26010953e-02 -7.83872306e-01 -1.81754097e-01 8.52121532e-01 -6.87340617e-01 1.19244385e+00 -2.49348855e+00 4.30086143e-02 3.53575915e-01 8.13024342e-01 -6.43200949e-02 3.48299205e-01 5.90396941e-01 2.92144287e-02 -1.27194017e-01 5.37534319e-02 -3.65412444e-01 -2.23631561e-01 1.86798736e-01 -3.78159732e-01 5.97527146e-01 -4.74866003e-01 4.01895106e-01 -8.57535839e-01 -3.22321981e-01 -2.58859813e-01 4.91366088e-01 -7.92841911e-01 3.03482234e-01 4.10189107e-02 6.85394347e-01 -1.12745297e+00 6.15390420e-01 8.89548838e-01 -6.89944148e-01 7.24157572e-01 -3.53149652e-01 2.62416482e-01 -9.35948566e-02 -1.29530191e+00 1.65753245e+00 -8.01723003e-02 -3.56202275e-01 4.49888945e-01 -9.67499912e-01 4.16831493e-01 8.59497309e-01 1.18643165e+00 1.36127621e-01 2.35508233e-01 3.81957032e-02 -2.56103069e-01 -5.18338084e-01 -5.94067536e-02 2.99391244e-02 -1.62363693e-01 9.56549942e-01 -1.60525665e-01 9.81377602e-01 -3.06209415e-01 6.14730239e-01 1.36072350e+00 -5.97212315e-01 3.68877798e-02 -1.81903020e-01 3.40965241e-01 -3.49493504e-01 8.82173836e-01 3.27093750e-01 1.63426638e-01 1.18804239e-01 5.06953180e-01 -7.91648626e-01 -6.89568996e-01 -9.22098398e-01 1.39815798e-02 7.14365363e-01 -1.55768365e-01 -8.45361710e-01 -5.17751515e-01 -8.72475564e-01 6.10029280e-01 -7.41783828e-02 -4.65650082e-01 6.10398240e-02 -8.05776641e-02 -1.08933723e+00 6.30133152e-01 2.75738537e-01 1.97199225e-01 7.42294043e-02 -3.80056143e-01 2.77400345e-01 -4.51413602e-01 -1.20638931e+00 -7.78959632e-01 -1.82716206e-01 -1.01715589e+00 -1.21191514e+00 -5.50282478e-01 -2.08006561e-01 8.64937723e-01 5.07326007e-01 5.56873202e-01 1.25011563e-01 -6.14533782e-01 6.80162072e-01 -1.46166071e-01 -1.92647368e-01 1.13992058e-01 -3.20914954e-01 5.46701729e-01 8.81860137e-01 8.86718109e-02 -6.25417411e-01 -9.82602000e-01 2.21414328e-01 -1.07721615e+00 -2.59976536e-01 1.49406627e-01 8.78082037e-01 5.69669425e-01 2.65575290e-01 4.76014078e-01 -1.11797488e+00 5.54482818e-01 -8.33073735e-01 -3.66755098e-01 3.56485695e-01 -5.69633245e-01 7.67652004e-04 8.38647127e-01 -2.08219185e-01 -8.01723301e-01 1.21745303e-01 4.49926436e-01 -5.67832768e-01 1.50886983e-01 3.72265607e-01 -1.72846854e-01 5.17226644e-02 1.76838562e-01 3.65005553e-01 1.93854496e-02 -8.44566584e-01 1.90320134e-01 1.01086974e+00 7.30275959e-02 -7.23377705e-01 5.83108366e-01 7.00531781e-01 2.76341170e-01 -5.99853814e-01 -3.69563103e-01 -7.44508743e-01 -1.01754703e-01 1.68938473e-01 5.49419999e-01 -1.25224972e+00 -1.65223825e+00 2.65336096e-01 -1.13658392e+00 5.30049801e-01 -6.26125112e-02 9.68985677e-01 -1.11749947e-01 8.51309001e-01 -9.35296297e-01 -6.91804588e-01 -7.97150433e-01 -8.74119520e-01 9.32385564e-01 -3.03703576e-01 2.56994486e-01 -6.02912486e-01 -8.66101757e-02 6.75370395e-01 1.78861886e-01 2.96092033e-01 1.20209563e+00 -5.36238015e-01 -8.70362997e-01 -2.35840842e-01 -2.02364251e-01 2.43196517e-01 2.76346207e-01 -3.75392944e-01 -6.39241159e-01 -7.20018864e-01 1.99248269e-01 -1.68710828e-01 1.54470623e-01 7.82779083e-02 1.47033072e+00 -8.97561252e-01 -5.83292782e-01 8.57620895e-01 1.19655430e+00 -1.03772886e-01 7.99134001e-02 -4.84203100e-01 8.76887202e-01 3.32448572e-01 4.28270429e-01 1.32064950e+00 5.19973755e-01 4.02469993e-01 1.06497012e-01 -8.41936991e-02 5.04669547e-01 -1.10373348e-01 -8.77736732e-02 1.21576142e+00 -4.23171781e-02 -1.08921133e-01 -5.96265972e-01 3.15246910e-01 -2.00255775e+00 -3.07036251e-01 -1.28310680e-01 2.43268561e+00 7.22324312e-01 -6.66082621e-01 2.47979701e-01 1.67577788e-01 4.83613729e-01 -4.37529474e-01 -5.83684862e-01 -1.05719932e-03 3.09117436e-01 -3.64857197e-01 7.27046549e-01 -2.62350112e-01 -8.20085168e-01 5.29775731e-02 5.27148676e+00 4.90101159e-01 -1.16004276e+00 5.55140793e-01 8.02167714e-01 -1.75757900e-01 -2.28718519e-01 -1.48845628e-01 -2.64935702e-01 6.27768993e-01 7.04596817e-01 -2.76450455e-01 4.44537342e-01 9.84637797e-01 1.32288024e-01 2.54639298e-01 -1.01763868e+00 1.36620378e+00 -4.28551704e-01 -1.23278487e+00 1.38674736e-01 3.16989928e-01 5.29167771e-01 4.59571835e-03 1.10385120e-01 -2.37223685e-01 2.13297620e-01 -4.15505648e-01 -5.80197461e-02 3.91308695e-01 8.99561882e-01 -9.11475539e-01 7.38449037e-01 4.42552090e-01 -1.09236228e+00 -4.02334481e-01 -4.94432390e-01 4.13149819e-02 4.66066003e-02 1.17618883e+00 -1.03659117e+00 1.10607803e+00 9.03937995e-01 3.08574468e-01 -7.38439485e-02 7.45843232e-01 4.93968010e-01 7.08923340e-01 -6.02962554e-01 2.50116080e-01 -8.85709152e-02 -2.00159580e-01 4.40423906e-01 8.88470769e-01 4.31654334e-01 5.18056989e-01 6.82679355e-01 2.49892741e-01 -2.73172885e-01 5.87687910e-01 -5.43597341e-01 -1.86751232e-01 3.76819968e-01 1.27677810e+00 -3.26310098e-01 -3.37366998e-01 -2.75131553e-01 8.99369240e-01 1.70854822e-01 2.95622826e-01 -5.71116030e-01 -2.65610814e-01 9.11665261e-01 2.72691101e-01 2.40244251e-03 -2.48486012e-01 -1.89619750e-01 -1.49844778e+00 2.42417231e-01 -8.90214264e-01 8.49593043e-01 -1.65508181e-01 -1.32296157e+00 3.97411376e-01 -4.30208802e-01 -1.25191915e+00 8.80897939e-02 -2.00742334e-01 -8.24897066e-02 5.84834397e-01 -9.96736765e-01 -1.06972325e+00 -4.31449935e-02 1.21559489e+00 -5.39169252e-01 -2.09973648e-01 1.10631859e+00 8.86631727e-01 -9.76696908e-01 1.01309133e+00 4.51513886e-01 3.28839064e-01 4.98194754e-01 -8.26982141e-01 -2.63627797e-01 6.46493495e-01 -2.02152014e-01 9.80895162e-01 3.31528097e-01 -5.86315870e-01 -2.25392413e+00 -1.29311299e+00 7.86788106e-01 -1.59714650e-02 4.52616394e-01 -5.11913061e-01 -7.07777858e-01 6.76183701e-01 -2.62326151e-01 4.62446362e-01 1.42335618e+00 3.00617844e-01 -6.22546911e-01 -7.71417201e-01 -1.43054557e+00 2.57039249e-01 7.03992605e-01 -5.80441117e-01 2.58937359e-01 9.32965159e-01 9.03859973e-01 -1.11952171e-01 -1.57031035e+00 2.48542711e-01 3.19522381e-01 -5.43573976e-01 7.62395799e-01 -1.07531750e+00 -3.15154761e-01 -5.79841316e-01 -1.42659083e-01 -9.16771770e-01 -4.25497651e-01 -1.18555021e+00 -2.87205577e-01 1.09902096e+00 1.17067218e-01 -9.57558930e-01 8.41245174e-01 7.77351558e-01 4.06396866e-01 -6.81292951e-01 -1.22563636e+00 -5.10188520e-01 -7.58260965e-01 -2.14450464e-01 1.03002524e+00 1.26251066e+00 4.21118945e-01 3.53122443e-01 -6.69781268e-01 5.84221005e-01 9.12818253e-01 1.90423071e-01 9.80228007e-01 -1.00992036e+00 -6.97364509e-01 5.30265927e-01 -5.71503460e-01 -7.03389943e-01 -4.08912957e-01 -9.10753429e-01 -5.83128452e-01 -1.01825154e+00 4.36177671e-01 -7.46644080e-01 -5.30085802e-01 6.63618267e-01 -7.52872527e-02 -1.70221552e-01 1.68615859e-02 2.74298370e-01 -5.98500252e-01 3.71888876e-01 1.18071175e+00 -9.19897333e-02 6.36738762e-02 2.03078389e-01 -7.17278481e-01 2.80938774e-01 4.60789412e-01 -6.31907105e-01 -7.80205190e-01 -6.38029277e-01 2.51363128e-01 7.25814104e-01 9.80933234e-02 -7.56184161e-01 3.85793537e-01 -6.00282662e-03 1.73840851e-01 -4.25640434e-01 1.70014456e-01 -1.17836511e+00 6.45473361e-01 8.46170306e-01 4.98106144e-02 1.66743353e-01 -2.18224764e-01 1.07528496e+00 -1.65443942e-02 7.33537912e-01 2.04700410e-01 7.59874508e-02 3.97927403e-01 8.99254560e-01 -1.11578278e-01 -2.84164935e-01 1.22568989e+00 4.30893600e-01 -1.44939408e-01 -1.36709765e-01 -7.67216146e-01 5.20256341e-01 5.11149541e-02 5.68430945e-02 5.62434435e-01 -1.12945247e+00 -8.43207181e-01 3.34286988e-01 5.08132130e-02 1.35537475e-01 8.72787595e-01 1.14639235e+00 -2.11824879e-01 5.29962897e-01 2.61130303e-01 -6.47126555e-01 -1.29616666e+00 1.08462560e+00 -1.53061107e-01 -2.76227742e-01 -8.50901365e-01 5.57008505e-01 3.13939989e-01 -4.02520686e-01 3.18568349e-01 -3.72185856e-01 3.71683270e-01 -1.05819136e-01 7.58508980e-01 5.47101617e-01 2.59288073e-01 -2.72413701e-01 -4.09812540e-01 7.94150084e-02 -2.20136166e-01 4.91905183e-01 1.22886896e+00 -2.53904641e-01 -6.04306161e-01 5.43889552e-02 1.31637573e+00 -3.14552486e-02 -5.03960669e-01 -5.82402468e-01 -3.02622080e-01 -6.24796569e-01 4.92501967e-02 -5.24813294e-01 -1.56486487e+00 5.14191806e-01 4.23519015e-01 2.74400972e-03 1.17017663e+00 -3.24027479e-01 1.28805113e+00 5.32456875e-01 9.08177316e-01 -5.27134001e-01 -3.51839870e-01 -2.17435416e-02 1.73323616e-01 -6.53553009e-01 5.69216646e-02 -6.63167238e-01 -5.93769789e-01 9.12835419e-01 1.31612808e-01 3.90783906e-01 1.16131973e+00 3.39284807e-01 -2.54412740e-02 -1.87047780e-01 -1.06087899e+00 6.37107670e-01 -3.59147370e-01 4.39669281e-01 1.88589782e-01 7.40537822e-01 -4.03960794e-01 1.06279898e+00 2.43974760e-01 3.60040009e-01 3.38439673e-01 9.41214204e-01 2.38304362e-01 -1.13397932e+00 -5.35384774e-01 6.68711483e-01 -9.45302308e-01 -2.93648411e-02 1.48770183e-01 -2.19880939e-01 2.80705720e-01 8.48808408e-01 -2.55845338e-01 -5.72098017e-01 2.24360585e-01 4.31415718e-03 -6.57783402e-03 -5.64012885e-01 -6.88389361e-01 2.06346095e-01 -1.12372965e-01 -7.91785359e-01 6.84915558e-02 -5.78066587e-01 -1.17065871e+00 -4.71050203e-01 -9.06844661e-02 6.08823478e-01 5.92435002e-01 7.03190744e-01 1.09210920e+00 3.64800245e-01 1.28400922e+00 1.67216331e-01 -9.36870515e-01 -5.79689384e-01 -1.08933771e+00 2.74655938e-01 4.59803849e-01 -2.01032713e-01 7.38154771e-03 -4.31407280e-02]
[6.2257843017578125, 6.3494768142700195]
ac93780b-8fc5-4f78-95aa-a1f90363238c
neumann-network-with-recursive-kernels-for
null
null
http://openaccess.thecvf.com//content/CVPR2023/html/Quan_Neumann_Network_With_Recursive_Kernels_for_Single_Image_Defocus_Deblurring_CVPR_2023_paper.html
http://openaccess.thecvf.com//content/CVPR2023/papers/Quan_Neumann_Network_With_Recursive_Kernels_for_Single_Image_Defocus_Deblurring_CVPR_2023_paper.pdf
Neumann Network With Recursive Kernels for Single Image Defocus Deblurring
Single image defocus deblurring (SIDD) refers to recovering an all-in-focus image from a defocused blurry one. It is a challenging recovery task due to the spatially-varying defocus blurring effects with significant size variation. Motivated by the strong correlation among defocus kernels of different sizes and the blob-type structure of defocus kernels, we propose a learnable recursive kernel representation (RKR) for defocus kernels that expresses a defocus kernel by a linear combination of recursive, separable and positive atom kernels, leading to a compact yet effective and physics-encoded parametrization of the spatially-varying defocus blurring process. Afterwards, a physics-driven and efficient deep model with a cross-scale fusion structure is presented for SIDD, with inspirations from the truncated Neumann series for approximating the matrix inversion of the RKR-based blurring operator. In addition, a reblurring loss is proposed to regularize the RKR learning. Extensive experiments show that, our proposed approach significantly outperforms existing ones, with a model size comparable to that of the top methods.
['Hui Ji', 'Zicong Wu', 'Yuhui Quan']
2023-01-01
null
null
null
cvpr-2023-1
['deblurring']
['computer-vision']
[ 3.17609280e-01 -5.16610682e-01 4.37441856e-01 -3.53153914e-01 -4.99895006e-01 -3.26916605e-01 4.06789213e-01 -4.51778114e-01 -2.00055629e-01 8.57211351e-01 7.23575294e-01 3.68910544e-02 -7.31950581e-01 -2.26148829e-01 -8.60661089e-01 -1.35895872e+00 -1.68608904e-01 1.60533488e-01 -2.10567731e-02 1.86916396e-01 3.42018932e-01 5.47083855e-01 -1.46495605e+00 6.91101253e-02 1.39475870e+00 8.65425706e-01 6.83730483e-01 8.66030574e-01 4.09236968e-01 1.32452893e+00 -2.82557726e-01 1.12695754e-01 9.91013721e-02 -3.07216138e-01 -6.49609327e-01 -1.77339707e-02 8.17622781e-01 -6.72162831e-01 -1.13642263e+00 1.31044924e+00 5.57542503e-01 4.14431542e-01 9.86729264e-01 -5.87724388e-01 -9.40664411e-01 4.31196630e-01 -7.17466176e-01 4.38470483e-01 2.18458056e-01 2.56931335e-01 4.60471690e-01 -9.13444340e-01 3.05519819e-01 1.12800038e+00 5.03431261e-01 4.44574237e-01 -1.20024610e+00 -2.90655762e-01 -3.47999215e-01 3.59426141e-01 -1.37371826e+00 -4.61284757e-01 6.06062710e-01 -6.19700849e-01 8.39377642e-01 2.73101211e-01 4.37327713e-01 7.03824759e-01 4.94825155e-01 4.58037525e-01 1.42144632e+00 -3.35237920e-01 2.55868733e-01 -3.22289020e-01 3.11057627e-01 6.80977166e-01 5.17464876e-01 4.27855760e-01 -5.45328498e-01 -3.96447361e-01 1.13027263e+00 1.15678370e-01 -1.41279697e+00 -6.19776785e-01 -1.29843271e+00 4.60595936e-01 7.20992863e-01 2.03604206e-01 -4.81262624e-01 2.88563460e-01 -4.32923287e-02 -5.30779287e-02 6.09145463e-01 6.57597184e-01 -2.99897105e-01 -5.04794233e-02 -8.57715905e-01 3.67604464e-01 7.39213705e-01 5.51722944e-01 1.08320725e+00 -1.43949628e-01 -4.71187919e-01 7.09002793e-01 2.77964562e-01 9.68785465e-01 2.88455904e-01 -1.02370608e+00 -1.40716314e-01 -7.62411067e-03 7.68896759e-01 -6.64630115e-01 -2.34343871e-01 -3.43493313e-01 -1.30778742e+00 1.50566250e-01 2.17279851e-01 -2.55695693e-02 -1.01434886e+00 1.61131060e+00 2.23849222e-01 1.19596136e+00 2.68914163e-01 1.54517412e+00 7.81930447e-01 6.68767035e-01 -3.85592461e-01 -4.14015859e-01 1.23996615e+00 -9.49514925e-01 -8.72613728e-01 -1.32229939e-01 1.69372022e-01 -7.60815024e-01 6.77756667e-01 2.08673924e-01 -1.16929340e+00 -4.03280258e-01 -8.70036066e-01 -3.56306881e-01 2.91975319e-01 -9.65854973e-02 8.14266026e-01 2.14370757e-01 -1.29925966e+00 4.41537112e-01 -5.67719758e-01 2.14498058e-01 3.90494972e-01 8.04740638e-02 -2.03799114e-01 -5.39032817e-01 -1.10246825e+00 1.01411080e+00 2.40948901e-01 2.76669472e-01 -9.53808606e-01 -1.42595637e+00 -7.91557789e-01 1.61081529e-03 4.31905246e-05 -1.13050020e+00 1.00797474e+00 -2.05962643e-01 -1.49286664e+00 7.88665235e-01 -2.77382851e-01 -4.93296742e-01 2.59053320e-01 -4.78771120e-01 -8.73874649e-02 4.15638119e-01 -3.39833677e-01 3.45624201e-02 1.58236659e+00 -1.30228305e+00 -2.74097651e-01 -1.50710389e-01 2.68100590e-01 5.07057786e-01 1.80328757e-01 -2.82929093e-01 6.85334802e-02 -4.40707713e-01 -1.51169360e-01 -6.63213253e-01 -1.25606492e-01 -3.33180755e-01 -7.09542409e-02 1.82263106e-01 6.81767583e-01 -6.55719995e-01 1.31868684e+00 -2.31061292e+00 7.20739484e-01 -5.75171411e-01 6.72876179e-01 3.01172733e-01 1.03745535e-01 1.68198466e-01 -1.29168659e-01 -6.39434695e-01 -5.86604774e-01 -2.64698714e-01 -1.21727429e-01 1.60394758e-02 -6.08901501e-01 8.30345750e-01 1.11715654e-02 9.52470958e-01 -1.25836527e+00 4.84297164e-02 5.44236243e-01 1.03463984e+00 -5.64313591e-01 7.65142620e-01 -4.55956720e-02 7.26248682e-01 -1.55552298e-01 2.57366866e-01 1.52942073e+00 -4.96114194e-01 -4.14834380e-01 -8.38697553e-01 -3.94199133e-01 -1.33591533e-01 -8.09908271e-01 1.98535860e+00 -5.64501643e-01 4.32468027e-01 5.96622467e-01 -8.29850674e-01 3.24919224e-01 2.66955830e-02 2.00158373e-01 -1.10074200e-01 3.38010630e-03 4.56262469e-01 -2.51248837e-01 -7.84899354e-01 5.15953124e-01 -3.78623068e-01 3.87193531e-01 3.57405603e-01 7.08900318e-02 -7.58016288e-01 -2.89397657e-01 1.17994130e-01 1.26665103e+00 1.42326960e-02 -3.23940255e-02 -7.51050889e-01 7.49873817e-01 -6.16285145e-01 -1.98098183e-01 9.28159058e-01 -1.34310082e-01 8.08033407e-01 -3.11777472e-01 -4.28322256e-01 -7.20879197e-01 -1.03859770e+00 -7.03230798e-01 4.17349786e-01 5.81319273e-01 2.11749867e-01 -9.21571612e-01 -2.17582569e-01 1.97687209e-01 5.23272455e-01 -6.03097916e-01 -4.00196642e-01 -5.00917733e-01 -1.04512942e+00 3.68086919e-02 -2.81077445e-01 7.74215639e-01 -6.44847691e-01 -7.45244920e-01 -1.61421046e-01 -4.07032490e-01 -1.07022560e+00 -8.24752748e-01 1.87745586e-01 -5.10252655e-01 -1.12551117e+00 -1.22229815e+00 -6.58476055e-01 4.52586889e-01 9.14283514e-01 9.28132713e-01 -3.17625612e-01 -5.18237054e-01 5.94078004e-01 4.92047481e-02 4.85006198e-02 -2.98931330e-01 -5.29309094e-01 2.16151834e-01 3.97318155e-01 5.15710302e-02 -7.04641879e-01 -1.16764867e+00 -1.63899004e-01 -1.15758610e+00 2.80961722e-01 6.50336623e-01 8.26429844e-01 1.32487223e-01 2.78551161e-01 3.40109579e-02 -2.40761548e-01 6.71748102e-01 -4.51289326e-01 -5.02115250e-01 1.77190825e-01 -3.01265597e-01 3.80439401e-01 2.70426601e-01 -5.05192816e-01 -1.47201681e+00 -5.97553432e-01 5.39479613e-01 -9.47791159e-01 -1.99611276e-01 2.77685136e-01 1.96899340e-01 -5.40500939e-01 7.34219134e-01 4.08445328e-01 -7.77337477e-02 -4.88132238e-01 5.31427026e-01 6.91640913e-01 8.34281743e-01 -4.83292103e-01 8.57284486e-01 7.15871990e-01 1.08118862e-01 -1.01690912e+00 -1.22492421e+00 -5.69843233e-01 -2.87752181e-01 -2.19835714e-01 8.84131491e-01 -1.10946512e+00 -8.61786544e-01 1.40763879e+00 -1.29472446e+00 -5.01250088e-01 -4.16304678e-01 9.16567147e-01 -7.55105019e-01 5.98732293e-01 -9.07157183e-01 -7.87698865e-01 -1.93404332e-01 -1.02976346e+00 1.19560885e+00 5.11901081e-01 5.79959810e-01 -1.10562074e+00 2.06196517e-01 3.73061389e-01 8.24222922e-01 5.41419908e-02 8.84114385e-01 3.33985895e-01 -9.86418605e-01 4.67763871e-01 -8.35779011e-01 4.94602650e-01 4.72423524e-01 -5.82667351e-01 -1.11906254e+00 -6.52764797e-01 9.21555519e-01 -3.19717079e-01 1.30038679e+00 1.05779946e+00 9.14358139e-01 -1.80143341e-01 -2.90545464e-01 1.18410289e+00 1.54810059e+00 -1.36843011e-01 6.07124627e-01 -2.23270461e-01 9.59868848e-01 2.27664366e-01 2.84784198e-01 6.37043417e-01 2.15765506e-01 3.55646670e-01 4.14598882e-01 -5.64022474e-02 -4.63289946e-01 2.43340313e-01 1.80813462e-01 6.52966857e-01 -1.77975729e-01 -1.25405669e-01 -4.70560551e-01 3.11382145e-01 -1.89174509e+00 -8.74799132e-01 -2.23343298e-01 2.37083030e+00 1.12571263e+00 -5.34181774e-01 -7.12590277e-01 -4.57237750e-01 7.43980229e-01 4.66562420e-01 -8.42287600e-01 2.57705390e-01 -4.02233601e-01 3.31538588e-01 4.15490180e-01 1.22725475e+00 -9.72823739e-01 6.76454544e-01 6.07906246e+00 7.92667866e-01 -1.24724650e+00 -1.37738930e-02 2.97899663e-01 4.82414216e-02 -4.09981549e-01 -1.54035747e-01 -5.50523996e-01 6.50287271e-01 6.86970353e-01 -1.76613897e-01 1.12860036e+00 2.19264012e-02 2.93484479e-01 -3.61329228e-01 -8.87968242e-01 1.49318624e+00 2.07606852e-01 -1.30870628e+00 -6.44397140e-02 -2.80560609e-02 8.57446134e-01 1.99081570e-01 1.74964979e-01 -2.26197585e-01 1.34019390e-01 -1.06861186e+00 4.03562278e-01 1.19641113e+00 9.47989702e-01 -4.79556322e-01 3.95734280e-01 2.56488174e-01 -7.56630123e-01 -4.65768948e-02 -5.82750082e-01 -6.26735538e-02 2.41551161e-01 1.29006624e+00 -2.65048236e-01 6.52067363e-01 8.27723444e-01 1.12725890e+00 -2.68598795e-02 1.24877584e+00 2.42895894e-02 3.88030380e-01 -8.60638022e-02 3.99039656e-01 -1.36001706e-01 -6.68627143e-01 8.10430706e-01 1.21446812e+00 6.25856221e-01 4.41051245e-01 -4.15109485e-01 1.12369919e+00 2.77588684e-02 -7.28932500e-01 -3.57877493e-01 2.28944749e-01 7.59541839e-02 1.27527809e+00 1.64136678e-01 -2.07344100e-01 -1.64860636e-01 1.30730128e+00 3.05570841e-01 9.86501873e-01 -9.05835390e-01 -2.98375070e-01 1.08534431e+00 1.14042595e-01 4.12693501e-01 -1.88407063e-01 3.73594105e-01 -1.79340160e+00 -2.61485219e-01 -3.78333300e-01 -2.67536491e-01 -1.33536673e+00 -1.56664956e+00 4.92233336e-01 2.97208786e-01 -9.05990005e-01 3.37687731e-01 -7.00513840e-01 -3.07272345e-01 1.50540400e+00 -2.31876206e+00 -9.86433804e-01 -7.20016599e-01 6.65916264e-01 5.12198396e-02 3.09764057e-01 6.13447070e-01 1.48688406e-01 -9.40808728e-02 -1.37225822e-01 3.43543649e-01 -5.59447229e-01 1.01780736e+00 -1.42448342e+00 -1.23145372e-01 7.72505581e-01 -6.51976347e-01 8.63968730e-01 9.08693194e-01 -4.50179100e-01 -1.70680082e+00 -9.92302895e-01 3.67766112e-01 -3.64657402e-01 9.38543499e-01 -1.53915241e-01 -1.15158880e+00 2.54457325e-01 4.93229151e-01 3.63948643e-01 3.01739335e-01 -6.09083712e-01 -2.91087270e-01 -7.88943395e-02 -1.14939213e+00 1.91695958e-01 1.14041984e+00 -6.98669612e-01 -5.97920954e-01 4.48071957e-01 8.66960287e-01 -8.83170724e-01 -6.55469179e-01 6.87496126e-01 1.48129702e-01 -1.33868325e+00 1.23108852e+00 -1.46923557e-01 5.63417017e-01 -5.86770236e-01 2.16450659e-03 -1.57172227e+00 -9.52077866e-01 -1.04416740e+00 -7.97267675e-01 6.34566545e-01 -4.43986416e-01 -8.20057213e-01 1.61659926e-01 1.78547025e-01 -4.66862708e-01 -5.54611206e-01 -9.16133404e-01 -3.31467450e-01 -1.57717869e-01 1.00126587e-01 4.10472512e-01 8.46078873e-01 -3.39987367e-01 2.51655906e-01 -5.97601950e-01 7.96105623e-01 1.27479470e+00 2.84103990e-01 1.88072950e-01 -1.04449558e+00 -5.60455203e-01 -1.38584629e-01 -9.56798941e-02 -1.78535056e+00 2.44898081e-01 -7.84731090e-01 3.44294369e-01 -1.21622741e+00 6.43320322e-01 -5.13711512e-01 -4.28152740e-01 -3.70046467e-01 -5.07934153e-01 5.54669835e-02 -2.60815471e-01 5.22615969e-01 -2.88728297e-01 5.48021317e-01 1.72196257e+00 -1.72984719e-01 -2.10222695e-02 -1.10984817e-01 -5.45655489e-01 5.07168412e-01 9.63190943e-02 1.16613001e-01 -6.72125041e-01 -6.30351484e-01 8.84295404e-02 2.62888372e-01 7.25580275e-01 -9.76960123e-01 3.48773152e-01 -2.70108163e-01 1.55765982e-02 -2.48199165e-01 4.42013711e-01 -6.12125397e-01 1.64684772e-01 2.42311656e-01 -1.39208540e-01 -8.77805650e-01 1.90948710e-01 7.75932074e-01 -5.67677021e-02 -3.42417598e-01 1.05220115e+00 -9.73497629e-02 -3.08218032e-01 5.11774004e-01 -2.14272410e-01 1.34603858e-01 5.94064593e-01 1.08259276e-01 -6.61501586e-01 -2.25774810e-01 -1.83972567e-01 -3.07789534e-01 6.60315871e-01 1.22232705e-01 7.23387122e-01 -1.12408483e+00 -8.60877514e-01 3.97037923e-01 -9.09162313e-02 4.51969445e-01 8.07026386e-01 1.14875758e+00 -6.39777660e-01 2.79391259e-01 1.50650889e-01 -6.27204001e-01 -7.35618830e-01 6.20493054e-01 8.18645835e-01 -1.55197144e-01 -6.62948906e-01 1.38179982e+00 9.82178390e-01 -3.82502228e-02 -5.47492318e-02 -8.30365300e-01 1.78592745e-02 -5.20217717e-01 9.58903909e-01 4.16390836e-01 -7.77140111e-02 -4.88659918e-01 -8.00588652e-02 8.06188166e-01 -4.37497161e-02 2.17264235e-01 1.27203596e+00 -3.67035329e-01 -7.69846916e-01 3.00000966e-01 1.12919176e+00 1.55092403e-01 -1.91891241e+00 -3.81314218e-01 -7.66508400e-01 -6.89025819e-01 4.66391683e-01 -8.24985206e-01 -7.49139488e-01 8.25430334e-01 5.71866632e-01 1.91596985e-01 1.32156301e+00 2.13561505e-01 9.07359719e-01 5.33512160e-02 3.69914323e-01 -2.23830700e-01 -2.19529141e-02 6.09435856e-01 1.12694168e+00 -1.05897677e+00 -8.52388889e-02 -3.92139316e-01 -1.45079466e-02 9.98541355e-01 1.64406076e-01 -2.43501768e-01 9.08775568e-01 4.42091674e-01 -2.55065829e-01 -1.08454011e-01 -4.46111500e-01 1.79088756e-01 4.53879237e-01 6.69407129e-01 1.82085723e-01 -8.80901292e-02 6.17805757e-02 8.30830112e-02 2.17363298e-01 2.80453682e-01 6.19378150e-01 4.56391394e-01 -5.86658835e-01 -5.16068876e-01 -2.52199620e-01 2.88124591e-01 -5.61663173e-02 -4.91624862e-01 3.06410789e-01 -1.25829250e-01 1.74967393e-01 7.74845600e-01 7.29901642e-02 2.57914625e-02 6.90249875e-02 -3.32347453e-01 1.06927049e+00 -3.80957574e-01 -1.01223905e-02 1.30985841e-01 -6.83215976e-01 -3.78692120e-01 -8.73165429e-01 -4.61326033e-01 -1.01936841e+00 -1.74045011e-01 -4.02508706e-01 1.74050152e-01 1.46201938e-01 7.72481084e-01 6.08843148e-01 3.21487278e-01 7.46300101e-01 -1.31426442e+00 -6.11427069e-01 -1.07465827e+00 -1.01215577e+00 4.59687382e-01 1.32648957e+00 -7.25865126e-01 -1.20886040e+00 8.26605596e-03]
[11.556395530700684, -2.7897775173187256]
ba469f4e-ec0a-4e56-a6f1-ac693bfd5f8f
analysis-of-drug-repurposing-knowledge-graphs
2212.03911
null
https://arxiv.org/abs/2212.03911v1
https://arxiv.org/pdf/2212.03911v1.pdf
Analysis of Drug repurposing Knowledge graphs for Covid-19
Knowledge graph (KG) is used to represent data in terms of entities and structural relations between the entities. This representation can be used to solve complex problems such as recommendation systems and question answering. In this study, a set of candidate drugs for COVID-19 are proposed by using Drug repurposing knowledge graph (DRKG). DRKG is a biological knowledge graph constructed using a vast amount of open source biomedical knowledge to understand the mechanism of compounds and the related biological functions. Node and relation embeddings are learned using knowledge graph embedding models and neural network and attention related models. Different models are used to get the node embedding by changing the objective of the model. These embeddings are later used to predict if a candidate drug is effective to treat a disease or how likely it is for a drug to bind to a protein associated to a disease which can be modelled as a link prediction task between two nodes. RESCAL performed the best on the test dataset in terms of MR, MRR and Hits@3.
['Ajay Kumar Gogineni']
2022-12-07
null
null
null
null
['knowledge-graph-embedding']
['graphs']
[ 1.80285722e-02 4.08112019e-01 -3.94114077e-01 -1.71251357e-01 2.07836822e-01 -4.57794458e-01 2.86441922e-01 8.89447570e-01 -2.52936184e-01 9.03962672e-01 4.26245421e-01 -4.05680895e-01 -4.84310210e-01 -1.11262643e+00 -6.71823382e-01 -5.79303563e-01 -2.63430327e-01 5.08407116e-01 1.62168935e-01 -8.90738294e-02 7.78778344e-02 6.07604086e-01 -8.25337052e-01 2.80720890e-01 7.33158469e-01 4.86045420e-01 8.58257413e-02 4.20631498e-01 -1.38114646e-01 6.08977199e-01 -3.94652665e-01 -5.16759634e-01 -1.39607042e-01 -3.01026165e-01 -9.45311010e-01 -8.12773228e-01 1.26608424e-02 2.92980880e-01 -4.26009059e-01 9.69056249e-01 6.76234782e-01 1.62102222e-01 9.21998739e-01 -1.06737828e+00 -1.13464046e+00 3.22084188e-01 -1.22738436e-01 3.33541483e-01 4.43013817e-01 -2.26886004e-01 1.22502065e+00 -9.26570892e-01 8.71669471e-01 1.32635498e+00 4.66384351e-01 5.66818833e-01 -1.16052961e+00 -2.69197077e-01 -2.23876610e-01 6.97665095e-01 -1.33530891e+00 2.34299317e-01 2.62176067e-01 -7.22843707e-01 1.43558419e+00 2.14348391e-01 5.77657223e-01 7.01121151e-01 5.99640131e-01 5.79995699e-02 5.05122006e-01 -1.36407629e-01 2.54506022e-01 3.29710841e-01 5.37342727e-01 8.83849204e-01 4.93709058e-01 -1.60931557e-01 -4.99283135e-01 -6.00190938e-01 3.65905374e-01 2.87226498e-01 -5.28737068e-01 -2.12834746e-01 -9.93654072e-01 1.24011028e+00 1.04486513e+00 4.28659081e-01 -4.62470919e-01 6.31837696e-02 3.22270930e-01 1.48302898e-01 3.17148119e-01 1.04229355e+00 -9.15166020e-01 5.46274722e-01 -9.96952876e-02 7.66816139e-02 7.87748575e-01 4.04631764e-01 5.31375945e-01 -2.20975578e-01 -2.27070466e-01 6.42747879e-01 7.00747430e-01 1.24091655e-02 5.62474847e-01 -8.77269059e-02 -1.36261001e-01 1.03776526e+00 -1.46353364e-01 -1.22205353e+00 -6.14762366e-01 -1.73823550e-01 -5.29312193e-01 -1.91921562e-01 1.47954375e-01 -3.03057767e-02 -9.27244961e-01 1.54605651e+00 5.69005013e-01 7.27561474e-01 2.12349162e-01 8.50236475e-01 1.50465322e+00 7.98523545e-01 4.50950593e-01 -3.27665417e-04 1.83363187e+00 -8.15024614e-01 -6.85869813e-01 2.39546612e-01 9.20913219e-01 -2.82648832e-01 3.14453810e-01 -5.78287691e-02 -3.53327692e-01 -1.60154536e-01 -1.04683471e+00 -1.14184082e-01 -1.19319081e+00 -8.36555138e-02 6.67715967e-01 2.53222555e-01 -7.21906245e-01 8.13962579e-01 -6.40092552e-01 -6.37633979e-01 5.42007923e-01 6.10208213e-01 -6.33361459e-01 -3.05229992e-01 -1.77092469e+00 1.26633179e+00 7.05588579e-01 -1.68791562e-01 -6.77360594e-01 -9.54727948e-01 -9.22341049e-01 2.98722893e-01 5.79844154e-02 -9.82381940e-01 1.61477804e-01 -2.35114485e-01 -1.05634320e+00 5.67914426e-01 7.37625062e-02 -5.36848545e-01 -4.33138907e-01 2.66418122e-02 -6.03110492e-01 -3.08639947e-02 -1.44001573e-01 5.66071451e-01 2.61214137e-01 -3.54032844e-01 -1.65439740e-01 -6.35354102e-01 1.67704776e-01 9.99632403e-02 -2.87772477e-01 -1.08539261e-01 -3.12490463e-01 -4.37961906e-01 -3.34452838e-01 -9.28895414e-01 -2.31191069e-01 -1.48286030e-01 -4.75094497e-01 -5.67541897e-01 6.68781638e-01 -7.94059515e-01 1.24276865e+00 -1.69285405e+00 5.56546390e-01 3.42336714e-01 5.09273171e-01 6.82669699e-01 -4.18938488e-01 7.75513172e-01 -5.73470414e-01 5.17807901e-01 8.76955241e-02 8.77002776e-01 -4.05667812e-01 3.82334255e-02 2.23207444e-01 4.34503347e-01 3.70388985e-01 1.06707394e+00 -9.73342240e-01 -2.94950120e-02 -1.53511450e-01 8.62151444e-01 -4.66820478e-01 1.71179399e-01 -5.00242591e-01 2.10154459e-01 -9.08333719e-01 3.71081829e-01 4.34505403e-01 -4.97053921e-01 5.11348128e-01 -6.33711159e-01 5.20970225e-01 1.36740029e-01 -5.59404075e-01 1.15436029e+00 2.38300487e-02 3.17081928e-01 -5.11179566e-01 -1.14494395e+00 8.89660418e-01 4.37797397e-01 3.59493136e-01 -3.15234691e-01 1.41244307e-01 -1.17276184e-01 4.86157119e-01 -6.66179895e-01 1.63853057e-02 7.75795653e-02 3.04067254e-01 2.81622633e-02 1.73917189e-01 6.21050775e-01 -4.82631885e-02 3.43910486e-01 1.50434530e+00 -2.37628117e-01 4.62847084e-01 -2.07407251e-01 6.70992434e-01 -4.96461503e-02 5.78378916e-01 3.68939713e-02 2.99127966e-01 -1.80266082e-01 6.95970833e-01 -6.27650142e-01 -8.50752950e-01 -6.21617019e-01 -3.10762256e-01 6.84982240e-01 -5.53255230e-02 -6.11958921e-01 -4.04293656e-01 -7.42444396e-01 2.94852942e-01 4.61680025e-01 -1.05035400e+00 -5.44057727e-01 6.78558573e-02 -8.83498788e-01 2.29441851e-01 1.26965582e-01 -6.40355796e-02 -9.26259220e-01 -1.71361174e-02 2.79560328e-01 2.32865602e-01 -7.11048961e-01 -3.68890643e-01 2.98028469e-01 -7.01207995e-01 -1.60983229e+00 -4.80716169e-01 -8.93542826e-01 8.04812431e-01 -2.00850412e-01 7.78734267e-01 2.29840904e-01 -4.75159526e-01 -8.65839124e-02 -4.11214262e-01 -5.76108694e-01 -2.07198352e-01 -2.09252954e-01 1.55937746e-01 -4.64995094e-02 6.20059192e-01 -2.67724127e-01 -7.07752883e-01 2.44420558e-01 -8.95460725e-01 -2.09233090e-01 3.69790941e-01 9.41886306e-01 6.54730499e-01 -5.85462786e-02 9.81388927e-01 -1.11382616e+00 1.04188681e+00 -9.96112287e-01 -4.01357085e-01 4.53056216e-01 -7.62903333e-01 4.53632414e-01 3.95159751e-01 -4.23913181e-01 -2.68733889e-01 -8.12343508e-02 -2.15710923e-02 -3.18941712e-01 2.04549611e-01 1.13904905e+00 -1.74567685e-01 -4.44448404e-02 6.09851122e-01 -1.66973889e-01 -1.84612408e-01 -4.84466463e-01 4.64040428e-01 5.02856076e-01 -2.03384563e-01 -1.14787996e-01 2.66919464e-01 -8.55199322e-02 3.54044676e-01 -5.61426342e-01 -6.06525123e-01 -4.34568346e-01 -2.73589253e-01 2.71824896e-01 1.26743376e+00 -6.32919133e-01 -1.03847647e+00 -1.73842072e-01 -1.17890096e+00 2.66287357e-01 2.73210108e-01 7.91109383e-01 1.59980029e-01 1.26338109e-01 -5.38575590e-01 -1.28952071e-01 -6.40594184e-01 -9.75896180e-01 4.70080972e-01 2.32172757e-01 -2.19584778e-01 -1.29426944e+00 5.68970799e-01 2.21847907e-01 1.67129889e-01 4.78101194e-01 1.69379425e+00 -1.35637534e+00 -3.12113374e-01 -2.22204715e-01 -3.18180799e-01 3.71835083e-02 3.84967357e-01 -1.11464791e-01 -4.72301781e-01 -1.03901997e-01 -6.57989740e-01 -1.57954134e-02 7.41809607e-01 4.04606223e-01 8.98790300e-01 -5.10897338e-01 -7.72756457e-01 3.93934637e-01 1.55368996e+00 4.36324775e-01 9.03703749e-01 1.20186523e-01 8.64489198e-01 4.22873288e-01 2.64528632e-01 4.10833508e-02 1.18072584e-01 9.45085526e-01 3.97354692e-01 1.30164707e-02 2.00307935e-01 -2.40588382e-01 1.36806294e-01 4.52421278e-01 1.82730821e-03 -4.68156666e-01 -9.62980986e-01 4.45385337e-01 -1.93729377e+00 -8.94490719e-01 -4.16110903e-01 2.10365558e+00 9.35724378e-01 -3.60765904e-01 -2.05777109e-01 -5.04775524e-01 7.66695023e-01 -4.17208165e-01 -7.83174753e-01 -7.52088845e-01 1.10086851e-01 4.87780452e-01 4.45179939e-01 5.05643845e-01 -7.23371685e-01 1.00466561e+00 5.87795591e+00 7.08468378e-01 -9.63602245e-01 -5.09897098e-02 6.00144029e-01 3.20972502e-01 -2.42338791e-01 1.05997063e-01 -6.91853404e-01 4.34660792e-01 1.33299720e+00 -3.84415597e-01 2.37252995e-01 5.39911151e-01 2.50989139e-01 1.15146279e-01 -1.23864985e+00 7.35480368e-01 4.52842526e-02 -1.78709781e+00 2.64944404e-01 2.06933707e-01 4.68412638e-01 1.15488783e-01 -3.85826349e-01 1.79890767e-01 3.57469618e-01 -1.35495484e+00 -5.33605337e-01 9.61081147e-01 4.45566207e-01 -5.79644144e-01 9.17224944e-01 1.03053354e-01 -9.77146864e-01 3.31789613e-01 -6.36122286e-01 3.30792934e-01 -2.01234654e-01 4.19646680e-01 -1.42794728e+00 7.17215478e-01 3.90656888e-01 9.55015600e-01 -5.37827313e-01 1.12493396e+00 -3.34174246e-01 5.30209005e-01 2.78280713e-02 -2.79941529e-01 7.96937346e-02 -2.62872547e-01 3.12911153e-01 8.63324583e-01 2.09490553e-01 2.02947423e-01 6.64122924e-02 9.07324493e-01 -4.27214354e-01 4.25441027e-01 -7.92897999e-01 -5.51773012e-01 3.96029502e-01 1.42445827e+00 -5.04313052e-01 -2.90967882e-01 -2.46385634e-01 8.22448432e-01 4.21085328e-01 4.04006273e-01 -7.98832476e-01 -4.63954031e-01 8.24097455e-01 2.25167349e-01 3.39580089e-01 3.15311670e-01 5.60422480e-01 -8.00575614e-01 -6.77146494e-01 -6.45825446e-01 5.45468509e-01 -1.01003432e+00 -1.45842326e+00 6.44479692e-01 -2.44131744e-01 -9.37137723e-01 2.72298396e-01 -7.98031151e-01 -4.58731204e-01 1.11007202e+00 -1.46728027e+00 -9.15801823e-01 5.04291691e-02 3.48962516e-01 -8.59381706e-02 -3.59059513e-01 1.37367558e+00 3.50404650e-01 -6.62841976e-01 2.75893211e-01 3.04586500e-01 8.27898830e-02 6.62370622e-01 -1.07352674e+00 6.60758540e-02 -6.42390177e-02 1.34309262e-01 9.08935368e-01 5.57558715e-01 -8.01723778e-01 -1.37454462e+00 -1.48393273e+00 8.84472191e-01 -5.79985678e-01 7.19924569e-01 -1.08703442e-01 -1.06061578e+00 4.20489967e-01 3.05720102e-02 3.32507074e-01 1.33571482e+00 1.83870327e-02 -4.06859517e-01 3.12098652e-01 -1.31313920e+00 3.63273799e-01 6.40833139e-01 -2.68215686e-01 -4.83235270e-01 7.04288483e-01 1.02974093e+00 1.89840384e-02 -1.44656312e+00 2.49181569e-01 3.02002668e-01 -7.16426894e-02 1.06229520e+00 -1.56487584e+00 3.47289443e-01 -6.39313281e-01 -1.98590346e-02 -1.53122354e+00 -7.39643931e-01 -2.20699329e-02 -2.11938962e-01 7.11379647e-01 8.90822887e-01 -6.94301844e-01 4.94754165e-01 4.01812434e-01 1.39474705e-01 -1.21400166e+00 -8.33723664e-01 -4.27294374e-01 -9.57590938e-02 3.63695741e-01 5.70140958e-01 1.47541928e+00 2.86794215e-01 1.03723347e+00 -2.92445272e-01 3.38621199e-01 8.89703166e-03 -1.53855279e-01 2.64767647e-01 -1.56935573e+00 -2.65690088e-01 -5.44432662e-02 -1.03827584e+00 -3.01561147e-01 6.36860952e-02 -1.46345866e+00 -7.42728770e-01 -1.99657297e+00 2.55754352e-01 -1.79331750e-01 -5.65493226e-01 7.01026440e-01 -1.50817782e-01 -9.00780782e-02 -1.85938686e-01 -1.13701209e-01 -4.07406092e-01 4.34243768e-01 9.93423462e-01 -4.57855046e-01 -2.07425401e-01 -3.04718107e-01 -7.80712664e-01 4.86392349e-01 6.82863474e-01 -6.68510914e-01 -5.50396919e-01 1.79974437e-02 6.54510200e-01 2.15592608e-01 1.14494987e-01 -2.83293098e-01 6.25900701e-02 -7.05671012e-02 3.45565349e-01 -1.58366621e-01 1.94874480e-01 -9.73962069e-01 7.32308924e-01 7.25750625e-01 -3.07483792e-01 -4.22524177e-02 3.19800079e-01 1.04833078e+00 -1.22836158e-01 -3.27002674e-01 6.11346185e-01 1.52094260e-01 -4.85868037e-01 6.41598523e-01 -1.12098411e-01 -2.27524638e-01 1.10888731e+00 -9.74214450e-02 -4.73994583e-01 -4.71834019e-02 -1.20469499e+00 2.36862391e-01 9.01503023e-03 5.86457491e-01 8.82859349e-01 -1.45477188e+00 -6.21414781e-01 -2.72391409e-01 5.10565519e-01 -6.61777854e-01 4.13589776e-02 7.24458635e-01 -6.11981571e-01 6.67252421e-01 3.59771727e-03 -8.70070755e-02 -1.59679353e+00 7.79607415e-01 5.47601044e-01 -6.00625515e-01 -2.75108039e-01 7.72961378e-01 1.80256918e-01 -3.85229260e-01 -6.87812939e-02 -1.88345686e-01 -9.27193105e-01 1.93216708e-02 4.55403715e-01 1.10795498e-01 1.18934447e-02 -6.01584256e-01 -7.57373571e-01 4.94607031e-01 -3.90941799e-01 8.31555486e-01 1.66841841e+00 6.90926611e-01 -3.95871371e-01 -4.70554642e-02 1.40674031e+00 -1.88476637e-01 -3.08904827e-01 -3.83725256e-01 1.37917995e-01 -1.93346277e-01 1.80092543e-01 -1.10414028e+00 -1.02466750e+00 5.86315572e-01 8.64298940e-01 -1.02723345e-01 3.91819477e-01 2.27567077e-01 4.59217966e-01 5.84199429e-01 3.00764497e-02 -7.05466986e-01 2.18856446e-02 5.33137977e-01 9.29901600e-01 -1.08374238e+00 2.57956833e-01 -3.83771777e-01 -5.04696310e-01 1.05987942e+00 4.52755034e-01 -2.26892028e-02 1.07014334e+00 -3.34242344e-01 -5.93413413e-01 -9.30644751e-01 -8.58614206e-01 2.07314659e-02 7.50067592e-01 5.19013464e-01 7.36187279e-01 6.78110346e-02 -7.09380209e-01 6.36352241e-01 4.99399215e-01 -3.25799105e-03 2.19972357e-01 3.91789764e-01 -4.95435983e-01 -1.33451056e+00 5.23445308e-02 7.27443814e-01 -5.55988431e-01 -4.90998000e-01 -8.02529812e-01 3.03369671e-01 2.71844029e-01 6.93164527e-01 -3.09236079e-01 -5.64527452e-01 2.32409388e-01 1.97354689e-01 1.90074995e-01 -1.01011515e+00 -5.16708493e-01 -3.26152325e-01 1.47004724e-01 -1.69096529e-01 -2.53982842e-01 -3.69342119e-02 -1.57815146e+00 -1.25888899e-01 -7.87992239e-01 4.68246996e-01 5.08504093e-01 6.68444753e-01 9.93920445e-01 7.28697360e-01 3.57807964e-01 5.31290732e-02 1.68273851e-01 -9.18718696e-01 -6.20486021e-01 4.05704409e-01 -2.52379209e-01 -6.34966075e-01 1.53006569e-01 4.94399369e-02]
[5.610896110534668, 5.9828314781188965]
986e161b-5319-4075-a972-96aa275f7e11
unical-a-single-branch-transformer-based
2304.09715
null
https://arxiv.org/abs/2304.09715v1
https://arxiv.org/pdf/2304.09715v1.pdf
UniCal: a Single-Branch Transformer-Based Model for Camera-to-LiDAR Calibration and Validation
We introduce a novel architecture, UniCal, for Camera-to-LiDAR (C2L) extrinsic calibration which leverages self-attention mechanisms through a Transformer-based backbone network to infer the 6-degree of freedom (DoF) relative transformation between the sensors. Unlike previous methods, UniCal performs an early fusion of the input camera and LiDAR data by aggregating camera image channels and LiDAR mappings into a multi-channel unified representation before extracting their features jointly with a single-branch architecture. This single-branch architecture makes UniCal lightweight, which is desirable in applications with restrained resources such as autonomous driving. Through experiments, we show that UniCal achieves state-of-the-art results compared to existing methods. We also show that through transfer learning, weights learned on the calibration task can be applied to a calibration validation task without re-training the backbone.
['Marius Bruehlmeier', 'Aaron Low', 'Mathieu Cocheteux']
2023-04-19
null
null
null
null
['camera-auto-calibration']
['computer-vision']
[ 2.24204630e-01 2.27769911e-01 -4.42042053e-01 -6.29169047e-01 -1.19490027e+00 -6.73103750e-01 5.44055641e-01 -3.98835003e-01 -5.20608723e-01 3.12608302e-01 4.91522439e-02 -5.12027919e-01 3.79870594e-01 -5.00641942e-01 -1.30900264e+00 -3.08704555e-01 3.51791024e-01 5.47457516e-01 1.77865267e-01 1.90301299e-01 8.96033421e-02 5.10705709e-01 -1.45402968e+00 6.32871687e-02 5.71964562e-01 1.08459938e+00 1.34179711e-01 7.98465908e-01 8.81094784e-02 6.85961902e-01 1.22640356e-01 -5.04685938e-01 5.36227882e-01 2.11626872e-01 -4.89015996e-01 -1.08135030e-01 1.14310944e+00 -6.57328606e-01 -4.56081063e-01 8.38481963e-01 3.48071158e-01 -1.04586996e-01 5.55520952e-01 -1.34777987e+00 -5.45112252e-01 2.49674842e-01 -6.16291761e-01 -1.43120095e-01 5.10542281e-02 2.99293876e-01 1.18699586e+00 -1.14864659e+00 4.49623644e-01 1.38953078e+00 1.14152217e+00 4.62674052e-01 -1.54524827e+00 -1.27206373e+00 2.56917834e-01 1.17084786e-01 -1.18161809e+00 -6.72662020e-01 6.78766549e-01 -4.84063774e-01 9.41579223e-01 -5.36878049e-01 3.57273489e-01 1.18681741e+00 6.35971203e-02 6.97588265e-01 7.68537939e-01 -7.08226562e-02 -4.28340435e-02 7.55262095e-03 -9.44301039e-02 6.45533085e-01 3.29261184e-01 2.40109429e-01 -6.66293144e-01 3.82828802e-01 8.65839958e-01 -1.70116410e-01 1.48144484e-01 -9.20697391e-01 -1.18365991e+00 9.64466572e-01 9.86796677e-01 -4.91131186e-01 5.10589369e-02 8.78930688e-01 2.25671127e-01 8.36941153e-02 1.11060582e-01 3.85597974e-01 -5.81537962e-01 -4.05076109e-02 -6.33428454e-01 -7.13272020e-02 3.49274993e-01 1.52939439e+00 1.30476224e+00 -2.19347281e-03 3.41101110e-01 3.09212863e-01 3.02318066e-01 1.07629859e+00 4.27511334e-02 -1.34198856e+00 6.88877940e-01 3.13008845e-01 2.67156437e-02 -3.39738160e-01 -3.10677409e-01 -4.19908196e-01 -5.19755483e-01 2.20724404e-01 5.99871725e-02 -3.20096254e-01 -9.09986377e-01 1.81204176e+00 1.88323349e-01 5.06932914e-01 3.99520695e-02 8.34200501e-01 5.14999449e-01 2.97342569e-01 9.16908234e-02 5.56588471e-01 1.11786652e+00 -1.18331707e+00 -8.00195187e-02 -7.25250065e-01 4.14840609e-01 -6.40011966e-01 9.46252167e-01 1.64050147e-01 -7.53287852e-01 -9.65608597e-01 -1.20311999e+00 -7.38147914e-01 -1.97236508e-01 1.53120637e-01 8.42738748e-01 3.28175604e-01 -1.12413621e+00 3.13131660e-01 -9.94532645e-01 -3.37929785e-01 5.37608147e-01 5.72066247e-01 -5.19853413e-01 -1.91113278e-01 -5.59866190e-01 9.50832367e-01 1.12435147e-01 -4.45585810e-02 -1.02435446e+00 -8.99143517e-01 -1.10818708e+00 -1.58620089e-01 3.31745028e-01 -8.73739600e-01 1.57465398e+00 -5.88430107e-01 -1.41428995e+00 6.72818065e-01 -3.63537103e-01 -5.62589288e-01 3.55881661e-01 -6.49961829e-01 -1.40785500e-01 2.89363638e-02 4.44510162e-01 1.50863600e+00 9.12130117e-01 -1.33142292e+00 -8.79032314e-01 -2.39779681e-01 1.25104249e-01 1.00035742e-01 -3.72612439e-02 -5.28514802e-01 -8.41026902e-01 -1.68104634e-01 2.00200677e-01 -1.12886691e+00 -1.68177724e-01 4.88184899e-01 -2.69353747e-01 1.52134210e-01 1.06017828e+00 -1.54086649e-01 4.24042434e-01 -2.26020551e+00 6.62450492e-02 3.62828211e-03 2.96561748e-01 -6.22145198e-02 -4.39567208e-01 8.57060309e-03 -7.79284164e-02 -1.90648213e-01 -1.55386791e-01 -8.35281312e-01 -1.23541695e-05 3.57449383e-01 -7.15178490e-01 4.18686360e-01 7.61383295e-01 1.19095910e+00 -9.24985945e-01 -4.33320582e-01 5.86182714e-01 7.18866169e-01 -6.87355161e-01 3.84304494e-01 -1.18850335e-01 7.15185106e-01 -2.86656350e-01 6.66447639e-01 8.57203543e-01 -2.27781415e-01 -1.06576411e-02 -6.01939201e-01 5.14817908e-02 3.18428129e-01 -7.24763215e-01 2.11887860e+00 -9.83218133e-01 7.96966314e-01 3.75524126e-02 -7.34946370e-01 9.41188157e-01 -1.15570659e-02 4.90118772e-01 -8.59430432e-01 1.35086387e-01 1.28071964e-01 -4.08969551e-01 -1.62132546e-01 3.96924168e-01 1.68327871e-03 -4.17182267e-01 2.87062734e-01 4.21906203e-01 -5.52925348e-01 -1.94943056e-01 2.25099176e-01 9.41188872e-01 7.43547201e-01 -1.33008454e-02 4.99359034e-02 2.91600198e-01 -1.69416249e-01 5.04798412e-01 5.28791189e-01 -5.91503568e-02 6.61473155e-01 2.03934133e-01 -4.25415039e-01 -1.25193846e+00 -1.42089033e+00 -2.28164405e-01 7.89500356e-01 1.05358884e-01 -3.69244397e-01 -4.80470836e-01 -4.76906270e-01 6.70280457e-01 6.86663389e-01 -5.81506014e-01 -2.06425086e-01 -5.38789928e-01 -8.78553241e-02 5.24415433e-01 1.05130875e+00 6.21215880e-01 -3.92840415e-01 -8.94006610e-01 2.73752481e-01 1.96310747e-02 -1.91158593e+00 -5.57402432e-01 6.29057646e-01 -8.89651239e-01 -1.07924712e+00 -7.85909444e-02 -4.59315002e-01 5.38732708e-01 5.54230630e-01 1.17392004e+00 -3.40746969e-01 -3.46079208e-02 4.27025974e-01 4.53660004e-02 -4.34294403e-01 1.06124356e-01 3.74128193e-01 -4.44496907e-02 -4.57745753e-02 6.62921667e-01 -8.67747307e-01 -3.68499488e-01 2.28646919e-01 -3.81432414e-01 1.63800836e-01 8.57981920e-01 7.44765818e-01 6.62274599e-01 -7.57492304e-01 2.38354251e-01 -7.94073999e-01 -9.79354978e-03 -2.13843733e-01 -1.16938436e+00 -2.36300051e-01 -6.58577383e-01 4.29170698e-01 4.88035589e-01 -3.04389477e-01 -7.12740898e-01 7.80455768e-01 1.75171718e-01 -9.90802526e-01 4.70199593e-04 1.13936365e-01 -2.66509950e-01 -3.84464622e-01 3.95856678e-01 -1.93567261e-01 -1.03925191e-01 -4.59439397e-01 8.67102563e-01 6.37665987e-01 1.33876729e+00 -8.57587695e-01 1.29180944e+00 5.41339099e-01 2.27174550e-01 -3.81684035e-01 -1.11463654e+00 -6.29125714e-01 -1.17336380e+00 -1.74690280e-02 1.09642017e+00 -1.70183241e+00 -9.24103558e-01 1.35374486e-01 -1.22577047e+00 -3.80445212e-01 -2.49318734e-01 6.25743389e-01 -7.12773025e-01 -8.28579254e-03 -2.70061821e-01 -6.19882882e-01 -8.40332881e-02 -1.40377235e+00 1.76041961e+00 3.56690548e-02 9.38115455e-03 -6.38599217e-01 6.62770728e-03 4.64663863e-01 3.70453477e-01 -2.24257330e-03 5.37685156e-01 -2.64648926e-02 -1.12441385e+00 -1.55448213e-01 -6.08169079e-01 3.29103887e-01 -2.04581782e-01 -6.99478090e-02 -1.55722463e+00 -3.64978284e-01 -5.55638731e-01 -7.25785077e-01 1.16322923e+00 1.12998679e-01 1.15864193e+00 2.30506703e-01 -4.31338251e-01 1.34453082e+00 1.39853621e+00 -1.99245363e-01 3.94572914e-01 1.52171642e-01 9.87946928e-01 6.54199868e-02 3.57656717e-01 -2.82473285e-02 9.22953606e-01 6.30741119e-01 8.14364731e-01 -1.87659130e-01 -2.69959450e-01 -7.38779306e-01 4.74121541e-01 7.10837722e-01 2.71722287e-01 2.50911087e-01 -7.25801349e-01 3.98801625e-01 -1.79326141e+00 -5.80105841e-01 1.73598498e-01 2.23954558e+00 4.70717162e-01 4.01054591e-01 -2.52919227e-01 -3.71128112e-01 3.87178034e-01 1.09236732e-01 -1.09218073e+00 -2.67797619e-01 -2.02263203e-02 4.32919890e-01 1.12587428e+00 6.99090898e-01 -1.21881735e+00 1.28429055e+00 6.96038771e+00 3.65591109e-01 -1.24678123e+00 2.72544503e-01 3.53427917e-01 -1.15623832e-01 -4.32968676e-01 3.65393430e-01 -1.06011248e+00 9.88833383e-02 1.04731178e+00 3.39390814e-01 4.20075506e-01 8.93078864e-01 -3.40566546e-01 5.56535609e-02 -1.45779407e+00 9.68404174e-01 -5.50779738e-02 -1.29025578e+00 -7.45898485e-02 6.28511161e-02 5.44849217e-01 8.40161026e-01 8.33344758e-02 4.48296279e-01 7.81119227e-01 -9.50993121e-01 9.28402185e-01 2.55905062e-01 1.45535886e+00 -6.09291673e-01 4.63355988e-01 1.34542704e-01 -1.46168756e+00 -2.56099105e-01 -5.12879670e-01 -1.40172662e-02 1.19125739e-01 4.59762484e-01 -8.17899287e-01 5.51428378e-01 7.01278925e-01 1.12749648e+00 -7.12868631e-01 6.22518122e-01 -3.78795832e-01 4.66716170e-01 -4.93245751e-01 5.05672276e-01 3.37481111e-01 -1.21220514e-01 1.44631058e-01 9.88270521e-01 3.65222037e-01 -3.21081191e-01 1.26731932e-01 1.02729023e+00 -4.30755734e-01 -5.02365768e-01 -7.60142803e-01 2.00994506e-01 7.54546940e-01 1.32011175e+00 -1.51477262e-01 -2.20029756e-01 -7.48511791e-01 7.37797439e-01 5.18150270e-01 3.39495450e-01 -9.48718250e-01 -1.81830287e-01 8.90131235e-01 -9.78021249e-02 8.51528645e-01 -4.16131407e-01 -3.83240879e-01 -1.21792006e+00 3.61302793e-02 -4.03413534e-01 -1.67480391e-02 -1.30237746e+00 -1.12835014e+00 4.54238892e-01 -1.04958117e-01 -1.36001623e+00 -1.90445453e-01 -7.46918440e-01 -4.40160096e-01 8.57109308e-01 -1.97871745e+00 -1.87536275e+00 -5.94981849e-01 6.19618952e-01 3.02853137e-01 -8.79729316e-02 7.62823462e-01 3.27067107e-01 -3.95301998e-01 7.64363468e-01 -2.67348170e-01 1.70829952e-01 1.08396149e+00 -1.20472538e+00 8.48904133e-01 8.84711146e-01 2.17995524e-01 3.59609127e-01 2.55778939e-01 -4.33659941e-01 -1.70458567e+00 -1.48202598e+00 6.69497907e-01 -8.32304835e-01 8.15184832e-01 -8.28839481e-01 -3.69183689e-01 1.19629180e+00 1.81087807e-01 4.69923705e-01 3.05728763e-01 2.86703467e-01 -1.13410711e+00 -6.16285443e-01 -6.90300226e-01 2.71472842e-01 1.11058021e+00 -9.85173702e-01 -2.95850098e-01 1.77116677e-01 1.00797606e+00 -6.04667783e-01 -7.21096933e-01 3.33400965e-01 6.87201858e-01 -7.76855826e-01 1.15176022e+00 -4.00700599e-01 3.80484134e-01 -4.44020748e-01 -4.86430675e-01 -9.90262628e-01 -3.79875213e-01 -6.11914575e-01 -2.47043744e-03 8.10802341e-01 4.88027126e-01 -5.52474082e-01 7.09049225e-01 3.55369687e-01 -4.99937832e-01 -3.80648404e-01 -1.11910784e+00 -5.06407976e-01 1.86717093e-01 -7.88915038e-01 6.99082911e-01 4.96055216e-01 -5.39909899e-01 9.92245197e-01 -3.90609860e-01 4.56847876e-01 8.22942138e-01 5.13037257e-02 1.32987475e+00 -1.39858794e+00 -1.86165601e-01 -1.89923808e-01 -5.61514974e-01 -1.47624195e+00 5.00827551e-01 -8.31101596e-01 2.36596912e-01 -1.09256661e+00 4.49359603e-02 -4.61083055e-01 -2.66455621e-01 6.91691875e-01 1.51558137e-02 5.00606537e-01 2.75122225e-01 1.41129345e-01 -5.72877645e-01 4.97582227e-01 1.00352669e+00 -2.97731549e-01 2.67992646e-01 -1.53139010e-01 -9.11038458e-01 4.17291790e-01 3.23739529e-01 -4.08886105e-01 -5.59476852e-01 -9.69187200e-01 9.81486365e-02 -1.44109279e-01 5.86946487e-01 -1.34594870e+00 4.86897051e-01 1.12067722e-01 4.80991215e-01 -8.31653714e-01 7.54478931e-01 -1.13544762e+00 9.60627978e-04 1.57183483e-01 -1.29623264e-01 2.48325914e-01 2.46272713e-01 7.06995428e-01 -1.46901980e-01 3.63377780e-01 7.94484317e-01 2.03194052e-01 -8.29864860e-01 4.43664014e-01 1.80105880e-01 -7.57695585e-02 7.72585273e-01 -2.66922802e-01 -2.34986544e-01 -5.13589919e-01 -2.03572363e-01 5.07424772e-01 5.38281441e-01 4.89352167e-01 3.21127534e-01 -1.41508269e+00 -3.97026449e-01 5.24617493e-01 5.09040117e-01 6.75098538e-01 -1.53879657e-01 7.87871361e-01 -3.07763696e-01 6.17334962e-01 -2.25925028e-01 -1.12002170e+00 -6.18234217e-01 6.68826103e-01 2.35715970e-01 1.16428815e-01 -6.93323195e-01 9.08426404e-01 2.60230362e-01 -8.60854864e-01 2.36342624e-01 -5.47712207e-01 3.39328796e-01 -2.13893339e-01 8.87386426e-02 -1.63079962e-01 1.26600549e-01 -7.70445585e-01 -4.15997416e-01 1.16938925e+00 6.90395162e-02 -1.17624961e-01 1.26513445e+00 -1.70127019e-01 2.97842324e-01 4.09207553e-01 1.38428521e+00 -1.54860899e-01 -1.99622762e+00 -4.33761209e-01 -2.48002693e-01 -3.15127254e-01 3.08891416e-01 -5.43598533e-01 -1.36020839e+00 1.36711240e+00 4.51053500e-01 -7.70090342e-01 8.73691916e-01 -9.93131921e-02 6.95417225e-01 8.31348181e-01 3.98828119e-01 -8.94846380e-01 -4.11812104e-02 6.03202522e-01 4.74821121e-01 -1.48450947e+00 -4.52343114e-02 -3.47781241e-01 -6.41778409e-01 1.12140203e+00 7.95498908e-01 -3.18242311e-01 7.01097310e-01 6.99645400e-01 3.10249358e-01 9.00832117e-02 -9.05944288e-01 -2.48704478e-01 1.60378173e-01 8.31639826e-01 1.77227631e-01 -5.16743809e-02 7.35650837e-01 2.40446731e-01 -2.82877088e-01 9.92155001e-02 3.16602379e-01 6.83391809e-01 -2.56058902e-01 -1.11132288e+00 -1.23092487e-01 1.80276275e-01 3.70022982e-01 -1.33813113e-01 -2.26685733e-01 9.78227735e-01 2.20895782e-01 6.74927294e-01 4.76582259e-01 -7.76245832e-01 4.08816427e-01 6.70405105e-03 4.19646829e-01 -4.27066207e-01 -2.28956342e-01 1.93563536e-01 -9.16955546e-02 -1.01243961e+00 -4.60083395e-01 -6.79845393e-01 -1.09989381e+00 -3.20390940e-01 -2.56289244e-01 -1.68620586e-01 8.26580644e-01 7.57939339e-01 7.19488442e-01 4.56256092e-01 6.03917658e-01 -1.16083717e+00 -6.45905733e-01 -7.55208910e-01 -3.44669819e-02 1.71016067e-01 7.26979554e-01 -9.02085125e-01 -9.41965133e-02 1.77312270e-01]
[7.918496131896973, -2.5315070152282715]
a8b6dda8-7e74-4f73-935e-54eaea46ced6
effects-of-mindfulness-on-perceived-stress
1708.08006
null
http://arxiv.org/abs/1708.08006v1
http://arxiv.org/pdf/1708.08006v1.pdf
Effects of mindfulness on perceived stress levels and heart rate variability
Mindfulness has become increasingly popular as a method for building resilience against stress in both clinical and healthy populations. This study sought to investigate the effects of mindfulness training on perceived levels of stress and heart rate variability in students.
[]
2017-08-26
null
null
null
null
['heart-rate-variability']
['medical']
[-9.14276913e-02 -2.80982666e-02 -8.49099338e-01 -1.92008555e-01 1.21297464e-01 1.31612360e-01 -3.37191224e-01 9.94758427e-01 -6.16462469e-01 3.11778545e-01 1.18646622e-01 -3.45625222e-01 2.66761005e-01 -3.01579714e-01 -8.14213976e-02 -4.24982309e-01 -2.08288245e-02 -4.83079404e-01 -3.66182148e-01 -4.14344043e-01 1.14206326e+00 4.88623440e-01 -9.82341766e-01 -3.93523246e-01 1.05008531e+00 1.11483105e-01 -2.06940114e-01 4.70777303e-01 2.89284457e-02 9.55423176e-01 -5.58780551e-01 -1.84481040e-01 -6.59679830e-01 -1.05151391e+00 -7.66758323e-01 -3.07841022e-02 4.32193547e-01 -1.50528610e-01 -2.49129534e-02 9.80884731e-01 9.33557928e-01 1.00746942e+00 -5.38622499e-01 -8.88004482e-01 -4.98498082e-01 4.33839291e-01 -3.36450547e-01 1.16813195e+00 5.68771183e-01 1.20699547e-01 -2.56093502e-01 -4.18444246e-01 -9.15779620e-02 8.41441751e-01 6.70332432e-01 7.59216785e-01 -1.45797598e+00 -1.04799831e+00 -1.77835181e-01 7.36619949e-01 -1.08185077e+00 -7.82359600e-01 6.74696147e-01 -8.86003152e-02 8.54691625e-01 1.79853871e-01 1.39362407e+00 5.63067019e-01 8.57135773e-01 -1.78940669e-01 1.46418428e+00 -4.77504224e-01 4.74480212e-01 6.12583578e-01 8.83893609e-01 2.05040619e-01 8.20519567e-01 3.00835259e-03 -7.40971267e-01 -2.14244314e-02 7.19317973e-01 -1.03590243e-01 -5.27414560e-01 3.82111967e-02 -5.50420403e-01 9.34443951e-01 5.74433288e-05 4.45652336e-01 -6.58631444e-01 4.32688445e-02 6.45635009e-01 1.16090141e-01 4.01142448e-01 4.53079194e-01 1.60807282e-01 -8.43522429e-01 -6.61100388e-01 -3.57111216e-01 3.99735570e-01 -3.86938393e-01 5.90412617e-02 9.86062109e-01 1.54188544e-01 9.27197993e-01 7.98793286e-02 4.03092086e-01 9.07763958e-01 -1.16316295e+00 -5.60884833e-01 3.21135432e-01 -1.14831708e-01 -1.26713610e+00 -4.79720235e-01 -3.89184713e-01 -3.59417945e-01 3.97639513e-01 -1.65528432e-01 1.38718009e-01 -3.41284245e-01 1.67540050e+00 5.74948668e-01 1.12115562e-01 -6.97672963e-02 8.50398839e-01 6.53317511e-01 2.19984531e-01 7.03719258e-01 -8.82361054e-01 1.36303616e+00 -6.63361847e-01 -1.49594414e+00 -6.09746516e-01 3.47735405e-01 -5.19294322e-01 1.06980169e+00 3.10811341e-01 -1.41755831e+00 -8.59486103e-01 -8.17584157e-01 -3.43506448e-02 1.45943537e-01 -7.55206883e-01 5.68966508e-01 1.59101880e+00 -8.17555130e-01 7.92137086e-01 -1.20859969e+00 -4.56194222e-01 3.47407103e-01 -3.45320404e-01 -3.38616014e-01 1.17773458e-01 -1.21051967e+00 1.45613790e+00 3.28775018e-01 2.34841928e-01 -2.39320219e-01 -1.12918949e+00 -1.04684114e+00 2.27398783e-01 1.18161544e-01 -5.40406823e-01 1.15957308e+00 -6.41047835e-01 -2.02303410e+00 7.52798975e-01 -5.57423569e-02 -1.67098895e-01 -5.07086575e-01 -4.69965905e-01 -6.26376152e-01 5.92231333e-01 5.18792234e-02 -2.48566508e-01 3.27594280e-01 -3.16658109e-01 3.37947071e-01 -6.83452010e-01 -6.24086201e-01 3.43845785e-01 -2.47747049e-01 5.38364388e-02 9.65325892e-01 -1.71948597e-01 5.36998928e-01 -7.34100282e-01 -3.03084642e-01 -5.45544386e-01 1.69771880e-01 -6.26726449e-02 6.09122030e-02 -5.82205415e-01 1.00907314e+00 -1.90688932e+00 -8.15195501e-01 8.30646381e-02 1.07802384e-01 3.31501365e-01 2.93350399e-01 4.87294078e-01 -3.11538994e-01 7.44791031e-01 5.35267293e-01 5.65430939e-01 -3.07600886e-01 2.33404636e-01 1.45704180e-01 9.23841119e-01 -1.27618000e-01 2.94040769e-01 -1.01629519e+00 -3.17407697e-01 4.68521446e-01 5.69952190e-01 -4.83577073e-01 2.53038228e-01 1.13244855e+00 4.72528636e-01 -9.74969268e-02 3.34803492e-01 7.24001169e-01 1.22736312e-01 4.20754224e-01 4.77952987e-01 -6.01333261e-01 6.64154947e-01 -4.56993610e-01 1.33597553e+00 -7.94686452e-02 7.37608016e-01 -1.41127363e-01 -7.43380606e-01 9.90260899e-01 7.16098964e-01 2.87962019e-01 -1.13196969e+00 3.18101496e-01 -4.24604028e-01 4.23162341e-01 -8.66064131e-01 3.74734998e-01 -1.48378932e+00 2.91083992e-01 4.59173650e-01 -8.37643594e-02 -2.61754960e-01 4.28355448e-02 1.72585919e-01 4.47784185e-01 -3.92174423e-01 7.74498880e-01 -7.81218171e-01 1.97834060e-01 -7.35671595e-02 6.04879975e-01 4.68126118e-01 -9.66393590e-01 -5.06445706e-01 5.08158147e-01 -4.01863635e-01 2.08894610e-01 -6.80102110e-01 -2.59674460e-01 9.20026720e-01 -3.91288698e-02 1.74051121e-01 -3.27786684e-01 3.79897565e-01 -2.33593404e-01 1.58991551e+00 -6.33734882e-01 -1.02301717e+00 -2.67140687e-01 -4.78399307e-01 9.88812447e-02 3.62930417e-01 3.76708478e-01 -1.07326078e+00 -1.37373197e+00 1.78672984e-01 -3.25385064e-01 -4.77590770e-01 -4.34325844e-01 1.39820114e-01 -1.65213907e+00 -8.65139842e-01 5.65360524e-02 -4.85219568e-01 3.86547476e-01 8.91402960e-01 1.12865591e+00 3.48455936e-01 -3.92812282e-01 5.28107166e-01 -3.94650409e-03 -7.71830738e-01 -1.26498029e-01 -3.47418398e-01 1.50614738e-01 -6.47892058e-01 3.25602144e-01 -8.81061614e-01 -1.01304960e+00 -1.22332685e-01 -4.88537014e-01 -4.65134233e-01 2.12679163e-01 1.20703839e-01 4.29424465e-01 -3.56906146e-01 7.85090506e-01 -8.38173926e-01 7.98053265e-01 -5.83181024e-01 2.91118860e-01 -5.81580997e-01 -1.06077921e+00 -8.86657357e-01 1.69933848e-02 -9.12146509e-01 -9.83371317e-01 -9.44492996e-01 -1.06102861e-01 2.77828455e-01 -2.71499932e-01 6.10055983e-01 5.57575047e-01 -3.48652065e-01 9.78142142e-01 5.45364758e-03 4.44957733e-01 -7.08015114e-02 -2.68863618e-01 -3.15841109e-01 3.98290515e-01 -1.89001516e-01 1.14258312e-01 -1.50134191e-01 2.17090189e-01 -1.39462280e+00 -1.18895650e+00 -3.68625820e-01 -2.11014494e-01 -5.51601231e-01 6.51288986e-01 -9.96413827e-01 -1.27414024e+00 -2.94173807e-01 -1.44936386e-02 -4.98196304e-01 -4.08266157e-01 1.27624011e+00 -9.46841016e-02 3.04576010e-01 -6.21408761e-01 -9.00958180e-01 -7.43553638e-01 -2.61788875e-01 -4.28876698e-01 1.01851285e+00 -4.83249664e-01 -1.34886014e+00 1.04888964e+00 4.99703109e-01 8.89392495e-01 8.39043558e-01 6.20976627e-01 -2.68664092e-01 5.35590172e-01 -6.08680069e-01 7.12134957e-01 1.07984833e-01 2.50792056e-01 -3.53062421e-01 -5.90578377e-01 -2.26336226e-01 1.02414346e+00 -4.24348176e-01 1.18256681e-01 7.81845748e-01 4.14556950e-01 -2.72889137e-01 2.40594804e-01 1.11233667e-01 1.39192045e+00 4.76101309e-01 8.64536464e-01 4.89550292e-01 3.84131402e-01 4.74757165e-01 2.93173969e-01 1.33305833e-01 2.30772913e-01 -1.94996506e-01 -1.56305641e-01 2.15793066e-02 2.22033978e-01 5.03619730e-01 3.06620032e-01 9.06249225e-01 -1.07870862e-01 5.06877720e-01 -1.04231858e+00 3.34044933e-01 -4.25669760e-01 -1.53377485e+00 -4.06178713e-01 2.25365806e+00 1.06942320e+00 -9.45628658e-02 3.85080427e-01 4.39026244e-02 6.60904944e-01 3.29435542e-02 -5.50532758e-01 -1.40545750e+00 4.74452674e-01 9.46389914e-01 6.76759407e-02 3.27025205e-01 -1.22770809e-01 2.47063726e-01 7.91091251e+00 -4.63121593e-01 -1.45184112e+00 1.10462695e-01 6.39691830e-01 -3.73106182e-01 -7.93038160e-02 1.00549616e-01 -2.32936606e-01 2.36018542e-02 1.95959127e+00 -6.13625348e-01 -1.09217562e-01 7.79461980e-01 8.40206385e-01 -7.47965932e-01 -3.50329489e-01 3.10109884e-01 8.32190961e-02 -7.25901663e-01 -8.00334156e-01 2.76264437e-02 2.19085336e-01 -9.13910925e-01 4.95417044e-03 8.30568314e-01 -4.19880748e-01 -1.02448285e+00 2.05942273e-01 4.79515374e-01 2.38507837e-01 -9.04251337e-01 6.01532340e-01 5.08964844e-02 -2.05147013e-01 2.97549844e-01 -1.16448984e-01 -7.55783141e-01 -5.20589352e-02 6.08782232e-01 -4.70041484e-01 -5.44099331e-01 1.57762885e-01 9.88709852e-02 -2.78990686e-01 1.02778137e+00 -2.46930510e-01 1.11268067e+00 2.60597199e-01 2.44591087e-02 -2.27646977e-01 -3.71518731e-02 1.30762905e-01 5.34295142e-01 -1.02617681e-01 5.87873459e-01 1.51323169e-01 7.05765009e-01 5.09263456e-01 4.05820310e-01 -2.15875894e-01 -1.78504616e-01 2.44687274e-01 1.23369122e+00 -1.16808391e+00 -4.78824288e-01 -1.11413479e-01 3.72676142e-02 -1.55671120e-01 9.03207511e-02 -5.48415124e-01 -3.59806389e-01 7.98547983e-01 3.09168428e-01 -2.18238503e-01 -2.19067082e-01 -6.87419116e-01 -5.63136995e-01 -7.57829905e-01 -7.90108502e-01 4.68900383e-01 -3.42067897e-01 -4.72859204e-01 -2.78510094e-01 -1.40041057e-02 -4.86948878e-01 3.89836013e-01 4.12169278e-01 -1.12582135e+00 1.26378977e+00 -1.65446091e+00 -3.96711938e-02 -4.11735743e-01 2.57282704e-01 3.75562906e-01 1.10311055e+00 9.97192860e-01 -2.34918714e-01 -1.33305109e+00 4.62276131e-01 -6.43293679e-01 -4.66138393e-01 8.75591218e-01 -1.24323785e+00 -5.26675105e-01 8.73829663e-01 -8.19449544e-01 1.30075252e+00 9.48526442e-01 -6.33921146e-01 -1.41635144e+00 -4.44387466e-01 6.70199335e-01 3.75893712e-01 4.62850705e-02 4.99733686e-01 -1.35739005e+00 2.94529676e-01 6.53782904e-01 -4.52303439e-01 1.74757266e+00 4.45146561e-01 2.68595487e-01 -6.15970455e-02 -1.26190305e+00 5.02046525e-01 -2.80006770e-02 -2.97434330e-01 -5.32362223e-01 8.00970867e-02 5.33831060e-01 -4.16986704e-01 -1.57749379e+00 -1.17164791e-01 4.66253728e-01 -9.22324002e-01 8.34439754e-01 -3.48980486e-01 3.15951109e-01 4.82419223e-01 9.19218838e-01 -1.28365374e+00 -5.23980737e-01 -9.35621321e-01 8.99240673e-02 1.02465916e+00 -4.39502388e-01 -1.01108634e+00 6.38423741e-01 1.40261209e+00 -4.81916606e-01 -1.03957973e-01 -5.61412036e-01 -1.79666415e-01 3.50695580e-01 3.89390290e-01 -1.35097370e-01 1.22525990e+00 1.37578475e+00 5.23107171e-01 9.74051841e-03 -1.70325294e-01 2.46571586e-01 -6.67334627e-03 3.72129858e-01 -9.51455891e-01 1.32335946e-01 -2.09057122e-01 -2.42402285e-01 1.62807941e-01 -1.70971304e-01 -3.48937333e-01 -7.51077905e-02 -1.42974591e+00 -4.39854190e-02 3.83205920e-01 -4.80059326e-01 6.05374277e-01 -7.57803202e-01 -3.52134649e-03 -1.13913622e-02 -5.77896535e-01 -3.07858232e-02 6.36118770e-01 1.44967520e+00 8.97009313e-01 -1.08838558e+00 -2.32109636e-01 -1.59845579e+00 7.93820545e-02 1.71009946e+00 -7.31473982e-01 -6.83390439e-01 4.64475960e-01 8.19526687e-02 9.58302677e-01 -1.65200889e-01 -9.57617581e-01 -4.02682930e-01 -9.43201959e-01 4.28617060e-01 -9.03497823e-03 3.84320374e-05 -1.82259992e-01 1.14871845e-01 1.05561602e+00 -4.11070406e-01 4.99465853e-01 8.91968191e-01 -1.33665651e-01 2.06756026e-01 -4.25409973e-01 1.16893435e+00 -2.21260592e-01 -2.73152322e-01 -6.67423129e-01 -1.23476171e+00 4.25127894e-01 9.50886548e-01 -3.15159410e-01 -5.63568771e-01 -4.74707484e-01 -8.33714008e-01 2.55565755e-02 3.54827046e-01 -6.61675408e-02 7.26317227e-01 -1.07315481e+00 -1.86143488e-01 3.38305603e-03 -4.43312615e-01 -5.71643054e-01 1.36239851e+00 1.89131045e+00 -6.48149848e-01 2.73032665e-01 -7.56955445e-01 -2.38463357e-02 -1.09675717e+00 6.15722597e-01 4.01993752e-01 5.97596169e-02 -8.14300179e-01 6.92085862e-01 -6.78167105e-01 7.38381088e-01 -2.16049626e-01 1.23557359e-01 -4.60509419e-01 -2.31198847e-01 1.12146986e+00 8.97501409e-01 -3.34180295e-01 -1.96902230e-01 -2.72091687e-01 7.26762339e-02 1.89210430e-01 5.73300608e-02 9.33658004e-01 -3.70722324e-01 -4.80145216e-01 7.94844508e-01 5.11592031e-01 1.83632210e-01 -6.43873274e-01 4.45100963e-01 -2.06271380e-01 -6.51600063e-01 4.11513120e-01 -5.67293823e-01 -5.54906785e-01 1.09124243e+00 1.25286663e+00 2.60801107e-01 1.15591168e+00 -6.73714995e-01 5.45033097e-01 -1.64632097e-01 -4.65909511e-01 -1.00042117e+00 6.13276064e-01 -2.39627466e-01 5.20042360e-01 -7.42233872e-01 1.27438173e-01 -1.45631775e-01 -8.03874493e-01 1.09183204e+00 5.80949426e-01 -4.21195745e-01 6.90793157e-01 -1.33220807e-01 3.59852254e-01 -4.38688070e-01 -7.91767955e-01 -1.57874115e-02 -8.18394274e-02 7.26300955e-01 1.00926542e+00 1.14239551e-01 -9.10132706e-01 -8.43993351e-02 -2.94666320e-01 3.38442385e-01 1.38592649e+00 1.04839671e+00 -9.41329181e-01 -6.48775995e-01 -7.18001962e-01 3.00135314e-01 -8.29236329e-01 1.26596689e-02 4.36696969e-02 7.82962143e-01 -1.03913441e-01 1.31922114e+00 -5.61180338e-03 1.61594659e-01 6.36344075e-01 7.34482646e-01 3.14357817e-01 -1.00972843e+00 -9.81393158e-01 2.01595724e-01 -2.72177346e-03 -4.10862178e-01 -4.75189745e-01 -9.14461732e-01 -1.17331815e+00 -8.58875215e-01 -4.48337644e-02 4.76197630e-01 7.41467834e-01 4.04753566e-01 2.43683904e-01 7.11774766e-01 4.45306033e-01 -2.04054981e-01 -3.84559125e-01 -1.16757512e+00 -7.99167633e-01 1.54910639e-01 1.75872892e-01 -2.84156889e-01 -9.95288119e-02 -2.34457329e-01]
[13.744756698608398, 3.0804243087768555]
538a69af-3d70-4be2-a28d-8b5db5be3669
self-supervised-robust-scene-flow-estimation
2203.12193
null
https://arxiv.org/abs/2203.12193v1
https://arxiv.org/pdf/2203.12193v1.pdf
Self-Supervised Robust Scene Flow Estimation via the Alignment of Probability Density Functions
In this paper, we present a new self-supervised scene flow estimation approach for a pair of consecutive point clouds. The key idea of our approach is to represent discrete point clouds as continuous probability density functions using Gaussian mixture models. Scene flow estimation is therefore converted into the problem of recovering motion from the alignment of probability density functions, which we achieve using a closed-form expression of the classic Cauchy-Schwarz divergence. Unlike existing nearest-neighbor-based approaches that use hard pairwise correspondences, our proposed approach establishes soft and implicit point correspondences between point clouds and generates more robust and accurate scene flow in the presence of missing correspondences and outliers. Comprehensive experiments show that our method makes noticeable gains over the Chamfer Distance and the Earth Mover's Distance in real-world environments and achieves state-of-the-art performance among self-supervised learning methods on FlyingThings3D and KITTI, even outperforming some supervised methods with ground truth annotations.
['Anand Rangarajan', 'Sanjay Ranka', 'Patrick Emami', 'Pan He']
2022-03-23
null
null
null
null
['scene-flow-estimation']
['computer-vision']
[-2.29838893e-01 -2.80567676e-01 -1.58654481e-01 -3.24968308e-01 -6.46266282e-01 -6.95669591e-01 7.72046566e-01 1.26863018e-01 -3.57455760e-01 4.62709159e-01 -8.31264555e-02 -1.39619857e-02 -2.85672545e-01 -7.73828328e-01 -7.30040014e-01 -4.78188068e-01 -3.38245094e-01 9.16004777e-01 6.18697405e-01 -1.24919862e-01 4.23735529e-01 6.48964584e-01 -1.73524070e+00 -2.14405298e-01 1.06159234e+00 6.79674149e-01 6.35758266e-02 8.74674618e-01 -2.57678062e-01 9.21000063e-01 -3.69767755e-01 -2.36886784e-01 4.78170514e-01 -2.74557471e-01 -8.90957355e-01 7.84612522e-02 9.35148537e-01 -2.12746069e-01 -4.53449696e-01 8.94448340e-01 6.80067316e-02 2.39304632e-01 9.42382157e-01 -1.70524526e+00 -2.31668219e-01 -1.48756653e-01 -8.29734623e-01 1.96347207e-01 6.50080264e-01 2.22670357e-03 8.20107162e-01 -6.65344834e-01 4.96673137e-01 1.14644063e+00 8.34110677e-01 -2.33967509e-02 -1.32538962e+00 -4.84972924e-01 3.22486870e-02 2.09816441e-01 -1.56369317e+00 -2.97434419e-01 7.49016881e-01 -8.66233945e-01 7.56072760e-01 5.84289283e-02 7.58795619e-01 3.85517269e-01 -8.23624432e-02 7.55025983e-01 6.14816546e-01 -2.71697700e-01 4.20556426e-01 -9.66473892e-02 -2.26585343e-01 7.62556851e-01 3.09468329e-01 1.07037216e-01 -6.17016375e-01 -2.98562020e-01 1.01500678e+00 1.39677197e-01 -2.39542544e-01 -1.14086366e+00 -1.70418894e+00 6.38475358e-01 5.96673906e-01 4.91056107e-02 -1.90224126e-01 1.76078036e-01 -3.37037407e-02 -7.18448088e-02 5.66900790e-01 3.70310545e-02 -1.29549831e-01 -3.08039516e-01 -1.28761256e+00 2.70742029e-01 1.00663960e+00 1.32204390e+00 1.11942184e+00 -1.11989386e-01 4.07263905e-01 1.50959849e-01 6.90272450e-01 7.51176834e-01 2.09386736e-01 -1.38371682e+00 3.51479590e-01 5.09314716e-01 3.93273324e-01 -1.26585543e+00 -1.21894956e-01 -1.16017265e-02 -7.78986096e-01 3.22567552e-01 7.36878395e-01 7.98095912e-02 -5.13714910e-01 1.18886173e+00 6.59434736e-01 9.92309570e-01 9.61090177e-02 9.34373438e-01 6.88966870e-01 6.58049405e-01 -3.31642687e-01 -4.90665575e-03 6.46173775e-01 -8.91320050e-01 -4.34434980e-01 -1.49197385e-01 5.86545348e-01 -7.36987710e-01 6.83670461e-01 1.38547421e-01 -9.67813849e-01 -7.09763706e-01 -1.04366863e+00 -6.02367297e-02 -5.99358492e-02 -1.53152704e-01 6.82878613e-01 3.96357745e-01 -1.04862666e+00 8.33708227e-01 -1.14753890e+00 -3.80071700e-01 3.39991033e-01 1.95158973e-01 -5.09002626e-01 3.24851424e-02 -3.74699533e-01 7.28715599e-01 1.84921905e-01 -6.22618422e-02 -6.13842130e-01 -1.04263711e+00 -1.12130487e+00 -4.04480815e-01 -1.57838672e-01 -9.83850360e-01 9.96205926e-01 -3.93339783e-01 -1.57434118e+00 1.07066464e+00 -4.41351205e-01 -4.32311863e-01 7.65269637e-01 -5.21224141e-01 7.53031578e-03 2.60089815e-01 3.38130802e-01 8.52384388e-01 7.64559686e-01 -1.37721431e+00 -8.36729288e-01 -2.59858489e-01 -2.06288606e-01 3.16082448e-01 1.76919863e-01 -4.20830041e-01 -2.42537677e-01 -2.47145444e-01 5.44065118e-01 -1.08439529e+00 -2.87785262e-01 3.84764373e-01 -2.36280873e-01 -1.88837662e-01 9.68829572e-01 -2.63949722e-01 6.43198133e-01 -1.90722454e+00 3.39669794e-01 1.40387654e-01 1.51196331e-01 -2.41003232e-03 2.10208833e-01 3.09186161e-01 1.03223249e-01 -2.56876469e-01 -5.90821981e-01 -4.93105024e-01 -4.52895127e-02 2.47733787e-01 -3.81117374e-01 1.16801202e+00 -1.33869033e-02 5.73445797e-01 -1.38067949e+00 -7.35102057e-01 9.09676373e-01 5.23304522e-01 -5.11775732e-01 1.69615403e-01 3.02876774e-02 7.16947317e-01 -1.59029737e-01 3.26602966e-01 1.02593517e+00 -1.90789267e-01 -3.61972123e-01 -1.08675770e-01 -2.61555701e-01 7.23828152e-02 -1.57649016e+00 2.33270574e+00 -2.19617039e-01 7.43253887e-01 -1.89482763e-01 -7.78167248e-01 9.67215419e-01 -4.22480069e-02 9.39432859e-01 4.01330404e-02 -3.62818763e-02 7.30360523e-02 -4.20768499e-01 -2.36674160e-01 5.80586553e-01 -1.04961358e-01 1.32847399e-01 8.38693976e-02 1.87318578e-01 -8.45540285e-01 4.96393405e-02 3.35743219e-01 9.27641571e-01 6.98909760e-01 3.29695970e-01 -2.98436075e-01 7.40143299e-01 4.49041963e-01 5.23521006e-01 4.28007364e-01 -3.44832122e-01 9.41814542e-01 -3.77063155e-02 -4.90172148e-01 -8.74408007e-01 -1.36727166e+00 -1.74460322e-01 2.79330432e-01 7.16625333e-01 -5.62359452e-01 -4.08098876e-01 -5.92071295e-01 3.20242494e-01 5.95229208e-01 -2.20970467e-01 9.73766372e-02 -4.91257042e-01 -4.51176196e-01 3.34526330e-01 4.23501939e-01 6.89635456e-01 -4.73493278e-01 -7.27006495e-01 -2.66234614e-02 -4.92048889e-01 -1.29855192e+00 -4.04239088e-01 -3.28257978e-01 -1.13865674e+00 -1.19116366e+00 -7.34135032e-01 -6.83597982e-01 6.25418842e-01 6.94722533e-01 1.08207190e+00 -2.24050209e-01 -2.98566997e-01 6.61841989e-01 -1.86597973e-01 -1.98589623e-01 -2.02802524e-01 -2.86499977e-01 2.26736933e-01 2.95936894e-02 5.04118800e-01 -7.13087380e-01 -6.14655316e-01 4.46202666e-01 -4.90965664e-01 -1.85045987e-01 9.61749777e-02 3.66053730e-01 8.57622266e-01 4.36092950e-02 -8.95438045e-02 -2.55130887e-01 -6.73828050e-02 -3.58493537e-01 -8.53224039e-01 -1.06443856e-02 -1.94639519e-01 6.79458156e-02 2.26639539e-01 -1.35941431e-01 -9.47280169e-01 5.79070389e-01 3.14791858e-01 -8.59988213e-01 -5.41528404e-01 -1.06736407e-01 1.00154109e-01 -5.00333786e-01 7.21746802e-01 8.33654627e-02 1.09321773e-01 -2.02301428e-01 7.39830911e-01 3.56917739e-01 8.93457472e-01 -3.86353672e-01 1.35857689e+00 1.28156841e+00 3.17967534e-01 -1.07744718e+00 -7.31483996e-01 -1.28031635e+00 -1.61739326e+00 -3.79230797e-01 9.99388456e-01 -1.08130944e+00 -8.45462561e-01 4.43732202e-01 -1.15874386e+00 -9.85992700e-02 -4.21555758e-01 7.70319045e-01 -1.01757467e+00 7.81830311e-01 -2.61840224e-01 -7.66780376e-01 1.47978157e-01 -8.91573846e-01 1.42204881e+00 1.17574982e-01 -2.55760521e-01 -1.41478658e+00 6.87122822e-01 3.07837605e-01 -1.55463964e-01 4.95088309e-01 1.95318256e-02 -1.27306178e-01 -8.09450984e-01 -2.69140720e-01 1.43960463e-02 1.50645018e-01 3.70152861e-01 2.80023247e-01 -9.60247338e-01 -2.99290448e-01 9.72474366e-02 8.87642875e-02 6.79475605e-01 5.24193466e-01 6.81354284e-01 -3.26186675e-03 -5.41256309e-01 1.03273690e+00 1.52783811e+00 -1.54323831e-01 6.08829618e-01 4.01541650e-01 1.00007236e+00 6.09563708e-01 8.48178804e-01 4.39326108e-01 7.79180229e-01 4.46428388e-01 5.38867891e-01 2.51578093e-02 1.33090522e-02 -5.33450067e-01 1.83010101e-01 7.08127499e-01 -1.88172251e-01 5.07478565e-02 -1.20533526e+00 6.82387054e-01 -2.07436872e+00 -1.17333364e+00 -6.90013409e-01 2.56525397e+00 2.99085975e-01 -7.34778419e-02 1.75735891e-01 9.84334722e-02 5.46977282e-01 7.28076845e-02 -3.61345053e-01 3.50103796e-01 -9.59409922e-02 -2.17769116e-01 7.75636852e-01 7.47894049e-01 -1.35770774e+00 1.03828347e+00 6.53477144e+00 2.67765313e-01 -7.88157284e-01 -8.56930614e-02 1.53998844e-02 -1.84265077e-02 1.06195107e-01 3.11552972e-01 -7.28479326e-01 4.08087939e-01 5.88190734e-01 -2.80974597e-01 5.23064099e-02 7.85668194e-01 5.35698235e-02 -3.78548592e-01 -1.30171227e+00 1.33856666e+00 2.15488538e-01 -1.44145477e+00 -2.84896791e-01 9.29021761e-02 8.79995227e-01 3.77019554e-01 -2.31180355e-01 -2.96735823e-01 6.74740136e-01 -6.03129208e-01 7.78804600e-01 5.69425106e-01 4.10086542e-01 -6.01025939e-01 4.46632266e-01 6.45737469e-01 -1.37735498e+00 3.62438262e-01 -5.38890004e-01 -2.18137458e-01 4.83959228e-01 5.84213793e-01 -8.12539995e-01 8.00691128e-01 9.93747413e-01 1.34167492e+00 -4.70104605e-01 1.63447547e+00 -2.94806939e-02 2.75490791e-01 -7.08943784e-01 3.80979210e-01 1.13193825e-01 -6.77820444e-01 8.52651536e-01 1.10730064e+00 3.13264638e-01 -7.54507408e-02 3.23414743e-01 9.34397936e-01 3.32857758e-01 -1.20421022e-01 -9.43056583e-01 5.70117354e-01 4.44690287e-01 1.17198789e+00 -9.28570569e-01 -3.51441383e-01 -3.27213854e-01 1.05150235e+00 1.32594407e-01 2.65243858e-01 -7.04130173e-01 -2.20120609e-01 7.89091408e-01 1.89667225e-01 1.57671839e-01 -6.47419393e-01 -1.35163516e-01 -1.39200330e+00 4.60406989e-02 -2.92103831e-02 3.07694167e-01 -7.64019728e-01 -1.38392198e+00 2.82669485e-01 4.15457517e-01 -1.97239423e+00 -2.37196296e-01 -3.88072938e-01 -8.23424160e-01 6.65995598e-01 -1.62299740e+00 -1.07163751e+00 -7.42591560e-01 6.61058247e-01 3.04224312e-01 1.01007193e-01 4.55769330e-01 1.21620037e-01 -7.23698884e-02 -1.21833019e-01 2.53803372e-01 2.83990372e-02 7.71696270e-01 -1.47553205e+00 5.13319075e-01 1.01456499e+00 5.56523442e-01 2.94420689e-01 7.77985275e-01 -7.08513677e-01 -1.22574854e+00 -1.01915610e+00 6.12429023e-01 -8.48783135e-01 7.62886584e-01 -1.58116549e-01 -8.54367256e-01 8.75605404e-01 7.07560824e-03 3.59272867e-01 5.24894297e-01 -1.98779643e-01 -9.26293060e-02 -6.77872226e-02 -1.17426562e+00 1.39475018e-01 1.27513063e+00 -3.70688945e-01 -8.09330642e-01 6.32289052e-01 4.33189422e-01 -6.27400517e-01 -8.32423329e-01 4.40395266e-01 4.40437943e-02 -1.15825534e+00 1.35401249e+00 -2.25931376e-01 5.29129431e-02 -8.23428035e-01 -1.36006698e-01 -1.34319913e+00 -3.24367613e-01 -8.67271364e-01 -1.74832746e-01 1.01250136e+00 -1.16653383e-01 -3.69661272e-01 1.16485381e+00 3.40387523e-01 -9.46538150e-02 7.77029842e-02 -9.72233176e-01 -1.02623343e+00 6.22489639e-02 -4.48807478e-01 4.17527169e-01 1.12217939e+00 4.11051065e-02 -1.55538153e-02 -1.12874232e-01 5.39802015e-01 1.33371556e+00 2.43156433e-01 1.21941638e+00 -1.47096407e+00 6.46742955e-02 -2.23148063e-01 -1.07635498e+00 -1.52667522e+00 4.75223809e-01 -8.75540316e-01 3.16768974e-01 -1.69400203e+00 -3.36911738e-01 -5.70854187e-01 4.16250736e-01 -2.00528465e-02 -2.34998856e-02 1.91872388e-01 1.47948340e-01 6.61911190e-01 -8.62736344e-01 7.55997419e-01 9.86836076e-01 -4.11870778e-02 -4.08926845e-01 2.57720083e-01 -4.74702753e-03 1.26387930e+00 4.65783536e-01 -3.63991141e-01 -5.19197404e-01 -5.73499560e-01 -2.36494113e-02 -2.08153203e-02 7.60314465e-01 -1.47892141e+00 8.25463593e-01 -2.53642291e-01 2.83464372e-01 -1.17190385e+00 4.26649570e-01 -9.71253693e-01 1.02332443e-01 2.47653455e-01 1.89956859e-01 1.94999382e-01 1.28926784e-01 9.43185866e-01 -3.71942222e-01 5.24315760e-02 6.81904256e-01 -1.38444036e-01 -9.06537414e-01 4.19072330e-01 -1.46774184e-02 1.75636232e-01 1.27578092e+00 -4.94134396e-01 -1.10281639e-01 -4.23783273e-01 -4.73022550e-01 2.60037899e-01 7.83148348e-01 4.38578993e-01 8.65223050e-01 -1.21863270e+00 -7.07295060e-01 1.30689472e-01 1.69337392e-01 8.08156610e-01 -7.25926785e-03 8.88844907e-01 -1.04685104e+00 3.87679815e-01 -9.70227495e-02 -1.44304049e+00 -1.00793397e+00 3.24912578e-01 4.40859169e-01 4.67628002e-01 -8.91614437e-01 9.40698922e-01 7.77699053e-02 -7.17531204e-01 1.26672581e-01 -4.72657919e-01 2.09865067e-02 -2.40021348e-01 2.83071041e-01 7.92320907e-01 -1.74610123e-01 -1.26946354e+00 -5.91186345e-01 1.04242194e+00 4.86034453e-01 -1.42893001e-01 1.13449895e+00 -1.99756220e-01 -8.65221098e-02 6.97213233e-01 1.23560560e+00 -1.10437900e-01 -1.56174171e+00 -2.60850042e-01 -1.61905345e-02 -9.61852074e-01 1.04978532e-01 -5.17731756e-02 -7.90431023e-01 9.50331628e-01 5.80788076e-01 -8.77969339e-02 6.07610881e-01 1.91444710e-01 5.11862397e-01 3.51273775e-01 6.07624054e-01 -5.88033199e-01 -1.20269254e-01 3.98670226e-01 4.40883607e-01 -1.50246966e+00 2.67830402e-01 -6.68342710e-01 -5.00268579e-01 1.01280749e+00 5.20076275e-01 -5.15046775e-01 9.13977027e-01 1.92144781e-01 -2.00928189e-02 -1.01035260e-01 -2.18210295e-01 -2.19728008e-01 4.72212940e-01 9.75811362e-01 7.74610117e-02 -1.08971015e-01 5.40122211e-01 -4.89245176e-01 -6.23493791e-01 9.20721889e-02 3.92768770e-01 1.03135240e+00 -4.69727635e-01 -7.41547823e-01 -5.50565064e-01 5.15683368e-02 3.16949189e-01 2.54501224e-01 -2.03725532e-01 8.36953044e-01 -5.70345856e-02 9.32936847e-01 6.39619112e-01 -2.25523338e-01 4.49864089e-01 -2.54247338e-01 6.77375853e-01 -4.55418736e-01 -2.16127187e-03 -2.74870358e-02 -3.97484034e-01 -7.65056610e-01 -1.14587140e+00 -1.08551431e+00 -1.42023218e+00 -4.94694978e-01 -2.75857866e-01 2.10020080e-01 5.75209439e-01 9.26304340e-01 2.56277263e-01 1.94066782e-02 5.69424510e-01 -1.29374433e+00 -2.01524004e-01 -5.03073871e-01 -4.05713916e-01 6.11852705e-01 6.01491153e-01 -7.66325176e-01 -7.58501768e-01 4.50038254e-01]
[8.540793418884277, -2.0172038078308105]
5c758a4b-eabd-4fef-821c-f3b790375d4f
multi-resolution-3d-convolutional-neural
1805.12254
null
https://arxiv.org/abs/1805.12254v2
https://arxiv.org/pdf/1805.12254v2.pdf
Multi-level 3D CNN for Learning Multi-scale Spatial Features
3D object recognition accuracy can be improved by learning the multi-scale spatial features from 3D spatial geometric representations of objects such as point clouds, 3D models, surfaces, and RGB-D data. Current deep learning approaches learn such features either using structured data representations (voxel grids and octrees) or from unstructured representations (graphs and point clouds). Learning features from such structured representations is limited by the restriction on resolution and tree depth while unstructured representations creates a challenge due to non-uniformity among data samples. In this paper, we propose an end-to-end multi-level learning approach on a multi-level voxel grid to overcome these drawbacks. To demonstrate the utility of the proposed multi-level learning, we use a multi-level voxel representation of 3D objects to perform object recognition. The multi-level voxel representation consists of a coarse voxel grid that contains volumetric information of the 3D object. In addition, each voxel in the coarse grid that contains a portion of the object boundary is subdivided into multiple fine-level voxel grids. The performance of our multi-level learning algorithm for object recognition is comparable to dense voxel representations while using significantly lower memory.
['Sambit Ghadai', 'Xian Lee', 'Adarsh Krishnamurthy', 'Aditya Balu', 'Soumik Sarkar']
2018-05-30
null
null
null
null
['3d-object-recognition']
['computer-vision']
[ 7.46382922e-02 2.28000619e-02 -6.56464100e-02 -3.87270242e-01 -8.32447112e-01 -1.41697362e-01 4.84793603e-01 6.75166368e-01 -1.69564381e-01 3.56170326e-01 -1.31689206e-01 7.83631504e-02 -3.36528748e-01 -1.31036472e+00 -8.39462459e-01 -4.01361257e-01 -5.03202617e-01 8.56587529e-01 5.40548980e-01 3.65621299e-01 2.64382064e-01 1.39184070e+00 -1.93107820e+00 5.57333827e-01 5.93319118e-01 1.59191477e+00 9.71222445e-02 4.75906402e-01 -7.37525642e-01 2.27364242e-01 -3.91830176e-01 5.22278666e-01 4.33757752e-01 2.12105542e-01 -5.51223159e-01 3.56539458e-01 9.02540207e-01 -1.66602001e-01 -5.17559946e-02 6.45165265e-01 2.09961519e-01 8.27329084e-02 9.62694347e-01 -1.08704519e+00 -1.84573621e-01 -1.98906630e-01 -8.49227905e-01 -1.23274587e-01 1.90068454e-01 -1.18893646e-01 8.55054617e-01 -1.09303939e+00 4.87421691e-01 1.47858918e+00 6.91633523e-01 9.43546072e-02 -1.20702004e+00 -4.21306580e-01 3.46359521e-01 2.35982165e-02 -1.49815595e+00 1.62527874e-01 9.42564368e-01 -6.69013083e-01 1.22717857e+00 2.04234958e-01 8.56221974e-01 4.30980206e-01 2.58668751e-01 6.20780289e-01 1.30168915e+00 -4.19544876e-01 4.45718259e-01 -2.18623802e-01 2.52206624e-01 9.48866546e-01 3.66023272e-01 1.05620719e-01 -4.14739996e-01 -3.61610502e-01 1.32371569e+00 3.17999810e-01 1.60138056e-01 -9.54639673e-01 -9.01131988e-01 8.58551562e-01 9.67813671e-01 2.65043348e-01 -5.90232432e-01 2.97832787e-01 3.51639390e-01 -5.70787080e-02 6.24361277e-01 -2.28198548e-03 -5.02167642e-01 1.11110687e-01 -9.91929770e-01 2.98462570e-01 6.14412010e-01 7.79021561e-01 1.24520600e+00 5.88666536e-02 -1.14317574e-01 7.30917692e-01 4.33332533e-01 2.39468217e-01 2.32920989e-01 -6.03011072e-01 3.13505590e-01 1.05161214e+00 -1.00762375e-01 -1.01905191e+00 -5.74129164e-01 -4.22224969e-01 -9.89838660e-01 8.26695919e-01 2.00602069e-01 4.63215947e-01 -1.38265979e+00 9.92506027e-01 8.08998525e-01 3.26960772e-01 -3.97193849e-01 9.77625966e-01 1.02064621e+00 8.37965786e-01 -1.54799357e-01 2.32474595e-01 1.19458616e+00 -7.52743483e-01 8.92902911e-02 6.07630759e-02 7.06394792e-01 -2.70334423e-01 8.63082051e-01 2.16688171e-01 -9.02932942e-01 -7.39559591e-01 -1.00500298e+00 -1.89660773e-01 -5.52440703e-01 -1.70065090e-01 6.66838408e-01 4.95936036e-01 -8.17193151e-01 4.72370803e-01 -9.36170757e-01 -1.32619545e-01 7.59652495e-01 6.32833660e-01 -3.93793702e-01 -3.48382980e-01 -4.70991552e-01 7.54234672e-01 3.40786874e-01 -3.24130028e-01 -8.65958214e-01 -8.24874282e-01 -1.13828468e+00 1.11267202e-01 1.09910276e-02 -5.64390838e-01 7.93506444e-01 -3.98612678e-01 -1.13364494e+00 9.71512139e-01 -1.00116804e-01 -1.09480500e-01 3.78971159e-01 -6.45005330e-02 2.81848311e-01 1.03756912e-01 1.11588864e-02 5.56335688e-01 8.31494272e-01 -1.44935668e+00 -5.32018960e-01 -9.48924959e-01 -1.05825244e-02 9.94388610e-02 1.13772988e-01 -4.48833972e-01 -2.09091038e-01 -3.58706146e-01 5.39812565e-01 -5.12135267e-01 -4.54326332e-01 4.32320327e-01 -2.45954901e-01 -3.41726333e-01 1.15848804e+00 -1.96002483e-01 3.74480933e-01 -2.14459276e+00 -8.78027827e-03 5.54505825e-01 3.15982759e-01 -1.68143421e-01 -2.14623094e-01 6.58329800e-02 2.11333215e-01 1.19910343e-03 -1.86739847e-01 -2.50672042e-01 -2.02555638e-02 3.13251883e-01 6.05265349e-02 5.74002087e-01 2.22307995e-01 7.08496511e-01 -6.60332978e-01 -4.84333456e-01 6.07316315e-01 7.91517675e-01 -5.62653601e-01 1.09964296e-01 -4.38296825e-01 3.41355592e-01 -9.46207404e-01 8.86650383e-01 8.27158570e-01 -2.77820826e-01 -3.36516201e-01 -1.63695514e-01 -2.79292166e-01 4.50484306e-01 -1.36912739e+00 1.77023733e+00 -6.12417400e-01 1.40527695e-01 2.18778521e-01 -1.00471222e+00 1.32888055e+00 5.71123660e-02 8.62090170e-01 -6.63640320e-01 -8.66456181e-02 1.89864561e-01 -4.60843265e-01 1.94435373e-01 2.07303077e-01 -1.31534263e-01 -1.26583874e-01 4.25842494e-01 -1.88354328e-01 -7.24485338e-01 -4.33929563e-01 -2.90826082e-01 1.04324532e+00 2.65516471e-02 4.46867406e-01 -5.13759516e-02 2.93122441e-01 1.75984710e-01 3.69200051e-01 7.54654467e-01 2.17910767e-01 5.26602626e-01 1.74680918e-01 -8.09037685e-01 -1.02860606e+00 -1.20860898e+00 -4.47927207e-01 5.94839811e-01 9.15922672e-02 -4.44140077e-01 -3.03620845e-01 -4.46580946e-01 6.98857486e-01 2.98435122e-01 -6.12670302e-01 1.56401083e-01 -5.57300866e-01 -3.19836617e-01 8.12454708e-03 7.38570273e-01 4.06747252e-01 -7.04361916e-01 -1.10399711e+00 1.49301544e-01 7.22968042e-01 -9.83722746e-01 3.62605453e-02 5.13749003e-01 -1.48309410e+00 -9.76882458e-01 -3.26892257e-01 -8.00817072e-01 7.44584858e-01 2.71216750e-01 1.08202827e+00 -1.66638717e-02 -6.70572698e-01 5.03483772e-01 -3.08417141e-01 -4.37117726e-01 2.30115633e-02 -7.11458027e-02 1.00890867e-01 -2.24890962e-01 2.29722008e-01 -8.05687845e-01 -3.31507474e-01 1.39351606e-01 -7.68233657e-01 2.23407120e-01 5.70681453e-01 6.79574847e-01 1.33075261e+00 4.94575575e-02 1.61514938e-01 -5.78524470e-01 5.54132136e-03 -3.91898692e-01 -8.67769718e-01 -1.62108660e-01 -7.34048709e-02 2.00586319e-02 3.28615487e-01 -3.52684617e-01 -4.16125834e-01 4.79185313e-01 -5.81958592e-02 -7.71705687e-01 -5.58258295e-01 4.45968181e-01 -4.71798182e-02 -4.49312329e-01 3.69632274e-01 1.79841384e-01 -8.82965922e-02 -8.34311783e-01 4.14768636e-01 5.38256586e-01 -2.35490105e-03 -8.35954905e-01 7.92570591e-01 3.69475871e-01 3.99913609e-01 -1.12491751e+00 -6.36327744e-01 -3.02988112e-01 -1.18608117e+00 -2.21068501e-01 7.06720352e-01 -8.04465950e-01 -5.30578494e-01 1.23564571e-01 -1.19229889e+00 -4.33237374e-01 -6.60425842e-01 5.15836656e-01 -7.49393046e-01 4.80884872e-02 -4.72606957e-01 -7.23656535e-01 -8.41736645e-02 -9.30496931e-01 1.51067412e+00 -4.24411036e-02 -5.56239206e-03 -7.25665271e-01 1.03412449e-01 1.63560942e-01 1.45845801e-01 7.63893545e-01 1.48511779e+00 -3.60408127e-01 -1.04064476e+00 -3.72377187e-01 -4.04132366e-01 -7.62165710e-02 2.67302722e-01 -3.39679897e-01 -7.81925738e-01 -3.42541635e-01 -1.80276826e-01 -5.54863513e-01 7.71538794e-01 4.62980866e-01 1.69456136e+00 1.55326277e-02 -5.41317165e-01 6.87327921e-01 1.64442861e+00 -6.80760518e-02 1.17447063e-01 2.49639824e-01 8.84237587e-01 4.45741832e-01 4.22821701e-01 6.74720824e-01 2.97547996e-01 7.09584832e-01 7.90222347e-01 -2.09638283e-01 -1.65933207e-01 -1.74642503e-02 -9.29923430e-02 5.74518025e-01 -8.47451203e-03 2.77186185e-01 -1.27245057e+00 4.16639358e-01 -1.45434833e+00 -5.25306284e-01 -2.35669494e-01 2.24755549e+00 5.09610713e-01 3.01785827e-01 1.66414887e-01 1.71057597e-01 3.24340761e-01 2.07728118e-01 -6.76169217e-01 -4.80390519e-01 1.74364477e-01 4.56770927e-01 2.76369423e-01 4.00857687e-01 -1.00800276e+00 6.03864670e-01 5.94181347e+00 6.86767042e-01 -1.04116559e+00 -1.36451289e-01 4.34513599e-01 -2.28128001e-01 -1.95427358e-01 -4.18978035e-01 -9.57874417e-01 -1.01590566e-02 5.67892253e-01 1.32816255e-01 -9.09309536e-02 1.12001979e+00 -5.07501401e-02 -2.94465804e-03 -1.49056041e+00 1.24018383e+00 -8.15678313e-02 -1.46989787e+00 3.06060374e-01 3.66921335e-01 7.17243075e-01 2.84631282e-01 -5.87913096e-02 2.90798306e-01 2.67737776e-01 -1.28549755e+00 6.00219727e-01 2.06415936e-01 1.02710128e+00 -7.07483649e-01 2.91314185e-01 5.79747021e-01 -1.52729344e+00 1.44201115e-01 -6.21292233e-01 -2.02370957e-01 -2.02980757e-01 9.44497526e-01 -7.98168957e-01 4.37645733e-01 7.04123616e-01 5.16067445e-01 -3.67855310e-01 1.33806002e+00 3.83494645e-01 1.69410333e-01 -7.30351627e-01 -3.92434560e-02 6.17576063e-01 -1.58104226e-01 3.57357800e-01 1.15159309e+00 2.72993922e-01 2.42977604e-01 6.56203628e-01 9.09792244e-01 -9.14446563e-02 1.20249845e-01 -9.89818692e-01 2.96090513e-01 3.46349508e-01 9.28018332e-01 -7.11399436e-01 -2.83881456e-01 -8.21168125e-01 6.82850540e-01 7.03246415e-01 6.06162474e-02 -2.72664696e-01 -2.07989931e-01 8.08443844e-01 4.57277179e-01 6.36329353e-01 -7.72330165e-01 -6.85029268e-01 -6.17846310e-01 7.68518522e-02 -3.49536985e-01 2.60003179e-01 -5.25764048e-01 -1.24552393e+00 4.97179151e-01 3.16633657e-02 -1.16392720e+00 4.62032884e-04 -8.56378913e-01 -2.92512745e-01 9.78878796e-01 -1.52693212e+00 -1.14251423e+00 -6.49157822e-01 6.32986784e-01 5.35258889e-01 3.06457579e-02 9.84885097e-01 -1.49275601e-01 1.91089585e-01 2.14302391e-01 1.90809131e-01 1.43743262e-01 -2.21710443e-01 -1.07373142e+00 3.27802122e-01 2.28548311e-02 2.51517892e-01 4.92483266e-02 -4.49584611e-02 -6.92811608e-01 -1.55460513e+00 -1.35839272e+00 4.65434432e-01 -3.25785458e-01 1.65062010e-01 -6.27444923e-01 -1.14689553e+00 5.72017074e-01 -6.69551611e-01 6.78437531e-01 6.20975196e-01 2.03157946e-01 -5.68523347e-01 -2.11556390e-01 -1.46473539e+00 5.47814481e-02 9.77124274e-01 -4.70609039e-01 -6.01466537e-01 3.53739887e-01 5.59322119e-01 -6.33412182e-01 -1.15174448e+00 6.72373235e-01 4.90665972e-01 -8.25381160e-01 1.18234634e+00 -7.02934086e-01 3.08928698e-01 -1.64918780e-01 -4.37578201e-01 -1.24973691e+00 -5.29880226e-01 2.92281002e-01 -2.84842372e-01 5.73712051e-01 -2.54771374e-02 -2.73918957e-01 1.13662028e+00 3.27068388e-01 -2.56706208e-01 -1.22492206e+00 -1.38367486e+00 -8.83537352e-01 2.87733912e-01 -5.93528330e-01 6.29877090e-01 7.30619788e-01 -4.97978181e-01 -3.76196112e-03 3.21339190e-01 3.93682510e-01 9.70003963e-01 8.57630253e-01 7.79261589e-01 -1.78791177e+00 5.20588383e-02 -4.62840378e-01 -1.02176237e+00 -1.18385792e+00 8.44637379e-02 -1.01389158e+00 -5.61523065e-02 -1.84298944e+00 8.91520083e-02 -1.01844585e+00 -2.33253896e-01 4.32992250e-01 4.14307386e-01 2.05749318e-01 1.08706743e-01 1.80949762e-01 -3.00602883e-01 7.64226139e-01 1.38783324e+00 -3.73579890e-01 -2.88693309e-01 -1.87791958e-01 9.39051658e-02 7.51002312e-01 4.98984873e-01 -4.11949396e-01 -7.45416656e-02 -6.06083751e-01 -4.76303905e-01 1.48158744e-01 5.41756570e-01 -1.09621036e+00 1.33553773e-01 -1.34013474e-01 8.85183454e-01 -1.17131150e+00 7.47264206e-01 -1.05108142e+00 -1.12817906e-01 3.73349994e-01 -1.18568033e-01 -3.20584506e-01 4.46397811e-01 5.27104080e-01 -1.62146911e-01 1.16057523e-01 9.53604996e-01 -4.49214041e-01 -5.92720807e-01 7.68237650e-01 -1.85248569e-01 -3.74275774e-01 1.19944108e+00 -8.46284151e-01 3.13319623e-01 1.87345564e-01 -7.46492922e-01 1.40564531e-01 6.02277398e-01 3.04919422e-01 1.09404588e+00 -1.57143521e+00 -6.08863175e-01 4.89539057e-01 1.41961083e-01 7.48902202e-01 2.12777965e-02 3.05674255e-01 -5.02630651e-01 5.73145568e-01 -5.37535787e-01 -1.01342940e+00 -1.12502885e+00 4.22453791e-01 4.19995338e-01 -1.66515946e-01 -1.06148589e+00 8.25147271e-01 3.44265103e-01 -5.16727626e-01 3.86659294e-01 -7.68562615e-01 1.24131128e-01 -1.57710955e-01 2.67493486e-01 3.10329199e-01 2.59653836e-01 -6.48322821e-01 -5.52752972e-01 1.10105371e+00 -9.46018547e-02 2.81785160e-01 1.64589846e+00 4.11737114e-01 -1.11662030e-01 7.47654319e-01 1.35687292e+00 -4.31655645e-01 -1.30751145e+00 -4.60383177e-01 1.65756959e-02 -8.33326101e-01 4.02237386e-01 -5.20688593e-01 -9.24542844e-01 1.07131159e+00 7.03815460e-01 6.72872830e-03 7.08928704e-01 3.87260675e-01 5.97587407e-01 3.85708243e-01 8.92990589e-01 -6.82657063e-01 2.69618750e-01 6.80828154e-01 9.93584692e-01 -1.08740532e+00 7.79499412e-02 -6.40298426e-01 8.00674930e-02 1.18349504e+00 6.63748860e-01 -3.83638322e-01 9.01245654e-01 3.59438866e-01 -3.30749124e-01 -6.48877501e-01 -5.01720369e-01 -1.99644700e-01 4.54776078e-01 8.34633529e-01 2.02352300e-01 1.25870809e-01 3.91272455e-01 3.52360755e-01 -2.35926434e-02 -2.10972667e-01 -7.03707412e-02 1.13001668e+00 -7.62173295e-01 -8.52859259e-01 -7.05230236e-01 8.90866220e-01 1.78691104e-01 3.31108153e-01 -2.44494408e-01 7.89140284e-01 1.50527596e-01 2.64513552e-01 6.19577825e-01 -1.81889594e-01 5.41878879e-01 -1.33214975e-02 9.02213633e-01 -1.10218394e+00 -2.58845806e-01 -1.55544460e-01 -3.25996965e-01 -8.22778344e-01 -1.03203081e-01 -6.66176260e-01 -1.53919566e+00 3.85423377e-02 7.65788555e-02 1.67593285e-02 9.06991422e-01 7.27532685e-01 3.68469954e-01 3.16579223e-01 6.35066509e-01 -1.58902419e+00 -3.95578861e-01 -6.83093965e-01 -8.74831200e-01 1.03490613e-01 4.73606974e-01 -1.02638316e+00 -8.67800787e-02 -3.37436229e-01]
[8.171210289001465, -3.6596851348876953]
dc84f481-9903-44de-9639-133e1e4548de
fighting-over-fitting-with-quantization-for
2303.11803
null
https://arxiv.org/abs/2303.11803v1
https://arxiv.org/pdf/2303.11803v1.pdf
Fighting over-fitting with quantization for learning deep neural networks on noisy labels
The rising performance of deep neural networks is often empirically attributed to an increase in the available computational power, which allows complex models to be trained upon large amounts of annotated data. However, increased model complexity leads to costly deployment of modern neural networks, while gathering such amounts of data requires huge costs to avoid label noise. In this work, we study the ability of compression methods to tackle both of these problems at once. We hypothesize that quantization-aware training, by restricting the expressivity of neural networks, behaves as a regularization. Thus, it may help fighting overfitting on noisy data while also allowing for the compression of the model at inference. We first validate this claim on a controlled test with manually introduced label noise. Furthermore, we also test the proposed method on Facial Action Unit detection, where labels are typically noisy due to the subtlety of the task. In all cases, our results suggests that quantization significantly improve the results compared with existing baselines, regularization as well as other compression methods.
['Kevin Bailly', 'Arnaud Dapogny', 'Edouard Yvinec', 'Gauthier Tallec']
2023-03-21
null
null
null
null
['action-unit-detection', 'facial-action-unit-detection']
['computer-vision', 'computer-vision']
[ 5.10844588e-01 5.07025838e-01 -2.43618235e-01 -5.09506941e-01 -5.82167029e-01 -2.39529207e-01 4.36208487e-01 -5.63901626e-02 -6.21108651e-01 6.07797682e-01 1.85287565e-01 -1.00370556e-01 1.19350627e-01 -6.08822584e-01 -7.72821248e-01 -6.53827965e-01 6.80803508e-02 2.23945200e-01 -1.33287445e-01 1.78898633e-01 -7.68450052e-02 3.56419712e-01 -1.73021007e+00 2.59849250e-01 6.21458173e-01 1.20520508e+00 1.19605467e-01 2.02733427e-01 2.41682440e-01 9.17749345e-01 -6.16032839e-01 -7.34305620e-01 2.91837513e-01 -2.43094444e-01 -6.71040475e-01 2.68032670e-01 6.98699951e-01 -6.12617552e-01 -2.69830614e-01 1.16718030e+00 3.96203548e-01 9.08253491e-02 5.20202518e-01 -1.00868785e+00 -2.79746562e-01 6.89885855e-01 -3.46248865e-01 -8.42821524e-02 -9.69018340e-02 1.68541968e-01 1.16567278e+00 -7.63965607e-01 5.78338981e-01 1.23528242e+00 6.90603852e-01 8.41647804e-01 -1.40363514e+00 -7.02079773e-01 -5.78168109e-02 -3.37326527e-03 -1.30670285e+00 -9.59014118e-01 6.30410790e-01 -1.29366547e-01 7.00740635e-01 1.55705169e-01 4.68292952e-01 1.34957671e+00 -3.61867368e-01 7.90565968e-01 9.45701241e-01 -4.24284935e-01 5.01107812e-01 1.71807900e-01 1.70365814e-03 7.16237664e-01 3.01625937e-01 -8.04666255e-04 -6.02407098e-01 -1.12452023e-01 3.74800265e-01 -2.30938092e-01 -4.08191264e-01 4.54021208e-02 -5.17202377e-01 7.57086277e-01 2.75848657e-01 2.55692393e-01 -3.89015794e-01 4.69810039e-01 5.88780463e-01 7.65423998e-02 6.32492304e-01 5.27314544e-01 -5.34212649e-01 -3.54231536e-01 -1.36015010e+00 6.30366728e-02 6.91537321e-01 7.61370540e-01 6.65673018e-01 2.05831140e-01 -1.91886380e-01 9.04305696e-01 2.71646112e-01 -2.30272990e-02 5.34236968e-01 -1.30080807e+00 4.02981997e-01 5.95918298e-01 -3.68272632e-01 -1.08541262e+00 -3.28799635e-01 -6.00997627e-01 -1.06093335e+00 1.37637541e-01 5.22279561e-01 -1.38129443e-01 -8.95638883e-01 2.14446688e+00 4.55904752e-02 2.13247225e-01 -1.52731672e-01 7.65686035e-01 4.45435643e-01 3.14237684e-01 2.88577169e-01 -3.32119793e-01 1.24822140e+00 -8.92845154e-01 -9.24099803e-01 -2.84212500e-01 9.78516698e-01 -4.59478766e-01 1.28557038e+00 5.59123814e-01 -9.73805249e-01 -4.84309465e-01 -9.16409612e-01 -9.78002176e-02 -4.97315228e-02 3.41114908e-01 6.64507270e-01 7.72861302e-01 -9.44730222e-01 9.87320840e-01 -9.33407009e-01 -1.77229971e-01 9.85269606e-01 3.69708031e-01 -2.25249678e-01 -2.61380434e-01 -1.02627981e+00 6.31057739e-01 2.98568606e-01 1.80768058e-01 -7.19446003e-01 -6.30340636e-01 -7.35550761e-01 4.00278538e-01 4.33413118e-01 -3.22472155e-01 1.20430243e+00 -1.12399065e+00 -1.27471745e+00 7.56250441e-01 -1.34252533e-01 -6.47628486e-01 5.23285627e-01 -2.89282233e-01 -1.91150908e-03 9.62382853e-02 -3.09463322e-01 1.02060103e+00 7.84610569e-01 -1.10586643e+00 -3.03181261e-01 -3.99279177e-01 1.28046930e-01 -2.16037065e-01 -8.54880512e-01 -2.20864475e-01 -3.05653125e-01 -5.53111136e-01 -2.36785654e-02 -8.81460428e-01 -1.02514617e-01 1.19853057e-01 -3.68959814e-01 -3.27587038e-01 7.19712079e-01 -5.47219992e-01 1.37533522e+00 -2.46458888e+00 -1.10281833e-01 1.31187856e-01 2.46549547e-01 4.43308830e-01 -1.73836052e-01 1.19546182e-01 -9.65373889e-02 5.84226370e-01 -2.57664770e-01 -9.29481685e-01 1.43690526e-01 5.22168398e-01 -1.89292476e-01 3.87585282e-01 3.72221947e-01 7.13984549e-01 -3.90205204e-01 -5.81716478e-01 -3.13664258e-01 5.47706127e-01 -7.59835958e-01 3.49883735e-02 -4.38400626e-01 1.94984034e-01 -6.88592270e-02 7.21189499e-01 5.81836104e-01 -3.99043113e-01 1.97910160e-01 -3.32787693e-01 3.28998119e-01 4.10883099e-01 -1.01787376e+00 1.61128891e+00 -2.40550295e-01 7.24359453e-01 1.08127758e-01 -1.25181532e+00 5.03927708e-01 5.16060412e-01 3.37058067e-01 -7.08417356e-01 3.24178308e-01 9.61296484e-02 8.69350806e-02 -5.03482342e-01 2.98731983e-01 -6.34319186e-02 3.21075082e-01 4.43726271e-01 1.24359034e-01 2.48639837e-01 3.81777167e-01 6.75190166e-02 1.21510029e+00 1.01122200e-01 -3.90924439e-02 2.25645900e-02 -8.23358446e-03 -2.31706575e-01 7.74659097e-01 6.54696643e-01 -3.39501739e-01 4.24417615e-01 6.41039133e-01 -3.76869023e-01 -1.09822750e+00 -2.87830532e-01 -2.86729664e-01 1.05549848e+00 -4.17826384e-01 -8.41446698e-01 -1.12964010e+00 -7.50128031e-01 -2.09311441e-01 6.36496305e-01 -7.01105058e-01 -3.61934781e-01 -4.01241302e-01 -8.42779696e-01 9.43923652e-01 4.45046335e-01 6.05666876e-01 -7.95650005e-01 -8.48785579e-01 2.45115068e-02 -2.75581479e-01 -1.29180682e+00 -5.34218214e-02 3.74924630e-01 -1.09986353e+00 -8.87316585e-01 -2.94444263e-01 -3.37117016e-01 5.48874199e-01 -5.19173630e-02 1.17895162e+00 5.92240989e-01 -1.42867714e-01 -6.28296658e-02 -3.75956804e-01 -4.01673585e-01 -4.74562615e-01 1.83171123e-01 -1.66831631e-02 -1.86840128e-02 6.20362341e-01 -7.21590877e-01 -4.75823790e-01 1.21007979e-01 -1.15024471e+00 3.95133682e-02 7.07273841e-01 8.41535032e-01 5.10039210e-01 2.45694295e-01 4.02430803e-01 -9.50458825e-01 5.62001348e-01 -2.61759877e-01 -4.74250793e-01 6.81665316e-02 -7.51376331e-01 2.49060273e-01 5.40464580e-01 -6.43114865e-01 -7.21979141e-01 1.15868948e-01 -1.89485252e-01 -6.42293870e-01 -3.08571815e-01 6.07016742e-01 -1.00841552e-01 7.72326961e-02 7.64902353e-01 -6.88052922e-02 5.24328910e-02 -6.01292729e-01 1.78761184e-01 6.42587900e-01 9.03711393e-02 -4.93129820e-01 4.44662482e-01 3.58947068e-01 1.58159912e-01 -8.02318275e-01 -1.02894056e+00 1.21130578e-01 -5.13610542e-01 -1.49701551e-01 6.75916314e-01 -7.72489250e-01 -5.55041909e-01 7.73370191e-02 -1.11302662e+00 -3.91760796e-01 -3.47700268e-01 3.33508879e-01 -4.20853078e-01 3.63215476e-01 -7.40352452e-01 -8.31526041e-01 -1.89931646e-01 -1.19071901e+00 1.01967561e+00 -1.01655677e-01 -3.93377602e-01 -7.88209081e-01 -2.23649040e-01 4.60868388e-01 3.99281532e-01 2.33263507e-01 9.79925632e-01 -7.92731047e-01 -5.16669691e-01 -2.78892964e-01 -2.85835028e-01 7.37064123e-01 -4.13816161e-02 -5.40210903e-02 -1.43963230e+00 -2.33442977e-01 2.03116909e-01 -8.81448209e-01 9.29238915e-01 1.93621561e-01 1.51746905e+00 -5.86472213e-01 -2.12969724e-02 4.96699840e-01 1.21270812e+00 -2.65443802e-01 6.36871099e-01 1.45435452e-01 5.51347554e-01 8.00167263e-01 3.43549341e-01 5.15366018e-01 -1.81053262e-02 6.34410620e-01 6.68086112e-01 -1.33161932e-01 -1.07473813e-01 -1.57639325e-01 1.51761234e-01 7.07079113e-01 -1.59986854e-01 -2.24617228e-01 -8.06039870e-01 2.47643217e-01 -1.63475764e+00 -6.97777271e-01 1.42665789e-01 1.99567974e+00 1.18134212e+00 2.04146266e-01 -7.00504659e-03 4.90741849e-01 3.25206906e-01 -3.22384275e-02 -3.77283931e-01 -2.06006572e-01 -2.41082674e-03 1.44910038e-01 4.45469469e-01 5.67077734e-02 -1.01650572e+00 7.48440564e-01 6.36693001e+00 1.14660490e+00 -1.20740879e+00 2.23150060e-01 1.09423649e+00 -4.02044296e-01 2.59816982e-02 -1.39936253e-01 -7.37350166e-01 5.85551083e-01 1.21876597e+00 3.62001181e-01 4.76887137e-01 9.01399076e-01 2.11371556e-01 -8.81599411e-02 -1.41965938e+00 1.01331043e+00 -1.94241423e-02 -1.12325633e+00 -1.87333878e-02 3.59445453e-01 5.18045008e-01 -5.55668846e-02 1.56719983e-01 1.99185714e-01 2.00289506e-02 -1.19571352e+00 7.31923044e-01 2.22628593e-01 6.35428607e-01 -6.23672247e-01 8.67833912e-01 5.81533074e-01 -6.16788805e-01 -7.74375945e-02 -6.02134764e-01 -8.32053274e-02 -6.22096062e-02 8.54216456e-01 -7.13899970e-01 5.00170551e-02 5.70421219e-01 4.25211102e-01 -7.53918290e-01 7.15650320e-01 -1.93008080e-01 1.12312579e+00 -5.62501848e-01 2.46112287e-01 3.26012045e-01 -4.00233306e-02 4.39296290e-02 1.15776014e+00 8.36143494e-02 1.47208804e-03 2.36191275e-03 1.01538539e+00 -4.36087728e-01 8.44781697e-02 -6.46130204e-01 -3.12407076e-01 4.61605668e-01 1.13002264e+00 -5.20321131e-01 -2.71602601e-01 -4.31549519e-01 6.74661577e-01 6.21990323e-01 3.36852252e-01 -8.40544462e-01 1.00751460e-01 3.57395738e-01 1.88429430e-01 5.81721254e-02 -7.85333514e-02 -3.87348354e-01 -9.06523585e-01 2.29514122e-01 -9.69353199e-01 1.42283291e-01 -4.64510262e-01 -9.09042358e-01 4.65600044e-01 -2.62515813e-01 -9.22937930e-01 -2.22753346e-01 -4.86386895e-01 -1.57043427e-01 4.30909991e-01 -1.54728591e+00 -8.81675422e-01 -2.27576077e-01 2.55737662e-01 5.66328168e-01 -1.24706356e-02 9.54350829e-01 6.36058331e-01 -7.93497920e-01 9.03789699e-01 -1.39338061e-01 2.52590030e-01 6.56845391e-01 -8.90392900e-01 -6.08803220e-02 7.49517500e-01 3.65664840e-01 5.91096640e-01 5.93405068e-01 -4.40572023e-01 -1.09110522e+00 -1.18924522e+00 7.81951308e-01 -1.78634495e-01 3.70691895e-01 -5.72853565e-01 -1.10940921e+00 5.72747171e-01 7.24424869e-02 1.05765209e-01 8.49120855e-01 1.85757831e-01 -4.25191373e-01 -3.85885499e-02 -1.13153934e+00 4.58088219e-01 1.02021992e+00 -4.33825850e-01 -1.40814647e-01 3.88941050e-01 7.55970776e-01 -8.36962312e-02 -8.55840147e-01 4.08826232e-01 4.62001294e-01 -1.00390303e+00 5.83223343e-01 -6.37269199e-01 8.64936531e-01 3.10101867e-01 -4.56720799e-01 -1.08267438e+00 -1.39219955e-01 -4.13286209e-01 -3.32756579e-01 1.43603146e+00 5.54823160e-01 -1.86065108e-01 1.06050849e+00 1.02881241e+00 6.91772327e-02 -9.22545850e-01 -1.07007658e+00 -7.20153093e-01 -5.64908497e-02 -5.87382197e-01 3.58523458e-01 1.05405986e+00 -1.69448584e-01 3.04832637e-01 -5.33681333e-01 -1.32187232e-01 6.31516159e-01 -5.39837778e-01 5.02403438e-01 -1.22028780e+00 -4.52388614e-01 -4.00229424e-01 -3.70869994e-01 -8.11934114e-01 4.65171129e-01 -7.72549868e-01 -2.38567162e-02 -9.79250193e-01 9.63031799e-02 -5.86485744e-01 -4.33065258e-02 7.82526374e-01 6.17897213e-02 4.91563648e-01 2.08283350e-01 3.26700360e-01 -4.85145718e-01 5.23890555e-01 1.04169226e+00 -1.27431691e-01 1.33890733e-01 -1.40857443e-01 -5.58386981e-01 9.21139836e-01 8.06762338e-01 -6.35793924e-01 -4.65856582e-01 -6.77046418e-01 4.32665318e-01 -3.54129195e-01 3.08750361e-01 -1.05395889e+00 1.14847295e-01 1.21160157e-01 2.47724384e-01 -6.93594739e-02 5.27939677e-01 -1.10184300e+00 -1.40975416e-01 3.29336584e-01 -5.86979806e-01 -3.04822803e-01 1.74007341e-01 5.14675617e-01 -3.71578395e-01 -4.37421203e-01 9.94199157e-01 -1.60205796e-01 -2.29635507e-01 1.21349677e-01 -1.94282278e-01 5.37015721e-02 5.70346832e-01 -2.92564601e-01 -1.17859557e-01 -4.15130258e-01 -7.34639287e-01 -2.50537582e-02 3.82057905e-01 1.18279248e-01 3.84411246e-01 -1.13081431e+00 -4.72814590e-01 2.55172282e-01 -1.21036388e-01 1.24637611e-01 -1.93206891e-02 9.36157703e-01 -2.13572502e-01 2.67395526e-01 1.26011565e-01 -5.88202655e-01 -1.32249904e+00 5.20574391e-01 2.52111465e-01 -2.60687113e-01 -5.23097157e-01 8.96665275e-01 -3.64654511e-02 -1.68865900e-02 9.28368986e-01 -4.16769266e-01 -1.44053638e-01 2.32244506e-01 4.95214164e-01 3.27056319e-01 2.80476719e-01 -3.80770594e-01 -6.42364249e-02 2.50223249e-01 -1.68443799e-01 1.31148636e-01 1.33100140e+00 -1.21314060e-02 -7.42570460e-02 2.80201882e-01 1.16634488e+00 -3.63047153e-01 -1.36899531e+00 -2.28772074e-01 1.43049225e-01 -4.19685811e-01 2.09201470e-01 -7.01414347e-01 -1.32698298e+00 1.08867383e+00 6.53079689e-01 1.91699892e-01 1.26278019e+00 -1.35809094e-01 6.64385140e-01 6.30440116e-01 2.51046002e-01 -1.13683057e+00 1.31911024e-01 2.31337085e-01 4.20768023e-01 -1.27061415e+00 7.95638189e-03 -4.00150478e-01 -4.22676742e-01 9.20767307e-01 4.65454251e-01 8.78866464e-02 4.41768140e-01 3.43514144e-01 -1.32945448e-01 -1.72404841e-01 -9.80548680e-01 3.00891213e-02 3.08352429e-02 3.05291831e-01 5.52381039e-01 -2.72262603e-01 -1.94571197e-01 5.68908215e-01 -1.75427049e-01 1.95142269e-01 3.77070546e-01 8.68904352e-01 -3.31640095e-01 -1.10459542e+00 -5.92829324e-02 5.60453355e-01 -8.39760125e-01 -2.23343238e-01 -3.99088144e-01 6.81015074e-01 4.03916717e-01 9.46330428e-01 5.63725382e-02 -3.75421077e-01 2.13095367e-01 3.49886924e-01 2.79062837e-01 -5.85148931e-01 -3.76942456e-01 1.37750849e-01 1.59169316e-01 -7.50158250e-01 -5.67882717e-01 -4.26488727e-01 -9.62646425e-01 -2.71197379e-01 -5.21022737e-01 -2.51991265e-02 7.58294821e-01 1.05487669e+00 4.77595747e-01 5.15614271e-01 2.57606477e-01 -6.37839615e-01 -9.88225460e-01 -1.27348804e+00 -5.94844460e-01 6.22544348e-01 1.88036591e-01 -6.66622758e-01 -5.65830767e-01 3.29337478e-01]
[8.997003555297852, 3.4872231483459473]
5503382b-67fe-4327-acc4-2971f863f720
crossing-generative-adversarial-networks-for
1801.01760
null
http://arxiv.org/abs/1801.01760v1
http://arxiv.org/pdf/1801.01760v1.pdf
Crossing Generative Adversarial Networks for Cross-View Person Re-identification
Person re-identification (\textit{re-id}) refers to matching pedestrians across disjoint yet non-overlapping camera views. The most effective way to match these pedestrians undertaking significant visual variations is to seek reliably invariant features that can describe the person of interest faithfully. Most of existing methods are presented in a supervised manner to produce discriminative features by relying on labeled paired images in correspondence. However, annotating pair-wise images is prohibitively expensive in labors, and thus not practical in large-scale networked cameras. Moreover, seeking comparable representations across camera views demands a flexible model to address the complex distributions of images. In this work, we study the co-occurrence statistic patterns between pairs of images, and propose to crossing Generative Adversarial Network (Cross-GAN) for learning a joint distribution for cross-image representations in a unsupervised manner. Given a pair of person images, the proposed model consists of the variational auto-encoder to encode the pair into respective latent variables, a proposed cross-view alignment to reduce the view disparity, and an adversarial layer to seek the joint distribution of latent representations. The learned latent representations are well-aligned to reflect the co-occurrence patterns of paired images. We empirically evaluate the proposed model against challenging datasets, and our results show the importance of joint invariant features in improving matching rates of person re-id with comparison to semi/unsupervised state-of-the-arts.
['Yang Wang', 'Chengyuan Zhang', 'Lin Wu']
2018-01-04
null
null
null
null
['cross-view-person-re-identification']
['computer-vision']
[ 2.07388297e-01 -2.23985359e-01 5.09035960e-02 -5.35312414e-01 -6.99316859e-01 -5.74685276e-01 8.16915989e-01 -3.51978719e-01 -3.56774539e-01 5.59603333e-01 2.73134112e-01 3.89758795e-01 7.36773340e-03 -8.33910823e-01 -8.52894664e-01 -5.70715368e-01 3.42988014e-01 5.26489496e-01 -7.32505694e-02 4.50064838e-02 -7.30508268e-02 5.09711981e-01 -1.66343927e+00 5.66281602e-02 6.74814880e-01 6.05289936e-01 3.05332206e-02 5.10630488e-01 2.36465052e-01 2.83499181e-01 -4.10464644e-01 -1.13703561e+00 6.77003860e-01 -5.58946669e-01 -4.40905511e-01 6.89757466e-01 1.01936913e+00 -3.78173262e-01 -7.63331652e-01 1.28109443e+00 4.38226163e-01 3.43325794e-01 9.09160197e-01 -1.62159228e+00 -9.48973775e-01 -6.55483007e-02 -6.88571393e-01 1.94941670e-01 5.63472211e-01 3.27398956e-01 7.94936895e-01 -6.08440876e-01 3.77012312e-01 1.41629112e+00 8.00461650e-01 6.96687281e-01 -1.32958007e+00 -7.34263420e-01 9.13612396e-02 3.30611259e-01 -1.63225961e+00 -5.83218455e-01 7.90852666e-01 -6.29384041e-01 5.04046977e-01 1.39147416e-01 7.68742323e-01 1.31866384e+00 -6.02339841e-02 4.47801113e-01 9.87133443e-01 -3.21246684e-01 -6.03118464e-02 2.92918950e-01 -1.60437495e-01 5.64200103e-01 4.63103980e-01 2.77518690e-01 -3.92639160e-01 -7.71751180e-02 9.09421206e-01 6.05720580e-01 -6.66714236e-02 -5.89891136e-01 -1.02067649e+00 7.17352331e-01 3.77334386e-01 9.21227038e-02 -2.75745481e-01 -1.15522258e-02 3.03786904e-01 -3.27894613e-02 2.23030463e-01 9.12865438e-03 1.00161962e-01 2.94019133e-01 -8.90468299e-01 4.65053678e-01 1.62527427e-01 1.14260209e+00 8.55193973e-01 -1.19447626e-01 -3.33191156e-01 9.07587588e-01 3.85553241e-01 7.60055125e-01 3.82291466e-01 -8.71483922e-01 6.80503428e-01 5.89420557e-01 2.79651970e-01 -1.26523626e+00 1.30654410e-01 -2.40344688e-01 -1.20570314e+00 9.83495712e-02 5.98940611e-01 1.29965926e-02 -7.04950571e-01 1.91300917e+00 1.83063105e-01 2.47561514e-01 1.32684827e-01 9.48371232e-01 6.59326971e-01 5.02811909e-01 1.49876863e-01 2.16766130e-02 1.36180890e+00 -9.41792071e-01 -4.29054469e-01 -4.13779944e-01 -1.73495308e-01 -8.32580745e-01 6.35877907e-01 -1.82533503e-01 -1.02729249e+00 -1.05186927e+00 -8.43287945e-01 -1.46301705e-02 -2.97478437e-01 1.61852121e-01 1.01550229e-01 8.53245497e-01 -9.95428085e-01 3.40108395e-01 -4.66835827e-01 -6.12866163e-01 3.26887995e-01 3.37026238e-01 -7.47207940e-01 -3.79362255e-01 -9.88057852e-01 7.48692811e-01 1.02122307e-01 1.26755297e-01 -9.32019234e-01 -4.71040189e-01 -1.12747657e+00 -3.59664951e-03 2.95554437e-02 -1.06342614e+00 5.98579288e-01 -1.11390984e+00 -1.19364166e+00 1.29752743e+00 -3.38807285e-01 -3.56666774e-01 7.62295842e-01 -8.98374766e-02 -4.80513573e-01 5.49837239e-02 5.86387873e-01 8.21080208e-01 1.10263777e+00 -1.40443587e+00 -6.33335412e-01 -5.23582697e-01 1.74747854e-02 1.73368052e-01 -3.60327065e-01 -2.12686509e-03 -7.96987236e-01 -6.47564054e-01 -1.56099126e-01 -1.06469107e+00 -2.18523756e-01 7.31476843e-02 -4.22090650e-01 -3.94284874e-02 5.57722092e-01 -8.75561595e-01 5.64614594e-01 -2.08474064e+00 2.43895054e-01 6.36458769e-02 1.05274640e-01 1.37055412e-01 -1.05526775e-01 2.18460977e-01 -1.93268120e-01 -1.88813895e-01 -1.19015165e-01 -7.21678972e-01 6.57882541e-02 1.71572864e-01 -1.30991906e-01 8.15069437e-01 1.00603312e-01 9.70665514e-01 -8.42046976e-01 -6.13812745e-01 5.81845939e-01 6.25007093e-01 -4.90817010e-01 7.47468233e-01 3.66060108e-01 6.31399035e-01 -2.38887250e-01 4.73485827e-01 8.34626079e-01 -1.24012046e-01 3.26092131e-02 -5.54048836e-01 9.50350538e-02 -4.22465801e-01 -1.25981951e+00 1.47654378e+00 -2.33656228e-01 4.62291837e-01 -2.94068605e-01 -1.20487273e+00 9.88232911e-01 1.50027812e-01 5.40784597e-01 -6.07527494e-01 -5.51347807e-02 -1.19422778e-01 -4.32363033e-01 -4.01683420e-01 3.34560990e-01 -3.96456867e-02 -2.36059263e-01 1.61467582e-01 1.03769742e-01 2.74304032e-01 2.94353575e-01 2.27737166e-02 4.59473729e-01 2.46011671e-02 2.72560090e-01 -3.72070167e-03 7.75098979e-01 -4.95528966e-01 5.93222380e-01 6.58565402e-01 -3.15053791e-01 1.01862979e+00 2.19314583e-02 -7.43236005e-01 -1.50267375e+00 -1.33257377e+00 -5.74297570e-02 5.67362785e-01 4.91195470e-01 -1.68315068e-01 -8.13636899e-01 -6.75775111e-01 6.37996569e-02 3.83873284e-01 -6.73457146e-01 -8.14048424e-02 -4.33157533e-01 -5.67414999e-01 5.20745575e-01 5.61335444e-01 8.42068553e-01 -6.69060111e-01 -3.28231931e-01 -8.54390338e-02 -3.73919815e-01 -1.38676560e+00 -9.48208272e-01 -3.44077319e-01 -3.68446678e-01 -1.13859010e+00 -1.18463755e+00 -8.36431086e-01 9.63118672e-01 5.77894568e-01 1.08655024e+00 -2.32847214e-01 -2.57214189e-01 6.88768506e-01 -1.09818481e-01 2.24761039e-01 -2.42096260e-01 -3.36685956e-01 3.68563503e-01 5.96440792e-01 4.72597539e-01 -5.92390120e-01 -7.74721563e-01 6.17616832e-01 -6.95743322e-01 9.47693065e-02 4.40655291e-01 1.01440656e+00 6.59876287e-01 -2.17078463e-03 1.93404704e-01 -5.72705925e-01 2.37208381e-02 -3.63600045e-01 -4.85402465e-01 4.87742066e-01 -2.72643179e-01 -1.04330145e-02 6.29146218e-01 -3.58598351e-01 -9.98339117e-01 1.84835836e-01 7.24610463e-02 -7.33410239e-01 -4.96128768e-01 -2.74890512e-01 -6.00523174e-01 3.49107273e-02 3.35459560e-01 4.26344991e-01 -7.77910324e-03 -1.74207106e-01 5.16113758e-01 5.24025798e-01 8.47956717e-01 -6.21103704e-01 1.12246335e+00 5.82068026e-01 -3.93648744e-02 -6.54920757e-01 -6.87085688e-01 -5.59551597e-01 -9.50126648e-01 -2.59674221e-01 1.14481294e+00 -1.16929960e+00 -5.32411814e-01 4.67412621e-01 -1.07877588e+00 2.74598509e-01 -3.55831504e-01 3.30442011e-01 -6.05863154e-01 6.14090443e-01 -1.84293330e-01 -6.81238890e-01 -1.69722378e-01 -1.28028440e+00 1.20662940e+00 4.94206846e-01 -8.27758238e-02 -9.38543200e-01 1.45557329e-01 7.74803698e-01 1.16712637e-01 2.79912829e-01 3.78835142e-01 -4.29845423e-01 -5.93834817e-01 -3.79327059e-01 -4.48251307e-01 3.55900168e-01 3.87354195e-01 -3.08923125e-01 -9.25059319e-01 -5.79079211e-01 -4.35969830e-01 -1.25202730e-01 6.49680853e-01 4.71651912e-01 9.56259847e-01 -3.08285385e-01 -4.85449791e-01 8.12407732e-01 1.38511181e+00 -4.65830043e-02 7.00289130e-01 3.24596375e-01 8.82130444e-01 7.00190425e-01 2.65974998e-01 4.22470063e-01 6.11048520e-01 1.06220675e+00 1.90778852e-01 -1.68087274e-01 -1.62188590e-01 -5.60816646e-01 3.05437982e-01 4.51848984e-01 -3.58293980e-01 -1.88599393e-01 -5.40566921e-01 6.68571651e-01 -1.82470024e+00 -1.40622854e+00 1.63877383e-01 2.33141708e+00 4.04826969e-01 -2.35527381e-01 3.98346215e-01 -8.16028491e-02 1.22719169e+00 6.12504454e-03 -4.24442142e-01 1.04103439e-01 -1.11967959e-01 -2.32453182e-01 4.88862157e-01 3.18047404e-01 -1.21049631e+00 5.91425359e-01 5.39671087e+00 6.34369314e-01 -6.51262581e-01 2.98339248e-01 7.67037272e-01 1.70854196e-01 -1.56716466e-01 -1.10124193e-01 -9.56857443e-01 7.59468496e-01 5.18768728e-01 -9.79393721e-02 3.24977249e-01 8.21250677e-01 -3.52497362e-02 3.05939525e-01 -1.34721506e+00 1.52619195e+00 5.38249016e-01 -1.18614495e+00 3.26621532e-01 2.39855826e-01 1.01355541e+00 -5.32302141e-01 3.16411138e-01 9.99062955e-02 2.73513466e-01 -1.07294965e+00 7.81556368e-01 8.53972435e-01 9.24769938e-01 -8.64440918e-01 8.77382517e-01 5.52815348e-02 -1.50117731e+00 1.33679118e-02 -5.33717394e-01 2.01217011e-01 2.63932139e-01 1.38442039e-01 -3.97927403e-01 7.31487215e-01 9.29815531e-01 9.22364354e-01 -7.44951546e-01 8.35844874e-01 1.42819524e-01 -4.16501760e-02 -3.86806317e-02 4.90862310e-01 -1.18381448e-01 -4.76823002e-01 3.79690528e-01 1.11435151e+00 4.24274296e-01 -1.69840664e-01 3.92879307e-01 9.87270117e-01 -1.75658669e-02 -1.55744106e-01 -8.00606310e-01 3.41737002e-01 3.33860338e-01 9.36996698e-01 -4.35006976e-01 -3.60325575e-01 -7.09655285e-01 1.54495430e+00 3.26958418e-01 4.07819241e-01 -8.88188124e-01 1.62503898e-01 8.41831386e-01 3.31817418e-01 3.66482735e-01 -4.79659587e-02 6.24383502e-02 -1.34965789e+00 1.30498677e-01 -8.16376865e-01 5.47353446e-01 -6.19006634e-01 -1.99725294e+00 6.38454080e-01 1.80836439e-01 -1.50442159e+00 -4.56608295e-01 -3.16383064e-01 -5.57596028e-01 1.07345963e+00 -1.33789718e+00 -1.75034249e+00 -6.42488062e-01 1.08124065e+00 6.43088341e-01 -6.25110030e-01 7.74902880e-01 5.13389826e-01 -5.92882395e-01 1.00271463e+00 1.99089229e-01 4.67880577e-01 8.44042122e-01 -1.09129465e+00 4.98049468e-01 1.15241301e+00 1.92457736e-01 6.46461725e-01 6.37653828e-01 -5.17664254e-01 -1.13589954e+00 -1.35654783e+00 6.76508486e-01 -5.12507975e-01 9.09588039e-02 -2.55384088e-01 -5.74143827e-01 8.32486391e-01 2.26114959e-01 4.55835015e-01 6.92558646e-01 -1.68552592e-01 -4.43660855e-01 -3.29166740e-01 -1.18281877e+00 4.59574908e-01 1.15048683e+00 -6.92915916e-01 -5.81252873e-01 2.40731850e-01 2.15200022e-01 -1.18190236e-01 -7.45375216e-01 1.65696383e-01 5.89951515e-01 -1.08841574e+00 1.59441805e+00 -4.22634482e-01 3.15365821e-01 -4.77962375e-01 -4.36858624e-01 -1.14851892e+00 -4.61501807e-01 -3.09952408e-01 3.40732157e-01 1.70894134e+00 -2.86010355e-01 -5.06808043e-01 8.19714785e-01 7.24942863e-01 2.35991314e-01 -1.76852033e-01 -9.94698405e-01 -6.53250992e-01 -1.68814555e-01 -2.94256601e-02 5.93797684e-01 8.10979068e-01 -5.06981075e-01 1.38746023e-01 -7.84400165e-01 5.39978266e-01 1.28222799e+00 1.65554389e-01 1.23952770e+00 -1.01966298e+00 -4.61910397e-01 -2.30753526e-01 -9.59119260e-01 -9.85112071e-01 3.68511707e-01 -8.25384378e-01 -2.18507200e-01 -1.17844617e+00 7.38653421e-01 -3.35756600e-01 -8.38677436e-02 4.63733748e-02 -4.23035264e-01 5.23213625e-01 3.14467221e-01 3.71854305e-01 -5.93452692e-01 7.22207367e-01 9.89958346e-01 -5.58724940e-01 2.32588977e-01 1.26081944e-01 -5.33103049e-01 6.39432132e-01 3.85166407e-01 -3.10810447e-01 -3.52071017e-01 -3.26023996e-01 -2.07571685e-01 -5.86757138e-02 9.03235972e-01 -1.31583965e+00 2.99527287e-01 -1.23298883e-01 9.77293313e-01 -4.31426644e-01 4.93562490e-01 -9.35865641e-01 5.30466080e-01 1.00038134e-01 -2.73551166e-01 3.22338581e-01 -1.28822163e-01 7.40645528e-01 -3.20481449e-01 -2.13993728e-01 9.05788720e-01 -2.30461657e-01 -6.51833594e-01 6.31309628e-01 -4.09393720e-02 1.19920164e-01 1.14257526e+00 -4.37334239e-01 1.05266362e-01 -4.62147385e-01 -5.87146401e-01 -3.95404100e-02 8.30975235e-01 4.48231816e-01 7.00579524e-01 -1.68468356e+00 -7.97509491e-01 5.40488958e-01 3.02536219e-01 -1.89695463e-01 5.81953704e-01 1.87593728e-01 -2.13679790e-01 2.61920750e-01 -7.82922566e-01 -7.80032754e-01 -1.25137627e+00 7.50415742e-01 5.30936778e-01 -3.03244710e-01 -5.41512728e-01 6.42568231e-01 5.50180435e-01 -4.56399918e-01 2.03117188e-02 2.83374578e-01 -2.54338145e-01 -1.25183031e-01 5.21417499e-01 3.13463926e-01 -3.43322009e-01 -1.27588224e+00 -2.61128128e-01 9.37773168e-01 -4.41646762e-02 -6.52452558e-02 1.07452905e+00 -4.39089060e-01 1.30501419e-01 7.89708942e-02 1.33630359e+00 -1.33426920e-01 -1.59253836e+00 -4.52022344e-01 -5.83262026e-01 -9.88252521e-01 -4.26778585e-01 -9.21983197e-02 -1.17357957e+00 7.77725220e-01 9.62650716e-01 -1.70408458e-01 8.61445010e-01 -2.64144316e-02 7.62590170e-01 -4.46106717e-02 4.71280843e-01 -7.53721476e-01 3.28871787e-01 -4.89272922e-02 6.67933881e-01 -1.46896911e+00 5.72055317e-02 -3.50632936e-01 -6.88833237e-01 9.80650783e-01 7.44072497e-01 -2.51323104e-01 4.53832358e-01 -1.74334556e-01 -1.74584791e-01 7.98779353e-02 -2.26062015e-02 -2.75592893e-01 4.46018100e-01 1.06747520e+00 1.42929778e-01 1.26560509e-01 3.53065193e-01 4.86471385e-01 -1.65394858e-01 -3.25124621e-01 6.60718977e-02 3.98510337e-01 1.48317277e-01 -1.24478853e+00 -7.60274172e-01 1.47540301e-01 -2.05161676e-01 2.21514389e-01 -3.54864933e-02 6.11086607e-01 5.37763178e-01 9.12214041e-01 3.32698226e-01 -3.88167590e-01 4.30840820e-01 -1.95098057e-01 6.03627622e-01 -2.73430973e-01 -1.71731561e-01 -2.51170218e-01 -2.96371728e-01 -3.00488651e-01 -7.27058470e-01 -9.63522494e-01 -4.77645874e-01 -4.72505033e-01 4.78707068e-02 -9.38541591e-02 1.56830385e-01 9.82414663e-01 1.70887157e-01 3.27673465e-01 7.12412119e-01 -1.09739435e+00 -4.68605876e-01 -7.06747234e-01 -4.44578916e-01 1.22344458e+00 1.58857077e-01 -9.43728983e-01 -1.62545860e-01 5.75914204e-01]
[14.643656730651855, 0.9946673512458801]
4a2eb40d-1715-4024-b885-9770a1bb983f
asking-clarifying-questions-based-on-negative
2107.05760
null
https://arxiv.org/abs/2107.05760v1
https://arxiv.org/pdf/2107.05760v1.pdf
Asking Clarifying Questions Based on Negative Feedback in Conversational Search
Users often need to look through multiple search result pages or reformulate queries when they have complex information-seeking needs. Conversational search systems make it possible to improve user satisfaction by asking questions to clarify users' search intents. This, however, can take significant effort to answer a series of questions starting with "what/why/how". To quickly identify user intent and reduce effort during interactions, we propose an intent clarification task based on yes/no questions where the system needs to ask the correct question about intents within the fewest conversation turns. In this task, it is essential to use negative feedback about the previous questions in the conversation history. To this end, we propose a Maximum-Marginal-Relevance (MMR) based BERT model (MMR-BERT) to leverage negative feedback based on the MMR principle for the next clarifying question selection. Experiments on the Qulac dataset show that MMR-BERT outperforms state-of-the-art baselines significantly on the intent identification task and the selected questions also achieve significantly better performance in the associated document retrieval tasks.
['W. Bruce Croft', 'Qingyao Ai', 'Keping Bi']
2021-07-12
null
null
null
null
['conversational-search', 'question-selection']
['natural-language-processing', 'natural-language-processing']
[ 2.38018662e-01 6.84958175e-02 -4.10521567e-01 -5.95866799e-01 -1.15386379e+00 -6.65495157e-01 5.80092490e-01 1.83569789e-01 -6.67273283e-01 4.49780315e-01 6.29051507e-01 -7.70119429e-01 -2.73474693e-01 -2.24301443e-01 1.59375638e-01 3.21363583e-02 4.43719655e-01 5.72564006e-01 2.80815423e-01 -7.25629926e-01 7.71909952e-01 2.08012331e-02 -1.27247190e+00 7.25046039e-01 1.13672078e+00 7.37450361e-01 6.98749185e-01 9.72440481e-01 -4.35488939e-01 9.96819913e-01 -4.94497389e-01 -3.85909766e-01 -1.54267237e-01 -5.05569756e-01 -1.42241716e+00 -1.72934905e-01 2.83822715e-01 -7.56824255e-01 -1.51252672e-01 5.39186001e-01 3.58884841e-01 5.78177571e-01 5.10426462e-01 -1.10426021e+00 -6.30903721e-01 2.99289674e-01 -2.22399443e-01 6.34391963e-01 9.87129986e-01 3.42230387e-02 1.53890789e+00 -1.16348517e+00 1.80822432e-01 1.25438046e+00 2.50012398e-01 6.35206282e-01 -9.57587183e-01 -3.32657605e-01 4.28329498e-01 5.33864200e-01 -8.99977386e-01 -5.06920934e-01 5.83299756e-01 -2.55222648e-01 1.11031008e+00 1.04131413e+00 2.07594976e-01 8.09095144e-01 -1.47044152e-01 1.11229956e+00 7.09699869e-01 -4.90490466e-01 2.02160701e-02 3.39313358e-01 9.62542653e-01 2.94444412e-01 -2.76990831e-01 -4.99259889e-01 -6.87685907e-01 -5.40523231e-01 -1.09526463e-01 2.67679453e-01 -6.20241463e-01 2.49278143e-01 -7.45405018e-01 9.32554781e-01 3.85166287e-01 3.50488782e-01 -6.48138523e-01 -3.83277923e-01 9.52028632e-02 4.33295190e-01 2.47144207e-01 1.05712700e+00 -6.61647439e-01 -4.96975511e-01 -5.16351104e-01 3.48271459e-01 1.26216638e+00 7.67490447e-01 9.16365802e-01 -1.06404185e+00 -8.62525403e-01 1.37624836e+00 3.61701727e-01 1.99122474e-01 5.06011903e-01 -1.11913049e+00 4.76844341e-01 8.26632977e-01 6.00300074e-01 -8.53767276e-01 -2.73864716e-01 -2.91396677e-01 -4.61493850e-01 -5.64889073e-01 7.74117410e-02 -1.64255619e-01 -6.16583645e-01 1.36562777e+00 2.63435572e-01 -5.21274924e-01 -1.48124236e-03 8.97417843e-01 9.16537046e-01 6.86197340e-01 -1.46915577e-02 -3.65475208e-01 1.81066990e+00 -1.17454660e+00 -1.09082973e+00 -5.81044495e-01 7.49162853e-01 -1.05757821e+00 1.59929371e+00 9.41010416e-02 -8.35175693e-01 -3.67178559e-01 -4.85390097e-01 -4.32948768e-01 -1.13364309e-01 2.84899652e-01 5.29554605e-01 5.27257502e-01 -1.07211947e+00 -4.90093306e-02 -2.94168353e-01 -5.07040858e-01 -2.99657911e-01 1.85211137e-01 3.65462154e-01 -1.26506880e-01 -1.52251863e+00 7.78648734e-01 -3.22625697e-01 1.83477789e-01 -3.06958050e-01 -7.87678003e-01 -5.94735563e-01 4.93091166e-01 6.68430746e-01 -7.83045411e-01 2.14487100e+00 -3.24648559e-01 -1.41387653e+00 4.39990193e-01 -1.02656376e+00 -1.55518234e-01 1.33160651e-01 -6.87053561e-01 -1.18794046e-01 2.32105717e-01 3.58306259e-01 6.01865649e-01 5.65637589e-01 -1.00723112e+00 -8.93271446e-01 -1.66729867e-01 5.42985141e-01 7.31572628e-01 -3.43970507e-01 1.07964553e-01 -8.44292700e-01 -1.33841246e-01 1.38576612e-01 -1.10132694e+00 -1.41055837e-01 -4.72073793e-01 -3.15532774e-01 -9.24638450e-01 9.66090679e-01 -9.03009117e-01 1.72070992e+00 -1.83941507e+00 -2.96376556e-01 -4.56669666e-02 2.62056917e-01 3.48260105e-01 -2.90861577e-02 9.19494152e-01 3.65460604e-01 2.84093291e-01 1.93816081e-01 -3.53612989e-01 6.54917359e-02 -1.84790909e-01 -4.70451653e-01 -3.76571804e-01 -2.18773976e-01 1.07358980e+00 -8.19908381e-01 -4.33012396e-01 -1.22544006e-01 6.27494752e-02 -6.76759899e-01 7.63236761e-01 -3.13300610e-01 8.66235793e-02 -5.36291003e-01 3.55995446e-01 2.96998233e-01 -8.18038404e-01 -3.37175205e-02 -9.58573148e-02 7.56099597e-02 1.12157702e+00 -4.91376579e-01 1.24471486e+00 -8.59867752e-01 6.67355776e-01 2.86756068e-01 -4.26506758e-01 4.35113966e-01 2.65031874e-01 1.71366319e-01 -9.24856186e-01 -2.21285313e-01 -3.87964360e-02 9.49129090e-03 -7.05282331e-01 6.70409739e-01 3.79319400e-01 -7.87093341e-02 9.03587937e-01 -4.92734015e-01 -4.67108563e-02 3.31166267e-01 8.26318085e-01 1.33362067e+00 -6.10031486e-01 3.45998973e-01 -1.01403542e-01 9.98424888e-01 -1.77403137e-01 7.11308718e-02 1.26425517e+00 -2.65981793e-01 3.13013703e-01 3.03042382e-01 1.78335682e-02 -1.84232980e-01 -5.73791564e-01 3.98867637e-01 1.74193788e+00 1.74351543e-01 -6.95899665e-01 -5.39964616e-01 -9.67219472e-01 -2.17232421e-01 1.21437144e+00 -4.35849786e-01 -2.32754916e-01 -6.04378998e-01 -2.49055490e-01 -1.26333803e-01 1.12768173e-01 6.84739590e-01 -1.03831291e+00 -4.17280108e-01 8.13709050e-02 -9.84749913e-01 -8.49765420e-01 -1.27226985e+00 -7.71069601e-02 -4.35039341e-01 -8.96820068e-01 -7.89841235e-01 -7.62326360e-01 3.93996954e-01 9.96748626e-01 1.14697242e+00 5.83060861e-01 1.11028507e-01 7.98177063e-01 -6.43369853e-01 -2.45699380e-02 -3.58361714e-02 4.79051828e-01 -4.39407200e-01 -2.06380524e-02 6.05278850e-01 -3.22290957e-01 -1.13843930e+00 8.31128538e-01 -6.34493172e-01 9.48422477e-02 5.31015158e-01 8.56652856e-01 -2.38031879e-01 -4.77219462e-01 8.86681974e-01 -8.94681036e-01 1.58035457e+00 -4.76173848e-01 -9.18052495e-02 6.81533933e-01 -9.28659439e-01 1.78364307e-01 3.66527706e-01 -2.74894804e-01 -1.31184542e+00 -4.91627961e-01 -4.88408417e-01 3.86723578e-01 1.47144988e-01 6.38954282e-01 1.04364485e-01 2.38640472e-01 7.32701480e-01 3.49671990e-01 -1.87803745e-01 -5.38495719e-01 2.08822593e-01 1.25923479e+00 1.50538504e-01 -9.18761864e-02 3.46315742e-01 -4.44738828e-02 -7.87165880e-01 -9.00288165e-01 -1.37240183e+00 -1.44031274e+00 -9.51540843e-02 -3.70861113e-01 6.23948455e-01 -5.35555720e-01 -1.33509505e+00 1.91483181e-02 -1.35400486e+00 -1.91968456e-01 2.91161150e-01 8.25922266e-02 -2.36839637e-01 5.81380010e-01 -6.27108693e-01 -1.22877550e+00 -7.38061666e-01 -1.06564581e+00 9.79930878e-01 3.93989176e-01 -8.54223728e-01 -8.00108552e-01 -1.74556132e-02 1.06718194e+00 6.94666386e-01 -1.05929923e+00 1.07882428e+00 -9.96573448e-01 -5.27647436e-01 -1.50314808e-01 -1.94034815e-01 -1.71790551e-02 5.15582800e-01 -7.08970070e-01 -7.51863360e-01 -6.76826015e-02 2.74809688e-01 -2.47641176e-01 8.41743350e-01 -3.07899024e-02 1.04298270e+00 -9.08388138e-01 -4.26874638e-01 -2.66478926e-01 5.85352719e-01 4.71669793e-01 3.40854973e-01 1.63163304e-01 2.18082234e-01 9.03369188e-01 8.51689339e-01 2.79886752e-01 9.47434783e-01 8.83671284e-01 3.75088006e-02 2.04858139e-01 5.04320255e-03 -2.98904002e-01 1.71481863e-01 5.90760171e-01 7.19735801e-01 -5.45635819e-01 -8.08709204e-01 4.97453600e-01 -1.90523911e+00 -8.36997747e-01 -8.63770545e-02 2.03642511e+00 1.09126675e+00 -1.35773178e-02 -6.68494031e-02 -7.41852671e-02 6.07192278e-01 2.31542110e-01 -6.59264624e-01 -1.96400493e-01 4.77284074e-01 -2.84119785e-01 -2.29215384e-01 1.19884706e+00 -6.59885585e-01 7.82000899e-01 5.76023054e+00 4.73887444e-01 -7.00422287e-01 -3.11583956e-03 7.82420039e-01 9.36272442e-02 -6.20689332e-01 1.90564767e-01 -1.07360244e+00 4.36204523e-01 7.91423798e-01 -2.98767298e-01 5.29042840e-01 8.43035519e-01 3.88239861e-01 -6.00587666e-01 -1.12502897e+00 1.01354635e+00 2.64036506e-01 -1.05811012e+00 1.18575394e-01 -2.78699130e-01 2.85201788e-01 -3.56579214e-01 9.43581574e-03 8.16335261e-01 2.93325454e-01 -5.82958221e-01 -1.69143155e-01 4.65556353e-01 1.31950185e-01 -4.17716086e-01 5.93871236e-01 9.82943892e-01 -8.90468895e-01 -2.28510395e-01 4.85967584e-02 -1.39767870e-01 4.26253855e-01 3.24596763e-01 -1.54213691e+00 -3.33047993e-02 6.06443524e-01 2.37037629e-01 -6.53666735e-01 7.95067489e-01 -1.89782917e-01 6.65010393e-01 -1.35517821e-01 -7.83023596e-01 1.92027345e-01 -9.93833132e-03 4.26068515e-01 1.05112910e+00 1.67969540e-01 6.19814694e-01 -6.71400083e-03 5.53029358e-01 -3.34679455e-01 3.27738017e-01 -2.49049589e-02 -1.02503657e-01 5.15518188e-01 1.43138158e+00 -2.23771408e-01 -4.40988451e-01 -3.03335994e-01 1.26812053e+00 3.72296214e-01 5.27261972e-01 -1.87499568e-01 -5.47107875e-01 5.02918720e-01 1.71647951e-01 -1.02448471e-01 -3.71164046e-02 -4.15181406e-02 -1.01622593e+00 2.61000842e-01 -1.07241631e+00 6.28495693e-01 -9.66239870e-01 -1.02677286e+00 4.43796635e-01 -2.07163915e-01 -7.02346921e-01 -8.06620181e-01 7.72797642e-03 -6.11033320e-01 1.15350485e+00 -1.55166876e+00 -4.13000047e-01 -4.23358560e-01 2.01643035e-01 1.14873743e+00 3.07943702e-01 6.42933965e-01 2.66768128e-01 -2.36073002e-01 4.85413939e-01 -4.06273931e-01 -1.77886069e-01 9.01598811e-01 -1.21809757e+00 4.05584872e-01 5.18922806e-01 -4.67409156e-02 1.15159488e+00 7.26955652e-01 -7.09570825e-01 -1.38165164e+00 -4.60806400e-01 1.64040697e+00 -5.75582743e-01 2.09356457e-01 -3.96548390e-01 -1.21968460e+00 5.17468274e-01 5.93461215e-01 -7.95742095e-01 9.76234674e-01 5.89302719e-01 -4.02459800e-02 3.71113094e-03 -7.47671008e-01 9.93528605e-01 7.46497393e-01 -7.46993184e-01 -9.03490424e-01 7.18800247e-01 1.00905359e+00 -1.43072292e-01 -2.54092157e-01 3.07858288e-01 4.02990848e-01 -8.24246347e-01 8.76882374e-01 -5.07885873e-01 3.14015709e-02 5.11436276e-02 1.56046584e-01 -1.06196427e+00 -2.45398685e-01 -1.14744318e+00 -1.76244751e-01 1.07259011e+00 7.77583599e-01 -5.95402241e-01 7.77702093e-01 1.35734618e+00 2.90200245e-02 -7.81624019e-01 -6.57481790e-01 -1.00084364e-01 -6.65225327e-01 -2.67369628e-01 4.24825877e-01 4.69308019e-01 5.41269898e-01 1.18472362e+00 -3.31393570e-01 2.64090225e-02 3.45361978e-02 2.37981975e-01 7.98786283e-01 -1.08055305e+00 -1.73854321e-01 -4.33424771e-01 7.89019942e-01 -2.22256613e+00 4.82407250e-02 -5.10078609e-01 3.44449133e-01 -1.87335181e+00 3.48395556e-01 -1.61151111e-01 -3.22753824e-02 2.73173720e-01 -8.95804226e-01 -5.03936052e-01 5.69342859e-02 5.16780853e-01 -1.16597366e+00 7.06754386e-01 1.03869903e+00 -7.61548057e-02 -6.10351086e-01 6.71667516e-01 -1.25333977e+00 4.24775243e-01 5.57448566e-01 -1.81395844e-01 -6.51788592e-01 -7.75637329e-02 5.85755646e-01 5.88503599e-01 -9.06095728e-02 -4.01339859e-01 7.71182537e-01 -2.80642420e-01 -1.70999557e-01 -1.10136473e+00 5.96474528e-01 -6.84246838e-01 -5.85902452e-01 3.88478339e-01 -1.12028360e+00 2.90810794e-01 -1.72387823e-01 4.73924965e-01 -2.23067611e-01 -6.24290705e-01 2.11395383e-01 -3.88489924e-02 -4.57209915e-01 7.47643635e-02 -8.60279739e-01 2.16470882e-01 4.09694761e-01 2.65228212e-01 -3.26556981e-01 -1.35499418e+00 -3.69408309e-01 7.94704258e-01 -7.83253163e-02 7.42361307e-01 6.71108723e-01 -9.80766773e-01 -4.05418485e-01 -1.93457231e-01 2.79384553e-01 -3.96689653e-01 3.42113554e-01 6.98244512e-01 1.37081981e-01 1.23171735e+00 6.49865627e-01 -4.23983634e-01 -1.68284965e+00 2.25467354e-01 2.45174840e-01 -6.41047478e-01 1.08437855e-02 1.02897584e+00 3.82239699e-01 -5.52521884e-01 4.35058773e-01 -1.65702254e-01 -7.56673932e-01 2.04771832e-01 9.30904627e-01 3.16212654e-01 2.69557029e-01 -2.31561549e-02 -3.01485777e-01 -5.33764139e-02 -7.34613359e-01 -5.47870636e-01 7.09896982e-01 -8.03393602e-01 -1.53576985e-01 2.27673903e-01 1.32928050e+00 1.07472152e-01 -7.23236382e-01 -5.71500003e-01 4.35532689e-01 -5.39796472e-01 -8.02267119e-02 -1.28864479e+00 -1.80936128e-01 6.65365458e-01 1.91365302e-01 6.94555223e-01 1.08836329e+00 1.34686306e-01 1.00639939e+00 1.08172488e+00 -1.20555043e-01 -9.61659074e-01 5.76618552e-01 9.02524710e-01 1.23982704e+00 -1.50395703e+00 -3.03230733e-01 -4.89646107e-01 -7.66462803e-01 8.70800793e-01 8.41450155e-01 6.85467184e-01 5.48424006e-01 -4.19784755e-01 3.01592320e-01 -2.90194482e-01 -1.26814318e+00 -2.69791365e-01 7.62489855e-01 -2.69417954e-03 7.57151723e-01 -4.05399799e-01 -6.76759183e-01 5.14399588e-01 -2.89597183e-01 -3.15238655e-01 2.71561712e-01 9.94749308e-01 -7.04047978e-01 -1.12013316e+00 -6.98627457e-02 8.59979272e-01 -4.33508545e-01 -5.51313698e-01 -9.07880127e-01 9.06743258e-02 -1.04815769e+00 1.81934619e+00 -2.32322350e-01 -3.19071740e-01 3.58557731e-01 3.33209753e-01 -2.98510790e-01 -8.11430752e-01 -9.62735295e-01 -7.58415088e-02 4.92570430e-01 -5.70595622e-01 -1.48929372e-01 -6.68548465e-01 -9.07638073e-01 8.48364159e-02 -5.98192573e-01 8.70469809e-01 4.40516531e-01 1.05094624e+00 8.52482080e-01 1.81878313e-01 8.05276334e-01 -1.94880202e-01 -8.68904591e-01 -1.30015957e+00 2.25134313e-01 3.51341039e-01 7.15520501e-01 -2.21927449e-01 -7.03007638e-01 -3.23162645e-01]
[12.180259704589844, 7.797821044921875]
949b80fd-3bab-49e3-b1c2-243eb64f39e7
correlation-clustering-of-bird-sounds
2306.09906
null
https://arxiv.org/abs/2306.09906v1
https://arxiv.org/pdf/2306.09906v1.pdf
Correlation Clustering of Bird Sounds
Bird sound classification is the task of relating any sound recording to those species of bird that can be heard in the recording. Here, we study bird sound clustering, the task of deciding for any pair of sound recordings whether the same species of bird can be heard in both. We address this problem by first learning, from a training set, probabilities of pairs of recordings being related in this way, and then inferring a maximally probable partition of a test set by correlation clustering. We address the following questions: How accurate is this clustering, compared to a classification of the test set? How do the clusters thus inferred relate to the clusters obtained by classification? How accurate is this clustering when applied to recordings of bird species not heard during training? How effective is this clustering in separating, from bird sounds, environmental noise not heard during training?
['Bjoern Andres', 'David Stein']
2023-06-16
null
null
null
null
['sound-classification', 'clustering', 'classification-1']
['audio', 'methodology', 'methodology']
[ 1.96656972e-01 -3.72869551e-01 7.61828959e-01 -2.56871074e-01 -3.86110187e-01 -1.01282346e+00 1.75706983e-01 4.06329185e-01 -4.88400519e-01 3.41016561e-01 8.92041922e-02 -3.30115020e-01 -3.61919105e-01 -7.95184374e-01 -5.36528528e-01 -7.59290159e-01 -5.75138390e-01 4.77566540e-01 4.15497273e-01 1.71746179e-01 1.58665136e-01 2.65722215e-01 -2.05518389e+00 1.39684394e-01 3.99354130e-01 5.14766276e-01 3.80583644e-01 1.24757242e+00 1.21337526e-01 6.64599597e-01 -1.14959192e+00 -3.39553617e-02 3.68786156e-01 -1.10296106e+00 -9.72652495e-01 1.19784057e-01 5.02976894e-01 2.43297204e-01 3.69892031e-01 1.14406300e+00 9.06982049e-02 6.09113872e-01 9.47677433e-01 -1.04470682e+00 -1.39640912e-01 8.49494040e-01 -2.76237637e-01 5.18820584e-01 4.55757469e-01 -2.37230167e-01 1.15018940e+00 -1.64062828e-01 -4.74792384e-02 1.00972593e+00 7.38623083e-01 4.21758533e-01 -1.71043634e+00 -8.22913110e-01 -9.58168181e-04 5.72915711e-02 -1.62640107e+00 -3.25919300e-01 6.65047944e-01 -8.64585698e-01 6.16960943e-01 7.29846179e-01 8.67576480e-01 2.42286727e-01 -1.10929430e-01 -7.45471492e-02 1.04416561e+00 -4.49532002e-01 5.94716966e-01 2.72352457e-01 1.93749443e-01 3.54708582e-01 -1.58519924e-01 2.70847589e-01 -4.74324524e-01 -3.06743771e-01 7.93031752e-02 -3.25549930e-01 -5.76289237e-01 3.90147716e-01 -1.15056348e+00 6.27390563e-01 6.09062016e-01 7.56157815e-01 -2.13429183e-01 3.84351648e-02 1.39687108e-02 2.20892370e-01 2.60993958e-01 9.75386262e-01 -3.56521368e-01 1.25218719e-01 -1.11401296e+00 -2.85088979e-02 1.22976720e+00 5.64386547e-01 8.68857801e-01 -6.65619448e-02 5.71967542e-01 8.48365664e-01 2.14546919e-01 5.17229259e-01 3.43066335e-01 -7.86012232e-01 -1.29877672e-01 1.36291429e-01 -2.59911746e-01 -1.10734057e+00 -3.29115570e-01 -3.86229217e-01 -5.76260865e-01 1.11706415e-02 3.85320693e-01 -2.21053094e-01 -3.56343150e-01 1.98559749e+00 2.10502908e-01 7.24338412e-01 1.42893985e-01 6.46733642e-01 8.47862720e-01 9.24909830e-01 4.19896841e-03 -5.06836653e-01 1.18129647e+00 -5.54976583e-01 -9.01692733e-03 -2.46219218e-01 2.51739323e-01 -8.54676306e-01 6.51531935e-01 4.48166728e-01 -4.24203426e-01 -1.05615163e+00 -1.00847137e+00 6.38399780e-01 -1.49089426e-01 -5.09131104e-02 9.95768085e-02 7.21521437e-01 -1.07168162e+00 5.40568888e-01 -5.95328331e-01 -4.97104406e-01 -3.77439767e-01 4.07782942e-01 -2.96390444e-01 4.39444840e-01 -7.78990328e-01 5.49160004e-01 2.60489196e-01 9.49676260e-02 -9.84549999e-01 -5.25927246e-01 -3.81323248e-01 1.18190281e-01 -7.73906931e-02 -4.12415057e-01 1.33557439e+00 -1.09336782e+00 -8.41034055e-01 6.17197037e-01 -1.14949517e-01 -2.65289247e-01 -1.14836775e-01 3.37455451e-01 -7.79568195e-01 1.51028901e-01 2.45258749e-01 4.06656921e-01 7.07500756e-01 -1.51634622e+00 -1.03772831e+00 -4.57767546e-01 -2.53934652e-01 6.51667044e-02 -7.52554322e-03 9.05567333e-02 1.61149669e-02 -2.35427573e-01 2.12619320e-01 -1.18597353e+00 -1.88567907e-01 -2.47546509e-01 -2.92007267e-01 -2.34400809e-01 3.93962801e-01 -6.41900420e-01 1.23516548e+00 -2.67550468e+00 -2.81512700e-02 4.01555806e-01 1.62633523e-01 -1.04496554e-01 -1.45366728e-01 2.74943709e-01 -4.05031741e-02 1.97906315e-01 -4.91580546e-01 -3.03179375e-03 -2.88435519e-01 3.37492883e-01 -4.67592716e-01 3.53023112e-01 -1.13101937e-01 -2.34211937e-01 -1.14451861e+00 -5.41219592e-01 3.28732841e-02 2.21490815e-01 -6.81600094e-01 7.96493173e-01 -1.73482695e-03 4.40355331e-01 1.66314077e-02 -1.59173146e-01 4.03011680e-01 6.44182622e-01 1.94656506e-01 8.71511549e-02 -2.21303523e-01 3.03902566e-01 -1.31117225e+00 8.10463130e-01 -4.53557342e-01 8.38397682e-01 1.32524073e-02 -1.01193774e+00 1.23732936e+00 3.55173767e-01 1.87659591e-01 6.15577139e-02 2.06772517e-02 7.71410344e-03 6.76489115e-01 -5.03494263e-01 2.19297528e-01 -7.41073728e-01 -1.55107453e-01 6.05789661e-01 3.11088562e-01 -5.14901340e-01 1.18745230e-01 -2.06721067e-01 9.12526309e-01 -4.87952352e-01 1.30686387e-01 -3.06219310e-01 1.59327284e-01 -1.96259618e-01 5.46120524e-01 8.76325130e-01 -9.50349346e-02 3.68732512e-01 3.00855547e-01 -1.00006744e-01 -5.02185345e-01 -1.36732507e+00 -4.39982414e-01 1.28058934e+00 2.06698135e-01 -5.28431356e-01 -8.35811377e-01 -4.02062982e-01 -1.24808691e-01 8.19865584e-01 -7.23873377e-01 -2.03186125e-01 -7.15095699e-02 -7.58549511e-01 6.55680835e-01 2.65407950e-01 2.27107674e-01 -1.12590289e+00 -6.94508731e-01 1.65133789e-01 -2.08113685e-01 -5.99738240e-01 -4.39713389e-01 8.30708146e-01 -5.56006789e-01 -1.10902774e+00 -1.94698907e-02 -1.17558193e+00 4.89616394e-01 3.89481694e-01 1.25771308e+00 3.38618487e-01 -1.93759739e-01 4.17480916e-01 -4.66937602e-01 -4.23274070e-01 -8.28689039e-01 -2.90857345e-01 5.64956844e-01 1.43415332e-01 4.11756486e-01 -9.00993049e-01 -1.23540647e-01 7.33193099e-01 -7.25438058e-01 -4.65651453e-01 1.74540848e-01 4.18262064e-01 4.73419189e-01 1.15687180e+00 3.31324160e-01 -6.08465374e-01 3.15390378e-01 -6.04452968e-01 -1.95779920e-01 3.03691030e-01 1.81224898e-01 -1.70555711e-01 1.02807105e+00 -5.88666379e-01 -4.38440859e-01 1.90352723e-01 -2.73001641e-01 -3.29799242e-02 -1.01784730e+00 2.38735661e-01 -1.13865614e-01 2.36476094e-01 9.55788970e-01 2.29914144e-01 -3.90332520e-01 -4.93117929e-01 2.00252950e-01 9.90238070e-01 9.22112405e-01 -4.28945601e-01 7.83676267e-01 -1.30651608e-01 -1.95858836e-01 -1.52345181e+00 -7.05115557e-01 -7.96245456e-01 -9.79116976e-01 -4.71798182e-01 1.11377001e+00 -4.37466592e-01 -5.94612837e-01 3.21530588e-02 -8.00767839e-01 -4.79926132e-02 -2.40953922e-01 8.23284447e-01 -3.32936913e-01 1.10633008e-01 2.12138653e-01 -1.24481547e+00 1.90946341e-01 -8.49216342e-01 5.74628413e-01 2.66544700e-01 -5.96846700e-01 -9.06634212e-01 5.78371108e-01 1.53229669e-01 -2.00523749e-01 2.35292856e-02 9.07298625e-01 -1.18812692e+00 6.24194890e-02 -1.34214714e-01 5.11811316e-01 5.30553341e-01 5.12020648e-01 4.73791271e-01 -1.11715162e+00 -4.01364237e-01 3.93558562e-01 -1.44911679e-02 7.64249206e-01 2.07972378e-01 6.99124813e-01 -5.45503318e-01 -2.92688131e-01 4.89312828e-01 1.16858053e+00 7.82016695e-01 -5.37854955e-02 -3.91652614e-01 5.30759394e-01 1.01821971e+00 2.84627706e-01 -2.50744134e-01 7.93454424e-03 6.12236917e-01 -1.10766117e-03 8.23599771e-02 4.59108353e-02 -5.26157796e-01 1.46470979e-01 1.15258324e+00 -4.92954142e-02 -1.98725268e-01 -8.92774940e-01 6.19395852e-01 -1.47392070e+00 -1.29281592e+00 -2.29289964e-01 2.48093510e+00 6.72455549e-01 -2.11764534e-04 6.27023399e-01 6.87726438e-01 1.02472901e+00 -1.47743538e-01 -1.17956668e-01 -5.53055048e-01 3.48850489e-01 2.77346820e-01 -1.48195431e-01 7.68590927e-01 -1.17528439e+00 4.77750659e-01 6.92423677e+00 4.89148349e-01 -9.10388052e-01 -2.76925772e-01 2.60961175e-01 1.14103081e-02 3.70474048e-02 2.12340578e-01 -4.16719824e-01 5.36674201e-01 9.84986842e-01 -2.28353336e-01 7.03383267e-01 4.90013540e-01 -5.67981899e-02 -4.14173365e-01 -1.60230637e+00 7.40150392e-01 1.42173469e-01 -5.91988087e-01 -1.39358610e-01 -1.31338790e-01 4.43897396e-01 -2.34980598e-01 -2.59294063e-01 1.74056575e-01 6.91194713e-01 -1.06911683e+00 8.07422876e-01 3.83595258e-01 2.55362898e-01 -6.62422419e-01 5.13730526e-01 7.99508333e-01 -1.49983490e+00 -7.12836757e-02 -4.18322414e-01 -4.65478659e-01 -3.11831772e-01 2.57978499e-01 -1.21709692e+00 2.75598705e-01 1.13861763e+00 6.16063595e-01 -7.79909790e-01 1.50844729e+00 -3.54968011e-01 1.37955785e+00 -6.52245402e-01 -3.58160198e-01 -1.07118420e-01 -2.05864459e-01 7.08480358e-01 1.40744925e+00 2.37016127e-01 1.01839572e-01 4.32020068e-01 8.65506649e-01 3.92896801e-01 -1.08767383e-01 -6.38839960e-01 1.38725504e-01 7.97138453e-01 1.05348051e+00 -1.16265488e+00 -1.16025053e-01 1.81456935e-02 5.72580814e-01 9.13597643e-03 9.72691700e-02 -4.17312801e-01 -4.29643095e-01 6.43907070e-01 -4.93035503e-02 2.78634995e-01 1.17964916e-01 2.10161641e-01 -5.20895183e-01 -3.15003812e-01 -4.43879783e-01 5.37005782e-01 -7.96864986e-01 -1.46999478e+00 1.05891776e+00 3.25972494e-03 -1.56329691e+00 -2.70065397e-01 2.24652495e-02 -8.77940178e-01 7.68944800e-01 -5.49985707e-01 -4.40352708e-01 4.18145284e-02 3.49422812e-01 1.11284994e-01 -1.03242174e-01 1.13156068e+00 -1.26331508e-01 -1.94913983e-01 2.37435237e-01 1.95508063e-01 3.95913184e-01 3.54842484e-01 -1.66954410e+00 2.78447848e-02 7.38871813e-01 1.13470352e+00 3.57746392e-01 9.93553281e-01 -2.38317311e-01 -5.72102368e-01 -1.08324409e+00 7.77535796e-01 -5.62231481e-01 7.50764847e-01 -4.96907800e-01 -9.97349679e-01 3.98528725e-01 -1.40928641e-01 -3.98542464e-01 1.40705979e+00 2.92430937e-01 -5.06410182e-01 -2.22895056e-01 -9.71776307e-01 1.44039303e-01 6.43294215e-01 -8.37295651e-01 -1.17160749e+00 1.11850947e-02 5.68737566e-01 2.94000417e-01 -6.57268584e-01 -6.31920546e-02 4.45473999e-01 -1.11949098e+00 6.65332675e-01 -5.26197672e-01 3.00313801e-01 -1.07087672e+00 -4.13514107e-01 -1.80423641e+00 -6.66636109e-01 -1.14200607e-01 1.03836775e+00 1.57540894e+00 4.90438014e-01 -2.54812002e-01 1.89443409e-01 -5.87091334e-02 8.21590647e-02 2.65185107e-02 -9.99767184e-01 -8.39463830e-01 1.23038992e-01 -5.53835869e-01 3.88391465e-01 9.56856310e-01 5.72909378e-02 7.34060287e-01 -1.46282628e-01 6.39785469e-01 6.23824239e-01 3.94662559e-01 7.93431938e-01 -1.62083292e+00 -7.63638616e-01 -3.32267195e-01 -9.34725821e-01 -6.23062849e-01 1.57747477e-01 -9.60081339e-01 8.89813244e-01 -1.24054265e+00 9.10957903e-03 -3.33331317e-01 3.17470841e-02 2.94535935e-01 -3.37502807e-01 6.19762719e-01 3.63873512e-01 3.40860665e-01 -3.38295490e-01 3.95429693e-03 5.14237821e-01 -6.84661493e-02 -2.70154446e-01 6.36389196e-01 -5.11520743e-01 8.71685624e-01 4.45089579e-01 -8.73428345e-01 -2.29392841e-01 -1.61622524e-01 -8.06682631e-02 4.66421247e-01 3.43749046e-01 -1.55773818e+00 3.00341666e-01 1.32984579e-01 2.84443140e-01 -1.68323264e-01 -1.47531837e-01 -1.14238954e+00 6.58318460e-01 5.11598587e-01 -6.34799659e-01 -1.90548778e-01 2.37493739e-01 6.30176663e-01 -2.95014441e-01 -5.94220459e-01 1.01727152e+00 -8.74735713e-02 -4.95994210e-01 -1.39731288e-01 -1.05198240e+00 1.04708686e-01 8.46126616e-01 -2.31356010e-01 2.57577866e-01 -3.99396747e-01 -1.13956392e+00 -1.87381003e-02 2.66975075e-01 2.70306140e-01 1.50007844e-01 -1.01454794e+00 -7.59020984e-01 3.66080493e-01 1.27793297e-01 -1.33088484e-01 4.60597910e-02 4.74465877e-01 -3.85018736e-01 -1.14726789e-01 6.16030805e-02 -8.01955998e-01 -1.61887002e+00 6.32066607e-01 6.75997019e-01 4.87822831e-01 -2.00048566e-01 1.12670994e+00 4.22303081e-01 -5.04547179e-01 2.88849741e-01 -4.11918044e-01 -3.15033555e-01 1.00374650e-02 5.15763700e-01 1.34810686e-01 -1.46485806e-01 -1.01648366e+00 -4.93946582e-01 8.46467912e-01 4.27271754e-01 -2.88081348e-01 1.21908915e+00 2.31420174e-01 -3.35284770e-01 1.19291008e+00 1.27141869e+00 3.14421356e-01 -7.99766839e-01 1.74546719e-01 -2.71670446e-02 -4.35857803e-01 -9.97671336e-02 -7.69890010e-01 -8.16474080e-01 9.03897941e-01 4.88573194e-01 1.04345894e+00 1.30329323e+00 3.47379059e-01 1.05559610e-01 4.76079494e-01 4.07626837e-01 -5.26384532e-01 -3.13838542e-01 1.75350219e-01 7.01216817e-01 -7.46394813e-01 -3.92840177e-01 -1.31835476e-01 -4.99403715e-01 9.50270295e-01 3.89711857e-01 -3.88652116e-01 1.05618584e+00 3.17278892e-01 1.87403858e-01 -4.79086787e-02 -7.01085448e-01 -3.18035066e-01 5.03678203e-01 9.51718271e-01 3.63550097e-01 4.10995424e-01 3.07420075e-01 4.28142548e-01 -1.18605626e+00 -6.26085341e-01 4.71200824e-01 5.46997786e-01 -7.74780095e-01 -3.84312510e-01 -6.26458108e-01 5.53178370e-01 -6.91701025e-02 1.22142211e-02 -1.14410055e+00 2.54327685e-01 7.13695705e-01 1.49405277e+00 6.67622745e-01 -1.18433797e+00 3.49677593e-01 -4.48216274e-02 3.52959365e-01 -1.09273982e+00 -9.26620603e-01 2.48514295e-01 -6.21162262e-03 2.55943984e-01 -7.25741923e-01 -7.26532161e-01 -1.01071632e+00 -1.53376147e-01 -6.80413961e-01 1.35944366e+00 3.01135749e-01 1.05172467e+00 -5.74583828e-01 3.73960972e-01 1.17724311e+00 -7.19292581e-01 -7.94262588e-02 -8.00377369e-01 -1.22153699e+00 2.70753860e-01 6.88051522e-01 -1.96699217e-01 -1.12158227e+00 3.88739258e-01]
[15.383846282958984, 5.411357879638672]
6db66f7f-b5d5-47d3-86a3-e7ecc7f85493
why-is-ai-hard-and-physics-simple
2104.00008
null
https://arxiv.org/abs/2104.00008v1
https://arxiv.org/pdf/2104.00008v1.pdf
Why is AI hard and Physics simple?
We discuss why AI is hard and why physics is simple. We discuss how physical intuition and the approach of theoretical physics can be brought to bear on the field of artificial intelligence and specifically machine learning. We suggest that the underlying project of machine learning and the underlying project of physics are strongly coupled through the principle of sparsity, and we call upon theoretical physicists to work on AI as physicists. As a first step in that direction, we discuss an upcoming book on the principles of deep learning theory that attempts to realize this approach.
['Daniel A. Roberts']
2021-03-31
null
null
null
null
['physical-intuition']
['reasoning']
[ 3.36252674e-02 6.43436968e-01 -3.57641250e-01 -5.08573115e-01 2.05351904e-01 -2.76913196e-01 7.34797657e-01 -9.38277096e-02 -1.90950319e-01 5.09305894e-01 1.35090277e-01 -6.12520576e-01 -3.98825318e-01 -1.10408759e+00 -7.07161486e-01 -7.27467120e-01 -5.96397407e-02 3.45103025e-01 -7.20834509e-02 -4.17869210e-01 4.71775264e-01 5.43367982e-01 -1.16382194e+00 2.96516985e-01 4.77491647e-01 6.70815647e-01 4.81423289e-02 5.44478476e-01 -3.08866143e-01 1.15942585e+00 5.94710372e-02 -4.96969789e-01 4.29996759e-01 -3.96678716e-01 -1.12501657e+00 8.54028109e-03 2.08005294e-01 1.80802241e-01 -7.40964949e-01 1.09904397e+00 -3.38986926e-02 1.67408064e-01 6.28935635e-01 -1.04293728e+00 -1.04446471e+00 7.34541774e-01 -4.49938953e-01 4.66888607e-01 -2.38659397e-01 3.19086075e-01 1.36114192e+00 -6.25067472e-01 2.17821777e-01 1.36294556e+00 6.83161616e-01 6.98236108e-01 -1.17323983e+00 -1.43086344e-01 9.68903974e-02 4.02037114e-01 -1.00951242e+00 -3.54111254e-01 9.95344102e-01 -5.95933557e-01 9.18912768e-01 3.82774845e-02 7.86142647e-01 7.03339815e-01 5.16196132e-01 8.85632694e-01 1.22757292e+00 -9.66271877e-01 4.44462776e-01 3.25912863e-01 6.35740519e-01 7.80205309e-01 4.33051229e-01 6.83082104e-01 -6.22309625e-01 1.12829223e-01 8.36177945e-01 -1.86299235e-01 1.30363822e-01 -2.54851013e-01 -1.02024221e+00 1.32023287e+00 6.42647088e-01 7.56946206e-01 -3.42117846e-01 4.83145952e-01 1.52800918e-01 1.83693320e-01 -3.17862928e-02 8.18911612e-01 -6.07953131e-01 2.72277117e-01 -4.95166838e-01 1.64845675e-01 1.01297057e+00 5.47337592e-01 6.54376447e-01 2.57110029e-01 5.21757126e-01 4.29869056e-01 5.38576543e-01 2.19825715e-01 1.23760544e-01 -1.35457170e+00 -1.97206393e-01 3.90287153e-02 -4.51500341e-02 -9.13125217e-01 -4.74567145e-01 -4.53638852e-01 -7.11202085e-01 3.84164661e-01 1.68882564e-01 -3.39900196e-01 -7.06022918e-01 1.62149858e+00 -1.66248679e-01 1.13137752e-01 2.80437708e-01 8.36334109e-01 9.54855204e-01 8.19416702e-01 6.41608909e-02 -1.85861051e-01 1.20427537e+00 -8.61867368e-01 -5.72505474e-01 -5.14700472e-01 6.32433653e-01 -4.85484451e-01 8.49057257e-01 6.00665569e-01 -1.31379616e+00 -6.36095285e-01 -9.60863471e-01 -1.48199856e-01 -3.66735846e-01 -2.00136378e-01 1.62007546e+00 6.71627164e-01 -1.02332067e+00 9.18023586e-01 -1.09570158e+00 -5.52251160e-01 2.31003642e-01 2.24231809e-01 -3.17815579e-02 3.71163607e-01 -1.23366117e+00 1.41902685e+00 6.75543487e-01 -7.15027703e-03 -2.56346732e-01 -7.33177185e-01 -6.00292206e-01 -5.18021435e-02 1.64181069e-01 -1.03993642e+00 1.42963922e+00 -9.51375306e-01 -1.54297268e+00 1.05268121e+00 -2.27787942e-01 -6.25558197e-01 -2.57525086e-01 -2.61436880e-01 -4.17172879e-01 2.14470048e-02 -3.98874074e-01 5.89656770e-01 4.83433127e-01 -1.43559492e+00 -5.59370697e-01 -3.10786486e-01 3.32282394e-01 -4.80073169e-02 -3.07338923e-01 -2.09907144e-01 1.85121804e-01 -2.59772509e-01 1.63799033e-01 -9.41572130e-01 -4.21189934e-01 -2.72199124e-01 -2.22666353e-01 -7.10439622e-01 2.58574486e-01 1.81898460e-01 6.72510684e-01 -1.94693756e+00 3.33419949e-01 1.49637997e-01 5.82072139e-01 1.06772482e-01 -2.84135016e-03 6.59686506e-01 -3.52923930e-01 2.04560354e-01 1.06586754e-01 3.19189221e-01 2.89715797e-01 4.87096816e-01 -5.13623416e-01 4.69692945e-01 7.19598830e-02 1.25845230e+00 -8.47289979e-01 -7.64948055e-02 2.96545029e-01 3.12556237e-01 -5.67930341e-01 -2.11085483e-01 -2.22851679e-01 3.22650701e-01 -7.77008057e-01 1.73109263e-01 5.13943613e-01 -6.75434351e-01 2.33678281e-01 -3.15037698e-01 -3.43249649e-01 8.88944983e-01 -7.88687825e-01 1.44031441e+00 -1.54567540e-01 8.29022467e-01 1.04011625e-01 -1.81239271e+00 5.42266488e-01 2.44714379e-01 5.69235623e-01 -5.97088218e-01 2.44388163e-01 1.80639904e-02 6.87624514e-01 -5.55975199e-01 -1.53522611e-01 -7.10661292e-01 2.73796827e-01 7.65238345e-01 1.18563265e-01 -5.25662184e-01 -2.56888792e-02 2.36212552e-01 9.40025330e-01 -7.80082121e-02 2.27888167e-01 -7.57546604e-01 3.48508567e-01 3.05791348e-01 1.86992049e-01 1.02606392e+00 -3.89641374e-02 -2.59109825e-01 7.04866648e-02 -1.28606629e+00 -1.30664647e+00 -1.03704739e+00 -3.90044034e-01 9.28348958e-01 -1.51188448e-01 -4.92566139e-01 -5.40512145e-01 -1.99494079e-01 -4.22288030e-02 1.02439559e+00 -6.28904045e-01 -2.57849097e-01 -5.42361617e-01 -1.13858581e+00 6.62241504e-02 4.75172162e-01 3.04963291e-01 -1.31505156e+00 -5.71196795e-01 1.60256829e-02 3.92138928e-01 -1.04017842e+00 7.27771759e-01 5.32049596e-01 -9.30122674e-01 -6.36673033e-01 -1.40094012e-01 -6.90682590e-01 2.65421152e-01 2.97083050e-01 1.49363804e+00 3.57821018e-01 -1.86558127e-01 6.70849562e-01 -1.73459724e-01 -8.53964984e-01 -6.22295678e-01 -3.38104591e-02 2.61780322e-01 -5.32350302e-01 1.09407103e+00 -1.10369015e+00 -4.19092894e-01 -4.65451956e-01 -7.92144358e-01 1.51860878e-01 7.31971443e-01 4.30609733e-01 -1.69379398e-01 4.17744815e-01 2.76210099e-01 -7.51035333e-01 4.12711293e-01 -3.67376119e-01 -4.32623237e-01 1.53525546e-01 -4.58673328e-01 2.42719471e-01 4.48466659e-01 6.07727915e-02 -8.42497170e-01 -1.09091513e-01 -2.27869198e-01 1.74622744e-01 -2.80974090e-01 6.40995860e-01 1.47804633e-01 -4.41592216e-01 8.06097269e-01 1.59858286e-01 -2.74266779e-01 -4.24688160e-01 6.36516094e-01 3.29961270e-01 3.85273695e-01 -7.50850141e-01 9.63622928e-01 6.79089725e-01 3.16864789e-01 -1.10010934e+00 -1.32339704e+00 1.25263885e-01 -8.34429562e-01 1.11368358e-01 8.18629682e-01 -7.33709693e-01 -8.50877404e-01 2.23276503e-02 -1.37219214e+00 -2.78549343e-01 -6.10329211e-01 7.98049033e-01 -6.93574965e-01 2.53661543e-01 -6.19516432e-01 -6.85206652e-01 -2.11448912e-02 -6.64491177e-01 4.74234849e-01 4.37124491e-01 -2.52142787e-01 -1.47693527e+00 2.78544843e-01 3.39938343e-01 2.22321615e-01 -4.71725404e-01 1.26727724e+00 -5.86349070e-01 -6.09891653e-01 6.55834675e-02 -1.36185989e-01 3.47076863e-01 5.22080623e-02 1.40949309e-01 -1.07707441e+00 9.47125852e-02 7.25871623e-01 -4.60330456e-01 1.12844718e+00 6.97938025e-01 1.11883783e+00 -1.22894906e-02 -3.06804538e-01 4.18930054e-01 1.71457744e+00 1.18357040e-01 5.32060564e-01 4.25176501e-01 7.06111073e-01 8.39317560e-01 -2.02802002e-01 3.10595065e-01 3.40644121e-01 2.12391242e-01 3.80663015e-02 -1.45883858e-01 -2.01547388e-02 9.66682956e-02 1.15018085e-01 1.19451225e+00 -4.32442844e-01 6.46685436e-02 -1.01878190e+00 4.27887380e-01 -1.78940833e+00 -1.20324886e+00 -4.51419324e-01 1.74669564e+00 8.45951557e-01 4.70678896e-01 -5.85015863e-02 4.86617535e-02 4.72095907e-01 -2.22270247e-02 -2.29204044e-01 -8.65673721e-01 -2.59089321e-02 4.63129669e-01 3.74720961e-01 6.80295289e-01 -9.70652699e-01 1.13027132e+00 8.57319546e+00 5.82927942e-01 -1.11525142e+00 1.11957252e-01 2.85240501e-01 2.39386812e-01 -5.06239176e-01 3.37862819e-01 -7.20962226e-01 2.67759442e-01 1.17676067e+00 -1.99872792e-01 6.42258108e-01 7.03709483e-01 2.36418530e-01 -3.17399316e-02 -1.50385237e+00 7.04040289e-01 -4.66384053e-01 -1.62978518e+00 3.73260840e-03 1.66475073e-01 7.29140699e-01 4.54781204e-01 1.16827168e-01 -1.05186757e-02 8.50134909e-01 -1.43742907e+00 2.24487960e-01 4.95556980e-01 -1.63706481e-01 -2.91359365e-01 3.55368555e-01 5.07125556e-01 -6.52381599e-01 -1.27902389e-01 -8.77009928e-01 -1.02141058e+00 -6.54648850e-03 9.98929739e-01 -2.76417047e-01 1.80920705e-01 6.03951514e-01 7.70161092e-01 -9.10272747e-02 7.82557786e-01 -3.72593910e-01 6.56279683e-01 -4.45416689e-01 -1.58967704e-01 4.77971047e-01 -1.22113697e-01 3.15665305e-01 1.13966942e+00 -1.14247039e-01 6.35679603e-01 1.85892597e-01 1.36085021e+00 2.80098975e-01 -3.43536317e-01 -7.26183593e-01 -3.56300086e-01 1.54821262e-01 1.23680544e+00 -7.39608824e-01 -5.01889884e-01 -8.64608288e-01 2.04721615e-01 1.89630479e-01 2.32537970e-01 -5.59744358e-01 -1.56892478e-01 5.40253580e-01 -2.88214058e-01 3.98640037e-01 -3.71188998e-01 -8.87218773e-01 -1.32753217e+00 -3.91037494e-01 -7.58844435e-01 -1.03433378e-01 -6.80779517e-01 -1.78384793e+00 1.02082625e-01 -2.13918805e-01 -5.64373255e-01 -2.45544463e-02 -1.18248153e+00 -1.00224590e+00 6.13879919e-01 -1.46280336e+00 -8.78731012e-01 1.30174205e-01 3.76183957e-01 2.39693388e-01 -1.82750449e-01 9.55132544e-01 -2.04730928e-01 -3.65898103e-01 4.23376784e-02 2.97012091e-01 2.36816272e-01 -1.90142114e-02 -1.24074256e+00 8.47764671e-01 3.59279841e-01 5.80671191e-01 7.21597731e-01 1.07615840e+00 -2.76636869e-01 -1.88420236e+00 -3.67165029e-01 8.01937878e-01 -6.27625048e-01 1.15090430e+00 -2.22313881e-01 -7.67731011e-01 8.16065073e-01 6.13421679e-01 3.27395126e-02 8.33634973e-01 4.74510193e-01 -2.61879236e-01 1.47653475e-01 -7.46341288e-01 3.29583734e-01 8.60942006e-01 -5.10835588e-01 -1.36051881e+00 8.70521843e-01 5.94181359e-01 1.65537164e-01 -8.72565508e-01 4.45293993e-01 6.50104403e-01 -9.79231775e-01 1.07923639e+00 -1.00607109e+00 6.01228058e-01 1.21535957e-01 -2.49452218e-02 -1.22197807e+00 -8.77985835e-01 -7.42295206e-01 -1.88079670e-01 5.18640041e-01 2.72714049e-01 -6.47979975e-01 8.49585831e-01 8.11668158e-01 9.86248851e-02 -5.76717257e-01 -5.09216607e-01 -8.20042551e-01 9.56681013e-01 -5.64594805e-01 -5.00968425e-03 1.09165430e+00 5.64409435e-01 7.40339816e-01 -1.40132174e-01 3.69008556e-02 6.57980740e-01 2.91858256e-01 4.79886591e-01 -1.67885339e+00 -5.95170259e-01 -5.46304584e-01 -3.94638062e-01 -1.01005530e+00 1.35303393e-01 -9.17926490e-01 -1.33771792e-01 -1.38884223e+00 3.86584729e-01 -1.58810973e-01 -5.88128567e-01 2.76177436e-01 3.31075221e-01 1.58205137e-01 2.42800996e-01 2.08445624e-01 -3.23660821e-01 2.65176266e-01 1.20894110e+00 6.25710562e-02 -4.07535173e-02 -1.18618116e-01 -1.19192564e+00 1.36319304e+00 1.02488053e+00 -3.45437378e-01 -3.47228169e-01 -6.40736461e-01 7.18491256e-01 -4.30110872e-01 5.63187182e-01 -9.54236627e-01 3.28710318e-01 -5.75532854e-01 5.20139754e-01 -1.70886323e-01 4.02312785e-01 -7.13443220e-01 -5.42269170e-01 6.68786108e-01 -5.53709626e-01 -2.96090961e-01 1.75081000e-01 9.10947844e-02 1.09994411e-01 -4.79430616e-01 8.57197821e-01 -7.00350642e-01 -7.78292060e-01 1.39522450e-02 -4.92643982e-01 1.01309991e-03 7.34608531e-01 2.00779527e-01 -4.31608617e-01 -3.74957770e-01 -8.39734733e-01 -5.16606532e-02 2.82786816e-01 6.43672720e-02 2.55160451e-01 -9.78450716e-01 -5.22519290e-01 -1.00366682e-01 -5.00608504e-01 -7.48525143e-01 -8.42674077e-02 8.40195596e-01 -2.48661026e-01 1.16627884e+00 -3.49913150e-01 -3.36076796e-01 -5.39253891e-01 8.58127832e-01 5.06607831e-01 2.68364072e-01 -9.72579658e-01 1.04022086e+00 6.70514166e-01 -2.65669227e-01 2.61437088e-01 -1.54651254e-01 -2.64313370e-01 -5.89053690e-01 7.32471704e-01 2.09327769e-02 -2.93686092e-01 -3.10547471e-01 -4.54341203e-01 9.09595191e-01 -1.33898661e-01 -1.23310849e-01 1.68040752e+00 -9.69794020e-03 -5.24379075e-01 7.92308688e-01 8.92625153e-01 -4.06994149e-02 -7.54344165e-01 -3.35057437e-01 1.27670363e-01 -1.84779279e-02 5.69441736e-01 -7.56160855e-01 -9.00024116e-01 1.38020682e+00 3.44691515e-01 9.48203623e-01 7.93149889e-01 4.41367567e-01 8.74365330e-01 9.89631593e-01 1.88425943e-01 -1.18182480e+00 -2.11513802e-01 6.29277408e-01 5.08744061e-01 -1.38195121e+00 3.75068098e-01 -3.83714616e-01 -2.81002492e-01 1.41256917e+00 4.13543195e-01 -7.30270088e-01 1.10240507e+00 3.43185604e-01 -2.69321889e-01 -6.51481569e-01 -1.04877543e+00 -3.51144284e-01 2.58823514e-01 5.55531263e-01 8.36324453e-01 1.90313309e-02 -5.39319098e-01 3.84581000e-01 -6.59338653e-01 3.82242709e-01 4.03958738e-01 9.07010794e-01 -1.07260656e+00 -1.26952457e+00 -2.59154052e-01 1.40599355e-01 -6.04966581e-01 -3.46400708e-01 -5.44103265e-01 5.68537235e-01 4.07820582e-01 8.67386758e-01 4.70549911e-02 -3.75619471e-01 -2.87480474e-01 1.59088448e-01 9.54873383e-01 -6.65919125e-01 -8.74242783e-02 -1.21990785e-01 -1.56897575e-01 -2.08328664e-01 -8.27311993e-01 -3.69176418e-01 -1.32568920e+00 -8.31276953e-01 -9.26170424e-02 5.09066820e-01 6.67328417e-01 1.51902938e+00 1.07482053e-01 4.35692400e-01 4.47531700e-01 -8.71010244e-01 -4.95879859e-01 -7.43429840e-01 -6.35628164e-01 -9.74940732e-02 3.55173290e-01 -5.26639938e-01 -5.02997756e-01 5.80536351e-02]
[9.071990013122559, 6.422181129455566]
067542c5-f290-4536-8a9e-ae5e319fa24a
varifocal-question-generation-for-fact
2210.12400
null
https://arxiv.org/abs/2210.12400v1
https://arxiv.org/pdf/2210.12400v1.pdf
Varifocal Question Generation for Fact-checking
Fact-checking requires retrieving evidence related to a claim under investigation. The task can be formulated as question generation based on a claim, followed by question answering. However, recent question generation approaches assume that the answer is known and typically contained in a passage given as input, whereas such passages are what is being sought when verifying a claim. In this paper, we present {\it Varifocal}, a method that generates questions based on different focal points within a given claim, i.e.\ different spans of the claim and its metadata, such as its source and date. Our method outperforms previous work on a fact-checking question generation dataset on a wide range of automatic evaluation metrics. These results are corroborated by our manual evaluation, which indicates that our method generates more relevant and informative questions. We further demonstrate the potential of focal points in generating sets of clarification questions for product descriptions.
['Andreas Vlachos', 'Zhangdie Yuan', 'Nedjma Ousidhoum']
2022-10-22
null
null
null
null
['question-generation']
['natural-language-processing']
[ 4.06565696e-01 5.72814107e-01 -4.14633512e-01 -6.79321364e-02 -1.52250636e+00 -1.15374231e+00 1.05713463e+00 9.64578569e-01 -4.18560356e-02 9.43519950e-01 7.81842232e-01 -7.39717424e-01 -3.68657529e-01 -7.83705473e-01 -7.90555477e-01 2.34241337e-01 6.89610183e-01 2.84012020e-01 5.36186278e-01 -5.17505825e-01 8.99733305e-01 -3.93385254e-02 -1.46006942e+00 6.43597722e-01 1.02498102e+00 9.41198230e-01 -9.85272750e-02 4.67359930e-01 -4.72128898e-01 1.33832455e+00 -1.13868928e+00 -1.20571053e+00 -2.04987720e-01 -8.45779836e-01 -1.48660088e+00 3.04368466e-01 6.27548039e-01 8.33761320e-03 2.08592311e-01 1.11009669e+00 -9.25735682e-02 -2.72773445e-01 5.83321691e-01 -1.31814981e+00 -9.66860354e-01 8.52076232e-01 -2.36458331e-01 3.44321728e-01 9.86265719e-01 -2.09129751e-02 1.30643332e+00 -5.11155128e-01 1.06402814e+00 7.18769491e-01 7.39099205e-01 1.97556496e-01 -1.02219915e+00 -2.27484390e-01 -1.12471029e-01 3.44625622e-01 -9.24822152e-01 -3.32494020e-01 8.66382658e-01 -7.10176528e-01 5.29877305e-01 5.67783952e-01 2.99061596e-01 7.79753804e-01 1.24950029e-01 6.56575680e-01 1.22162044e+00 -6.25118434e-01 3.29318404e-01 4.10992473e-01 5.21833301e-01 2.81207919e-01 6.09832704e-01 -5.18700480e-01 -4.10658181e-01 -8.02974284e-01 1.92979097e-01 -4.86711800e-01 -2.89894819e-01 2.61888295e-01 -1.08856785e+00 1.18320894e+00 -1.96824282e-01 5.47088563e-01 -8.44073832e-01 -2.10607573e-01 5.42180717e-01 2.60778308e-01 4.29813176e-01 8.57035816e-01 -5.59147596e-01 -7.81174749e-02 -1.20396626e+00 8.48489702e-01 1.41398513e+00 8.73946071e-01 5.97483814e-01 -6.74553275e-01 -8.03354561e-01 4.77405459e-01 8.58423412e-02 2.38443717e-01 1.50986269e-01 -1.17726970e+00 7.27230489e-01 9.84280586e-01 7.40433753e-01 -1.30966401e+00 1.35506824e-01 -3.91056687e-01 -3.96491364e-02 4.93032672e-02 5.56024015e-01 8.76710117e-02 -3.41716915e-01 1.42063069e+00 5.28344452e-01 -2.15744436e-01 2.76188374e-01 4.62360382e-01 9.90527511e-01 2.45908782e-01 1.48868551e-02 -3.17599952e-01 1.88767600e+00 -5.16123772e-01 -1.21754813e+00 1.35688066e-01 4.92614329e-01 -1.08567691e+00 1.03880072e+00 3.14458042e-01 -1.02462542e+00 -2.46508390e-01 -9.97082353e-01 -8.76864120e-02 -3.17186356e-01 1.32933363e-01 3.37106824e-01 6.55553520e-01 -6.55569971e-01 3.45652044e-01 6.60221428e-02 -1.75035551e-01 4.62854594e-01 -5.80202222e-01 -3.84234004e-02 -2.06997450e-02 -1.37097502e+00 6.44218028e-01 4.50790405e-01 -5.46336889e-01 -4.70650524e-01 -8.91830266e-01 -8.57347131e-01 -1.43546388e-01 1.01437306e+00 -7.71598160e-01 1.61574435e+00 -3.33462507e-01 -8.42496991e-01 1.00150347e+00 -2.45264068e-01 -8.20870280e-01 2.42887437e-01 -9.84272659e-02 -8.14896047e-01 6.03652477e-01 8.70756865e-01 1.32790238e-01 7.08896399e-01 -1.38316357e+00 -7.15925157e-01 -8.32972229e-02 7.49885201e-01 -4.87070054e-01 2.72944510e-01 6.64303184e-01 -2.26078480e-02 -7.51666486e-01 2.42098905e-02 -4.83936310e-01 -3.26578738e-03 -3.52970511e-01 -7.88784623e-01 -6.12123013e-01 8.37162435e-01 -1.22350502e+00 1.79837239e+00 -1.43000948e+00 -9.13692236e-01 -3.05835400e-02 1.80506393e-01 2.74503622e-02 2.47269392e-01 8.86834860e-01 -3.19574736e-02 8.28256905e-01 -2.68876731e-01 4.55798268e-01 3.34100693e-01 -1.32853046e-01 -8.21003020e-01 4.32438869e-03 3.77484262e-01 1.07359302e+00 -9.59966898e-01 -8.46241236e-01 -6.58909798e-01 4.74206284e-02 -3.86590540e-01 -3.15881342e-01 -9.32746530e-01 1.16288565e-01 -5.23318946e-01 6.15830302e-01 5.18974245e-01 -5.04925072e-01 -2.29773581e-01 -3.13895494e-01 1.17100189e-02 9.15338397e-01 -1.04556859e+00 1.41887772e+00 -1.06782012e-01 4.71869618e-01 -1.68084562e-01 -3.46164078e-01 9.73642290e-01 5.83030939e-01 1.89964339e-01 -5.23159206e-01 -1.56116456e-01 2.35586807e-01 -4.07929242e-01 -7.08723009e-01 1.00067222e+00 -1.68566898e-01 -4.01434571e-01 1.01189697e+00 -2.73749441e-01 -3.93570423e-01 7.47387350e-01 7.09941685e-01 1.16915762e+00 9.11355093e-02 5.08874238e-01 -1.39148325e-01 6.50366962e-01 7.09181547e-01 2.44052544e-01 9.21031177e-01 9.47220027e-02 5.60888827e-01 9.64990556e-01 -4.04260717e-02 -9.96730208e-01 -5.97910881e-01 -1.70929298e-01 2.71505564e-01 -3.66850570e-02 -8.20373356e-01 -9.23331559e-01 -1.15295959e+00 4.38484177e-02 1.23019278e+00 -9.07789588e-01 2.84944206e-01 -3.54815960e-01 -3.36015932e-02 6.35180056e-01 1.08010359e-01 3.18176746e-01 -1.01851928e+00 -9.11276758e-01 3.39556873e-01 -7.65262187e-01 -1.11388183e+00 -3.46123368e-01 -5.58708668e-01 -4.18651581e-01 -1.75304651e+00 -2.82924265e-01 -3.50032032e-01 2.40075931e-01 1.31494015e-01 1.49151039e+00 4.16328460e-01 1.76229328e-01 3.94815654e-01 -6.26523435e-01 -6.36722982e-01 -9.34030175e-01 -7.64709190e-02 -7.18588233e-01 -8.08150694e-02 2.87789464e-01 -1.54880285e-01 -3.06195140e-01 1.51809335e-01 -1.48772788e+00 -4.51348066e-01 1.02526397e-01 5.62119722e-01 5.85521042e-01 1.98235452e-01 8.42668593e-01 -1.22724402e+00 1.38371873e+00 -8.09176326e-01 -2.49256685e-01 6.06572449e-01 -6.77603900e-01 1.07146747e-01 1.34717032e-01 -2.75340796e-01 -1.08133638e+00 -7.48909235e-01 -9.42263305e-02 2.31087148e-01 -2.32477739e-01 9.92511153e-01 -1.15356416e-01 5.48966467e-01 1.00141501e+00 4.32294002e-03 -6.49906024e-02 -3.10185909e-01 3.89429629e-01 4.18314815e-01 4.97902244e-01 -6.72643840e-01 8.48695517e-01 4.64728296e-01 -2.88149804e-01 -2.17113629e-01 -1.34592330e+00 -5.37424028e-01 -3.30605716e-01 -2.69186437e-01 5.59175849e-01 -4.22638118e-01 -7.38603234e-01 -1.93453133e-01 -1.43965316e+00 2.27531284e-01 -8.09501052e-01 1.90889105e-01 -5.06846309e-01 4.74002540e-01 -3.39522928e-01 -9.39703882e-01 -2.08866358e-01 -7.21714258e-01 1.01607192e+00 -7.46049210e-02 -8.25655639e-01 -9.69918311e-01 1.50825635e-01 9.90811169e-01 2.70205796e-01 7.21282303e-01 1.15599298e+00 -1.07069492e+00 -5.02570271e-01 -4.34828132e-01 -1.03043981e-01 9.99439657e-02 2.67016649e-01 5.06960824e-02 -6.60877049e-01 4.84611511e-01 3.95487607e-01 -4.33950543e-01 5.52563310e-01 3.52333374e-02 7.04642892e-01 -1.17514324e+00 -1.34225488e-01 -5.38666308e-01 1.44718623e+00 -1.99267752e-02 6.98551118e-01 7.15066373e-01 -1.22022936e-02 9.87382174e-01 8.96707237e-01 2.75347769e-01 6.96793139e-01 5.67849338e-01 3.71158928e-01 3.18359911e-01 -2.84670144e-01 -5.05021751e-01 -1.39552578e-01 2.78592438e-01 3.48235399e-01 -1.17619865e-01 -7.57281661e-01 1.00260818e+00 -1.66137779e+00 -1.64975250e+00 -6.10293388e-01 1.79485798e+00 1.12172019e+00 2.76572973e-01 2.74854928e-01 5.36813736e-01 7.21555471e-01 2.20670566e-01 -2.88492411e-01 -1.95840895e-01 -3.48473370e-01 1.58652306e-01 2.47249231e-02 6.09945118e-01 -4.95827556e-01 4.63569045e-01 6.28837872e+00 7.65027046e-01 -3.47129703e-01 3.89660180e-01 4.59589303e-01 2.86199123e-01 -1.11091280e+00 4.96621549e-01 -8.29423547e-01 5.41571140e-01 8.56935382e-01 -6.36611402e-01 -4.10723001e-01 5.49646914e-01 2.51011282e-01 -4.88558292e-01 -9.04147327e-01 3.10233593e-01 5.08400798e-01 -1.93700564e+00 2.51736701e-01 1.64626181e-01 8.91781807e-01 -7.98590660e-01 -2.45504543e-01 1.59373209e-01 4.02805924e-01 -5.30113339e-01 1.28999782e+00 6.61230862e-01 4.89528924e-01 -5.73637962e-01 8.98481607e-01 4.62262034e-01 -7.05594122e-01 1.15772948e-01 5.32211773e-02 1.47177249e-01 6.26560271e-01 9.69976783e-01 -1.13968492e+00 1.00137782e+00 4.31822807e-01 1.69405326e-01 -7.57032454e-01 9.42670166e-01 -6.58874273e-01 9.34169173e-01 1.29968375e-01 -2.39317402e-01 1.07863292e-01 -4.99150753e-02 5.59208810e-01 1.17193031e+00 2.91891724e-01 3.14290345e-01 5.07486649e-02 1.17694366e+00 -1.57853395e-01 1.50065571e-01 -4.43064243e-01 -1.26214966e-01 5.34313083e-01 1.03968191e+00 -6.53590798e-01 -5.51011980e-01 -1.09461650e-01 3.36652040e-01 -5.54247238e-02 2.49677360e-01 -8.89878750e-01 -4.92129177e-01 9.22656134e-02 6.35878146e-01 5.32119274e-01 7.62309954e-02 -3.39858294e-01 -8.53332937e-01 6.01313770e-01 -1.20467365e+00 6.75888658e-01 -1.14072824e+00 -1.05518413e+00 6.61516249e-01 1.69436321e-01 -1.22031999e+00 -4.11591053e-01 -7.40987882e-02 -5.42843938e-01 8.48859310e-01 -1.69022262e+00 -1.02324986e+00 -3.07947192e-02 3.32579523e-01 3.75794470e-01 3.49192888e-01 4.67751652e-01 2.25273762e-02 5.45556024e-02 2.33239159e-01 -7.66071916e-01 9.82409045e-02 4.95892853e-01 -1.08822417e+00 2.89640278e-01 9.50435042e-01 4.27822441e-01 8.92709196e-01 1.00651073e+00 -1.09029293e+00 -7.77890325e-01 -9.04498100e-01 1.76286769e+00 -8.93197954e-01 1.25600624e+00 1.40380457e-01 -1.23022091e+00 2.52899885e-01 8.10458124e-01 -6.53742492e-01 1.00782561e+00 9.02608335e-02 -7.44767070e-01 2.68037915e-01 -1.36856115e+00 2.98390418e-01 7.64086604e-01 -8.91855538e-01 -1.34343600e+00 6.70912981e-01 1.05379295e+00 -2.12134719e-01 -9.10117090e-01 -3.76658738e-02 6.22228831e-02 -7.40570009e-01 4.68915462e-01 -8.84517908e-01 7.32180297e-01 -5.25918424e-01 -1.51772559e-01 -7.55664766e-01 -7.29832947e-02 -8.10445249e-01 -3.19513828e-01 1.76687813e+00 1.06230521e+00 -3.87013584e-01 4.47817028e-01 6.03333533e-01 6.50782436e-02 -6.36407852e-01 -7.81802297e-01 -7.51727283e-01 -9.06949341e-02 -6.60536408e-01 1.03759599e+00 1.15397012e+00 2.54657805e-01 4.45619583e-01 2.30738342e-01 9.37616378e-02 3.81626219e-01 5.35977662e-01 5.73722959e-01 -1.37186635e+00 -7.78016225e-02 -3.00967813e-01 7.54152909e-02 -6.03155673e-01 5.38415946e-02 -7.71762252e-01 -1.43186659e-01 -1.99273884e+00 2.66362429e-01 -2.39816368e-01 2.97451138e-01 2.99975276e-01 -4.12529498e-01 -1.43045215e-02 1.02726556e-01 3.77614379e-01 -6.18681192e-01 9.09073204e-02 1.10620165e+00 -1.66111559e-01 6.78940862e-02 9.51454192e-02 -1.45595384e+00 5.69580078e-01 8.46243858e-01 -7.28038967e-01 -2.55425572e-01 2.53613181e-02 8.45744848e-01 1.87610358e-01 7.87629545e-01 -7.97716975e-01 3.44521910e-01 -3.12656373e-01 -2.86430895e-01 -8.47940207e-01 -8.67540166e-02 -4.67503965e-01 1.66821182e-01 3.47977668e-01 -6.07832670e-01 3.22445035e-01 -2.58310102e-02 5.02190173e-01 -4.32279289e-01 -8.26451719e-01 -4.57042875e-03 -2.49151185e-01 -2.58952528e-01 -4.90815639e-02 -2.70000219e-01 5.75202942e-01 7.55673110e-01 -4.98600990e-01 -9.16719079e-01 -6.67508721e-01 -4.57985938e-01 5.24585508e-02 3.28898519e-01 3.59676749e-01 5.01961172e-01 -1.43102479e+00 -1.26404011e+00 -5.06754398e-01 5.28669000e-01 -5.87114871e-01 -1.90170144e-03 8.47450495e-01 -7.46492073e-02 5.22691607e-01 3.59448761e-01 -1.84496433e-01 -1.02762878e+00 7.22989500e-01 -2.91596740e-01 -5.96251488e-01 -3.59614432e-01 2.63627261e-01 -4.83688086e-01 -1.37638226e-01 -2.90275723e-01 -5.91062188e-01 -5.98803878e-01 3.46395910e-01 9.21191633e-01 3.38186681e-01 1.83653772e-01 -6.66552961e-01 -9.91810784e-02 6.63371161e-02 -4.06502150e-02 -5.50884068e-01 1.08241916e+00 -1.40636966e-01 -2.17847928e-01 3.00623327e-01 7.29163766e-01 7.38428831e-01 -8.23416829e-01 -3.72146189e-01 6.40677989e-01 -4.00776327e-01 -3.57093990e-01 -1.20798063e+00 -5.30718803e-01 4.96669151e-02 -3.36080819e-01 1.10764110e+00 7.34247267e-01 7.28126943e-01 8.71596634e-01 1.56256944e-01 4.24717367e-01 -1.05083454e+00 1.79967180e-01 3.74796897e-01 1.43078172e+00 -1.06400621e+00 1.03765829e-02 -6.86981857e-01 -7.65234649e-01 8.77589643e-01 9.09239799e-02 3.46106410e-01 3.69754583e-01 -3.06602586e-02 1.41662911e-01 -9.74655747e-01 -6.24261558e-01 -3.03983718e-01 4.07229155e-01 4.85945880e-01 3.29752266e-01 -3.64546448e-01 -8.51423919e-01 9.22827005e-01 -5.93751967e-01 1.18028596e-01 9.06907380e-01 1.07337630e+00 -4.59703624e-01 -1.02569246e+00 -6.43281400e-01 5.51198006e-01 -9.53856885e-01 -1.49901658e-01 -6.73769236e-01 8.37680757e-01 2.31801093e-01 1.59831345e+00 -4.40689296e-01 -1.19303323e-01 6.00098073e-01 7.94188306e-02 2.08590209e-01 -8.57466221e-01 -9.66366589e-01 -5.07700920e-01 8.19157660e-01 -2.14746431e-01 -7.43063211e-01 -1.09734905e+00 -1.07042134e+00 -1.04927085e-01 -6.67583466e-01 9.10782039e-01 6.58238471e-01 1.22074974e+00 5.14926314e-01 2.34337822e-01 3.27251524e-01 3.38827670e-01 -6.12114787e-01 -9.18191552e-01 -2.50429451e-01 7.87313938e-01 2.99459249e-01 -3.82422477e-01 -2.95451969e-01 5.76710165e-01]
[8.910386085510254, 9.641571998596191]
07cde1d0-8d81-41c8-84ea-0f0e5fb1af03
low-weight-and-learnable-image-denoising
1911.07167
null
https://arxiv.org/abs/1911.07167v2
https://arxiv.org/pdf/1911.07167v2.pdf
LIDIA: Lightweight Learned Image Denoising with Instance Adaptation
Image denoising is a well studied problem with an extensive activity that has spread over several decades. Despite the many available denoising algorithms, the quest for simple, powerful and fast denoisers is still an active and vibrant topic of research. Leading classical denoising methods are typically designed to exploit the inner structure in images by modeling local overlapping patches, while operating in an unsupervised fashion. In contrast, recent newcomers to this arena are supervised and universal neural-network-based methods that bypass this modeling altogether, targeting the inference goal directly and globally, while tending to be very deep and parameter heavy. This work proposes a novel lightweight learnable architecture for image denoising, and presents a combination of supervised and unsupervised training of it, the first aiming for a universal denoiser and the second for adapting it to the incoming image. Our architecture embeds in it several of the main concepts taken from classical methods, relying on patch processing, leveraging non-local self-similarity, exploiting representation sparsity and providing a multiscale treatment. Our proposed universal denoiser achieves near state-of-the-art results, while using a small fraction of the typical number of parameters. In addition, we introduce and demonstrate two highly effective ways for further boosting the denoising performance, by adapting this universal network to the input image.
['Michael Elad', 'Peyman Milanfar', 'Gregory Vaksman']
2019-11-17
null
null
null
null
['grayscale-image-denoising']
['computer-vision']
[ 4.18540061e-01 -7.45920911e-02 2.40249336e-01 -3.28237921e-01 -7.89737821e-01 -1.98639244e-01 6.01669431e-01 2.42518917e-01 -4.87170219e-01 4.38523978e-01 3.04658145e-01 1.38545230e-01 -1.88973293e-01 -8.66459250e-01 -6.87635660e-01 -1.27888668e+00 -9.43802670e-02 2.74054632e-02 1.02479510e-01 -5.42780340e-01 7.76140764e-03 3.57385546e-01 -1.70154393e+00 1.35997444e-01 9.14471149e-01 1.17112517e+00 2.64218777e-01 5.14946997e-01 1.94580227e-01 6.01379335e-01 -3.51698339e-01 -2.87403286e-01 3.53212655e-01 -6.56888187e-01 -3.47835690e-01 2.15784773e-01 4.38297868e-01 -5.87991951e-03 -2.66142517e-01 1.03055096e+00 6.58569932e-01 2.82450944e-01 3.70796502e-01 -4.14727718e-01 -6.28708839e-01 4.52315807e-01 -4.19007450e-01 2.07170203e-01 1.35259986e-01 1.32471278e-01 7.05804527e-01 -6.75074756e-01 4.54761624e-01 1.03452623e+00 1.06888449e+00 3.75996053e-01 -1.46244967e+00 -2.11502194e-01 -4.48760241e-02 2.78655380e-01 -1.27407694e+00 -5.49981892e-01 1.00253808e+00 -9.25413147e-02 7.12778866e-01 2.70782113e-01 5.54501832e-01 1.27287865e+00 2.53682166e-01 4.13987190e-01 1.44174302e+00 -5.67499280e-01 4.44076806e-01 -1.98300257e-01 -1.49711370e-01 3.28992099e-01 -1.04638591e-01 9.12833139e-02 -3.77853185e-01 8.60850066e-02 8.25511992e-01 1.74226820e-01 -3.07375908e-01 -3.64636302e-01 -9.59067523e-01 8.74740124e-01 6.08405888e-01 7.98670113e-01 -7.68243909e-01 1.14281878e-01 5.40653229e-01 5.13843000e-01 8.10759902e-01 1.42162949e-01 -1.70625985e-01 9.49961022e-02 -1.30817485e+00 2.91529447e-01 7.05920398e-01 2.21759632e-01 1.01942253e+00 1.83258846e-01 -6.35900646e-02 9.18937624e-01 1.49676636e-01 2.54294008e-01 6.67021453e-01 -9.19365168e-01 -5.29892221e-02 4.34313089e-01 -1.88906133e-01 -1.38389242e+00 -3.15022320e-01 -9.42165434e-01 -1.58001947e+00 4.96539086e-01 2.46900767e-01 4.13263328e-02 -9.61315691e-01 1.70448482e+00 4.31775033e-01 1.48685664e-01 -1.81485847e-01 8.34519029e-01 4.11193818e-01 5.94360113e-01 6.17213920e-02 -2.38909855e-01 1.25864637e+00 -1.01675475e+00 -6.83287263e-01 -2.69753456e-01 1.45255432e-01 -8.47663164e-01 6.93785310e-01 7.66031325e-01 -1.21405268e+00 -7.69362390e-01 -1.00205755e+00 -2.24204659e-01 -3.48917037e-01 -7.80146793e-02 4.54439253e-01 6.02198780e-01 -1.39968467e+00 1.14197993e+00 -8.25610340e-01 -4.64864045e-01 5.42378008e-01 1.48552030e-01 -5.66694558e-01 -2.96408862e-01 -9.73892570e-01 9.23939824e-01 1.12310946e-01 5.38977146e-01 -9.52753663e-01 -4.79227334e-01 -7.54040837e-01 1.67521521e-01 3.23174059e-01 -9.53039289e-01 7.45965004e-01 -1.33632994e+00 -1.65793264e+00 7.85714567e-01 -8.06299001e-02 -8.13547134e-01 6.07323170e-01 -3.30630958e-01 -3.04932028e-01 2.67582715e-01 -7.58662298e-02 3.39053810e-01 1.56494677e+00 -1.29776859e+00 -2.37999424e-01 -4.38790709e-01 -1.35930285e-01 -2.68275160e-02 -6.08988941e-01 -1.68521509e-01 -1.24861188e-01 -1.21975541e+00 3.01570445e-01 -3.35553408e-01 -5.98100185e-01 -2.75764428e-02 -5.96914515e-02 -5.27015328e-02 6.11570001e-01 -8.70839417e-01 1.13951957e+00 -2.20313072e+00 6.65655553e-01 2.03084156e-01 3.09517443e-01 3.40647191e-01 -2.47629285e-01 6.24547064e-01 -1.88676387e-01 -1.77176401e-01 -6.28878117e-01 -8.31089914e-01 -5.07223681e-02 3.64858538e-01 -2.60286480e-01 6.55437827e-01 1.33581772e-01 7.51474559e-01 -8.39969754e-01 -1.19122684e-01 4.78409916e-01 9.52330589e-01 -4.70826656e-01 2.81762481e-01 -3.17729861e-02 7.18916953e-01 -1.02774985e-01 4.30166394e-01 7.88313150e-01 9.82857794e-02 1.54790610e-01 -4.74939883e-01 -1.98092714e-01 -8.42097774e-02 -1.40837944e+00 1.88251221e+00 -6.87133372e-01 4.02590722e-01 7.26612091e-01 -1.69712639e+00 9.57475901e-01 4.12145704e-01 4.68799412e-01 -7.48985529e-01 2.92365044e-01 4.71622080e-01 -3.83224636e-01 -4.60825652e-01 1.60677105e-01 -5.40768266e-01 2.56089091e-01 1.29361138e-01 3.60667199e-01 1.41815171e-01 2.05283478e-01 -1.96782053e-01 1.29287505e+00 2.12480575e-01 3.53686512e-01 -3.30041409e-01 8.41369271e-01 -4.24604625e-01 3.32858682e-01 8.05757582e-01 7.29398895e-03 9.10641909e-01 9.50457230e-02 -6.65007174e-01 -1.00895882e+00 -8.09170067e-01 -1.41311586e-01 9.44061995e-01 -1.28522217e-01 -3.19097489e-01 -1.03229117e+00 -2.31948376e-01 -2.32751578e-01 1.50701776e-01 -8.76374722e-01 -5.78017533e-02 -6.62310779e-01 -5.74875772e-01 3.39101464e-01 2.39672899e-01 7.16240108e-01 -1.04606187e+00 -5.52208543e-01 4.66801792e-01 -1.06979489e-01 -8.91620994e-01 -5.30521525e-03 4.22830671e-01 -1.19602358e+00 -6.28846169e-01 -1.15529335e+00 -7.48811841e-01 5.37263811e-01 3.97102952e-01 1.27684581e+00 1.94232881e-01 -2.97309130e-01 3.19202274e-01 -4.62594599e-01 2.34314054e-02 -4.43297565e-01 6.09004535e-02 -1.57795921e-01 4.97015953e-01 -4.71484698e-02 -1.35278773e+00 -8.82716238e-01 6.33227006e-02 -1.38810802e+00 -2.08194405e-01 9.35527921e-01 9.44873393e-01 6.04670346e-01 4.89678830e-02 5.82635105e-01 -6.73710465e-01 5.51906645e-01 -4.64065135e-01 -1.27859652e-01 -8.48893672e-02 -4.33893889e-01 7.17720538e-02 9.91356254e-01 -2.18081012e-01 -9.50073004e-01 1.16332881e-02 -6.55140340e-01 -4.02408779e-01 -5.16073287e-01 5.87473094e-01 -8.73675868e-02 -3.13752800e-01 9.30156648e-01 3.96306366e-01 2.14988098e-01 -9.49096978e-01 4.95173752e-01 3.47355485e-01 7.08369374e-01 -4.02408242e-01 9.69953775e-01 8.91624928e-01 1.24040104e-01 -1.04883754e+00 -7.79870391e-01 -6.31859601e-01 -5.19515276e-01 -1.57817423e-01 7.73414493e-01 -8.25783730e-01 -4.14806068e-01 7.10780680e-01 -9.59536314e-01 -2.87336349e-01 -6.25660777e-01 1.36124805e-01 -4.82901126e-01 7.27882326e-01 -5.80813587e-01 -6.04786098e-01 -4.15057033e-01 -8.88930440e-01 1.06593680e+00 -2.35913340e-02 4.79437225e-02 -1.18580687e+00 3.23564082e-01 2.33633876e-01 9.84228969e-01 4.04755414e-01 5.65855742e-01 -2.94586897e-01 -3.26617181e-01 -3.84077668e-01 -7.81530738e-02 1.05040836e+00 1.12727411e-01 -6.59043670e-01 -1.06737733e+00 -4.10488605e-01 7.53754020e-01 -2.03990385e-01 1.33395159e+00 5.07300138e-01 1.05948460e+00 -2.62137771e-01 7.04165772e-02 7.45724142e-01 1.67775881e+00 -4.97709721e-01 8.35968316e-01 4.05963719e-01 5.05920887e-01 5.85931718e-01 -1.37966305e-01 3.36155832e-01 7.64526799e-02 5.95306218e-01 7.93335140e-01 -3.60377163e-01 -1.87628508e-01 1.39328435e-01 3.33744764e-01 9.69434440e-01 -6.10804617e-01 5.44345081e-02 -5.31952381e-01 7.13893950e-01 -1.95081460e+00 -9.77693319e-01 -6.28775544e-03 2.18743539e+00 8.05584013e-01 1.90808326e-02 -2.19784845e-02 3.51103276e-01 4.05889302e-01 5.41228652e-01 -1.64894819e-01 -2.84532547e-01 -4.57089037e-01 7.95964241e-01 2.84821719e-01 4.82911885e-01 -1.31600094e+00 5.17424822e-01 5.69370699e+00 1.05786669e+00 -1.16603339e+00 2.04286307e-01 5.86220264e-01 2.77393997e-01 -1.39765203e-01 -4.15165946e-02 -1.75858960e-01 2.84014672e-01 8.97722661e-01 1.67214379e-01 6.23430252e-01 6.04454875e-01 3.20696890e-01 -1.20326422e-01 -8.23122919e-01 9.75015044e-01 2.78290093e-01 -1.34835792e+00 9.84330699e-02 -1.46174520e-01 7.62463868e-01 -3.13506015e-02 4.34225462e-02 1.23498447e-01 -2.08202854e-01 -1.15930879e+00 6.20140016e-01 8.03047895e-01 1.23893045e-01 -6.33780062e-01 9.38490629e-01 4.25652593e-01 -9.82063651e-01 -1.89834446e-01 -2.83854067e-01 -3.33508819e-01 1.84654668e-01 1.18153882e+00 2.95367777e-01 9.28176641e-01 9.89856541e-01 9.94255185e-01 -4.82302397e-01 1.06651688e+00 -2.63841987e-01 6.87009990e-01 -3.45201075e-01 4.96701628e-01 3.70184928e-01 -5.68364859e-01 7.81748950e-01 1.40394676e+00 3.33757013e-01 -2.77631819e-01 4.94720936e-02 7.16836810e-01 -5.59706241e-02 5.36015108e-02 -5.09388387e-01 3.35101545e-01 -2.71418422e-01 1.54271305e+00 -7.11641908e-01 -2.16051161e-01 -2.33586431e-01 1.23933601e+00 1.22299232e-01 3.03375185e-01 -4.57283080e-01 -2.78617084e-01 4.16548520e-01 1.53214559e-01 7.17980623e-01 -3.40145797e-01 -3.98894131e-01 -1.12170231e+00 2.82468975e-01 -1.16821730e+00 1.73665091e-01 -2.99534857e-01 -1.59847736e+00 7.02304661e-01 -3.66633922e-01 -1.12375546e+00 -1.18738137e-01 -5.42702913e-01 -6.99172735e-01 9.92728353e-01 -1.76599610e+00 -1.19588625e+00 -4.71156240e-01 5.93791306e-01 5.20235717e-01 1.13531917e-01 9.70018983e-01 6.81159735e-01 -3.62116188e-01 2.44113475e-01 3.75716686e-01 -1.77479997e-01 7.60238826e-01 -1.16672111e+00 1.25551000e-01 9.90106881e-01 9.86846238e-02 7.94729173e-01 9.76275444e-01 -2.13538244e-01 -1.35083818e+00 -8.49318385e-01 7.50003874e-01 4.98918928e-02 6.57236695e-01 -4.19802666e-01 -1.23764718e+00 3.16357434e-01 5.58811486e-01 2.19112873e-01 3.03261787e-01 -9.19986330e-03 -3.73796999e-01 -5.28073013e-01 -1.14049733e+00 5.34636736e-01 9.41810369e-01 -4.37918037e-01 -5.03341436e-01 2.23763600e-01 1.56877518e-01 -1.82198569e-01 -8.09560180e-01 2.67050833e-01 3.91399920e-01 -1.51556611e+00 1.14339876e+00 -1.03228420e-01 7.06755221e-01 -2.64243633e-01 -8.76463950e-02 -1.45887375e+00 -3.98907572e-01 -8.88022482e-01 -2.00853825e-01 1.14942205e+00 -1.54090717e-01 -6.69533730e-01 6.05491042e-01 -1.78798005e-01 -2.99151748e-01 -9.04909432e-01 -1.18066072e+00 -6.85566366e-01 8.42573941e-02 -2.94147491e-01 5.14952578e-02 6.81341290e-01 -5.67307949e-01 2.26808220e-01 -5.59217453e-01 4.70595621e-02 1.05883408e+00 -1.23188183e-01 6.26947045e-01 -1.14957690e+00 -3.97228301e-01 -6.20466113e-01 -5.58331668e-01 -1.20833838e+00 -1.49405614e-01 -7.02154994e-01 1.71675533e-01 -1.55323446e+00 -1.35532930e-01 -9.06531438e-02 -3.80113751e-01 2.54053026e-01 -8.59769061e-02 7.48497486e-01 6.27805653e-04 1.82039961e-01 -4.09753412e-01 6.39362752e-01 1.01936352e+00 -2.08712205e-01 7.14371502e-02 -9.22145769e-02 -7.41443813e-01 8.43879700e-01 6.21668875e-01 -2.68670768e-01 -3.87564823e-02 -5.35988629e-01 2.65606701e-01 -4.14450735e-01 8.11028600e-01 -1.36367738e+00 4.42216247e-01 3.80035698e-01 3.14140856e-01 -1.04873307e-01 3.29498410e-01 -9.00779247e-01 1.27903104e-01 3.53425443e-01 -4.93395999e-02 -2.20810264e-01 6.27262443e-02 8.02171886e-01 -7.04493105e-01 -2.42582202e-01 9.78927910e-01 -3.42233747e-01 -6.81266189e-01 1.28386682e-02 -2.67755359e-01 -2.54938304e-01 7.22376645e-01 -3.25697422e-01 1.52969435e-01 -4.27906245e-01 -9.87898111e-01 -2.39872515e-01 3.11251700e-01 9.35919657e-02 4.75108564e-01 -9.92368698e-01 -9.53623652e-01 2.27377445e-01 -2.92466253e-01 -4.53077070e-02 7.54815400e-01 1.02969646e+00 -5.14401853e-01 -1.17401816e-01 -2.91421950e-01 -6.82885468e-01 -8.67341161e-01 5.26780903e-01 3.73159379e-01 -5.56725025e-01 -9.17154372e-01 7.11405396e-01 1.20593626e-02 -3.40809256e-01 1.23082615e-01 -1.00438252e-01 -1.45693824e-01 1.64661929e-01 4.85450923e-01 5.12939215e-01 4.18660522e-01 -6.01121724e-01 4.98969778e-02 6.85595155e-01 1.82263210e-01 1.75138295e-01 1.85625494e+00 -1.59401625e-01 -6.31635785e-01 3.13387930e-01 1.13799703e+00 8.13218486e-03 -1.24253094e+00 -2.78837174e-01 -1.01558827e-01 -2.53447145e-01 3.36978525e-01 -5.59471130e-01 -1.17147911e+00 9.79913950e-01 8.43165517e-01 6.25728130e-01 1.52143621e+00 -2.75966167e-01 9.22036171e-01 1.31117970e-01 3.00555646e-01 -9.77441609e-01 7.50988722e-02 3.62603128e-01 1.14934361e+00 -1.17586994e+00 1.06597908e-01 -2.96750575e-01 1.15428835e-01 1.27959764e+00 -7.37359226e-02 -7.54949450e-01 6.51906967e-01 1.71552777e-01 1.90191329e-01 -1.36160806e-01 -2.32995003e-01 -1.74214989e-01 2.03364626e-01 5.94933629e-01 3.43739688e-01 -2.40432635e-01 -4.75537568e-01 4.09170657e-01 2.47636884e-01 3.38621736e-02 4.86592278e-02 8.13800812e-01 -4.17521089e-01 -1.42747474e+00 -5.47061324e-01 2.55116343e-01 -5.99131644e-01 -1.04699150e-01 8.86463523e-02 4.66308922e-01 3.20492268e-01 8.31641376e-01 -2.26529717e-01 -3.04609928e-02 4.90074277e-01 -1.35652021e-01 3.93274426e-01 -3.99388760e-01 -1.06498051e+00 1.64409459e-01 -3.79695624e-01 -6.43046856e-01 -8.36275935e-01 -4.21952963e-01 -4.69110489e-01 -1.19612053e-01 -7.91969374e-02 -5.23170233e-02 6.64876282e-01 1.03112769e+00 3.20183903e-01 6.87892914e-01 7.16126084e-01 -1.57609928e+00 -7.65067816e-01 -1.07027066e+00 -5.20920277e-01 6.24548674e-01 5.95551491e-01 -4.35037166e-01 -6.24329746e-01 1.60048857e-01]
[11.52619743347168, -2.320706605911255]
36a4b8d6-b5d2-4713-903d-0bf226461a7b
implementation-of-the-vbm3d-video-denoising
2001.01802
null
https://arxiv.org/abs/2001.01802v1
https://arxiv.org/pdf/2001.01802v1.pdf
Implementation of the VBM3D Video Denoising Method and Some Variants
VBM3D is an extension to video of the well known image denoising algorithm BM3D, which takes advantage of the sparse representation of stacks of similar patches in a transform domain. The extension is rather straightforward: the similar 2D patches are taken from a spatio-temporal neighborhood which includes neighboring frames. In spite of its simplicity, the algorithm offers a good trade-off between denoising performance and computational complexity. In this work we revisit this method, providing an open-source C++ implementation reproducing the results. A detailed description is given and the choice of parameters is thoroughly discussed. Furthermore, we discuss several extensions of the original algorithm: (1) a multi-scale implementation, (2) the use of 3D patches, (3) the use of optical flow to guide the patch search. These extensions allow to obtain results which are competitive with even the most recent state of the art.
['Thibaud Ehret', 'Pablo Arias']
2020-01-06
null
null
null
null
['video-denoising']
['computer-vision']
[ 1.97501525e-01 -2.91740090e-01 3.46025467e-01 4.18126136e-02 -5.86305320e-01 -3.88092458e-01 5.70150554e-01 -4.84226197e-02 -4.71247703e-01 6.42633796e-01 2.67746806e-01 -9.04627517e-02 -1.21839076e-01 -7.31046259e-01 -5.27030766e-01 -9.60891068e-01 -1.77733958e-01 -2.02069599e-02 8.64956975e-01 -3.97880018e-01 4.94772851e-01 7.88121462e-01 -1.74890554e+00 2.35719278e-01 6.80091500e-01 1.07037115e+00 3.19475263e-01 6.38885677e-01 -1.33525059e-01 4.68477249e-01 -2.85522521e-01 -3.42320919e-01 5.95875323e-01 -6.99430585e-01 -8.42420638e-01 6.33744895e-01 6.94716334e-01 -3.09508353e-01 -1.91216052e-01 9.88587737e-01 5.17855108e-01 4.46310937e-01 4.93371934e-01 -7.91952431e-01 -2.11105570e-01 -3.33008260e-01 -5.22157133e-01 5.24338782e-01 4.68422353e-01 2.71581151e-02 5.55349410e-01 -8.52353275e-01 9.63194132e-01 1.14475501e+00 8.85758519e-01 3.70088845e-01 -1.44895315e+00 -1.99412648e-03 -4.66563068e-02 4.40494061e-01 -1.34627116e+00 -3.11293066e-01 7.35677421e-01 -4.17767495e-01 1.09964561e+00 2.32067749e-01 9.67317402e-01 6.20624125e-01 2.55868703e-01 4.89526123e-01 1.33446300e+00 -7.84534216e-01 3.06912184e-01 -1.19078860e-01 3.42804082e-02 6.42932355e-01 6.86420053e-02 2.04601482e-01 -5.01633584e-01 -3.84834617e-01 1.08243048e+00 -2.32454211e-01 -5.03859699e-01 -5.94541728e-01 -1.08142352e+00 8.02323997e-01 2.38132998e-01 7.00715125e-01 -6.31053627e-01 5.45638762e-02 3.87529641e-01 4.12931502e-01 8.14240158e-01 -1.47979567e-03 -1.60359249e-01 -1.33951828e-01 -1.32726371e+00 4.40298945e-01 7.52485991e-01 5.82900882e-01 1.14202893e+00 9.77024660e-02 5.91153428e-02 7.72116899e-01 7.58640468e-02 3.09083164e-01 2.14824066e-01 -1.55788171e+00 3.18137929e-03 2.99462490e-02 1.54916581e-03 -1.15235078e+00 -1.06881127e-01 -1.32418707e-01 -8.96029770e-01 6.81802332e-01 2.93016940e-01 2.32210547e-01 -6.60730124e-01 1.21003115e+00 4.10226047e-01 5.82199156e-01 -1.69851705e-01 7.96677411e-01 5.73309541e-01 6.99564278e-01 -3.85220855e-01 -4.27006781e-01 9.57789004e-01 -1.06482446e+00 -6.97432399e-01 3.59321892e-01 1.46606907e-01 -1.25461805e+00 4.63142723e-01 6.29981101e-01 -1.39468145e+00 -7.99487770e-01 -7.79365420e-01 -1.24760374e-01 -2.14645848e-01 -3.69760811e-01 5.61281741e-01 6.38865590e-01 -1.67343926e+00 1.09684181e+00 -8.58736992e-01 -7.57701099e-01 1.71486571e-01 2.20684946e-01 -5.29200017e-01 -2.83864170e-01 -7.15883613e-01 7.98949718e-01 3.35684568e-02 5.12960888e-02 -5.79161346e-01 -5.17514944e-01 -9.31303501e-01 -1.87557444e-01 1.55766666e-01 -8.73316526e-01 1.00216877e+00 -1.12779021e+00 -1.58163381e+00 1.14476192e+00 -5.99607825e-01 -4.48631674e-01 6.82837665e-01 -4.45853621e-02 -1.10919952e-01 6.50711894e-01 8.40255916e-02 6.40518665e-01 1.10210299e+00 -1.10399115e+00 -4.71196204e-01 -9.97535586e-02 4.96088155e-02 6.01782985e-02 8.84666890e-02 -1.89214163e-02 -6.39144957e-01 -1.08716738e+00 2.80450463e-01 -7.59440601e-01 -4.38301742e-01 3.03584009e-01 8.84203389e-02 5.16154692e-02 7.92333961e-01 -7.38648653e-01 1.08334339e+00 -2.39099598e+00 4.44002092e-01 1.22782685e-01 3.33282575e-02 2.78944343e-01 -1.71033606e-01 7.14964807e-01 -1.89897254e-01 -1.60582691e-01 -3.12413812e-01 -5.34424663e-01 -3.54773253e-01 3.73422295e-01 -1.29161030e-01 7.93911815e-01 1.27624571e-01 4.43990648e-01 -7.09568024e-01 -4.05969352e-01 7.27332175e-01 7.37223566e-01 -6.79143846e-01 -4.26143641e-03 2.49505013e-01 5.05188048e-01 -2.80427605e-01 5.59157133e-01 1.05989718e+00 2.47829705e-01 -2.97161818e-01 -4.40241486e-01 -5.78904331e-01 1.99595928e-01 -1.63031805e+00 1.74480581e+00 -1.27653316e-01 5.97555339e-01 5.29772997e-01 -1.19771588e+00 8.08597267e-01 4.85937566e-01 7.45261431e-01 -5.10149837e-01 -1.11220874e-01 4.45088655e-01 -4.21749830e-01 -5.31784713e-01 3.73961806e-01 -3.40748966e-01 6.42080128e-01 2.95511603e-01 2.18726680e-01 -3.85031164e-01 5.59166551e-01 1.99150387e-02 1.07356632e+00 3.88228238e-01 5.13104439e-01 -6.31457210e-01 9.45771337e-01 1.39273882e-01 4.30597275e-01 7.40042210e-01 -3.71861190e-01 9.84036028e-01 4.33874905e-01 -6.64404869e-01 -1.01532435e+00 -8.78357112e-01 -2.37057462e-01 3.76224577e-01 2.76458085e-01 -5.13864398e-01 -6.89523160e-01 -3.29981476e-01 -1.68807372e-01 1.71077430e-01 -7.04158366e-01 3.69238973e-01 -8.05614769e-01 -6.57639980e-01 3.26513592e-03 1.48440912e-01 6.33839369e-01 -1.08694470e+00 -6.20586932e-01 2.42372274e-01 -2.35792115e-01 -8.91355276e-01 -2.58440346e-01 1.37083977e-01 -1.45594263e+00 -1.11565709e+00 -1.08216977e+00 -8.98033738e-01 4.59941208e-01 6.47738159e-01 1.29084659e+00 3.88651192e-01 -2.80320078e-01 7.66885400e-01 -5.01338243e-01 1.41498283e-01 -3.75975072e-01 -2.67838240e-01 -2.34810039e-01 -2.51626372e-02 1.66638881e-01 -8.57583284e-01 -6.31409705e-01 3.08506817e-01 -1.23828554e+00 -2.39654869e-01 2.23354653e-01 7.52269804e-01 9.46976721e-01 3.17458600e-01 2.93297898e-02 -6.43137097e-01 2.21524686e-01 -1.40843734e-01 -6.03668988e-01 -1.91021696e-01 -3.75872135e-01 -1.98067501e-01 3.54948461e-01 -5.89962937e-02 -8.87606204e-01 2.67461717e-01 -5.83281398e-01 -3.19298744e-01 -2.96643317e-01 2.81862635e-02 1.63874179e-01 -5.37110567e-01 4.43733335e-01 2.98757672e-01 3.10571730e-01 -8.34955573e-01 2.80103922e-01 1.99701801e-01 4.31238055e-01 -3.25924098e-01 7.61533618e-01 9.40693021e-01 1.24385856e-01 -1.18555892e+00 -3.54649782e-01 -8.20793271e-01 -8.50650549e-01 -2.76656985e-01 8.53565812e-01 -6.44163430e-01 -3.15124422e-01 6.89901829e-01 -1.26746118e+00 -2.41947472e-01 -4.42896873e-01 4.16527927e-01 -8.67849052e-01 8.49357724e-01 -8.47912669e-01 -5.01288235e-01 -1.90385543e-02 -1.19710302e+00 1.12320960e+00 5.61290681e-02 7.93356895e-02 -1.09820402e+00 2.61438042e-01 2.29627788e-01 5.08822978e-01 2.02274844e-01 6.82704628e-01 6.76923469e-02 -5.99969566e-01 1.16344929e-01 1.49888573e-02 7.52783954e-01 1.02349922e-01 -1.82616021e-02 -8.55913937e-01 -4.10394102e-01 3.61444354e-01 2.56508887e-01 1.05550194e+00 8.77113283e-01 7.59176433e-01 1.93673998e-01 -1.83274657e-01 7.21253633e-01 1.76972592e+00 7.04376847e-02 1.04793310e+00 7.07624435e-01 1.59753621e-01 6.28257334e-01 5.35193801e-01 2.72789478e-01 -7.86724240e-02 8.02653432e-01 6.29427135e-01 -2.99949557e-01 -4.36766833e-01 3.40806007e-01 3.09186816e-01 7.17100859e-01 -5.50156832e-01 1.13309294e-01 -3.87852997e-01 6.36821866e-01 -1.81506217e+00 -1.18179977e+00 -4.38715547e-01 2.23687649e+00 3.14713329e-01 -1.68301567e-01 2.78692573e-01 5.82506001e-01 6.74151242e-01 4.07613963e-01 4.61933985e-02 -4.94973689e-01 -3.41600090e-01 4.74220663e-01 4.46868896e-01 9.28603530e-01 -1.19266927e+00 5.66327929e-01 7.27935839e+00 8.98380280e-01 -1.03209865e+00 1.87276080e-01 3.55444103e-01 1.33425340e-01 -7.55769759e-02 2.54122578e-02 -6.08136594e-01 3.27732861e-01 4.96928364e-01 3.43746841e-02 4.37866956e-01 4.24752504e-01 4.05893594e-01 -5.58686316e-01 -6.44166231e-01 1.02937829e+00 2.66392261e-01 -1.48647249e+00 -1.12553313e-01 7.84702748e-02 7.56014287e-01 1.49995223e-01 -2.99493730e-01 -2.42436707e-01 -3.77387196e-01 -6.58464074e-01 7.07721889e-01 4.38836068e-01 3.23703259e-01 -7.22910225e-01 6.89122796e-01 6.65688515e-02 -1.18771899e+00 9.77875516e-02 -4.89719540e-01 -1.46753550e-01 5.47359765e-01 8.15509140e-01 -8.84777494e-03 8.52359056e-01 1.21264851e+00 1.11238027e+00 -3.69005561e-01 1.40319526e+00 -6.99675158e-02 4.67811674e-01 -4.28708136e-01 5.40030539e-01 4.29218233e-01 -6.80171549e-01 7.64000118e-01 1.28437018e+00 5.59786022e-01 1.43585261e-02 1.08270086e-02 5.10473132e-01 5.82460225e-01 1.51604742e-01 -6.59174502e-01 6.26660526e-01 -1.62749946e-01 1.12673748e+00 -8.40493977e-01 -3.48015368e-01 -8.01104009e-01 1.33353364e+00 -4.49134335e-02 3.33245456e-01 -4.24810946e-01 -2.38925606e-01 7.45663583e-01 4.11010534e-01 9.15186882e-01 -3.86702061e-01 5.09106070e-02 -1.01515353e+00 1.03941217e-01 -7.30769932e-01 3.69046777e-01 -7.19463348e-01 -1.27065778e+00 8.15204501e-01 1.21301517e-01 -1.50744748e+00 -1.98840275e-01 -6.21670723e-01 -3.74702841e-01 9.09761667e-01 -1.79843712e+00 -6.35985255e-01 -2.90191233e-01 7.63837636e-01 5.37965894e-01 1.94702238e-01 8.11433554e-01 5.86337924e-01 -1.54613853e-01 -1.83030441e-01 3.79689991e-01 -2.20953941e-01 6.30933821e-01 -1.03357875e+00 1.88909486e-01 1.05855417e+00 9.03582349e-02 3.23398590e-01 7.82191694e-01 -4.01181072e-01 -1.04456437e+00 -7.10838377e-01 1.11008048e+00 -2.25545038e-02 4.93726313e-01 2.29405947e-02 -1.06778932e+00 5.02982557e-01 4.23419178e-01 2.19615206e-01 4.40433741e-01 -3.29712957e-01 -9.17412415e-02 -1.97301418e-01 -1.26252174e+00 3.43591064e-01 9.60226655e-01 -2.25269258e-01 -6.79164112e-01 8.46753269e-02 1.26042813e-01 -3.38862985e-01 -8.15173566e-01 1.55324101e-01 2.66670316e-01 -1.80805326e+00 1.28041184e+00 1.04589090e-01 2.23653972e-01 -5.67159116e-01 -2.59743631e-01 -1.15094209e+00 -5.25338590e-01 -8.99280310e-01 7.13009164e-02 9.49054897e-01 -3.08354974e-01 -6.75151289e-01 6.57589197e-01 -5.45718819e-02 -1.93373397e-01 -6.39623940e-01 -1.34013510e+00 -8.87508690e-01 -9.21719968e-02 -3.73773634e-01 8.45052451e-02 5.33020496e-01 -3.40969980e-01 3.98315564e-02 -4.40824091e-01 -3.59194688e-02 7.09836721e-01 2.24940628e-01 4.53245401e-01 -1.24126828e+00 -3.83185059e-01 -4.35975611e-01 -8.47821534e-01 -1.37243223e+00 -1.64581120e-01 -5.18281341e-01 -1.89604253e-01 -1.50505245e+00 3.05619650e-03 -1.53175756e-01 -1.10651128e-01 1.36728242e-01 1.06527694e-01 9.05305207e-01 1.75866753e-01 2.33258665e-01 -3.54820222e-01 3.30689520e-01 1.27740705e+00 1.33559093e-01 -1.13504954e-01 5.28097004e-02 -2.96361506e-01 7.11678922e-01 4.67468172e-01 -3.44027042e-01 -1.87104747e-01 -5.23187160e-01 -1.79687664e-01 -9.32473987e-02 5.92095792e-01 -9.28758979e-01 -1.63390990e-02 2.29778722e-01 1.79947525e-01 -7.87686050e-01 6.16585910e-01 -1.06885386e+00 2.39155710e-01 6.10219657e-01 2.97368169e-01 2.06137449e-01 1.50306135e-01 5.61632693e-01 -6.89453065e-01 -5.92279971e-01 1.16043997e+00 -4.83043492e-01 -1.00513184e+00 7.84835219e-02 -7.76095212e-01 -3.45155448e-01 9.47271287e-01 -7.59427428e-01 1.32143572e-01 -4.46730852e-01 -9.00616109e-01 -4.17381197e-01 8.56179655e-01 2.59206444e-02 3.63681138e-01 -1.24629474e+00 -6.99881911e-01 2.67002285e-01 -2.95726627e-01 -1.47669345e-01 3.88873160e-01 1.16763973e+00 -1.00467789e+00 1.85852244e-01 -3.62386733e-01 -8.08331251e-01 -1.45871484e+00 5.63100219e-01 4.54493701e-01 -5.48514649e-02 -9.86115754e-01 6.65118873e-01 1.20170079e-01 3.21591608e-02 3.55803259e-02 -1.12470888e-01 -4.53646295e-02 -7.29926154e-02 7.56794751e-01 6.95829272e-01 3.18023294e-01 -8.52251887e-01 -3.75973880e-01 1.05084467e+00 4.06700015e-01 -1.70181897e-02 1.61856294e+00 -4.12429363e-01 -6.53604746e-01 2.03170069e-02 1.21657050e+00 1.40287846e-01 -1.39512348e+00 -1.00836143e-01 -2.01893121e-01 -8.02918315e-01 2.59582847e-01 -2.53102273e-01 -1.34329474e+00 7.51122832e-01 6.63040340e-01 3.97915035e-01 1.66048205e+00 -1.39091581e-01 7.59630024e-01 -1.40448781e-02 4.02186841e-01 -7.63837159e-01 -9.41454470e-02 5.19616604e-01 8.18605006e-01 -8.86406839e-01 2.86952734e-01 -8.69967341e-01 -1.02105886e-01 1.44826090e+00 5.17967828e-02 -5.74454129e-01 9.14789736e-01 3.10350150e-01 2.45574579e-01 -8.64821076e-02 -3.22309196e-01 -3.83640319e-01 1.41169235e-01 9.44069803e-01 4.70765501e-01 -6.45999253e-01 -7.26569831e-01 -2.15966016e-01 3.59176219e-01 7.21640736e-02 5.05408764e-01 1.06686032e+00 -3.62222105e-01 -1.47632623e+00 -6.72312140e-01 -4.35328782e-02 -6.01369202e-01 -1.07179098e-01 6.92662597e-02 7.81297505e-01 3.58862758e-01 9.93615150e-01 -4.03956994e-02 1.83052480e-01 5.46800435e-01 -9.91623849e-02 8.56082916e-01 -3.74660432e-01 -6.50998116e-01 4.88687992e-01 -1.69569805e-01 -9.74798918e-01 -1.11288083e+00 -8.11321139e-01 -7.70466268e-01 -3.95132601e-01 2.57346518e-02 1.25594482e-01 5.03047764e-01 7.09300220e-01 3.71170670e-01 1.32467672e-01 5.28186917e-01 -1.56876159e+00 -1.70154616e-01 -6.32647216e-01 -9.68111694e-01 4.44466442e-01 5.18029571e-01 -7.17144847e-01 -4.94195908e-01 4.08363163e-01]
[11.325738906860352, -2.339233160018921]
00d0cf4e-b329-430f-99bc-5488471f5ac6
ensemble-based-offline-to-online
2306.06871
null
https://arxiv.org/abs/2306.06871v1
https://arxiv.org/pdf/2306.06871v1.pdf
Ensemble-based Offline-to-Online Reinforcement Learning: From Pessimistic Learning to Optimistic Exploration
Offline reinforcement learning (RL) is a learning paradigm where an agent learns from a fixed dataset of experience. However, learning solely from a static dataset can limit the performance due to the lack of exploration. To overcome it, offline-to-online RL combines offline pre-training with online fine-tuning, which enables the agent to further refine its policy by interacting with the environment in real-time. Despite its benefits, existing offline-to-online RL methods suffer from performance degradation and slow improvement during the online phase. To tackle these challenges, we propose a novel framework called Ensemble-based Offline-to-Online (E2O) RL. By increasing the number of Q-networks, we seamlessly bridge offline pre-training and online fine-tuning without degrading performance. Moreover, to expedite online performance enhancement, we appropriately loosen the pessimism of Q-value estimation and incorporate ensemble-based exploration mechanisms into our framework. Experimental results demonstrate that E2O can substantially improve the training stability, learning efficiency, and final performance of existing offline RL methods during online fine-tuning on a range of locomotion and navigation tasks, significantly outperforming existing offline-to-online RL methods.
['Zhaopeng Meng', 'Yan Zheng', 'Jinyi Liu', 'Yi Ma', 'Kai Zhao']
2023-06-12
null
null
null
null
['offline-rl']
['playing-games']
[-2.50311911e-01 -1.51790768e-01 -2.25688368e-01 -1.26074210e-01 -5.87729752e-01 -6.10302746e-01 2.94679821e-01 1.76414579e-01 -8.33214819e-01 1.16568327e+00 -1.92499354e-01 -2.58967102e-01 -3.70618701e-01 -8.61078978e-01 -8.21131706e-01 -7.06673443e-01 -3.41433048e-01 2.59174764e-01 3.92007679e-01 -4.82986450e-01 1.73506692e-01 2.25824714e-01 -1.72759986e+00 -2.43429735e-01 1.32743394e+00 9.85572100e-01 4.24282968e-01 5.13039768e-01 3.15966420e-02 7.38542616e-01 -4.51185167e-01 1.19125351e-01 4.49981838e-01 -5.37899554e-01 -4.21710432e-01 -1.94597125e-01 -3.38720649e-01 -5.90593517e-01 -4.46704209e-01 8.09648156e-01 7.43669868e-01 5.65595865e-01 -2.42404407e-03 -1.11768508e+00 -3.25970232e-01 5.94817936e-01 -3.70999366e-01 4.04901318e-02 2.30544627e-01 6.94101691e-01 7.26421535e-01 -5.41292667e-01 3.54268134e-01 1.13891411e+00 6.78730488e-01 5.86348355e-01 -1.08544362e+00 -7.63765097e-01 7.01781392e-01 9.98972133e-02 -9.91711199e-01 -3.33177924e-01 5.97535074e-01 -2.23740041e-02 8.53908181e-01 -1.83000565e-01 1.00687313e+00 9.41107512e-01 6.79191062e-03 9.12986696e-01 1.16322994e+00 -1.24891020e-01 7.72686779e-01 -3.03046167e-01 -5.65665960e-01 7.97282815e-01 1.83121338e-01 8.57172966e-01 -7.24264503e-01 9.05616507e-02 1.04775369e+00 -1.74639523e-01 -6.97503760e-02 -7.30925024e-01 -8.82033288e-01 6.87911510e-01 5.88342607e-01 -2.30050728e-01 -3.14410627e-01 3.57136726e-01 6.60177410e-01 6.23972297e-01 -4.76241745e-02 7.91976273e-01 -5.07870317e-01 -6.67688966e-01 -5.83432734e-01 3.08092982e-01 5.81193328e-01 6.68374002e-01 8.96487296e-01 3.31185669e-01 -1.13849267e-02 8.33230495e-01 8.83851871e-02 2.99510956e-01 8.49407971e-01 -1.15671885e+00 5.85199237e-01 7.95035183e-01 4.87015694e-01 -5.71629763e-01 -4.14819509e-01 -7.52038956e-01 -3.72701675e-01 5.93441904e-01 2.99510419e-01 -4.20233041e-01 -6.17662668e-01 1.88959193e+00 6.37643635e-01 2.39889175e-02 9.90384296e-02 9.10295844e-01 1.55229419e-01 3.54927272e-01 8.89938921e-02 -3.83229405e-01 7.04786062e-01 -1.36006749e+00 -6.88415766e-01 -3.33827674e-01 5.67669451e-01 7.93502256e-02 1.26169896e+00 5.51175714e-01 -9.44664776e-01 -5.95048904e-01 -1.15578973e+00 3.71043295e-01 -2.37412870e-01 -1.98125876e-02 6.98364139e-01 5.19033849e-01 -7.36072361e-01 1.01493025e+00 -1.04726100e+00 -5.24946079e-02 4.58152235e-01 6.46623671e-01 -1.68477654e-01 2.28983670e-01 -1.17613864e+00 7.61944592e-01 7.59080112e-01 1.67311162e-01 -1.16013837e+00 -5.81631005e-01 -5.52844524e-01 -2.04926915e-02 9.62108076e-01 -6.49310589e-01 1.45026243e+00 -1.02806723e+00 -2.42783141e+00 -7.99416006e-02 2.35527664e-01 -6.15549266e-01 8.39973450e-01 -4.32062477e-01 -1.14420928e-01 -8.81441310e-02 -2.08039954e-01 5.80303729e-01 7.77121186e-01 -1.31455600e+00 -7.89266348e-01 -2.71916062e-01 3.84904057e-01 6.70443594e-01 -4.56335336e-01 -7.25966215e-01 -2.92188764e-01 -4.64883387e-01 -1.16360724e-01 -8.67220402e-01 -5.05906284e-01 1.08487448e-02 4.34119165e-01 -1.46322146e-01 7.26254702e-01 -2.04608858e-01 1.32682049e+00 -1.90975857e+00 2.39240602e-02 9.03263167e-02 -1.14604626e-02 3.90338540e-01 -4.30833876e-01 5.52077532e-01 4.80555266e-01 -2.54798412e-01 -1.79417267e-01 -1.86488330e-01 5.17781153e-02 6.45082057e-01 -2.26757959e-01 1.07776016e-01 -1.47993013e-03 1.10817122e+00 -1.50417137e+00 -3.00043344e-01 1.32375672e-01 9.72478986e-02 -9.02422607e-01 4.57044929e-01 -4.38378066e-01 9.35429513e-01 -5.76528013e-01 5.03004253e-01 2.98008710e-01 -1.10013470e-01 5.09377897e-01 3.73911262e-01 -3.24115038e-01 1.50014907e-01 -1.19478106e+00 1.90474689e+00 -1.00415170e+00 1.41382322e-01 1.46067441e-01 -1.02407801e+00 8.98730576e-01 -5.13008423e-02 5.63772261e-01 -1.14371347e+00 4.17865738e-02 4.17504936e-01 4.07667905e-02 -4.41226035e-01 5.40475249e-01 2.05229327e-01 2.76790932e-02 4.93879497e-01 5.92649244e-02 1.69916227e-01 3.35613549e-01 -2.92521358e-01 1.02872193e+00 8.15533578e-01 4.76159394e-01 -1.35781085e-02 3.92381519e-01 -1.53505027e-01 8.12602341e-01 1.01611185e+00 -4.42048788e-01 -9.85147804e-02 2.26884022e-01 -3.85494232e-01 -8.35916817e-01 -9.76077855e-01 2.50115901e-01 1.59238255e+00 3.89844507e-01 -3.63188863e-01 -5.94616830e-01 -8.40037048e-01 3.34006935e-01 4.51406300e-01 -5.99782884e-01 -3.56100827e-01 -7.98623085e-01 -5.92714608e-01 4.57068086e-01 7.61448264e-01 7.73982882e-01 -1.10472023e+00 -1.27310109e+00 7.27686405e-01 -3.45870480e-02 -6.93826258e-01 -4.36334193e-01 4.82272744e-01 -1.22294581e+00 -8.22287560e-01 -4.07985508e-01 -4.17221308e-01 6.15022302e-01 1.08076386e-01 7.81317353e-01 1.08199760e-01 -1.96627434e-02 2.46788248e-01 -4.54652637e-01 -4.40293215e-02 -1.72433838e-01 3.66704613e-02 2.81455994e-01 -3.68946522e-01 -2.09491611e-01 -8.74738812e-01 -9.31906700e-01 4.95993048e-01 -6.37947857e-01 -1.66203201e-01 7.18222499e-01 1.31608105e+00 5.50009787e-01 1.58385515e-01 1.03122079e+00 -4.87675428e-01 6.69010758e-01 -4.14759010e-01 -9.18354273e-01 1.56915307e-01 -1.06081951e+00 5.10783672e-01 1.13612056e+00 -8.19269300e-01 -1.00835931e+00 3.56964320e-02 -6.35253685e-03 -3.80236506e-01 2.94646442e-01 5.80553472e-01 5.14418297e-02 -4.57407176e-01 6.08624399e-01 5.57821751e-01 2.41426706e-01 -2.91446328e-01 4.49269205e-01 2.78823197e-01 4.85046059e-01 -8.14187527e-01 7.17831492e-01 2.67423391e-01 -1.33947596e-01 -1.46147490e-01 -7.58388579e-01 -2.56113380e-01 -3.52197319e-01 -4.97876346e-01 1.61368906e-01 -9.13249314e-01 -1.24407554e+00 3.05156320e-01 -3.65884811e-01 -9.46108520e-01 -5.72226405e-01 3.99816304e-01 -8.45840335e-01 1.93211243e-01 -2.31894284e-01 -1.09487522e+00 -2.26224318e-01 -1.07063651e+00 4.92157668e-01 5.52906871e-01 1.64595470e-01 -9.43026781e-01 2.27688625e-01 5.94443157e-02 6.38564050e-01 1.21998005e-01 4.33670789e-01 -1.52794793e-01 -4.10597116e-01 2.68417537e-01 8.20541978e-02 6.20643049e-02 6.86134249e-02 -4.58613992e-01 -6.87781453e-01 -8.08079600e-01 -2.64461428e-01 -9.70455229e-01 7.09713995e-01 -4.08164412e-03 1.34030795e+00 -5.33240438e-01 -1.63099036e-01 6.16472781e-01 1.47627437e+00 4.41240489e-01 3.34146231e-01 8.56496155e-01 2.49574140e-01 3.14164370e-01 9.67288613e-01 9.91516113e-01 5.41757882e-01 5.56943774e-01 6.63731396e-01 3.55100900e-01 5.53533398e-02 -8.50398540e-01 6.88150704e-01 7.53289461e-01 -3.07208877e-02 -2.60513485e-03 -5.54835558e-01 5.45165241e-01 -2.26270509e+00 -7.46748865e-01 8.11468899e-01 2.48831987e+00 1.14030898e+00 1.94903314e-01 4.14101094e-01 -6.73231035e-02 2.89866090e-01 -6.12878688e-02 -1.32750201e+00 -2.04808459e-01 2.53198266e-01 -3.91480625e-02 5.01031220e-01 3.85234952e-01 -8.72743487e-01 9.08707500e-01 6.09688330e+00 9.03152347e-01 -1.28727889e+00 -8.30742121e-02 2.84940809e-01 -3.16949397e-01 -1.76569089e-01 -1.31551409e-02 -6.48061156e-01 5.14893711e-01 8.46197903e-01 -5.13171926e-02 1.15368354e+00 1.11540508e+00 2.74636000e-01 -3.48846078e-01 -1.00752831e+00 7.55300343e-01 -4.08146977e-01 -1.08129907e+00 -2.89884418e-01 -9.98039544e-02 9.06990230e-01 6.10554442e-02 3.42083238e-02 9.07857895e-01 5.60961664e-01 -8.50893199e-01 7.96420872e-01 3.13185900e-01 6.95612669e-01 -1.00423312e+00 4.69188690e-01 7.16076314e-01 -1.35593688e+00 -7.89928913e-01 -1.79147705e-01 -2.41928414e-01 -4.72702719e-02 1.59740716e-01 -5.32813311e-01 5.95217943e-01 5.38508475e-01 4.44680721e-01 -4.52722400e-01 1.08395529e+00 -3.46674591e-01 3.83185208e-01 -3.42886090e-01 -4.44609821e-01 4.23914552e-01 -2.70260453e-01 5.09734094e-01 6.60843849e-01 2.50759900e-01 7.73491189e-02 4.94218916e-01 7.06442714e-01 1.30248800e-01 -1.16514519e-01 -3.18689793e-01 -1.51753291e-01 8.57057571e-01 9.96297300e-01 -4.79715586e-01 -1.00039504e-01 4.50527743e-02 7.24270701e-01 8.03235888e-01 2.84113228e-01 -8.17023933e-01 -5.31048596e-01 4.22325879e-01 -2.63319582e-01 5.05422533e-01 -4.42070752e-01 -9.45971310e-02 -9.98769224e-01 1.82150099e-02 -9.68098283e-01 1.29022390e-01 -2.34828651e-01 -1.01288569e+00 3.48623037e-01 -2.37128139e-01 -1.40808165e+00 -3.65283817e-01 -2.55699366e-01 -4.18321878e-01 2.13398859e-01 -1.61682200e+00 -6.77787840e-01 -3.21437806e-01 2.70963609e-01 5.26730299e-01 -1.96932420e-01 6.30965114e-01 7.15354085e-02 -6.05441272e-01 8.31976950e-01 5.84802032e-01 -3.55320692e-01 7.21405387e-01 -1.31732547e+00 1.30055040e-01 4.15196776e-01 -2.88797051e-01 5.57663620e-01 5.77329695e-01 -4.52746809e-01 -1.62204790e+00 -1.07130814e+00 5.11413403e-02 7.09040696e-03 7.16414154e-01 -2.85560429e-01 -7.52937615e-01 1.39982373e-01 -2.15548813e-01 5.77193163e-02 4.07963842e-01 1.64357141e-01 -1.33032441e-01 -4.55299616e-01 -9.73699808e-01 8.78812134e-01 1.19341767e+00 -3.34645003e-01 -2.76623935e-01 -7.93199092e-02 6.57022059e-01 -6.37176096e-01 -9.78967726e-01 4.33721244e-01 8.20087671e-01 -9.44429517e-01 8.29611897e-01 -4.77415115e-01 2.03480303e-01 -4.38285828e-01 2.29902819e-01 -1.65364301e+00 -1.47505462e-01 -1.00894570e+00 -4.65727150e-01 7.41405427e-01 5.39173819e-02 -9.07977402e-01 8.43126416e-01 1.33820206e-01 -9.96298343e-02 -1.16911829e+00 -1.05397832e+00 -1.21906877e+00 1.58851162e-01 -1.87293574e-01 7.83148289e-01 5.40930510e-01 2.58229226e-01 -1.19299658e-01 -3.97483081e-01 6.65656999e-02 5.77500761e-01 3.39999855e-01 9.48271394e-01 -8.43878984e-01 -7.10784614e-01 -3.62859607e-01 1.57460093e-01 -1.53719413e+00 -2.60620955e-02 -4.83761191e-01 5.02565205e-01 -1.21725845e+00 -1.05788127e-01 -7.55513191e-01 -5.85262835e-01 6.56872332e-01 -3.84858638e-01 -1.52144790e-01 1.19488411e-01 2.69192785e-01 -1.08944631e+00 1.30577672e+00 1.58649421e+00 3.33405942e-01 -6.41467273e-01 -2.06899315e-01 -3.75160664e-01 3.92067581e-01 9.60950673e-01 -3.77290398e-01 -6.95426762e-01 -3.90812069e-01 2.23044023e-01 4.55349535e-01 9.77783054e-02 -1.24449301e+00 3.50936145e-01 -4.93758142e-01 2.19792172e-01 -2.29331478e-01 1.96990207e-01 -5.44594646e-01 -2.66562462e-01 8.02533865e-01 -3.67599845e-01 1.04974203e-01 2.25876108e-01 1.05136681e+00 -1.56086624e-01 -4.72463071e-02 6.68813050e-01 -8.22491497e-02 -8.80721569e-01 2.56983876e-01 -4.35856909e-01 1.66621029e-01 9.70869005e-01 -3.82923424e-01 -3.08035940e-01 -3.02644044e-01 -5.49479961e-01 9.10637796e-01 4.87192571e-01 2.82431513e-01 4.91026819e-01 -1.21094429e+00 -1.05529062e-01 3.25529128e-01 7.20547587e-02 6.17768355e-02 3.01006913e-01 9.36322451e-01 -3.21574122e-01 2.14596242e-01 -5.68192303e-01 -2.95609772e-01 -6.16488159e-01 4.04516876e-01 4.11161244e-01 -5.56210339e-01 -4.94713575e-01 7.07915246e-01 -1.50499836e-01 -7.86765456e-01 4.72500533e-01 -9.26321819e-02 -1.68367490e-01 -2.10340083e-01 2.33527660e-01 6.43347025e-01 -2.18504399e-01 1.68871239e-01 -1.68442219e-01 3.01716328e-01 1.11465417e-02 -1.96301416e-01 1.31877935e+00 -3.35170865e-01 3.29446405e-01 5.53935528e-01 6.58636153e-01 -1.76790670e-01 -2.13785481e+00 -9.38708782e-02 -4.32872064e-02 -3.52020353e-01 1.18136294e-01 -1.08544016e+00 -9.39657271e-01 5.22546053e-01 6.55372918e-01 -2.26044372e-01 1.24849665e+00 -7.41406679e-01 8.82439435e-01 9.76153195e-01 8.15622211e-01 -1.82477069e+00 5.62190533e-01 7.26144195e-01 6.53446972e-01 -1.39117384e+00 8.07830244e-02 2.33774066e-01 -7.39159048e-01 1.25269544e+00 1.00650287e+00 -3.17754865e-01 2.95178682e-01 1.45529717e-01 -4.65673581e-02 3.72719258e-01 -1.16042852e+00 -2.27893487e-01 -2.00859740e-01 4.41808850e-01 -9.95096490e-02 -9.30853412e-02 -1.86114877e-01 6.24090075e-01 -1.05828643e-01 2.94838071e-01 1.97784364e-01 1.27075016e+00 -6.38461828e-01 -1.21396470e+00 -3.42739634e-02 1.64761320e-01 3.73787433e-02 3.00424814e-01 4.79322635e-02 7.42716730e-01 5.27268238e-02 8.70469809e-01 -1.71380013e-01 -5.30778229e-01 1.19166791e-01 -3.41417968e-01 5.63022673e-01 -1.75095543e-01 -8.34347069e-01 -4.99476679e-02 2.45579612e-02 -9.48380768e-01 -1.16134882e-01 -3.75212431e-01 -1.42243826e+00 -3.52733396e-02 -5.11802316e-01 3.59520495e-01 4.35498834e-01 9.09416616e-01 5.64534724e-01 6.82083964e-01 9.69368100e-01 -7.59232581e-01 -1.19765878e+00 -6.76771820e-01 -2.69920588e-01 -7.86179155e-02 5.21698475e-01 -1.10302579e+00 -2.71661162e-01 -3.44011992e-01]
[4.1002912521362305, 2.187426805496216]
e6beb651-171c-4137-961e-66f992ff55e4
multiple-imputation-using-chained-equations
null
null
https://doi.org/10.1002/sim.4067
https://onlinelibrary.wiley.com/doi/epdf/10.1002/sim.4067
Multiple imputation using chained equations: issues and guidance for practice
Multiple imputation by chained equations (MICE) is a flexible and practical approach to handling missing data. We describe the principles of the method and show how to impute categorical and quantitative variables, including skewed variables. We give guidance on how to specify the imputation model and how many imputations are needed. We describe the practical analysis of multiply imputed data, including model building and model checking. We stress the limitations of the method and discuss the possible pitfalls. We illustrate the ideas using a data set in mental health, giving Stata code fragments.
['Ian R. White', 'Patrick Royston', 'Angela M. Wood']
2010-11-30
null
null
null
statistics-in-medicine-304377399-2011-2010-11
['multivariate-time-series-imputation']
['time-series']
[ 2.42353693e-01 -8.53450447e-02 -6.93105102e-01 -1.19870460e+00 -6.99065685e-01 -3.22278380e-01 -4.12581354e-01 3.07090551e-01 -4.04177338e-01 1.47757697e+00 6.66230559e-01 -6.69231355e-01 -5.23083091e-01 -6.26137614e-01 -5.92697144e-01 -3.98218870e-01 -1.67880598e-02 7.36963034e-01 -8.15479815e-01 1.13495782e-01 2.92318821e-01 3.04682344e-01 -9.60915983e-01 6.63135409e-01 1.27414107e+00 1.04668111e-01 -4.40193802e-01 4.65987355e-01 -1.52622938e-01 1.07196164e+00 -3.33817422e-01 -7.77583778e-01 -2.35621348e-01 -4.95818943e-01 -5.96876383e-01 -3.87459278e-01 1.59349799e-01 -9.53515828e-01 2.55072951e-01 3.49451452e-01 7.05581784e-01 -2.72735953e-01 9.22113299e-01 -1.60452902e+00 -8.63859296e-01 8.37761641e-01 -5.98168194e-01 2.87709408e-03 6.38053775e-01 -1.79073572e-01 1.32489681e-01 -8.84937823e-01 7.82223821e-01 1.26126540e+00 1.44650948e+00 5.00068069e-01 -1.84033763e+00 -9.85646129e-01 -4.87616599e-01 -1.51957005e-01 -1.29234743e+00 -1.02111387e+00 -5.23796715e-02 -6.97153151e-01 8.61400425e-01 4.07163829e-01 4.76283222e-01 1.02442157e+00 4.77237403e-01 3.59480768e-01 1.28650427e+00 -3.03004086e-01 4.27184254e-01 -1.45121530e-01 6.60246730e-01 5.69257811e-02 7.12099969e-01 7.23449588e-02 -3.93541425e-01 -1.28288114e+00 9.47725594e-01 3.25318366e-01 3.78145844e-01 -2.86820561e-01 -1.14346611e+00 1.06613231e+00 -5.12678474e-02 -2.69542754e-01 -6.42034054e-01 5.28133400e-02 3.21982473e-01 2.09754944e-01 1.22629352e-01 -2.69312292e-01 -7.22523451e-01 -3.31331104e-01 -1.10563064e+00 3.30299616e-01 7.26328790e-01 9.98954833e-01 2.86868036e-01 -9.22080576e-02 -3.53228033e-01 1.16700757e+00 1.82693735e-01 5.72413146e-01 -3.01108867e-01 -1.33662701e+00 5.23955822e-01 2.31363595e-01 4.30518806e-01 -5.31755149e-01 -5.41761100e-01 7.38266930e-02 -1.17637479e+00 1.37204483e-01 6.70524657e-01 -6.30131364e-01 -8.46847534e-01 1.56749487e+00 1.24297351e-01 -2.96161562e-01 -1.69710740e-01 2.19240755e-01 6.11053288e-01 -2.77502656e-01 5.91747105e-01 -7.15033531e-01 1.21796560e+00 -8.56601074e-02 -1.21521628e+00 8.43765512e-02 6.79670572e-01 -5.89230776e-01 4.27354962e-01 3.10565501e-01 -1.62250030e+00 2.34756693e-02 1.75991133e-01 -3.79884541e-01 5.02568632e-02 -3.03456277e-01 9.40965593e-01 1.10453832e+00 -8.08604598e-01 2.46503636e-01 -1.15500951e+00 -4.11302656e-01 5.57944417e-01 6.53497934e-01 -6.54529214e-01 -2.05208927e-01 -8.35997283e-01 8.49068463e-01 -1.29913557e-02 9.32289585e-02 5.11747003e-02 -8.83046448e-01 -1.00660276e+00 5.89087531e-02 -4.51039612e-01 -1.19746685e+00 1.02391255e+00 -5.94813704e-01 -5.59645593e-01 7.46936202e-01 -9.35111225e-01 1.81045413e-01 5.84621787e-01 2.73236871e-01 -5.01722932e-01 -3.12669098e-01 4.75523978e-01 2.93978482e-01 2.28263717e-02 -9.36037481e-01 -3.82408381e-01 -7.11182237e-01 -7.82511234e-01 -9.98262838e-02 3.32026809e-01 6.99573815e-01 3.28014404e-01 -7.35661864e-01 3.69843423e-01 -4.17659342e-01 -5.28671563e-01 -1.68250114e-01 -2.22219840e-01 1.75339192e-01 -3.93951356e-01 -9.55712974e-01 1.48852432e+00 -1.92315614e+00 -5.19684613e-01 2.43329421e-01 4.12676595e-02 -6.85652614e-01 8.09597820e-02 7.32541978e-01 -4.13455009e-01 2.31033772e-01 -5.20567656e-01 -5.25366783e-01 -5.03897481e-02 3.81066293e-01 -1.15845509e-01 5.60552180e-01 -2.37988278e-01 1.06731915e+00 -6.17271483e-01 -6.43774748e-01 1.29422009e-01 6.17074490e-01 -7.62825787e-01 -2.80349720e-02 8.17194581e-01 6.33755863e-01 1.06928134e-02 1.19457269e+00 1.43974471e+00 1.48353666e-01 3.73033822e-01 5.77761412e-01 -4.15783614e-01 2.31052727e-01 -1.11562407e+00 1.16613746e+00 1.52792066e-01 -6.50500283e-02 6.03513122e-01 -4.25618201e-01 1.01224399e+00 1.23552613e-01 4.05500710e-01 -2.47966498e-01 -2.77608335e-01 5.45241833e-01 -1.47848874e-01 -6.93844914e-01 3.96916382e-02 -6.64354146e-01 -1.74212843e-01 4.23475921e-01 -3.79188061e-01 2.24402145e-01 -1.64059624e-01 -1.52809381e-01 1.15839124e+00 -1.29886146e-03 7.87274003e-01 -2.59981811e-01 -9.86294970e-02 1.53856084e-01 1.09638417e+00 1.34443784e+00 -1.31865144e-01 7.16941655e-01 6.60976529e-01 -6.79908812e-01 -7.62359321e-01 -1.27868319e+00 -1.09950161e+00 9.34575379e-01 -7.09584117e-01 -1.02340236e-01 -2.21870661e-01 -4.00943682e-02 7.26046205e-01 9.75636601e-01 -8.30369532e-01 4.49972540e-01 -2.35069692e-01 -1.23888254e+00 6.79225206e-01 6.77151918e-01 -2.20889390e-01 -1.06742084e+00 -3.63173962e-01 3.55425835e-01 -4.79379296e-01 -2.68041670e-01 8.77376944e-02 3.96472275e-01 -1.44474065e+00 -9.40600514e-01 -5.02032399e-01 -5.58808684e-01 8.30855370e-01 -1.26639560e-01 1.21906221e+00 2.43896306e-01 -1.48362713e-02 1.34452343e-01 -9.59220994e-03 -3.32873553e-01 -3.37853134e-01 -2.46053711e-01 1.22861929e-01 -5.14853656e-01 1.09009135e+00 -6.46418810e-01 -4.58329529e-01 -3.55880149e-02 -6.70719981e-01 -4.65739955e-04 2.62378603e-01 1.01379907e+00 3.67456585e-01 -5.00054240e-01 6.79208934e-01 -1.35415256e+00 7.36789167e-01 -1.10533321e+00 -2.27584392e-01 2.67056257e-01 -5.15091300e-01 -3.82426292e-01 1.77941732e-02 -4.70899092e-03 -9.06451643e-01 1.31814957e-01 -4.53270167e-01 4.48411644e-01 -6.08980894e-01 6.34849072e-01 -5.09092165e-03 9.91995931e-02 4.74956125e-01 -1.97563186e-01 2.57595241e-01 -7.76949406e-01 -1.40692949e-01 9.52216744e-01 4.34338272e-01 -6.21636808e-01 -5.25548495e-02 4.25050855e-01 2.75582192e-03 -4.20559347e-01 1.38129413e-01 -2.26249367e-01 -7.61017621e-01 3.65669459e-01 6.10760331e-01 -9.43909049e-01 -1.36042464e+00 2.53453285e-01 -9.95041430e-01 -3.31515431e-01 -7.90273324e-02 9.74021971e-01 -7.94611633e-01 -7.58241769e-03 -6.76833034e-01 -1.15363336e+00 -1.23954765e-01 -9.31572020e-01 6.17438912e-01 -1.37459442e-01 -8.07610333e-01 -9.70971465e-01 4.56823945e-01 4.07893151e-01 5.43303311e-01 6.59156859e-01 1.03808177e+00 -4.83127892e-01 2.54281610e-01 5.07790335e-02 -1.41284794e-01 -5.14937103e-01 3.38640928e-01 1.34499565e-01 -4.71194357e-01 -2.13816151e-01 3.71196158e-02 -2.97843993e-01 5.22511780e-01 1.06136274e+00 1.17461777e+00 -3.96596402e-01 -3.67342919e-01 8.81130338e-01 1.31455398e+00 4.01313789e-02 8.75108421e-01 1.63848758e-01 2.90728331e-01 8.21098745e-01 1.97452784e-01 7.39051044e-01 9.40872490e-01 3.90932500e-01 9.65270922e-02 -1.96522519e-01 5.48525512e-01 -1.55061081e-01 4.67966422e-02 3.51671427e-01 -1.15620740e-01 4.18740213e-01 -1.14890587e+00 6.30479097e-01 -1.97977209e+00 -1.21464860e+00 -1.00213408e+00 2.30432534e+00 9.60551083e-01 -3.41033310e-01 3.83822024e-01 2.54769661e-02 7.15595126e-01 -8.61913621e-01 -3.58570963e-01 -1.19546866e+00 -2.37669870e-01 1.10840000e-01 5.16645491e-01 7.07630575e-01 -6.00960970e-01 9.30686146e-02 9.44443703e+00 -1.51519561e-02 1.90498848e-02 8.22490007e-02 7.70776689e-01 -7.37768188e-02 -5.61431348e-01 2.79433489e-01 -6.40988171e-01 7.65982270e-01 9.22160685e-01 1.18902840e-01 4.29814965e-01 4.78097111e-01 7.86452234e-01 -4.12611544e-01 -9.14299965e-01 7.51712441e-01 -3.12658012e-01 -1.00335193e+00 -3.52821767e-01 2.03572243e-01 5.03947258e-01 -8.28253329e-02 -2.69626319e-01 2.52220869e-01 6.14111662e-01 -1.38319325e+00 1.49881124e-01 8.54464412e-01 1.18915498e+00 -7.53175795e-01 8.89248431e-01 1.91630796e-01 -3.12195957e-01 -2.77999789e-01 -5.54256558e-01 -8.74151349e-01 4.63226557e-01 8.09357703e-01 -5.07008553e-01 3.09682012e-01 9.20030415e-01 4.34483200e-01 -3.45316529e-01 1.40257502e+00 5.25457524e-02 7.22360849e-01 -3.28412503e-01 7.23900914e-01 -4.82515603e-01 -5.30263245e-01 -1.64743960e-01 1.23235726e+00 5.76206088e-01 4.94295895e-01 -5.27387023e-01 9.35529292e-01 3.05592775e-01 -5.57266958e-02 -5.74201941e-01 5.70542514e-01 7.20757604e-01 7.36360908e-01 -3.41985285e-01 -9.41131040e-02 -7.42930353e-01 5.33508778e-01 3.23089391e-01 4.59642291e-01 -4.48301852e-01 -1.26188383e-01 6.17157459e-01 2.38686249e-01 -1.35871351e-01 8.93332437e-03 -1.20233214e+00 -1.23820066e+00 -2.95500129e-01 -9.89579082e-01 1.01225853e+00 -8.70880008e-01 -1.64030719e+00 -2.86437482e-01 3.35278630e-01 -7.64703810e-01 -3.13002884e-01 8.88389722e-03 -3.03355783e-01 1.30262935e+00 -7.80764461e-01 -1.00604892e+00 -1.54599279e-01 7.11120486e-01 -1.79017872e-01 2.53302217e-01 1.25288320e+00 2.98100382e-01 -4.34981316e-01 7.90589750e-01 7.30460227e-01 1.49735361e-01 1.06660795e+00 -1.15790677e+00 2.85492778e-01 2.14735135e-01 -9.31281686e-01 1.34826660e+00 7.15002000e-01 -1.24764550e+00 -1.08979809e+00 -5.16084194e-01 1.79558647e+00 -7.69894063e-01 1.84808314e-01 -3.73312473e-01 -8.62252772e-01 1.30668044e+00 2.17303205e-02 -2.63144970e-01 1.59769547e+00 7.94697404e-01 -3.54179479e-02 9.67671126e-02 -1.91339850e+00 2.84877121e-01 8.20755303e-01 1.35436915e-02 -8.20289254e-01 1.61368743e-01 1.06718548e-01 -3.79824996e-01 -1.23660064e+00 3.74493122e-01 1.26374710e+00 -1.20397949e+00 9.44914937e-01 -9.83386636e-01 3.46317083e-01 1.65382912e-03 -5.85660934e-02 -6.93828940e-01 -8.60482275e-01 -4.45743471e-01 1.24435268e-01 1.26167727e+00 2.46075392e-01 -8.77068996e-01 6.63835943e-01 1.70259106e+00 1.13439359e-01 -2.88561016e-01 -1.19850945e+00 -4.06220973e-01 5.36551535e-01 -4.41283137e-01 9.85365212e-01 1.38199115e+00 4.69766229e-01 -4.01066333e-01 -6.09978676e-01 -1.10420845e-01 1.07174134e+00 -2.76551917e-02 6.62245035e-01 -1.13029206e+00 -9.58520025e-02 3.79872769e-01 -9.62300077e-02 -1.26203910e-01 -1.72749609e-01 -5.64400792e-01 -6.89911783e-01 -1.63801181e+00 9.05322969e-01 -6.86631024e-01 -3.25767219e-01 1.01068175e+00 -1.50851637e-01 2.53012449e-01 -8.81845579e-02 1.97657123e-01 -6.94932863e-02 -8.12990367e-02 6.34425282e-01 2.02134416e-01 -5.11874378e-01 2.97146142e-01 -1.22317672e+00 3.98506850e-01 9.93928492e-01 -9.84635711e-01 1.35580814e-02 -7.66041756e-01 5.05742788e-01 7.07805932e-01 4.71297860e-01 -2.40484938e-01 1.77390233e-01 -7.64622152e-01 1.21131575e+00 -9.42624867e-01 1.01639396e-02 -1.05424428e+00 1.13643169e+00 5.35878181e-01 -2.87646592e-01 4.43087339e-01 1.34307235e-01 -9.77658182e-02 2.10628808e-01 -3.03247631e-01 4.80870634e-01 3.27164270e-02 2.04154104e-01 -1.33476719e-01 -5.13526559e-01 -4.60445046e-01 6.99524641e-01 -4.49280500e-01 -3.92221361e-01 -4.13973480e-01 -1.37376082e+00 5.43134212e-01 8.42001259e-01 -3.72697823e-02 6.44748151e-01 -1.57435870e+00 -1.05397511e+00 6.56507730e-01 1.36070391e-02 -5.61953545e-01 3.81445587e-01 1.35527849e+00 -5.05648315e-01 1.81123689e-01 -6.64280951e-01 -7.34043643e-02 -1.28162694e+00 5.68651676e-01 6.40471205e-02 3.31308872e-01 -5.12674749e-01 6.16662428e-02 -2.05059022e-01 -8.19544494e-01 2.60723531e-02 -8.90052468e-02 -6.22169906e-03 -1.05156131e-01 7.00651705e-01 9.59357023e-01 -1.49826393e-01 -3.91494393e-01 -1.01217008e+00 2.43392974e-01 2.23462254e-01 -1.54738445e-02 1.67308366e+00 -6.75809264e-01 -1.00011969e+00 4.81029958e-01 7.36618876e-01 2.33277649e-01 -7.77735531e-01 4.33898568e-01 -1.25008628e-01 -5.95984459e-01 -7.12835491e-01 -1.06071281e+00 -1.87154651e-01 4.85287100e-01 3.10619086e-01 -1.78793207e-01 1.17344487e+00 -4.42992985e-01 1.71097852e-02 -1.39541313e-01 5.64890563e-01 -7.63430774e-01 -1.38727975e+00 2.86193609e-01 6.33967221e-01 -1.10398602e+00 2.71542400e-01 -2.70528942e-01 -4.91651714e-01 1.01688898e+00 1.15623541e-01 1.74439117e-01 5.68014503e-01 5.62790334e-01 6.05979338e-02 -2.70293225e-02 -6.38391316e-01 5.31234860e-01 -4.09611464e-01 1.16600919e+00 8.93470466e-01 2.99663246e-01 -9.67220366e-01 1.07464421e+00 -3.03493112e-01 8.35969090e-01 8.58479798e-01 1.01406527e+00 1.08131617e-01 -1.50342119e+00 -1.13689601e+00 1.12050843e+00 -7.00961649e-01 -1.19196780e-01 -3.50537837e-01 7.02170491e-01 2.66531914e-01 1.35205579e+00 1.50960296e-01 2.87342489e-01 1.86985657e-01 2.89215595e-01 2.34124914e-01 -3.19888085e-01 -6.98037088e-01 1.06491499e-01 2.11418644e-01 -3.44580978e-01 -4.06126380e-01 -1.29476714e+00 -9.34796453e-01 -1.16723096e+00 -2.58699823e-02 2.73441464e-01 3.47809911e-01 5.44303894e-01 4.65446115e-01 -4.22395915e-02 4.19483811e-01 -2.14761853e-01 -3.29474777e-01 -7.52790570e-01 -8.21416080e-01 2.55010456e-01 6.62612021e-01 -4.74171549e-01 -3.26309651e-01 9.70182288e-03]
[7.835690975189209, 4.913211822509766]
e509197d-3b8d-43c4-83e6-c3a310a7d12d
fighting-noise-and-imbalance-in-action-unit
2303.02994
null
https://arxiv.org/abs/2303.02994v1
https://arxiv.org/pdf/2303.02994v1.pdf
Fighting noise and imbalance in Action Unit detection problems
Action Unit (AU) detection aims at automatically caracterizing facial expressions with the muscular activations they involve. Its main interest is to provide a low-level face representation that can be used to assist higher level affective computing tasks learning. Yet, it is a challenging task. Indeed, the available databases display limited face variability and are imbalanced toward neutral expressions. Furthermore, as AU involve subtle face movements they are difficult to annotate so that some of the few provided datapoints may be mislabeled. In this work, we aim at exploiting label smoothing ability to mitigate noisy examples impact by reducing confidence [1]. However, applying label smoothing as it is may aggravate imbalance-based pre-existing under-confidence issue and degrade performance. To circumvent this issue, we propose Robin Hood Label Smoothing (RHLS). RHLS principle is to restrain label smoothing confidence reduction to the majority class. In that extent, it alleviates both the imbalance-based over-confidence issue and the negative impact of noisy majority class examples. From an experimental standpoint, we show that RHLS provides a free performance improvement in AU detection. In particular, by applying it on top of a modern multi-task baseline we get promising results on BP4D and outperform state-of-the-art methods on DISFA.
['Kevin Bailly', 'Arnaud Dapogny', 'Gauthier Tallec']
2023-03-06
null
null
null
null
['action-unit-detection']
['computer-vision']
[ 1.99674413e-01 3.09549093e-01 -1.62386551e-01 -5.42604744e-01 -7.77684093e-01 -3.02071035e-01 2.67934293e-01 -4.60314751e-02 -2.73588598e-01 7.24895000e-01 3.07347775e-01 6.07656538e-01 2.52941310e-01 -3.63807023e-01 -3.87132853e-01 -9.82760429e-01 1.42313004e-01 -1.34242207e-01 -3.73726010e-01 -2.27851644e-01 -2.06660122e-01 4.58253145e-01 -1.66884589e+00 6.48949623e-01 5.62144637e-01 1.30264425e+00 -4.90867496e-01 -1.28909603e-01 2.27202699e-01 7.54016161e-01 -7.53816664e-01 -6.03577673e-01 1.98079765e-01 -3.28603685e-01 -4.78018135e-01 7.18678311e-02 6.29404008e-01 -2.41804510e-01 2.86244780e-01 1.22279119e+00 8.22831035e-01 1.02546737e-01 7.09199429e-01 -1.51156318e+00 -2.49491602e-01 1.67888358e-01 -1.07837784e+00 -1.00586295e-01 3.25121373e-01 -8.15517530e-02 9.31435585e-01 -1.22659051e+00 5.61599493e-01 1.44574809e+00 6.37364984e-01 1.11670434e+00 -1.40104783e+00 -9.67626393e-01 2.52807111e-01 1.07110679e-01 -1.49874318e+00 -6.90658510e-01 9.71975863e-01 -3.21305871e-01 6.18442893e-01 3.00270140e-01 4.51057732e-01 1.49303102e+00 -1.23278759e-01 8.34404945e-01 1.29806387e+00 -2.45270580e-01 2.45010898e-01 3.74856710e-01 -3.34539637e-02 5.03885925e-01 -1.76207617e-01 -2.34589055e-01 -8.07184398e-01 -1.96074858e-01 2.78257251e-01 -3.17641348e-01 -2.04473048e-01 -9.59335640e-02 -5.66901326e-01 6.75297797e-01 5.15128911e-01 3.09071481e-01 -4.98060524e-01 6.04409277e-02 7.36208200e-01 7.42536858e-02 7.93251157e-01 2.71466285e-01 -2.58031368e-01 -5.51016219e-02 -9.49630082e-01 1.61153212e-01 4.03135538e-01 4.60110068e-01 6.83037281e-01 9.56901833e-02 -6.52720928e-01 1.21929550e+00 2.00763326e-02 2.24628687e-01 4.70408171e-01 -9.26502407e-01 1.96029283e-02 6.47225976e-01 -1.37230173e-01 -1.05830193e+00 -5.70860267e-01 -3.64581317e-01 -1.01874578e+00 5.91099262e-01 3.53316575e-01 -4.02787894e-01 -7.12762415e-01 2.27878547e+00 4.22753125e-01 9.81380418e-02 -1.90266550e-01 1.19025183e+00 8.51265907e-01 3.38255227e-01 4.99645978e-01 -6.63631380e-01 1.57411456e+00 -7.07857370e-01 -1.08655798e+00 -3.68378967e-01 6.57854378e-01 -6.70703828e-01 1.04463077e+00 5.52823365e-01 -9.71765697e-01 -5.11698186e-01 -8.90696526e-01 1.99499324e-01 -1.93748623e-01 4.19197023e-01 7.36791074e-01 7.55829513e-01 -9.30079818e-01 4.84196097e-01 -5.04485130e-01 -2.02339694e-01 8.09717953e-01 3.37173641e-01 -7.01377988e-01 2.94643253e-01 -1.22731614e+00 9.62415636e-01 -5.51404012e-03 3.45180094e-01 -4.59121287e-01 -8.02868724e-01 -7.53189802e-01 -3.15474570e-02 5.13042569e-01 -1.59540877e-01 1.10642660e+00 -1.42802298e+00 -1.36475110e+00 1.16206014e+00 -1.98171526e-01 -1.15617923e-01 5.79385102e-01 -3.21421176e-01 -4.69052762e-01 6.52832985e-02 2.65631936e-02 8.83668184e-01 1.15734243e+00 -1.19740224e+00 -1.18233889e-01 -6.32719398e-01 -1.77280501e-01 3.58179778e-01 -6.30368650e-01 1.80469513e-01 -2.19839066e-01 -8.26939642e-01 -1.56972870e-01 -1.03757668e+00 2.99999490e-02 2.19883278e-01 -2.16560796e-01 -4.86005723e-01 8.57218683e-01 -3.71529311e-01 1.15604794e+00 -2.45034981e+00 5.61673706e-03 1.58285379e-01 2.04744652e-01 4.31347609e-01 -9.88675207e-02 -1.01767644e-01 -4.22675967e-01 4.49572206e-02 -4.38716598e-02 -5.64099193e-01 -2.21923701e-02 -9.39932186e-03 -5.17866239e-02 5.62458992e-01 6.32009089e-01 6.31625652e-01 -6.26698792e-01 -5.42018771e-01 1.03431262e-01 5.81106305e-01 -6.79360449e-01 2.47583047e-01 -8.30878988e-02 5.29339015e-01 -1.02039017e-01 8.72216880e-01 7.86289692e-01 2.57001370e-01 4.92510423e-02 -6.65221810e-01 2.03791827e-01 -2.34147653e-01 -1.28297806e+00 1.52758932e+00 -4.01547819e-01 4.48368311e-01 3.47848475e-01 -9.69119966e-01 1.03863895e+00 4.71390516e-01 6.38169348e-01 -5.01370728e-01 4.39137310e-01 -2.99557708e-02 -9.19035152e-02 -4.17916477e-01 1.47887737e-01 -4.40393329e-01 3.45776863e-02 1.85522899e-01 2.67413240e-02 5.09558171e-02 -1.06981799e-01 3.14398967e-02 7.57444561e-01 1.36286825e-01 3.92076552e-01 -2.23683819e-01 5.99184930e-01 -6.29636705e-01 8.46722662e-01 2.06564859e-01 -7.08173275e-01 5.02277255e-01 7.24444985e-01 -7.30294064e-02 -3.83483529e-01 -9.05790508e-01 -3.62315655e-01 1.28418255e+00 -2.47861519e-01 -4.63135421e-01 -1.03202677e+00 -8.05813193e-01 -2.82535702e-02 4.84828055e-01 -9.74524319e-01 -4.83231753e-01 -2.03155965e-01 -1.07029736e+00 6.79951966e-01 5.18491507e-01 4.07224000e-01 -1.22830904e+00 -3.61741036e-01 2.36023907e-02 -3.27623397e-01 -1.13170183e+00 -3.92425358e-01 2.10341766e-01 -2.40437627e-01 -8.15789223e-01 -6.46035552e-01 -4.49251831e-01 6.48234367e-01 -8.38560462e-02 1.00275898e+00 -1.76721737e-01 -5.10888755e-01 9.89157781e-02 -2.83433110e-01 -6.46956503e-01 -7.67081305e-02 -4.18960631e-01 3.84708822e-01 5.37603438e-01 7.23050773e-01 -5.61682463e-01 -6.94136024e-01 2.42949575e-01 -6.51004672e-01 -2.94408083e-01 4.03647870e-01 8.16827178e-01 4.99637723e-01 -2.76277721e-01 1.06744778e+00 -7.54536629e-01 6.70612812e-01 -3.67080867e-01 7.74894208e-02 -7.76465004e-03 -4.61773634e-01 -3.64754587e-01 3.20301533e-01 -5.51769137e-01 -1.28650212e+00 2.08217576e-01 -2.49942005e-01 -7.17363060e-01 -2.98425168e-01 1.01608895e-01 -2.16360673e-01 -1.22108795e-01 7.87236333e-01 -3.57262313e-01 1.74597800e-01 -1.52159229e-01 2.34092593e-01 8.36231470e-01 2.94483483e-01 -3.97112280e-01 2.14515850e-01 5.02764523e-01 8.98201615e-02 -8.24884832e-01 -1.26445973e+00 -3.98109496e-01 -5.23529828e-01 -6.66020691e-01 8.74732077e-01 -1.13815010e+00 -9.42072153e-01 2.86521167e-01 -1.04254842e+00 -7.73390383e-02 -9.14705992e-02 1.86038852e-01 -4.66152549e-01 2.06004858e-01 -5.99499702e-01 -1.08049321e+00 -5.27894676e-01 -1.01426494e+00 1.14315915e+00 2.57346421e-01 -7.50076354e-01 -5.00639021e-01 -6.66626841e-02 3.80352646e-01 1.38082638e-01 6.34630322e-01 5.48851132e-01 -4.81351078e-01 3.81363720e-01 -1.29743725e-01 -3.43380839e-01 7.84619749e-01 2.39789769e-01 1.02893889e-01 -1.62182474e+00 -1.75529554e-01 1.37469530e-01 -9.01200950e-01 6.71496570e-01 3.37531179e-01 1.44741797e+00 -1.28661484e-01 -7.64464065e-02 2.78401017e-01 9.16303754e-01 -2.23679543e-01 4.84982967e-01 -1.07317232e-01 5.13085365e-01 9.97839928e-01 8.94547880e-01 6.99062109e-01 -4.92835827e-02 9.80883598e-01 4.65864420e-01 -4.22194451e-01 -1.15966611e-01 1.76440075e-01 4.86537814e-01 1.85276881e-01 -1.63193762e-01 1.76196858e-01 -4.50720787e-01 2.98765242e-01 -1.90626717e+00 -9.00527120e-01 -1.41432300e-01 1.96727550e+00 1.16163564e+00 -3.79215516e-02 3.46652269e-01 1.89074606e-01 6.44181252e-01 1.11766882e-01 -4.66492712e-01 -5.90359688e-01 -2.41790026e-01 2.60040700e-01 -7.97789916e-02 1.09973662e-01 -1.23980618e+00 9.37984586e-01 4.90240192e+00 1.00419343e+00 -1.11249197e+00 1.65311471e-01 8.79112899e-01 -4.42769945e-01 3.05208355e-01 -7.00423360e-01 -7.65329540e-01 4.09530848e-01 6.52665317e-01 6.79919794e-02 1.69592246e-01 1.06883681e+00 4.14083451e-01 -2.34741181e-01 -1.15873134e+00 1.40110242e+00 2.52309412e-01 -6.48845196e-01 -2.93735445e-01 -2.22146958e-01 4.94290560e-01 -3.86357635e-01 1.16637982e-01 4.57325906e-01 -3.63888711e-01 -1.13520443e+00 6.51175916e-01 3.31053585e-01 8.79397571e-01 -8.82179677e-01 8.15755010e-01 9.47254375e-02 -9.42735374e-01 9.32018980e-02 -2.68942922e-01 -8.94143134e-02 -1.83887035e-02 7.89200842e-01 -5.96212626e-01 1.39854416e-01 8.06150854e-01 5.64039290e-01 -4.29268867e-01 4.40549672e-01 -3.94276500e-01 2.90928572e-01 -2.11677194e-01 1.09166503e-01 5.47357388e-02 -1.35673642e-01 4.13471818e-01 1.34455240e+00 2.09380314e-02 2.43400574e-01 8.23898688e-02 6.82489932e-01 -3.55784297e-01 3.45747799e-01 -5.94413280e-01 2.32651696e-01 9.68733281e-02 1.76346910e+00 -4.50757504e-01 -1.51266381e-01 -3.65799308e-01 1.21684492e+00 6.53647244e-01 2.05853745e-01 -9.59738970e-01 -4.38906290e-02 1.00140500e+00 -9.73414928e-02 -2.02902272e-01 3.78080904e-01 -2.66089857e-01 -1.04674232e+00 1.14010647e-01 -9.51373577e-01 4.34287906e-01 -6.49250865e-01 -1.38243282e+00 6.06019080e-01 -3.18909198e-01 -1.06991494e+00 -1.24445163e-01 -4.26355779e-01 -5.26884079e-01 6.56326592e-01 -1.35636795e+00 -1.06605172e+00 -4.76652920e-01 6.84222877e-01 5.23360968e-01 2.59540707e-01 1.03284729e+00 5.65338254e-01 -7.45390475e-01 1.00589216e+00 -5.08651078e-01 -9.69878584e-03 1.32500648e+00 -1.09430087e+00 -5.27447045e-01 5.16375363e-01 -2.20012479e-03 3.58186483e-01 7.50004232e-01 -4.49841231e-01 -8.45826805e-01 -1.09492135e+00 5.25152624e-01 -2.99517930e-01 4.74775940e-01 -5.70833147e-01 -1.04747009e+00 3.62908334e-01 5.70901064e-03 3.50928724e-01 9.43129539e-01 2.17354476e-01 -4.50739145e-01 -2.67151982e-01 -1.42066312e+00 6.82684064e-01 1.02724981e+00 -5.09924114e-01 -2.77895898e-01 3.43595207e-01 -1.11352270e-02 -2.73767173e-01 -9.29544389e-01 7.27465749e-01 5.63780129e-01 -1.22871959e+00 7.48089731e-01 -4.47836995e-01 4.03562337e-01 -2.14740112e-02 7.16187581e-02 -1.44334412e+00 -2.04035878e-01 -5.35460234e-01 -1.67501926e-01 1.72776365e+00 2.28436351e-01 -1.23845011e-01 9.15290236e-01 8.30743551e-01 1.14684463e-01 -9.54689085e-01 -1.04294086e+00 -5.60418844e-01 -1.29944474e-01 -3.57875884e-01 7.35229999e-02 1.12776220e+00 3.03154975e-01 3.67598742e-01 -6.12116814e-01 -9.88077074e-02 5.82343280e-01 -2.15477422e-01 5.13077140e-01 -1.08536601e+00 1.54638797e-01 -4.03222173e-01 -2.97544181e-01 -2.77302235e-01 5.48403442e-01 -6.99269652e-01 2.04581350e-01 -8.06424439e-01 9.05277282e-02 -1.94702119e-01 -4.21207249e-01 8.75464916e-01 -3.65999579e-01 8.14329445e-01 1.12147510e-01 -1.57126546e-01 -4.69871312e-01 7.66376138e-01 1.03465021e+00 6.52548596e-02 -7.64664114e-02 -8.67283866e-02 -7.02351213e-01 9.68909442e-01 8.36833894e-01 -3.67094278e-01 -2.90713847e-01 1.61492899e-01 5.25866188e-02 -3.46156359e-01 3.25369418e-01 -8.15029383e-01 -1.63520962e-01 7.46158361e-02 5.17699480e-01 -1.74747318e-01 7.72818565e-01 -7.64857292e-01 -1.79774001e-01 2.96769083e-01 -2.13044107e-01 -4.32674468e-01 3.70142490e-01 3.33834916e-01 -2.30489790e-01 -3.82282212e-02 1.28646362e+00 -1.44203734e-02 -5.65576077e-01 1.62095875e-01 -4.01026756e-01 1.15467580e-02 1.23426938e+00 -5.58181144e-02 -1.19478881e-01 -3.40599298e-01 -8.77725303e-01 2.32838511e-01 1.73176304e-01 5.91316581e-01 2.86948532e-01 -1.43063903e+00 -5.56512237e-01 2.69875675e-01 3.36555123e-01 -3.77142787e-01 4.41794068e-01 1.08853376e+00 2.88385928e-01 -1.31423408e-02 -3.45116079e-01 -4.95915383e-01 -1.67761266e+00 2.66237199e-01 2.89040715e-01 1.88638251e-02 -1.75289363e-01 1.18934035e+00 3.46252382e-01 -1.41554818e-01 5.15661716e-01 1.67842284e-01 -4.76635307e-01 7.91240573e-01 7.18345761e-01 2.56146759e-01 3.98077041e-01 -6.86794639e-01 -6.34123862e-01 4.89180356e-01 -1.38029367e-01 2.21863568e-01 1.23277330e+00 -3.35404649e-02 -2.04075560e-01 4.50966746e-01 1.09549761e+00 2.56427322e-02 -1.19026542e+00 -5.86068258e-03 -7.19614103e-02 -3.49312842e-01 1.76767409e-01 -7.79183745e-01 -1.13437533e+00 8.61346185e-01 8.34311426e-01 -6.90757036e-02 1.34167767e+00 2.73389723e-02 3.69776875e-01 9.62922648e-02 4.76422682e-02 -1.47613525e+00 3.16801697e-01 -1.11942543e-02 1.12253916e+00 -1.50832748e+00 -7.81776905e-02 -6.33254409e-01 -9.75207806e-01 8.72711539e-01 1.11783016e+00 1.10266156e-01 5.55454731e-01 2.98944652e-01 2.57831305e-01 -2.75686026e-01 -7.09265947e-01 -3.19620609e-01 3.05014521e-01 4.62650120e-01 6.51947975e-01 -9.73914005e-03 -4.91118908e-01 9.20266688e-01 -9.13673732e-03 -8.37411955e-02 2.27133989e-01 5.33006847e-01 -1.27314419e-01 -8.52771640e-01 -3.28306496e-01 4.96341944e-01 -8.45570862e-01 1.21596441e-01 -5.47052264e-01 5.61838686e-01 3.40610027e-01 9.79781628e-01 3.75459716e-02 -2.43294507e-01 4.67483014e-01 4.05751377e-01 4.45638835e-01 -5.92508256e-01 -7.19230533e-01 2.24943131e-01 2.74488628e-01 -7.64400959e-01 -5.72440982e-01 -6.47975028e-01 -1.17898154e+00 -1.19512513e-01 -3.30443442e-01 -8.02873001e-02 5.58166623e-01 8.24591637e-01 2.66858608e-01 4.20475066e-01 6.33846283e-01 -8.82078230e-01 -5.22951722e-01 -1.14561367e+00 -6.82897627e-01 8.36810410e-01 2.34144956e-01 -1.04980695e+00 -5.67270219e-01 -1.50880262e-01]
[13.613018035888672, 1.6957881450653076]
ef48d4f2-8336-4b54-8158-ab4aa65d5dec
harnessing-spatial-homogeneity-of
2007.11899
null
https://arxiv.org/abs/2007.11899v1
https://arxiv.org/pdf/2007.11899v1.pdf
Harnessing spatial homogeneity of neuroimaging data: patch individual filter layers for CNNs
Neuroimaging data, e.g. obtained from magnetic resonance imaging (MRI), is comparably homogeneous due to (1) the uniform structure of the brain and (2) additional efforts to spatially normalize the data to a standard template using linear and non-linear transformations. Convolutional neural networks (CNNs), in contrast, have been specifically designed for highly heterogeneous data, such as natural images, by sliding convolutional filters over different positions in an image. Here, we suggest a new CNN architecture that combines the idea of hierarchical abstraction in neural networks with a prior on the spatial homogeneity of neuroimaging data: Whereas early layers are trained globally using standard convolutional layers, we introduce for higher, more abstract layers patch individual filters (PIF). By learning filters in individual image regions (patches) without sharing weights, PIF layers can learn abstract features faster and with fewer samples. We thoroughly evaluated PIF layers for three different tasks and data sets, namely sex classification on UK Biobank data, Alzheimer's disease detection on ADNI data and multiple sclerosis detection on private hospital data. We demonstrate that CNNs using PIF layers result in higher accuracies, especially in low sample size settings, and need fewer training epochs for convergence. To the best of our knowledge, this is the first study which introduces a prior on brain MRI for CNN learning.
['Jan Philipp Albrecht', 'Martin Weygandt', 'Kerstin Ritter', 'Friedemann Paul', 'Fabian Eitel']
2020-07-23
null
null
null
null
['alzheimer-s-disease-detection']
['medical']
[ 3.81176680e-01 4.68559027e-01 1.40291050e-01 -7.96301126e-01 -4.70455945e-01 -2.03014284e-01 7.08375514e-01 2.20213234e-01 -9.73940611e-01 7.07987309e-01 4.30378318e-01 -7.48256743e-02 -1.31486997e-01 -5.90401590e-01 -8.90952706e-01 -4.59764212e-01 -6.43853724e-01 5.49775064e-01 2.69572645e-01 5.52242212e-02 -7.23674148e-02 7.37614989e-01 -1.25042582e+00 5.75450778e-01 4.81276453e-01 1.13121855e+00 1.25175221e-02 5.40390790e-01 2.40114376e-01 5.85609376e-01 -4.86436129e-01 -2.22663105e-01 3.29063714e-01 -2.33457237e-01 -8.69548380e-01 -4.23598439e-02 9.34732914e-01 -3.90312999e-01 -3.80481839e-01 9.82547939e-01 7.39896715e-01 -2.24557981e-01 6.85104191e-01 -8.79202962e-01 -8.35048735e-01 8.09635878e-01 -5.30665934e-01 6.47816002e-01 -4.10635799e-01 1.85367435e-01 8.52910221e-01 -7.31711686e-01 7.79180467e-01 1.22298479e+00 1.14429605e+00 6.12161756e-01 -1.70056200e+00 -7.13728011e-01 6.73017427e-02 5.04556969e-02 -1.14927745e+00 -4.13506567e-01 1.69917777e-01 -6.03517115e-01 1.06135654e+00 1.50152117e-01 6.76022887e-01 1.18232012e+00 6.31597340e-01 5.74378490e-01 1.30064368e+00 -2.71841921e-02 1.71236783e-01 -1.58670247e-01 3.68561894e-01 4.91309434e-01 2.68013567e-01 -6.20429926e-02 -1.41414374e-01 -2.24932626e-01 1.07767344e+00 2.11724207e-01 -7.34086037e-02 -2.76664913e-01 -1.48531234e+00 8.83507013e-01 9.41638112e-01 4.73973453e-01 -6.57832921e-01 1.06296353e-01 6.06127977e-01 2.82324553e-01 4.43077773e-01 3.85977089e-01 -5.05749643e-01 5.77424705e-01 -1.26125193e+00 3.20857197e-01 5.02728581e-01 5.93025863e-01 4.81933028e-01 -4.62424383e-02 -3.30736130e-01 8.63100410e-01 -1.70049518e-01 8.10801089e-02 9.17121470e-01 -7.90915847e-01 2.64181197e-01 3.14304590e-01 -4.93991673e-01 -6.25234425e-01 -1.05139065e+00 -7.53447354e-01 -1.22804892e+00 5.04671574e-01 5.54219484e-01 -2.42083758e-01 -1.37660420e+00 1.82458973e+00 -5.62279895e-02 -7.43883327e-02 -2.91546792e-01 8.31975996e-01 7.99762070e-01 8.85204040e-03 2.28803635e-01 2.85325229e-01 1.76984906e+00 -5.54868817e-01 -3.95901322e-01 -3.32632720e-01 4.50566679e-01 -2.83702910e-01 9.12267745e-01 2.81517982e-01 -1.16139424e+00 -4.11912084e-01 -1.08364999e+00 -3.59432012e-01 -6.14958823e-01 -2.98785102e-02 6.97344959e-01 6.14792883e-01 -1.48917627e+00 6.08597934e-01 -8.95519555e-01 -3.78062040e-01 1.16107738e+00 6.53847396e-01 -6.58570945e-01 -6.13520201e-03 -1.18009496e+00 1.04229832e+00 2.61745661e-01 4.86042798e-02 -6.80346191e-01 -1.27816772e+00 -7.89515972e-01 2.19194829e-01 -7.76608884e-02 -7.87799120e-01 1.11728632e+00 -1.21003604e+00 -9.84789610e-01 1.02979910e+00 2.22436309e-01 -1.02124536e+00 5.46106160e-01 1.04243144e-01 -2.36461222e-01 2.20147118e-01 8.77961591e-02 1.21637511e+00 7.44219303e-01 -7.64004827e-01 -3.21380258e-01 -6.56333208e-01 -1.47528917e-01 -3.40843946e-01 -2.98675597e-01 2.55908281e-01 2.94759631e-01 -7.78348446e-01 2.00599849e-01 -9.31088150e-01 -4.60234374e-01 1.67072147e-01 -1.71337321e-01 -8.88144672e-02 4.67632085e-01 -9.59561229e-01 7.23735154e-01 -1.94890988e+00 9.49496999e-02 3.63721997e-01 9.24843192e-01 5.29740080e-02 -8.75054747e-02 -2.91104078e-01 -5.36876380e-01 1.14902921e-01 -5.00956357e-01 -2.40363568e-01 -4.18000296e-02 1.06462426e-01 1.49740577e-01 6.98173285e-01 4.45423603e-01 1.19061553e+00 -5.49708605e-01 -2.84993380e-01 -1.54119611e-01 6.24492288e-01 -8.61596704e-01 -3.50013286e-01 2.44338825e-01 4.63361204e-01 8.95032883e-02 3.75901461e-01 6.66383743e-01 -2.87252843e-01 1.42197767e-02 -3.54833513e-01 -1.14020482e-02 1.51582405e-01 -7.70862639e-01 1.69086528e+00 -3.71969938e-01 6.25490963e-01 3.28811973e-01 -1.37738419e+00 5.91275930e-01 4.16635752e-01 6.80220664e-01 -1.09058511e+00 1.86904788e-01 3.17855150e-01 6.61453962e-01 -2.31372312e-01 7.96448886e-02 -2.10548565e-01 2.04411130e-02 3.88741136e-01 5.44033349e-01 2.26679236e-01 1.08229950e-01 -2.24377252e-02 1.29515243e+00 -3.83100778e-01 1.32566318e-01 -7.38291681e-01 1.91069111e-01 -3.69016021e-01 5.00533521e-01 9.63724136e-01 -2.85961151e-01 1.08945417e+00 7.06194282e-01 -8.12076211e-01 -1.46081948e+00 -1.09132528e+00 -8.00050437e-01 9.01407719e-01 -7.51954615e-01 -1.82465091e-01 -9.14578140e-01 -5.26446998e-01 1.21991113e-01 1.36144519e-01 -1.20562792e+00 -6.46911711e-02 -9.14634049e-01 -1.16542614e+00 7.36140490e-01 8.20197940e-01 5.70301890e-01 -1.07000756e+00 -7.78848886e-01 2.65476614e-01 3.25955987e-01 -1.05425775e+00 -3.60048145e-01 4.89821374e-01 -1.13694441e+00 -8.22030008e-01 -1.22952747e+00 -7.35334873e-01 6.91736877e-01 -3.40645969e-01 1.25991940e+00 -4.07544971e-02 -7.41195023e-01 1.58318564e-01 1.14060163e-01 -5.60731113e-01 6.83970973e-02 4.33579624e-01 -8.28542933e-02 -8.25095102e-02 2.59122372e-01 -7.60004222e-01 -9.26852882e-01 -1.82347093e-02 -1.12230372e+00 -1.08965516e-01 1.01838398e+00 1.10654521e+00 4.33929116e-01 -2.01643154e-01 4.88829613e-01 -1.08183658e+00 7.06656635e-01 -6.58036411e-01 -9.38877016e-02 -1.31046697e-01 -2.79434949e-01 1.81216866e-01 4.68737990e-01 -4.70015287e-01 -6.80708826e-01 -1.05904900e-01 -1.43152773e-01 -1.00001134e-01 -4.54359442e-01 4.61100191e-01 2.52698749e-01 -1.02895677e-01 1.07958031e+00 -1.66488290e-01 3.14286679e-01 -5.76501369e-01 1.66395664e-01 4.07681316e-01 6.98896587e-01 -4.30638433e-01 2.03606844e-01 6.00225449e-01 1.25018775e-01 -7.90311515e-01 -4.00673926e-01 -3.98936123e-02 -1.06580043e+00 1.99598819e-01 1.07420087e+00 -7.57662058e-01 -4.12151217e-01 3.59298617e-01 -8.37026834e-01 -4.81647313e-01 -2.38376975e-01 7.49777079e-01 -3.07461530e-01 -4.01057564e-02 -7.60717213e-01 -1.75672844e-01 -3.90625656e-01 -1.27190256e+00 1.04863513e+00 -1.05418570e-01 -6.14160597e-01 -9.04341161e-01 -2.18001619e-01 -7.26651028e-02 7.91917861e-01 3.59482855e-01 1.09156072e+00 -7.43003488e-01 -1.91720337e-01 -1.21555984e-01 -5.31631052e-01 4.59313363e-01 -1.82211548e-02 -4.86394286e-01 -1.01610327e+00 -3.42194617e-01 -8.68374705e-02 -3.11435163e-01 1.23000300e+00 8.28508973e-01 1.45415568e+00 -3.38657141e-01 -1.27595142e-01 7.81019688e-01 1.03545737e+00 -3.96377146e-02 8.72484684e-01 5.74755251e-01 5.51801562e-01 7.05249071e-01 -4.39564526e-01 1.90378383e-01 1.96304008e-01 5.09310603e-01 1.66585177e-01 -6.43284559e-01 -4.08367217e-01 6.29993439e-01 -2.38227751e-02 2.91067362e-01 -3.26287001e-01 5.67270637e-01 -1.05303192e+00 5.19784749e-01 -1.57312179e+00 -7.91410029e-01 4.37858663e-02 1.96833646e+00 8.86075199e-01 1.55133739e-01 3.39502573e-01 -1.43297032e-01 5.85603416e-01 -1.82587244e-02 -5.89926958e-01 -3.22164059e-01 -2.60855705e-01 8.78122091e-01 8.48028719e-01 2.18307078e-01 -1.27385676e+00 3.60121429e-01 6.65263081e+00 2.47379273e-01 -1.35168171e+00 4.93392020e-01 1.00963712e+00 -4.71114218e-01 3.50324973e-03 -5.38274288e-01 -3.53069752e-01 3.66861552e-01 1.03947830e+00 2.28999227e-01 3.72448266e-01 5.03798187e-01 -5.81011586e-02 2.91917562e-01 -1.19878328e+00 9.12530899e-01 -1.01478465e-01 -1.35586429e+00 2.79646087e-02 1.82914376e-01 6.05993450e-01 5.28202116e-01 3.47694367e-01 1.04585350e-01 2.04713851e-01 -1.51194787e+00 8.14644337e-01 7.06132889e-01 6.20617509e-01 -6.62323356e-01 8.73188019e-01 -2.31062114e-01 -7.37638593e-01 -2.10789129e-01 -3.96920919e-01 1.09395705e-01 -2.58967817e-01 6.36789501e-01 -6.64012253e-01 1.13402128e-01 1.14252496e+00 5.67955792e-01 -8.72601032e-01 1.09560585e+00 3.12060446e-01 3.41537505e-01 -1.69637308e-01 2.51429647e-01 5.27891934e-01 1.18120089e-01 1.45049781e-01 1.43117571e+00 2.53163427e-01 -8.69624410e-03 -2.38665752e-02 1.02086103e+00 -1.83295146e-01 1.18565246e-01 -3.33415598e-01 1.69263393e-01 -5.81984110e-02 1.37180948e+00 -9.51718509e-01 -4.15861666e-01 -6.35920525e-01 6.38405323e-01 3.74662161e-01 3.24803919e-01 -5.99127173e-01 -4.41704661e-01 6.52093589e-01 4.65961128e-01 3.52343142e-01 -2.32508212e-01 -7.95883477e-01 -1.02653623e+00 -3.80968153e-02 -8.42809200e-01 4.44064528e-01 -5.65321207e-01 -1.46126556e+00 8.47111225e-01 7.06095994e-03 -5.42147517e-01 -1.04728542e-01 -8.83040130e-01 -4.57889557e-01 1.07566965e+00 -1.33512604e+00 -1.03864872e+00 -1.29824907e-01 7.16952622e-01 1.38486013e-01 -2.39710510e-01 7.65245378e-01 5.66070080e-01 -3.05958778e-01 5.24626315e-01 -9.89296511e-02 3.76419336e-01 7.23348916e-01 -1.42049217e+00 5.67409694e-01 4.01355743e-01 -1.25702232e-01 1.05567110e+00 4.38547194e-01 -5.14629900e-01 -6.79151535e-01 -9.38465834e-01 7.39260554e-01 -1.44926921e-01 6.99778318e-01 -6.65467203e-01 -1.11252117e+00 9.67887759e-01 3.69689077e-01 2.65375018e-01 6.40074313e-01 2.16698483e-01 -5.16415298e-01 -7.66754299e-02 -1.31085968e+00 4.74877656e-01 1.03128433e+00 -4.81287152e-01 -6.12824976e-01 4.15145427e-01 3.29369843e-01 -2.80259758e-01 -1.24118078e+00 5.70919693e-01 9.10005569e-01 -9.17917371e-01 1.27792239e+00 -8.80494118e-01 4.62800086e-01 1.78349972e-01 6.67605177e-02 -1.45568180e+00 -6.99196815e-01 -7.46500716e-02 3.40464294e-01 6.26083493e-01 5.83712876e-01 -9.33944702e-01 6.25879526e-01 7.93984830e-01 -2.46600643e-01 -7.97948420e-01 -1.07286859e+00 -5.60252190e-01 6.09806359e-01 -1.52813876e-02 7.30241776e-01 8.93198013e-01 -4.01144117e-01 6.01980723e-02 -1.66927483e-02 -3.07066627e-02 6.49582744e-01 -3.78651500e-01 1.98764905e-01 -1.55006266e+00 -6.00437336e-02 -8.72188568e-01 -7.93770730e-01 -2.16079369e-01 1.47259012e-01 -1.15730190e+00 -2.66129851e-01 -1.29463995e+00 3.00912499e-01 -4.19084340e-01 -5.84472358e-01 6.33303165e-01 9.98150930e-02 6.62230194e-01 4.37963009e-02 6.77867904e-02 -4.78367321e-02 -1.32796299e-02 1.17822838e+00 -3.10208678e-01 4.46541794e-03 -3.36672843e-01 -8.44847381e-01 5.76696157e-01 9.96051848e-01 -3.58752489e-01 -2.81802267e-01 -5.97403824e-01 -2.06685826e-01 -5.73156774e-01 7.68963397e-01 -1.18480051e+00 1.41223505e-01 4.74932879e-01 1.13169658e+00 -1.04277998e-01 4.30665575e-02 -6.16300344e-01 7.55389407e-02 5.43236017e-01 -6.30764663e-01 2.42214084e-01 2.77686447e-01 -5.83555177e-02 -3.78369428e-02 -1.63149517e-02 9.23352420e-01 -5.61568379e-01 -3.45829070e-01 5.68944097e-01 -4.54302311e-01 -5.37887104e-02 4.99630421e-01 -1.52366221e-01 -3.04148674e-01 -6.67703971e-02 -1.26943934e+00 -1.75010145e-01 3.79477292e-02 3.34467918e-01 4.21060830e-01 -1.39167130e+00 -8.98770750e-01 3.50230098e-01 -1.24583259e-01 -8.46926421e-02 3.02794904e-01 1.28788054e+00 -5.97418964e-01 4.96486604e-01 -8.34134996e-01 -8.13290060e-01 -1.02240837e+00 2.44544193e-01 5.27907252e-01 -1.19228922e-01 -1.03813398e+00 6.10700905e-01 5.07151186e-01 -5.73519230e-01 2.22750768e-01 -8.11788261e-01 -1.01812117e-01 3.41092646e-01 7.78044522e-01 1.38726598e-02 5.70363224e-01 -6.27647996e-01 -2.65161484e-01 4.73871231e-02 -4.84830886e-01 -1.49947271e-01 1.82883215e+00 2.26308286e-01 -3.59993309e-01 7.02267289e-02 1.45068550e+00 -5.40837109e-01 -1.19259346e+00 -3.30666393e-01 1.87789604e-01 -9.46086273e-02 3.16779792e-01 -7.86093295e-01 -1.39972222e+00 9.03697789e-01 9.06608641e-01 1.70930892e-01 8.59549761e-01 1.04317077e-01 6.76445484e-01 2.73134768e-01 8.74614269e-02 -1.02281535e+00 -9.50233564e-02 5.00118077e-01 1.05818713e+00 -8.78178656e-01 -8.02283511e-02 1.85912520e-01 -3.67394924e-01 1.34975731e+00 4.17451084e-01 -3.70334536e-01 7.74827003e-01 5.10232210e-01 -4.92770225e-02 -4.90082711e-01 -3.27668995e-01 -2.63539664e-02 3.48929405e-01 6.30180836e-01 5.85831285e-01 4.74928319e-02 -2.54496217e-01 7.70279050e-01 -3.36013108e-01 1.03992410e-01 1.95030689e-01 9.41172719e-01 -3.09833974e-01 -7.98543990e-01 -3.06200683e-01 9.86193419e-01 -7.44926035e-01 -3.32584321e-01 -2.45153993e-01 7.84199417e-01 3.72335285e-01 6.11838922e-02 5.85156798e-01 9.47257280e-02 3.58470112e-01 2.77922243e-01 6.92216575e-01 -5.39662719e-01 -8.18698585e-01 -1.87231395e-02 -5.55322617e-02 -6.41629100e-01 -3.98103744e-01 -9.10565615e-01 -1.10381186e+00 -1.70690760e-01 3.19261134e-01 -3.30086559e-01 6.13217831e-01 8.10401261e-01 3.81408989e-01 1.01927817e+00 -1.37304356e-02 -1.25986874e+00 -3.96164894e-01 -1.19183028e+00 -8.19678485e-01 5.56382239e-01 3.75286788e-01 -6.51823640e-01 1.77831221e-02 -4.53078523e-02]
[14.293983459472656, -2.0069308280944824]
f0f75550-8bdc-4b27-b082-a5db643bd1c1
ced-catalog-extraction-from-documents
2304.14662
null
https://arxiv.org/abs/2304.14662v1
https://arxiv.org/pdf/2304.14662v1.pdf
CED: Catalog Extraction from Documents
Sentence-by-sentence information extraction from long documents is an exhausting and error-prone task. As the indicator of document skeleton, catalogs naturally chunk documents into segments and provide informative cascade semantics, which can help to reduce the search space. Despite their usefulness, catalogs are hard to be extracted without the assist from external knowledge. For documents that adhere to a specific template, regular expressions are practical to extract catalogs. However, handcrafted heuristics are not applicable when processing documents from different sources with diverse formats. To address this problem, we build a large manually annotated corpus, which is the first dataset for the Catalog Extraction from Documents (CED) task. Based on this corpus, we propose a transition-based framework for parsing documents into catalog trees. The experimental results demonstrate that our proposed method outperforms baseline systems and shows a good ability to transfer. We believe the CED task could fill the gap between raw text segments and information extraction tasks on extremely long documents. Data and code are available at \url{https://github.com/Spico197/CatalogExtraction}
['Wenliang Chen', 'Pingfu Chao', 'Baoxing Huai', 'Zhefeng Wang', 'Mengsong Wu', 'Junfei Ren', 'Zijian Yu', 'Zechang Li', 'Guoliang Zhang', 'Tong Zhu']
2023-04-28
null
null
null
null
['catalog-extraction']
['natural-language-processing']
[-2.14897189e-03 -5.69568425e-02 -4.99782264e-01 -4.72500652e-01 -1.19604707e+00 -9.71367955e-01 5.88894844e-01 2.63491571e-01 -3.37450594e-01 7.23137259e-01 3.37551773e-01 -2.07653761e-01 1.14598431e-01 -6.50426209e-01 -5.46199024e-01 -3.51614356e-01 4.28973258e-01 6.50480270e-01 5.40452659e-01 -2.12060735e-01 4.06727731e-01 1.84775934e-01 -1.26391983e+00 6.99185073e-01 9.12468374e-01 8.48728478e-01 5.98728061e-01 3.52839530e-01 -7.60392249e-01 6.44333959e-01 -6.29288018e-01 -6.30923390e-01 -5.55741526e-02 -3.51449788e-01 -1.18876147e+00 5.31684384e-02 8.48372355e-02 -2.21905857e-01 -3.71211208e-02 1.04033780e+00 1.46264836e-01 -1.94948375e-01 6.45380735e-01 -9.67558861e-01 -3.85624796e-01 1.34171605e+00 -4.63946909e-01 5.68840057e-02 4.79607910e-01 -2.67424583e-01 1.51366806e+00 -1.08767676e+00 9.05737638e-01 1.10875106e+00 4.35651630e-01 4.51189756e-01 -7.51809657e-01 -5.46535373e-01 4.10243899e-01 1.25937372e-01 -1.28193450e+00 -4.09284383e-01 6.89462185e-01 -3.63764584e-01 9.92920518e-01 3.03428650e-01 5.96848726e-01 1.26761878e+00 -2.59710342e-01 1.54917514e+00 5.60798943e-01 -5.68107903e-01 -2.05971915e-02 3.48180145e-01 7.07413971e-01 4.24153715e-01 4.18733627e-01 -6.20428741e-01 -5.58063567e-01 5.92889525e-02 2.52072453e-01 -1.88168436e-01 -2.82950610e-01 1.82606056e-01 -9.41605330e-01 9.14846838e-01 6.71598986e-02 3.79786700e-01 -1.07771657e-01 -1.97060972e-01 4.83555436e-01 9.96958241e-02 3.88568282e-01 6.14961505e-01 -5.94166517e-01 -2.16531634e-01 -9.89499629e-01 3.21634352e-01 1.07849216e+00 1.72238183e+00 4.53707963e-01 -5.14126241e-01 -1.47303075e-01 9.06857252e-01 2.69948035e-01 2.29704112e-01 4.47157770e-01 -5.42526186e-01 8.24702203e-01 7.91752696e-01 5.41998297e-02 -5.50660372e-01 -3.38876784e-01 -2.99676806e-01 -6.16462886e-01 -6.35766327e-01 4.66859967e-01 -1.87845364e-01 -5.24983525e-01 1.18740547e+00 3.37826282e-01 -5.33137023e-01 -1.39621142e-02 6.90425992e-01 9.63698328e-01 7.98387110e-01 1.55764185e-02 -3.43159229e-01 1.84720457e+00 -1.08292484e+00 -8.39524806e-01 -3.34689021e-01 7.41441309e-01 -1.09220517e+00 1.38504410e+00 5.73424697e-01 -1.00946021e+00 -2.32145950e-01 -1.10862207e+00 -2.68930167e-01 -5.42089522e-01 3.30258757e-01 7.38742709e-01 4.29399431e-01 -4.56885427e-01 3.09075058e-01 -7.46818483e-01 -2.77405262e-01 4.47966099e-01 -2.78301686e-02 4.52899411e-02 -2.31334940e-02 -1.30084920e+00 5.77646077e-01 6.61743045e-01 8.48620832e-02 -3.38468194e-01 -4.56154346e-01 -6.04549289e-01 1.57501236e-01 7.17356920e-01 -5.77402592e-01 1.64317262e+00 -6.89572334e-01 -1.20153487e+00 6.43527865e-01 -3.14923197e-01 -4.25061345e-01 3.43145967e-01 -4.72227067e-01 -1.86744422e-01 2.90111691e-01 2.89437592e-01 4.42713827e-01 7.24750817e-01 -1.13766110e+00 -9.43872213e-01 -2.84944743e-01 -7.81298280e-02 1.35571077e-01 -5.49543381e-01 4.31381494e-01 -1.02884579e+00 -6.83447003e-01 1.00254603e-02 -8.28775764e-01 6.48344308e-02 -7.04560280e-01 -8.84723961e-01 -6.29770875e-01 6.72820628e-01 -6.26899898e-01 1.57096410e+00 -1.99572837e+00 -1.01656713e-01 -3.48218679e-02 9.21946298e-03 -5.96726686e-03 6.04075938e-02 7.80999124e-01 4.24723923e-01 4.19883460e-01 -1.74330607e-01 -3.02807748e-01 6.74142316e-02 1.63168609e-02 -6.70393407e-01 -1.61797807e-01 1.43312767e-01 9.38009202e-01 -8.97611737e-01 -9.85077977e-01 -4.84571785e-01 3.13418135e-02 -2.68974364e-01 2.49154449e-01 -7.08590508e-01 1.66338846e-01 -9.33262944e-01 8.27479899e-01 4.70445752e-01 -3.69417697e-01 2.29083136e-01 -1.25101386e-02 -1.46834143e-02 5.66303015e-01 -1.15870214e+00 1.70138359e+00 -3.37096334e-01 5.80830038e-01 -8.41770694e-02 -7.55531192e-01 7.17003644e-01 2.29568928e-01 2.48909891e-01 -3.16395462e-01 1.21421553e-01 2.98047304e-01 -2.49679685e-01 -5.87365389e-01 8.56913507e-01 2.37653479e-01 -5.38724661e-01 4.90025222e-01 1.43536165e-01 -5.22301257e-01 9.76850808e-01 6.01000428e-01 1.13482463e+00 -2.99315043e-02 1.90121964e-01 -1.04273833e-01 5.49319625e-01 3.79719526e-01 5.65093994e-01 8.03348362e-01 2.71013051e-01 5.69798827e-01 7.67839491e-01 -1.35353133e-01 -8.38020265e-01 -7.70538330e-01 -1.70218319e-01 1.10210490e+00 5.51387668e-02 -8.63086641e-01 -9.64525700e-01 -9.97793853e-01 -2.37383738e-01 7.51492858e-01 -8.34716856e-02 2.85238981e-01 -7.10713983e-01 -7.75532484e-01 5.61737001e-01 6.27166033e-01 3.20177794e-01 -1.00586784e+00 -1.82539061e-01 3.63247097e-01 -5.46320200e-01 -1.32817721e+00 -6.91864550e-01 1.40131965e-01 -7.33678520e-01 -1.06959283e+00 -4.46460426e-01 -9.69039619e-01 5.85784495e-01 1.30347356e-01 1.30145586e+00 1.75234735e-01 -8.15369338e-02 2.26310939e-02 -7.61310518e-01 -6.79930389e-01 -4.76718187e-01 8.28515410e-01 -2.24180549e-01 -3.40525270e-01 7.67390072e-01 -2.15680510e-01 -5.19743204e-01 2.60010600e-01 -8.82348716e-01 1.86213657e-01 7.11556375e-01 8.16590190e-01 5.85455000e-01 1.69730663e-01 7.08749652e-01 -1.26599157e+00 9.82052803e-01 -3.78613174e-01 -7.96429515e-01 2.92785406e-01 -7.49404073e-01 2.42166996e-01 8.12193036e-01 -3.04864556e-01 -1.45728993e+00 2.88406104e-01 -1.93767443e-01 2.60514885e-01 -2.43223771e-01 7.17054605e-01 -3.75460267e-01 7.50372171e-01 3.53303254e-01 2.77120054e-01 -4.70450073e-01 -8.78620028e-01 5.71186185e-01 1.08604538e+00 4.81576532e-01 -6.76553726e-01 6.56599462e-01 2.25900218e-01 -5.66502810e-01 -6.98272824e-01 -1.35954523e+00 -8.31191123e-01 -8.32475364e-01 -2.84580165e-03 7.30146229e-01 -8.41643989e-01 -3.08748662e-01 2.01710239e-01 -1.43690860e+00 1.61564443e-02 -2.14721903e-01 2.14163497e-01 -1.68680549e-01 3.70979935e-01 -1.00638652e+00 -5.92382193e-01 -7.40786552e-01 -9.62018669e-01 1.13050735e+00 2.49705344e-01 -4.61753845e-01 -6.62544608e-01 -1.22338533e-01 5.33534408e-01 -5.20896837e-02 -2.82542706e-01 1.04715884e+00 -9.83038425e-01 -8.01162899e-01 -4.83140618e-01 -2.38398269e-01 1.32917866e-01 3.85397226e-02 1.40625998e-01 -9.04692411e-01 7.79456366e-03 -2.73698773e-02 -4.00695056e-01 1.12224007e+00 6.68172687e-02 1.12380016e+00 -5.53713799e-01 -4.90209043e-01 3.05508316e-01 1.10525393e+00 2.75741369e-01 3.53896558e-01 4.82486546e-01 6.65651023e-01 8.49802732e-01 7.48133898e-01 4.58563656e-01 4.70164746e-01 4.87242937e-01 -3.11628748e-02 2.14875057e-01 -9.12039876e-02 -5.23876548e-01 3.87921810e-01 1.35596526e+00 2.47863948e-01 -5.11046171e-01 -9.12208498e-01 5.60872376e-01 -1.73513687e+00 -9.03918147e-01 -3.90885681e-01 1.66659510e+00 1.46515548e+00 4.82951045e-01 -2.91374419e-02 8.48961920e-02 5.38644969e-01 1.19497262e-01 -1.98270261e-01 -1.04605503e-01 -4.67972718e-02 -1.64270893e-01 2.39221841e-01 3.49035800e-01 -1.18808424e+00 1.36213255e+00 5.53971958e+00 1.19471717e+00 -6.08095169e-01 9.19375103e-03 2.88493961e-01 1.01189334e-02 -3.53515834e-01 3.15077782e-01 -1.67530286e+00 6.23413384e-01 9.44428861e-01 -2.26774007e-01 3.76186916e-04 9.37319934e-01 -4.25009541e-02 -1.77643821e-01 -1.27525568e+00 5.87839305e-01 -1.40478879e-01 -1.38810027e+00 1.43062100e-01 -2.49876142e-01 4.17653620e-01 -8.95794183e-02 -4.05719727e-01 4.68125701e-01 3.77190441e-01 -6.42846882e-01 9.15814936e-01 2.31171370e-01 5.74486911e-01 -5.28863788e-01 7.87308335e-01 5.71291149e-01 -1.30335188e+00 8.16967115e-02 -4.15140897e-01 2.51691699e-01 3.94095898e-01 9.04518604e-01 -9.92800355e-01 4.72714067e-01 5.84283352e-01 7.40207314e-01 -7.33417094e-01 8.21536362e-01 -6.10100985e-01 8.17654371e-01 -2.97923565e-01 -5.23747563e-01 2.43515387e-01 -2.19585761e-01 4.34763998e-01 1.77139306e+00 1.32487372e-01 8.54643583e-02 1.32567674e-01 8.01887453e-01 -3.06048691e-01 3.10662270e-01 -3.98898989e-01 -2.29079545e-01 6.38384581e-01 1.51340353e+00 -1.09183204e+00 -5.21432579e-01 -5.36041915e-01 8.03642869e-01 2.35157490e-01 3.99926037e-01 -5.58313131e-01 -8.41844440e-01 1.68194130e-01 -9.21355262e-02 3.76845479e-01 -2.38061711e-01 -5.07135451e-01 -1.43792748e+00 3.99997324e-01 -8.62719536e-01 6.40352845e-01 -4.88202840e-01 -1.25340021e+00 7.17282414e-01 1.52776465e-01 -1.25855148e+00 -4.59465235e-01 -3.57076138e-01 -3.74553442e-01 4.55719829e-01 -1.36999261e+00 -1.04065394e+00 -1.71561629e-01 3.72781843e-01 1.06895149e+00 -1.04668193e-01 5.33639431e-01 1.23755209e-01 -7.22238719e-01 4.87139970e-01 6.53891116e-02 4.54276979e-01 6.06305778e-01 -1.38532603e+00 4.95071501e-01 1.04368627e+00 4.47060972e-01 7.56453872e-01 5.85004508e-01 -7.75848806e-01 -1.49426126e+00 -9.28821743e-01 1.16420734e+00 -6.20277107e-01 8.36376965e-01 -7.85197377e-01 -1.02992129e+00 6.66230321e-01 4.95396703e-01 -5.52685142e-01 7.00030506e-01 1.95942372e-01 -3.97282749e-01 -9.10255015e-02 -5.48510253e-01 5.49146414e-01 9.80358958e-01 -3.58775377e-01 -7.76168942e-01 6.24728918e-01 9.54340875e-01 -2.98147142e-01 -7.68832088e-01 7.05141798e-02 4.50843751e-01 -4.40475047e-01 7.21502364e-01 -3.31705183e-01 5.72721720e-01 -7.42589161e-02 1.09710738e-01 -9.94874716e-01 6.23731874e-02 -6.70270503e-01 -2.31519282e-01 1.80859900e+00 8.93011689e-01 -2.88994044e-01 6.82464719e-01 5.16842186e-01 -1.43277764e-01 -7.93519020e-01 -5.40340602e-01 -9.30655181e-01 1.40966654e-01 -6.73019826e-01 6.92662299e-01 5.04996240e-01 3.99208486e-01 8.91631782e-01 3.33498456e-02 -2.99546067e-02 4.40786719e-01 5.35696328e-01 6.68859780e-01 -1.19695413e+00 -1.46981969e-01 -5.68931222e-01 4.11563575e-01 -1.36613321e+00 2.65368074e-01 -1.08037889e+00 2.96612471e-01 -1.63411736e+00 4.39314634e-01 -4.70479459e-01 1.45221889e-01 5.84841847e-01 -1.93254068e-01 -2.56253302e-01 1.28277019e-01 5.16384721e-01 -7.90469587e-01 5.60042024e-01 1.06575561e+00 -3.22835326e-01 -2.99207836e-01 2.07100391e-01 -1.01489413e+00 7.99543560e-01 8.71774077e-01 -7.06955910e-01 -3.90417159e-01 -4.27057177e-01 4.19374555e-01 3.99717912e-02 -3.65498632e-01 -4.03495878e-01 4.23123330e-01 -1.62431121e-01 2.51853794e-01 -9.48052824e-01 1.07353620e-01 -4.99540806e-01 -3.07803243e-01 1.19090013e-01 -3.76198769e-01 -4.85208258e-02 -1.33454492e-02 3.79080325e-01 -4.66608405e-01 -7.44162321e-01 3.23593616e-01 -2.66505748e-01 -6.53905571e-01 1.60606503e-01 -5.62031627e-01 4.19834435e-01 5.75567663e-01 7.51825795e-02 -4.66940403e-01 -1.41772747e-01 -2.40449160e-01 4.75114107e-01 1.98108360e-01 4.98050511e-01 5.77150047e-01 -9.91099060e-01 -6.34245992e-01 1.10528052e-01 2.80378640e-01 3.05334628e-01 -2.53998399e-01 5.89690149e-01 -4.38729644e-01 6.96247160e-01 2.96720296e-01 -3.75228941e-01 -1.13483727e+00 5.51441312e-01 -2.78315663e-01 -6.21951699e-01 -7.14542449e-01 8.52197886e-01 -2.42815465e-02 -1.56679004e-01 3.90418947e-01 -5.36981344e-01 -4.79872137e-01 5.92579663e-01 7.34180748e-01 -3.45272645e-02 2.72192210e-01 -1.68461248e-01 -2.40412086e-01 3.47672552e-01 -6.93829775e-01 -2.62796015e-01 1.30008900e+00 -3.23918462e-01 -1.91705436e-01 2.17558339e-01 1.00063121e+00 1.64488867e-01 -9.53382969e-01 -3.16113174e-01 8.28345358e-01 -2.21151367e-01 -1.08576596e-01 -6.63624346e-01 -8.79280508e-01 6.99555278e-01 -2.06608415e-01 4.77774888e-01 1.13564253e+00 3.55676442e-01 1.12148893e+00 5.36642194e-01 3.09761882e-01 -1.36876881e+00 -1.85915586e-02 7.15412915e-01 9.73130703e-01 -1.24651992e+00 6.23869859e-02 -6.46171987e-01 -7.66079545e-01 1.17418253e+00 5.57968736e-01 3.16062421e-01 5.50383449e-01 4.64108139e-01 -5.00124693e-02 -2.20083088e-01 -9.38490331e-01 -3.02195221e-01 2.89514184e-01 2.16770083e-01 4.92089421e-01 -2.12177128e-01 -5.83426535e-01 1.21546865e+00 -4.85981256e-01 -3.21884513e-01 6.11117244e-01 9.81988013e-01 -5.42001963e-01 -1.40844309e+00 -7.03896433e-02 4.64735359e-01 -6.72913671e-01 -3.13622475e-01 -6.81033850e-01 8.23080063e-01 -2.58092910e-01 1.03767395e+00 -1.99904144e-01 -3.85194644e-02 2.93014556e-01 2.93292403e-01 1.82785735e-01 -8.79183590e-01 -4.62660313e-01 4.04557824e-01 3.97734851e-01 -2.70606518e-01 -2.18718037e-01 -7.89427996e-01 -1.33786225e+00 6.35868451e-03 -4.82292175e-01 3.88287872e-01 5.83720446e-01 8.72554541e-01 2.44464800e-01 3.85867625e-01 4.89529580e-01 -4.85009879e-01 -5.86429060e-01 -1.17010891e+00 -4.88895237e-01 2.60244876e-01 -4.30137664e-02 -2.45534405e-01 -2.20285401e-01 4.95954156e-01]
[10.982760429382324, 8.669939994812012]
cf74a701-789a-4b1f-b6b0-ffe719142de3
unispeech-sat-universal-speech-representation
2110.05752
null
https://arxiv.org/abs/2110.05752v1
https://arxiv.org/pdf/2110.05752v1.pdf
UniSpeech-SAT: Universal Speech Representation Learning with Speaker Aware Pre-Training
Self-supervised learning (SSL) is a long-standing goal for speech processing, since it utilizes large-scale unlabeled data and avoids extensive human labeling. Recent years witness great successes in applying self-supervised learning in speech recognition, while limited exploration was attempted in applying SSL for modeling speaker characteristics. In this paper, we aim to improve the existing SSL framework for speaker representation learning. Two methods are introduced for enhancing the unsupervised speaker information extraction. First, we apply the multi-task learning to the current SSL framework, where we integrate the utterance-wise contrastive loss with the SSL objective function. Second, for better speaker discrimination, we propose an utterance mixing strategy for data augmentation, where additional overlapped utterances are created unsupervisely and incorporate during training. We integrate the proposed methods into the HuBERT framework. Experiment results on SUPERB benchmark show that the proposed system achieves state-of-the-art performance in universal representation learning, especially for speaker identification oriented tasks. An ablation study is performed verifying the efficacy of each proposed method. Finally, we scale up training dataset to 94 thousand hours public audio data and achieve further performance improvement in all SUPERB tasks.
['Xiangzhan Yu', 'Jinyu Li', 'Furu Wei', 'Yao Qian', 'Jian Wu', 'Shujie Liu', 'Zhuo Chen', 'Zhengyang Chen', 'Chengyi Wang', 'Yu Wu', 'Sanyuan Chen']
2021-10-12
null
null
null
null
['speaker-identification']
['speech']
[ 5.04964054e-01 8.47612098e-02 -3.08091223e-01 -8.74867499e-01 -1.30692422e+00 -2.65431523e-01 5.31200945e-01 2.10087951e-02 -3.83297652e-01 6.00141406e-01 5.23729563e-01 -4.39610571e-01 2.64453422e-02 -7.00186566e-02 -4.86587107e-01 -7.86177993e-01 6.43215049e-03 2.18259394e-01 4.91697788e-02 -3.01693469e-01 -2.37872321e-02 2.57957488e-01 -1.60235262e+00 1.60721272e-01 8.47535491e-01 9.92598593e-01 1.25813186e-01 4.66215879e-01 -2.82013059e-01 8.01574647e-01 -7.32874811e-01 -1.91348329e-01 4.40266840e-02 -3.09142530e-01 -9.91123855e-01 3.81801754e-01 4.63872582e-01 -6.42686933e-02 -5.95006086e-02 8.19356680e-01 8.41987610e-01 3.61873209e-01 5.57132423e-01 -1.30378950e+00 -4.40598875e-01 1.16861951e+00 -6.44650042e-01 5.12400985e-01 2.16366455e-01 -1.50424421e-01 9.38999057e-01 -9.57130611e-01 -1.07933916e-01 1.49451935e+00 5.74071586e-01 7.02094138e-01 -9.92180228e-01 -7.85138547e-01 3.98818851e-01 3.00293237e-01 -1.40647399e+00 -1.16435134e+00 1.08930898e+00 -2.16249943e-01 8.50573123e-01 3.73849303e-01 -1.13640778e-01 1.13459170e+00 -6.43177390e-01 1.14849496e+00 1.22982657e+00 -8.35397184e-01 9.53763500e-02 4.26227182e-01 6.62376940e-01 5.56800187e-01 -3.53579402e-01 8.41846317e-02 -8.21683705e-01 -2.02458389e-02 3.00260007e-01 -1.75329626e-01 -1.19377181e-01 -3.97356264e-02 -1.05527890e+00 7.26370037e-01 1.16202086e-01 3.96109611e-01 -5.94912283e-02 -3.56936514e-01 6.65927112e-01 4.03375685e-01 7.84609914e-01 -1.36282975e-02 -4.05636430e-01 -7.66752437e-02 -1.09857559e+00 -1.78024501e-01 4.98528361e-01 9.26247597e-01 7.10975051e-01 5.70210397e-01 -3.36196393e-01 1.60360086e+00 4.54911739e-01 2.94481546e-01 1.07900965e+00 -4.76547211e-01 5.33167958e-01 3.93978298e-01 -3.40206981e-01 -3.91622871e-01 -3.32229674e-01 -6.88228607e-01 -9.10977066e-01 -1.42678425e-01 -3.15873288e-02 -1.59937233e-01 -9.54988182e-01 1.73714757e+00 2.77467728e-01 5.57244122e-01 3.82756412e-01 7.09929883e-01 1.07948256e+00 7.89357245e-01 1.63220719e-01 -5.09179890e-01 1.34799647e+00 -1.33193409e+00 -9.48596418e-01 -3.19115967e-01 5.19436061e-01 -7.85543561e-01 1.05275548e+00 3.49028885e-01 -1.00826907e+00 -8.91135156e-01 -1.00988317e+00 1.91899702e-01 -2.81171203e-01 3.66769999e-01 5.72064221e-01 1.31653619e+00 -8.55906129e-01 1.66350156e-01 -6.49462521e-01 -3.25802118e-01 3.88663769e-01 4.11792129e-01 -1.55452028e-01 2.71606922e-01 -1.25474298e+00 6.81822896e-01 4.55843836e-01 5.04241623e-02 -1.14530587e+00 -5.56254804e-01 -9.50279772e-01 -6.08852040e-03 4.31982487e-01 -4.41559777e-02 1.44522130e+00 -8.49662840e-01 -1.91814315e+00 6.66022241e-01 -4.63531703e-01 -8.00407946e-01 1.51353657e-01 -9.76293087e-02 -7.75420606e-01 -1.90921113e-01 -1.35729134e-01 5.42610288e-01 1.04748082e+00 -1.36736810e+00 -6.64518178e-01 -4.73532170e-01 -2.47447327e-01 3.01984519e-01 -9.78792906e-01 2.54142672e-01 -2.86396921e-01 -8.49687696e-01 2.82000214e-01 -6.31370842e-01 -2.14506537e-01 -6.40518427e-01 -4.47529197e-01 -6.31923676e-01 1.00054359e+00 -7.44898975e-01 1.37250364e+00 -2.34295869e+00 -6.41075745e-02 2.25321278e-01 -2.30966568e-01 4.71550137e-01 -1.94409803e-01 2.00265497e-01 -1.40651196e-01 -5.52789383e-02 -4.90545332e-01 -1.06922579e+00 1.63113512e-02 8.25020000e-02 -3.66170496e-01 3.05745602e-01 1.16129950e-01 4.76384699e-01 -5.21313965e-01 -5.02305984e-01 2.67745972e-01 4.54492748e-01 -4.17682201e-01 3.79653424e-01 1.34890780e-01 5.27046919e-01 -1.27973691e-01 7.87018597e-01 7.34462559e-01 1.77687749e-01 -8.10186043e-02 -1.53100833e-01 -9.30343196e-02 6.56227231e-01 -1.36365759e+00 1.86051893e+00 -7.47264683e-01 2.14339420e-01 2.39284381e-01 -1.35803711e+00 1.14037097e+00 5.93289733e-01 3.00096720e-01 -2.97367036e-01 1.96707267e-02 1.81275681e-01 -2.48780637e-03 -2.94181377e-01 2.34939814e-01 -3.29198569e-01 1.16788082e-01 3.48549932e-01 4.80802476e-01 -1.03235938e-01 9.58236773e-03 1.10367417e-01 6.10101283e-01 -1.21982925e-01 3.98809314e-01 -2.31461927e-01 1.08587325e+00 -4.84731615e-01 6.34308398e-01 7.13154972e-01 -4.41059500e-01 4.03772295e-01 -7.98843950e-02 7.45171830e-02 -5.83781540e-01 -8.69119823e-01 -2.38144070e-01 1.73262918e+00 -2.54356712e-01 -1.21033169e-01 -7.50859857e-01 -7.68815875e-01 -1.55709580e-01 7.47512102e-01 -3.00980449e-01 -1.15494922e-01 -4.73191828e-01 -9.32905197e-01 8.03345859e-01 5.49099565e-01 6.64081454e-01 -1.11740386e+00 1.07672252e-01 2.85234481e-01 -1.58020213e-01 -1.15068388e+00 -6.65233254e-01 3.42066199e-01 -8.08546722e-01 -4.28790241e-01 -7.81882405e-01 -1.26833546e+00 3.85961264e-01 4.19830710e-01 6.96284890e-01 -3.08293849e-01 -5.07270917e-02 3.77434462e-01 -5.98545611e-01 -4.55157131e-01 -6.86260760e-01 3.27452213e-01 4.38648999e-01 5.42154849e-01 2.68433362e-01 -5.76276481e-01 -1.85973391e-01 4.47994232e-01 -5.82104623e-01 -2.24682033e-01 6.08875215e-01 1.06663120e+00 3.23069930e-01 8.11922029e-02 1.46529913e+00 -8.24365258e-01 5.28514802e-01 -2.71284223e-01 -1.81870267e-01 2.41252616e-01 -5.55403888e-01 2.30451282e-02 3.68547559e-01 -4.49139088e-01 -1.55090988e+00 1.94833443e-01 -6.07308686e-01 -2.76375413e-01 -3.31899554e-01 4.45109248e-01 -4.29131061e-01 2.79066600e-02 4.86655235e-01 5.54947019e-01 1.59765497e-01 -7.58826792e-01 3.65365684e-01 1.36513162e+00 4.47545052e-01 -6.05278015e-01 5.45686364e-01 2.07273155e-01 -6.86553180e-01 -1.19452155e+00 -1.05052972e+00 -1.00113916e+00 -5.06353080e-01 7.75612099e-03 4.25206810e-01 -1.04802656e+00 -7.30479121e-01 5.62913895e-01 -8.35984051e-01 -7.06086978e-02 -2.80653566e-01 6.24687731e-01 -4.35650200e-01 4.58857417e-01 -4.05019879e-01 -1.44141388e+00 -5.29260039e-01 -1.25627768e+00 1.08209491e+00 6.62806816e-03 1.43131420e-01 -8.45843434e-01 -2.39516608e-02 8.85288477e-01 4.69030559e-01 -6.83349609e-01 5.46481729e-01 -1.23438239e+00 -1.08597018e-02 5.69952875e-02 -1.80136040e-02 8.91168416e-01 5.90468824e-01 -4.72972095e-01 -1.57372880e+00 -5.54639935e-01 2.31391236e-01 -6.02296710e-01 1.07467151e+00 1.39450639e-01 1.43608165e+00 -3.18477511e-01 -7.57547244e-02 4.13501859e-01 8.07704031e-01 2.22685710e-01 2.72918642e-01 8.56681541e-02 7.46807218e-01 8.19550872e-01 4.35683221e-01 3.32168013e-01 2.84410059e-01 7.59084344e-01 1.10738911e-01 -1.05357982e-01 -4.00347799e-01 -1.28580406e-01 6.00590110e-01 1.41955698e+00 8.74165222e-02 -5.16701154e-02 -8.54513824e-01 6.37522995e-01 -1.59198654e+00 -9.38943684e-01 2.61336893e-01 2.30575895e+00 1.02385569e+00 2.82712370e-01 4.64037389e-01 5.78327835e-01 8.59598100e-01 3.08313370e-01 -5.30602276e-01 -1.08713932e-01 -3.16838592e-01 1.66790202e-01 2.77429014e-01 6.63368225e-01 -1.43456066e+00 1.16388106e+00 5.75799036e+00 1.19540429e+00 -1.17092490e+00 5.05734205e-01 6.98308945e-01 8.75161812e-02 7.79756978e-02 -3.99808437e-01 -1.14965892e+00 1.86954945e-01 1.09620976e+00 -1.27452105e-01 3.00692886e-01 9.06286240e-01 3.35194439e-01 4.83391792e-01 -8.11639011e-01 1.04918802e+00 5.18466234e-01 -9.92038190e-01 5.08702211e-02 -2.95280457e-01 6.18365049e-01 -7.88641423e-02 2.26185158e-01 7.52109110e-01 1.02711074e-01 -7.15475082e-01 5.77448666e-01 -1.23544283e-01 6.26289129e-01 -7.92625785e-01 7.59462714e-01 4.07597393e-01 -1.18792355e+00 -2.76769221e-01 -2.46143818e-01 1.99538797e-01 2.29990095e-01 2.93418020e-01 -1.34432292e+00 5.48362076e-01 5.14256239e-01 6.06597006e-01 -6.43687606e-01 8.78769100e-01 1.24321887e-02 1.23382080e+00 -2.29192778e-01 -1.53138256e-02 1.33852392e-01 7.64165074e-02 5.83080411e-01 1.45637429e+00 4.28543016e-02 -1.74948126e-01 3.53703141e-01 2.81580091e-01 -1.68444723e-01 5.99104166e-01 -3.32735330e-01 2.97864452e-02 5.89761615e-01 1.10098517e+00 -3.19702357e-01 -4.21452403e-01 -4.51524854e-01 8.38194609e-01 3.04833442e-01 3.76873225e-01 -6.07464194e-01 -1.13884948e-01 3.62931520e-01 -1.31471261e-01 8.34206492e-02 2.83484664e-02 -2.90188402e-01 -9.61972058e-01 -1.57105923e-01 -1.09499228e+00 4.40173030e-01 -2.88862169e-01 -1.52544236e+00 8.00888658e-01 -8.51054341e-02 -1.15321016e+00 -3.34389359e-01 -4.18102920e-01 -6.67987466e-01 8.02741408e-01 -1.71204865e+00 -1.48964322e+00 -3.37039642e-02 6.39117420e-01 1.35483968e+00 -7.98415244e-01 8.80106151e-01 5.35780489e-01 -8.72570574e-01 1.00557268e+00 8.88541713e-02 3.54511142e-02 8.76695395e-01 -1.15963674e+00 3.50136667e-01 8.43913019e-01 4.47421014e-01 6.56968474e-01 4.16105717e-01 -3.27709854e-01 -1.14540017e+00 -1.15873575e+00 6.69028461e-01 -3.76974680e-02 5.92795074e-01 -5.54791510e-01 -1.01960015e+00 6.88516378e-01 2.83064395e-01 -1.25104889e-01 1.00506318e+00 5.25356174e-01 -4.12007600e-01 -4.23837721e-01 -1.07433116e+00 2.91938722e-01 7.99819112e-01 -6.45628572e-01 -8.68417799e-01 3.90859514e-01 9.33374643e-01 9.70988348e-03 -5.39948642e-01 5.62488019e-01 2.85868645e-01 -6.44996345e-01 1.00127721e+00 -6.98351264e-01 -1.97120085e-01 -2.20735759e-01 -4.31749791e-01 -1.39257205e+00 1.90085731e-02 -7.90823102e-01 -4.19694148e-02 1.91495430e+00 6.12740695e-01 -6.39984548e-01 8.17253411e-01 -1.95372440e-02 -5.44804573e-01 -5.47522426e-01 -1.20134282e+00 -9.81867731e-01 -8.89465660e-02 -5.05614221e-01 5.96350253e-01 1.09659600e+00 7.93104172e-02 6.71752036e-01 -6.63178742e-01 3.30560476e-01 8.34033728e-01 -1.86502218e-01 6.63144410e-01 -1.01587880e+00 -3.70887607e-01 -3.11763465e-01 -2.00381264e-01 -1.32149935e+00 4.83661562e-01 -1.01745987e+00 2.55572766e-01 -1.12298632e+00 1.42010599e-01 -5.47524750e-01 -6.73802435e-01 4.79489654e-01 -3.70746940e-01 1.86982155e-01 8.49399120e-02 6.54339641e-02 -6.26455665e-01 8.50302279e-01 7.72616386e-01 -4.10123616e-01 -4.01467592e-01 5.18367290e-01 -6.60367548e-01 7.23495305e-01 8.67949486e-01 -1.88091502e-01 -5.91444552e-01 -2.45412663e-01 -7.54893482e-01 -7.34882802e-02 2.07867334e-03 -9.62849319e-01 3.45375597e-01 2.35678583e-01 -1.70413122e-01 -7.50420690e-01 6.00898564e-01 -6.46361411e-01 -5.58322906e-01 1.21658631e-01 -6.24302804e-01 -5.36549509e-01 2.13682592e-01 4.23933595e-01 -4.92077261e-01 -4.06832755e-01 9.28608835e-01 2.17010722e-01 -7.07393050e-01 9.57920402e-02 -1.36309728e-01 -1.14857808e-01 6.27809346e-01 7.52583519e-02 -9.60185453e-02 -3.64430547e-01 -9.16044116e-01 1.90853700e-01 -3.90730530e-01 6.13855779e-01 5.55338562e-01 -1.22042191e+00 -9.54196036e-01 5.05911052e-01 2.74556071e-01 -2.20775217e-01 4.88474339e-01 5.98645210e-01 2.38657862e-01 4.80357856e-01 1.63781792e-01 -6.67990565e-01 -1.41714227e+00 4.96300071e-01 1.22041062e-01 -1.20964669e-01 -1.96205840e-01 1.22125304e+00 1.68204114e-01 -8.64316583e-01 7.57825315e-01 -1.72296003e-01 -4.80783463e-01 1.50886610e-01 6.63067818e-01 4.21070039e-01 3.41056317e-01 -8.49535763e-01 -4.21513349e-01 2.00867370e-01 -3.78456146e-01 -3.01797658e-01 1.37464011e+00 -3.52349967e-01 2.28805214e-01 8.35579574e-01 1.12896478e+00 1.51047751e-01 -1.07305264e+00 -4.63942528e-01 8.35050717e-02 -4.90603559e-02 3.06710929e-01 -7.98084378e-01 -7.94022202e-01 1.00553358e+00 7.98256338e-01 3.43735009e-01 1.04588389e+00 7.20619410e-02 6.73514783e-01 3.43510181e-01 2.95587271e-01 -9.35612082e-01 1.15953341e-01 4.17814583e-01 8.56313586e-01 -1.64647365e+00 -2.29290530e-01 -6.30128503e-01 -8.71707380e-01 6.11951828e-01 6.73961401e-01 4.04695958e-01 7.34944344e-01 1.60398200e-01 2.55459905e-01 2.65404165e-01 -6.08005166e-01 -2.91647464e-01 2.96199888e-01 5.28318286e-01 6.27624452e-01 1.21804485e-02 -9.75373685e-02 8.42286706e-01 -2.71766812e-01 -4.18032438e-01 2.38517016e-01 7.65539050e-01 -5.87056994e-01 -1.32871580e+00 -5.31090438e-01 3.13890785e-01 -3.44700694e-01 -3.06860954e-01 -1.17804006e-01 2.17664629e-01 -1.41479328e-01 1.32883751e+00 -1.01267084e-01 -2.40526035e-01 3.60583931e-01 5.47300100e-01 1.99937508e-01 -8.28665495e-01 -6.37948334e-01 4.35583740e-01 1.12356119e-01 4.15358804e-02 -7.04208612e-01 -7.56959140e-01 -1.02860868e+00 3.94047290e-01 -6.28975451e-01 5.18646777e-01 9.89319503e-01 9.13062692e-01 9.78494510e-02 7.88561523e-01 1.08546972e+00 -7.17738271e-01 -9.87949252e-01 -1.46358764e+00 -5.77340841e-01 2.13457406e-01 5.73634326e-01 -7.00999439e-01 -6.92587435e-01 2.70784557e-01]
[14.477108001708984, 6.301933288574219]
36d5701a-82f1-441a-b0b7-a7d09f783dce
opal-occlusion-pattern-aware-loss-for
2203.02231
null
https://arxiv.org/abs/2203.02231v3
https://arxiv.org/pdf/2203.02231v3.pdf
OPAL: Occlusion Pattern Aware Loss for Unsupervised Light Field Disparity Estimation
Light field disparity estimation is an essential task in computer vision with various applications. Although supervised learning-based methods have achieved both higher accuracy and efficiency than traditional optimization-based methods, the dependency on ground-truth disparity for training limits the overall generalization performance not to say for real-world scenarios where the ground-truth disparity is hard to capture. In this paper, we argue that unsupervised methods can achieve comparable accuracy, but, more importantly, much higher generalization capacity and efficiency than supervised methods. Specifically, we present the Occlusion Pattern Aware Loss, named OPAL, which successfully extracts and encodes the general occlusion patterns inherent in the light field for loss calculation. OPAL enables: i) accurate and robust estimation by effectively handling occlusions without using any ground-truth information for training and ii) much efficient performance by significantly reducing the network parameters required for accurate inference. Besides, a transformer-based network and a refinement module are proposed for achieving even more accurate results. Extensive experiments demonstrate our method not only significantly improves the accuracy compared with the SOTA unsupervised methods, but also possesses strong generalization capacity, even for real-world data, compared with supervised methods. Our code will be made publicly available.
['Tao Yu', 'Haoqian Wang', 'Chao Deng', 'Jingyao Wu', 'Jiayin Zhao', 'Peng Li']
2022-03-04
null
null
null
null
['disparity-estimation']
['computer-vision']
[ 2.89479494e-01 -2.87865400e-01 -1.65499225e-01 -6.02357328e-01 -6.73429012e-01 6.03913143e-02 2.00015366e-01 1.98034272e-02 -3.93268108e-01 8.74918103e-01 -2.03633443e-01 -1.74396664e-01 -1.37505665e-01 -9.50403035e-01 -5.15967429e-01 -9.26129580e-01 3.14515054e-01 3.55040818e-01 4.77371365e-01 2.64811546e-01 2.65381575e-01 5.53815544e-01 -2.05835724e+00 -3.64920288e-01 1.37133980e+00 1.32446909e+00 3.78685564e-01 1.28298432e-01 -1.50315002e-01 9.27022636e-01 -2.54842401e-01 -4.14865643e-01 4.34534937e-01 -7.49378502e-02 -3.20813566e-01 3.34524751e-01 8.95651937e-01 -5.42478442e-01 -2.95308441e-01 1.21287131e+00 5.40067554e-01 7.93116465e-02 5.31320691e-01 -1.10461140e+00 -1.31075516e-01 -2.03583669e-02 -6.62067652e-01 -9.34158936e-02 6.95065781e-03 4.42228049e-01 8.98919284e-01 -7.01512754e-01 4.35756624e-01 9.01533365e-01 6.81232870e-01 2.98940510e-01 -1.20395386e+00 -7.09119022e-01 -4.73611429e-02 3.66435915e-01 -1.39552605e+00 -5.23752928e-01 1.01863635e+00 -5.62108457e-01 5.82354188e-01 -9.54185799e-02 6.72113895e-01 5.64761400e-01 -9.68614519e-02 6.87109828e-01 1.26077378e+00 -3.92236352e-01 2.72613257e-01 1.15268916e-01 2.56558824e-02 9.23505664e-01 3.84282798e-01 3.24240506e-01 -5.14179885e-01 1.99127257e-01 8.18874478e-01 -2.96157673e-02 -4.29899931e-01 -6.60825729e-01 -1.02735758e+00 7.46025801e-01 5.93738794e-01 -7.67900199e-02 -4.78239834e-01 8.19073021e-02 1.18516095e-01 -1.14691213e-01 4.92471337e-01 2.61405587e-01 -1.92642823e-01 3.44913416e-02 -1.03074443e+00 4.89106402e-02 4.37257469e-01 9.34528291e-01 1.31268251e+00 1.27169058e-01 -1.09316289e-01 8.99361134e-01 2.03867003e-01 7.43786216e-01 1.38794556e-01 -1.10909593e+00 4.95521456e-01 8.23474705e-01 1.76488921e-01 -1.18103826e+00 -4.04085964e-01 -5.80735326e-01 -9.84602392e-01 4.70344663e-01 5.47062576e-01 6.52114451e-02 -8.22298586e-01 1.75362551e+00 3.75618488e-01 1.01734832e-01 -9.57969204e-02 1.00924182e+00 7.70362377e-01 4.44771916e-01 -1.56252176e-01 -4.87416118e-01 1.09585738e+00 -7.77459383e-01 -6.89007699e-01 -1.71879500e-01 1.88500896e-01 -7.66516447e-01 9.57936287e-01 5.33192992e-01 -9.49796498e-01 -4.92317975e-01 -1.00792372e+00 -2.79876739e-01 7.57199824e-02 1.79081246e-01 9.99998093e-01 6.03373408e-01 -8.10307980e-01 3.91179115e-01 -7.74242103e-01 -2.10590333e-01 6.89495981e-01 4.60453331e-01 -1.84514791e-01 -2.36954242e-01 -7.27086067e-01 7.03132808e-01 2.60223269e-01 2.65650362e-01 -6.01282716e-01 -6.96133912e-01 -1.07100844e+00 -8.72453302e-02 4.22185630e-01 -6.14510357e-01 9.04774904e-01 -7.57516921e-01 -1.57556689e+00 7.38057375e-01 -4.04599547e-01 -5.61905861e-01 4.64034110e-01 -5.19295149e-02 8.40015896e-03 2.77748942e-01 -1.95202709e-03 7.38736451e-01 8.93556476e-01 -1.27391195e+00 -7.47533977e-01 -3.34692746e-01 1.18537009e-01 7.82470554e-02 -3.51345927e-01 -2.81465679e-01 -5.37880838e-01 -4.17203784e-01 5.02030790e-01 -6.90864742e-01 -1.76862076e-01 5.63193440e-01 -3.98714423e-01 -1.07647188e-01 6.00956500e-01 -3.88138980e-01 1.03247368e+00 -1.89708507e+00 -2.68376321e-01 2.16119245e-01 4.16999489e-01 4.29782897e-01 7.52929524e-02 5.39161339e-02 4.39593464e-01 -4.79135156e-01 -6.07846797e-01 -4.20025706e-01 -2.56620854e-01 3.61081153e-01 -9.22354385e-02 6.01086557e-01 2.00823739e-01 6.69512451e-01 -8.75160396e-01 -7.71500945e-01 7.33865321e-01 4.83574182e-01 -7.36465216e-01 1.20426640e-01 -2.51897961e-01 7.55488575e-01 -4.21292365e-01 7.32715011e-01 8.97376537e-01 -3.22535157e-01 1.55367227e-02 -5.83391964e-01 -2.10947976e-01 1.94249481e-01 -1.17603874e+00 1.54100335e+00 -6.80101931e-01 8.59269202e-01 -1.14254519e-01 -1.16829979e+00 1.15562272e+00 -1.34713635e-01 5.97515345e-01 -9.64953125e-01 1.68185160e-01 2.81512171e-01 -1.17957816e-01 -5.88442564e-01 3.10486197e-01 -1.15822546e-01 3.86993140e-01 1.03555910e-01 -7.19379559e-02 -2.53857255e-01 2.18327165e-01 -1.32604018e-01 7.11601973e-01 1.25901505e-01 1.80890128e-01 -3.50172997e-01 8.10789227e-01 -1.72602803e-01 1.07072067e+00 5.47341287e-01 -1.43849641e-01 5.41341126e-01 9.81192738e-02 -4.30705845e-01 -8.00512493e-01 -1.06079662e+00 -5.72283208e-01 3.79054785e-01 6.14432812e-01 -9.15724635e-02 -6.12201512e-01 -3.92628849e-01 9.47825834e-02 4.91407633e-01 -2.70372331e-01 2.02487662e-01 -5.13573587e-01 -8.26294541e-01 1.33533970e-01 4.46591794e-01 1.04313087e+00 -6.98695540e-01 -5.65077960e-01 6.27151951e-02 -5.03241718e-01 -1.52351773e+00 -1.58076659e-01 -2.05305085e-01 -1.10778761e+00 -1.14455545e+00 -4.06948179e-01 -5.69070876e-01 9.36770320e-01 3.51053596e-01 1.01089871e+00 2.20817596e-01 -4.60661173e-01 2.07764193e-01 -4.34401780e-02 -5.33037484e-01 6.01423346e-02 -2.46783748e-01 -8.47933069e-02 2.65217751e-01 4.25248295e-01 -8.20425212e-01 -8.45057011e-01 5.25514662e-01 -7.26387620e-01 1.88036442e-01 5.82909405e-01 8.00618649e-01 9.16348934e-01 2.31134936e-01 3.06766987e-01 -7.64296949e-01 -1.64536998e-01 1.90536663e-01 -1.27941740e+00 3.26834794e-04 -9.59988713e-01 1.00319825e-01 5.78787267e-01 -9.96477380e-02 -1.27573717e+00 1.83039203e-01 -9.71978828e-02 -2.26339325e-01 -6.53719902e-02 1.51033387e-01 -2.49728993e-01 -4.18150336e-01 5.70217490e-01 2.56669492e-01 6.43699914e-02 -4.64409977e-01 -2.49729510e-02 5.22361040e-01 6.14933491e-01 -6.22439742e-01 1.06780827e+00 9.00971234e-01 4.96454179e-01 -9.30064738e-01 -9.76292789e-01 -5.64692140e-01 -5.78840256e-01 -4.05673444e-01 6.97775841e-01 -9.00731683e-01 -8.32355618e-01 7.63847649e-01 -9.66900766e-01 -2.10154429e-01 -1.94423318e-01 8.00436735e-01 -6.07959270e-01 5.16327918e-01 -4.08375353e-01 -8.64316881e-01 -1.45827413e-01 -1.26704013e+00 8.28507841e-01 4.00724709e-01 5.10804713e-01 -1.06577706e+00 -1.96974516e-01 7.12507308e-01 4.19892013e-01 2.72086799e-01 8.21901202e-01 1.30721524e-01 -1.22640193e+00 2.33623199e-02 -6.24235094e-01 5.83924949e-01 2.80665249e-01 -2.72253007e-02 -1.18155634e+00 -2.18869895e-01 4.71277460e-02 -2.79594272e-01 8.46366882e-01 6.59792304e-01 1.38350010e+00 -5.46248704e-02 -1.18347295e-01 9.69345629e-01 1.87653673e+00 -7.69974515e-02 9.12088394e-01 3.05139989e-01 7.27813959e-01 7.37929761e-01 6.93500400e-01 3.25177163e-01 5.45037508e-01 7.97171593e-01 6.85247600e-01 -2.79493004e-01 -4.49442625e-01 -2.52887130e-01 -4.54956666e-02 8.51435065e-01 -2.83242315e-01 -7.31525570e-02 -6.82832599e-01 4.64409381e-01 -1.76056552e+00 -8.16245317e-01 -3.22948486e-01 2.58735728e+00 8.23122084e-01 1.75674468e-01 -1.80117711e-01 2.69094586e-01 6.10169470e-01 6.29162714e-02 -8.48000526e-01 1.59350768e-01 -2.19800934e-01 1.95507392e-01 6.04771197e-01 7.05928743e-01 -9.13575828e-01 8.27530026e-01 5.82823133e+00 8.05346251e-01 -1.26920247e+00 6.56813383e-02 4.32835430e-01 5.34302816e-02 -1.94929406e-01 4.39997055e-02 -8.79765034e-01 4.42873329e-01 1.62527576e-01 1.72717497e-01 3.09070081e-01 6.78770840e-01 3.03406864e-01 -5.07951319e-01 -9.87904787e-01 1.28900456e+00 2.99995653e-02 -1.19780242e+00 3.83131132e-02 1.39401376e-01 7.92026401e-01 -1.31482035e-02 -1.18460968e-01 -1.70379609e-01 -7.40836486e-02 -6.74360514e-01 4.34967816e-01 6.15925491e-01 8.47336113e-01 -6.70579910e-01 7.66882181e-01 4.16534424e-01 -1.02465093e+00 2.38726730e-03 -6.71622038e-01 -1.21098235e-01 5.53396866e-02 1.34348357e+00 -3.72431725e-01 6.18771195e-01 5.76211631e-01 8.74489903e-01 -5.00667036e-01 1.37851143e+00 -5.07633746e-01 5.04962564e-01 -4.97071028e-01 9.09388214e-02 -9.38357860e-02 -4.55602705e-01 6.01104736e-01 8.67075801e-01 8.61342773e-02 8.52857344e-03 1.70026794e-01 8.37297380e-01 6.47813678e-02 9.89940539e-02 -3.93186480e-01 4.59220946e-01 4.23490733e-01 1.16585052e+00 -6.11221075e-01 -1.32722780e-01 -4.89915669e-01 4.92718637e-01 3.49480689e-01 4.69124287e-01 -7.10935116e-01 -4.09307986e-01 5.79602718e-01 2.94231534e-01 1.48434877e-01 -1.80626929e-01 -3.02385300e-01 -1.32239306e+00 3.40046257e-01 -4.70816165e-01 8.53558522e-05 -6.87273681e-01 -1.33659351e+00 4.52464640e-01 -7.61434436e-02 -1.54771745e+00 -9.21152383e-02 -7.39237964e-01 -4.64270890e-01 6.53350592e-01 -2.07724690e+00 -1.09375942e+00 -7.97857165e-01 6.84974492e-01 3.04007679e-01 -2.35682484e-02 4.05658811e-01 6.09153807e-01 -7.22540915e-01 6.17042840e-01 -2.34410702e-03 1.74030781e-01 6.61781490e-01 -1.02758813e+00 -2.77791500e-01 1.03608799e+00 6.98018968e-02 4.35569376e-01 6.08416021e-01 -2.87051439e-01 -1.29873049e+00 -1.02442813e+00 6.67464375e-01 -4.35609594e-02 3.60366881e-01 -1.73045337e-01 -9.19410884e-01 3.38469803e-01 -3.50522786e-01 1.13508739e-01 2.85096943e-01 2.26393975e-02 -3.00335288e-01 -7.94980466e-01 -1.29930961e+00 3.66045922e-01 1.14970350e+00 -4.02701944e-01 -3.07193339e-01 3.09471250e-01 3.37961406e-01 -3.97985846e-01 -6.81008935e-01 7.28626430e-01 5.22381008e-01 -1.59308970e+00 1.02826917e+00 1.94517806e-01 4.66909200e-01 -6.53690636e-01 -1.64416045e-01 -7.49141634e-01 -6.58532381e-02 -4.05379951e-01 -1.47639558e-01 1.29528427e+00 1.49911523e-01 -1.14275789e+00 9.31810498e-01 4.07113433e-01 5.09952269e-02 -8.13597083e-01 -9.39044595e-01 -9.14157510e-01 -2.97535926e-01 -6.73331141e-01 5.51222444e-01 7.10849524e-01 -6.00087762e-01 1.82222486e-01 -2.74662793e-01 3.95045787e-01 1.37273037e+00 3.68112326e-01 9.70870018e-01 -1.43031597e+00 -1.36126444e-01 -3.55038494e-01 -8.68595004e-01 -1.36841285e+00 7.55441785e-02 -6.13968849e-01 2.55204260e-01 -1.57149816e+00 1.62714422e-01 -9.64988410e-01 -3.04429717e-02 3.40418190e-01 -6.34344891e-02 6.25144780e-01 -1.25774413e-01 4.09740448e-01 -4.13112074e-01 6.02732301e-01 1.41851830e+00 -1.72386959e-01 -1.40297264e-01 7.83044696e-02 -4.02771503e-01 1.03155982e+00 6.14609003e-01 -2.93920606e-01 -6.06062651e-01 -6.37903035e-01 6.48454726e-02 -1.10290304e-01 5.78546464e-01 -1.29317582e+00 2.96609700e-01 -2.26338819e-01 2.62969792e-01 -6.34152830e-01 3.82778138e-01 -7.91976213e-01 -2.67451666e-02 3.36299986e-01 1.00145675e-01 -6.83187306e-01 3.55883539e-02 4.37344193e-01 -4.09658790e-01 -3.06236714e-01 1.08060229e+00 2.48087391e-01 -7.75302410e-01 6.16225541e-01 4.36552651e-02 -5.28959893e-02 7.69999921e-01 -4.78504032e-01 -2.48880237e-01 -4.82905120e-01 -2.54387766e-01 1.88000694e-01 6.67206466e-01 -1.22192912e-01 6.43248022e-01 -1.26704895e+00 -6.33219361e-01 3.41067374e-01 3.55415910e-01 4.07459199e-01 1.76068559e-01 9.23206568e-01 -6.30265713e-01 2.09083527e-01 -7.38151148e-02 -9.83340859e-01 -9.30252194e-01 3.24791104e-01 3.26887935e-01 -6.44620955e-02 -8.27607691e-01 6.06073856e-01 4.78008598e-01 -2.88626850e-01 3.85702759e-01 -3.85723799e-01 -1.86412353e-02 -1.74318567e-01 6.06546104e-01 4.38503951e-01 3.29923704e-02 -5.98153651e-01 -2.65785068e-01 1.20771754e+00 2.70261824e-01 3.29933614e-01 1.16510582e+00 -2.83788025e-01 -2.71957695e-01 2.26302937e-01 1.00018418e+00 1.39172718e-01 -1.49079442e+00 -5.87719917e-01 -2.49682069e-01 -8.19769800e-01 4.69618201e-01 -6.03479743e-01 -1.46766210e+00 1.08224881e+00 6.77979171e-01 -1.84006110e-01 1.49713242e+00 -3.62410754e-01 9.01179314e-01 4.86078352e-01 7.71284223e-01 -1.03209615e+00 -1.23433836e-01 1.23112552e-01 4.50276017e-01 -1.59844899e+00 3.47583473e-01 -8.59754205e-01 -2.04621717e-01 1.03531086e+00 6.99726224e-01 4.58924891e-03 6.18691802e-01 1.43453524e-01 1.03275172e-01 -7.70020708e-02 -2.85447389e-01 -5.02612352e-01 3.81093204e-01 9.17878211e-01 3.76832560e-02 -2.01358229e-01 -1.40117675e-01 -6.54842928e-02 -9.32204798e-02 1.76067442e-01 2.50563473e-01 5.65942466e-01 -5.86386800e-01 -1.02814913e+00 -2.29352579e-01 5.01345277e-01 -2.21087888e-01 -7.08653927e-02 2.22632214e-01 7.35217631e-01 3.11211973e-01 1.01047254e+00 2.36657336e-01 -8.24271664e-02 2.76302010e-01 -4.22535211e-01 7.91286528e-01 -3.59136492e-01 6.92775324e-02 -1.20052777e-01 -1.31930321e-01 -6.90959871e-01 -8.64315927e-01 -4.25290704e-01 -1.07862759e+00 -3.29749078e-01 -5.81677914e-01 -1.85493931e-01 6.63875997e-01 1.02444279e+00 1.99146748e-01 1.98835999e-01 8.58525276e-01 -7.74755895e-01 -1.42875969e-01 -7.28168249e-01 -4.86745149e-01 2.88680375e-01 4.16413695e-01 -1.05847299e+00 -5.59952855e-01 5.72922342e-02]
[9.307449340820312, -2.470959424972534]
cd447d01-1019-430f-8615-2564e94ac66b
image-steganography-based-on-style-transfer
2203.04500
null
https://arxiv.org/abs/2203.04500v1
https://arxiv.org/pdf/2203.04500v1.pdf
Image Steganography based on Style Transfer
Image steganography is the art and science of using images as cover for covert communications. With the development of neural networks, traditional image steganography is more likely to be detected by deep learning-based steganalysis. To improve upon this, we propose image steganography network based on style transfer, and the embedding of secret messages can be disguised as image stylization. We embed secret information while transforming the content image style. In latent space, the secret information is integrated into the latent representation of the cover image to generate the stego images, which are indistinguishable from normal stylized images. It is an end-to-end unsupervised model without pre-training. Extensive experiments on the benchmark dataset demonstrate the reliability, quality and security of stego images generated by our steganographic network.
['Yaofei Wang', 'Jian Wang', 'Cong Yu', 'Yu Zhang', 'Donghui Hu']
2022-03-09
null
null
null
null
['steganalysis', 'image-stylization', 'image-steganography']
['computer-vision', 'computer-vision', 'computer-vision']
[ 1.16384995e+00 6.00608528e-01 2.94993129e-02 5.86726293e-02 7.18988776e-02 -4.72580910e-01 6.72357559e-01 -1.00522995e+00 2.14215256e-02 4.10338789e-01 1.46255419e-01 -5.19763529e-01 8.06079447e-01 -1.01406741e+00 -1.18560994e+00 -8.69154155e-01 -7.80669749e-02 -6.43299073e-02 -1.95591524e-01 -3.43762904e-01 2.15912148e-01 -1.40643999e-01 -9.56665397e-01 5.52399158e-01 6.82174802e-01 6.23557627e-01 -5.16626704e-03 6.90119207e-01 -8.60301778e-02 7.68580437e-01 -6.47814929e-01 -4.36822861e-01 4.31481093e-01 -1.10148442e+00 -7.17335641e-01 7.08853543e-01 -6.44469410e-02 -4.68964368e-01 -7.92547286e-01 1.58479977e+00 7.91717544e-02 -6.95899248e-01 4.91832048e-01 -1.18328512e+00 -1.38079429e+00 7.80387759e-01 -5.14234304e-01 -3.93074125e-01 2.14411914e-01 7.17181981e-01 3.33169937e-01 -3.51681948e-01 7.68600821e-01 1.59097254e+00 3.79567713e-01 7.90493727e-01 -1.17680907e+00 -1.04485118e+00 -3.76809567e-01 -6.94992542e-02 -8.94156873e-01 -4.15501922e-01 1.05536175e+00 -2.23240890e-02 3.07021439e-01 2.06156790e-01 9.63475406e-01 1.47551584e+00 8.02744925e-01 9.25695837e-01 1.50828445e+00 -5.19742906e-01 -2.77556330e-01 2.68608004e-01 -1.01109159e+00 9.61067319e-01 5.52874446e-01 7.07027078e-01 -1.52518630e-01 2.09088087e-01 1.11484206e+00 4.35826220e-02 -2.77518868e-01 -4.18866992e-01 -1.36020327e+00 1.17410851e+00 6.55861914e-01 7.52826333e-02 1.15504004e-01 5.82717001e-01 2.25567743e-01 9.92690861e-01 5.09889007e-01 3.78014266e-01 2.63238311e-01 4.55056071e-01 -8.15449357e-01 -3.17301601e-01 8.72359335e-01 9.91111100e-01 5.80390096e-01 6.36173129e-01 1.87271953e-01 5.13128154e-02 6.14176989e-01 1.10624611e+00 7.17212617e-01 -6.84164286e-01 2.63238013e-01 4.08517599e-01 -3.07244599e-01 -1.39786375e+00 4.14147973e-01 -3.38953763e-01 -1.31177831e+00 4.24934387e-01 -9.13536921e-02 -2.09036127e-01 -1.37255108e+00 1.17526340e+00 -1.97791591e-01 4.51681376e-01 4.55470026e-01 3.96214873e-01 6.46368861e-01 7.81763077e-01 -2.93472558e-01 -2.72916369e-02 1.19044578e+00 -1.07603276e+00 -9.83063459e-01 -5.33739209e-01 6.06851995e-01 -8.88798714e-01 5.97573876e-01 2.35101096e-02 -8.58036816e-01 -5.74086964e-01 -1.33295846e+00 4.96939868e-01 -2.68202901e-01 -6.15194440e-01 2.91016877e-01 1.18732655e+00 -9.83217359e-01 4.42378312e-01 -3.05133730e-01 9.79663990e-03 8.64246964e-01 3.57353628e-01 -5.31178832e-01 8.66445899e-02 -1.57397258e+00 3.86934429e-01 9.39269841e-01 -1.01834148e-01 -1.11406612e+00 1.05774542e-02 -1.09307766e+00 -1.87434107e-01 1.32332429e-01 -5.19389868e-01 6.28406048e-01 -1.64821100e+00 -1.70437622e+00 1.21808481e+00 1.53448999e-01 -7.06034720e-01 7.56141126e-01 3.57100278e-01 -6.76924765e-01 3.10744822e-01 -6.95570856e-02 8.01684260e-01 1.82230461e+00 -1.65207279e+00 -3.83121878e-01 3.56539518e-01 -3.41063887e-01 -9.82593298e-02 -1.48714244e-01 -3.18799853e-01 -1.67398840e-01 -9.35832024e-01 1.58196777e-01 -1.36733341e+00 -9.30018723e-02 -2.54759938e-01 -6.35306537e-01 5.57783425e-01 1.51686156e+00 -8.22113574e-01 6.18826091e-01 -2.24384785e+00 2.86300499e-02 5.67235351e-01 4.37103450e-01 5.45129776e-01 -3.12095284e-01 3.10311884e-01 -3.34720880e-01 6.63079798e-01 -3.28369051e-01 1.68443054e-01 -2.94353008e-01 2.31350496e-01 -5.14966667e-01 6.43143892e-01 7.43417861e-03 1.67186272e+00 -9.87035453e-01 -4.74550545e-01 1.86621651e-01 3.21660042e-01 -2.48783886e-01 1.27775967e-01 -1.00193962e-01 9.10417497e-01 -2.66816586e-01 2.30160683e-01 8.24356437e-01 -4.79231238e-01 2.71388978e-01 4.45904955e-02 6.32368565e-01 -2.12632924e-01 -5.33795118e-01 1.24316669e+00 -2.37730905e-01 9.63290930e-01 -3.84025276e-01 -8.58521402e-01 1.00277781e+00 5.64124405e-01 8.59675556e-02 -5.44286072e-01 5.47019422e-01 2.91423649e-01 -5.63050248e-02 -6.97670460e-01 8.51335302e-02 -1.98246896e-01 -2.40169734e-01 8.72545362e-01 -1.88753158e-01 -3.59919339e-01 -4.99727309e-01 5.60999736e-02 6.30266726e-01 1.81733130e-03 -1.03737852e-02 1.55770481e-01 5.30630112e-01 -4.07920659e-01 -2.74956264e-02 7.47995496e-01 1.97498918e-01 3.65553856e-01 4.99228865e-01 -4.67992246e-01 -1.64054573e+00 -7.87339807e-01 3.96318823e-01 2.78198659e-01 4.09284532e-01 2.09569544e-01 -1.18714619e+00 -9.97423768e-01 -3.26579213e-01 4.70626265e-01 -5.46123028e-01 -8.31465840e-01 -6.08835042e-01 -3.11103761e-01 9.26561117e-01 -2.66642570e-01 1.41434264e+00 -1.65834141e+00 -1.78709000e-01 2.03960761e-01 -3.68077695e-01 -1.11663795e+00 -6.16361618e-01 -4.54930007e-01 -7.66198993e-01 -8.92302334e-01 -9.09087420e-01 -1.27468133e+00 1.04529154e+00 4.55660164e-01 7.52645671e-01 5.79042137e-01 1.66545391e-01 -2.24003810e-02 -4.74949837e-01 -3.66250813e-01 -1.54041851e+00 -4.98761050e-03 -1.70507714e-01 3.74419719e-01 7.16613829e-02 -4.93579775e-01 -5.51448584e-01 3.47717315e-01 -1.54450679e+00 5.78150868e-01 1.10225618e+00 1.11221766e+00 7.04780295e-02 5.34264922e-01 5.58391027e-02 -1.44571602e+00 4.23080623e-01 -2.58582830e-01 -3.97305280e-01 1.43949926e-01 -7.30906963e-01 3.56209308e-01 3.75103056e-01 -6.59686744e-01 -7.27056682e-01 -2.16588750e-01 5.15826568e-02 -4.51317847e-01 -3.70750437e-03 3.28611314e-01 -3.14612150e-01 -7.16573775e-01 3.57979655e-01 9.70473588e-01 7.88301229e-01 5.80419973e-03 4.86878812e-01 7.51740515e-01 4.63131189e-01 2.70464391e-01 1.71014786e+00 8.45782280e-01 -2.88932472e-02 -7.15386927e-01 -4.23618734e-01 4.10636723e-01 -3.25095832e-01 -5.90556599e-02 8.28599513e-01 -7.37132132e-01 -5.46140313e-01 1.09966314e+00 -1.14521790e+00 -2.21776709e-01 -1.38299331e-01 -1.01305433e-02 -6.11090004e-01 7.90299356e-01 -5.06500006e-01 -2.60441095e-01 -3.30033362e-01 -1.31670189e+00 5.88623762e-01 -4.20882143e-02 -1.81426704e-02 -1.25816119e+00 -4.17322874e-01 3.34526896e-01 3.96476924e-01 7.25730836e-01 8.54581177e-01 -2.54811764e-01 -8.60235751e-01 -4.83553141e-01 -1.79086015e-01 6.08323336e-01 4.20008004e-01 -5.52473485e-01 -6.60347283e-01 -7.61162937e-01 4.54290330e-01 -3.85444611e-02 1.06969392e+00 -3.71234561e-03 1.05723143e+00 -1.07073152e+00 -3.58667612e-01 1.15958428e+00 1.45837963e+00 3.19361120e-01 1.39625585e+00 4.13360029e-01 1.18843186e+00 5.79957366e-01 -1.87968686e-01 -2.21176058e-01 7.71172270e-02 -6.22959249e-02 4.87462610e-01 -5.61878800e-01 -1.69923410e-01 -6.68593884e-01 6.28627479e-01 7.48135388e-01 1.65675744e-01 -6.59007132e-01 -4.62990582e-01 2.62803972e-01 -1.28339410e+00 -1.19614029e+00 -4.40965779e-03 1.59754157e+00 7.74308920e-01 5.37419200e-01 -4.03180718e-01 2.23111898e-01 1.01580954e+00 6.96443141e-01 -4.98795509e-01 -4.32118624e-01 -3.71459097e-01 -3.15435641e-02 1.11979139e+00 3.78409058e-01 -1.13322520e+00 1.38315105e+00 6.35417747e+00 9.62229073e-01 -1.38981390e+00 1.15490533e-01 8.99070382e-01 5.45769215e-01 -6.59128964e-01 1.42881021e-01 -2.06486389e-01 9.09129620e-01 7.58833647e-01 -1.69153195e-02 7.87011206e-01 3.65409613e-01 -1.26628932e-02 3.29400361e-01 -6.28860414e-01 8.78794730e-01 2.95980930e-01 -1.54994297e+00 5.23751497e-01 5.30224741e-01 1.23877811e+00 -3.61095786e-01 8.66716802e-01 -9.82802287e-02 3.67675930e-01 -1.40864861e+00 6.81366086e-01 2.56361902e-01 1.40445650e+00 -6.82388723e-01 7.55642891e-01 9.24204662e-02 -7.48465359e-01 1.30766645e-01 -2.54955053e-01 2.49091372e-01 6.28838241e-02 3.85094993e-02 -7.65082419e-01 3.20442885e-01 2.94653922e-01 8.72798324e-01 -5.06875634e-01 2.96500206e-01 -7.31642306e-01 1.00901592e+00 2.02825382e-01 1.48999944e-01 6.28709018e-01 -1.36195257e-01 8.84925187e-01 9.04453158e-01 4.14663464e-01 -3.78078222e-01 -1.68204829e-01 9.21666503e-01 -4.18949574e-01 -4.51309294e-01 -1.30325866e+00 -5.51621795e-01 8.28894414e-03 7.69583523e-01 -9.61436272e-01 -5.24548709e-01 6.17557652e-02 1.59867823e+00 -6.31397724e-01 5.23352921e-01 -7.39777327e-01 -5.40708959e-01 -1.95548311e-02 1.63479909e-01 6.25636339e-01 -3.59580740e-02 -3.25401247e-01 -1.21412921e+00 -4.18753564e-01 -1.49024820e+00 -2.98270255e-01 -6.93044662e-01 -9.25371110e-01 5.90471208e-01 -2.79066443e-01 -1.56588709e+00 -1.65220544e-01 -3.78695101e-01 -8.35082114e-01 6.86341763e-01 -1.48355675e+00 -1.64043379e+00 -1.53328106e-01 4.29236799e-01 2.29943514e-01 -7.72442281e-01 7.32234240e-01 -3.69117230e-01 3.14999036e-02 7.99047530e-01 4.88790035e-01 7.06467867e-01 4.01967466e-01 -7.43587017e-01 1.15983117e+00 1.00238621e+00 -1.16107762e-02 3.00747842e-01 1.02238810e+00 -1.10789287e+00 -1.14076269e+00 -1.30546546e+00 4.11181062e-01 -1.91538036e-01 4.51017827e-01 -4.96272355e-01 -7.39203274e-01 9.26149487e-01 7.89400220e-01 -3.47433358e-01 4.13138330e-01 -9.97802079e-01 -4.24413741e-01 4.66467857e-01 -1.35530710e+00 6.67607963e-01 9.24205363e-01 -6.37367725e-01 -4.57907706e-01 1.61817327e-01 1.07998538e+00 -3.53936791e-01 -3.24197799e-01 -8.82455800e-03 6.10161483e-01 -7.00581133e-01 1.01008213e+00 -2.89554507e-01 8.56924057e-01 -1.01651661e-01 3.45953375e-01 -1.50221324e+00 -2.12126628e-01 -1.25257218e+00 -3.08822673e-02 7.87665486e-01 1.80959329e-01 -9.04683471e-01 9.35334980e-01 -2.65518337e-01 5.19032180e-01 -1.84877947e-01 -4.95259941e-01 -8.04576755e-01 1.33809865e-01 1.77338243e-01 1.08088183e+00 1.03878462e+00 -5.38727164e-01 1.25044640e-02 -1.26668847e+00 6.96193501e-02 1.22105396e+00 -1.62731782e-01 8.83723617e-01 -7.95989633e-01 -1.99485570e-01 -1.74112245e-01 -5.68065763e-01 -9.60942805e-01 2.40882695e-01 -1.00497460e+00 -9.77784917e-02 -1.09478378e+00 -1.71748340e-01 -8.67826268e-02 4.00149748e-02 1.29977748e-01 3.45813222e-02 9.62593615e-01 7.69293755e-02 5.45430839e-01 -3.37140888e-01 4.25680131e-01 2.06151295e+00 -6.55612051e-01 3.58816907e-02 -1.45366102e-01 -7.72059619e-01 8.19694221e-01 9.41697836e-01 -8.54844689e-01 -3.68101746e-01 -3.31828713e-01 3.01334113e-01 6.20285654e-03 5.45715988e-01 -9.03095841e-01 -2.65031248e-01 -1.18989378e-01 6.00422084e-01 -2.34619811e-01 -7.76956463e-03 -9.36614454e-01 3.28571677e-01 1.27243483e+00 -4.20543253e-01 -4.21395540e-01 -2.33103305e-01 8.61067414e-01 -9.44397226e-02 -2.02895239e-01 9.76700008e-01 -5.86837053e-01 -6.69056833e-01 3.55497867e-01 -6.49715602e-01 -2.04065740e-01 1.00195050e+00 -6.56170487e-01 -9.36473161e-02 -7.92638779e-01 -4.85585779e-01 -1.73593357e-01 7.46141553e-01 5.25335670e-01 1.07938278e+00 -1.50191522e+00 -7.27607846e-01 9.33422029e-01 -2.70823948e-03 -3.28184575e-01 7.95336887e-02 3.49801153e-01 -1.05181062e+00 2.20495135e-01 -5.67874610e-01 -4.44502920e-01 -1.26919043e+00 7.08390594e-01 2.62373954e-01 -3.32268864e-01 -8.06777060e-01 6.09924078e-01 3.32244694e-01 -1.02723546e-01 -2.83611536e-01 2.09204242e-01 -1.62404314e-01 -4.68832135e-01 4.79656488e-01 -1.31171897e-01 -8.50008667e-01 -8.73376310e-01 3.64177257e-01 4.18100655e-01 -2.92153418e-01 -1.62769318e-01 1.12109232e+00 -4.88357902e-01 -4.23242599e-01 -6.40596002e-02 1.57677960e+00 -2.61985630e-01 -1.11838579e+00 -4.30263460e-01 -2.63929844e-01 -7.12813199e-01 4.15760167e-02 -3.14068168e-01 -1.23195159e+00 7.63237536e-01 9.56206262e-01 5.64972222e-01 7.74791777e-01 -3.65481675e-01 1.47857809e+00 3.70887160e-01 2.91753799e-01 -7.90087759e-01 4.57085758e-01 4.28012013e-01 7.47237265e-01 -1.43332899e+00 -2.40154952e-01 -2.43206158e-01 -6.95832789e-01 1.05082428e+00 6.72807172e-02 -5.76480448e-01 6.67306662e-01 -4.10432257e-02 2.74812788e-01 -3.26195061e-01 -2.57286113e-02 3.15752119e-01 1.79294184e-01 7.80939698e-01 -3.27126086e-01 7.09907338e-02 1.80470109e-01 -3.58714491e-01 -4.74707782e-01 -5.95089570e-02 7.29457796e-01 9.36061978e-01 -4.04748261e-01 -1.14132380e+00 -6.78070962e-01 2.93901294e-01 -6.99369550e-01 -2.24524617e-01 -3.46028060e-01 4.42420393e-01 1.85325950e-01 9.10292089e-01 -1.53117493e-01 -8.46442461e-01 -4.89293694e-01 -1.84965461e-01 4.11053389e-01 -3.55770171e-01 -1.23644635e-01 1.74272448e-01 -4.70182210e-01 -7.43446872e-02 -5.80362916e-01 3.64372432e-02 -8.26924503e-01 -8.64525497e-01 -3.02359670e-01 2.18896531e-02 5.82865953e-01 1.05381191e+00 2.29951199e-02 6.53202593e-01 1.19750869e+00 -7.34867871e-01 -2.57971257e-01 -7.53781617e-01 -4.74522740e-01 7.71159649e-01 5.28985620e-01 -6.18939959e-02 -6.02993250e-01 6.89303696e-01]
[4.306353569030762, 8.051715850830078]
b367d379-6264-48b9-981e-e2510f230109
direct-photometric-alignment-by-mesh
null
null
http://openaccess.thecvf.com/content_cvpr_2017/html/Lin_Direct_Photometric_Alignment_CVPR_2017_paper.html
http://openaccess.thecvf.com/content_cvpr_2017/papers/Lin_Direct_Photometric_Alignment_CVPR_2017_paper.pdf
Direct Photometric Alignment by Mesh Deformation
The choice of motion models is vital in applications like image/video stitching and video stabilization. Conventional methods explored different approaches ranging from simple global parametric models to complex per-pixel optical flow. Mesh-based warping methods achieve a good balance between computational complexity and model flexibility. However, they typically require high quality feature correspondences and suffer from mismatches and low-textured image content. In this paper, we propose a mesh-based photometric alignment method that minimizes pixel intensity difference instead of Euclidean distance of known feature correspondences. The proposed method combines the superior performance of dense photometric alignment with the efficiency of mesh-based image warping. It achieves better global alignment quality than the feature-based counterpart in textured images, and more importantly, it is also robust to low-textured image content. Abundant experiments show that our method can handle a variety of images and videos, and outperforms representative state-of-the-art methods in both image stitching and video stabilization tasks.
['Loong-Fah Cheong', 'Shuaicheng Liu', 'Nianjuan Jiang', 'Kaimo Lin', 'Minh Do', 'Jiangbo Lu']
2017-07-01
null
null
null
cvpr-2017-7
['image-stitching', 'video-stabilization']
['computer-vision', 'computer-vision']
[ 3.93860161e-01 -6.00267589e-01 -2.23179594e-01 1.12712957e-01 -5.13102591e-01 -6.16701901e-01 5.79373002e-01 -3.76273036e-01 -1.89176366e-01 4.12325591e-01 -6.88243983e-03 4.42966633e-02 -1.37675991e-02 -4.39947873e-01 -6.47782743e-01 -9.66643810e-01 4.62107688e-01 1.53363317e-01 5.36264658e-01 -3.39824677e-01 5.21179795e-01 3.96245539e-01 -1.46013021e+00 -3.49777520e-01 8.16269100e-01 1.01148057e+00 1.28045633e-01 3.35115522e-01 7.19262064e-02 3.82707298e-01 -1.97231412e-01 -1.78690240e-01 5.62826335e-01 -4.08664465e-01 -5.13953209e-01 5.37571669e-01 9.55964983e-01 -2.91429549e-01 -3.81857604e-01 1.09402609e+00 4.78164494e-01 2.27771312e-01 2.76935995e-01 -1.09355783e+00 -4.30159837e-01 -1.65052786e-01 -1.10112143e+00 3.84227671e-02 5.13744414e-01 3.09512585e-01 5.50659120e-01 -8.13762307e-01 7.07286656e-01 1.30003345e+00 7.59705722e-01 3.11679721e-01 -1.48267114e+00 -6.12657845e-01 -1.71970174e-01 2.22721659e-02 -1.31718397e+00 -6.44439995e-01 1.01839960e+00 -5.36479831e-01 2.98474759e-01 3.47482622e-01 6.24685943e-01 6.96477354e-01 4.50676531e-01 2.53903210e-01 1.10691142e+00 -4.42401469e-01 -1.12932526e-01 -4.91196632e-01 -2.91710466e-01 8.50356340e-01 2.49840707e-01 2.63705194e-01 -8.01280677e-01 -2.97526032e-01 1.52367759e+00 7.58384615e-02 -6.58419549e-01 -7.11290121e-01 -1.79788041e+00 5.70004046e-01 5.70212454e-02 2.63752550e-01 -3.83894555e-02 3.29277575e-01 2.20573306e-01 2.31693029e-01 3.31774592e-01 3.40581030e-01 1.82835758e-02 -7.12760687e-02 -1.08789623e+00 1.68592930e-01 4.49387878e-01 9.81328785e-01 1.01534784e+00 2.29073703e-01 9.46872234e-02 6.80283368e-01 1.72932371e-01 6.02304220e-01 5.26375651e-01 -1.44080055e+00 3.14895600e-01 2.34849229e-01 2.46905029e-01 -1.55881357e+00 -9.38348100e-02 1.20228052e-01 -8.81883264e-01 3.27015400e-01 3.22062880e-01 1.68540537e-01 -6.56859756e-01 1.62581503e+00 6.53799176e-01 5.14394760e-01 -3.24409008e-01 9.52911079e-01 5.25145829e-01 5.68088830e-01 -5.57834804e-01 -5.11452198e-01 1.23242569e+00 -1.19377279e+00 -1.01063979e+00 -1.23694301e-01 1.87907770e-01 -1.31073272e+00 8.64980817e-01 2.96751317e-02 -1.18666506e+00 -6.40003562e-01 -1.02324712e+00 -1.72532693e-01 3.70961964e-01 -5.24343662e-02 1.94787040e-01 5.70759237e-01 -1.21827793e+00 6.92681074e-01 -9.18642581e-01 -3.38951498e-01 -2.34491765e-01 3.70558262e-01 -5.07910728e-01 -6.63761422e-02 -7.96786070e-01 7.40979433e-01 7.88034946e-02 -5.40014431e-02 -4.16643858e-01 -6.96550310e-01 -9.74881709e-01 -2.00882107e-01 3.15671235e-01 -9.46434081e-01 9.85182703e-01 -1.10363448e+00 -2.03237438e+00 7.80252635e-01 -3.96265745e-01 -3.31571847e-02 6.20527565e-01 -1.61283568e-01 -6.95686787e-02 2.73679256e-01 -4.87065837e-02 5.85611105e-01 1.55472970e+00 -1.18482804e+00 -2.46995196e-01 -2.30591483e-02 -3.01165283e-01 3.15102458e-01 -2.41571903e-01 1.37862470e-02 -6.40167832e-01 -8.85147095e-01 3.07931393e-01 -1.06845653e+00 -1.69216812e-01 3.71892989e-01 -1.58836976e-01 3.99935484e-01 1.29620087e+00 -3.98066431e-01 1.21000779e+00 -2.17504954e+00 3.48192930e-01 3.15226912e-02 9.07472745e-02 3.10230821e-01 -1.80305600e-01 4.82724965e-01 3.38215171e-03 -3.10562491e-01 -1.03818074e-01 -3.39505464e-01 -3.77651781e-01 1.32557794e-01 -2.72467077e-01 8.36171448e-01 -6.01326153e-02 7.31289327e-01 -8.15650582e-01 -7.31055677e-01 6.34744883e-01 5.19531727e-01 -4.84706461e-01 8.77132919e-03 1.29469350e-01 8.73823702e-01 -4.04451996e-01 5.60168386e-01 8.60885262e-01 -8.68175104e-02 8.05815682e-02 -6.02206171e-01 -4.90448356e-01 -2.13085353e-01 -1.28967500e+00 1.81917632e+00 -1.72691271e-01 7.79193938e-01 4.08689529e-01 -6.79635227e-01 1.06159377e+00 2.36977145e-01 8.85017097e-01 -3.52403671e-01 2.97186315e-01 3.50503772e-01 -2.83822566e-01 -3.14884812e-01 7.39469290e-01 3.70984338e-02 2.32035518e-01 2.34552309e-01 -2.57441789e-01 -6.06161177e-01 5.47353849e-02 -1.88646138e-01 7.63446510e-01 3.69794250e-01 2.43169338e-01 -7.24869907e-01 8.72089446e-01 -1.64269075e-01 8.57994318e-01 2.48904914e-01 -2.08221853e-01 1.01892579e+00 1.65535092e-01 -6.42700255e-01 -1.33006501e+00 -6.77152872e-01 -1.55871734e-01 4.16550517e-01 8.93092513e-01 -6.37576997e-01 -8.69271636e-01 8.66491050e-02 -1.87388957e-01 -1.00135542e-02 -4.23897713e-01 6.92919791e-02 -9.71815646e-01 -2.86620080e-01 2.51435220e-01 2.71300264e-02 7.12837934e-01 -4.80716884e-01 -6.29475176e-01 1.90632597e-01 -4.50484216e-01 -1.23084009e+00 -1.20056880e+00 -7.05551088e-01 -1.07424378e+00 -1.15410936e+00 -8.35303128e-01 -9.57509100e-01 8.02991092e-01 9.61548209e-01 9.31442797e-01 2.20663264e-01 -1.26236767e-01 4.49711412e-01 -1.87072843e-01 2.69016773e-01 -3.31451714e-01 -1.73482388e-01 1.97211623e-01 4.97785330e-01 -3.05546880e-01 -5.10887384e-01 -9.34643805e-01 1.11761487e+00 -1.15994883e+00 2.79670537e-01 1.66569069e-01 1.09237874e+00 8.19894195e-01 -8.02980438e-02 -1.96666911e-01 -2.63055205e-01 3.08167696e-01 3.40025455e-01 -8.86234581e-01 2.05776602e-01 -3.29370528e-01 5.54723442e-02 3.37068796e-01 -5.85654855e-01 -9.39471841e-01 2.76354790e-01 2.31289431e-01 -8.03589940e-01 3.45694691e-01 1.35192871e-01 6.39484078e-02 -8.23955178e-01 5.03336430e-01 2.19891474e-01 5.90255260e-01 -3.09944928e-01 7.98540786e-02 1.35120645e-01 7.66190290e-01 -6.68913484e-01 1.15667498e+00 8.72026682e-01 2.72837400e-01 -1.05736816e+00 -3.46826226e-01 -4.19314861e-01 -6.72643661e-01 -3.74610931e-01 7.15552509e-01 -6.83831096e-01 -6.57022059e-01 9.70061004e-01 -1.11603820e+00 -1.56035438e-01 1.47141479e-02 5.00298321e-01 -8.71453047e-01 1.04179156e+00 -5.34665883e-01 -2.32499212e-01 -3.72032702e-01 -1.58952820e+00 1.25427306e+00 1.44495606e-01 -2.16291454e-02 -1.04650080e+00 2.66985357e-01 2.76256561e-01 5.77762783e-01 5.53321064e-01 3.96162271e-01 5.59036791e-01 -8.14834893e-01 -1.24869578e-01 8.23263898e-02 4.36711758e-02 5.29591739e-01 4.79942292e-01 -5.44351339e-01 -6.38680518e-01 3.07633847e-01 -2.71868538e-02 5.13896346e-01 6.12041593e-01 8.09603810e-01 -2.01215193e-01 -2.79136717e-01 8.73861849e-01 1.46177781e+00 -9.51208025e-02 8.45402777e-01 5.70216656e-01 7.91298985e-01 4.61305648e-01 1.01949275e+00 2.96190947e-01 2.70813610e-02 1.26221108e+00 4.46114093e-01 -2.78235942e-01 -9.02771577e-02 -1.63125284e-02 4.62499559e-01 9.24431205e-01 -2.13640764e-01 3.61796618e-02 -5.98852038e-01 3.26723814e-01 -2.18634129e+00 -9.03762460e-01 -2.56450862e-01 2.49965930e+00 8.08463037e-01 -3.27063054e-01 2.44283974e-02 5.40596172e-02 1.15134180e+00 4.43458587e-01 -2.95322597e-01 -1.87152818e-01 -3.08095872e-01 -1.93311751e-01 6.75302625e-01 6.93736553e-01 -1.08134472e+00 9.98864532e-01 6.39933872e+00 9.45124447e-01 -1.23307228e+00 3.01403031e-02 2.82379180e-01 8.78374279e-02 -2.72153169e-01 7.35325292e-02 -5.87170899e-01 5.48686385e-01 3.43215019e-01 -7.98612759e-02 2.53750205e-01 3.56534332e-01 4.09257323e-01 -1.87821701e-01 -7.97804713e-01 1.41683483e+00 1.92014307e-01 -1.49325597e+00 -8.94346647e-03 9.76687372e-02 1.06538689e+00 -2.77667671e-01 1.53453276e-01 -7.33064353e-01 -1.87879220e-01 -6.65704846e-01 7.92224586e-01 3.13030064e-01 8.67311597e-01 -5.59668243e-01 5.35585940e-01 7.79697895e-02 -1.40801966e+00 2.56558388e-01 -2.88771957e-01 1.47328556e-01 4.16316420e-01 4.33049738e-01 8.97171125e-02 5.99341273e-01 5.99244356e-01 8.82305443e-01 -9.63732824e-02 1.16572046e+00 2.08804011e-02 3.49082798e-02 -4.20679599e-01 4.21407491e-01 2.06513643e-01 -7.44814157e-01 7.17746139e-01 7.19068587e-01 6.28119290e-01 1.43145978e-01 2.54431516e-01 4.08431530e-01 2.34043449e-01 9.63503718e-02 -6.08878255e-01 3.98021162e-01 4.72785205e-01 1.19883740e+00 -6.08582616e-01 -2.08745241e-01 -4.56800967e-01 1.01563382e+00 -1.84554175e-01 2.43750796e-01 -9.11099434e-01 -9.72705930e-02 9.20539796e-01 3.22577775e-01 3.58616531e-01 -7.27589250e-01 -4.41453420e-02 -1.42919123e+00 -7.76239252e-03 -1.01778471e+00 -2.26825606e-02 -7.01561511e-01 -8.77210081e-01 5.34729362e-01 -1.07277840e-01 -1.91716051e+00 -9.60700512e-02 -4.22532350e-01 -6.21370614e-01 5.93578041e-01 -1.53566027e+00 -1.09006619e+00 -6.35652781e-01 9.36920345e-01 5.78201234e-01 2.30747312e-01 4.90931243e-01 2.76753485e-01 -4.94423032e-01 4.04789716e-01 5.27948141e-01 -8.73977765e-02 1.20456755e+00 -5.30339539e-01 3.82808685e-01 1.01708794e+00 -2.80248582e-01 5.93845308e-01 7.37461269e-01 -4.99253482e-01 -1.75322008e+00 -8.06862354e-01 4.81297642e-01 -1.51374489e-01 5.96430600e-01 1.42507866e-01 -8.31072032e-01 4.63371277e-01 4.33823586e-01 8.80063549e-02 2.46477038e-01 -7.08079040e-01 -2.59341866e-01 -2.63173819e-01 -1.08881783e+00 7.31610835e-01 8.47676992e-01 -3.08085114e-01 -2.50161171e-01 1.19334981e-01 3.94637555e-01 -8.04512858e-01 -9.84876394e-01 3.52476567e-01 6.03131473e-01 -1.08459854e+00 1.05353761e+00 2.98356652e-01 2.86292076e-01 -8.28359425e-01 1.15268447e-01 -1.06873572e+00 -6.24071121e-01 -1.47194266e+00 9.20014605e-02 1.22448039e+00 -2.07627013e-01 -6.66186094e-01 6.46672249e-01 5.19889116e-01 -5.11482498e-03 -3.92643720e-01 -1.01087785e+00 -8.98750305e-01 -2.45707229e-01 2.05928162e-01 4.30005163e-01 1.12969387e+00 -1.42037794e-01 1.29537201e-02 -8.28933418e-01 1.08302243e-01 9.45854545e-01 3.38554144e-01 1.03873181e+00 -9.52173114e-01 -7.40440562e-02 -3.91984701e-01 -7.27953911e-01 -1.28845024e+00 2.84971949e-02 -2.16721192e-01 -6.44365251e-02 -9.59511340e-01 -1.23706430e-01 -3.64093274e-01 2.55194724e-01 1.34391800e-01 -1.20404363e-01 5.39309680e-01 1.67143464e-01 7.09630251e-01 -1.46320939e-01 5.58792651e-01 1.56105113e+00 -6.47977069e-02 -3.16707730e-01 -2.21451387e-01 -5.96471271e-03 6.31549656e-01 7.16690242e-01 -1.77326784e-01 -2.60567605e-01 -7.81996608e-01 -8.10287669e-02 2.05734104e-01 2.66958654e-01 -1.01276362e+00 2.85375357e-01 -5.11473298e-01 7.86227640e-03 -3.61169100e-01 4.84827220e-01 -9.38965321e-01 6.47859514e-01 5.33261597e-01 9.01056379e-02 3.26774895e-01 -1.68322418e-02 5.25355279e-01 -4.79499817e-01 -1.34861797e-01 1.25021064e+00 -2.80490436e-04 -6.11579239e-01 4.93018568e-01 -1.85066491e-01 -2.90321052e-01 1.03249085e+00 -7.06165195e-01 -2.83277929e-01 -4.17595595e-01 -1.03835620e-01 -1.14379205e-01 1.30456054e+00 4.38618541e-01 6.01586342e-01 -1.51735723e+00 -5.35635650e-01 4.15539443e-01 -1.19485065e-01 1.95388496e-02 1.60193622e-01 1.23280406e+00 -1.13922942e+00 1.43358946e-01 -4.40935612e-01 -1.08454752e+00 -1.53312433e+00 5.21130621e-01 3.80899459e-01 8.34971070e-02 -6.03913128e-01 4.91053253e-01 4.22056794e-01 -7.61901811e-02 -9.99263823e-02 -2.02748328e-01 2.56840736e-01 -2.56096393e-01 5.90408623e-01 5.30075788e-01 1.27471862e-02 -1.10484576e+00 -1.77151278e-01 1.63445115e+00 1.96167499e-01 -1.69882759e-01 9.53022122e-01 -4.68476355e-01 -3.48041803e-01 -3.07890289e-02 1.17757642e+00 9.27213803e-02 -1.68922865e+00 -3.68856609e-01 -3.69670093e-01 -1.20697975e+00 1.37039721e-01 2.34248057e-01 -1.41247284e+00 6.19487107e-01 4.27841216e-01 5.98158734e-03 1.17460966e+00 -4.18469310e-01 1.06765807e+00 9.89838839e-02 4.75222707e-01 -9.46326494e-01 3.72463495e-01 4.88905370e-01 7.56317139e-01 -1.05306780e+00 3.24526668e-01 -7.93714106e-01 -2.74359375e-01 1.35824120e+00 4.69795465e-01 -3.54411989e-01 3.70259434e-01 2.11911783e-01 2.37947941e-01 8.82223323e-02 -3.96313727e-01 1.43435866e-01 3.39217484e-01 5.16920745e-01 2.33285993e-01 -4.66199845e-01 -4.28269207e-01 -6.17637992e-01 1.06044330e-01 -1.48961276e-01 4.52157229e-01 1.02306807e+00 -3.91663224e-01 -1.35144377e+00 -8.97767961e-01 -1.84810981e-01 -3.51788491e-01 5.98745868e-02 -8.92983824e-02 6.88628972e-01 -3.46743971e-01 8.81509662e-01 2.14921385e-02 -3.08687717e-01 1.10088162e-01 -4.03808713e-01 6.91083789e-01 4.24873792e-02 -4.83633608e-01 6.38219595e-01 -1.54058114e-01 -8.17582488e-01 -1.08624017e+00 -7.53091395e-01 -7.95686364e-01 -8.09964895e-01 -5.63705981e-01 -1.28498897e-01 4.85149890e-01 6.12960339e-01 5.23019314e-01 -1.42659470e-01 9.29498672e-01 -1.44178927e+00 -2.18886063e-01 -4.24465537e-01 -4.11318511e-01 5.16710043e-01 5.80922544e-01 -7.81102836e-01 -3.25643450e-01 3.53198349e-01]
[9.349893569946289, -2.336008071899414]