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
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
60e49953-f2a6-4032-9e3e-79daca5dd76e
|
problem-decomposition-and-multi-shot-asp
|
2205.07537
| null |
https://arxiv.org/abs/2205.07537v2
|
https://arxiv.org/pdf/2205.07537v2.pdf
|
Problem Decomposition and Multi-shot ASP Solving for Job-shop Scheduling
|
The Job-shop Scheduling Problem (JSP) is a well-known and challenging combinatorial optimization problem in which tasks sharing a machine are to be arranged in a sequence such that encompassing jobs can be completed as early as possible. In this paper, we propose problem decomposition into time windows whose operations can be successively scheduled and optimized by means of multi-shot Answer Set Programming (ASP) solving. Decomposition aims to split highly complex scheduling tasks into better manageable sub-problems with a balanced number of operations so that good quality or even optimal partial solutions can be reliably found in a small fraction of runtime. Problem decomposition must respect the precedence of operations within their jobs and partial schedules optimized by time windows should yield better global solutions than obtainable in similar runtime on the entire instance. We devise and investigate a variety of decomposition strategies in terms of the number and size of time windows as well as heuristics for choosing their operations. Moreover, we incorporate time window overlapping and compression techniques into the iterative scheduling process to counteract window-wise optimization limitations restricted to partial schedules. Our experiments on JSP benchmark sets of several sizes show that successive optimization by multi-shot ASP solving leads to substantially better schedules within the runtime limit than global optimization on the full problem, where the gap increases with the number of operations to schedule. While the obtained solution quality still remains behind a state-of-the-art Constraint Programming system, our multi-shot solving approach comes closer the larger the instance size, demonstrating good scalability by problem decomposition.
|
['Konstantin Schekotihin', 'Martin Gebser', 'Mohammed M. S. El-Kholany']
|
2022-05-16
| null | null | null | null |
['problem-decomposition']
|
['miscellaneous']
|
[ 3.32341641e-01 4.26838040e-01 -2.29031488e-01 -2.27246240e-01
-3.78469795e-01 -5.90505958e-01 -6.74749389e-02 4.60654050e-01
-2.84816176e-01 8.66278112e-01 -3.49645436e-01 -3.04012656e-01
-8.96295190e-01 -5.99885941e-01 -5.22126079e-01 -6.14375472e-01
-5.32999456e-01 1.24029112e+00 3.66722554e-01 -2.54769295e-01
1.09995194e-01 5.59869468e-01 -1.53768599e+00 1.31908715e-01
8.07205379e-01 9.43699539e-01 6.57766402e-01 5.16040206e-01
-2.34658659e-01 2.22084984e-01 -6.84510291e-01 2.22686410e-01
7.97822952e-01 -1.94808424e-01 -1.10484946e+00 7.52330422e-01
-4.96798277e-01 1.18436463e-01 2.08399341e-01 6.69562459e-01
-1.02016270e-01 4.77372468e-01 -8.87897622e-04 -1.74829197e+00
5.10272607e-02 5.89884758e-01 -7.79986978e-01 1.82697594e-01
4.01302427e-01 -2.60024029e-03 9.24791753e-01 -1.42833665e-01
8.94852400e-01 8.29579949e-01 9.36717391e-02 2.35659420e-01
-1.50117314e+00 -1.33685768e-01 5.03578842e-01 2.13645712e-01
-1.32491457e+00 -1.14953727e-01 3.19399446e-01 -1.19460765e-02
1.47189105e+00 8.41097534e-01 4.52444255e-01 7.56109133e-02
2.71714866e-01 3.77468437e-01 8.36486816e-01 -5.96246302e-01
4.95299488e-01 4.60802279e-02 9.43772644e-02 3.21063727e-01
2.13197008e-01 -1.96732268e-01 -3.10478449e-01 -1.68281838e-01
3.50023091e-01 1.77015141e-01 -3.13838609e-02 -4.46031690e-01
-1.41215873e+00 7.00997531e-01 -1.15628995e-01 4.12020296e-01
-5.11192501e-01 -9.42135230e-02 5.46470821e-01 6.03502929e-01
4.00035232e-01 8.39612424e-01 -9.11747992e-01 3.98533270e-02
-1.16580057e+00 5.93341947e-01 1.20732081e+00 1.32167602e+00
5.81739545e-01 -4.22736466e-01 -4.30820882e-01 6.28211081e-01
-2.99485356e-01 1.64003998e-01 -3.17022979e-01 -1.24818528e+00
8.07302237e-01 4.18001503e-01 5.44500411e-01 -7.75719762e-01
-5.86465418e-01 -2.39725560e-01 -4.16027635e-01 1.31689966e-01
6.56895101e-01 -2.83420142e-02 -8.48224580e-01 1.41328657e+00
5.38300216e-01 -2.99982071e-01 -3.66210252e-01 9.87805307e-01
-2.97765553e-01 1.02975667e+00 -2.09726244e-01 -1.04750073e+00
1.75260472e+00 -1.24521422e+00 -7.41054714e-01 -3.58154207e-01
5.52729547e-01 -8.55991185e-01 3.33585590e-01 5.78304768e-01
-1.41916096e+00 -2.86848396e-01 -8.70661259e-01 2.96492875e-01
-2.42846861e-01 -4.62698132e-01 7.17128098e-01 4.61681396e-01
-7.96800077e-01 7.33857870e-01 -9.52414513e-01 -1.96943745e-01
-5.61019629e-02 7.27236032e-01 -2.93212533e-01 -3.52401555e-01
-7.42274284e-01 1.00206876e+00 5.00743866e-01 1.94407895e-01
-5.50838113e-01 -7.61713862e-01 -7.26434171e-01 4.33816463e-01
1.44545281e+00 -4.57879782e-01 1.31457806e+00 -4.94541287e-01
-1.06094038e+00 5.47324836e-01 -2.89379209e-01 -4.16217566e-01
2.42343172e-01 4.07546043e-01 -1.97127894e-01 7.91142508e-02
6.95509389e-02 1.89929992e-01 5.98791778e-01 -1.10202324e+00
-9.82783914e-01 -3.50083321e-01 4.96810496e-01 3.42476755e-01
-4.84134108e-02 3.49319160e-01 -6.38276279e-01 -4.25053574e-02
2.72307843e-01 -1.14657187e+00 -9.50263739e-01 -4.55571353e-01
-3.50567281e-01 -2.54026711e-01 1.89789906e-01 -4.76438314e-01
1.50683069e+00 -1.88028359e+00 7.36376405e-01 4.05613601e-01
2.54438221e-01 -2.26305053e-01 -1.39227197e-01 8.76834214e-01
-6.10559732e-02 -3.93865220e-02 -1.54509455e-01 -3.44300985e-01
4.03207779e-01 7.92010248e-01 -7.72750452e-02 3.51224393e-01
-1.22938596e-01 4.26807821e-01 -7.98889577e-01 -3.95702928e-01
-4.67459392e-03 -4.62159425e-01 -5.04595697e-01 1.95403934e-01
-6.18552446e-01 2.51405668e-02 -3.87140960e-01 7.17051506e-01
7.72313356e-01 -6.84013963e-02 5.77192605e-01 2.69936800e-01
-3.86361748e-01 4.29996513e-02 -1.44767666e+00 1.61009252e+00
-5.56897461e-01 9.40204784e-02 7.41255581e-01 -1.26824558e+00
6.49936199e-01 1.65017277e-01 9.26049232e-01 -6.78465068e-01
-1.29712567e-01 8.10855478e-02 2.83561852e-02 -4.36531007e-01
6.01213872e-01 -8.50600153e-02 -3.31619352e-01 5.67951977e-01
-4.14422959e-01 -6.86545223e-02 1.20761001e+00 3.92713910e-03
1.36297691e+00 -2.24029288e-01 2.09644943e-01 -4.35705215e-01
4.00293678e-01 5.70375443e-01 9.00500178e-01 5.97634971e-01
-1.67582207e-03 1.97978348e-01 9.56831396e-01 -6.71690047e-01
-1.09179401e+00 -5.22074640e-01 -8.72286111e-02 1.37936318e+00
2.97036290e-01 -4.57558900e-01 -5.99169612e-01 -4.34004217e-01
1.12620294e-01 6.26087844e-01 -4.77775812e-01 3.79631579e-01
-5.17717898e-01 -8.08644712e-01 -3.89027029e-01 2.49990985e-01
-2.76962042e-01 -9.56886351e-01 -9.93459105e-01 7.14771509e-01
7.16784671e-02 -1.27287960e+00 -6.12930000e-01 7.40084589e-01
-6.98109925e-01 -1.09095454e+00 -4.90309954e-01 -6.89381182e-01
1.22348857e+00 4.59928542e-01 9.72725391e-01 1.76676750e-01
-5.74724376e-01 -2.31447797e-02 -4.52000141e-01 -2.75752664e-01
-7.64249936e-02 1.68307349e-01 -6.56483248e-02 -4.00636733e-01
-4.47598994e-02 -4.10561681e-01 -1.43282458e-01 5.23102701e-01
-9.43032801e-01 1.45429567e-01 4.36641276e-01 7.85986602e-01
8.55163991e-01 8.90616298e-01 2.04544738e-01 -9.42276716e-01
5.66844344e-01 -3.70232671e-01 -9.70787585e-01 6.49974108e-01
-8.09740603e-01 2.18249410e-01 8.14600229e-01 -2.32505292e-01
-9.49936211e-01 1.38628528e-01 5.20167768e-01 -3.90121043e-01
-2.39770394e-02 6.86679184e-01 1.96035609e-01 1.40661895e-01
1.69322100e-02 9.64726845e-04 -8.09147954e-02 -1.97026044e-01
5.93752339e-02 -4.49440349e-03 1.86529756e-01 -8.86557877e-01
6.88272417e-01 1.78973362e-01 3.98705184e-01 -3.11236411e-01
-5.79800010e-01 -6.55323505e-01 -4.60397542e-01 -2.07554534e-01
4.64394271e-01 -7.43811876e-02 -9.97172952e-01 -2.94459641e-01
-1.06940401e+00 -3.61980051e-01 -3.28791648e-01 5.38498862e-03
-6.84059978e-01 2.93967396e-01 -4.81630743e-01 -1.10417628e+00
9.74740386e-02 -1.16331792e+00 8.20512533e-01 3.63343582e-02
-3.53022248e-01 -5.34733951e-01 2.80537028e-02 4.74508822e-01
2.07091585e-01 4.33842391e-01 7.64846683e-01 -6.19326770e-01
-7.74534285e-01 -2.65608847e-01 8.00264347e-03 -2.65999317e-01
-6.23217151e-02 -8.82229209e-02 -2.04176996e-02 -4.78268862e-01
5.22817783e-02 1.73001096e-01 3.41347903e-01 4.68390256e-01
1.08566201e+00 -2.73670703e-01 -5.63024461e-01 2.97180623e-01
1.57222736e+00 5.03118992e-01 2.65879005e-01 5.12798488e-01
1.48009449e-01 1.03809524e+00 1.17837286e+00 8.78645658e-01
1.40946105e-01 8.44484150e-01 5.17916858e-01 5.33615649e-02
5.86031854e-01 5.50273180e-01 -6.25106841e-02 6.34954870e-01
-3.84714097e-01 -3.83916855e-01 -1.00519347e+00 6.76708639e-01
-2.13119960e+00 -8.69006455e-01 -1.96094111e-01 2.44899726e+00
6.20162785e-01 5.36690235e-01 2.80868739e-01 4.83663082e-01
5.79338431e-01 3.69016193e-02 -2.14573458e-01 -1.11104918e+00
5.28331757e-01 2.05366790e-01 9.84253764e-01 5.34822822e-01
-7.91837454e-01 4.37398195e-01 5.59876060e+00 6.35290265e-01
-5.24210215e-01 7.47893006e-02 4.52611864e-01 -5.91040194e-01
-9.49590281e-03 2.65817970e-01 -6.93675041e-01 4.22902733e-01
1.05071712e+00 -5.38338959e-01 1.24942422e+00 8.35955262e-01
3.31328243e-01 -4.99210984e-01 -1.30219316e+00 4.43467706e-01
-3.63791347e-01 -1.20889652e+00 -8.80593419e-01 3.01753312e-01
9.39974368e-01 -4.21184480e-01 -3.90100151e-01 1.42989337e-01
1.26570851e-01 -8.06874990e-01 7.68570125e-01 1.88527837e-01
2.36707479e-01 -1.08799124e+00 8.03051829e-01 6.22545421e-01
-1.43829381e+00 -4.24355358e-01 -4.70239103e-01 -4.47085261e-01
6.26694500e-01 7.24935532e-01 -7.74706483e-01 1.07086658e+00
6.10936165e-01 -1.08686939e-01 1.57930583e-01 1.14597201e+00
2.61998415e-01 4.42678249e-03 -5.94869256e-01 -3.29979882e-02
1.81274787e-01 -2.77061313e-01 4.08711821e-01 8.58502507e-01
5.69354929e-02 4.44772035e-01 7.94220805e-01 7.45428145e-01
5.09970546e-01 -1.52009681e-01 -2.28490263e-01 -1.93073243e-01
5.41605651e-01 1.36823177e+00 -1.47004437e+00 -1.58548564e-01
-3.07778120e-01 6.17298961e-01 2.60624796e-01 3.32457483e-01
-9.97565269e-01 -5.57539999e-01 6.25125110e-01 5.01674786e-02
4.43071872e-01 -6.02014482e-01 -5.18334270e-01 -5.15073955e-01
4.51378167e-01 -4.62623656e-01 4.49362785e-01 -2.94614345e-01
-8.15255344e-01 6.23592377e-01 3.70048076e-01 -9.04081821e-01
-1.80108741e-01 -4.87662911e-01 -4.70195532e-01 7.82779634e-01
-1.44587195e+00 -4.35224771e-01 1.16749629e-01 2.12167248e-01
9.16764617e-01 4.55740005e-01 6.45772278e-01 9.70821381e-02
-8.16722691e-01 5.14650606e-02 -1.96602172e-03 -7.82609046e-01
2.72612870e-01 -1.19529986e+00 -3.94829400e-02 7.41788149e-01
-4.41198885e-01 4.57861096e-01 1.11652291e+00 -5.27624369e-01
-1.81287086e+00 -4.85788196e-01 1.20051479e+00 -7.97970816e-02
6.51184559e-01 -3.62558216e-01 -7.21423805e-01 4.84269291e-01
2.80867010e-01 -9.88006964e-02 3.06807220e-01 5.21138370e-01
2.78866738e-01 -4.28398997e-01 -9.75066662e-01 2.60996193e-01
8.85952771e-01 1.55837402e-01 -4.96479660e-01 9.07823682e-01
7.73025274e-01 -6.75958931e-01 -9.41134036e-01 2.23256245e-01
9.74142179e-02 -6.24364793e-01 6.94468796e-01 -6.22131348e-01
4.13246870e-01 -3.45117718e-01 1.68168142e-01 -1.16580296e+00
-5.33498049e-01 -8.95067215e-01 -2.82393629e-03 9.62482810e-01
5.85806608e-01 -4.46964800e-01 6.87290013e-01 1.36279190e+00
-4.32675749e-01 -1.19539213e+00 -1.03376865e+00 -1.22615004e+00
-7.19815612e-01 -3.64610940e-01 7.64654875e-01 8.35911632e-01
5.64636052e-01 -1.34134933e-01 -2.81973928e-01 3.14414710e-01
5.55456579e-01 6.88516617e-01 3.49676818e-01 -1.10935390e+00
-5.72219551e-01 -2.78600037e-01 2.95492291e-01 -5.40461063e-01
5.46412058e-02 -7.07029879e-01 3.37808162e-01 -1.70400536e+00
1.55119047e-01 -7.26644576e-01 -2.72588909e-01 8.00378144e-01
1.79747894e-01 -4.94576454e-01 5.74264288e-01 -1.49525581e-02
-1.11107755e+00 -7.59225115e-02 1.43126106e+00 -5.54756895e-02
-4.21821147e-01 1.05690733e-01 -3.07498366e-01 -9.54798795e-03
5.44262707e-01 -6.71745360e-01 -3.65047932e-01 -4.32269573e-01
4.69811350e-01 1.01495767e+00 -4.23425674e-01 -5.12446702e-01
5.79848409e-01 -8.37930679e-01 -2.73974687e-01 -3.97278726e-01
1.72370762e-01 -1.30398798e+00 7.18725324e-01 5.13358235e-01
-3.66336435e-01 3.65888268e-01 2.47982726e-01 5.07916093e-01
3.10363304e-02 -5.86598575e-01 2.71646708e-01 -2.12264329e-01
-5.27171791e-01 2.84983337e-01 -3.62497687e-01 -4.10058200e-01
1.64466333e+00 -2.68488824e-01 -7.89831653e-02 1.68822914e-01
-1.06856668e+00 1.05503488e+00 2.64500737e-01 3.31674129e-01
9.04554650e-02 -8.33830953e-01 -5.02928972e-01 -6.61316067e-02
-9.83349904e-02 2.02105850e-01 4.46892649e-01 1.20810854e+00
-5.81200480e-01 8.19997966e-01 -3.38327706e-01 -3.22974592e-01
-1.30018365e+00 1.17965531e+00 -3.66615593e-01 -1.01413751e+00
-4.07336116e-01 7.40365028e-01 -1.71191543e-01 -4.62795049e-02
2.06489339e-01 -4.36699778e-01 2.89844632e-01 1.04279377e-01
4.81420338e-01 7.26635754e-01 3.27510446e-01 1.59192592e-01
-5.86836040e-01 9.99396220e-02 -2.69548148e-01 1.50415137e-01
1.57269657e+00 -1.60428435e-01 -6.24274075e-01 -1.44110173e-01
4.77351815e-01 -1.71714604e-01 -1.10519791e+00 -5.71586285e-03
3.43471557e-01 -6.40628934e-01 -7.52373636e-02 -7.09403217e-01
-1.16310954e+00 2.75515635e-02 -2.46122122e-01 1.06384313e+00
1.41955650e+00 7.90171176e-02 7.81983912e-01 2.53552586e-01
7.57525742e-01 -1.45501733e+00 -5.68683386e-01 4.45593029e-01
7.15442777e-01 -7.21949100e-01 4.11225587e-01 -9.81561899e-01
-5.59883773e-01 1.17159426e+00 4.91375387e-01 -5.81657663e-02
1.62818655e-01 5.93317926e-01 -6.57010198e-01 -1.81817561e-01
-1.26228356e+00 -2.73032546e-01 -9.48830247e-02 2.46453285e-01
-6.66308478e-02 3.39279324e-01 -9.68256891e-01 7.19567716e-01
6.48617968e-02 -1.47209048e-01 4.52241659e-01 1.36121190e+00
-5.61218321e-01 -1.45002592e+00 -6.96527839e-01 4.56841886e-01
-2.51869112e-01 3.30529273e-01 1.11701638e-01 7.49545395e-01
1.63704216e-01 1.06130147e+00 5.46746776e-02 6.78639859e-03
4.98109013e-01 5.63977361e-02 6.66802645e-01 -9.37261939e-01
-6.52347565e-01 2.04378039e-01 5.46431661e-01 -8.26379597e-01
-2.42333904e-01 -6.15557671e-01 -1.41355300e+00 -3.24622065e-01
-4.33296263e-01 4.37156260e-01 7.13305354e-01 1.24613428e+00
2.16463864e-01 9.68698978e-01 8.45193207e-01 -1.14281261e+00
-6.91492260e-01 -3.59417051e-01 -8.29956055e-01 -5.24800038e-03
-1.28270900e-02 -6.20040417e-01 -9.99966189e-02 -3.85002941e-02]
|
[5.103593349456787, 2.641881227493286]
|
58738236-6bee-4b9e-9f28-3023693590e1
|
an-investigation-of-machine-translation
| null | null |
https://aclanthology.org/W15-3057
|
https://aclanthology.org/W15-3057.pdf
|
An Investigation of Machine Translation Evaluation Metrics in Cross-lingual Question Answering
| null |
['Kyoshiro Sugiyama', 'Masahiro Mizukami', 'Koichiro Yoshino', 'Tomoki Toda', 'Satoshi Nakamura', 'Sakriani Sakti', 'Graham Neubig']
|
2015-09-01
| null | null | null |
ws-2015-9
|
['cross-lingual-question-answering']
|
['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]
|
[-7.356884956359863, 3.7960174083709717]
|
b316dc36-441f-40b9-ab01-03ff703fe660
|
circlenet-reciprocating-feature-adaptation
|
2212.05691
| null |
https://arxiv.org/abs/2212.05691v1
|
https://arxiv.org/pdf/2212.05691v1.pdf
|
CircleNet: Reciprocating Feature Adaptation for Robust Pedestrian Detection
|
Pedestrian detection in the wild remains a challenging problem especially when the scene contains significant occlusion and/or low resolution of the pedestrians to be detected. Existing methods are unable to adapt to these difficult cases while maintaining acceptable performance. In this paper we propose a novel feature learning model, referred to as CircleNet, to achieve feature adaptation by mimicking the process humans looking at low resolution and occluded objects: focusing on it again, at a finer scale, if the object can not be identified clearly for the first time. CircleNet is implemented as a set of feature pyramids and uses weight sharing path augmentation for better feature fusion. It targets at reciprocating feature adaptation and iterative object detection using multiple top-down and bottom-up pathways. To take full advantage of the feature adaptation capability in CircleNet, we design an instance decomposition training strategy to focus on detecting pedestrian instances of various resolutions and different occlusion levels in each cycle. Specifically, CircleNet implements feature ensemble with the idea of hard negative boosting in an end-to-end manner. Experiments on two pedestrian detection datasets, Caltech and CityPersons, show that CircleNet improves the performance of occluded and low-resolution pedestrians with significant margins while maintaining good performance on normal instances.
|
['Qixiang Ye', 'Baochang Zhang', 'Huijuan Xu', 'Zhenjun Han', 'Tianliang Zhang']
|
2022-12-12
| null | null | null | null |
['pedestrian-detection']
|
['computer-vision']
|
[ 4.21061069e-02 -3.02602857e-01 2.36618802e-01 -2.50091881e-01
-4.04777169e-01 -1.99002072e-01 5.01347721e-01 9.83268321e-02
-5.81980944e-01 6.17262781e-01 1.02036230e-01 3.99436196e-03
2.12844878e-01 -8.28174710e-01 -5.80261588e-01 -6.32967293e-01
-1.65059909e-01 7.04258606e-02 8.10329854e-01 -2.22286031e-01
-1.52626038e-01 5.19475639e-01 -1.74853861e+00 5.51395953e-01
7.39851654e-01 7.25905478e-01 3.09026778e-01 7.45032191e-01
2.27959260e-01 4.61712360e-01 -4.62652564e-01 -4.74391580e-01
5.07928729e-01 5.39500974e-02 -2.05364779e-01 4.73818928e-01
7.35828876e-01 -2.94118971e-01 -1.94464490e-01 8.80916238e-01
6.65020466e-01 -1.75262094e-02 3.71359646e-01 -1.17323160e+00
-3.80562574e-01 1.42450824e-01 -1.13640785e+00 6.61719501e-01
3.76052320e-01 5.95887184e-01 6.50217175e-01 -1.25605595e+00
1.74985602e-01 1.58960795e+00 8.11419785e-01 4.19684529e-01
-1.36592722e+00 -5.42075455e-01 5.48033059e-01 5.93798935e-01
-1.32399368e+00 -3.17439228e-01 6.45119965e-01 -3.73225987e-01
9.10462439e-01 2.23495513e-01 8.01150084e-01 1.07946706e+00
6.69998676e-02 9.65160370e-01 1.04795480e+00 -3.53732139e-01
2.82048061e-02 2.43806541e-01 2.95034856e-01 5.51177382e-01
6.06480718e-01 4.33607936e-01 -4.30536538e-01 -6.18323907e-02
5.21689177e-01 1.70016170e-01 -1.12068608e-01 -5.14160693e-01
-9.46494401e-01 6.77278221e-01 9.63758647e-01 1.84440240e-01
-6.53909326e-01 -2.24881843e-01 3.79026800e-01 9.06739105e-03
3.17235172e-01 -5.16074486e-02 -4.52860028e-01 3.64422500e-01
-6.73458576e-01 3.03030968e-01 3.13829929e-01 6.39182746e-01
6.63026273e-01 -6.18223324e-02 -5.34417272e-01 6.02298141e-01
3.59391063e-01 4.01856571e-01 3.28800082e-01 -4.32453334e-01
4.47471231e-01 8.68480384e-01 2.03748912e-01 -9.17222440e-01
-6.00444853e-01 -1.05007350e+00 -8.36428881e-01 6.80201411e-01
5.82436204e-01 -1.67598173e-01 -1.07431936e+00 1.57369745e+00
6.84610069e-01 1.64519876e-01 -1.03213400e-01 1.14922154e+00
7.75163114e-01 4.92554545e-01 5.92890322e-01 1.26332000e-01
1.88293660e+00 -1.07624984e+00 -1.21886961e-01 -5.26807368e-01
2.57461369e-01 -7.51129150e-01 7.64033079e-01 2.75809854e-01
-8.69924843e-01 -1.13796544e+00 -1.04498386e+00 1.87831163e-01
-4.71070558e-01 4.55875278e-01 5.31621337e-01 9.69189703e-01
-8.89148414e-01 2.51334459e-01 -7.42700934e-01 -6.53122962e-01
6.16675675e-01 1.80310220e-01 -4.20770198e-01 -3.25557351e-01
-8.39243293e-01 9.10974920e-01 4.28124249e-01 2.69348711e-01
-6.18296146e-01 -7.45044589e-01 -8.98844361e-01 1.69343844e-01
3.45298976e-01 -1.10989487e+00 6.80422127e-01 -8.21549594e-01
-9.32993948e-01 5.70088983e-01 -1.25368044e-01 -7.14000583e-01
8.73035014e-01 -4.07124221e-01 -3.57299238e-01 -8.12027976e-02
3.20502341e-01 8.62438023e-01 1.09079993e+00 -1.21713376e+00
-1.23017299e+00 -6.22492552e-01 3.43348645e-02 1.39233291e-01
-2.37876922e-01 1.42670184e-01 -2.84896970e-01 -4.48404729e-01
1.11311547e-01 -5.93907893e-01 -4.69427317e-01 7.50427693e-02
-3.11835617e-01 -2.58962482e-01 1.06932116e+00 -5.38931251e-01
7.84444809e-01 -1.99211204e+00 -1.62405238e-01 1.21443242e-01
2.94267446e-01 7.07111835e-01 -1.43327013e-01 1.19537152e-02
-1.15814902e-01 -3.62565517e-01 4.37554121e-02 -3.61025482e-01
-2.31601328e-01 -4.67917547e-02 1.51107043e-01 4.76050407e-01
7.28540957e-01 7.30173469e-01 -8.42739344e-01 -5.06577730e-01
5.49366176e-01 7.90756702e-01 -5.47857225e-01 1.06883094e-01
2.49704510e-01 4.09459174e-01 -5.10671258e-01 7.79459119e-01
9.90997195e-01 -1.06644958e-01 -2.71420151e-01 -4.21183228e-01
-4.50799614e-01 -4.47912246e-01 -1.62562120e+00 1.07054269e+00
-2.18735293e-01 5.22252142e-01 1.28541142e-01 -7.69546747e-01
7.19697177e-01 -1.80014130e-02 9.17016640e-02 -7.51041234e-01
2.24214736e-02 -9.80869234e-02 2.23261818e-01 -3.91434252e-01
4.70892221e-01 2.14828148e-01 2.15557232e-01 -2.16114238e-01
-7.89229795e-02 5.90122759e-01 6.05387866e-01 1.23332605e-01
9.54951644e-01 5.33997789e-02 3.82101387e-01 1.76333766e-02
8.77048135e-01 -7.09429458e-02 5.55417418e-01 1.00081968e+00
-5.36684334e-01 4.57976729e-01 -2.25526523e-02 -6.89649403e-01
-1.01644778e+00 -1.09198463e+00 -1.99886709e-01 1.56542873e+00
5.85749485e-02 -2.91087776e-01 -6.00946665e-01 -5.44248521e-01
1.88292608e-01 3.80137235e-01 -7.83183753e-01 -1.33965373e-01
-7.35326111e-01 -1.06751180e+00 1.29446492e-01 6.55288398e-01
9.76900220e-01 -9.82982159e-01 -1.09230602e+00 3.36366951e-01
-2.11820100e-03 -1.18498540e+00 -2.69587845e-01 1.43400952e-01
-3.43843281e-01 -1.13482869e+00 -7.48668849e-01 -6.77409410e-01
5.41800022e-01 7.27129996e-01 9.61789131e-01 -5.27616404e-02
-7.36219108e-01 3.64621073e-01 -3.61491203e-01 -4.46217924e-01
-7.91863799e-02 -1.63607374e-01 2.79905181e-02 4.70307857e-01
2.88863510e-01 -4.09418076e-01 -9.59806025e-01 3.81270885e-01
-4.44472164e-01 2.82221604e-02 1.01068854e+00 9.16372180e-01
2.32469007e-01 1.03815429e-01 4.36793029e-01 -2.89704055e-01
2.32293099e-01 -2.73075312e-01 -4.57221150e-01 2.10935399e-01
-1.92043468e-01 -3.59050393e-01 5.31916380e-01 -4.68945473e-01
-1.24723291e+00 3.43799174e-01 -1.63314879e-01 -1.32585347e-01
-4.63168144e-01 -1.90250441e-01 -3.26420486e-01 -2.14100540e-01
9.74955201e-01 2.93759108e-01 -2.11197153e-01 -5.40486753e-01
4.42492276e-01 3.76999199e-01 5.73127449e-01 -3.88460904e-01
9.14957166e-01 6.10863209e-01 -6.62299842e-02 -8.86933625e-01
-6.93698525e-01 -7.54077971e-01 -8.38187039e-01 -4.64487910e-01
7.89405704e-01 -1.11108041e+00 -5.98855615e-01 5.34108460e-01
-9.56411839e-01 1.55077130e-01 -4.63913321e-01 3.49727750e-01
-6.80554956e-02 4.57182914e-01 -2.99733758e-01 -9.51727927e-01
-3.96771252e-01 -9.97185469e-01 1.05562174e+00 6.64889634e-01
2.49968484e-01 -3.87067735e-01 -4.23526525e-01 1.52822986e-01
6.71215773e-01 3.29336494e-01 4.00153458e-01 -5.05234063e-01
-6.69451356e-01 -4.13953900e-01 -7.29822993e-01 2.28477627e-01
-1.51267787e-02 -1.46934196e-01 -1.05507207e+00 -5.79144001e-01
-3.99574578e-01 -4.18290533e-02 1.19923651e+00 4.79325265e-01
6.36830807e-01 6.85562864e-02 -5.90268672e-01 3.09374839e-01
1.23670352e+00 -2.09150866e-01 4.20577496e-01 4.87045437e-01
5.27974486e-01 4.82078373e-01 5.93403935e-01 5.96248448e-01
3.89217436e-01 8.50386202e-01 5.14257193e-01 -4.59308147e-01
-6.11750782e-01 -2.74954233e-02 2.51743913e-01 -2.77262300e-01
-8.63462538e-02 2.12851495e-01 -6.54450595e-01 5.06105959e-01
-1.89063120e+00 -1.15135658e+00 -2.23419309e-01 2.07188797e+00
3.24946344e-01 4.55682844e-01 6.66022122e-01 2.67170370e-01
9.79606271e-01 -1.26124933e-01 -4.93733197e-01 7.84738883e-02
-2.78245568e-01 -1.18681416e-01 3.20100009e-01 9.84615535e-02
-1.46770012e+00 8.34036291e-01 5.10596418e+00 6.57130599e-01
-9.26922739e-01 1.50213912e-01 6.97530448e-01 6.31825160e-03
6.38157427e-01 -1.88104793e-01 -1.14989471e+00 4.15581584e-01
4.26172495e-01 1.48442417e-01 -1.16910785e-01 9.59634006e-01
2.11455345e-01 -1.92810029e-01 -8.87081921e-01 8.76694381e-01
-1.41991705e-01 -8.64787340e-01 -9.86598879e-02 -1.76812395e-01
4.27138835e-01 -6.05869703e-02 8.37360546e-02 5.90910554e-01
2.41857708e-01 -6.91389680e-01 7.72503793e-01 2.53721565e-01
1.82386577e-01 -6.43307805e-01 7.65731812e-01 4.59276319e-01
-1.66076934e+00 -6.17745042e-01 -4.46703464e-01 -1.75744027e-01
2.48801664e-01 3.83764625e-01 -8.86101961e-01 4.69775021e-01
9.87036586e-01 4.20716316e-01 -1.09573340e+00 1.72371781e+00
-2.90562306e-02 3.05167019e-01 -4.41111565e-01 9.66255739e-02
2.62220323e-01 1.96997508e-01 7.21890450e-01 1.68689966e+00
1.18647695e-01 3.65106426e-02 6.83965504e-01 6.60261869e-01
4.01552707e-01 -2.41962150e-02 -4.34187174e-01 9.37687039e-01
2.52938032e-01 1.53509247e+00 -6.14000559e-01 -2.86211938e-01
-5.77333987e-01 9.26592648e-01 3.67534250e-01 2.77518779e-01
-8.35474908e-01 -8.90882388e-02 5.50710559e-01 3.86049539e-01
8.25937808e-01 -1.53790504e-01 -2.59176195e-02 -8.66441607e-01
2.08333656e-01 -5.58921337e-01 5.58872581e-01 -4.43275839e-01
-1.40554500e+00 5.87630212e-01 -3.41587141e-02 -1.12754929e+00
4.94152196e-02 -7.12875068e-01 -7.04412818e-01 8.73819411e-01
-1.69010854e+00 -1.66757977e+00 -6.31259918e-01 7.93827951e-01
7.32212067e-01 -9.95706618e-02 3.72448444e-01 4.06464130e-01
-7.29842365e-01 6.54696643e-01 -3.13615322e-01 1.08318351e-01
4.35499847e-01 -1.26854777e+00 3.82066041e-01 1.25263667e+00
-2.15726763e-01 2.74603903e-01 8.67821455e-01 -5.29498160e-01
-1.11653495e+00 -1.48338509e+00 5.52444637e-01 -3.31499040e-01
3.46845955e-01 -3.03496063e-01 -8.72301280e-01 3.06641251e-01
1.04745980e-02 3.86249125e-01 3.87575537e-01 1.25354886e-01
-4.24536705e-01 -2.41303965e-01 -1.41954410e+00 5.45478761e-01
1.05992937e+00 9.06438380e-02 -4.55232948e-01 3.57026339e-01
3.29993546e-01 -2.08627492e-01 -3.45590413e-01 5.54908156e-01
5.10813892e-01 -1.20353484e+00 1.46615541e+00 -6.60886288e-01
-1.53636470e-01 -6.01380110e-01 -1.63154870e-01 -1.12551689e+00
-9.45985377e-01 -1.31145641e-01 -1.69642523e-01 1.24257731e+00
1.51891857e-01 -5.47518790e-01 7.44230747e-01 4.29313272e-01
-2.37052403e-02 -5.09738684e-01 -1.18014443e+00 -7.87792981e-01
-3.18437934e-01 -2.07162291e-01 3.44054550e-01 3.83910567e-01
-4.74803299e-01 4.85821813e-01 -3.27833921e-01 5.41536629e-01
1.10823655e+00 -3.01673282e-02 9.83715355e-01 -1.29112005e+00
-1.97148964e-01 -5.20235181e-01 -9.02067065e-01 -8.53546739e-01
-4.09661591e-01 -4.52677071e-01 -4.18475680e-02 -1.41472602e+00
5.17452359e-01 -2.21819714e-01 -2.76499808e-01 3.15929085e-01
-6.39121175e-01 5.11638463e-01 4.91869420e-01 -9.99197960e-02
-8.49556088e-01 4.55084622e-01 1.02928996e+00 -2.91457146e-01
-2.31766269e-01 2.71033674e-01 -7.36954689e-01 8.44035625e-01
5.63259006e-01 -1.15860723e-01 3.28413397e-02 -1.68514520e-01
-5.04262388e-01 -3.45650196e-01 1.02628887e+00 -1.60627913e+00
4.08495456e-01 3.11285764e-01 1.21326053e+00 -8.21129918e-01
4.63671923e-01 -8.54154050e-01 -7.44544491e-02 7.83385336e-01
8.07829499e-02 1.02412082e-01 3.37224603e-01 7.35128582e-01
1.12643592e-01 1.57088682e-01 1.16265571e+00 -1.69379339e-01
-1.16435587e+00 2.82994449e-01 -3.37221652e-01 -3.48315090e-01
1.22336793e+00 -4.89074916e-01 -3.26247960e-01 9.48492512e-02
-1.02264333e+00 3.82114172e-01 1.75347358e-01 5.25333881e-01
7.66308248e-01 -1.25336337e+00 -1.16627753e+00 4.19886529e-01
1.88516870e-01 -2.85830855e-01 4.21748161e-01 7.99221814e-01
8.81753191e-02 3.51913095e-01 -3.50007206e-01 -9.54577863e-01
-1.56878114e+00 6.73177719e-01 5.08641481e-01 -3.97567689e-01
-7.39213109e-01 8.23249638e-01 4.70704645e-01 -3.54823694e-02
2.79822707e-01 -5.10951094e-02 -5.26520371e-01 5.16935140e-02
9.55857933e-01 4.82475013e-01 5.96455522e-02 -8.60243082e-01
-3.60061646e-01 5.89909732e-01 -4.06816751e-01 3.91934663e-01
1.19998360e+00 -2.62209564e-01 3.26015204e-01 -1.89223990e-01
7.44196713e-01 -2.42429033e-01 -1.63972104e+00 -5.76546252e-01
-1.28835723e-01 -7.74358332e-01 -1.50454819e-01 -8.24261844e-01
-9.61014092e-01 7.05832064e-01 1.34832621e+00 1.74726233e-01
1.07487428e+00 -6.51077703e-02 5.33783853e-01 1.85575947e-01
2.94897646e-01 -8.02173018e-01 1.85935318e-01 3.31212401e-01
7.61704564e-01 -1.56303060e+00 -9.51298848e-02 -4.20159400e-01
-3.92629772e-01 1.01579201e+00 9.23047900e-01 -9.14358199e-02
5.05195200e-01 4.26144637e-02 -2.51513839e-01 -3.62414196e-02
-5.26173532e-01 -8.51900160e-01 3.13624978e-01 1.00658464e+00
9.79929939e-02 1.67157650e-01 -5.23789972e-03 4.55837101e-01
1.52926609e-01 -1.87951118e-01 6.09810390e-02 7.54597545e-01
-9.15149033e-01 -7.79221594e-01 -9.01615322e-01 3.99872750e-01
-2.76156694e-01 5.43110678e-03 -1.27908275e-01 8.33689868e-01
6.34375095e-01 1.10626960e+00 4.41310108e-02 -2.50524610e-01
7.80432224e-01 -1.20474443e-01 4.86320347e-01 -3.39061022e-01
-6.93567157e-01 2.54957993e-02 -4.77022268e-02 -5.00464737e-01
-3.36205661e-01 -8.85402024e-01 -7.79891789e-01 -9.25423801e-02
-2.59201884e-01 -1.35457665e-01 4.17400867e-01 7.46126354e-01
3.64855796e-01 7.04284668e-01 5.90203941e-01 -1.35540092e+00
-4.90136892e-01 -1.05858982e+00 -2.81544000e-01 5.67616463e-01
4.02214438e-01 -8.41908395e-01 1.20516323e-01 -6.80912808e-02]
|
[8.042346954345703, -0.6201605200767517]
|
7b8e6068-ad49-40a0-abac-6c8dafb5a4ed
|
discogen-learning-to-discover-gene-regulatory
|
2304.05823
| null |
https://arxiv.org/abs/2304.05823v1
|
https://arxiv.org/pdf/2304.05823v1.pdf
|
DiscoGen: Learning to Discover Gene Regulatory Networks
|
Accurately inferring Gene Regulatory Networks (GRNs) is a critical and challenging task in biology. GRNs model the activatory and inhibitory interactions between genes and are inherently causal in nature. To accurately identify GRNs, perturbational data is required. However, most GRN discovery methods only operate on observational data. Recent advances in neural network-based causal discovery methods have significantly improved causal discovery, including handling interventional data, improvements in performance and scalability. However, applying state-of-the-art (SOTA) causal discovery methods in biology poses challenges, such as noisy data and a large number of samples. Thus, adapting the causal discovery methods is necessary to handle these challenges. In this paper, we introduce DiscoGen, a neural network-based GRN discovery method that can denoise gene expression measurements and handle interventional data. We demonstrate that our model outperforms SOTA neural network-based causal discovery methods.
|
['Danilo Rezende', 'Mike Mozer', 'Anirudh Goyal', 'Matthew Botvinick', 'David Barrett', 'Theophane Weber', 'Jane Wang', 'Albin Cassirer', 'Jean-Baptiste Lespiau', 'Melanie Rey', 'Silvia Chiappa', 'Jorg Bornschein', 'Sara-Jane Dunn', 'Nan Rosemary Ke']
|
2023-04-12
| null | null | null | null |
['causal-discovery']
|
['knowledge-base']
|
[ 5.29577851e-01 -1.77739143e-01 -5.80274045e-01 -3.28932852e-01
-5.36758423e-01 -3.34834784e-01 4.42207962e-01 2.68661082e-01
5.90451621e-02 1.24284601e+00 4.92201954e-01 -5.68813026e-01
-7.59515285e-01 -9.96295512e-01 -1.08250642e+00 -1.02709460e+00
-5.44888854e-01 5.17102897e-01 -2.90933400e-01 5.98005354e-02
-2.24185176e-02 2.90377796e-01 -1.41157603e+00 2.25220874e-01
7.76137650e-01 4.36059594e-01 -3.97472948e-01 8.50642979e-01
-4.04717121e-03 3.79354417e-01 -3.46883416e-01 6.08358234e-02
-1.54150680e-01 -7.13452518e-01 -4.22161251e-01 -1.07665980e+00
3.00984234e-01 2.67982893e-02 -5.14713407e-01 8.99789870e-01
9.73800600e-01 2.89482679e-02 5.59308112e-01 -1.34936023e+00
-9.51319933e-01 1.14839542e+00 -5.53585589e-01 2.80790180e-01
1.28752306e-01 1.13563165e-01 1.05941045e+00 -4.17956412e-01
6.16262317e-01 1.81660903e+00 8.48060012e-01 7.57867217e-01
-1.61387694e+00 -8.76689970e-01 6.37014657e-02 3.86959724e-02
-1.36557877e+00 -4.49145913e-01 3.56980205e-01 -2.78264314e-01
7.75099516e-01 5.16147256e-01 6.54784679e-01 1.65066278e+00
3.28007549e-01 3.66721720e-01 8.38079810e-01 -2.90497452e-01
6.27120912e-01 -1.29062033e+00 -6.98922127e-02 5.37605345e-01
4.31406915e-01 5.77439845e-01 -7.81595051e-01 -7.31536210e-01
9.43108380e-01 7.75135309e-02 -1.38379693e-01 2.56771982e-01
-1.23866642e+00 7.56797016e-01 2.26746961e-01 1.17136888e-01
-4.84253436e-01 1.01416361e+00 5.14039934e-01 -1.57348476e-02
5.91252267e-01 5.78983724e-01 -7.73965240e-01 -1.87452614e-01
-7.86450148e-01 4.74682927e-01 7.12766826e-01 6.40122056e-01
3.51228602e-02 1.02055117e-01 -4.12690967e-01 9.07859981e-01
2.56189644e-01 3.56516927e-01 2.21685812e-01 -8.63163531e-01
-3.88467461e-01 5.46622872e-01 -2.42258608e-01 -9.25443530e-01
-6.58358276e-01 -2.77908206e-01 -1.39332592e+00 -3.53586823e-01
4.15366918e-01 -3.79881799e-01 -1.12854576e+00 2.08669519e+00
6.57363474e-01 9.55746174e-01 -3.00216168e-01 7.89364696e-01
9.76747096e-01 3.78712565e-01 5.36421478e-01 -3.58904511e-01
1.24408710e+00 -4.59209364e-03 -9.79161322e-01 1.57947183e-01
3.88219625e-01 -2.61151463e-01 6.18740916e-01 2.70109206e-01
-6.64466798e-01 3.70024368e-02 -6.00947142e-01 -5.82490154e-02
-4.83978361e-01 -2.83549577e-01 1.32849991e+00 3.94890964e-01
-7.86297381e-01 7.11448133e-01 -8.48274052e-01 -4.89293337e-01
6.77424073e-01 5.59884429e-01 -3.38682085e-01 -2.09632292e-01
-1.76418853e+00 3.11532080e-01 4.42471862e-01 3.86979818e-01
-1.50045872e+00 -1.36529505e+00 -4.56756711e-01 3.60219806e-01
3.91315073e-01 -1.06178129e+00 9.42701221e-01 -1.87415808e-01
-1.23806894e+00 3.24830025e-01 -1.00003183e-01 -1.89322501e-01
-1.36355478e-02 -4.82350215e-02 -4.22313184e-01 -4.08849955e-01
-8.28985944e-02 5.93987107e-01 1.81467414e-01 -8.44791353e-01
-4.11066264e-01 -4.92170662e-01 -3.25565249e-01 -4.22617763e-01
2.45644040e-02 -6.43244311e-02 -8.93619135e-02 -6.59949958e-01
3.99072878e-02 -6.59690797e-01 -4.76999491e-01 -1.50328323e-01
-7.42974460e-01 -4.20759648e-01 6.10179186e-01 -3.08256000e-01
1.12337625e+00 -1.76008117e+00 2.48133764e-01 1.88055709e-02
5.58356464e-01 7.48668388e-02 -3.91912103e-01 2.85790294e-01
-6.12882912e-01 6.68734193e-01 -5.01912124e-02 2.77203709e-01
-1.55859748e-02 4.46978629e-01 -1.99318469e-01 4.66690302e-01
3.26647729e-01 1.15564251e+00 -1.21890008e+00 -7.96100646e-02
-1.33957803e-01 5.97962797e-01 -5.81768334e-01 2.03520358e-01
-7.70278156e-01 6.75215483e-01 -4.90634143e-01 9.68090594e-01
3.56251031e-01 -8.92973691e-02 7.23420501e-01 -1.43830944e-02
-1.00906678e-01 1.55489102e-01 -7.90242195e-01 1.57455766e+00
-9.90665182e-02 2.55445868e-01 -9.48460922e-02 -1.44229007e+00
7.48132110e-01 4.86884117e-01 5.47968268e-01 -3.10673445e-01
2.82149047e-01 -4.82114404e-03 2.92938262e-01 -6.90353811e-01
-3.33429217e-01 -2.91313648e-01 -1.24380767e-01 1.24019779e-01
1.64069980e-02 3.83906960e-01 1.12596974e-01 -1.82459932e-02
1.78793418e+00 9.01515712e-04 4.75781292e-01 -2.94749916e-01
-2.83509225e-01 -1.00189494e-02 1.26881099e+00 1.07928109e+00
-9.17869955e-02 3.06623936e-01 1.21010005e+00 -4.98726606e-01
-7.76793242e-01 -9.30686831e-01 -2.01406598e-01 1.22911274e+00
-4.61653829e-01 -1.53617650e-01 -5.79132736e-01 -4.80644196e-01
3.51544321e-01 5.41889906e-01 -1.18084288e+00 -4.03853863e-01
-2.99770385e-01 -1.72044826e+00 1.36007655e+00 3.42109382e-01
-4.05408740e-02 -6.51215792e-01 2.67928958e-01 3.38976413e-01
-1.17895782e-01 -5.16955078e-01 -1.19967729e-01 5.99177778e-01
-7.65687704e-01 -1.51788700e+00 -2.95623224e-02 -6.75322115e-02
5.84973037e-01 -1.00986943e-01 1.04103363e+00 -6.11473806e-03
-7.07537711e-01 -3.32634568e-01 -4.86025475e-02 -7.09982395e-01
-4.25277799e-01 -9.33111235e-02 4.06061143e-01 -4.93008316e-01
5.87580919e-01 -1.05539227e+00 -3.80971849e-01 7.29717268e-03
-1.11837506e+00 -2.72780508e-01 6.83485687e-01 1.26906335e+00
7.98914194e-01 3.12783420e-01 1.18517697e+00 -1.03955948e+00
7.40360618e-01 -9.47864830e-01 -8.05214107e-01 2.45993540e-01
-6.87324405e-01 1.41782761e-01 5.61837554e-01 -5.10717750e-01
-1.10229850e+00 3.20161283e-02 -2.05697164e-01 -2.32399464e-01
-4.79892373e-01 1.11165309e+00 -4.11026746e-01 3.03496093e-01
8.94223213e-01 -2.90466636e-01 -9.14770141e-02 -4.29043621e-01
5.39987981e-01 3.40049446e-01 6.10373437e-01 -7.64627516e-01
1.35276884e-01 3.43792468e-01 8.12640727e-01 -3.92409414e-01
-9.14315641e-01 -3.44509929e-01 -5.50872862e-01 2.41533550e-03
5.83524764e-01 -8.12508643e-01 -1.37420285e+00 1.71861902e-01
-1.08065760e+00 -4.42480534e-01 1.29894823e-01 4.53605443e-01
-1.95174918e-01 -2.15206459e-01 -6.43011153e-01 -7.12538779e-01
-3.46735895e-01 -7.53587306e-01 1.14049971e+00 -1.71808135e-02
-3.51590931e-01 -1.09392738e+00 5.87335527e-01 -4.14887965e-02
2.47986555e-01 6.55369341e-01 1.21695018e+00 -5.70075810e-01
-2.99456656e-01 -2.46629477e-01 -3.28343242e-01 -4.68924850e-01
3.05998623e-01 4.43746477e-01 -9.06108081e-01 2.38777623e-01
-7.74123132e-01 -1.73902452e-01 1.04833722e+00 1.07383907e+00
1.43025148e+00 -3.38750005e-01 -7.45159984e-01 6.44117117e-01
1.07494938e+00 1.12423591e-01 6.72512352e-01 -3.33998919e-01
9.02837217e-01 4.50655192e-01 2.07687259e-01 3.04983139e-01
2.02774592e-02 3.23093712e-01 7.15640545e-01 -1.93473548e-01
-1.57952383e-02 -5.08023798e-01 -1.33490618e-02 3.03511024e-01
-1.78687468e-01 -6.73109710e-01 -9.19578016e-01 4.49703932e-01
-2.36245060e+00 -8.26141715e-01 -8.57795417e-01 1.89480579e+00
1.37939692e+00 -4.43977118e-01 -1.49177849e-01 -1.60738215e-01
8.16076756e-01 -4.12783265e-01 -8.10491920e-01 -2.62577772e-01
-3.42448950e-01 2.15095818e-01 5.25741339e-01 1.56402394e-01
-1.02825117e+00 7.02683389e-01 7.63394403e+00 8.81261587e-01
-8.90799582e-01 1.72408804e-01 8.19858551e-01 -3.80178690e-01
-2.92966783e-01 -5.34147620e-02 -6.62335992e-01 4.29429352e-01
1.31162953e+00 -1.26419067e-01 4.26133037e-01 4.69612747e-01
9.46924031e-01 -2.55787130e-02 -1.24137473e+00 5.83944440e-01
-5.15181065e-01 -1.65283370e+00 5.53923845e-02 1.65733039e-01
8.29581439e-01 3.04501946e-03 -1.95673376e-01 1.22957431e-01
1.26046419e+00 -1.53923392e+00 -2.04923168e-01 7.69644678e-01
9.25143063e-01 -7.89068520e-01 9.16538358e-01 -3.38270068e-02
-3.89496356e-01 5.31061888e-02 -6.69864535e-01 -3.45447987e-01
-5.59269264e-02 1.84756517e+00 -8.61923397e-01 4.71894473e-01
8.40226948e-01 6.88696623e-01 3.20718959e-02 9.66465294e-01
-4.87340212e-01 1.38056624e+00 -2.55054563e-01 -6.69731498e-02
-3.48062545e-01 6.87451810e-02 3.99874568e-01 1.13677800e+00
3.95124376e-01 4.59583551e-01 2.11447887e-02 1.18460584e+00
-4.99320120e-01 -4.24380660e-01 -6.12602115e-01 -3.80654126e-01
6.62029564e-01 1.21837211e+00 -4.32380915e-01 -2.19697550e-01
1.36997133e-01 3.67027521e-01 2.71858752e-01 3.68731797e-01
-8.68687689e-01 1.49451680e-02 1.18492556e+00 -2.02355519e-01
-2.83627540e-01 1.39909923e-01 -3.73299301e-01 -7.73993075e-01
-8.69808614e-01 -8.04556847e-01 7.81244934e-01 -5.27602792e-01
-1.67008364e+00 -4.09644008e-01 -4.63643782e-02 -3.72589946e-01
-6.45936728e-02 -5.38064241e-01 -3.49233985e-01 7.79330671e-01
-9.62737441e-01 -1.11991119e+00 -1.03231214e-01 -3.36346738e-02
7.13341087e-02 3.46030653e-01 1.07981431e+00 4.84865457e-01
-1.16645885e+00 3.27315748e-01 3.57772797e-01 -6.37903437e-02
8.42898250e-01 -1.34819543e+00 3.07998240e-01 6.33226573e-01
-4.85122353e-01 1.11379743e+00 8.15665483e-01 -1.15730250e+00
-1.74649119e+00 -1.50749147e+00 6.56311095e-01 -3.34070474e-01
9.71439719e-01 -5.44236064e-01 -8.95196438e-01 6.77176952e-01
-3.36584061e-01 2.64448375e-01 9.80483413e-01 8.49834323e-01
-5.00316620e-01 -5.47179254e-03 -9.13181603e-01 6.85124695e-01
1.47857237e+00 -1.18313171e-01 -2.00364366e-01 2.98812896e-01
8.20121884e-01 -2.78127730e-01 -1.08899045e+00 4.23262209e-01
7.25292087e-01 -3.36357355e-01 1.05068767e+00 -1.18112814e+00
7.56038308e-01 -5.30300677e-01 2.37590477e-01 -1.77211869e+00
-5.89022934e-01 -5.50021648e-01 -1.56625077e-01 1.20921469e+00
2.52347052e-01 -4.09434348e-01 5.63420177e-01 4.73545253e-01
-2.36688107e-02 -3.67869467e-01 -1.22153938e+00 -5.18096089e-01
1.21039055e-01 -3.99251252e-01 7.75112271e-01 1.53569520e+00
1.44468978e-01 3.79356831e-01 -5.38506329e-01 1.85561597e-01
7.19470918e-01 -2.38431573e-01 4.84009713e-01 -1.64324951e+00
-6.30046055e-02 -3.64156961e-01 -2.07743958e-01 -1.57262817e-01
2.46012703e-01 -8.63783479e-01 4.01130646e-01 -1.28269005e+00
4.92961377e-01 -4.28110808e-01 -5.23042083e-01 1.12643635e+00
-6.63714111e-01 3.53566632e-02 -1.01112521e+00 -3.19608718e-01
-2.78317630e-01 5.22343755e-01 8.70986581e-01 -2.20030531e-01
-5.97324371e-02 -5.14221847e-01 -9.41312850e-01 5.04417598e-01
8.69018316e-01 -8.58923137e-01 -2.79096246e-01 -1.69706896e-01
6.37837648e-01 1.43128783e-01 7.76044309e-01 -1.74384743e-01
2.24967793e-01 -5.68167448e-01 6.45766497e-01 -2.21599147e-01
-3.02547723e-01 -2.02654555e-01 8.65788341e-01 4.44522560e-01
-7.27903128e-01 -2.49506250e-01 3.89498234e-01 9.48578060e-01
3.23714763e-02 3.45559776e-01 3.64858836e-01 -9.07156393e-02
-2.67200053e-01 4.07750100e-01 -5.55357456e-01 -2.57009029e-01
3.59495819e-01 4.61181313e-01 -7.14036107e-01 -1.18443586e-01
-6.62599504e-01 2.74469167e-01 -2.00199276e-01 2.36893311e-01
3.88637692e-01 -1.32778084e+00 -9.68650222e-01 -3.35328311e-01
9.15907174e-02 1.25674739e-01 4.22113806e-01 9.56754267e-01
2.11782139e-02 5.74287117e-01 6.84719682e-02 -3.11787754e-01
-1.18863869e+00 6.47552371e-01 2.19539121e-01 -5.70958033e-02
-1.25952229e-01 9.97146189e-01 3.56643438e-01 -8.86287808e-01
-5.68358414e-02 -2.21605912e-01 -9.46682394e-02 7.60381669e-02
5.48852205e-01 5.19882977e-01 -4.68170121e-02 3.30287099e-01
-3.21098745e-01 -2.11827263e-01 3.87119561e-01 4.24719721e-01
1.53986061e+00 3.50174367e-01 -1.01135075e+00 4.95001167e-01
8.39581490e-01 -3.53361189e-01 -7.75751173e-01 1.72678664e-01
5.77127226e-02 -1.78993806e-01 4.27268118e-01 -1.23940277e+00
-8.64209771e-01 4.66121644e-01 4.28430021e-01 1.26014501e-02
9.40480888e-01 -1.12856977e-01 2.23510742e-01 2.50005990e-01
7.46799558e-02 -7.36830235e-01 -5.11026859e-01 2.94400334e-01
7.48455763e-01 -9.62394655e-01 6.49825707e-02 -6.35984480e-01
6.01932406e-01 9.60476220e-01 4.66777116e-01 2.08431542e-01
4.73314792e-01 5.51973164e-01 -1.98687255e-01 -5.33028841e-01
-1.36182737e+00 1.71219520e-02 1.13632597e-01 7.03528047e-01
9.07849729e-01 4.56214935e-01 -5.72385788e-01 7.76914001e-01
1.97725967e-01 6.12739027e-01 4.66538638e-01 4.00894761e-01
8.95538777e-02 -1.01279843e+00 -5.48604369e-01 8.80072117e-01
-7.78247774e-01 -6.08613193e-01 -5.93662381e-01 5.72399139e-01
2.48386994e-01 1.10010278e+00 -2.77765859e-02 -2.18340009e-01
1.32078469e-01 7.86860883e-02 1.69401333e-01 -4.46800351e-01
-1.97688788e-01 2.56868362e-01 3.53444368e-01 -6.81400299e-01
-3.48307610e-01 -6.77096188e-01 -1.35522926e+00 -7.17859924e-01
-3.49784642e-01 1.33023500e-01 5.73986471e-01 9.21165586e-01
8.27691257e-01 1.29403925e+00 3.02844252e-02 -2.87212551e-01
-9.43960920e-02 -1.05442882e+00 -6.29618645e-01 -9.14648101e-02
7.60685131e-02 -8.60956430e-01 -2.96175838e-01 1.73627734e-01]
|
[7.878788471221924, 5.3837103843688965]
|
e6711b3c-eb52-4f64-a8a3-544e152cd059
|
attentive-modality-hopping-mechanism-for
|
1912.00846
| null |
https://arxiv.org/abs/1912.00846v2
|
https://arxiv.org/pdf/1912.00846v2.pdf
|
Attentive Modality Hopping Mechanism for Speech Emotion Recognition
|
In this work, we explore the impact of visual modality in addition to speech and text for improving the accuracy of the emotion detection system. The traditional approaches tackle this task by fusing the knowledge from the various modalities independently for performing emotion classification. In contrast to these approaches, we tackle the problem by introducing an attention mechanism to combine the information. In this regard, we first apply a neural network to obtain hidden representations of the modalities. Then, the attention mechanism is defined to select and aggregate important parts of the video data by conditioning on the audio and text data. Furthermore, the attention mechanism is again applied to attend important parts of the speech and textual data, by considering other modality. Experiments are performed on the standard IEMOCAP dataset using all three modalities (audio, text, and video). The achieved results show a significant improvement of 3.65% in terms of weighted accuracy compared to the baseline system.
|
['Hwanhee Lee', 'Seunghyun Yoon', 'Subhadeep Dey', 'Kyomin Jung']
|
2019-11-29
| null | null | null | null |
['multimodal-emotion-recognition', 'multimodal-emotion-recognition']
|
['computer-vision', 'speech']
|
[ 2.75242299e-01 -6.64324360e-03 5.04671736e-03 -2.25314781e-01
-7.99771190e-01 -1.32325634e-01 6.87611938e-01 2.53941149e-01
-7.53711879e-01 5.09551525e-01 4.91583437e-01 2.46176943e-01
2.26607233e-01 -2.91926116e-01 -4.09248918e-01 -7.54806519e-01
1.43228173e-01 -1.13478586e-01 2.21159253e-02 -2.78791450e-02
1.49461702e-01 1.77938849e-01 -1.94733560e+00 6.60165608e-01
7.05013871e-01 1.42310894e+00 1.81693241e-01 6.15061939e-01
-2.96604514e-01 1.01367092e+00 -4.98469353e-01 -2.27243051e-01
-1.90765321e-01 -4.04625595e-01 -7.13588357e-01 3.74846339e-01
7.83460662e-02 -1.21048532e-01 -1.27416074e-01 9.53499198e-01
6.48388863e-01 4.59243953e-01 5.66515863e-01 -1.29446840e+00
-2.11833254e-01 4.42263722e-01 -7.35048056e-01 1.49385363e-01
6.45223618e-01 -2.15446755e-01 9.61462080e-01 -1.20391095e+00
5.32293558e-01 1.07238221e+00 1.25296995e-01 4.57210094e-01
-7.74635494e-01 -3.70389014e-01 4.06310707e-01 6.78269863e-01
-1.36245799e+00 -7.13047862e-01 1.00929213e+00 -3.54753315e-01
1.02479982e+00 2.82276899e-01 3.46668392e-01 1.26111829e+00
-1.00612879e-01 1.05198777e+00 8.50399673e-01 -7.05781758e-01
1.98363468e-01 4.31569427e-01 4.21426952e-01 2.99424976e-01
-4.00558352e-01 -3.30426037e-01 -7.31456280e-01 -4.80017252e-02
-1.73304304e-02 2.26829536e-02 -4.68679339e-01 -2.20726579e-02
-9.58093703e-01 5.51545501e-01 2.93740958e-01 5.08357823e-01
-8.79669785e-01 -1.45861074e-01 6.57904148e-01 1.25695392e-01
5.48697650e-01 1.35991666e-02 -3.19513410e-01 -1.23661287e-01
-8.70818853e-01 -3.07376117e-01 6.11378968e-01 6.57176912e-01
5.06145716e-01 3.40384804e-02 -3.46104234e-01 8.78535211e-01
5.95055699e-01 1.20261878e-01 4.66190219e-01 -6.80719435e-01
7.61679769e-01 7.36484528e-01 -4.23746295e-02 -8.90101969e-01
-2.93143034e-01 -1.91035524e-01 -7.41273820e-01 2.25739524e-01
3.68844792e-02 -4.52554733e-01 -1.11250877e+00 1.65875113e+00
3.57964545e-01 2.25280657e-01 3.36119324e-01 1.01918840e+00
1.17684507e+00 8.65532994e-01 4.47981924e-01 -3.63604665e-01
1.61185718e+00 -1.05771375e+00 -1.07929683e+00 -1.02620758e-01
2.47277170e-01 -6.89056098e-01 8.03638816e-01 4.31796789e-01
-1.07372844e+00 -6.25954270e-01 -1.05185008e+00 -2.53719091e-02
-6.16972387e-01 3.57647926e-01 6.09594099e-02 2.41266027e-01
-8.73549759e-01 1.00500047e-01 -6.34122372e-01 -5.22699058e-01
2.16807365e-01 3.53428125e-01 -4.49347615e-01 1.76733270e-01
-1.27634060e+00 8.07174206e-01 5.22043407e-01 2.57234246e-01
-6.14111960e-01 -1.75620362e-01 -8.59690845e-01 3.38676006e-01
3.59945923e-01 -3.68136376e-01 1.01625526e+00 -1.64613724e+00
-1.68556643e+00 5.48121870e-01 -3.41914684e-01 -2.44225413e-01
2.29627877e-01 -3.23887259e-01 -5.22443175e-01 5.89983821e-01
-4.57884878e-01 5.91561794e-01 9.44656193e-01 -1.17963207e+00
-7.96990335e-01 -3.31732035e-01 1.21433467e-01 5.24463296e-01
-7.01738358e-01 4.39371884e-01 -8.59750509e-01 -4.45065469e-01
-7.68702701e-02 -6.51139557e-01 1.21108159e-01 -3.62282366e-01
-2.75240988e-01 -2.18731716e-01 1.07684636e+00 -9.23841536e-01
1.28014827e+00 -2.44210672e+00 6.74193561e-01 1.88620731e-01
1.37730643e-01 1.54795289e-01 -1.24219999e-01 3.63895178e-01
-1.48754969e-01 8.33156854e-02 2.50667776e-03 -7.91150272e-01
4.26846966e-02 1.35366034e-04 -9.74693298e-02 2.29593322e-01
3.93621117e-01 6.51820898e-01 -3.77778351e-01 -6.59561098e-01
3.16699207e-01 9.02682364e-01 -4.30757195e-01 2.96229184e-01
-4.70275208e-02 5.23633778e-01 -4.57687885e-01 5.81541896e-01
4.32002604e-01 -1.93091519e-02 1.28508851e-01 -3.10902745e-01
-1.81257874e-02 1.71923533e-01 -1.16823065e+00 1.58195496e+00
-3.69733483e-01 6.69710755e-01 4.24587250e-01 -1.07783985e+00
6.79564655e-01 9.09987092e-01 5.20670116e-01 -6.26812875e-01
6.67856276e-01 -2.44680524e-01 -1.15638100e-01 -8.40036213e-01
5.06750286e-01 -1.39356494e-01 -1.87889770e-01 1.99501410e-01
5.39829493e-01 5.48702478e-01 1.00428388e-01 1.98483005e-01
6.65886164e-01 8.04603472e-03 1.79323658e-01 2.96353608e-01
8.72941315e-01 -2.72962004e-01 2.85945177e-01 2.99850166e-01
-2.45217159e-01 4.19058770e-01 5.66493154e-01 3.40610333e-02
-5.98675609e-01 -5.56663513e-01 2.05326363e-01 1.33983541e+00
2.14193344e-01 -4.98834103e-01 -6.55967236e-01 -7.58759379e-01
-4.69554394e-01 5.34300327e-01 -8.27670574e-01 -2.07412764e-01
-1.59930959e-01 -6.26624286e-01 2.65006065e-01 6.74240589e-01
4.63198751e-01 -1.21521556e+00 -7.16720939e-01 -3.18101086e-02
-5.75859666e-01 -1.19502795e+00 -1.45189002e-01 2.37720236e-01
-3.90719980e-01 -7.80854106e-01 -5.90960324e-01 -5.59937358e-01
3.86443913e-01 -8.75266194e-02 6.33082271e-01 -3.46974842e-02
1.87224392e-02 5.84214628e-01 -6.93806469e-01 -5.13116717e-01
-1.64555460e-02 6.16891943e-02 -1.96017697e-01 8.40036392e-01
4.12487149e-01 -2.19312042e-01 -3.31047356e-01 -4.71270196e-02
-9.41066384e-01 4.82423278e-03 5.45098901e-01 7.50222743e-01
3.79277050e-01 1.16655059e-01 4.17116463e-01 -3.41756046e-01
4.91365224e-01 -6.49639487e-01 -1.14381686e-01 2.80109376e-01
-4.89081964e-02 -6.48554787e-02 3.50626558e-01 -6.24956131e-01
-1.40933609e+00 3.20385039e-01 -1.84584036e-01 -5.60533583e-01
-4.88353342e-01 7.46853292e-01 -5.00086367e-01 1.60200641e-01
9.22075883e-02 2.92355027e-02 -2.22160086e-01 -4.53339756e-01
9.65965539e-02 1.11158502e+00 2.68792063e-01 -2.55899966e-01
2.05589518e-01 3.92882228e-01 -4.32055682e-01 -9.94830251e-01
-4.92271334e-01 -5.89363515e-01 -5.55469275e-01 -5.57708681e-01
1.25509441e+00 -7.96225190e-01 -7.75931716e-01 4.27581489e-01
-1.07728398e+00 1.71371713e-01 -4.63371761e-02 8.70784104e-01
-3.10696751e-01 3.08020055e-01 -5.46956182e-01 -1.18208754e+00
-3.18034500e-01 -1.20520627e+00 1.12108004e+00 3.69985998e-01
-2.10739419e-01 -8.22771907e-01 -2.24454120e-01 3.26425612e-01
2.75109261e-01 1.60850570e-01 5.59834957e-01 -9.74512339e-01
-4.15653139e-02 -3.80773336e-01 -2.37274915e-01 3.33367854e-01
-1.25356048e-01 -1.25558476e-03 -1.33554769e+00 -2.99255531e-02
1.66680768e-01 -4.12651569e-01 1.09819293e+00 1.41587108e-01
8.95587444e-01 -9.34474021e-02 -1.74825251e-01 2.26125389e-01
1.20433426e+00 3.83676738e-01 6.10047221e-01 1.77179128e-01
6.24693871e-01 8.10984254e-01 5.44745743e-01 6.29475415e-01
3.36150646e-01 6.82410359e-01 6.93859160e-01 -2.02691808e-01
7.45438263e-02 2.13739142e-01 4.44888502e-01 7.78169990e-01
-2.13582665e-01 -4.89252597e-01 -7.84888566e-01 4.94282752e-01
-1.92686033e+00 -1.11125565e+00 -4.23862115e-02 1.96815073e+00
4.04776365e-01 -5.32635562e-02 1.79445803e-01 4.19910043e-01
8.30395222e-01 1.30301267e-01 -1.84414700e-01 -4.78080094e-01
-7.41778016e-02 -2.57109087e-02 -1.08791828e-01 4.31609601e-01
-1.33319426e+00 6.07142091e-01 6.07308102e+00 6.11695170e-01
-1.45312655e+00 1.75004676e-01 4.58936274e-01 -4.16123688e-01
-1.62061173e-02 -2.84290731e-01 -5.80753267e-01 4.90286112e-01
1.06878459e+00 1.42616659e-01 3.38436007e-01 4.26917106e-01
2.05323532e-01 -2.89253086e-01 -8.77892315e-01 9.85366344e-01
4.59530205e-01 -5.98869562e-01 -1.59437265e-02 -2.09607854e-01
2.99173117e-01 -2.19603643e-01 2.36031562e-02 4.85738456e-01
-3.76422584e-01 -8.73839974e-01 8.26566279e-01 8.21195185e-01
3.52778107e-01 -8.44481945e-01 1.06604564e+00 3.08262080e-01
-1.17706919e+00 -1.93457142e-01 1.06319077e-01 -4.05029878e-02
2.39741206e-01 1.53533846e-01 -5.04614234e-01 8.34340155e-01
9.59956288e-01 5.21605551e-01 -2.80979365e-01 8.88979137e-01
-1.70035303e-01 5.86226285e-01 -2.49955401e-01 1.75287537e-02
1.66466415e-01 1.73881486e-01 4.60631371e-01 1.34826529e+00
2.81073958e-01 6.52340651e-02 1.24189690e-01 4.22842890e-01
-1.50338277e-01 3.90049309e-01 -2.69816041e-01 -3.52308489e-02
1.03764616e-01 1.42487633e+00 -5.49525201e-01 -6.49312794e-01
-6.62139118e-01 1.09348512e+00 2.86761314e-01 7.10678339e-01
-1.03405643e+00 -6.65947318e-01 3.52229327e-01 -4.15288597e-01
5.67535639e-01 1.12657979e-01 5.15762344e-03 -1.17874873e+00
1.07296325e-01 -6.65095150e-01 5.92174888e-01 -1.04189146e+00
-1.00840378e+00 8.73550236e-01 -1.30920142e-01 -1.21425354e+00
-1.98145017e-01 -5.13218939e-01 -5.14545083e-01 1.02779424e+00
-1.46047747e+00 -1.05104530e+00 -3.34331483e-01 8.11689794e-01
6.37937188e-01 -6.80734927e-04 7.40411341e-01 5.06012082e-01
-7.01160073e-01 3.93609822e-01 -1.94663912e-01 1.54942900e-01
7.76859641e-01 -9.63534951e-01 -5.12132883e-01 8.12919438e-01
1.34593546e-01 3.48361671e-01 5.56306243e-01 -3.89699101e-01
-1.24265039e+00 -7.98607945e-01 9.77068365e-01 -7.61492997e-02
4.98888254e-01 -2.42975071e-01 -8.97933125e-01 5.78734100e-01
8.84214044e-01 -1.06034882e-01 8.13424826e-01 8.08001533e-02
-3.25020283e-01 7.45026097e-02 -1.00343692e+00 3.78746599e-01
3.93177271e-01 -6.77484453e-01 -8.10188949e-01 -2.57879347e-01
6.25888109e-01 -1.98973984e-01 -8.30218256e-01 4.54100728e-01
5.39845288e-01 -7.28331506e-01 7.10045099e-01 -6.50890112e-01
5.62961578e-01 -2.68758416e-01 -4.03083503e-01 -1.28553128e+00
-1.30122498e-01 -2.27375031e-01 -3.20040643e-01 1.52567923e+00
4.19372886e-01 -3.77191037e-01 3.54419410e-01 5.04636705e-01
-9.54523906e-02 -4.99508321e-01 -9.98352706e-01 -1.33750096e-01
-4.92484540e-01 -5.60181558e-01 3.05250555e-01 8.85146201e-01
5.13669312e-01 6.26086175e-01 -6.90174103e-01 2.42857039e-01
1.75380692e-01 4.03554440e-02 3.66165072e-01 -1.10554957e+00
-2.20939100e-01 -4.18421298e-01 -3.91875446e-01 -5.79176366e-01
2.79524356e-01 -7.51073420e-01 1.90958500e-01 -1.53447258e+00
2.83725947e-01 3.64733994e-01 -8.61365438e-01 5.79297125e-01
-3.05172294e-01 3.40282112e-01 4.75128353e-01 -7.14909732e-02
-1.00399959e+00 7.18768895e-01 7.85187781e-01 -2.35620871e-01
-2.30482355e-01 -2.58386105e-01 -6.41225576e-01 7.24638641e-01
8.42929840e-01 -2.03392133e-01 -2.75182068e-01 -4.55084980e-01
-1.31053388e-01 1.55535549e-01 3.68693888e-01 -9.16553974e-01
4.23232287e-01 8.86103660e-02 5.28124452e-01 -6.84008658e-01
7.43096650e-01 -1.17437077e+00 -9.31394324e-02 -4.72877324e-02
-4.85684603e-01 -2.38451377e-01 4.78882402e-01 4.77193594e-01
-6.36862636e-01 -7.93827549e-02 5.59401512e-01 2.56718993e-01
-8.33365083e-01 -4.04383019e-02 -7.62903273e-01 -3.35267574e-01
1.13641834e+00 -1.43206507e-01 -1.00255720e-01 -4.79735702e-01
-1.30826032e+00 3.05917919e-01 -3.32837589e-02 5.56614578e-01
6.91955209e-01 -1.34820211e+00 -5.40573359e-01 2.16508657e-01
2.94456482e-01 -6.55664623e-01 5.28190553e-01 1.25670719e+00
1.78509176e-01 2.77256906e-01 -3.46618414e-01 -4.92476761e-01
-1.71264553e+00 7.81628728e-01 2.17294022e-01 -1.03832960e-01
-1.52287453e-01 5.88235617e-01 1.32387653e-01 -4.41241376e-02
6.73255503e-01 -1.50807425e-01 -8.64471793e-01 6.80720389e-01
7.09448099e-01 2.17280269e-01 7.82344416e-02 -1.10162270e+00
-5.20661533e-01 5.36557913e-01 1.64022774e-01 -4.61361259e-01
1.32467151e+00 -5.35681844e-01 3.24489474e-02 6.24672592e-01
1.22653818e+00 3.09229158e-02 -9.13266063e-01 -3.11287016e-01
-5.87493852e-02 -4.85238805e-02 1.70865119e-01 -8.07905853e-01
-1.13765037e+00 1.18812025e+00 6.56513512e-01 2.99050122e-01
1.54590178e+00 4.37093042e-02 3.01560730e-01 3.82283470e-03
-1.40711978e-01 -1.14880884e+00 3.43950726e-02 5.75399876e-01
9.18666840e-01 -1.24835289e+00 -3.46251041e-01 -1.69995397e-01
-9.78050590e-01 1.01639473e+00 5.97685039e-01 1.05721317e-01
6.33691728e-01 2.42862687e-01 1.43275782e-01 -1.11663260e-01
-1.01471996e+00 -4.79999036e-01 6.16551876e-01 1.84828818e-01
5.42530298e-01 -3.33527952e-01 -3.13634217e-01 9.28716063e-01
4.96274024e-01 3.23858997e-03 1.14940479e-01 8.90416026e-01
-4.85788584e-01 -8.74638736e-01 -6.28040373e-01 1.64927095e-01
-7.97584295e-01 -4.70905602e-02 -6.61812961e-01 4.73599404e-01
1.39769152e-01 1.30610442e+00 8.08923244e-02 -5.46374321e-01
3.71249974e-01 6.07488453e-01 1.00496396e-01 -3.00027430e-01
-7.62300909e-01 4.51504618e-01 2.55215853e-01 -4.55922902e-01
-7.17646658e-01 -6.20072842e-01 -1.28761411e+00 1.89600199e-01
-3.33216161e-01 2.57590592e-01 7.59390652e-01 1.03028584e+00
4.95428115e-01 9.51404154e-01 5.20209908e-01 -1.08686626e+00
-4.83341441e-02 -1.17523515e+00 -3.06573242e-01 5.16285598e-01
4.75049555e-01 -7.11282670e-01 -4.32101548e-01 7.59745464e-02]
|
[13.273224830627441, 5.203298091888428]
|
62c1045d-7eef-4749-9e74-eeb0c9635192
|
pina-leveraging-side-information-in-extreme
|
2305.12349
| null |
https://arxiv.org/abs/2305.12349v1
|
https://arxiv.org/pdf/2305.12349v1.pdf
|
PINA: Leveraging Side Information in eXtreme Multi-label Classification via Predicted Instance Neighborhood Aggregation
|
The eXtreme Multi-label Classification~(XMC) problem seeks to find relevant labels from an exceptionally large label space. Most of the existing XMC learners focus on the extraction of semantic features from input query text. However, conventional XMC studies usually neglect the side information of instances and labels, which can be of use in many real-world applications such as recommendation systems and e-commerce product search. We propose Predicted Instance Neighborhood Aggregation (PINA), a data enhancement method for the general XMC problem that leverages beneficial side information. Unlike most existing XMC frameworks that treat labels and input instances as featureless indicators and independent entries, PINA extracts information from the label metadata and the correlations among training instances. Extensive experimental results demonstrate the consistent gain of PINA on various XMC tasks compared to the state-of-the-art methods: PINA offers a gain in accuracy compared to standard XR-Transformers on five public benchmark datasets. Moreover, PINA achieves a $\sim 5\%$ gain in accuracy on the largest dataset LF-AmazonTitles-1.3M. Our implementation is publicly available.
|
['Hsiang-Fu Yu', 'Olgica Milenkovic', 'Wei-Cheng Chang', 'Jyun-Yu Jiang', 'Cho-Jui Hsieh', 'Jiong Zhang', 'Eli Chien']
|
2023-05-21
| null | null | null | null |
['extreme-multi-label-classification']
|
['methodology']
|
[ 2.30912685e-01 -2.58129448e-01 -7.65900433e-01 -6.44654751e-01
-1.15198123e+00 -4.49985415e-01 4.15920734e-01 4.62307572e-01
-5.02306223e-01 7.00595379e-01 -1.01727478e-01 -2.89455920e-01
-5.89447141e-01 -5.84311604e-01 -3.84511679e-01 -6.31007254e-01
3.75678726e-02 6.84519053e-01 6.86466787e-03 2.49914955e-02
5.12195468e-01 -2.21507866e-02 -1.97470009e+00 7.80294418e-01
9.32177186e-01 1.53117919e+00 -4.93408553e-02 3.25761288e-02
-6.00497842e-01 8.09633315e-01 -3.33262593e-01 -5.19983053e-01
2.22752750e-01 -4.99280691e-02 -1.05054712e+00 -1.92588359e-01
5.37044287e-01 3.11055392e-01 3.23320895e-01 1.06716168e+00
4.80300695e-01 -8.38153716e-03 8.70072722e-01 -1.58874369e+00
-5.98663092e-01 7.03111649e-01 -9.12019551e-01 9.15808454e-02
4.39991951e-01 -2.84145445e-01 1.51477587e+00 -1.16590774e+00
6.82263136e-01 1.30873120e+00 8.85509431e-01 4.11157697e-01
-1.13363039e+00 -9.81658280e-01 3.66567940e-01 3.62226874e-01
-1.43286133e+00 -7.80242831e-02 5.90896487e-01 -1.72684804e-01
8.28477561e-01 5.02981901e-01 -9.97735411e-02 9.39660847e-01
-2.07963064e-01 1.20487249e+00 1.48506331e+00 -7.12486804e-01
2.60155797e-01 4.95299578e-01 6.65397584e-01 5.15663922e-01
-9.47496071e-02 -6.90432563e-02 -7.91660249e-01 -3.47826183e-01
-1.44578353e-01 2.12701380e-01 1.25817927e-02 -2.38981038e-01
-1.16751182e+00 9.73613083e-01 3.35031182e-01 1.00819781e-01
-1.62512869e-01 9.84213129e-02 4.35080051e-01 3.55204999e-01
8.31980169e-01 7.83610582e-01 -8.17766547e-01 6.20365180e-02
-6.45883024e-01 3.05950135e-01 6.87601924e-01 1.27702129e+00
6.04742825e-01 -6.07190073e-01 -2.59071708e-01 1.10966110e+00
4.67930943e-01 4.91954446e-01 5.67250431e-01 -1.02460825e+00
6.59174502e-01 1.03011751e+00 -2.17835903e-01 -7.41253793e-01
-6.66087925e-01 -6.42785907e-01 -4.27373976e-01 3.44185196e-02
1.85488597e-01 6.49079457e-02 -5.57080865e-01 1.64964521e+00
5.00806570e-01 1.65093631e-01 -8.37525278e-02 5.50057471e-01
9.98021185e-01 3.73590887e-01 3.88339102e-01 -2.51248926e-01
1.30691910e+00 -9.66617763e-01 -6.82580590e-01 -9.82396975e-02
1.03209865e+00 -8.08976889e-01 1.26006281e+00 3.05483490e-01
-5.57438552e-01 -4.70202774e-01 -8.83997262e-01 6.44847602e-02
-8.54177713e-01 1.96916535e-02 9.86793339e-01 5.18457294e-01
-6.92993999e-01 6.51251137e-01 -1.39547437e-01 -1.40531048e-01
4.62220758e-01 4.96512115e-01 -1.13216758e-01 -3.44629049e-01
-1.20314014e+00 7.49738455e-01 5.57873547e-01 -3.91326636e-01
-3.36503267e-01 -1.05712259e+00 -6.21909559e-01 3.70873176e-02
8.27526867e-01 -2.38498986e-01 1.29803669e+00 -6.57426894e-01
-8.78007948e-01 8.51441383e-01 -2.07734749e-01 -1.49580002e-01
2.46274337e-01 -5.02626121e-01 -7.24955738e-01 -4.83887382e-02
6.66660726e-01 9.59661186e-01 5.47793984e-01 -1.49344599e+00
-1.33958590e+00 -3.90863627e-01 -8.77183024e-03 3.53478938e-01
-4.41123247e-01 3.27611677e-02 -4.33448017e-01 -5.74552357e-01
1.00213036e-01 -1.02333307e+00 -9.42760110e-02 -2.48962343e-01
-4.30457294e-01 -7.99476504e-01 9.36208844e-01 9.32011306e-02
1.55418587e+00 -2.08142185e+00 -1.94924220e-01 2.98791379e-01
2.24198908e-01 1.75623223e-01 -7.44246095e-02 4.06960249e-01
-1.25728741e-01 1.75087765e-01 5.22840060e-02 -5.32968819e-01
3.18519622e-01 -1.05688579e-01 -3.10666710e-01 1.88512519e-01
-1.12998888e-01 1.10618925e+00 -1.04002893e+00 -7.78535068e-01
6.77696019e-02 -8.43412057e-03 -3.22250992e-01 -2.25702673e-01
-5.50020039e-01 1.20146506e-01 -4.59966838e-01 1.13392675e+00
2.85482317e-01 -8.65112305e-01 3.95694934e-03 1.83328941e-01
1.79630414e-01 1.24865636e-01 -1.22167945e+00 1.79058814e+00
-4.14753735e-01 1.49798423e-01 -4.94643003e-01 -7.81901479e-01
5.21464765e-01 3.65007758e-01 8.38941157e-01 -8.38206828e-01
9.84831899e-02 5.30867517e-01 -4.04771119e-01 -1.63312152e-01
3.89663011e-01 1.37112960e-02 -3.39275450e-01 7.70091951e-01
3.53051014e-02 3.75373632e-01 3.35085928e-01 3.02087247e-01
8.04560125e-01 -2.76853014e-02 4.05424953e-01 -5.81391096e-01
4.98299897e-01 1.95621058e-01 5.92117965e-01 8.64736319e-01
-1.77190006e-01 2.99022108e-01 2.29023948e-01 -4.05438900e-01
-4.98936087e-01 -7.87920296e-01 -4.47723597e-01 1.75605059e+00
3.92334849e-01 -7.63713121e-01 -4.56640869e-01 -1.32859707e+00
4.56279367e-01 9.96272922e-01 -6.50191426e-01 -1.11720793e-01
-2.91486055e-01 -9.58285511e-01 2.53684998e-01 5.26871860e-01
3.90646011e-01 -1.08709645e+00 -3.68866026e-02 1.10183537e-01
-1.81393489e-01 -9.01398420e-01 -5.09855628e-01 4.35573816e-01
-7.14252532e-01 -1.11293864e+00 -5.13046421e-02 -7.50373125e-01
4.65706527e-01 3.91786128e-01 1.46366203e+00 2.40012705e-02
-3.02691251e-01 2.82208741e-01 -5.78772843e-01 -6.93230927e-01
4.02596816e-02 5.03685832e-01 1.54455945e-01 6.36333227e-02
1.01396096e+00 -2.41788700e-01 -5.94910502e-01 4.72194642e-01
-5.98457217e-01 -4.78253886e-02 3.94690365e-01 7.28561878e-01
7.64016867e-01 2.14047268e-01 1.14316666e+00 -1.74284029e+00
4.72310811e-01 -8.32268178e-01 -4.52807665e-01 5.85716963e-01
-1.64491308e+00 1.11570373e-01 5.03332078e-01 -2.85517603e-01
-1.06724560e+00 1.12575531e-01 4.54356559e-02 -8.88948441e-02
-1.19135663e-01 5.09193003e-01 -1.37749523e-01 1.66646808e-01
6.37071848e-01 -1.26907796e-01 -3.61146003e-01 -6.99017465e-01
5.89230657e-01 8.00559938e-01 4.24661428e-01 -7.65355170e-01
3.89955670e-01 3.51265192e-01 5.91466650e-02 -4.81984392e-02
-1.58061004e+00 -1.25752938e+00 -6.86907589e-01 -2.08210230e-01
6.63432598e-01 -8.46513450e-01 -8.50754559e-01 2.66564395e-02
-6.15000188e-01 1.65483668e-01 -3.20731819e-01 3.16538543e-01
-4.43275183e-01 -1.42215401e-01 -5.86875260e-01 -5.56804597e-01
-3.42944264e-01 -1.12159443e+00 1.43268800e+00 4.69585434e-02
-1.91762581e-01 -1.17592525e+00 -1.28871962e-01 4.86730903e-01
2.61895478e-01 4.82950844e-02 1.36079097e+00 -9.10189927e-01
-5.76030016e-01 -2.66665936e-01 -3.06622386e-01 -4.48055342e-02
2.53434759e-02 -5.95660269e-01 -1.16351986e+00 -3.20545554e-01
-3.10205728e-01 -6.95212483e-01 1.02016723e+00 1.49533704e-01
1.47235835e+00 1.62292831e-02 -7.20521092e-01 4.53558892e-01
1.59383655e+00 4.19016898e-01 8.21379423e-02 3.49027961e-01
6.33663535e-01 5.68555832e-01 1.15900505e+00 4.13034111e-01
3.48820806e-01 7.22813189e-01 5.20763159e-01 -8.24312791e-02
6.49335887e-03 -2.03318819e-01 -2.17856213e-01 7.02827573e-01
1.69989824e-01 -1.65564358e-01 -8.78372252e-01 3.15913379e-01
-2.04289055e+00 -7.21533954e-01 -2.03151867e-01 1.92463994e+00
9.90585446e-01 5.03904335e-02 -1.82550535e-01 2.06811115e-01
6.26130760e-01 -6.69912025e-02 -7.10338414e-01 1.40563667e-01
-1.39417887e-01 1.84970319e-01 5.64722240e-01 3.06708306e-01
-1.51704252e+00 8.65225792e-01 6.02467680e+00 1.22392499e+00
-7.23176003e-01 3.89495075e-01 7.53343582e-01 -4.56184655e-01
-2.81953454e-01 -1.97664842e-01 -1.46780860e+00 4.95903909e-01
9.34419870e-01 2.82763578e-02 1.48805231e-01 1.19916618e+00
-4.95722175e-01 -1.03342995e-01 -1.36367905e+00 1.16516089e+00
1.54371157e-01 -1.30594432e+00 -1.52743891e-01 2.82284617e-01
9.80906904e-01 -4.72110286e-02 4.80547816e-01 8.81783664e-01
5.54951668e-01 -9.13972795e-01 6.04668796e-01 4.43807840e-01
9.67380941e-01 -9.03324842e-01 8.16549420e-01 4.59895313e-01
-1.11715138e+00 -4.44022000e-01 -2.47325271e-01 1.53960720e-01
-6.61201179e-02 5.97401738e-01 -6.68845952e-01 5.84709764e-01
7.74306476e-01 8.06714594e-01 -1.04765427e+00 6.84201539e-01
9.48419794e-02 6.08624041e-01 -5.97459413e-02 -1.11506768e-01
3.25888515e-01 -9.17783380e-02 -2.44433656e-02 1.06826711e+00
7.61537105e-02 6.79117218e-02 4.12661016e-01 4.03474361e-01
-4.00283486e-01 5.82715452e-01 -5.43917656e-01 3.04977298e-01
7.36666858e-01 1.24678349e+00 -7.84925938e-01 -4.35069978e-01
-8.18814099e-01 7.10578442e-01 6.28854454e-01 1.69245705e-01
-8.59620869e-01 -3.34587961e-01 4.53236669e-01 -9.97416452e-02
1.79874256e-01 5.90649426e-01 -6.46904171e-01 -8.93910110e-01
1.53358757e-01 -9.16763961e-01 7.67477214e-01 -2.67931074e-01
-1.65871954e+00 3.40832710e-01 1.67012922e-02 -1.36351156e+00
-3.05201888e-01 -6.43909335e-01 -4.49259542e-02 6.43213570e-01
-1.55498052e+00 -1.16854632e+00 -8.24761912e-02 3.62552077e-01
6.32131696e-01 -4.26723450e-01 1.04827785e+00 5.42903364e-01
-5.35420299e-01 7.22681940e-01 5.78063011e-01 -2.23424926e-01
9.47316587e-01 -1.65952098e+00 2.35756952e-02 2.84332097e-01
4.77239341e-01 7.03620732e-01 5.29124551e-02 -5.73356509e-01
-9.80396569e-01 -1.42984438e+00 1.18777096e+00 -9.36515152e-01
3.82990927e-01 -4.73178178e-01 -7.53040791e-01 6.26108468e-01
9.06691626e-02 2.79331714e-01 1.21624494e+00 7.64060140e-01
-6.79882705e-01 -3.98556113e-01 -9.00371253e-01 2.12731421e-01
1.14307141e+00 -6.56167269e-01 -2.81854361e-01 6.91785336e-01
8.47018540e-01 -8.07125028e-03 -1.11759460e+00 7.02937365e-01
4.90731537e-01 -6.78748071e-01 1.11673081e+00 -8.51751566e-01
3.90968949e-01 -2.60152459e-01 -4.87344772e-01 -1.15711284e+00
-4.47831720e-01 5.28573170e-02 -1.67187497e-01 1.39621294e+00
8.71858120e-01 -3.89970869e-01 7.04099596e-01 6.01600051e-01
-1.10348478e-01 -1.00136864e+00 -6.47168279e-01 -7.43834853e-01
1.09422162e-01 -6.48244560e-01 8.44251454e-01 1.30646420e+00
1.81278989e-01 6.72847927e-01 -1.88058123e-01 1.08307945e-02
7.45100021e-01 6.39488876e-01 3.07021707e-01 -1.67641389e+00
-2.02375159e-01 -3.89913470e-01 5.19588776e-02 -6.67419970e-01
5.38842499e-01 -1.41847777e+00 -1.20018855e-01 -1.27577245e+00
4.35938388e-01 -1.08137083e+00 -8.65740180e-01 5.01480341e-01
-6.37028456e-01 3.11231256e-01 1.92680717e-01 4.24709678e-01
-1.46962118e+00 2.64937431e-01 8.29244733e-01 -1.93129629e-01
-1.04225697e-02 1.91447049e-01 -9.64751005e-01 6.07007205e-01
6.09261155e-01 -7.09872961e-01 -8.49821925e-01 -2.16111004e-01
3.46328467e-01 -1.93691686e-01 -1.05005018e-01 -8.50811243e-01
3.68992895e-01 -1.41549960e-01 3.05032849e-01 -9.08778191e-01
1.35003135e-01 -9.51648891e-01 2.05349084e-02 1.44767001e-01
-8.86087656e-01 4.59661037e-02 -2.74526924e-01 7.94054806e-01
-2.16431215e-01 -4.54566240e-01 4.94969785e-01 -2.65769750e-01
-8.60327959e-01 2.65193969e-01 3.57946813e-01 3.34299117e-01
1.13931596e+00 2.93258488e-01 -6.55577660e-01 5.56720719e-02
-4.82779026e-01 5.50923228e-01 1.75167710e-01 5.35812616e-01
3.71557355e-01 -1.68090796e+00 -5.24614871e-01 5.38415797e-02
8.05585504e-01 -6.33920133e-02 -1.50406748e-01 6.22159183e-01
5.83199263e-02 8.65128815e-01 2.94788986e-01 -5.93211710e-01
-1.26127458e+00 1.01732290e+00 -1.03093222e-01 -5.38055599e-01
-3.05397570e-01 8.15253675e-01 1.52251288e-01 -6.54392362e-01
5.39701760e-01 3.08280177e-02 -3.75616401e-01 3.14457297e-01
6.78945601e-01 5.00113189e-01 2.81301141e-01 -3.75114053e-01
-3.82143050e-01 2.93205380e-01 -4.96988118e-01 6.70768097e-02
1.19221973e+00 -1.30618140e-01 -9.80701596e-02 7.62134612e-01
1.38203669e+00 -2.87078649e-01 -8.19589794e-01 -7.99700260e-01
7.90802598e-01 -5.07585049e-01 -1.06907487e-01 -1.15231240e+00
-9.49351847e-01 6.42163873e-01 6.65198267e-01 3.19311470e-02
1.06175196e+00 2.53258824e-01 5.36582530e-01 4.09061760e-01
7.19709158e-01 -1.39481068e+00 1.24669850e-01 2.93854356e-01
5.59008896e-01 -1.64812851e+00 1.11539148e-01 -5.37065268e-01
-7.69956112e-01 5.70975780e-01 8.74581218e-01 4.93880987e-01
9.40465093e-01 -2.74474616e-03 3.07597160e-01 -5.16937315e-01
-1.12929320e+00 -3.89613897e-01 4.32436258e-01 3.61476868e-01
5.09385407e-01 9.88964736e-02 -4.06685740e-01 5.89316905e-01
6.81413636e-02 -3.64474088e-01 -2.02077463e-01 1.08753252e+00
-3.97616833e-01 -1.24613357e+00 -7.84661695e-02 9.69047725e-01
-6.32354856e-01 -3.59702736e-01 -3.08637738e-01 8.42832148e-01
4.41522419e-01 1.03600192e+00 -1.81120798e-01 -3.27502817e-01
2.19806790e-01 5.70567846e-01 -3.55080888e-02 -7.87890375e-01
-5.99924505e-01 8.77489219e-04 -1.75776314e-02 -6.23210847e-01
-5.48699081e-01 -8.98345709e-01 -1.42936659e+00 -1.07252173e-01
-6.19028270e-01 3.20097953e-01 7.45930970e-01 8.17035973e-01
5.14143705e-01 4.67890978e-01 7.63044536e-01 -7.96535239e-02
-6.79135025e-01 -1.00832403e+00 -7.60073721e-01 9.00370598e-01
-1.68587156e-02 -1.02190220e+00 -1.23556919e-01 -1.90210208e-01]
|
[9.5662841796875, 4.390312194824219]
|
a2ad9f50-cd39-4fa4-870f-dccfc617bd84
|
towards-tokenized-human-dynamics
|
2111.11433
| null |
https://arxiv.org/abs/2111.11433v1
|
https://arxiv.org/pdf/2111.11433v1.pdf
|
Towards Tokenized Human Dynamics Representation
|
For human action understanding, a popular research direction is to analyze short video clips with unambiguous semantic content, such as jumping and drinking. However, methods for understanding short semantic actions cannot be directly translated to long human dynamics such as dancing, where it becomes challenging even to label the human movements semantically. Meanwhile, the natural language processing (NLP) community has made progress in solving a similar challenge of annotation scarcity by large-scale pre-training, which improves several downstream tasks with one model. In this work, we study how to segment and cluster videos into recurring temporal patterns in a self-supervised way, namely acton discovery, the main roadblock towards video tokenization. We propose a two-stage framework that first obtains a frame-wise representation by contrasting two augmented views of video frames conditioned on their temporal context. The frame-wise representations across a collection of videos are then clustered by K-means. Actons are then automatically extracted by forming a continuous motion sequence from frames within the same cluster. We evaluate the frame-wise representation learning step by Kendall's Tau and the lexicon building step by normalized mutual information and language entropy. We also study three applications of this tokenization: genre classification, action segmentation, and action composition. On the AIST++ and PKU-MMD datasets, actons bring significant performance improvements compared to several baselines.
|
['Stephen Lin', 'Fangyun Wei', 'Zhirong Wu', 'Xiao Sun', 'Kenneth Li']
|
2021-11-22
| null | null | null | null |
['action-understanding', 'human-dynamics', 'genre-classification']
|
['computer-vision', 'computer-vision', 'computer-vision']
|
[ 5.24854720e-01 -7.79415146e-02 -6.15104198e-01 -3.59952658e-01
-7.97982395e-01 -6.23396754e-01 6.21353090e-01 1.73883528e-01
-5.88467836e-01 5.12006879e-01 5.44033229e-01 5.57081811e-02
9.53753386e-03 -3.42409372e-01 -4.44576263e-01 -6.62789702e-01
-2.39295438e-01 2.87906438e-01 2.57053733e-01 9.40236300e-02
2.65150368e-01 4.00686488e-02 -1.67915356e+00 7.92211235e-01
6.50108457e-01 1.04690456e+00 1.07440360e-01 5.99737048e-01
-1.36625618e-01 1.33719361e+00 -3.05166662e-01 -1.85612738e-01
1.30689114e-01 -8.57534230e-01 -1.34579194e+00 4.97848779e-01
2.52156556e-01 -1.82700872e-01 -2.82508790e-01 8.83731663e-01
2.18268111e-01 5.20173132e-01 5.51852584e-01 -1.38006020e+00
-2.41886422e-01 6.65381074e-01 -5.75752497e-01 3.02705824e-01
7.66873002e-01 5.62547743e-02 1.25672221e+00 -5.39115429e-01
9.98533964e-01 1.24972367e+00 5.16150773e-01 5.62593222e-01
-9.42757964e-01 -4.17820603e-01 4.15748894e-01 7.97083735e-01
-1.16965342e+00 -2.44037420e-01 6.67204142e-01 -7.07444608e-01
9.48014915e-01 2.46583432e-01 7.82066762e-01 1.19887400e+00
-3.05658251e-01 1.39494765e+00 8.45391631e-01 -2.61579216e-01
1.62735730e-01 -4.47154433e-01 1.19047858e-01 5.70518792e-01
-2.75236189e-01 -5.17370939e-01 -7.63985872e-01 2.31501266e-01
6.10760629e-01 1.00152351e-01 -1.41163662e-01 -1.21749170e-01
-1.73090243e+00 6.59006178e-01 8.08782727e-02 5.68263233e-01
-4.77549285e-01 1.54467776e-01 8.45899820e-01 1.64047822e-01
6.28328800e-01 4.22975659e-01 -3.94361645e-01 -7.64434576e-01
-1.01992416e+00 2.96786636e-01 6.81886792e-01 8.05249155e-01
5.72724402e-01 -3.70839119e-01 -1.61772057e-01 8.21758449e-01
-3.10621839e-02 1.56670034e-01 8.29340279e-01 -1.27045012e+00
4.56212431e-01 6.51147246e-01 -1.90484319e-02 -1.03854692e+00
-3.67177069e-01 3.86544168e-01 -5.69704473e-01 -3.03328961e-01
5.87905526e-01 -1.34283886e-03 -7.91558683e-01 1.81563711e+00
4.68988895e-01 5.05366623e-01 -2.69114766e-02 9.61296797e-01
6.31218195e-01 6.47723556e-01 4.31255937e-01 -4.94796813e-01
1.59352779e+00 -1.16802478e+00 -8.61563563e-01 -6.55968785e-02
9.93982255e-01 -5.36671638e-01 8.54212165e-01 4.51988459e-01
-9.18856084e-01 -6.64937019e-01 -6.01801872e-01 -2.02908471e-01
-2.79794008e-01 9.02357176e-02 6.19257450e-01 1.99465632e-01
-6.38400793e-01 6.77704930e-01 -1.13071263e+00 -6.17345214e-01
6.06456041e-01 4.58571222e-03 -6.73519313e-01 3.55096087e-02
-1.11709535e+00 7.03078210e-01 6.48946524e-01 -1.84203506e-01
-7.66158581e-01 -4.64822531e-01 -1.09380555e+00 -3.18991125e-01
7.75366306e-01 -2.81002134e-01 1.17828631e+00 -1.45740902e+00
-1.45648456e+00 1.13924527e+00 -3.59730870e-01 -6.41811371e-01
4.84606028e-01 -5.00132501e-01 -2.51346171e-01 4.46461082e-01
3.97474825e-01 7.77502120e-01 7.27112830e-01 -8.99628997e-01
-1.14442265e+00 -3.06297123e-01 2.71255821e-01 5.69487989e-01
-2.93550551e-01 2.91605145e-01 -7.16507316e-01 -8.97389948e-01
1.57918200e-01 -1.08684492e+00 -2.40673646e-01 -1.74527958e-01
-3.10701936e-01 -4.24326450e-01 6.84592962e-01 -8.73043597e-01
1.41016126e+00 -2.20404768e+00 5.55824637e-01 -6.33263262e-03
1.12113208e-01 -4.62622475e-03 -1.07803136e-01 2.38052964e-01
-2.39069402e-01 2.05479413e-02 -2.50526488e-01 -3.43013138e-01
3.04857884e-02 3.61158401e-01 -1.75652578e-01 4.85461503e-01
4.87782881e-02 7.77418673e-01 -1.26406968e+00 -8.34347427e-01
2.02908427e-01 1.35752410e-01 -6.75730050e-01 1.63967833e-01
-2.92692155e-01 6.78274095e-01 -4.27719682e-01 5.59635103e-01
-7.09423497e-02 -3.07438523e-01 6.16307855e-01 -2.75526941e-01
-1.00056008e-01 3.21730733e-01 -1.17578852e+00 2.05676937e+00
-1.71362951e-01 8.17196190e-01 -4.14606661e-01 -1.46336949e+00
3.49672675e-01 4.51296419e-01 1.33896172e+00 -5.79902649e-01
1.34430379e-01 -9.62672457e-02 -1.80219710e-01 -1.05065358e+00
5.51818311e-01 -9.03662071e-02 -3.73924881e-01 6.77918494e-01
1.86954081e-01 3.30894947e-01 7.04057753e-01 2.02196926e-01
1.17439270e+00 6.06450915e-01 5.47238708e-01 1.29036009e-01
4.74856645e-01 2.47170582e-01 6.66576743e-01 3.92690182e-01
-3.87763351e-01 6.88535869e-01 6.29799783e-01 -5.26089668e-01
-7.76590347e-01 -7.93174922e-01 1.76557586e-01 1.47636008e+00
2.61539847e-01 -8.10680568e-01 -9.58229005e-01 -8.59712362e-01
-2.04833120e-01 3.46084714e-01 -7.11928844e-01 -1.02383696e-01
-8.07283223e-01 -6.11114383e-01 6.00547194e-01 6.39388323e-01
5.51880240e-01 -1.31313765e+00 -8.24066818e-01 2.23186344e-01
-9.07504201e-01 -1.48517656e+00 -5.21519542e-01 -5.36550023e-02
-6.19480550e-01 -1.28784883e+00 -5.99930882e-01 -7.90522099e-01
4.86009866e-01 2.25445539e-01 9.71562207e-01 -8.04622993e-02
-3.02970797e-01 5.80494940e-01 -9.17073548e-01 -4.34144363e-02
-2.02423438e-01 -7.60229751e-02 2.20136032e-01 3.65998179e-01
5.74076772e-01 -3.99098396e-01 -6.02840185e-01 5.00152588e-01
-7.64821947e-01 2.70147949e-01 3.76153678e-01 4.29967940e-01
8.80801499e-01 -2.28615552e-02 2.22143173e-01 -7.21886337e-01
1.33526564e-01 -5.47913611e-01 1.13259420e-01 2.26058960e-01
-2.46357732e-02 -1.88497767e-01 3.37638348e-01 -6.83493853e-01
-9.69836354e-01 3.87436986e-01 1.17611602e-01 -5.11811614e-01
-3.53713334e-01 4.96595144e-01 -1.40920177e-01 4.41058546e-01
3.52512985e-01 2.99222380e-01 -1.05404831e-01 -4.01391685e-01
6.02849066e-01 4.63436395e-01 7.94103146e-01 -5.83844960e-01
4.43674296e-01 6.60264730e-01 -4.03240204e-01 -9.82334137e-01
-1.09349120e+00 -8.91158700e-01 -1.09722519e+00 -5.44488311e-01
1.59667206e+00 -9.10684586e-01 -8.39938939e-01 4.65561450e-01
-1.01217842e+00 -4.18363363e-01 -4.08797204e-01 6.99028432e-01
-9.82253015e-01 7.26127148e-01 -6.14104867e-01 -5.39821267e-01
1.64317280e-01 -9.43795919e-01 1.13755679e+00 3.88317332e-02
-7.01458454e-01 -8.53699267e-01 1.89229801e-01 7.33163416e-01
-3.75523835e-01 4.18974817e-01 5.42632997e-01 -7.61592448e-01
-2.01746732e-01 4.23853286e-02 -2.09609400e-02 4.19456989e-01
2.16883495e-01 -1.61066145e-01 -7.03239560e-01 3.55568081e-02
-3.27705108e-02 -4.48860884e-01 8.87250364e-01 5.06861746e-01
1.32951522e+00 -3.24548960e-01 -3.26129138e-01 4.26773995e-01
9.02977705e-01 3.31997335e-01 7.14030027e-01 3.69400293e-01
9.95194495e-01 7.43187129e-01 9.51088071e-01 4.68880832e-01
4.88565356e-01 7.69717932e-01 3.07548009e-02 1.37970731e-01
1.61422137e-02 -3.09818804e-01 5.79457879e-01 9.66731548e-01
-6.43262744e-01 -1.56171933e-01 -8.55385840e-01 5.32503366e-01
-2.28917575e+00 -1.57348633e+00 4.25483920e-02 1.82646000e+00
8.75604689e-01 -1.91010684e-02 4.39960986e-01 2.55201995e-01
6.97841048e-01 3.67659360e-01 -3.57917279e-01 9.27419364e-02
-3.50168347e-03 -5.78292646e-02 2.02451020e-01 1.72233701e-01
-1.56379020e+00 1.13355744e+00 5.62279606e+00 9.78738427e-01
-8.66159856e-01 2.33852431e-01 6.11540318e-01 -1.59114346e-01
3.39327902e-01 1.05239125e-02 -4.50451881e-01 5.86990535e-01
7.90652573e-01 -6.30812049e-02 2.60891378e-01 8.08243275e-01
4.86143172e-01 -3.78781885e-01 -1.35129762e+00 1.24865544e+00
3.60105246e-01 -1.22506642e+00 5.38942739e-02 -1.32724509e-01
7.80294120e-01 -1.64571851e-01 -3.59738469e-01 3.47538710e-01
1.44562110e-01 -8.34394336e-01 9.20174420e-01 4.56542462e-01
6.36097372e-01 -4.72172111e-01 4.31655884e-01 2.48056173e-01
-1.60774803e+00 -1.07362553e-01 5.64819276e-02 -2.14095473e-01
5.25579393e-01 1.39684200e-01 -3.99664998e-01 6.03769660e-01
7.89781988e-01 1.42093718e+00 -3.48190427e-01 7.74650156e-01
-1.83392152e-01 5.86549044e-01 -1.10041641e-01 2.07805172e-01
4.13873255e-01 -3.84859949e-01 4.25220728e-01 1.29932237e+00
1.40833393e-01 4.45132554e-01 5.53610206e-01 2.29165852e-01
-1.58768222e-02 1.51705667e-01 -5.02875268e-01 -3.02785605e-01
1.12531148e-01 1.11401176e+00 -1.20249486e+00 -6.61165476e-01
-4.33463573e-01 1.21331465e+00 1.12720780e-01 2.33904243e-01
-1.17782974e+00 -8.07796195e-02 7.84613967e-01 -9.80449766e-02
2.38315165e-01 -2.42178306e-01 -5.61820418e-02 -1.40396547e+00
8.80752802e-02 -9.61314976e-01 8.21095467e-01 -5.70449889e-01
-1.08003342e+00 3.49509239e-01 2.01705411e-01 -1.62640953e+00
-4.37971681e-01 -3.40362102e-01 -3.43656152e-01 -3.08062434e-02
-9.99295592e-01 -1.07131684e+00 -3.13840747e-01 7.14507759e-01
1.10653520e+00 1.20017484e-01 5.95817745e-01 5.78719258e-01
-6.12894654e-01 1.86030447e-01 -1.79526627e-01 4.20834541e-01
6.06938303e-01 -1.15712738e+00 5.10478839e-02 8.14511955e-01
6.22678101e-01 1.60456836e-01 5.99880576e-01 -7.38040149e-01
-1.14464271e+00 -1.09466112e+00 9.85048354e-01 -6.10962272e-01
8.16862285e-01 -2.31617063e-01 -8.34538102e-01 8.84898424e-01
-1.74612943e-02 -9.99826863e-02 7.47765660e-01 1.95138920e-02
-2.50020444e-01 2.44622856e-01 -6.41738534e-01 5.50034344e-01
1.54641259e+00 -7.08940268e-01 -7.92991698e-01 6.39276445e-01
4.60574120e-01 -2.84764320e-01 -8.15555573e-01 1.75465465e-01
6.79109633e-01 -7.42890179e-01 1.03784907e+00 -1.06915855e+00
7.28770554e-01 -3.45266700e-01 -1.58656031e-01 -9.76881683e-01
-3.14356349e-02 -7.22210467e-01 4.99099605e-02 1.16423965e+00
1.05287902e-01 6.67067617e-02 7.65168548e-01 4.03181314e-01
-2.05072656e-01 -5.37975490e-01 -8.58339310e-01 -7.73240387e-01
-3.59913766e-01 -9.31871176e-01 1.12457193e-01 1.21479559e+00
5.16234994e-01 2.33688623e-01 -6.88501477e-01 -2.94829607e-01
3.08238298e-01 1.61017224e-01 8.19285929e-01 -9.45641279e-01
-1.22085191e-01 -4.82827365e-01 -7.02260196e-01 -1.20369947e+00
4.72826838e-01 -9.36986446e-01 2.97401935e-01 -1.39151406e+00
5.12935936e-01 3.48458961e-02 -2.51209170e-01 6.51895046e-01
-1.47913903e-01 4.50731963e-01 2.80014932e-01 3.91453564e-01
-1.35694408e+00 3.82910252e-01 1.04751623e+00 -1.68115243e-01
-2.09991604e-01 -1.91173628e-01 -2.06163391e-01 1.17603135e+00
4.53745246e-01 -4.30427462e-01 -4.95009452e-01 -3.10314536e-01
3.51611301e-02 5.02925366e-02 2.33638972e-01 -1.14916778e+00
7.05544427e-02 -4.14716601e-01 4.11197729e-02 -4.45645809e-01
3.18969846e-01 -6.87758923e-01 7.62936473e-02 3.47526520e-01
-6.71795368e-01 3.49698327e-02 -2.13806376e-01 6.85389161e-01
-4.86521125e-01 2.18608957e-02 6.20283246e-01 -2.31656358e-01
-1.28041196e+00 3.62137884e-01 -6.78326249e-01 3.46267939e-01
1.38560021e+00 -3.46644789e-01 6.65515661e-02 -4.40061569e-01
-1.12354565e+00 3.01651984e-01 2.56028056e-01 6.37065887e-01
4.54859525e-01 -1.48281181e+00 -5.01089692e-01 -2.19267562e-01
1.38811678e-01 -1.94801301e-01 5.23077846e-01 1.17435575e+00
-3.06448489e-01 1.64788112e-01 -3.36505771e-01 -6.93083942e-01
-1.54926407e+00 4.81947035e-01 3.75066213e-02 -3.45711261e-01
-8.11030567e-01 7.03766167e-01 3.50574076e-01 9.88120064e-02
2.68655866e-01 -5.14529824e-01 -6.45492196e-01 6.42515063e-01
6.16870046e-01 3.70905221e-01 -3.81563723e-01 -1.18247604e+00
-4.38638300e-01 5.61426222e-01 1.78610742e-01 -4.71884049e-02
1.29430878e+00 -2.94311970e-01 -2.76857633e-02 6.16101980e-01
1.33177161e+00 -5.17547846e-01 -1.29472888e+00 -1.82293415e-01
3.59417289e-01 -4.68876451e-01 -4.55392599e-01 -2.47899532e-01
-1.08561230e+00 5.93018234e-01 3.28219444e-01 1.38594463e-01
1.09980178e+00 3.04830730e-01 1.05094266e+00 3.70168388e-01
1.83741376e-01 -1.53403556e+00 4.82777297e-01 5.74278355e-01
5.22306800e-01 -1.31950772e+00 -4.10031751e-02 -3.41237396e-01
-1.06753683e+00 9.36253071e-01 4.30679381e-01 8.73481855e-02
4.87269700e-01 -1.18152112e-01 -1.09848514e-01 -2.85431415e-01
-5.59039235e-01 -6.33745611e-01 4.76556361e-01 4.25602585e-01
3.96765977e-01 8.42490643e-02 -4.78020668e-01 7.45030582e-01
-7.18354341e-03 8.65491033e-02 2.34893367e-01 1.11392939e+00
-3.49890828e-01 -9.25307631e-01 -2.14535370e-02 4.16053295e-01
-5.64585567e-01 2.34859392e-01 -3.47188741e-01 7.53186643e-01
3.61888170e-01 7.22057819e-01 2.69665748e-01 -5.67401290e-01
1.96636632e-01 2.91909128e-01 3.82472873e-01 -6.55804574e-01
-3.55299056e-01 9.42702591e-02 3.12182039e-01 -9.93507981e-01
-1.16297245e+00 -1.12608194e+00 -1.55619550e+00 3.30355801e-02
1.57570437e-01 1.00301601e-01 2.99150527e-01 1.35324931e+00
4.69343103e-02 4.14081216e-01 4.40138280e-01 -1.05146968e+00
7.38563091e-02 -7.43959844e-01 -4.37397748e-01 1.19090652e+00
-9.65860784e-02 -8.15705419e-01 -2.61317700e-01 9.86669600e-01]
|
[8.476055145263672, 0.5671378970146179]
|
2a7a2761-24a5-4b85-885e-5793257396ef
|
life-net-data-driven-modelling-of-time
|
2212.08403
| null |
https://arxiv.org/abs/2212.08403v1
|
https://arxiv.org/pdf/2212.08403v1.pdf
|
LiFe-net: Data-driven Modelling of Time-dependent Temperatures and Charging Statistics Of Tesla's LiFePo4 EV Battery
|
Modelling the temperature of Electric Vehicle (EV) batteries is a fundamental task of EV manufacturing. Extreme temperatures in the battery packs can affect their longevity and power output. Although theoretical models exist for describing heat transfer in battery packs, they are computationally expensive to simulate. Furthermore, it is difficult to acquire data measurements from within the battery cell. In this work, we propose a data-driven surrogate model (LiFe-net) that uses readily accessible driving diagnostics for battery temperature estimation to overcome these limitations. This model incorporates Neural Operators with a traditional numerical integration scheme to estimate the temperature evolution. Moreover, we propose two further variations of the baseline model: LiFe-net trained with a regulariser and LiFe-net trained with time stability loss. We compared these models in terms of generalization error on test data. The results showed that LiFe-net trained with time stability loss outperforms the other two models and can estimate the temperature evolution on unseen data with a relative error of 2.77 % on average.
|
['Nico Hoffmann', 'Luisa Fennert', 'Jeyhun Rustamov']
|
2022-12-16
| null | null | null | null |
['numerical-integration']
|
['miscellaneous']
|
[-3.23180825e-01 -1.61478221e-01 -1.81354843e-02 -4.58248496e-01
-5.54059267e-01 -4.03875172e-01 2.82430053e-01 -1.43002033e-01
-3.78246099e-01 1.02022552e+00 -7.92689145e-01 -4.12935615e-01
4.42655087e-02 -6.97886825e-01 -9.09476995e-01 -1.03473723e+00
-2.27443874e-02 5.47559679e-01 7.14930072e-02 -1.43546179e-01
-6.89278916e-02 6.91230297e-01 -1.90031981e+00 -3.47082824e-01
1.28360665e+00 1.51556969e+00 3.90022136e-02 4.30174202e-01
4.94515628e-01 4.53166723e-01 -5.40391088e-01 -7.13188052e-02
-9.41728204e-02 1.47182448e-02 -2.31833994e-01 -7.34365463e-01
-2.66453058e-01 -2.01629654e-01 -4.58671063e-01 8.43996406e-01
6.08729422e-01 2.96691179e-01 8.08751941e-01 -1.84646833e+00
-3.59840170e-02 1.97731704e-01 2.32233316e-01 2.75881998e-02
-2.84829259e-01 1.84168652e-01 1.36180922e-01 -6.86115921e-01
-4.66618426e-02 6.62914038e-01 1.00732291e+00 8.90140235e-01
-1.03597927e+00 -7.72542179e-01 -4.38099414e-01 4.15193409e-01
-1.61139870e+00 -5.91773927e-01 1.05738616e+00 -1.75715476e-01
1.28640747e+00 6.71870768e-01 9.45977509e-01 9.84692633e-01
8.25962663e-01 5.49182236e-01 1.28944099e+00 3.41968313e-02
6.89798236e-01 5.58126926e-01 1.05711199e-01 2.47483134e-01
2.21353099e-01 5.18943012e-01 -1.54319957e-01 -1.70323879e-01
-5.87593764e-02 -3.77122641e-01 -1.38946593e-01 -3.31279576e-01
-2.42962465e-01 4.31339383e-01 2.17082828e-01 1.32455453e-01
-6.61661383e-03 5.44255376e-01 6.10331237e-01 1.36179745e-01
6.79734707e-01 3.31145495e-01 -5.28954744e-01 -4.86971736e-01
-1.07057405e+00 2.21281245e-01 8.99736702e-01 9.57447588e-01
5.55023193e-01 3.81025970e-01 9.59148034e-02 6.15807056e-01
4.69438165e-01 7.76523113e-01 5.18647671e-01 -6.10226989e-01
3.11256181e-02 3.74232799e-01 1.79852486e-01 -1.48875996e-01
-7.22225428e-01 -2.70051181e-01 -8.93002510e-01 2.55985469e-01
-8.31352845e-02 -2.25221455e-01 -1.24008632e+00 1.46400547e+00
6.04945421e-02 1.31805822e-01 2.08514974e-01 7.06276655e-01
7.34679043e-01 8.67051482e-01 1.42856181e-01 -3.62298995e-01
8.10872197e-01 -8.62429798e-01 -1.06881952e+00 -1.30956009e-01
7.83683896e-01 -1.86861902e-02 7.59632349e-01 3.92731011e-01
-1.30142105e+00 -3.53494257e-01 -1.69085479e+00 3.18490006e-02
-9.01699543e-01 -6.25887439e-02 3.29999655e-01 9.42744493e-01
-1.31634426e+00 1.00280654e+00 -1.03646398e+00 -5.76055348e-02
1.76083431e-01 5.95434487e-01 2.37878859e-01 4.48308378e-01
-1.60017848e+00 1.47202492e+00 3.77017647e-01 5.13355315e-01
-1.27962077e+00 -8.49717021e-01 -8.84882629e-01 -2.04264745e-01
-2.97949165e-02 -2.92665005e-01 1.38486862e+00 -4.98199552e-01
-1.77148950e+00 2.80853868e-01 -2.23960802e-01 -5.43585420e-01
6.49449766e-01 -8.90713781e-02 -8.07870150e-01 -3.95957649e-01
-5.93291223e-01 3.40600997e-01 7.77263045e-01 -1.33469915e+00
-1.92620769e-01 -1.48548171e-01 -5.56006968e-01 -2.17659786e-01
-3.57760578e-01 -4.70343977e-01 -1.06072694e-01 -1.82817444e-01
-3.79371613e-01 -1.02943182e+00 1.01871230e-02 -1.77400649e-01
-1.21333353e-01 -5.66517830e-01 1.32593894e+00 -8.86893511e-01
1.05386758e+00 -1.94749808e+00 -1.11890100e-01 2.22202674e-01
-1.97286427e-01 1.93758309e-01 4.20643896e-01 1.37264505e-01
2.10783370e-02 7.66990259e-02 -4.56358433e-01 -4.77984935e-01
2.83146203e-01 3.96460325e-01 -1.96745947e-01 6.75676942e-01
2.25698322e-01 1.19646823e+00 -6.83244824e-01 -1.41398594e-01
3.20447564e-01 8.26515317e-01 3.29203531e-02 2.15361208e-01
-1.73408732e-01 1.14157684e-01 -1.31142497e-01 6.44964576e-01
8.91237199e-01 3.49198759e-01 -1.23329930e-01 -3.86664689e-01
-1.08225636e-01 9.44949389e-02 -5.19074738e-01 1.32423329e+00
-6.48015261e-01 7.62487471e-01 1.76175430e-01 -1.01280558e+00
1.14677966e+00 3.38675290e-01 3.34469467e-01 -1.10621369e+00
5.44743657e-01 5.52129924e-01 -3.21747214e-01 -4.16116506e-01
7.68617094e-01 -4.27935481e-01 -1.36132807e-01 -8.76973718e-02
-8.10785443e-02 -6.23002291e-01 -1.39123827e-01 -4.51509088e-01
7.70972490e-01 1.74518362e-01 -6.35855436e-01 -6.44282699e-01
5.97918212e-01 -6.96710497e-02 6.77146494e-01 1.12370789e-01
-3.24239731e-01 1.32520914e-01 3.29249054e-01 -2.49755219e-01
-1.24168611e+00 -8.69973123e-01 -5.69795430e-01 5.07933259e-01
6.09016955e-01 -1.28454432e-01 -8.58483553e-01 -4.12476093e-01
2.30972409e-01 1.39490080e+00 -4.86122936e-01 -8.64344954e-01
-5.59148908e-01 -9.51332331e-01 5.47685623e-01 1.02398252e+00
2.90910482e-01 -6.22353792e-01 -5.95626354e-01 8.67304858e-03
1.68924868e-01 -8.64501774e-01 6.63608359e-03 8.08507383e-01
-9.99221265e-01 -7.12881088e-01 -3.63299400e-01 -5.28499544e-01
6.85001194e-01 -6.95465744e-01 1.12069905e+00 -3.34163196e-02
-2.58182734e-01 8.92410651e-02 9.27874073e-02 -9.13819551e-01
-7.19745398e-01 1.79364935e-01 3.37656200e-01 -4.22571599e-01
3.56862277e-01 -2.81649232e-01 -5.22579908e-01 7.11023331e-01
-6.92969918e-01 -4.54739593e-02 2.02936620e-01 5.86295485e-01
4.95579302e-01 5.43510258e-01 8.28072071e-01 -9.81660560e-02
5.85192382e-01 -5.51879406e-01 -9.11571801e-01 1.70196414e-01
-1.26780987e+00 1.97505549e-01 7.54183173e-01 -4.18183118e-01
-9.23731804e-01 -6.75094351e-02 -4.94976431e-01 -6.17483199e-01
2.51769423e-01 -2.04473268e-02 -4.22428876e-01 -1.93262115e-01
6.94712400e-02 2.12136850e-01 2.99459696e-01 -3.20054382e-01
-2.12594748e-01 7.49677062e-01 6.39388502e-01 -4.80642378e-01
6.76749110e-01 1.93793580e-01 3.23089749e-01 -5.56993306e-01
-9.98940915e-02 8.15200806e-02 -1.62999779e-01 -5.69893539e-01
6.31007493e-01 -8.88483584e-01 -1.00182724e+00 8.36895823e-01
-7.11950362e-01 -7.57586300e-01 -3.30970854e-01 1.73392296e-01
-5.54709733e-01 -5.19059300e-02 -4.34986830e-01 -1.20313990e+00
-8.77370656e-01 -1.26640964e+00 9.06630635e-01 3.36215764e-01
-1.29616400e-02 -1.28150451e+00 -2.54631072e-01 1.36703029e-01
8.73804331e-01 4.64775264e-01 7.07254052e-01 -3.17990005e-01
-1.11903444e-01 -4.67880577e-01 3.79141957e-01 8.78682554e-01
-2.59333163e-01 -9.59539860e-02 -1.49752510e+00 -7.66820967e-01
4.28243637e-01 -2.49055058e-01 6.62797749e-01 3.29198837e-01
1.60152054e+00 1.11266881e-01 -7.68388867e-01 6.12681508e-01
1.49379170e+00 7.12614357e-01 8.56787324e-01 1.89545900e-01
4.85837370e-01 2.37033442e-01 6.09135509e-01 6.30429909e-02
2.37094745e-01 5.48259020e-01 6.93747163e-01 -5.52592203e-02
5.61396718e-01 -2.02802736e-02 5.66354334e-01 9.97370481e-01
1.05013803e-01 -3.96426857e-01 -8.19241524e-01 5.67193568e-01
-1.55206203e+00 -2.91252702e-01 -2.18818828e-01 2.34914422e+00
6.19604647e-01 4.28176284e-01 -1.15816467e-01 5.23590565e-01
3.99115831e-01 -2.70223886e-01 -1.05929446e+00 -1.04500926e+00
-2.85733994e-02 2.16751453e-02 1.02608907e+00 3.06278616e-01
-6.86125040e-01 1.68557122e-01 6.79838896e+00 9.71076846e-01
-1.29820228e+00 2.65737414e-01 8.74062777e-01 -3.69965345e-01
-4.98693138e-01 -2.08477855e-01 -6.65212691e-01 8.61220002e-01
1.80379748e+00 -1.12141319e-01 3.93788844e-01 8.36174667e-01
1.12047739e-01 -2.73699701e-01 -1.45837891e+00 8.26422870e-01
-2.65396703e-02 -7.74666607e-01 -7.21359313e-01 -8.32866281e-02
4.68668461e-01 1.27829254e-01 -1.11722820e-01 6.48044109e-01
-3.16308320e-01 -1.28644729e+00 8.92994285e-01 9.91035283e-01
1.14635444e+00 -1.24854231e+00 1.10035026e+00 4.49535996e-01
-1.02104509e+00 -1.30989775e-01 -2.51504660e-01 6.43822104e-02
3.90185148e-01 6.28126144e-01 -7.29552329e-01 4.97903973e-01
9.81549442e-01 2.67095715e-01 -4.89994377e-01 8.31682861e-01
1.95017710e-01 7.94072270e-01 -7.67015159e-01 -5.54816842e-01
-1.95085287e-01 -2.92256325e-01 2.69963652e-01 8.34508777e-01
6.32483900e-01 -5.34527481e-01 -4.01217937e-01 1.01022720e+00
1.62989162e-02 -5.72838306e-01 -4.34675276e-01 6.68203160e-02
5.08722663e-01 1.42447913e+00 -3.23971391e-01 -1.47518843e-01
4.19091657e-02 7.29877234e-01 -1.67525783e-01 3.28443497e-01
-1.29593742e+00 -6.39552772e-01 6.95148885e-01 3.86141807e-01
-3.89431976e-02 9.17069763e-02 -3.72720093e-01 -4.59082425e-01
3.06412041e-01 -1.44896936e-03 -2.72180319e-01 -9.49253798e-01
-1.00587273e+00 4.95788336e-01 3.13857853e-01 -8.60633135e-01
-2.60124356e-01 -8.85646701e-01 -7.94766486e-01 9.01354909e-01
-1.70976102e+00 -7.17820942e-01 -4.38171536e-01 -2.31261309e-02
2.29974627e-01 1.04198225e-01 6.08339906e-01 4.89066988e-01
-9.05648172e-01 8.14989865e-01 7.36059070e-01 -4.71166730e-01
2.44691208e-01 -1.25301659e+00 4.22135621e-01 2.51205385e-01
-1.07990074e+00 1.05429448e-01 1.01643741e+00 -4.81698722e-01
-2.05145645e+00 -1.16090739e+00 5.83442330e-01 -3.81131679e-01
4.11375254e-01 -8.34278166e-01 -1.14490855e+00 2.71386623e-01
3.26565385e-01 1.38599679e-01 1.65624365e-01 -4.91975695e-01
5.00692248e-01 -7.49477625e-01 -1.48162329e+00 4.40770425e-02
5.78096211e-01 -5.59383810e-01 -2.03935891e-01 1.26994222e-01
4.25777942e-01 -5.82833946e-01 -1.28365278e+00 9.61083949e-01
5.91175914e-01 -5.14135718e-01 6.19496107e-01 1.63368568e-01
-2.22591057e-01 -3.18856359e-01 2.83272892e-01 -1.32103252e+00
1.92765951e-01 -7.14987516e-01 -7.22372055e-01 1.34855890e+00
4.35316920e-01 -1.09480715e+00 6.99130714e-01 1.05891323e+00
-4.85757321e-01 -1.32973266e+00 -1.43755043e+00 -1.19967675e+00
5.14415085e-01 -8.10546279e-01 1.01138091e+00 2.14733765e-01
5.60011789e-02 -5.98406494e-02 -6.61129579e-02 6.75866157e-02
4.57006514e-01 -6.16229236e-01 2.48003960e-01 -1.32541180e+00
3.05197954e-01 -1.65562496e-01 -3.06749552e-01 -4.03044105e-01
5.92009366e-01 -7.83215225e-01 5.75195670e-01 -1.38693583e+00
-6.49287784e-03 -7.04923570e-01 -5.84490120e-01 1.04059264e-01
1.78156629e-01 8.77857879e-02 -2.97531009e-01 -2.96894431e-01
-1.50310844e-01 9.69809592e-01 7.88571060e-01 -3.36605161e-01
2.51645930e-02 -1.25042424e-01 1.47777461e-02 4.75540847e-01
1.10708177e+00 -3.99150759e-01 -6.45903111e-01 5.67739680e-02
5.03415823e-01 -1.34268239e-01 6.20390952e-01 -1.42844069e+00
2.51968682e-01 1.58160761e-01 4.97220784e-01 -1.02330029e+00
5.01056910e-01 -1.10505056e+00 7.48252451e-01 4.55943823e-01
1.69826150e-01 2.17090622e-01 7.63565302e-01 2.81710804e-01
1.13953091e-01 -3.86584461e-01 6.98298752e-01 5.11170626e-01
-4.81156617e-01 7.03539476e-02 -5.65461099e-01 -3.22227210e-01
1.45400441e+00 -4.13889050e-01 -2.71261960e-01 -9.41490084e-02
-2.57603288e-01 6.47966981e-01 6.63413763e-01 5.71096599e-01
5.28356194e-01 -1.54510617e+00 -1.88587695e-01 3.42220008e-01
-9.48789064e-03 2.93893993e-01 1.20676965e-01 8.95522833e-01
-3.67366970e-01 3.66720170e-01 2.54667222e-01 -7.40009069e-01
-9.94795918e-01 6.46423876e-01 9.92443502e-01 -2.74978839e-02
-2.42708683e-01 5.80236912e-01 -3.91406953e-01 -4.37368751e-01
1.57739252e-01 -6.28612041e-01 1.92027569e-01 -1.26108333e-01
3.37711014e-02 7.83890784e-01 6.33806586e-01 -5.14416277e-01
-5.45999527e-01 4.84508753e-01 3.44983160e-01 1.00144416e-01
1.10091567e+00 1.27865925e-01 -2.77504385e-01 9.39015269e-01
1.52751172e+00 -7.62607098e-01 -1.28000510e+00 4.51518387e-01
-8.44265744e-02 2.06832379e-01 5.27196169e-01 -9.74634409e-01
-1.10734403e+00 7.19450235e-01 1.15706527e+00 2.46302724e-01
1.51927602e+00 -2.13401005e-01 1.01659548e+00 1.77002743e-01
2.85424978e-01 -1.81951940e+00 -5.44542611e-01 3.08763832e-01
7.86547363e-01 -8.95596206e-01 -1.20815933e-01 4.25987430e-02
-1.22244783e-01 1.04787874e+00 8.14538300e-01 1.90725639e-01
7.84480929e-01 7.82324553e-01 -1.19958006e-01 -6.22118153e-02
-8.06123674e-01 5.61173499e-01 3.24035197e-01 3.39311242e-01
1.96858775e-03 3.66582796e-02 -2.87308484e-01 6.21496439e-01
-1.37032270e-01 1.72139511e-01 5.32898679e-02 1.19124496e+00
-6.64344206e-02 -6.69885576e-01 -3.83609712e-01 5.89244723e-01
-2.42942467e-01 3.60947162e-01 1.51068121e-01 6.84176862e-01
1.47949502e-01 9.50539231e-01 3.44273239e-01 -6.66556418e-01
4.84672844e-01 5.38934946e-01 2.65542954e-01 3.99261147e-01
-4.01771098e-01 -4.03878897e-01 2.08954915e-01 -4.10182059e-01
-6.60062060e-02 -5.10862052e-01 -1.53338742e+00 -3.14942002e-01
-7.19329774e-01 3.65611017e-01 1.44002569e+00 9.65898275e-01
2.93146670e-01 9.50062215e-01 8.14595699e-01 -1.05186915e+00
-6.24011695e-01 -1.02471471e+00 -8.09430420e-01 5.35332737e-03
5.94602346e-01 -7.71939695e-01 -7.56100118e-01 -4.77685124e-01]
|
[6.310601711273193, 2.774564504623413]
|
afd38a9d-70b5-4c40-ad0f-f70a8b28f82f
|
feature-imitating-networks-enhance-the
|
2306.14572
| null |
https://arxiv.org/abs/2306.14572v1
|
https://arxiv.org/pdf/2306.14572v1.pdf
|
Feature Imitating Networks Enhance The Performance, Reliability And Speed Of Deep Learning On Biomedical Image Processing Tasks
|
Feature-Imitating-Networks (FINs) are neural networks with weights that are initialized to approximate closed-form statistical features. In this work, we perform the first-ever evaluation of FINs for biomedical image processing tasks. We begin by training a set of FINs to imitate six common radiomics features, and then compare the performance of networks with and without the FINs for three experimental tasks: COVID-19 detection from CT scans, brain tumor classification from MRI scans, and brain-tumor segmentation from MRI scans; we find that FINs provide best-in-class performance for all three tasks, while converging faster and more consistently when compared to networks with similar or greater representational power. The results of our experiments provide evidence that FINs may provide state-of-the-art performance for a variety of other biomedical image processing tasks.
|
['Tuka Alhanai', 'Mohammad Mahdi Ghassemi', 'Shangyang Min']
|
2023-06-26
| null | null | null | null |
['tumor-segmentation', 'brain-tumor-segmentation']
|
['computer-vision', 'medical']
|
[ 4.32355076e-01 -1.62547529e-02 -2.50181518e-02 -6.62035823e-01
-8.21904659e-01 -7.20516890e-02 5.97778440e-01 7.74348229e-02
-1.03732979e+00 4.20253903e-01 1.48189843e-01 -3.61102462e-01
-1.70484513e-01 -2.80354798e-01 -5.50218403e-01 -5.67756414e-01
-5.50233185e-01 6.06922448e-01 2.19150439e-01 1.43776551e-01
-9.22445133e-02 7.87509620e-01 -1.07214451e+00 4.64405358e-01
3.24733526e-01 1.18313265e+00 4.44269553e-02 6.64683640e-01
3.68892550e-01 6.77501857e-01 -4.53561872e-01 -1.28745481e-01
6.13632537e-02 -7.46565908e-02 -8.61458063e-01 -2.26282328e-01
3.88131857e-01 -2.67972440e-01 -4.79191631e-01 9.19891000e-01
6.11452818e-01 8.18469152e-02 1.21503329e+00 -1.02515280e+00
-4.45467651e-01 5.06045997e-01 -5.91355920e-01 8.22381139e-01
-2.36879177e-02 2.09835038e-01 7.48569787e-01 -9.12079215e-01
5.63846111e-01 1.13891101e+00 1.07573569e+00 7.04643667e-01
-1.21272397e+00 -7.24624455e-01 -3.59499127e-01 -1.16075747e-01
-1.31090426e+00 -6.23267233e-01 1.61077291e-01 -3.04221243e-01
1.09241402e+00 2.03835696e-01 3.85155678e-01 1.07408094e+00
7.98909307e-01 8.99601161e-01 9.63368952e-01 -1.90629512e-01
7.85951316e-03 -2.15233475e-01 4.59463149e-01 1.02942050e+00
2.38047525e-01 2.03492329e-01 -2.58347988e-01 -3.37135494e-01
8.35142970e-01 -2.14449912e-02 -1.12863310e-01 3.58335488e-02
-1.58117115e+00 8.97029877e-01 6.72977090e-01 5.89637280e-01
-5.78681767e-01 4.07623738e-01 5.70453823e-01 1.53834641e-01
6.03596210e-01 7.03638375e-01 -3.76818895e-01 1.39616519e-01
-1.14381087e+00 5.53564653e-02 5.03992736e-01 4.54116374e-01
1.99291036e-01 2.57749259e-01 -3.41948897e-01 9.31762695e-01
7.46577382e-02 2.77474463e-01 1.10707104e+00 -8.60780120e-01
-1.64880883e-02 1.04812972e-01 -4.33367401e-01 -6.42615676e-01
-1.23222589e+00 -5.34011245e-01 -1.11994350e+00 1.56069666e-01
5.85964978e-01 -3.86738360e-01 -1.39893138e+00 1.70514750e+00
-1.67718872e-01 5.51702790e-02 -1.21338569e-01 7.08496094e-01
1.13403249e+00 1.90337256e-01 2.25338265e-01 7.55697563e-02
1.60393977e+00 -7.45423436e-01 -5.03329813e-01 -2.09293023e-01
5.82991958e-01 -4.59544688e-01 7.57119119e-01 2.50919044e-01
-9.17006671e-01 -4.13777381e-01 -9.43598986e-01 1.08098887e-01
-2.20909953e-01 5.02844714e-02 9.00950313e-01 6.82689071e-01
-1.30265868e+00 7.72003889e-01 -1.22967958e+00 -4.20552552e-01
8.42637002e-01 5.47967792e-01 -4.87842321e-01 5.06836222e-03
-8.69773984e-01 9.62497532e-01 1.81211621e-01 4.56034541e-02
-1.04186308e+00 -9.79279935e-01 -8.14446330e-01 -1.13558643e-01
-1.76521361e-01 -8.40481102e-01 1.43125367e+00 -9.59771395e-01
-1.12336612e+00 9.86811101e-01 -8.17750767e-02 -6.25940442e-01
4.36685532e-01 9.61280391e-02 -4.87400740e-01 3.30158532e-01
7.38446340e-02 1.08976471e+00 7.08113551e-01 -8.84046793e-01
-7.30300173e-02 -3.78423542e-01 -5.85314035e-01 2.16992535e-02
-3.76209468e-01 3.76373947e-01 -2.11695075e-01 -7.88692296e-01
5.66294640e-02 -8.62317085e-01 -5.98545730e-01 1.56064972e-01
-6.89511836e-01 -1.79651514e-01 5.25361001e-01 -5.23474276e-01
5.86065650e-01 -2.13003874e+00 -2.01520711e-01 3.49913627e-01
5.64461470e-01 2.16259912e-01 -4.59845275e-01 -2.13458791e-01
-5.08107662e-01 1.76959231e-01 -1.83103859e-01 -3.91623348e-01
-2.06859782e-01 1.67164639e-01 4.08485800e-01 6.62040710e-01
2.95917302e-01 1.17761934e+00 -7.00652182e-01 -6.44662321e-01
-2.43300796e-02 4.55307603e-01 -3.57932240e-01 -1.29550576e-01
1.73893824e-01 3.32455993e-01 -5.00265956e-01 5.95925212e-01
1.80935085e-01 -4.77735460e-01 -8.63726437e-02 -3.78036141e-01
3.38810384e-01 -2.61458784e-01 -4.88489866e-01 1.52012885e+00
-2.16881454e-01 7.51315594e-01 -1.68939512e-02 -9.70131993e-01
3.93553555e-01 3.25019419e-01 7.79789150e-01 -5.58025718e-01
5.89188397e-01 8.39711949e-02 5.59558213e-01 -4.24290240e-01
2.40607232e-01 -4.49828655e-01 1.31685093e-01 6.74557090e-01
3.93270344e-01 -1.52844191e-01 2.10187986e-01 -8.11764598e-03
1.46302176e+00 -5.44202864e-01 2.48579055e-01 -5.61234176e-01
2.25094035e-01 -1.99259341e-01 1.67486981e-01 1.05006814e+00
-4.46472615e-01 7.85254598e-01 3.85624737e-01 -7.19480217e-01
-8.29695165e-01 -1.20554268e+00 -5.44017434e-01 1.10064030e+00
-5.24616361e-01 -8.43720883e-03 -8.36365819e-01 -7.65228331e-01
1.26088947e-01 5.32452047e-01 -1.07881069e+00 -1.10076386e-02
-3.42933655e-01 -1.30583560e+00 9.26332414e-01 8.63143921e-01
1.62342981e-01 -1.25394166e+00 -6.73568487e-01 1.51785865e-01
2.24034667e-01 -1.18113542e+00 -5.34040928e-01 6.87664509e-01
-1.03300846e+00 -1.19288850e+00 -1.13719261e+00 -8.63157511e-01
9.03977394e-01 -1.35718271e-01 1.11555159e+00 3.06912452e-01
-7.51064122e-01 3.44368756e-01 6.73301741e-02 -7.16404498e-01
-4.42104131e-01 1.71083510e-01 2.24194765e-01 -3.51615071e-01
2.02906489e-01 -4.33337450e-01 -5.77394307e-01 2.89299756e-01
-1.13366318e+00 -3.76113057e-02 7.49216139e-01 1.12723875e+00
5.05135953e-01 -1.87939495e-01 4.21842486e-01 -1.04285073e+00
9.09957290e-01 -3.52691740e-01 -4.86198217e-02 5.10215797e-02
-2.82676041e-01 5.81177957e-02 6.45891607e-01 -3.97730798e-01
-6.14988387e-01 1.34516507e-01 -3.94043505e-01 -4.95738357e-01
-2.61666596e-01 6.02732301e-01 6.65147185e-01 -2.56559014e-01
1.14189458e+00 3.39961350e-02 3.67025524e-01 -1.29787073e-01
2.01300398e-01 3.25999916e-01 7.32798219e-01 -5.05182207e-01
5.26031017e-01 4.99273241e-01 1.68712705e-01 -9.06475842e-01
-8.57228458e-01 -3.85906667e-01 -5.48188150e-01 -8.53054225e-02
9.59066153e-01 -5.74839413e-01 -6.20696902e-01 6.50033474e-01
-8.79219115e-01 -4.71074641e-01 -3.83903146e-01 5.97997725e-01
-5.63216627e-01 3.63867618e-02 -1.04786396e+00 -2.30595484e-01
-7.28430927e-01 -1.39636052e+00 9.23410654e-01 2.45998472e-01
-3.71371984e-01 -1.04187393e+00 9.91268829e-03 -2.82953419e-02
6.93746150e-01 3.36855501e-01 1.04456139e+00 -9.34096873e-01
2.12467179e-01 -2.39829183e-01 -5.15499949e-01 3.15252155e-01
1.98994026e-01 -1.89752042e-01 -1.16075337e+00 -2.99797565e-01
-2.46880934e-01 -5.80314815e-01 1.20243895e+00 9.73229051e-01
1.58816028e+00 2.66849875e-01 -4.76929545e-01 8.77519369e-01
1.05822873e+00 1.34815201e-01 4.59222674e-01 9.60604995e-02
3.98325890e-01 2.40453079e-01 -1.41214266e-01 1.98776990e-01
1.56574305e-02 2.26179019e-01 1.35509431e-01 -6.17103755e-01
-2.77133793e-01 3.48621458e-01 -4.71591130e-02 6.17585838e-01
-4.06924449e-02 -1.17264554e-01 -1.14489758e+00 5.32123089e-01
-1.47599518e+00 -6.56310320e-01 1.63734078e-01 1.59831750e+00
8.11347723e-01 1.72526434e-01 1.86477333e-01 -2.44310215e-01
5.07168770e-01 1.19625755e-01 -6.78553402e-01 -2.50148684e-01
1.67077512e-01 5.75995684e-01 6.20912552e-01 -4.57759574e-03
-1.25963962e+00 5.69136679e-01 8.31141376e+00 6.61178529e-01
-1.21881890e+00 2.29659766e-01 1.20586884e+00 -8.73234197e-02
5.19444533e-02 -8.95132244e-01 -2.73772806e-01 2.37216018e-02
1.17204487e+00 6.45692870e-02 2.00368866e-01 5.75823784e-01
5.99111430e-03 -9.58782658e-02 -1.38112378e+00 8.10830176e-01
1.49739191e-01 -1.50836742e+00 -1.22167610e-01 -9.42755044e-02
6.72286332e-01 7.41674304e-01 3.14396709e-01 1.19837612e-01
5.75382173e-01 -1.48421907e+00 4.89066064e-01 4.95467544e-01
1.11564493e+00 -6.15222752e-01 1.02151978e+00 -1.98406894e-02
-7.23054349e-01 1.67415664e-01 -2.92140841e-01 4.09162343e-01
1.24579072e-02 6.69751585e-01 -1.15119052e+00 3.05772781e-01
7.50310659e-01 5.39876819e-01 -6.85149014e-01 1.23676097e+00
1.79442167e-01 6.96506321e-01 -4.24828827e-01 -2.15770483e-01
3.92228216e-01 4.92331058e-01 1.95265964e-01 1.60937285e+00
4.97437604e-02 7.67614320e-02 1.11219458e-01 7.18077838e-01
-3.44935507e-01 -2.13994016e-03 -5.06062806e-01 -1.05949819e-01
-4.01348136e-02 1.44483471e+00 -1.30492175e+00 -3.64420652e-01
-1.24440700e-01 5.16755819e-01 1.70543939e-01 2.66162664e-01
-6.45499170e-01 -4.64075655e-01 4.97588336e-01 -4.28471155e-02
7.25053251e-02 -8.50675255e-03 -4.24131691e-01 -8.27073336e-01
-5.79573095e-01 -8.08259606e-01 4.61222976e-01 -8.75847518e-01
-1.56816792e+00 1.01352596e+00 6.35540634e-02 -7.04112828e-01
-3.73526126e-01 -1.09134841e+00 -7.71248281e-01 6.68919086e-01
-1.29192781e+00 -8.12011123e-01 -2.03694001e-01 6.82888627e-01
2.48146564e-01 -3.82885396e-01 1.05321705e+00 4.12817448e-02
-5.89941740e-01 7.04084575e-01 1.67033076e-01 5.24628639e-01
5.73900163e-01 -1.09974563e+00 5.38812876e-01 4.67722267e-01
1.77166879e-01 6.87342048e-01 5.51546931e-01 -3.69805068e-01
-9.77830112e-01 -1.00741696e+00 2.99972355e-01 -2.59628743e-01
8.29779923e-01 -1.08872026e-01 -6.16824508e-01 7.87712932e-01
2.23899975e-01 4.32215929e-01 7.72090197e-01 2.43979797e-01
-3.63661945e-01 1.03103094e-01 -1.41362989e+00 4.13901269e-01
8.40344846e-01 -3.50363642e-01 -5.93674064e-01 7.69401312e-01
3.00569683e-01 -5.18517613e-01 -1.06195533e+00 4.43422467e-01
8.55428755e-01 -5.95932543e-01 1.06945777e+00 -8.77465487e-01
6.37160778e-01 2.22775400e-01 4.70870640e-03 -1.62576830e+00
-5.14204204e-01 -1.67445585e-01 5.62129021e-01 4.55922157e-01
6.99686170e-01 -6.23834729e-01 9.43735540e-01 4.46483582e-01
-3.33450049e-01 -9.30644691e-01 -1.03335738e+00 -7.28740871e-01
4.10997957e-01 -6.14001989e-01 2.21245199e-01 8.59992862e-01
-5.39379679e-02 1.68603495e-01 1.66893765e-01 -3.22496384e-01
6.55718148e-01 -4.18476462e-01 1.93597257e-01 -1.23989105e+00
-8.05535913e-02 -7.14032948e-01 -6.10105991e-01 -3.36742818e-01
4.53123391e-01 -1.26276278e+00 2.88051635e-01 -1.32161009e+00
5.72876930e-01 -2.92363405e-01 -6.48828566e-01 9.19571698e-01
1.60623412e-03 6.51115060e-01 1.23336129e-01 -1.12217844e-01
-5.05527496e-01 1.56127885e-01 1.30073905e+00 -2.97583073e-01
1.85584724e-01 -8.21948424e-02 -7.93523848e-01 9.38811898e-01
6.86087191e-01 -6.21553838e-01 -2.68098265e-01 -4.32838649e-01
-5.89092195e-01 9.33127478e-02 3.13022435e-01 -1.22108126e+00
2.41298541e-01 1.05387911e-01 1.02875316e+00 -3.81622910e-01
2.20499620e-01 -4.40021098e-01 -2.03136817e-01 6.97478712e-01
-4.47479993e-01 3.94997597e-01 4.85324264e-01 1.70080170e-01
4.07501310e-03 -2.68099368e-01 1.11910963e+00 -2.74824947e-01
-4.13098752e-01 3.74662340e-01 -8.57634425e-01 1.51474506e-01
6.90961421e-01 -7.87691250e-02 -4.13716644e-01 -2.94813275e-01
-8.04970026e-01 1.99099351e-02 -4.10238169e-02 2.14337751e-01
7.57866561e-01 -1.18628883e+00 -8.08745801e-01 3.25466841e-01
-2.93524079e-02 -2.70919919e-01 1.11496709e-01 9.73677516e-01
-5.57553649e-01 4.42322224e-01 -4.39017326e-01 -8.47634256e-01
-1.08628750e+00 1.86125681e-01 7.22312570e-01 -5.64414144e-01
-6.69172466e-01 1.00219345e+00 3.90432745e-01 -4.55010295e-01
2.44496942e-01 -6.55153692e-01 -1.60336167e-01 -2.14682057e-01
6.05349362e-01 -1.32694066e-01 3.13930631e-01 -5.17722189e-01
-5.75198710e-01 1.87090695e-01 -4.69855607e-01 -2.55444467e-01
1.62659419e+00 7.08528280e-01 8.11415464e-02 2.04605907e-01
1.42935002e+00 -6.91747904e-01 -8.20014834e-01 -2.23998502e-01
-2.21682843e-02 7.97722861e-02 4.19943541e-01 -1.04256487e+00
-1.50613976e+00 6.44916117e-01 8.82894993e-01 -1.99328795e-01
1.01083946e+00 3.43478248e-02 6.18689358e-01 6.66521192e-01
2.37644091e-01 -7.39684880e-01 2.73167104e-01 5.79851747e-01
7.45424092e-01 -1.05507815e+00 1.54194877e-01 7.10970722e-03
-4.80519235e-01 1.13742077e+00 4.28951502e-01 -3.19293052e-01
9.05499697e-01 6.81216180e-01 1.22756980e-01 -5.79872727e-01
-6.31254435e-01 6.82858899e-02 5.86714625e-01 6.77813172e-01
6.09006941e-01 1.07180767e-01 -1.60535257e-02 5.83409786e-01
-1.17723189e-01 1.76118627e-01 3.46075207e-01 8.29228163e-01
-3.64292264e-01 -6.65991247e-01 -3.12669188e-01 1.30661380e+00
-8.15875232e-01 -3.47882003e-01 -1.73011974e-01 1.01839864e+00
-3.61289829e-01 5.28304815e-01 1.55116767e-01 -3.83760899e-01
2.58618027e-01 1.47481933e-01 6.42539501e-01 -5.92201948e-01
-8.85933936e-01 -6.91097453e-02 1.40355557e-01 -6.90916300e-01
-2.77467728e-01 -8.68435919e-01 -1.29585314e+00 -6.83527812e-02
-2.69566119e-01 -6.48819506e-02 6.95370495e-01 1.05507886e+00
-9.09979865e-02 9.93177474e-01 2.32101068e-01 -9.89572585e-01
-5.93125105e-01 -1.12174284e+00 -6.96051180e-01 2.83146113e-01
3.48155439e-01 -6.86933577e-01 -9.35267210e-02 -2.20887780e-01]
|
[14.651325225830078, -2.2742044925689697]
|
ddf0af5d-0f49-4a5f-90a9-7ae290404a71
|
matching-web-tables-with-knowledge-base
| null | null |
https://link.springer.com/chapter/10.1007/978-3-319-68288-4_16
|
https://iswc2017.ai.wu.ac.at/wp-content/uploads/papers/MainProceedings/98.pdf
|
Matching Web Tables with Knowledge Base Entities: From Entity Lookups to Entity Embeddings
|
Web tables constitute valuable sources of information for various applications, ranging from Web search to Knowledge Base (KB) augmentation. An underlying common requirement is to annotate the rows of Web tables with semantically rich descriptions of entities published in Web KBs. In this paper, we evaluate three unsupervised annotation methods: (a) a lookup-based method which relies on the minimal entity context provided in Web tables to discover correspondences to the KB, (b) a semantic embeddings method that exploits a vectorial representation of the rich entity context in a KB to identify the most relevant subset of entities in the Web table, and (c) an ontology matching method, which exploits schematic and instance information of entities available both in a KB and a Web table. Our experimental evaluation is conducted using two existing benchmark data sets in addition to a new large-scale benchmark created using Wikipedia tables. Our results show that: (1) our novel lookup-based method outperforms state-of-the-art lookup-based methods, (2) the semantic embeddings method outperforms lookup-based methods in one benchmark data set, and (3) the lack of a rich schema in Web tables can limit the ability of ontology matching tools in performing high-quality table annotation. As a result, we propose a hybrid method that significantly outperforms individual methods on all the benchmarks.
|
['Vassilis Christophides', 'Mariano Rodriguez-Muro', 'Oktie Hassanzadeh', 'Vasilis Efthymiou']
|
2017-10-01
| null | null | null |
the-semantic-web-iswc-2017-10
|
['ontology-matching', 'table-annotation', 'entity-embeddings', 'table-annotation', 'cell-entity-annotation']
|
['knowledge-base', 'knowledge-base', 'methodology', 'natural-language-processing', 'natural-language-processing']
|
[-3.62863570e-01 2.54709214e-01 -3.85346383e-01 -1.95280522e-01
-8.37636232e-01 -8.07714581e-01 6.22608721e-01 9.63361681e-01
-4.05677110e-01 8.20973098e-01 5.84855795e-01 1.32526025e-01
-3.24632168e-01 -1.27477551e+00 -8.40693414e-01 -2.89034154e-02
-4.49426565e-03 8.75229478e-01 8.50035369e-01 -6.50800884e-01
3.75738293e-02 1.25294790e-01 -1.88318062e+00 4.90420014e-01
1.01159346e+00 1.40125644e+00 -6.94591105e-02 -1.91934630e-01
-9.44434822e-01 7.65553057e-01 -1.46623686e-01 -9.17224765e-01
2.44431004e-01 3.42032425e-02 -9.86060560e-01 -4.38089728e-01
3.91446501e-01 1.94366828e-01 -4.92297977e-01 1.22874820e+00
2.03180313e-01 2.52999086e-02 5.07556200e-01 -1.29510498e+00
-7.57626534e-01 8.28847528e-01 4.94109057e-02 2.05300143e-03
6.96698248e-01 -5.25167823e-01 1.39124250e+00 -1.00157380e+00
1.31056559e+00 9.48308885e-01 8.06121349e-01 1.62703127e-01
-8.73834908e-01 -3.84841651e-01 -1.10780649e-01 4.08579439e-01
-1.67418861e+00 -3.34480703e-01 6.13384664e-01 -3.91090900e-01
1.15635967e+00 2.16134265e-01 4.47869480e-01 8.42460275e-01
-2.61845767e-01 3.29680204e-01 7.88851142e-01 -5.26580691e-01
3.51170361e-01 7.67336607e-01 3.54688019e-01 7.82081187e-01
8.57707798e-01 -4.12905753e-01 -6.35134697e-01 -4.94165063e-01
3.18423390e-01 -3.32456023e-01 -1.02241606e-01 -7.92711616e-01
-1.36605656e+00 7.15879440e-01 5.20249128e-01 2.99451858e-01
-3.97863477e-01 -2.82642037e-01 7.58838296e-01 -2.62753796e-02
1.99976102e-01 7.93004215e-01 -6.21981323e-01 3.79932187e-02
-3.18581372e-01 4.27078307e-01 1.26715040e+00 1.43064117e+00
1.09866226e+00 -4.75309879e-01 4.86458512e-03 9.14832592e-01
2.42951766e-01 2.79993266e-01 6.56339169e-01 -6.11235678e-01
9.73678708e-01 1.51478684e+00 3.44324201e-01 -1.17587674e+00
-3.55210841e-01 9.63834897e-02 -2.83160865e-01 -5.17003834e-01
3.15491289e-01 4.93960738e-01 -3.76255333e-01 1.38907528e+00
6.43719614e-01 -2.41173163e-01 4.75665241e-01 6.26523256e-01
1.38684165e+00 3.45044017e-01 -4.56124581e-02 1.36253864e-01
1.73590529e+00 -7.60582328e-01 -9.89435434e-01 -1.65195391e-02
8.10757756e-01 -4.67747360e-01 1.21104157e+00 -3.24421138e-01
-5.02998590e-01 -3.86514932e-01 -1.19790578e+00 -2.47606888e-01
-1.50908065e+00 -7.17557669e-02 7.74108410e-01 4.29258645e-01
-5.84523976e-01 4.06388521e-01 -4.16149795e-01 -8.42249155e-01
1.22090332e-01 1.91741973e-01 -9.48773623e-01 -1.20411582e-01
-1.60241759e+00 9.83315289e-01 9.90227163e-01 -5.06756544e-01
-9.24038962e-02 -8.64089787e-01 -1.23764336e+00 2.77289033e-01
8.26658070e-01 -4.96351600e-01 6.86169088e-01 -3.31086129e-01
-7.22247183e-01 8.85091126e-01 3.57583873e-02 -4.60100949e-01
7.60135949e-02 -6.39511719e-02 -8.56026530e-01 2.97906194e-02
5.18241107e-01 4.99110967e-01 -1.51495993e-01 -1.35236907e+00
-7.52701044e-01 -4.86333013e-01 2.95761317e-01 1.76024333e-01
-8.44771564e-01 -1.55355692e-01 -8.85666788e-01 -6.00354552e-01
1.87202856e-01 -6.26580060e-01 1.82673275e-01 -5.43510094e-02
-6.76782846e-01 -4.00828063e-01 5.12765527e-01 -8.11049998e-01
1.72208643e+00 -1.80177832e+00 -7.55988294e-03 2.83355087e-01
1.26155853e-01 1.48119556e-03 2.13688642e-01 1.00273347e+00
1.64624065e-01 3.90070647e-01 -1.83428481e-01 1.69349968e-01
3.39423627e-01 4.34718698e-01 -3.04207444e-01 -1.96114272e-01
-5.96217774e-02 8.28345716e-01 -1.00844872e+00 -8.57221305e-01
-1.12655841e-01 2.56076902e-01 -4.85233903e-01 9.44199972e-03
-2.63262302e-01 -2.86204576e-01 -4.70758915e-01 9.99022722e-01
1.81518555e-01 -2.92371988e-01 5.55525124e-01 -8.74310493e-01
-2.94841751e-02 6.34904444e-01 -1.53734195e+00 1.66610730e+00
-2.13942394e-01 1.73637211e-01 -6.91690445e-01 -6.34918213e-01
9.24571335e-01 3.30110759e-01 4.54282075e-01 -6.65682077e-01
-1.84848070e-01 4.53991592e-01 -4.81941283e-01 -5.43994665e-01
7.29247928e-01 3.47719401e-01 -3.66135925e-01 7.86722824e-02
4.43007767e-01 3.35738450e-01 7.21188962e-01 3.48820269e-01
1.26502764e+00 7.18396828e-02 6.37287498e-01 -4.63459700e-01
7.41565466e-01 3.78620178e-01 7.17733264e-01 4.90472317e-01
1.07322939e-01 9.11093131e-02 5.29840887e-01 -7.61843622e-01
-1.17837524e+00 -1.03944504e+00 -1.67662516e-01 7.28472650e-01
4.46773201e-01 -1.11292577e+00 -5.76412201e-01 -1.08011818e+00
5.19076109e-01 6.23543978e-01 -7.71458685e-01 3.16791050e-02
-3.19096029e-01 -4.59604532e-01 5.00798285e-01 8.11003923e-01
5.61530530e-01 -7.92552054e-01 -1.10173091e-01 2.28957891e-01
-5.85004389e-01 -1.50886190e+00 -3.36656392e-01 7.84185305e-02
-5.73894322e-01 -1.40341091e+00 1.75494447e-01 -7.97269404e-01
5.50069809e-01 -1.03985049e-01 1.26730001e+00 -1.47763789e-01
-3.02813831e-03 4.81589824e-01 -5.76123476e-01 -3.80534470e-01
-3.02171797e-01 2.70709157e-01 2.87348866e-01 -1.09107569e-01
8.27921093e-01 -3.22286367e-01 -1.21165194e-01 5.30190825e-01
-1.01070893e+00 -1.93761259e-01 3.42412800e-01 6.58182442e-01
9.49800611e-01 3.14554811e-01 4.79229122e-01 -1.11200750e+00
3.46117646e-01 -7.11398721e-01 -6.95036829e-01 6.90916598e-01
-9.61968660e-01 4.93745744e-01 5.94886661e-01 -2.11259276e-01
-8.44224215e-01 5.86123951e-02 7.54769817e-02 -2.00896561e-01
1.44530922e-01 8.99567723e-01 -5.73904335e-01 4.59237061e-02
7.39134014e-01 1.46236479e-01 -3.50428969e-01 -9.14761543e-01
3.51932764e-01 5.48033893e-01 5.96005797e-01 -7.15356767e-01
1.08122814e+00 4.22818005e-01 -1.01897746e-01 -4.35084760e-01
-7.24773347e-01 -7.65417218e-01 -8.67083788e-01 8.29743370e-02
8.29784214e-01 -1.02278864e+00 -3.86998445e-01 -2.89069086e-01
-8.06019962e-01 2.88752347e-01 -3.78114700e-01 3.51289511e-01
-2.74834365e-01 2.53464043e-01 -4.26408440e-01 -2.56071746e-01
-1.55234545e-01 -8.29255641e-01 9.10891771e-01 1.51345178e-01
-1.62416011e-01 -9.61157799e-01 2.79159188e-01 4.93859887e-01
1.57575250e-01 3.32341611e-01 1.18423736e+00 -1.40339911e+00
-6.53394103e-01 -5.27113497e-01 -2.78831869e-01 -1.55644655e-01
3.55799288e-01 -1.07370749e-01 -6.36924803e-01 1.82351738e-01
-7.38291562e-01 -1.81002840e-01 4.11958635e-01 -5.56931734e-01
8.65777969e-01 -6.03095114e-01 -5.80242157e-01 5.16406059e-01
1.86304307e+00 1.52460560e-01 6.69065714e-01 1.02576625e+00
8.40803027e-01 7.05308139e-01 7.01273501e-01 2.24822983e-01
8.19444954e-01 1.08393872e+00 3.01157892e-01 3.42731148e-01
1.68717708e-02 -7.35041261e-01 -9.96795017e-03 1.01825535e+00
1.19084850e-01 -1.24761865e-01 -1.13138855e+00 7.19721079e-01
-2.02316546e+00 -8.53633821e-01 4.14222777e-02 2.33626151e+00
1.05554700e+00 1.76919803e-01 1.89910848e-02 1.34234115e-01
5.44751942e-01 -2.46304080e-01 -2.50961334e-01 1.27449229e-01
-1.87925756e-01 3.18985507e-02 7.42202938e-01 5.01861200e-02
-1.23976517e+00 8.59061480e-01 5.67382193e+00 6.21873081e-01
-3.54297161e-01 1.81982756e-01 -1.15941249e-01 4.68759120e-01
-5.07612288e-01 8.87831897e-02 -1.22726882e+00 6.20489478e-01
8.77371848e-01 -5.55569351e-01 2.94217974e-01 1.07451022e+00
-6.22652352e-01 2.83837616e-01 -1.19778800e+00 7.57946849e-01
4.13287207e-02 -1.67777336e+00 4.99642849e-01 8.58878195e-02
5.41667223e-01 -2.53236532e-01 -4.74989712e-01 5.55911183e-01
2.05472812e-01 -6.79518700e-01 6.34142458e-01 4.73221511e-01
7.29116321e-01 -6.48109853e-01 1.11354792e+00 -1.07329890e-01
-1.66621685e+00 1.01297379e-01 -4.64343518e-01 5.29149950e-01
-2.78091937e-01 3.52088153e-01 -6.38650179e-01 1.00083208e+00
1.13163495e+00 4.84210134e-01 -8.64101291e-01 9.03378010e-01
-2.38150686e-01 1.99549079e-01 -3.04412544e-01 -6.18610382e-02
6.02098331e-02 -8.81901979e-02 5.17849922e-01 1.14619029e+00
1.94009483e-01 -1.05547093e-01 6.12293035e-02 8.58749568e-01
-5.75014830e-01 6.04807556e-01 -7.98389077e-01 -2.55000472e-01
9.90772545e-01 1.34241641e+00 -6.32778227e-01 -4.74962771e-01
-7.86316812e-01 6.93179727e-01 5.33169210e-01 1.07283384e-01
-6.59269392e-01 -8.00794303e-01 8.07214499e-01 2.03222230e-01
4.80282933e-01 1.31508216e-01 -5.23653515e-02 -1.24032807e+00
4.45271432e-01 -6.00261152e-01 8.88480365e-01 -8.43209743e-01
-1.33379793e+00 7.80039489e-01 1.38956666e-01 -1.34735763e+00
-2.71659404e-01 -5.57706475e-01 4.83679026e-02 5.69937885e-01
-1.41811657e+00 -1.07088935e+00 -6.13600910e-01 4.71490800e-01
2.24150583e-01 -3.88418913e-01 1.15222597e+00 7.25134313e-01
-4.57368731e-01 6.33510590e-01 2.37254426e-01 6.63270175e-01
8.05948615e-01 -1.51247823e+00 3.27907473e-01 5.96589446e-01
3.89081568e-01 8.75658512e-01 4.58741516e-01 -9.51304734e-01
-1.71875560e+00 -1.23617709e+00 1.24870539e+00 -8.73864293e-01
9.68030870e-01 -4.50193405e-01 -1.24756086e+00 8.79329383e-01
-1.63691044e-01 2.47492969e-01 1.01165712e+00 2.54944593e-01
-8.34143162e-01 -4.16950256e-01 -1.26729953e+00 4.40263629e-01
1.01094866e+00 -6.85596645e-01 -9.35229182e-01 4.14209932e-01
9.64691281e-01 -4.12124157e-01 -1.57134092e+00 4.01232392e-01
6.90529168e-01 -4.80940789e-01 1.13190150e+00 -8.11134279e-01
4.53510016e-01 -6.12978280e-01 -7.09519386e-01 -1.10866952e+00
-1.17395908e-01 2.57312059e-02 -5.74447632e-01 1.68473434e+00
6.31689608e-01 -6.41530335e-01 6.09598160e-01 6.06966674e-01
5.92249446e-02 -6.76534772e-01 -8.28991234e-01 -1.07486141e+00
-3.56373250e-01 4.55887243e-02 1.07330966e+00 1.28196299e+00
2.40257248e-01 1.46119297e-01 1.79413706e-01 3.46158147e-01
4.53694761e-01 1.35565540e-02 7.07868636e-01 -1.55375910e+00
2.28129417e-01 -1.97819605e-01 -8.44660223e-01 -2.92623252e-01
7.12299123e-02 -1.17984140e+00 -2.84361959e-01 -1.74798465e+00
2.83856481e-01 -5.86817920e-01 -5.13232768e-01 5.97357571e-01
-1.63125783e-01 1.18527515e-02 -1.46649867e-01 3.65128696e-01
-6.96204364e-01 3.65372330e-01 3.03954124e-01 -2.07908422e-01
2.41953563e-02 -7.12316513e-01 -8.98060203e-01 6.03274465e-01
3.97609562e-01 -6.24239862e-01 -3.38257700e-01 -1.46687895e-01
6.53882027e-01 -2.39985853e-01 1.46413311e-01 -1.17528844e+00
4.61908221e-01 -6.69094175e-02 2.01491788e-01 -4.68138188e-01
1.45615280e-01 -9.75145817e-01 1.49468213e-01 2.26978026e-02
-1.66352034e-01 1.31344885e-01 2.68585086e-01 7.65843689e-01
-4.88548219e-01 -2.55840749e-01 3.63294721e-01 -1.50162190e-01
-1.25268483e+00 1.60337836e-01 -2.23971643e-02 4.79044825e-01
7.16681004e-01 -1.27438933e-01 -5.65937817e-01 1.65265366e-01
-2.66454339e-01 2.29825988e-01 7.81404555e-01 7.25567341e-01
2.51553714e-01 -1.78585637e+00 -1.94528490e-01 -5.37960455e-02
1.04550111e+00 -2.82040358e-01 -2.40564585e-01 4.85851318e-01
-5.67906201e-01 6.24280751e-01 -2.73580253e-01 -1.52487651e-01
-1.02950799e+00 8.09024215e-01 -1.13321841e-02 -4.14821684e-01
-6.09322906e-01 5.01104712e-01 -1.48545980e-01 -8.65490258e-01
3.09453368e-01 -2.02698320e-01 -4.62575823e-01 2.20491201e-01
5.11966228e-01 2.80497581e-01 5.80873787e-01 -7.61821151e-01
-6.57445490e-01 3.88735175e-01 2.78221704e-02 1.05313174e-01
1.24232519e+00 1.36732413e-02 -1.87383100e-01 5.90888143e-01
1.09396720e+00 1.51466146e-01 -3.16805571e-01 -6.28607810e-01
5.80991924e-01 -4.36328620e-01 -2.11763114e-01 -9.33457613e-01
-7.41276443e-01 1.47968858e-01 3.52688551e-01 2.52515912e-01
8.57384562e-01 1.16505831e-01 7.83852160e-01 8.71693313e-01
6.74362957e-01 -1.33063698e+00 -3.07487249e-01 3.96429718e-01
7.36048102e-01 -1.33225298e+00 3.47669423e-02 -8.18326771e-01
-6.96949184e-01 1.08413792e+00 8.27326715e-01 3.75871658e-01
6.39639556e-01 6.12266287e-02 -1.39923066e-01 -5.39267063e-01
-7.47684240e-01 -5.79515278e-01 7.46900737e-01 5.67789435e-01
3.04156780e-01 -2.41921157e-01 -4.15318787e-01 1.06585765e+00
-1.27121523e-01 -1.56173050e-01 2.12370992e-01 1.09655130e+00
-4.01444912e-01 -1.15126646e+00 -1.05255544e-01 6.60485446e-01
-2.09874943e-01 -1.58812702e-01 -6.86412513e-01 1.23892725e+00
2.54853159e-01 5.34527719e-01 2.33001281e-02 -4.30691957e-01
6.35930657e-01 5.47852874e-01 5.06499223e-02 -6.45301342e-01
-4.78799462e-01 -6.30278409e-01 4.92286742e-01 -8.66897345e-01
-2.56742656e-01 -4.31548953e-01 -1.26073742e+00 -2.14323446e-01
-3.31219047e-01 4.29317623e-01 6.78132236e-01 7.85715342e-01
5.78000605e-01 2.91926265e-01 8.75975788e-02 -1.06448337e-01
-3.13251853e-01 -7.66779542e-01 -5.26186943e-01 7.40009069e-01
-5.08489370e-01 -1.14771581e+00 -1.13094665e-01 3.10361329e-02]
|
[9.277871131896973, 8.058921813964844]
|
f850a042-f088-4382-b637-8de23caaa253
|
ensemble-transfer-learning-for-multilingual
|
2301.09175
| null |
https://arxiv.org/abs/2301.09175v1
|
https://arxiv.org/pdf/2301.09175v1.pdf
|
Ensemble Transfer Learning for Multilingual Coreference Resolution
|
Entity coreference resolution is an important research problem with many applications, including information extraction and question answering. Coreference resolution for English has been studied extensively. However, there is relatively little work for other languages. A problem that frequently occurs when working with a non-English language is the scarcity of annotated training data. To overcome this challenge, we design a simple but effective ensemble-based framework that combines various transfer learning (TL) techniques. We first train several models using different TL methods. Then, during inference, we compute the unweighted average scores of the models' predictions to extract the final set of predicted clusters. Furthermore, we also propose a low-cost TL method that bootstraps coreference resolution models by utilizing Wikipedia anchor texts. Leveraging the idea that the coreferential links naturally exist between anchor texts pointing to the same article, our method builds a sizeable distantly-supervised dataset for the target language that consists of tens of thousands of documents. We can pre-train a model on the pseudo-labeled dataset before finetuning it on the final target dataset. Experimental results on two benchmark datasets, OntoNotes and SemEval, confirm the effectiveness of our methods. Our best ensembles consistently outperform the baseline approach of simple training by up to 7.68% in the F1 score. These ensembles also achieve new state-of-the-art results for three languages: Arabic, Dutch, and Spanish.
|
['Heng Ji', 'Tuan Manh Lai']
|
2023-01-22
| null | null | null | null |
['coreference-resolution']
|
['natural-language-processing']
|
[-1.87200140e-02 1.03132159e-01 -5.87964952e-01 -4.08831328e-01
-1.38357055e+00 -5.94974220e-01 5.29179335e-01 -8.17781687e-02
-6.43249273e-01 1.12900805e+00 5.25093794e-01 -1.82349846e-01
-4.83960509e-02 -6.26684129e-01 -8.01122546e-01 -5.71590781e-01
2.14815557e-01 9.15191412e-01 3.04610521e-01 -4.14512664e-01
1.77493945e-01 4.88235950e-02 -1.43591881e+00 6.15620315e-01
1.33092940e+00 4.27788615e-01 2.87661761e-01 1.07091211e-01
-3.42731893e-01 6.50770485e-01 -3.72885972e-01 -7.47139096e-01
-2.73512036e-01 -2.04532072e-01 -1.07689571e+00 -7.55073011e-01
3.58831555e-01 2.37183303e-01 -5.13439327e-02 1.10853732e+00
5.39264977e-01 1.36535764e-01 5.21807671e-01 -9.76101160e-01
-4.00746852e-01 1.31142640e+00 -7.19490349e-01 1.59032136e-01
5.92318773e-01 -6.17209733e-01 1.33441567e+00 -1.07186520e+00
9.39049602e-01 1.26060486e+00 8.35818529e-01 6.53759301e-01
-1.07625461e+00 -1.16781747e+00 3.90825383e-02 6.13782167e-01
-1.33607721e+00 -4.81150150e-01 7.13599741e-01 -1.05631709e-01
8.32203984e-01 2.69040346e-01 -2.17884839e-01 1.33105218e+00
-1.85507298e-01 8.04095030e-01 9.25458372e-01 -8.11190844e-01
-9.03395936e-02 2.04382762e-01 2.20165655e-01 3.24302942e-01
9.55789983e-02 -5.24024889e-02 -6.74742162e-01 -2.41520301e-01
3.26904990e-02 -3.37790608e-01 -5.77946782e-01 -2.84123719e-01
-1.26517820e+00 8.98569226e-01 2.59011149e-01 6.59995735e-01
-1.84802502e-01 -5.86834073e-01 4.04380947e-01 1.54997230e-01
3.34750324e-01 4.09209996e-01 -7.63537109e-01 -3.28129791e-02
-8.55530858e-01 1.90068424e-01 1.18929148e+00 1.02003717e+00
6.11324728e-01 -8.23079944e-01 -2.83993687e-02 1.05289054e+00
1.89808607e-01 4.99430507e-01 3.91148150e-01 -8.96382034e-01
1.03421915e+00 5.80412447e-01 2.65239567e-01 -9.15159881e-01
-4.23992008e-01 -2.04960138e-01 -7.92426765e-01 -2.55513936e-01
4.76608425e-01 -3.16185623e-01 -1.95693746e-01 1.98517859e+00
4.84405786e-01 3.45556140e-01 4.82951909e-01 6.86203241e-01
8.71160626e-01 5.68447113e-01 3.06812406e-01 -4.40963864e-01
1.41913557e+00 -1.08846664e+00 -8.59988451e-01 -1.50669679e-01
7.39027262e-01 -7.84743428e-01 7.68297434e-01 1.87366173e-01
-7.34926999e-01 -3.15862447e-01 -8.53716969e-01 5.90647124e-02
-3.42794955e-01 7.75099993e-02 4.59341794e-01 2.58648843e-01
-5.68142414e-01 5.81061900e-01 -5.76439619e-01 -4.51477796e-01
6.69923723e-02 4.04982001e-01 -6.48903131e-01 -1.83010891e-01
-1.58016181e+00 1.08924592e+00 8.48465025e-01 -8.04847702e-02
2.09933352e-02 -7.82115817e-01 -7.00302422e-01 1.24801107e-01
4.62675840e-01 -5.40331960e-01 1.18299127e+00 -8.06765676e-01
-1.29774642e+00 9.10394669e-01 -3.54301542e-01 -3.58101636e-01
2.55592078e-01 -5.36367655e-01 -7.70184517e-01 -7.50379860e-02
3.70736480e-01 4.71186280e-01 1.94485649e-01 -1.32973373e+00
-1.05313945e+00 -3.18856627e-01 -2.11168349e-01 3.38854998e-01
-4.23323363e-01 3.09065402e-01 -5.64696074e-01 -4.86127973e-01
-1.28209010e-01 -9.58350480e-01 1.91858634e-01 -7.34304667e-01
-4.26678091e-01 -7.02537715e-01 6.88040018e-01 -7.57179499e-01
1.55429339e+00 -2.04746127e+00 3.60228598e-01 5.24758697e-02
-1.73721924e-01 4.70378608e-01 -1.99215248e-01 4.01839763e-01
-1.67150527e-01 -3.88164446e-02 -3.06708395e-01 -1.75954968e-01
6.80208504e-02 6.19851612e-02 -4.39717561e-01 -3.68763693e-02
-1.04310006e-01 6.32524192e-01 -1.10592484e+00 -8.81933153e-01
-9.05602649e-02 3.97234112e-01 -6.75339460e-01 2.63313651e-01
-1.17464632e-01 5.28252363e-01 -4.55857605e-01 3.64677548e-01
6.32831335e-01 -1.41258180e-01 8.19753885e-01 -4.48240399e-01
-2.04347923e-01 5.67575157e-01 -1.28448725e+00 1.99014199e+00
-4.27928448e-01 3.57699513e-01 -8.98935795e-02 -1.09688330e+00
8.40577126e-01 4.44924563e-01 3.13020885e-01 -5.29352188e-01
-9.64484811e-02 5.04214823e-01 -8.40850547e-03 -6.63493872e-01
3.79729539e-01 -6.78713024e-02 -4.04965550e-01 5.37163734e-01
2.33952045e-01 4.45719779e-01 2.61048436e-01 1.36072785e-01
9.82500076e-01 3.18607539e-01 3.96585822e-01 -2.32206404e-01
8.57569456e-01 6.74495399e-02 8.99422109e-01 4.13311064e-01
1.16803966e-01 2.81469166e-01 8.72183517e-02 -1.94395527e-01
-6.79810882e-01 -8.98475349e-01 -3.38153750e-01 1.53017879e+00
2.19936490e-01 -5.38267970e-01 -8.57325196e-01 -1.03954673e+00
-1.43921360e-01 1.05225134e+00 -5.42952776e-01 -2.09718309e-02
-9.93511200e-01 -7.40100026e-01 6.78616285e-01 5.25821269e-01
5.43226480e-01 -1.44410706e+00 -1.09956726e-01 2.70748764e-01
-1.03033531e+00 -1.20504045e+00 -5.30057430e-01 -1.15804128e-01
-5.84085166e-01 -1.27702129e+00 -2.34844029e-01 -1.20652878e+00
2.86101431e-01 6.14117831e-02 1.34822488e+00 8.47411677e-02
2.84299493e-01 8.40174779e-02 -4.31535602e-01 -2.34492466e-01
-3.20777595e-01 5.95601737e-01 2.51016468e-01 -7.65892118e-02
1.20946884e+00 -5.37205577e-01 -1.91280276e-01 1.89777076e-01
-4.36121851e-01 -1.16134301e-01 5.77202737e-01 1.07726979e+00
5.13582408e-01 -2.67537057e-01 7.25167274e-01 -1.32884324e+00
5.20562053e-01 -6.73850596e-01 -2.17744499e-01 7.65700579e-01
-5.21423101e-01 2.83023775e-01 4.49409693e-01 -4.90942270e-01
-1.60516953e+00 -8.05846751e-02 -3.05815414e-02 -9.33566913e-02
-1.54993281e-01 6.77199364e-01 -3.68638337e-01 3.87264520e-01
5.49641967e-01 -1.82052210e-01 -4.14371490e-01 -6.59622610e-01
4.49116558e-01 1.06301224e+00 9.84119952e-01 -9.24861073e-01
8.13320637e-01 1.42461702e-01 -5.98067403e-01 -4.68114108e-01
-1.39076972e+00 -5.54231524e-01 -9.88312185e-01 1.83094785e-01
7.33143806e-01 -9.59426224e-01 -6.77870035e-01 2.36096215e-02
-1.40972388e+00 5.19343168e-02 1.76706925e-01 6.42282784e-01
-2.66648412e-01 3.47681791e-01 -4.74299967e-01 -4.41745073e-01
-6.91710353e-01 -6.85467184e-01 8.01802278e-01 3.32893610e-01
-5.22276103e-01 -1.04973304e+00 6.44837379e-01 6.24654055e-01
7.98592418e-02 7.57530555e-02 9.96638834e-01 -1.12400162e+00
-2.03448907e-01 1.35584891e-01 -1.02518894e-01 -1.24196991e-01
2.34279167e-02 -2.24967659e-01 -8.99437487e-01 -2.45443553e-01
-3.19451600e-01 -4.09343362e-01 8.02244246e-01 6.22596741e-02
8.58679831e-01 -1.82990283e-01 -9.72833872e-01 4.14805830e-01
1.22669554e+00 3.53935659e-02 5.74621499e-01 7.96845376e-01
5.20452499e-01 7.34707713e-01 9.93809938e-01 -5.51827671e-03
7.08528996e-01 8.32494974e-01 -8.33034217e-02 2.25682363e-01
-3.90864536e-02 -4.11120296e-01 2.86839902e-01 1.28764927e+00
-3.92694056e-01 -2.36162934e-02 -9.91650045e-01 6.39126718e-01
-2.18337417e+00 -1.33562911e+00 -2.58398652e-02 2.19825411e+00
1.40913916e+00 -1.38344560e-02 -2.58569837e-01 -7.95504600e-02
1.03270531e+00 -2.43062869e-01 -3.30436826e-01 1.61190424e-02
-2.50001937e-01 3.74193788e-01 -6.46535400e-03 5.79517901e-01
-1.34004295e+00 1.16082680e+00 5.53151035e+00 6.02482378e-01
-8.51777792e-01 1.88210338e-01 9.45242867e-03 1.66266277e-01
-2.80047536e-01 5.72277308e-02 -1.27254486e+00 4.63352084e-01
1.11156869e+00 -2.82646596e-01 2.04069346e-01 5.81850469e-01
-2.87804753e-01 2.54485399e-01 -1.17160869e+00 7.47672796e-01
3.13988805e-01 -1.17375457e+00 -4.14357819e-02 -2.21077740e-01
7.08070934e-01 2.51087606e-01 -4.07405972e-01 7.63482153e-01
5.66655278e-01 -8.27863038e-01 3.12920034e-01 4.48055208e-01
7.62511134e-01 -9.16170001e-01 1.07319891e+00 6.03922963e-01
-1.06320667e+00 -5.86778075e-02 -3.91865015e-01 3.50462019e-01
2.42102936e-01 3.59518796e-01 -6.17027521e-01 8.11741710e-01
1.01721787e+00 6.61751032e-01 -2.95980573e-01 9.09673035e-01
-5.98437309e-01 5.46944916e-01 -1.97968036e-01 7.61300027e-02
-1.69752195e-01 -8.61972198e-02 4.90624547e-01 1.50667250e+00
3.32989961e-01 3.51225644e-01 1.17969858e-02 6.50925338e-01
-4.39291626e-01 3.59776318e-01 -4.85123754e-01 4.20746177e-01
1.21914732e+00 1.40705347e+00 -1.36924507e-02 -1.86030671e-01
-6.55126095e-01 8.45273733e-01 9.50743437e-01 1.83096513e-01
-8.25205743e-01 -6.24684215e-01 3.67711186e-01 -3.92768055e-01
4.20778483e-01 3.58621746e-01 1.12679467e-01 -1.39631712e+00
-6.36410117e-02 -1.12563467e+00 8.45010459e-01 -4.25851196e-01
-1.44606102e+00 7.45391309e-01 -4.88719568e-02 -1.20998657e+00
-4.82153535e-01 -3.15456867e-01 -6.46289110e-01 6.81933939e-01
-1.67036331e+00 -1.25480473e+00 -1.60640523e-01 6.44544184e-01
3.91659379e-01 -3.38841796e-01 1.29964054e+00 5.41562796e-01
-7.26287723e-01 7.62307644e-01 2.19800532e-01 5.85338175e-01
1.41498899e+00 -1.22763550e+00 5.23724630e-02 6.59187198e-01
3.82016242e-01 8.12888861e-01 6.62408292e-01 -4.56001163e-01
-9.41551208e-01 -9.56301272e-01 1.63306916e+00 -7.00403869e-01
6.74102664e-01 -1.40621498e-01 -1.25202060e+00 1.05825782e+00
6.12794161e-01 -2.18106538e-01 1.05964112e+00 7.07346499e-01
-6.78202391e-01 -5.04760332e-02 -9.52044189e-01 4.42448050e-01
1.10526323e+00 -3.99425834e-01 -1.52917814e+00 1.21599913e-01
6.56674504e-01 -4.70939845e-01 -1.10122442e+00 6.07818842e-01
4.39237058e-01 -7.87189722e-01 9.36980784e-01 -8.22411537e-01
3.73449564e-01 -1.73101485e-01 -3.31766814e-01 -1.45465171e+00
-4.50131148e-01 -4.04565930e-01 -2.52072185e-01 1.75077713e+00
7.22952366e-01 -3.81865263e-01 4.57584590e-01 5.14741480e-01
3.53852659e-02 -2.44399711e-01 -8.91646683e-01 -4.58877891e-01
3.52420270e-01 -1.70413218e-02 6.47257209e-01 1.47543228e+00
5.93185186e-01 9.43948090e-01 -8.85472149e-02 2.52701342e-01
8.71139407e-01 5.54600716e-01 6.96283638e-01 -1.79477262e+00
4.88253273e-02 -1.59477875e-01 3.58968407e-01 -7.90398836e-01
9.27809298e-01 -1.16726983e+00 1.10885046e-01 -1.16292632e+00
6.09795988e-01 -7.08141983e-01 -5.36065161e-01 5.78139305e-01
-4.91586089e-01 -6.54991865e-02 -6.20089285e-03 3.44466448e-01
-9.14232671e-01 3.34063172e-01 7.52667665e-01 -5.07884286e-02
-1.53363973e-01 -1.74354672e-01 -8.10151637e-01 8.60349238e-01
6.68103337e-01 -6.65268898e-01 3.43174599e-02 -4.39157009e-01
1.22158237e-01 -9.02110860e-02 -1.66038960e-01 -6.93352759e-01
5.71070373e-01 -1.50976047e-01 9.81205702e-02 -7.09005475e-01
5.29847760e-03 -7.27463722e-01 -3.31631377e-02 1.49347439e-01
-3.66686344e-01 -5.13454750e-02 2.06880778e-01 4.81501251e-01
-4.34499413e-01 -3.27582628e-01 5.33190787e-01 6.42097294e-02
-8.72626960e-01 -5.21201603e-02 2.76360303e-01 5.84264636e-01
8.93010676e-01 3.96856040e-01 -4.25569683e-01 -1.29263662e-03
-5.67761004e-01 4.29733038e-01 1.21988706e-01 8.37506711e-01
1.98631182e-01 -1.45743477e+00 -1.10206842e+00 -1.48738861e-01
2.38486215e-01 -6.87997043e-02 1.20156474e-01 8.17696750e-01
4.91493568e-02 6.89748764e-01 -1.54704869e-01 -4.97251451e-01
-1.37179792e+00 5.34802437e-01 1.54336169e-01 -5.45305669e-01
-5.23816407e-01 5.67550123e-01 -1.30446211e-01 -1.07238090e+00
2.66451299e-01 2.22554579e-01 -7.42302775e-01 2.42848575e-01
7.60366738e-01 2.03657448e-01 6.71395436e-02 -7.34127581e-01
-5.64386249e-01 6.71456873e-01 -3.64128858e-01 4.83939573e-02
1.67388535e+00 -1.31454021e-01 -3.00802290e-01 3.00943106e-01
9.81487155e-01 3.96133691e-01 -7.12898552e-01 -6.17078662e-01
6.79514170e-01 2.12261155e-02 -2.19154224e-01 -8.71835351e-01
-7.96017766e-01 6.46082163e-01 2.37832710e-01 -2.41609842e-01
8.54999125e-01 2.33130664e-01 7.96669662e-01 5.71027040e-01
5.10993004e-01 -1.12692857e+00 -3.39880377e-01 7.05917418e-01
8.18794191e-01 -1.38741136e+00 -2.39230767e-01 -4.36428010e-01
-6.23340905e-01 9.59910691e-01 7.66805232e-01 1.41148210e-01
5.53994834e-01 3.47497135e-01 1.49287909e-01 4.51103970e-02
-8.75021696e-01 3.36294877e-03 3.58539224e-01 5.89299977e-01
8.59401405e-01 1.25290304e-01 -5.56146204e-01 1.18527853e+00
-3.22574735e-01 6.26573414e-02 9.24897566e-02 5.26934505e-01
-3.84386986e-01 -1.51522958e+00 -3.89074326e-01 5.78852631e-02
-6.15751803e-01 -3.36626649e-01 -3.45641971e-01 9.59273875e-01
6.82867467e-02 8.74274135e-01 -5.18064713e-03 -3.73056501e-01
3.80153775e-01 3.65218580e-01 3.75649422e-01 -5.33794165e-01
-4.58578855e-01 -2.71098316e-01 3.68765622e-01 -3.65968972e-01
-9.70142066e-01 -1.00099933e+00 -1.37778914e+00 -2.47752324e-01
-5.98044932e-01 7.14051485e-01 2.08619580e-01 1.14712250e+00
3.89928490e-01 1.13733754e-01 4.25036043e-01 -5.07402122e-01
-5.82382917e-01 -1.22684181e+00 -2.22318754e-01 7.45373487e-01
-6.93415701e-02 -8.44258845e-01 -4.11465168e-01 -3.33419931e-03]
|
[9.28255844116211, 9.500099182128906]
|
1594b56c-c7f2-4f1b-a72a-62da5cefc378
|
simultaneously-updating-all-persistence
|
2211.11620
| null |
https://arxiv.org/abs/2211.11620v1
|
https://arxiv.org/pdf/2211.11620v1.pdf
|
Simultaneously Updating All Persistence Values in Reinforcement Learning
|
In reinforcement learning, the performance of learning agents is highly sensitive to the choice of time discretization. Agents acting at high frequencies have the best control opportunities, along with some drawbacks, such as possible inefficient exploration and vanishing of the action advantages. The repetition of the actions, i.e., action persistence, comes into help, as it allows the agent to visit wider regions of the state space and improve the estimation of the action effects. In this work, we derive a novel All-Persistence Bellman Operator, which allows an effective use of both the low-persistence experience, by decomposition into sub-transition, and the high-persistence experience, thanks to the introduction of a suitable bootstrap procedure. In this way, we employ transitions collected at any time scale to update simultaneously the action values of the considered persistence set. We prove the contraction property of the All-Persistence Bellman Operator and, based on it, we extend classic Q-learning and DQN. After providing a study on the effects of persistence, we experimentally evaluate our approach in both tabular contexts and more challenging frameworks, including some Atari games.
|
['Marcello Restelli', 'Alberto Maria Metelli', 'Lorenzo Bisi', 'Luca Al Daire', 'Luca Sabbioni']
|
2022-11-21
| null | null | null | null |
['atari-games']
|
['playing-games']
|
[-2.30477706e-01 2.94015333e-02 -2.10125580e-01 4.14719522e-01
-4.40966010e-01 -6.41827822e-01 7.74530709e-01 5.95770538e-01
-1.01793015e+00 1.26133299e+00 -2.94870853e-01 -9.63894129e-02
-7.03203917e-01 -9.04449642e-01 -6.24988616e-01 -1.21984982e+00
-4.12251651e-01 5.56865096e-01 4.56353754e-01 -3.08068454e-01
1.00879222e-01 5.61480403e-01 -1.70268905e+00 -4.56631392e-01
1.21567667e+00 7.78992057e-01 1.57393873e-01 4.04418141e-01
7.03056827e-02 5.61580181e-01 -5.45496106e-01 7.64019936e-02
1.94893509e-01 -5.44646084e-01 -4.16124165e-01 1.13105021e-01
-3.57728183e-01 -8.60612094e-02 1.29201010e-01 1.01336908e+00
3.72405738e-01 5.77012181e-01 4.58662778e-01 -1.00300264e+00
3.15473408e-01 7.72847891e-01 -2.78214961e-01 1.64245203e-01
1.42534122e-01 2.31682017e-01 7.55429983e-01 -1.76746860e-01
9.26474869e-01 9.15537357e-01 3.85056794e-01 3.30001771e-01
-1.48312151e+00 -1.94054410e-01 1.88277125e-01 1.18572995e-01
-1.12249660e+00 -4.42537591e-02 4.80074137e-01 -4.78722811e-01
5.40163338e-01 1.37466850e-04 1.12311995e+00 9.08052087e-01
2.50369698e-01 4.51645076e-01 1.54567182e+00 -5.52244902e-01
9.57405210e-01 3.00948054e-01 1.79164193e-03 4.24287796e-01
2.38967419e-01 4.76381600e-01 -9.69211310e-02 -2.49694452e-01
7.81405866e-01 -1.06479898e-01 -2.81909883e-01 -8.50098073e-01
-6.17828488e-01 9.17260170e-01 3.37867528e-01 5.65420687e-01
-5.78053117e-01 1.07237332e-01 2.78690547e-01 5.12576342e-01
3.73045653e-01 5.15932024e-01 -1.34305000e-01 -4.60001022e-01
-6.01786077e-01 4.49359000e-01 8.06927204e-01 2.10557148e-01
7.79235423e-01 -6.86622113e-02 -1.67922929e-01 4.20255125e-01
-2.24852741e-01 3.33051443e-01 3.97054225e-01 -7.28892565e-01
1.94780543e-01 3.25044483e-01 5.37803769e-01 -6.39590025e-01
-5.44525325e-01 -7.75059402e-01 -6.74903333e-01 7.48515666e-01
7.58630097e-01 -4.11931664e-01 -6.06989682e-01 2.05961251e+00
5.76147914e-01 8.66995752e-03 2.62730032e-01 6.25803649e-01
-4.68246371e-01 4.84030008e-01 1.49009362e-01 -7.66130686e-01
1.03975761e+00 -4.57014829e-01 -7.32331634e-01 2.15909153e-01
6.62965238e-01 -1.31375566e-01 8.98521483e-01 5.88574111e-01
-1.16060686e+00 -3.49422485e-01 -7.63305008e-01 6.79390848e-01
-3.48703355e-01 -2.09176447e-02 3.65935564e-01 5.39367259e-01
-8.14172149e-01 1.19032788e+00 -8.79564345e-01 -2.27053836e-01
5.85649535e-02 2.77410358e-01 -1.36366650e-01 3.60117048e-01
-1.21475840e+00 9.97258842e-01 5.73546529e-01 1.24128647e-01
-7.27367461e-01 -3.85076493e-01 -4.14278597e-01 2.09431022e-01
1.01394725e+00 -3.29780370e-01 1.05203855e+00 -1.09119105e+00
-1.75767875e+00 1.16958670e-01 4.72100139e-01 -8.39401186e-01
1.25219727e+00 1.76355392e-02 2.26518482e-01 1.92318469e-01
-1.39191240e-01 2.08536282e-01 9.52387691e-01 -9.71970201e-01
-4.55671728e-01 -4.30601895e-01 3.61479461e-01 3.77689391e-01
-3.54286581e-01 -6.63703144e-01 1.28707811e-01 -3.68022054e-01
-3.75011593e-01 -1.18629813e+00 -5.22683740e-01 -4.79868770e-01
5.83353601e-02 -2.40670159e-01 1.29228234e-01 -2.14721233e-01
1.05829418e+00 -2.21675801e+00 7.40099907e-01 4.34656709e-01
-7.60439634e-02 1.61821127e-01 1.31903529e-01 9.01797175e-01
1.38843313e-01 -1.43961340e-01 -2.64167964e-01 -2.46686384e-01
6.11171760e-02 3.12463164e-01 -2.77003944e-01 4.33148086e-01
-7.83260763e-02 4.48353618e-01 -9.29943323e-01 -3.79863769e-01
2.40724325e-01 2.82278687e-01 -6.48366630e-01 -3.14212553e-02
-5.80142558e-01 6.05884314e-01 -7.10068107e-01 -3.22249383e-01
4.20712352e-01 8.99773836e-02 4.27028000e-01 4.73173529e-01
-6.75883174e-01 -1.18656838e-02 -1.52498829e+00 1.39207196e+00
-6.94559872e-01 -9.98519212e-02 -6.16207309e-02 -9.81810331e-01
7.21550703e-01 2.45350286e-01 6.69422746e-01 -7.28563786e-01
9.75757837e-02 4.35852647e-01 5.37689589e-02 -2.15183660e-01
1.98553279e-01 -1.36327028e-01 1.28201723e-01 4.56012130e-01
-3.55831310e-02 -3.73415165e-02 6.00962579e-01 -2.78979614e-02
8.34308803e-01 3.08532178e-01 4.34187621e-01 -5.48655450e-01
5.58059394e-01 -2.46269122e-01 2.95635581e-01 8.72603893e-01
1.94879934e-01 -1.27199501e-01 1.07460105e+00 -2.14680612e-01
-7.09225595e-01 -8.35251391e-01 -2.15586260e-01 8.75185251e-01
1.12164542e-01 -2.04822734e-01 -7.20844328e-01 -6.13238275e-01
1.16025008e-01 5.75580657e-01 -9.08952773e-01 -2.84607738e-01
-6.45192087e-01 -7.01882064e-01 -6.14159517e-02 4.75273803e-02
3.54079664e-01 -1.27779329e+00 -1.40310276e+00 3.48202527e-01
1.47644997e-01 -5.64079821e-01 1.04217511e-02 6.86638176e-01
-1.01121926e+00 -8.66863072e-01 -6.75566673e-01 -7.48362392e-02
1.40283555e-01 -3.17899168e-01 7.03785241e-01 -9.16710943e-02
3.35522071e-02 3.18929642e-01 -3.58559221e-01 -2.14360356e-02
-6.29479229e-01 2.84861445e-01 -5.99312745e-02 -1.86219742e-03
-2.53755569e-01 -7.60249019e-01 -3.81077230e-01 6.46296144e-02
-9.51398492e-01 -2.03514934e-01 4.95682210e-01 9.03569758e-01
4.53991234e-01 2.47331843e-01 6.05327368e-01 -7.40909815e-01
6.31233513e-01 -2.16171846e-01 -1.25865519e+00 7.67292455e-02
-6.40848398e-01 5.43100238e-01 8.89881074e-01 -6.79591238e-01
-8.36729586e-01 -7.48794749e-02 -2.64582634e-02 -9.49737877e-02
1.69848174e-01 4.72861320e-01 1.02691613e-01 -6.64603859e-02
6.06553972e-01 2.15909287e-01 1.23338558e-01 -4.97399509e-01
3.60675663e-01 1.39419854e-01 -9.57735628e-02 -6.77263737e-01
4.71291363e-01 2.48268530e-01 2.97839344e-01 -9.29210842e-01
-3.18584651e-01 -5.13725355e-02 -3.27724606e-01 -3.95583212e-01
4.86161649e-01 -3.80135983e-01 -9.41745460e-01 4.45429683e-01
-8.04840147e-01 -5.37109077e-01 -1.04850340e+00 5.34932971e-01
-8.70353162e-01 2.68846065e-01 -5.73407233e-01 -1.13656962e+00
1.86924845e-01 -1.21162975e+00 4.02705312e-01 3.42462301e-01
3.70010316e-01 -9.15927947e-01 5.74838281e-01 -3.58453214e-01
2.32756391e-01 1.41138658e-01 8.22362125e-01 -5.35911620e-01
-4.87853080e-01 1.19322479e-01 4.14668679e-01 2.07389906e-01
-2.47497395e-01 -2.17041388e-01 -6.36224926e-01 -4.79606867e-01
1.70519233e-01 -1.92534626e-01 9.99626040e-01 4.70791221e-01
6.33203328e-01 -3.05295467e-01 -1.70645699e-01 2.07303688e-02
1.52221024e+00 3.74034077e-01 5.31951308e-01 4.37117368e-01
-2.43629850e-02 6.72344446e-01 6.37459695e-01 9.24911022e-01
1.56194391e-02 1.17422903e+00 4.86003816e-01 2.89162755e-01
4.66675282e-01 -8.36675987e-03 3.42267722e-01 3.08440089e-01
-4.14027303e-01 -1.06036469e-01 -6.28008008e-01 2.57739544e-01
-2.01162934e+00 -9.35584188e-01 2.03766093e-01 2.79108334e+00
9.29353714e-01 4.87820387e-01 6.75160706e-01 2.45626867e-01
5.27990103e-01 6.89633116e-02 -4.12561625e-01 -5.37386775e-01
8.55551139e-02 2.02483729e-01 4.07928765e-01 7.95237839e-01
-8.43162179e-01 5.00825882e-01 5.18019915e+00 9.70084488e-01
-1.06297231e+00 1.75275952e-01 1.59613550e-01 -2.58776508e-02
-1.37767985e-01 -3.60320173e-02 -5.86069286e-01 7.06593156e-01
7.44943380e-01 -6.34698346e-02 7.15757251e-01 7.42634296e-01
2.64182299e-01 -4.59648043e-01 -7.65923202e-01 3.70891273e-01
-4.11120981e-01 -8.70441854e-01 -4.54486281e-01 4.77276534e-01
5.76365530e-01 -3.96639615e-01 -4.37094085e-02 3.35070133e-01
9.52349007e-02 -6.00417078e-01 7.66920030e-01 5.33378243e-01
1.94054261e-01 -1.08219481e+00 8.31144750e-01 4.74054962e-01
-9.22213674e-01 -5.92298865e-01 -1.01322904e-01 -1.99324951e-01
2.98676975e-02 5.81495225e-01 -2.63452202e-01 6.62555218e-01
8.00958648e-02 3.23543638e-01 -3.08076501e-01 1.01837933e+00
-8.30655470e-02 3.49604845e-01 -5.27835906e-01 -4.02839243e-01
3.97627324e-01 -7.71880567e-01 5.48312604e-01 6.59464240e-01
3.73652935e-01 -9.35692713e-02 2.35881686e-01 8.52689505e-01
3.93443704e-01 1.90434262e-01 -5.69906294e-01 1.45908548e-02
3.15297931e-01 9.96406734e-01 -9.27792847e-01 -1.21497102e-01
5.53222233e-03 5.58787882e-01 4.52908397e-01 3.19722921e-01
-7.27911234e-01 -2.14624166e-01 3.87868613e-01 9.72330868e-02
5.71955025e-01 -3.32751662e-01 2.88298398e-01 -9.56354976e-01
8.49819258e-02 -7.75885820e-01 4.28080499e-01 2.04919670e-02
-7.04936743e-01 4.54230815e-01 2.20935553e-01 -1.15903020e+00
-6.92928851e-01 -1.35439545e-01 -1.79274291e-01 5.17063200e-01
-1.26607966e+00 -4.66997147e-01 2.25592420e-01 4.93490875e-01
1.57313198e-01 2.38385364e-01 5.80905676e-01 1.75323993e-01
-5.53296089e-01 1.56686828e-02 4.79452461e-01 -5.11555612e-01
1.44033954e-01 -1.46175122e+00 -3.15793723e-01 5.13913095e-01
1.05610013e-01 3.21423352e-01 1.12776363e+00 -4.61099356e-01
-1.04799438e+00 -3.91732454e-01 4.08401996e-01 5.39808758e-02
8.72216046e-01 -2.72491574e-01 -9.12774086e-01 2.96468377e-01
3.02653071e-02 -3.08674991e-01 -8.12325180e-02 1.72071740e-01
2.13231906e-01 -3.42881888e-01 -8.34108889e-01 6.37356877e-01
6.41107380e-01 -2.53387988e-01 -3.38235825e-01 -3.93532701e-02
2.81852156e-01 -9.64369550e-02 -7.21861959e-01 1.29297197e-01
5.72006166e-01 -1.23949134e+00 6.20756209e-01 -3.29601854e-01
6.98903799e-02 -7.96975717e-02 4.18617219e-01 -1.51542616e+00
-3.09130340e-03 -6.82245076e-01 -1.94349810e-01 9.99933720e-01
8.81376490e-02 -8.83382499e-01 5.87163508e-01 -9.55225155e-02
3.62039328e-01 -8.62311721e-01 -1.45573652e+00 -9.20936048e-01
2.78339297e-01 1.18385337e-01 2.20624402e-01 5.58724701e-01
2.47912273e-01 1.47188082e-01 -3.94842267e-01 -1.50966942e-01
5.55461884e-01 4.11359400e-01 3.78034681e-01 -1.22146833e+00
-9.09492731e-01 -5.20322561e-01 -2.43649125e-01 -6.86439753e-01
-8.71841237e-02 -3.15418810e-01 1.79274082e-01 -8.83852243e-01
-4.66269813e-02 -5.87787867e-01 -4.31712210e-01 1.24527097e-01
-9.99607053e-03 -2.00367808e-01 5.48808038e-01 -1.74264889e-02
-5.12050211e-01 7.81553566e-01 1.31193674e+00 3.55427653e-01
-6.66663408e-01 2.25513756e-01 1.35305345e-01 4.35488194e-01
7.38842249e-01 -5.45258641e-01 -5.07957399e-01 1.93117097e-01
5.00562906e-01 5.66947818e-01 3.04735214e-01 -1.11207938e+00
-8.53825659e-02 -2.99189717e-01 -1.70917004e-01 -9.77699533e-02
3.33376259e-01 -7.15987802e-01 3.28475744e-01 1.11573315e+00
-5.18459141e-01 -1.63857743e-01 1.16513828e-02 5.93976796e-01
7.93811902e-02 -6.73644245e-01 8.45925331e-01 -9.62133929e-02
-2.76089609e-01 -1.05306342e-01 -7.08641708e-01 4.94699599e-03
9.64195013e-01 1.19547114e-01 4.34668325e-02 -3.69889259e-01
-9.80425119e-01 2.13606879e-01 5.07238925e-01 -1.52021814e-02
-8.35683197e-02 -9.20338273e-01 -2.10124776e-01 8.80701691e-02
-1.83073521e-01 -5.28412819e-01 4.46970820e-01 1.19256496e+00
-8.84638950e-02 1.23943955e-01 -4.96008784e-01 -3.16713631e-01
-8.51134896e-01 7.83393145e-01 6.04831457e-01 -6.65501118e-01
-6.63754880e-01 4.73671034e-02 -7.92546570e-02 1.11730181e-01
2.34470367e-01 -5.40918291e-01 -4.81325895e-01 5.32743573e-01
2.94969469e-01 4.92492199e-01 -6.84201121e-02 3.42455581e-02
-1.93457171e-01 5.70976198e-01 2.50303507e-01 -4.30066735e-01
1.14023161e+00 -7.75689036e-02 1.94231123e-01 7.32061267e-01
5.72657824e-01 1.51914373e-01 -1.49503422e+00 1.29086941e-01
3.63868959e-02 -1.88149095e-01 -6.31008074e-02 -5.50001979e-01
-6.48218453e-01 7.43023574e-01 7.53906071e-01 8.77914369e-01
1.08909392e+00 -2.66005993e-01 1.29666641e-01 2.58043140e-01
6.58208668e-01 -1.15031147e+00 2.01248424e-03 3.57468337e-01
6.04489088e-01 -6.43196821e-01 -2.07501456e-01 -2.92166807e-02
-5.08985817e-01 1.20551693e+00 1.48506090e-01 -2.81111598e-01
3.04544032e-01 1.33546844e-01 -2.44093522e-01 2.14523926e-01
-8.23463023e-01 -5.85888803e-01 -4.09246773e-01 1.20072417e-01
-6.17239289e-02 1.71337321e-01 -1.05125809e+00 1.90602973e-01
1.84993178e-01 1.55997738e-01 6.70791149e-01 7.92671144e-01
-3.61322612e-01 -1.38698709e+00 -1.29690036e-01 1.47848679e-02
-2.29803875e-01 2.47784525e-01 4.54345569e-02 1.09329772e+00
2.41197377e-01 5.30391693e-01 -3.83908786e-02 2.23568290e-01
2.90244013e-01 4.44530845e-02 7.32215643e-01 -2.41030902e-01
-6.00294530e-01 1.35764003e-01 3.78787182e-02 -5.57362497e-01
-3.77643615e-01 -8.36654723e-01 -9.77057695e-01 -1.70207962e-01
-3.83638769e-01 7.20796049e-01 5.79370737e-01 1.05345321e+00
4.22625467e-02 4.43410993e-01 7.49121785e-01 -7.38833427e-01
-1.34546959e+00 -7.83729434e-01 -8.35729480e-01 2.55623460e-01
4.75586027e-01 -9.59142923e-01 -4.93317127e-01 -6.58234060e-01]
|
[4.371739387512207, 2.3125481605529785]
|
f641b0eb-c77e-4f09-909b-28cd5ef4fbf9
|
is-dataset-condensation-a-silver-bullet-for
|
2305.03711
| null |
https://arxiv.org/abs/2305.03711v1
|
https://arxiv.org/pdf/2305.03711v1.pdf
|
Is dataset condensation a silver bullet for healthcare data sharing?
|
Safeguarding personal information is paramount for healthcare data sharing, a challenging issue without any silver bullet thus far. We study the prospect of a recent deep-learning advent, dataset condensation (DC), in sharing healthcare data for AI research, and the results are promising. The condensed data abstracts original records and irreversibly conceals individual-level knowledge to achieve a bona fide de-identification, which permits free sharing. Moreover, the original deep-learning utilities are well preserved in the condensed data with compressed volume and accelerated model convergences. In PhysioNet-2012, a condensed dataset of 20 samples can orient deep models attaining 80.3% test AUC of mortality prediction (versus 85.8% of 5120 original records), an inspiring discovery generalised to MIMIC-III and Coswara datasets. We also interpret the inhere privacy protections of DC through theoretical analysis and empirical evidence. Dataset condensation opens a new gate to sharing healthcare data for AI research with multiple desirable traits.
|
['David Clifton', 'Tingting Zhu', 'Li Shang', 'Stavros Petridis', 'Pingchuan Ma', 'Mingzhi Dong', 'Anshul Thakur', 'Yujiang Wang']
|
2023-05-05
| null | null | null | null |
['mortality-prediction', 'de-identification']
|
['medical', 'natural-language-processing']
|
[ 1.56217128e-01 5.38626969e-01 -3.62616330e-01 -4.10800695e-01
-8.14961731e-01 -1.94412246e-01 5.41834086e-02 5.41784763e-01
-5.42451084e-01 1.11569214e+00 5.19238174e-01 -2.88756102e-01
-2.38550827e-01 -6.52301431e-01 -8.02416027e-01 -8.48284304e-01
-4.54612643e-01 3.48993778e-01 -7.49543726e-01 -9.88906845e-02
-2.18193099e-01 3.01649570e-01 -1.10596776e+00 6.58859074e-01
9.06380236e-01 1.23979998e+00 -4.37190861e-01 2.86712945e-01
3.20047855e-01 6.90029979e-01 -4.51243758e-01 -1.03372085e+00
4.93930966e-01 -2.38054156e-01 -8.06745350e-01 -6.47280395e-01
7.88162425e-02 -4.05995905e-01 -5.85837781e-01 9.19544399e-01
9.11099553e-01 -4.77091849e-01 3.81565928e-01 -1.34549546e+00
-1.12115848e+00 8.31633687e-01 -2.44510934e-01 -3.85328382e-01
1.28406659e-01 1.83582872e-01 8.10367227e-01 -2.65311927e-01
5.69475114e-01 7.12773621e-01 1.13514423e+00 8.87006938e-01
-9.47245121e-01 -8.32371891e-01 -6.44534171e-01 1.13601208e-01
-1.66910195e+00 -3.63618970e-01 3.97709042e-01 -2.60233015e-01
4.13835108e-01 8.67461801e-01 9.05742407e-01 1.46924758e+00
8.02367389e-01 9.32774544e-01 7.49053478e-01 1.00043222e-01
2.88426757e-01 3.26845795e-01 7.99252186e-03 4.51872319e-01
7.80719697e-01 4.02513891e-01 -5.59897780e-01 -5.58151603e-01
2.64853209e-01 6.96391284e-01 -3.85052919e-01 -5.71059346e-01
-1.22961867e+00 6.91294551e-01 5.30513227e-01 8.58754888e-02
-3.17435384e-01 -7.63142705e-02 5.69322824e-01 4.97220814e-01
2.99697578e-01 6.21513844e-01 -6.91126943e-01 2.99808532e-01
-7.39911377e-01 5.28806269e-01 7.92603016e-01 1.10091984e+00
7.58756176e-02 -2.32471317e-01 -3.40290666e-01 1.66198239e-01
3.37352827e-02 4.69493032e-01 4.55882639e-01 -9.27097082e-01
2.63313830e-01 6.25891566e-01 1.49067016e-02 -1.13100147e+00
-7.72253096e-01 -7.19998240e-01 -1.79150236e+00 -1.99573204e-01
2.90647686e-01 -3.66806298e-01 -4.95825738e-01 1.74351668e+00
1.96638152e-01 -3.26207399e-01 4.20851618e-01 6.15625501e-01
6.96314216e-01 1.87958062e-01 2.60093212e-01 -2.23116517e-01
1.50823939e+00 -3.62827152e-01 -1.00487363e+00 5.42865753e-01
1.03927493e+00 8.22304636e-02 5.73485434e-01 5.22212982e-01
-9.60587978e-01 -1.93129182e-01 -9.20630932e-01 -2.85946280e-01
-4.75998163e-01 -4.18048948e-01 8.68877351e-01 9.34599280e-01
-9.46537554e-01 8.68084192e-01 -5.62772214e-01 -1.01522654e-01
1.55259073e+00 4.67039734e-01 -6.56401396e-01 -8.64182115e-02
-1.71613443e+00 4.45750594e-01 3.75846773e-01 -1.63078815e-01
-8.97741556e-01 -1.43964052e+00 -6.02495134e-01 3.32016274e-02
5.83128035e-02 -1.07285655e+00 6.12454236e-01 -4.08138812e-01
-7.78566718e-01 1.22789335e+00 3.87087554e-01 -1.09927857e+00
1.03375518e+00 -5.01616657e-01 -6.12078965e-01 1.38989286e-02
-1.84713408e-01 4.90130991e-01 4.66044843e-01 -9.20528054e-01
-3.86027813e-01 -8.14030647e-01 -5.25429130e-01 -1.72102720e-01
-7.18883753e-01 -3.82195413e-01 1.43231630e-01 -8.34613502e-01
-5.32460093e-01 -7.71966457e-01 -4.53943133e-01 5.57515264e-01
-5.96397758e-01 1.30750462e-01 3.93472403e-01 -8.72664511e-01
1.25242984e+00 -2.20385551e+00 4.36721966e-02 8.66256133e-02
9.05862689e-01 5.18384337e-01 1.22216128e-01 4.50915575e-01
-1.50966629e-01 3.41820508e-01 -5.97244740e-01 -3.53832990e-01
-2.05713421e-01 -7.26638129e-03 -2.35261500e-01 8.06693852e-01
-1.48435026e-01 1.47062957e+00 -7.84171462e-01 -3.29474568e-01
-1.38136432e-01 5.77312171e-01 -7.59107471e-01 1.69830337e-01
1.63279235e-01 5.01243412e-01 -4.66389686e-01 8.11879933e-01
9.25899804e-01 -3.71779799e-01 4.18950468e-01 -1.64924897e-02
3.39163959e-01 -2.39136502e-01 -3.73744428e-01 1.96986079e+00
2.89329410e-01 2.68551707e-01 -1.41538763e-02 -9.09795702e-01
8.51209521e-01 4.33503509e-01 9.65880215e-01 -9.83628213e-01
2.27258533e-01 5.38946688e-02 -1.00624792e-01 -5.40687025e-01
2.75291204e-01 -1.72430858e-01 -3.96716893e-01 2.90717661e-01
-3.85250479e-01 4.76978540e-01 -9.31119144e-01 1.86768368e-01
9.74398196e-01 -4.10166323e-01 3.35046440e-01 -6.67689979e-01
-5.04873879e-02 -3.29931855e-01 6.92777574e-01 8.72512639e-01
-5.97703099e-01 6.08757854e-01 4.79089707e-01 -1.10238075e+00
-1.32281673e+00 -9.57192361e-01 -7.39047706e-01 5.52966118e-01
-2.93466002e-01 -4.44350362e-01 -7.88400710e-01 -6.60019994e-01
5.79985797e-01 6.13319695e-01 -1.09014654e+00 -7.93765485e-01
-1.14634477e-01 -9.84357357e-01 1.02129054e+00 3.43333960e-01
4.11824107e-01 -9.02341306e-01 -9.06234920e-01 -1.04135685e-01
-1.69545904e-01 -4.40686494e-01 -2.82031149e-01 8.04665536e-02
-6.52264535e-01 -1.01099467e+00 -9.42839086e-01 -2.95585483e-01
3.65590066e-01 -2.38476798e-01 9.73890543e-01 2.60139138e-01
-7.03304350e-01 -1.15243137e-01 -2.26603806e-01 -8.52269113e-01
-4.03008044e-01 3.19813371e-01 3.25011820e-01 -1.02871001e-01
6.46996021e-01 -4.11735058e-01 -9.63851571e-01 -1.32360145e-01
-9.35210586e-01 -9.64366496e-02 4.55780119e-01 7.81542897e-01
4.01247054e-01 -1.73543021e-01 9.14003849e-01 -1.14675701e+00
4.29523379e-01 -9.40873325e-01 -1.22254163e-01 3.16027313e-01
-9.62860584e-01 -1.31436855e-01 5.77972233e-01 1.26054540e-01
-4.37003911e-01 -6.09625801e-02 1.69247966e-02 -5.96008182e-01
4.55512218e-02 2.47961506e-01 -3.24216276e-01 3.00994039e-01
6.86092317e-01 2.81983584e-01 5.60024798e-01 -5.24596572e-01
2.92842090e-01 7.93627799e-01 4.74191099e-01 -2.21779838e-01
2.83851802e-01 5.40733576e-01 1.24366686e-01 -3.70830774e-01
-9.28294301e-01 1.07295990e-01 -5.92211425e-01 4.01791662e-01
1.14133060e+00 -1.10874021e+00 -1.35005736e+00 7.21043766e-01
-8.26800048e-01 -1.55439565e-07 -7.36538172e-01 1.32362023e-01
-4.47736055e-01 3.03086579e-01 -4.12602603e-01 -7.71837890e-01
-7.63967156e-01 -5.91758251e-01 7.91931748e-01 -1.31990403e-01
-5.16029835e-01 -8.36375892e-01 -7.05379248e-02 6.88660443e-01
4.33342665e-01 8.63413692e-01 9.55913186e-01 -1.03695369e+00
-3.30699831e-01 -4.45002228e-01 -9.59189385e-02 3.39375257e-01
2.12518126e-01 -7.60881543e-01 -1.19925678e+00 -6.71350002e-01
5.01118600e-02 -4.51705784e-01 9.11201656e-01 1.96477801e-01
2.06758213e+00 -7.92691469e-01 -5.44733524e-01 1.19633961e+00
1.07504106e+00 2.76110649e-01 1.04418337e+00 2.56922226e-02
8.85250688e-01 6.17261648e-01 4.54487652e-02 9.81283426e-01
5.00270605e-01 2.02179149e-01 5.69479048e-01 -3.48721832e-01
3.31879646e-01 -3.19518268e-01 -3.02455455e-01 5.64183295e-01
1.72755316e-01 -3.22700828e-01 -9.40935552e-01 6.30691767e-01
-1.63188064e+00 -8.34601462e-01 -5.97148873e-02 2.38401294e+00
1.12578368e+00 -4.10009712e-01 3.20857167e-02 7.62169017e-03
3.54393393e-01 -2.25367490e-02 -1.10696280e+00 -1.89930782e-01
-4.20601785e-01 1.24286920e-01 7.85729527e-01 -1.42457247e-01
-1.11320126e+00 2.89100707e-01 6.86274910e+00 5.46797931e-01
-6.51140451e-01 1.26228169e-01 1.25430775e+00 -4.96594191e-01
-6.65281534e-01 -6.32002711e-01 -3.82112682e-01 7.73670495e-01
1.27203703e+00 -5.15238881e-01 1.67042509e-01 6.68260455e-01
-3.83821309e-01 6.57703161e-01 -1.25445569e+00 1.08465528e+00
-1.00704007e-01 -1.80140090e+00 4.01048481e-01 5.71258485e-01
7.44558156e-01 -1.39145732e-01 6.20307624e-01 -8.27321317e-03
2.79077113e-01 -1.52057374e+00 3.19134653e-01 1.14572537e+00
1.31278348e+00 -1.13899720e+00 1.01943743e+00 7.33343959e-02
-3.72834742e-01 -4.21639979e-01 -4.24868643e-01 1.09731384e-01
-1.09372988e-01 6.39603257e-01 -6.64213300e-01 9.88839984e-01
9.83227968e-01 9.55339670e-01 -4.34155136e-01 7.02531099e-01
7.21599340e-01 3.05741996e-01 3.33592333e-02 1.86892137e-01
-2.77435869e-01 2.37095460e-01 3.49916875e-01 8.88938546e-01
1.76475942e-01 2.68885285e-01 -3.29461306e-01 8.15430582e-01
-6.44870281e-01 -3.23563232e-04 -8.35515797e-01 -2.79306382e-01
5.18257320e-01 6.85443819e-01 1.60202622e-01 -1.81908846e-01
-2.38013032e-04 9.54250991e-01 1.54231623e-01 -8.35825577e-02
-5.30753195e-01 -3.99380505e-01 1.03216279e+00 2.24895090e-01
5.59977908e-03 4.44580495e-01 -6.64811373e-01 -1.19243145e+00
-1.42889112e-01 -1.07029569e+00 7.91900635e-01 -3.23816180e-01
-1.76240540e+00 2.97421545e-01 -3.87563437e-01 -1.16078436e+00
1.30790696e-01 -3.87237221e-01 1.08945489e-01 7.78889358e-01
-1.18629706e+00 -1.15743792e+00 -3.41020599e-02 8.19302797e-01
-4.34826225e-01 -5.55931985e-01 1.26647031e+00 4.91058230e-01
-5.39971948e-01 1.61107206e+00 7.69015431e-01 1.95593178e-01
6.23944283e-01 -9.04266357e-01 4.61093515e-01 2.11968720e-01
-3.24036777e-01 6.85713232e-01 3.65008324e-01 -7.00751364e-01
-1.66852939e+00 -1.43505061e+00 8.64838243e-01 -9.78460670e-01
1.90000340e-01 -5.10799468e-01 -1.14237559e+00 7.13887811e-01
1.17829286e-01 -5.04557490e-02 1.41160452e+00 -1.65630095e-02
-3.98936957e-01 -3.46595079e-01 -1.72312260e+00 3.21765453e-01
1.14331353e+00 -5.72604537e-01 -3.78047526e-01 4.42201853e-01
1.31073391e+00 -2.19448730e-01 -1.31766391e+00 3.59373122e-01
9.27018702e-01 -8.40432167e-01 1.16574371e+00 -1.24849045e+00
5.32613933e-01 3.85184646e-01 -4.06697482e-01 -8.42390478e-01
-3.45967084e-01 -9.61939931e-01 -2.78701961e-01 8.61267328e-01
2.33588353e-01 -7.33382404e-01 7.89611340e-01 1.14265454e+00
9.47131403e-03 -1.01277447e+00 -1.14896882e+00 -4.16889101e-01
5.91357946e-01 -1.26787484e-01 1.33359003e+00 1.60212135e+00
6.15418190e-03 -4.33064401e-01 -1.00984359e+00 -1.21660367e-01
1.07427180e+00 -3.16044688e-01 5.41216552e-01 -1.58366787e+00
8.77404511e-02 7.24368356e-03 -4.46725279e-01 -1.89905182e-01
-1.10607475e-01 -1.31876421e+00 -7.85807729e-01 -9.41668630e-01
5.30880272e-01 -5.94634354e-01 -9.34640050e-01 7.46909916e-01
-3.12832631e-02 8.74266699e-02 1.04454361e-01 2.77068406e-01
-5.31260431e-01 6.83497846e-01 1.34225512e+00 -1.18880309e-01
6.74190074e-02 -7.11059198e-02 -1.49952769e+00 3.84396203e-02
7.95505524e-01 -5.01139641e-01 -2.48588443e-01 -3.12176317e-01
2.88844258e-01 1.25157863e-01 5.89146137e-01 -8.71082902e-01
5.06337360e-02 -1.45826107e-02 7.03558803e-01 -5.74182451e-01
-5.12914285e-02 -1.07816613e+00 6.11364007e-01 9.48877096e-01
-8.28946888e-01 -2.34239221e-01 2.38967761e-01 8.53940308e-01
3.57489526e-01 4.58365530e-01 6.88805997e-01 -2.02835381e-01
-8.28673765e-02 8.03476572e-01 1.26165554e-01 1.14706591e-01
1.25124323e+00 -4.33395617e-02 -2.48047292e-01 -1.16610937e-01
-7.72499979e-01 4.56744403e-01 4.66298819e-01 5.31753004e-01
6.08243406e-01 -1.34650552e+00 -9.60593104e-01 5.84552288e-01
2.97819287e-01 1.09470837e-01 8.11997056e-01 4.35519636e-01
-2.70121843e-01 4.74643171e-01 -5.48185289e-01 -3.85855585e-01
-8.38560283e-01 1.16700900e+00 3.39984208e-01 -5.47731156e-03
-9.36158001e-01 8.81413162e-01 1.52890116e-01 -4.09729749e-01
3.94041479e-01 -4.71797213e-02 1.01910554e-01 1.23147644e-01
9.57575798e-01 4.57708031e-01 -1.71390194e-02 -2.24025950e-01
-5.03685832e-01 -8.60586762e-02 -4.94483262e-01 6.01502299e-01
1.62207830e+00 -1.71824992e-01 -2.26881176e-01 2.05839977e-01
1.62411952e+00 -3.26428592e-01 -1.00053453e+00 -1.41421452e-01
-3.48565429e-01 -5.15956879e-01 -1.54501721e-01 -1.15066695e+00
-1.23392045e+00 8.61317456e-01 9.91793096e-01 1.75668493e-01
9.45831656e-01 -1.73861936e-01 1.15656412e+00 3.91584635e-01
4.67719108e-01 -6.96553349e-01 -2.82818466e-01 -2.18152776e-01
8.80898297e-01 -1.20768046e+00 9.06562805e-02 4.49424982e-01
-8.08611393e-01 5.06676853e-01 2.16508076e-01 3.22610945e-01
8.41349483e-01 7.15015233e-02 -6.26118481e-02 -2.90996343e-01
-7.00432718e-01 7.55646825e-01 4.15360518e-02 9.76880908e-01
1.03801243e-01 4.37709421e-01 -1.19658947e-01 1.15483320e+00
-3.89157623e-01 2.72033840e-01 2.16166928e-01 6.62639380e-01
5.86022586e-02 -8.38472843e-01 9.23623815e-02 7.35683084e-01
-7.83399105e-01 -1.87989846e-01 -3.48043978e-01 5.03246009e-01
2.06063703e-01 4.28722471e-01 1.25711873e-01 -5.96072912e-01
1.56166509e-01 1.51582584e-01 -1.30573183e-01 -4.18421924e-02
-8.28571558e-01 -6.83681428e-01 -2.17794865e-01 -7.57853806e-01
8.50581564e-03 -7.21211612e-01 -9.26883221e-01 -9.22043025e-01
1.86857462e-01 1.16311096e-01 2.88610399e-01 5.83415806e-01
1.15560424e+00 4.18222696e-01 6.53874815e-01 -8.91213771e-03
-6.71007872e-01 -3.13095093e-01 -9.03271794e-01 5.49886346e-01
8.35940361e-01 -6.60084784e-02 -7.53672421e-02 -1.44870713e-01]
|
[6.232334613800049, 6.7000732421875]
|
cdc0bf08-2165-432f-ae3f-642936ca95b8
|
towards-coherent-visual-storytelling-with-1
| null | null |
https://openreview.net/forum?id=kO0y8t9PMEf
|
https://openreview.net/pdf?id=kO0y8t9PMEf
|
Towards Coherent Visual Storytelling with Ordered Image Attention
|
We address the problem of visual storytelling, i.e., generating a story for a given sequence of images. While each story sentence should describe a corresponding image, a coherent story also needs to be consistent and relate to both future and past images. Current approaches encode images independently, disregarding relations between images. Our approach learns to encode images with different interactions based on the story position (i.e., past image or future image). To this end, we develop a novel message-passing-like algorithm for ordered image attention (OIA) that collects interactions across all the images in the sequence. Finally, to generate the story's sentences, a second attention mechanism picks the important image attention vectors with an Image-Sentence Attention (ISA). The obtained results improve the METEOR score on the VIST dataset by 1%. Furthermore, a thorough human study confirms improvements and demonstrates that order-based Interactions significantly improve coherency (64.20% vs. 28.70%).
|
['Anonymous']
|
2021-11-16
| null | null | null |
acl-arr-november-2021-11
|
['visual-storytelling']
|
['natural-language-processing']
|
[ 5.77012539e-01 1.65495008e-01 1.76595449e-01 -5.67890882e-01
-5.80468416e-01 -3.89677793e-01 9.77581382e-01 5.09356149e-02
-2.52897114e-01 7.37668216e-01 5.29141843e-01 8.81641731e-02
3.02964777e-01 -7.46933103e-01 -1.21285379e+00 -7.40356088e-01
3.86218391e-02 1.67471513e-01 2.50657946e-01 -2.40083903e-01
3.85898173e-01 2.12098137e-01 -1.65325940e+00 8.12208474e-01
4.80286300e-01 5.79284608e-01 7.41932690e-01 9.81148899e-01
-1.82456765e-02 1.28983057e+00 -7.46361613e-01 -5.64385176e-01
-8.27007592e-02 -8.43834519e-01 -8.59351575e-01 6.62364483e-01
5.69384933e-01 -5.38521528e-01 -4.28705156e-01 7.17504859e-01
2.01312423e-01 1.22208104e-01 6.05653882e-01 -1.45732796e+00
-9.08679426e-01 7.19872713e-01 -9.44213629e-01 1.96948081e-01
4.74432081e-01 4.19695795e-01 9.51031208e-01 -8.67801845e-01
9.93068635e-01 1.09945142e+00 -6.25538304e-02 4.83045846e-01
-1.12727249e+00 -4.39830601e-01 3.85104775e-01 3.90159935e-01
-1.12074661e+00 -4.09394354e-01 7.72698998e-01 -4.69216645e-01
8.65434229e-01 3.74817044e-01 8.34568441e-01 1.17580926e+00
3.73334050e-01 9.82235968e-01 7.98446655e-01 -2.91194350e-01
-3.05314995e-02 2.19186991e-01 4.03689174e-03 4.00138229e-01
7.82319903e-02 -1.30566463e-01 -7.73226738e-01 4.45491999e-01
7.99640834e-01 3.96261588e-02 -2.00530931e-01 -2.62526095e-01
-1.55973256e+00 4.93197471e-01 4.67430979e-01 3.40785176e-01
-7.22416639e-01 4.38514858e-01 4.73992229e-02 3.56121510e-02
2.80808389e-01 4.54868376e-01 1.63260087e-01 2.52769887e-01
-8.41218054e-01 3.70639235e-01 2.74491549e-01 9.25537109e-01
6.98207676e-01 -8.69496912e-02 -6.85413420e-01 5.49163759e-01
1.26469254e-01 3.28491360e-01 9.62596983e-02 -8.78706992e-01
4.90991652e-01 2.56538093e-01 3.22399110e-01 -1.21754766e+00
2.21842509e-02 -4.41444904e-01 -9.48666453e-01 1.27133086e-01
1.90910965e-01 2.76655424e-02 -8.87088418e-01 1.77720571e+00
2.61480600e-01 2.81658232e-01 1.88233569e-01 1.05257058e+00
7.88042963e-01 1.17900014e+00 2.57957190e-01 -3.85258436e-01
1.43211615e+00 -1.25177467e+00 -8.10003698e-01 -4.05812830e-01
1.15122914e-01 -7.90267706e-01 9.49919343e-01 2.50292778e-01
-1.50922775e+00 -7.02337682e-01 -9.76788104e-01 -1.83804855e-01
-2.09489726e-02 7.70281702e-02 3.95895869e-01 -5.14334142e-02
-1.29633188e+00 2.53069341e-01 -2.90312201e-01 -4.29930747e-01
1.98239595e-01 -5.19985072e-02 -3.06950152e-01 -1.60401791e-01
-1.02046192e+00 6.24747872e-01 5.23903728e-01 -2.43383318e-01
-1.18991351e+00 -5.72720587e-01 -7.06562817e-01 1.43011570e-01
3.75157557e-02 -9.72201288e-01 1.17553949e+00 -1.40029776e+00
-1.07243478e+00 1.01520407e+00 -4.55005080e-01 -7.69349039e-01
5.17020106e-01 -7.74480477e-02 -2.27459893e-01 4.17125583e-01
2.51264572e-01 1.29337585e+00 9.71143842e-01 -1.76594901e+00
-7.70616651e-01 2.66010128e-02 3.21279705e-01 5.72463870e-01
-3.21788311e-01 1.88063290e-02 -8.44310701e-01 -5.51156998e-01
-2.07575828e-01 -7.48826921e-01 -1.10369302e-01 -1.25109538e-01
-4.79350477e-01 -2.28341315e-02 7.83969462e-01 -7.56556809e-01
1.11754870e+00 -2.21908212e+00 2.66905457e-01 -2.95476705e-01
3.14980179e-01 -2.20011115e-01 -3.12853992e-01 5.60930490e-01
-4.15878803e-01 5.82964644e-02 -2.79651314e-01 -6.43911839e-01
-3.92543465e-01 2.38658234e-01 -4.79104131e-01 1.40846789e-01
4.26791131e-01 1.09431660e+00 -1.07018757e+00 -7.09517419e-01
4.00363117e-01 4.76440638e-01 -4.28742230e-01 3.32230836e-01
-3.59614611e-01 6.20695710e-01 -6.11914136e-02 1.13487206e-01
6.30231738e-01 -6.80238903e-01 1.30459711e-01 -2.42561966e-01
-3.00024122e-01 -1.11083299e-01 -6.82071447e-01 1.74329901e+00
-4.51391846e-01 9.64732468e-01 -4.20009553e-01 -6.48599267e-01
6.85565233e-01 3.61777157e-01 4.73610729e-01 -8.15166593e-01
-1.97985750e-02 -2.77545959e-01 -1.93161011e-01 -5.89226067e-01
8.71052742e-01 1.03140876e-01 -1.34494156e-01 6.95401311e-01
-8.82460922e-02 3.12993452e-02 3.86948138e-01 6.51085854e-01
8.13082337e-01 -5.42576388e-02 1.92221448e-01 1.07401080e-01
4.25259352e-01 5.62445819e-02 9.38692391e-02 8.80988061e-01
-2.40440462e-02 1.15821028e+00 5.47800064e-01 -5.18673778e-01
-1.34551823e+00 -1.00121045e+00 3.98606598e-01 9.92582917e-01
6.68352127e-01 -2.62854159e-01 -6.40270352e-01 -4.24776614e-01
-3.68645519e-01 9.82026875e-01 -8.67802382e-01 -5.80468103e-02
-6.57158256e-01 -5.33367097e-01 -1.11190826e-01 3.41310889e-01
4.76387888e-01 -1.41336262e+00 -8.87459695e-01 9.09247696e-02
-5.90271354e-01 -1.11355174e+00 -6.86186492e-01 -3.27425629e-01
-3.62263441e-01 -7.82771587e-01 -1.01324105e+00 -7.87858307e-01
8.36424768e-01 7.41648257e-01 1.31750858e+00 6.82763308e-02
-6.45076409e-02 4.12283212e-01 -4.64824349e-01 -2.33407512e-01
-4.32974279e-01 -2.83041626e-01 -4.48886126e-01 5.03121138e-01
-3.30407560e-01 -4.05597121e-01 -8.47376406e-01 -2.42043417e-02
-1.10275805e+00 9.79324341e-01 8.98019075e-01 8.48857403e-01
5.91812432e-01 4.65146229e-02 2.12390453e-01 -6.95258439e-01
4.31702435e-01 -6.30840480e-01 -9.72325429e-02 4.73975182e-01
-1.69929653e-01 1.38117298e-02 3.28289926e-01 -4.42196250e-01
-1.32863581e+00 3.38804603e-01 1.19128942e-01 -4.07579899e-01
-2.15251476e-01 8.86737034e-02 5.81853352e-02 4.15262431e-01
3.69838238e-01 6.57466352e-01 -2.33438879e-01 2.19620541e-01
5.14810264e-01 3.09964478e-01 7.14217126e-01 -7.21201673e-02
5.60701609e-01 6.80806458e-01 -3.98489565e-01 -7.83689022e-01
-7.86849856e-01 -3.41121852e-01 -4.46468174e-01 -7.79925525e-01
1.22806180e+00 -9.36862111e-01 -5.92272937e-01 3.68811965e-01
-1.75541735e+00 -2.34324262e-01 -2.29669034e-01 2.14391544e-01
-6.71370149e-01 2.71057427e-01 -4.18207645e-01 -9.33131397e-01
-2.76914388e-01 -1.09175098e+00 1.31995034e+00 2.62068659e-01
-3.82145852e-01 -5.61825454e-01 -2.31635511e-01 3.67088974e-01
1.19740725e-01 4.36096996e-01 5.24042130e-01 -6.28480464e-02
-1.13976479e+00 1.47675604e-01 -4.84457850e-01 -6.84517771e-02
-1.73682049e-01 4.14433815e-02 -8.52560103e-01 -9.35848355e-02
5.82160987e-03 -4.15796153e-02 9.27352011e-01 4.82719302e-01
1.02983630e+00 -4.07515585e-01 -2.51269996e-01 1.96010008e-01
1.38349783e+00 6.25737190e-01 9.24389899e-01 2.42121384e-01
6.98542178e-01 7.32612073e-01 6.23292327e-01 5.36420166e-01
4.40981150e-01 6.68534815e-01 5.99379003e-01 -4.58719432e-01
-5.71603060e-01 -4.12698030e-01 3.83005470e-01 5.04000485e-01
1.38948008e-01 -6.84948683e-01 -7.66922951e-01 8.38397145e-01
-2.00189447e+00 -1.26932812e+00 -4.65044409e-01 1.89385629e+00
5.80531418e-01 -2.43960395e-02 -1.09809585e-01 -8.31311941e-02
9.40766573e-01 5.25480747e-01 -5.83373666e-01 -1.68313816e-01
-4.33283895e-01 -3.24344367e-01 2.46690392e-01 5.32525122e-01
-8.47826302e-01 9.32605207e-01 6.26857376e+00 6.75658882e-01
-9.68460083e-01 1.18414246e-01 1.19143212e+00 -2.72864521e-01
-5.43560207e-01 -6.84819892e-02 -4.32668000e-01 5.34478247e-01
4.62271541e-01 -3.37787896e-01 3.40375423e-01 5.66200972e-01
3.21096599e-01 -4.77729321e-01 -8.79713774e-01 1.08654845e+00
5.12218595e-01 -1.59248662e+00 4.68424678e-01 -2.18647167e-01
1.13380718e+00 -4.62512404e-01 2.66854227e-01 -5.23365848e-02
1.58379719e-01 -8.54133248e-01 1.32286942e+00 8.10859144e-01
6.39118016e-01 -7.83993483e-01 5.65003932e-01 2.84842521e-01
-1.13026226e+00 -3.66113037e-02 -4.38060015e-02 -1.37213096e-01
6.51940107e-01 4.95056152e-01 -8.08876276e-01 5.83327353e-01
6.66630745e-01 9.08665657e-01 -5.94311595e-01 6.87221766e-01
-3.73089403e-01 2.88519591e-01 3.30739439e-01 -8.25018063e-02
3.09364587e-01 -1.46598825e-02 5.56837440e-01 1.11941183e+00
3.46318871e-01 1.60212547e-01 -2.30788857e-01 9.79864538e-01
3.41075063e-02 -4.20128703e-02 -6.81399643e-01 6.11606613e-02
1.99961767e-01 1.12733173e+00 -7.67591178e-01 -7.14412928e-01
-3.29986602e-01 1.57966042e+00 1.52050763e-01 5.08143067e-01
-1.15172279e+00 -4.93768938e-02 4.43633288e-01 2.52211511e-01
2.58773685e-01 -1.67521894e-01 -3.62397045e-01 -9.08301473e-01
2.30556484e-02 -5.35772085e-01 2.27356657e-01 -1.62091422e+00
-1.01033878e+00 8.57278347e-01 5.60864843e-02 -1.24960935e+00
-2.95055211e-01 8.43107179e-02 -7.38432646e-01 7.63649523e-01
-1.37471139e+00 -1.19269121e+00 -5.48577845e-01 4.14591670e-01
1.00440896e+00 2.24870920e-01 2.21548721e-01 1.34657487e-01
-3.94184381e-01 1.87780097e-01 -3.18418264e-01 -6.44833520e-02
6.75682306e-01 -1.05194020e+00 6.54311538e-01 1.12043571e+00
4.02354330e-01 2.69592643e-01 8.78196359e-01 -6.61241233e-01
-1.17917347e+00 -1.24878728e+00 1.28207970e+00 -2.72643298e-01
4.25207555e-01 -3.24388027e-01 -6.66108191e-01 6.28826678e-01
9.13426459e-01 -4.70883787e-01 1.55316770e-01 -4.90311623e-01
-1.40565902e-01 -1.00918852e-01 -6.71122730e-01 1.01352012e+00
1.13672554e+00 -3.11645359e-01 -3.30583960e-01 3.26184183e-01
1.07779706e+00 -3.29273492e-01 -3.32612306e-01 6.08363487e-02
5.25378406e-01 -1.30079329e+00 1.14914131e+00 -2.50880927e-01
1.18493915e+00 -5.57879746e-01 6.89904094e-02 -1.02965677e+00
-4.87524658e-01 -4.90038127e-01 4.61748913e-02 1.19860816e+00
2.68107414e-01 -1.02601252e-01 5.57146847e-01 4.28699911e-01
-1.02352515e-01 -5.42776585e-01 -5.15210450e-01 -5.12501538e-01
-4.78444576e-01 -2.97902554e-01 5.91318786e-01 8.48460078e-01
-1.67566240e-01 5.46262562e-01 -9.44618762e-01 1.98497593e-01
5.22560358e-01 3.27734381e-01 8.87063503e-01 -5.39015293e-01
-4.11054969e-01 -3.10961813e-01 -2.91285694e-01 -1.08611703e+00
-8.05436298e-02 -7.00580537e-01 2.00594500e-01 -1.89020789e+00
6.53480887e-01 1.13386080e-01 -7.76667446e-02 2.53539473e-01
-3.87487680e-01 4.55283672e-01 5.98118544e-01 2.53863186e-01
-8.57653916e-01 5.10425925e-01 1.62121630e+00 -3.22794974e-01
5.22180535e-02 -5.38979828e-01 -7.13940084e-01 3.03892463e-01
8.35252941e-01 -3.31418425e-01 -6.04508936e-01 -7.19459772e-01
1.25751868e-01 4.78597701e-01 6.16902649e-01 -1.00170350e+00
3.71069759e-01 -2.81313449e-01 5.99556744e-01 -8.96866977e-01
5.27558744e-01 -5.16612351e-01 6.06261909e-01 5.25030315e-01
-6.21411979e-01 1.96004242e-01 -3.42013240e-02 6.90179646e-01
-3.44817162e-01 1.21793248e-01 6.79570019e-01 -2.50213027e-01
-9.22683954e-01 2.50977933e-01 -4.32719260e-01 -2.82955170e-01
1.45323110e+00 -3.21341723e-01 -2.56975919e-01 -8.16571295e-01
-6.30017340e-01 3.50583702e-01 4.34084147e-01 5.53362191e-01
9.43377912e-01 -1.44602346e+00 -1.00047839e+00 1.14914902e-01
3.22685182e-01 -1.45873040e-01 8.31772745e-01 6.36448383e-01
-3.27992260e-01 2.59261578e-01 -3.77259701e-01 -8.14987063e-01
-1.52887166e+00 6.19592130e-01 -8.65415931e-02 -2.81580567e-01
-5.99807322e-01 1.00191474e+00 8.97798538e-01 4.46694672e-01
3.20536271e-02 -1.63383856e-02 -3.56574416e-01 1.21075772e-01
7.88262844e-01 -8.57288688e-02 -4.02163595e-01 -8.93080175e-01
-4.21254784e-02 4.77641940e-01 -1.94379911e-01 -4.94523436e-01
1.26398563e+00 -4.05765057e-01 -4.35044616e-02 5.61489940e-01
1.12954807e+00 -4.01496112e-01 -1.52538204e+00 3.69872451e-02
-3.40984136e-01 -7.16872811e-01 -3.10580909e-01 -7.19927728e-01
-1.04408014e+00 8.22462082e-01 2.63195217e-01 2.08395377e-01
1.35681474e+00 2.85670877e-01 7.39634812e-01 -2.79789448e-01
2.30459377e-01 -5.75675070e-01 6.52272642e-01 3.49552453e-01
1.42117071e+00 -1.13683915e+00 -1.54765606e-01 -3.24554414e-01
-1.21774244e+00 8.09881628e-01 6.61219060e-01 -1.45385284e-02
3.48902345e-02 -2.90922187e-02 -1.61896750e-01 -1.56650946e-01
-1.11738074e+00 -2.53760368e-01 2.89876312e-01 4.42379773e-01
4.34227437e-01 1.18664801e-02 -3.28265280e-01 2.46685058e-01
-1.62144471e-02 -2.22060293e-01 6.68435395e-01 5.03864348e-01
-2.75859982e-01 -7.65727222e-01 -5.06597161e-01 1.36988878e-01
4.85969931e-02 -1.43368766e-01 -4.65833783e-01 5.94556630e-01
1.24342501e-01 9.97984052e-01 4.75334823e-01 -4.19098347e-01
1.47755861e-01 -2.18189210e-01 4.12194788e-01 -5.50350010e-01
-4.41635340e-01 5.55102713e-02 -1.65188275e-02 -5.26771784e-01
-5.54177821e-01 -7.39865482e-01 -1.18356931e+00 -2.54197747e-01
1.55785188e-01 -8.05908144e-02 5.14335692e-01 6.56117737e-01
4.14472610e-01 9.64107871e-01 7.14180827e-01 -9.50549960e-01
3.78182203e-01 -6.32271111e-01 -3.48530740e-01 7.60250509e-01
3.08450162e-01 -2.59528071e-01 -1.13334872e-01 6.54244602e-01]
|
[11.097826957702637, 0.620402455329895]
|
e5c5c80b-a2b8-437f-98ba-7267b0d4eb77
|
neural-morphological-disambiguation-using
| null | null |
https://aclanthology.org/W17-7559
|
https://aclanthology.org/W17-7559.pdf
|
Neural Morphological Disambiguation Using Surface and Contextual Morphological Awareness
| null |
['Anil Kumar Singh', 'Akhilesh Sudhakar']
|
2017-12-01
| null | null | null |
ws-2017-12
|
['morphological-disambiguation']
|
['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]
|
[-7.244236469268799, 3.7718560695648193]
|
d625667b-e0d8-4fdd-b326-6e21a842ff66
|
the-dual-information-bottleneck-1
|
2006.04641
| null |
https://arxiv.org/abs/2006.04641v1
|
https://arxiv.org/pdf/2006.04641v1.pdf
|
The Dual Information Bottleneck
|
The Information Bottleneck (IB) framework is a general characterization of optimal representations obtained using a principled approach for balancing accuracy and complexity. Here we present a new framework, the Dual Information Bottleneck (dualIB), which resolves some of the known drawbacks of the IB. We provide a theoretical analysis of the dualIB framework; (i) solving for the structure of its solutions (ii) unraveling its superiority in optimizing the mean prediction error exponent and (iii) demonstrating its ability to preserve exponential forms of the original distribution. To approach large scale problems, we present a novel variational formulation of the dualIB for Deep Neural Networks. In experiments on several data-sets, we compare it to a variational form of the IB. This exposes superior Information Plane properties of the dualIB and its potential in improvement of the error.
|
['Ravid Shwartz-Ziv', 'Zoe Piran', 'Naftali Tishby']
|
2020-06-08
| null |
https://openreview.net/forum?id=B1xZD1rtPr
|
https://openreview.net/pdf?id=B1xZD1rtPr
| null |
['information-plane']
|
['methodology']
|
[ 3.30282331e-01 3.80938053e-01 -2.33957991e-01 -9.28647295e-02
-8.71551394e-01 -4.62621629e-01 3.18380564e-01 1.68933034e-01
-4.80115294e-01 8.71023715e-01 2.02629223e-01 -4.07264352e-01
-7.91437864e-01 -3.78053069e-01 -8.99978936e-01 -8.98134112e-01
-4.62843990e-03 5.53602934e-01 1.12397790e-01 -1.25941291e-01
5.59693933e-01 5.62002480e-01 -1.59292400e+00 2.18579546e-01
7.76205778e-01 1.44190216e+00 3.22619945e-01 5.22372901e-01
-1.04460254e-01 6.90642178e-01 -3.49842429e-01 -4.28074151e-01
4.42111015e-01 -3.36818069e-01 -1.01711428e+00 -3.07768226e-01
4.94552076e-01 -2.76536793e-01 -3.01593542e-01 1.25456548e+00
3.48230034e-01 1.54555440e-01 9.09146905e-01 -1.44749546e+00
-5.62420487e-01 4.16952521e-01 -5.62052011e-01 6.17388129e-01
1.81195736e-02 -5.23221195e-01 1.53009641e+00 -6.57581568e-01
5.15555978e-01 1.16401803e+00 9.63840604e-01 7.01467693e-01
-1.57799470e+00 -2.44968817e-01 1.81527242e-01 3.65302682e-01
-1.51979494e+00 -5.61197340e-01 5.97928822e-01 -4.96385545e-01
9.08842921e-01 1.70694023e-01 4.45918053e-01 9.40352023e-01
3.25563759e-01 1.25876200e+00 9.85940099e-01 -5.44291019e-01
2.95521021e-01 4.05335009e-01 6.42431080e-01 6.52102411e-01
5.17056584e-01 2.63983786e-01 -6.70506060e-01 -2.87762552e-01
6.80602491e-01 -3.63130361e-01 -4.60167199e-01 -7.80762196e-01
-6.95157230e-01 9.85904396e-01 3.28811824e-01 7.30764940e-02
-3.05222124e-01 1.39038742e-01 4.38260525e-01 2.66374975e-01
6.02564394e-01 2.45378107e-01 -5.23127496e-01 2.72612292e-02
-9.54891443e-01 4.23597455e-01 8.65069926e-01 9.58210111e-01
7.19752192e-01 -9.19360667e-02 -1.95383951e-01 7.51246929e-01
4.48170155e-01 3.10605794e-01 5.23796141e-01 -1.20047688e+00
4.44019079e-01 2.66725957e-01 7.41197839e-02 -8.91266704e-01
-6.24662042e-01 -5.28887153e-01 -9.33439195e-01 1.23158328e-01
5.44097960e-01 1.29406169e-01 -4.55438823e-01 2.06139016e+00
-8.40780661e-02 -2.66269982e-01 6.67203739e-02 6.59415364e-01
3.12054366e-01 6.61136150e-01 -2.43378893e-01 -4.82129544e-01
8.92590940e-01 -6.78836405e-01 -6.99374139e-01 7.38403527e-03
2.99445659e-01 -1.69612929e-01 7.03614593e-01 4.17763114e-01
-1.38266218e+00 -3.84547770e-01 -1.14192045e+00 -1.95545867e-01
-2.17813924e-01 -1.74371496e-01 4.48941886e-01 7.54471838e-01
-1.44376171e+00 1.09536302e+00 -9.22525465e-01 7.71305151e-03
3.94267529e-01 5.50251901e-01 -2.04247236e-01 2.85353780e-01
-1.06344712e+00 8.36640060e-01 4.20950502e-01 -1.73301119e-02
-6.55230343e-01 -1.18192017e+00 -8.22741151e-01 4.38793272e-01
1.87780127e-01 -8.11607718e-01 1.31671202e+00 -1.08435619e+00
-1.43932652e+00 5.93028963e-01 -2.99383759e-01 -5.31671584e-01
7.21676886e-01 -3.58248681e-01 3.48199815e-01 2.59869009e-01
1.40980410e-03 6.30438149e-01 7.70741165e-01 -1.14172530e+00
-3.42177093e-01 -7.00525701e-01 -1.57996058e-01 -2.50513777e-02
-4.74153757e-01 -4.71181482e-01 -1.36797562e-01 -5.81324458e-01
5.85103109e-02 -7.19412148e-01 -1.85091689e-01 8.64475146e-02
-4.14076626e-01 -2.54228830e-01 2.92109519e-01 -6.08074665e-01
1.30685318e+00 -1.98531389e+00 5.42817056e-01 2.08342075e-01
3.99053007e-01 2.02895045e-01 -1.92766398e-01 4.61178780e-01
-2.59178042e-01 2.89728999e-01 -3.56041074e-01 -4.88928050e-01
2.68420398e-01 3.62879932e-01 -3.99016619e-01 5.19283295e-01
2.05486074e-01 7.39713132e-01 -6.64083302e-01 -3.55690539e-01
-2.01970980e-01 6.52285218e-01 -9.41418886e-01 3.20327766e-02
1.17394671e-01 7.18300715e-02 -3.73854160e-01 9.84825566e-02
8.46531034e-01 -3.17375302e-01 1.85707346e-01 -2.30303198e-01
-7.92701021e-02 2.87591368e-01 -1.24559033e+00 1.41218996e+00
-1.70324400e-01 8.38816583e-01 3.34140599e-01 -1.38490510e+00
5.70921540e-01 1.82971984e-01 7.03678310e-01 -6.61887586e-01
4.48771156e-02 2.00631261e-01 -3.16125631e-01 -2.44384423e-01
4.46523517e-01 -3.49184155e-01 3.45644861e-01 4.45241690e-01
3.45880717e-01 4.68983293e-01 3.05938780e-01 3.54362875e-01
8.80236447e-01 4.00579907e-02 3.50706577e-01 -1.09051335e+00
4.50039506e-01 -4.03435707e-01 4.82301682e-01 1.10428536e+00
-5.12241483e-01 7.13645041e-01 8.47025633e-01 -4.22074705e-01
-1.31480110e+00 -1.23401368e+00 -5.43379009e-01 9.39128280e-01
-8.02102387e-02 -4.16194499e-01 -8.48415434e-01 -5.21193027e-01
1.67003781e-01 5.34787774e-01 -8.58762860e-01 -1.65047899e-01
-1.92981020e-01 -1.01284254e+00 4.19961095e-01 5.54117620e-01
4.92616087e-01 -5.26440978e-01 -7.99255371e-01 -1.78911891e-02
-4.21150476e-01 -8.18222225e-01 -3.42510283e-01 4.27570939e-01
-1.06286597e+00 -9.22159731e-01 -7.22058535e-01 -1.34182066e-01
2.92883784e-01 6.14797436e-02 1.19397318e+00 -1.47391915e-01
-1.28562659e-01 4.77195323e-01 5.34174405e-02 -3.76265079e-01
-4.27171469e-01 3.55547011e-01 1.79019958e-01 -5.55247515e-02
1.14891432e-01 -5.86447835e-01 -5.63571095e-01 8.84010792e-02
-1.00063455e+00 2.78686136e-02 5.68577290e-01 8.81477892e-01
4.90958273e-01 -2.49442339e-01 5.10987341e-01 -7.65804529e-01
7.10733593e-01 -5.51228523e-01 -8.01461160e-01 2.18208969e-01
-1.22713888e+00 6.85185075e-01 2.96300650e-01 -2.08780915e-02
-9.86548901e-01 -1.98104411e-01 -3.42215210e-01 -3.93945754e-01
3.41600537e-01 4.52284604e-01 5.93365878e-02 -6.83519570e-03
5.25837004e-01 1.66530028e-01 3.81998628e-01 -7.07012951e-01
2.08513603e-01 4.67150688e-01 3.05330694e-01 -7.48364687e-01
9.80009660e-02 4.47493523e-01 2.67820299e-01 -8.73212934e-01
-1.13994563e+00 -3.65325540e-01 -8.35208833e-01 3.74209099e-02
5.67474008e-01 -4.58638489e-01 -9.57984805e-01 1.67919904e-01
-1.25171959e+00 -1.64977551e-01 -5.71123004e-01 2.24084988e-01
-1.02004969e+00 2.62084723e-01 -7.03009307e-01 -1.09386027e+00
-2.07973704e-01 -1.14124036e+00 9.13043678e-01 -1.50644584e-02
1.32256672e-01 -1.26706350e+00 2.72096038e-01 1.21847153e-01
4.07537669e-01 2.30556149e-02 1.11094689e+00 -6.99772298e-01
-3.82147938e-01 1.88839570e-01 -5.14752626e-01 5.61215878e-01
-5.38552880e-01 -1.30733037e-02 -1.10934901e+00 -4.72612172e-01
9.49374288e-02 -1.31039888e-01 1.23843062e+00 7.47791648e-01
9.74023998e-01 -4.11763281e-01 -2.28540033e-01 7.72416055e-01
1.78357255e+00 -1.06706820e-01 4.75030035e-01 2.81426370e-01
1.78142115e-01 7.09260046e-01 -9.41647142e-02 6.37558103e-01
1.36350885e-01 7.12115943e-01 2.26609781e-01 2.80608714e-01
-1.14114292e-01 -3.27085942e-01 3.67476702e-01 8.77877295e-01
-1.09726176e-01 1.96773577e-02 -6.61927462e-01 4.86992031e-01
-2.07655001e+00 -9.05443430e-01 -3.21252481e-03 2.20930219e+00
6.19406283e-01 -6.40869513e-02 2.07643628e-01 1.85584977e-01
5.47281325e-01 3.34206931e-02 -8.47755253e-01 -5.73742747e-01
-8.77977461e-02 -3.80925871e-02 5.64216733e-01 7.09872901e-01
-8.67466986e-01 3.68583351e-01 8.54372120e+00 8.45656276e-01
-5.01455903e-01 2.71095961e-01 6.65049374e-01 5.25058806e-03
-3.77094954e-01 -3.95430267e-01 -1.15250063e+00 1.71557546e-01
1.26332438e+00 -4.63223994e-01 5.70150256e-01 7.98720241e-01
-4.92331274e-02 1.36304900e-01 -1.29895687e+00 9.68776941e-01
1.20745401e-03 -1.59303880e+00 2.91921377e-01 4.09890026e-01
5.84601521e-01 3.15943658e-01 1.88332409e-01 2.42103040e-01
1.12959944e-01 -8.12004209e-01 1.05438137e+00 3.59774560e-01
5.66794991e-01 -8.27615380e-01 6.42536104e-01 3.70511055e-01
-8.21452320e-01 -3.70435387e-01 -7.61542618e-01 -6.52105808e-02
-1.20590813e-01 5.52397847e-01 -1.14976257e-01 6.30359769e-01
6.64927602e-01 7.25251913e-01 -4.79557186e-01 7.85181284e-01
2.58658409e-01 1.59038946e-01 -3.88624221e-01 9.85960364e-02
1.02762759e-01 -3.74823749e-01 7.65974164e-01 1.28546178e+00
1.24740601e-01 -3.40457112e-01 -2.29404673e-01 1.25224137e+00
-1.48760406e-02 -1.51357707e-02 -7.17582166e-01 -2.47490853e-02
1.78048670e-01 7.31635153e-01 -4.85078603e-01 -9.63352174e-02
-1.99593067e-01 7.95840800e-01 8.52511346e-01 4.57258075e-01
-4.65104580e-01 -2.99019635e-01 8.80718172e-01 -1.82874769e-01
3.97121757e-01 9.40065160e-02 -4.34291512e-01 -1.12584424e+00
1.07857898e-01 -5.63150704e-01 5.91082931e-01 -5.04621804e-01
-1.28100264e+00 6.03738189e-01 4.90792483e-01 -6.99627697e-01
-2.02755466e-01 -1.10404944e+00 3.32091525e-02 7.35500097e-01
-1.61605930e+00 -4.30648893e-01 1.53831154e-01 5.26723444e-01
2.47088134e-01 6.03246801e-02 7.33235419e-01 3.80678207e-01
-6.80994034e-01 7.24742234e-01 7.59124577e-01 -3.98458183e-01
5.02860807e-02 -1.48036253e+00 2.11108923e-02 5.60017407e-01
1.09347058e-02 6.54730558e-01 9.17530358e-01 -3.30812782e-02
-1.25624812e+00 -5.78156352e-01 7.50927567e-01 -4.59605813e-01
6.45349741e-01 -1.71237722e-01 -9.14753377e-01 7.26775646e-01
3.13959308e-02 -3.41476858e-01 7.02850223e-01 2.11408138e-01
-4.87029225e-01 -2.55424857e-01 -1.00334692e+00 1.59781262e-01
9.58003938e-01 -4.34314579e-01 -5.45869768e-01 2.64844000e-01
5.53153872e-01 4.86590248e-03 -7.18468189e-01 3.24559897e-01
9.30488288e-01 -1.54043746e+00 9.80809450e-01 -7.87015140e-01
5.37233293e-01 3.84107351e-01 -5.03544390e-01 -1.32904196e+00
-2.91447103e-01 -7.52819717e-01 -1.86674789e-01 8.91915739e-01
3.60573173e-01 -7.38546550e-01 6.39089823e-01 6.96560264e-01
-5.51658645e-02 -9.75125015e-01 -1.28782010e+00 -9.66638029e-01
6.27507448e-01 -4.03075993e-01 5.16567230e-02 3.85147899e-01
1.46288231e-01 2.78887808e-01 -1.61289424e-01 -1.16670586e-01
8.50626349e-01 -1.06784686e-01 1.43813923e-01 -1.42763972e+00
-2.94047564e-01 -7.91995347e-01 -4.41006124e-01 -1.34080017e+00
1.77109107e-01 -1.13847744e+00 -8.91827494e-02 -1.21310496e+00
5.54225087e-01 -1.64319992e-01 -6.64219558e-01 5.03675006e-02
5.87961413e-02 -2.71377731e-02 2.48492479e-01 3.09741497e-01
-5.20017624e-01 7.21129060e-01 8.33465219e-01 2.38319323e-03
-3.89060453e-02 -9.15723480e-03 -8.99529696e-01 6.45578563e-01
5.25567830e-01 -6.21675372e-01 -5.48644304e-01 -4.99863446e-01
6.53118193e-01 -7.76933208e-02 2.38909468e-01 -8.95595968e-01
1.06546491e-01 2.79153883e-01 1.54230833e-01 -4.59978521e-01
1.91660181e-01 -5.81712961e-01 -1.49806634e-01 6.13112390e-01
-8.02172840e-01 1.35582149e-01 3.25812876e-01 8.41186881e-01
-1.30678564e-01 -5.86418808e-01 1.13252890e+00 3.22330371e-02
-3.36444438e-01 2.74744689e-01 -3.14626753e-01 4.26403195e-01
6.39324963e-01 -1.17385965e-02 -3.48871291e-01 -3.49539191e-01
-6.51859581e-01 5.35718873e-02 1.13491036e-01 7.75242969e-02
5.79091132e-01 -1.20770419e+00 -6.22449279e-01 3.86528999e-01
-8.68862495e-02 -4.62000847e-01 7.34070465e-02 1.07880914e+00
-3.84291083e-01 6.69746697e-01 -2.48674661e-01 -5.70649445e-01
-9.21034157e-01 7.49188840e-01 4.94099528e-01 -6.04572773e-01
-6.38496220e-01 8.74988794e-01 4.94188011e-01 -2.35662252e-01
4.36330110e-01 -1.05887562e-01 -7.92703480e-02 1.97125107e-01
8.03558886e-01 5.81804872e-01 8.54998678e-02 -4.71369356e-01
-3.29071909e-01 3.82089496e-01 -9.06134993e-02 -3.51031363e-01
1.40161383e+00 -4.00810421e-01 -1.47758499e-01 6.24634802e-01
1.62918949e+00 -4.96521980e-01 -1.60056615e+00 -1.83061332e-01
3.00121814e-01 -2.91294068e-01 4.62077320e-01 -4.06063676e-01
-1.05272830e+00 1.17225301e+00 6.75466597e-01 4.83965337e-01
1.01807511e+00 -1.08298287e-01 4.72908258e-01 6.01756632e-01
2.13346049e-01 -1.03274107e+00 -9.74342600e-02 4.36534077e-01
1.00008702e+00 -9.71018553e-01 -1.07807159e-01 -4.38796654e-02
-4.05597091e-01 1.37140822e+00 2.18598060e-02 -4.23647203e-02
1.05478942e+00 3.35485697e-01 -2.45356396e-01 -1.93325490e-01
-9.02487636e-01 -1.23576894e-01 3.06661665e-01 6.00727022e-01
2.32967243e-01 -2.77102113e-01 -3.57067108e-01 6.66631341e-01
2.72506122e-02 -1.48263291e-01 3.27728182e-01 6.49637222e-01
-4.28495735e-01 -8.17411602e-01 -8.54552537e-02 2.65669912e-01
-6.35419250e-01 -1.36509165e-01 -2.87830114e-01 7.67948031e-01
-3.30398411e-01 8.00722718e-01 1.46215364e-01 -1.18769139e-01
2.86684215e-01 3.20999801e-01 6.70957327e-01 2.37829946e-02
-1.14639796e-01 -2.88358442e-02 -2.10706741e-01 -8.58498633e-01
-4.18410838e-01 -7.99884558e-01 -6.64412320e-01 -3.78474593e-01
-2.93342650e-01 3.04273337e-01 8.90948296e-01 9.50805485e-01
4.99075860e-01 4.85884994e-01 4.49742138e-01 -1.03469884e+00
-1.18391621e+00 -7.95107245e-01 -7.84640133e-01 4.41111267e-01
7.78766513e-01 -8.10185075e-01 -5.64133167e-01 -1.02600522e-01]
|
[7.820838451385498, 3.6699864864349365]
|
fadab1ea-ec54-45da-9be3-9d626cbf2594
|
on-the-effects-of-different-types-of-label
|
2207.13975
| null |
https://arxiv.org/abs/2207.13975v2
|
https://arxiv.org/pdf/2207.13975v2.pdf
|
On the Effects of Different Types of Label Noise in Multi-Label Remote Sensing Image Classification
|
The development of accurate methods for multi-label classification (MLC) of remote sensing (RS) images is one of the most important research topics in RS. To address MLC problems, the use of deep neural networks that require a high number of reliable training images annotated by multiple land-cover class labels (multi-labels) has been found popular in RS. However, collecting such annotations is time-consuming and costly. A common procedure to obtain annotations at zero labeling cost is to rely on thematic products or crowdsourced labels. As a drawback, these procedures come with the risk of label noise that can distort the learning process of the MLC algorithms. In the literature, most label noise robust methods are designed for single-label classification (SLC) problems in computer vision (CV), where each image is annotated by a single label. Unlike SLC, label noise in MLC can be associated with: 1) subtractive label-noise (a land cover class label is not assigned to an image while that class is present in the image); 2) additive label-noise (a land cover class label is assigned to an image although that class is not present in the given image); and 3) mixed label-noise (a combination of both). In this paper, we investigate three different noise robust CV SLC methods and adapt them to be robust for multi-label noise scenarios in RS. During experiments, we study the effects of different types of multi-label noise and evaluate the adapted methods rigorously. To this end, we also introduce a synthetic multi-label noise injection strategy that is more adequate to simulate operational scenarios compared to the uniform label noise injection strategy, in which the labels of absent and present classes are flipped at uniform probability. Further, we study the relevance of different evaluation metrics in MLC problems under noisy multi-labels.
|
['Begüm Demir', 'Mahdyar Ravanbakhsh', 'Tom Burgert']
|
2022-07-28
| null | null | null | null |
['remote-sensing-image-classification']
|
['miscellaneous']
|
[ 5.68923593e-01 -2.88906604e-01 2.35940740e-01 -3.94661784e-01
-8.99520040e-01 -7.39660144e-01 4.99738634e-01 4.66167718e-01
-5.77990353e-01 7.32068837e-01 -4.29821938e-01 -3.21386546e-01
-1.25478029e-01 -1.04987550e+00 -6.61199689e-01 -1.08872652e+00
2.44444251e-01 3.37975502e-01 1.77400425e-01 3.51785980e-02
-2.00411454e-01 6.39818251e-01 -1.97928548e+00 8.40470493e-02
7.78764904e-01 1.00932765e+00 3.89113128e-01 4.57382172e-01
-4.15199548e-01 6.85885072e-01 -7.77028620e-01 -8.75566229e-02
5.14300108e-01 -4.47559625e-01 -6.66544378e-01 1.20110236e-01
4.35750157e-01 8.78670067e-02 6.26603723e-01 1.53898358e+00
5.55523336e-01 5.28308116e-02 8.82314622e-01 -1.18769979e+00
-1.15526691e-01 3.52617353e-01 -7.31670141e-01 -2.47659072e-01
-2.62747318e-01 -5.86898550e-02 7.11720645e-01 -6.76365018e-01
3.40147167e-01 1.20179796e+00 9.26901042e-01 4.21411127e-01
-1.48264813e+00 -6.02951109e-01 1.15492247e-01 -2.83978790e-01
-1.51651752e+00 -1.45319805e-01 5.84415793e-01 -7.07897961e-01
1.99624315e-01 3.99232090e-01 3.23844045e-01 8.03422391e-01
2.71497089e-02 2.77157545e-01 1.64140034e+00 -6.09653056e-01
6.38516128e-01 1.51186571e-01 1.51878297e-01 2.73225516e-01
3.67243707e-01 6.94388971e-02 1.58018678e-01 -2.07086891e-01
3.17989498e-01 -8.86943787e-02 -1.89075738e-01 -2.01903041e-02
-7.95483649e-01 8.65383685e-01 4.37733710e-01 4.40383494e-01
-4.08151865e-01 3.78610015e-01 3.64222318e-01 1.05777048e-01
7.28499532e-01 1.47499755e-01 -3.52242023e-01 3.90090048e-01
-9.74417806e-01 2.05619633e-01 5.85836709e-01 5.88961422e-01
1.32352328e+00 -5.25658280e-02 -3.11319590e-01 1.02253687e+00
3.32616299e-01 9.31756377e-01 1.20357931e-01 -8.98733139e-01
1.64656803e-01 4.54442054e-01 3.05981249e-01 -1.13135815e+00
-5.54709077e-01 -5.99302709e-01 -1.01949489e+00 5.41237652e-01
3.04308891e-01 -2.54625678e-01 -1.11238635e+00 1.66562569e+00
2.88201183e-01 5.80536090e-02 1.43581793e-01 8.19792867e-01
8.53503764e-01 7.01722026e-01 5.36575794e-01 -2.56260097e-01
1.16966486e+00 -5.73186100e-01 -8.94284666e-01 -3.67469549e-01
7.53144264e-01 -6.39129043e-01 9.71764803e-01 1.54663669e-02
-3.18008780e-01 -5.56417048e-01 -8.31625819e-01 3.85914296e-01
-6.81299567e-01 4.37817454e-01 1.29139811e-01 9.13307726e-01
-9.65316236e-01 4.29900885e-01 -4.69868481e-01 -3.28604132e-01
3.40653747e-01 2.56214403e-02 -2.51293331e-01 -1.39267266e-01
-1.15844834e+00 9.08125103e-01 2.28421241e-01 3.57079178e-01
-9.07006085e-01 -3.30903143e-01 -6.99430406e-01 -4.00249362e-02
5.08640945e-01 9.71182883e-02 1.00026107e+00 -1.43996060e+00
-1.00507224e+00 1.01994681e+00 -7.42277652e-02 -5.43719195e-02
5.46747208e-01 8.07928070e-02 -4.71680373e-01 -1.53867960e-01
4.87181067e-01 4.72318649e-01 7.45395184e-01 -1.95817590e+00
-6.70371830e-01 -3.17120552e-01 1.53342066e-02 1.59990758e-01
2.36005738e-01 1.23562314e-01 1.64929122e-01 -6.48983538e-01
3.21870893e-01 -1.02240288e+00 -2.73720741e-01 6.18583895e-02
-2.37946689e-01 -6.09744675e-02 7.72016585e-01 -4.27813977e-01
8.48344266e-01 -2.19938612e+00 -3.66784275e-01 3.55989128e-01
-1.01580746e-01 4.73342031e-01 -1.62417293e-01 1.83137089e-01
-2.25582737e-02 5.58697939e-01 -6.62005842e-01 -3.71724963e-01
-2.51871169e-01 6.00898921e-01 6.69057891e-02 5.35662055e-01
2.20384762e-01 4.95550215e-01 -8.93482268e-01 -3.99999827e-01
2.72315562e-01 4.03036207e-01 1.53161481e-01 -1.74969225e-03
-2.54625618e-01 4.24121588e-01 -2.77570516e-01 6.69157863e-01
1.08124721e+00 -8.30742344e-03 5.16593531e-02 -3.60401928e-01
-2.13562086e-01 -3.07747245e-01 -1.69085836e+00 8.93941224e-01
-6.16346776e-01 3.55888665e-01 2.68996507e-01 -1.12625217e+00
1.14310241e+00 3.54001790e-01 2.71616995e-01 -3.47412109e-01
1.53758779e-01 5.59313178e-01 -4.16822314e-01 -5.79264998e-01
2.91837573e-01 -5.32897472e-01 1.05471514e-01 4.28458989e-01
-1.59313962e-01 -2.11479485e-01 1.14318645e-02 -4.05372590e-01
8.11133564e-01 -4.83056530e-02 2.72419870e-01 -4.72265899e-01
5.91093481e-01 -5.83478017e-03 8.38879049e-01 1.05514359e+00
-4.36968595e-01 6.65620089e-01 2.04950988e-01 -4.82195318e-01
-7.52075374e-01 -7.52800345e-01 -3.38685989e-01 1.02953649e+00
1.85734898e-01 2.84751803e-01 -6.74689651e-01 -6.69518173e-01
-1.64604753e-01 4.75070357e-01 -5.30542791e-01 2.59562671e-01
-2.35060692e-01 -1.23906386e+00 7.55142272e-01 2.30035231e-01
8.82098854e-01 -1.01916003e+00 -6.93900406e-01 2.56965131e-01
-4.58767742e-01 -9.23295677e-01 1.09934442e-01 6.62143350e-01
-4.49689776e-01 -1.21236813e+00 -6.98304594e-01 -7.17362523e-01
6.79267228e-01 2.97796905e-01 9.82989848e-01 1.08285867e-01
-7.87304863e-02 8.86141062e-02 -6.83314145e-01 -4.06339586e-01
-7.08340168e-01 -1.41531423e-01 -1.38869956e-01 4.17688042e-01
3.99158567e-01 -1.05940431e-01 -1.70261964e-01 4.30571407e-01
-1.23993933e+00 -1.66060939e-01 4.02428389e-01 6.43722475e-01
7.27678716e-01 4.36986566e-01 6.90746486e-01 -9.55148757e-01
3.59292239e-01 -4.54493314e-01 -8.10021639e-01 4.81198251e-01
-3.67469609e-01 -1.89631492e-01 3.85890752e-01 -4.31765378e-01
-1.02943766e+00 5.52746773e-01 -2.13350788e-01 8.26740637e-03
-7.20172763e-01 7.03390002e-01 -3.10845554e-01 -2.59752870e-01
9.87693369e-01 -6.31932467e-02 -1.41922235e-01 -4.05935556e-01
2.05174074e-01 1.02621555e+00 2.00268880e-01 -5.00060260e-01
6.50722027e-01 5.20339906e-01 2.96098679e-01 -8.96385908e-01
-1.10778105e+00 -6.19632542e-01 -4.71755624e-01 -6.11319065e-01
1.03148293e+00 -1.06750882e+00 -1.18746839e-01 8.71052921e-01
-1.01603746e+00 -3.79621178e-01 -1.78217158e-01 2.68564492e-01
-1.15237892e-01 2.50696301e-01 -2.57273912e-01 -1.16468716e+00
-1.01183861e-01 -1.47519219e+00 1.12428010e+00 3.38596880e-01
2.58976609e-01 -8.36948097e-01 -1.06328428e-01 5.07964231e-02
5.18772066e-01 5.42286694e-01 7.30388820e-01 -4.93171066e-01
-1.17714562e-01 -1.49575457e-01 -4.72954512e-01 8.28251481e-01
3.37481201e-01 2.81319600e-02 -1.27347922e+00 -1.72771752e-01
7.25049824e-02 -3.83627534e-01 7.41351783e-01 4.55120653e-01
8.41464281e-01 -9.49095655e-03 -1.93631798e-01 1.51841059e-01
1.97812760e+00 2.37603307e-01 4.41618890e-01 3.51914704e-01
6.76706672e-01 8.05603206e-01 5.28139651e-01 1.89492494e-01
3.24412882e-02 5.41504502e-01 6.43769801e-01 -3.77407670e-01
9.60025843e-03 3.02018642e-01 -4.40272614e-02 1.93327144e-01
-1.08525403e-01 -4.99493152e-01 -1.04664814e+00 5.81050754e-01
-1.69688082e+00 -7.31115937e-01 -6.22064710e-01 2.22141862e+00
7.51589656e-01 -2.67080724e-01 -2.15532973e-01 4.51206714e-01
1.21930170e+00 1.48787841e-01 -2.03985229e-01 1.14574078e-02
-5.35266042e-01 5.06121777e-02 8.93099129e-01 5.25600493e-01
-1.60010552e+00 6.61247909e-01 5.52933073e+00 9.61806774e-01
-1.03215611e+00 3.78672838e-01 7.58891582e-01 4.71136302e-01
-1.83150753e-01 -8.08825791e-02 -7.98599124e-01 4.41449106e-01
5.37733078e-01 5.58517754e-01 2.52588689e-01 7.43913710e-01
3.04650545e-01 -6.51763797e-01 -5.48767388e-01 9.02444363e-01
-9.43948999e-02 -8.23288739e-01 -1.78564817e-01 -2.98028179e-02
1.07823288e+00 6.31773099e-02 -2.88806409e-01 7.97160640e-02
4.77787107e-01 -7.99629867e-01 7.62315512e-01 3.59265417e-01
8.05107176e-01 -5.45580924e-01 1.23549449e+00 4.15254951e-01
-1.21014595e+00 -1.80083990e-01 -4.49471682e-01 -4.10701856e-02
6.46507442e-02 1.05618346e+00 -3.48507352e-02 6.68092012e-01
7.79440582e-01 3.32960516e-01 -6.29404366e-01 1.00958657e+00
-3.41274440e-01 5.95606446e-01 -5.54342449e-01 6.06547296e-02
3.66540253e-01 -2.85262406e-01 2.89285362e-01 1.03531528e+00
4.55822438e-01 -1.18351970e-02 4.90834385e-01 8.82536769e-01
1.33631930e-01 1.69700071e-01 -6.41541898e-01 3.77600461e-01
5.34525990e-01 1.27894783e+00 -1.09296238e+00 -1.60862118e-01
-3.45079929e-01 7.26321638e-01 2.59890389e-02 5.84188819e-01
-6.03424788e-01 -1.00002542e-01 3.28529209e-01 1.10124610e-02
-3.29485945e-02 1.13371216e-01 -3.71637493e-01 -6.25225484e-01
7.48645468e-03 -5.81848681e-01 9.46579501e-02 -7.49529302e-01
-1.31414390e+00 5.47992647e-01 2.40682680e-02 -1.23265243e+00
2.53365368e-01 -4.87862766e-01 -2.14888617e-01 1.15398264e+00
-1.63349271e+00 -1.19673097e+00 -7.87390649e-01 1.87666237e-01
1.94676772e-01 1.23497881e-01 9.15961683e-01 6.13076806e-01
-3.24925661e-01 9.72151011e-02 3.69649261e-01 6.33713901e-02
4.58623141e-01 -1.12649703e+00 -9.46208090e-02 8.83476615e-01
-7.40675777e-02 -1.30246710e-02 6.05696559e-01 -7.60996282e-01
-3.08489412e-01 -1.63843584e+00 7.94521809e-01 2.83794496e-02
3.34971935e-01 -2.47277975e-01 -8.89224946e-01 4.37496990e-01
-3.90685946e-01 3.73415291e-01 5.53321183e-01 -3.45500857e-01
-7.18696639e-02 -2.60575145e-01 -1.38403845e+00 1.68737471e-01
4.82142091e-01 -4.98891443e-01 -5.86126037e-02 3.95134717e-01
4.30682003e-01 -6.32926002e-02 -6.00238144e-01 6.90780759e-01
2.20996067e-01 -8.72618794e-01 6.66506171e-01 1.23838693e-01
1.53973088e-01 -8.85565758e-01 -4.64609116e-01 -1.39364457e+00
-1.64558873e-01 2.56328374e-01 7.51917303e-01 1.45291281e+00
3.42215955e-01 -5.22946477e-01 2.96140045e-01 4.03359056e-01
-1.84531081e-02 -1.52272150e-01 -1.05783808e+00 -9.23682690e-01
8.44630152e-02 -4.26848054e-01 5.95127106e-01 1.02818167e+00
-8.29504550e-01 -6.68605119e-02 -3.52827936e-01 4.98384833e-01
5.31696916e-01 -2.25742519e-01 4.82142299e-01 -1.69661105e+00
1.60712823e-01 -1.61234781e-01 -1.50879145e-01 -2.64373243e-01
1.69563040e-01 -6.49787903e-01 6.74977064e-01 -1.69378972e+00
4.08831798e-02 -1.04496825e+00 -2.62837578e-02 7.20013738e-01
-2.06797123e-01 5.12469053e-01 1.80653811e-01 3.06979448e-01
-1.30444884e-01 3.42253417e-01 8.18907797e-01 -2.83182621e-01
-6.33247495e-02 1.36539161e-01 -2.56449610e-01 6.70837104e-01
8.62793684e-01 -8.91132295e-01 -6.42958879e-02 -2.56496370e-01
4.67057705e-01 -2.03734577e-01 5.27707994e-01 -1.22728074e+00
-2.66280062e-02 -1.21446088e-01 -1.07043967e-01 -5.04459023e-01
-2.29960345e-02 -1.29029155e+00 5.59054554e-01 3.77189130e-01
-1.98954552e-01 -3.62930328e-01 1.27167851e-01 5.96182644e-01
-1.64251015e-01 -8.58189940e-01 1.17820203e+00 -3.96669298e-01
-6.85522318e-01 -8.65426809e-02 -8.70207191e-01 -1.74746558e-01
1.03369284e+00 -9.65504125e-02 -3.28211010e-01 -1.27369419e-01
-6.75265133e-01 -3.44427563e-02 4.18231189e-01 9.00911838e-02
4.78560477e-02 -1.24973500e+00 -6.94085062e-01 6.21297918e-02
3.38844985e-01 2.90735722e-01 1.94789767e-01 4.79397655e-01
-6.84133649e-01 -1.56153768e-01 1.13677166e-01 -7.30484486e-01
-1.16528380e+00 3.04411471e-01 7.89664984e-01 -1.45993918e-01
-2.30298787e-01 6.44863009e-01 2.16487069e-02 -9.45817471e-01
1.71445310e-01 -1.87212303e-01 -4.82732326e-01 3.44354779e-01
4.34127122e-01 3.96148443e-01 3.29336435e-01 -1.03147733e+00
-2.13687181e-01 8.48479331e-01 5.50123155e-01 1.80507209e-02
9.67378914e-01 -1.48897752e-01 -3.41370612e-01 7.78907001e-01
8.97147954e-01 -3.01322132e-01 -9.51716781e-01 -2.08729416e-01
9.15481672e-02 -2.20952988e-01 3.11266303e-01 -1.04178536e+00
-1.10362196e+00 5.90345562e-01 1.15197933e+00 2.96636760e-01
1.25560188e+00 -2.66918629e-01 3.63970026e-02 3.07325274e-01
4.62821573e-01 -1.26902330e+00 -2.96550095e-01 2.87652820e-01
6.44299269e-01 -1.62310672e+00 -2.26188436e-01 -5.13698578e-01
-4.67907727e-01 8.67216527e-01 3.40908885e-01 1.85425207e-01
9.47528243e-01 3.46692473e-01 6.81322157e-01 -3.59332114e-01
1.31945997e-01 -5.93500674e-01 -3.83320898e-02 6.97676897e-01
2.07159624e-01 3.41658950e-01 -4.76379305e-01 2.49466762e-01
4.89566028e-01 9.79539379e-02 6.06100202e-01 9.30355847e-01
-5.31820774e-01 -8.69295001e-01 -8.28323007e-01 5.06579161e-01
-5.49390435e-01 -3.05398107e-02 -1.35994330e-01 5.03683031e-01
8.07422459e-01 1.47204232e+00 -8.00890699e-02 -1.68342248e-01
3.08322787e-01 7.55764991e-02 -2.09040076e-01 -6.53518319e-01
-6.45829380e-01 1.87841803e-01 1.77526660e-02 -2.23262876e-01
-1.19109321e+00 -6.42613828e-01 -9.68763411e-01 2.12019458e-02
-6.66650891e-01 1.54885696e-02 9.71404135e-01 1.06609571e+00
-1.90794230e-01 3.84844720e-01 5.15777826e-01 -9.07437444e-01
-3.87345940e-01 -1.19622970e+00 -1.01452565e+00 4.48246270e-01
3.33155245e-01 -9.51775074e-01 -8.61871600e-01 1.58489212e-01]
|
[9.517462730407715, -1.1152726411819458]
|
cce7b537-b898-40c1-a432-cefc0162ab14
|
interpretable-video-captioning-via-trajectory
| null | null |
http://openaccess.thecvf.com/content_cvpr_2018/html/Wu_Interpretable_Video_Captioning_CVPR_2018_paper.html
|
http://openaccess.thecvf.com/content_cvpr_2018/papers/Wu_Interpretable_Video_Captioning_CVPR_2018_paper.pdf
|
Interpretable Video Captioning via Trajectory Structured Localization
|
Automatically describing open-domain videos with natural language are attracting increasing interest in the field of artificial intelligence. Most existing methods simply borrow ideas from image captioning and obtain a compact video representation from an ensemble of global image feature before feeding to an RNN decoder which outputs a sentence of variable length. However, it is not only arduous for the generator to focus on specific salient objects at different time given the global video representation, it is more formidable to capture the fine-grained motion information and the relation between moving instances for more subtle linguistic descriptions. In this paper, we propose a Trajectory Structured Attentional Encoder-Decoder (TSA-ED) neural network framework for more elaborate video captioning which works by integrating local spatial-temporal representation at trajectory level through structured attention mechanism. Our proposed method is based on a LSTM-based encoder-decoder framework, which incorporates an attention modeling scheme to adaptively learn the correlation between sentence structure and the moving objects in videos, and consequently generates more accurate and meticulous statement description in the decoding stage. Experimental results demonstrate that the feature representation and structured attention mechanism based on the trajectory cluster can efficiently obtain the local motion information in the video to help generate a more fine-grained video description, and achieve the state-of-the-art performance on the well-known Charades and MSVD datasets.
|
['Xian Wu', 'Qingxing Cao', 'Liang Lin', 'Qingge Ji', 'Guanbin Li']
|
2018-06-01
| null | null | null |
cvpr-2018-6
|
['video-description']
|
['computer-vision']
|
[ 4.53712717e-02 -3.99292022e-01 -2.74113506e-01 -4.09755737e-01
-6.36936724e-01 -2.65263557e-01 4.63655949e-01 -2.21456766e-01
-1.95176378e-01 6.22270286e-01 6.22905850e-01 4.33488637e-02
6.95351660e-02 -5.52069902e-01 -9.00952220e-01 -6.82304680e-01
-9.75329429e-02 1.17545605e-01 1.94948822e-01 -9.41314846e-02
2.84769148e-01 8.79095569e-02 -1.44135118e+00 6.58295989e-01
6.88966095e-01 9.88504469e-01 9.58113194e-01 7.17098951e-01
-3.26047361e-01 1.22740757e+00 -4.29829121e-01 -1.26485884e-01
-2.66159385e-01 -7.33699799e-01 -6.86046720e-01 5.09920776e-01
1.79461598e-01 -4.44412559e-01 -6.76972270e-01 9.94726479e-01
1.96353793e-01 3.73732239e-01 6.92067742e-01 -9.05472219e-01
-1.10063314e+00 5.21471560e-01 -5.48174977e-01 5.88711023e-01
4.03813422e-01 3.75044167e-01 8.44953001e-01 -7.27876782e-01
6.38604522e-01 1.33362842e+00 1.50547668e-01 5.44624805e-01
-6.64584279e-01 -4.22699630e-01 5.81853628e-01 8.23070645e-01
-1.54828966e+00 -2.58997202e-01 7.35135615e-01 -5.03366768e-01
9.15558398e-01 2.16336176e-01 5.88291049e-01 1.27878654e+00
2.67891824e-01 9.08956170e-01 3.61535490e-01 6.79381937e-03
-2.78537278e-04 -1.34232044e-01 -1.31510466e-01 6.78631067e-01
-1.55045092e-01 -1.35841176e-01 -3.10141921e-01 4.17005837e-01
1.06220913e+00 3.55495542e-01 -5.29123902e-01 -1.11349329e-01
-1.64366543e+00 7.54288256e-01 5.95195889e-01 6.01682305e-01
-6.45538747e-01 3.80814105e-01 7.16695428e-01 -5.43512627e-02
4.25011575e-01 -9.79908928e-03 -2.58547664e-01 -4.30290431e-01
-9.51037645e-01 8.23798180e-02 2.58759826e-01 1.24664843e+00
6.26613975e-01 1.85950175e-01 -5.44986784e-01 4.27529335e-01
2.84002006e-01 4.06252384e-01 8.66795421e-01 -8.82957518e-01
7.68512666e-01 3.88665348e-01 1.46291465e-01 -1.30554914e+00
-1.21397659e-01 -3.92646670e-01 -9.60777700e-01 -5.28033614e-01
-1.30039215e-01 -8.14290834e-04 -7.46626914e-01 1.68970799e+00
-1.42630830e-01 5.66219628e-01 1.76655933e-01 1.44515610e+00
9.14007962e-01 1.42771327e+00 2.47639894e-01 -5.66022873e-01
1.37358379e+00 -1.27297997e+00 -1.04492986e+00 -9.52270105e-02
5.76692343e-01 -2.84284651e-01 1.01646793e+00 -8.12169388e-02
-9.90137458e-01 -9.04076993e-01 -8.48047853e-01 -2.62791395e-01
-1.03850409e-01 1.49953276e-01 2.96956927e-01 -2.13244393e-01
-8.22578609e-01 3.42229366e-01 -7.65875995e-01 -3.69949847e-01
3.74848932e-01 1.55226842e-01 -3.22667360e-01 -2.67393827e-01
-1.28984725e+00 6.38653934e-01 7.78143227e-01 1.83540851e-01
-9.69588578e-01 -2.71389693e-01 -1.18762171e+00 2.65928298e-01
2.86666870e-01 -6.79608345e-01 1.05321014e+00 -1.39821458e+00
-1.14687288e+00 4.03233260e-01 -5.30949295e-01 -4.77001727e-01
1.83325157e-01 4.21459675e-02 -3.91338646e-01 4.84577030e-01
2.19959408e-01 9.50832963e-01 8.80762279e-01 -9.04506981e-01
-6.39740288e-01 3.41159925e-02 9.10117030e-02 4.82039243e-01
-3.17416191e-01 7.88295045e-02 -6.40208900e-01 -8.89037013e-01
-3.49184364e-01 -4.79594976e-01 -9.97047946e-02 -1.52736306e-01
-1.61757305e-01 -3.44192237e-01 1.02669621e+00 -8.82906318e-01
1.48047006e+00 -2.33837724e+00 5.42286456e-01 -6.14942908e-01
-8.95567387e-02 3.49958420e-01 -1.93481445e-01 3.34009260e-01
-7.94687867e-03 3.80242616e-02 -3.11863542e-01 -8.37420002e-02
-2.09044859e-01 3.02160680e-01 -4.82006311e-01 3.05624455e-01
3.70812178e-01 1.10987484e+00 -1.22421479e+00 -6.15148306e-01
3.35491389e-01 5.35800338e-01 -5.85108042e-01 3.96929532e-01
-4.42344785e-01 5.18299878e-01 -7.27039993e-01 2.85988927e-01
3.84236097e-01 -4.72828984e-01 -2.48119220e-01 -4.59455490e-01
-1.43443108e-01 5.67957349e-02 -7.21203685e-01 2.07777739e+00
-3.61597091e-01 8.43172193e-01 -1.85686514e-01 -1.24391460e+00
7.56592035e-01 5.18991053e-01 3.24421674e-01 -5.86716771e-01
9.60999578e-02 -8.34513977e-02 -2.21660301e-01 -1.13613009e+00
3.74106705e-01 -1.32221341e-01 -2.75887281e-01 1.35582224e-01
9.54924822e-02 4.76303816e-01 2.40038037e-01 2.19964191e-01
7.64680207e-01 4.88802940e-01 2.47939289e-01 -1.82175804e-02
8.56102884e-01 1.06471308e-01 5.09459257e-01 1.50770947e-01
-2.14748651e-01 7.98934281e-01 2.03083515e-01 -5.82796633e-01
-1.13329828e+00 -7.38331616e-01 1.79863423e-01 8.93707037e-01
4.80975628e-01 -3.71672630e-01 -7.34997988e-01 -5.50651371e-01
-5.05384862e-01 7.63969898e-01 -5.56591630e-01 -3.43466550e-01
-5.78138351e-01 -2.90025383e-01 1.64478227e-01 6.10392570e-01
9.05633986e-01 -1.36426711e+00 -5.25678337e-01 4.04167742e-01
-7.72422075e-01 -1.26608825e+00 -9.97421503e-01 -3.12360495e-01
-6.28863096e-01 -5.99025905e-01 -1.00574064e+00 -1.20368087e+00
6.69467390e-01 4.63337153e-01 7.51938462e-01 -1.15750976e-01
-4.16191295e-02 1.61458552e-01 -7.11876512e-01 1.27132446e-01
-1.73310131e-01 -5.79158813e-02 -6.47085682e-02 4.39122826e-01
4.49029446e-01 -2.97496110e-01 -5.95395029e-01 2.08814308e-01
-1.08655906e+00 6.03653610e-01 6.32814705e-01 7.65227914e-01
5.66272914e-01 2.03904510e-02 6.30631089e-01 -3.38419557e-01
4.02448863e-01 -9.10656810e-01 -1.67225197e-01 2.27139831e-01
1.47822440e-01 1.89824596e-01 9.45503831e-01 -5.36476493e-01
-1.11003172e+00 9.30760205e-02 -1.14880510e-01 -9.99990523e-01
-2.51361310e-01 5.78799129e-01 -1.90945208e-01 4.11708146e-01
9.89849865e-02 9.85492587e-01 -1.38118461e-01 -3.11509460e-01
4.62574422e-01 7.93970704e-01 6.91163838e-01 -3.18906128e-01
5.72679937e-01 4.02909666e-01 -3.25340331e-01 -7.05975771e-01
-7.66400695e-01 -3.97077739e-01 -6.65817678e-01 -2.20921800e-01
1.47139430e+00 -1.13629580e+00 -5.45702219e-01 1.51298881e-01
-1.62852514e+00 8.49066884e-04 -1.93708204e-02 5.71562111e-01
-8.93596649e-01 5.28197825e-01 -6.51210308e-01 -4.67646122e-01
-1.32023960e-01 -1.31042099e+00 1.19971454e+00 2.28870511e-01
8.80850777e-02 -1.02353179e+00 -2.17724130e-01 2.75418699e-01
2.70082921e-01 2.30980441e-01 6.18010223e-01 -3.42259437e-01
-1.06961846e+00 6.19800463e-02 -4.79125917e-01 8.54761451e-02
1.52599782e-01 -2.94551641e-01 -5.03733099e-01 -1.40843019e-01
1.45181611e-01 -1.89767718e-01 8.76418412e-01 4.73966837e-01
1.41964471e+00 -7.89245844e-01 -2.83725321e-01 7.35381603e-01
1.41969955e+00 3.47897053e-01 7.40421116e-01 2.85834193e-01
9.67168629e-01 5.80145061e-01 8.84624183e-01 3.81942481e-01
5.49508452e-01 7.45131850e-01 5.54426551e-01 1.24047190e-01
-8.23977962e-02 -3.85422170e-01 5.95083654e-01 1.11219776e+00
-1.19582228e-01 -3.18189293e-01 -4.46881473e-01 8.00601721e-01
-2.18304300e+00 -1.52337050e+00 8.21644515e-02 1.73066282e+00
5.43022513e-01 -1.70952380e-01 -2.61508767e-02 -3.36864233e-01
1.05816889e+00 4.24447060e-01 -4.79884058e-01 -3.36025774e-01
-7.35786930e-02 -6.39670610e-01 2.25096956e-01 1.99396938e-01
-9.56298828e-01 9.59557652e-01 4.80782843e+00 1.07037711e+00
-1.20030749e+00 2.58067220e-01 7.79575944e-01 -2.36576408e-01
-2.07574725e-01 -3.17262203e-01 -5.72204351e-01 9.38305259e-01
1.07861328e+00 -3.05483401e-01 4.24520224e-01 6.28596961e-01
7.13544607e-01 3.30889851e-01 -9.75032926e-01 1.31061363e+00
3.04332078e-01 -1.62913084e+00 6.50963366e-01 -2.27685004e-01
6.53984129e-01 -6.18531108e-02 -3.72942351e-02 2.96294898e-01
-3.82192343e-01 -1.04707730e+00 8.78438115e-01 7.73526669e-01
9.26317513e-01 -7.14529634e-01 6.77419960e-01 6.34138703e-01
-1.63918412e+00 -2.62232363e-01 -6.47669375e-01 -2.15805382e-01
5.22596180e-01 -1.68151502e-02 -4.72598881e-01 6.92832649e-01
5.74112058e-01 1.30668890e+00 -2.35555857e-01 9.14570272e-01
1.45845443e-01 4.61749673e-01 2.25056544e-01 -2.78653860e-01
7.95201004e-01 -1.36997417e-01 6.16743207e-01 1.39005458e+00
5.93555987e-01 4.47678566e-01 1.74019620e-01 8.90511513e-01
9.31601301e-02 9.40588266e-02 -6.34939730e-01 -2.74430782e-01
2.66497493e-01 1.02713513e+00 -5.38523316e-01 -5.46202064e-01
-6.86564803e-01 1.24674416e+00 3.20308477e-01 4.86827999e-01
-1.17451441e+00 -3.00976425e-01 4.34507996e-01 7.97593221e-02
8.20651650e-01 -3.59229356e-01 3.52928042e-01 -1.51592946e+00
1.06520504e-01 -5.96402347e-01 3.05232167e-01 -1.17577565e+00
-9.65783417e-01 7.95738637e-01 -1.02649331e-02 -1.61135554e+00
-4.95519042e-01 -3.13551575e-01 -7.32647181e-01 7.81457126e-01
-1.48306632e+00 -1.13574326e+00 -3.73814374e-01 7.02396452e-01
1.35501838e+00 -1.02753714e-01 4.68985677e-01 3.81913453e-01
-6.41726971e-01 1.74288943e-01 1.17054656e-01 1.78949669e-01
3.81435603e-01 -7.42011964e-01 2.46005535e-01 8.00878823e-01
5.54558337e-02 4.03809696e-01 6.60579503e-01 -5.53499460e-01
-1.38685679e+00 -1.60561371e+00 8.03108513e-01 -2.04182446e-01
5.34480989e-01 -1.64056405e-01 -9.65329230e-01 8.00881684e-01
4.32025343e-01 4.50178199e-02 2.36689433e-01 -7.75623083e-01
1.00864701e-01 -3.29123922e-02 -6.68552577e-01 5.81905067e-01
1.10837686e+00 -5.04775226e-01 -7.84425259e-01 3.87831926e-01
1.17119765e+00 -3.03355813e-01 -5.40495872e-01 1.01188995e-01
7.74431229e-02 -6.79896533e-01 8.18775356e-01 -5.53078532e-01
8.35985005e-01 -6.24268115e-01 -2.76566625e-01 -1.10326827e+00
-6.80740595e-01 -5.99960148e-01 -1.60032228e-01 1.29847300e+00
1.66586470e-02 -2.30713021e-02 4.10379052e-01 1.77445605e-01
-3.77075076e-01 -8.36134076e-01 -8.58853757e-01 -6.25667572e-01
-2.97654748e-01 -2.62010992e-01 4.08911496e-01 7.75862455e-01
4.89303060e-02 5.52097499e-01 -7.28779197e-01 2.03445718e-01
2.88043588e-01 2.58070797e-01 3.02206397e-01 -5.96710384e-01
-1.69293612e-01 -3.26134354e-01 -7.40530968e-01 -1.72652292e+00
4.01655614e-01 -9.32170451e-01 1.91598043e-01 -1.77496028e+00
5.49481452e-01 2.81868279e-01 -2.20061988e-01 1.21861309e-01
-2.35273063e-01 1.86362267e-01 3.00463974e-01 4.10544872e-01
-1.20745075e+00 9.32728827e-01 1.54261458e+00 -3.20553243e-01
3.82269844e-02 -4.07681346e-01 -5.67400217e-01 4.83825415e-01
2.72022873e-01 -1.10697918e-01 -5.43366551e-01 -8.49898040e-01
-1.42571524e-01 5.61667144e-01 4.85775411e-01 -1.01202106e+00
2.47780845e-01 -3.32732886e-01 4.28184271e-01 -7.27093577e-01
3.47223043e-01 -8.62842560e-01 1.13518022e-01 4.63173419e-01
-5.06637633e-01 1.33770302e-01 1.71706229e-01 8.57834399e-01
-5.32198727e-01 -4.25395146e-02 4.98602748e-01 -3.83322567e-01
-1.31834567e+00 6.97144747e-01 -4.22906309e-01 -1.11491226e-01
1.32518888e+00 -2.51122475e-01 -1.31202027e-01 -6.31319463e-01
-6.73408210e-01 3.59536946e-01 3.01264316e-01 8.38800192e-01
9.40676808e-01 -1.70565820e+00 -7.92882025e-01 1.09247223e-01
1.41476423e-01 -5.37682604e-03 6.83187306e-01 5.60807705e-01
-5.76826096e-01 8.28469574e-01 -3.06097060e-01 -7.72184491e-01
-8.24421883e-01 9.79405105e-01 2.47913361e-01 1.32557675e-01
-7.56449044e-01 7.53946722e-01 8.87001991e-01 2.83887416e-01
1.89788145e-04 -3.25204879e-01 -5.19445598e-01 -2.01437384e-01
8.77368093e-01 -5.69923520e-02 -5.70109487e-01 -1.26001465e+00
-2.42949903e-01 7.55560577e-01 -4.92566973e-02 1.94211632e-01
1.16818058e+00 -6.83493078e-01 2.80898046e-02 4.76460695e-01
1.59917986e+00 -6.00100577e-01 -1.66709757e+00 -1.93912297e-01
-3.68144721e-01 -4.66052383e-01 6.44057151e-03 -3.35180402e-01
-1.01193416e+00 1.03835475e+00 2.87031442e-01 3.33851241e-02
1.21650755e+00 9.81676728e-02 1.16310263e+00 6.45848885e-02
2.97969282e-01 -6.15680933e-01 3.06876570e-01 4.97449219e-01
1.00970030e+00 -1.07294953e+00 -3.89084607e-01 -8.29750672e-02
-9.57478106e-01 1.32396257e+00 7.41633534e-01 -1.89731121e-01
1.39505535e-01 -2.58161962e-01 -2.44829714e-01 -8.04131627e-02
-9.94932830e-01 -9.10290182e-02 3.60283196e-01 4.90402341e-01
2.31322333e-01 -2.25544065e-01 -2.45752096e-01 9.86477911e-01
2.69436121e-01 7.54237175e-02 4.58389997e-01 4.63901490e-01
-7.26907969e-01 -5.81857264e-01 -2.03469187e-01 2.92832166e-01
-2.95810938e-01 -1.90835625e-01 1.28804252e-01 4.17598397e-01
2.45971292e-01 6.72686219e-01 5.31916916e-01 -3.30450177e-01
-3.87389623e-02 -1.54238462e-01 2.87203372e-01 -6.71141922e-01
3.98598947e-02 1.64688766e-01 -1.48221403e-01 -6.14942670e-01
-6.47724152e-01 -6.59910560e-01 -1.47519135e+00 -9.80242565e-02
-1.90722682e-02 3.44736308e-01 3.39451432e-01 1.21468735e+00
5.92228711e-01 8.61309409e-01 6.15963042e-01 -1.11268675e+00
-2.59264857e-01 -9.76575851e-01 -4.91589278e-01 5.74787021e-01
5.63438118e-01 -5.80658972e-01 -1.48866579e-01 5.90700030e-01]
|
[10.416010856628418, 0.6891830563545227]
|
4cd3614c-f24c-4d6c-8a74-279db0a1bd16
|
empathetic-response-generation-via-emotion
|
2302.11787
| null |
https://arxiv.org/abs/2302.11787v1
|
https://arxiv.org/pdf/2302.11787v1.pdf
|
Empathetic Response Generation via Emotion Cause Transition Graph
|
Empathetic dialogue is a human-like behavior that requires the perception of both affective factors (e.g., emotion status) and cognitive factors (e.g., cause of the emotion). Besides concerning emotion status in early work, the latest approaches study emotion causes in empathetic dialogue. These approaches focus on understanding and duplicating emotion causes in the context to show empathy for the speaker. However, instead of only repeating the contextual causes, the real empathic response often demonstrate a logical and emotion-centered transition from the causes in the context to those in the responses. In this work, we propose an emotion cause transition graph to explicitly model the natural transition of emotion causes between two adjacent turns in empathetic dialogue. With this graph, the concept words of the emotion causes in the next turn can be predicted and used by a specifically designed concept-aware decoder to generate the empathic response. Automatic and human experimental results on the benchmark dataset demonstrate that our method produces more empathetic, coherent, informative, and specific responses than existing models.
|
['Yongbin Li', 'Yuchuan Wu', 'Yuexian Hou', 'Dongming Zhao', 'Ying Zhu', 'Yinhe Zheng', 'Ting-En Lin', 'Bo wang', 'Yushan Qian']
|
2023-02-23
| null | null | null | null |
['response-generation', 'empathetic-response-generation']
|
['natural-language-processing', 'natural-language-processing']
|
[-8.99876356e-02 3.92886758e-01 9.85512063e-02 -8.14976275e-01
-4.67847027e-02 -4.65556026e-01 5.53878367e-01 2.12451935e-01
1.21665239e-01 8.14596415e-01 9.72319901e-01 2.72712827e-01
1.98597237e-01 -5.15983343e-01 -1.20029211e-01 -4.06618625e-01
3.80633563e-01 4.31461215e-01 -5.32039940e-01 -9.35044646e-01
4.89041835e-01 8.53709802e-02 -1.19594526e+00 8.40854824e-01
7.64204502e-01 5.82591891e-01 -6.95988685e-02 5.26209056e-01
-4.27087843e-01 1.71789265e+00 -9.15976524e-01 -7.47442186e-01
-3.11737210e-01 -1.46875668e+00 -1.27063382e+00 -1.44990563e-01
-2.89082080e-01 -2.68863857e-01 1.82414770e-01 8.54512870e-01
3.94254357e-01 4.26020086e-01 6.90815926e-01 -1.41711819e+00
-6.37395918e-01 1.17543912e+00 -2.35223040e-01 -2.38960296e-01
1.28678334e+00 -6.36995733e-02 1.00334692e+00 -6.44467294e-01
7.79341280e-01 1.56112766e+00 6.36327744e-01 9.95564878e-01
-6.74605310e-01 -5.16507030e-01 2.16597156e-03 7.32906997e-01
-7.29581177e-01 -2.39082247e-01 1.14975882e+00 -3.37012321e-01
7.03333616e-01 3.84621382e-01 7.87546754e-01 1.39792001e+00
2.90226042e-01 6.32521272e-01 1.14274704e+00 -3.92826468e-01
2.66216844e-01 2.47050673e-01 1.64547354e-01 1.54539570e-01
-9.03357089e-01 -2.49726951e-01 -8.35436225e-01 -2.49379024e-01
3.96342903e-01 -3.06092650e-01 -2.19828457e-01 1.82299823e-01
-1.10794687e+00 1.10631096e+00 5.73951602e-01 3.25851262e-01
-9.85344410e-01 -8.75780284e-02 6.76100194e-01 3.17792922e-01
1.09252140e-01 8.28073084e-01 -1.78699136e-01 -5.42138934e-01
-3.35182309e-01 3.62643689e-01 1.31347966e+00 8.08013856e-01
4.96861041e-01 -3.34025800e-01 -3.49467486e-01 1.00661600e+00
-3.02184448e-02 1.59368441e-01 4.11062866e-01 -1.20085955e+00
-7.04619065e-02 7.40328372e-01 4.58780438e-01 -1.23969078e+00
-6.08830988e-01 3.42397355e-02 -4.05676037e-01 -3.18364143e-01
3.83694172e-01 -5.86852610e-01 1.41713217e-01 2.16607618e+00
6.60733521e-01 -8.90325457e-02 5.41922688e-01 1.26343012e+00
1.24633646e+00 8.84822965e-01 3.44654322e-01 -3.90932590e-01
1.50740683e+00 -1.07604480e+00 -1.13513720e+00 -3.01815510e-01
8.65744472e-01 -9.19291973e-01 1.13269234e+00 2.82994270e-01
-9.21212196e-01 -2.32746974e-01 -5.59786499e-01 -1.66515738e-01
6.22902885e-02 2.04870637e-04 6.51604474e-01 -4.80685420e-02
-5.96459627e-01 3.15987706e-01 2.33446747e-01 -6.54466689e-01
-3.89928371e-01 -2.30529442e-01 -1.75362781e-01 1.79883078e-01
-1.87159812e+00 1.29925740e+00 1.80172801e-01 6.39404804e-02
-3.10292423e-01 -7.30232477e-01 -7.13008463e-01 1.35241756e-02
-8.61851349e-02 -5.63282967e-01 1.56357753e+00 -1.47716510e+00
-1.85542858e+00 7.96879828e-01 -1.30510613e-01 -2.55259454e-01
2.11557150e-01 -4.10116792e-01 -3.11541200e-01 3.43615770e-01
2.71944907e-02 1.02931368e+00 4.41360772e-01 -1.41331434e+00
-4.71898913e-01 3.83826531e-02 1.93901137e-01 5.94375730e-01
-6.58592805e-02 5.57397425e-01 3.44202191e-01 -3.25626880e-01
-2.45174691e-01 -7.64362395e-01 -7.13682473e-02 -2.53015429e-01
-3.62784207e-01 -5.25083899e-01 4.74055648e-01 -6.48570716e-01
1.05823934e+00 -2.18409967e+00 2.71279305e-01 -1.49592325e-01
1.89912915e-01 -3.13176870e-01 1.18021097e-03 1.15469348e+00
-4.23985392e-01 -3.92937362e-01 7.00510219e-02 -8.20637215e-03
1.84517160e-01 1.94622487e-01 -7.85017252e-01 1.62081942e-01
-1.38508379e-01 6.72092736e-01 -1.14017904e+00 -6.78163886e-01
2.33398098e-02 2.76927173e-01 -7.87695348e-01 7.43763149e-01
-2.52462357e-01 7.05021679e-01 -3.90791506e-01 2.11943891e-02
3.05141687e-01 1.66628584e-01 1.86201110e-01 -1.65131718e-01
1.31851926e-01 4.81469035e-01 -5.01779318e-01 1.38605797e+00
-7.05392241e-01 4.65390772e-01 -1.24729462e-01 -4.83957022e-01
1.38618088e+00 4.77034032e-01 3.23998332e-01 -5.81618845e-01
4.20933455e-01 -3.05615403e-02 3.77017379e-01 -8.74319315e-01
7.17251360e-01 -8.21816266e-01 -6.60480976e-01 9.30678725e-01
-3.60671997e-01 -3.86832327e-01 -1.17771827e-01 5.36644220e-01
7.39533901e-01 6.94938973e-02 6.08295858e-01 4.60436493e-02
6.46553040e-01 3.19808215e-01 5.89545846e-01 3.40822488e-01
-4.74847168e-01 -3.09317023e-03 9.56464112e-01 -6.32603705e-01
-6.39799058e-01 -6.14000738e-01 2.89145380e-01 1.34669733e+00
3.84683400e-01 -4.11072165e-01 -8.53029072e-01 -3.96747410e-01
-4.97741550e-01 1.65918136e+00 -8.52849543e-01 -5.00189185e-01
-5.71695685e-01 -1.05220355e-01 5.52146852e-01 1.98955730e-01
4.36358869e-01 -1.76399076e+00 -8.60160053e-01 4.72597361e-01
-9.32409644e-01 -1.02322221e+00 -3.47903669e-01 -2.80035716e-02
-3.13785195e-01 -7.67108023e-01 -1.04290903e-01 -7.17661083e-01
5.23576379e-01 -8.98778737e-02 1.02295005e+00 3.30983708e-03
1.30659118e-01 4.18696582e-01 -9.43830729e-01 -4.52186823e-01
-9.70616281e-01 -5.78795314e-01 -1.28583819e-01 1.26933843e-01
3.88920903e-01 -5.36974907e-01 -5.72539985e-01 3.21973711e-01
-4.69332129e-01 4.59402025e-01 2.75480421e-03 8.19467306e-01
-1.16084240e-01 -5.59103191e-01 1.06016707e+00 -7.09198058e-01
1.30443907e+00 -9.10774887e-01 6.61032140e-01 2.25747436e-01
-8.93908069e-02 -2.43559375e-01 1.06321895e+00 -6.66553736e-01
-1.57605326e+00 -1.01820283e-01 -1.66114509e-01 -1.35220766e-01
-3.56043667e-01 4.29128915e-01 6.76082671e-02 5.53251088e-01
8.99319708e-01 1.38049439e-01 8.81004855e-02 1.04239643e-01
7.18568861e-01 6.90120995e-01 7.75380790e-01 -9.16093886e-01
7.68864602e-02 1.24481626e-01 -2.91745216e-01 -3.41956526e-01
-1.07035542e+00 -2.67827362e-01 -2.39484370e-01 -9.35583532e-01
8.19090128e-01 -7.23759174e-01 -1.01666391e+00 1.89495608e-01
-1.76121986e+00 -3.38396311e-01 -2.50910491e-01 4.21020091e-01
-9.38323855e-01 3.89659137e-01 -8.48277092e-01 -7.45716929e-01
-4.59059000e-01 -6.90695047e-01 6.68332398e-01 4.89467382e-01
-1.54076552e+00 -1.02441466e+00 1.76233232e-01 4.36048061e-01
1.39730498e-01 2.33923689e-01 1.29878449e+00 -9.92056131e-01
5.58523595e-01 -2.03076139e-01 1.12519257e-01 -5.49024567e-02
2.74827093e-01 -3.00631635e-02 -7.01219380e-01 6.96221948e-01
4.38348860e-01 -7.22788572e-01 -1.02657400e-01 -2.01980844e-01
5.78579545e-01 -9.05978978e-01 6.12516552e-02 7.12813484e-03
6.22931480e-01 4.04910088e-01 8.99229646e-01 -1.02510504e-01
1.65362239e-01 1.45106053e+00 1.11774170e+00 9.28255141e-01
6.35774434e-01 5.21894872e-01 3.82676482e-01 -1.63946338e-02
2.17140555e-01 -4.51939195e-01 5.82448661e-01 6.67858958e-01
2.94340074e-01 -4.27174792e-02 -5.47975957e-01 3.97264212e-01
-1.81773150e+00 -1.55940664e+00 -7.08102763e-01 1.41031599e+00
1.35396338e+00 -6.61270559e-01 1.51666366e-02 -2.49702901e-01
8.85822117e-01 1.98411822e-01 -1.91838846e-01 -1.40222037e+00
-6.03989325e-02 -1.25984490e-01 -7.16646791e-01 7.54403472e-01
-3.56738180e-01 1.32329786e+00 5.51189995e+00 4.57282513e-01
-9.31866109e-01 -1.23307608e-01 5.02709508e-01 -5.25597408e-02
-4.58253056e-01 1.40222907e-01 -1.38055608e-01 2.60556251e-01
4.41270679e-01 -4.01761860e-01 4.38040912e-01 8.83214295e-01
5.96469939e-01 -1.69082925e-01 -1.36896336e+00 9.20111835e-01
1.94604337e-01 -8.09261501e-01 -5.36197312e-02 -6.67270482e-01
3.06360662e-01 -1.05669355e+00 -4.02707100e-01 4.31195408e-01
3.02283674e-01 -8.53616655e-01 8.60015154e-01 6.02444530e-01
1.87266633e-01 -7.30296671e-01 7.90127754e-01 4.20744389e-01
-5.21193981e-01 -5.55181243e-02 -1.53555393e-01 -6.71581507e-01
4.08968717e-01 5.87740354e-02 -1.04460776e+00 2.54852414e-01
3.58066201e-01 6.62002683e-01 5.40865175e-02 4.35807407e-01
-9.43308175e-01 4.93859977e-01 1.29551604e-01 -6.60236537e-01
1.00940131e-01 -2.29609787e-01 6.41835988e-01 1.18577814e+00
1.67112604e-01 7.94706345e-01 -1.46064177e-01 1.04025793e+00
1.31969377e-01 5.16353846e-01 -4.06240076e-01 1.49538100e-01
5.84205031e-01 1.48389554e+00 -2.87044644e-01 -2.36232474e-01
2.07553253e-01 1.18804419e+00 3.46336842e-01 1.00587539e-01
-1.21701252e+00 -1.83273762e-01 5.75804710e-01 -4.04378146e-01
-3.20892543e-01 4.47033107e-01 -3.53945285e-01 -5.48098028e-01
-3.23431134e-01 -1.02804554e+00 3.86599243e-01 -1.47067690e+00
-1.55208218e+00 6.65394783e-01 -2.44960058e-02 -1.04438365e+00
-6.08986735e-01 -4.40649390e-02 -1.41173494e+00 8.39397550e-01
-8.09998035e-01 -9.00023818e-01 -4.27228153e-01 5.23313403e-01
4.44781661e-01 3.77805144e-01 1.18672311e+00 -2.30791584e-01
-3.58441353e-01 3.36594671e-01 -8.94489169e-01 -1.21563282e-02
1.20926929e+00 -8.65405619e-01 -3.58698100e-01 3.54392231e-01
-5.32106757e-01 6.23677075e-01 1.42075467e+00 -6.74330711e-01
-8.97557616e-01 -5.13516724e-01 1.50293207e+00 -2.12360278e-01
8.45126271e-01 2.04600468e-02 -8.88778269e-01 4.09482569e-01
8.43575001e-01 -8.44624758e-01 1.14369047e+00 6.12086765e-02
-4.47467923e-01 3.18389922e-01 -1.31852651e+00 1.00213468e+00
7.99734533e-01 -1.94691509e-01 -1.22997046e+00 4.82629478e-01
7.14060009e-01 -4.14130092e-01 -6.24428749e-01 -1.82345524e-01
4.76526946e-01 -1.16528392e+00 4.63391304e-01 -1.09556389e+00
1.39650500e+00 5.92074096e-02 4.17596959e-02 -1.62652445e+00
-1.49177805e-01 -9.34682846e-01 3.44674379e-01 1.48927462e+00
2.69478738e-01 -3.72720093e-01 2.02923179e-01 1.04204726e+00
-4.39271837e-01 -7.00420916e-01 -7.00511873e-01 -1.16161682e-01
1.08256243e-01 -1.80789694e-01 7.04296231e-01 1.24016666e+00
1.13535392e+00 7.56029665e-01 -7.54346192e-01 -2.38979176e-01
7.88104534e-02 5.71217537e-01 8.34023893e-01 -8.72658432e-01
-1.82158798e-01 -4.84592468e-01 1.14843696e-01 -8.72297764e-01
5.91650546e-01 -8.07379127e-01 5.17043233e-01 -1.74538898e+00
1.85351133e-01 -2.52490014e-01 4.44541276e-01 5.75825214e-01
-4.88818794e-01 -4.48156416e-01 3.94885182e-01 1.21977493e-01
-4.94232893e-01 8.62637162e-01 1.45015264e+00 2.67536342e-01
-3.92423362e-01 -3.67147326e-01 -8.54167879e-01 9.11267281e-01
1.02944636e+00 -5.92698574e-01 -2.84165293e-01 1.82119191e-01
5.39056659e-01 7.15968549e-01 5.29940665e-01 -4.03230518e-01
3.95351291e-01 -6.92231297e-01 -2.93505657e-02 -3.18601727e-01
5.26089251e-01 -6.36579275e-01 2.00854633e-02 4.58771825e-01
-9.36042786e-01 1.50314510e-01 -9.44147557e-02 7.45115653e-02
-4.05073464e-01 -4.31998730e-01 8.40172887e-01 -1.06302395e-01
-6.32362843e-01 -4.62198228e-01 -6.59803391e-01 1.12732567e-01
1.16846263e+00 -1.69779122e-01 -4.02418196e-01 -1.32282758e+00
-6.91063583e-01 3.50954503e-01 2.84983069e-01 5.07048070e-01
7.82929420e-01 -1.32360494e+00 -8.84099782e-01 -6.04275346e-01
2.06611723e-01 -4.61762547e-01 7.90318847e-01 9.54908013e-01
-2.49216929e-01 -1.35186897e-03 -4.91161525e-01 -2.95777861e-02
-1.25430322e+00 4.33372557e-01 5.57586133e-01 -6.07502721e-02
-5.60968637e-01 9.19227004e-01 4.20608729e-01 -4.65469003e-01
5.80647103e-02 1.64504617e-01 -7.45214224e-01 3.65261495e-01
5.20362675e-01 6.88454807e-02 -6.75922871e-01 -7.19310462e-01
-3.35467130e-01 1.57727405e-01 7.66681135e-02 -2.18432441e-01
1.16469157e+00 -2.00384587e-01 -5.73003173e-01 6.48490429e-01
8.54763806e-01 1.58551201e-01 -7.24377990e-01 5.80540672e-02
-1.54866055e-01 -1.02873705e-01 -6.86078012e-01 -1.13306952e+00
-4.95681733e-01 7.74974287e-01 -2.89485395e-01 2.27357879e-01
1.13491011e+00 2.20122755e-01 9.98082936e-01 3.00977230e-01
2.66041040e-01 -1.21908414e+00 3.88537973e-01 7.27256179e-01
1.68208694e+00 -6.39936507e-01 -2.73280531e-01 -5.72349787e-01
-1.48576641e+00 1.37892294e+00 9.83562529e-01 2.25611869e-02
7.56493211e-02 2.30621994e-01 6.38780773e-01 -4.21754509e-01
-1.18926477e+00 3.04931641e-01 -2.59083867e-01 4.80803698e-01
6.55319095e-01 2.06734806e-01 -7.18194008e-01 1.26977146e+00
-9.14481103e-01 -2.83681303e-01 9.53693151e-01 5.36138356e-01
-3.06252658e-01 -8.72997701e-01 -4.62537825e-01 -9.35858339e-02
-1.30126894e-01 -1.04899697e-01 -1.49431884e+00 5.33349991e-01
-5.48106059e-02 1.36985207e+00 -1.04015945e-02 -4.69398618e-01
3.73060197e-01 3.45683873e-01 2.63132066e-01 -5.60544789e-01
-1.35414171e+00 -4.46744114e-01 6.73252702e-01 -5.12608051e-01
-4.82986063e-01 -6.58229053e-01 -2.04210806e+00 -5.37503302e-01
7.65900314e-02 4.34318691e-01 4.78974730e-01 1.19876635e+00
2.25431502e-01 2.67178625e-01 1.07155454e+00 -4.55340385e-01
-5.63009322e-01 -1.17193723e+00 -2.66584873e-01 9.76894259e-01
-3.17913055e-01 -4.34637398e-01 -6.03698611e-01 -2.08518412e-02]
|
[13.14785099029541, 7.629980564117432]
|
9de1fe16-b436-4bf6-b425-c1faf522f23c
|
a-simple-information-based-approach-to
| null | null |
https://openreview.net/forum?id=ULHJwUO0AUx
|
https://openreview.net/pdf?id=ULHJwUO0AUx
|
A Simple Information-Based Approach to Unsupervised Domain-Adaptive Aspect-Based Sentiment Analysis
|
Aspect-based sentiment analysis (ABSA) is a fine-grained sentiment analysis task which aims to extract the aspects from sentences and identify their corresponding sentiments. Aspect term extraction (ATE) is the crucial step for ABSA. Due to the expensive annotation for aspect terms, we often lack labeled target domain data for fine-tuning. To address this problem, many approaches have been proposed recently to transfer common knowledge in an unsupervised way, but such methods have too many modules and require expensive multi-stage preprocessing. In this paper, we propose a simple but effective technique based on mutual information maximization, which can serve as an additional component to enhance any kind of model for cross-domain ABSA and ATE. Furthermore, we provide some analysis of this approach. Experiment results show that our proposed method outperforms the state-of-the-art methods for cross-domain ABSA by 4.32\% Micro-F1 on average over 10 different domain pairs. Apart from that, our method can be extended to other sequence labeling tasks, such as named entity recognition (NER). Codes will be released.
|
['Anonymous']
|
2021-11-16
| null | null | null |
acl-arr-november-2021-11
|
['term-extraction', 'aspect-based-sentiment-analysis']
|
['natural-language-processing', 'natural-language-processing']
|
[ 2.40584329e-01 -1.65456563e-01 -4.21133079e-02 -6.87553525e-01
-1.07582033e+00 -8.27538192e-01 6.66349292e-01 2.02803999e-01
-4.49593753e-01 6.53223574e-01 1.97342187e-02 -2.66175598e-01
6.52876049e-02 -8.38579297e-01 -4.40347582e-01 -6.55179560e-01
4.52625483e-01 4.68807906e-01 3.89281482e-01 -6.20118618e-01
3.55811864e-01 7.60433897e-02 -1.29633546e+00 4.35993284e-01
9.54683006e-01 9.13452685e-01 6.71586916e-02 2.93315798e-01
-8.76173854e-01 4.20870364e-01 -7.60736287e-01 -8.69063795e-01
-1.51452914e-01 -5.42349935e-01 -8.85293663e-01 7.77290463e-02
-2.84639180e-01 3.08802545e-01 4.20039415e-01 1.19700563e+00
5.24456859e-01 -2.78189145e-02 7.79937148e-01 -1.14516377e+00
-4.03366715e-01 7.05380380e-01 -8.43275070e-01 -1.37608320e-01
3.70878458e-01 -2.50062644e-01 1.14726532e+00 -9.18229759e-01
4.43302393e-01 9.80603576e-01 7.04223752e-01 5.85322022e-01
-6.74315035e-01 -6.25302732e-01 3.51591498e-01 -4.02144194e-02
-1.01761019e+00 -2.90982872e-01 1.00055051e+00 -2.53983915e-01
1.03852248e+00 2.23384470e-01 5.79540133e-01 9.79141533e-01
-2.83810981e-02 1.05788326e+00 1.49164987e+00 -4.90269095e-01
2.93514490e-01 5.02376199e-01 4.08736140e-01 3.46308500e-01
3.09294522e-01 -5.79070866e-01 -5.47524929e-01 -3.17556173e-01
1.77731410e-01 -3.17845076e-01 -3.76396589e-02 -1.81171343e-01
-1.09672868e+00 8.53364527e-01 -2.42317155e-01 5.87813914e-01
-2.58142024e-01 -5.38868248e-01 6.04648709e-01 3.33431631e-01
7.15482414e-01 6.18933022e-01 -1.10879230e+00 -2.96439469e-01
-7.58610070e-01 1.39281616e-01 1.16130722e+00 1.12871957e+00
7.43794978e-01 -1.01434946e-01 -9.89850760e-02 1.07197487e+00
3.71747345e-01 5.60061276e-01 7.94479728e-01 -1.65583000e-01
5.65486670e-01 9.70574379e-01 -1.51211051e-02 -8.90040100e-01
-3.30671638e-01 -2.15123966e-01 -7.01835692e-01 -8.90619233e-02
3.06159884e-01 -2.67452389e-01 -9.39307809e-01 1.41261971e+00
4.76440907e-01 -2.69395828e-01 2.62351245e-01 6.50128901e-01
9.08699691e-01 5.85294068e-01 9.56957713e-02 -2.48592034e-01
2.01942015e+00 -1.06142616e+00 -7.84402788e-01 -4.11008298e-01
5.49884439e-01 -1.28443992e+00 1.09145677e+00 4.02345598e-01
-7.46203959e-01 -3.14952821e-01 -1.09351802e+00 2.31295615e-01
-8.60470116e-01 2.56060690e-01 8.73414457e-01 1.01811159e+00
-7.10557997e-01 2.40215376e-01 -6.64801657e-01 -4.27546740e-01
1.68331727e-01 2.67727584e-01 -4.74843532e-01 6.20863736e-02
-1.31644416e+00 6.42403483e-01 3.73046100e-01 -4.95401509e-02
-3.32612693e-01 -4.79992568e-01 -1.01402056e+00 -9.28888395e-02
4.31232750e-01 -7.28468478e-01 1.27805555e+00 -1.02670443e+00
-1.66282856e+00 9.62918758e-01 -4.00381088e-01 -2.11445406e-01
-1.90334674e-02 -4.22086984e-01 -6.82864487e-01 -9.23320800e-02
3.42922896e-01 1.57456324e-01 9.16563690e-01 -1.02085471e+00
-6.70355260e-01 -5.64987242e-01 3.21958125e-01 2.04610422e-01
-6.30247653e-01 4.81352270e-01 -6.07900143e-01 -9.27172542e-01
-1.50874346e-01 -9.24546301e-01 -4.32772577e-01 -6.40703857e-01
-2.79305786e-01 -3.07028979e-01 7.43237436e-01 -5.52761734e-01
1.40138435e+00 -1.98184705e+00 -9.12487656e-02 1.23589598e-01
-1.29307583e-01 3.87389034e-01 -1.06983148e-01 4.46647078e-01
-5.10852337e-02 1.19043134e-01 -6.93632305e-01 -2.94362634e-01
9.77319404e-02 -1.61227837e-01 -1.52283326e-01 -7.61568546e-02
3.84355634e-01 6.55454099e-01 -7.75133491e-01 -5.92022479e-01
-1.46046594e-01 4.69206035e-01 -3.14840764e-01 1.89119130e-01
-3.02627504e-01 3.61608297e-01 -6.68924987e-01 6.90804899e-01
8.28037918e-01 -2.13590592e-01 2.64805168e-01 -2.23940387e-01
-1.13021964e-02 5.25302589e-01 -1.26509881e+00 1.85986841e+00
-6.99738085e-01 2.92275459e-01 -4.49815728e-02 -1.11318612e+00
1.14244533e+00 3.96858275e-01 4.23988998e-01 -4.77026850e-01
3.56919974e-01 2.62230009e-01 -1.71492442e-01 -4.21927124e-01
7.40456522e-01 -4.76651132e-01 -6.62451327e-01 5.04781425e-01
2.14206025e-01 -3.02379876e-01 4.38965172e-01 -6.15339400e-03
9.67233777e-01 1.92587957e-01 7.01075017e-01 -3.20008278e-01
1.07319307e+00 2.49836236e-01 6.90779984e-01 2.61557966e-01
-1.73686936e-01 5.23606479e-01 3.82719189e-01 -2.65600026e-01
-8.86395335e-01 -4.85527933e-01 -9.19870734e-02 9.75056231e-01
6.37395978e-02 -6.93288445e-01 -9.05306041e-01 -1.13355935e+00
-4.65285987e-01 6.05308235e-01 -3.50500375e-01 1.36376247e-01
-3.97690147e-01 -1.17710721e+00 5.61915934e-01 4.32307780e-01
7.12942779e-01 -1.06571114e+00 3.21791694e-02 2.62175769e-01
-3.78694862e-01 -1.34166968e+00 -4.04906005e-01 2.51681030e-01
-7.45975912e-01 -9.21330035e-01 -8.20116341e-01 -1.01090360e+00
5.25737584e-01 1.67691097e-01 1.25201821e+00 -4.12027299e-01
1.05101839e-01 2.41402581e-01 -7.44784355e-01 -5.16536117e-01
-3.13052326e-01 4.69453752e-01 -1.17213614e-01 1.74905926e-01
1.05775416e+00 -5.24512589e-01 -5.89266956e-01 3.79964352e-01
-9.85939980e-01 -2.09911510e-01 1.03135467e+00 7.32762456e-01
8.11319053e-01 2.70244390e-01 7.29705632e-01 -1.53211284e+00
9.47525144e-01 -3.55452478e-01 -5.83617568e-01 3.25458467e-01
-6.81750774e-01 2.93703884e-01 7.56833494e-01 -1.70066938e-01
-1.56236970e+00 2.04750538e-01 -4.69237268e-01 2.85482198e-01
-4.27105010e-01 6.52611256e-01 -6.45304799e-01 1.55161709e-01
3.10786188e-01 4.76212651e-01 -3.72329563e-01 -5.79458594e-01
3.63097131e-01 1.00736499e+00 -1.65855400e-02 -4.89204198e-01
6.67363822e-01 3.73889118e-01 -1.90606222e-01 -8.87838006e-01
-1.11809886e+00 -9.26162362e-01 -5.26628792e-01 1.82318673e-01
8.63523781e-01 -1.06612849e+00 -3.62325668e-01 6.99470818e-01
-1.19636977e+00 2.48231605e-01 -1.05976742e-02 4.19297487e-01
-3.15050572e-01 5.78016579e-01 -4.53972489e-01 -6.12976432e-01
-8.16313744e-01 -1.06981885e+00 1.18802822e+00 3.93068403e-01
-3.67166579e-01 -1.03703845e+00 3.83999914e-01 6.74369812e-01
3.58053207e-01 -1.27356753e-01 7.63278008e-01 -1.02776182e+00
-1.71940356e-01 -1.96434215e-01 -9.34090931e-03 5.28650880e-01
4.00458246e-01 -2.64526218e-01 -1.11150622e+00 2.04612464e-02
3.37915272e-01 -4.55998816e-02 6.09566987e-01 1.05908036e-01
7.51315534e-01 -9.58560482e-02 -1.93694398e-01 3.20464551e-01
1.35945106e+00 3.02615434e-01 5.61770558e-01 7.72663891e-01
6.05798662e-01 5.92440724e-01 1.25157797e+00 3.50283206e-01
4.31589127e-01 4.15525168e-01 -6.97491765e-02 3.09378356e-02
1.19400866e-01 2.60453727e-02 5.88774264e-01 1.53762352e+00
2.40487903e-02 -1.43159553e-01 -6.36668324e-01 8.63584161e-01
-1.61597061e+00 -7.14509010e-01 -2.90296733e-01 1.92490864e+00
1.13663936e+00 3.15746099e-01 1.24781981e-01 2.13030145e-01
5.78357875e-01 3.77849907e-01 -2.86232740e-01 -7.50985026e-01
-2.54088163e-01 4.28845376e-01 1.82488278e-01 2.15888575e-01
-1.46611226e+00 1.04485476e+00 5.30372715e+00 1.21814275e+00
-7.47868121e-01 1.86631054e-01 4.09806043e-01 4.46744263e-01
-5.78056931e-01 1.04591317e-01 -9.98467088e-01 3.29158783e-01
9.28868651e-01 -1.11477308e-01 -2.16539204e-02 1.21393776e+00
-1.56921461e-01 5.45289041e-03 -6.57980800e-01 8.15405011e-01
2.91907132e-01 -8.37209344e-01 1.22290231e-01 -2.98200935e-01
9.31073010e-01 -2.76701868e-01 -2.52664655e-01 5.81094921e-01
1.56167001e-01 -4.91666853e-01 7.92505294e-02 1.24015309e-01
5.32645345e-01 -1.03394091e+00 1.14584839e+00 1.86938018e-01
-1.49998665e+00 5.33503711e-01 -3.91559660e-01 9.23818499e-02
1.12318873e-01 9.68662798e-01 -5.74901462e-01 9.96870041e-01
7.39834726e-01 6.78188205e-01 -4.34691489e-01 8.50956500e-01
-4.98882979e-01 6.27967536e-01 -5.69690168e-02 -4.88336712e-01
2.01795265e-01 -3.26185316e-01 6.32984281e-01 1.36962712e+00
3.32831115e-01 -1.44726112e-01 -1.25353485e-01 3.71378988e-01
-7.89213032e-02 6.52822554e-01 -5.95767558e-01 -2.81680554e-01
1.21004947e-01 1.67116952e+00 -9.34473455e-01 -4.46129084e-01
-8.17547798e-01 1.22807777e+00 6.21826313e-02 7.63499215e-02
-7.35500753e-01 -9.87189651e-01 8.10749412e-01 -2.33327851e-01
7.21837342e-01 3.17634791e-02 -3.67879242e-01 -1.49216354e+00
3.04327488e-01 -1.14727497e+00 5.25074542e-01 -5.31661212e-01
-1.56782722e+00 1.08897889e+00 -3.31122726e-01 -1.70115530e+00
-2.88784951e-01 -6.63747013e-01 -4.42058891e-01 7.61022568e-01
-1.85258889e+00 -1.24454904e+00 -4.76737022e-02 6.88742578e-01
7.50960231e-01 -2.57223278e-01 9.40291882e-01 4.65658337e-01
-3.39700490e-01 6.24819160e-01 -5.30189686e-02 2.10996240e-01
1.04424345e+00 -1.30347633e+00 4.65274185e-01 9.82965529e-01
1.52463987e-01 7.73673594e-01 5.94859958e-01 -5.77956736e-01
-1.29635942e+00 -1.07593250e+00 1.39459252e+00 -5.46100497e-01
9.02846515e-01 -2.66378462e-01 -7.54260600e-01 5.44381738e-01
5.04855573e-01 -4.78115410e-01 1.18713462e+00 5.04239738e-01
-4.18625772e-01 -2.16956750e-01 -9.84230042e-01 4.37691867e-01
6.66402280e-01 -5.26912808e-01 -9.96497273e-01 -2.93178726e-02
9.21818733e-01 -1.30116358e-01 -9.88780975e-01 4.45978910e-01
3.35491985e-01 -9.12958145e-01 6.47657812e-01 -4.67336446e-01
4.10082608e-01 -6.41833961e-01 -5.15913703e-02 -1.46934426e+00
4.44421507e-02 -4.72033769e-01 2.87688911e-01 1.93554854e+00
7.15138853e-01 -6.69588208e-01 6.49866581e-01 4.89636570e-01
-6.40498400e-02 -5.90948522e-01 -5.50665915e-01 -7.62962341e-01
-9.60280895e-02 -6.65441692e-01 8.56956899e-01 1.05075657e+00
1.91077322e-01 9.17928815e-01 -2.26049796e-01 1.77999020e-01
4.07055527e-01 5.86857259e-01 6.31039441e-01 -9.70924258e-01
-4.67705041e-01 -3.75045151e-01 -3.13948780e-01 -8.69811058e-01
1.10992588e-01 -6.38398588e-01 1.02549858e-01 -1.43891275e+00
3.62591058e-01 -3.58283997e-01 -4.39592689e-01 3.05475414e-01
-5.47067285e-01 1.79870889e-01 -6.19851649e-02 -9.62978750e-02
-8.02402377e-01 7.25004613e-01 1.13646007e+00 -2.27657765e-01
-1.90980405e-01 3.54749054e-01 -1.11137092e+00 1.00742590e+00
9.75229144e-01 -6.40219152e-01 -3.98875237e-01 -2.60900110e-01
4.54443306e-01 -3.93764049e-01 -4.08431292e-01 -7.89294839e-01
7.47129917e-02 -4.69303131e-02 5.80476113e-02 -7.26658702e-01
2.66514748e-01 -9.58520055e-01 -3.20704490e-01 7.86731094e-02
9.48999450e-02 1.52311876e-01 3.36274773e-01 5.05898476e-01
-8.34484637e-01 -4.66195643e-01 3.71489704e-01 -1.80460557e-01
-9.10162508e-01 1.10827737e-01 -4.95188773e-01 3.04188013e-01
8.86241019e-01 2.34579474e-01 -1.99122399e-01 -1.89519987e-01
-4.23957586e-01 -2.55787987e-02 2.88358063e-01 4.25056010e-01
3.96020263e-01 -1.25498903e+00 -4.73115772e-01 9.95056555e-02
5.71924686e-01 -5.18231206e-02 3.35431993e-01 6.91988945e-01
-1.97376251e-01 5.09679437e-01 -2.36641876e-02 -2.70300120e-01
-1.34898078e+00 5.56668162e-01 -2.61775434e-01 -8.44360650e-01
-6.57250509e-02 7.01632857e-01 1.38964549e-01 -8.78175139e-01
-2.75552094e-01 -1.03463173e-01 -6.75525546e-01 3.20331573e-01
6.04392350e-01 -1.51665015e-02 3.65022689e-01 -6.79442883e-01
-6.73719704e-01 8.43557715e-01 -2.52721518e-01 -1.30144641e-01
1.32945275e+00 -2.42725164e-01 -3.88345629e-01 3.88927937e-01
1.15497863e+00 3.62941980e-01 -4.40302223e-01 -1.99201137e-01
1.55091792e-01 -1.67580530e-01 -3.23761523e-01 -7.87409246e-01
-9.86439407e-01 8.48184049e-01 2.32566461e-01 3.26044500e-01
1.45879364e+00 2.81561725e-02 9.65108454e-01 3.60812187e-01
3.13064307e-01 -1.22138417e+00 -2.49355718e-01 6.50736272e-01
4.84257162e-01 -1.45750165e+00 -7.42388219e-02 -6.04551613e-01
-9.37869430e-01 8.55400741e-01 6.36534154e-01 2.70079495e-03
7.62154639e-01 3.00138235e-01 3.89559269e-01 -1.87437579e-01
-5.08589804e-01 -4.88224357e-01 2.80055940e-01 5.41229188e-01
6.41829789e-01 9.56427976e-02 -8.17486584e-01 1.23819435e+00
-2.64772594e-01 -2.04962954e-01 1.94131166e-01 1.03108287e+00
-2.06699401e-01 -1.71160662e+00 -2.34244972e-01 3.30384403e-01
-9.10494328e-01 -4.45839375e-01 -3.71904820e-01 7.74885833e-01
-2.07988381e-01 1.09878993e+00 -4.77187395e-01 -4.01255697e-01
5.21887600e-01 3.27286929e-01 1.62853003e-01 -5.95357358e-01
-7.56968498e-01 2.64125973e-01 4.97875184e-01 -3.48312169e-01
-8.76832008e-01 -6.47002995e-01 -9.69153821e-01 -2.09798012e-02
-4.01241302e-01 5.13155282e-01 8.67577136e-01 1.07009029e+00
4.52477217e-01 5.48762381e-01 6.90568924e-01 -2.01280758e-01
-9.19778496e-02 -1.08770776e+00 -5.80597878e-01 5.25996029e-01
-5.77450357e-02 -3.53412718e-01 -2.84066379e-01 1.02512598e-01]
|
[11.361104965209961, 6.72358512878418]
|
da019ff3-5e92-48c4-b026-7aa2cb7a2880
|
consumer-side-fairness-in-recommender-systems
|
2305.09330
| null |
https://arxiv.org/abs/2305.09330v1
|
https://arxiv.org/pdf/2305.09330v1.pdf
|
Consumer-side Fairness in Recommender Systems: A Systematic Survey of Methods and Evaluation
|
In the current landscape of ever-increasing levels of digitalization, we are facing major challenges pertaining to scalability. Recommender systems have become irreplaceable both for helping users navigate the increasing amounts of data and, conversely, aiding providers in marketing products to interested users. The growing awareness of discrimination in machine learning methods has recently motivated both academia and industry to research how fairness can be ensured in recommender systems. For recommender systems, such issues are well exemplified by occupation recommendation, where biases in historical data may lead to recommender systems relating one gender to lower wages or to the propagation of stereotypes. In particular, consumer-side fairness, which focuses on mitigating discrimination experienced by users of recommender systems, has seen a vast number of diverse approaches for addressing different types of discrimination. The nature of said discrimination depends on the setting and the applied fairness interpretation, of which there are many variations. This survey serves as a systematic overview and discussion of the current research on consumer-side fairness in recommender systems. To that end, a novel taxonomy based on high-level fairness interpretation is proposed and used to categorize the research and their proposed fairness evaluation metrics. Finally, we highlight some suggestions for the future direction of the field.
|
['Helge Langseth', 'Bjørnar Vassøy']
|
2023-05-16
| null | null | null | null |
['marketing']
|
['miscellaneous']
|
[ 2.29858056e-01 2.05369622e-01 -6.44980788e-01 -7.40134537e-01
-1.12857349e-01 -4.50483918e-01 5.03259301e-01 3.25582743e-01
-4.42777395e-01 6.56027257e-01 4.62640166e-01 -5.31373262e-01
-6.13445640e-01 -7.64495373e-01 1.10194094e-01 -2.58661300e-01
4.06626523e-01 2.41223127e-01 -4.67084050e-01 -6.08751476e-01
8.88255239e-01 4.09766763e-01 -1.94599783e+00 1.42726198e-01
1.28972590e+00 9.73071635e-01 -2.83519268e-01 3.45156044e-01
-1.70726761e-01 6.44684255e-01 -4.97105271e-01 -1.30229461e+00
2.04776719e-01 -3.88522297e-01 -6.82540894e-01 -2.71374613e-01
7.85975993e-01 -3.06817830e-01 -1.03738733e-01 1.20171118e+00
4.96495575e-01 3.86029392e-01 6.75068438e-01 -1.49337661e+00
-1.20123518e+00 8.07288229e-01 -4.83445913e-01 2.04117388e-01
2.39490822e-01 -3.30538630e-01 1.20380318e+00 -5.40433288e-01
3.94999981e-01 1.37387347e+00 5.71777701e-01 5.92679799e-01
-1.09058166e+00 -8.52521181e-01 4.47317451e-01 2.82115042e-01
-1.14400601e+00 -6.14568114e-01 2.47968033e-01 -6.14809573e-01
3.96664470e-01 1.03603792e+00 4.73229408e-01 7.72124112e-01
9.86655131e-02 2.29283482e-01 1.36000693e+00 -2.28528574e-01
2.29940563e-01 5.74851453e-01 5.06691635e-01 9.96818617e-02
6.84349775e-01 2.52196252e-01 -4.37565982e-01 -4.15396869e-01
2.66890168e-01 1.63145542e-01 1.52686447e-01 -2.56161541e-01
-2.86318779e-01 1.17660153e+00 3.64617616e-01 1.31448686e-01
-3.57148439e-01 -2.68850386e-01 3.93836051e-01 3.34782928e-01
6.78356707e-01 8.09610903e-01 -6.54928982e-02 -3.01176727e-01
-9.59323108e-01 4.97267514e-01 9.48652744e-01 6.17457032e-01
3.83906871e-01 -7.90957734e-02 -2.94212699e-01 1.11812794e+00
3.33884984e-01 3.40149432e-01 3.81163836e-01 -1.26919889e+00
3.46489847e-02 3.27650279e-01 2.33357430e-01 -1.55397558e+00
-3.29849869e-01 -5.40235460e-01 -6.14468813e-01 3.77175361e-01
6.24669552e-01 1.54918991e-03 -1.98024631e-01 1.60135102e+00
2.10089549e-01 -4.61781830e-01 -3.13390017e-01 1.12293005e+00
5.28093278e-01 1.10399142e-01 3.53192717e-01 -3.83204639e-01
1.25886917e+00 -6.77278817e-01 -7.18601584e-01 -6.19324967e-02
2.66635209e-01 -9.39254701e-01 9.58036780e-01 4.36240137e-01
-9.65286076e-01 -4.77591246e-01 -7.01104045e-01 -1.85387164e-01
-5.38572133e-01 -2.47787163e-01 9.24948514e-01 1.58957386e+00
-6.61327541e-01 9.61970508e-01 -9.89852399e-02 -5.88209569e-01
6.07908964e-01 3.02248597e-01 1.93534285e-01 -1.53778195e-01
-1.36532974e+00 1.11088908e+00 -3.37490320e-01 1.33380601e-02
-8.14109221e-02 -9.63879168e-01 -4.58660930e-01 1.65797234e-01
3.07297140e-01 -5.81533432e-01 1.16657448e+00 -1.32626915e+00
-1.30028665e+00 8.30444217e-01 3.60049784e-01 -1.80215374e-01
6.26180947e-01 -3.94883394e-01 -9.43883717e-01 -6.72420382e-01
2.61038423e-01 2.17813879e-01 5.54985881e-01 -1.05035889e+00
-1.09869611e+00 -6.85581625e-01 3.26620579e-01 5.18617392e-01
-5.98292112e-01 4.81925547e-01 3.44467551e-01 -5.38448691e-01
-3.85738224e-01 -7.52987087e-01 -3.28850508e-01 1.43281026e-02
-9.96180847e-02 -1.35664359e-01 4.86376345e-01 -3.81832480e-01
1.70176804e+00 -2.04833674e+00 -2.25225687e-01 3.35841596e-01
3.44694763e-01 1.87800497e-01 7.84437582e-02 4.69519734e-01
1.69595405e-01 3.89835089e-01 2.64316171e-01 -8.83183070e-03
3.79837781e-01 3.17443870e-02 -3.50741357e-01 5.41778326e-01
-3.62053007e-01 3.36569160e-01 -9.27177250e-01 -1.06837884e-01
5.84468246e-02 4.16506290e-01 -7.67427146e-01 -3.32831442e-02
3.71492058e-01 1.37577534e-01 -2.43231758e-01 5.41301847e-01
7.36093402e-01 1.13840751e-01 3.88318956e-01 6.69890121e-02
-5.10817766e-01 4.45328772e-01 -1.16287541e+00 8.51714909e-01
-3.56722534e-01 2.39444941e-01 1.05467871e-01 -7.71903574e-01
8.11788738e-01 -2.73132443e-01 3.32945973e-01 -9.56672549e-01
1.40845299e-01 3.83225679e-01 4.21109110e-01 -1.58483773e-01
1.24689913e+00 -2.19059512e-01 -1.54113904e-01 5.84734201e-01
-4.78736669e-01 1.08799532e-01 8.28056782e-02 2.53551006e-01
4.90918100e-01 -2.27199331e-01 4.87078369e-01 -6.47492588e-01
4.48914826e-01 -1.13379292e-01 5.84046900e-01 8.34447622e-01
-6.09596670e-01 1.24186695e-01 3.65913570e-01 -3.19910109e-01
-6.29356027e-01 -7.92245388e-01 -3.49770278e-01 1.76213968e+00
3.13571185e-01 -3.39036465e-01 -5.92481911e-01 -6.47328973e-01
5.56467414e-01 1.05299473e+00 -6.56160533e-01 -3.39771748e-01
6.68499470e-02 -5.10565162e-01 4.04584855e-01 1.48510590e-01
1.55335188e-01 -6.62010014e-01 -6.00991607e-01 -9.55329761e-02
-1.20174132e-01 -6.26643181e-01 -6.25942409e-01 -3.99167776e-01
-8.66653442e-01 -9.35077906e-01 -5.06468475e-01 -1.61315128e-01
3.92364442e-01 6.14808381e-01 1.29864049e+00 3.23140591e-01
5.61358184e-02 4.23737019e-01 -3.02854806e-01 -6.30126178e-01
-3.30396652e-01 -1.35650607e-02 3.33673209e-01 -1.09368049e-01
6.39868319e-01 -3.71201396e-01 -7.32043505e-01 6.05864227e-01
-5.56506634e-01 -3.54647964e-01 3.07955772e-01 5.26155055e-01
1.27798140e-01 -1.50979338e-02 9.51653779e-01 -1.60629261e+00
1.20964575e+00 -8.05944860e-01 -2.66686022e-01 -1.21737337e-02
-1.51083183e+00 -5.60229123e-01 5.11400282e-01 -2.24875852e-01
-1.25951242e+00 -6.74369335e-01 -1.54319569e-01 3.18084866e-01
-2.32083984e-02 5.03047407e-01 -2.22692708e-03 -1.14002906e-01
1.01937997e+00 -6.53534651e-01 7.72658810e-02 -3.95916671e-01
6.60368800e-01 1.03183722e+00 2.00726837e-01 -5.58211803e-01
3.71758610e-01 1.95129126e-01 -3.36104244e-01 -7.50883281e-01
-9.72685039e-01 -3.23749334e-01 -3.03219914e-01 -4.94378984e-01
3.83498341e-01 -4.44835633e-01 -7.29081869e-01 -8.18809122e-03
-3.98091793e-01 1.19473888e-02 -4.18555737e-01 3.98784876e-01
-3.15354586e-01 2.77768552e-01 -5.42085707e-01 -1.26662421e+00
-2.06472561e-01 -9.89903033e-01 3.64260256e-01 5.30559123e-01
-6.99723125e-01 -7.97641933e-01 -1.70112744e-01 8.39377880e-01
8.16633403e-01 -1.59021154e-01 8.76167834e-01 -6.94899678e-01
1.18214078e-01 -1.27205595e-01 -4.02857423e-01 2.04570249e-01
1.85637325e-01 -3.83373499e-02 -9.65908110e-01 -6.71804473e-02
-2.26505265e-01 -1.08631887e-01 3.37926984e-01 5.71484685e-01
1.07084548e+00 -3.90281320e-01 -1.09876096e-01 3.99959534e-02
1.09227514e+00 1.80597156e-01 5.66879511e-01 3.00726801e-01
3.71981472e-01 1.09783089e+00 1.15545392e+00 6.83302879e-01
5.49961686e-01 7.42683232e-01 3.18228334e-01 -7.87391365e-02
1.15088053e-01 -1.68556452e-01 2.46852804e-02 5.44462383e-01
-5.06744921e-01 -9.94070084e-04 -3.96540850e-01 -3.50599401e-02
-1.79233539e+00 -1.10544717e+00 -3.31394285e-01 2.59213758e+00
4.66593355e-01 -9.85549986e-02 6.72573924e-01 1.26050413e-01
9.45703626e-01 -9.61844251e-02 -5.83597660e-01 -1.07148182e+00
2.83394456e-01 -1.35704547e-01 5.97901404e-01 6.07092500e-01
-7.96211481e-01 7.85643280e-01 7.02027893e+00 5.17946780e-01
-8.75216126e-01 -1.83512960e-02 9.22216713e-01 -1.23942763e-01
-5.27132213e-01 -2.02655122e-01 -5.60437381e-01 5.28809726e-01
9.23783302e-01 -7.01475680e-01 5.64967632e-01 8.20430934e-01
4.47436243e-01 -1.79144219e-01 -8.02445710e-01 9.19174254e-01
1.47951767e-02 -9.07939553e-01 -1.08146332e-01 5.00122368e-01
6.91103816e-01 -4.25254554e-01 5.46229780e-01 4.93986905e-01
3.59848410e-01 -1.00255752e+00 7.05618918e-01 5.08468747e-01
6.72959328e-01 -1.06648159e+00 5.69439113e-01 1.18476696e-01
-5.02629936e-01 -4.02086169e-01 -6.88853979e-01 -8.99634361e-01
4.40804549e-02 8.25891674e-01 -1.62709951e-02 3.79164636e-01
7.00914621e-01 4.39741373e-01 -2.78492689e-01 1.01372278e+00
2.83994794e-01 4.02069181e-01 2.48484880e-01 -3.22294829e-04
-2.15458676e-01 -6.20351493e-01 6.85461238e-02 9.76536751e-01
4.39434916e-01 2.09443316e-01 -6.26150519e-02 6.50490224e-01
-5.48239574e-02 6.63342655e-01 -5.86600244e-01 -2.65927464e-01
7.70137250e-01 1.57558262e+00 -5.69511831e-01 -1.11969151e-01
-6.11586332e-01 6.54487669e-01 2.03835368e-01 1.00663424e-01
-6.51047707e-01 -2.02459753e-01 1.43361437e+00 4.56407845e-01
-3.84503096e-01 1.38686478e-01 -6.77706122e-01 -7.70535052e-01
-7.24046648e-01 -1.28393042e+00 6.61470890e-01 -2.51254797e-01
-1.61295640e+00 1.19664893e-01 -2.79048890e-01 -8.95272136e-01
1.47089869e-01 -3.97530556e-01 -1.43136516e-01 9.35894251e-01
-1.27997756e+00 -7.78895378e-01 -1.93452939e-01 1.94401905e-01
1.96854040e-01 -1.49000600e-01 9.38202858e-01 7.35140502e-01
-4.43638235e-01 8.16394687e-01 1.29215091e-01 -5.94157815e-01
1.20524192e+00 -1.05352628e+00 8.05737376e-02 5.05546331e-01
-1.04792811e-01 8.90659094e-01 8.73368263e-01 -5.87665558e-01
-1.09107399e+00 -8.27035248e-01 1.05843616e+00 -5.00340879e-01
4.35420960e-01 -9.96994376e-02 -4.57878590e-01 3.11681539e-01
4.01668474e-02 -4.83371258e-01 1.52964830e+00 8.62316966e-01
-4.34875309e-01 -2.90644079e-01 -1.50952303e+00 7.70351291e-01
1.35576260e+00 -4.93012577e-01 -6.75331727e-02 -6.01585023e-04
6.14816174e-02 -2.74138719e-01 -1.07342899e+00 -7.95742422e-02
1.20657432e+00 -1.39469337e+00 7.33178616e-01 -9.31698859e-01
7.65762269e-01 1.59027949e-01 -3.19392323e-01 -1.33049262e+00
-9.77878511e-01 -4.17902946e-01 2.32775778e-01 1.40841937e+00
3.72327596e-01 -7.58689463e-01 7.56490588e-01 1.44282484e+00
-7.60783777e-02 -6.31110728e-01 -2.90813893e-01 -4.68286276e-01
3.07809591e-01 -3.00842106e-01 7.23398268e-01 1.30765522e+00
2.54789054e-01 3.79483283e-01 -6.75130665e-01 -2.22631246e-01
4.79899406e-01 2.51278281e-01 7.42157221e-01 -1.71786988e+00
-6.00638986e-02 -9.72688735e-01 -3.51173609e-01 -7.92152941e-01
-1.16315380e-01 -8.85811210e-01 -3.27159017e-01 -1.21690941e+00
3.48445147e-01 -7.01508999e-01 -5.17764568e-01 -8.85147303e-02
-2.37496495e-01 5.69268525e-01 3.88598204e-01 1.02170736e-01
-4.17806923e-01 -1.81355625e-02 1.21314669e+00 1.63927689e-01
-4.59615327e-02 4.56516147e-01 -2.03605533e+00 5.82693875e-01
8.14360797e-01 -2.76985914e-01 -5.55940747e-01 -5.30869588e-02
7.10118949e-01 -2.89816260e-01 -1.69059515e-01 -5.57630002e-01
-6.74336851e-02 -6.99467123e-01 2.29375094e-01 8.83768499e-02
7.67760798e-02 -8.74365091e-01 2.81002790e-01 3.60249579e-01
-6.73482180e-01 1.35171995e-01 -1.63523212e-01 4.96371001e-01
1.98382095e-01 -3.39305133e-01 8.36110294e-01 1.50572911e-01
-5.69215059e-01 3.28235120e-01 -5.24846435e-01 1.17847130e-01
7.98784375e-01 -3.02837938e-01 -5.02655864e-01 -8.05949152e-01
-5.00047803e-01 1.31401449e-01 4.84341532e-01 7.74187744e-01
1.10094704e-01 -1.42980254e+00 -4.80334818e-01 -1.31339490e-01
2.02100903e-01 -1.15384912e+00 3.57674181e-01 8.60351920e-01
9.35131460e-02 1.86451703e-01 -5.47516406e-01 2.54037172e-01
-1.42381537e+00 4.51356024e-01 1.64557561e-01 1.81369707e-01
3.25852111e-02 7.81831920e-01 1.42854258e-01 -4.80047792e-01
1.35121614e-01 2.09248483e-01 -6.20082498e-01 2.87807643e-01
6.40638530e-01 1.14546216e+00 -9.59911123e-02 -8.92861068e-01
-3.33625644e-01 -2.92687751e-02 -1.82065353e-01 1.36144787e-01
9.88461614e-01 -4.22954977e-01 -1.65059090e-01 3.75262350e-01
6.13758981e-01 5.02781153e-01 -7.37799108e-01 2.09560946e-01
3.53545807e-02 -1.14732277e+00 8.37308690e-02 -1.10626876e+00
-1.10324907e+00 5.27180970e-01 6.84774876e-01 7.23136604e-01
9.62377191e-01 -4.33067352e-01 3.35269362e-01 -8.47652256e-02
3.63540590e-01 -1.60432923e+00 -3.39107245e-01 3.00311655e-01
6.75963640e-01 -1.14254582e+00 2.84777045e-01 -5.58053911e-01
-8.96753311e-01 9.36704695e-01 5.64645529e-01 7.89219290e-02
8.86733711e-01 -1.33625060e-01 3.05649400e-01 -7.75099695e-02
-2.60263085e-01 -1.77199200e-01 3.37252855e-01 8.79605532e-01
1.21641636e+00 5.45134306e-01 -1.17989612e+00 1.01156914e+00
-5.69628179e-01 -3.17528956e-02 5.57825923e-01 3.73940498e-01
-4.67181236e-01 -1.55514157e+00 -1.76793590e-01 1.17608571e+00
-7.29099274e-01 -2.72203772e-03 -7.37931609e-01 4.93991196e-01
2.06882209e-01 1.43889999e+00 -5.55404946e-02 -4.54290181e-01
6.20122015e-01 -3.79958361e-01 3.38703603e-01 -6.38669074e-01
-7.11640894e-01 -2.94309646e-01 6.51149154e-01 -6.81745887e-01
-3.97632718e-01 -8.95807743e-01 -7.47901201e-01 -1.01295757e+00
-5.34210980e-01 3.33148986e-01 5.69225073e-01 7.06011713e-01
3.75229239e-01 3.59904736e-01 5.88469088e-01 -4.85277683e-01
-8.99179697e-01 -6.27302945e-01 -1.10618269e+00 5.07082224e-01
-2.61879176e-01 -9.61861730e-01 -1.28719091e-01 -5.87216198e-01]
|
[9.63028621673584, 5.687957763671875]
|
75973039-8922-444c-b05f-e330da40299a
|
sparse-spatial-transformers-for-few-shot
|
2109.12932
| null |
https://arxiv.org/abs/2109.12932v3
|
https://arxiv.org/pdf/2109.12932v3.pdf
|
Sparse Spatial Transformers for Few-Shot Learning
|
Learning from limited data is challenging because data scarcity leads to a poor generalization of the trained model. A classical global pooled representation will probably lose useful local information. Many few-shot learning methods have recently addressed this challenge using deep descriptors and learning a pixel-level metric. However, using deep descriptors as feature representations may lose image contextual information. Moreover, most of these methods independently address each class in the support set, which cannot sufficiently use discriminative information and task-specific embeddings. In this paper, we propose a novel transformer-based neural network architecture called sparse spatial transformers (SSFormers), which finds task-relevant features and suppresses task-irrelevant features. Particularly, we first divide each input image into several image patches of different sizes to obtain dense local features. These features retain contextual information while expressing local information. Then, a sparse spatial transformer layer is proposed to find spatial correspondence between the query image and the full support set to select task-relevant image patches and suppress task-irrelevant image patches. Finally, we propose using an image patch-matching module to calculate the distance between dense local representations, thus determining which category the query image belongs to in the support set. Extensive experiments on popular few-shot learning benchmarks demonstrate the superiority of our method over state-of-the-art methods. Our source code is available at \url{https://github.com/chenhaoxing/ssformers}.
|
['Chunlin Chen', 'Yaohui Li', 'Huaxiong Li', 'Haoxing Chen']
|
2021-09-27
| null | null | null | null |
['patch-matching']
|
['computer-vision']
|
[ 3.32954377e-01 -3.57817382e-01 -4.76884663e-01 -4.10414428e-01
-8.03545177e-01 -2.82662604e-02 4.03824896e-01 2.19924748e-01
-3.39131415e-01 3.63764316e-01 1.19363144e-01 4.62380707e-01
-3.91524553e-01 -9.50600147e-01 -6.52741611e-01 -9.03636158e-01
2.82301426e-01 5.71412891e-02 5.65995693e-01 -1.12867430e-01
3.23662668e-01 2.65206188e-01 -1.84935677e+00 4.58064675e-01
7.39306569e-01 1.44992113e+00 6.36437774e-01 -1.61497127e-02
-1.46688119e-01 7.05089509e-01 -4.30884033e-01 2.87618518e-01
3.04122686e-01 -5.07868409e-01 -5.33191860e-01 2.40253210e-02
7.41434276e-01 -1.95026860e-01 -5.38747847e-01 1.30358183e+00
5.67489207e-01 5.83383203e-01 5.23890197e-01 -1.05288792e+00
-8.07310402e-01 1.35111347e-01 -5.84340215e-01 6.12985790e-01
6.44798502e-02 1.26368955e-01 9.56716120e-01 -1.26974332e+00
6.11465633e-01 9.13301229e-01 4.03178573e-01 4.08963263e-01
-1.21184969e+00 -7.12318599e-01 1.33416653e-01 5.93809307e-01
-1.79177427e+00 -4.81591254e-01 1.15855670e+00 -3.42683613e-01
8.17885101e-01 2.01729238e-01 5.74427366e-01 8.40745330e-01
1.21281870e-01 8.81553888e-01 9.82501805e-01 -3.17613959e-01
3.32636923e-01 7.19362870e-02 2.14372113e-01 7.37210512e-01
-3.20276469e-02 -6.66830614e-02 -6.50164604e-01 -2.08281074e-02
7.13350952e-01 8.71703804e-01 -4.80477184e-01 -5.87579012e-01
-1.12624943e+00 8.37936640e-01 9.43475127e-01 8.47732306e-01
-4.99227971e-01 -4.15886380e-02 3.87030184e-01 2.91338861e-01
4.21685249e-01 9.74237546e-02 -2.38992870e-01 2.84781873e-01
-1.01332843e+00 1.26041815e-01 1.91390723e-01 8.12216759e-01
1.40606785e+00 -2.53250837e-01 -6.19135797e-01 1.26712978e+00
-6.92710429e-02 2.89671093e-01 7.30928302e-01 -6.94946945e-01
3.43008906e-01 8.36466849e-01 -2.85268933e-01 -1.43325961e+00
-1.26289368e-01 -4.79559571e-01 -1.10582089e+00 -3.74915227e-02
3.81631628e-02 4.24521744e-01 -9.70930398e-01 1.45262802e+00
1.78305134e-01 3.87302488e-01 -3.69674742e-01 1.14857137e+00
1.01630843e+00 7.87917674e-01 1.71372276e-02 -1.22038394e-01
1.25786912e+00 -1.08081746e+00 -3.78828198e-01 -4.11070317e-01
5.05156338e-01 -6.00318551e-01 1.22502089e+00 8.47319886e-03
-7.46351123e-01 -8.90584350e-01 -1.07706976e+00 -1.68122977e-01
-5.51479578e-01 7.41174668e-02 2.40688071e-01 7.27909803e-02
-7.49878228e-01 5.73662758e-01 -5.40462852e-01 -3.36584419e-01
7.28698432e-01 1.34153664e-01 -3.97114784e-01 -7.19228208e-01
-1.20820212e+00 5.68057179e-01 3.46438229e-01 -1.75814688e-01
-9.58255112e-01 -7.23706543e-01 -1.19694543e+00 2.83083856e-01
4.49305773e-01 -3.96676153e-01 8.56901705e-01 -9.54566061e-01
-1.02087510e+00 7.71735311e-01 -3.67938906e-01 -6.97981194e-02
-6.11615479e-02 7.40719438e-02 -2.64187723e-01 2.18484610e-01
5.39543927e-01 5.73608875e-01 1.14808416e+00 -8.74113500e-01
-7.38843143e-01 -5.88038087e-01 -2.15198338e-01 2.15136468e-01
-6.71655476e-01 -1.26173481e-01 -6.97513461e-01 -7.86056519e-01
3.16963583e-01 -4.52277571e-01 -2.31329054e-01 2.90156901e-01
-1.50695041e-01 -4.33984697e-01 9.73847151e-01 -2.46348456e-01
1.31642413e+00 -2.43492293e+00 4.71195281e-02 1.58015087e-01
2.50134230e-01 3.02548975e-01 -4.29725528e-01 2.53245115e-01
-6.27845153e-02 -2.39073128e-01 -2.74629116e-01 -7.15926811e-02
-1.94260240e-01 1.35518715e-01 -2.23137245e-01 5.73465109e-01
2.05001622e-01 9.60854292e-01 -8.69426489e-01 -5.28121769e-01
5.56663573e-01 5.33767104e-01 -2.74843961e-01 1.10450543e-01
8.98037478e-02 1.26284227e-01 -7.41638064e-01 7.74514496e-01
7.47344315e-01 -3.08662951e-01 -3.61540347e-01 -3.82464141e-01
-1.32940382e-01 -1.68265253e-02 -1.03578222e+00 2.04547596e+00
-5.78730524e-01 4.46850270e-01 -1.90992132e-01 -1.48896766e+00
1.14977419e+00 -1.34561405e-01 5.91734350e-01 -1.28430200e+00
1.63904428e-01 2.35486627e-01 -3.81779969e-01 -4.67644781e-01
7.72221833e-02 -1.71776295e-01 -5.31365536e-02 2.62261868e-01
2.90488839e-01 1.05323456e-01 9.42025781e-02 2.01083906e-02
1.11014628e+00 -1.61883950e-01 3.79718572e-01 -2.07008615e-01
6.71722591e-01 -1.30375281e-01 8.41032147e-01 7.27470398e-01
-5.13441086e-01 9.00441706e-01 3.91093642e-02 -7.05067575e-01
-7.24183202e-01 -9.72115040e-01 -2.04911381e-01 1.29896855e+00
5.92865229e-01 -4.82902139e-01 -4.03439999e-01 -7.56220639e-01
6.04799856e-03 4.00955588e-01 -9.13800418e-01 -5.51676571e-01
-5.96396148e-01 -3.77905041e-01 -4.22393493e-02 3.83556813e-01
5.77573955e-01 -1.11606860e+00 -6.61654770e-01 1.30746886e-01
-1.10624060e-01 -6.89629078e-01 -7.11550176e-01 1.93365097e-01
-7.30703473e-01 -1.05382502e+00 -1.16612816e+00 -1.25284898e+00
8.02321255e-01 9.29432809e-01 8.54337513e-01 4.67303582e-02
-5.78360558e-01 4.56335619e-02 -4.73035336e-01 -7.43845338e-03
3.29074562e-01 -1.94951892e-02 -1.97017923e-01 2.52404809e-01
7.04700947e-01 -5.61001897e-01 -1.05015469e+00 4.74314600e-01
-9.51671124e-01 -2.13498577e-01 7.34595776e-01 1.11780035e+00
1.03251755e+00 -3.32495682e-02 3.66116047e-01 -6.35271728e-01
3.43425453e-01 -5.04603744e-01 -1.93805188e-01 2.27908254e-01
-2.67079473e-01 -1.81349721e-02 7.31615424e-01 -2.88712323e-01
-7.80467272e-01 1.56669408e-01 2.05163389e-01 -8.77919972e-01
-2.62608737e-01 4.15280402e-01 -1.34214625e-01 -1.65917680e-01
6.58334672e-01 7.10089326e-01 -1.42266508e-02 -5.31478107e-01
1.73911080e-01 5.63447535e-01 2.29766637e-01 -2.49649361e-01
5.63543200e-01 6.08649433e-01 -2.72242934e-01 -8.43791127e-01
-1.05146301e+00 -8.98261487e-01 -5.72477043e-01 -4.01158235e-04
5.59356451e-01 -8.79369795e-01 -1.07781991e-01 2.22733065e-01
-8.12344849e-01 -9.26055089e-02 -6.08000219e-01 4.06260997e-01
-4.67810333e-01 2.94727832e-01 -3.01920325e-01 -2.71573961e-01
-3.69787931e-01 -1.07376695e+00 1.20603538e+00 4.01759237e-01
6.89921603e-02 -6.52424097e-01 2.05725998e-01 1.75688207e-01
5.25414407e-01 7.10113160e-03 7.46441245e-01 -4.74474430e-01
-4.91312742e-01 -3.67095679e-01 -5.43805897e-01 3.68162245e-01
3.11745405e-01 -5.53346395e-01 -8.73829126e-01 -4.93395627e-01
1.12849645e-01 -4.68228549e-01 1.33569121e+00 5.62459648e-01
1.50122559e+00 -1.26375794e-01 -5.12519300e-01 7.33671010e-01
1.55382144e+00 -3.91753502e-02 5.88169217e-01 2.67171264e-01
7.14971900e-01 4.42962378e-01 8.91518593e-01 4.36779350e-01
1.42783806e-01 7.96077549e-01 1.81416944e-01 -6.11872226e-02
-2.99968541e-01 -2.86053091e-01 1.45707563e-01 7.07722187e-01
2.11415514e-01 2.48344019e-01 -6.75617635e-01 8.08907628e-01
-1.92531538e+00 -1.09348965e+00 4.01488483e-01 2.20859241e+00
6.70568943e-01 -6.40103966e-02 -1.59649268e-01 6.44725654e-03
8.00582170e-01 6.63848042e-01 -6.71404123e-01 6.02763072e-02
-1.01234429e-01 4.13238466e-01 2.73190975e-01 7.20398277e-02
-1.35217428e+00 8.49113703e-01 4.54061222e+00 1.43629992e+00
-1.22679293e+00 3.74160498e-01 7.01634049e-01 -4.41049397e-01
-2.03696162e-01 -2.46974863e-02 -7.39605784e-01 5.23645759e-01
3.78261477e-01 -2.59229153e-01 1.53932333e-01 1.00372255e+00
1.39573529e-01 -8.77948329e-02 -8.94097567e-01 1.39092648e+00
4.97073323e-01 -1.44676828e+00 1.55368403e-01 -1.68130174e-01
8.32254708e-01 1.87434658e-01 1.10585302e-01 5.53123415e-01
-3.72408390e-01 -9.30580437e-01 4.15019870e-01 6.25649571e-01
8.44460666e-01 -7.23952293e-01 5.99016845e-01 3.32921118e-01
-1.45451784e+00 -3.30659658e-01 -9.77064729e-01 3.05433702e-02
-3.11678201e-01 7.38643050e-01 -1.88008174e-01 2.95628637e-01
1.01456726e+00 1.23385656e+00 -6.89814329e-01 1.22550607e+00
-5.73515147e-03 1.92288354e-01 -1.89192295e-01 3.42735238e-02
4.26388294e-01 -9.78629366e-02 3.80104601e-01 1.06363833e+00
3.89138758e-01 9.70466137e-02 3.62293392e-01 9.11018133e-01
-4.99328598e-02 3.88384998e-01 -8.60074103e-01 3.90933454e-01
4.27869380e-01 1.26631474e+00 -6.57900810e-01 -3.89098316e-01
-5.79096735e-01 1.16496384e+00 4.89336580e-01 3.58743668e-01
-4.47455704e-01 -8.07422638e-01 5.57929993e-01 2.02629372e-01
5.72463810e-01 1.05549000e-01 -1.78417619e-02 -1.26618218e+00
2.97702521e-01 -6.84973478e-01 5.73848546e-01 -4.48128134e-01
-1.37266946e+00 3.97113740e-01 -1.59247264e-01 -1.62462044e+00
1.39225572e-01 -2.95544416e-01 -8.19519043e-01 8.42264950e-01
-1.55947745e+00 -1.14212072e+00 -6.99494898e-01 8.56606901e-01
8.70308757e-01 -2.57268280e-01 6.47658885e-01 4.21274036e-01
-5.21212220e-01 5.67156017e-01 3.77333015e-01 8.73354077e-02
8.26305866e-01 -7.57630706e-01 -4.32115309e-02 6.18939817e-01
2.45531872e-01 6.00802302e-01 2.08201051e-01 -4.30581629e-01
-1.28922665e+00 -1.31226158e+00 7.20418096e-01 -8.36388469e-02
4.14218903e-01 -2.67369151e-01 -1.17323959e+00 1.78409576e-01
-7.26159140e-02 7.96058655e-01 6.14968598e-01 -1.48965910e-01
-5.22060275e-01 -6.02149487e-01 -1.00676227e+00 2.38340944e-01
1.01218700e+00 -7.43671298e-01 -5.70911229e-01 3.15664381e-01
4.71235126e-01 1.58843577e-01 -6.28615558e-01 3.46232891e-01
3.38475436e-01 -1.05279493e+00 9.60095584e-01 -2.63989508e-01
3.09300184e-01 -3.71066451e-01 -3.59354049e-01 -1.30322194e+00
-7.26888239e-01 -6.95585534e-02 9.15439650e-02 9.54136431e-01
6.35749474e-03 -3.84329468e-01 9.02237177e-01 1.92627832e-01
-1.81043789e-01 -1.01538086e+00 -1.04006910e+00 -8.17475736e-01
1.43386489e-02 -1.29057273e-01 3.75843614e-01 1.02246177e+00
-7.40031600e-02 2.81159043e-01 -2.34069556e-01 -1.53001115e-01
7.53998399e-01 6.29213631e-01 4.08932179e-01 -1.14711750e+00
-1.39572229e-02 -5.73106229e-01 -7.68658519e-01 -9.59605932e-01
1.41630679e-01 -1.00770497e+00 1.43065676e-01 -1.60063124e+00
5.13964772e-01 -4.26795214e-01 -8.95548582e-01 6.56368673e-01
-1.46354839e-01 5.24538457e-01 1.15500808e-01 4.66673523e-01
-8.77394319e-01 9.63740051e-01 1.23723030e+00 -5.44774592e-01
-1.16754472e-01 -1.87837481e-01 -5.19790411e-01 4.99914110e-01
8.20445538e-01 -5.39514601e-01 -4.26886022e-01 -2.84016103e-01
-2.98378021e-01 -3.58001649e-01 5.37080944e-01 -1.26537490e+00
4.48215485e-01 -2.24250719e-01 6.54281557e-01 -4.31074947e-01
2.68976033e-01 -7.13488400e-01 -2.04160750e-01 5.01173913e-01
-3.38414818e-01 -5.63693106e-01 2.09479686e-02 6.28513277e-01
-6.30510509e-01 -2.80740112e-01 1.04416490e+00 -2.09169328e-01
-1.29891729e+00 7.41281450e-01 1.19450111e-02 9.75925401e-02
1.15590119e+00 -3.41203570e-01 -1.98078305e-01 7.32142329e-02
-5.76953292e-01 7.99344108e-02 4.56964523e-01 6.09895647e-01
1.16724813e+00 -1.56849241e+00 -5.34864008e-01 5.27089298e-01
7.18293905e-01 -1.15657859e-01 8.02889049e-01 8.69722009e-01
-1.40838161e-01 3.84791344e-01 -4.59301591e-01 -6.02657855e-01
-1.17498541e+00 7.94633090e-01 3.63205373e-01 1.48169816e-01
-7.27866292e-01 9.63652909e-01 6.29179120e-01 -1.79727808e-01
1.82304442e-01 -1.50393903e-01 -3.12872738e-01 1.90141991e-01
9.10787404e-01 1.85705587e-01 5.63605167e-02 -8.26412559e-01
-6.03375256e-01 9.76393521e-01 -2.90068597e-01 4.44766879e-01
1.42111433e+00 -8.64588097e-02 -7.83522874e-02 3.34219813e-01
1.76404810e+00 -4.77928817e-01 -1.19468915e+00 -9.03389812e-01
-1.96226865e-01 -8.08987796e-01 4.58585650e-01 -2.49459654e-01
-1.36748719e+00 1.17323768e+00 9.02739942e-01 -1.23260245e-01
1.31715143e+00 2.70209759e-01 8.84539723e-01 3.72975856e-01
2.50668168e-01 -1.17928922e+00 3.33405733e-01 4.16207641e-01
9.83753026e-01 -1.54788744e+00 -1.50168184e-02 -1.19230740e-01
-4.67194885e-01 8.72071505e-01 8.02463949e-01 -4.88273233e-01
9.29731190e-01 -2.80619264e-01 -1.72240689e-01 -4.29032415e-01
-6.17428541e-01 -4.31038797e-01 4.88539338e-01 5.61043739e-01
1.66153595e-01 -7.25867972e-02 -1.89585656e-01 6.73943341e-01
3.67777139e-01 6.19634725e-02 -6.89271614e-02 1.03192163e+00
-8.74544799e-01 -7.61787653e-01 -2.15486154e-01 7.06682682e-01
-5.54798096e-02 -1.48062095e-01 -2.16627479e-01 3.62749636e-01
3.79320353e-01 6.13330245e-01 2.77194738e-01 -5.09646118e-01
5.07228971e-01 -2.42440566e-01 3.05135816e-01 -8.84733737e-01
-2.57772446e-01 6.41805232e-02 -6.39774144e-01 -8.07667673e-01
-2.73544997e-01 -4.64873940e-01 -1.06906831e+00 7.02641830e-02
-6.78627789e-02 1.24428049e-01 2.06214353e-01 6.48789823e-01
5.45664191e-01 3.13434839e-01 8.72252285e-01 -9.92320776e-01
-4.85083759e-01 -8.09789658e-01 -7.28901148e-01 5.46020985e-01
3.85485739e-01 -8.63038301e-01 -3.36364776e-01 -1.65113911e-01]
|
[9.733154296875, 2.076828718185425]
|
9250a5c5-89a1-46ae-b5e4-d4bd9ec762a7
|
macro-action-selection-with-deep
|
1812.00336
| null |
https://arxiv.org/abs/1812.00336v3
|
https://arxiv.org/pdf/1812.00336v3.pdf
|
Macro action selection with deep reinforcement learning in StarCraft
|
StarCraft (SC) is one of the most popular and successful Real Time Strategy (RTS) games. In recent years, SC is also widely accepted as a challenging testbed for AI research because of its enormous state space, partially observed information, multi-agent collaboration, and so on. With the help of annual AIIDE and CIG competitions, a growing number of SC bots are proposed and continuously improved. However, a large gap remains between the top-level bot and the professional human player. One vital reason is that current SC bots mainly rely on predefined rules to select macro actions during their games. These rules are not scalable and efficient enough to cope with the enormous yet partially observed state space in the game. In this paper, we propose a deep reinforcement learning (DRL) framework to improve the selection of macro actions. Our framework is based on the combination of the Ape-X DQN and the Long-Short-Term-Memory (LSTM). We use this framework to build our bot, named as LastOrder. Our evaluation, based on training against all bots from the AIIDE 2017 StarCraft AI competition set, shows that LastOrder achieves an 83% winning rate, outperforming 26 bots in total 28 entrants.
|
['Hongyu Kuang', 'Renjie Hu', 'Huyang Sun', 'Yang Liu', 'Sijia Xu', 'Zhi Zhuang']
|
2018-12-02
| null | null | null | null |
['real-time-strategy-games']
|
['playing-games']
|
[-4.55819815e-01 -3.89155895e-01 -2.09731340e-01 3.45874071e-01
-2.55301982e-01 -6.62826896e-01 7.49264121e-01 -4.01523620e-01
-8.84718657e-01 7.85415351e-01 -2.04329081e-02 -6.36002198e-02
-2.11077303e-01 -7.69903302e-01 -4.15604949e-01 -6.42078340e-01
-3.63063440e-02 9.44290757e-01 9.95568812e-01 -9.08997059e-01
3.48625749e-01 4.05938298e-01 -1.30664957e+00 9.77512896e-02
8.42952609e-01 8.16285789e-01 5.08172452e-01 5.88403165e-01
1.51098475e-01 1.49369192e+00 -9.93323922e-01 -4.59584564e-01
5.28855741e-01 -4.21371669e-01 -7.45932043e-01 -4.07072574e-01
-2.97456861e-01 -5.95910430e-01 -6.59287989e-01 8.33653927e-01
7.50015259e-01 3.12177837e-01 1.43005192e-01 -1.45510423e+00
-6.95977956e-02 1.03372467e+00 -3.41735423e-01 3.36322993e-01
-1.05120637e-01 9.04294670e-01 1.00688791e+00 -1.48330107e-01
8.18255544e-01 1.10114408e+00 3.45951557e-01 7.57723689e-01
-7.13842273e-01 -7.98936546e-01 5.37352450e-02 7.61191547e-01
-9.28211987e-01 -1.42174199e-01 6.62063301e-01 -3.45911294e-01
1.16008830e+00 -3.50738287e-01 9.94354010e-01 1.56862700e+00
4.49996173e-01 9.19520438e-01 9.85399067e-01 1.09177560e-01
4.68559474e-01 -3.92762780e-01 -3.80993605e-01 7.25421369e-01
-2.79799965e-03 5.05605280e-01 -3.32075745e-01 -7.18055591e-02
1.02421761e+00 -5.67183867e-02 3.72785866e-01 -3.08360040e-01
-1.07589161e+00 9.88003492e-01 6.38422966e-01 5.19637525e-01
-7.54797459e-01 4.27526742e-01 5.81895888e-01 4.10738409e-01
-5.07707112e-02 8.95168364e-01 -2.12251067e-01 -8.93852293e-01
-4.62313086e-01 6.40118539e-01 7.57822752e-01 3.50515574e-01
3.41529518e-01 4.91084218e-01 -2.21742883e-01 6.29341006e-01
-4.24435101e-02 4.50621963e-01 5.99280298e-01 -1.39304721e+00
3.98005605e-01 7.86720455e-01 2.02607542e-01 -9.04728293e-01
-5.22684813e-01 -6.62769914e-01 -3.88039738e-01 5.53355396e-01
3.49670202e-01 -4.72884923e-01 -6.69300973e-01 1.87675369e+00
-2.28477623e-02 2.68968612e-01 4.77311164e-02 9.08869147e-01
5.08130252e-01 6.38212323e-01 4.42915484e-02 1.40798077e-01
9.37662899e-01 -1.15362120e+00 -4.83519614e-01 -4.34279740e-01
4.32991534e-01 -2.57447530e-02 7.75533617e-01 5.17350554e-01
-1.03140271e+00 -3.52787167e-01 -7.83733606e-01 5.14852941e-01
-2.76986927e-01 -1.72149524e-01 5.77445090e-01 2.53931612e-01
-9.05275464e-01 7.54654825e-01 -1.00004327e+00 -3.24549288e-01
3.74618322e-01 5.72786808e-01 -5.24676368e-02 3.36333036e-01
-1.34668326e+00 1.25238764e+00 4.74031508e-01 -2.88963109e-01
-1.60737824e+00 -1.96557239e-01 -2.63087004e-01 1.85596999e-02
1.08022594e+00 -2.62836635e-01 1.55588424e+00 -5.85107863e-01
-2.02791142e+00 3.98216248e-01 6.26289666e-01 -9.94597435e-01
6.17209256e-01 -1.07188988e-02 -2.03801170e-01 1.10040180e-01
2.10961193e-01 6.06504440e-01 8.42404485e-01 -9.38852906e-01
-9.82718110e-01 -1.34928823e-01 5.24207950e-01 2.61969000e-01
-2.00178131e-01 1.02981940e-01 -1.46905363e-01 -2.65530169e-01
-6.87222600e-01 -1.12984765e+00 -4.26379800e-01 -7.13764608e-01
-9.79223847e-02 -6.95241988e-01 8.74578595e-01 -4.52143341e-01
1.03326488e+00 -1.98438704e+00 4.74973679e-01 -2.02741921e-01
4.29963589e-01 7.82456696e-01 -3.13220501e-01 6.72479630e-01
5.64440727e-01 -2.13052154e-01 1.27120376e-01 4.65023471e-03
2.95966893e-01 3.69143516e-01 -3.36777568e-01 8.17777961e-02
9.95734893e-03 1.07738066e+00 -1.23811674e+00 -2.96389401e-01
2.56915063e-01 5.86614572e-02 -6.15441561e-01 2.83992380e-01
-5.67188263e-01 7.51884043e-01 -5.99206388e-01 3.57320011e-01
5.22409901e-02 -5.18337898e-02 5.76416999e-02 4.40132976e-01
-4.20622766e-01 4.44550991e-01 -7.46901870e-01 1.52577317e+00
-1.60339117e-01 4.45729226e-01 5.69801442e-02 -9.97231305e-01
8.48978937e-01 3.06026310e-01 7.83654809e-01 -8.88345063e-01
5.82950532e-01 3.13589275e-01 7.50720680e-01 -2.18484700e-01
4.58498687e-01 1.85205087e-01 -2.25098491e-01 5.07392347e-01
2.39197835e-01 -2.29327828e-01 7.70638943e-01 2.78689653e-01
1.58713782e+00 6.39305413e-02 7.67141879e-02 1.68144420e-01
2.67978698e-01 2.38114089e-01 8.67270648e-01 1.02831054e+00
-7.45863199e-01 -1.25462428e-01 8.34847152e-01 -5.70595920e-01
-6.83144212e-01 -7.94501424e-01 7.26798475e-01 1.25813711e+00
2.11783633e-01 -5.04811525e-01 -8.46153975e-01 -8.37559998e-01
-3.37566733e-01 6.50811911e-01 -5.07369697e-01 -4.10062879e-01
-9.04698849e-01 -1.68220580e-01 8.57226968e-01 3.74889493e-01
1.00916398e+00 -1.95198214e+00 -1.32626915e+00 6.90473139e-01
-1.97587073e-01 -1.30198300e+00 -2.15934858e-01 1.97420627e-01
-3.07362705e-01 -1.15131164e+00 -3.50210279e-01 -4.34516340e-01
-2.69571394e-01 1.70119703e-01 8.61355841e-01 -2.79403832e-02
4.83650900e-02 1.39759466e-01 -6.01472199e-01 -4.11933601e-01
-6.23615861e-01 2.86439568e-01 3.01413000e-01 -3.87052178e-01
1.25485823e-01 -5.48812568e-01 -3.48885119e-01 4.28903908e-01
-6.40925288e-01 7.84028843e-02 8.02236915e-01 8.73575985e-01
-4.91098054e-02 3.59186232e-01 5.74258387e-01 -3.81234080e-01
8.39750767e-01 -4.64918524e-01 -8.96502852e-01 1.77797284e-02
-2.13001832e-01 -1.21855780e-01 9.50935304e-01 -5.53151608e-01
-8.54924202e-01 -3.09003443e-01 -1.36916429e-01 -4.08858478e-01
2.29247175e-02 2.72921056e-01 5.53840175e-02 -1.94314107e-01
7.13141739e-01 2.95676351e-01 -3.40550505e-02 -1.07477851e-01
5.86155616e-02 5.97347498e-01 3.63785446e-01 -5.73397279e-01
7.17588127e-01 2.05936626e-01 -7.84471557e-02 -5.18058360e-01
-4.97379720e-01 -7.06256330e-02 -2.45569021e-01 -4.82878745e-01
8.05130899e-01 -5.40557444e-01 -1.24054384e+00 9.13656354e-01
-1.21907389e+00 -1.03549683e+00 -3.81815225e-01 3.17811757e-01
-8.23194385e-01 7.08418563e-02 -7.48898685e-01 -7.19796240e-01
-2.28688642e-01 -1.56450868e+00 4.72928345e-01 4.18379307e-01
1.91864595e-01 -6.43250406e-01 2.83649564e-01 4.47998166e-01
7.66336143e-01 4.77762707e-02 5.66590428e-01 -6.51614428e-01
-4.42311585e-01 -5.61185144e-02 -2.15307157e-02 2.38236696e-01
-6.78376406e-02 -2.14537531e-01 -2.33600631e-01 -2.48088941e-01
-1.13311976e-01 -7.54453838e-01 7.30332673e-01 2.58872420e-01
6.56472504e-01 -1.48419335e-01 -9.40642282e-02 1.51145920e-01
9.76930439e-01 6.43284321e-01 3.63021135e-01 5.75802803e-01
4.50133592e-01 4.09382135e-01 6.39221728e-01 7.05903113e-01
4.80535865e-01 8.49593639e-01 1.00963986e+00 4.79139805e-01
6.26327544e-02 -4.47473317e-01 8.51548731e-01 6.72061563e-01
-5.40135801e-01 -3.28166008e-01 -9.48663473e-01 4.03765291e-01
-2.03889799e+00 -1.30873883e+00 2.67971277e-01 1.82356775e+00
6.13723040e-01 4.50612634e-01 7.62649119e-01 1.07821785e-02
5.90302229e-01 1.34170592e-01 -8.49805117e-01 -4.27187264e-01
-5.92404343e-02 2.65916139e-01 5.58781385e-01 1.63684040e-01
-9.58337009e-01 1.75791037e+00 5.64948845e+00 1.23937845e+00
-1.15357149e+00 1.87626258e-01 1.34672169e-02 -1.09927699e-01
4.93869215e-01 -6.38940111e-02 -7.93528259e-01 6.06496155e-01
9.45110321e-01 -2.69981652e-01 9.82257068e-01 9.31512952e-01
2.81530738e-01 2.11966969e-02 -3.04078937e-01 7.60385633e-01
-2.52456993e-01 -1.37813151e+00 -2.89752722e-01 2.82995790e-01
6.75577462e-01 7.56644964e-01 1.54718474e-01 9.05777991e-01
1.26619995e+00 -9.81681526e-01 7.43146479e-01 1.09941801e-02
2.95047522e-01 -7.45236516e-01 7.34758675e-01 8.18469465e-01
-9.90124047e-01 -7.93504298e-01 -4.27975357e-01 -3.93413782e-01
5.21271490e-02 -1.94088474e-01 -6.63051486e-01 4.44724947e-01
6.09367073e-01 7.53269732e-01 -3.69285047e-01 1.07686770e+00
-6.25091434e-01 6.39746368e-01 -2.80980140e-01 -4.84014601e-01
8.68439674e-01 -1.91295549e-01 8.14032137e-01 5.60711265e-01
5.72491065e-02 1.11879714e-01 5.96892595e-01 7.61792481e-01
1.05367951e-01 -3.61838788e-01 -6.24703825e-01 -4.62367386e-01
4.69294280e-01 1.21956003e+00 -7.09714532e-01 -1.02834426e-01
-4.27677184e-02 6.88564599e-01 5.26662052e-01 6.85128719e-02
-1.05133271e+00 -1.64583176e-01 6.93724990e-01 -2.39681169e-01
3.80433738e-01 -5.57407856e-01 1.91359118e-01 -9.50551391e-01
-5.20885646e-01 -1.30336118e+00 2.17611730e-01 -4.28987026e-01
-1.07778037e+00 8.95210922e-01 -8.70472640e-02 -1.03051615e+00
-5.42354584e-01 -5.85247695e-01 -7.45924234e-01 2.01114103e-01
-1.15960741e+00 -1.03439271e+00 -9.94938612e-02 6.33798301e-01
7.01671720e-01 -8.64909351e-01 5.53755581e-01 -2.33901218e-02
-6.84939861e-01 1.21420652e-01 -1.06324246e-02 2.37683684e-01
2.44672045e-01 -1.09657478e+00 4.20158476e-01 6.95392966e-01
1.17853902e-01 7.08333179e-02 7.04415858e-01 -6.55723035e-01
-1.46293926e+00 -7.11546898e-01 1.18111163e-01 -3.82909656e-01
1.06270647e+00 -2.66104847e-01 -3.78406912e-01 5.23503482e-01
4.28138584e-01 -3.01094681e-01 -7.26660341e-02 -3.08230489e-01
-7.55522475e-02 -8.63827094e-02 -9.29827988e-01 9.40726459e-01
1.04661822e+00 -2.64688712e-02 -4.84928727e-01 1.43138766e-01
5.66588640e-01 -4.26420122e-01 -3.88673544e-01 3.62829596e-01
3.38911533e-01 -1.23035002e+00 5.52629888e-01 -6.25125170e-01
2.85377681e-01 -1.13404296e-01 7.58726448e-02 -1.62789965e+00
-4.31823134e-01 -8.55551064e-01 -1.25657022e-01 7.30232298e-01
-6.97926581e-02 -6.49096489e-01 9.43463743e-01 1.15553237e-01
-1.13663130e-01 -5.79684854e-01 -1.11819696e+00 -1.18918669e+00
5.76139018e-02 -2.42115602e-01 3.89726758e-01 4.73476589e-01
1.84615850e-01 4.41204607e-01 -6.10391080e-01 -4.45750028e-01
5.29438198e-01 -6.69616759e-02 1.10707545e+00 -1.19117975e+00
-6.59437835e-01 -7.66006708e-01 -4.07019347e-01 -1.02682555e+00
3.00006062e-01 -5.31816185e-01 1.44147903e-01 -1.51664829e+00
2.34918185e-02 -4.15624022e-01 -3.70943099e-01 7.30396569e-01
1.46417320e-01 2.15926636e-02 5.35965443e-01 1.73577875e-01
-1.02902901e+00 7.84971952e-01 1.44883668e+00 -6.01493716e-02
-2.28543848e-01 8.46628696e-02 -3.51476461e-01 7.19244242e-01
1.18607247e+00 -4.37758088e-01 -2.91501731e-01 -4.51193154e-01
1.69894308e-01 1.06697612e-01 2.77455479e-01 -1.41521084e+00
4.29365575e-01 -5.93119204e-01 -4.36067343e-01 -2.60800362e-01
6.48866534e-01 -5.87526679e-01 -1.30132154e-01 1.10131311e+00
-5.92724904e-02 4.87967916e-02 8.21411908e-02 3.23516190e-01
-1.39027387e-02 -6.82286993e-02 7.82142937e-01 -3.51911306e-01
-9.51844811e-01 4.58514720e-01 -8.42759788e-01 2.78653443e-01
1.21506393e+00 1.22926891e-01 -6.04403973e-01 -6.42929733e-01
-2.10589767e-01 4.67102468e-01 2.20362797e-01 4.82451677e-01
3.49268734e-01 -9.07434165e-01 -7.60091066e-01 -1.32459298e-01
-2.57592261e-01 -3.51032019e-01 2.24083126e-01 8.91856372e-01
-4.87033129e-01 3.86220992e-01 -8.51591468e-01 -1.06919043e-01
-9.32635725e-01 3.04696798e-01 3.87451589e-01 -9.32155371e-01
-7.46244490e-01 6.31019115e-01 -8.35848525e-02 -4.75149781e-01
2.50182182e-01 2.51626581e-01 -4.50554878e-01 -3.64154041e-01
3.16354573e-01 5.72254717e-01 -3.24448973e-01 -6.31000102e-01
-1.80997863e-01 3.57784405e-02 -1.01252820e-03 -4.26419169e-01
1.58431041e+00 4.98909086e-01 -1.98178794e-02 1.56627759e-01
2.75395483e-01 -3.08390677e-01 -1.52387953e+00 -2.20272705e-01
8.72095302e-02 -2.06641763e-01 -5.18164113e-02 -9.98856246e-01
-1.13857865e+00 7.86274672e-01 2.32303455e-01 4.75467563e-01
6.77438617e-01 -2.28849918e-01 1.08505630e+00 6.66134119e-01
8.76018405e-01 -1.28444660e+00 5.80520749e-01 1.23912024e+00
7.26936221e-01 -9.44976985e-01 -5.08085370e-01 2.90186852e-01
-1.06686628e+00 8.30483973e-01 1.08383381e+00 -6.06996119e-01
1.28140047e-01 1.93629339e-01 -6.26097992e-02 -1.44572228e-01
-1.18570173e+00 -5.49841046e-01 -4.64100868e-01 8.02510917e-01
-3.30647588e-01 3.59149999e-03 -3.15901399e-01 8.20133805e-01
-3.40000987e-01 -4.18318510e-02 6.72434151e-01 8.75043929e-01
-7.10610867e-01 -1.28173172e+00 -1.57400575e-02 3.44415456e-01
-2.89918154e-01 3.11291188e-01 -7.83394158e-01 7.11183369e-01
2.53165960e-01 1.20806336e+00 -2.42365822e-01 -7.68087745e-01
3.25304091e-01 -4.03771788e-01 4.75511551e-01 -4.84931827e-01
-1.08867574e+00 -3.53259623e-01 3.10840234e-02 -7.26483226e-01
-5.89224435e-02 -3.92603606e-01 -1.31902242e+00 -6.88417077e-01
-1.60358027e-01 3.22350562e-01 6.07052505e-01 1.10028636e+00
3.39478463e-01 4.90743786e-01 6.75440788e-01 -8.49736691e-01
-8.78616095e-01 -1.10233164e+00 -5.30339599e-01 1.04697086e-01
-1.58298686e-01 -1.17427552e+00 -1.19923189e-01 -7.60860205e-01]
|
[3.6227059364318848, 1.5493600368499756]
|
416e1f29-d711-4651-b082-8df9e7824843
|
using-positive-matching-contrastive-loss-with
|
2303.04896
| null |
https://arxiv.org/abs/2303.04896v1
|
https://arxiv.org/pdf/2303.04896v1.pdf
|
Using Positive Matching Contrastive Loss with Facial Action Units to mitigate bias in Facial Expression Recognition
|
Machine learning models automatically learn discriminative features from the data, and are therefore susceptible to learn strongly-correlated biases, such as using protected attributes like gender and race. Most existing bias mitigation approaches aim to explicitly reduce the model's focus on these protected features. In this work, we propose to mitigate bias by explicitly guiding the model's focus towards task-relevant features using domain knowledge, and we hypothesize that this can indirectly reduce the dependence of the model on spurious correlations it learns from the data. We explore bias mitigation in facial expression recognition systems using facial Action Units (AUs) as the task-relevant feature. To this end, we introduce Feature-based Positive Matching Contrastive Loss which learns the distances between the positives of a sample based on the similarity between their corresponding AU embeddings. We compare our approach with representative baselines and show that incorporating task-relevant features via our method can improve model fairness at minimal cost to classification performance.
|
['Desmond C. Ong', 'Varsha Suresh']
|
2023-03-08
| null | null | null | null |
['facial-expression-recognition']
|
['computer-vision']
|
[ 3.63942653e-01 4.10610497e-01 -3.77768666e-01 -1.14396739e+00
-6.57133162e-01 -4.22411323e-01 8.37152541e-01 1.73896790e-01
-7.61303008e-01 5.46593010e-01 4.55024570e-01 9.58023667e-02
1.69888169e-01 -6.15588725e-01 -6.83219612e-01 -6.21604025e-01
-4.82869744e-02 2.60048267e-02 -2.76903510e-01 -1.68641105e-01
2.76614755e-01 4.73370373e-01 -1.43369198e+00 5.68558931e-01
6.48827910e-01 1.16601968e+00 -6.45576596e-01 2.26595923e-01
3.91411394e-01 8.99276912e-01 -4.22959417e-01 -6.57631993e-01
4.08806205e-01 -3.88538212e-01 -6.67851806e-01 -2.83665478e-01
8.82284999e-01 -4.65562850e-01 -9.71221924e-02 1.00535679e+00
4.66526508e-01 1.85692813e-02 1.05174696e+00 -1.67967319e+00
-5.26443183e-01 3.13502222e-01 -6.77233040e-01 1.35310233e-01
-1.43629700e-01 1.24014601e-01 1.35980439e+00 -9.97846544e-01
6.04871690e-01 1.47402418e+00 7.73091853e-01 9.58352983e-01
-1.62360620e+00 -1.06343627e+00 3.20915014e-01 1.66000053e-01
-1.21731710e+00 -8.04759383e-01 1.00143623e+00 -5.64980626e-01
5.38487494e-01 1.36202261e-01 2.38325745e-01 1.32023609e+00
-4.71944697e-02 7.38403022e-01 1.25049913e+00 -4.05032963e-01
2.17642263e-01 4.16840494e-01 1.80447951e-01 8.35179210e-01
6.22087233e-02 3.17840725e-01 -7.59695232e-01 -5.11629939e-01
3.11956942e-01 -1.21396124e-01 6.79332688e-02 -7.66764879e-01
-5.68647444e-01 1.16681004e+00 5.41430414e-01 1.04078585e-02
-2.12150961e-01 2.88012832e-01 3.98846686e-01 1.63000986e-01
7.66171396e-01 5.97097456e-01 -5.30181050e-01 3.44973877e-02
-6.63507879e-01 4.73368436e-01 3.05465460e-01 5.19535661e-01
1.16460145e+00 -2.03424022e-01 -4.45756704e-01 9.99322593e-01
3.60011399e-01 3.51138592e-01 3.78522426e-01 -1.10776305e+00
9.90939792e-03 6.01465523e-01 -6.55127615e-02 -1.20625830e+00
-2.33550057e-01 -4.28466350e-02 -2.46133924e-01 3.71224910e-01
4.66454864e-01 -3.08251888e-01 -7.43831396e-01 2.50861096e+00
4.07162935e-01 -2.31534958e-01 -3.02126229e-01 9.24775958e-01
3.32687914e-01 1.28977880e-01 4.94834453e-01 9.54853594e-02
1.14587188e+00 -5.12137175e-01 -3.02270621e-01 -4.63168859e-01
9.99136984e-01 -4.20766830e-01 1.17423689e+00 2.21936349e-02
-8.35711062e-01 -2.71293581e-01 -9.04509127e-01 -2.70010352e-01
-3.73142242e-01 7.92774111e-02 7.40159690e-01 8.65450621e-01
-7.27718294e-01 7.62935042e-01 -6.25821948e-01 -2.57453501e-01
9.57473278e-01 5.14377415e-01 -5.54753482e-01 6.87495619e-02
-1.10255182e+00 1.07394087e+00 -2.64757723e-01 -1.21183358e-01
-7.75465548e-01 -9.68201637e-01 -9.11784530e-01 1.62314363e-02
2.85223499e-02 -3.98127526e-01 1.13543832e+00 -1.61266565e+00
-1.22246349e+00 1.36294031e+00 -2.82955319e-01 -3.68823469e-01
4.20233458e-01 -3.31843913e-01 1.55505724e-02 -2.25759745e-01
-7.74398744e-02 1.09106922e+00 1.14812660e+00 -1.13148797e+00
-4.97419119e-01 -6.34199739e-01 6.41958416e-02 9.03426260e-02
-8.38708758e-01 1.12346284e-01 2.95382798e-01 -4.75620955e-01
-4.97644722e-01 -9.71713424e-01 -5.78163043e-02 5.58969140e-01
-5.88330105e-02 -3.90378863e-01 6.94444239e-01 -4.52105135e-01
1.00900567e+00 -2.25106382e+00 -4.62291352e-02 3.35430294e-01
1.24006778e-01 2.72270650e-01 -4.18560833e-01 -8.82467926e-02
-9.14699063e-02 1.79464489e-01 -1.36237249e-01 -5.06922901e-01
8.92625228e-02 2.39219233e-01 -3.23073924e-01 6.92196965e-01
7.58254409e-01 6.91936195e-01 -7.78570771e-01 -4.51427877e-01
-1.64777398e-01 5.54055750e-01 -1.12166786e+00 3.13815922e-01
2.01444197e-02 8.96798074e-02 -3.03427041e-01 3.81829113e-01
6.67690814e-01 3.50789994e-01 2.26791278e-01 -2.50948042e-01
2.18986899e-01 5.34215152e-01 -5.63078046e-01 1.19136763e+00
-5.70106506e-01 6.51471794e-01 1.00755207e-01 -1.10092723e+00
1.08158064e+00 -1.95057482e-01 2.39078239e-01 -6.79121733e-01
1.19733997e-01 7.54711553e-02 2.67813474e-01 -1.34270370e-01
1.80236295e-01 -4.41131175e-01 5.47597371e-02 4.69453275e-01
1.58703655e-01 1.07134536e-01 -4.66633499e-01 2.17754301e-02
8.27472687e-01 8.09112191e-02 2.49765277e-01 -6.07605934e-01
4.12887424e-01 -4.04999763e-01 8.40514004e-01 5.85941732e-01
-5.65248311e-01 2.91908145e-01 8.39188337e-01 -3.48645955e-01
-9.22654986e-01 -1.00017262e+00 -3.30663383e-01 1.48490453e+00
-4.27143902e-01 -4.17584777e-01 -7.78981805e-01 -1.23620296e+00
4.59948123e-01 6.82431877e-01 -1.31754446e+00 -8.51734221e-01
-5.07774413e-01 -6.56737626e-01 6.56830430e-01 6.40314817e-01
5.36797605e-02 -7.14319944e-01 -5.50735116e-01 -4.01561052e-01
2.09426358e-01 -6.77440345e-01 -4.14584368e-01 3.11576486e-01
-7.70844996e-01 -8.99234354e-01 -4.91117179e-01 -3.80147845e-01
9.43022370e-01 -1.45347081e-02 1.09292066e+00 4.91941944e-02
-3.29198897e-01 3.08092117e-01 -1.19846545e-01 -7.19579637e-01
-1.81357726e-01 8.14210847e-02 1.87054202e-01 3.98508936e-01
1.04643524e+00 -4.24892217e-01 -6.37219131e-01 3.20074946e-01
-6.51233852e-01 -1.63154781e-01 2.76741266e-01 9.96324897e-01
1.04862861e-01 -8.49613249e-01 5.81970394e-01 -1.12548101e+00
4.45888519e-01 -4.63928878e-01 -3.22349101e-01 2.14477237e-02
-8.32956851e-01 2.66074568e-01 4.06375766e-01 -4.90722567e-01
-1.04478240e+00 -4.78123501e-02 3.20741236e-01 -3.80084753e-01
-4.43128720e-02 -3.54971439e-02 -2.97792077e-01 -2.48492703e-01
9.21134710e-01 -3.49476278e-01 2.78705508e-01 -3.27168494e-01
3.26261401e-01 7.30231583e-01 -4.41548079e-02 -9.23029780e-01
5.64256489e-01 6.62677884e-01 2.06502303e-01 -6.12944722e-01
-1.14888752e+00 1.35975573e-02 -5.23635685e-01 -1.65412411e-01
4.86811787e-01 -8.31649125e-01 -6.17362201e-01 2.13774726e-01
-9.42938626e-01 -5.11416018e-01 -1.85287654e-01 4.59788471e-01
-5.42074502e-01 -1.71882603e-02 -3.44379097e-01 -9.26633060e-01
-9.20115411e-02 -8.38854373e-01 1.11948693e+00 1.28225712e-02
-8.82819235e-01 -8.71131241e-01 1.48120448e-01 1.71071231e-01
4.22121912e-01 2.32060030e-01 1.12895465e+00 -9.40415263e-01
3.97899561e-02 -9.21227783e-02 -4.60423678e-01 7.93815553e-01
2.82104850e-01 3.43956351e-02 -1.57577562e+00 -2.31587872e-01
-1.11691736e-01 -6.31721556e-01 1.12700152e+00 2.93790907e-01
1.30032766e+00 -4.06177193e-01 -2.25693688e-01 6.25035703e-01
9.53803778e-01 -2.12506592e-01 4.37231630e-01 1.32395163e-01
6.50058329e-01 1.14753485e+00 7.01742709e-01 4.73122090e-01
1.40782416e-01 7.41793513e-01 2.91597277e-01 -1.63637266e-01
1.43404320e-01 -4.97759551e-01 5.38665473e-01 -1.14091039e-01
1.25126854e-01 3.63043427e-01 -7.24783480e-01 5.96061707e-01
-1.73121679e+00 -8.20789397e-01 2.18361422e-01 2.19341612e+00
1.17834687e+00 -1.06027946e-01 2.71266609e-01 -1.71215340e-01
4.35589612e-01 3.63554299e-01 -5.99954426e-01 -7.66622961e-01
-2.32426357e-03 3.93852800e-01 3.34701061e-01 6.48704708e-01
-1.12982893e+00 9.48292851e-01 6.56019306e+00 4.69692558e-01
-1.24388337e+00 -9.09846276e-02 1.01376510e+00 -6.18455589e-01
-4.08589780e-01 -8.71450529e-02 -6.45834804e-01 3.22045833e-01
7.83206701e-01 -1.63021535e-01 1.11756466e-01 1.10705149e+00
6.64384142e-02 4.89947945e-02 -1.80096042e+00 7.14851379e-01
2.70302176e-01 -8.40301454e-01 1.54647425e-01 3.61680090e-01
5.22937655e-01 -3.01851124e-01 5.70911765e-01 3.73411179e-01
3.47599179e-01 -1.24770665e+00 6.84360206e-01 3.11462462e-01
5.41685522e-01 -8.40419531e-01 5.02414525e-01 -1.57228529e-01
-2.45045438e-01 -1.36107922e-01 -4.66448992e-01 -4.18677807e-01
-4.62165356e-01 5.82609594e-01 -9.33787644e-01 -3.88378918e-01
5.91166079e-01 6.80838108e-01 -6.22524083e-01 1.84766978e-01
-2.19570398e-01 7.25681305e-01 -1.88250110e-01 -7.40494952e-02
5.93310408e-02 -6.91053942e-02 1.88801721e-01 1.18777323e+00
3.30213122e-02 -1.18887238e-01 -2.31205925e-01 9.58060563e-01
-2.53681749e-01 2.19246313e-01 -8.42542946e-01 4.51777950e-02
1.59373268e-01 1.24871743e+00 9.48246866e-02 -1.61504284e-01
-3.88743818e-01 8.24186623e-01 7.24067390e-01 1.41810119e-01
-6.65385365e-01 -1.92185074e-01 1.54246151e+00 1.10738412e-01
1.14367511e-02 2.12585181e-01 -4.65691984e-01 -9.95777547e-01
9.49447304e-02 -9.62143242e-01 4.22644675e-01 -4.67748970e-01
-1.46583748e+00 1.71382904e-01 -1.49628595e-01 -6.62203729e-01
-2.55434901e-01 -8.44347060e-01 -3.49831194e-01 9.98750389e-01
-1.67663276e+00 -1.21870339e+00 1.02386944e-01 5.91562569e-01
-5.70527539e-02 -1.21350259e-01 9.13826168e-01 1.28259644e-01
-4.20385450e-01 1.16675246e+00 -2.55722493e-01 4.55038667e-01
1.33646262e+00 -1.06912422e+00 7.78110996e-02 4.49196517e-01
1.64103210e-01 9.16318357e-01 8.30034971e-01 -2.53654987e-01
-8.66836131e-01 -8.55256975e-01 1.07228470e+00 -8.31256270e-01
5.16186476e-01 -6.87581062e-01 -8.24305117e-01 7.95962811e-01
-2.35978469e-01 3.33681673e-01 1.09524012e+00 7.05590487e-01
-1.26104414e+00 -5.17086685e-01 -1.38615024e+00 6.17705345e-01
1.13007534e+00 -8.21395278e-01 -2.99609482e-01 -4.90632355e-02
1.54634863e-01 1.71077237e-01 -6.34350240e-01 4.07524079e-01
9.22664702e-01 -1.02460849e+00 7.56065845e-01 -1.33271837e+00
6.69475377e-01 2.02660382e-01 -4.10205871e-01 -1.48881114e+00
-3.67027819e-01 -2.97725260e-01 1.68392748e-01 1.37866402e+00
4.81730282e-01 -5.01842201e-01 9.69938397e-01 1.03513455e+00
3.83435458e-01 -6.23358727e-01 -9.36750710e-01 -5.23289204e-01
5.89542866e-01 -1.11386076e-01 6.10913277e-01 1.13364005e+00
1.65415436e-01 2.58017510e-01 -5.34848452e-01 8.88352469e-02
5.82766593e-01 -8.73392597e-02 9.40815210e-01 -1.25475478e+00
-7.67987221e-02 -4.49424982e-01 -4.63374197e-01 -4.09577161e-01
7.46008754e-01 -9.14089918e-01 2.73956778e-03 -4.67913717e-01
4.67395782e-01 -4.89782363e-01 -5.04474223e-01 7.36052990e-01
-2.55337864e-01 1.81015983e-01 1.91078469e-01 -2.15103716e-01
-3.31697494e-01 7.88654327e-01 7.56833017e-01 -1.93878077e-02
1.59891814e-01 -3.39425236e-01 -9.96127009e-01 8.31707954e-01
6.90080702e-01 -7.73178339e-01 -2.58005947e-01 -4.55781162e-01
1.86071634e-01 -7.88438737e-01 5.51876903e-01 -4.91874337e-01
-2.68843919e-01 -3.93165797e-01 6.14845395e-01 5.25378466e-01
4.95911121e-01 -7.70748377e-01 -5.93819737e-01 3.50407541e-01
-1.06705976e+00 -2.89064944e-01 1.51716664e-01 3.48503441e-01
-5.50690144e-02 -1.62718371e-02 1.01610827e+00 1.46070153e-01
-3.00019532e-01 1.86058596e-01 -1.03508927e-01 2.75344878e-01
8.12136650e-01 2.21756428e-01 -3.47324938e-01 -4.37549889e-01
-3.05672050e-01 1.32451504e-02 6.25121891e-01 5.27879000e-01
2.04699948e-01 -1.48376834e+00 -7.11726546e-01 3.19504678e-01
5.74045777e-01 -6.71834171e-01 -1.01095594e-01 5.95476270e-01
2.63486892e-01 2.44245246e-01 -4.43663985e-01 -3.74116123e-01
-1.64250445e+00 4.13843840e-01 5.42707264e-01 5.22578023e-02
1.86762407e-01 1.06068540e+00 6.35265291e-01 -6.18821323e-01
1.17502384e-01 -1.06548086e-01 1.61414221e-01 3.16892296e-01
5.37509441e-01 3.01106185e-01 -5.08531928e-02 -7.51767337e-01
-6.84530973e-01 3.90390933e-01 -4.47838634e-01 -1.67401761e-01
1.30420327e+00 2.49141693e-01 1.05915451e-02 3.39227617e-01
1.65894079e+00 6.98451698e-02 -1.48812902e+00 -3.80773187e-01
1.16557308e-01 -8.65989387e-01 1.31344214e-01 -7.92700589e-01
-9.03636038e-01 1.18320012e+00 7.40957201e-01 -3.54809105e-01
9.27898645e-01 -8.11879784e-02 2.92532831e-01 2.24646628e-01
1.14843287e-01 -1.33511758e+00 1.33304358e-01 2.13263169e-01
6.22675002e-01 -1.48924696e+00 2.81357709e-02 -2.41080850e-01
-6.44310296e-01 6.96129680e-01 8.67363453e-01 -2.93886334e-01
5.70387304e-01 6.25632033e-02 2.98485398e-01 -4.41849269e-02
-9.68012571e-01 3.02057574e-03 3.68341684e-01 8.63409340e-01
7.48328030e-01 9.94251445e-02 -3.80886316e-01 6.94534779e-01
-1.35912612e-01 -1.52907269e-02 1.33494988e-01 7.54454732e-01
-1.91267014e-01 -1.24674118e+00 -1.02794975e-01 5.92891634e-01
-6.05131865e-01 -6.20025583e-02 -7.65368342e-01 5.21526515e-01
1.25281230e-01 5.18714011e-01 3.92527312e-01 -3.16969961e-01
2.52729028e-01 2.97741085e-01 6.09602988e-01 -5.82060456e-01
-6.01941466e-01 -5.17181575e-01 2.06478924e-01 -8.21642637e-01
-4.18755472e-01 -1.03274679e+00 -7.67915666e-01 -2.57251889e-01
1.33703174e-02 -1.14046760e-01 4.41618562e-01 6.90027773e-01
6.18896484e-01 -1.96571231e-01 8.49504650e-01 -5.57776570e-01
-1.00933301e+00 -7.46017277e-01 -4.53256577e-01 9.38567162e-01
5.41598678e-01 -8.38123560e-01 -4.72011983e-01 -2.38246769e-01]
|
[13.068617820739746, 1.3263792991638184]
|
1780cd09-7645-4dce-b885-95df67fd1fd6
|
model-discovery-in-the-sparse-sampling-regime
|
2105.00400
| null |
https://arxiv.org/abs/2105.00400v1
|
https://arxiv.org/pdf/2105.00400v1.pdf
|
Model discovery in the sparse sampling regime
|
To improve the physical understanding and the predictions of complex dynamic systems, such as ocean dynamics and weather predictions, it is of paramount interest to identify interpretable models from coarsely and off-grid sampled observations. In this work, we investigate how deep learning can improve model discovery of partial differential equations when the spacing between sensors is large and the samples are not placed on a grid. We show how leveraging physics informed neural network interpolation and automatic differentiation, allow to better fit the data and its spatiotemporal derivatives, compared to more classic spline interpolation and numerical differentiation techniques. As a result, deep learning-based model discovery allows to recover the underlying equations, even when sensors are placed further apart than the data's characteristic length scale and in the presence of high noise levels. We illustrate our claims on both synthetic and experimental data sets where combinations of physical processes such as (non)-linear advection, reaction, and diffusion are correctly identified.
|
['Remy Kusters', 'Georges Tod', 'Gert-Jan Both']
|
2021-05-02
| null | null | null | null |
['model-discovery']
|
['miscellaneous']
|
[-1.41603678e-01 -3.29414338e-01 1.66166693e-01 2.05133930e-01
-3.15489680e-01 -6.30739808e-01 6.79943025e-01 5.31921625e-01
-1.77325353e-01 1.18174982e+00 -1.77423805e-01 -4.45346981e-01
-2.94401765e-01 -7.30564058e-01 -8.21167409e-01 -7.04802513e-01
-6.94717228e-01 3.75202954e-01 1.07744224e-01 -2.17669755e-01
7.21620768e-02 9.28629875e-01 -1.23906672e+00 -3.18414748e-01
1.08984208e+00 1.06820083e+00 -2.77080834e-01 6.65257812e-01
-5.59950545e-02 6.33742809e-01 -5.73829859e-02 5.05684316e-01
3.16798091e-01 -1.82065114e-01 -2.76622176e-01 -1.63837671e-01
2.03218445e-01 -3.39260280e-01 -9.27869305e-02 8.62374663e-01
8.86583328e-02 5.63393176e-01 8.05826962e-01 -7.15950608e-01
-4.42100286e-01 1.23331353e-01 -5.23241103e-01 2.60459810e-01
-1.85501948e-01 5.91413856e-01 5.46515346e-01 -7.85555422e-01
1.40777826e-01 1.05417180e+00 1.22756147e+00 6.62390366e-02
-1.70920825e+00 -4.12000358e-01 4.98569570e-02 -3.92808378e-01
-1.37794363e+00 -3.57310683e-01 7.59218097e-01 -1.00761652e+00
4.83457237e-01 2.09377274e-01 7.77476549e-01 7.13996708e-01
3.67755979e-01 -1.07566983e-01 1.16638279e+00 -5.05417623e-02
4.99629110e-01 -5.70448190e-02 1.57442912e-01 4.38058168e-01
3.92878264e-01 6.61886692e-01 -2.12304264e-01 -6.14893258e-01
1.15374541e+00 9.25675780e-02 -4.52857286e-01 -1.20109946e-01
-9.57824290e-01 7.86596239e-01 4.16469038e-01 1.48844451e-01
-8.14453304e-01 2.68125206e-01 -6.50649369e-02 -3.32265720e-03
8.69279563e-01 8.43156934e-01 -8.50205839e-01 -6.66767592e-03
-1.14661551e+00 3.83924931e-01 7.87369728e-01 4.72983330e-01
9.55409765e-01 4.41537738e-01 2.88577229e-01 4.60553765e-01
2.35834926e-01 7.06342816e-01 1.99621439e-01 -1.28076971e+00
-1.82858631e-01 4.21471536e-01 7.15445757e-01 -8.90495598e-01
-3.73044223e-01 -5.77439606e-01 -1.41580236e+00 5.87441564e-01
6.43104076e-01 -7.75585413e-01 -8.27323914e-01 1.36890817e+00
4.47470576e-01 6.79978549e-01 1.37770638e-01 1.00013602e+00
1.73104048e-01 8.89207244e-01 -8.33954737e-02 -3.11476916e-01
1.07458615e+00 -4.36465800e-01 -5.48217356e-01 -2.56897155e-02
3.82863402e-01 -3.43615383e-01 9.94998455e-01 4.37900163e-02
-1.13698947e+00 -5.30058503e-01 -7.35047519e-01 1.33716986e-01
-4.28087682e-01 -3.61209393e-01 4.66793716e-01 -3.47917199e-01
-9.15656567e-01 1.32947552e+00 -1.23949480e+00 5.26878573e-02
7.12103099e-02 -2.99340505e-02 7.58386925e-02 4.54890460e-01
-1.08455503e+00 5.51946282e-01 -2.52251774e-01 4.63305295e-01
-7.37561464e-01 -1.26811230e+00 -6.75734401e-01 2.60014713e-01
-1.14376910e-01 -6.34840250e-01 1.01469564e+00 -7.84311235e-01
-1.31555533e+00 1.20391659e-01 -3.43609333e-01 -6.72428608e-01
6.79384530e-01 -3.64165045e-02 -2.52461672e-01 -9.92923602e-02
-1.35360926e-01 1.38838038e-01 5.96819878e-01 -1.33440161e+00
-3.14409405e-01 -1.37071714e-01 -2.16994733e-01 -9.89347398e-02
2.69184798e-01 -3.86894464e-01 3.66028249e-01 -4.56601202e-01
8.45201463e-02 -9.90417957e-01 -6.59883618e-01 6.33382618e-01
-2.93603241e-01 1.79221094e-01 5.06759167e-01 -6.97992146e-01
6.60173535e-01 -1.98197162e+00 -1.79557174e-01 2.81656444e-01
3.20921600e-01 2.14563936e-01 3.77933979e-01 6.02596283e-01
1.24432929e-01 3.55127037e-01 -5.93850255e-01 -2.60393798e-01
-2.91595250e-01 3.26639384e-01 -6.14497066e-01 6.97011530e-01
3.31425697e-01 4.13091302e-01 -8.16848457e-01 2.01556664e-02
1.31088704e-01 6.31270528e-01 -2.93100864e-01 1.25287995e-01
-4.45147544e-01 1.16933560e+00 -3.78679305e-01 1.81617573e-01
8.92041743e-01 -5.73653340e-01 -2.74468899e-01 1.95849493e-01
-8.10970068e-01 2.92866558e-01 -1.23543155e+00 1.00175595e+00
-6.80276096e-01 9.20453489e-01 5.15833616e-01 -7.89607942e-01
8.82054746e-01 3.31948191e-01 5.08556962e-01 -6.00993633e-01
-2.81415790e-01 3.70251149e-01 1.43983394e-01 -3.47323328e-01
3.69794488e-01 -2.88969457e-01 4.73459393e-01 2.72104442e-01
-6.80030107e-01 -1.47622377e-01 -1.32651135e-01 -1.60172731e-01
6.37503445e-01 -1.70739785e-01 9.85407904e-02 -7.94472575e-01
3.40587765e-01 2.67793745e-01 4.82630789e-01 6.21400774e-01
1.72850266e-01 2.56051213e-01 5.52790940e-01 -5.67544580e-01
-1.18571734e+00 -7.46138692e-01 -5.86711586e-01 3.60633701e-01
1.69465095e-01 1.78688541e-01 -3.69577944e-01 2.47648224e-01
6.39817119e-01 4.14675653e-01 -8.52188408e-01 -6.04654811e-02
-6.44883871e-01 -6.88685060e-01 3.38149875e-01 4.42986518e-01
3.35662544e-01 -6.08285487e-01 -4.17194843e-01 4.07073319e-01
3.73743862e-01 -1.02624369e+00 4.96073589e-02 2.66090065e-01
-1.00422502e+00 -8.04735065e-01 -6.54736161e-01 -1.93595886e-01
5.66662729e-01 -1.76554739e-01 9.79576766e-01 1.34462342e-01
1.44327849e-01 1.97521728e-02 2.18243048e-01 -1.69010416e-01
-5.39031923e-01 -1.45269230e-01 1.34615183e-01 8.39026794e-02
-3.24069887e-01 -1.03144610e+00 -6.99269295e-01 4.45121713e-02
-6.41511738e-01 -9.75768939e-02 4.91129607e-02 5.91096461e-01
5.21514714e-01 -4.61182483e-02 2.70994663e-01 -3.66633683e-01
7.00038493e-01 -8.16749275e-01 -1.29819632e+00 -1.86844200e-01
-6.77224517e-01 2.93385029e-01 8.88962924e-01 -6.23210728e-01
-7.67203152e-01 -1.29117340e-01 4.62683365e-02 -3.70614409e-01
-3.04559052e-01 8.30594897e-01 6.25377059e-01 -3.65046859e-01
8.03843081e-01 3.03149879e-01 1.65860113e-02 -8.75716209e-01
-1.05040327e-01 1.77466169e-01 2.86867023e-01 -7.09932864e-01
6.56777740e-01 6.04114056e-01 4.82546866e-01 -1.26843119e+00
-2.13132143e-01 -9.21708345e-02 -6.47387624e-01 -5.87079152e-02
5.20528078e-01 -8.02738249e-01 -7.59380639e-01 5.97593606e-01
-1.30526376e+00 -8.50968301e-01 -5.19514740e-01 6.31559908e-01
-1.34415194e-01 8.61427635e-02 -6.82708800e-01 -1.19467890e+00
4.23398577e-02 -8.27475131e-01 7.74258554e-01 3.27922761e-01
-2.53523022e-01 -1.57314873e+00 3.84692401e-01 -6.27135098e-01
7.66467571e-01 7.48187065e-01 7.15430737e-01 -2.06933811e-01
-5.41519344e-01 7.42118731e-02 -2.06377760e-01 2.29116865e-02
1.93503037e-01 6.63317323e-01 -8.06624055e-01 6.15079363e-04
2.12440535e-01 1.52321517e-01 7.73886919e-01 8.17203462e-01
9.26621795e-01 -6.39378130e-01 -3.20654571e-01 1.01828909e+00
1.37422049e+00 1.21132016e-01 6.61221743e-02 -1.05086990e-01
4.93209273e-01 5.64577579e-01 -9.38812792e-02 5.70551693e-01
2.75818676e-01 3.54227841e-01 2.72270858e-01 -5.33852756e-01
2.81187147e-01 -6.60127774e-02 -5.91510497e-02 4.66562301e-01
-1.35939881e-01 1.75388574e-04 -1.29098260e+00 6.36256933e-01
-1.49738431e+00 -6.26385808e-01 -6.16593242e-01 2.14535952e+00
1.03491747e+00 6.64050058e-02 -2.00758455e-03 -1.87725350e-01
4.78613943e-01 3.22149061e-02 -9.23224568e-01 -4.43400741e-01
-7.41650984e-02 1.99017584e-01 8.90477598e-01 1.08117461e+00
-8.53942871e-01 3.44216883e-01 6.65292168e+00 1.36335537e-01
-1.73228669e+00 -3.20351161e-02 5.59779823e-01 2.28877246e-01
-1.98406711e-01 1.06976345e-01 -7.24334538e-01 5.94496727e-01
1.15338480e+00 -1.31346911e-01 5.44438064e-01 3.30542654e-01
8.25407982e-01 -1.81011274e-01 -8.19931626e-01 3.34374040e-01
-8.69307816e-01 -1.66343999e+00 -4.57779497e-01 3.21967691e-01
9.12694573e-01 2.62101769e-01 -2.71346103e-02 -1.15796208e-01
5.99199057e-01 -1.13065708e+00 4.34898704e-01 1.06085849e+00
3.96464735e-01 -7.09533319e-02 3.54054242e-01 7.72000492e-01
-1.25603807e+00 2.20947936e-01 -1.30990863e-01 -7.07862794e-01
3.04556549e-01 8.13718736e-01 -5.47019839e-01 1.81670591e-01
5.74177861e-01 8.20617676e-01 -3.45279910e-02 1.18197513e+00
1.76077828e-01 1.01952016e+00 -8.27771425e-01 9.45378765e-02
2.90220231e-01 -4.66661900e-01 6.40141129e-01 8.33487630e-01
5.21864116e-01 4.12872553e-01 1.18666448e-01 1.26485837e+00
2.31602043e-01 -3.19822222e-01 -5.25843501e-01 9.85526666e-02
4.38719720e-01 8.62021506e-01 -3.83639723e-01 -2.55875796e-01
-1.57482535e-01 2.93373257e-01 9.67408121e-02 9.78115678e-01
-6.58087134e-01 6.84388131e-02 1.29582202e+00 4.37855870e-01
2.70714402e-01 -8.83250952e-01 -5.16372263e-01 -8.95026982e-01
-7.07229227e-02 -3.62225950e-01 -6.65857829e-03 -7.18062282e-01
-1.29000413e+00 3.32185328e-01 2.24130619e-02 -1.06187320e+00
-2.99782783e-01 -5.86398244e-01 -9.97146547e-01 1.32561135e+00
-1.81685269e+00 -4.66260940e-01 -2.59942561e-01 1.90807909e-01
-1.85594428e-02 4.19149995e-01 6.60735130e-01 8.68004337e-02
-5.76932192e-01 -1.47989005e-01 1.00216043e+00 1.09978169e-01
1.21355057e-01 -1.17607820e+00 5.63872099e-01 7.47636735e-01
-4.18625295e-01 4.53231066e-01 1.04035294e+00 -6.47441387e-01
-1.17010176e+00 -1.12263310e+00 5.83649516e-01 4.34978455e-02
1.10342169e+00 -1.78145558e-01 -1.72806239e+00 4.10923004e-01
6.20955490e-02 6.20205402e-01 1.94237590e-01 -1.25370368e-01
1.84622839e-01 -4.39328223e-01 -1.08360422e+00 4.05739695e-01
3.87663543e-01 -4.10497308e-01 -2.33407855e-01 4.30738240e-01
4.90846336e-01 -4.16065961e-01 -1.01784682e+00 5.53481579e-01
4.03566241e-01 -5.95319450e-01 8.06935906e-01 -8.17431808e-01
4.28332239e-01 -3.92042041e-01 1.23746306e-01 -1.57581234e+00
-1.21685736e-01 -8.12588930e-01 -2.22430915e-01 1.10618293e+00
4.83672857e-01 -1.09603035e+00 2.85240382e-01 9.72290337e-01
1.05304822e-01 -7.40732253e-01 -9.57695901e-01 -8.28692675e-01
6.30232573e-01 -1.28836527e-01 5.94643474e-01 1.20572078e+00
-3.93119723e-01 -1.32748783e-01 -2.77362972e-01 8.10782909e-01
7.90229499e-01 2.00018033e-01 5.27150333e-01 -1.70659816e+00
-2.38963917e-01 -5.39172709e-01 2.56833304e-02 -1.09851611e+00
4.68550026e-02 -2.13613495e-01 2.10088581e-01 -1.08623207e+00
-7.58204341e-01 -8.74721587e-01 -2.91126240e-02 6.68269321e-02
-1.11617439e-03 1.17505714e-02 -2.49510661e-01 3.88874859e-01
3.72468978e-01 5.75596154e-01 1.30337560e+00 8.87773186e-02
-3.33488196e-01 1.69717550e-01 -1.00162201e-01 7.58952200e-01
8.42134535e-01 -4.11579221e-01 -7.26024583e-02 -4.24582541e-01
8.40025768e-02 4.81826037e-01 6.68974876e-01 -1.06819391e+00
3.28942299e-01 -6.85614645e-01 4.23171014e-01 -3.09778541e-01
3.37092191e-01 -7.07296491e-01 4.36114132e-01 4.23274636e-01
-6.24860168e-01 1.17488295e-01 7.43598640e-01 4.51237708e-01
-1.51572809e-01 1.83620080e-01 8.98023248e-01 -2.40570739e-01
-2.35061482e-01 4.52524304e-01 -7.30082810e-01 2.05520511e-01
7.44996190e-01 6.35824054e-02 -1.88822895e-01 -4.08067465e-01
-6.45802498e-01 4.25776333e-01 6.13953054e-01 -3.03489506e-01
1.54243305e-01 -9.85247195e-01 -7.88500845e-01 4.07431573e-01
-4.57174331e-01 3.52019936e-01 1.10902444e-01 9.80822384e-01
-7.81114459e-01 6.97069243e-02 1.15694053e-01 -6.75112724e-01
-5.67773521e-01 2.39888415e-01 9.47330713e-01 -3.37303840e-02
-6.26597285e-01 4.88387316e-01 1.23173654e-01 -2.92663097e-01
-1.20203495e-01 -1.14865661e+00 2.21337214e-01 -1.60203263e-01
1.43371731e-01 4.47097212e-01 -2.04565555e-01 -4.44993049e-01
-6.56576678e-02 6.92718208e-01 7.12373435e-01 -2.67755371e-02
1.21152723e+00 -1.99590579e-01 -6.59024492e-02 9.23464239e-01
9.75866258e-01 1.89160645e-01 -2.12296343e+00 -2.45945707e-01
-1.48075119e-01 -1.56338811e-01 1.89719185e-01 -5.36528826e-01
-9.90479469e-01 1.19258761e+00 3.29360068e-01 9.60309207e-01
6.36868358e-01 -3.04109037e-01 9.13500190e-01 2.45453775e-01
-3.10291290e-01 -5.52615762e-01 -6.17320716e-01 7.76491523e-01
7.27348745e-01 -1.15822816e+00 -5.91789279e-03 -2.91383341e-02
-7.50410706e-02 1.21007884e+00 2.66964108e-01 -4.66103017e-01
1.12930524e+00 6.90405965e-01 1.25189587e-01 7.79061988e-02
-6.68369830e-01 9.71904844e-02 2.94494092e-01 5.49575761e-02
2.00098976e-01 -1.19901681e-02 2.11320356e-01 1.96643263e-01
7.71506205e-02 -3.90702747e-02 4.48644042e-01 6.15110695e-01
-3.62322062e-01 -4.92119282e-01 -4.43203211e-01 3.24369609e-01
-2.18221426e-01 -1.79076344e-01 -4.70975153e-02 6.79666340e-01
-1.71732530e-02 5.07010758e-01 4.54146296e-01 2.87501812e-01
1.61037490e-01 1.94217300e-03 -2.13132754e-01 -1.95894793e-01
-2.78209269e-01 -9.81676206e-02 -1.68771833e-01 -1.84130237e-01
-3.43556345e-01 -7.44754076e-01 -1.45047176e+00 -5.35197914e-01
-1.24477692e-01 5.26245892e-01 5.00811577e-01 1.08549416e+00
6.08498931e-01 4.17661220e-01 6.16368175e-01 -1.25313640e+00
-5.34595132e-01 -8.55073631e-01 -8.14915776e-01 2.42369883e-02
1.38663363e+00 -6.94216967e-01 -8.84331882e-01 1.16550080e-01]
|
[6.536596298217773, 3.3599233627319336]
|
9a1dbe7d-c38c-442e-b268-988a40cb34e0
|
um-iuling-at-semeval-2019-task-6-identifying
|
1904.03450
| null |
http://arxiv.org/abs/1904.03450v1
|
http://arxiv.org/pdf/1904.03450v1.pdf
|
UM-IU@LING at SemEval-2019 Task 6: Identifying Offensive Tweets Using BERT and SVMs
|
This paper describes the UM-IU@LING's system for the SemEval 2019 Task 6:
OffensEval. We take a mixed approach to identify and categorize hate speech in
social media. In subtask A, we fine-tuned a BERT based classifier to detect
abusive content in tweets, achieving a macro F1 score of 0.8136 on the test
data, thus reaching the 3rd rank out of 103 submissions. In subtasks B and C,
we used a linear SVM with selected character n-gram features. For subtask C,
our system could identify the target of abuse with a macro F1 score of 0.5243,
ranking it 27th out of 65 submissions.
|
['Sandra Kübler', 'Zuoyu Tian', 'Jian Zhu']
|
2019-04-06
|
um-iuling-at-semeval-2019-task-6-identifying-1
|
https://aclanthology.org/S19-2138
|
https://aclanthology.org/S19-2138.pdf
|
semeval-2019-6
|
['abuse-detection']
|
['natural-language-processing']
|
[-2.74257034e-01 -3.01955082e-02 -2.44271368e-01 -1.12169638e-01
-9.27338243e-01 -8.95837963e-01 9.07961249e-01 4.89337444e-01
-6.74936712e-01 8.36843193e-01 1.06485277e-01 -4.34748709e-01
2.29520351e-01 -3.73455018e-01 -3.07310641e-01 -4.51251298e-01
-1.26665263e-02 2.49565229e-01 3.40220958e-01 -3.02891225e-01
7.44688392e-01 4.58041906e-01 -9.44193065e-01 7.40178466e-01
7.46717393e-01 7.23471820e-01 -5.79104722e-01 9.95103061e-01
2.30732895e-02 1.20811081e+00 -1.35235655e+00 -1.08039856e+00
-2.99524307e-01 -3.32290292e-01 -1.19332850e+00 -4.66432363e-01
5.58833599e-01 -3.35790277e-01 -4.81625974e-01 1.13410199e+00
3.28155965e-01 1.16815813e-01 8.28132689e-01 -1.37726307e+00
-6.52641952e-01 8.40813875e-01 -5.86590171e-01 9.21508968e-01
7.68457890e-01 2.43603140e-01 1.07453048e+00 -9.53577757e-01
4.77229595e-01 1.04197085e+00 4.31022793e-01 7.60787547e-01
-1.14549124e+00 -1.06917167e+00 -3.98036361e-01 3.11805397e-01
-1.24402678e+00 -3.69783461e-01 6.23281538e-01 -9.57524657e-01
9.75996614e-01 4.54817146e-01 3.97781014e-01 1.60821068e+00
-1.65651832e-02 9.07214701e-01 1.25248027e+00 -1.78739056e-01
1.37951002e-01 3.53244901e-01 7.14365721e-01 5.50143361e-01
1.17451765e-01 -2.11760506e-01 -5.22252977e-01 -6.92542970e-01
2.48773079e-02 -4.83935684e-01 1.51206538e-01 7.99879432e-01
-7.24630535e-01 1.37983549e+00 1.19842760e-01 7.28068411e-01
-5.42666167e-02 -2.82899439e-01 8.49197030e-01 1.53180152e-01
6.38313115e-01 9.41076577e-01 -4.94410247e-02 -5.64720333e-01
-8.96040738e-01 4.63211268e-01 7.89139688e-01 3.38603735e-01
1.69324145e-01 3.30087356e-02 -5.73679149e-01 1.09992599e+00
-2.60624737e-01 5.09264708e-01 5.26076972e-01 -6.15520477e-01
6.29054666e-01 1.69663548e-01 -4.15626504e-02 -9.76783931e-01
-3.60882431e-01 -1.92342028e-01 -4.26333100e-01 -2.92888451e-02
5.90254724e-01 -5.41539311e-01 -6.44695759e-01 1.34072256e+00
-6.40716329e-02 -2.48540025e-02 -3.89956832e-01 5.58696389e-01
8.35140646e-01 8.25880229e-01 3.39144558e-01 -2.79925346e-01
1.14929843e+00 -7.62553811e-01 -6.47173762e-01 -7.07549304e-02
9.31040704e-01 -7.25058913e-01 9.60190117e-01 6.55409873e-01
-8.13459873e-01 -1.22400500e-01 -9.95942056e-01 2.16386199e-01
-6.05693817e-01 -3.59196246e-01 1.46103084e-01 9.28600848e-01
-5.90532184e-01 6.75605416e-01 -2.15375766e-01 -2.17846915e-01
4.52395231e-01 1.16751641e-02 -2.99674928e-01 2.51271307e-01
-1.51980364e+00 1.23867011e+00 3.36920083e-01 -5.09218037e-01
-8.33302975e-01 -6.41225636e-01 -5.57071328e-01 -1.51022181e-01
4.53309976e-02 3.12759072e-01 1.25259340e+00 -3.28023106e-01
-1.14307773e+00 1.20656800e+00 1.10928640e-02 -4.63763028e-01
4.49735254e-01 -4.28557634e-01 -8.04306567e-01 1.18668757e-01
2.21179739e-01 -4.39273976e-02 8.37042987e-01 -9.92578268e-01
-4.59731340e-01 -2.53351390e-01 1.81323156e-01 -2.39084020e-01
-7.52857924e-01 8.13348889e-01 4.49696988e-01 -4.90147322e-01
-7.56678820e-01 -7.50364661e-01 2.70430028e-01 -9.30811763e-01
-8.32373619e-01 -4.75034624e-01 1.02120459e+00 -1.01809144e+00
2.02220535e+00 -2.09922075e+00 -1.05320863e-01 2.11565942e-01
5.06336987e-01 8.56532276e-01 2.78077543e-01 5.60962558e-01
-1.94833606e-01 5.42781115e-01 -9.31329429e-02 -4.05008554e-01
-5.97623065e-02 -3.00404131e-01 -5.10601282e-01 4.71535712e-01
1.58591181e-01 5.79391479e-01 -1.19618654e+00 -3.33797544e-01
-1.29851392e-02 7.74318278e-02 -3.61823589e-01 4.14445043e-01
3.71859163e-01 1.17629290e-01 -3.48097682e-01 4.61697161e-01
2.89719254e-01 2.96092123e-01 -4.62282896e-01 2.85189629e-01
-1.41709551e-01 5.03048658e-01 -3.42724472e-01 8.27597201e-01
-3.21871758e-01 1.17045426e+00 -2.02454198e-02 -4.44596320e-01
9.22090650e-01 3.21765721e-01 2.19431117e-01 -4.07876641e-01
3.09925705e-01 3.01513314e-01 1.50916025e-01 -7.33573258e-01
4.03137475e-01 -1.90981537e-01 -5.15840709e-01 3.97099733e-01
2.03621954e-01 -8.50725695e-02 3.18947911e-01 4.88640755e-01
1.57994485e+00 -5.04579067e-01 1.92828953e-01 -1.61815062e-01
7.44869888e-01 -1.22269288e-01 -9.71720442e-02 8.74424577e-01
-5.90859234e-01 4.98066992e-01 9.94729996e-01 -4.52037543e-01
-9.38275456e-01 -7.16991067e-01 -2.07511678e-01 1.28946698e+00
-4.58974093e-01 -6.14644527e-01 -1.09759510e+00 -1.14408231e+00
-2.40886584e-02 1.34861660e+00 -7.09999979e-01 -3.27963918e-01
-5.51299930e-01 -7.88172543e-01 1.20712376e+00 -3.10799349e-02
2.62439549e-01 -1.35745466e+00 -4.16263700e-01 -9.38741956e-04
-3.03846300e-01 -1.18625331e+00 -4.53264892e-01 3.04738969e-01
-3.39466855e-02 -1.06478083e+00 -3.72520745e-01 -2.88193554e-01
8.59516487e-02 -2.48007700e-01 6.92642808e-01 1.79265901e-01
-1.98375911e-01 -3.83230329e-01 -6.65071785e-01 -2.61055708e-01
-5.65576136e-01 4.17674750e-01 2.24047258e-01 7.52821937e-02
7.22414970e-01 -3.34357440e-01 8.97470787e-02 -1.62456945e-01
-6.87838972e-01 -5.09161472e-01 -8.53167549e-02 6.56286836e-01
-4.56427664e-01 -2.89760619e-01 5.72135746e-01 -1.07811153e+00
1.12953222e+00 -8.54288518e-01 -3.86667028e-02 -2.44028687e-01
-2.91314334e-01 -5.01248896e-01 1.07816041e+00 -5.62257051e-01
-5.83182037e-01 -3.90695602e-01 -5.75669885e-01 -3.59230816e-01
-3.02084684e-01 1.19583167e-01 1.26988098e-01 2.12122574e-01
1.05531156e+00 2.54075285e-02 -3.47274572e-01 -5.81360161e-01
-7.40169585e-02 1.15986311e+00 3.76471907e-01 -4.27884191e-01
1.10110843e+00 -3.76074404e-01 -4.68679637e-01 -1.25599849e+00
-1.52698672e+00 -6.75296783e-01 -5.94572127e-01 -4.05745536e-01
1.05057156e+00 -3.96580338e-01 -9.18231010e-01 7.87746251e-01
-1.39947689e+00 -2.88297862e-01 2.79815912e-01 2.13527292e-01
-1.43750235e-01 1.94449559e-01 -1.01086247e+00 -1.05307078e+00
-5.19657850e-01 -9.06306148e-01 6.92046881e-01 -1.31302297e-01
-9.87331510e-01 -8.77538383e-01 3.36744368e-01 7.41346240e-01
1.64077356e-01 5.51236272e-01 8.65750909e-01 -1.58702648e+00
7.58078158e-01 -5.46235383e-01 -1.29199445e-01 5.22189915e-01
-9.22977626e-02 1.40534744e-01 -1.32805264e+00 -2.13398114e-01
-2.03792870e-01 -7.53402770e-01 6.73802197e-01 -2.38195077e-01
1.24840987e+00 -6.38400197e-01 -1.03055820e-01 8.88026059e-02
9.49353337e-01 3.70158762e-01 5.17536998e-01 3.39336336e-01
8.74111176e-01 4.47452515e-01 3.74427527e-01 6.31476700e-01
-2.31911018e-01 7.85072148e-01 2.71925688e-01 4.47773457e-01
2.66841441e-01 -5.28856397e-01 6.74441695e-01 5.32193363e-01
8.57490972e-02 -2.46554077e-01 -1.19043100e+00 5.57440042e-01
-1.44063795e+00 -1.38663340e+00 -5.74088395e-01 1.89041901e+00
8.91818285e-01 6.09130144e-01 7.41016805e-01 3.71498466e-01
8.93330336e-01 4.92172450e-01 -3.99332158e-02 -1.29796875e+00
2.37501979e-01 2.35734746e-01 3.18000644e-01 7.74209976e-01
-1.36679113e+00 1.09129000e+00 6.53964663e+00 1.26277936e+00
-8.17488074e-01 3.03699762e-01 6.86845660e-01 -4.55512136e-01
1.08453773e-01 -3.97934973e-01 -9.48051810e-01 1.16788769e+00
1.61063993e+00 -1.72716871e-01 3.63990515e-01 7.49893725e-01
-1.25068784e-01 -1.10067725e-01 -8.08071017e-01 8.06278706e-01
5.00668883e-01 -1.07467234e+00 -1.93236813e-01 2.20302045e-01
5.31237304e-01 -3.41312364e-02 -3.17766592e-02 6.66003823e-01
5.89671195e-01 -1.41804838e+00 7.47802913e-01 6.94951639e-02
5.09904683e-01 -1.15042150e+00 9.06061649e-01 8.69022429e-01
-2.50972718e-01 -1.54984504e-01 -1.19092010e-01 -2.85698652e-01
1.64428890e-01 7.01176047e-01 -1.01742530e+00 -8.45866557e-03
7.32658386e-01 4.46095318e-01 -6.76549137e-01 8.43352079e-01
-5.17447472e-01 1.29854798e+00 -7.74040073e-02 -4.39028949e-01
4.09204662e-01 2.62010425e-01 8.90577495e-01 1.95471311e+00
-2.24111781e-01 1.33590952e-01 -5.56391664e-02 4.61273849e-01
-2.81575501e-01 8.43213424e-02 -7.15984881e-01 -2.70681560e-01
6.39932334e-01 1.26640224e+00 -2.87152827e-01 -3.51669401e-01
1.43808067e-01 9.88903701e-01 6.23570800e-01 -6.08136095e-02
-1.05744064e+00 -7.83519864e-01 5.25910139e-01 2.03544363e-01
-1.13283657e-01 -2.82093212e-02 -4.01217580e-01 -1.13390446e+00
-4.07368660e-01 -9.11559045e-01 4.33110684e-01 -4.48336393e-01
-1.55379224e+00 7.96724856e-01 1.22260049e-01 -5.66286683e-01
-3.19765061e-01 -6.24426901e-01 -7.32938528e-01 7.52110422e-01
-5.84686339e-01 -8.39081705e-01 2.60902464e-01 2.97838032e-01
5.17641306e-01 -3.87223840e-01 7.98770249e-01 3.12264055e-01
-8.51598620e-01 7.94950902e-01 -8.74382034e-02 6.95088446e-01
7.11714864e-01 -1.26891291e+00 3.14145893e-01 6.43275619e-01
-4.13480662e-02 5.83491921e-01 9.75912809e-01 -6.66694939e-01
-5.33127010e-01 -9.18327808e-01 1.51311862e+00 -1.04819810e+00
1.36834943e+00 -5.37286460e-01 -1.01080072e+00 5.43664813e-01
2.42553920e-01 -3.93553495e-01 1.00764573e+00 1.74257070e-01
-7.60350883e-01 4.41542387e-01 -1.31662893e+00 3.48674655e-01
6.83990300e-01 -8.21229279e-01 -8.47690642e-01 5.19084930e-01
4.66062605e-01 -2.48313367e-01 -9.53746200e-01 -7.79692903e-02
4.49218124e-01 -8.43599260e-01 6.24730885e-01 -1.15379381e+00
1.08858061e+00 3.03582788e-01 -9.64744575e-03 -1.27556324e+00
-4.55896676e-01 -6.86450541e-01 -3.70381236e-01 1.50385857e+00
5.18873870e-01 -3.33727866e-01 3.98025602e-01 5.36936343e-01
1.05350398e-01 -7.89914012e-01 -9.32594240e-01 -8.23218226e-01
6.66526020e-01 -3.28529209e-01 7.51148462e-02 1.32924938e+00
4.71813381e-01 7.30344534e-01 -6.35962248e-01 -3.71662319e-01
4.87073243e-01 -4.21250939e-01 4.40671265e-01 -1.07766175e+00
-6.09403141e-02 -6.73072159e-01 -2.12185949e-01 -3.93276185e-01
6.06187761e-01 -8.86232078e-01 -2.71219552e-01 -8.23307753e-01
5.24915695e-01 1.25252053e-01 -1.20636769e-01 6.08841479e-01
-2.52670377e-01 7.85283804e-01 4.93918300e-01 1.69523105e-01
-5.93173862e-01 1.24119550e-01 6.69154942e-01 -1.33667827e-01
-1.47343464e-02 7.18564317e-02 -6.97505057e-01 7.86547542e-01
9.67415035e-01 -6.85568094e-01 2.27595538e-01 1.16134211e-01
-1.48919283e-03 -2.16944337e-01 2.24747732e-01 -6.97141111e-01
-2.23496512e-01 -2.68709362e-01 3.03255737e-01 -4.38211530e-01
3.30424756e-01 -3.01432073e-01 -3.42860281e-01 6.16840661e-01
-6.79219902e-01 -9.47266519e-02 8.54671523e-02 1.82768442e-02
-1.13851026e-01 -7.46631503e-01 9.28229213e-01 9.73735154e-02
-1.28657103e-01 4.52703461e-02 -9.89180207e-01 4.87433314e-01
1.08902812e+00 6.49657333e-03 -5.93084633e-01 -3.80219012e-01
-7.23966897e-01 8.20830348e-04 1.80245757e-01 5.07748485e-01
3.49811077e-01 -9.79662180e-01 -9.85308528e-01 -1.75429732e-01
1.54263586e-01 -8.73869717e-01 -5.69801517e-02 5.40280521e-01
-3.39383870e-01 3.57139498e-01 -1.48806781e-01 5.17816022e-02
-1.48767316e+00 3.13827932e-01 1.08475484e-01 -4.90912735e-01
-1.05377480e-01 9.49153423e-01 -3.57582897e-01 -2.19284952e-01
-1.00815795e-01 7.37099826e-01 -6.76208794e-01 4.61227119e-01
9.22660887e-01 7.28444397e-01 8.44909623e-02 -1.12482345e+00
-6.08900726e-01 -1.10895634e-01 -3.51232708e-01 -2.31711298e-01
1.25678337e+00 4.68891531e-01 -3.19181561e-01 5.92802525e-01
1.70441496e+00 4.44702774e-01 -3.58992308e-01 1.39599517e-01
3.33428800e-01 -6.40935659e-01 -6.02176860e-02 -9.50993896e-01
-5.56131244e-01 8.09599221e-01 -1.42308041e-01 8.15798342e-01
4.16665345e-01 1.56286228e-02 9.93798673e-01 2.53783613e-01
2.59550720e-01 -9.77669120e-01 2.51011759e-01 1.14001739e+00
9.16665494e-01 -1.07829762e+00 -9.07607377e-02 -1.47262171e-01
-1.03726113e+00 1.08555245e+00 7.57880211e-01 -3.06022286e-01
3.86162072e-01 1.08196206e-01 -1.89024016e-01 -2.97770172e-01
-6.41202450e-01 2.65371293e-01 3.44108492e-01 3.99263620e-01
5.78605592e-01 2.54644215e-01 -6.69094324e-01 6.53249979e-01
-6.78132653e-01 -6.58332586e-01 6.74995780e-01 4.94818717e-01
-6.50739789e-01 -7.32552350e-01 -3.79247993e-01 6.65685236e-01
-1.13155472e+00 -9.45159942e-02 -1.21404862e+00 5.51124334e-01
1.50086926e-02 1.19415832e+00 -7.55467415e-02 -9.25538957e-01
2.73007751e-01 4.44428295e-01 1.04779430e-01 -7.54104376e-01
-1.27336562e+00 -4.04213250e-01 7.57979512e-01 -3.78820181e-01
1.28619164e-01 -8.55117381e-01 -8.29220414e-01 -7.90952802e-01
-1.55744746e-01 3.86409730e-01 3.81357253e-01 1.13011909e+00
-1.63470522e-01 -1.53731182e-02 9.15653586e-01 -3.79584819e-01
-6.83335602e-01 -1.20740914e+00 -6.19132459e-01 6.79393172e-01
4.98743236e-01 -5.77700317e-01 -8.30068886e-01 -1.75731063e-01]
|
[8.81168270111084, 10.575645446777344]
|
69052275-7d17-4608-b61e-446d25608adb
|
model-reduction-of-swing-equations-with
|
2110.14066
| null |
https://arxiv.org/abs/2110.14066v2
|
https://arxiv.org/pdf/2110.14066v2.pdf
|
Towards Model Reduction for Power System Transients with Physics-Informed PDE
|
This manuscript reports the first step towards building a robust and efficient model reduction methodology to capture transient dynamics in a transmission level electric power system. Such dynamics is normally modeled on seconds-to-tens-of-seconds time scales by the so-called swing equations, which are ordinary differential equations defined on a spatially discrete model of the power grid. Following Seymlyen (1974) and Thorpe, Seyler, and Phadke (1999), we suggest to map the swing equations onto a linear, inhomogeneous Partial Differential Equation (PDE) of parabolic type in two space and one time dimensions with time-independent coefficients and properly defined boundary conditions. We illustrate our method on the synchronous transmission grid of continental Europe. We show that, when properly coarse-grained, i.e., with the PDE coefficients and source terms extracted from a spatial convolution procedure of the respective discrete coefficients in the swing equations, the resulting PDE reproduces faithfully and efficiently the original swing dynamics. We finally discuss future extensions of this work, where the presented PDE-based modeling will initialize a physics-informed machine learning approach for real-time modeling, $n-1$ feasibility assessment and transient stability analysis of power systems.
|
['Philippe Jacquod', 'Julian Fritzsch', 'Michael Chertkov', 'Laurent Pagnier']
|
2021-10-26
| null | null | null | null |
['physics-informed-machine-learning']
|
['graphs']
|
[-4.03761148e-01 -3.08724135e-01 4.61578131e-01 2.34500736e-01
-2.83815712e-01 -7.36465514e-01 6.05257809e-01 1.23799458e-01
-5.92047460e-02 1.14036155e+00 -5.51132441e-01 -3.70834827e-01
-5.46794593e-01 -7.59072840e-01 -1.98773101e-01 -1.08020961e+00
-5.59442699e-01 4.52480257e-01 -2.08542496e-01 -3.52076173e-01
-1.30757153e-01 9.85331237e-01 -1.01431692e+00 -2.74880081e-01
1.00622356e+00 9.85218406e-01 -8.77418518e-02 6.87660933e-01
4.83063161e-01 6.19411588e-01 -5.75265884e-01 3.00961792e-01
6.37098923e-02 -4.34963018e-01 -7.73551762e-01 1.62458032e-01
-5.05098760e-01 9.00443345e-02 -4.40611750e-01 8.25550139e-01
2.52942204e-01 4.19476062e-01 9.28547144e-01 -1.39314198e+00
-4.59263027e-02 3.40325177e-01 -4.67396677e-01 4.81598884e-01
2.33154763e-02 2.11560562e-01 4.95218575e-01 -6.69655502e-01
2.72404462e-01 5.86017728e-01 1.16022563e+00 2.54558865e-02
-1.60116208e+00 -2.33269915e-01 -1.87763244e-01 4.78503555e-02
-1.64630306e+00 1.39506772e-01 8.31665277e-01 -6.38842404e-01
1.29195976e+00 5.30218244e-01 9.94479358e-01 2.61382699e-01
7.44442940e-01 2.46269181e-01 1.28213906e+00 -4.02727515e-01
4.79926020e-01 6.73502237e-02 5.19228801e-02 4.49557394e-01
-3.76433618e-02 3.29449177e-01 1.29896760e-01 -6.69919789e-01
6.19381011e-01 -4.54407424e-01 -3.19752544e-01 -2.76215166e-01
-8.10083568e-01 8.83439243e-01 1.45907924e-01 8.61295760e-01
-6.68246269e-01 -5.51730692e-02 3.04088861e-01 1.60544291e-01
8.02275598e-01 3.52109551e-01 -5.79769373e-01 -1.55673191e-01
-1.35439539e+00 5.06445348e-01 1.09944522e+00 7.61898935e-01
4.88117248e-01 6.53465211e-01 1.33810535e-01 4.06029642e-01
7.94390962e-02 6.53963625e-01 1.67471051e-01 -9.31714714e-01
-3.05306911e-01 6.47412762e-02 5.69862068e-01 -5.98103583e-01
-7.48338521e-01 -6.14427745e-01 -1.11242056e+00 3.36149037e-01
2.16709360e-01 -8.49322617e-01 -2.28684470e-01 1.48031235e+00
2.88250864e-01 4.52200651e-01 3.96756530e-02 5.98006606e-01
5.20727076e-02 1.25701928e+00 -1.41947595e-02 -8.91088784e-01
1.06492054e+00 -5.27683794e-01 -7.67528892e-01 4.46902156e-01
8.35646868e-01 -6.26288950e-01 4.23633754e-01 4.17188138e-01
-1.47519445e+00 -2.27700666e-01 -1.00493324e+00 6.36827409e-01
-4.40436751e-01 2.22715512e-01 3.43914121e-01 3.70852321e-01
-1.33390522e+00 1.28261316e+00 -1.05319929e+00 -2.21341088e-01
-1.17629960e-01 8.32790136e-02 1.28310725e-01 8.68649125e-01
-1.36979973e+00 1.21619201e+00 -2.34075591e-01 3.04976046e-01
-6.49429560e-01 -1.34797537e+00 -5.30273736e-01 1.12345986e-01
-1.66297033e-01 -5.02886593e-01 1.28456867e+00 -4.55360442e-01
-1.66983330e+00 3.52726460e-01 1.38470624e-02 -5.73842883e-01
5.61260700e-01 4.26720440e-01 -4.85395789e-01 5.37588224e-02
-9.03997719e-02 -4.99031931e-01 5.55419505e-01 -1.21351409e+00
-5.01167119e-01 1.45705119e-01 -2.19774798e-01 2.12900992e-02
-2.40825713e-02 1.17681593e-01 3.35430890e-01 -7.43881524e-01
-1.64788410e-01 -1.04753351e+00 -5.07378280e-01 -2.27327511e-01
-1.40883937e-01 -8.85114297e-02 9.50983167e-01 -1.11569047e+00
1.20852840e+00 -1.73909843e+00 5.92224479e-01 4.43780363e-01
-8.56684148e-02 2.65395939e-01 5.15495121e-01 1.03367829e+00
-5.22874415e-01 -2.12103561e-01 -8.50778461e-01 -7.54233673e-02
1.59346879e-01 1.67020097e-01 -4.51012462e-01 9.59475994e-01
1.34003922e-01 4.93395388e-01 -4.64978904e-01 1.53284045e-02
4.60331678e-01 5.56713879e-01 -7.93225691e-02 -1.07440598e-01
-3.19174826e-02 7.90058434e-01 -5.73679268e-01 -5.65083139e-02
8.25355828e-01 -1.60479322e-01 -1.20319523e-01 -3.66686612e-01
-7.12840617e-01 -2.25877360e-01 -1.38892615e+00 1.24501932e+00
-9.66599703e-01 5.01848757e-01 7.93573916e-01 -1.62584162e+00
5.92697740e-01 5.49266279e-01 1.16547847e+00 -5.73883116e-01
1.91707835e-01 2.17154533e-01 -2.58605689e-01 -2.45246798e-01
1.95481122e-01 -4.39061671e-01 -1.64899528e-01 6.35979950e-01
-4.81589884e-02 -7.49281347e-01 4.68016893e-01 5.07390089e-02
1.04826212e+00 -5.29705696e-02 2.46237248e-01 -1.17952478e+00
7.76427329e-01 4.55796808e-01 5.46816647e-01 1.66896567e-01
3.27048600e-02 9.05690268e-02 5.86327255e-01 -4.44242001e-01
-1.13973653e+00 -7.70721912e-01 -8.37947190e-01 2.52515942e-01
-2.70146370e-01 -2.59754956e-01 -9.48173344e-01 1.31296575e-01
1.94632873e-01 9.41072404e-01 -4.09583151e-01 -1.01040132e-01
-8.05029750e-01 -1.28966546e+00 3.90177101e-01 2.36473963e-01
3.82323861e-01 -6.89612687e-01 -6.16886616e-01 6.04215682e-01
3.40585947e-01 -8.93068552e-01 1.71173643e-02 3.97658944e-01
-7.58382261e-01 -9.55708921e-01 -8.96328330e-01 -5.91290057e-01
3.89207125e-01 -6.36537492e-01 1.07082736e+00 2.34104916e-02
-6.71535730e-01 4.92694110e-01 3.38505544e-02 6.68945536e-02
-5.36266267e-01 -3.56568366e-01 2.91448653e-01 -1.68605089e-01
-3.31519574e-01 -1.01353776e+00 -5.99077046e-02 1.25629500e-01
-7.00801015e-01 -4.02959660e-02 -8.52638036e-02 6.36618018e-01
4.21308815e-01 8.60644281e-01 3.59183460e-01 -2.55167395e-01
7.28707373e-01 -5.42804420e-01 -1.39658797e+00 5.50712161e-02
-5.91885448e-01 -1.48641467e-01 1.08732665e+00 -4.07164484e-01
-1.15182376e+00 6.83031790e-03 -4.51227427e-02 -2.36200690e-01
-7.46813640e-02 6.27492368e-01 2.53560156e-01 -5.88798463e-01
2.60287344e-01 3.64537567e-01 -2.31015399e-01 -6.44092083e-01
9.86268558e-03 3.20076704e-01 2.20633686e-01 -8.15202892e-01
1.13425994e+00 6.31304026e-01 6.08448923e-01 -1.22896397e+00
9.45053846e-02 4.71624807e-02 -8.34621191e-01 -1.08096503e-01
6.41911089e-01 -6.59777164e-01 -6.88033819e-01 9.34951842e-01
-1.11594474e+00 -7.60658681e-01 -6.78599417e-01 3.34153086e-01
-7.32202470e-01 1.64541110e-01 -1.12016678e+00 -9.65029299e-01
-1.16998099e-01 -8.53527308e-01 6.50655091e-01 1.54053733e-01
-2.29555950e-01 -1.71592402e+00 4.38898981e-01 -6.21367991e-01
8.38529587e-01 5.78639030e-01 1.13716245e+00 -6.71288446e-02
-1.45341665e-01 2.17155516e-02 3.19089800e-01 2.47464120e-01
-1.79205254e-01 3.47638011e-01 -5.83713174e-01 -4.79945689e-01
6.44351184e-01 4.25452530e-01 8.01739842e-02 7.28040218e-01
6.05791926e-01 -2.31249735e-01 -5.58702886e-01 6.80641353e-01
1.89019704e+00 4.51032609e-01 1.05453832e-02 8.16133469e-02
1.29464120e-01 2.62855947e-01 2.02402130e-01 8.21790278e-01
1.54799104e-01 6.12591028e-01 -1.06257282e-01 -4.06955004e-01
4.11621392e-01 5.72931767e-01 -1.61121953e-02 7.37365186e-01
-2.05267280e-01 1.21408785e-02 -9.67589080e-01 6.65062487e-01
-1.59279096e+00 -8.49831820e-01 -3.35487276e-01 1.69201708e+00
6.88230038e-01 -5.50381504e-02 5.25767021e-02 4.52413112e-01
5.92357993e-01 -5.86467944e-02 -3.10021460e-01 -6.85961545e-01
-1.55383036e-01 5.71529925e-01 5.04506767e-01 8.97898793e-01
-8.44283164e-01 1.68824762e-01 6.49730206e+00 6.87779605e-01
-1.23661077e+00 1.01397075e-01 3.55758905e-01 1.71375126e-01
-1.79646686e-02 1.49376005e-01 -4.61118102e-01 4.98175293e-01
1.39128351e+00 -1.05889976e+00 5.62126338e-01 5.21843612e-01
7.96481371e-01 -2.28775516e-01 -7.41546988e-01 6.35031998e-01
-4.77829874e-01 -1.32800436e+00 -6.93781674e-01 4.94285300e-02
9.57404077e-01 -7.05153048e-02 -1.75065964e-01 1.96104527e-01
9.23171118e-02 -7.66690552e-01 5.36595643e-01 7.31307805e-01
2.39807203e-01 -6.91002727e-01 4.46864605e-01 6.83412790e-01
-1.57232189e+00 -7.10281283e-02 4.16198112e-02 -3.11120719e-01
7.34972537e-01 8.64651024e-01 -4.08692770e-02 7.44334877e-01
6.82741046e-01 8.58202219e-01 1.01681083e-01 1.04960120e+00
3.27224284e-01 6.92850530e-01 -7.68948793e-01 9.59394798e-02
3.08188379e-01 -7.34077156e-01 6.32719696e-01 1.06155527e+00
6.02284372e-01 5.70432842e-01 -1.01911267e-02 1.22512722e+00
5.54710090e-01 -4.09044355e-01 -2.61963367e-01 2.31100827e-01
1.02426499e-01 1.51968992e+00 -7.22804427e-01 -2.74439931e-01
-4.27882671e-01 5.46339333e-01 -4.02042538e-01 8.31524968e-01
-1.14154422e+00 -4.81379122e-01 7.35123217e-01 3.07966352e-01
2.78359324e-01 -4.34857696e-01 -1.88250318e-01 -9.45590258e-01
1.01063596e-02 -3.33740294e-01 2.26859450e-01 -8.29306901e-01
-1.29993558e+00 7.87621677e-01 5.94059527e-01 -1.16783035e+00
-5.52690089e-01 -6.07839525e-01 -1.09046543e+00 1.48889112e+00
-1.34902394e+00 -7.71607697e-01 3.32460046e-01 6.59302175e-01
6.05722144e-02 9.56243798e-02 9.52750027e-01 3.67579818e-01
-6.56926513e-01 -2.45055869e-01 8.29343855e-01 -1.04270637e-01
-2.74604887e-01 -1.29541659e+00 2.32539728e-01 7.14079082e-01
-5.51707625e-01 8.21826397e-04 1.14952469e+00 -4.89387602e-01
-1.52856803e+00 -7.67984927e-01 7.58082032e-01 1.28487125e-01
1.10004759e+00 -6.33042380e-02 -1.11913788e+00 6.48184121e-01
7.17320800e-01 2.70654261e-01 2.26316795e-01 -5.29440761e-01
8.00212383e-01 -2.49845341e-01 -1.31563866e+00 1.78217694e-01
1.92712888e-01 -4.56450969e-01 -4.09752190e-01 5.69868445e-01
-4.96090529e-03 -4.20282543e-01 -1.34078503e+00 4.89982426e-01
-8.89113992e-02 -6.47046089e-01 7.67195761e-01 -3.63403082e-01
-3.36865574e-01 -2.95467377e-01 2.73052931e-01 -1.72338820e+00
-3.74107301e-01 -1.09382200e+00 -2.91624647e-02 1.01743901e+00
3.17749351e-01 -7.83812582e-01 3.80021840e-01 6.75254643e-01
-1.94590047e-01 -8.47648084e-01 -1.41802478e+00 -7.73398340e-01
8.03265452e-01 -2.75789917e-01 4.51050490e-01 1.14367914e+00
5.69850624e-01 -2.96554446e-01 1.21587746e-01 5.23920596e-01
6.45236015e-01 2.43071273e-01 -4.67696153e-02 -1.22485340e+00
-1.91835791e-01 -4.28483844e-01 1.61923207e-02 -4.50322419e-01
4.39491093e-01 -4.60383475e-01 -6.25775456e-02 -1.40714133e+00
-3.76780927e-01 -5.71297109e-01 -1.43197486e-02 -3.81025001e-02
4.38675046e-01 -6.28811568e-02 -1.20044202e-01 4.39800099e-02
2.59230494e-01 5.65340877e-01 8.69267881e-01 8.05334225e-02
-1.75694451e-01 1.95585132e-01 3.33114445e-01 7.45690703e-01
7.51293004e-01 -2.51255512e-01 -2.64659673e-01 -9.20837373e-02
-1.82047531e-01 6.64879918e-01 4.84440625e-01 -1.21210492e+00
4.62694705e-01 -2.57079124e-01 1.20397151e-01 -4.74645674e-01
3.59543949e-01 -9.19384897e-01 6.45147324e-01 7.20948696e-01
1.79254949e-01 5.25992513e-01 8.29004049e-01 1.97610892e-02
-2.47957185e-01 -8.71604383e-02 1.02129042e+00 -8.64660889e-02
-3.52245629e-01 1.81779951e-01 -1.03912413e+00 -6.96178824e-02
1.30132711e+00 2.57815301e-01 -6.00081086e-02 -2.99117684e-01
-1.10993469e+00 1.46599710e-01 3.06810170e-01 -3.29952270e-01
-1.19131364e-01 -1.20611668e+00 -7.22244799e-01 3.51646483e-01
-9.40371394e-01 -3.59527022e-01 4.37303305e-01 1.21137714e+00
-6.72113836e-01 5.89922011e-01 -1.36031389e-01 -6.30075872e-01
-6.78090513e-01 2.74792701e-01 1.09422660e+00 -4.54672217e-01
-6.92268491e-01 4.72754896e-01 -3.22633386e-01 -7.62405545e-02
-3.29866737e-01 -5.81057072e-01 1.70923099e-02 6.57812729e-02
1.37457654e-01 5.91325343e-01 2.92629778e-01 -9.09833729e-01
-4.29524690e-01 8.69131863e-01 7.91255832e-01 -2.67985135e-01
1.51908684e+00 -2.14141235e-02 -3.37318122e-01 3.12190741e-01
1.11336613e+00 -1.88894749e-01 -1.23412788e+00 2.43422017e-02
-1.95548654e-01 1.82842255e-01 4.17780489e-01 -6.34321511e-01
-1.27965188e+00 6.94298923e-01 2.80383855e-01 1.10300791e+00
1.24652052e+00 -3.53091449e-01 5.48892617e-01 6.19259803e-03
3.88774008e-01 -1.12397635e+00 -8.01731884e-01 4.88267303e-01
8.90980601e-01 -3.42997909e-01 1.49114296e-01 -4.86338675e-01
-1.79431215e-01 1.24269187e+00 1.11559086e-01 -4.90010649e-01
1.45041668e+00 1.17936909e+00 -2.43087888e-01 1.61696002e-01
-7.31709480e-01 2.58531928e-01 8.15126747e-02 4.81501311e-01
3.18499565e-01 -6.86736628e-02 -5.59026957e-01 7.34934866e-01
-3.48111354e-02 2.13691220e-02 5.78006029e-01 1.18374813e+00
-1.28378004e-01 -9.28164005e-01 -4.89426911e-01 1.22441553e-01
-3.30974340e-01 1.13058500e-01 2.81660050e-01 1.15214634e+00
-1.02302767e-01 8.68499398e-01 2.31518149e-01 2.68742830e-01
6.54152453e-01 1.83260486e-01 2.95411706e-01 -1.82041854e-01
-6.74239457e-01 2.28554204e-01 -1.15866981e-01 -2.54456758e-01
-2.01830298e-01 -8.23541939e-01 -1.49589586e+00 -6.94540739e-01
-1.88200936e-01 9.11343098e-01 5.11270106e-01 1.24019098e+00
-6.41893875e-03 7.37695932e-01 7.86229551e-01 -1.12270379e+00
-7.60105431e-01 -9.16863441e-01 -1.31707907e+00 -2.54046500e-01
3.95390898e-01 -5.61040580e-01 -9.09091115e-01 2.61813030e-02]
|
[5.959691047668457, 2.887789249420166]
|
889f9b48-3756-4587-89b6-59c4f842df73
|
a-self-adjusting-fusion-representation
|
2212.11772
| null |
https://arxiv.org/abs/2212.11772v1
|
https://arxiv.org/pdf/2212.11772v1.pdf
|
A Self-Adjusting Fusion Representation Learning Model for Unaligned Text-Audio Sequences
|
Inter-modal interaction plays an indispensable role in multimodal sentiment analysis. Due to different modalities sequences are usually non-alignment, how to integrate relevant information of each modality to learn fusion representations has been one of the central challenges in multimodal learning. In this paper, a Self-Adjusting Fusion Representation Learning Model (SA-FRLM) is proposed to learn robust crossmodal fusion representations directly from the unaligned text and audio sequences. Different from previous works, our model not only makes full use of the interaction between different modalities but also maximizes the protection of the unimodal characteristics. Specifically, we first employ a crossmodal alignment module to project different modalities features to the same dimension. The crossmodal collaboration attention is then adopted to model the inter-modal interaction between text and audio sequences and initialize the fusion representations. After that, as the core unit of the SA-FRLM, the crossmodal adjustment transformer is proposed to protect original unimodal characteristics. It can dynamically adapt the fusion representations by using single modal streams. We evaluate our approach on the public multimodal sentiment analysis datasets CMU-MOSI and CMU-MOSEI. The experiment results show that our model has significantly improved the performance of all the metrics on the unaligned text-audio sequences.
|
['Kai Gao', 'Hua Xu', 'Ruxuan Zhang', 'Kaicheng Yang']
|
2022-11-12
| null | null | null | null |
['multimodal-sentiment-analysis', 'multimodal-sentiment-analysis']
|
['computer-vision', 'natural-language-processing']
|
[ 1.78049222e-01 -2.11018994e-01 5.90242669e-02 -3.14119101e-01
-1.04084885e+00 -4.56714630e-01 6.56942010e-01 -2.13191241e-01
-4.51000512e-01 3.56014848e-01 4.93467391e-01 2.16614738e-01
-1.37022614e-01 -4.34402287e-01 -5.93054056e-01 -9.45077538e-01
4.39187616e-01 2.92759594e-02 -2.96903644e-02 -5.61325312e-01
7.08092451e-02 -2.42167842e-02 -1.82854474e+00 8.15913081e-01
7.72427320e-01 1.24411130e+00 8.62724856e-02 6.60217166e-01
-4.04395580e-01 6.14837527e-01 -2.83431947e-01 -4.42515135e-01
-7.93949738e-02 -5.86751103e-01 -7.43862450e-01 8.88138041e-02
1.73353687e-01 4.66962829e-02 -2.34177202e-01 1.08441734e+00
8.56975138e-01 3.39018971e-01 7.29607403e-01 -1.47314477e+00
-3.92826855e-01 1.07353115e+00 -9.05058384e-01 1.03410237e-01
4.66595501e-01 -2.23420858e-01 8.81227314e-01 -1.06176853e+00
2.09932625e-01 1.53772891e+00 3.56839389e-01 4.37549174e-01
-9.32133198e-01 -8.35455656e-01 4.44857061e-01 5.39462507e-01
-1.34322870e+00 -6.39574766e-01 1.04440963e+00 -2.99599171e-01
4.00714070e-01 2.64093250e-01 2.78180867e-01 1.06044650e+00
-2.80483775e-02 1.05812120e+00 7.99806237e-01 -5.65138698e-01
-2.91962922e-01 2.60890543e-01 2.14376971e-01 1.91750988e-01
-5.82093716e-01 -1.70734465e-01 -7.79425621e-01 9.96464714e-02
1.68741614e-01 2.05942705e-01 -4.11801159e-01 -2.15456337e-01
-1.43869770e+00 6.42625451e-01 2.55768538e-01 6.47363365e-01
-2.19014063e-01 -3.02073866e-01 6.22074723e-01 3.48060668e-01
2.40076650e-02 -1.79389074e-01 -3.45015764e-01 -1.10751078e-01
-5.51095128e-01 -1.57518357e-01 2.61028796e-01 7.71372378e-01
7.59684980e-01 8.27433392e-02 -1.91748396e-01 1.12447500e+00
5.57789564e-01 7.87702620e-01 8.80137503e-01 -5.91344476e-01
9.05112505e-01 6.79849863e-01 -1.60872653e-01 -1.09781945e+00
-5.54120362e-01 -1.43879384e-01 -1.05772138e+00 -7.50652328e-02
6.32535815e-02 -2.72476077e-01 -6.29410028e-01 1.97668231e+00
3.87663901e-01 7.56919459e-02 4.68822807e-01 8.92069757e-01
1.18446791e+00 8.60966146e-01 1.27708718e-01 -4.29815382e-01
1.42710936e+00 -8.51076663e-01 -1.04662693e+00 7.08424523e-02
3.33953619e-01 -1.12315369e+00 7.99242258e-01 3.84885997e-01
-1.14106166e+00 -9.67888474e-01 -1.07125890e+00 2.87058018e-02
-4.91582185e-01 2.15618700e-01 1.27082979e-02 4.60963905e-01
-5.26863933e-01 1.78061631e-02 -5.00043035e-01 -1.92016050e-01
5.96539043e-02 4.02099013e-01 -5.86123645e-01 -7.63968378e-02
-1.57259238e+00 5.46457469e-01 8.11788499e-01 4.93353456e-01
-4.10668701e-01 -3.33422333e-01 -1.01124370e+00 5.36379330e-02
1.76553816e-01 -3.69780272e-01 8.97677481e-01 -1.47412503e+00
-1.51673603e+00 3.84950608e-01 -1.19669378e-01 1.57806799e-02
1.22998297e-01 -7.67955780e-02 -9.61751342e-01 2.40023971e-01
-2.64849842e-01 7.22997367e-01 1.15871441e+00 -1.45717335e+00
-1.02023959e+00 -2.56702513e-01 -1.04524732e-01 7.30596125e-01
-7.35415161e-01 -6.01425543e-02 -5.15622437e-01 -7.35454142e-01
2.81207144e-01 -7.72224545e-01 3.22261810e-01 -7.93868661e-01
-2.21230075e-01 7.90469162e-03 8.04403543e-01 -6.32235825e-01
1.43789220e+00 -2.32818604e+00 8.16886067e-01 3.12889636e-01
-1.87583581e-01 1.55830588e-02 -3.63160789e-01 3.40318173e-01
-3.84261698e-01 -2.12901622e-01 -2.35968322e-01 -5.12132943e-01
1.41798973e-01 2.59141456e-02 -4.86826837e-01 2.48616114e-01
4.82128039e-02 7.45160997e-01 -5.11824608e-01 -6.96043372e-01
4.23615754e-01 7.99844503e-01 -2.58898526e-01 2.84317404e-01
3.38872880e-01 6.44474983e-01 -1.53616771e-01 5.93964458e-01
9.49240744e-01 1.17267683e-01 8.09878260e-02 -9.69097137e-01
5.95568679e-02 -3.73254657e-01 -1.53222001e+00 2.06857491e+00
-3.68983805e-01 1.60375968e-01 1.82311907e-01 -1.09887064e+00
7.71540701e-01 4.69122916e-01 6.00225329e-01 -8.60431314e-01
4.63651717e-01 -3.41075137e-02 -9.06675905e-02 -5.97612619e-01
4.69970226e-01 -2.65493453e-01 -3.93320709e-01 3.71969134e-01
5.00678003e-01 1.02256544e-01 9.56551582e-02 1.86840728e-01
3.03762347e-01 -6.27238005e-02 5.21326438e-02 1.45316869e-01
1.34426093e+00 -5.85946977e-01 5.68530917e-01 2.02463061e-01
-9.67633910e-03 8.16237390e-01 1.85044646e-01 -9.36799422e-02
-5.62256455e-01 -1.04522443e+00 -7.61179775e-02 1.51961076e+00
4.73351657e-01 -4.24159557e-01 -5.21270633e-01 -6.90297782e-01
-3.11554015e-01 4.39342350e-01 -7.13190138e-01 -5.02634287e-01
-4.49314713e-01 -7.57095873e-01 3.81971091e-01 4.81771082e-01
4.39388365e-01 -9.23296630e-01 -1.08449198e-01 4.22593802e-02
-7.46630788e-01 -9.39032555e-01 -6.51166856e-01 7.85519555e-02
-4.55550343e-01 -8.39203358e-01 -7.42958426e-01 -8.25190127e-01
4.47475940e-01 3.37335914e-01 5.31307876e-01 -2.79774755e-01
2.57456392e-01 6.96754694e-01 -6.57689035e-01 -2.38439232e-01
-2.23341435e-01 2.43528157e-01 4.70734015e-03 8.92723501e-01
2.65457332e-01 -5.85255325e-01 -2.94819713e-01 4.42785054e-01
-1.39006376e+00 1.04081109e-01 5.10581851e-01 1.09398091e+00
6.19664490e-01 8.97466242e-02 7.79298961e-01 -4.63387370e-01
4.38741148e-01 -7.31005609e-01 -1.25091359e-01 4.35610950e-01
-1.11224346e-01 4.11999933e-02 5.84999621e-01 -8.77572775e-01
-1.33670115e+00 1.41083732e-01 -7.36732855e-02 -6.32345319e-01
6.33455953e-03 7.60694623e-01 -8.42206776e-01 1.94521397e-01
1.58721745e-01 3.07153344e-01 -4.27859947e-02 -2.77679205e-01
5.50107419e-01 8.80541265e-01 7.11115003e-01 -5.48860669e-01
7.60376811e-01 3.88891369e-01 -2.72671998e-01 -7.02466428e-01
-4.95899558e-01 -3.11172366e-01 -7.34600484e-01 -4.19425219e-01
9.01453793e-01 -1.01966321e+00 -8.23003292e-01 6.64208651e-01
-9.57711101e-01 4.03579712e-01 -3.49698663e-02 6.89840257e-01
-3.68203551e-01 6.36123419e-01 -3.40716600e-01 -7.91788340e-01
-3.39069337e-01 -1.34697878e+00 1.11536968e+00 5.79365373e-01
4.00792286e-02 -8.12525690e-01 1.30691350e-01 6.07666969e-01
2.71078795e-01 -2.86336429e-02 8.07588518e-01 -6.63928986e-01
-3.52159329e-02 -1.18358985e-01 3.14028710e-02 4.89808232e-01
2.97363579e-01 -2.15361193e-02 -1.18617737e+00 -2.56264538e-01
-4.80551273e-02 -4.72298026e-01 1.01429224e+00 1.19809709e-01
8.47163618e-01 -3.74546349e-02 -1.88377162e-04 5.81918478e-01
9.09340501e-01 2.57285208e-01 6.55938864e-01 1.96136534e-01
8.11420619e-01 7.64374316e-01 7.16619968e-01 5.47117054e-01
6.16791725e-01 4.09640253e-01 4.94666278e-01 6.93030581e-02
1.71946988e-01 -7.99208730e-02 7.02707171e-01 1.39365911e+00
1.48795322e-01 -1.59903541e-01 -5.24050057e-01 4.14640397e-01
-2.17489767e+00 -1.13061762e+00 1.99585214e-01 2.03829074e+00
7.57837355e-01 -1.86222076e-01 9.02997702e-02 3.58492076e-01
9.37952518e-01 9.27729383e-02 -3.82870942e-01 -9.49365944e-02
-6.39023066e-01 -1.20572664e-01 -1.54319435e-01 4.47606653e-01
-1.41815937e+00 5.05540073e-01 4.63803816e+00 1.06971431e+00
-1.15905154e+00 2.05268651e-01 2.40364715e-01 -2.99614727e-01
-4.51051414e-01 -2.86561757e-01 -5.91841340e-01 6.83833420e-01
6.64462388e-01 1.16596192e-01 5.97515404e-01 1.88527331e-01
-2.05658361e-01 6.96241707e-02 -9.44106579e-01 1.38624060e+00
4.00583386e-01 -7.41783023e-01 2.82058746e-01 -4.46063280e-01
5.86141348e-01 -4.14474130e-01 3.52277815e-01 4.93412822e-01
-1.78577974e-01 -9.19671059e-01 9.24358070e-01 1.02729750e+00
7.01196015e-01 -1.23679399e+00 9.90923166e-01 2.42177442e-01
-1.53278995e+00 -3.02269191e-01 -1.42060965e-01 5.37038028e-01
3.40402752e-01 2.21405938e-01 -4.02628817e-02 1.07283127e+00
8.43696177e-01 8.54060769e-01 -6.11216068e-01 5.24783313e-01
2.11985409e-01 1.94439143e-01 -3.21064532e-01 5.85635364e-01
-2.30150282e-01 -1.37477383e-01 4.33170795e-01 1.25425017e+00
4.74131882e-01 -6.78147078e-02 1.17189519e-01 2.71296859e-01
-1.88099861e-03 3.07549506e-01 -3.04561853e-01 -6.16859570e-02
4.14743751e-01 1.41268468e+00 -3.00088793e-01 -1.37066782e-01
-6.28738523e-01 8.16397667e-01 -2.97861430e-03 4.64490384e-01
-9.91243541e-01 -4.48746443e-01 4.93836939e-01 -6.89478576e-01
4.05819416e-01 2.08517104e-01 -6.25876486e-02 -1.46015346e+00
-1.06539950e-01 -1.15118325e+00 8.46022367e-01 -8.66498590e-01
-1.47287285e+00 7.54826128e-01 -1.16167322e-03 -1.59074461e+00
-2.08676726e-01 -2.84048259e-01 -4.13367629e-01 9.84670758e-01
-1.44497144e+00 -1.74394059e+00 -2.51595348e-01 1.24368918e+00
3.45068157e-01 -5.04169285e-01 6.55661106e-01 5.95993876e-01
-7.49848604e-01 9.19626117e-01 -1.09285913e-01 -3.50675546e-02
1.06375802e+00 -7.76525319e-01 -6.91430271e-01 8.09212744e-01
-2.30943388e-03 5.20586789e-01 4.04579550e-01 -3.29362273e-01
-1.52367830e+00 -8.08762908e-01 3.61863405e-01 -1.21028908e-01
5.87467015e-01 -1.77660540e-01 -1.06281006e+00 5.55207014e-01
7.24200189e-01 -2.26479828e-01 9.47155654e-01 -1.22522354e-01
-4.31280494e-01 -5.80552578e-01 -7.22457647e-01 3.93546164e-01
4.59612310e-01 -7.38166809e-01 -8.46654832e-01 -2.20733806e-01
7.50797868e-01 -3.25009048e-01 -9.43890989e-01 6.86659098e-01
7.00944006e-01 -7.13048756e-01 1.01873636e+00 -4.83285189e-01
2.51822621e-01 -6.96474135e-01 -6.82376206e-01 -1.31143022e+00
-1.03799187e-01 -3.20263624e-01 -5.67031391e-02 1.90566075e+00
3.13027322e-01 -4.58684772e-01 3.25324051e-02 2.69039214e-01
-7.72596151e-02 -2.19135016e-01 -1.07569146e+00 -7.00925849e-03
-8.42582285e-02 -5.65058887e-01 7.52519429e-01 1.06459773e+00
4.58712667e-01 7.14782238e-01 -6.74573958e-01 3.65124226e-01
3.62203360e-01 2.60396481e-01 6.32433355e-01 -9.75676000e-01
-1.89521059e-01 -5.86537123e-01 -2.18573213e-01 -7.34286249e-01
4.05552745e-01 -9.29596305e-01 -8.71293694e-02 -1.01370609e+00
1.91048279e-01 7.44705498e-02 -8.62103999e-01 4.63705003e-01
-2.88719207e-01 9.86365005e-02 3.87019068e-01 4.92443964e-02
-7.80541241e-01 1.03618920e+00 1.09459090e+00 -4.55657482e-01
-1.87444344e-01 -1.15224950e-01 -7.96796143e-01 5.97858429e-01
4.91182894e-01 -1.19904317e-01 -6.22358382e-01 -3.61614406e-01
3.94192398e-01 1.60065070e-01 -5.56738935e-02 -9.66902077e-01
4.80083019e-01 -1.55706987e-01 4.57836747e-01 -1.06580341e+00
6.36940479e-01 -1.26580787e+00 1.95164174e-01 -1.04275398e-01
-4.20968562e-01 -2.13668169e-03 3.42445761e-01 4.58024979e-01
-6.98920369e-01 -1.02951467e-01 7.60251522e-01 2.75713265e-01
-6.23688519e-01 5.53091839e-02 -3.00860137e-01 -1.43644094e-01
9.30279613e-01 1.25591755e-02 -3.29852879e-01 -4.58490044e-01
-7.85125375e-01 5.51011562e-01 8.57930109e-02 6.47870839e-01
7.73765624e-01 -1.81203985e+00 -5.56600511e-01 3.58569622e-01
2.78338671e-01 -2.43268356e-01 1.08827305e+00 1.09544635e+00
2.91678548e-01 -3.63605320e-02 -3.79670948e-01 -7.15497673e-01
-1.36085832e+00 7.90641963e-01 3.85406077e-01 9.30260681e-03
5.12153767e-02 6.18363857e-01 4.97489035e-01 -6.97642863e-01
4.05740947e-01 -1.58419777e-02 -9.52098310e-01 7.73851871e-01
7.69543827e-01 3.06529164e-01 -1.89044792e-02 -1.31617057e+00
-3.49103838e-01 8.97363305e-01 -6.26108497e-02 -3.55198205e-01
1.10373533e+00 -7.24454939e-01 -2.67352015e-01 8.39618146e-01
1.30921316e+00 5.90794124e-02 -9.31507409e-01 -5.01454532e-01
-6.04248643e-01 -1.29837885e-01 -4.31032415e-04 -6.75486803e-01
-1.32842898e+00 1.14789259e+00 8.99726927e-01 -9.58144944e-03
1.65333891e+00 -1.78167343e-01 7.06387460e-01 6.14590682e-02
-1.13253491e-02 -1.23036528e+00 1.88197926e-01 5.75423896e-01
1.00920212e+00 -1.24428451e+00 -3.05015564e-01 4.67392243e-02
-1.10425436e+00 1.22655809e+00 6.32177830e-01 3.91958505e-01
7.58299351e-01 1.30711108e-01 2.00324342e-01 1.73822001e-01
-4.77031827e-01 -1.67910069e-01 6.53300226e-01 5.05225301e-01
2.49755472e-01 -1.77635431e-01 -1.67997573e-02 1.24363232e+00
9.52994004e-02 -3.50358069e-01 2.34592959e-01 8.26687634e-01
-2.73635030e-01 -1.10794675e+00 -8.72674465e-01 5.47643080e-02
-2.32795760e-01 7.17957169e-02 -9.65023339e-02 4.59500015e-01
3.19085836e-01 1.27050877e+00 -1.09713212e-01 -9.81165826e-01
4.51663077e-01 4.12329614e-01 2.54228801e-01 -2.00385507e-03
-6.13137007e-01 5.62303841e-01 -3.15522105e-01 -3.58119458e-01
-8.71210277e-01 -8.26640904e-01 -1.34367180e+00 -1.39383748e-01
-2.53154039e-01 1.52569085e-01 4.72673327e-01 1.01968920e+00
3.43220472e-01 6.90210521e-01 9.68831360e-01 -1.09551883e+00
-2.63449579e-01 -1.01627076e+00 -5.29209554e-01 6.73260033e-01
5.11043668e-01 -8.01663816e-01 -3.39718252e-01 2.05804422e-01]
|
[13.192227363586426, 4.9931159019470215]
|
b2569d88-e08c-4d3d-8e08-8c586cdd7a19
|
dfuc2020-analysis-towards-diabetic-foot-ulcer
|
2004.11853
| null |
https://arxiv.org/abs/2004.11853v3
|
https://arxiv.org/pdf/2004.11853v3.pdf
|
DFUC2020: Analysis Towards Diabetic Foot Ulcer Detection
|
Every 20 seconds, a limb is amputated somewhere in the world due to diabetes. This is a global health problem that requires a global solution. The MICCAI challenge discussed in this paper, which concerns the automated detection of diabetic foot ulcers using machine learning techniques, will accelerate the development of innovative healthcare technology to address this unmet medical need. In an effort to improve patient care and reduce the strain on healthcare systems, recent research has focused on the creation of cloud-based detection algorithms. These can be consumed as a service by a mobile app that patients (or a carer, partner or family member) could use themselves at home to monitor their condition and to detect the appearance of a diabetic foot ulcer (DFU). Collaborative work between Manchester Metropolitan University, Lancashire Teaching Hospital and the Manchester University NHS Foundation Trust has created a repository of 4,000 DFU images for the purpose of supporting research toward more advanced methods of DFU detection. Based on a joint effort involving the lead scientists of the UK, US, India and New Zealand, this challenge will solicit original work, and promote interactions between researchers and interdisciplinary collaborations. This paper presents a dataset description and analysis, assessment methods, benchmark algorithms and initial evaluation results. It facilitates the challenge by providing useful insights into state-of-the-art and ongoing research. This grand challenge takes on even greater urgency in a peri and post-pandemic period, where stresses on resource utilization will increase the need for technology that allows people to remain active, healthy and intact in their home.
|
["Claire O'Shea", 'Bijan Najafi', 'Arun G. Maiya', 'Satyan Rajbhandari', 'Justina Wu', 'Eibe Frank', 'Andrew Boulton', 'Neil D. Reeves', 'David Armstrong', 'Bill Cassidy', 'Pappachan Joseph', 'Moi Hoon Yap', 'David Gillespie']
|
2020-04-24
| null | null | null | null |
['diabetic-foot-ulcer-detection']
|
['medical']
|
[ 4.29454982e-01 -2.34940276e-01 -2.82304019e-01 -3.20171118e-02
-6.97600126e-01 -3.03706050e-01 3.05441841e-02 7.62285888e-01
-4.25440758e-01 8.57591510e-01 5.02330601e-01 -4.96901840e-01
-3.18462938e-01 -8.22070479e-01 -3.83080617e-02 -4.22198474e-01
-3.05651724e-01 5.39203227e-01 1.25994265e-01 -1.73391744e-01
3.06787252e-01 3.30735624e-01 -1.35657334e+00 6.38664305e-01
8.83342028e-01 7.39127100e-01 3.68000925e-01 8.58574569e-01
2.71110505e-01 7.44148374e-01 -1.69924825e-01 -7.64165297e-02
3.18491131e-01 -6.29311323e-01 -5.95477462e-01 -1.39950931e-01
1.47880195e-02 -4.96704549e-01 2.40361556e-01 4.91755277e-01
1.01183403e+00 -4.34319675e-01 4.13457662e-01 -7.33553350e-01
8.53330195e-02 -3.92303318e-01 -2.23858878e-01 2.80395895e-01
6.57584369e-01 4.00433868e-01 2.35064596e-01 -2.86068827e-01
9.48148429e-01 7.73576140e-01 1.06360364e+00 2.82487273e-01
-1.00057697e+00 -2.88070887e-01 -2.97464758e-01 5.41742206e-01
-8.50853503e-01 -2.14988470e-01 5.81505941e-03 -4.61648524e-01
1.03401041e+00 6.51310861e-01 1.19826233e+00 1.10181391e+00
5.48071206e-01 5.79827189e-01 1.18493235e+00 -5.53286672e-01
4.39235628e-01 -9.23534855e-02 -2.98164815e-01 4.47369128e-01
4.30143237e-01 -1.28426597e-01 -5.18836789e-02 -5.22682667e-01
9.80676949e-01 4.29222524e-01 -3.05328727e-01 -4.11799461e-01
-1.29071355e+00 9.27657425e-01 2.06495762e-01 5.26958331e-03
-7.74064720e-01 -2.29508132e-01 7.42485464e-01 2.39973381e-01
2.51922935e-01 1.77429557e-01 -6.57002270e-01 -7.93998003e-01
-3.64883512e-01 3.60059887e-01 7.77642906e-01 2.25552127e-01
2.26041228e-02 -7.12196350e-01 2.09222898e-01 9.24211204e-01
3.03546250e-01 2.66935766e-01 2.93626428e-01 -9.92007494e-01
-7.09291920e-03 9.25335407e-01 1.38913870e-01 -6.69868112e-01
-3.81786466e-01 -1.29116654e-01 -6.58758461e-01 4.91400242e-01
2.41341069e-01 -5.65029681e-01 -1.04826820e+00 7.49296606e-01
4.83311296e-01 -9.23330188e-02 5.43390848e-02 1.01232731e+00
1.60398349e-01 1.94132119e-01 6.30202889e-02 -8.45583007e-02
1.38356793e+00 -5.20264030e-01 -4.33471233e-01 -6.63016587e-02
8.85431111e-01 -9.57405806e-01 6.28556132e-01 7.67465651e-01
-9.14003611e-01 1.09321900e-01 -9.11811769e-01 3.78582388e-01
-3.45419735e-01 -4.45286244e-01 5.39935231e-01 6.55947506e-01
-7.55184591e-01 3.62642199e-01 -1.14055216e+00 -1.48440421e+00
7.31585026e-01 1.86539143e-01 -4.38979954e-01 -8.66410732e-01
-9.27343369e-01 1.35540354e+00 -1.95136145e-01 -1.46958977e-01
-2.16111556e-01 -8.03547859e-01 -5.74075520e-01 -7.34653711e-01
2.59715728e-02 -1.05355430e+00 8.38556588e-01 -5.33942640e-01
-8.14083636e-01 7.94274986e-01 1.45668879e-01 -5.02010822e-01
1.08320665e+00 -2.35594586e-01 -3.97813857e-01 7.69078955e-02
4.21991348e-01 2.54093289e-01 -1.52520905e-03 -6.34608567e-01
-1.42401206e+00 -7.16796696e-01 -4.39975470e-01 1.49506390e-01
1.19379677e-01 2.67228782e-01 4.04036604e-02 -3.49454343e-01
-1.00727484e-01 -8.48768651e-01 -5.88091731e-01 9.21722874e-02
7.79825449e-02 2.02013507e-01 9.06139970e-01 -9.51849878e-01
9.56882060e-01 -1.75864756e+00 -2.03871474e-01 2.55727917e-01
-1.06011264e-01 4.06295419e-01 3.30406129e-01 9.17664230e-01
2.34594047e-01 -1.06637329e-02 -1.01483263e-01 5.15464187e-01
-4.07895535e-01 5.02512872e-01 4.66277748e-01 5.47647655e-01
1.97995380e-01 4.95390683e-01 -1.10590708e+00 -1.26536295e-01
5.44076800e-01 8.21700513e-01 -3.54648829e-01 -1.25166580e-01
3.28946710e-01 4.71629500e-01 -1.51669100e-01 8.28093231e-01
3.05501312e-01 -9.09699500e-02 4.42659020e-01 1.70354843e-01
-3.32846940e-01 -2.48175524e-02 -1.23709583e+00 1.33033812e+00
-2.66606718e-01 4.02117580e-01 1.42138600e-01 -8.55956137e-01
6.30383790e-01 3.78744841e-01 8.96129429e-01 -7.38035142e-01
8.15570131e-02 5.62014580e-01 6.58224076e-02 -1.06092298e+00
-3.03003281e-01 -6.26618043e-02 3.80979359e-01 3.23091626e-01
-6.53033733e-01 7.00983822e-01 1.54060543e-01 -1.77309215e-02
1.54864740e+00 3.70509624e-01 4.83795762e-01 -2.29339018e-01
-2.53242180e-02 3.51764917e-01 2.96809465e-01 4.22261357e-01
-6.18885040e-01 4.31806147e-01 2.26812959e-01 -9.54872668e-01
-1.09955740e+00 -9.77643013e-01 -5.79563320e-01 6.39194667e-01
-2.02336460e-01 -2.82503068e-01 -3.58278364e-01 -2.05447987e-01
4.35694337e-01 1.17348760e-01 -6.22835219e-01 -1.89609881e-02
-5.08118987e-01 -9.38959777e-01 1.76785767e-01 4.94552851e-01
4.47240084e-01 -8.63200247e-01 -1.58082449e+00 7.46407151e-01
-3.06846768e-01 -5.89904368e-01 2.54278094e-01 3.81868444e-02
-7.68131495e-01 -1.11512423e+00 -1.35377526e+00 -8.92143428e-01
3.83993566e-01 5.85587397e-02 4.13859934e-01 5.22528812e-02
-1.32281363e+00 4.43697870e-01 -4.80993956e-01 -8.62540901e-01
-3.12203258e-01 -4.04479980e-01 -3.80283855e-02 -3.96474987e-01
7.41781056e-01 -5.09208739e-01 -1.28741407e+00 2.07608566e-01
-5.37644088e-01 1.81082804e-02 6.31812990e-01 4.55008894e-01
5.47368348e-01 -2.77419537e-01 7.98297822e-01 -7.37512469e-01
7.04249203e-01 -7.42978930e-01 1.23230487e-01 2.34656893e-02
-8.71540368e-01 -5.76528490e-01 -3.10122073e-01 -9.94449630e-02
-4.76556063e-01 1.31057933e-01 -1.02467975e-02 3.49447280e-01
-2.39024177e-01 6.53185010e-01 4.07177627e-01 -8.81721918e-03
9.64793980e-01 -1.75579950e-01 6.42785788e-01 -4.92692202e-01
-2.39049330e-01 1.23468196e+00 3.47754002e-01 2.40971837e-02
-2.68952012e-01 5.69281876e-01 -5.33212628e-03 -9.66088891e-01
2.78262477e-02 -1.00737345e+00 -3.04390222e-01 -5.32031476e-01
7.18548298e-01 -9.23520029e-01 -2.95555562e-01 5.32340348e-01
-7.20372140e-01 -3.88200074e-01 1.87702756e-02 4.41332817e-01
-3.15866232e-01 1.34769022e-01 -3.25462669e-01 -8.69840086e-01
-3.49814266e-01 -7.92409182e-01 6.57754421e-01 -1.05894461e-01
-8.03842604e-01 -8.34406078e-01 3.94348860e-01 5.43347418e-01
9.08030748e-01 1.13944185e+00 9.74685013e-01 -1.81271676e-02
-1.99493662e-01 -7.46516645e-01 -3.80929619e-01 3.24345201e-01
7.67346561e-01 -1.08185254e-01 -3.37209463e-01 -2.37282887e-01
-1.79364458e-01 7.19444007e-02 5.58077693e-01 6.95520043e-01
2.36769930e-01 -1.79459929e-01 -6.60337985e-01 -3.21369767e-02
1.64273155e+00 4.63204235e-01 1.00117493e+00 9.24102306e-01
1.21775694e-01 4.89619374e-01 6.29045546e-01 6.29083693e-01
5.58383882e-01 2.04849467e-01 4.39500213e-01 -4.87554342e-01
-2.16812029e-01 5.00278652e-01 4.75233085e-02 9.98683348e-02
-8.53586376e-01 2.50968367e-01 -1.27125227e+00 8.68802428e-01
-2.01685143e+00 -7.80684888e-01 -5.50224602e-01 2.55056310e+00
5.67359090e-01 7.36563876e-02 6.02692008e-01 2.09750369e-01
5.30472815e-01 -8.71656358e-01 -4.65547144e-01 -5.09018004e-01
3.07673633e-01 3.66638571e-01 6.63643420e-01 1.63234383e-01
-7.70007789e-01 1.52050152e-01 5.92064095e+00 -2.44688541e-01
-1.19892001e+00 3.54383849e-02 5.02197206e-01 -1.45066395e-01
3.62373412e-01 -2.71924645e-01 -1.83034912e-01 5.63275456e-01
8.74562681e-01 -1.61769122e-01 2.07135320e-01 7.32840955e-01
5.24287164e-01 -5.45377970e-01 -7.13784039e-01 5.92809021e-01
-1.45239502e-01 -1.45150840e+00 -4.91946936e-01 4.78827208e-01
5.46795666e-01 5.37991166e-01 -3.85347515e-01 -3.19958001e-01
-3.27935304e-05 -8.57149422e-01 8.03066865e-02 7.33993769e-01
6.74205184e-01 -8.80445421e-01 1.09182978e+00 6.46829903e-02
-7.36759961e-01 -2.79066145e-01 1.83308303e-01 -6.03049457e-01
4.64309931e-01 3.84549618e-01 -1.07524049e+00 2.22337604e-01
1.04566836e+00 4.03552055e-01 -2.85686374e-01 1.63847864e+00
4.03728157e-01 2.76680440e-01 -3.27397764e-01 -6.89165220e-02
1.34708017e-01 -7.14703044e-03 2.82192886e-01 1.30283916e+00
1.12484448e-01 5.33719808e-02 1.42774820e-01 -2.71440744e-01
8.63270223e-01 4.47861403e-01 -5.30779779e-01 1.81929022e-02
-6.57425076e-02 9.61643159e-01 -8.09276879e-01 -6.74452558e-02
-5.01253068e-01 1.13365841e+00 -3.92667651e-01 3.32140289e-02
-2.48505190e-01 -5.04564047e-01 9.45810497e-01 9.71719027e-01
-5.72667494e-02 1.63189039e-01 -3.68796438e-01 -4.08465683e-01
-7.81757161e-02 -1.11662459e+00 5.64413905e-01 -3.79577160e-01
-1.06252229e+00 -7.67254923e-03 -4.21557039e-01 -9.73403275e-01
-3.12711447e-01 -5.76889873e-01 -3.01457554e-01 9.62810040e-01
-1.22765386e+00 -1.01118517e+00 -3.61677229e-01 4.33755070e-01
3.02673668e-01 6.68238327e-02 1.41933513e+00 4.29504782e-01
-2.17661947e-01 7.35816061e-02 1.80840090e-01 -1.54052630e-01
8.54771376e-01 -8.94093037e-01 4.42355126e-02 2.53479183e-01
-8.08882236e-01 5.24493933e-01 5.60328364e-01 -1.08253562e+00
-1.41923571e+00 -1.06937587e+00 9.08838689e-01 -5.21388412e-01
4.42727417e-01 9.38220918e-02 -4.91407037e-01 5.07800102e-01
3.94496135e-02 -3.49834055e-01 1.08893311e+00 -7.10945129e-02
3.52328539e-01 1.70436889e-01 -1.55985487e+00 4.07412469e-01
8.05016279e-01 -1.32031187e-01 -1.95835471e-01 7.98496485e-01
-1.09495051e-01 -3.02796245e-01 -1.09794128e+00 5.97415984e-01
1.25710464e+00 -9.35978770e-01 7.34134197e-01 -7.54801810e-01
6.01599216e-01 4.82658446e-02 -1.55628651e-01 -8.47017229e-01
-2.30864555e-01 -5.73527336e-01 1.97897002e-01 4.09246922e-01
1.01152569e-01 -8.98375392e-01 8.03300440e-01 6.20566964e-01
1.40144944e-01 -1.23302686e+00 -9.04675364e-01 -4.35395837e-01
-2.13636070e-01 -2.48328313e-01 5.39487340e-02 9.69483376e-01
3.54454756e-01 -6.17407374e-02 -2.31268462e-02 -5.40405624e-02
5.95612288e-01 -3.39972973e-01 6.67395353e-01 -1.47502995e+00
1.12267107e-01 -2.18660265e-01 -1.22722685e+00 1.64136395e-01
-1.41945338e+00 -4.81269956e-01 -3.39087963e-01 -2.51665211e+00
1.57957554e-01 -3.24586809e-01 -2.89855152e-01 5.27270615e-01
8.63040015e-02 3.29681098e-01 -2.81163037e-01 1.85972497e-01
-5.38868771e-04 -5.60203910e-01 1.13997352e+00 1.46168500e-01
-4.62593585e-01 1.97677851e-01 -9.29356933e-01 3.57007354e-01
9.14097965e-01 -2.61192352e-01 -1.31472498e-01 -6.64568618e-02
2.40393281e-01 -1.76894754e-01 4.67193067e-01 -1.22646558e+00
1.43932402e-01 -2.09251106e-01 5.77136397e-01 -1.96024969e-01
-8.54810793e-03 -7.80316234e-01 4.11358297e-01 1.35371137e+00
2.77391762e-01 -6.93659559e-02 1.18592337e-01 3.39449942e-01
2.27566332e-01 2.66943216e-01 6.83040142e-01 -6.17061615e-01
-5.71905315e-01 -1.43338576e-01 -7.35714197e-01 -5.22750616e-01
1.44631517e+00 -5.23403287e-01 -2.89751947e-01 -2.64574140e-01
-9.50316191e-01 4.20205772e-01 6.52862072e-01 3.43548298e-01
4.97085631e-01 -1.00306821e+00 -1.16553855e+00 2.92931408e-01
3.51304173e-01 -7.93430731e-02 3.62732381e-01 1.29565024e+00
-1.13667512e+00 2.38140523e-01 -6.75583303e-01 -5.76359332e-01
-1.55488586e+00 2.88902909e-01 2.33203635e-01 2.46381491e-01
-1.14465582e+00 5.79895318e-01 -7.79636085e-01 4.02152166e-02
2.61005163e-01 -5.51232584e-02 7.39936084e-02 8.53768513e-02
9.45025325e-01 7.45547652e-01 3.05226058e-01 -2.20852420e-01
-5.28600574e-01 1.81350023e-01 -2.66984671e-01 1.42920047e-01
1.62447739e+00 -9.99813899e-02 -2.28539445e-02 1.25846177e-01
1.02309012e+00 -3.41295272e-01 -8.51328254e-01 4.95866954e-01
-2.22332492e-01 -4.50192511e-01 2.20546931e-01 -1.39873457e+00
-4.61242199e-01 3.76048774e-01 1.54182601e+00 3.01487207e-01
1.08196867e+00 -6.46435320e-02 1.04438114e+00 -6.68632984e-02
4.93970543e-01 -1.16625607e+00 -2.85209686e-01 -1.91533715e-01
8.41060579e-01 -1.15202606e+00 6.27407655e-02 -3.88260603e-01
-5.37368119e-01 1.04940557e+00 -4.19191010e-02 -7.94761851e-02
5.03769159e-01 3.05774093e-01 5.88260710e-01 -1.74389556e-01
-4.90691066e-01 -1.58315212e-01 -6.29604936e-01 1.00452244e+00
3.62821370e-01 3.51742983e-01 -8.87554944e-01 1.31543577e-01
3.31750900e-01 8.01463246e-01 4.12907720e-01 1.82056940e+00
-5.92085660e-01 -1.48847556e+00 -3.12689751e-01 1.05390012e+00
-5.07361531e-01 5.12317643e-02 -5.13239682e-01 7.57386327e-01
4.36976641e-01 6.96329474e-01 -1.11304700e-01 -7.56882653e-02
5.60838282e-01 1.04397677e-01 6.34268403e-01 -3.39459896e-01
-2.87001729e-01 2.29766294e-01 4.58497316e-01 -4.95866567e-01
-3.44931334e-01 -9.50626850e-01 -1.16252518e+00 -2.04069152e-01
1.20630503e-01 -2.84426242e-01 1.24688280e+00 6.81580007e-01
6.82560027e-01 4.41932857e-01 2.12670103e-01 -5.05063653e-01
-2.59495229e-02 -7.09398508e-01 -4.35051471e-01 -1.19113863e-01
3.23897034e-01 -4.75123137e-01 5.35930954e-02 9.85255763e-02]
|
[14.45710563659668, -1.8804094791412354]
|
ed2f5b51-9b87-4618-a35e-5731ea3f149c
|
fast-contextual-scene-graph-generation-with
| null | null |
http://openaccess.thecvf.com//content/CVPR2023/html/Jin_Fast_Contextual_Scene_Graph_Generation_With_Unbiased_Context_Augmentation_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Jin_Fast_Contextual_Scene_Graph_Generation_With_Unbiased_Context_Augmentation_CVPR_2023_paper.pdf
|
Fast Contextual Scene Graph Generation With Unbiased Context Augmentation
|
Scene graph generation (SGG) methods have historically suffered from long-tail bias and slow inference speed. In this paper, we notice that humans can analyze relationships between objects relying solely on context descriptions,and this abstract cognitive process may be guided by experience. For example, given descriptions of cup and table with their spatial locations, humans can speculate possible relationships < cup, on, table > or < table, near, cup >. Even without visual appearance information, some impossible predicates like flying in and looking at can be empirically excluded. Accordingly, we propose a contextual scene graph generation (C-SGG) method without using visual information and introduce a context augmentation method. We propose that slight perturbations in the position and size of objects do not essentially affect the relationship between objects. Therefore, at the context level, we can produce diverse context descriptions by using a context augmentation method based on the original dataset. These diverse context descriptions can be used for unbiased training of C-SGG to alleviate long-tail bias. In addition, we also introduce a context guided visual scene graph generation (CV-SGG) method, which leverages the C-SGG experience to guide vision to focus on possible predicates. Through extensive experiments on the publicly available dataset, C-SGG alleviates long-tail bias and omits the huge computation of visual feature extraction to realize real-time SGG. CV-SGG achieves a great trade-off between common predicates and tail predicates.
|
['Wei Song', 'Zonghao Mu', 'Wen Wang', 'Xiangming Xi', 'Shiqiang Zhu', 'Qiwei Meng', 'Fangtai Guo', 'Tianlei Jin']
|
2023-01-01
| null | null | null |
cvpr-2023-1
|
['scene-graph-generation']
|
['computer-vision']
|
[ 1.85385868e-01 9.01957452e-02 3.16689909e-01 -5.54890215e-01
-1.70876682e-01 -7.52885878e-01 7.33605862e-01 3.66064399e-01
-8.17668959e-02 4.70903248e-01 2.37212464e-01 -5.07110059e-01
1.54222026e-01 -9.03617263e-01 -8.91907334e-01 -4.01147693e-01
1.65948823e-01 1.92846626e-01 5.54853439e-01 -2.73165017e-01
3.40935230e-01 2.83347934e-01 -1.68569636e+00 4.23297256e-01
8.82058501e-01 9.18754995e-01 5.58883131e-01 4.53271002e-01
-1.57784134e-01 7.36407936e-01 -5.88900924e-01 -3.39404404e-01
2.58391619e-01 -3.93342108e-01 -5.46905577e-01 3.34915310e-01
6.31665409e-01 -3.05817306e-01 -2.25167021e-01 1.08444047e+00
2.77276903e-01 4.48686540e-01 5.12347281e-01 -1.32980597e+00
-9.15524244e-01 4.83525723e-01 -4.28495526e-01 3.57009135e-02
6.35897875e-01 4.91492599e-01 1.21094871e+00 -9.39848006e-01
6.55839503e-01 1.35871065e+00 2.23987341e-01 4.53544527e-01
-1.28382623e+00 -4.52278972e-01 9.21016574e-01 3.50895971e-01
-1.43054378e+00 -9.93200019e-02 1.10585225e+00 -3.49463612e-01
7.85774827e-01 3.49716395e-01 9.65424180e-01 1.09258771e+00
-2.28707314e-01 8.30636740e-01 9.99910712e-01 -4.23545897e-01
5.27951360e-01 -5.68220690e-02 -8.68587792e-02 7.94629157e-01
4.67076421e-01 9.00047272e-03 -5.49163401e-01 4.37796377e-02
8.80655348e-01 -5.01319654e-02 -3.70192319e-01 -5.53045034e-01
-1.45904398e+00 5.80765367e-01 7.99379408e-01 -9.57926810e-02
-8.68538022e-02 7.62154832e-02 7.71219507e-02 -1.32389277e-01
-6.59953011e-03 5.49090266e-01 -1.71059534e-01 3.35986823e-01
-4.34214979e-01 5.26422441e-01 4.76724058e-01 1.51137733e+00
9.54237938e-01 -7.96389058e-02 -5.31467259e-01 4.23503727e-01
3.07516634e-01 4.91840065e-01 1.20397024e-01 -8.06290448e-01
5.94589233e-01 8.45139563e-01 1.97851032e-01 -1.30239308e+00
-3.77800941e-01 -2.30289668e-01 -5.78055918e-01 1.76163614e-02
4.32477802e-01 -1.49553241e-02 -1.10961163e+00 1.80730247e+00
5.07254839e-01 1.27530038e-01 -2.11676136e-01 1.03844655e+00
9.32808876e-01 3.54804426e-01 1.99495360e-01 1.63378697e-02
1.46957910e+00 -8.68163645e-01 -4.50236201e-01 -7.53427684e-01
4.65684235e-01 -3.89777660e-01 1.62603939e+00 1.58789068e-01
-7.08588481e-01 -6.64638519e-01 -9.94656026e-01 -3.32398146e-01
-5.66928327e-01 1.16828546e-01 9.93082702e-01 2.69990504e-01
-8.43884110e-01 2.75164455e-01 -6.69255078e-01 -3.98161173e-01
3.85383964e-01 7.86485374e-02 -2.53309608e-01 -3.37050140e-01
-8.45008314e-01 5.48930109e-01 7.10573792e-01 1.40204072e-01
-7.71991551e-01 -3.12192619e-01 -1.24391639e+00 3.26826721e-02
7.78742492e-01 -8.46135199e-01 9.97068644e-01 -7.46523499e-01
-9.39348698e-01 7.07849801e-01 -3.35732430e-01 -1.83494329e-01
3.23871642e-01 -1.57407835e-01 -2.85611480e-01 9.74564850e-02
2.53085792e-01 8.76037002e-01 7.58184373e-01 -1.55961120e+00
-4.55051422e-01 -3.01141173e-01 4.79751885e-01 4.23065305e-01
1.45780891e-01 -4.25097734e-01 -7.61704504e-01 -7.20944643e-01
5.28980255e-01 -9.55726445e-01 -2.59794593e-01 2.66135335e-01
-7.77535796e-01 -2.11543173e-01 6.73948526e-01 -2.21416712e-01
1.00488591e+00 -2.26965284e+00 -2.88811773e-01 1.31441548e-01
1.69494390e-01 4.38210368e-02 -2.59349316e-01 3.46493036e-01
4.95700613e-02 9.04320478e-02 -2.06173375e-01 -3.77787113e-01
-6.20877855e-02 4.96463478e-01 -5.13660848e-01 7.61280805e-02
2.74472713e-01 1.12049103e+00 -1.31560087e+00 -6.39842033e-01
3.79976332e-01 1.89486578e-01 -7.85903811e-01 1.32603005e-01
-6.95114970e-01 4.31245476e-01 -4.94945109e-01 8.97891164e-01
6.44715607e-01 -4.67478454e-01 3.25095773e-01 -4.10934687e-01
7.30177015e-02 3.30659211e-01 -1.01855588e+00 1.74650824e+00
-1.27323166e-01 4.68403220e-01 -4.03156370e-01 -8.24882209e-01
8.47221196e-01 -1.68745250e-01 -2.31018424e-01 -6.63641036e-01
5.99186346e-02 -1.90864354e-01 8.82431418e-02 -4.20603931e-01
4.16933358e-01 1.35791764e-01 -7.98015948e-03 1.39508635e-01
-2.76576340e-01 -3.92903537e-01 2.72607386e-01 5.49133122e-01
8.01715672e-01 4.10832644e-01 4.56744015e-01 -1.18781440e-01
3.67327511e-01 4.57991548e-02 6.85255527e-01 9.67814684e-01
-3.07921886e-01 8.51979613e-01 4.99969065e-01 -3.50394487e-01
-6.10239565e-01 -1.22988319e+00 3.44274938e-01 9.81589854e-01
7.88172960e-01 -6.90384746e-01 -4.66102183e-01 -8.39779317e-01
-3.79912667e-02 1.02814400e+00 -7.84564376e-01 -8.26099217e-02
-5.00939250e-01 -4.52546984e-01 -5.42823561e-02 7.62239158e-01
5.96223533e-01 -1.25290406e+00 -6.96439981e-01 -1.81932412e-02
-2.54603893e-01 -1.39663887e+00 -5.81660688e-01 -2.65593678e-02
-6.95762455e-01 -1.15325463e+00 -1.73805535e-01 -7.54474401e-01
1.20644426e+00 6.38831079e-01 1.17790842e+00 2.71725625e-01
-1.90837875e-01 3.52026463e-01 -5.03050625e-01 -4.80969310e-01
7.92000145e-02 -4.86439884e-01 -6.61661476e-02 -2.58029606e-02
2.87702262e-01 -8.51646483e-01 -9.25805449e-01 2.76241750e-01
-5.62962532e-01 5.64929128e-01 4.76837158e-01 6.02333009e-01
8.47487032e-01 -5.71294650e-02 3.34116340e-01 -9.79703426e-01
3.69849592e-01 -2.18712762e-01 -6.41248226e-01 3.16995144e-01
-4.80737597e-01 1.15482405e-01 7.30291307e-01 -6.06579363e-01
-1.03707862e+00 1.96801901e-01 2.77232319e-01 -5.92936516e-01
-2.67131031e-01 3.41977656e-01 -5.91808915e-01 1.08681999e-01
6.87133908e-01 3.87020499e-01 -5.53611934e-01 -2.55013168e-01
6.87448740e-01 -1.12330988e-02 7.09866285e-01 -6.88045323e-01
8.51007521e-01 6.32581115e-01 6.03980906e-02 -5.93716621e-01
-1.16625607e+00 -2.32223198e-01 -5.59287190e-01 -6.77725524e-02
7.98266768e-01 -8.44134629e-01 -7.21195519e-01 7.02375993e-02
-1.05482888e+00 -4.78956223e-01 -1.67647362e-01 3.26222241e-01
-4.35787499e-01 4.55673128e-01 -1.07412651e-01 -8.21165860e-01
1.23375736e-01 -8.15436900e-01 1.09570956e+00 3.44531447e-01
-7.15371072e-02 -8.88092577e-01 -4.62155938e-01 1.72704399e-01
7.95166939e-02 4.20402259e-01 9.68282342e-01 -5.62320113e-01
-9.65558350e-01 1.70781597e-01 -5.67560971e-01 -1.06497332e-01
3.80074948e-01 -1.32817104e-01 -8.59240353e-01 -1.84314057e-01
-3.23438287e-01 -1.20504469e-01 6.88571751e-01 2.12691098e-01
1.49793983e+00 -4.11965638e-01 -3.89269590e-01 7.25559056e-01
1.37456572e+00 2.68113047e-01 4.60393429e-01 2.38319814e-01
1.12479532e+00 5.73794365e-01 8.92793834e-01 3.96835029e-01
7.33976066e-01 4.76984292e-01 6.17909491e-01 1.18639119e-01
-3.18753630e-01 -7.32925713e-01 1.93135291e-02 1.28550962e-01
-1.34578660e-01 -3.66422504e-01 -8.03709567e-01 6.18848443e-01
-1.95099676e+00 -8.26966405e-01 -1.03916205e-01 2.16298461e+00
6.42993093e-01 4.12996858e-01 -1.58386171e-01 -6.64258972e-02
6.00232005e-01 2.12682873e-01 -7.71109581e-01 2.03842930e-02
-1.18015185e-01 -1.61667719e-01 2.29154378e-01 3.28648388e-01
-1.03748262e+00 1.20411682e+00 5.86328077e+00 5.00339925e-01
-9.68356371e-01 -3.56929272e-01 3.80883515e-01 3.30024324e-02
-4.82124746e-01 3.79568934e-01 -8.31481159e-01 3.94123256e-01
-5.87289073e-02 -2.07820222e-01 5.19533336e-01 1.02401578e+00
4.81717438e-02 -2.79481441e-01 -1.26906323e+00 1.22085822e+00
1.25763193e-01 -1.11547506e+00 4.25157964e-01 -9.98398885e-02
4.99515444e-01 -2.69140303e-01 -6.82868510e-02 3.94125283e-01
3.66370916e-01 -6.84373260e-01 9.59735692e-01 2.18930528e-01
6.44073904e-01 -4.73809332e-01 2.21133977e-01 3.28982115e-01
-1.40009606e+00 -1.79787874e-02 -4.01851743e-01 -2.87543476e-01
1.50620848e-01 5.58101058e-01 -9.58431959e-01 5.98437488e-01
5.74638486e-01 6.06535017e-01 -1.03154683e+00 8.57848525e-01
-8.36401761e-01 4.33316678e-01 -2.67275274e-01 -1.28059417e-01
5.49827330e-02 -1.22837864e-01 5.77007294e-01 9.42499518e-01
1.53692573e-01 4.17342067e-01 4.08570945e-01 1.20901132e+00
2.14078933e-01 -1.56150758e-01 -7.28969395e-01 7.61301741e-02
7.64822304e-01 1.16894746e+00 -9.45920169e-01 -5.09555936e-01
-3.88637543e-01 1.00793719e+00 4.62248504e-01 6.69641018e-01
-8.34375381e-01 -3.06321681e-01 5.05504906e-01 2.46652648e-01
5.23680687e-01 -3.57502252e-01 -4.14352924e-01 -1.31324148e+00
2.44355768e-01 -5.76354325e-01 3.99604887e-01 -1.17003572e+00
-1.18766820e+00 4.89915699e-01 2.77415574e-01 -1.26868474e+00
-5.56068495e-02 -5.39582253e-01 -7.36931443e-01 6.59845948e-01
-1.35852110e+00 -1.30224347e+00 -6.60884202e-01 8.21156085e-01
5.63463628e-01 3.41392100e-01 3.71863276e-01 -3.15062553e-01
-3.12922746e-01 4.36593860e-01 -7.28271663e-01 1.29420668e-01
5.35215676e-01 -1.40515554e+00 7.28275716e-01 1.11857581e+00
3.46666604e-01 9.87226725e-01 7.32896447e-01 -9.15896833e-01
-1.21022224e+00 -1.31029308e+00 6.85104311e-01 -6.13127410e-01
3.76860946e-01 -7.24081218e-01 -9.51099932e-01 8.62895012e-01
-1.72988638e-01 3.00776720e-01 4.43022132e-01 2.22325504e-01
-5.89383423e-01 5.34163974e-02 -9.85392213e-01 1.12956786e+00
1.54747427e+00 -5.60844600e-01 -7.76138723e-01 4.57860470e-01
9.79634881e-01 -4.57013309e-01 -1.65963933e-01 3.97847146e-01
2.67391413e-01 -1.11323857e+00 1.05321801e+00 -4.37238216e-01
2.81580329e-01 -7.92934716e-01 -2.81447649e-01 -1.14461529e+00
-3.77977282e-01 -3.92165810e-01 -3.31339389e-01 1.16467452e+00
1.44247830e-01 -5.22449315e-01 6.89594150e-01 8.17795694e-01
-2.13839367e-01 -6.31775200e-01 -3.69638652e-01 -8.83440614e-01
-4.98048276e-01 -5.03709614e-01 7.94091105e-01 9.47405934e-01
-1.95308737e-02 3.53066891e-01 -1.16295181e-01 4.53805864e-01
7.08322167e-01 6.21884525e-01 9.09098744e-01 -9.49175835e-01
-3.83081526e-01 -1.40529215e-01 -3.80462557e-01 -1.32589161e+00
-1.57733068e-01 -6.32824957e-01 2.07849145e-01 -1.83413160e+00
2.18163542e-02 -3.98929298e-01 -8.57203603e-02 7.60860145e-01
-6.43581748e-01 5.56870736e-02 3.84890735e-01 2.57184287e-03
-6.74126148e-01 5.89631438e-01 1.74406636e+00 3.28422873e-03
-2.84071743e-01 -2.84192562e-01 -1.05521953e+00 9.09336627e-01
6.18162751e-01 -1.12587750e-01 -8.84070039e-01 -3.92833620e-01
3.47361118e-01 -1.47932291e-01 6.42560363e-01 -1.01126528e+00
1.61532328e-01 -5.40162265e-01 4.68834430e-01 -7.61444449e-01
4.28886265e-01 -5.20545781e-01 -1.58017635e-01 1.11311890e-01
-5.61670922e-02 -9.75811407e-02 1.29779890e-01 8.61997128e-01
-7.57599249e-02 1.84807181e-01 3.52270007e-01 -3.76663119e-01
-1.09176695e+00 2.66512901e-01 -1.28543731e-02 1.04413435e-01
9.07085121e-01 -4.41746414e-01 -4.14726853e-01 -3.94991726e-01
-7.05165446e-01 4.66820598e-01 5.94815612e-01 3.62498939e-01
7.55350709e-01 -1.33429837e+00 -2.24581182e-01 2.36566365e-01
5.31579792e-01 5.52500725e-01 2.47540548e-01 4.33209091e-01
-2.68531650e-01 1.95884034e-01 4.11493182e-02 -5.94856560e-01
-1.14869046e+00 1.05977285e+00 -2.83076242e-02 2.14991033e-01
-7.90432572e-01 9.80545640e-01 9.15913761e-01 -1.38955638e-01
3.53432149e-02 -6.55390382e-01 -1.70237646e-01 -2.48731047e-01
5.22436142e-01 -1.93197146e-01 -2.00173974e-01 -3.15706432e-01
-5.26953220e-01 4.86354887e-01 -7.87896737e-02 -3.68115194e-02
9.35819864e-01 -2.68164217e-01 2.36147448e-01 4.35905963e-01
6.42846465e-01 1.70921966e-01 -1.51383543e+00 -9.21788812e-02
-2.50873208e-01 -7.64567614e-01 -2.27473080e-01 -9.07802165e-01
-8.11267018e-01 8.07186246e-01 9.80392545e-02 1.28122911e-01
1.34438181e+00 1.81011438e-01 4.17724758e-01 5.44724643e-01
6.66664541e-01 -6.51035666e-01 2.91754991e-01 3.90283495e-01
1.04170012e+00 -1.37262630e+00 1.41852409e-01 -9.70792830e-01
-8.67043376e-01 7.77237654e-01 1.13753855e+00 -1.32850021e-01
3.42370123e-01 -2.07193911e-01 -8.74890387e-02 -4.18816030e-01
-7.51000881e-01 -6.00446522e-01 4.86468971e-01 8.84642959e-01
1.00645065e-01 1.41230687e-01 -6.20257519e-02 4.65771019e-01
-3.74305993e-01 -2.82416105e-01 4.38686281e-01 1.07244956e+00
-3.27628732e-01 -8.81451964e-01 -2.21802965e-01 2.94661373e-01
1.74689457e-01 -3.53354573e-01 -5.85055768e-01 7.86976933e-01
3.82665455e-01 8.63148153e-01 4.63229679e-02 -2.82088637e-01
3.48833799e-01 -1.76252812e-01 6.28092706e-01 -8.09229553e-01
-1.68510690e-01 -7.83542171e-02 7.55445436e-02 -8.15342963e-01
-2.95349777e-01 -5.73993802e-01 -1.41155124e+00 1.19284086e-01
-3.50672543e-01 -2.17269465e-01 4.11300480e-01 8.97695184e-01
4.32309955e-01 5.16040027e-01 2.68104851e-01 -8.38488400e-01
8.68758932e-03 -6.32864177e-01 -4.89518285e-01 7.38319635e-01
2.63167202e-01 -1.01146138e+00 -4.26630676e-01 1.92854330e-01]
|
[10.35105037689209, 1.6419466733932495]
|
50cf744f-b30d-403c-9440-f2a399852d8f
|
shall-we-trust-all-relational-tuples-by-open
|
2305.04181
| null |
https://arxiv.org/abs/2305.04181v1
|
https://arxiv.org/pdf/2305.04181v1.pdf
|
Shall We Trust All Relational Tuples by Open Information Extraction? A Study on Speculation Detection
|
Open Information Extraction (OIE) aims to extract factual relational tuples from open-domain sentences. Downstream tasks use the extracted OIE tuples as facts, without examining the certainty of these facts. However, uncertainty/speculation is a common linguistic phenomenon. Existing studies on speculation detection are defined at sentence level, but even if a sentence is determined to be speculative, not all tuples extracted from it may be speculative. In this paper, we propose to study speculations in OIE and aim to determine whether an extracted tuple is speculative. We formally define the research problem of tuple-level speculation detection and conduct a detailed data analysis on the LSOIE dataset which contains labels for speculative tuples. Lastly, we propose a baseline model OIE-Spec for this new research task.
|
['XiaoLi Li', 'Jung-jae Kim', 'Aixin Sun', 'Kuicai Dong']
|
2023-05-07
| null | null | null | null |
['open-information-extraction', 'speculation-detection']
|
['natural-language-processing', 'natural-language-processing']
|
[ 2.24805549e-02 1.11095059e+00 -8.71219933e-01 -4.48318809e-01
-9.52603638e-01 -5.25020719e-01 6.58155382e-01 5.86268544e-01
1.58150285e-01 1.14131165e+00 7.67899632e-01 -5.54166675e-01
5.37116349e-01 -9.93944824e-01 -1.05415475e+00 1.93722129e-01
-1.92475632e-01 5.04381418e-01 4.17275846e-01 -1.84227437e-01
2.51264870e-01 2.56165266e-01 -1.29223084e+00 1.07545006e+00
7.34958410e-01 1.21756518e+00 -5.03003836e-01 2.31879890e-01
-5.38119912e-01 1.48986149e+00 -1.03499353e+00 -1.05497134e+00
3.23750705e-01 -2.57011931e-02 -1.17139757e+00 -8.73080492e-02
3.31112385e-01 -2.54265457e-01 -7.34406337e-02 9.61215854e-01
-2.94613123e-01 -2.63498098e-01 6.82592094e-01 -1.49703169e+00
-4.68485206e-01 1.64433253e+00 -4.27773058e-01 7.07444608e-01
5.89319289e-01 -2.23432928e-01 1.76690912e+00 -1.14173806e+00
1.15665078e+00 1.29236531e+00 2.99957871e-01 2.02973112e-01
-8.39161038e-01 -6.23083532e-01 4.45785582e-01 1.85165517e-02
-1.18944538e+00 -5.52530706e-01 6.53119981e-01 -2.31263533e-01
1.37459922e+00 3.88372838e-01 3.02916944e-01 1.35953867e+00
7.68284082e-01 1.14619637e+00 9.87453282e-01 -2.21675381e-01
4.53554779e-01 2.16245115e-01 6.46629810e-01 2.30697736e-01
8.71497989e-01 1.14899039e-01 -1.39431834e+00 -2.70273149e-01
6.36392906e-02 -5.26373088e-01 -1.74313009e-01 5.14113307e-01
-1.01581156e+00 7.92745590e-01 7.19769672e-02 -7.70502351e-03
-4.65476304e-01 -2.46144325e-01 6.88078940e-01 4.25830454e-01
7.15080738e-01 4.98077780e-01 -9.59503770e-01 -1.04521655e-01
-5.68559468e-01 6.08078957e-01 1.57704842e+00 1.36163020e+00
4.82316345e-01 -4.33406502e-01 -2.93869495e-01 1.89038888e-01
4.94778156e-01 6.70318976e-02 3.77035290e-01 -7.75459647e-01
1.10570550e+00 8.36879253e-01 3.54804069e-01 -9.94385421e-01
-2.95163810e-01 -1.58696860e-01 -9.84448195e-02 -4.52732861e-01
3.67374197e-02 -3.68821472e-01 -5.88967919e-01 1.34896171e+00
2.36080274e-01 1.34667769e-01 6.93813205e-01 7.42337167e-01
1.22081077e+00 5.68713963e-01 1.62547380e-01 -4.77162987e-01
1.64188290e+00 -5.46350121e-01 -9.14480925e-01 -7.04062462e-01
5.15668154e-01 -4.90844339e-01 6.63789153e-01 5.86273193e-01
-6.05407596e-01 2.73583420e-02 -1.46149802e+00 -4.69812006e-01
-4.46090370e-01 -2.48453990e-01 7.88468480e-01 4.60685641e-01
-2.43092611e-01 3.55496377e-01 -7.77568161e-01 3.60166401e-01
3.60586941e-01 -7.52594769e-02 -1.91189513e-01 4.07297492e-01
-1.98233187e+00 7.78707385e-01 1.16398346e+00 -3.53741981e-02
-5.39353907e-01 -7.83148587e-01 -1.11623871e+00 2.51773626e-01
1.31617391e+00 -3.84431899e-01 1.59940553e+00 -9.73870084e-02
-1.02912509e+00 7.17918515e-01 -5.70809722e-01 -1.00128818e+00
3.00522178e-01 -5.30544162e-01 -8.25390160e-01 3.98394465e-03
6.17890775e-01 3.34871590e-01 5.00095606e-01 -1.36215973e+00
-1.11866033e+00 -3.86917561e-01 2.97204375e-01 -1.87063053e-01
1.95677444e-01 2.30105236e-01 6.17554598e-02 -5.38844943e-01
4.40122068e-01 -6.37587249e-01 4.30734642e-02 -9.40035522e-01
-1.14010954e+00 -6.26288712e-01 6.39441967e-01 -4.84930426e-01
1.67126596e+00 -1.87374246e+00 -5.90927899e-01 -2.99319588e-02
5.20903230e-01 -1.90744013e-01 6.63018942e-01 4.93426979e-01
-1.18684843e-01 6.37455046e-01 -7.14596733e-02 -1.28881168e-02
1.89303085e-01 3.61079633e-01 -1.51915228e+00 -2.41626769e-01
6.20381057e-01 1.12692070e+00 -6.94846034e-01 -8.82589281e-01
-5.03711581e-01 -5.66770971e-01 -1.20463006e-01 1.36345401e-01
-1.18853092e+00 -1.53497577e-01 -6.28494442e-01 1.04899895e+00
6.71846747e-01 -1.46909073e-01 2.12155998e-01 -2.54567593e-01
-9.85851809e-02 1.18890738e+00 -1.18251991e+00 9.51031029e-01
-1.86857760e-01 5.18558741e-01 -4.58328724e-01 -3.56221139e-01
8.67183983e-01 4.07584369e-01 2.71203160e-01 -3.10756117e-01
3.78346927e-02 2.22421303e-01 -3.54826659e-01 -4.81394559e-01
1.03519654e+00 -3.24002624e-01 -8.16587865e-01 4.66580123e-01
-1.75370857e-01 -7.39323571e-02 5.27295649e-01 7.89238572e-01
1.22545755e+00 -7.51507059e-02 7.98343658e-01 -1.77740335e-01
3.97370040e-01 4.91114736e-01 1.30062366e+00 6.37324512e-01
-1.67889223e-01 2.33261794e-01 1.53745055e+00 -2.79205233e-01
-5.93268692e-01 -1.16938484e+00 -4.74402994e-01 6.95248544e-01
2.47310534e-01 -1.25002527e+00 -7.35368878e-02 -1.28229094e+00
3.51359963e-01 1.29105198e+00 -5.68977952e-01 2.39446029e-01
-4.81239527e-01 -5.75679004e-01 5.33239782e-01 4.63582009e-01
2.63206780e-01 -8.30756187e-01 -6.75833166e-01 2.65023887e-01
-6.32416129e-01 -1.61279738e+00 -1.35202214e-01 3.14347386e-01
-5.28250754e-01 -1.11660159e+00 7.34093666e-01 -2.57458150e-01
1.08578473e-01 -3.54598254e-01 1.72076714e+00 -4.90635306e-01
4.59396422e-01 -4.30567801e-01 -4.89917904e-01 -9.61858630e-01
-6.13106310e-01 -3.52635072e-03 -6.81466237e-02 -3.13975394e-01
1.02491128e+00 -2.15287134e-01 5.99626265e-02 -1.28057748e-02
-7.16510594e-01 -2.52938960e-02 2.79507667e-01 4.72537369e-01
6.65420890e-01 3.19930583e-01 7.01225460e-01 -1.71336329e+00
8.82303894e-01 -8.33506048e-01 -4.31659639e-01 3.59575748e-01
-6.37925208e-01 3.87559742e-01 5.81728697e-01 8.70989561e-02
-1.61402810e+00 -3.97457063e-01 1.40636817e-01 2.30860710e-02
-4.87470329e-02 1.10730708e+00 -4.00621414e-01 8.79963219e-01
6.82387650e-01 -4.61631089e-01 -7.08341718e-01 1.87776022e-04
2.94937044e-01 7.33754754e-01 7.09402919e-01 -8.93040121e-01
3.98757041e-01 5.44004261e-01 -7.60249496e-02 -4.70321268e-01
-1.69907713e+00 -1.05225138e-01 -3.69225919e-01 8.63243118e-02
2.63109058e-01 -1.13719690e+00 -8.89395118e-01 -2.21758276e-01
-1.31142640e+00 3.32929939e-01 -3.77081573e-01 1.33929208e-01
-3.02976459e-01 3.37550908e-01 -8.91785622e-01 -8.92428577e-01
-2.68173367e-01 -8.27252507e-01 1.06506979e+00 1.02433793e-01
-8.66387010e-01 -5.76729119e-01 -1.49487719e-01 4.58099127e-01
-5.96768498e-01 2.60816872e-01 8.36819768e-01 -1.23276329e+00
-9.24052238e-01 -7.32258633e-02 -1.36685446e-01 -8.93773660e-02
4.33241613e-02 2.60016710e-01 -7.76283920e-01 4.92763907e-01
5.55220127e-01 -5.73114336e-01 1.04018593e+00 -8.89730826e-02
5.83083510e-01 -7.43282735e-01 -4.55509096e-01 1.13152556e-01
1.11333978e+00 1.84487224e-01 3.48089427e-01 2.66555130e-01
1.71087980e-01 9.17767346e-01 1.22117853e+00 7.26621747e-01
5.53069413e-01 -1.17855081e-02 3.53547484e-02 9.53112304e-01
2.65426964e-01 -7.60751724e-01 4.25175607e-01 3.36927354e-01
5.00949621e-01 -6.35894775e-01 -9.82282519e-01 4.28266048e-01
-1.86170626e+00 -8.25462937e-01 -4.52925354e-01 1.51236939e+00
1.14745235e+00 1.01204062e+00 -5.50495148e-01 3.20649929e-02
4.10055220e-01 4.38213289e-01 -4.09313411e-01 -5.12843013e-01
-3.83452475e-01 -9.25148949e-02 3.90274286e-01 4.66917247e-01
-1.08895719e+00 1.20532095e+00 6.42641687e+00 4.55868602e-01
-7.67233551e-01 2.18854193e-03 6.57396078e-01 -3.40804122e-02
-7.96384692e-01 4.13306057e-01 -1.53073311e+00 5.06541073e-01
1.22073960e+00 -4.77197707e-01 -6.21613085e-01 9.43379045e-01
-6.48507774e-02 -7.95901239e-01 -1.64241385e+00 4.12501395e-01
-4.51231515e-03 -1.66057062e+00 1.28230438e-01 -1.49585456e-01
6.88848257e-01 -2.78633446e-01 -2.91412830e-01 5.12936175e-01
5.28769553e-01 -7.84100235e-01 1.04547727e+00 2.66264856e-01
2.49986425e-01 -6.28187835e-01 9.11905706e-01 5.62540889e-01
-7.61306226e-01 -8.60605761e-02 -1.40712962e-01 -2.65899956e-01
4.85735029e-01 1.16157269e+00 -1.07213259e+00 6.99271441e-01
4.96789992e-01 6.60863817e-01 -5.61211288e-01 -1.14974119e-01
-7.90820956e-01 7.35500991e-01 -2.95906663e-01 -1.29086748e-01
3.58930044e-02 -4.01879326e-02 6.18046761e-01 1.13686240e+00
-1.73714951e-01 4.81420577e-01 9.77116227e-02 1.24271965e+00
-3.34120601e-01 -2.44497001e-01 -4.58845198e-01 -4.48403686e-01
5.90574682e-01 5.70503652e-01 -6.39059186e-01 -6.36892259e-01
-6.77253067e-01 5.29686153e-01 4.82757092e-01 7.12024271e-02
-6.07352912e-01 -1.15907595e-01 6.47181273e-01 -9.11535472e-02
2.52798975e-01 -6.04716949e-02 -1.06445062e+00 -1.54677522e+00
6.25527382e-01 -7.01649964e-01 7.66538978e-01 -5.93892932e-01
-1.45122433e+00 5.04453838e-01 1.88392490e-01 -7.68258929e-01
-7.44422019e-01 -5.07298172e-01 -5.19073188e-01 5.84843159e-01
-1.23004365e+00 -5.61455309e-01 3.15145999e-01 -2.00439095e-02
5.89066386e-01 6.04679948e-03 4.59018499e-01 -2.62168944e-01
-6.39293373e-01 1.27155676e-01 -6.61081254e-01 3.53389025e-01
6.81202710e-01 -1.23868775e+00 8.81889224e-01 1.40846622e+00
2.27045134e-01 1.00553143e+00 1.18469644e+00 -1.68556762e+00
-1.32344294e+00 -8.26904595e-01 1.75166297e+00 -9.13115680e-01
1.21789467e+00 -5.13367951e-01 -1.11872637e+00 1.36778009e+00
8.37799087e-02 -2.15740763e-02 5.11602461e-01 5.98913908e-01
-6.64096713e-01 1.36672467e-01 -8.98713529e-01 5.69870174e-01
1.00443780e+00 -6.47563040e-01 -1.56763244e+00 3.38126183e-01
1.45028234e+00 -8.09514880e-01 -1.04319584e+00 7.44295716e-01
1.94879413e-01 -9.88594532e-01 5.44103742e-01 -8.08576941e-01
1.05148840e+00 -2.01253667e-01 -1.58413202e-01 -9.71456647e-01
3.85625094e-01 -5.63782275e-01 -8.51227403e-01 1.46849489e+00
1.08647096e+00 -5.92544556e-01 9.09725666e-01 1.23884714e+00
-2.47064352e-01 -7.98209310e-01 -9.13845718e-01 -6.08441353e-01
-1.74436435e-01 -1.02072346e+00 7.57039487e-01 6.28997386e-01
9.02547598e-01 6.80096567e-01 9.71423537e-02 4.88628179e-01
4.78491843e-01 7.26788759e-01 2.75353640e-01 -1.00279164e+00
1.55764986e-02 3.31900418e-02 -8.40975699e-05 -9.59678233e-01
7.81683385e-01 -5.79900503e-01 1.19687371e-01 -1.28306699e+00
-1.25539556e-01 -3.71649593e-01 8.63180980e-02 2.64854372e-01
-1.83244526e-01 -4.88954604e-01 -6.73032627e-02 1.66135564e-01
-4.01528001e-01 4.58237469e-01 8.99965048e-01 -1.81583777e-01
-3.34221929e-01 2.29420006e-01 -1.15038729e+00 6.89407766e-01
6.16747499e-01 -5.24147093e-01 -4.06306446e-01 2.15067491e-01
8.28583956e-01 5.65457582e-01 -6.56556338e-03 -4.33230191e-01
2.56881624e-01 -4.25673604e-01 -5.90709895e-02 -1.10696578e+00
1.11868538e-01 -5.05854785e-01 -4.33961861e-02 7.10186139e-02
-6.51279926e-01 -2.43085220e-01 -4.25061844e-02 6.71266973e-01
-6.77959323e-01 -4.30199146e-01 -5.57601452e-02 -1.81573913e-01
-9.11865056e-01 4.80647460e-02 -3.93862307e-01 5.88135362e-01
9.02267873e-01 4.26087111e-01 -9.07776058e-01 7.57229179e-02
-6.54971361e-01 3.45141113e-01 -9.51618887e-03 5.14656842e-01
7.05036879e-01 -8.40396881e-01 -6.63156331e-01 3.94000299e-02
4.49751884e-01 3.18561077e-01 -1.27460554e-01 3.65180492e-01
-3.24493617e-01 8.33909035e-01 3.79674613e-01 -1.46144861e-02
-6.86933756e-01 8.84110451e-01 -2.00248852e-01 -4.06078100e-01
-5.73203743e-01 8.49597037e-01 -1.58004761e-01 8.66511278e-03
4.54704873e-02 -1.02060461e+00 -1.56765133e-01 4.73739088e-01
4.44077134e-01 1.14652269e-01 3.12989593e-01 -4.04625565e-01
-5.41744292e-01 -7.03772187e-01 -3.97793025e-01 -3.27845901e-01
7.99236476e-01 -1.22661784e-01 -5.10244250e-01 8.64280522e-01
7.89915025e-01 5.49734473e-01 -9.42169309e-01 -4.51583177e-01
8.41146111e-01 -5.96763492e-01 -2.15823203e-01 -7.76641667e-01
-4.67469931e-01 1.19749583e-01 -8.54473293e-01 4.95271921e-01
4.33937311e-01 6.18958771e-01 9.68235910e-01 3.97298336e-01
5.75118899e-01 -1.09240627e+00 -2.60505736e-01 7.96242714e-01
1.05714035e+00 -1.49734783e+00 2.21167341e-01 -1.25241578e+00
-1.01844239e+00 1.02551997e+00 1.01394403e+00 1.67873964e-01
7.93229342e-01 7.98057258e-01 -7.05381185e-02 -4.90613252e-01
-1.52268612e+00 5.48235774e-02 2.12495029e-02 -5.07805264e-03
6.56302452e-01 4.57323283e-01 -5.76903403e-01 1.27302957e+00
-8.91524076e-01 -4.40289862e-02 1.08138871e+00 1.03861606e+00
-3.37729514e-01 -8.37953329e-01 -3.85340780e-01 8.94028664e-01
-6.67666793e-01 -2.52037555e-01 -6.81101203e-01 4.42682028e-01
6.71500042e-02 1.25339878e+00 2.63062716e-02 -5.12938201e-01
2.38946542e-01 2.50187993e-01 -5.20761572e-02 -9.26184237e-01
-4.99152452e-01 -5.20155311e-01 9.98322129e-01 -6.38495028e-01
-2.17882007e-01 -9.29799378e-01 -1.70839334e+00 -9.74804610e-02
-2.15298891e-01 4.87229973e-01 3.49631220e-01 1.28687596e+00
3.55759501e-01 4.04174656e-01 3.18342447e-01 4.37492043e-01
-4.59961236e-01 -7.00107336e-01 -7.78944731e-01 5.09726256e-02
3.28561395e-01 -6.17347598e-01 -3.53504509e-01 5.19319177e-02]
|
[9.632503509521484, 8.639388084411621]
|
a55ff26f-d31f-4c83-bc57-2d7f4d5ad9c8
|
sketchparse-towards-rich-descriptions-for
|
1709.01295
| null |
http://arxiv.org/abs/1709.01295v1
|
http://arxiv.org/pdf/1709.01295v1.pdf
|
SketchParse : Towards Rich Descriptions for Poorly Drawn Sketches using Multi-Task Hierarchical Deep Networks
|
The ability to semantically interpret hand-drawn line sketches, although very
challenging, can pave way for novel applications in multimedia. We propose
SketchParse, the first deep-network architecture for fully automatic parsing of
freehand object sketches. SketchParse is configured as a two-level fully
convolutional network. The first level contains shared layers common to all
object categories. The second level contains a number of expert sub-networks.
Each expert specializes in parsing sketches from object categories which
contain structurally similar parts. Effectively, the two-level configuration
enables our architecture to scale up efficiently as additional categories are
added. We introduce a router layer which (i) relays sketch features from shared
layers to the correct expert (ii) eliminates the need to manually specify
object category during inference. To bypass laborious part-level annotation, we
sketchify photos from semantic object-part image datasets and use them for
training. Our architecture also incorporates object pose prediction as a novel
auxiliary task which boosts overall performance while providing supplementary
information regarding the sketch. We demonstrate SketchParse's abilities (i) on
two challenging large-scale sketch datasets (ii) in parsing unseen,
semantically related object categories (iii) in improving fine-grained
sketch-based image retrieval. As a novel application, we also outline how
SketchParse's output can be used to generate caption-style descriptions for
hand-drawn sketches.
|
['Sahil Manocha', 'R. Venkatesh Babu', 'Abhijat Biswas', 'Ravi Kiran Sarvadevabhatla', 'Isht Dwivedi']
|
2017-09-05
| null | null | null | null |
['sketch-based-image-retrieval']
|
['computer-vision']
|
[ 2.64898658e-01 1.75744712e-01 -1.24252573e-01 -5.56255400e-01
-8.18573534e-01 -9.38336790e-01 6.24012530e-01 -3.09440285e-01
-1.51678622e-01 2.05974415e-01 1.16506182e-01 -2.44969532e-01
1.28131688e-01 -9.32794333e-01 -1.05230546e+00 -1.72529563e-01
2.26483315e-01 6.76940799e-01 5.00915051e-01 -1.32313579e-01
3.18447918e-01 1.02217174e+00 -1.54171824e+00 7.76625633e-01
2.82360882e-01 1.20493484e+00 4.22596365e-01 7.55253077e-01
-6.62970662e-01 2.20392302e-01 -6.36564732e-01 -7.70964444e-01
4.53548640e-01 7.11755902e-02 -7.18830228e-01 2.03539386e-01
1.20071912e+00 -7.89767206e-01 -2.99037337e-01 5.37575603e-01
2.83787042e-01 -3.51927020e-02 7.35632360e-01 -1.42783403e+00
-6.86343372e-01 5.04029036e-01 -2.67868072e-01 -5.21462679e-01
1.92346171e-01 2.46284559e-01 1.26358545e+00 -1.26301873e+00
8.05561721e-01 1.60457492e+00 6.17291689e-01 8.61181974e-01
-1.10367465e+00 -7.97740161e-01 5.92176735e-01 -3.53032202e-01
-1.28895712e+00 -3.63580197e-01 9.24145460e-01 -1.81516334e-01
7.99713850e-01 1.22586779e-01 5.51024079e-01 1.13053489e+00
-3.96238416e-01 1.35194373e+00 4.73593086e-01 -1.33771688e-01
1.39675856e-01 -1.64245069e-03 -1.12595320e-01 1.01541364e+00
1.96772404e-02 -2.13727698e-01 -3.41007143e-01 -1.56367883e-01
1.33238745e+00 1.87595263e-01 1.27982736e-01 -7.24219203e-01
-1.19101429e+00 5.26194453e-01 7.11800933e-01 8.56620818e-03
-2.98075676e-01 7.91492462e-01 2.53882259e-01 7.01770335e-02
-4.82081212e-02 4.64328617e-01 -5.67040205e-01 3.31124187e-01
-1.05607414e+00 4.06687886e-01 8.11097562e-01 1.46957564e+00
6.64566338e-01 -7.33910725e-02 -1.60497040e-01 1.05025911e+00
2.07877353e-01 6.28190100e-01 -1.40442744e-01 -1.11467779e+00
4.67147142e-01 6.27537072e-01 4.35463935e-02 -8.36379468e-01
-1.47415444e-01 -9.52060744e-02 -5.54848909e-01 3.47089052e-01
4.45987165e-01 3.08572054e-01 -1.22382057e+00 1.61262953e+00
3.93216684e-02 -2.59858191e-01 -2.25328520e-01 1.07320535e+00
1.17341018e+00 4.80894834e-01 5.11429191e-01 8.76338124e-01
1.68037248e+00 -1.06092048e+00 -1.33074850e-01 -3.11021119e-01
-8.88765007e-02 -7.49636412e-01 1.43388999e+00 3.76984000e-01
-1.21606517e+00 -7.78770745e-01 -1.05457628e+00 -5.45267820e-01
-8.09178054e-01 6.07992828e-01 8.79481435e-01 4.42215025e-01
-1.10221899e+00 7.61354148e-01 -4.92670596e-01 -4.51890677e-01
1.04664969e+00 4.08826709e-01 -6.16868973e-01 -1.42605692e-01
-7.76504278e-01 5.10122538e-01 1.11921340e-01 -1.11847281e-01
-9.35396194e-01 -7.96554506e-01 -9.57338870e-01 4.80643481e-01
5.03106117e-01 -1.07444370e+00 1.17598665e+00 -1.01052356e+00
-1.39605856e+00 9.58979726e-01 -5.04655503e-02 1.75808053e-02
5.36402285e-01 -1.79562256e-01 -3.21883708e-02 6.13819122e-01
8.90381038e-02 1.77124465e+00 1.20390320e+00 -1.45792711e+00
-5.91862440e-01 -1.17328778e-01 3.14951152e-01 -7.70798028e-02
-2.07199212e-02 -3.55866849e-01 -9.93848920e-01 -1.05368721e+00
1.34717390e-01 -7.71166325e-01 5.10889813e-02 9.25149918e-01
-4.69512999e-01 -5.18318415e-01 9.34855163e-01 -4.57490921e-01
7.64843404e-01 -2.01231861e+00 3.90875489e-02 1.93637714e-01
1.12215981e-01 2.41185263e-01 -7.31036782e-01 5.62425911e-01
-3.81650683e-03 3.43790680e-01 -2.87730575e-01 -5.82581937e-01
3.12810957e-01 2.61304379e-01 -7.36538649e-01 -3.67344499e-01
8.35479319e-01 1.45208788e+00 -7.15231061e-01 -4.14163917e-01
4.19177204e-01 3.40691090e-01 -7.20323682e-01 3.07224423e-01
-6.74865901e-01 -1.45610482e-01 -3.88531953e-01 1.02271128e+00
7.64538765e-01 -3.15880507e-01 7.06503540e-02 -6.01983726e-01
2.45488539e-01 1.16422862e-01 -1.07538688e+00 2.11664701e+00
-5.66390157e-01 6.36737585e-01 1.05416149e-01 -6.97490513e-01
9.17351723e-01 2.16557048e-02 7.67122395e-03 -4.20174330e-01
-2.35881299e-01 4.88901557e-03 -4.76891786e-01 -2.65074611e-01
4.40071940e-01 -1.06178075e-01 -4.05147523e-01 6.29841387e-01
2.68976301e-01 -5.74250340e-01 2.12104730e-02 5.62938273e-01
7.98325181e-01 5.60951829e-01 3.29207815e-02 -1.47779405e-01
4.48771209e-01 -2.13434517e-01 -1.05909452e-01 8.57149839e-01
1.89773738e-01 9.89342928e-01 6.14146650e-01 -8.13215673e-01
-1.29556906e+00 -1.42343056e+00 2.22329453e-01 1.30311120e+00
1.72544956e-01 -3.77783686e-01 -4.37857717e-01 -1.06312180e+00
4.10454005e-01 4.71275449e-01 -6.39935195e-01 2.68618643e-01
-6.74276352e-01 2.92479157e-01 6.26390159e-01 1.04450989e+00
4.24918264e-01 -1.41519582e+00 -4.61906701e-01 -5.67389466e-02
1.66170359e-01 -1.15608740e+00 -6.68215454e-01 -1.31869689e-01
-6.22695625e-01 -1.21379685e+00 -9.87986267e-01 -8.10619235e-01
9.49692428e-01 3.22886020e-01 1.46196985e+00 5.64894915e-01
-5.76550484e-01 9.23395693e-01 -2.07776073e-02 -3.92195523e-01
-2.35957086e-01 2.27671310e-01 -4.05428559e-01 -1.75760925e-01
4.59752008e-02 -4.99511719e-01 -7.98673034e-01 4.13857102e-01
-9.66772497e-01 3.58529896e-01 9.43762064e-01 6.88017905e-01
5.43160677e-01 -5.65832317e-01 6.09152377e-01 -7.45760679e-01
3.47691089e-01 -1.73160597e-03 -6.15023553e-01 5.13866186e-01
6.46483898e-02 3.91426206e-01 5.33278883e-01 -4.84361917e-01
-9.23900843e-01 4.48944241e-01 -3.27822804e-01 -6.01714969e-01
-3.86189163e-01 -1.96385905e-01 -3.99754465e-01 -5.88404872e-02
1.38887227e-01 4.39632274e-02 -1.62391700e-02 -6.08776629e-01
8.38079751e-01 4.82865542e-01 7.23461807e-01 -1.12037134e+00
8.94954443e-01 5.00967741e-01 2.68314872e-02 -7.42043436e-01
-7.33990550e-01 -1.40833676e-01 -7.61984706e-01 3.78207956e-03
8.43499422e-01 -8.48962486e-01 -1.02714825e+00 1.24706954e-01
-1.32084215e+00 -5.59995651e-01 -2.75430888e-01 -3.11441779e-01
-6.58158302e-01 1.49199799e-01 -4.50453550e-01 -5.56212366e-01
-5.30579925e-01 -1.02335775e+00 1.93392372e+00 1.00832753e-01
-2.48410329e-01 -6.54888272e-01 -6.04417384e-01 1.74178079e-01
3.19415957e-01 5.25174215e-02 1.15299594e+00 -5.25152147e-01
-1.17592907e+00 -1.81889638e-01 -7.68353820e-01 2.19463706e-01
-2.06737339e-01 7.92208984e-02 -9.98470962e-01 1.86405014e-02
-1.02534103e+00 -6.03538096e-01 9.41140294e-01 -3.12592760e-02
1.73342395e+00 -1.81517810e-01 -3.67906392e-01 5.07741511e-01
1.29218292e+00 -2.07172737e-01 6.34487271e-01 -2.14786708e-01
8.23031545e-01 5.84616184e-01 3.22086811e-01 1.20381247e-02
3.61938477e-01 5.91579199e-01 3.54363292e-01 -2.48067379e-01
-5.49974918e-01 -8.44208777e-01 -1.61619425e-01 1.06603555e-01
2.13373467e-01 -2.41586551e-01 -5.64748526e-01 4.10472035e-01
-1.64461660e+00 -8.99504006e-01 3.24752837e-01 1.92848456e+00
5.35370588e-01 1.02073103e-01 2.59601474e-01 -3.18089575e-01
4.30275947e-01 2.96270512e-02 -5.82938194e-01 -3.14237952e-01
6.07836917e-02 5.72754204e-01 1.95384905e-01 3.10893416e-01
-1.20244694e+00 1.42428505e+00 5.96922922e+00 8.16187620e-01
-9.69758749e-01 -3.05822343e-01 4.52490479e-01 5.11281043e-02
-5.09019613e-01 -5.94140701e-02 -6.30283058e-01 1.59897625e-01
1.65801466e-01 4.70256269e-01 4.24364835e-01 1.02284420e+00
-4.47214484e-01 7.40799904e-02 -1.56045473e+00 1.05175281e+00
1.02968551e-02 -1.64858103e+00 7.74766445e-01 -1.99657544e-01
3.25741798e-01 -3.99875224e-01 -1.25887528e-01 5.64746320e-01
1.27426669e-01 -9.84268188e-01 9.92127419e-01 5.95376670e-01
1.25497293e+00 -7.52683878e-01 1.65234163e-01 4.61522751e-02
-1.42084599e+00 -1.42857777e-02 -4.41753507e-01 2.12496862e-01
1.54612169e-01 -4.88709062e-02 -7.57509887e-01 1.88400924e-01
4.45070744e-01 4.45047706e-01 -6.30569875e-01 7.10031152e-01
-4.21405911e-01 -5.06740063e-04 -3.37848693e-01 -2.05411259e-02
3.80967766e-01 1.05816394e-01 2.04287380e-01 1.42942131e+00
4.14589196e-02 3.62862438e-01 9.99702215e-02 1.22649527e+00
-4.67480123e-01 -3.66140038e-01 -5.18194079e-01 -3.73430252e-01
6.77755475e-01 1.45869994e+00 -1.04508674e+00 -6.75071776e-01
-2.88657308e-01 1.37227178e+00 4.52383250e-01 3.74377191e-01
-5.65614402e-01 -6.61635637e-01 7.87972212e-01 2.11539507e-01
7.50112891e-01 -2.08386138e-01 -3.08027834e-01 -9.06388581e-01
4.23973147e-03 -5.72959721e-01 2.17858687e-01 -1.21301901e+00
-1.39458835e+00 4.99350131e-01 -1.31474078e-01 -9.23507273e-01
-5.24054915e-02 -1.11113560e+00 -5.28484344e-01 7.28329897e-01
-1.33893394e+00 -1.89697945e+00 -5.06715775e-01 3.48371983e-01
9.42017138e-01 -1.74479887e-01 9.41704333e-01 2.72759616e-01
-1.51486844e-01 8.13072264e-01 -8.28688204e-01 5.95921040e-01
7.24252343e-01 -1.35845423e+00 9.49984670e-01 3.44001621e-01
4.14856374e-01 8.19705725e-01 -2.02177349e-03 -6.47044241e-01
-1.30031347e+00 -9.50030625e-01 6.10853851e-01 -7.35140800e-01
3.24091703e-01 -9.20117259e-01 -6.47281229e-01 5.97698152e-01
-1.91332608e-01 1.53184071e-01 3.65255147e-01 -7.22723305e-02
-9.73415613e-01 -2.43025795e-02 -1.03286839e+00 7.73868144e-01
1.35348833e+00 -7.20109642e-01 -6.69649482e-01 1.65946230e-01
6.42857373e-01 -2.63033539e-01 -7.00187325e-01 2.32415646e-01
1.28142965e+00 -7.16199398e-01 1.51849198e+00 -9.01258886e-01
6.77314520e-01 -1.64586902e-01 -1.86786279e-01 -6.58522069e-01
-2.08596021e-01 -3.22296560e-01 -5.15885092e-02 1.07491171e+00
2.46410906e-01 -5.88653497e-02 1.05117941e+00 8.69505644e-01
-5.47600202e-02 -8.56716931e-01 -2.97898501e-01 -5.96474230e-01
4.04170938e-02 -4.46558625e-01 7.84167051e-01 5.25781870e-01
-4.99102592e-01 3.02902758e-01 -1.30409405e-01 6.38424605e-02
4.81008410e-01 5.37101448e-01 1.21786714e+00 -1.36229515e+00
-1.53876916e-01 -8.07717860e-01 -2.10027218e-01 -1.43919897e+00
3.07827502e-01 -8.62162769e-01 -1.02273829e-01 -1.68711674e+00
6.84128627e-02 -8.72621775e-01 1.93406820e-01 8.58775377e-01
-1.38526931e-01 6.30009949e-01 7.51011610e-01 -1.40426159e-02
-6.80784106e-01 2.76428014e-01 1.21136534e+00 -3.17574918e-01
5.15222251e-02 -1.66780371e-02 -6.73638463e-01 6.82443380e-01
4.33232725e-01 -1.47801220e-01 -4.19589937e-01 -5.75522959e-01
4.71817814e-02 -2.00808467e-03 8.07446301e-01 -8.60999763e-01
1.11933298e-01 6.29686415e-02 8.47440064e-01 -9.20714259e-01
5.60876131e-01 -1.03699863e+00 1.73412426e-03 1.80024147e-01
-5.15995443e-01 -2.16129929e-01 3.65204394e-01 4.30096507e-01
1.61399931e-01 -1.77765876e-01 6.41425133e-01 -4.80513573e-01
-8.55125189e-01 5.46345294e-01 4.24281582e-02 -2.91813910e-01
6.49280071e-01 -2.37698957e-01 -1.38117433e-01 -3.62450391e-01
-7.84661591e-01 1.98358193e-01 5.77734590e-01 8.28608513e-01
8.78840804e-01 -1.41022623e+00 -2.51059055e-01 2.66091019e-01
2.81862140e-01 9.17504355e-02 3.49178195e-01 -7.68319815e-02
-5.55313170e-01 4.61378932e-01 -3.37336659e-01 -4.56409514e-01
-1.09990561e+00 8.15023482e-01 1.29254028e-01 2.00365379e-01
-7.47053146e-01 1.04749894e+00 6.14723027e-01 -6.14329934e-01
4.91902620e-01 -4.36006427e-01 3.22392583e-01 -7.29510188e-02
5.40282428e-01 6.61809519e-02 -1.80135220e-01 -1.82722449e-01
-3.38530898e-01 8.46226275e-01 -8.54749605e-02 -2.22279608e-01
1.27694380e+00 1.69479057e-01 1.39653713e-01 4.87769246e-02
1.24486911e+00 -1.14986282e-02 -1.62496102e+00 3.41036581e-02
4.36100438e-02 -3.13861907e-01 -3.22653472e-01 -1.23882842e+00
-9.95396852e-01 1.24850130e+00 2.30355203e-01 -3.56594563e-01
8.28100741e-01 2.92763770e-01 8.35702538e-01 7.67762303e-01
4.72084731e-01 -8.03794563e-01 5.06150901e-01 1.22648105e-01
1.36288631e+00 -1.24399710e+00 -5.09059243e-03 -5.12916565e-01
-6.16533339e-01 1.59638870e+00 6.93992496e-01 -3.65973145e-01
3.81355286e-01 3.11377704e-01 -7.52526969e-02 -3.21092665e-01
-5.95086217e-01 -2.73112595e-01 6.85012162e-01 5.81364930e-01
1.30731449e-01 5.53502627e-02 2.60719478e-01 7.44853437e-01
-2.14900039e-02 -1.23180918e-01 -1.91866364e-02 7.20983267e-01
-2.94020295e-01 -1.32302618e+00 -1.23417772e-01 3.71741682e-01
1.03340402e-01 -1.07495695e-01 -8.02483141e-01 1.03205681e+00
1.13884397e-01 3.02959234e-01 4.49657440e-01 -2.81354524e-02
4.13517535e-01 2.24819824e-01 6.63362265e-01 -5.77440262e-01
-4.34567094e-01 -1.40024155e-01 -2.96421573e-02 -8.37520063e-01
7.54225766e-03 -1.57103032e-01 -1.22328579e+00 3.04963980e-02
1.11138262e-01 -2.29799882e-01 9.68403578e-01 6.12397611e-01
6.13533616e-01 3.80781144e-01 9.88520682e-02 -1.32822931e+00
-2.27956355e-01 -6.12971008e-01 -1.36952326e-01 4.34460759e-01
2.02351645e-01 -6.10026777e-01 1.43440291e-01 3.99546139e-02]
|
[11.686574935913086, 0.453192800283432]
|
016cb2e2-ce60-473e-b979-34ca92156ac3
|
generative-neural-networks-for-anomaly
| null | null |
https://ieeexplore.ieee.org/abstract/document/8513816
|
https://ieeexplore.ieee.org/abstract/document/8513816
|
Generative Neural Networks for Anomaly Detection in Crowded Scenes
|
Security surveillance is critical to social harmony and people's peaceful life. It has a great impact on strengthening social stability and life safeguarding. Detecting anomaly timely, effectively and efficiently in video surveillance remains challenging. This paper proposes a new approach, called S 2 -VAE, for anomaly detection from video data. The S 2 -VAE consists of two proposed neural networks: a Stacked Fully Connected Variational AutoEncoder (S F -VAE) and a Skip Convolutional VAE (S C -VAE). The S F -VAE is a shallow generative network to obtain a model like Gaussian mixture to fit the distribution of the actual data. The S C -VAE, as a key component of S 2 -VAE, is a deep generative network to take advantages of CNN, VAE and skip connections. Both S F -VAE and S C -VAE are efficient and effective generative networks and they can achieve better performance for detecting both local abnormal events and global abnormal events. The proposed S 2 -VAE is evaluated using four public datasets. The experimental results show that the S 2 -VAE outperforms the state-of-the-art algorithms. The code is available publicly at https://github.com/tianwangbuaa/.
|
['Chang Choi', 'Zhe Liu', 'Hichem Snoussi', 'Ce Li', 'Zhiwei Lin', 'Meina Qiao', 'Tian Wang']
|
2018-10-29
| null | null | null | null |
['abnormal-event-detection-in-video', 'semi-supervised-anomaly-detection', 'abnormal-event-detection-in-video']
|
['computer-vision', 'computer-vision', 'methodology']
|
[-5.45742989e-01 -2.80036569e-01 2.58971304e-01 -1.19927712e-01
-2.45133191e-01 -8.87903273e-02 4.65546966e-01 -4.28965986e-01
-1.12158947e-01 4.34723943e-01 2.28769898e-01 -3.84077191e-01
1.64002389e-01 -8.83317053e-01 -6.93369508e-01 -7.67962217e-01
-2.17802048e-01 9.25014690e-02 5.03089190e-01 -5.34797966e-01
-1.59433261e-01 3.90010357e-01 -1.24671006e+00 2.13073447e-01
7.37302899e-01 1.28658593e+00 -3.64621699e-01 9.83797133e-01
4.04564738e-02 1.03028047e+00 -5.49198627e-01 -5.56957722e-01
2.69554317e-01 -5.43825388e-01 -2.88410276e-01 7.79221654e-02
-1.53947085e-01 -8.27674866e-01 -6.85693741e-01 1.17674232e+00
3.81627053e-01 2.85053849e-01 7.30708122e-01 -1.78666317e+00
-6.76778436e-01 1.59038335e-01 -7.64565647e-01 6.21075034e-01
1.40084280e-02 3.59799594e-01 3.91167670e-01 -6.48463130e-01
1.55758202e-01 1.38729322e+00 8.53368163e-01 7.61580110e-01
-4.14399803e-01 -8.62311363e-01 1.87512234e-01 4.42270935e-01
-1.26649249e+00 -1.72676399e-01 7.61094093e-01 -6.34707153e-01
8.51146102e-01 1.34992465e-01 8.87475312e-01 1.43434083e+00
4.98673528e-01 9.26044345e-01 3.86208683e-01 2.50150442e-01
-1.11920878e-01 -4.60724309e-02 2.36371592e-01 9.14211869e-01
2.89364010e-01 2.48916075e-02 5.51219136e-02 -3.11575979e-01
8.91997457e-01 6.37982190e-01 -3.33324581e-01 3.02567720e-01
-6.63512766e-01 8.11946273e-01 4.19176906e-01 3.17371011e-01
-7.42610931e-01 6.89046830e-02 6.70240104e-01 2.38220200e-01
5.62784612e-01 -2.92893082e-01 -1.47922084e-01 -9.62733328e-02
-8.74470592e-01 2.36169472e-01 4.28301096e-01 3.94351155e-01
2.13381797e-01 4.46707517e-01 -4.46485549e-01 5.23284554e-01
5.74277818e-01 5.88092208e-01 5.81612587e-01 -7.04525352e-01
1.98951080e-01 7.36836970e-01 5.58759691e-03 -1.32608509e+00
-3.59271616e-01 -4.45610613e-01 -1.21065044e+00 1.06944382e-01
-4.73553985e-02 -7.11988449e-01 -8.50982428e-01 1.81527150e+00
4.16072667e-01 8.60494077e-01 3.11898738e-01 6.22078061e-01
1.25731623e+00 1.23705220e+00 -4.90534566e-02 -2.30523825e-01
1.00531673e+00 -9.23092723e-01 -9.68989968e-01 5.24827428e-02
1.33041531e-01 -5.50977170e-01 3.94195080e-01 2.55453050e-01
-9.85290527e-01 -6.98829889e-01 -8.46267104e-01 3.18267465e-01
-3.49383533e-01 2.91888446e-01 1.20018035e-01 4.42975849e-01
-1.12291324e+00 2.60285616e-01 -1.08548248e+00 -4.06821996e-01
8.17800343e-01 4.73087691e-02 -3.82648587e-01 2.93300092e-01
-1.37667036e+00 5.28575420e-01 4.86334890e-01 5.26367486e-01
-1.27721298e+00 -1.25789464e-01 -9.96707916e-01 2.54802108e-01
4.37722743e-01 -6.63335025e-01 1.07257104e+00 -8.32192600e-01
-1.23864579e+00 5.97297192e-01 -1.31628275e-01 -5.31107008e-01
4.18344259e-01 -2.95667291e-01 -7.54809976e-01 -4.38425988e-02
-1.15120411e-01 1.67746261e-01 7.51832426e-01 -1.09011722e+00
-4.88770694e-01 -3.49654496e-01 -4.52548563e-01 -4.46429327e-02
-3.11416477e-01 2.38415986e-01 -3.73354584e-01 -7.76101232e-01
-3.74928772e-01 -7.04799056e-01 -1.01954184e-01 -9.49557200e-02
-4.69163179e-01 -5.32646298e-01 1.55761445e+00 -1.23764062e+00
1.57648885e+00 -2.07664871e+00 -4.21833582e-02 1.93612561e-01
4.64285433e-01 1.09124684e+00 1.08481087e-01 4.05939519e-01
-1.36912778e-01 -4.05949764e-02 -2.15700716e-01 -3.31569344e-01
-2.95919359e-01 2.94534892e-01 -1.38542458e-01 5.18178403e-01
3.56318116e-01 9.55369592e-01 -7.69937098e-01 -2.46626019e-01
1.69863328e-01 6.49986863e-01 -5.35939932e-01 6.19848907e-01
4.96697612e-03 3.52100730e-01 -6.18947268e-01 6.67216480e-01
7.86514759e-01 -2.33401686e-01 -2.84504503e-01 7.09543154e-02
-6.91794883e-03 -4.07639027e-01 -1.06661844e+00 1.01673341e+00
4.79550868e-01 7.22007215e-01 1.94864348e-01 -1.24496305e+00
1.01617932e+00 4.78986651e-01 3.60737324e-01 -3.05676579e-01
7.69217193e-01 3.02834045e-02 -1.04464263e-01 -9.34603512e-01
1.74583405e-01 2.52074838e-01 1.57544985e-01 1.54855460e-01
1.97440058e-01 6.16837561e-01 1.13431700e-01 4.70257103e-01
8.94772410e-01 -3.83543037e-02 3.61455411e-01 -6.40533492e-02
9.12338495e-01 -4.85707730e-01 1.17538297e+00 4.86275703e-01
-7.20445693e-01 1.73665509e-01 7.08447158e-01 -6.96865141e-01
-9.39950347e-01 -9.95470941e-01 2.25842759e-01 6.14607155e-01
6.97674155e-02 -2.96521157e-01 -1.03702343e+00 -7.22739518e-01
-2.02798039e-01 5.68764448e-01 -8.46953213e-01 -5.61097085e-01
-4.83443379e-01 -8.71929884e-01 5.17723858e-01 6.26019657e-01
1.04578102e+00 -1.21449339e+00 -5.12604892e-01 1.73535012e-02
-3.60821187e-01 -8.71575356e-01 -5.53349853e-01 -7.68139303e-01
-4.17113245e-01 -1.33430398e+00 -7.70701230e-01 -4.49160606e-01
4.27480638e-01 1.98861271e-01 9.27442372e-01 2.29538515e-01
-1.27064466e-01 3.76739532e-01 -4.77128506e-01 -9.66184616e-01
-4.70672429e-01 -4.18028176e-01 2.08562598e-01 5.74771047e-01
7.03364313e-01 -5.56959212e-01 -6.27468944e-01 1.89507827e-01
-8.24374020e-01 -2.38734066e-01 2.69020945e-01 7.58998096e-01
4.32650328e-01 2.21311420e-01 4.50030625e-01 -6.48385763e-01
7.49962389e-01 -1.13314402e+00 -5.52111745e-01 7.18350112e-02
-2.12063819e-01 -4.47796315e-01 5.00486791e-01 -2.83254087e-01
-9.34098244e-01 -5.28497458e-01 -5.78279197e-01 -1.11512172e+00
-3.27968568e-01 3.04069936e-01 -1.55185550e-01 4.36102092e-01
2.48106062e-01 4.76253867e-01 1.37608021e-01 -2.56611645e-01
-1.72009647e-01 7.28298068e-01 9.30887997e-01 1.09120034e-01
7.06653655e-01 2.25752056e-01 -1.19338490e-01 -1.03801262e+00
-6.87375247e-01 -4.00056064e-01 -2.56915689e-01 -6.48346066e-01
1.35678601e+00 -9.55762684e-01 -5.82141101e-01 1.23450983e+00
-1.22334099e+00 -1.23896457e-01 1.42231300e-01 3.93903583e-01
-6.41175807e-02 4.34139580e-01 -6.71734095e-01 -1.13157439e+00
-7.64016688e-01 -1.15661967e+00 9.98953760e-01 8.26191187e-01
3.35170925e-01 -1.07829130e+00 2.64162719e-01 2.82871187e-01
3.94243002e-01 9.81195629e-01 2.56985128e-01 -8.91437292e-01
-1.86668545e-01 -2.34292224e-01 -8.72500017e-02 8.16114187e-01
5.95754981e-02 4.60984141e-01 -8.45573962e-01 -4.04033095e-01
1.22340292e-01 -1.56723373e-02 9.39301968e-01 7.86186993e-01
1.32934475e+00 -5.58390081e-01 -1.73741803e-01 8.29840600e-01
1.05517530e+00 6.12956703e-01 9.24598217e-01 1.47913590e-01
7.70703137e-01 6.95312619e-02 2.83596963e-01 5.68861187e-01
4.92871016e-01 1.84015825e-01 6.52929962e-01 -2.46229410e-01
4.39021178e-02 -7.99077451e-02 6.89927518e-01 7.66038597e-01
-3.40953290e-01 -5.24456859e-01 -8.44777167e-01 6.38994873e-01
-2.09766269e+00 -1.55124283e+00 -2.83455402e-01 1.70552611e+00
8.79705176e-02 8.86846557e-02 4.13042307e-01 2.17938572e-01
9.77361858e-01 2.46968880e-01 -5.76893687e-01 -4.75104898e-01
-9.28811077e-03 -1.75414547e-01 5.09731052e-03 2.18537614e-01
-1.36971653e+00 7.07645416e-01 4.79686832e+00 7.85772026e-01
-1.12604916e+00 2.64807224e-01 7.45183468e-01 -1.29557744e-01
2.41613135e-01 -5.99377930e-01 -7.09373474e-01 9.32463825e-01
9.49698627e-01 1.12867005e-01 1.28039941e-01 9.50401008e-01
2.16843069e-01 2.91540802e-01 -5.17645836e-01 1.07415867e+00
1.88232854e-01 -1.21593523e+00 1.27075640e-02 -6.60319999e-02
6.14380002e-01 7.84703195e-02 1.12856440e-01 5.09102762e-01
1.50282547e-01 -9.12568271e-01 4.89956588e-01 7.92968690e-01
4.73925740e-01 -1.09908235e+00 1.17719960e+00 4.67812151e-01
-1.41273010e+00 -2.37265244e-01 -3.46151352e-01 1.71249628e-01
2.63969779e-01 4.18398440e-01 -1.32579014e-01 5.74668169e-01
1.14864922e+00 9.69855130e-01 -4.64488626e-01 8.82222593e-01
-1.92463204e-01 9.46482897e-01 -8.54956508e-02 1.18559422e-02
6.71888828e-01 -2.43564859e-01 1.12348711e+00 1.29428566e+00
5.01023710e-01 3.35907727e-01 2.36282021e-01 5.45903921e-01
3.29174171e-03 -1.72025263e-01 -5.39017141e-01 -5.29000983e-02
2.93104053e-01 1.19936562e+00 -4.00825381e-01 -5.74895442e-01
-3.11764479e-01 8.08420002e-01 6.19852468e-02 2.30050728e-01
-1.21451604e+00 -4.59168971e-01 8.74424040e-01 -4.12987135e-02
3.55336994e-01 1.01454943e-01 4.61375207e-01 -1.33316004e+00
-6.94371713e-03 -9.81529236e-01 7.64111340e-01 -7.88362265e-01
-1.36756623e+00 9.79253352e-01 7.22198635e-02 -1.12335253e+00
-2.50070333e-01 -7.14582682e-01 -1.39563322e+00 7.84303904e-01
-1.11565351e+00 -1.20900917e+00 -7.57955253e-01 1.19013762e+00
6.28744781e-01 -6.87832594e-01 4.98177975e-01 3.76665890e-01
-1.37951910e+00 5.74650288e-01 9.31782126e-02 7.00127602e-01
3.54739219e-01 -8.92072022e-01 4.89957869e-01 1.58590662e+00
-2.50441432e-01 3.80600959e-01 5.04773080e-01 -8.73278201e-01
-7.52096057e-01 -1.53468263e+00 3.76823366e-01 -3.05978537e-01
3.12312514e-01 -1.66280001e-01 -1.14164090e+00 1.01843846e+00
4.03751999e-01 2.21921131e-01 7.59442091e-01 -4.58791673e-01
-1.03285886e-01 -5.72496988e-02 -1.33490038e+00 3.41551989e-01
6.93632483e-01 -2.01518208e-01 -5.24243355e-01 1.62376955e-01
5.67256987e-01 -3.90309691e-01 -6.02442026e-01 4.03654397e-01
4.23607618e-01 -1.43881547e+00 9.05412734e-01 -6.95989549e-01
4.73808587e-01 -2.96589077e-01 -5.02825081e-02 -1.18351734e+00
-4.76598769e-01 -5.17040730e-01 -6.08629405e-01 1.15906382e+00
3.08531709e-03 -8.16262543e-01 4.97693956e-01 3.42662632e-01
-3.38742256e-01 -1.03868544e+00 -7.00765133e-01 -5.60653090e-01
-2.04478398e-01 -1.94248438e-01 7.16719091e-01 1.07090390e+00
-7.43024290e-01 -6.38736486e-02 -9.15154338e-01 5.16879916e-01
5.51743031e-01 -4.21619445e-01 8.39439154e-01 -1.21424425e+00
-2.78215051e-01 -5.81099510e-01 -6.83395982e-01 -6.78765357e-01
9.92229860e-03 -4.28068966e-01 -3.70102048e-01 -1.43346012e+00
2.62675196e-01 3.21986198e-01 -4.52427328e-01 3.87285650e-01
-5.81406713e-01 1.46915615e-01 -5.76040149e-02 -1.98613927e-02
-6.12201512e-01 8.38640869e-01 1.01289046e+00 1.68456919e-02
-2.99090475e-01 1.91102579e-01 -4.53130245e-01 1.01081359e+00
1.06542444e+00 -1.72689959e-01 -3.52367252e-01 -2.54492164e-01
-3.08528662e-01 3.98251712e-02 5.47979295e-01 -1.07345080e+00
2.36691803e-01 -7.93025568e-02 7.30041146e-01 -8.98214996e-01
1.35336474e-01 -5.52650809e-01 1.07222095e-01 7.74562597e-01
1.59649014e-01 4.22427326e-01 2.65720338e-01 6.42363369e-01
-3.90632272e-01 8.07872787e-02 8.73392463e-01 -6.34488314e-02
-7.96663463e-01 7.75895298e-01 -4.70497698e-01 -1.06111705e-01
1.37927008e+00 -1.39157800e-02 -3.92761976e-01 -6.65734351e-01
-5.71313202e-01 6.46911621e-01 -7.03654364e-02 5.27754366e-01
9.69990611e-01 -1.59151852e+00 -9.93152142e-01 4.92319047e-01
-2.85245955e-01 1.19171590e-01 7.83508956e-01 7.53414869e-01
-7.20403254e-01 1.28971547e-01 -3.99385750e-01 -5.06797910e-01
-1.48228002e+00 5.76766014e-01 6.04131341e-01 -3.47043306e-01
-5.95198154e-01 1.20004201e+00 1.25603959e-01 -5.49306832e-02
2.78831840e-01 -1.22418683e-02 -7.41498411e-01 -2.50052780e-01
8.00406158e-01 7.33864963e-01 -5.39647758e-01 -1.11209083e+00
-3.69024664e-01 3.03068429e-01 -1.55317010e-02 2.27342144e-01
1.41253221e+00 4.02614921e-02 -3.47176939e-01 1.33681327e-01
1.03443098e+00 -2.09751129e-01 -1.11167467e+00 4.74661738e-02
-8.17270339e-01 -3.34227711e-01 2.98924983e-01 -2.36662626e-01
-1.55569208e+00 8.50411057e-01 7.77229548e-01 3.60253990e-01
1.57857978e+00 -2.50764340e-01 1.22507858e+00 1.71844020e-01
-1.76517546e-01 -6.90207422e-01 2.12130234e-01 3.98117930e-01
8.61959517e-01 -1.42653513e+00 -3.45003426e-01 1.68376099e-02
-7.66205549e-01 9.47075903e-01 9.83164728e-01 -4.25780445e-01
1.01796651e+00 -4.73925434e-02 8.77073482e-02 -4.06400383e-01
-5.04352391e-01 -2.59148460e-02 5.47842801e-01 4.70940292e-01
7.82015026e-02 -6.69885650e-02 2.46578343e-02 1.10629106e+00
-1.16274506e-02 -1.04918234e-01 3.00666302e-01 7.34762967e-01
-4.08162564e-01 -6.83831513e-01 -4.30604368e-01 6.41480207e-01
-6.15829706e-01 3.62678468e-01 -2.56839186e-01 6.78384125e-01
5.17177820e-01 9.89264071e-01 3.74886125e-01 -8.42757344e-01
1.60754785e-01 7.13918284e-02 -8.56912509e-02 1.48241594e-02
-5.63770175e-01 1.35475144e-01 -3.08801591e-01 -8.14550459e-01
-3.06543946e-01 -5.55804193e-01 -9.15852010e-01 -6.07899427e-01
-2.83100493e-02 1.61437824e-01 3.21842074e-01 6.95712090e-01
3.09516132e-01 8.19878995e-01 7.33312786e-01 -6.11910641e-01
-2.05275059e-01 -1.03048575e+00 -5.09288132e-01 3.57938349e-01
6.10022247e-01 -6.59912765e-01 -2.38669246e-01 -1.72859699e-01]
|
[7.884372711181641, 1.5129941701889038]
|
cc672e14-fd96-4e6e-aacf-cfb022fd5377
|
lightweight-high-performance-blind-image
|
2303.13057
| null |
https://arxiv.org/abs/2303.13057v1
|
https://arxiv.org/pdf/2303.13057v1.pdf
|
Lightweight High-Performance Blind Image Quality Assessment
|
Blind image quality assessment (BIQA) is a task that predicts the perceptual quality of an image without its reference. Research on BIQA attracts growing attention due to the increasing amount of user-generated images and emerging mobile applications where reference images are unavailable. The problem is challenging due to the wide range of content and mixed distortion types. Many existing BIQA methods use deep neural networks (DNNs) to achieve high performance. However, their large model sizes hinder their applicability to edge or mobile devices. To meet the need, a novel BIQA method with a small model, low computational complexity, and high performance is proposed and named "GreenBIQA" in this work. GreenBIQA includes five steps: 1) image cropping, 2) unsupervised representation generation, 3) supervised feature selection, 4) distortion-specific prediction, and 5) regression and decision ensemble. Experimental results show that the performance of GreenBIQA is comparable with that of state-of-the-art deep-learning (DL) solutions while demanding a much smaller model size and significantly lower computational complexity.
|
['C. -C. Jay Kuo', 'Yong Yan', 'Xingze He', 'Yun-Cheng Wang', 'Zhanxuan Mei']
|
2023-03-23
| null | null | null | null |
['blind-image-quality-assessment', 'image-quality-assessment', 'image-cropping']
|
['computer-vision', 'computer-vision', 'computer-vision']
|
[ 1.77171677e-01 -6.25748873e-01 1.60614073e-01 -2.70015895e-01
-9.65695083e-01 -1.11114465e-01 3.97627383e-01 -2.14863688e-01
-2.58779377e-01 5.42789400e-01 3.04874271e-01 -2.92951465e-01
-9.07156095e-02 -6.98633075e-01 -3.08818936e-01 -7.86365688e-01
2.26985529e-01 -6.05226643e-02 3.93975765e-01 -9.86515656e-02
3.21676999e-01 4.21452850e-01 -1.56243086e+00 2.82865226e-01
1.32076955e+00 1.37346911e+00 2.79574603e-01 7.58780837e-01
4.01937142e-02 1.02903140e+00 -6.06770813e-01 -6.07588649e-01
3.43227029e-01 -6.30364776e-01 -4.69528884e-01 1.98807627e-01
4.75234210e-01 -7.69936085e-01 -5.20154238e-01 1.29013002e+00
1.08548772e+00 5.51929325e-02 4.63674366e-01 -1.23720682e+00
-1.11625803e+00 -7.19962418e-02 -5.52733839e-01 3.42143565e-01
8.02329928e-02 3.90026569e-01 7.51186073e-01 -1.05657208e+00
9.79358554e-02 1.00573838e+00 5.51834345e-01 6.08105540e-01
-7.90308714e-01 -6.45908594e-01 -9.66851190e-02 8.58392596e-01
-1.32609022e+00 -6.92331851e-01 8.03713083e-01 -4.40073967e-01
7.79445410e-01 9.74834934e-02 3.24010462e-01 7.80573368e-01
7.50946850e-02 7.50152469e-01 1.18756628e+00 -4.57427412e-01
4.18724656e-01 -7.19144344e-02 -2.34964952e-01 6.17568612e-01
-1.64118558e-02 2.32385993e-01 -4.60101008e-01 2.27367207e-01
6.74798489e-01 -7.85519853e-02 -4.68021065e-01 -3.91306341e-01
-9.81711864e-01 6.49329126e-01 7.26015925e-01 1.57446817e-01
-5.77750862e-01 -5.92970699e-02 3.05646062e-01 2.65678763e-01
3.65043521e-01 2.52512783e-01 -1.74147755e-01 -1.67572945e-01
-1.09971690e+00 9.28474665e-02 3.52265656e-01 7.82292068e-01
5.26242912e-01 3.38533223e-01 -3.42724353e-01 1.16323268e+00
2.82150596e-01 4.80685651e-01 7.31497586e-01 -9.85044479e-01
5.70666730e-01 5.76587021e-01 2.94890881e-01 -1.19693792e+00
-1.72637120e-01 -5.63833714e-01 -1.23764277e+00 5.53254008e-01
1.86301962e-01 7.18776956e-02 -1.08233047e+00 1.21088111e+00
-3.72872651e-02 -1.13878688e-02 3.29365619e-02 1.31435919e+00
1.08815265e+00 8.22433949e-01 -1.24058016e-01 -2.28088453e-01
9.89195585e-01 -1.18624139e+00 -7.09437549e-01 -2.12607220e-01
1.47315800e-01 -7.86520898e-01 1.02543259e+00 6.69257402e-01
-1.24410844e+00 -9.04902637e-01 -1.27771211e+00 -2.76409030e-01
-2.92505682e-01 4.50679123e-01 4.99114871e-01 8.68806601e-01
-1.32307386e+00 2.90673763e-01 -3.43936443e-01 -2.36106245e-03
7.19472945e-01 3.18214238e-01 -1.85195953e-01 -5.20135105e-01
-1.07185435e+00 7.72067666e-01 -2.54882015e-02 3.35662603e-01
-1.03872406e+00 -3.47611308e-01 -7.99677908e-01 2.09506959e-01
2.40849182e-01 -6.64092302e-01 1.20573890e+00 -1.31286454e+00
-1.65812922e+00 5.33908427e-01 -1.84911400e-01 -4.10851926e-01
6.45045102e-01 -2.88073212e-01 -7.26712644e-01 2.15838358e-01
7.71665340e-03 6.26664400e-01 1.24243760e+00 -1.34124148e+00
-7.50065267e-01 -4.26308006e-01 1.02878310e-01 4.21525568e-01
-3.80550832e-01 7.97733665e-02 -8.69236767e-01 -6.25669301e-01
1.37398213e-01 -5.77170491e-01 -2.00142667e-01 1.69314653e-01
-1.18562281e-01 -1.59912005e-01 6.00434721e-01 -1.01880705e+00
1.48931301e+00 -2.02442265e+00 -6.22627176e-02 -1.08002774e-01
3.29465449e-01 8.27897847e-01 -3.13092291e-01 8.62221792e-02
9.67945009e-02 -1.13836698e-01 -3.15561295e-01 -2.12487310e-01
-6.53654486e-02 -1.70957312e-01 3.23550925e-02 3.29984337e-01
2.51431614e-01 7.67663002e-01 -8.49982977e-01 -3.91375691e-01
4.74462926e-01 4.44238275e-01 -5.35461724e-01 4.47986215e-01
1.42857775e-01 1.64064452e-01 -1.26254693e-01 8.94794881e-01
9.87500489e-01 -2.75397688e-01 -1.47481486e-01 -6.22981608e-01
-7.92734250e-02 1.60014674e-01 -1.29975522e+00 1.42237580e+00
-5.29089928e-01 7.96852291e-01 -4.13781293e-02 -8.12725306e-01
9.50939655e-01 2.81074315e-01 2.70818532e-01 -1.34412372e+00
9.73862261e-02 3.25980604e-01 2.14961275e-01 -6.41132772e-01
4.98763591e-01 1.27093241e-01 4.00987238e-01 2.35807166e-01
4.60512079e-02 5.60204610e-02 2.44125441e-01 -3.09276562e-02
8.68295908e-01 7.95088336e-03 1.36708543e-01 1.72733396e-01
7.45239258e-01 -3.98054630e-01 7.67556667e-01 5.11388898e-01
-6.26638591e-01 8.56512487e-01 4.80847172e-02 -4.23028529e-01
-1.01502967e+00 -1.01819456e+00 1.52413070e-01 7.92286515e-01
5.30847132e-01 -1.44250467e-01 -8.28314304e-01 -5.28834462e-01
-3.34505618e-01 3.83599341e-01 -3.59622449e-01 -2.25091532e-01
-3.99342299e-01 -7.77903855e-01 1.28477946e-01 5.94318748e-01
1.09495318e+00 -1.35446525e+00 -4.37886477e-01 1.29513428e-01
-3.72546911e-01 -9.51339781e-01 -4.74650830e-01 -3.11845541e-01
-7.49182045e-01 -9.62740779e-01 -1.12877154e+00 -9.61782515e-01
6.02775812e-01 5.79420328e-01 1.14847231e+00 1.94928572e-02
-8.58765468e-02 6.66226223e-02 -4.05753762e-01 -3.48785013e-01
-2.30331302e-01 -3.63986284e-01 -3.16231512e-02 2.97616571e-01
4.18314904e-01 -3.04436535e-01 -1.09835184e+00 3.41611326e-01
-1.03790426e+00 5.80448508e-02 9.61355567e-01 1.03495145e+00
5.68626285e-01 5.22126257e-01 7.18792796e-01 -3.75798017e-01
8.17754924e-01 -1.80665031e-01 -6.42336369e-01 4.16965723e-01
-7.15880096e-01 -3.11009109e-01 5.61587095e-01 -2.64364541e-01
-1.31803858e+00 -2.03628033e-01 -2.42551625e-01 -2.52173096e-01
-6.95974082e-02 3.95083845e-01 -5.35398483e-01 -1.77283898e-01
8.21902394e-01 3.89722466e-01 -6.47223070e-02 -4.39903438e-01
2.59799451e-01 1.03062654e+00 6.56850994e-01 8.63965079e-02
7.03200638e-01 2.27518469e-01 -2.25221857e-01 -6.69423401e-01
-6.15164340e-01 -4.24016356e-01 -4.62751418e-01 -4.00098711e-01
7.02763259e-01 -1.08006620e+00 -4.81901973e-01 9.33517814e-01
-9.95008647e-01 -2.53758460e-01 4.14753817e-02 4.85500306e-01
-3.77534330e-01 5.38251162e-01 -5.20092249e-01 -7.24988222e-01
-6.85663760e-01 -1.44556499e+00 7.16995478e-01 4.67931420e-01
3.08741003e-01 -5.99073708e-01 -2.99367130e-01 7.22805500e-01
5.87141752e-01 -2.24988505e-01 7.24870563e-01 8.64974037e-02
-7.88140833e-01 -1.77978307e-01 -9.22747374e-01 8.42214525e-01
3.31201226e-01 -3.17090303e-01 -1.05776787e+00 -2.69998014e-01
5.33786020e-04 -2.72984475e-01 7.47270584e-01 6.41211092e-01
1.38998985e+00 -2.67703295e-01 1.62930250e-01 6.72885358e-01
1.57467186e+00 5.74400187e-01 9.35117066e-01 4.55678314e-01
6.90946400e-01 3.17095965e-01 5.75828731e-01 2.13544041e-01
3.99104804e-01 6.93219125e-01 6.48463011e-01 -3.34371448e-01
-5.34162283e-01 -8.20979178e-02 4.00082827e-01 9.99811709e-01
5.46697080e-02 -3.71265382e-01 -8.66744637e-01 9.13366795e-01
-1.72459710e+00 -8.81789923e-01 -9.95892212e-02 2.03163695e+00
6.79601133e-01 1.17031761e-01 2.31294278e-02 5.11939049e-01
6.61588788e-01 8.60483870e-02 -8.08019280e-01 -3.85030806e-01
-1.47595286e-01 3.67186740e-02 3.09322685e-01 1.94955081e-01
-1.11528909e+00 5.90263486e-01 5.82599831e+00 8.90299320e-01
-1.11115503e+00 1.10776655e-01 8.58469486e-01 1.83980688e-01
4.32257801e-02 -4.86269981e-01 -3.56471211e-01 6.78797603e-01
7.25477517e-01 1.51096985e-01 5.50487876e-01 6.81684375e-01
2.67551273e-01 -9.94220003e-02 -6.69648111e-01 1.51186001e+00
3.92272383e-01 -1.14322639e+00 2.16001645e-01 -1.49338886e-01
8.97970676e-01 -1.60391666e-02 2.80500799e-01 1.06077150e-01
1.04287103e-01 -1.05672145e+00 6.44342244e-01 5.74532270e-01
1.02976346e+00 -7.59163022e-01 9.43146825e-01 4.44559306e-02
-8.96865904e-01 -4.57732797e-01 -5.58851779e-01 7.41711110e-02
1.02903254e-01 6.91557467e-01 -3.84376526e-01 4.75592762e-01
9.98683929e-01 7.85942435e-01 -7.07608283e-01 1.59259760e+00
-2.44550884e-01 7.25451410e-01 2.67785251e-01 2.15449810e-01
1.67953297e-01 -1.34723529e-01 3.48221809e-01 9.65143085e-01
5.12206137e-01 8.13530758e-03 -1.45223290e-01 6.21182680e-01
-2.30029061e-01 6.18575737e-02 -2.21677125e-01 1.14446603e-01
2.90069610e-01 1.06711388e+00 -3.21913481e-01 -2.34175920e-01
-6.55445457e-01 1.22970068e+00 -3.62960398e-02 4.71288085e-01
-5.99968314e-01 -5.69448590e-01 6.16610944e-01 -4.83671203e-02
3.57518733e-01 -1.25591248e-01 -3.88512313e-01 -1.02186632e+00
2.07498312e-01 -1.11605406e+00 2.09477514e-01 -1.22638857e+00
-1.40347934e+00 8.92988086e-01 -6.98350310e-01 -1.70344663e+00
-9.92739946e-02 -6.79406404e-01 -4.69585240e-01 1.16834021e+00
-1.93802893e+00 -1.06312108e+00 -6.94454372e-01 7.33098686e-01
8.73503983e-01 -4.39269692e-01 6.61008954e-01 6.27954483e-01
-5.11445403e-01 7.33936727e-01 3.31883669e-01 1.93961233e-01
7.64829397e-01 -1.24264181e+00 3.14879984e-01 1.23362350e+00
-1.56334311e-01 1.25668928e-01 2.74696738e-01 -2.95733601e-01
-1.12149608e+00 -1.15029442e+00 9.25330341e-01 -1.14675760e-01
3.13751101e-01 4.56105210e-02 -9.12671387e-01 -7.06321150e-02
1.27867281e-01 1.60204738e-01 5.01617730e-01 -1.80289984e-01
-2.35854968e-01 -6.39329612e-01 -1.12513494e+00 6.51186049e-01
8.81972373e-01 -6.47087276e-01 -3.10971528e-01 -2.58357804e-02
3.89763325e-01 -3.52084100e-01 -5.14543176e-01 3.70621085e-01
5.42071879e-01 -1.33803594e+00 9.64924872e-01 3.02697700e-02
5.85438371e-01 -6.45753205e-01 -1.54827729e-01 -1.42260993e+00
-6.49075568e-01 -2.55019844e-01 -3.19685876e-01 1.23074162e+00
1.56907141e-01 -1.22801982e-01 4.69519764e-01 5.24532437e-01
-1.29694924e-01 -6.97872221e-01 -7.94039190e-01 -5.80663383e-01
-2.96477586e-01 -4.03440237e-01 7.77637899e-01 6.33045197e-01
-4.18592483e-01 2.26176217e-01 -6.74393415e-01 2.00926900e-01
5.92442751e-01 1.51876107e-01 5.53922892e-01 -1.01682115e+00
-1.42276227e-01 -5.06601214e-01 -6.43135309e-01 -1.14045095e+00
-4.55374897e-01 -4.32677448e-01 2.15123951e-01 -2.03923631e+00
1.09995313e-01 -4.42661524e-01 -5.31118989e-01 1.38661847e-01
-4.33490604e-01 6.20544970e-01 1.01924725e-01 3.07024390e-01
-8.50453854e-01 7.52217174e-01 1.39896667e+00 -4.16691363e-01
-2.35588148e-01 -1.10044805e-02 -7.61933088e-01 5.61809599e-01
7.39482224e-01 1.76377036e-02 -5.84433019e-01 -8.17230999e-01
1.46933645e-01 -1.10767717e-02 3.73658329e-01 -1.43967402e+00
2.66131699e-01 1.23773813e-01 7.02609122e-01 -5.08467317e-01
2.74231642e-01 -8.04536283e-01 -1.62204862e-01 2.67695814e-01
-5.45194298e-02 2.18145140e-02 6.46558031e-02 3.71696591e-01
-6.20530367e-01 -1.53716907e-01 9.34048474e-01 3.19694988e-02
-1.13674474e+00 5.11440635e-01 -2.21205235e-01 -2.68166393e-01
6.85385644e-01 -4.22028035e-01 -2.88735747e-01 -7.10129559e-01
-4.41318661e-01 -1.27438288e-02 3.08432907e-01 6.19130731e-01
1.04455912e+00 -1.50748122e+00 -7.77573824e-01 3.00368011e-01
1.36500686e-01 -1.11924835e-01 6.34988427e-01 5.27927697e-01
-5.47574282e-01 2.63041556e-01 -4.61887866e-01 -4.24490869e-01
-1.19447088e+00 5.51436901e-01 4.67188925e-01 -1.31945148e-01
-3.56663108e-01 8.09042454e-01 2.70067215e-01 -1.71920899e-02
2.96120763e-01 1.82507131e-02 -4.96335596e-01 -2.03757267e-02
8.66992295e-01 6.42574012e-01 3.70754570e-01 -9.14954662e-01
-1.57105893e-01 4.33670253e-01 -4.71462347e-02 1.60943389e-01
1.22089183e+00 -4.22269613e-01 5.85380457e-02 -7.48584345e-02
1.00782144e+00 -2.44162545e-01 -1.47294140e+00 -3.81536394e-01
-2.61157840e-01 -8.53591084e-01 5.52204370e-01 -1.23134768e+00
-1.37963879e+00 1.29323769e+00 1.32599187e+00 1.00077756e-01
1.83074725e+00 -4.25927222e-01 9.46444035e-01 2.32561976e-02
3.50119770e-01 -1.12157726e+00 2.82024235e-01 2.05679521e-01
1.13385177e+00 -1.46648812e+00 -5.74129410e-02 -6.23606779e-02
-7.25010276e-01 8.88560355e-01 8.02107692e-01 9.54486057e-02
6.08570814e-01 -3.24821919e-01 4.11907673e-01 1.37694865e-01
-3.82614166e-01 -3.41249764e-01 7.29731858e-01 1.00189960e+00
2.92131245e-01 -1.17747948e-01 -1.81416243e-01 7.31323898e-01
3.56217399e-02 2.81307757e-01 3.78739387e-01 4.95221555e-01
-2.64102072e-01 -9.45898533e-01 -3.48837644e-01 6.77329719e-01
-5.70784330e-01 -3.28323632e-01 -7.61546986e-03 2.11371750e-01
4.33332831e-01 1.45561993e+00 -1.38867497e-01 -4.79448766e-01
2.97359318e-01 -4.88907814e-01 2.44657367e-01 -2.44998634e-01
-3.04052830e-01 -6.51643425e-02 -1.80462465e-01 -6.62989199e-01
-6.12626851e-01 -3.04878473e-01 -6.22449219e-01 -1.68207750e-01
-3.53806168e-01 -1.38102621e-01 6.44366145e-01 7.68972695e-01
4.30035114e-01 5.80502033e-01 8.02413464e-01 -8.22381854e-01
-2.79572785e-01 -1.07811368e+00 -5.10546207e-01 4.82640088e-01
6.18987918e-01 -5.73471367e-01 -9.54927281e-02 3.08317810e-01]
|
[11.911710739135742, -1.7829917669296265]
|
63eebd1f-3044-4110-9e66-3fc3de6467df
|
category-level-6d-object-pose-and-size
|
2207.05444
| null |
https://arxiv.org/abs/2207.05444v2
|
https://arxiv.org/pdf/2207.05444v2.pdf
|
Category-Level 6D Object Pose and Size Estimation using Self-Supervised Deep Prior Deformation Networks
|
It is difficult to precisely annotate object instances and their semantics in 3D space, and as such, synthetic data are extensively used for these tasks, e.g., category-level 6D object pose and size estimation. However, the easy annotations in synthetic domains bring the downside effect of synthetic-to-real (Sim2Real) domain gap. In this work, we aim to address this issue in the task setting of Sim2Real, unsupervised domain adaptation for category-level 6D object pose and size estimation. We propose a method that is built upon a novel Deep Prior Deformation Network, shortened as DPDN. DPDN learns to deform features of categorical shape priors to match those of object observations, and is thus able to establish deep correspondence in the feature space for direct regression of object poses and sizes. To reduce the Sim2Real domain gap, we formulate a novel self-supervised objective upon DPDN via consistency learning; more specifically, we apply two rigid transformations to each object observation in parallel, and feed them into DPDN respectively to yield dual sets of predictions; on top of the parallel learning, an inter-consistency term is employed to keep cross consistency between dual predictions for improving the sensitivity of DPDN to pose changes, while individual intra-consistency ones are used to enforce self-adaptation within each learning itself. We train DPDN on both training sets of the synthetic CAMERA25 and real-world REAL275 datasets; our results outperform the existing methods on REAL275 test set under both the unsupervised and supervised settings. Ablation studies also verify the efficacy of our designs. Our code is released publicly at https://github.com/JiehongLin/Self-DPDN.
|
['Kui Jia', 'Changxing Ding', 'Zewei Wei', 'Jiehong Lin']
|
2022-07-12
| null | null | null | null |
['6d-pose-estimation-1']
|
['computer-vision']
|
[ 8.21860209e-02 1.35467499e-01 -6.16182089e-02 -4.71262664e-01
-7.92976916e-01 -6.43147171e-01 5.97434103e-01 -2.34021962e-01
-2.41107389e-01 3.76894593e-01 -7.73168951e-02 3.19320560e-01
-9.39493440e-03 -6.38317704e-01 -1.11561251e+00 -7.34093606e-01
2.93008953e-01 7.19599187e-01 4.10171062e-01 1.96074292e-01
5.09898923e-02 5.67989111e-01 -1.45530200e+00 -2.33659763e-02
7.67099321e-01 1.17037153e+00 3.62235874e-01 1.85092449e-01
2.74645597e-01 3.64659131e-01 -2.52240539e-01 -3.09675574e-01
7.20423758e-01 -7.36099184e-02 -4.82858330e-01 4.93047565e-01
5.07115722e-01 -3.24608058e-01 -3.48063856e-01 1.08463466e+00
4.82854724e-01 -6.54604565e-03 9.07917917e-01 -1.36550200e+00
-7.25268126e-01 1.08763911e-01 -7.81419516e-01 -2.33719662e-01
1.54483736e-01 4.56155270e-01 8.20050895e-01 -1.06364787e+00
5.91816127e-01 1.22715724e+00 6.69151068e-01 5.13735116e-01
-1.44790494e+00 -7.32937515e-01 1.89216197e-01 -1.72903821e-01
-1.35712421e+00 -3.46052259e-01 9.71427679e-01 -6.76148951e-01
4.13560897e-01 -1.15265839e-01 4.40533519e-01 1.39632726e+00
-4.89928052e-02 6.63085520e-01 1.08535314e+00 -1.01821326e-01
1.89041406e-01 6.69272020e-02 -1.69878677e-01 5.31354904e-01
2.80686140e-01 7.85362422e-02 -3.13338310e-01 -1.14200385e-02
1.19146490e+00 7.09795654e-02 -1.68092445e-01 -9.40110505e-01
-1.27098095e+00 6.43164515e-01 4.46949452e-01 -1.01817690e-01
-1.58071622e-01 4.46742885e-02 3.46176744e-01 1.13516733e-01
5.90198100e-01 3.33738357e-01 -7.03786910e-01 1.63490832e-01
-2.82912910e-01 5.67797959e-01 4.49853390e-01 1.30495536e+00
8.28330755e-01 -1.95736155e-01 -2.24526703e-01 9.33637798e-01
5.36996424e-01 7.86965907e-01 5.72223663e-01 -9.35323894e-01
6.45819724e-01 6.90128744e-01 2.64212430e-01 -1.04472303e+00
-3.64580512e-01 -4.25111443e-01 -8.41563284e-01 1.11276820e-01
4.56716150e-01 1.67074487e-01 -9.73330677e-01 1.92585802e+00
6.46032333e-01 1.85422152e-01 -8.88883993e-02 1.11647999e+00
7.32118785e-01 4.06403631e-01 -1.41829895e-02 3.55384266e-03
1.22884393e+00 -8.79974008e-01 -1.86501771e-01 -3.57395440e-01
4.98618633e-01 -7.56721079e-01 1.38005424e+00 9.08617005e-02
-9.29233789e-01 -7.24011123e-01 -8.43861520e-01 -3.47492956e-02
-1.11225806e-02 4.48330104e-01 4.05817986e-01 1.54713333e-01
-4.26185489e-01 5.06962419e-01 -9.70156431e-01 -1.89148545e-01
5.54999232e-01 4.00687307e-01 -4.12541449e-01 1.99903734e-02
-9.48882341e-01 7.39256561e-01 5.07153988e-01 1.46747977e-01
-8.52727771e-01 -8.21702123e-01 -9.92019057e-01 -3.31114352e-01
6.07135296e-01 -7.07329929e-01 1.22605014e+00 -8.59428167e-01
-1.61684668e+00 1.33632672e+00 2.68103689e-01 -1.76784337e-01
8.65639985e-01 -2.92393148e-01 -1.27437443e-01 -1.89911678e-01
2.62755781e-01 7.11418748e-01 9.28518355e-01 -1.51212668e+00
-2.67157406e-01 -6.73412502e-01 7.43598640e-02 2.90864199e-01
-2.16779932e-01 -3.61220032e-01 -8.37737441e-01 -7.75953472e-01
3.45986158e-01 -1.25765800e+00 -5.59106804e-02 3.78280282e-01
-4.43420291e-01 -2.77832419e-01 7.49060810e-01 -4.11869884e-01
6.30605042e-01 -2.26876140e+00 2.52222568e-01 7.53851011e-02
1.48089334e-01 1.18649997e-01 -2.49799818e-01 -1.29272267e-01
-8.73392355e-03 -1.75200060e-01 -4.78525490e-01 -5.34965038e-01
1.12117194e-01 3.99372578e-01 -2.07970411e-01 7.66663015e-01
4.88643199e-01 8.50872099e-01 -8.10824096e-01 -5.02864420e-01
1.63931161e-01 2.98365265e-01 -6.02191150e-01 5.94187796e-01
-4.35893357e-01 7.74172485e-01 -7.37839282e-01 5.67723930e-01
1.02304339e+00 -2.79920757e-01 -2.35025696e-02 -4.62807983e-01
1.39977232e-01 -1.30956816e-02 -1.25445414e+00 1.86610460e+00
-4.67037082e-01 1.60786659e-01 2.08283383e-02 -1.12641060e+00
1.11726499e+00 3.66302207e-02 6.55328572e-01 -5.28905153e-01
6.51421323e-02 1.96677506e-01 -2.00197965e-01 -5.24181306e-01
1.11695714e-01 -8.19247514e-02 -1.38584182e-01 1.86901897e-01
9.85850543e-02 -5.77947259e-01 -8.52863565e-02 -2.25937501e-01
6.10289216e-01 6.56633794e-01 1.53230220e-01 -2.09976941e-01
4.80110705e-01 -1.86497703e-01 7.42926240e-01 3.21185380e-01
-7.31837973e-02 1.07146573e+00 4.64741319e-01 -4.14542705e-01
-1.36745536e+00 -1.27019584e+00 -3.24675679e-01 7.38458931e-01
5.80622971e-01 -4.45316955e-02 -6.00669444e-01 -9.44794059e-01
3.61802012e-01 3.87735993e-01 -6.27272069e-01 -2.38714069e-01
-6.91024184e-01 -7.47429252e-01 2.38218814e-01 6.17900550e-01
5.74035823e-01 -8.51077557e-01 -3.48098695e-01 -1.34206610e-02
5.09061366e-02 -1.41854358e+00 -5.17348528e-01 6.01382293e-02
-6.67619884e-01 -1.08568227e+00 -7.73036718e-01 -8.13188732e-01
8.30003917e-01 4.59126756e-02 9.54759240e-01 -3.27351332e-01
-8.82224813e-02 3.25298011e-01 -2.16701642e-01 -4.08509344e-01
-3.70557398e-01 7.33212605e-02 2.99690425e-01 1.27193540e-01
1.24723576e-01 -7.62966216e-01 -7.57623851e-01 7.52030313e-01
-8.64488721e-01 2.13259399e-01 6.11106157e-01 8.13649237e-01
9.73167181e-01 -3.59081447e-01 3.44851166e-01 -7.48535156e-01
1.59330890e-01 -3.79307270e-01 -8.92938375e-01 6.94151074e-02
-4.27053839e-01 1.66253626e-01 6.69911861e-01 -6.87111318e-01
-9.32347775e-01 3.78183246e-01 -6.56157061e-02 -8.48955512e-01
-2.25947216e-01 1.17899291e-01 -5.34099281e-01 -5.61279207e-02
6.26456678e-01 1.25649646e-01 1.66548371e-01 -5.40951669e-01
5.88406846e-02 4.78685588e-01 5.92662513e-01 -8.41517150e-01
1.12360203e+00 3.59690249e-01 4.08475427e-03 -3.96088958e-01
-1.08434963e+00 -2.61479676e-01 -7.68621266e-01 -2.41532661e-02
7.44746923e-01 -1.01321292e+00 -5.57983220e-01 6.93408966e-01
-1.07479274e+00 -4.72420454e-01 -4.02085841e-01 4.71719682e-01
-8.94512475e-01 3.38203698e-01 -2.62584299e-01 -4.34978783e-01
-3.83872651e-02 -1.20276034e+00 1.42499483e+00 5.35927601e-02
1.25478312e-01 -7.49660730e-01 3.60899903e-02 3.84661019e-01
-8.64229724e-02 4.58295017e-01 6.15161717e-01 -5.88058770e-01
-4.50727612e-01 -7.75794759e-02 -4.13067073e-01 6.14147544e-01
1.30361751e-01 -1.06498979e-01 -9.12551165e-01 -4.08121884e-01
1.19476311e-01 -6.04151785e-01 4.89846319e-01 3.08359683e-01
1.59972322e+00 -1.04580700e-01 -2.34405845e-01 7.89303839e-01
1.35381663e+00 -1.84020892e-01 4.45004165e-01 3.34164798e-01
9.75074172e-01 5.40829122e-01 9.19892669e-01 5.01876414e-01
3.77671868e-01 9.82566416e-01 6.62086427e-01 2.79872585e-02
-1.50129765e-01 -4.14481908e-01 2.13317543e-01 6.04769886e-01
-1.44270137e-01 -2.36867547e-01 -8.98741961e-01 3.28674436e-01
-1.84866786e+00 -4.39450920e-01 3.69084664e-02 2.39502788e+00
8.26071203e-01 3.03670913e-01 9.16751698e-02 -2.90714055e-01
7.93478966e-01 6.21005856e-02 -8.95863175e-01 3.01340491e-01
4.93524224e-02 -1.58669695e-01 5.32576740e-01 1.13589965e-01
-1.25390041e+00 8.25693071e-01 4.57743025e+00 8.37428212e-01
-1.14341426e+00 -1.58616174e-02 7.05956578e-01 1.41501233e-01
-1.98130250e-01 -1.82677045e-01 -8.20419490e-01 5.94307661e-01
2.71247894e-01 7.94307142e-02 2.04708576e-01 1.02226341e+00
1.44802481e-01 2.26192459e-01 -1.34440386e+00 1.15591216e+00
-1.32904984e-02 -1.00773251e+00 6.08886071e-02 -3.95357609e-02
8.55680466e-01 -4.36476246e-02 1.77593648e-01 3.16399336e-01
1.17900841e-01 -7.48657823e-01 8.99764240e-01 3.65600616e-01
1.04369593e+00 -5.02621770e-01 5.68555892e-01 5.06400168e-01
-1.09224486e+00 5.70505634e-02 -4.69850123e-01 2.67169744e-01
1.12526417e-02 4.15705442e-01 -7.04597116e-01 6.03476703e-01
6.47685349e-01 8.79092276e-01 -5.00985324e-01 6.90540254e-01
-2.41616264e-01 3.34362924e-01 -4.32479084e-01 3.08283478e-01
-4.36142683e-02 -2.05589384e-01 5.97533464e-01 8.35469723e-01
2.90609062e-01 -4.19861861e-02 2.60435164e-01 1.14167690e+00
-1.13408715e-01 -9.12398547e-02 -4.40602183e-01 2.46602073e-01
4.49808449e-01 1.01494908e+00 -5.24824142e-01 -7.37288222e-02
-2.55062580e-01 9.45672214e-01 3.68232280e-01 1.98065087e-01
-1.13123977e+00 -3.38377208e-02 6.32680237e-01 3.13144684e-01
3.64693940e-01 -2.81577319e-01 -3.75530511e-01 -1.46180177e+00
3.80782038e-01 -7.77343035e-01 1.82594165e-01 -7.17832923e-01
-1.62307549e+00 4.38996851e-01 2.26582721e-01 -1.77586031e+00
-1.47146747e-01 -8.41682434e-01 -4.00418431e-01 6.75329566e-01
-1.28020632e+00 -1.44555545e+00 -6.02005780e-01 4.96798813e-01
5.79538822e-01 -6.60854504e-02 6.17327631e-01 2.37070844e-01
-5.19929767e-01 8.30694437e-01 1.98356390e-01 1.45083085e-01
9.61282492e-01 -1.08481681e+00 2.70492017e-01 5.20135105e-01
-5.13659455e-02 3.31080258e-01 6.18430853e-01 -5.41601658e-01
-1.38182521e+00 -1.42327058e+00 1.31668210e-01 -5.55332303e-01
6.03706896e-01 -7.33434975e-01 -9.28694248e-01 6.92281127e-01
-4.99951810e-01 5.64132273e-01 1.08998746e-01 -2.41204903e-01
-3.72990519e-01 -3.27497303e-01 -1.19289052e+00 4.62969899e-01
1.34898424e+00 -3.34498525e-01 -5.30967653e-01 5.16531587e-01
9.21696603e-01 -8.71896505e-01 -1.12521231e+00 7.65767932e-01
5.40076792e-01 -7.52607048e-01 1.08187985e+00 -3.27267975e-01
6.51639819e-01 -4.83166128e-01 -1.01884395e-01 -1.25700188e+00
-1.76093698e-01 -3.16532075e-01 -7.23952949e-02 1.33934212e+00
2.36133680e-01 -6.74018860e-01 8.84814203e-01 4.79503423e-01
-2.89823830e-01 -9.17213500e-01 -8.59766483e-01 -9.67415988e-01
1.41340852e-01 -3.37803513e-01 5.11002660e-01 8.72386158e-01
-6.88631117e-01 3.08888793e-01 -1.85831755e-01 3.97590101e-01
7.53391862e-01 3.41658145e-01 1.19643962e+00 -1.13846910e+00
-4.45141941e-01 -1.92233652e-01 -5.61721206e-01 -1.44156802e+00
3.22376877e-01 -7.69337714e-01 7.08638132e-02 -9.79757726e-01
2.74824351e-01 -7.87978888e-01 6.17597029e-02 3.62488806e-01
-1.23919537e-02 3.44603181e-01 1.09221779e-01 5.05704224e-01
-4.33153212e-01 9.55748260e-01 1.69527280e+00 -8.05365816e-02
-1.87015072e-01 2.52117127e-01 -3.12435061e-01 8.15054357e-01
5.87013185e-01 -4.39101458e-01 -4.58521426e-01 -5.04608691e-01
-7.20241740e-02 -1.14557400e-01 6.22883737e-01 -9.33389425e-01
-7.38249347e-02 -3.34749520e-01 4.61052239e-01 -5.51583230e-01
2.88328946e-01 -9.69749451e-01 3.07659619e-02 2.18573213e-01
-2.23147824e-01 -2.62160987e-01 1.91952452e-01 6.53129756e-01
-1.50915697e-01 -9.49715171e-03 9.82290328e-01 2.66914479e-02
-6.02847278e-01 7.38456130e-01 3.21149915e-01 2.25569457e-01
1.17019808e+00 -2.43163705e-01 -1.19400984e-02 -8.69396329e-03
-6.98233187e-01 3.84002030e-01 7.82627702e-01 5.43010235e-01
4.11157399e-01 -1.49335837e+00 -7.00971425e-01 2.99750954e-01
5.28571010e-01 9.23930824e-01 2.15210274e-01 8.47373068e-01
-3.60809892e-01 3.14137563e-02 -1.87026232e-01 -1.11191988e+00
-9.58342075e-01 6.60265088e-01 3.88924867e-01 -2.09789559e-01
-6.51438773e-01 7.68214941e-01 7.13200033e-01 -9.05758381e-01
2.69728750e-01 -4.87787127e-01 1.16283529e-01 -3.26490581e-01
2.69479994e-02 7.75824562e-02 -8.14708043e-03 -6.32556379e-01
-3.06216002e-01 8.73434544e-01 -3.66840474e-02 2.25157425e-01
1.47938132e+00 -1.39256582e-01 1.94927633e-01 4.14641708e-01
1.33743846e+00 -1.74974903e-01 -1.89246655e+00 -3.42470676e-01
-2.05685496e-01 -5.49546897e-01 -3.87258679e-01 -4.97001380e-01
-1.03401220e+00 5.90875268e-01 7.09348679e-01 -1.98471054e-01
9.79839623e-01 3.23291481e-01 6.15791857e-01 3.02917689e-01
2.54403800e-01 -9.49859798e-01 3.46201748e-01 4.37846035e-01
1.15393436e+00 -1.54496717e+00 1.06557021e-02 -6.37154877e-01
-6.79384589e-01 1.00975001e+00 1.07366681e+00 -3.67021441e-01
4.07270223e-01 1.88773781e-01 -1.68147281e-01 -1.39050320e-01
-3.21595460e-01 1.23655736e-01 4.38255847e-01 5.84534526e-01
1.51820064e-01 -2.61114910e-02 -4.47918512e-02 6.78681612e-01
-7.26687089e-02 -8.69280398e-02 -4.25569760e-03 6.79614365e-01
4.26898263e-02 -9.81902540e-01 -4.19886231e-01 2.29862690e-01
-6.15556315e-02 4.06013221e-01 -2.62864411e-01 9.47119892e-01
3.40584815e-01 2.35792145e-01 1.75146714e-01 -3.72864127e-01
6.09799325e-01 -3.22459459e-01 6.05667830e-01 -7.07030177e-01
-1.08044967e-01 3.88103016e-02 -1.76141798e-01 -4.83753502e-01
-5.00837088e-01 -6.85187221e-01 -1.27685380e+00 4.23261225e-02
-3.27210873e-01 -2.85023957e-01 4.17511553e-01 9.06207561e-01
3.08793902e-01 2.29300573e-01 8.77636790e-01 -1.10408604e+00
-1.02302277e+00 -9.89718616e-01 -5.37404537e-01 9.07512426e-01
1.58203140e-01 -1.05686021e+00 -3.76185894e-01 1.84147403e-01]
|
[7.689124584197998, -2.7616403102874756]
|
e902d651-b4c2-4afc-9723-b3c1c174749e
|
action-guidance-getting-the-best-of-sparse-1
|
2010.03956
| null |
https://arxiv.org/abs/2010.03956v1
|
https://arxiv.org/pdf/2010.03956v1.pdf
|
Action Guidance: Getting the Best of Sparse Rewards and Shaped Rewards for Real-time Strategy Games
|
Training agents using Reinforcement Learning in games with sparse rewards is a challenging problem, since large amounts of exploration are required to retrieve even the first reward. To tackle this problem, a common approach is to use reward shaping to help exploration. However, an important drawback of reward shaping is that agents sometimes learn to optimize the shaped reward instead of the true objective. In this paper, we present a novel technique that we call action guidance that successfully trains agents to eventually optimize the true objective in games with sparse rewards while maintaining most of the sample efficiency that comes with reward shaping. We evaluate our approach in a simplified real-time strategy (RTS) game simulator called $\mu$RTS.
|
['Santiago Ontañón', 'Shengyi Huang']
|
2020-10-05
|
action-guidance-getting-the-best-of-sparse
|
https://openreview.net/forum?id=1OQ90khuUGZ
|
https://openreview.net/pdf?id=1OQ90khuUGZ
| null |
['real-time-strategy-games']
|
['playing-games']
|
[-1.62313908e-01 1.74170583e-01 -2.27273442e-02 2.32744917e-01
-6.31180644e-01 -5.68185806e-01 2.69605130e-01 2.25821972e-01
-1.00768268e+00 1.35055161e+00 -3.30934703e-01 -3.18914652e-01
-3.09432775e-01 -9.00461674e-01 -6.01656497e-01 -8.27556610e-01
-3.39318931e-01 5.52400589e-01 2.87253827e-01 -6.15950525e-01
5.78335881e-01 2.94978857e-01 -1.60089672e+00 -3.31159055e-01
1.07888973e+00 6.96725786e-01 5.63030720e-01 7.83444822e-01
-1.06906414e-01 1.00718474e+00 -1.03886819e+00 5.09295799e-02
5.52556217e-01 -6.82674110e-01 -6.72245979e-01 -8.19698796e-02
-5.69095254e-01 -5.92395782e-01 -1.13162994e-02 1.00345004e+00
5.34138560e-01 5.88875890e-01 2.00805932e-01 -1.04387617e+00
2.81560749e-01 7.50371337e-01 -4.85461593e-01 2.54947960e-01
3.41279805e-01 3.87925059e-01 7.94437170e-01 -1.14681013e-01
5.14325857e-01 9.54369843e-01 3.96289770e-03 6.90549314e-01
-1.17184341e+00 -6.26598239e-01 1.66324690e-01 2.90761869e-02
-1.09262335e+00 -7.90372640e-02 4.36219335e-01 4.80456613e-02
1.07371068e+00 4.61564437e-02 9.24893737e-01 5.93233049e-01
5.00576496e-02 8.39008927e-01 1.42168677e+00 -2.72054970e-01
8.23825240e-01 8.81004855e-02 -4.50941890e-01 4.65788960e-01
1.30273938e-01 6.98227704e-01 -1.60847872e-01 -2.01795056e-01
8.94916058e-01 -1.72178820e-01 3.37362289e-02 -5.57038248e-01
-7.15787947e-01 1.08378172e+00 3.49956214e-01 1.19148336e-01
-6.29947662e-01 5.09089351e-01 9.55972746e-02 6.20537877e-01
-4.48171012e-02 9.96895790e-01 -2.14517996e-01 -9.97155190e-01
-6.83664382e-01 7.20758259e-01 8.76068532e-01 5.43935239e-01
8.25564444e-01 6.09225750e-01 1.15634613e-01 6.96698368e-01
1.09563164e-01 4.63460386e-01 3.97528529e-01 -1.14572406e+00
3.70245069e-01 4.95063215e-01 4.88090843e-01 -3.42742652e-01
-3.19543988e-01 -6.08472586e-01 -2.53915340e-01 1.11580467e+00
5.43865204e-01 -7.33421743e-01 -5.93269944e-01 1.63243234e+00
3.42275500e-01 -6.41782135e-02 3.35135758e-01 8.47639978e-01
2.05117583e-01 4.55657959e-01 -1.02558650e-01 -2.95072049e-01
7.19809532e-01 -7.32834280e-01 -3.38611901e-01 -3.14692378e-01
5.62621534e-01 -2.53045171e-01 9.13683951e-01 5.35684049e-01
-1.38315439e+00 -1.09037653e-01 -9.34939623e-01 7.23357499e-01
-3.34944203e-02 -2.77426273e-01 6.90145791e-01 6.78565383e-01
-9.64137375e-01 1.00414646e+00 -7.54417241e-01 -4.00525741e-02
1.87013164e-01 8.38988483e-01 1.17856272e-01 2.51768529e-01
-8.92597198e-01 8.61217558e-01 7.12882459e-01 -3.05856645e-01
-1.21415555e+00 -3.59342813e-01 -5.79454601e-01 3.11004907e-01
1.05917811e+00 -4.79459018e-02 1.57308877e+00 -9.02224660e-01
-2.22328043e+00 8.36522728e-02 3.69928837e-01 -6.86230004e-01
5.05216360e-01 -1.25980645e-01 3.63652557e-01 -2.64038723e-02
-7.10349753e-02 5.84872603e-01 8.10179591e-01 -1.28171599e+00
-7.79083908e-01 9.48445126e-02 6.36337399e-01 5.88118911e-01
4.20528464e-02 -1.87633723e-01 1.17470078e-01 -2.98730612e-01
-4.67954636e-01 -9.46679354e-01 -8.18022728e-01 -7.02105165e-01
-4.55332063e-02 -1.99121714e-01 1.90959185e-01 -5.81387915e-02
9.90986943e-01 -1.89097166e+00 3.28921735e-01 5.89280427e-01
1.49631649e-01 3.82785797e-01 -2.71015882e-01 6.89615667e-01
1.86412841e-01 -2.45855212e-01 -1.69482678e-01 -3.95072810e-02
-5.08375578e-02 4.91332620e-01 -3.92966360e-01 9.29668099e-02
6.08190708e-02 7.93966830e-01 -1.27963507e+00 -9.65544283e-02
1.27160370e-01 -6.78172410e-02 -8.03153694e-01 3.81234437e-01
-4.24721897e-01 5.78811228e-01 -7.45808661e-01 3.90276283e-01
2.89991945e-01 4.91239391e-02 3.78427446e-01 8.51117671e-01
-4.52448756e-01 2.16394827e-01 -1.55800462e+00 1.30039132e+00
-2.84083188e-01 1.49391085e-01 2.58289963e-01 -1.08436811e+00
1.02090919e+00 4.41218307e-03 8.49194586e-01 -1.00524747e+00
4.40614194e-01 3.40558350e-01 3.93156111e-01 -1.12334482e-01
8.67762029e-01 -1.71494097e-01 -7.97856897e-02 8.35702658e-01
-2.34025717e-01 -4.35476720e-01 5.77903628e-01 3.16491649e-02
1.27834702e+00 2.22462609e-01 4.59218770e-01 -8.09071139e-02
2.47878745e-01 2.03563660e-01 4.84178692e-01 1.02748907e+00
5.37179178e-03 2.84864511e-02 1.03039885e+00 -1.73463970e-01
-9.55617428e-01 -8.17784548e-01 5.44044912e-01 1.03574359e+00
4.63876985e-02 -2.70475119e-01 -4.78499800e-01 -7.41724730e-01
-1.40934275e-03 7.63706326e-01 -5.33183396e-01 -1.43895417e-01
-7.55566955e-01 -5.79425335e-01 2.27379501e-01 2.56090164e-01
3.85086060e-01 -1.40946972e+00 -1.46775198e+00 5.45781493e-01
2.18870401e-01 -4.95904803e-01 -2.23160341e-01 5.28754890e-01
-9.33059335e-01 -1.10346007e+00 -9.12505329e-01 -3.28337848e-01
5.75089514e-01 1.49691984e-01 1.02641571e+00 3.76719475e-01
-2.00517058e-01 3.60904485e-01 -5.60138643e-01 -3.21254343e-01
-4.30150717e-01 2.45177094e-02 -1.08088419e-01 -4.80866939e-01
-1.13344111e-01 -5.10613263e-01 -3.97242665e-01 2.44847760e-01
-8.71984243e-01 -3.19865495e-01 6.11329198e-01 9.89090979e-01
5.02017140e-01 2.60685593e-01 7.21079588e-01 -5.03360629e-01
1.18747890e+00 -3.41901511e-01 -1.20633864e+00 -8.86418372e-02
-4.66870815e-01 5.32615542e-01 7.99540162e-01 -6.80517614e-01
-6.39526784e-01 4.60838377e-02 -1.94274709e-01 -2.67877191e-01
1.90973118e-01 5.18943965e-01 4.58891153e-01 -2.73201257e-01
6.86126649e-01 3.52063924e-01 2.96002209e-01 -2.17034310e-01
3.11400741e-01 2.36889809e-01 -5.43724112e-02 -7.43589401e-01
6.23514771e-01 1.08337611e-01 8.60742405e-02 -5.83980203e-01
-2.31148183e-01 -1.90521240e-01 1.41169786e-01 -3.04484457e-01
2.36564219e-01 -4.15470749e-01 -1.52531576e+00 1.06785253e-01
-5.76948285e-01 -9.33271289e-01 -1.01590395e+00 5.44417322e-01
-9.64573324e-01 1.49920598e-01 -1.92467377e-01 -1.36248112e+00
-9.55119282e-02 -1.31943011e+00 5.42323470e-01 6.54102981e-01
1.24644451e-01 -5.94941556e-01 4.83354837e-01 -4.40495074e-01
6.41829729e-01 1.17512167e-01 4.91097897e-01 -3.42509151e-01
-7.40118682e-01 1.13544939e-02 1.52872533e-01 -1.93872042e-02
-4.68201004e-02 -3.46933186e-01 -2.22505435e-01 -6.07740998e-01
-6.50713965e-02 -7.30583251e-01 4.98032004e-01 4.21892494e-01
7.44654715e-01 -1.26210630e-01 8.14863667e-02 2.58217245e-01
1.37030339e+00 7.42836058e-01 7.27883279e-01 6.07737958e-01
-2.17806250e-02 3.48345697e-01 1.02606785e+00 1.08418429e+00
1.18721485e-01 7.06461012e-01 7.90174246e-01 2.67347127e-01
4.04113561e-01 -2.29376093e-01 5.32712698e-01 1.39679536e-01
-4.58469868e-01 -4.90277335e-02 -5.54849148e-01 3.52991045e-01
-2.00297236e+00 -1.05901217e+00 5.04837155e-01 2.59501529e+00
8.85909617e-01 3.38722438e-01 7.19369590e-01 1.13962285e-01
1.01310857e-01 -2.42389143e-01 -7.65971184e-01 -6.94226563e-01
2.59620368e-01 5.56463182e-01 6.25370026e-01 7.67481267e-01
-5.68118572e-01 1.04960823e+00 6.72135019e+00 7.86552489e-01
-1.07355821e+00 -2.54765064e-01 1.86386824e-01 -5.13638854e-01
-3.11509985e-02 1.08064391e-01 -7.09709466e-01 4.12024885e-01
8.09080005e-01 -4.06120300e-01 1.08431971e+00 1.12407339e+00
3.31699431e-01 -8.03357303e-01 -6.11119986e-01 6.73794270e-01
-3.96519572e-01 -1.03364766e+00 -5.19768119e-01 4.30248618e-01
5.37633836e-01 -1.88817874e-01 7.84489736e-02 8.23048472e-01
1.10334635e+00 -1.16412604e+00 4.96130943e-01 2.68340856e-01
5.52439809e-01 -1.15696692e+00 7.10108817e-01 6.51421487e-01
-1.02350342e+00 -3.65856141e-01 -3.95141751e-01 -4.45675701e-01
3.52816135e-02 -4.78154942e-02 -1.25825739e+00 2.87113339e-01
3.56023580e-01 2.98281372e-01 -2.58984536e-01 1.49507356e+00
-3.40597957e-01 3.03245246e-01 -4.64710325e-01 -6.34486318e-01
6.30915642e-01 -4.35037553e-01 5.84696054e-01 4.39730376e-01
4.96295691e-01 1.67574391e-01 3.92800093e-01 8.31801951e-01
3.74875963e-01 1.28562069e-02 -5.93910694e-01 -2.02445343e-01
1.91809133e-01 1.04855144e+00 -9.33201909e-01 -1.56649083e-01
1.67482898e-01 5.93179405e-01 6.05362356e-01 2.33314991e-01
-7.81300366e-01 -3.66812885e-01 6.18696928e-01 -2.07423329e-01
4.68804240e-01 -3.69875342e-01 1.39505789e-01 -7.17077971e-01
-3.54374975e-01 -1.01180696e+00 2.45755628e-01 -2.87990570e-01
-5.87192297e-01 5.15899003e-01 1.40923876e-02 -1.37465847e+00
-9.48425710e-01 -2.60520190e-01 -4.99205709e-01 7.80154943e-01
-1.53074706e+00 -1.17416896e-01 -2.56923176e-02 5.52437246e-01
5.49855471e-01 -3.42130959e-01 6.06482446e-01 -2.28315592e-02
-3.87684554e-01 4.57438916e-01 2.34266356e-01 -3.51547688e-01
2.00964287e-01 -1.61737740e+00 -9.33542252e-02 3.99667442e-01
-1.71638235e-01 3.14752191e-01 9.81688440e-01 -7.21133828e-01
-1.41429901e+00 -3.80118370e-01 1.08056590e-02 1.24516688e-01
5.98455071e-01 5.63882999e-02 -5.21865785e-01 1.93887055e-01
2.69898146e-01 -2.92476237e-01 3.08964640e-01 2.44825445e-02
2.95919001e-01 1.82783417e-02 -1.02442682e+00 8.09535205e-01
4.86629575e-01 7.47901574e-02 -2.81856209e-01 -1.72959343e-01
2.65380591e-01 -7.12228119e-01 -4.41421360e-01 5.67671238e-03
4.41779017e-01 -9.16041791e-01 7.94430435e-01 -5.64883709e-01
1.81969136e-01 -3.56437951e-01 2.15168353e-02 -1.94654882e+00
3.96740474e-02 -1.00199127e+00 -1.65482596e-01 5.51909268e-01
2.20117673e-01 -6.45024121e-01 1.23600876e+00 4.13827181e-01
8.71938616e-02 -7.71288991e-01 -1.07414770e+00 -9.85634804e-01
1.82660386e-01 -2.87587404e-01 5.34204066e-01 3.45192432e-01
3.72195601e-01 5.49923703e-02 -6.64353490e-01 -2.45948568e-01
6.48940742e-01 6.45873472e-02 9.27136064e-01 -7.59156525e-01
-8.22892308e-01 -6.03863597e-01 -4.75128368e-02 -1.08191085e+00
-1.92562312e-01 -3.54688227e-01 3.58904272e-01 -1.32362938e+00
-9.14568156e-02 -7.77875781e-01 -1.79002270e-01 3.33171368e-01
-7.84214884e-02 -2.87232816e-01 5.64678729e-01 -1.51107848e-01
-8.61438751e-01 9.01827276e-01 1.47066569e+00 1.16585724e-01
-8.17359090e-01 3.42198938e-01 -5.24576664e-01 3.87386918e-01
1.08428335e+00 -6.10182941e-01 -6.04149044e-01 -3.63457836e-02
5.25479376e-01 6.91975057e-01 1.12031206e-01 -1.06398249e+00
1.12281241e-01 -6.52895868e-01 -1.43049449e-01 -5.28259933e-01
4.54374731e-01 -6.90225840e-01 -4.86264266e-02 8.71393383e-01
-2.90053427e-01 1.41837984e-01 2.56561846e-01 4.53531265e-01
-6.22126795e-02 -7.22897708e-01 6.06520832e-01 -3.27572316e-01
-4.67730939e-01 -2.61479001e-02 -7.46730864e-01 1.48356751e-01
1.25516975e+00 -9.51271206e-02 -1.51968390e-01 -7.84157038e-01
-6.03576422e-01 5.41504800e-01 3.91022444e-01 4.11048681e-02
7.54141033e-01 -9.59800541e-01 -5.01284063e-01 -4.25413996e-02
-3.66300285e-01 -2.23819360e-01 -1.62998047e-02 7.23097384e-01
-3.60609502e-01 2.45846495e-01 -4.68978673e-01 -1.69381499e-01
-1.05349660e+00 4.47988600e-01 5.42057276e-01 -8.03115427e-01
-5.52983403e-01 5.66831172e-01 -2.58110434e-01 -1.98625743e-01
2.33887672e-01 -1.76226512e-01 -4.53465819e-01 -6.35682419e-02
6.43444419e-01 4.51252699e-01 -2.91497678e-01 3.73844355e-02
-2.79653631e-02 1.33694783e-01 -1.81556437e-02 -5.68948448e-01
1.34861779e+00 1.33474931e-01 2.85998374e-01 5.63289560e-02
4.15258497e-01 3.02747525e-02 -1.72257018e+00 -3.87851335e-03
-5.52433729e-02 -7.59727597e-01 -1.73260227e-01 -7.58820713e-01
-1.03962314e+00 5.68436682e-01 6.53590024e-01 6.30039930e-01
9.12850916e-01 -3.95187259e-01 4.90445882e-01 7.82912314e-01
7.96571851e-01 -1.43139291e+00 4.03059334e-01 7.64570296e-01
6.91403747e-01 -1.05893493e+00 -1.39688388e-01 1.29828155e-01
-1.02464569e+00 1.03538346e+00 7.57155240e-01 -5.69357574e-01
1.73759148e-01 3.24364275e-01 -2.61850238e-01 -6.23190999e-02
-6.92530632e-01 -8.92608285e-01 -5.10100007e-01 5.43457747e-01
-1.01380460e-02 -5.73323574e-03 -3.77924621e-01 1.87931821e-01
-2.63056874e-01 6.99548647e-02 8.11734200e-01 1.32041216e+00
-7.84830689e-01 -1.53930581e+00 -4.79156941e-01 5.08503079e-01
-4.32969362e-01 1.32177860e-01 -1.35096699e-01 7.29723394e-01
-3.40312153e-01 1.12797689e+00 -8.64583105e-02 -2.64587909e-01
3.79365802e-01 -3.91274601e-01 6.10033333e-01 -6.79070592e-01
-8.66350174e-01 1.67409718e-01 1.12664357e-01 -7.99903572e-01
-1.76658183e-01 -6.46249712e-01 -1.58110142e+00 -2.96505421e-01
-1.14333525e-01 6.67814136e-01 6.02692544e-01 9.76308107e-01
7.26177916e-02 6.82475388e-01 8.60777020e-01 -7.79991210e-01
-9.14601982e-01 -6.19320810e-01 -8.20856869e-01 -3.52202021e-02
2.94342548e-01 -9.69463944e-01 -8.23161379e-02 -8.76752615e-01]
|
[3.8394851684570312, 1.7551366090774536]
|
2671502b-e323-446b-bd2d-6dd388c74653
|
enhancing-few-shot-ner-with-prompt-ordering
|
2305.11791
| null |
https://arxiv.org/abs/2305.11791v1
|
https://arxiv.org/pdf/2305.11791v1.pdf
|
Enhancing Few-shot NER with Prompt Ordering based Data Augmentation
|
Recently, data augmentation (DA) methods have been proven to be effective for pre-trained language models (PLMs) in low-resource settings, including few-shot named entity recognition (NER). However, conventional NER DA methods are mostly aimed at sequence labeling models, i.e., token-level classification, and few are compatible with unified autoregressive generation frameworks, which can handle a wider range of NER tasks, such as nested NER. Furthermore, these generation frameworks have a strong assumption that the entities will appear in the target sequence with the same left-to-right order as the source sequence. In this paper, we claim that there is no need to keep this strict order, and more diversified but reasonable target entity sequences can be provided during the training stage as a novel DA method. Nevertheless, a naive mixture of augmented data can confuse the model since one source sequence will then be paired with different target sequences. Therefore, we propose a simple but effective Prompt Ordering based Data Augmentation (PODA) method to improve the training of unified autoregressive generation frameworks under few-shot NER scenarios. Experimental results on three public NER datasets and further analyses demonstrate the effectiveness of our approach.
|
['Lidong Bing', 'De Wen Soh', 'Wenxuan Zhang', 'Liying Cheng', 'Huiming Wang']
|
2023-05-19
| null | null | null | null |
['few-shot-ner', 'named-entity-recognition-ner']
|
['natural-language-processing', 'natural-language-processing']
|
[ 1.63535714e-01 -1.13159701e-01 -7.28665814e-02 -4.12321866e-01
-4.79829431e-01 -4.80047882e-01 5.40084600e-01 8.60167518e-02
-8.85766923e-01 8.06212425e-01 3.82075906e-01 -3.75221312e-01
3.10482949e-01 -9.61885691e-01 -4.37233537e-01 -5.96363544e-01
4.21992898e-01 4.45221126e-01 1.67074248e-01 -4.64054018e-01
-2.40029082e-01 3.07607263e-01 -1.26512182e+00 -2.87445001e-02
1.15749323e+00 3.98489624e-01 2.71768838e-01 4.15834218e-01
-6.33835852e-01 8.44283044e-01 -6.48253620e-01 -6.54759824e-01
5.31515386e-03 -4.17537779e-01 -7.52543867e-01 -1.89486220e-01
-1.26935095e-01 -2.59412616e-01 -5.15500784e-01 8.78802896e-01
7.21707642e-01 6.47633076e-01 5.04176617e-01 -1.07155585e+00
-7.46313274e-01 9.94253337e-01 -2.31383979e-01 1.16522878e-01
2.79598776e-03 -3.32831629e-02 1.10259855e+00 -1.17536438e+00
5.67517519e-01 1.02910399e+00 6.77184284e-01 8.80923688e-01
-8.84315670e-01 -5.90167761e-01 2.74368733e-01 1.32990167e-01
-1.18178380e+00 -4.34012681e-01 6.34005964e-01 -1.51142597e-01
7.90001333e-01 2.55376250e-01 1.81129709e-01 1.32655907e+00
-3.81581217e-01 7.97843874e-01 8.44102740e-01 -5.73174953e-01
2.23089516e-01 6.73255846e-02 4.98113275e-01 2.65364885e-01
2.59662032e-01 -3.04377109e-01 -1.27118856e-01 -1.34724379e-01
5.46363115e-01 7.49219134e-02 -1.29173160e-01 -2.80169956e-02
-1.20391583e+00 9.65180874e-01 1.95981801e-01 7.95827925e-01
-6.81622744e-01 -3.22409898e-01 6.61982656e-01 -1.82495564e-01
4.54175591e-01 3.76059353e-01 -6.28423631e-01 -8.38860422e-02
-7.04139531e-01 -2.84946207e-02 6.23137534e-01 9.77717400e-01
6.17885470e-01 3.15327108e-01 -5.29084504e-01 1.12144661e+00
8.64754692e-02 1.74368232e-01 9.29286361e-01 -2.17080310e-01
4.99715954e-01 4.87218231e-01 2.33822033e-01 -6.04790568e-01
-3.80801857e-01 -3.26072812e-01 -1.17144620e+00 -3.97715807e-01
2.60944575e-01 -6.15212739e-01 -1.11521745e+00 1.90387750e+00
3.91535789e-01 4.10647988e-01 4.11181957e-01 5.77543259e-01
7.73906827e-01 9.97396350e-01 5.73150754e-01 -3.72864097e-01
1.43452764e+00 -1.00613558e+00 -1.07824516e+00 -3.03284675e-01
9.70403731e-01 -5.56106925e-01 1.19574094e+00 -1.40190005e-01
-6.93581641e-01 -6.31088912e-01 -7.95252144e-01 -7.58769661e-02
-5.14586329e-01 2.41534695e-01 5.86214542e-01 8.29315782e-01
-5.55509269e-01 4.52354997e-01 -7.26083338e-01 -2.44019538e-01
2.87227333e-02 -1.46334529e-01 -2.82438546e-01 -2.22366259e-01
-1.81448889e+00 9.02265966e-01 9.12431777e-01 2.81179518e-01
-6.18206561e-01 -6.21393383e-01 -1.06831801e+00 2.01208994e-01
5.07601142e-01 -4.36093777e-01 1.24225843e+00 -4.75124925e-01
-1.27250934e+00 4.14960891e-01 -3.74123931e-01 -5.91065764e-01
2.90202618e-01 -3.98490399e-01 -7.37756252e-01 -3.21126193e-01
1.03921063e-01 4.32972252e-01 3.47117364e-01 -1.13007319e+00
-5.60810328e-01 -1.02136359e-01 -6.09061914e-03 1.18303210e-01
-7.44301915e-01 3.04650605e-01 -1.97477579e-01 -8.09996009e-01
-1.64726377e-01 -7.56709814e-01 -7.91909814e-01 -7.78840363e-01
-4.64463949e-01 -4.47337687e-01 4.90576565e-01 -7.47528434e-01
1.61566007e+00 -2.13580370e+00 -1.84866160e-01 -2.22665258e-02
-1.99561253e-01 7.73662210e-01 -3.38180214e-01 4.72926676e-01
-2.02565759e-01 5.27140617e-01 -2.80988365e-01 -4.20819968e-01
8.29910412e-02 3.17558646e-01 -4.56166267e-01 -8.81988406e-02
3.74231189e-01 8.72370958e-01 -1.08511710e+00 -4.72918510e-01
5.40629737e-02 4.51191545e-01 -2.04371139e-01 4.44900036e-01
-5.95578030e-02 3.32125872e-01 -3.22496653e-01 1.91081181e-01
5.82529545e-01 -4.36860286e-02 2.11445674e-01 -1.24304824e-01
-2.64690936e-01 5.20908654e-01 -1.29502523e+00 1.31352973e+00
-5.60612977e-01 1.07412569e-01 -4.82806742e-01 -7.84322858e-01
1.09366941e+00 6.64698422e-01 2.49792993e-01 -3.45922649e-01
1.51074022e-01 1.30532995e-01 1.61679387e-01 -2.33122960e-01
8.59878540e-01 -3.51143777e-01 -2.49971122e-01 1.81894779e-01
2.20842466e-01 4.02501971e-01 6.24462545e-01 6.07711934e-02
9.07634616e-01 5.74831925e-02 5.23829281e-01 2.79099822e-01
4.67490256e-01 -6.07844666e-02 1.06350052e+00 9.98726606e-01
-1.30341843e-01 4.49412376e-01 1.70483708e-01 -1.96273625e-01
-1.30947852e+00 -6.41295373e-01 -2.80814059e-02 1.29459500e+00
-1.01888955e-01 -5.09854436e-01 -6.80388153e-01 -8.09960723e-01
-7.92354286e-01 1.14426529e+00 -2.86245495e-01 -5.40451556e-02
-6.63136899e-01 -9.77500141e-01 9.04125631e-01 7.36658394e-01
4.03732687e-01 -1.30710661e+00 -2.35131934e-01 6.76013172e-01
-4.38543975e-01 -1.36484146e+00 -5.79021871e-01 3.32123131e-01
-5.95865488e-01 -5.67898989e-01 -8.15728724e-01 -8.40032756e-01
5.58447957e-01 2.92216182e-01 8.34256530e-01 -5.79174757e-02
2.25759000e-01 -1.57251097e-02 -8.21785152e-01 -3.86431515e-01
-6.58896208e-01 1.89082757e-01 2.65224159e-01 3.28840502e-02
4.85569388e-01 -3.26828390e-01 -1.17199883e-01 2.27158368e-01
-1.20110667e+00 -1.31610306e-02 6.49308741e-01 1.13576055e+00
4.66808170e-01 1.11128919e-01 8.09832633e-01 -1.09332132e+00
6.44509852e-01 -5.02152205e-01 -1.62186086e-01 5.61506152e-01
-3.53201896e-01 1.41220465e-01 1.06840575e+00 -7.58768499e-01
-1.39978933e+00 6.52154461e-02 -5.31453550e-01 -4.49549198e-01
-4.39624250e-01 8.47759485e-01 -5.60391128e-01 5.30241132e-01
6.40361547e-01 4.92169678e-01 -5.85439920e-01 -6.27681851e-01
6.77098274e-01 6.29755676e-01 5.76348066e-01 -4.61412460e-01
8.67514491e-01 -2.42026094e-02 -3.78700376e-01 -8.21522772e-01
-8.32267702e-01 -6.12514734e-01 -7.64889538e-01 1.56021029e-01
7.36699700e-01 -9.47983801e-01 -1.06472202e-01 5.02782881e-01
-1.46880424e+00 -5.99327274e-02 -2.92703062e-01 5.89774668e-01
-5.70800416e-02 6.59943104e-01 -7.53396988e-01 -1.13781357e+00
-4.71059620e-01 -7.60070324e-01 6.13287687e-01 4.08280492e-01
-1.88943893e-01 -8.60504806e-01 1.33320361e-01 7.28456452e-02
2.86885649e-01 -1.30934194e-01 1.01841664e+00 -1.49894965e+00
3.08761075e-02 -2.67585009e-01 1.16342977e-01 4.40262944e-01
2.17381239e-01 -4.65216599e-02 -9.67631459e-01 1.02797989e-02
-1.26786325e-02 -8.78341943e-02 6.09467387e-01 -4.19314429e-02
9.02651787e-01 -4.67357516e-01 -5.35887219e-02 1.14330783e-01
1.19666636e+00 3.72338742e-01 8.79006386e-01 2.30317488e-01
8.39471161e-01 5.20103455e-01 8.15026283e-01 5.22133172e-01
2.65880734e-01 5.10764718e-01 -1.66650640e-03 -1.77607834e-01
1.03561893e-01 -5.37340164e-01 4.18574750e-01 1.15633476e+00
1.29952403e-02 -6.00964785e-01 -1.00793159e+00 7.60867894e-01
-1.72188067e+00 -1.05270624e+00 -2.96746612e-01 2.23171687e+00
1.07494903e+00 1.42725095e-01 1.16939507e-02 1.12597503e-01
1.11335051e+00 1.63211018e-01 -3.34203035e-01 -1.76982731e-01
-2.28568405e-01 2.21538886e-01 3.38341504e-01 1.25097319e-01
-1.20575047e+00 1.14728940e+00 5.44885349e+00 1.11740017e+00
-8.50332618e-01 2.52991170e-01 5.26126623e-01 5.20822048e-01
-4.01273459e-01 7.68459961e-02 -1.23434961e+00 7.06361115e-01
1.30576611e+00 -3.46213251e-01 8.50818958e-03 9.93852377e-01
1.37662381e-01 4.39681441e-01 -7.47388184e-01 6.59127593e-01
-1.32333830e-01 -1.09299517e+00 3.21001381e-01 -1.28709450e-01
6.78985596e-01 -8.52606148e-02 -4.86150831e-01 8.57534766e-01
5.76160669e-01 -7.00711787e-01 2.69597650e-01 3.55317980e-01
6.24531686e-01 -8.13508570e-01 1.06039524e+00 7.48113155e-01
-1.22120261e+00 8.95829499e-02 -5.89601159e-01 1.79094464e-01
5.43245137e-01 7.18144596e-01 -9.81894433e-01 9.64089870e-01
3.23000878e-01 3.11277837e-01 -2.76538432e-01 1.11778879e+00
-4.50452834e-01 7.44128644e-01 -2.30842263e-01 -1.82165697e-01
1.94084585e-01 -1.72148034e-01 4.24097210e-01 1.30298388e+00
3.73538047e-01 2.82839686e-01 1.58450559e-01 4.63158399e-01
-3.05202454e-01 5.67771137e-01 -5.76688766e-01 -4.85318512e-01
8.06347549e-01 1.32977378e+00 -6.42903507e-01 -6.02391779e-01
-5.72841704e-01 9.86748576e-01 3.79492879e-01 2.35981151e-01
-8.52662623e-01 -6.59467459e-01 3.40306342e-01 -2.68969387e-01
3.48249823e-01 -2.90253937e-01 -8.62306356e-02 -1.44799674e+00
-1.00257665e-01 -8.01652372e-01 5.78408360e-01 -6.00251734e-01
-1.57628572e+00 8.67790639e-01 -3.01131189e-01 -1.35590971e+00
-2.79166490e-01 -3.00592840e-01 -7.26213872e-01 7.83733010e-01
-1.48916554e+00 -1.10076344e+00 -4.04698923e-02 3.50887597e-01
7.40242720e-01 1.16072819e-01 1.03023076e+00 5.12804925e-01
-8.94526839e-01 6.42882824e-01 1.34473816e-01 6.53210402e-01
6.74386799e-01 -1.07122326e+00 6.29813731e-01 1.42429829e+00
3.26297730e-01 8.50406528e-01 5.86305797e-01 -8.58517528e-01
-9.61145580e-01 -1.37239659e+00 1.34969461e+00 -1.92389503e-01
6.99486077e-01 -3.67057472e-01 -1.36645818e+00 7.42974997e-01
5.27278557e-02 -8.71959105e-02 9.01830614e-01 2.30163410e-01
-2.61592269e-01 2.88690865e-01 -7.94937313e-01 8.40502799e-01
8.84154320e-01 -4.29764152e-01 -8.77608538e-01 1.20500788e-01
1.07225740e+00 -1.89517558e-01 -8.40731025e-01 6.18017316e-01
8.89945775e-02 -3.61755878e-01 7.07173049e-01 -1.06158578e+00
2.54161268e-01 -3.89558166e-01 -6.40200451e-02 -1.42910111e+00
-3.15310001e-01 -3.99887592e-01 -1.98163241e-01 1.91236591e+00
4.84084576e-01 -3.69345397e-01 3.11166495e-01 7.16220319e-01
-3.27753514e-01 -4.18171912e-01 -7.84542322e-01 -1.12703133e+00
-4.62145098e-02 -4.27641571e-01 7.61712432e-01 1.14635742e+00
-7.29274973e-02 7.23915994e-01 -8.36424828e-01 3.04941595e-01
1.75724387e-01 -3.16029429e-01 6.20011270e-01 -1.24922967e+00
4.11502235e-02 -2.00291440e-01 -1.32269099e-01 -1.04469001e+00
2.79198587e-01 -6.54426575e-01 4.62647617e-01 -1.58308470e+00
9.86548960e-02 -5.44922829e-01 -5.58836639e-01 7.61859298e-01
-7.73176372e-01 -2.31998563e-01 1.09201282e-01 1.22563154e-01
-4.57400262e-01 9.61345017e-01 6.85513675e-01 1.58749104e-01
-3.56541663e-01 -3.97289768e-02 -6.04710102e-01 6.36593401e-01
7.38100171e-01 -5.84774375e-01 -2.39085734e-01 -1.67865098e-01
8.42025876e-02 -1.06586203e-01 -1.10462897e-01 -7.37277210e-01
3.00350636e-01 -1.82367757e-01 1.87220350e-01 -4.80870634e-01
1.10211819e-01 -7.73240268e-01 8.90158117e-02 2.32063919e-01
-4.04773951e-01 -1.17254682e-01 6.63931966e-02 4.76296544e-01
-1.75930500e-01 -6.37455702e-01 6.52461410e-01 -7.36192837e-02
-1.10213959e+00 4.61275995e-01 -2.66995966e-01 7.91459382e-02
8.53969455e-01 -5.06367460e-02 -2.02392697e-01 -1.93883628e-01
-6.31146789e-01 1.71332836e-01 1.64618939e-01 5.86178243e-01
4.02094573e-01 -1.45967340e+00 -8.48156273e-01 -7.38148764e-02
2.87325978e-01 1.93635803e-02 5.19939959e-01 5.95992446e-01
1.34312406e-01 3.14536303e-01 -9.95500851e-03 2.69999094e-02
-1.08538115e+00 7.39311993e-01 5.09858131e-03 -6.34718537e-01
-5.09848595e-01 5.40342093e-01 2.02686206e-01 -7.45035470e-01
1.40923643e-02 -6.93493709e-02 -5.72833776e-01 2.03269795e-01
7.71884620e-01 2.31023267e-01 1.56461880e-01 -7.52897024e-01
-1.85576156e-01 -1.34150833e-01 -3.19339067e-01 7.29984343e-02
1.35687959e+00 -1.09256096e-01 6.31389171e-02 6.62078500e-01
6.80071056e-01 1.40481845e-01 -8.77447307e-01 -5.15012145e-01
4.02356744e-01 -1.68822646e-01 -2.50728160e-01 -5.09332836e-01
-6.49908304e-01 8.24037671e-01 2.65810370e-01 2.60960877e-01
9.21466768e-01 -1.57750696e-01 1.03053355e+00 4.49273258e-01
2.75107026e-01 -1.00264692e+00 -3.35946590e-01 8.23209047e-01
4.04837310e-01 -1.01122487e+00 -5.12895823e-01 -4.39619571e-01
-1.02679598e+00 9.17118669e-01 8.50732327e-01 3.08207482e-01
1.52965665e-01 6.10431023e-02 -6.55128956e-02 2.94926941e-01
-6.31685019e-01 -5.61217487e-01 2.50831787e-02 4.87460643e-01
6.71246409e-01 -2.57713981e-02 -6.22480512e-01 1.05849493e+00
6.53863922e-02 -1.11436374e-01 6.59666836e-01 8.14637244e-01
-3.98862541e-01 -1.30493605e+00 -1.96363792e-01 2.61080325e-01
-5.50068855e-01 -5.00761867e-01 -1.43670365e-02 6.29292130e-01
1.86849236e-02 9.71604288e-01 -1.25953555e-01 -1.92873299e-01
4.65418786e-01 6.61676228e-01 -1.67384520e-01 -7.76192725e-01
-4.97790188e-01 6.95248619e-02 2.07774132e-01 1.35830939e-01
-2.65999943e-01 -4.32532072e-01 -1.44154418e+00 -1.61578953e-02
-5.85678220e-01 5.02877653e-01 5.35966992e-01 1.11234128e+00
3.40235084e-01 5.00555515e-01 7.58746922e-01 -4.24293399e-01
-6.53933287e-01 -1.21527088e+00 -5.01530528e-01 5.66986740e-01
-3.63743715e-02 -4.93072391e-01 -1.76789686e-01 2.56514270e-02]
|
[9.752989768981934, 9.443215370178223]
|
4ae958c1-36c6-4c5b-8e0c-8046400e37bd
|
geneface-generalized-and-high-fidelity-audio
|
2301.13430
| null |
https://arxiv.org/abs/2301.13430v1
|
https://arxiv.org/pdf/2301.13430v1.pdf
|
GeneFace: Generalized and High-Fidelity Audio-Driven 3D Talking Face Synthesis
|
Generating photo-realistic video portrait with arbitrary speech audio is a crucial problem in film-making and virtual reality. Recently, several works explore the usage of neural radiance field in this task to improve 3D realness and image fidelity. However, the generalizability of previous NeRF-based methods to out-of-domain audio is limited by the small scale of training data. In this work, we propose GeneFace, a generalized and high-fidelity NeRF-based talking face generation method, which can generate natural results corresponding to various out-of-domain audio. Specifically, we learn a variaitional motion generator on a large lip-reading corpus, and introduce a domain adaptative post-net to calibrate the result. Moreover, we learn a NeRF-based renderer conditioned on the predicted facial motion. A head-aware torso-NeRF is proposed to eliminate the head-torso separation problem. Extensive experiments show that our method achieves more generalized and high-fidelity talking face generation compared to previous methods.
|
['Zhou Zhao', 'Jinzheng He', 'Jinglin Liu', 'Yi Ren', 'Ziyue Jiang', 'Zhenhui Ye']
|
2023-01-31
| null | null | null | null |
['talking-face-generation', 'face-generation']
|
['computer-vision', 'computer-vision']
|
[ 2.85814494e-01 1.15278117e-01 1.08986564e-01 -4.67543662e-01
-1.07910383e+00 -3.30742687e-01 6.44410193e-01 -1.05748343e+00
1.32097080e-01 7.36550689e-01 5.45763314e-01 1.29880786e-01
2.97525674e-01 -7.27575779e-01 -8.32050562e-01 -7.23071694e-01
3.24564159e-01 -1.35377068e-02 1.92147158e-02 -4.33911264e-01
-6.20182827e-02 5.47930717e-01 -1.77883983e+00 4.63310301e-01
7.89341629e-01 1.06190515e+00 1.97521880e-01 8.74434590e-01
1.24431863e-01 5.89697361e-01 -8.42803955e-01 -3.45690012e-01
1.54697359e-01 -8.39607537e-01 -4.00287062e-01 4.12607729e-01
7.14617670e-01 -4.43415105e-01 -4.37807262e-01 7.24016607e-01
1.16466475e+00 2.83486754e-01 8.49685252e-01 -1.24826491e+00
-7.21398413e-01 2.04669982e-01 -5.88185847e-01 -2.01517969e-01
9.25782382e-01 2.92974979e-01 5.17952919e-01 -8.30575466e-01
9.00471091e-01 1.59383857e+00 6.35693252e-01 1.06725872e+00
-9.70808566e-01 -9.46164668e-01 7.92231178e-04 9.38827246e-02
-1.72852051e+00 -9.28427041e-01 1.18291891e+00 -2.37282395e-01
5.10710478e-01 4.22479630e-01 7.64094532e-01 1.61540854e+00
-5.77545352e-02 6.48203015e-01 8.61839294e-01 -4.30546463e-01
1.69240594e-01 4.59076129e-02 -9.48018909e-01 5.81230462e-01
-5.19906104e-01 9.69331488e-02 -7.89126575e-01 4.64698896e-02
1.07585835e+00 -5.68027616e-01 -6.89957798e-01 -1.22098766e-01
-9.67594385e-01 5.72646916e-01 3.18151355e-01 1.47074237e-01
-2.82977045e-01 1.90442979e-01 1.71046749e-01 -1.22357801e-01
6.46193981e-01 9.35479403e-02 2.17197850e-01 -3.14669460e-02
-1.14411426e+00 3.21774453e-01 5.53416908e-01 8.66989970e-01
3.00294548e-01 6.14033818e-01 -4.07771498e-01 1.14826107e+00
1.32246733e-01 6.01200402e-01 5.66838920e-01 -1.14755523e+00
4.11960781e-01 -1.21664532e-01 9.73474011e-02 -7.96662331e-01
-2.18400091e-01 -7.93690979e-02 -1.00179005e+00 2.44670793e-01
-2.64064322e-04 -4.28394079e-01 -8.68951619e-01 1.89617002e+00
6.46635354e-01 6.61315262e-01 9.34038684e-02 1.33208537e+00
9.66099083e-01 9.34545100e-01 -1.96975708e-01 -3.88250738e-01
1.09741056e+00 -7.92628884e-01 -1.16607118e+00 1.59288645e-01
4.95071039e-02 -6.78495705e-01 1.17304552e+00 4.54368949e-01
-1.08998501e+00 -7.96234190e-01 -7.18715310e-01 -2.21181229e-01
2.05482081e-01 2.00028837e-01 4.96874183e-01 7.75176287e-01
-1.13824785e+00 3.15545112e-01 -2.91439682e-01 -2.13703528e-01
3.32010448e-01 4.41564880e-02 -3.86890203e-01 8.49911124e-02
-1.27799082e+00 3.91479552e-01 3.69216874e-02 1.65472806e-01
-1.04543698e+00 -6.05447710e-01 -1.00178730e+00 -1.66582793e-01
2.45230556e-01 -1.03379524e+00 1.31525409e+00 -1.32953203e+00
-2.37080288e+00 5.82015276e-01 -2.54582375e-01 -6.11005314e-02
7.44394898e-01 2.39742491e-02 -5.44020355e-01 3.58597547e-01
-1.83310941e-01 1.12564826e+00 1.22001529e+00 -1.48330081e+00
-2.44486481e-01 -1.31165415e-01 -1.03761598e-01 5.89411676e-01
-4.10423577e-01 -9.79938060e-02 -5.71337461e-01 -1.01002526e+00
-2.16079816e-01 -8.03798676e-01 1.24256119e-01 3.75314474e-01
-4.30132300e-01 1.55877993e-02 8.04033458e-01 -8.71431947e-01
1.05034542e+00 -2.18200469e+00 1.79872587e-01 -1.23227634e-01
-1.06964201e-01 1.45942613e-01 -2.99600810e-01 1.66372150e-01
-1.47886410e-01 9.30033773e-02 -1.78653628e-01 -5.33406258e-01
-1.14218101e-01 -3.58093455e-02 -3.08200866e-01 4.17510927e-01
2.61672527e-01 6.29016936e-01 -7.18053699e-01 -6.19662464e-01
3.04679483e-01 1.13144696e+00 -7.87110686e-01 3.13857466e-01
-2.31482834e-01 7.88160801e-01 -3.19291711e-01 5.18265903e-01
7.87318766e-01 2.52721071e-01 8.36345460e-03 -2.45303467e-01
6.72725067e-02 -2.71274298e-02 -1.24551988e+00 2.03065157e+00
-7.60340393e-01 7.59007573e-01 1.96797609e-01 -5.23623824e-01
9.52942848e-01 6.22174084e-01 4.90991116e-01 -6.22579396e-01
3.31804037e-01 -5.37516922e-02 -4.16104615e-01 -7.99209952e-01
5.34522355e-01 -5.39335489e-01 8.30496997e-02 -1.14637231e-02
1.46478772e-01 -6.46715701e-01 -2.51436234e-01 -1.42218217e-01
5.15216351e-01 4.52256918e-01 -9.15973485e-02 1.89685211e-01
5.91873586e-01 -6.03366435e-01 5.75766385e-01 1.10491432e-01
1.74958128e-02 1.33321798e+00 3.11142087e-01 4.71189544e-02
-9.97427583e-01 -9.61340070e-01 -1.89574957e-02 7.77097523e-01
1.70085371e-01 -2.69651115e-01 -1.26922953e+00 -4.13399398e-01
-3.14112186e-01 7.81142235e-01 -4.16685492e-01 -1.34163752e-01
-4.92581844e-01 -4.70120102e-01 7.77492762e-01 3.33839625e-01
7.23359227e-01 -9.40034568e-01 -1.10917255e-01 3.50785479e-02
-7.50616550e-01 -1.33273220e+00 -8.80889952e-01 -7.00405955e-01
-3.58119100e-01 -5.67379832e-01 -1.31886148e+00 -6.82568312e-01
3.42386425e-01 1.35502830e-01 7.12325215e-01 -3.90208781e-01
-2.58667678e-01 2.16575623e-01 -1.98152751e-01 -2.96496272e-01
-5.65373540e-01 -2.13632315e-01 2.01869100e-01 4.77823019e-01
-3.67505193e-01 -5.52633464e-01 -6.73410892e-01 4.09266770e-01
-8.62426281e-01 2.90129274e-01 1.01454683e-01 7.19532967e-01
5.36231875e-01 -3.01573500e-02 6.89521790e-01 -3.89025778e-01
6.84656560e-01 -2.44930163e-01 -3.46913755e-01 2.85082124e-02
1.35100126e-01 -1.47134259e-01 6.81572556e-01 -9.46896911e-01
-1.57216299e+00 1.81492031e-01 -6.25671089e-01 -7.95603096e-01
-1.56962663e-01 -3.01798433e-01 -7.33394563e-01 -5.51467463e-02
8.17341626e-01 1.71243921e-01 -6.72083944e-02 -2.15819448e-01
4.29778546e-01 8.81249726e-01 7.98391819e-01 -3.75842690e-01
6.79677904e-01 5.28132737e-01 -3.81553397e-02 -1.16870165e+00
-3.33004922e-01 7.45099410e-02 -1.62395626e-01 -6.83189392e-01
1.04373384e+00 -1.19543648e+00 -8.30347002e-01 5.97176909e-01
-1.29385030e+00 -5.85289299e-01 -3.14369053e-01 5.72767496e-01
-9.27240431e-01 3.49878699e-01 -4.67865050e-01 -9.90770817e-01
-2.33289480e-01 -1.19125712e+00 1.58872616e+00 2.27114990e-01
-7.75150806e-02 -5.60258389e-01 1.70926124e-01 5.38569748e-01
2.28085861e-01 4.54157889e-01 2.38339707e-01 3.24589640e-01
-5.18435121e-01 1.09902725e-01 3.51131782e-02 4.63870496e-01
5.21311574e-02 3.00746616e-02 -1.33727598e+00 -5.11073507e-03
-3.43290009e-02 -2.19688043e-01 5.75110734e-01 6.10061586e-01
1.30791485e+00 -4.88482714e-01 -2.36777723e-01 9.14937496e-01
9.93295014e-01 1.35019675e-01 9.31319833e-01 -3.55354965e-01
8.43776464e-01 6.51788235e-01 5.25266588e-01 6.79076433e-01
3.37161213e-01 1.15591168e+00 1.97045952e-01 -1.72951728e-01
-7.65698254e-01 -6.21099532e-01 6.30737305e-01 5.21066427e-01
-2.58466214e-01 -7.65109360e-01 -3.79995912e-01 3.75777990e-01
-1.37826073e+00 -1.16180670e+00 1.14712872e-01 1.96087492e+00
8.51639926e-01 -3.51703644e-01 2.60016143e-01 5.17853238e-02
9.86918032e-01 1.83445692e-01 -5.01809597e-01 -2.30007529e-01
-2.84172714e-01 1.09551124e-01 3.12902480e-02 6.51465952e-01
-7.67966926e-01 1.10692298e+00 6.01261568e+00 1.22279000e+00
-1.58324873e+00 1.45148188e-01 6.83975220e-01 -3.05714339e-01
-6.30839527e-01 -4.53346968e-01 -6.62568510e-01 3.59548599e-01
8.38037610e-01 2.62235068e-02 4.70788985e-01 6.63164675e-01
6.90897822e-01 1.64956182e-01 -8.62574041e-01 1.38386941e+00
6.34955943e-01 -1.18423164e+00 1.24757253e-01 7.84289092e-02
7.85936236e-01 -6.61389410e-01 1.62855178e-01 3.35709639e-02
-1.22034721e-01 -9.91785884e-01 1.05158317e+00 7.24950969e-01
1.44378603e+00 -9.27170813e-01 1.73055887e-01 2.69988030e-01
-1.17177665e+00 8.14785510e-02 -1.49154261e-01 3.33689541e-01
6.64414167e-01 3.10201317e-01 -9.26766694e-01 5.39829254e-01
5.73691308e-01 3.61687571e-01 -4.68064733e-02 7.86923409e-01
-2.78467685e-01 3.42644036e-01 -3.27345371e-01 1.62535831e-01
-2.13070244e-01 7.27904588e-02 7.02662170e-01 1.01039219e+00
7.38423347e-01 3.23458701e-01 -1.37922361e-01 7.98200369e-01
-3.36989492e-01 1.10275432e-01 -7.96741426e-01 1.56134441e-01
3.26348156e-01 1.21231127e+00 -2.92321682e-01 -1.07693359e-01
1.62217885e-01 1.40751648e+00 -1.85209990e-01 4.96294945e-01
-1.16503417e+00 -3.74339998e-01 6.83826387e-01 4.91883039e-01
2.49671608e-01 -9.97642502e-02 1.47397906e-01 -1.23356998e+00
3.53011228e-02 -8.61305237e-01 -1.09875284e-01 -1.45283628e+00
-8.15875649e-01 9.49743032e-01 1.68430842e-02 -1.33920825e+00
-7.18192816e-01 -2.89104223e-01 -5.15417397e-01 7.48817205e-01
-1.41347563e+00 -1.25530434e+00 -5.44444621e-01 8.41975689e-01
8.09915066e-01 -2.20537782e-01 5.83238900e-01 4.13944751e-01
-4.42345887e-01 8.87795627e-01 -1.84208244e-01 -2.53622755e-02
9.56709683e-01 -5.85872293e-01 3.23835492e-01 6.39696181e-01
8.04196671e-03 1.83078293e-02 7.36740112e-01 -4.99593347e-01
-1.36900389e+00 -1.14251232e+00 5.76678276e-01 -2.30294898e-01
7.04468712e-02 -7.44156957e-01 -6.17538095e-01 2.99399078e-01
2.19298095e-01 7.21166725e-04 4.55492139e-01 -5.64182639e-01
-1.13566756e-01 -3.53738308e-01 -1.46262801e+00 7.91442513e-01
1.39967918e+00 -5.77581465e-01 -1.79334432e-01 2.15140194e-01
7.85279334e-01 -5.13908327e-01 -7.77249336e-01 3.68083447e-01
6.29656255e-01 -1.00382042e+00 1.02726722e+00 2.60731182e-03
4.62412953e-01 -3.47215831e-01 -7.49181286e-02 -1.41797018e+00
9.10101458e-02 -1.20487797e+00 1.65518329e-01 1.57008338e+00
-2.74259262e-02 -2.57662356e-01 7.26941407e-01 1.62705064e-01
-2.20257267e-01 -2.59934306e-01 -9.68660116e-01 -5.89311838e-01
-1.11957639e-01 -5.30208468e-01 8.26800466e-01 8.64050686e-01
-1.87042668e-01 5.58822751e-01 -8.50659668e-01 1.51780307e-01
3.85475099e-01 -2.03799650e-01 1.12149584e+00 -9.12260711e-01
-3.02783668e-01 -5.56640439e-02 -3.38047713e-01 -1.19027114e+00
4.36463982e-01 -6.16003096e-01 3.56808424e-01 -1.39258039e+00
-2.43043035e-01 -1.71785682e-01 5.54764152e-01 1.06438458e-01
3.44530046e-02 4.25289303e-01 2.40319192e-01 -1.73634902e-01
-1.68725759e-01 1.11151600e+00 1.77047503e+00 6.22171983e-02
-2.55083591e-01 -8.25923681e-02 -4.79702115e-01 6.94678605e-01
4.97426242e-01 -1.97027668e-01 -7.42503703e-01 -3.83284152e-01
-1.82509750e-01 7.12354600e-01 4.68171924e-01 -1.18446076e+00
1.14252502e-02 -1.40907556e-01 4.47346300e-01 -3.36666167e-01
1.05011904e+00 -6.29241824e-01 4.81432587e-01 -9.80915129e-02
-2.96263993e-01 -2.21041471e-01 1.51580557e-01 4.41305786e-01
-2.67997324e-01 1.36270210e-01 9.67957377e-01 2.93820649e-02
-4.29550111e-01 4.86041456e-01 -2.29099780e-01 9.66890529e-02
8.60512078e-01 -3.33245486e-01 5.25687151e-02 -1.09251451e+00
-5.19522011e-01 -3.22645485e-01 4.09863681e-01 5.35337269e-01
8.80239666e-01 -1.64586627e+00 -9.09353316e-01 3.15177590e-01
-5.19985743e-02 -4.17241566e-02 8.18227112e-01 2.50597775e-01
-4.00537878e-01 4.93054138e-03 -2.03159600e-01 -6.64207757e-01
-1.25147462e+00 4.78239685e-01 5.19987047e-01 3.64239484e-01
-6.53432012e-01 8.45078111e-01 5.23602962e-01 -3.72868747e-01
2.17283577e-01 -5.80065548e-02 -1.65299684e-01 -1.72944769e-01
6.49135709e-01 1.56945691e-01 -1.08382069e-01 -9.62073684e-01
-1.64943591e-01 8.05641174e-01 5.42022884e-01 -6.34310722e-01
1.08372164e+00 -2.28933126e-01 5.09226739e-01 2.37780452e-01
1.30323792e+00 1.18248492e-01 -1.56764281e+00 2.81866044e-01
-9.45876956e-01 -6.39849544e-01 7.09141642e-02 -4.24617440e-01
-1.30514181e+00 1.03178990e+00 7.03752816e-01 -3.44886273e-01
1.49511814e+00 -2.16368124e-01 8.79891276e-01 -1.00298792e-01
2.22764030e-01 -1.01964700e+00 3.36958319e-01 -2.11019116e-03
1.33193970e+00 -9.05232906e-01 -3.31105292e-01 -6.88060284e-01
-1.01965809e+00 9.72972512e-01 6.70466304e-01 1.97137117e-01
3.48394930e-01 2.96689689e-01 4.77819815e-02 2.57895321e-01
-5.39662480e-01 7.77602941e-03 2.85395414e-01 9.60835993e-01
3.40135753e-01 -1.80787727e-01 -2.12366320e-02 3.88171643e-01
-6.12998366e-01 3.69657725e-01 4.81651306e-01 4.23480779e-01
-8.52324590e-02 -7.58826613e-01 -6.66241467e-01 -3.18589121e-01
-3.27131391e-01 1.10142700e-01 -2.86265761e-01 6.67111039e-01
2.07088470e-01 9.62484479e-01 1.51161343e-01 -5.05250812e-01
3.73138428e-01 -5.60631976e-03 7.07700193e-01 -2.01345593e-01
-3.10772300e-01 4.98089999e-01 2.25637257e-01 -4.89564925e-01
-3.53376865e-01 -3.54419023e-01 -1.16540253e+00 -4.19366360e-01
-3.88945192e-01 -1.50657803e-01 7.87733853e-01 5.07150173e-01
5.48682630e-01 6.35716915e-01 8.98820639e-01 -1.35022724e+00
-1.48857608e-01 -9.06260729e-01 -6.19310200e-01 4.24711078e-01
3.47888976e-01 -6.90116465e-01 -2.98710793e-01 3.16232026e-01]
|
[13.191535949707031, -0.4533918499946594]
|
28451b88-8af3-470c-9ecc-873684c0279a
|
xiezhi-an-ever-updating-benchmark-for
|
2306.05783
| null |
https://arxiv.org/abs/2306.05783v2
|
https://arxiv.org/pdf/2306.05783v2.pdf
|
Xiezhi: An Ever-Updating Benchmark for Holistic Domain Knowledge Evaluation
|
New Natural Langauge Process~(NLP) benchmarks are urgently needed to align with the rapid development of large language models (LLMs). We present Xiezhi, the most comprehensive evaluation suite designed to assess holistic domain knowledge. Xiezhi comprises multiple-choice questions across 516 diverse disciplines ranging from 13 different subjects with 249,587 questions and accompanied by Xiezhi-Specialty and Xiezhi-Interdiscipline, both with 15k questions. We conduct evaluation of the 47 cutting-edge LLMs on Xiezhi. Results indicate that LLMs exceed average performance of humans in science, engineering, agronomy, medicine, and art, but fall short in economics, jurisprudence, pedagogy, literature, history, and management. We anticipate Xiezhi will help analyze important strengths and shortcomings of LLMs, and the benchmark is released in~\url{https://github.com/MikeGu721/XiezhiBenchmark}.
|
['Shusen Wang', 'Zili Wang', 'Yanghua Xiao', 'Hongwei Feng', 'Weiguo Zheng', 'Wenhao Huang', 'Rui Xu', 'Qianyu He', 'Zihan Li', 'Zhuozhi Xiong', 'Sihang Jiang', 'Jianchen Wang', 'Lin Zhang', 'Haoning Ye', 'Xiaoxuan Zhu', 'Zhouhong Gu']
|
2023-06-09
| null | null | null | null |
['jurisprudence']
|
['miscellaneous']
|
[-2.11435318e-01 1.06986910e-01 -2.05854088e-01 5.44401556e-02
-1.10714650e+00 -1.22282290e+00 8.38635206e-01 2.48711199e-01
-4.23307121e-01 9.33779955e-01 7.02365339e-01 -8.17763329e-01
-7.29130507e-01 -6.59683108e-01 -7.42765307e-01 6.88624457e-02
4.61210459e-01 3.93421501e-01 -3.54524761e-01 -2.55784929e-01
5.48934817e-01 3.84979993e-02 -1.04350591e+00 2.39115745e-01
1.14983702e+00 5.07455170e-01 1.66830868e-01 5.74753463e-01
-4.19627130e-01 7.11219847e-01 -5.57934701e-01 -4.71184880e-01
-2.45058030e-01 -2.19262153e-01 -1.23524976e+00 -7.91512847e-01
3.63211036e-01 8.87789577e-02 -3.44807506e-01 8.48429024e-01
5.09232223e-01 3.52019928e-02 5.61360657e-01 -1.19052708e+00
-1.21654880e+00 8.91867578e-01 -3.29631507e-01 2.41158903e-01
7.96227396e-01 4.12096381e-01 1.39425957e+00 -1.10613227e+00
1.05421579e+00 1.41022730e+00 3.78618389e-01 3.86886239e-01
-9.76964533e-01 -1.08688402e+00 4.16156203e-02 2.76488990e-01
-1.34706688e+00 -3.66402984e-01 2.52075076e-01 -6.15224302e-01
8.34570467e-01 1.18326992e-01 1.88133061e-01 1.23288906e+00
4.18334514e-01 8.96691620e-01 1.30743980e+00 -2.09795982e-01
4.96217944e-02 -2.56821513e-03 3.39728802e-01 4.13155019e-01
3.50138456e-01 -1.73017919e-01 -1.10615456e+00 -1.44883141e-01
5.45324326e-01 -2.71593809e-01 -3.09993386e-01 6.70856535e-01
-1.77250993e+00 4.47290570e-01 -1.87359020e-01 5.94724178e-01
-4.28744256e-01 1.25535745e-02 3.48036289e-02 6.10760450e-01
3.90473574e-01 1.11949277e+00 -9.23148155e-01 -6.37079358e-01
-5.85320711e-01 6.57187700e-01 1.19699776e+00 1.41259849e+00
6.67888939e-01 -5.25921106e-01 -3.50144655e-01 8.02586913e-01
3.62079591e-01 6.78351760e-01 2.15203464e-01 -1.32620060e+00
5.05259514e-01 7.77131796e-01 5.37634641e-02 -7.45331109e-01
-2.09263906e-01 -3.38384598e-01 -7.31411695e-01 -4.81376141e-01
7.04065681e-01 -3.75987262e-01 -2.16554135e-01 1.71465814e+00
1.69480257e-02 1.09435633e-01 5.34444638e-02 3.41282189e-01
1.55672514e+00 9.29575503e-01 5.96602380e-01 2.23146200e-01
1.50090408e+00 -5.57195067e-01 -6.76070273e-01 -4.20318812e-01
6.14581943e-01 -1.08260667e+00 1.63215363e+00 6.84888780e-01
-1.29123330e+00 -3.34368497e-01 -4.83099252e-01 -4.45259094e-01
-7.25857615e-01 1.88648000e-01 4.50109899e-01 3.23274672e-01
-1.23168445e+00 4.14895058e-01 -6.06094524e-02 -6.13195598e-01
1.80269271e-01 -6.50523826e-02 -2.31814295e-01 -3.22290719e-01
-1.54880452e+00 1.05988979e+00 3.79304916e-01 -2.00945035e-01
-9.13535655e-01 -1.26111221e+00 -6.39518023e-01 4.28595357e-02
5.95242977e-01 -4.09492284e-01 1.38174999e+00 3.44584256e-01
-1.47208571e+00 1.01034856e+00 -1.85896322e-01 7.67337754e-02
8.76966417e-02 -5.33677757e-01 -5.09440422e-01 -1.32561207e-01
3.35166663e-01 5.96196592e-01 -2.25249112e-01 -7.05515921e-01
-5.65136254e-01 -9.09422189e-02 2.06410855e-01 -3.96047272e-02
-2.70069361e-01 4.20634240e-01 -4.74226356e-01 -4.03232872e-01
-1.77632958e-01 -7.28882253e-01 2.41621695e-02 -2.17410460e-01
-2.83884794e-01 -7.76041567e-01 5.90916932e-01 -9.20642555e-01
1.45171881e+00 -1.73823333e+00 -1.65033326e-01 -1.60025597e-01
4.96680468e-01 2.75716316e-02 -4.86335158e-01 1.05006719e+00
1.92946702e-01 6.26693964e-01 8.56213942e-02 9.43230018e-02
3.28547537e-01 2.08373759e-02 -3.74060035e-01 2.59658366e-01
8.17943960e-02 1.43901467e+00 -1.31649423e+00 -3.94774497e-01
1.30900592e-01 -7.39574134e-02 -3.72016579e-01 3.63331288e-02
-5.80227137e-01 5.43674469e-01 -6.11525953e-01 8.04381311e-01
3.62576127e-01 -6.92768395e-01 -8.58092457e-02 4.20110703e-01
-5.30666292e-01 5.98542452e-01 -7.78312266e-01 1.98885584e+00
-5.92916846e-01 9.13217783e-01 2.98109114e-01 -3.33782136e-01
8.36985648e-01 5.27136087e-01 3.98409665e-01 -4.11008984e-01
5.74123114e-02 2.09890038e-01 -9.48501527e-02 -5.04683733e-01
3.52715194e-01 1.12069800e-01 -4.47285831e-01 6.48068964e-01
2.90538907e-01 -6.15451217e-01 3.02715749e-01 3.52152616e-01
1.14124060e+00 3.21036339e-01 5.42757928e-01 -7.27344811e-01
5.61873794e-01 -4.65699583e-02 3.73172253e-01 8.97761643e-01
-2.37411201e-01 1.24246310e-02 5.47732353e-01 3.70527171e-02
-7.64372587e-01 -1.02618146e+00 -1.97053030e-01 1.40532041e+00
-4.81144749e-02 -7.81472147e-01 -4.92941022e-01 -2.12797672e-01
-1.29927740e-01 1.27164757e+00 -2.90065259e-01 2.34226122e-01
-2.19396904e-01 -3.30714345e-01 7.77916849e-01 1.82367802e-01
6.23009145e-01 -1.43087685e+00 -1.64641634e-01 2.56251812e-01
-4.19936776e-01 -1.14685631e+00 -2.37835377e-01 -2.30788454e-01
-3.05560946e-01 -8.03789496e-01 -5.58511496e-01 -7.37131417e-01
-2.98457742e-01 -2.13700742e-01 1.56592703e+00 -3.49257290e-01
-4.97110076e-02 7.04526246e-01 -3.16466540e-01 -8.71322513e-01
-5.23975492e-01 4.98596609e-01 -2.03844011e-01 -8.94253075e-01
7.51413047e-01 -4.69434887e-01 -4.23907399e-01 9.52823311e-02
-8.13800812e-01 -8.30079988e-02 6.48219049e-01 3.24254006e-01
2.08143473e-01 -2.46585995e-01 9.94461298e-01 -8.24136853e-01
1.01183069e+00 -8.80538702e-01 -4.84333247e-01 6.81551576e-01
-6.61140323e-01 -2.43468776e-01 4.54936564e-01 -2.65730083e-01
-1.21714664e+00 -8.52238953e-01 -3.62034798e-01 3.35184038e-01
-6.31704330e-01 9.97219384e-01 -3.94773185e-01 1.50715366e-01
5.94075263e-01 1.02867082e-01 -3.89840007e-01 -6.03128850e-01
7.10074842e-01 6.59423232e-01 5.54028392e-01 -1.08739614e+00
6.30368173e-01 -1.78717762e-01 -2.22291023e-01 -1.13620591e+00
-7.70769715e-01 -5.57896614e-01 -1.31679356e-01 -4.30760026e-01
7.29375541e-01 -8.81681859e-01 -1.12303650e+00 3.55212986e-01
-1.44694054e+00 -3.86459678e-01 -1.33497372e-01 5.15775561e-01
-1.67788103e-01 1.96956709e-01 -1.05554426e+00 -4.48945403e-01
-4.78779495e-01 -8.34000826e-01 9.95669663e-01 3.68491769e-01
-6.19154990e-01 -1.18502593e+00 6.42014593e-02 5.08457720e-01
3.09619099e-01 5.49040958e-02 1.32856059e+00 -8.22504878e-01
-5.11524320e-01 7.12925661e-03 -1.91330895e-01 3.14884603e-01
-4.10583057e-03 -7.66575858e-02 -7.17276692e-01 9.35065597e-02
-8.73623192e-02 -5.99156320e-01 2.13945314e-01 2.75450349e-01
1.10781217e+00 -2.08439425e-01 -1.53942127e-02 7.12660179e-02
1.45188403e+00 3.74925762e-01 5.54274619e-01 4.34854358e-01
4.51301873e-01 9.10707176e-01 4.57663745e-01 4.10262793e-01
7.13310122e-01 -2.80913085e-01 -1.17169864e-01 4.95298505e-01
1.36910481e-02 -5.05956292e-01 4.72908646e-01 1.11527765e+00
8.45837444e-02 -5.19076884e-01 -1.67341661e+00 7.68101394e-01
-1.42114639e+00 -7.57565916e-01 -4.05317038e-01 1.80692470e+00
1.08979857e+00 -2.64997166e-02 -4.66489315e-01 -4.92529809e-01
2.86455810e-01 2.09621806e-02 -4.80122864e-01 -2.14814335e-01
-2.30302125e-01 5.37342191e-01 2.89004207e-01 6.55121267e-01
-5.95347643e-01 1.35282850e+00 6.54832649e+00 1.14852762e+00
-5.79085708e-01 -1.52711524e-02 4.75995064e-01 -2.29490362e-02
-8.94582450e-01 2.41815239e-01 -1.09316325e+00 2.70830542e-01
1.26726294e+00 -9.35777366e-01 4.58338797e-01 5.23951828e-01
3.06398839e-01 -3.90545875e-02 -1.08440518e+00 8.13692093e-01
-2.93715596e-01 -1.57472849e+00 -1.58605233e-01 5.84078282e-02
8.71012330e-01 2.74261296e-01 8.01159814e-02 6.49170637e-01
7.46499121e-01 -1.25557351e+00 5.20070314e-01 7.90468216e-01
6.98267102e-01 -3.56470466e-01 4.28802818e-01 5.57895005e-01
-7.87624061e-01 1.68089658e-01 -1.86822563e-01 -2.94611961e-01
6.52747899e-02 4.53229010e-01 -8.14173579e-01 5.69180131e-01
6.65267944e-01 8.20041001e-01 -5.27475953e-01 6.82745397e-01
-4.82837170e-01 1.16826391e+00 -1.68552265e-01 -2.65461951e-01
3.24261844e-01 -4.47894752e-01 5.43753088e-01 1.27423263e+00
2.66945839e-01 4.94819164e-01 2.44357549e-02 1.11374044e+00
-1.62072957e-01 3.35482091e-01 -6.36644006e-01 -6.34950638e-01
8.32369864e-01 9.85539079e-01 -2.49759644e-01 -3.03407192e-01
-6.93153501e-01 4.61555779e-01 -1.46131858e-01 6.17245615e-01
-4.89424855e-01 -4.57869738e-01 7.25521028e-01 -7.39067495e-02
-5.15950203e-01 -3.57279509e-01 -7.48733521e-01 -1.10293794e+00
-3.54130149e-01 -1.27176714e+00 3.60324562e-01 -8.49398017e-01
-1.57484233e+00 2.45017380e-01 2.48584419e-01 -7.91086555e-01
-2.25070924e-01 -7.25399375e-01 -3.13256800e-01 1.11204827e+00
-1.30538201e+00 -1.30443192e+00 -1.29345983e-01 2.28874058e-01
6.09903872e-01 -1.43458007e-03 6.92322195e-01 2.66086340e-01
-4.11346078e-01 2.04868421e-01 2.42060184e-01 -9.32987928e-02
1.02982128e+00 -1.16168094e+00 6.60916388e-01 4.43018407e-01
-7.57444426e-02 1.05415428e+00 6.01874948e-01 -8.18525910e-01
-1.47372687e+00 -9.19693708e-01 1.68903267e+00 -1.05823207e+00
1.15977776e+00 -3.17671597e-01 -8.27441573e-01 9.51901317e-01
7.17724264e-01 -1.06090379e+00 1.04637516e+00 2.17222303e-01
-5.38776182e-02 1.64405912e-01 -9.06050682e-01 8.65579665e-01
1.03234756e+00 -8.30492377e-01 -7.67874599e-01 4.77435917e-01
1.03366494e+00 -2.38642603e-01 -1.46145582e+00 1.38240546e-01
5.14633477e-01 -1.99717894e-01 8.19769919e-01 -7.32609510e-01
8.33320737e-01 -3.05383429e-02 -5.06783538e-02 -9.34060514e-01
-3.27791661e-01 -8.51916552e-01 2.69637138e-01 1.50395513e+00
7.37741888e-01 -7.64690101e-01 5.54517508e-01 9.80914831e-01
-3.38712394e-01 -7.54605591e-01 -4.88175392e-01 -7.76320517e-01
9.42477584e-01 -7.65031576e-01 5.63616693e-01 1.06554508e+00
1.08937688e-01 5.78291297e-01 -1.94305271e-01 -1.06158301e-01
5.20708740e-01 1.33313248e-02 5.48707664e-01 -1.41825080e+00
-6.53700233e-02 -7.83153772e-01 2.57073760e-01 -1.23550248e+00
3.05720359e-01 -8.94977033e-01 -1.23149306e-01 -1.97085881e+00
8.12462568e-02 -3.25858407e-02 -4.52630804e-04 5.74934125e-01
-2.31535807e-01 -2.41248205e-01 3.86043475e-03 7.74357980e-03
-5.39086580e-01 5.57867229e-01 1.57987225e+00 8.78210738e-03
-2.66104769e-02 -2.27271661e-01 -1.31279624e+00 7.61688471e-01
9.37948823e-01 -2.19558805e-01 -5.26124358e-01 -6.80456817e-01
6.23734832e-01 1.12848192e-01 9.50720310e-02 -7.13107646e-01
5.89930773e-01 -6.49111688e-01 6.82850182e-02 -5.66315114e-01
-9.44335163e-02 -2.73315161e-01 -6.51270673e-02 1.38088331e-01
-8.29980969e-01 3.64027470e-01 3.79628837e-01 1.03236787e-01
-2.14045420e-01 -2.00345144e-01 3.26514572e-01 -4.43808436e-01
-9.41800117e-01 3.70973796e-01 -6.33392036e-01 7.05815971e-01
7.57137954e-01 2.40492910e-01 -6.31287873e-01 -5.65316617e-01
-4.56733644e-01 6.15447879e-01 2.36393347e-01 6.48036420e-01
6.27215743e-01 -8.04071307e-01 -1.08155227e+00 -3.84563297e-01
2.64109910e-01 3.54320593e-02 3.42289925e-01 3.61794293e-01
-4.77518082e-01 9.24785316e-01 1.13624811e-01 1.51389688e-01
-7.83982038e-01 1.11419454e-01 -3.67794484e-01 -3.66881698e-01
-1.45987689e-01 1.02957892e+00 1.86047181e-01 -7.02058673e-01
1.84970140e-01 -5.51868200e-01 -2.60962754e-01 -8.43902975e-02
4.96803343e-01 7.63940394e-01 -2.17057630e-01 -9.92681086e-02
-2.98078328e-01 3.36391449e-01 9.29458067e-02 -4.78455991e-01
1.23338294e+00 -1.07192352e-01 -5.90299726e-01 5.90754747e-01
1.03320360e+00 6.47895560e-02 -3.72595847e-01 -2.94697106e-01
6.35655701e-01 -2.10832357e-02 -1.55061230e-01 -1.28570044e+00
-1.98279709e-01 7.93922484e-01 -1.69800743e-02 -2.43064880e-01
8.31651807e-01 2.44495794e-01 8.46022725e-01 4.19835746e-01
4.11086440e-01 -9.78384316e-01 -3.41995284e-02 9.90897238e-01
1.17842543e+00 -8.32428694e-01 -1.44738093e-01 -3.22100431e-01
-6.25787795e-01 7.24409819e-01 7.44115949e-01 3.81252706e-01
9.43654180e-01 4.84826714e-02 -7.57085113e-03 -4.58737791e-01
-1.05972874e+00 6.18740357e-02 3.46320093e-01 2.23772824e-01
8.12204778e-01 -2.90674698e-02 -7.85105169e-01 7.70613253e-01
-5.35237432e-01 3.40142876e-01 5.25814891e-01 8.99584353e-01
-2.00101003e-01 -1.32172859e+00 -3.62612367e-01 4.85013127e-01
-6.26527071e-01 -5.20293593e-01 -5.64168394e-01 7.92678535e-01
2.50555813e-01 1.19027090e+00 -4.11466539e-01 -1.98342457e-01
2.48051345e-01 3.88705134e-01 1.67170301e-01 -7.87375093e-01
-7.72887766e-01 -1.79533988e-01 2.15759933e-01 -3.25792164e-01
-7.71447690e-03 -6.60884857e-01 -8.41083407e-01 -6.37609899e-01
1.88977063e-01 4.11827892e-01 5.63044667e-01 8.87302160e-01
5.28108060e-01 1.07654147e-01 -1.73873961e-01 1.04925781e-01
-4.44966972e-01 -1.14302349e+00 -5.99188209e-01 -8.16006660e-02
-1.88010894e-02 -1.96983919e-01 -2.37836212e-01 -2.83595026e-02]
|
[10.756387710571289, 8.723339080810547]
|
31174e1d-4f0b-4b2d-8b32-0f77553c83b4
|
hybrid-space-learning-for-language-based
|
2009.05381
| null |
https://arxiv.org/abs/2009.05381v2
|
https://arxiv.org/pdf/2009.05381v2.pdf
|
Dual Encoding for Video Retrieval by Text
|
This paper attacks the challenging problem of video retrieval by text. In such a retrieval paradigm, an end user searches for unlabeled videos by ad-hoc queries described exclusively in the form of a natural-language sentence, with no visual example provided. Given videos as sequences of frames and queries as sequences of words, an effective sequence-to-sequence cross-modal matching is crucial. To that end, the two modalities need to be first encoded into real-valued vectors and then projected into a common space. In this paper we achieve this by proposing a dual deep encoding network that encodes videos and queries into powerful dense representations of their own. Our novelty is two-fold. First, different from prior art that resorts to a specific single-level encoder, the proposed network performs multi-level encoding that represents the rich content of both modalities in a coarse-to-fine fashion. Second, different from a conventional common space learning algorithm which is either concept based or latent space based, we introduce hybrid space learning which combines the high performance of the latent space and the good interpretability of the concept space. Dual encoding is conceptually simple, practically effective and end-to-end trained with hybrid space learning. Extensive experiments on four challenging video datasets show the viability of the new method.
|
['Meng Wang', 'Xun Yang', 'Gang Yang', 'Xun Wang', 'Xirong Li', 'Jianfeng Dong', 'Chaoxi Xu']
|
2020-09-10
| null | null | null | null |
['ad-hoc-video-search']
|
['computer-vision']
|
[ 3.49799603e-01 -3.48404318e-01 -4.05189872e-01 -2.68062979e-01
-1.01380444e+00 -6.62976027e-01 9.18282509e-01 -7.86932409e-02
-2.45530143e-01 3.93933475e-01 2.60125339e-01 -4.38272953e-02
-1.47918537e-01 -4.72569138e-01 -8.50473583e-01 -6.59855723e-01
2.63971761e-02 1.03931189e-01 2.67996080e-02 7.22420216e-03
7.49304444e-02 1.81061476e-01 -1.68161082e+00 4.95907485e-01
5.47485590e-01 1.26118267e+00 3.23979080e-01 5.36700130e-01
-7.15468004e-02 1.02486849e+00 -2.08184958e-01 -4.24426734e-01
4.07333195e-01 -5.21243215e-01 -6.73719108e-01 4.20265377e-01
6.80074573e-01 -5.59020400e-01 -9.06603456e-01 1.08745384e+00
2.03771174e-01 2.00072169e-01 6.82894111e-01 -1.33709586e+00
-8.48326087e-01 1.42955065e-01 -4.01261002e-01 -5.35201952e-02
7.23116338e-01 1.65062204e-01 1.17593920e+00 -1.07739568e+00
7.37587690e-01 1.24606860e+00 2.62988746e-01 4.90698338e-01
-1.18012667e+00 -4.43871319e-01 1.09352402e-01 5.10202646e-01
-1.59593260e+00 -5.26010156e-01 9.34186697e-01 -7.40334153e-01
7.44113982e-01 1.79839984e-01 6.48703337e-01 1.31034458e+00
2.45101936e-03 1.07998753e+00 6.30215049e-01 -4.03511167e-01
2.35707611e-01 -4.83197793e-02 -2.31227309e-01 6.09787703e-01
-1.12721413e-01 -9.27689746e-02 -6.34819031e-01 -1.39464095e-01
8.97067368e-01 6.52906001e-01 -4.96342093e-01 -9.78291631e-01
-1.42421997e+00 9.20206010e-01 4.15826082e-01 4.39365566e-01
-4.07200158e-01 2.55481422e-01 4.62356567e-01 3.72154832e-01
2.43839413e-01 4.27330956e-02 2.41721179e-02 -2.02961266e-02
-1.24221313e+00 1.46073267e-01 5.93167961e-01 1.06196773e+00
6.30671620e-01 -3.53925973e-02 -2.37170368e-01 5.75393796e-01
5.20248175e-01 3.45828384e-01 8.22389245e-01 -8.36444139e-01
5.87461710e-01 4.59853053e-01 1.38967246e-01 -1.27837515e+00
3.23536575e-01 -2.66475737e-01 -8.48352015e-01 -7.63319209e-02
1.23147994e-01 3.19775194e-01 -9.08489108e-01 1.78487206e+00
-1.07963383e-03 4.39888656e-01 1.63440585e-01 1.09695661e+00
5.06593525e-01 9.23708975e-01 9.88738835e-02 -3.45780432e-01
1.30952787e+00 -1.20991230e+00 -9.35409188e-01 -1.33492440e-01
3.44553977e-01 -5.96078992e-01 9.26968217e-01 1.87847108e-01
-1.23485875e+00 -7.55029440e-01 -1.23156846e+00 -2.77189136e-01
-4.32846546e-01 -8.08853656e-02 3.35429370e-01 2.24400744e-01
-9.77270305e-01 3.36948276e-01 -6.75428629e-01 -3.83705109e-01
3.49136502e-01 1.54644534e-01 -6.09781504e-01 -2.88972020e-01
-1.30318642e+00 4.17620003e-01 6.41002715e-01 -3.02308183e-02
-8.59300017e-01 -3.99660468e-01 -1.12585473e+00 2.89656490e-01
5.74395657e-01 -8.51297319e-01 1.08581138e+00 -1.31609595e+00
-1.33278465e+00 7.81910419e-01 -1.71736747e-01 -5.24158478e-01
5.09775877e-01 -4.70604718e-01 -4.28614855e-01 8.09101224e-01
2.41677418e-01 5.98471522e-01 1.44184804e+00 -1.21184802e+00
-4.18628573e-01 -4.05171931e-01 2.97052860e-01 2.47028783e-01
-5.85474312e-01 -6.72405809e-02 -9.64717269e-01 -8.13252807e-01
1.49086803e-01 -7.79472768e-01 9.65125412e-02 4.61472392e-01
8.14036205e-02 -2.02049881e-01 1.07121778e+00 -5.79359353e-01
1.20754504e+00 -2.41285205e+00 4.49579090e-01 1.16893232e-01
4.08987671e-01 3.03025275e-01 -2.28677824e-01 6.60459518e-01
-2.46098310e-01 -1.49857998e-01 -2.34827459e-01 -3.07809055e-01
9.30624828e-02 1.06778383e-01 -4.67140943e-01 6.61953330e-01
2.57881999e-01 1.15695179e+00 -1.20594239e+00 -7.32731581e-01
3.34150732e-01 6.51553035e-01 -4.78032410e-01 5.20940125e-01
-1.23603955e-01 9.61459279e-02 -3.53891641e-01 7.18766928e-01
3.79640788e-01 -5.37791610e-01 2.68050134e-01 -4.97626424e-01
5.35339452e-02 -1.71871752e-01 -1.03404772e+00 2.28465176e+00
-2.69939989e-01 6.29448056e-01 -2.27399971e-02 -1.39242363e+00
6.15829229e-01 5.57334960e-01 5.56125402e-01 -7.64522433e-01
-1.26617983e-01 2.94245124e-01 -6.52577281e-01 -6.90844536e-01
3.50481570e-01 -1.41414642e-01 -1.32806703e-01 2.23875701e-01
4.48921591e-01 2.34660372e-01 1.73061073e-01 4.11533952e-01
8.08070242e-01 3.67667884e-01 4.66915131e-01 1.35518521e-01
6.19434178e-01 -3.39788198e-01 3.37037623e-01 6.43293738e-01
-9.54586118e-02 8.44030142e-01 2.67061144e-01 -3.01182985e-01
-1.09175014e+00 -1.02481651e+00 8.10400173e-02 9.79077280e-01
3.94565076e-01 -6.25659943e-01 -4.55214411e-01 -7.08899856e-01
-2.19149906e-02 1.88314840e-01 -6.68763757e-01 -1.77379668e-01
-4.04052496e-01 1.05540723e-01 2.47374222e-01 5.29485583e-01
4.11110133e-01 -7.60380149e-01 -6.90496206e-01 1.33120194e-01
-2.95288563e-01 -1.19560003e+00 -8.36694121e-01 -9.12640542e-02
-6.95285380e-01 -9.33761477e-01 -1.08053875e+00 -8.86620462e-01
4.44224060e-01 6.33249879e-01 1.01578093e+00 1.06281206e-01
-2.81837821e-01 7.81465590e-01 -6.26024246e-01 1.89203888e-01
6.87408149e-02 -4.06301200e-01 5.28594889e-02 6.70082629e-01
4.23677862e-01 -5.07213831e-01 -8.13011050e-01 2.61105895e-02
-1.44330478e+00 -2.10992619e-02 6.90988839e-01 1.13834572e+00
5.21272659e-01 -1.79916203e-01 2.81528771e-01 -4.93543625e-01
3.29804718e-01 -8.12140942e-01 -3.93902957e-01 5.99038243e-01
-1.56880349e-01 5.91627136e-02 5.37043095e-01 -4.57030118e-01
-6.90054536e-01 4.48226988e-01 2.48536810e-01 -1.19783115e+00
-7.14077801e-02 6.23145103e-01 -2.66243517e-01 2.13952512e-01
4.16937053e-01 5.69506049e-01 1.97828621e-01 -4.13247049e-01
6.51052594e-01 6.13864243e-01 6.47735059e-01 -2.50976861e-01
9.50840235e-01 5.77524781e-01 -1.94288433e-01 -8.43397439e-01
-5.56590438e-01 -9.32231486e-01 -8.82294774e-01 -2.37875625e-01
1.05209231e+00 -1.18375003e+00 -5.27204990e-01 1.34999618e-01
-1.07369626e+00 2.03888655e-01 -2.88358092e-01 5.61196387e-01
-8.18342626e-01 8.79660606e-01 -3.49180758e-01 -7.80935287e-01
-8.33339468e-02 -1.07542467e+00 1.43739545e+00 -4.79237363e-02
-5.67254387e-02 -1.00096202e+00 -4.75989655e-02 3.32452595e-01
1.90539330e-01 1.22410029e-01 6.13327146e-01 -6.00730658e-01
-6.75319195e-01 -2.95184791e-01 -3.18752646e-01 2.36905336e-01
6.88208938e-02 -2.20825791e-01 -8.64861846e-01 -6.74161434e-01
2.86492050e-01 -5.88309407e-01 9.59623218e-01 6.49576560e-02
1.26917267e+00 -5.28156996e-01 -3.03183794e-01 6.70724630e-01
1.79700089e+00 2.29285285e-01 6.10234737e-01 2.93010831e-01
6.12280369e-01 5.24576962e-01 4.85006869e-01 4.08871114e-01
1.72546327e-01 9.41138446e-01 5.01822591e-01 -3.08756381e-02
4.95377705e-02 -4.25201714e-01 4.46028352e-01 8.25203955e-01
2.16304839e-01 -2.19781220e-01 -7.12116838e-01 6.31805718e-01
-2.14540482e+00 -1.43301129e+00 4.67422932e-01 2.07200813e+00
6.78619981e-01 -1.52158856e-01 1.67916983e-01 2.12300673e-01
6.24134183e-01 3.47443640e-01 -4.60868597e-01 1.39979243e-01
-1.68572441e-01 -2.14527190e-01 6.40507415e-02 2.55845398e-01
-1.21411753e+00 5.35209656e-01 5.27219963e+00 9.87658262e-01
-1.20506513e+00 1.09038889e-01 2.62741148e-01 -1.98618099e-01
-3.88252914e-01 2.81151850e-03 -3.41732144e-01 5.53294539e-01
6.93642795e-01 -1.45179957e-01 3.30225885e-01 7.83678651e-01
-2.19550997e-01 2.90376872e-01 -1.41471219e+00 1.60312450e+00
3.93377542e-01 -1.44972694e+00 4.34585780e-01 -1.09934241e-01
5.78735828e-01 -3.68833333e-01 1.76924050e-01 3.87515724e-01
-4.08890843e-01 -9.12735641e-01 9.85888541e-01 7.83594489e-01
1.06936383e+00 -4.47059989e-01 4.77661431e-01 2.02441901e-01
-1.41226912e+00 -2.82814801e-01 -2.83059150e-01 1.87968671e-01
1.99772939e-01 1.30319744e-02 -2.90923238e-01 8.53072166e-01
6.22361898e-01 1.07141232e+00 -5.09673893e-01 1.00318038e+00
2.83051640e-01 2.87139058e-01 -4.31263000e-02 3.28488201e-01
4.97239053e-01 -1.09936856e-01 4.95314121e-01 1.38421750e+00
3.15419137e-01 1.00382783e-01 3.74437869e-01 7.74790287e-01
6.77101361e-03 1.56601399e-01 -1.02000868e+00 -3.84386659e-01
2.84590214e-01 9.98289585e-01 -4.82727468e-01 -5.36494613e-01
-8.79926801e-01 1.25740683e+00 1.19141370e-01 6.73512101e-01
-6.60820484e-01 -3.65324914e-01 3.78301978e-01 1.11972533e-01
5.23755610e-01 -1.36465341e-01 2.89675385e-01 -1.55807734e+00
3.31192732e-01 -9.47068453e-01 3.79025936e-01 -8.11918974e-01
-1.51454568e+00 6.46875978e-01 1.31703585e-01 -1.91566288e+00
-6.33357644e-01 -5.29526770e-01 -1.59252822e-01 7.33646750e-01
-1.59258771e+00 -1.40522671e+00 -3.21584225e-01 9.94114518e-01
9.09254909e-01 -4.74420041e-01 7.21741915e-01 3.98604989e-01
-1.64390892e-01 5.28934002e-01 4.36960638e-01 2.39322633e-01
6.18863821e-01 -1.07981849e+00 9.70712304e-03 7.79174805e-01
4.52805966e-01 8.42101812e-01 3.05460274e-01 -4.12434638e-01
-1.82298970e+00 -8.74750853e-01 7.21538007e-01 -3.06594908e-01
8.84193242e-01 -4.60532129e-01 -9.76522565e-01 5.29548168e-01
4.34096098e-01 2.61400759e-01 8.35454226e-01 -3.44245017e-01
-6.13009930e-01 -8.14261287e-02 -6.95547462e-01 4.60022122e-01
8.14735293e-01 -1.17715287e+00 -8.70665431e-01 3.21734548e-01
8.05263579e-01 -9.08743292e-02 -6.99335039e-01 3.61502647e-01
7.86820471e-01 -8.74087155e-01 1.15778720e+00 -7.81026483e-01
5.40317655e-01 -2.86699414e-01 -6.48087263e-01 -7.64202893e-01
-3.49709719e-01 -5.97295582e-01 -5.58120668e-01 9.54436302e-01
-1.26576811e-01 -8.26605558e-02 5.71027040e-01 3.88607532e-01
1.09053165e-01 -7.63134778e-01 -7.96525359e-01 -8.69021177e-01
-2.98766524e-01 -2.70686865e-01 4.22544360e-01 1.08796549e+00
1.01011433e-01 4.97855246e-01 -6.95508361e-01 -4.27848101e-02
7.13231862e-01 3.61272097e-01 5.62768340e-01 -1.09107733e+00
-4.38452065e-01 -3.24443340e-01 -7.33496368e-01 -1.52350736e+00
1.90828055e-01 -8.11069608e-01 8.34566541e-03 -1.30771637e+00
4.14965510e-01 2.20620483e-01 -5.14973104e-01 2.25927263e-01
1.50968274e-03 3.27667952e-01 4.25694525e-01 6.03740811e-01
-1.09866619e+00 7.51041770e-01 8.72640252e-01 -3.49077880e-01
1.27888501e-01 -3.42015982e-01 -3.27106178e-01 4.54258770e-01
2.03109294e-01 -1.26114056e-01 -7.23584294e-01 -5.50893068e-01
1.44773930e-01 5.01265883e-01 7.00384021e-01 -9.22396004e-01
4.32454705e-01 -4.27268520e-02 2.91619807e-01 -7.05888391e-01
6.58563972e-01 -1.21417439e+00 6.97127506e-02 2.49336675e-01
-5.02124965e-01 -7.04582781e-02 -9.70042720e-02 1.07327330e+00
-8.19079995e-01 -6.88146353e-02 4.92790937e-01 -1.75019324e-01
-1.00800896e+00 4.76252407e-01 -1.30075574e-01 -1.54729664e-01
1.19961214e+00 -5.41927099e-01 9.20146108e-02 -6.77798986e-01
-7.19850898e-01 2.10295245e-01 4.96737480e-01 6.46431923e-01
9.92538393e-01 -1.68350649e+00 -5.10090232e-01 3.16009641e-01
4.67331260e-01 -3.24840367e-01 4.17491466e-01 6.79553151e-01
-2.31416225e-01 7.96897531e-01 -3.31757575e-01 -8.82357836e-01
-1.05298185e+00 1.04601204e+00 6.26470894e-02 2.37131268e-02
-5.61350286e-01 7.09386230e-01 5.37161946e-01 7.78083652e-02
4.72285032e-01 1.74725190e-01 -3.22547108e-01 2.90655136e-01
6.59971297e-01 -2.59231217e-02 -2.54226923e-01 -9.86967564e-01
-2.64633000e-01 6.55555964e-01 2.78318115e-03 -8.76563042e-02
1.36090970e+00 -3.16541374e-01 -2.16841344e-02 7.04432964e-01
1.77346313e+00 -3.89101803e-01 -1.40399098e+00 -6.32889986e-01
-1.09486775e-02 -8.73485565e-01 -7.36202300e-02 -2.33276159e-01
-9.41557884e-01 9.46248770e-01 7.10301995e-01 2.72166550e-01
1.27000737e+00 -2.17896625e-02 7.60643244e-01 4.27624702e-01
1.40917033e-01 -8.96739304e-01 5.36295414e-01 1.48312435e-01
1.01799726e+00 -1.40178204e+00 -7.41031244e-02 -7.84987882e-02
-4.77185905e-01 1.19259071e+00 3.82813990e-01 -2.98616644e-02
5.23104429e-01 -3.51025373e-01 -2.48218894e-01 -2.20003188e-01
-8.36585283e-01 -1.76433414e-01 6.15305722e-01 3.97700310e-01
3.05338383e-01 -3.58569086e-01 -1.40120298e-01 2.97600567e-01
3.84661913e-01 1.88409656e-01 4.01408523e-02 1.15705824e+00
-2.04456002e-01 -9.64579642e-01 -2.44620815e-01 2.00326025e-01
-4.01747078e-01 -7.89210498e-02 -1.57017112e-01 5.93888283e-01
-1.59201533e-01 6.90871835e-01 8.00745189e-02 -3.07459772e-01
1.35857344e-01 2.59432703e-01 3.91044259e-01 -5.74191213e-01
-2.13693455e-02 2.91801065e-01 -5.23396611e-01 -7.41059780e-01
-7.96380937e-01 -5.50510406e-01 -7.98657775e-01 2.68354505e-01
-9.81561765e-02 2.43543029e-01 5.61878324e-01 9.74979043e-01
2.68377632e-01 1.00605555e-01 6.89519882e-01 -9.53388155e-01
-7.11192131e-01 -5.96775889e-01 -5.26358008e-01 9.34524059e-01
7.25681722e-01 -7.92845964e-01 -2.53795028e-01 4.85945076e-01]
|
[10.243576049804688, 0.9854130744934082]
|
c1488fb3-2188-47ef-a51e-9f0263c6a31c
|
detecting-signatures-of-early-stage-dementia
|
2007.03615
| null |
https://arxiv.org/abs/2007.03615v1
|
https://arxiv.org/pdf/2007.03615v1.pdf
|
Detecting Signatures of Early-stage Dementia with Behavioural Models Derived from Sensor Data
|
There is a pressing need to automatically understand the state and progression of chronic neurological diseases such as dementia. The emergence of state-of-the-art sensing platforms offers unprecedented opportunities for indirect and automatic evaluation of disease state through the lens of behavioural monitoring. This paper specifically seeks to characterise behavioural signatures of mild cognitive impairment (MCI) and Alzheimer's disease (AD) in the \textit{early} stages of the disease. We introduce bespoke behavioural models and analyses of key symptoms and deploy these on a novel dataset of longitudinal sensor data from persons with MCI and AD. We present preliminary findings that show the relationship between levels of sleep quality and wandering can be subtly different between patients in the early stages of dementia and healthy cohabiting controls.
|
['Raul Santos-Rodriguez', 'Yoav Ben-Shlomo', 'Niall Twomey', 'Weisong Yang', 'Rafael Poyiadzi', 'James Selwood', 'Liz Coulthard', 'Ian Craddock']
|
2020-07-03
| null | null | null | null |
['sleep-quality-prediction']
|
['medical']
|
[ 3.10316920e-01 -2.84965992e-01 1.87840387e-01 -5.81996322e-01
-2.17283189e-01 -6.13223054e-02 7.12253571e-01 1.62336320e-01
-8.95865858e-01 7.60074556e-01 7.08299220e-01 -2.65912652e-01
-7.24727988e-01 -3.10965955e-01 2.65259296e-01 -6.40965343e-01
-7.77706742e-01 6.02984309e-01 1.48845375e-01 -1.48677960e-01
-2.32037410e-01 1.70636252e-01 -1.19536150e+00 1.97384566e-01
6.55608833e-01 7.19348431e-01 6.18613958e-01 6.83294654e-01
4.80057955e-01 3.25232893e-01 -3.63080919e-01 -3.73585452e-03
-2.55957633e-01 -4.58127081e-01 -4.59497273e-01 -9.03380588e-02
-2.20319852e-02 -5.59789538e-01 -3.10745716e-01 8.13686550e-01
7.92272925e-01 -1.11353621e-01 3.61876637e-01 -1.00613976e+00
-3.60934943e-01 -2.33028784e-01 3.92321981e-02 1.04252827e+00
2.66855806e-01 1.75154269e-01 5.55078447e-01 -3.12715352e-01
3.06523800e-01 1.03015566e+00 1.06366003e+00 6.97089791e-01
-1.49121606e+00 -2.71750718e-01 1.37343286e-02 8.68355036e-01
-9.58050132e-01 -1.10303724e+00 1.82236254e-01 -7.89206803e-01
1.02659118e+00 2.03096300e-01 1.25506973e+00 1.34251487e+00
4.03110832e-01 -1.24216139e-01 1.20686936e+00 -2.44788323e-02
7.18125045e-01 -3.75308156e-01 7.12939382e-01 1.31308943e-01
3.12666208e-01 3.98351103e-02 -3.38982612e-01 -7.51614153e-01
3.75613660e-01 4.39500034e-01 -1.64589494e-01 -7.48634189e-02
-1.37501669e+00 5.92971087e-01 1.58515960e-01 4.43050265e-01
-4.37224418e-01 -1.09248452e-01 3.58617216e-01 3.30865920e-01
4.01128918e-01 -2.67134279e-01 -5.44573009e-01 -4.64770943e-01
-1.04966140e+00 -2.63949111e-02 3.17152441e-01 1.82699665e-01
1.93140015e-01 -4.71859843e-01 -1.98209673e-01 9.06208158e-01
8.11207652e-01 6.37280285e-01 6.97440028e-01 -1.12436664e+00
2.19314575e-01 6.78676784e-01 1.03154287e-01 -4.44572151e-01
-8.86859715e-01 -3.21320802e-01 -9.29338813e-01 4.64517891e-01
2.30781257e-01 -1.56270817e-01 -4.76994634e-01 1.66075909e+00
-1.06104929e-02 -4.60163280e-02 -2.43102923e-01 5.06848931e-01
1.99914843e-01 -2.74434924e-01 3.04119349e-01 -6.45619273e-01
1.64779174e+00 -9.36508030e-02 -5.27113259e-01 -8.96837771e-01
5.05701661e-01 3.02514464e-01 8.03788483e-01 6.96651787e-02
-8.32290351e-01 -2.28277043e-01 -7.66134620e-01 1.04273677e-01
-2.04548344e-01 -2.31750220e-01 2.40501449e-01 1.07416654e+00
-1.59715271e+00 4.14391398e-01 -1.63084757e+00 -1.01140523e+00
9.42028999e-01 5.98013759e-01 -3.96935701e-01 -2.00987682e-01
-7.73230076e-01 1.01197827e+00 -1.61517844e-01 4.73477952e-02
-4.69105095e-01 -6.99929297e-01 -2.38202125e-01 -4.58572090e-01
-5.06312311e-01 -1.06670022e+00 8.77615035e-01 -2.95444131e-01
-7.32018948e-01 1.20845044e+00 -4.85884994e-01 -3.10879022e-01
3.43533486e-01 -2.94156611e-01 -6.52239025e-01 1.12418830e-01
2.82505661e-01 3.02640945e-01 7.95399621e-02 -3.00618589e-01
-2.93310195e-01 -1.45077693e+00 -4.40840304e-01 -6.59891590e-02
-2.24930912e-01 5.89267194e-01 7.09829330e-01 -1.98509321e-01
-5.77219427e-02 -8.54439735e-01 -2.93587983e-01 6.78494334e-01
-1.62275553e-01 -7.65635371e-02 7.39490032e-01 -6.82603717e-01
1.13535869e+00 -1.94025922e+00 -2.34747216e-01 -3.29845786e-01
7.84802794e-01 2.62984425e-01 3.76084149e-01 8.63801241e-02
2.28909075e-01 7.20041841e-02 -4.26380128e-01 -5.60293198e-01
8.39894563e-02 5.10629714e-01 3.94947976e-01 8.70714724e-01
-1.58526272e-01 1.07372022e+00 -8.65714073e-01 -9.80092287e-02
1.40885636e-01 4.67863768e-01 -3.34806800e-01 -3.31468438e-03
2.47784033e-01 6.83742285e-01 -3.44479740e-01 4.14952785e-01
2.28933662e-01 -2.62267470e-01 2.77817905e-01 3.10025662e-01
-3.49534541e-01 5.23626506e-01 -2.23039195e-01 1.28235900e+00
3.87092650e-01 6.88760102e-01 3.18859518e-01 -9.73392427e-01
5.30105591e-01 3.34592968e-01 3.84537876e-01 -1.18780911e+00
2.35953793e-01 -4.33884449e-02 4.63497102e-01 -7.66644001e-01
-5.66932380e-01 -4.24214423e-01 1.25107259e-01 5.83823860e-01
-4.47560996e-01 8.55677843e-01 3.27233300e-02 -3.32323194e-01
1.89731407e+00 -4.90590841e-01 5.79067707e-01 -4.99193490e-01
2.00959757e-01 -2.65359223e-01 5.36982596e-01 6.35211051e-01
-1.08288121e+00 3.71538550e-01 3.88577133e-01 -3.18512470e-01
-9.34019327e-01 -1.44279826e+00 -3.51838797e-01 9.92413342e-01
-7.31261849e-01 -4.34510112e-01 -3.02830994e-01 -9.07072499e-02
-1.41017959e-01 4.64506686e-01 -7.22644031e-01 -5.28841138e-01
-3.41423690e-01 -1.42794836e+00 4.56812322e-01 6.12459064e-01
8.77802312e-01 -7.25664377e-01 -1.07742453e+00 1.14105225e-01
-1.33282751e-01 -9.49987411e-01 3.23003620e-01 2.24431455e-01
-1.27124560e+00 -1.20880497e+00 -6.96469605e-01 -7.19168425e-01
5.67699492e-01 1.29568249e-01 9.11310017e-01 -1.43183600e-02
-4.44050342e-01 4.87708211e-01 -2.60250956e-01 -2.36316264e-01
-5.64140379e-02 -2.41329268e-01 3.93287957e-01 -2.79061943e-01
8.60302508e-01 -1.26897430e+00 -1.14040101e+00 2.25589931e-01
-3.01350653e-01 -4.26859021e-01 5.25380194e-01 -2.02176601e-01
2.64411598e-01 -1.96780264e-01 5.60714424e-01 -1.02846622e-01
5.44464350e-01 -1.01169181e+00 1.61850736e-01 1.00204917e-02
-8.00597727e-01 -3.06240201e-01 1.10981800e-02 -2.53826648e-01
-6.23236299e-01 -1.47833422e-01 -2.46371534e-02 4.17127490e-01
-7.41402805e-01 1.04456849e-01 -3.45593005e-01 5.40790319e-01
5.60427368e-01 1.57834053e-01 3.02122623e-01 -6.63398862e-01
-2.34542057e-01 1.06950939e+00 6.19305789e-01 1.06343716e-01
9.07979012e-02 8.74682844e-01 -1.23927258e-01 -1.22005022e+00
-5.48524916e-01 -5.08075655e-01 -8.22934031e-01 -3.29387426e-01
1.15361249e+00 -8.99696410e-01 -4.82903451e-01 7.70410359e-01
-4.90277797e-01 -9.91583049e-01 -5.85126206e-02 6.28804088e-01
-4.84468311e-01 1.59299865e-01 -2.03479230e-01 -7.95093298e-01
-4.96717721e-01 -4.98836726e-01 8.83681953e-01 -1.42318472e-01
-9.77661908e-01 -1.25551462e+00 8.45305800e-01 5.26925683e-01
5.71692109e-01 2.56889820e-01 1.08305836e+00 -5.51258564e-01
1.37402967e-01 -6.01461045e-02 -1.87275723e-01 1.14171401e-01
3.79285097e-01 -7.42072403e-01 -4.44790840e-01 -2.13013574e-01
6.52317226e-01 1.62779577e-02 4.05548692e-01 5.66374421e-01
-1.44020572e-01 -1.00383954e-02 -5.50305784e-01 1.30023703e-01
7.94857204e-01 2.67259151e-01 1.05227971e+00 8.89322877e-01
1.40162826e-01 3.45356554e-01 -2.02408940e-01 3.82825315e-01
6.58069074e-01 7.79314697e-01 1.41271979e-01 6.03271663e-01
-1.13057576e-01 8.06232035e-01 7.27931261e-01 2.78505027e-01
-2.22725794e-01 3.60332988e-02 -1.04665482e+00 5.16780674e-01
-1.59928179e+00 -1.15977049e+00 -6.60268009e-01 2.08772135e+00
5.00996470e-01 3.84101309e-02 5.13355553e-01 3.18179786e-01
5.34830570e-01 -3.43578756e-02 -7.55996823e-01 1.29843988e-02
-1.38910294e-01 9.55256671e-02 1.17099304e-02 -1.03063053e-02
-6.13901138e-01 -1.49629980e-01 7.52767038e+00 -4.14405197e-01
-5.49052477e-01 5.25571585e-01 5.29295087e-01 -7.32136309e-01
2.67227918e-01 -2.55470783e-01 -4.72173870e-01 6.18185878e-01
1.81171525e+00 2.75971502e-01 4.45856541e-01 1.04249686e-01
1.00727677e+00 -4.91558731e-01 -1.10015380e+00 5.86504340e-01
-7.01713562e-02 -5.30457556e-01 -8.28345895e-01 5.55673242e-01
4.72400039e-02 7.55136669e-01 -2.49206468e-01 -4.13945727e-02
4.11956087e-02 -6.83432519e-01 4.83432502e-01 1.25469828e+00
6.77850127e-01 -1.15991026e-01 6.43703163e-01 3.04900020e-01
-8.11510086e-01 -4.61517632e-01 1.14772812e-01 -7.14509249e-01
5.95389545e-01 7.02173829e-01 -5.56633174e-01 -4.65655416e-01
1.04987442e+00 8.37028623e-01 -9.23912883e-01 1.20702982e+00
1.56166047e-01 6.47509754e-01 -2.87511766e-01 1.61890939e-01
-3.53830338e-01 -1.99166387e-01 5.14357924e-01 6.93121135e-01
6.40729249e-01 2.95224398e-01 -3.48698854e-01 8.49505603e-01
7.69027174e-01 -6.47383928e-01 -1.11401394e-01 -2.63743401e-01
1.98635176e-01 8.52061987e-01 -7.50612676e-01 -2.11334765e-01
-4.13896710e-01 7.68135428e-01 2.51932647e-02 -1.83681533e-01
-8.01612735e-02 4.92876261e-01 9.51394558e-01 5.72721422e-01
1.91814318e-01 -7.87332475e-01 -4.49402064e-01 -7.73775578e-01
1.81147709e-01 -3.90359610e-01 3.37393671e-01 -1.03532231e+00
-1.28227735e+00 1.20326430e-01 3.14044161e-03 -6.31896496e-01
-1.71563417e-01 -3.55731308e-01 -8.86759102e-01 5.68733931e-01
-9.75157797e-01 -4.42724913e-01 -6.41653717e-01 7.97966361e-01
2.73062498e-03 -9.76121873e-02 1.18424761e+00 3.38721335e-01
-8.19118679e-01 -2.39691898e-01 4.82949585e-01 -1.48438513e-01
2.63294816e-01 -1.08926094e+00 3.89853925e-01 3.71923804e-01
-6.05444610e-01 6.43523693e-01 6.28363907e-01 -1.00885761e+00
-7.95846760e-01 -1.12598813e+00 1.02067709e+00 -8.64328861e-01
8.52372706e-01 -5.44061780e-01 -4.84903604e-01 6.00637615e-01
-2.13692278e-01 -4.72753614e-01 1.20086694e+00 -3.67286019e-02
1.79408193e-01 -1.00398138e-01 -1.39477968e+00 5.69016814e-01
1.57284105e+00 -8.19352984e-01 -1.12613678e+00 3.44381362e-01
5.66863082e-02 8.95829499e-01 -9.03783381e-01 1.66317016e-01
8.57998729e-01 -1.61156952e+00 1.31840205e+00 -4.12658483e-01
-2.36616001e-01 9.73586068e-02 -3.70268613e-01 -8.69558990e-01
-5.87962270e-01 -3.31831634e-01 -3.39212179e-01 1.09613633e+00
-2.41932780e-01 -6.90280616e-01 6.73378408e-01 1.24545765e+00
-1.48185221e-02 -3.38087112e-01 -1.33102405e+00 -9.19227481e-01
6.27808124e-02 -4.59446609e-01 9.60480571e-02 2.38923103e-01
3.46168876e-01 3.40186447e-01 4.76028353e-01 5.96988872e-02
6.03135407e-01 -7.41346121e-01 -1.50441825e-01 -2.12106323e+00
2.27226630e-01 -2.88120538e-01 -1.07837045e+00 5.95294759e-02
-1.22843049e-01 -8.23161244e-01 -3.57929587e-01 -2.06756282e+00
5.93321204e-01 -2.16836050e-01 -4.32196945e-01 3.49220604e-01
1.30332962e-01 2.55156398e-01 -3.33837003e-01 3.96857858e-01
-6.72769368e-01 5.20096719e-01 4.65913385e-01 -2.22356291e-03
-4.23397332e-01 2.44904771e-01 -9.24874783e-01 5.39907217e-01
1.35495567e+00 -4.71012920e-01 -6.66906953e-01 -2.30410248e-01
2.82325029e-01 -2.10430607e-01 1.23175502e+00 -1.47326016e+00
5.32107279e-02 1.74279839e-01 6.17080569e-01 -3.33633840e-01
6.31229222e-01 -7.11913109e-01 3.57162207e-01 7.82579064e-01
-1.06013022e-01 1.95986837e-01 -2.85329074e-01 8.74652505e-01
8.15154850e-01 8.01051334e-02 7.99223065e-01 -3.66795897e-01
-1.38501763e-01 7.26967752e-02 -1.14252841e+00 2.25853160e-01
9.05014753e-01 -5.19784153e-01 -5.80036163e-01 6.36784956e-02
-1.29097462e+00 -1.82668224e-01 4.02591497e-01 2.31033131e-01
3.56614888e-01 -1.22787356e+00 -4.89753962e-01 1.80240318e-01
-5.35184294e-02 -6.91675901e-01 1.30110815e-01 1.65163672e+00
1.06735058e-01 4.01022464e-01 -5.55563390e-01 -4.16048020e-01
-1.34727907e+00 1.77050382e-01 5.15194118e-01 5.44698574e-02
-9.35660183e-01 3.03512633e-01 -2.78124303e-01 2.40089908e-01
4.19370346e-02 -3.12602639e-01 -1.78539187e-01 5.01399696e-01
1.08366168e+00 1.02559090e+00 1.48354813e-01 -6.06524587e-01
-6.73721254e-01 -1.52138963e-01 8.99949819e-02 -3.23026299e-01
1.81878543e+00 -7.07188845e-01 -3.91446084e-01 8.24607790e-01
1.16920698e+00 -7.03286171e-01 -1.09230411e+00 1.46259621e-01
3.12764406e-01 2.70130605e-01 2.79948860e-01 -8.47758830e-01
-5.51809013e-01 7.20509112e-01 1.71854246e+00 4.56723362e-01
7.83147573e-01 5.39895713e-01 1.03627098e+00 5.28332293e-01
3.30788016e-01 -8.33220720e-01 -3.36681485e-01 2.26957165e-02
5.66052914e-01 -6.70397460e-01 1.57925978e-01 5.92269778e-01
1.93403587e-02 7.40139723e-01 -2.85245091e-01 -1.59799546e-01
1.00920546e+00 1.91839114e-01 -7.35206679e-02 -5.58351576e-01
-6.76293075e-01 -4.76485670e-01 -5.04907481e-02 1.26244569e+00
1.25723839e-01 3.35491091e-01 -8.31507668e-02 1.01470661e+00
-4.26233798e-01 4.31903183e-01 3.06899041e-01 1.25762212e+00
-9.12116051e-01 -9.71846461e-01 -3.99738133e-01 1.29988062e+00
-3.41611743e-01 2.14758590e-02 -8.33025634e-01 3.79358798e-01
2.65848964e-01 1.32305479e+00 5.03616810e-01 -1.52257666e-01
4.32734668e-01 6.98521197e-01 2.54967809e-01 -5.64959347e-01
-1.67283505e-01 -3.49673659e-01 5.15524268e-01 -4.39966708e-01
-1.03077888e+00 -1.69392288e+00 -1.04531932e+00 -9.60077047e-02
5.75169444e-01 -4.54987794e-01 2.84294903e-01 1.47930288e+00
6.78086400e-01 4.95636880e-01 -9.20226276e-02 -8.95595610e-01
6.57032132e-02 -1.13170147e+00 -7.53398657e-01 -7.17046708e-02
6.85243070e-01 -8.08524966e-01 -4.72367615e-01 3.50420624e-01]
|
[13.496330261230469, 3.3401854038238525]
|
252e33f9-d187-41e0-86ea-1e8ac53e267b
|
a-generalised-multi-factor-deep-learning
|
2304.10686
| null |
https://arxiv.org/abs/2304.10686v1
|
https://arxiv.org/pdf/2304.10686v1.pdf
|
A generalised multi-factor deep learning electricity load forecasting model for wildfire-prone areas
|
This paper proposes a generalised and robust multi-factor Gated Recurrent Unit (GRU) based Deep Learning (DL) model to forecast electricity load in distribution networks during wildfire seasons. The flexible modelling methods consider data input structure, calendar effects and correlation-based leading temperature conditions. Compared to the regular use of instantaneous temperature, the Mean Absolute Percentage Error (MAPE) is decreased by 30.73% by using the proposed input feature selection and leading temperature relationships. Our model is generalised and applied to eight real distribution networks in Victoria, Australia, during the wildfire seasons of 2015-2020. We demonstrate that the GRU-based model consistently outperforms another DL model, Long Short-Term Memory (LSTM), at every step, giving average improvements in Mean Squared Error (MSE) and MAPE of 10.06% and 12.86%, respectively. The sensitivity to large-scale climate variability in training data sets, e.g. El Ni\~no or La Ni\~na years, is considered to understand the possible consequences for load forecasting performance stability, showing minimal impact. Other factors such as regional poverty rate and large-scale off-peak electricity use are potential factors to further improve forecast performance. The proposed method achieves an average forecast MAPE of around 3%, giving a potential annual energy saving of AU\$80.46 million for the state of Victoria.
|
['David C. H. Wallom', 'Sarah N. Sparrow', 'Weijia Yang']
|
2023-04-21
| null | null | null | null |
['load-forecasting']
|
['miscellaneous']
|
[-8.13434720e-02 -1.38321027e-01 9.18278545e-02 -2.44779810e-01
-3.59992564e-01 -2.31400430e-01 8.22927594e-01 -1.02150008e-01
-3.21752429e-01 1.22710896e+00 1.51704639e-01 -9.11889613e-01
-4.12634134e-01 -1.31826842e+00 -3.85418952e-01 -1.01407027e+00
-7.51566112e-01 1.14570007e-01 -6.98207200e-01 -4.07300860e-01
-7.92410672e-02 6.73487961e-01 -1.27560198e+00 -1.31632194e-01
1.02445543e+00 8.39249611e-01 3.75972599e-01 5.88049054e-01
2.54131049e-01 5.97640634e-01 -8.56357157e-01 3.54585469e-01
2.77068347e-01 -2.28324920e-01 -5.14200747e-01 -4.84132469e-01
-5.20768642e-01 -6.20179176e-01 -8.87291655e-02 5.75201929e-01
8.47296953e-01 5.54049850e-01 6.36060536e-01 -1.07389534e+00
-3.23823303e-01 7.93493509e-01 -5.46323180e-01 4.93013531e-01
7.11349491e-03 6.35047406e-02 5.28739572e-01 -5.91760635e-01
-9.38048288e-02 1.01092124e+00 9.04683530e-01 -7.82415792e-02
-1.31060445e+00 -9.73617017e-01 1.56004885e-02 1.14764489e-01
-1.55646157e+00 -1.75251648e-01 4.95445967e-01 -3.63000274e-01
1.94129109e+00 3.85974139e-01 4.90108848e-01 5.51771224e-01
8.14666450e-01 1.02609433e-01 1.26394880e+00 -4.12813127e-01
2.29744151e-01 -1.99155942e-01 -2.96431690e-01 -2.32108533e-02
2.04664096e-02 5.31341493e-01 1.20837875e-02 -9.61241405e-03
5.06598532e-01 -5.82345005e-04 -1.13305792e-01 8.25728655e-01
-8.44821811e-01 8.93087268e-01 7.32251406e-01 3.92389089e-01
-7.80864298e-01 1.72617033e-01 3.66601616e-01 4.10259128e-01
1.01742828e+00 7.50919878e-02 -7.84603119e-01 -1.80406287e-01
-1.34980893e+00 3.06207299e-01 6.39618814e-01 5.03801823e-01
5.56502104e-01 9.57466483e-01 7.52004758e-02 7.36016870e-01
3.00766170e-01 1.18225145e+00 3.71338099e-01 -5.84253132e-01
4.57971662e-01 1.51782945e-01 3.17217767e-01 -9.68964696e-01
-1.00311804e+00 -8.31046700e-01 -1.51538754e+00 2.43381724e-01
5.08877449e-02 -7.75590181e-01 -9.02386308e-01 1.74800825e+00
-3.36926877e-01 -4.58557624e-03 1.18970774e-01 4.86781985e-01
1.69189826e-01 1.33295071e+00 2.49543756e-01 -6.50510848e-01
8.51157248e-01 -3.70231479e-01 -7.51523197e-01 -1.83845058e-01
7.67512560e-01 -5.23874819e-01 3.78336817e-01 5.35330661e-02
-7.81342447e-01 -4.29697603e-01 -8.04547548e-01 7.39540160e-01
-8.96579862e-01 -1.51729994e-02 3.00523371e-01 5.68130672e-01
-1.10141206e+00 9.09913957e-01 -8.57005477e-01 -4.85450268e-01
-3.70627716e-02 3.46894354e-01 3.26130718e-01 3.85128796e-01
-1.84248412e+00 1.33147049e+00 4.72172737e-01 9.78480577e-01
-5.58775961e-01 -8.74556720e-01 -7.61991382e-01 3.86263162e-01
-3.46764624e-01 -2.73077250e-01 8.27588320e-01 -5.56868732e-01
-1.14231575e+00 -1.05132367e-02 -1.78908110e-01 -8.19304943e-01
3.22189122e-01 -3.02225277e-02 -9.40080345e-01 -5.18272340e-01
-9.98037606e-02 2.95572758e-01 1.82781190e-01 -8.10896754e-01
-5.94952822e-01 -1.62931353e-01 -4.42460537e-01 2.51702160e-01
9.66838449e-02 2.15026721e-01 8.84066522e-01 -8.58202994e-01
-3.49839717e-01 -8.00991714e-01 -6.17238581e-01 -9.89121675e-01
8.38396177e-02 -2.99634337e-01 7.20950007e-01 -1.22295141e+00
1.48885393e+00 -1.50605810e+00 -3.31477910e-01 5.95127404e-01
-5.09704232e-01 2.28673518e-01 1.90965906e-01 6.95963383e-01
-3.29336017e-01 3.08909923e-01 -5.04411221e-01 2.19853237e-01
3.04502361e-02 6.58552110e-01 -4.97073114e-01 3.65098983e-01
2.94230253e-01 9.38178658e-01 -8.78903329e-01 5.62162697e-01
6.41896844e-01 7.23880649e-01 2.76377857e-01 -3.07907853e-02
3.86452228e-02 2.55828321e-01 3.84789854e-02 3.87332678e-01
9.54132318e-01 2.23479524e-01 1.08188510e-01 2.84100026e-01
-5.05810618e-01 2.32676953e-01 -1.01614356e+00 9.17905092e-01
-8.60537887e-01 7.24455714e-01 -1.86621919e-01 -9.64337528e-01
1.36754596e+00 4.92525458e-01 3.00414383e-01 -1.18416321e+00
-2.16260299e-01 5.09316385e-01 1.96074232e-01 -3.96813080e-02
4.27225888e-01 -2.91458458e-01 5.33637404e-02 7.80097663e-01
-3.08460265e-01 1.86394170e-01 1.00824535e-01 -2.79335111e-01
4.41903472e-01 8.31551403e-02 1.35989100e-01 -9.50861096e-01
2.42277429e-01 -3.88080418e-01 7.90486872e-01 5.20658910e-01
1.94447190e-02 1.14505254e-01 2.67173558e-01 -8.30883801e-01
-1.11705959e+00 -7.91552544e-01 -5.87892592e-01 9.70296204e-01
-7.30818748e-01 -2.23978311e-02 -1.63743526e-01 1.94295752e-03
5.90348616e-02 1.47143424e+00 -5.19224524e-01 -1.67685617e-02
-7.22811997e-01 -1.74700522e+00 5.32539725e-01 7.17202187e-01
6.37291789e-01 -1.36828113e+00 -9.05644715e-01 7.03113616e-01
-9.43195969e-02 -4.68843281e-01 2.69303977e-01 7.91427374e-01
-8.21983635e-01 -2.62684405e-01 -9.91066158e-01 -2.06516787e-01
4.78368461e-01 -2.28802085e-01 1.39463055e+00 -3.42509151e-01
1.10728383e-01 -2.44473994e-01 -1.43683136e-01 -4.63461906e-01
-6.44872710e-02 2.37017155e-01 2.47623548e-01 -6.23889089e-01
4.10780579e-01 -8.31540167e-01 -5.34759223e-01 1.89947143e-01
-7.20317960e-01 -5.07430881e-02 3.44241410e-01 8.04867387e-01
6.92759007e-02 4.08200771e-01 1.32148147e+00 -4.34699327e-01
7.39613771e-01 -8.74650121e-01 -5.79550028e-01 1.37253910e-01
-1.29688704e+00 -1.33307323e-01 6.83676183e-01 1.20606281e-01
-1.15346205e+00 -2.37731993e-01 -5.75194806e-02 3.76506262e-02
-1.97597548e-01 1.11817908e+00 2.91058868e-01 4.03668642e-01
3.61190826e-01 3.64945859e-01 -3.71815830e-01 -4.35040832e-01
4.24425043e-02 6.68888628e-01 4.51833308e-01 -2.96863705e-01
7.20279872e-01 -1.24234438e-01 4.15919386e-02 -7.12566316e-01
-2.01325104e-01 -1.06977541e-02 -6.27450943e-01 -4.78877842e-01
4.54879880e-01 -1.23681617e+00 -4.98256445e-01 8.17560673e-01
-8.51041436e-01 -7.42411077e-01 -3.78635935e-02 5.16084909e-01
-2.51714170e-01 -3.35141480e-01 -4.75164205e-01 -1.33025539e+00
-8.52797747e-01 -8.01019847e-01 4.22280043e-01 2.29562208e-01
-1.35311499e-01 -1.42802954e+00 1.37540460e-01 -3.05691540e-01
1.10840023e+00 7.79792249e-01 1.01338255e+00 -2.35314742e-01
2.01657906e-01 9.65942144e-02 -2.51498818e-01 4.04101282e-01
4.56654161e-01 -1.93142109e-02 -1.12612009e+00 -6.38849974e-01
-2.80936331e-01 1.18682832e-01 7.67503262e-01 6.31868184e-01
8.40436399e-01 -5.32662630e-01 -1.36435833e-02 4.85102713e-01
1.67934787e+00 6.43789470e-01 7.41502225e-01 4.83902723e-01
2.69272655e-01 2.77585179e-01 1.60403654e-01 7.72834063e-01
3.99618149e-01 6.74397796e-02 3.65462393e-01 -4.79507238e-01
2.75514483e-01 2.51839340e-01 4.73806500e-01 8.30610454e-01
-4.31494623e-01 -2.12228730e-01 -1.33648944e+00 8.33884299e-01
-1.66661215e+00 -1.17028177e+00 -1.82961673e-01 2.26335883e+00
4.19787705e-01 1.34568751e-01 8.43478814e-02 3.38352710e-01
4.96410400e-01 5.46352088e-01 -4.67012286e-01 -1.31390381e+00
-3.85128468e-01 4.75748003e-01 1.09703553e+00 5.82449794e-01
-9.07099664e-01 3.35999668e-01 6.39293242e+00 5.76448917e-01
-1.25813746e+00 -2.26506010e-01 1.05846453e+00 -2.14948040e-02
-2.54465640e-01 -2.87537098e-01 -4.17654812e-01 4.96541768e-01
1.94039524e+00 -5.72396457e-01 6.65918171e-01 2.56171435e-01
1.17612731e+00 -3.90228927e-01 -3.22107404e-01 3.10849935e-01
-3.69765788e-01 -9.21841145e-01 -1.76606953e-01 3.99123967e-01
9.57877219e-01 6.38049662e-01 -1.74761191e-01 5.18850207e-01
5.41412413e-01 -1.47119534e+00 2.86046803e-01 8.01793635e-01
9.13488925e-01 -1.51196718e+00 1.19934881e+00 3.48847151e-01
-1.46192598e+00 -2.40020692e-01 -4.53671515e-01 -7.59409070e-01
3.22601467e-01 9.65720654e-01 -6.06766403e-01 1.05258822e+00
9.76332426e-01 8.35975349e-01 -1.59921288e-01 4.95491624e-01
4.85932194e-02 1.02386427e+00 -8.79239440e-01 2.28268504e-01
6.63262427e-01 -3.77519608e-01 2.07182944e-01 1.33655727e+00
4.83709663e-01 4.21063267e-02 -1.11334726e-01 6.15661323e-01
1.95725545e-01 -2.70559609e-01 -6.38730228e-01 4.40619916e-01
4.78513956e-01 1.06381834e+00 -3.95796686e-01 -1.79906458e-01
-2.78443635e-01 5.49575627e-01 -2.28403300e-01 7.72781610e-01
-7.43224978e-01 -6.19989932e-01 6.31647468e-01 -1.78915188e-01
2.34367803e-01 -2.91199744e-01 -4.83708054e-01 -7.50792265e-01
-1.03426710e-01 -3.54870796e-01 1.92187294e-01 -7.67804980e-01
-9.40169990e-01 4.90211368e-01 1.04842223e-01 -9.64098334e-01
-8.50629985e-01 -1.41388208e-01 -1.18751299e+00 1.73489296e+00
-1.95365870e+00 -1.11325371e+00 2.58919358e-01 1.84725538e-01
4.14080083e-01 -8.61204136e-03 1.23298287e+00 2.56260544e-01
-8.25943649e-01 2.68953860e-01 9.70386267e-01 5.04363477e-02
-2.14855354e-02 -1.26425755e+00 6.89241111e-01 9.07622337e-01
-6.29458368e-01 3.28604758e-01 6.43916965e-01 -5.72111726e-01
-8.27996790e-01 -1.56176817e+00 1.27840221e+00 2.50657618e-01
4.87089217e-01 -3.10331464e-01 -9.16966200e-01 7.47759581e-01
5.99624336e-01 -3.36438268e-01 5.04558802e-01 2.10005924e-01
8.88861492e-02 -2.92174727e-01 -1.39649951e+00 1.45573884e-01
2.99594611e-01 -3.69635224e-01 -2.16837019e-01 1.45496622e-01
1.84296489e-01 1.21234000e-01 -1.44487202e+00 7.86401749e-01
7.73738563e-01 -5.47202706e-01 5.88960767e-01 -3.31352085e-01
-6.30041361e-02 -1.94733933e-01 -3.13819677e-01 -2.00287771e+00
-6.69398069e-01 -5.76777458e-01 1.05597433e-02 1.34584188e+00
7.18958974e-01 -9.82035100e-01 1.39920384e-01 7.22384214e-01
4.57685702e-02 -7.99766362e-01 -1.27324831e+00 -6.98128045e-01
7.16043293e-01 -6.58399940e-01 9.10826981e-01 9.99647856e-01
-7.91368727e-03 -8.67091939e-02 -5.32420635e-01 3.17870885e-01
3.68593216e-01 1.60864085e-01 4.72350866e-02 -1.03982937e+00
4.36959803e-01 -4.27663088e-01 -6.05291873e-02 -2.17092991e-01
1.44433081e-01 -7.08411753e-01 -3.92049193e-01 -1.66006064e+00
-2.11900637e-01 -4.11481112e-01 -8.13243508e-01 8.26255441e-01
4.60230187e-02 -6.29963260e-03 2.09539071e-01 -3.28267626e-02
4.48844582e-01 8.89402926e-01 4.48092639e-01 -7.09205121e-02
-2.21789852e-01 2.35040486e-01 -6.03560396e-02 3.03814024e-01
1.49704242e+00 -2.24048555e-01 -1.16426289e-01 -4.10117626e-01
4.41626519e-01 2.00354643e-02 1.21489152e-01 -9.57756758e-01
-1.56204090e-01 -4.27105457e-01 8.70173991e-01 -9.32816863e-01
-3.55773270e-01 -7.58464634e-01 6.74472451e-01 8.31702352e-01
-1.29698114e-02 6.73516333e-01 6.62098825e-01 1.33763760e-01
8.39988217e-02 2.81741947e-01 4.70658004e-01 -9.35033858e-02
-5.93523979e-01 -1.31354317e-01 -9.90595520e-01 -6.13809705e-01
7.57910252e-01 -1.51800353e-03 -2.52145827e-01 -4.78878915e-01
-7.27873027e-01 7.72989392e-01 -6.05935045e-02 4.44183260e-01
1.22508407e-01 -1.32185113e+00 -1.21323323e+00 3.34439278e-01
-4.61349934e-01 -2.12151170e-01 2.12142438e-01 5.21791041e-01
-3.32619309e-01 7.17527390e-01 -1.85216978e-01 -3.01509857e-01
-4.58017230e-01 -2.77414396e-02 8.41551602e-01 -4.49547797e-01
-4.88049239e-01 3.83851618e-01 -4.70774114e-01 -8.07269812e-01
-2.14365050e-01 -6.15148902e-01 -7.76809230e-02 5.06854057e-01
4.26606089e-01 8.24064016e-01 4.98713940e-01 -7.28720069e-01
-4.13975060e-01 3.58472019e-01 4.15712237e-01 -4.38744687e-02
1.68484914e+00 -3.89159739e-01 -1.79871634e-01 8.08209658e-01
9.94964898e-01 -8.35927844e-01 -1.05795145e+00 -3.76129672e-02
9.43513215e-02 5.80315553e-02 6.67980194e-01 -1.49347830e+00
-1.35375130e+00 7.20004261e-01 1.04382932e+00 3.30616504e-01
1.53028452e+00 -7.40712047e-01 7.60603487e-01 2.60135084e-01
1.41076356e-01 -1.04503548e+00 -1.39743817e+00 6.86147511e-01
1.04063344e+00 -8.77057433e-01 1.13988988e-01 7.77428210e-01
-1.83209270e-01 1.19039404e+00 5.41632697e-02 6.23674877e-03
1.01830447e+00 4.67232764e-01 5.77773377e-02 1.90716982e-01
-9.75655317e-01 -1.33853540e-01 -1.41869754e-01 4.59388018e-01
5.15287697e-01 8.01812053e-01 -3.59162211e-01 2.31464416e-01
-3.51783395e-01 7.35951811e-02 4.38203007e-01 7.31375813e-01
-2.93206453e-01 -5.21016300e-01 -5.54492235e-01 9.28141713e-01
-5.20630836e-01 -2.94970900e-01 4.04064536e-01 5.74805856e-01
-5.09151965e-02 1.11357164e+00 5.51393390e-01 -2.33672962e-01
1.53739959e-01 2.79446274e-01 -3.25225085e-01 2.57945340e-02
-8.88789833e-01 2.97952145e-01 2.24696398e-02 -1.61272749e-01
-5.68147242e-01 -8.33779216e-01 -1.07624483e+00 -9.00640905e-01
-4.17787552e-01 1.74769998e-01 8.16638172e-01 1.07510638e+00
2.56768197e-01 6.83196366e-01 1.20893216e+00 -1.12641132e+00
-3.32961351e-01 -1.53021121e+00 -8.63079727e-01 -3.22921455e-01
4.39188987e-01 -3.56767088e-01 -6.57869935e-01 -4.92615730e-01]
|
[6.231354236602783, 2.8554182052612305]
|
131dc278-a690-4205-8119-770cacb974d9
|
learning-spatial-temporal-implicit-neural
|
2303.13767
| null |
https://arxiv.org/abs/2303.13767v2
|
https://arxiv.org/pdf/2303.13767v2.pdf
|
Learning Spatial-Temporal Implicit Neural Representations for Event-Guided Video Super-Resolution
|
Event cameras sense the intensity changes asynchronously and produce event streams with high dynamic range and low latency. This has inspired research endeavors utilizing events to guide the challenging video superresolution (VSR) task. In this paper, we make the first attempt to address a novel problem of achieving VSR at random scales by taking advantages of the high temporal resolution property of events. This is hampered by the difficulties of representing the spatial-temporal information of events when guiding VSR. To this end, we propose a novel framework that incorporates the spatial-temporal interpolation of events to VSR in a unified framework. Our key idea is to learn implicit neural representations from queried spatial-temporal coordinates and features from both RGB frames and events. Our method contains three parts. Specifically, the Spatial-Temporal Fusion (STF) module first learns the 3D features from events and RGB frames. Then, the Temporal Filter (TF) module unlocks more explicit motion information from the events near the queried timestamp and generates the 2D features. Lastly, the SpatialTemporal Implicit Representation (STIR) module recovers the SR frame in arbitrary resolutions from the outputs of these two modules. In addition, we collect a real-world dataset with spatially aligned events and RGB frames. Extensive experiments show that our method significantly surpasses the prior-arts and achieves VSR with random scales, e.g., 6.5. Code and dataset are available at https: //vlis2022.github.io/cvpr23/egvsr.
|
['Lin Wang', 'Hongjian Wang', 'Minjie Liu', 'Zipeng Wang', 'Yunfan Lu']
|
2023-03-24
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Lu_Learning_Spatial-Temporal_Implicit_Neural_Representations_for_Event-Guided_Video_Super-Resolution_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Lu_Learning_Spatial-Temporal_Implicit_Neural_Representations_for_Event-Guided_Video_Super-Resolution_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['video-super-resolution']
|
['computer-vision']
|
[ 3.02495360e-01 -5.11542559e-01 -2.24877466e-02 -2.87945956e-01
-9.35972750e-01 -4.14840549e-01 5.95216036e-01 -2.30885103e-01
-3.11251581e-01 6.36128247e-01 3.16860467e-01 1.92741603e-01
-2.98913959e-02 -9.08262432e-01 -8.10053647e-01 -7.43040800e-01
-5.50590828e-02 -2.22892210e-01 5.50071001e-01 -1.20523795e-01
4.42222841e-02 7.60060728e-01 -1.80394399e+00 4.66511607e-01
5.88035822e-01 1.29278755e+00 3.88405204e-01 8.47495496e-01
1.42665699e-01 1.11352634e+00 -3.04005295e-01 1.59385517e-01
2.68353671e-01 -3.74939024e-01 -6.08782709e-01 -3.48241962e-02
2.29035735e-01 -6.66045666e-01 -9.33486223e-01 7.09143341e-01
5.04830062e-01 5.38696885e-01 9.14487056e-03 -1.18089342e+00
-5.09196877e-01 1.79493591e-01 -7.53257930e-01 5.07735789e-01
6.12343490e-01 2.53259480e-01 5.78955472e-01 -1.02564871e+00
7.27833092e-01 1.06413734e+00 4.66583341e-01 3.96794498e-01
-1.02175117e+00 -7.11387396e-01 2.29656637e-01 3.88888359e-01
-1.76950836e+00 -5.90465307e-01 7.60499120e-01 -1.84180126e-01
8.50090086e-01 3.13879281e-01 6.79926276e-01 9.89225864e-01
-1.07548073e-01 6.62182689e-01 9.82073009e-01 8.14910010e-02
2.96532005e-01 -5.12378097e-01 -1.46942914e-01 3.64721626e-01
-1.79848790e-01 2.46532083e-01 -1.17212415e+00 -3.35933291e-04
1.39913034e+00 4.01089698e-01 -5.53787112e-01 7.35581294e-02
-1.43309677e+00 4.53578800e-01 5.93296885e-01 2.80582815e-01
-6.62938297e-01 4.60074961e-01 1.07783146e-01 7.37354979e-02
4.37406659e-01 -2.30538115e-01 -3.99532259e-01 -3.23704541e-01
-1.07097149e+00 6.26050830e-02 3.73680592e-01 8.90363336e-01
8.98221672e-01 1.29797697e-01 -2.98222713e-02 4.43448305e-01
1.01725176e-01 4.56754118e-01 3.35432351e-01 -1.45314193e+00
2.34655693e-01 2.01613963e-01 3.94552261e-01 -9.63765085e-01
-1.60145298e-01 1.34339377e-01 -8.67562354e-01 1.46962017e-01
3.37736040e-01 1.07652940e-01 -8.33788455e-01 1.60919142e+00
5.28966725e-01 9.53052521e-01 2.06758808e-02 1.17767930e+00
1.00522339e+00 1.03571796e+00 4.39620130e-02 -4.04874325e-01
1.22377050e+00 -5.37891984e-01 -7.58077323e-01 7.61578158e-02
-1.32583037e-01 -6.14765644e-01 8.05003643e-01 2.36141890e-01
-1.33064866e+00 -7.21225202e-01 -1.03365481e+00 -4.67798382e-01
-1.61983773e-01 -7.72345513e-02 5.66014647e-01 -2.68931717e-01
-1.15901768e+00 6.53152585e-01 -1.26049256e+00 -1.85727671e-01
2.68936992e-01 2.95303553e-01 -2.58867353e-01 -8.73245150e-02
-1.28131330e+00 3.23858052e-01 2.05998346e-01 1.14434280e-01
-8.54914486e-01 -6.87959611e-01 -8.26875806e-01 -1.70526639e-01
2.90991902e-01 -6.01110458e-01 1.23816407e+00 -1.05923176e+00
-1.53594697e+00 6.54893339e-01 -6.26078010e-01 -3.33241582e-01
4.18551117e-01 -2.29852617e-01 -3.51961702e-01 5.70106268e-01
5.87746454e-03 5.18489897e-01 8.83720398e-01 -1.10013449e+00
-8.62703085e-01 -3.17413509e-01 3.27134579e-01 2.66979903e-01
3.04813515e-02 2.91421656e-02 -8.15421283e-01 -8.07913065e-01
3.21438283e-01 -6.53352916e-01 -1.82384327e-01 2.05297083e-01
9.49032158e-02 3.37954275e-02 6.38162017e-01 -7.31652558e-01
1.03944397e+00 -2.40938473e+00 2.81353623e-01 -1.75971061e-01
2.88235098e-01 -3.31577212e-02 3.49259861e-02 1.47012949e-01
-7.55784288e-02 -2.94558138e-01 6.78633898e-03 -3.48138779e-01
-3.59708279e-01 2.06514940e-01 -7.11813450e-01 5.18519521e-01
1.55697197e-01 7.93178380e-01 -1.14736450e+00 -4.75377560e-01
6.15338624e-01 1.12785137e+00 -3.11177164e-01 3.70657086e-01
-7.13655651e-02 8.49012375e-01 -6.12134159e-01 6.75747037e-01
5.68989515e-01 -4.53665018e-01 -1.24970235e-01 -5.77222764e-01
-5.23639977e-01 4.04755205e-01 -1.47652292e+00 2.03801823e+00
-2.69558400e-01 6.12435877e-01 -2.76986137e-02 -5.91252625e-01
7.42494404e-01 4.86672431e-01 9.14080203e-01 -8.62242877e-01
-1.95159912e-02 1.10519722e-01 -8.58384073e-01 -4.57815319e-01
7.89163470e-01 1.18452892e-01 8.19059741e-03 1.35413170e-01
4.38190028e-02 1.06409736e-01 -3.49078588e-02 3.28082323e-01
1.11623776e+00 5.93589604e-01 4.00046319e-01 3.53723109e-01
3.84126157e-01 -1.11038655e-01 7.72191048e-01 5.24706483e-01
-2.34826177e-01 9.40062881e-01 -9.35984552e-02 -6.50328696e-01
-9.40983772e-01 -1.44041514e+00 6.82098279e-03 8.39376032e-01
6.37442648e-01 -4.19691980e-01 -3.16525847e-01 -9.88194048e-02
-2.54677534e-01 4.95528251e-01 -6.26456499e-01 7.86581263e-02
-9.47806239e-01 -5.07214248e-01 1.90204903e-01 7.60249138e-01
6.12159073e-01 -9.00543630e-01 -8.88118207e-01 2.77482182e-01
-5.84082365e-01 -1.37752295e+00 -5.28501809e-01 -1.10239759e-01
-8.61474872e-01 -8.25425804e-01 -5.74388087e-01 -3.37253332e-01
2.74637669e-01 6.85119331e-01 1.02145350e+00 -2.02736869e-01
-2.90741533e-01 6.00884259e-01 -4.90348101e-01 5.47574013e-02
1.94217905e-01 -3.25837493e-01 -3.07818875e-02 3.07287872e-01
2.84537673e-01 -9.28755283e-01 -9.11441505e-01 2.56907254e-01
-1.05126786e+00 4.37739193e-01 2.52760798e-01 3.81580234e-01
1.13782012e+00 -3.59130949e-02 4.02258813e-01 -1.97888181e-01
-1.62643269e-01 -5.96544743e-01 -6.15771592e-01 -5.08148745e-02
1.01386078e-01 -2.27607116e-01 5.72030187e-01 -6.27390265e-01
-1.15826142e+00 3.09353203e-01 4.21523303e-03 -7.90191233e-01
-1.84416637e-01 4.37412113e-02 -1.47301853e-01 1.73647583e-01
4.78764623e-01 5.28599977e-01 -4.41747904e-01 -3.04230541e-01
4.68933046e-01 4.88032162e-01 9.40264225e-01 -5.80890894e-01
7.58405507e-01 1.19019198e+00 -2.30555117e-01 -6.96359098e-01
-8.12662005e-01 -4.19001877e-01 -4.66322243e-01 -4.66156483e-01
9.38412011e-01 -1.35436428e+00 -7.07270324e-01 3.67064118e-01
-1.15258956e+00 -5.21184802e-01 -6.10385478e-01 5.53076267e-01
-7.98623443e-01 2.05991134e-01 -9.06199217e-01 -7.00590611e-01
-2.60710061e-01 -1.00115883e+00 1.40912294e+00 6.02680922e-01
-5.88384233e-02 -4.73986149e-01 -2.54702033e-03 7.63175040e-02
3.83899957e-01 6.15687072e-01 -1.76099017e-01 1.17344342e-01
-1.25159192e+00 1.29041284e-01 -4.22702014e-01 -1.45363063e-01
2.32995793e-01 3.95088941e-02 -1.01874685e+00 -2.29860947e-01
2.08605662e-01 -6.85278550e-02 8.17446351e-01 5.76412678e-01
1.20914400e+00 -8.99150521e-02 -1.21953741e-01 1.11695635e+00
1.55587852e+00 2.30089575e-01 6.99313283e-01 3.83863062e-01
7.53077149e-01 2.96064913e-01 7.62161314e-01 8.45887899e-01
6.61753535e-01 8.26851130e-01 3.70325416e-01 3.22006643e-02
-4.33747709e-01 -2.53435940e-01 5.48008442e-01 6.11966014e-01
-5.20019889e-01 4.31608828e-03 -5.28422713e-01 5.53378463e-01
-2.08570552e+00 -1.17419159e+00 2.36792937e-02 2.14907122e+00
9.77108717e-01 -2.26570234e-01 -6.70225695e-02 7.83552527e-02
7.40436852e-01 5.44178009e-01 -6.83236957e-01 1.09964229e-01
-2.81781495e-01 1.46991923e-01 4.44543272e-01 4.81789231e-01
-9.91227090e-01 8.21264744e-01 5.00794029e+00 6.67190969e-01
-1.21230531e+00 2.31177211e-01 3.64195228e-01 -6.18776083e-01
-1.44486517e-01 -2.46804524e-02 -7.26012111e-01 5.73787630e-01
1.01531804e+00 -2.24793419e-01 8.91308427e-01 3.59062999e-01
4.76265460e-01 -1.45651639e-01 -1.02164674e+00 1.29141343e+00
-1.39744863e-01 -1.49939549e+00 -1.61338881e-01 -2.30810106e-01
6.44809783e-01 2.23754779e-01 3.95317338e-02 -7.59330764e-02
4.46554601e-01 -7.91417420e-01 1.03183603e+00 9.18526232e-01
1.15639341e+00 -6.51610792e-01 1.83592558e-01 1.65009946e-02
-1.82750070e+00 7.67094046e-02 -2.04184085e-01 -9.87483561e-02
5.50836682e-01 6.93781853e-01 -3.25038224e-01 7.62060761e-01
1.15426600e+00 1.16784942e+00 -2.04604611e-01 6.77408993e-01
-1.91254094e-01 3.23529541e-01 -5.59612930e-01 6.43422246e-01
-4.44982797e-02 -1.00325227e-01 5.41229129e-01 9.95761275e-01
5.34552395e-01 7.03731179e-01 -4.90966765e-03 7.93668449e-01
7.80386329e-02 -3.94185185e-01 -4.80913252e-01 2.04472810e-01
7.34459400e-01 1.27121282e+00 -6.03930354e-01 -4.83701289e-01
-6.38829887e-01 1.19246578e+00 2.36008659e-01 6.04809582e-01
-1.20247626e+00 -1.26549378e-01 7.47526824e-01 5.15226983e-02
4.10858363e-01 -5.28108299e-01 -6.17433712e-02 -1.53996801e+00
1.53811276e-01 -5.52098632e-01 4.87130046e-01 -1.15005100e+00
-1.06820881e+00 6.68594778e-01 -6.55992925e-02 -1.38171530e+00
-1.96825385e-01 -1.06536843e-01 -3.31869662e-01 7.55897880e-01
-1.70821750e+00 -1.09407747e+00 -8.67027819e-01 1.16898835e+00
6.60311043e-01 4.75254416e-01 4.35576141e-01 4.77965862e-01
-4.65745628e-01 3.34780850e-02 -1.55905783e-01 1.44862413e-01
7.07016706e-01 -9.58547413e-01 3.49624932e-01 9.89907086e-01
2.59125475e-02 3.03102970e-01 6.15023494e-01 -5.29496551e-01
-1.62120676e+00 -1.27220929e+00 6.25836074e-01 -3.96799952e-01
6.53290927e-01 -7.14368969e-02 -9.08722520e-01 8.46478164e-01
-1.13018349e-01 5.84984779e-01 3.00047129e-01 -5.94216228e-01
-3.11776936e-01 -3.61662030e-01 -9.79626060e-01 5.49385965e-01
1.26937735e+00 -8.12837124e-01 -4.35120314e-01 -2.48799883e-02
9.57270145e-01 -7.38459706e-01 -1.02402794e+00 4.69216228e-01
4.76766199e-01 -9.91007686e-01 1.40985823e+00 -3.56527530e-02
3.77930105e-01 -8.54918063e-01 -6.45369053e-01 -6.45113051e-01
-1.16921343e-01 -6.27706707e-01 -6.34813011e-01 1.23309088e+00
-3.70490283e-01 -5.65738022e-01 4.23340321e-01 6.62306309e-01
9.89791751e-03 -6.06529474e-01 -1.04323697e+00 -5.02325416e-01
-5.15401304e-01 -6.27462864e-01 6.53851688e-01 8.79604459e-01
-3.62671614e-01 -1.35230899e-01 -4.11548138e-01 5.84691584e-01
8.56878400e-01 5.41225195e-01 6.28511429e-01 -7.92415917e-01
-2.90411949e-01 1.05592199e-02 -2.99727708e-01 -1.32646549e+00
-1.93916053e-01 -5.33342838e-01 7.86869600e-02 -1.40506375e+00
6.68308288e-02 -3.18079144e-01 -4.37532365e-01 3.10984910e-01
-1.97222158e-01 4.46165442e-01 3.44003886e-01 4.29591298e-01
-8.74881506e-01 5.33295333e-01 1.05328906e+00 2.65516847e-01
-3.25975060e-01 -4.63899940e-01 -3.64500940e-01 7.64577687e-01
8.76936436e-01 -2.66134351e-01 -2.26909921e-01 -6.76692784e-01
1.18228294e-01 5.98047793e-01 8.17395627e-01 -1.03394437e+00
4.41815972e-01 -2.97262937e-01 6.65353060e-01 -8.32418025e-01
7.55859077e-01 -7.55270720e-01 5.99138141e-01 -4.24553901e-02
-1.61412865e-01 1.44550234e-01 1.01973146e-01 6.07058883e-01
-2.07768679e-01 4.74306911e-01 7.71947563e-01 -1.55954361e-01
-1.03970003e+00 6.95472300e-01 -1.48688465e-01 -8.27386901e-02
9.42857921e-01 -2.16950476e-01 -2.61479944e-01 -3.68228376e-01
-7.48593569e-01 1.28727868e-01 7.61174738e-01 4.51787502e-01
8.98241043e-01 -1.41570473e+00 -5.69711626e-01 1.88607126e-01
-2.09920242e-01 4.03431535e-01 6.60700262e-01 9.03864980e-01
-3.81019115e-01 1.51660545e-02 -9.07488838e-02 -7.87927985e-01
-9.68406081e-01 6.54899538e-01 2.39336282e-01 2.74862554e-02
-1.00600016e+00 7.21221387e-01 2.79705375e-01 2.18101382e-01
1.40372604e-01 -2.37620473e-01 -5.34140319e-02 -8.03270787e-02
1.08857250e+00 5.21545827e-01 -3.45090210e-01 -9.07707274e-01
-2.96322763e-01 7.08674371e-01 2.21227303e-01 -2.77965546e-01
1.57722270e+00 -5.13426781e-01 1.26653397e-03 6.09023750e-01
1.23031008e+00 -8.03561509e-02 -1.91435933e+00 -5.71657717e-01
-3.69576484e-01 -7.58431375e-01 3.31989862e-02 -3.10396671e-01
-1.21750116e+00 6.73124790e-01 5.87969363e-01 3.78976502e-02
1.72389364e+00 7.89721608e-02 1.11464691e+00 -1.64811313e-01
7.26888895e-01 -7.80574858e-01 5.50125018e-02 2.41601750e-01
7.56181717e-01 -1.01603544e+00 -3.37069333e-02 -2.81026453e-01
-4.02907163e-01 1.11903548e+00 4.49993908e-01 -3.33316237e-01
4.03314680e-01 5.11551321e-01 -2.05201671e-01 -1.55358845e-02
-8.05164039e-01 -3.18915606e-01 3.33560854e-02 4.42113578e-01
1.88928083e-01 -8.14424083e-02 5.95414303e-02 5.67916393e-01
1.48409024e-01 5.21826029e-01 3.98209721e-01 9.94089544e-01
-2.63820380e-01 -6.38240457e-01 -4.37887609e-01 -6.57641813e-02
-4.00738448e-01 9.95065942e-02 2.24498376e-01 4.80639637e-01
1.93906754e-01 8.28412950e-01 1.70731634e-01 -4.08018649e-01
2.92199939e-01 -3.07605803e-01 4.19308424e-01 -8.88691321e-02
-1.56251445e-01 1.43500313e-01 -3.42479318e-01 -1.38456178e+00
-8.42317104e-01 -7.82560289e-01 -1.68557012e+00 -3.49532932e-01
1.98106721e-01 -1.78105280e-01 4.83279854e-01 5.28994501e-01
5.38844705e-01 7.38750815e-01 7.65828371e-01 -1.46823168e+00
5.16228937e-02 -4.08967853e-01 -4.21937734e-01 4.97031420e-01
6.92339480e-01 -4.65203851e-01 -4.88443673e-01 5.33610761e-01]
|
[10.73806095123291, -1.772188425064087]
|
cca16b43-b56e-479c-8d96-f28a2f72ad03
|
the-unfairness-of-fair-machine-learning
|
2302.02404
| null |
https://arxiv.org/abs/2302.02404v3
|
https://arxiv.org/pdf/2302.02404v3.pdf
|
The Unfairness of Fair Machine Learning: Levelling down and strict egalitarianism by default
|
In recent years fairness in machine learning (ML) has emerged as a highly active area of research and development. Most define fairness in simple terms, where fairness means reducing gaps in performance or outcomes between demographic groups while preserving as much of the accuracy of the original system as possible. This oversimplification of equality through fairness measures is troubling. Many current fairness measures suffer from both fairness and performance degradation, or "levelling down," where fairness is achieved by making every group worse off, or by bringing better performing groups down to the level of the worst off. When fairness can only be achieved by making everyone worse off in material or relational terms through injuries of stigma, loss of solidarity, unequal concern, and missed opportunities for substantive equality, something would appear to have gone wrong in translating the vague concept of 'fairness' into practice. This paper examines the causes and prevalence of levelling down across fairML, and explore possible justifications and criticisms based on philosophical and legal theories of equality and distributive justice, as well as equality law jurisprudence. We find that fairML does not currently engage in the type of measurement, reporting, or analysis necessary to justify levelling down in practice. We propose a first step towards substantive equality in fairML: "levelling up" systems by design through enforcement of minimum acceptable harm thresholds, or "minimum rate constraints," as fairness constraints. We likewise propose an alternative harms-based framework to counter the oversimplified egalitarian framing currently dominant in the field and push future discussion more towards substantive equality opportunities and away from strict egalitarianism by default. N.B. Shortened abstract, see paper for full abstract.
|
['Chris Russell', 'Sandra Wachter', 'Brent Mittelstadt']
|
2023-02-05
| null | null | null | null |
['jurisprudence']
|
['miscellaneous']
|
[ 1.40757829e-01 4.37788844e-01 -6.00197136e-01 -8.23434472e-01
-2.62631059e-01 -5.04312754e-01 6.56744838e-01 5.75290143e-01
-8.51504445e-01 1.14041805e+00 9.22310352e-01 -8.03167462e-01
-5.05405247e-01 -7.71678865e-01 -2.10884079e-01 -4.52763677e-01
6.62434816e-01 2.43533894e-01 -6.69249535e-01 -2.48236060e-01
6.48047447e-01 3.87243718e-01 -1.35598302e+00 5.14766052e-02
1.24885833e+00 1.94283992e-01 -6.95841432e-01 2.25440115e-01
-2.02858165e-01 1.20445383e+00 -6.29704833e-01 -1.03435862e+00
4.64895874e-01 -8.08626413e-01 -9.65800881e-01 -4.92249340e-01
8.40940118e-01 -7.46466398e-01 -5.38093597e-02 1.14206624e+00
5.69362402e-01 5.23758978e-02 6.54761970e-01 -1.34222937e+00
-1.07102442e+00 9.73356485e-01 -3.82088631e-01 -1.63494810e-01
7.14912415e-02 2.08790153e-01 1.04830205e+00 -2.51107752e-01
3.60475749e-01 1.55782521e+00 6.92952275e-01 8.04089308e-01
-1.27710831e+00 -9.68361020e-01 -8.15858841e-02 -1.89576849e-01
-9.86889064e-01 -9.56496477e-01 6.87991977e-02 -7.27840900e-01
6.01940632e-01 9.54721093e-01 7.76583970e-01 3.45246226e-01
4.77994829e-01 -1.63215697e-01 1.15831530e+00 -4.74415511e-01
6.29764870e-02 3.19998115e-02 2.23228261e-01 2.48523727e-01
1.26897287e+00 2.09452257e-01 -5.76037802e-02 -4.28075105e-01
5.17411768e-01 -1.07977197e-01 -1.64553821e-01 4.13336121e-02
-7.46561050e-01 9.18153107e-01 2.75040895e-01 2.22505778e-01
-8.86946470e-02 5.69808483e-02 6.42315269e-01 3.97176743e-01
4.78082567e-01 6.63000107e-01 -1.40887722e-01 -2.10706368e-01
-1.11108756e+00 5.74276924e-01 7.62279451e-01 1.79082915e-01
4.19831723e-01 1.02828806e-02 -4.64178234e-01 6.31087422e-01
3.31941575e-01 4.35433000e-01 9.72805172e-03 -1.68216598e+00
3.72194707e-01 4.82189566e-01 3.58961165e-01 -1.10484970e+00
-3.10351700e-01 -3.49704295e-01 -6.98044717e-01 7.93911099e-01
8.13865364e-01 -2.24914968e-01 -3.36955965e-01 2.10097957e+00
9.24707726e-02 -8.29833090e-01 -1.24398582e-01 9.72399652e-01
5.23092985e-01 2.95740902e-01 5.90888202e-01 -4.14964795e-01
9.02941227e-01 -2.15033695e-01 -8.37419987e-01 -9.75649953e-02
7.38924086e-01 -7.81789958e-01 1.12362337e+00 -3.67218964e-02
-1.29109681e+00 -8.60994533e-02 -6.14054501e-01 -4.24849153e-01
1.58104822e-02 -6.70441926e-01 7.41435409e-01 1.45743537e+00
-7.88130760e-01 9.09989893e-01 -3.43800128e-01 -2.23089606e-01
7.95941472e-01 7.83361420e-02 -3.37926120e-01 1.59489334e-01
-9.86613572e-01 1.22431898e+00 -2.31508352e-02 1.50676623e-01
-1.71538834e-02 -9.95840609e-01 -7.62908995e-01 4.51591879e-01
2.04805627e-01 -8.05164337e-01 7.68802524e-01 -1.42593265e+00
-8.05000424e-01 1.36303544e+00 2.14514092e-01 -4.21253592e-01
1.00450802e+00 -3.31061393e-01 -3.63080114e-01 -4.34695214e-01
5.11915803e-01 3.02014410e-01 -3.13182659e-02 -1.19194043e+00
-4.72095370e-01 -6.58617675e-01 2.13728487e-01 4.37409073e-01
-1.65596366e-01 7.52461493e-01 8.41788650e-01 -2.40941435e-01
-1.97804958e-01 -5.70756078e-01 -1.33046493e-01 1.93782121e-01
-2.79711783e-01 1.23904273e-02 3.64381641e-01 -6.54036224e-01
1.43894804e+00 -1.94505703e+00 -8.21476817e-01 7.13029802e-02
5.16411066e-01 3.94156694e-01 2.55340695e-01 3.45505327e-01
-2.94960886e-02 7.75607228e-01 -6.39355481e-02 9.62749422e-02
5.75507879e-01 3.54833186e-01 -1.22822501e-01 8.97541404e-01
-3.20317745e-01 7.11857975e-01 -9.76030290e-01 -4.63686109e-01
1.72207147e-01 1.25964701e-01 -8.05951416e-01 -2.99580187e-01
6.31296575e-01 1.98826522e-01 1.80983633e-01 4.46206540e-01
7.74922550e-01 2.70931572e-01 5.06308794e-01 2.10510522e-01
-7.63787389e-01 6.48046434e-01 -1.03896058e+00 7.81067073e-01
6.29557343e-03 5.54182053e-01 1.79223925e-01 -9.99526799e-01
7.63914406e-01 1.19976804e-01 1.90076351e-01 -9.86993849e-01
2.30667993e-01 4.72338319e-01 7.19810069e-01 -3.95335376e-01
7.03164876e-01 -8.86833251e-01 -1.61400810e-01 4.99233276e-01
-5.16433239e-01 -2.80925006e-01 -1.05477154e-01 4.00966071e-02
4.92110670e-01 -9.59146097e-02 6.34217620e-01 -7.73473382e-01
2.14335099e-01 -1.14827484e-01 9.84919369e-01 8.90607178e-01
-9.07711506e-01 1.44323081e-01 6.03436053e-01 -4.05058533e-01
-1.27363539e+00 -1.16523349e+00 -3.57294917e-01 1.11973333e+00
-1.08323072e-03 6.83871582e-02 -5.21070778e-01 -5.57411432e-01
4.34411913e-01 1.23936450e+00 -5.95407307e-01 -3.40166360e-01
-3.35680068e-01 -6.42282546e-01 8.75688493e-01 -3.03334021e-03
5.84186435e-01 -6.26365423e-01 -1.08794940e+00 -1.23809353e-01
5.08823954e-02 -4.69530225e-01 -2.17236772e-01 -4.25638109e-01
-6.96673870e-01 -1.01174927e+00 -4.47257549e-01 1.14635192e-02
3.11990380e-01 9.86712873e-02 9.77423370e-01 4.44219559e-01
2.27609519e-02 -3.22560854e-02 1.51076326e-02 -7.80437648e-01
-6.70754313e-01 -5.85487068e-01 9.89191905e-02 -5.33349514e-01
4.25223738e-01 -3.55219245e-01 -5.84535480e-01 -1.01844333e-01
-5.85094452e-01 -6.79904819e-02 3.23396921e-02 6.53066576e-01
-1.11747496e-01 -1.70319140e-01 1.03165781e+00 -1.18208563e+00
8.29058349e-01 -3.46111178e-01 -2.18035102e-01 4.00749892e-02
-1.21190381e+00 -4.16629583e-01 4.29565042e-01 2.53088083e-02
-1.26470602e+00 -1.21922147e+00 -1.03115328e-01 2.76364267e-01
-4.85177301e-02 2.62541682e-01 -2.78135478e-01 6.33813664e-02
9.60767686e-01 -4.71071362e-01 3.30809236e-01 -1.35690197e-01
3.79272193e-01 7.27942109e-01 4.29032207e-01 -8.29559505e-01
5.01630247e-01 4.88580525e-01 -1.31176621e-01 -5.15288413e-01
-9.84997690e-01 1.16508044e-02 -2.07140028e-01 -3.59439909e-01
6.68496430e-01 -7.17372239e-01 -9.12644148e-01 6.03849813e-02
-6.58174753e-01 -3.11979145e-01 -8.04378271e-01 7.20439732e-01
-3.40421826e-01 3.12991083e-01 -3.62025499e-01 -1.17002523e+00
-4.49820548e-01 -5.50318480e-01 1.49326310e-01 4.75550383e-01
-7.87256658e-01 -9.02694941e-01 8.74905586e-02 8.15164030e-01
6.94656610e-01 6.11967385e-01 9.87422585e-01 -6.02433383e-01
2.17238262e-01 -1.68706611e-01 -3.89007956e-01 4.90467697e-01
6.27710402e-01 1.72444597e-01 -7.05908060e-01 -7.86880106e-02
1.34894490e-01 -2.52175182e-01 3.17918330e-01 3.79687399e-01
6.70180738e-01 -9.56767082e-01 1.44993186e-01 3.83719683e-01
1.52561605e+00 3.26103717e-01 7.02267170e-01 3.87085348e-01
5.65168679e-01 1.07920706e+00 4.68301088e-01 5.84846914e-01
7.09037006e-01 3.98182034e-01 1.90836936e-01 -2.91680753e-01
-3.37376148e-02 -2.01114103e-01 -5.10249734e-02 2.15596914e-01
-4.31300610e-01 3.86185795e-01 -1.09036458e+00 4.87418652e-01
-1.60288799e+00 -1.61914110e+00 -4.76172000e-01 2.63580585e+00
9.03607070e-01 3.17713886e-01 3.67805481e-01 1.35161832e-01
7.93112278e-01 9.93445963e-02 -2.37828046e-01 -1.46232867e+00
4.43168841e-02 3.50485742e-02 5.02459288e-01 9.01452184e-01
-5.03597617e-01 8.05682778e-01 6.26391029e+00 3.93458843e-01
-9.06581700e-01 7.91179761e-02 8.89400423e-01 -2.65298843e-01
-7.67966390e-01 3.66930395e-01 -2.16559872e-01 4.30575103e-01
5.93213677e-01 -1.05203676e+00 -5.11322021e-02 3.89096826e-01
7.87561119e-01 -2.28973955e-01 -9.30995524e-01 4.53529716e-01
-3.09129179e-01 -1.11914027e+00 -7.90976435e-02 3.16308200e-01
7.21603632e-01 -4.34480697e-01 9.54216644e-02 1.82010159e-01
5.83762527e-01 -1.43283677e+00 1.09138167e+00 4.44975913e-01
9.18293536e-01 -7.78296649e-01 7.50538528e-01 3.19150597e-01
-4.89505231e-01 -1.07373454e-01 -4.37596321e-01 -1.10598338e+00
-1.11554407e-01 7.64711916e-01 -1.32871017e-01 4.67292607e-01
3.54145736e-01 1.51247885e-02 6.92003593e-02 9.70062852e-01
3.17896269e-02 4.21192497e-01 1.38708279e-01 2.38527566e-01
2.68959582e-01 -3.81475359e-01 3.42775643e-01 1.15696490e+00
-8.45759548e-03 2.37653971e-01 -3.38867486e-01 1.02200568e+00
-7.97981396e-02 3.83628279e-01 -6.63707197e-01 -4.62026522e-03
7.76432633e-01 8.81961346e-01 -3.88582975e-01 -3.52277070e-01
-4.30387557e-01 2.81180143e-01 2.41005123e-02 3.71936224e-02
-6.97059810e-01 -1.39077961e-01 9.45009291e-01 3.82898241e-01
-9.16172445e-01 2.00507283e-01 -9.59337473e-01 -1.06462562e+00
-4.36226964e-01 -1.01566911e+00 3.23194087e-01 -6.42568022e-02
-1.11406815e+00 -3.43550920e-01 -7.08463117e-02 -6.68957055e-01
2.55175650e-01 -1.44922197e-01 -6.34721100e-01 1.41023278e+00
-1.04671156e+00 -8.94531190e-01 7.44405538e-02 2.22559422e-02
-8.60736892e-02 3.56924176e-01 7.38519311e-01 2.58329004e-01
-1.77413747e-01 8.40920269e-01 4.63511907e-02 -3.88399959e-02
1.04244542e+00 -1.05156708e+00 -2.05777679e-02 8.46321762e-01
-4.99969214e-01 9.12005365e-01 8.79245102e-01 -6.56542838e-01
-4.68313038e-01 -6.88582599e-01 1.58337891e+00 -4.10661697e-01
3.58865321e-01 -7.62761980e-02 -8.55057657e-01 6.03954315e-01
8.07138160e-02 -6.24106348e-01 1.04342294e+00 4.39167082e-01
-5.42319953e-01 -1.53779238e-01 -1.70072520e+00 8.41777563e-01
1.20223713e+00 -5.83007932e-01 -6.33821845e-01 9.10271928e-02
4.61086124e-01 -9.84304994e-02 -7.27665603e-01 3.03653359e-01
1.19569218e+00 -1.36061943e+00 8.01311493e-01 -8.62229109e-01
6.47711039e-01 1.46898597e-01 -3.56729329e-01 -9.35260773e-01
-5.24031341e-01 -4.11463290e-01 7.09851742e-01 1.52059209e+00
2.11629048e-01 -9.88375962e-01 4.67075586e-01 1.38031852e+00
-2.40984708e-01 -5.88849545e-01 -9.78578627e-01 -5.81355631e-01
9.32635903e-01 -2.31760979e-01 8.65991652e-01 1.89693856e+00
3.64902616e-01 8.63579661e-02 -4.66342390e-01 -2.08841681e-01
7.53237665e-01 -4.24072631e-02 8.04872811e-01 -1.26701641e+00
1.26264825e-01 -1.03223896e+00 -2.80099332e-01 9.18522328e-02
5.00748269e-02 -8.14449787e-01 -3.53752583e-01 -1.75377131e+00
4.96508747e-01 -5.29086888e-01 -2.90556997e-02 3.93412560e-01
-2.86911488e-01 -1.36931822e-01 8.29711556e-01 3.23190093e-01
-8.22758526e-02 1.57883778e-01 1.29655242e+00 7.56077319e-02
-1.81080505e-01 -1.96156099e-01 -1.83698785e+00 7.90899932e-01
8.74704182e-01 -4.79145467e-01 -2.84249425e-01 -3.08745980e-01
4.76559222e-01 1.42841950e-01 4.82711434e-01 -6.36575580e-01
-7.17709139e-02 -1.01797116e+00 3.13748538e-01 4.10742342e-01
-1.74974382e-01 -6.51657999e-01 4.97115642e-01 9.03031528e-01
-5.95374525e-01 -2.97913611e-01 4.66034450e-02 -4.00279850e-01
3.52393270e-01 -1.72479317e-01 1.13854504e+00 -8.84732604e-02
1.73869561e-02 -1.73738241e-01 -1.07455239e-01 2.52187550e-01
8.69094670e-01 -4.42390323e-01 -1.01316202e+00 -3.75949323e-01
-5.46858072e-01 3.04582447e-01 1.03939188e+00 2.56717633e-02
5.14582470e-02 -1.15571904e+00 -1.11532092e+00 -3.02470416e-01
-2.03441471e-01 -6.74163401e-01 4.09467578e-01 8.68699789e-01
-5.44748545e-01 3.86995256e-01 -5.69762111e-01 2.54743904e-01
-1.29649842e+00 2.49409869e-01 8.32873225e-01 7.41369575e-02
-4.70151961e-01 5.90915859e-01 3.62432152e-01 -4.07401562e-01
-5.70511557e-02 1.95843980e-01 -5.52006885e-02 8.19504168e-03
5.30908108e-01 9.01178598e-01 -3.72839570e-01 -8.52832139e-01
-5.76276898e-01 1.35544941e-01 2.38145068e-01 -1.82408810e-01
8.26808393e-01 -3.91108155e-01 -6.79860711e-01 4.69230384e-01
6.39861941e-01 5.46296775e-01 -8.80677879e-01 4.32013243e-01
-2.40826264e-01 -1.21444750e+00 -3.39723408e-01 -1.07364154e+00
-3.86257499e-01 7.41700292e-01 3.50802928e-01 4.67752576e-01
8.39352489e-01 -3.16356301e-01 5.34996912e-02 -1.66258097e-01
1.95621490e-01 -1.32044220e+00 -6.78627253e-01 -2.16375235e-02
8.29055965e-01 -1.01554716e+00 3.13758850e-01 -3.26036736e-02
-5.29784083e-01 5.77293813e-01 6.56277776e-01 -1.38538405e-02
3.07541430e-01 -4.01524678e-02 1.83484033e-01 4.91958968e-02
-4.95696008e-01 2.59651206e-02 -1.06048740e-01 5.38006365e-01
8.87288511e-01 8.41368377e-01 -1.64853096e+00 5.12063265e-01
-6.51880443e-01 4.15570512e-02 7.56780565e-01 6.50561988e-01
-7.38391399e-01 -1.12879694e+00 -7.08699107e-01 8.54719222e-01
-1.02301216e+00 -1.54420078e-01 -6.01757646e-01 1.08676910e+00
8.16568434e-01 1.07368684e+00 1.83042094e-01 8.57879072e-02
1.90109298e-01 -1.95817389e-02 3.63971472e-01 -5.62465966e-01
-7.08203495e-01 -1.74648806e-01 6.62635744e-01 -3.36112320e-01
-4.61074293e-01 -6.67136073e-01 -1.21367860e+00 -1.34177363e+00
-1.20155893e-01 2.09648788e-01 2.97927082e-01 1.09887683e+00
6.98885545e-02 1.16711140e-01 3.57174158e-01 -1.77696437e-01
-7.67759442e-01 -4.39160645e-01 -7.26614475e-01 4.99194413e-01
3.32358927e-01 -3.29323620e-01 -3.52408290e-01 -5.06983638e-01]
|
[8.878751754760742, 5.5943474769592285]
|
15bf843f-e101-4068-97da-fa0dff941b1b
|
a-unified-generative-framework-for-aspect
|
2106.04300
| null |
https://arxiv.org/abs/2106.04300v1
|
https://arxiv.org/pdf/2106.04300v1.pdf
|
A Unified Generative Framework for Aspect-Based Sentiment Analysis
|
Aspect-based Sentiment Analysis (ABSA) aims to identify the aspect terms, their corresponding sentiment polarities, and the opinion terms. There exist seven subtasks in ABSA. Most studies only focus on the subsets of these subtasks, which leads to various complicated ABSA models while hard to solve these subtasks in a unified framework. In this paper, we redefine every subtask target as a sequence mixed by pointer indexes and sentiment class indexes, which converts all ABSA subtasks into a unified generative formulation. Based on the unified formulation, we exploit the pre-training sequence-to-sequence model BART to solve all ABSA subtasks in an end-to-end framework. Extensive experiments on four ABSA datasets for seven subtasks demonstrate that our framework achieves substantial performance gain and provides a real unified end-to-end solution for the whole ABSA subtasks, which could benefit multiple tasks.
|
['Zheng Zhang', 'Xipeng Qiu', 'Tuo ji', 'Junqi Dai', 'Hang Yan']
|
2021-06-08
|
a-unified-generative-framework-for-aspect-1
|
https://aclanthology.org/2021.acl-long.188
|
https://aclanthology.org/2021.acl-long.188.pdf
|
acl-2021-5
|
['aspect-term-extraction-and-sentiment', 'aspect-oriented-opinion-extraction', 'aspect-sentiment-triplet-extraction']
|
['natural-language-processing', 'natural-language-processing', 'natural-language-processing']
|
[ 1.30026937e-01 -2.29852736e-01 9.57898349e-02 -5.73524415e-01
-1.04019952e+00 -9.14120555e-01 5.64563096e-01 -3.14400017e-01
3.17655392e-02 3.67486477e-01 3.70942563e-01 -3.70748162e-01
-8.74326658e-03 -6.42248929e-01 -5.04122078e-01 -6.25378430e-01
4.15901184e-01 4.47653085e-01 6.88560829e-02 -6.83617473e-01
1.82852507e-01 -2.89057046e-01 -1.26525760e+00 7.05131054e-01
9.52247024e-01 1.02856100e+00 1.30642205e-01 4.89049435e-01
-5.61502516e-01 6.35004938e-01 -6.91223145e-01 -9.45762038e-01
-1.67423841e-02 -5.00949740e-01 -9.76815164e-01 3.71246487e-02
8.64287540e-02 1.10949486e-01 1.98826835e-01 1.08149862e+00
5.25420725e-01 1.03573367e-01 4.66705143e-01 -1.32997704e+00
-7.61958361e-01 6.84425294e-01 -6.93219066e-01 -1.88226715e-01
5.14905691e-01 2.16796696e-01 1.72330725e+00 -9.55581009e-01
2.52283990e-01 1.21235478e+00 5.73022723e-01 3.94257247e-01
-9.14824843e-01 -5.41905403e-01 8.25679183e-01 -3.51159684e-02
-8.68085325e-01 -1.44377664e-01 7.14563489e-01 -3.08419108e-01
1.24091268e+00 5.04815102e-01 1.00682938e+00 8.82815301e-01
3.44554454e-01 1.24182832e+00 1.09504437e+00 1.83084626e-02
2.27312997e-01 -3.45122337e-01 7.10269570e-01 6.09427631e-01
3.00779134e-01 -4.78556931e-01 -7.85192013e-01 -3.09756279e-01
1.97478920e-01 4.25522961e-02 4.54402119e-02 -3.22972447e-01
-1.25835574e+00 8.52941036e-01 3.12362254e-01 6.81788921e-02
-3.74775827e-01 -1.30498886e-01 4.66480047e-01 2.66129375e-01
5.35342455e-01 6.14625931e-01 -8.86944354e-01 -2.48115540e-01
-4.33113664e-01 6.13599777e-01 1.04741001e+00 1.00418532e+00
9.24648464e-01 5.93808703e-02 -5.97919822e-01 8.01991463e-01
3.63122255e-01 6.85426354e-01 5.45764148e-01 -4.75600541e-01
4.88125801e-01 1.02408886e+00 -5.74080856e-04 -8.20454776e-01
-4.66871858e-01 -5.73328495e-01 -5.50585687e-01 -1.55194685e-01
4.86511737e-02 -3.70820403e-01 -1.04966950e+00 1.64691067e+00
4.07719642e-01 -1.02954082e-01 1.38838753e-01 8.97951007e-01
8.41542661e-01 7.81187415e-01 2.72182040e-02 -4.46732566e-02
2.00810504e+00 -1.45272338e+00 -6.45313323e-01 -7.12575912e-01
6.63534582e-01 -1.08376431e+00 1.39635348e+00 3.52785021e-01
-9.59076762e-01 -3.66593093e-01 -9.51226115e-01 -2.39153326e-01
-3.48809034e-01 9.11585391e-02 8.87926579e-01 4.36368048e-01
-1.08326674e+00 -9.88335162e-02 -7.03169048e-01 1.08203828e-01
1.31881729e-01 1.97498173e-01 1.87133640e-01 1.50527969e-01
-1.20757663e+00 3.37654620e-01 1.39941692e-01 4.49442834e-01
-7.78606951e-01 -6.09418452e-01 -1.10142207e+00 1.99443161e-01
6.06429040e-01 -1.23224998e+00 1.61204541e+00 -8.38197112e-01
-1.57785916e+00 7.96707749e-01 -7.04506099e-01 -2.21303757e-02
-4.85529602e-02 -4.83718008e-01 -1.77579284e-01 -6.02016926e-01
2.98964590e-01 1.32100299e-01 9.38222229e-01 -1.18469906e+00
-8.07401896e-01 -5.78685760e-01 6.10356390e-01 5.72868168e-01
-3.32473665e-01 3.39753687e-01 -7.39077330e-01 -8.66372228e-01
1.30026815e-02 -1.09677505e+00 -4.05735373e-01 -9.00686145e-01
-7.03935623e-01 -4.06582564e-01 3.71893764e-01 -3.61015141e-01
1.40909457e+00 -1.93692791e+00 5.00275970e-01 -1.18966833e-01
4.25375611e-01 7.27873296e-02 -4.04070199e-01 2.40967572e-01
-9.87128764e-02 -1.82430536e-01 -2.32142910e-01 -5.48109651e-01
2.45251611e-01 3.36610898e-02 -6.66457355e-01 -3.31014246e-01
9.14518908e-02 1.34232914e+00 -1.14599311e+00 -1.77286044e-01
-2.88420022e-01 3.44691247e-01 -5.62978923e-01 4.69852477e-01
-6.04244947e-01 1.07941769e-01 -9.68277335e-01 8.20028663e-01
6.34337783e-01 -4.43989247e-01 1.55145470e-02 -5.57754524e-02
1.16681099e-01 6.03556097e-01 -8.14867139e-01 1.73379791e+00
-5.32011688e-01 2.98135936e-01 -3.51354219e-02 -7.08218634e-01
9.21378374e-01 2.83726305e-02 3.48322093e-01 -6.06793165e-01
1.83291063e-01 1.38655379e-01 -5.01738973e-02 -4.26894635e-01
8.45798135e-01 -2.37337664e-01 -6.03341699e-01 7.51928866e-01
5.85856289e-02 -3.23048204e-01 4.62522358e-01 1.93850040e-01
7.39979267e-01 2.25564688e-01 2.66835511e-01 -2.46638060e-01
5.67536116e-01 8.91156942e-02 5.22085071e-01 6.31039560e-01
7.02507123e-02 8.47962081e-01 7.69729257e-01 -7.39030719e-01
-5.49736679e-01 -7.99064696e-01 3.91900718e-01 1.62225342e+00
1.80962488e-01 -8.34774613e-01 -7.60816813e-01 -8.49656940e-01
-4.13049608e-01 5.43998003e-01 -7.56396949e-01 -7.59260505e-02
-4.31753010e-01 -1.24993038e+00 1.17631771e-01 4.82142776e-01
3.08972090e-01 -1.19397700e+00 -7.29607120e-02 7.53320903e-02
-3.95488232e-01 -9.27526832e-01 -8.35294306e-01 3.84757970e-03
-6.09726608e-01 -9.43923116e-01 -7.87151039e-01 -8.86702001e-01
4.92978901e-01 6.60099387e-01 1.56452203e+00 -2.40666047e-01
1.65425852e-01 -2.91080195e-02 -4.70190853e-01 -6.58942342e-01
-2.74205357e-02 4.42280173e-01 -4.33359444e-01 2.05651313e-01
6.64645553e-01 -4.35938030e-01 -5.96582234e-01 1.96272656e-01
-8.09486330e-01 2.76410222e-01 6.36011481e-01 6.12239182e-01
7.44732499e-01 -2.33044699e-01 3.67485940e-01 -1.16173983e+00
1.13104784e+00 -4.39432621e-01 -5.39397061e-01 2.89293110e-01
-7.35502362e-01 1.75975412e-02 6.33252203e-01 -1.38357401e-01
-1.04320776e+00 -4.40459363e-02 -2.48864204e-01 -1.73264191e-01
6.72953799e-02 1.04262424e+00 -4.49189633e-01 4.31820929e-01
3.03003311e-01 7.09342778e-01 -8.61022770e-02 -3.32546294e-01
3.66350204e-01 4.47137415e-01 1.20565407e-01 -6.86683357e-01
6.44079387e-01 1.74952924e-01 -3.31719875e-01 -3.17964882e-01
-1.79393923e+00 -6.68263257e-01 -1.08584411e-01 -2.75068600e-02
7.51915455e-01 -1.23553038e+00 -7.18726218e-01 8.31163645e-01
-1.22160041e+00 -7.56390765e-02 -1.20712236e-01 9.89477336e-03
-4.83910948e-01 1.55546039e-01 -5.72648227e-01 -5.21630883e-01
-9.56241906e-01 -1.68259358e+00 1.41048336e+00 3.88548970e-01
-3.84706318e-01 -7.78761148e-01 4.58063483e-01 8.06795776e-01
4.82538730e-01 -3.11947107e-01 9.28114057e-01 -6.74492240e-01
-4.90576506e-01 -1.67603999e-01 -1.72729697e-02 2.28749081e-01
1.41833976e-01 -8.26691464e-02 -8.55604112e-01 -2.02292740e-01
2.15884835e-01 -3.36163014e-01 9.37818766e-01 3.82813245e-01
9.78715360e-01 -2.33526826e-01 8.00476316e-03 7.22514808e-01
9.71124768e-01 2.85881579e-01 3.83123815e-01 5.23808420e-01
1.05054975e+00 4.32382166e-01 8.35456789e-01 1.99095681e-01
9.10716653e-01 4.49172616e-01 3.55488181e-01 2.38256380e-02
1.19524166e-01 -6.90863654e-02 6.62126005e-01 1.52415085e+00
-6.68352395e-02 -2.99324483e-01 -7.51100421e-01 6.89668715e-01
-2.00840020e+00 -6.07131124e-01 -2.01805696e-01 1.71122622e+00
1.05528748e+00 3.74252424e-02 2.18545422e-01 -3.83609980e-01
4.09330964e-01 8.06249619e-01 -6.29850686e-01 -5.26618361e-01
3.36690992e-02 -1.41088171e-02 -2.43365169e-01 4.21094269e-01
-1.15035975e+00 1.00444782e+00 6.17576170e+00 9.12949622e-01
-9.70490396e-01 1.68560773e-01 6.38666332e-01 -1.81789726e-01
-8.75953734e-01 1.98851332e-01 -9.05411780e-01 4.76179719e-01
5.00254929e-01 -4.65941072e-01 4.89286929e-01 1.11735046e+00
-1.73144713e-01 4.10157412e-01 -7.62276053e-01 7.47764826e-01
2.34048933e-01 -9.95874047e-01 6.26613736e-01 -1.98515162e-01
1.05526030e+00 6.53997138e-02 1.91034660e-01 8.13489020e-01
5.61480880e-01 -8.13802719e-01 7.73530364e-01 1.31459132e-01
3.05350274e-01 -8.38920593e-01 9.29022193e-01 -6.02270141e-02
-1.37570143e+00 1.27577573e-01 -2.73606420e-01 -1.81131214e-01
3.47543001e-01 6.04079723e-01 -3.25626403e-01 8.86116743e-01
5.69407463e-01 8.08066070e-01 -4.86506552e-01 7.65682042e-01
-6.11651540e-01 6.27565026e-01 1.79943740e-01 -4.79393303e-01
5.66291749e-01 -5.39146483e-01 6.41846836e-01 1.04363084e+00
2.56534904e-01 -1.53861433e-01 1.49073511e-01 6.16542459e-01
-9.43266153e-02 2.91068286e-01 -2.03447893e-01 -2.45657891e-01
5.41043542e-02 1.59035861e+00 -6.20116770e-01 -2.39857480e-01
-6.15725219e-01 8.24212849e-01 2.41639614e-01 6.23631120e-01
-8.36609483e-01 -5.31485915e-01 9.61352944e-01 -3.42644662e-01
4.12425995e-01 1.06042244e-01 -6.23093486e-01 -1.62790430e+00
2.39812136e-01 -1.54106236e+00 5.32138228e-01 -9.26738262e-01
-1.30690217e+00 1.07256997e+00 -2.59047508e-01 -1.15737510e+00
-3.76103610e-01 -5.14299870e-01 -8.15406322e-01 1.04670084e+00
-1.45453572e+00 -1.55492020e+00 -2.22632945e-01 5.20531058e-01
8.61586392e-01 -1.24171123e-01 7.34524965e-01 -2.19651703e-02
-6.49794579e-01 4.49126095e-01 -1.36594102e-01 7.29981884e-02
6.15890741e-01 -1.43862951e+00 9.19837534e-01 9.85224009e-01
3.88185345e-02 1.04427576e+00 6.19446278e-01 -5.47428727e-01
-1.58444440e+00 -9.92797196e-01 1.12496018e+00 -7.24533558e-01
1.01104963e+00 -4.75222111e-01 -7.72372186e-01 9.49725568e-01
4.30390060e-01 -5.60507655e-01 6.93080723e-01 9.53129768e-01
-4.12219703e-01 -1.51410669e-01 -5.55733852e-02 7.80527174e-01
9.09162521e-01 -5.93477190e-01 -8.63810539e-01 2.70737380e-01
1.09225714e+00 -5.90408742e-01 -3.48073572e-01 3.38605702e-01
4.96729523e-01 -7.60127306e-01 9.14256871e-01 -7.10475862e-01
8.76723647e-01 -4.59996700e-01 1.48619935e-01 -1.50959897e+00
-3.12571377e-01 -7.35405207e-01 -2.60753423e-01 1.26639175e+00
7.62399435e-01 -7.24651933e-01 4.87361223e-01 6.22450829e-01
-2.87061840e-01 -1.26962924e+00 -5.50011158e-01 -3.33256483e-01
-1.08954951e-01 -5.76595664e-01 9.99867618e-01 8.40200305e-01
-1.52650982e-01 1.17067361e+00 -3.54400992e-01 -1.52280834e-02
2.55061090e-01 1.18835592e+00 8.45311642e-01 -9.13294494e-01
-4.97674435e-01 -7.12384105e-01 7.22975805e-02 -1.25838661e+00
1.01684198e-01 -9.92058516e-01 1.90117106e-01 -1.63892269e+00
7.48161256e-01 -2.70896941e-01 -5.63793540e-01 6.98666155e-01
-1.03142214e+00 1.29696652e-01 1.46663949e-01 1.06702946e-01
-1.07605124e+00 7.80736268e-01 1.39368200e+00 -3.96942705e-01
-2.44041011e-01 1.96820378e-01 -1.54280090e+00 7.00688243e-01
5.63808739e-01 -4.16240156e-01 -6.82673752e-01 -8.74336720e-01
9.94378209e-01 -2.85766304e-01 -1.11581102e-01 -2.95116842e-01
4.14275527e-02 -2.73382962e-01 -1.87063307e-01 -6.90528512e-01
2.93648005e-01 -2.66677290e-01 -2.00918674e-01 2.94846557e-02
-1.94804564e-01 1.69102132e-01 1.06020652e-01 5.04635870e-01
-5.85711181e-01 -1.19729988e-01 -1.02057792e-02 -5.84644005e-02
-6.37882531e-01 3.48503530e-01 -1.69481292e-01 3.55953336e-01
8.32875967e-01 1.43261924e-01 -3.60296994e-01 -3.70725572e-01
-6.11294389e-01 5.27760386e-01 2.76838541e-01 7.04848409e-01
3.46362054e-01 -1.19467127e+00 -6.96947217e-01 1.43715605e-01
2.84802914e-01 4.81639296e-01 5.56339562e-01 7.06494749e-01
-1.45840064e-01 5.61791062e-01 2.81050745e-02 -3.92824560e-01
-1.26653147e+00 5.73815286e-01 1.80707887e-01 -8.27533901e-01
-3.26253138e-02 1.14369977e+00 7.69210994e-01 -7.47248113e-01
-4.75352742e-02 -1.20691173e-01 -4.76467431e-01 3.29586446e-01
5.34274280e-01 -1.02848627e-01 1.27699271e-01 -5.26076019e-01
-3.67553920e-01 6.31226063e-01 -5.19464135e-01 1.50063694e-01
1.31231964e+00 -1.77396953e-01 -4.64956135e-01 3.41094643e-01
9.37132895e-01 1.98210850e-01 -8.61813128e-01 -2.86390126e-01
-3.51300627e-01 -2.90779769e-01 -1.01961605e-01 -8.59600246e-01
-1.27588320e+00 8.31151843e-01 -2.45001674e-01 3.84034187e-01
1.34163523e+00 1.37405142e-01 1.10062444e+00 4.61332560e-01
2.92972952e-01 -6.90989017e-01 1.97750274e-02 1.01278281e+00
9.50152516e-01 -1.01947379e+00 -9.25720781e-02 -4.17613357e-01
-1.14032066e+00 6.63114667e-01 6.17489815e-01 1.53994814e-01
3.43064547e-01 1.49423480e-01 4.88843530e-01 -4.38887119e-01
-9.90570426e-01 -3.26261669e-01 3.79419714e-01 1.53033391e-01
5.54246485e-01 2.16354623e-01 -5.80757141e-01 1.40579319e+00
-5.64066827e-01 -1.51222214e-01 1.42528757e-01 9.81053114e-01
-6.12032339e-02 -1.15161884e+00 1.64980199e-02 4.39353377e-01
-7.23803341e-01 -5.41946352e-01 -3.90764147e-01 3.29814881e-01
-2.67333359e-01 8.41144502e-01 -1.89363495e-01 -5.41463733e-01
6.53887451e-01 1.40018016e-01 -5.50727323e-02 -6.17201388e-01
-9.47850883e-01 2.56580174e-01 1.92999914e-01 -7.16378331e-01
-2.40069687e-01 -4.72957313e-01 -1.01402783e+00 -8.42195749e-02
-2.82119930e-01 2.99097151e-01 3.62037152e-01 1.08835196e+00
7.39543796e-01 7.86296964e-01 7.07968295e-01 -4.27738547e-01
-5.00142872e-01 -9.78062451e-01 -3.50971252e-01 4.41070348e-01
2.16472611e-01 -2.21620888e-01 1.61720868e-02 5.85800074e-02]
|
[11.504339218139648, 6.652796745300293]
|
be20946d-7cbe-4c1c-9ba3-d40fdb18ae82
|
multi-modal-representation-learning-with-text
|
2304.00719
| null |
https://arxiv.org/abs/2304.00719v1
|
https://arxiv.org/pdf/2304.00719v1.pdf
|
Multi-Modal Representation Learning with Text-Driven Soft Masks
|
We propose a visual-linguistic representation learning approach within a self-supervised learning framework by introducing a new operation, loss, and data augmentation strategy. First, we generate diverse features for the image-text matching (ITM) task via soft-masking the regions in an image, which are most relevant to a certain word in the corresponding caption, instead of completely removing them. Since our framework relies only on image-caption pairs with no fine-grained annotations, we identify the relevant regions to each word by computing the word-conditional visual attention using multi-modal encoder. Second, we encourage the model to focus more on hard but diverse examples by proposing a focal loss for the image-text contrastive learning (ITC) objective, which alleviates the inherent limitations of overfitting and bias issues. Last, we perform multi-modal data augmentations for self-supervised learning via mining various examples by masking texts and rendering distortions on images. We show that the combination of these three innovations is effective for learning a pretrained model, leading to outstanding performance on multiple vision-language downstream tasks.
|
['Bohyung Han', 'Jaeyoo Park']
|
2023-04-03
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Park_Multi-Modal_Representation_Learning_With_Text-Driven_Soft_Masks_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Park_Multi-Modal_Representation_Learning_With_Text-Driven_Soft_Masks_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['text-matching']
|
['natural-language-processing']
|
[ 6.77162468e-01 2.62053609e-01 -1.85379848e-01 -5.18048286e-01
-1.02179492e+00 -3.10654819e-01 9.14330065e-01 -9.65971593e-03
-6.42502487e-01 3.77326041e-01 5.06508231e-01 -1.51074737e-01
3.17542642e-01 -4.63089764e-01 -1.16327858e+00 -5.93130231e-01
3.84146422e-01 2.28010952e-01 3.85138392e-02 -1.14492022e-01
2.00634539e-01 7.94450492e-02 -1.68375838e+00 7.94603646e-01
8.54598999e-01 1.03084862e+00 6.11945331e-01 3.25402051e-01
-2.67867267e-01 9.81190205e-01 -4.00005996e-01 -6.08798504e-01
3.04619700e-01 -5.89354098e-01 -6.67395711e-01 4.49769408e-01
9.09400105e-01 -1.89698055e-01 -3.01935047e-01 1.13675761e+00
3.82476330e-01 -4.54033762e-02 8.14413726e-01 -1.16299641e+00
-1.22888541e+00 4.27922547e-01 -9.69733119e-01 9.58932415e-02
1.17668532e-01 3.34786177e-01 1.21159470e+00 -1.34679627e+00
5.37317276e-01 1.35078108e+00 3.60707104e-01 6.52379513e-01
-1.31160724e+00 -5.38376391e-01 3.23460668e-01 1.33944601e-01
-1.23202753e+00 -6.75497532e-01 8.89249682e-01 -5.05705297e-01
8.04222643e-01 1.52131245e-01 2.30309546e-01 1.20682287e+00
-2.32543331e-02 9.57375824e-01 1.05345273e+00 -7.42798805e-01
-3.10680699e-02 4.18836653e-01 -2.73629844e-01 8.50477159e-01
-6.98569864e-02 1.36853606e-01 -4.66863990e-01 1.86069876e-01
4.95052606e-01 7.81350061e-02 -2.06776127e-01 -4.18943793e-01
-1.21931684e+00 9.39727247e-01 5.02013683e-01 2.02421322e-01
-1.65844321e-01 1.41894683e-01 3.46686304e-01 1.67171046e-01
7.04667449e-01 3.07344854e-01 -3.09327006e-01 4.55458224e-01
-1.04484200e+00 -5.09347580e-02 6.79348260e-02 9.70344424e-01
1.09685254e+00 3.98544362e-03 -5.47204196e-01 1.07056665e+00
4.02568430e-01 5.94971240e-01 6.50076807e-01 -7.56462455e-01
7.98907340e-01 6.25124753e-01 -1.35814458e-01 -8.68122995e-01
-1.42166615e-01 -5.10162294e-01 -9.21342969e-01 2.69021600e-01
1.36256263e-01 7.15462118e-02 -1.15762401e+00 1.94670737e+00
5.30598313e-02 -1.64660349e-01 1.16802350e-01 9.82478023e-01
8.02418411e-01 7.11298227e-01 3.08755487e-01 -1.46095976e-01
1.40110064e+00 -1.35054219e+00 -5.64748406e-01 -7.71390021e-01
4.89442796e-01 -8.03995848e-01 1.51896167e+00 1.18872030e-02
-1.12229872e+00 -7.10638046e-01 -9.18001175e-01 -3.94692242e-01
-2.84875810e-01 2.96689510e-01 3.62105668e-01 2.00542822e-01
-1.00487292e+00 1.79815695e-01 -3.76366079e-01 -1.31130829e-01
6.54708326e-01 -1.05783649e-01 -2.86154240e-01 -3.57216805e-01
-1.13490033e+00 1.01389480e+00 2.83048362e-01 -2.33165711e-01
-9.53408003e-01 -6.31321251e-01 -1.27097690e+00 -4.89538312e-02
2.85297781e-01 -7.89896727e-01 8.27551484e-01 -1.58571899e+00
-9.61544693e-01 1.40333211e+00 -3.27017844e-01 -4.73537624e-01
4.96501356e-01 8.43691360e-03 -3.73134643e-01 3.25017065e-01
4.65633363e-01 1.20755732e+00 1.49394357e+00 -1.53474629e+00
-5.29605091e-01 -2.42572084e-01 -5.75009594e-03 4.03796762e-01
-6.61765873e-01 3.73979136e-02 -7.82655299e-01 -1.01101923e+00
-1.40418932e-01 -6.97944701e-01 -2.99090534e-01 1.89382643e-01
-3.96464884e-01 -5.61726019e-02 6.14210844e-01 -6.67411387e-01
9.25986528e-01 -2.30184269e+00 1.04497761e-01 -1.08722530e-01
2.71707743e-01 2.22668082e-01 -6.55445457e-01 1.47559568e-01
-2.02641264e-01 1.03576370e-01 -5.46214163e-01 -7.93567061e-01
-1.63253412e-01 1.69774994e-01 -5.05028546e-01 2.52748996e-01
8.00323308e-01 1.33902276e+00 -7.12099791e-01 -7.88832307e-01
3.75828624e-01 3.70784640e-01 -6.28283322e-01 3.07335764e-01
-5.01911759e-01 3.95794570e-01 -1.43664554e-01 4.23456579e-01
6.10931754e-01 -4.90374744e-01 -3.38029325e-01 -4.18384761e-01
-7.03217611e-02 1.12986460e-01 -7.73337007e-01 1.86006951e+00
-7.52485931e-01 6.44757390e-01 -1.39299110e-01 -1.18902147e+00
7.52746880e-01 -9.20347869e-02 2.04019710e-01 -1.09782147e+00
-1.60552904e-01 -3.38805206e-02 -3.20666820e-01 -8.13304245e-01
2.61018097e-01 -8.92048404e-02 -1.06258616e-02 5.52308559e-01
2.19342664e-01 1.03594633e-02 1.68178469e-01 3.71858686e-01
7.00718641e-01 2.07339332e-01 6.73260838e-02 -9.80679691e-02
6.21766925e-01 -1.07745796e-01 2.58027732e-01 6.29149735e-01
-7.26405606e-02 9.51089919e-01 2.64854461e-01 -1.59860179e-01
-1.20449960e+00 -9.25590336e-01 -6.97952434e-02 1.37228358e+00
5.32314442e-02 -3.11153978e-01 -5.65487802e-01 -6.54882610e-01
2.18545478e-02 6.15822017e-01 -8.22931767e-01 -3.39480549e-01
-4.02579874e-01 -5.86085320e-01 3.32933635e-01 4.83443379e-01
4.52959687e-01 -1.18947184e+00 -4.23702687e-01 -1.01913214e-01
-3.99494261e-01 -1.26084828e+00 -9.63930547e-01 3.04917276e-01
-4.90623921e-01 -8.26817274e-01 -9.24197853e-01 -1.27500129e+00
9.92603540e-01 4.59436446e-01 1.22013652e+00 7.11363256e-02
-4.51874614e-01 4.33803529e-01 -1.90031275e-01 -2.04941928e-01
-5.84539711e-01 -2.82802492e-01 -2.78561473e-01 3.89019549e-01
2.07037389e-01 -3.62151831e-01 -6.74047768e-01 -1.76185125e-03
-1.06116462e+00 4.17661250e-01 9.10629451e-01 1.00568664e+00
6.64585590e-01 -2.89569199e-01 5.23275137e-01 -7.52805829e-01
4.57778662e-01 -3.51823956e-01 -3.59667033e-01 3.62897784e-01
-5.30707955e-01 2.32033625e-01 5.99812746e-01 -5.80999017e-01
-1.02839196e+00 3.49534065e-01 -2.49609575e-02 -7.35521555e-01
-1.45413578e-01 3.24724138e-01 -2.42339373e-01 8.76294822e-02
6.81059301e-01 5.12213647e-01 1.34234309e-01 -2.43115306e-01
8.57574522e-01 6.43835783e-01 6.60332620e-01 -4.33546185e-01
1.01937985e+00 5.57223916e-01 -3.25787693e-01 -7.74011314e-01
-1.26517892e+00 -4.44436461e-01 -5.59243143e-01 -8.93877670e-02
1.15719783e+00 -1.18862605e+00 -4.43506360e-01 1.93889588e-01
-1.17767274e+00 -1.80674195e-01 -4.24291074e-01 2.69693553e-01
-5.94821930e-01 5.97417295e-01 -3.69540423e-01 -7.35754311e-01
-3.61290634e-01 -1.14189267e+00 1.34324789e+00 -4.95717004e-02
8.13441649e-02 -8.84892762e-01 -2.55489856e-01 6.60381258e-01
3.64309460e-01 -5.66544523e-03 1.01280904e+00 -5.30450463e-01
-5.76051474e-01 1.02490619e-01 -6.57421708e-01 7.09883809e-01
1.22959301e-01 -3.68370146e-01 -1.11713636e+00 -3.81047070e-01
3.14762443e-02 -7.68382132e-01 1.30067515e+00 2.85181999e-01
1.40640318e+00 -3.85349095e-01 -1.14523068e-01 7.43416250e-01
1.40438235e+00 -2.10135862e-01 4.75933194e-01 2.78859079e-01
9.39043403e-01 8.71941686e-01 4.01847899e-01 1.21404819e-01
6.06583536e-01 5.36501765e-01 5.69639921e-01 -7.53338635e-01
-4.49579358e-01 -3.72892022e-01 4.38100368e-01 4.24272031e-01
4.72692460e-01 -1.21588662e-01 -7.57397175e-01 6.77944303e-01
-1.76409829e+00 -8.78439724e-01 9.47406739e-02 1.96741796e+00
1.11193275e+00 1.00090608e-01 -7.97404274e-02 -1.52352199e-01
8.46152663e-01 3.54397833e-01 -6.08839869e-01 -7.35008940e-02
-4.42801803e-01 1.08111829e-01 3.82373065e-01 5.06086826e-01
-1.24205315e+00 1.17020309e+00 5.51962090e+00 7.30738223e-01
-1.01343215e+00 1.75917104e-01 9.37053204e-01 -1.03175774e-01
-6.10874176e-01 -1.72425374e-01 -6.14000916e-01 4.51953828e-01
6.46153748e-01 8.19658116e-02 4.59422529e-01 6.13995075e-01
2.92065203e-01 2.62635369e-02 -1.22982955e+00 1.22197998e+00
6.52938724e-01 -1.45387328e+00 4.89831269e-01 -1.28137156e-01
8.48461211e-01 6.44768402e-02 2.96227366e-01 1.32500947e-01
7.11632743e-02 -1.10112321e+00 8.63010585e-01 3.86990607e-01
1.08314657e+00 -5.23354352e-01 2.02927560e-01 2.51139045e-01
-8.98878336e-01 -2.25289181e-01 -2.82421619e-01 2.54901081e-01
4.48106490e-02 5.96903563e-01 -4.68617111e-01 2.84983069e-01
6.59867823e-01 6.88278377e-01 -8.84753764e-01 6.50919557e-01
-1.49372518e-01 2.97094256e-01 1.17287934e-01 2.48134598e-01
3.20344716e-01 -9.71008539e-02 2.87234396e-01 1.37593591e+00
-5.81761859e-02 -2.97422051e-01 1.58123434e-01 1.23409092e+00
-2.95907408e-01 1.92967087e-01 -6.98074937e-01 5.81082664e-02
1.89038709e-01 1.22979295e+00 -3.02531004e-01 -3.51234436e-01
-7.68235803e-01 1.37820804e+00 4.77760643e-01 5.11904359e-01
-6.58735335e-01 -2.14217186e-01 3.72215033e-01 9.24856737e-02
2.88854152e-01 8.72879103e-02 -2.99323857e-01 -1.21318972e+00
3.60832810e-01 -9.05872583e-01 2.36721918e-01 -9.81799364e-01
-1.49625170e+00 5.46019435e-01 -3.02048862e-01 -1.39496207e+00
-2.11143628e-01 -5.51442564e-01 -4.20439124e-01 9.79858100e-01
-2.05628562e+00 -1.49355602e+00 -1.75171822e-01 9.02408361e-01
8.13462973e-01 -2.35099763e-01 5.16546130e-01 4.29829687e-01
-2.98104584e-01 7.79242575e-01 -1.62185058e-01 1.13470495e-01
1.00605559e+00 -1.16788423e+00 4.73205268e-01 9.78809059e-01
3.21733832e-01 4.68621999e-01 4.00804192e-01 -4.32262957e-01
-1.12542665e+00 -1.54931676e+00 9.49084342e-01 -4.88966316e-01
6.61281049e-01 -6.65263891e-01 -9.11657929e-01 6.07532740e-01
2.93438554e-01 2.07688972e-01 4.22655523e-01 -3.34990263e-01
-5.07412851e-01 -3.34519669e-02 -8.15701723e-01 7.79956758e-01
1.08124220e+00 -8.81241441e-01 -6.82381928e-01 6.66045249e-01
1.04391921e+00 -9.36570391e-02 -3.56138945e-01 2.25321323e-01
1.13302790e-01 -7.70130098e-01 1.10196173e+00 -7.47966230e-01
8.92391026e-01 -2.26579607e-01 -2.00872093e-01 -1.18794191e+00
-2.21047789e-01 -4.91606146e-01 6.16087299e-03 1.56265962e+00
5.89938879e-01 -3.27888280e-01 5.64382672e-01 2.10497186e-01
-2.10299164e-01 -5.40436208e-01 -7.87730336e-01 -5.15867233e-01
2.35761702e-01 -4.44590658e-01 1.69275671e-01 1.09376836e+00
-1.95927545e-01 6.11865342e-01 -6.92385972e-01 -1.69898644e-02
7.28719413e-01 1.39757872e-01 5.85405588e-01 -8.62276614e-01
-3.38592678e-01 -3.55612516e-01 -1.52979335e-02 -1.24348235e+00
4.10154432e-01 -1.18023813e+00 2.41995737e-01 -1.46308243e+00
5.87100089e-01 -7.21193254e-02 -2.61411399e-01 4.52877015e-01
-3.32436115e-01 5.24782658e-01 2.33825713e-01 3.39089185e-01
-7.48898208e-01 6.56850934e-01 1.48124146e+00 -4.87822711e-01
8.59514400e-02 -2.34070405e-01 -1.00881672e+00 6.98951125e-01
3.85432810e-01 -3.24092776e-01 -3.87851059e-01 -7.93505192e-01
3.85028683e-02 -1.38035312e-01 5.82243800e-01 -5.96046984e-01
2.00914234e-01 -1.26999184e-01 6.05858088e-01 -4.99417156e-01
2.13907868e-01 -8.07593286e-01 -5.86133838e-01 2.70202041e-01
-8.64073157e-01 3.89657281e-02 7.95740783e-02 6.18803382e-01
-2.35892087e-01 -1.56411901e-01 1.06610954e+00 -1.91848651e-01
-7.48461664e-01 3.35699975e-01 -6.38663545e-02 3.62924486e-01
8.28510284e-01 -6.65864125e-02 -2.89765924e-01 -4.49667811e-01
-6.34632945e-01 4.26290482e-01 5.30566752e-01 7.40490735e-01
8.27628255e-01 -1.38640559e+00 -8.65893424e-01 5.01943290e-01
6.16472721e-01 -5.46875671e-02 2.08062068e-01 5.20017445e-01
-5.04146181e-02 3.17196786e-01 -3.59646417e-02 -8.10198605e-01
-9.81785238e-01 8.85628879e-01 3.11122805e-01 -1.18950556e-04
-7.77759910e-01 7.47629762e-01 8.61987054e-01 -2.13044479e-01
4.76161450e-01 -1.27391219e-01 -2.00722530e-01 1.67947561e-02
5.48432052e-01 -3.25856566e-01 2.91622002e-02 -8.60923469e-01
-3.57533008e-01 7.43720651e-01 -1.75622657e-01 -2.37101465e-01
1.19821060e+00 -4.75355625e-01 3.98667902e-02 3.40301454e-01
1.54775834e+00 -2.93353088e-02 -1.50267684e+00 -6.70060515e-01
1.84905510e-02 -3.44866276e-01 1.16995908e-01 -9.03573096e-01
-1.15229666e+00 9.64982331e-01 6.19403541e-01 -1.80645093e-01
1.18196690e+00 2.64454961e-01 6.37833953e-01 1.91192374e-01
-1.70808867e-01 -1.13597655e+00 6.15897298e-01 4.03414965e-01
1.09221220e+00 -1.70957780e+00 -2.49387845e-01 -1.00157574e-01
-9.91455853e-01 8.43015373e-01 7.06015825e-01 1.63293872e-02
4.06511933e-01 1.63071916e-01 2.70250320e-01 -8.41613561e-02
-7.63020754e-01 -5.87956905e-01 6.45713627e-01 7.81122446e-01
3.44650090e-01 -3.66288811e-01 -5.05645648e-02 3.02105129e-01
6.22408390e-02 -3.53951186e-01 2.63547897e-01 5.72674751e-01
-4.60953683e-01 -9.18451905e-01 -2.64482409e-01 4.05783683e-01
-2.61945605e-01 -6.55147135e-01 -3.68804783e-01 5.53190231e-01
5.66466637e-02 8.97476375e-01 3.84584427e-01 -2.18687266e-01
2.34100252e-01 1.03811286e-01 3.86502266e-01 -7.93562770e-01
-2.16334596e-01 1.89683005e-01 -2.05530107e-01 -2.81261742e-01
-4.22609419e-01 -4.63412821e-01 -1.07548428e+00 2.90190428e-01
-1.14061296e-01 -2.30797231e-01 5.91045558e-01 1.06157410e+00
3.59282434e-01 5.36479652e-01 9.56935525e-01 -8.40125263e-01
-4.66944069e-01 -8.77691388e-01 -2.41423950e-01 8.82230997e-01
6.29302800e-01 -5.11795044e-01 -4.70867962e-01 3.60324472e-01]
|
[10.800907135009766, 1.5144157409667969]
|
34945a54-d87c-4daf-b45d-7ab3346242f7
|
referring-video-object-segmentation-with
|
2307.00536
| null |
https://arxiv.org/abs/2307.00536v1
|
https://arxiv.org/pdf/2307.00536v1.pdf
|
Referring Video Object Segmentation with Inter-Frame Interaction and Cross-Modal Correlation
|
Referring video object segmentation (RVOS) aims to segment the target object in a video sequence described by a language expression. Typical query-based methods process the video sequence in a frame-independent manner to reduce the high computational cost, which however affects the performance due to the lack of inter-frame interaction for temporal coherence modeling and spatio-temporal representation learning of the referred object. Besides, they directly adopt the raw and high-level sentence feature as the language queries to decode the visual features, where the weak correlation between visual and linguistic features also increases the difficulty of decoding the target information and limits the performance of the model. In this paper, we proposes a novel RVOS framework, dubbed IFIRVOS, to address these issues. Specifically, we design a plug-and-play inter-frame interaction module in the Transformer decoder to efficiently learn the spatio-temporal features of the referred object, so as to decode the object information in the video sequence more precisely and generate more accurate segmentation results. Moreover, we devise the vision-language interaction module before the multimodal Transformer to enhance the correlation between the visual and linguistic features, thus facilitating the process of decoding object information from visual features by language queries in Transformer decoder and improving the segmentation performance. Extensive experimental results on three benchmarks validate the superiority of our IFIRVOS over state-of-the-art methods and the effectiveness of our proposed modules.
|
['Lefei Zhang', 'Fu Rong', 'Meng Lan']
|
2023-07-02
| null | null | null | null |
['referring-video-object-segmentation', 'video-object-segmentation', 'video-semantic-segmentation']
|
['computer-vision', 'computer-vision', 'computer-vision']
|
[-4.47246172e-02 -2.72452980e-01 -2.37415522e-01 -3.16824466e-01
-5.08990765e-01 -2.90246338e-01 3.13859850e-01 -6.85819313e-02
-4.25713301e-01 1.10140666e-01 8.44770968e-02 4.26436178e-02
1.16965376e-01 -7.21673608e-01 -5.98719835e-01 -6.09809279e-01
4.27606970e-01 8.68218169e-02 7.84887791e-01 -5.52540794e-02
2.15940595e-01 1.49907976e-01 -1.42214656e+00 4.56627727e-01
8.89525473e-01 1.08192146e+00 7.54582644e-01 3.19396645e-01
-5.31708181e-01 1.00957251e+00 -3.44037324e-01 -1.99025974e-01
-1.93017036e-01 -6.13164961e-01 -6.97154999e-01 5.58215678e-01
9.44770034e-03 -4.89780724e-01 -6.02384746e-01 1.04762244e+00
2.67721146e-01 1.33359551e-01 4.32174861e-01 -1.13091326e+00
-3.27068061e-01 3.68296832e-01 -7.50020564e-01 3.32771540e-01
6.19817197e-01 2.01165363e-01 7.36297846e-01 -8.54953945e-01
5.62931478e-01 1.40289021e+00 1.94243550e-01 2.47338951e-01
-6.77603722e-01 -4.57722992e-01 4.85148251e-01 7.05796003e-01
-1.59734714e+00 -4.27919328e-01 1.03417695e+00 -4.22082752e-01
5.25486827e-01 1.87195390e-01 8.09818685e-01 6.83330476e-01
-3.93192098e-02 1.38086307e+00 6.19546652e-01 -1.93690881e-01
-2.85066348e-02 -1.13336451e-01 7.71530941e-02 9.96661961e-01
-3.94577026e-01 -1.73395365e-01 -5.07316411e-01 3.45459402e-01
8.25715184e-01 1.97448656e-01 -4.14466262e-01 -1.83646157e-01
-1.27810836e+00 4.96315509e-01 4.14045095e-01 4.78025019e-01
-3.72955173e-01 4.10760827e-02 6.01882696e-01 -8.99584070e-02
2.49997064e-01 -5.06393611e-01 -2.20032722e-01 -1.76045299e-01
-9.45571303e-01 -1.94586247e-01 4.29304957e-01 1.06753850e+00
7.65671790e-01 -1.22379340e-01 -3.75439286e-01 6.77015543e-01
5.11239588e-01 4.47169662e-01 4.30882037e-01 -9.16538119e-01
6.75983787e-01 7.75775790e-01 -2.99372002e-02 -1.13768995e+00
-1.52999550e-01 -2.08857939e-01 -6.04934454e-01 -4.69455719e-01
1.24530576e-01 1.46272853e-01 -8.79933536e-01 1.37227881e+00
4.15270507e-01 3.87449652e-01 9.84785846e-04 1.26808679e+00
1.10607183e+00 1.14133608e+00 1.58015162e-01 -5.79238296e-01
1.55760503e+00 -1.06543398e+00 -9.36675787e-01 -2.74775237e-01
4.97291863e-01 -8.81748319e-01 1.04170609e+00 5.11956401e-02
-9.82281804e-01 -8.94924045e-01 -6.83489144e-01 -1.78689435e-01
1.02816157e-01 3.32682759e-01 2.68695801e-01 1.29082456e-01
-6.26201689e-01 -4.50323597e-02 -8.71802926e-01 -2.33452663e-01
3.81773770e-01 1.84827507e-01 -1.72291726e-01 -1.98139384e-01
-1.11215496e+00 4.65876997e-01 7.76986361e-01 4.02414858e-01
-9.05403733e-01 -3.09004784e-01 -8.89233232e-01 3.52838524e-02
7.96424150e-01 -4.89052981e-01 1.06326091e+00 -1.21377039e+00
-1.30086744e+00 4.96750146e-01 -5.14686286e-01 -7.34702274e-02
5.41295409e-01 -1.54009730e-01 -2.88201392e-01 6.14156723e-01
1.98239401e-01 7.62119412e-01 7.47779906e-01 -1.19824779e+00
-1.00505459e+00 -2.35597417e-01 2.06259623e-01 4.92275804e-01
-2.17526257e-01 7.06532374e-02 -1.52621579e+00 -5.01998842e-01
2.24539459e-01 -5.71447253e-01 -1.04297011e-03 -9.69338343e-02
-2.16002896e-01 -4.41127330e-01 1.15211701e+00 -8.20324898e-01
1.35476899e+00 -2.50674582e+00 2.60030359e-01 -3.38452905e-02
2.20581125e-02 3.24577481e-01 -1.61436766e-01 2.33593568e-01
2.24521160e-01 -2.07404658e-01 -1.40805230e-01 -6.29183501e-02
-3.19365025e-01 3.74436945e-01 -3.91821079e-02 3.59033316e-01
6.90073892e-02 9.78955448e-01 -9.38631654e-01 -1.08670104e+00
5.39288580e-01 3.85868609e-01 -4.40860778e-01 5.06645679e-01
-3.52453411e-01 5.89278698e-01 -9.03542280e-01 4.91812378e-01
4.43168819e-01 -1.80389345e-01 -1.63566731e-02 -7.07731426e-01
-9.88556370e-02 -1.28671810e-01 -1.06981361e+00 1.83018911e+00
-3.92002106e-01 4.11995590e-01 -4.73998412e-02 -1.23098314e+00
7.77792275e-01 2.05869198e-01 6.29011750e-01 -8.96645427e-01
2.20998824e-01 -5.16582727e-02 -2.84216166e-01 -1.14555109e+00
1.39769718e-01 1.75853670e-01 4.60932329e-02 -7.45493844e-02
-1.67261258e-01 1.92704991e-01 3.54087621e-01 2.97947526e-01
5.05433202e-01 5.09682775e-01 7.44594336e-02 1.17580757e-01
1.10719907e+00 -1.06188118e-01 7.51510680e-01 3.02776694e-01
-1.30186304e-01 5.23482382e-01 4.45626318e-01 -1.40225619e-01
-6.44094169e-01 -8.78142834e-01 1.99956253e-01 9.75976825e-01
8.77634764e-01 -4.77683872e-01 -8.00707161e-01 -6.53465867e-01
-4.81041104e-01 6.12221360e-01 -3.21024776e-01 -1.57665193e-01
-6.64857805e-01 -4.40128148e-01 2.40912229e-01 5.69034874e-01
8.74468565e-01 -9.33567703e-01 -5.44060469e-01 2.42112786e-01
-8.55731010e-01 -1.54909897e+00 -8.03064108e-01 -4.04010952e-01
-7.31716990e-01 -1.03914571e+00 -6.33379638e-01 -1.14288807e+00
7.33957171e-01 4.79647696e-01 6.25763059e-01 2.10747480e-01
3.91811542e-02 4.27031279e-01 -5.94881594e-01 2.42005914e-01
-1.64059162e-01 -1.69707522e-01 -3.95654142e-01 4.04045612e-01
3.21393281e-01 -2.12020934e-01 -7.75723517e-01 4.94760305e-01
-1.12312686e+00 5.17251134e-01 5.88660598e-01 6.27921164e-01
6.62157118e-01 1.12548016e-01 2.63085246e-01 -4.37941343e-01
2.32650891e-01 -2.96358317e-01 -5.90039849e-01 4.59252089e-01
-1.23270854e-01 -8.47172961e-02 5.98405480e-01 -4.50821251e-01
-1.01302302e+00 2.54330367e-01 -1.00941993e-01 -7.05309927e-01
2.69119348e-02 6.46502852e-01 -4.99169827e-01 1.66074663e-01
-1.65311173e-01 8.78967583e-01 4.78148870e-02 -3.88030678e-01
3.70087862e-01 7.06562638e-01 6.10427022e-01 -4.17265713e-01
5.45919478e-01 3.34906012e-01 -1.82524711e-01 -7.66932249e-01
-7.85545170e-01 -7.21158624e-01 -6.08482361e-01 -4.28534687e-01
1.25994205e+00 -1.10585904e+00 -8.40831399e-01 3.59910846e-01
-1.32868052e+00 8.47393721e-02 4.45575602e-02 5.69757462e-01
-5.96910357e-01 8.90437007e-01 -6.03244781e-01 -5.73695481e-01
-2.43284509e-01 -1.57039416e+00 1.31874812e+00 5.08982241e-01
2.35612586e-01 -7.08011508e-01 -5.79091609e-01 6.21115386e-01
-1.76903293e-01 -1.76660776e-01 8.32213402e-01 -2.63144940e-01
-9.79234815e-01 5.34587875e-02 -6.35256290e-01 2.83018827e-01
-1.13525591e-03 2.14779619e-02 -6.19646847e-01 -6.03742786e-02
6.39840513e-02 7.30648190e-02 6.66748583e-01 3.78564656e-01
1.21821094e+00 -6.68982640e-02 -4.11942601e-01 5.18251657e-01
1.36269832e+00 5.07745504e-01 5.92559934e-01 7.49435052e-02
9.76465821e-01 5.04680812e-01 1.04156101e+00 2.94037879e-01
6.40997648e-01 8.43790710e-01 3.88760656e-01 -1.27084911e-01
-8.18910003e-02 -2.84223408e-01 5.88433444e-01 1.22750771e+00
-9.04994607e-02 -3.35071892e-01 -7.38677740e-01 5.09514511e-01
-2.14449239e+00 -8.78103852e-01 -2.56860197e-01 1.68229461e+00
6.08370304e-01 -5.22593744e-02 5.82231954e-02 -4.05371301e-02
8.09882820e-01 2.39018753e-01 -3.71907145e-01 1.46445632e-01
2.72723194e-03 -4.80351239e-01 1.14753775e-01 3.98219436e-01
-9.25780952e-01 1.05337429e+00 4.74171162e+00 1.31323659e+00
-1.15236139e+00 1.74966469e-01 6.19098365e-01 1.26446307e-01
-6.95945770e-02 -5.67177013e-02 -6.32751524e-01 5.86168230e-01
4.21091914e-01 -9.82501209e-02 3.09508353e-01 6.11875892e-01
6.86484754e-01 -2.78585911e-01 -1.02910352e+00 1.37689602e+00
2.76017010e-01 -1.12416089e+00 1.97534591e-01 -2.71556407e-01
3.36006016e-01 -1.87955260e-01 -2.31115162e-01 1.53263167e-01
-5.49284816e-01 -4.86749232e-01 1.03719366e+00 6.92776382e-01
6.76107287e-01 -8.08418214e-01 7.84693658e-01 4.63791281e-01
-1.62093282e+00 -4.82752658e-02 -2.08557487e-01 3.29946727e-01
3.26926440e-01 3.10583621e-01 -5.22438288e-01 8.46985638e-01
7.09347665e-01 1.00173295e+00 -6.38974547e-01 9.53904092e-01
-1.05004534e-01 4.65886891e-01 -2.65334696e-01 2.81746257e-02
4.93696064e-01 -3.89612496e-01 4.81070846e-01 1.10891604e+00
1.75556660e-01 3.59180361e-01 7.18400657e-01 6.54014409e-01
1.82377756e-01 3.33808988e-01 -2.54419953e-01 -1.29565388e-01
2.26971939e-01 1.05985999e+00 -9.05251384e-01 -4.21005368e-01
-7.02097535e-01 1.06808090e+00 -3.23722139e-02 5.72469473e-01
-1.11545348e+00 -2.18245342e-01 1.19563557e-01 1.20782815e-01
6.15885973e-01 -3.16345572e-01 5.83186559e-02 -1.35212672e+00
3.55645955e-01 -8.11086774e-01 2.65267968e-01 -9.10400093e-01
-8.49659979e-01 5.13374925e-01 1.23277128e-01 -1.42092335e+00
-1.12690963e-01 -1.78923592e-01 -3.66315365e-01 5.79783320e-01
-1.36240244e+00 -1.31226134e+00 -4.54528123e-01 8.33783150e-01
1.00776732e+00 2.09410004e-02 7.75898471e-02 4.28780526e-01
-6.73223078e-01 1.15423702e-01 -9.16972533e-02 3.53337139e-01
3.52851629e-01 -5.61420500e-01 -1.54501170e-01 9.09484565e-01
8.40291083e-02 3.25859219e-01 4.35735434e-01 -7.39405334e-01
-1.50930142e+00 -1.10224819e+00 5.37887752e-01 2.67881695e-02
4.27049845e-01 -1.67063460e-01 -8.76467466e-01 3.75193954e-01
1.49804071e-01 -4.66298871e-03 2.98021972e-01 -5.20803094e-01
1.12202562e-01 -2.91912407e-01 -6.62444353e-01 6.70928061e-01
1.03680062e+00 -6.97908342e-01 -7.01652169e-01 2.34703094e-01
8.53411853e-01 -3.83810818e-01 -6.56872571e-01 5.22265077e-01
3.98601323e-01 -8.87835860e-01 9.82140779e-01 -1.00863203e-01
4.44498450e-01 -8.22131395e-01 -1.91139042e-01 -5.70731044e-01
-1.26142316e-02 -2.42434457e-01 3.03533021e-02 1.58403254e+00
-1.02824487e-01 -1.24322094e-01 5.48846841e-01 1.96402252e-01
7.42879044e-03 -8.56869578e-01 -8.45598400e-01 -3.64621043e-01
-6.23356700e-01 -6.13401353e-01 2.95689911e-01 5.48135340e-01
-2.36645326e-01 4.79888588e-01 -3.24381590e-01 2.35342950e-01
3.79836053e-01 3.81950408e-01 5.57500005e-01 -6.91949248e-01
-4.00405861e-02 -2.13856891e-01 -6.10437036e-01 -1.70056069e+00
2.22411186e-01 -6.65257573e-01 1.37839198e-01 -1.60094094e+00
3.39594871e-01 -8.79104212e-02 -1.16470501e-01 2.50385255e-01
-4.16626453e-01 1.32518128e-01 5.18194914e-01 3.63406181e-01
-1.10445678e+00 7.91475475e-01 1.57000470e+00 -3.24722439e-01
-1.51819006e-01 -1.99725822e-01 -2.55805969e-01 7.77065158e-01
2.11714044e-01 -3.10142249e-01 -6.69763505e-01 -6.32568181e-01
-2.86981493e-01 5.57162404e-01 4.28483546e-01 -8.06227028e-01
3.97896588e-01 -2.53449261e-01 3.88711303e-01 -9.30918872e-01
3.89444649e-01 -1.05744421e+00 -3.52027528e-02 5.12508214e-01
-9.27862078e-02 -1.15383841e-01 4.63981554e-03 6.89009428e-01
-5.60151219e-01 -2.27045432e-01 6.30181491e-01 -9.29184631e-02
-1.21938753e+00 4.25941885e-01 -4.09092695e-01 5.02359383e-02
1.23966551e+00 -3.95276427e-01 -2.89808940e-02 -3.45921189e-01
-6.62051976e-01 5.68273187e-01 3.86435091e-01 6.15898132e-01
9.10390675e-01 -1.15432131e+00 -4.93318439e-01 2.74167210e-01
1.25565782e-01 1.79822698e-01 5.59144020e-01 1.09982669e+00
-7.13056266e-01 2.30744660e-01 9.97454277e-04 -1.08739555e+00
-1.33522022e+00 7.31928289e-01 3.31756681e-01 7.87739754e-02
-7.63204038e-01 5.09718716e-01 7.01909423e-01 2.47277126e-01
2.10248187e-01 -2.42605135e-01 -5.43502390e-01 1.78249866e-01
4.44101870e-01 6.42287657e-02 -3.56608212e-01 -1.08968544e+00
-4.14686948e-01 9.50368226e-01 -1.12765640e-01 1.57331408e-03
9.06066000e-01 -7.33400524e-01 -2.30616674e-01 5.06161749e-01
1.23542845e+00 -2.06537396e-01 -1.26351261e+00 -4.32590485e-01
-2.06864998e-01 -5.06513536e-01 1.64672017e-01 -2.76933014e-01
-1.31916535e+00 9.00567651e-01 6.22869551e-01 -2.79738773e-02
1.36160004e+00 7.54362866e-02 1.07097566e+00 1.32853389e-01
2.12468430e-01 -9.62168634e-01 1.78569511e-01 2.81920791e-01
5.20474136e-01 -9.78020608e-01 -7.04373345e-02 -7.67637372e-01
-8.95725846e-01 1.20912182e+00 6.83744669e-01 1.23042747e-01
4.47849751e-01 4.94045205e-02 1.79353476e-01 2.07682531e-02
-4.82921779e-01 -3.37929159e-01 4.99586552e-01 4.49124455e-01
1.91874638e-01 -3.31439078e-01 -4.64168876e-01 5.87456167e-01
3.89003485e-01 2.51506329e-01 1.70426235e-01 7.23346651e-01
-4.56162632e-01 -8.93803120e-01 -2.76156485e-01 9.13974941e-02
-3.56479466e-01 -9.71797388e-03 5.06240427e-02 5.27971029e-01
3.43055874e-01 1.09777772e+00 9.28464308e-02 -3.60819310e-01
1.95565224e-01 -1.70699328e-01 2.81030089e-01 -2.86830902e-01
-2.77513981e-01 7.17663765e-01 -8.37448165e-02 -6.61568105e-01
-8.68767679e-01 -4.88532186e-01 -1.68690705e+00 8.95100683e-02
-1.91880763e-01 3.13432574e-01 2.91249871e-01 1.34786308e+00
2.02717319e-01 6.99540734e-01 7.19244182e-01 -6.88601375e-01
-4.12933938e-02 -6.08382285e-01 -2.05174416e-01 4.11229283e-01
1.68464303e-01 -5.22226095e-01 1.43248573e-01 4.40364152e-01]
|
[9.540692329406738, 0.253795325756073]
|
9d80dde2-2e5b-4203-b652-4edf1f4f56bd
|
an-empirical-study-of-end-to-end-video
|
2209.01540
| null |
https://arxiv.org/abs/2209.01540v5
|
https://arxiv.org/pdf/2209.01540v5.pdf
|
An Empirical Study of End-to-End Video-Language Transformers with Masked Visual Modeling
|
Masked visual modeling (MVM) has been recently proven effective for visual pre-training. While similar reconstructive objectives on video inputs (e.g., masked frame modeling) have been explored in video-language (VidL) pre-training, previous studies fail to find a truly effective MVM strategy that can largely benefit the downstream performance. In this work, we systematically examine the potential of MVM in the context of VidL learning. Specifically, we base our study on a fully end-to-end VIdeO-LanguagE Transformer (VIOLET), where the supervision from MVM training can be backpropagated to the video pixel space. In total, eight different reconstructive targets of MVM are explored, from low-level pixel values and oriented gradients to high-level depth maps, optical flow, discrete visual tokens, and latent visual features. We conduct comprehensive experiments and provide insights into the factors leading to effective MVM training, resulting in an enhanced model VIOLETv2. Empirically, we show VIOLETv2 pre-trained with MVM objective achieves notable improvements on 13 VidL benchmarks, ranging from video question answering, video captioning, to text-to-video retrieval.
|
['Zicheng Liu', 'Lijuan Wang', 'William Yang Wang', 'Kevin Lin', 'Zhe Gan', 'Linjie Li', 'Tsu-Jui Fu']
|
2022-09-04
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Fu_An_Empirical_Study_of_End-to-End_Video-Language_Transformers_With_Masked_Visual_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Fu_An_Empirical_Study_of_End-to-End_Video-Language_Transformers_With_Masked_Visual_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['video-question-answering']
|
['computer-vision']
|
[ 4.05597657e-01 1.41647875e-01 -3.79468024e-01 -2.91548222e-01
-9.85551834e-01 -4.16973859e-01 7.75744081e-01 -2.30925947e-01
-3.37009519e-01 4.50047314e-01 5.14340639e-01 -5.18422663e-01
3.44742805e-01 -4.55775619e-01 -1.22569060e+00 -4.19982672e-01
1.00999493e-02 9.79677364e-02 1.74033895e-01 1.11197531e-01
2.90637612e-01 1.85944855e-01 -1.27210534e+00 8.86610627e-01
4.86953437e-01 7.98771143e-01 3.36262286e-01 8.35172296e-01
-2.73063630e-01 1.49236619e+00 -4.31438863e-01 -5.22854686e-01
2.19157934e-01 -4.82093722e-01 -8.53265941e-01 7.17306659e-02
8.74821186e-01 -8.23552549e-01 -7.12052643e-01 7.19848990e-01
3.34237903e-01 1.20279409e-01 5.87930679e-01 -1.35010779e+00
-7.16061056e-01 7.33495235e-01 -5.02845466e-01 2.86357343e-01
4.26172942e-01 6.26970351e-01 1.09372354e+00 -1.36326349e+00
8.95269513e-01 1.62621105e+00 4.38760340e-01 7.06793904e-01
-1.29944634e+00 -4.12636191e-01 3.98259550e-01 4.08026904e-01
-1.06393278e+00 -5.91475785e-01 7.31102407e-01 -5.86743772e-01
1.06204069e+00 1.16044462e-01 5.26295960e-01 1.28566098e+00
3.05772454e-01 1.17881942e+00 9.73876774e-01 -3.16680014e-01
-9.43049975e-03 6.57157600e-02 -3.06436598e-01 9.06735063e-01
-2.27072258e-02 2.23943070e-01 -1.06422353e+00 2.51604110e-01
8.33765805e-01 -1.60177112e-01 -4.64745909e-01 -3.28354836e-01
-1.21199226e+00 6.44279957e-01 4.13510919e-01 -8.78459364e-02
-2.83457041e-01 6.20445907e-01 3.02595884e-01 2.15005532e-01
3.74021173e-01 2.74947733e-01 -2.13550150e-01 -8.91373083e-02
-1.12919736e+00 -1.90481078e-03 2.77609766e-01 8.09799075e-01
8.32587361e-01 4.58499432e-01 -7.42791593e-01 4.57437813e-01
4.85958815e-01 4.55239683e-01 1.31829664e-01 -1.41037178e+00
7.27655709e-01 2.89970785e-01 -7.80027434e-02 -7.48351872e-01
3.31043527e-02 -1.26719177e-01 -4.73532438e-01 2.99406379e-01
2.60773152e-01 1.04799923e-02 -1.08333361e+00 1.71742165e+00
7.81490579e-02 5.70649385e-01 1.50895715e-01 1.07344294e+00
1.11726582e+00 9.71867204e-01 2.11285070e-01 -4.17174064e-02
9.79505658e-01 -1.45781386e+00 -6.28829539e-01 -3.73657465e-01
5.30568004e-01 -6.05474710e-01 1.24096477e+00 1.91707984e-01
-1.53627062e+00 -6.03729725e-01 -7.70789444e-01 -5.16265094e-01
1.03162397e-02 -1.08002700e-01 4.81017172e-01 2.25690752e-01
-1.38265300e+00 4.26592201e-01 -8.61737847e-01 -1.91258952e-01
5.82802892e-01 1.62692398e-01 -4.49559301e-01 -4.34609145e-01
-1.03279150e+00 6.31620824e-01 7.09293187e-02 2.91298747e-01
-1.64451826e+00 -8.91984999e-01 -9.16276038e-01 1.75541099e-02
2.59152979e-01 -9.84109640e-01 1.15075386e+00 -1.15006852e+00
-1.34677553e+00 8.09387922e-01 -4.03417021e-01 -7.51421988e-01
7.18975723e-01 -1.50969386e-01 8.96593034e-02 6.96241140e-01
7.45833889e-02 1.37922239e+00 1.18079448e+00 -1.38761365e+00
-4.12117630e-01 1.89069077e-01 3.39700311e-01 1.78513229e-01
-4.23312813e-01 3.44443731e-02 -9.85646427e-01 -7.71341324e-01
-4.33318704e-01 -6.56693876e-01 -1.57852378e-02 2.51732349e-01
-1.73189089e-01 1.05614610e-01 6.65854096e-01 -1.02864194e+00
1.08082998e+00 -1.99399519e+00 3.26492369e-01 -1.79149687e-01
3.30836385e-01 2.51734585e-01 -3.66545886e-01 2.44620860e-01
6.91055262e-04 3.12739238e-02 -1.98255211e-01 -6.84113562e-01
-1.35724470e-01 2.56185740e-01 -4.81449544e-01 2.84617871e-01
4.89603549e-01 1.53555048e+00 -8.68126512e-01 -8.29523027e-01
5.31757593e-01 7.28441834e-01 -9.39961433e-01 2.56168455e-01
-4.54571694e-01 4.61352348e-01 -1.27925202e-01 9.24438298e-01
3.89898449e-01 -5.08423328e-01 -8.85244459e-03 -4.14018422e-01
1.02309778e-01 2.09946677e-01 -4.20601457e-01 1.70794535e+00
-5.08804321e-01 1.27822900e+00 1.19019367e-01 -9.48866367e-01
3.64195257e-01 2.70167023e-01 6.98382854e-01 -1.00029123e+00
-7.75857568e-02 -1.38954833e-01 -3.59383523e-01 -5.50347686e-01
5.56926906e-01 1.48687288e-01 3.37936878e-01 5.44618890e-02
2.79680163e-01 2.77994722e-01 8.47944152e-03 5.24291575e-01
9.92953122e-01 5.35473585e-01 -1.07279375e-01 1.02198891e-01
3.55494529e-01 4.60046604e-02 3.38437229e-01 8.79394472e-01
-1.71075568e-01 8.15162480e-01 5.71187496e-01 -9.37380865e-02
-1.06269979e+00 -1.20763803e+00 2.58120149e-01 1.17531335e+00
1.98371977e-01 -6.18762255e-01 -7.74697840e-01 -5.72751939e-01
-5.46443239e-02 6.24693036e-01 -5.48300624e-01 -2.87274539e-01
-7.59811640e-01 -2.29140863e-01 5.18600643e-01 6.16458654e-01
3.25685740e-01 -1.16686797e+00 -5.61364651e-01 1.71798229e-01
-3.79734129e-01 -1.62633669e+00 -6.30988240e-01 -4.84197587e-03
-9.59049046e-01 -6.54123187e-01 -7.23020315e-01 -7.55027473e-01
6.57116950e-01 6.23845279e-01 1.43095100e+00 7.88066462e-02
-1.84527904e-01 8.25403273e-01 -2.81419635e-01 -6.33301120e-03
-5.23867965e-01 -2.00391382e-01 -2.81844139e-01 1.81709632e-01
-9.26931873e-02 -3.40928167e-01 -8.69341314e-01 1.20167144e-01
-9.80350733e-01 5.70790291e-01 7.51515687e-01 8.20204675e-01
5.98708630e-01 -6.66327596e-01 7.92218223e-02 -7.52551258e-01
1.15489557e-01 -3.95618081e-01 -3.88556182e-01 5.06520987e-01
-5.32218575e-01 8.30879211e-02 4.79702145e-01 -3.61865848e-01
-9.12381351e-01 6.47634119e-02 -2.39136890e-01 -1.21834040e+00
2.48064861e-01 4.39974695e-01 -9.06142294e-02 -1.57493591e-01
3.09869140e-01 4.27637458e-01 -2.73176115e-02 -2.24098906e-01
5.85216165e-01 1.81817353e-01 7.18651712e-01 -5.26274562e-01
8.22116077e-01 6.70834899e-01 -3.94461490e-02 -7.53292382e-01
-7.28528738e-01 -2.69033462e-01 -1.97770149e-01 -5.67948818e-01
1.09636116e+00 -1.21919167e+00 -6.16868615e-01 1.82221338e-01
-1.19165766e+00 -7.83831537e-01 -2.54880756e-01 2.67963916e-01
-5.36825955e-01 4.87513781e-01 -7.53250360e-01 -4.79864568e-01
-3.81492049e-01 -1.40265322e+00 1.32791770e+00 -1.70965835e-01
8.40644178e-04 -1.02475929e+00 -4.07559305e-01 7.73824334e-01
3.41889322e-01 8.30033198e-02 8.85698736e-01 4.29473817e-02
-1.22426486e+00 4.48314607e-01 -3.23452681e-01 4.41892743e-01
-2.72241384e-01 1.18415125e-01 -1.03393960e+00 -4.84216511e-01
-2.16921091e-01 -3.64168078e-01 1.34384441e+00 6.09435022e-01
1.19752479e+00 -3.72721076e-01 -2.46312886e-01 1.04464889e+00
1.31033099e+00 -8.82123485e-02 7.05216587e-01 1.95430949e-01
1.17612302e+00 4.12790358e-01 5.48200011e-01 1.69668585e-01
4.97841209e-01 7.44622409e-01 7.27287591e-01 -3.25391799e-01
-6.82625294e-01 -6.37909353e-01 9.92028415e-01 6.61594629e-01
3.68870869e-02 -4.24472511e-01 -7.32223034e-01 3.80237371e-01
-1.72201514e+00 -9.70787466e-01 1.32596731e-01 1.89406443e+00
5.69764435e-01 2.68654436e-01 -1.22695059e-01 -3.21626723e-01
4.80396062e-01 4.65626627e-01 -6.51564062e-01 -1.24887906e-01
-3.71704787e-01 6.67634457e-02 5.42482853e-01 8.76395345e-01
-8.98156285e-01 1.18177462e+00 6.42549515e+00 6.13163590e-01
-1.32750881e+00 3.31036717e-01 7.50481844e-01 -4.24450278e-01
-8.44995856e-01 2.47900680e-01 -6.82621956e-01 3.97679031e-01
7.59199083e-01 2.11239934e-01 5.13361454e-01 6.47353828e-01
4.21289772e-01 1.50243938e-01 -1.22649539e+00 1.29765081e+00
2.50883341e-01 -1.84189296e+00 6.30006015e-01 -3.20729427e-02
8.44726622e-01 2.36004293e-01 2.88482606e-01 4.97508228e-01
-3.24307308e-02 -1.19267535e+00 1.09197390e+00 4.82061744e-01
9.41104591e-01 -3.18271846e-01 1.71993539e-01 -3.64034809e-02
-1.16879582e+00 -1.70182854e-01 -1.56509325e-01 2.06626236e-01
4.87239420e-01 2.68904179e-01 -8.01343560e-01 3.46374363e-01
6.46009028e-01 1.08984804e+00 -6.77297056e-01 6.76359594e-01
-8.50718096e-02 9.17334855e-01 7.18809143e-02 2.92334020e-01
5.17933547e-01 5.24056470e-03 5.21756589e-01 1.24904728e+00
3.41616981e-02 -2.51065522e-01 1.07468516e-01 9.47410166e-01
-2.44543150e-01 -9.41345990e-02 -5.46494544e-01 -1.28259033e-01
3.90192680e-02 9.35188651e-01 -5.23795485e-01 -3.70649040e-01
-7.85479963e-01 1.26358926e+00 2.94602066e-01 8.05807710e-01
-1.09813368e+00 3.86971951e-01 7.98413932e-01 4.59854692e-01
4.39070493e-01 -2.47465059e-01 9.42185707e-03 -1.38161147e+00
6.92781806e-02 -9.29766834e-01 2.39916086e-01 -1.11936033e+00
-8.92191827e-01 3.99695307e-01 3.35018858e-02 -1.15458953e+00
-4.48613942e-01 -6.73350871e-01 -4.59906250e-01 5.91072083e-01
-1.84509623e+00 -1.15826166e+00 -4.81918901e-01 7.22729564e-01
9.41009641e-01 -1.59594581e-01 2.72336841e-01 5.26798487e-01
-4.07711685e-01 6.48655653e-01 6.39991164e-02 1.53739482e-01
7.33320117e-01 -8.67836595e-01 6.06015861e-01 1.19097888e+00
5.43410778e-01 3.78585070e-01 5.65557718e-01 -4.70094740e-01
-1.86321318e+00 -1.32476807e+00 4.52685148e-01 -6.05029166e-01
4.19905990e-01 -5.12019932e-01 -7.45568752e-01 7.44291067e-01
3.75292510e-01 2.53881365e-01 6.10608198e-02 -5.99901378e-01
-3.47899377e-01 -5.57072572e-02 -5.60507774e-01 7.23591089e-01
1.09902942e+00 -8.39381337e-01 -9.55405161e-02 2.86054492e-01
9.90621269e-01 -5.12225866e-01 -5.39610445e-01 4.04016793e-01
4.61173117e-01 -1.07237077e+00 1.40673316e+00 -7.79940248e-01
8.95710051e-01 -3.41635704e-01 -2.83805609e-01 -8.60765934e-01
-9.10452083e-02 -7.37093210e-01 -5.31177163e-01 1.10290909e+00
2.84826726e-01 -1.20688081e-01 8.46447110e-01 3.38773042e-01
-1.75290406e-01 -9.65821981e-01 -7.07022369e-01 -3.88722241e-01
6.59746584e-04 -6.88359439e-01 7.99218267e-02 7.39662647e-01
-5.56562960e-01 3.61817300e-01 -8.56184661e-01 -4.76689711e-02
6.23970628e-01 6.10747598e-02 7.88711905e-01 -6.18910789e-01
-6.42173171e-01 -5.26865900e-01 -2.81658500e-01 -1.67780554e+00
4.05647606e-01 -1.06411445e+00 -2.96021756e-02 -1.88213122e+00
3.19252789e-01 1.25270665e-01 -2.85804331e-01 4.00650471e-01
-3.11853468e-01 3.30945313e-01 4.95328546e-01 2.32130602e-01
-7.71791041e-01 5.11535943e-01 1.45356774e+00 -4.43339586e-01
7.02568814e-02 -4.57383394e-01 -5.66108704e-01 5.15811622e-01
4.13210362e-01 -2.85928696e-01 -6.29283965e-01 -9.49515104e-01
1.77177459e-01 4.41827267e-01 7.48810947e-01 -6.28955305e-01
1.27425209e-01 -1.80513620e-01 3.40057045e-01 -4.88847047e-01
5.95787883e-01 -5.56338668e-01 -1.24215275e-01 3.57454002e-01
-5.69005191e-01 1.35584503e-01 4.80534881e-02 5.20034075e-01
-4.10999417e-01 -6.81267157e-02 6.17150187e-01 -1.00248486e-01
-1.12702274e+00 4.40974414e-01 -2.52967179e-01 3.54657620e-01
7.35356331e-01 -3.75017494e-01 -2.84387201e-01 -6.78346217e-01
-5.80699980e-01 3.07397097e-01 5.12814343e-01 6.19982839e-01
9.88321960e-01 -1.15004253e+00 -8.26086521e-01 3.17683034e-02
-2.10883934e-03 -1.04794510e-01 2.99465030e-01 9.00725007e-01
-6.69620454e-01 5.25284946e-01 -1.16913579e-01 -9.37827229e-01
-1.20527112e+00 7.24011481e-01 3.85674030e-01 -4.62238602e-02
-9.14180636e-01 9.32144403e-01 6.48786962e-01 1.34775117e-01
5.89833438e-01 -5.99787414e-01 1.28478065e-01 -1.60293400e-01
3.67653579e-01 6.83896095e-02 -1.35397211e-01 -6.66245818e-01
-2.99295515e-01 5.21363139e-01 -1.89850815e-02 -2.90954262e-01
1.05210912e+00 -2.60093361e-01 5.55191636e-02 2.70120323e-01
1.55074441e+00 -1.29454151e-01 -1.81941020e+00 -1.80410206e-01
-3.57939780e-01 -5.11267602e-01 1.42965555e-01 -3.88898969e-01
-1.44304204e+00 1.20071483e+00 3.94847155e-01 -3.53693068e-01
1.01835370e+00 1.41568389e-02 8.44867826e-01 1.83716446e-01
1.40921891e-01 -7.47026324e-01 5.63335896e-01 3.38179976e-01
6.84235275e-01 -1.36752629e+00 -1.17855974e-01 -2.79099137e-01
-8.55949819e-01 8.99654448e-01 6.19368017e-01 8.93765762e-02
3.65224659e-01 2.53013760e-01 9.89576802e-02 7.34692365e-02
-1.15380478e+00 -6.56064376e-02 3.46932679e-01 3.74345601e-01
2.64210194e-01 -3.01608264e-01 2.46010482e-01 -5.14018640e-04
2.32950360e-01 -2.87397504e-02 4.48741794e-01 7.43977010e-01
-7.05664605e-02 -7.11296797e-01 -3.00735563e-01 2.86893725e-01
-4.87353563e-01 -4.43775356e-01 -2.55354375e-01 6.17049694e-01
-6.33103326e-02 7.65569389e-01 9.41429585e-02 -3.76107961e-01
1.93775967e-02 -1.15683384e-01 5.82766831e-01 -2.47523665e-01
-4.40163493e-01 8.07006136e-02 -1.65759362e-02 -8.71799707e-01
-5.85454404e-01 -5.80091059e-01 -1.30401540e+00 -1.76140532e-01
1.81010038e-01 -1.09579138e-01 4.90173966e-01 9.24148500e-01
3.58693033e-01 5.97396612e-01 3.38685751e-01 -9.11830068e-01
-8.72781575e-02 -5.82991064e-01 1.83700562e-01 5.31728387e-01
6.02119505e-01 -5.65144956e-01 -3.30282182e-01 3.62843871e-01]
|
[10.275664329528809, 0.8019721508026123]
|
9d293755-ddd7-490c-9420-bf329b0aa902
|
points-to-patches-enabling-the-use-of-self
|
2204.03957
| null |
https://arxiv.org/abs/2204.03957v1
|
https://arxiv.org/pdf/2204.03957v1.pdf
|
Points to Patches: Enabling the Use of Self-Attention for 3D Shape Recognition
|
While the Transformer architecture has become ubiquitous in the machine learning field, its adaptation to 3D shape recognition is non-trivial. Due to its quadratic computational complexity, the self-attention operator quickly becomes inefficient as the set of input points grows larger. Furthermore, we find that the attention mechanism struggles to find useful connections between individual points on a global scale. In order to alleviate these problems, we propose a two-stage Point Transformer-in-Transformer (Point-TnT) approach which combines local and global attention mechanisms, enabling both individual points and patches of points to attend to each other effectively. Experiments on shape classification show that such an approach provides more useful features for downstream tasks than the baseline Transformer, while also being more computationally efficient. In addition, we also extend our method to feature matching for scene reconstruction, showing that it can be used in conjunction with existing scene reconstruction pipelines.
|
["Mark O'Connor", 'Magnus Oskarsson', 'Axel Berg']
|
2022-04-08
| null | null | null | null |
['3d-feature-matching', '3d-shape-recognition']
|
['computer-vision', 'computer-vision']
|
[-2.05726101e-04 7.40287034e-03 1.12414517e-01 -1.98934644e-01
-8.96121383e-01 -7.67079175e-01 5.16166151e-01 3.16283643e-01
-7.96491429e-02 1.68074071e-01 2.13506535e-01 -3.83608341e-01
-9.03706551e-02 -9.09317553e-01 -8.09014618e-01 -3.41674805e-01
1.02771735e-02 7.19561756e-01 6.26668811e-01 -1.37524664e-01
4.11948472e-01 1.06805110e+00 -1.54064012e+00 2.65217155e-01
4.09331679e-01 1.03921461e+00 3.40203762e-01 5.11339843e-01
-1.91261843e-01 4.21279311e-01 -2.66228110e-01 -4.27704245e-01
4.11508828e-01 4.43776064e-02 -1.09922385e+00 1.27621591e-01
9.44052279e-01 -3.67557526e-01 -2.67013460e-01 6.65073633e-01
4.79961485e-01 4.02385026e-01 4.72083658e-01 -9.77385283e-01
-7.29608655e-01 -1.38875404e-02 -6.91887975e-01 3.74112278e-01
3.80399317e-01 3.70472260e-02 1.34173834e+00 -1.26079154e+00
6.33402765e-01 1.32764947e+00 9.50736582e-01 1.30354062e-01
-1.37785208e+00 -3.11936200e-01 2.44105950e-01 8.07770789e-02
-1.10313904e+00 -6.08758986e-01 8.40378821e-01 -2.26750106e-01
1.61415970e+00 1.29993185e-01 8.08389962e-01 5.76637805e-01
-1.31087676e-01 9.63404000e-01 4.18799669e-01 -3.40115398e-01
5.52618019e-02 -3.93364221e-01 -3.61447223e-02 7.98578143e-01
-6.81906939e-02 -1.45899907e-01 -5.21735311e-01 -1.09055966e-01
1.32313371e+00 1.55015081e-01 -1.19615428e-01 -7.80256808e-01
-1.11512172e+00 6.55975997e-01 9.64905560e-01 4.22235876e-01
-4.43989486e-01 3.75389069e-01 2.03684807e-01 3.40850115e-01
6.06450677e-01 5.92214942e-01 -4.80685800e-01 -1.57139748e-01
-7.16330886e-01 2.55376995e-01 4.64372158e-01 1.01458859e+00
9.00446177e-01 -1.98030725e-01 1.35548748e-02 8.62605453e-01
1.64146990e-01 3.61916237e-02 1.21686541e-01 -9.96808171e-01
3.85842890e-01 8.43043208e-01 -5.07604443e-02 -1.08389795e+00
-3.54105622e-01 -4.95072901e-01 -4.09049511e-01 2.68103391e-01
3.90011013e-01 4.97569531e-01 -9.13080513e-01 1.49690485e+00
3.86968583e-01 3.94696295e-01 -4.09197390e-01 8.42649102e-01
8.22531879e-01 4.15720195e-01 1.11414328e-01 4.06184226e-01
1.13259816e+00 -9.89290476e-01 -5.14214784e-02 -3.01694989e-01
5.22162855e-01 -8.06072831e-01 1.15065491e+00 -2.50647552e-02
-1.46262813e+00 -5.41931450e-01 -6.57343030e-01 -7.37309277e-01
-3.83838326e-01 -2.71076083e-01 8.46719503e-01 1.06112704e-01
-1.34936094e+00 7.17115939e-01 -9.29284513e-01 -6.81767166e-01
8.79652560e-01 5.87295890e-01 -5.58988333e-01 -1.06120162e-01
-3.50111187e-01 9.13120270e-01 -5.87513931e-02 -8.69296193e-02
-4.71364558e-01 -9.16021407e-01 -1.04488945e+00 2.61171311e-01
1.89725347e-02 -9.77133512e-01 1.47724068e+00 -7.31707096e-01
-1.22402704e+00 9.96287227e-01 -4.70065057e-01 -1.39838666e-01
1.94872200e-01 -2.54464626e-01 3.51031750e-01 3.21653485e-01
2.72436470e-01 9.47829545e-01 7.79310763e-01 -1.10547936e+00
-5.69160223e-01 -6.21334612e-01 1.92041084e-01 5.19822717e-01
-1.49632618e-01 -3.28409346e-03 -7.07164884e-01 -6.09972358e-01
5.81229389e-01 -5.06117105e-01 -2.67717838e-01 4.82102215e-01
-8.06050375e-02 -5.28226733e-01 1.27303827e+00 -3.38175297e-01
5.70879340e-01 -2.26500845e+00 3.06397080e-01 2.89748728e-01
2.49645829e-01 8.18012431e-02 -3.04500639e-01 3.97995085e-01
-1.72751591e-01 9.82449949e-02 -2.40177661e-01 -6.76754415e-01
-2.79527634e-01 3.18448842e-01 -2.38047153e-01 3.63236904e-01
5.70110261e-01 1.37784374e+00 -9.65531945e-01 -2.69642323e-01
4.36938554e-01 5.64987838e-01 -8.18153381e-01 -1.10671278e-02
-1.72219779e-02 2.22581625e-01 -4.83799398e-01 8.99295747e-01
3.94670874e-01 -4.93400395e-01 -2.50563443e-01 -1.04890481e-01
6.40949141e-03 5.97837329e-01 -9.52339351e-01 1.91816306e+00
-4.59280819e-01 6.90217376e-01 1.68169588e-01 -1.12211025e+00
7.20836401e-01 1.72595859e-01 6.65805042e-01 -6.17795527e-01
-7.87108764e-02 1.65453047e-01 -2.19409287e-01 -2.59374142e-01
5.28880060e-01 -1.57092705e-01 2.31129110e-01 6.04913592e-01
1.54007316e-01 -3.80651176e-01 -1.49626538e-01 1.29021883e-01
1.30221200e+00 3.92798960e-01 2.87063479e-01 -1.65954933e-01
1.09346911e-01 9.05603021e-02 1.06466301e-01 6.42363489e-01
-1.10943094e-01 8.76241922e-01 1.86944753e-01 -5.91165781e-01
-1.26467955e+00 -1.02747655e+00 -1.32935449e-01 1.17081702e+00
1.70880198e-01 -4.20233965e-01 -3.43825519e-01 -7.74631262e-01
3.51782739e-01 3.92155021e-01 -3.90947461e-01 7.54477307e-02
-7.31327474e-01 -8.95061493e-02 2.22745851e-01 9.58162844e-01
2.33079612e-01 -1.26941025e+00 -8.22108269e-01 3.63787413e-01
1.06519170e-01 -9.23783183e-01 -4.11246836e-01 2.50792533e-01
-1.17759979e+00 -1.12110424e+00 -6.56070292e-01 -9.43404794e-01
8.62736285e-01 7.65880644e-01 1.33147728e+00 4.39532042e-01
-2.55761266e-01 7.01781690e-01 -2.60007024e-01 -2.55016208e-01
1.03330821e-01 2.06597939e-01 -3.95491540e-01 -2.50883937e-01
4.68631148e-01 -8.26882422e-01 -6.29605532e-01 1.97991222e-01
-5.99257052e-01 -9.86298546e-03 5.05706847e-01 8.37514818e-01
5.46528697e-01 -1.97048530e-01 3.53633195e-01 -5.57855964e-01
5.19290864e-01 -2.79781610e-01 -3.44171256e-01 -5.17078340e-02
3.25645349e-04 -4.34223078e-02 3.56897622e-01 -2.15879485e-01
-6.56369746e-01 2.33211204e-01 -4.01109070e-01 -6.38148010e-01
-2.33956724e-01 4.21836942e-01 7.18220845e-02 -4.49762136e-01
4.83901888e-01 1.09080993e-01 1.84880704e-01 -7.06428766e-01
3.36327523e-01 2.11559042e-01 4.63678062e-01 -5.40527642e-01
8.57420206e-01 5.36414802e-01 1.74172241e-02 -8.95981371e-01
-7.15023637e-01 -6.50838196e-01 -8.80000055e-01 -3.51677090e-02
5.78427434e-01 -6.84151530e-01 -7.60817111e-01 1.77021265e-01
-1.25581610e+00 -5.21535218e-01 -5.44879019e-01 3.35825793e-02
-7.46565938e-01 4.09758925e-01 -6.43489420e-01 -5.49922526e-01
-1.87303156e-01 -1.16345930e+00 1.62341022e+00 -5.94420210e-02
-2.68947572e-01 -1.08121371e+00 -1.74422055e-01 1.11276753e-01
3.77556443e-01 -6.59682381e-04 9.32342350e-01 -4.22975570e-01
-9.25710022e-01 -3.20572853e-01 -4.95956510e-01 -2.33897701e-01
-3.20163032e-04 -1.38240322e-01 -9.45772111e-01 -3.10333073e-01
-2.43377090e-01 -3.23696256e-01 8.91345561e-01 3.42541456e-01
1.38442516e+00 -7.67594352e-02 -4.17462021e-01 6.62871420e-01
1.29743874e+00 -1.93313941e-01 6.61504388e-01 3.52750063e-01
7.19635665e-01 4.31797504e-01 2.29173601e-01 2.68129297e-02
4.81833994e-01 8.04653227e-01 5.51932454e-01 -4.03799236e-01
-3.11107218e-01 -4.32567298e-01 -4.15324941e-02 4.93842393e-01
-2.19261736e-01 6.59962147e-02 -1.13127577e+00 6.75181866e-01
-1.89134169e+00 -1.01580691e+00 -5.94145395e-02 2.24041772e+00
3.54972482e-01 7.22545534e-02 4.14342210e-02 1.84792668e-01
3.37635159e-01 -5.66025190e-02 -5.15772223e-01 -3.95194679e-01
-7.51513466e-02 5.84197462e-01 2.54571825e-01 5.16118646e-01
-1.06162477e+00 1.13273859e+00 7.38927555e+00 5.68736970e-01
-9.98730481e-01 7.93754496e-03 5.63141823e-01 5.63061871e-02
-4.52456921e-01 -1.92615464e-02 -4.63175297e-01 -5.67251369e-02
2.67589092e-01 2.13193074e-01 4.51919883e-01 7.53128111e-01
-7.89429098e-02 1.27751641e-02 -1.26646376e+00 1.03206658e+00
-1.44442907e-02 -1.43500590e+00 1.32610530e-01 8.46525133e-02
3.71018171e-01 3.11329097e-01 -5.21791726e-02 1.51576042e-01
1.63065851e-01 -1.06796348e+00 6.18380547e-01 2.40840152e-01
7.36109257e-01 -7.52873540e-01 3.79767388e-01 3.62661391e-01
-1.46227217e+00 -1.19205326e-01 -5.51898420e-01 -2.31560513e-01
7.82364830e-02 2.60086030e-01 -7.94096053e-01 4.36736166e-01
8.24061275e-01 9.38160956e-01 -7.09286630e-01 1.32643723e+00
-4.09689173e-02 1.13088839e-01 -7.00831294e-01 9.12248790e-02
4.00183350e-01 8.04031342e-02 6.12087011e-01 9.89681304e-01
4.35901552e-01 1.97781414e-01 6.10490814e-02 9.54861104e-01
-8.21311846e-02 1.24435849e-01 -1.10708666e+00 1.91030920e-01
3.38779986e-01 1.25049102e+00 -9.51679647e-01 -2.09260598e-01
-4.94181901e-01 1.07177162e+00 9.89793658e-01 4.07649815e-01
-3.30307633e-01 -2.72463411e-01 6.87776208e-01 2.87252098e-01
6.40288234e-01 -4.77015346e-01 -5.82036436e-01 -1.06521142e+00
7.50552416e-02 -5.61296999e-01 3.35666835e-01 -9.81849670e-01
-1.39035594e+00 3.73722047e-01 -1.97187170e-01 -1.08879435e+00
7.36661181e-02 -6.51552498e-01 -7.67328382e-01 7.89716363e-01
-1.43516791e+00 -1.34288740e+00 -2.59824246e-01 6.38581574e-01
7.55624413e-01 1.38462812e-01 8.29273403e-01 2.23425061e-01
-1.62430927e-01 4.98656720e-01 -3.73984277e-01 -3.75947952e-02
2.80492753e-01 -1.22465920e+00 8.81759465e-01 7.40800798e-01
4.94460791e-01 8.22485685e-01 2.22036913e-01 -4.59667623e-01
-1.37702894e+00 -7.94498980e-01 1.03031087e+00 -6.74772501e-01
4.44897801e-01 -4.04052645e-01 -9.46238697e-01 9.02393281e-01
4.25871573e-02 1.52696848e-01 3.93272012e-01 5.11080742e-01
-5.29313982e-01 1.85190365e-01 -1.07063746e+00 7.05371976e-01
1.40905654e+00 -6.41298354e-01 -6.53001547e-01 3.64730328e-01
5.80741465e-01 -5.87877452e-01 -8.36880922e-01 3.85028332e-01
3.82251322e-01 -1.10027885e+00 1.39654815e+00 -5.99379718e-01
4.59491760e-01 -2.31837586e-01 -1.54635608e-01 -1.10720158e+00
-7.38057554e-01 -5.01478434e-01 -1.59544066e-01 9.37024772e-01
3.93033892e-01 -6.70533180e-01 8.57937813e-01 6.75726116e-01
-4.22142953e-01 -9.60573852e-01 -1.13920867e+00 -4.00169700e-01
4.91271652e-02 -6.26379251e-01 5.75810552e-01 8.27361047e-01
-1.95987057e-02 5.29911101e-01 1.30982056e-01 1.29819125e-01
4.90042567e-01 3.68122190e-01 8.19098771e-01 -1.37117064e+00
-3.08357179e-01 -8.82483780e-01 -6.73602998e-01 -1.65086532e+00
-6.99312836e-02 -1.10683441e+00 -6.86166435e-02 -1.77787626e+00
1.50746912e-01 -7.43329048e-01 -1.75443292e-01 8.13413918e-01
-1.36630878e-01 6.35500133e-01 2.59822965e-01 3.21722955e-01
-4.88459647e-01 4.15905774e-01 1.40445721e+00 5.68578392e-02
-1.70835480e-01 -3.58396992e-02 -6.45387709e-01 6.02555335e-01
6.42877340e-01 -1.84682161e-01 -2.04682752e-01 -7.76381791e-01
5.40196756e-03 -2.70834267e-01 6.30219102e-01 -8.35309327e-01
3.52057278e-01 1.58591568e-01 6.48672760e-01 -8.48438799e-01
6.05047643e-01 -8.68917406e-01 -1.66154966e-01 9.16465372e-02
-8.23925585e-02 2.77760446e-01 4.23817694e-01 4.39172953e-01
-1.33038744e-01 -1.23857349e-01 6.90567255e-01 -3.54072332e-01
-5.98388314e-01 5.42434633e-01 -7.75464624e-02 -2.37441257e-01
8.94246340e-01 -6.92038596e-01 -1.90299877e-03 -3.44150335e-01
-6.44279420e-01 1.14578240e-01 7.21843660e-01 3.63614827e-01
7.87671208e-01 -1.46586978e+00 -3.20738822e-01 5.52275836e-01
5.32035604e-02 5.19163847e-01 -8.15296825e-03 8.05831730e-01
-5.86180687e-01 3.54370385e-01 -1.44988507e-01 -9.06370223e-01
-1.18852019e+00 4.72603858e-01 3.39006722e-01 2.38039792e-02
-1.12521410e+00 1.02657652e+00 2.42840081e-01 -4.78437304e-01
2.99363554e-01 -2.95021832e-01 3.47902887e-02 -1.68176621e-01
2.49326646e-01 1.36769310e-01 1.96368411e-01 -6.25151813e-01
-2.94674516e-01 1.02051830e+00 4.20500413e-02 -7.41058588e-03
1.64724934e+00 -1.69843007e-02 -2.67120779e-01 2.33515725e-01
1.11315906e+00 -7.90751874e-02 -1.27664793e+00 -3.59158456e-01
1.45170167e-01 -6.96517229e-01 1.00455796e-02 -3.90489370e-01
-9.80493844e-01 9.70977783e-01 1.08711943e-01 4.47515965e-01
1.04753458e+00 2.83265799e-01 8.45577478e-01 4.54522997e-01
4.15819377e-01 -5.43355286e-01 1.46459594e-01 7.04294622e-01
1.15164196e+00 -1.17185271e+00 -8.26705620e-02 -4.86255676e-01
-1.81323245e-01 1.24431217e+00 5.66321135e-01 -3.81474346e-01
5.81555724e-01 2.14592889e-01 -1.07963458e-01 -5.28096080e-01
-6.95568621e-01 -3.70131850e-01 4.58570570e-01 6.98056400e-01
5.22159934e-01 -4.14513707e-01 3.28496009e-01 -5.52290119e-02
-1.47491887e-01 -2.63225287e-01 -4.96387258e-02 1.02912605e+00
-4.88215119e-01 -1.14624834e+00 -1.82134762e-01 5.07192552e-01
-2.06601188e-01 -7.85545260e-02 -4.40668821e-01 7.70337582e-01
-1.38314307e-01 5.36568642e-01 4.05967206e-01 -4.17107306e-02
6.20095849e-01 6.38122857e-02 6.17469132e-01 -8.03177834e-01
-8.26546550e-01 2.63913810e-01 -1.52250633e-01 -9.05149341e-01
-3.18055630e-01 -6.25431120e-01 -1.16918755e+00 -2.57845938e-01
-2.01431572e-01 -3.24332893e-01 3.51827890e-01 7.41854548e-01
6.88950777e-01 3.26382756e-01 2.89989471e-01 -1.33648920e+00
-2.35659197e-01 -7.00519741e-01 -1.22590668e-01 6.59500659e-01
4.38886166e-01 -8.08349431e-01 -6.53922856e-02 -2.60916442e-01]
|
[7.9796624183654785, -3.50689697265625]
|
b49242a7-6658-4343-8d25-fdd923d10df4
|
searching-for-waveforms-on-spatially-filtered
|
2103.13853
| null |
https://arxiv.org/abs/2103.13853v1
|
https://arxiv.org/pdf/2103.13853v1.pdf
|
Searching for waveforms on spatially-filtered epileptic ECoG
|
Seizures are one of the defining symptoms in patients with epilepsy, and due to their unannounced occurrence, they can pose a severe risk for the individual that suffers it. New research efforts are showing a promising future for the prediction and preemption of imminent seizures, and with those efforts, a vast and diverse set of features have been proposed for seizure prediction algorithms. However, the data-driven discovery of nonsinusoidal waveforms for seizure prediction is lacking in the literature, which is in stark contrast with recent works that show the close connection between the waveform morphology of neural oscillations and the physiology and pathophysiology of the brain, and especially its use in effectively discriminating between normal and abnormal oscillations in electrocorticographic (ECoG) recordings of epileptic patients. Here, we explore a scalable, energy-guided waveform search strategy on spatially-projected continuous multi-day ECoG data sets. Our work shows that data-driven waveform learning methods have the potential to not only contribute features with predictive power for seizure prediction, but also to facilitate the discovery of oscillatory patterns that could contribute to our understanding of the pathophysiology and etiology of seizures.
|
['Austin J. Brockmeier', 'Carlos H. Mendoza-Cardenas']
|
2021-03-25
| null | null | null | null |
['seizure-prediction']
|
['medical']
|
[ 1.58435091e-01 -4.88504797e-01 1.58692271e-01 -1.07871249e-01
-7.00743258e-01 -6.22960210e-01 3.54372412e-01 2.89099216e-01
-1.02538869e-01 5.34219682e-01 4.63207334e-01 -2.31167197e-01
-7.72265673e-01 -3.19660932e-01 1.26858056e-01 -9.22376394e-01
-1.06409025e+00 1.35622069e-01 -9.85439271e-02 -9.46961567e-02
4.42577332e-01 7.46555805e-01 -1.34399652e+00 1.87249407e-01
7.81191170e-01 1.29237366e+00 3.21018249e-01 2.39076331e-01
1.37081012e-01 2.40873456e-01 -5.26511431e-01 3.01664710e-01
-2.12251648e-01 -7.88779736e-01 -2.71626800e-01 -2.11174473e-01
-4.71174449e-01 1.99920624e-01 -3.30004394e-01 6.86079264e-01
9.65962887e-01 -2.71192521e-01 4.97663379e-01 -9.17997539e-01
-1.70422703e-01 2.10817799e-01 -3.30493629e-01 9.21169043e-01
3.28336000e-01 5.80462217e-02 6.63723409e-01 -8.71900439e-01
4.43380654e-01 1.72046795e-01 7.04148173e-01 3.02882701e-01
-1.09188318e+00 -9.14860725e-01 -2.92328626e-01 7.33364999e-01
-1.31785583e+00 -5.59962749e-01 9.87255216e-01 -7.74092734e-01
9.48612154e-01 1.66484609e-01 1.38408947e+00 1.08182669e+00
7.02723026e-01 1.74842596e-01 1.05210745e+00 -1.55291125e-01
3.36666375e-01 -3.42274189e-01 -5.79743460e-02 -6.24648668e-02
2.66560912e-02 3.13116401e-01 -1.09605992e+00 -5.78203976e-01
1.27330914e-01 -1.61065370e-01 -9.42079782e-01 -6.42139465e-02
-1.16572511e+00 7.36459672e-01 -5.86667024e-02 7.54138231e-01
-6.73278391e-01 -2.48562455e-01 3.09258163e-01 1.57961383e-01
5.21908224e-01 7.64801204e-01 -5.22318065e-01 -6.73791587e-01
-1.30137300e+00 -2.04198770e-02 4.88068491e-01 -4.25373353e-02
1.00510336e-01 2.40981817e-01 1.15572698e-01 5.48817337e-01
-4.86202165e-02 2.99732536e-01 1.04358852e+00 -3.34349751e-01
-1.91009492e-01 5.11626422e-01 -1.26884580e-01 -9.48170483e-01
-8.42033327e-01 -5.42484999e-01 -5.95757842e-01 2.48250794e-02
2.47803098e-03 -1.48982152e-01 -2.98218787e-01 1.52435374e+00
-1.02347776e-01 5.78379989e-01 -1.34856418e-01 6.90532506e-01
5.91645241e-01 2.46978328e-01 -3.79711017e-02 -6.13590896e-01
1.22521567e+00 2.24196717e-01 -5.82413793e-01 -2.79367268e-01
4.45989966e-01 -5.41161478e-01 4.89839524e-01 6.91086650e-01
-8.08007717e-01 2.94965208e-01 -9.22500730e-01 7.68372774e-01
-5.57630509e-02 -1.34551123e-01 6.58035278e-01 5.05016565e-01
-9.91596401e-01 6.51051223e-01 -1.26407862e+00 -2.97893316e-01
5.79333901e-01 4.27031279e-01 -4.26760823e-01 3.94051343e-01
-9.35113907e-01 1.26087976e+00 2.01325580e-01 1.01806760e-01
-4.19656008e-01 -9.01645005e-01 -4.56556410e-01 8.32035467e-02
-2.97871351e-01 1.37870833e-01 6.40067339e-01 -2.72710472e-01
-1.04031348e+00 6.28901660e-01 -2.20093802e-01 -4.20408815e-01
-1.23545446e-01 3.25634360e-01 -8.02423894e-01 3.21360439e-01
-2.72020996e-01 1.54140547e-01 6.77752137e-01 -5.18435001e-01
-2.96279371e-01 -6.67414248e-01 -9.18618917e-01 -1.51521023e-02
-4.12517637e-01 3.09845388e-01 5.54290056e-01 -7.63896763e-01
4.38527733e-01 -6.65805221e-01 2.20904723e-02 -3.50329489e-01
1.51645005e-01 -2.00985998e-01 7.91839659e-01 -8.15355003e-01
1.31132305e+00 -2.36133957e+00 7.60020018e-02 2.15280771e-01
2.59216487e-01 1.02102406e-01 9.55516025e-02 6.55775607e-01
-3.82833362e-01 -1.70672964e-02 -4.43812758e-01 4.53924805e-01
-3.44714820e-01 -3.68015379e-01 -6.48251653e-01 7.77846813e-01
2.86875457e-01 9.47022438e-01 -7.30120897e-01 1.17237672e-01
1.52214542e-01 7.17889965e-01 -1.94893643e-01 1.44183472e-01
2.88965195e-01 8.08676362e-01 -3.46058100e-01 4.02594298e-01
3.30441505e-01 -2.84978956e-01 2.01646477e-01 1.70733690e-01
-2.81438231e-01 5.39347827e-01 -6.02682471e-01 1.34350550e+00
2.09287032e-02 1.09319997e+00 -2.43403643e-01 -1.20510483e+00
9.14023876e-01 6.81096435e-01 9.41093862e-01 -8.39298904e-01
2.06931666e-01 5.70895791e-01 6.90003633e-01 -6.85605288e-01
-2.98356473e-01 -1.99425109e-02 2.64819622e-01 7.60466337e-01
4.74831177e-04 -1.33369237e-01 2.04274356e-02 -3.00501704e-01
1.30894685e+00 -1.61495041e-02 4.59053040e-01 -4.44702953e-01
2.62382388e-01 -2.04682238e-02 5.40359378e-01 2.41739422e-01
-1.37639776e-01 4.83524919e-01 3.94151449e-01 -3.81220013e-01
-5.46303153e-01 -8.26799393e-01 -8.35719287e-01 4.74642873e-01
-9.80461296e-03 -1.58207178e-01 -3.45029444e-01 1.67687878e-01
-1.64539874e-01 5.59426188e-01 -3.36323380e-01 -4.06465441e-01
-4.81677532e-01 -1.35413003e+00 3.05652380e-01 3.77153158e-01
-7.43976086e-02 -1.33539438e+00 -1.12743282e+00 5.70747077e-01
-2.27522314e-01 -9.42296803e-01 -1.35981664e-01 6.12421513e-01
-8.58633637e-01 -1.37919724e+00 -5.38278401e-01 -7.75131702e-01
3.72217387e-01 -2.45412737e-01 6.61913574e-01 2.83915680e-02
-7.37410665e-01 4.08055604e-01 -3.05700600e-01 -4.47010547e-01
2.71556955e-02 -2.94814587e-01 1.94586992e-01 -1.16117090e-01
7.37527788e-01 -1.26091993e+00 -9.17128325e-01 1.09681726e-01
-6.03173971e-01 -3.28285694e-01 5.64195395e-01 5.78873754e-01
5.11090577e-01 3.21882591e-02 1.22104013e+00 1.95445842e-03
8.99774611e-01 -8.98414314e-01 -4.87028599e-01 8.86366665e-02
-8.41102242e-01 -2.43019447e-01 4.37345177e-01 -5.24020672e-01
-5.05868018e-01 -1.97120890e-01 7.82573503e-03 4.20541689e-03
-2.42681757e-01 6.76325142e-01 2.50867367e-01 -3.51494789e-01
5.38264155e-01 7.51092851e-01 -7.04282299e-02 -3.06733459e-01
-3.25668752e-01 5.90651751e-01 4.62470025e-01 5.70579544e-02
3.61614078e-01 3.37874144e-01 6.19975477e-02 -9.62946773e-01
-7.78140500e-02 -4.92349684e-01 -3.23684096e-01 -9.15782377e-02
8.20201635e-01 -6.48692906e-01 -5.16575336e-01 5.49190700e-01
-8.80923331e-01 -2.05714300e-01 9.64244753e-02 9.26045060e-01
-5.33976614e-01 1.43048450e-01 -2.41675898e-01 -7.45095015e-01
-5.79480112e-01 -1.06440580e+00 5.62345982e-01 9.07276273e-02
-5.97043455e-01 -9.16276753e-01 5.04193783e-01 -2.28130519e-01
6.62745953e-01 5.14815032e-01 1.27979302e+00 -1.13901365e+00
-3.02919775e-01 -2.74633348e-01 2.19304547e-01 -1.02944963e-01
2.90851772e-01 -3.99229884e-01 -8.08386445e-01 -2.28249371e-01
5.36826491e-01 -1.47068694e-01 4.18267787e-01 7.19156325e-01
8.73917937e-01 1.73780113e-01 -4.96285588e-01 7.01911271e-01
1.20632768e+00 7.57126331e-01 3.95640731e-01 1.68441728e-01
-2.42815197e-01 4.98491943e-01 1.79872215e-02 5.51425338e-01
1.44046381e-01 5.92797220e-01 1.99452028e-01 3.98753941e-01
-4.26123254e-02 2.67071366e-01 2.02332616e-01 7.43829846e-01
4.06248644e-02 4.12004776e-02 -1.23370540e+00 6.78272903e-01
-1.31558347e+00 -1.06645656e+00 -9.79875177e-02 2.17090034e+00
7.13380754e-01 -2.18677178e-01 -2.31915750e-02 3.23568970e-01
4.78713185e-01 3.76919843e-02 -7.05096960e-01 -1.16565593e-01
-2.72734612e-01 6.71492517e-01 -1.12842001e-01 -1.71333939e-01
-6.36184156e-01 2.10021645e-01 7.15109444e+00 2.99172044e-01
-1.73116326e+00 8.24064612e-02 2.40186736e-01 -3.20256650e-01
-1.61615282e-01 -2.35999092e-01 -3.13229591e-01 7.07736731e-01
1.22397137e+00 -8.01534414e-01 7.75692463e-01 3.63880724e-01
5.86327732e-01 -1.79535791e-01 -9.22773540e-01 1.20765972e+00
1.25172153e-01 -1.34113431e+00 -6.58384025e-01 2.81583041e-01
5.17477691e-01 5.22430837e-01 -4.72449921e-02 -3.87695521e-01
-5.49058974e-01 -1.14982820e+00 4.35496360e-01 6.04083061e-01
7.84217894e-01 -7.44899571e-01 5.87475240e-01 4.01275218e-01
-1.19034648e+00 -2.71633863e-01 4.07615416e-02 -4.65313978e-02
2.49545246e-01 6.74773633e-01 -7.25030422e-01 1.76085383e-01
6.93470776e-01 7.63861060e-01 -3.74028772e-01 1.67024136e+00
-1.79835558e-01 7.86550999e-01 -2.38177344e-01 1.77679006e-02
-2.12180823e-01 -1.55366272e-01 7.79583395e-01 7.73381412e-01
1.01204455e+00 3.80044788e-01 -3.11294407e-01 8.53887081e-01
3.68714601e-01 1.24826513e-01 -5.30696750e-01 -4.98621523e-01
5.91911793e-01 1.17310071e+00 -9.57708418e-01 3.88241649e-01
-3.64116251e-01 3.72493297e-01 4.42775451e-02 2.17371464e-01
-2.48912722e-01 -4.30482835e-01 5.36902308e-01 1.75255403e-01
1.02717608e-01 -3.10150206e-01 -4.82810706e-01 -1.03766632e+00
2.30476722e-01 -6.87621176e-01 1.76341996e-01 -5.35893619e-01
-1.02535832e+00 8.64058733e-01 -1.52595311e-01 -1.26789379e+00
-5.95913887e-01 -4.31474924e-01 -1.17509282e+00 9.07293856e-01
-1.50412524e+00 -5.12985170e-01 -8.29178020e-02 5.20485878e-01
2.05826074e-01 -3.00776660e-01 1.30337226e+00 5.46520352e-02
-1.91641852e-01 -1.24921434e-01 1.56152993e-01 -2.39008978e-01
3.77542853e-01 -8.96746635e-01 1.12331405e-01 6.03141725e-01
3.29261363e-01 2.31377259e-01 6.99892998e-01 -6.00053012e-01
-1.24584234e+00 -6.57266259e-01 9.70936775e-01 -3.27945709e-01
9.96471584e-01 -2.42794469e-01 -1.05957007e+00 2.09132403e-01
6.09005839e-02 -4.85372506e-02 1.15330291e+00 -9.78111774e-02
3.02811414e-01 8.88149142e-02 -8.71658981e-01 2.91746020e-01
6.81886733e-01 -4.61565077e-01 -9.76869047e-01 3.63469660e-01
-1.84215024e-01 1.33819729e-02 -9.82205570e-01 5.21968007e-01
6.84048533e-01 -9.45414305e-01 6.29694819e-01 -2.30258033e-01
1.67348366e-02 1.57087922e-01 3.39378536e-01 -1.69943368e+00
-2.16954023e-01 -7.93748081e-01 -1.24461345e-01 5.96992850e-01
3.85470152e-01 -9.08621252e-01 6.78791940e-01 3.97954136e-01
-2.26640105e-01 -1.22147357e+00 -1.16225040e+00 -7.21517920e-01
2.31755242e-01 -6.31152809e-01 5.65228343e-01 4.92514074e-01
8.34439576e-01 -4.16895568e-01 1.68910027e-02 1.06003024e-01
3.70182484e-01 1.91400349e-01 -3.10365796e-01 -1.50355089e+00
-1.48372725e-02 -8.04558814e-01 -7.53559351e-01 -3.78421023e-02
1.76714078e-01 -1.21050000e+00 -4.94969636e-02 -1.48564839e+00
9.27111730e-02 -2.50966579e-01 -4.99841541e-01 3.58037114e-01
1.08828202e-01 3.87898475e-01 -3.38729918e-01 3.23403955e-01
1.71754837e-01 4.79591310e-01 6.63527489e-01 2.30275705e-01
-5.77877104e-01 1.27257332e-01 -6.78305268e-01 7.64655411e-01
8.54204655e-01 -6.52169585e-01 -4.99658465e-01 2.50138473e-02
2.79137135e-01 2.61587054e-01 2.40922496e-01 -1.06867695e+00
4.25745666e-01 -1.14680305e-02 5.47912121e-01 -5.24970949e-01
2.91855842e-01 -5.30552566e-01 2.29959890e-01 4.76833880e-01
6.29035756e-02 6.83072135e-02 3.34361464e-01 3.93019825e-01
-2.55219847e-01 -1.19783431e-02 5.89026332e-01 2.45638654e-01
-5.25195301e-01 3.67291480e-01 -8.41107368e-01 2.43939057e-01
1.14318740e+00 -2.94460118e-01 -1.22460507e-01 -3.16576064e-01
-8.39414120e-01 -1.74926534e-01 3.36277708e-02 4.59300071e-01
7.90795505e-01 -1.13421440e+00 -7.15004146e-01 4.63547289e-01
7.68329427e-02 -7.74937987e-01 1.81787625e-01 1.42705214e+00
-9.00312290e-02 6.53238535e-01 -4.71224695e-01 -6.22721255e-01
-9.20601547e-01 -4.81599430e-03 4.43289548e-01 1.47177577e-01
-1.02307105e+00 6.67517126e-01 -1.27486736e-01 3.42299253e-01
4.34153192e-02 6.07493073e-02 -5.08388758e-01 2.66732126e-01
8.33027661e-01 2.48115793e-01 5.42423546e-01 -5.32030702e-01
-7.01539159e-01 3.40532511e-01 3.42940003e-01 -2.65215207e-02
1.96774828e+00 2.25321278e-01 -3.91625255e-01 3.72554690e-01
7.45410323e-01 -4.19581421e-02 -9.39199269e-01 1.85127094e-01
4.59084094e-01 -2.20664397e-01 2.23481148e-01 -9.58037257e-01
-1.05449629e+00 8.47166538e-01 8.52748811e-01 3.17302406e-01
1.39458799e+00 1.72135636e-01 6.72946751e-01 8.49831849e-02
6.91483498e-01 -7.30654955e-01 -1.07130870e-01 9.44352746e-02
9.47836280e-01 -7.61575997e-01 -1.32046759e-01 1.77927405e-01
-3.56513828e-01 1.35843694e+00 -7.47917667e-02 -1.65446967e-01
1.04011297e+00 5.80259681e-01 -3.51495370e-02 -4.17933077e-01
-9.07324612e-01 2.18446720e-02 5.00650465e-01 7.68841505e-01
5.27766645e-01 -1.44057572e-01 -7.31655955e-01 9.88143802e-01
-3.96270812e-01 1.22112662e-01 4.15722877e-01 7.90380716e-01
-6.17702603e-01 -1.26288760e+00 -2.34724134e-01 9.81399715e-01
-5.28740227e-01 -2.90057123e-01 -2.30156600e-01 4.12497699e-01
4.73017991e-02 9.54025984e-01 8.67142454e-02 -6.31368011e-02
-4.83439042e-04 6.42529488e-01 4.39813524e-01 -6.11552536e-01
-4.07866001e-01 1.74638510e-01 -1.67936146e-01 -4.39405918e-01
-1.63151249e-01 -1.06782389e+00 -1.32423794e+00 4.16855454e-01
-3.51372629e-01 2.73500085e-01 7.92060673e-01 1.18568003e+00
7.51386642e-01 4.21351045e-01 6.31595254e-01 -7.07313597e-01
-2.16352955e-01 -9.95241284e-01 -9.21065629e-01 -4.99487184e-02
3.25677693e-01 -7.04497337e-01 -7.21945822e-01 -1.07307136e-01]
|
[13.1749849319458, 3.549422025680542]
|
8f83af14-4689-4005-8a50-cfa1e30c91c0
|
protein-dna-binding-sites-prediction-based-on
|
2306.15912
| null |
https://arxiv.org/abs/2306.15912v1
|
https://arxiv.org/pdf/2306.15912v1.pdf
|
Protein-DNA binding sites prediction based on pre-trained protein language model and contrastive learning
|
Protein-DNA interaction is critical for life activities such as replication, transcription, and splicing. Identifying protein-DNA binding residues is essential for modeling their interaction and downstream studies. However, developing accurate and efficient computational methods for this task remains challenging. Improvements in this area have the potential to drive novel applications in biotechnology and drug design. In this study, we propose a novel approach called CLAPE, which combines a pre-trained protein language model and the contrastive learning method to predict DNA binding residues. We trained the CLAPE-DB model on the protein-DNA binding sites dataset and evaluated the model performance and generalization ability through various experiments. The results showed that the AUC values of the CLAPE-DB model on the two benchmark datasets reached 0.871 and 0.881, respectively, indicating superior performance compared to other existing models. CLAPE-DB showed better generalization ability and was specific to DNA-binding sites. In addition, we trained CLAPE on different protein-ligand binding sites datasets, demonstrating that CLAPE is a general framework for binding sites prediction. To facilitate the scientific community, the benchmark datasets and codes are freely available at https://github.com/YAndrewL/clape.
|
['Boxue Tian', 'Yufan Liu']
|
2023-06-28
| null | null | null | null |
['contrastive-learning', 'protein-language-model', 'contrastive-learning']
|
['computer-vision', 'medical', 'methodology']
|
[ 1.21064499e-01 -4.42750812e-01 -4.24881905e-01 -3.63859236e-01
-8.63000393e-01 -6.03715777e-01 2.04682127e-01 2.25513652e-01
-4.17703331e-01 1.12440503e+00 -4.75660898e-02 -4.77226406e-01
6.54997379e-02 -5.90904951e-01 -6.90724790e-01 -1.24393666e+00
1.45738095e-01 3.59454900e-01 4.41100359e-01 -1.46915570e-01
3.41188908e-01 5.67212462e-01 -1.30559158e+00 5.49652457e-01
1.03286290e+00 3.79350305e-01 5.55979609e-01 4.25395936e-01
-1.72876015e-01 1.58788368e-01 -1.00543007e-01 -2.51117200e-01
-1.91426575e-01 -3.63810658e-01 -4.66939628e-01 -7.40102232e-01
-4.47919577e-01 2.26521224e-01 6.45386577e-02 6.08551383e-01
6.78357124e-01 -3.62815261e-02 8.03372145e-01 -7.88884699e-01
-4.91251588e-01 1.28886417e-01 -5.20558059e-01 6.90572336e-02
4.53490049e-01 2.51843303e-01 1.09338725e+00 -1.30939615e+00
5.76870561e-01 1.09848106e+00 5.01133621e-01 6.57552183e-01
-1.23611856e+00 -8.06402862e-01 -2.71436095e-01 3.37535143e-01
-1.51364291e+00 4.65832651e-03 2.07535356e-01 -5.53667903e-01
1.25015676e+00 2.58071423e-01 3.09773564e-01 8.70162547e-01
5.04821360e-01 6.50830507e-01 8.59940767e-01 -3.19117904e-01
1.96276195e-02 -3.14344972e-01 3.12838197e-01 5.80676258e-01
1.92217201e-01 -4.42367792e-02 -5.38004935e-01 -5.32186866e-01
3.27557921e-01 2.26973206e-01 -1.51872173e-01 -2.78074801e-01
-9.24882948e-01 6.26869261e-01 3.25846791e-01 4.93652731e-01
-3.30386460e-01 -3.56801033e-01 3.89955759e-01 -2.01398209e-01
1.05162166e-01 3.64564247e-02 -5.27889192e-01 -3.83835025e-02
-1.63237721e-01 2.23554358e-01 6.35086417e-01 4.32122976e-01
5.85714579e-01 -6.97375953e-01 -8.81789550e-02 1.18732703e+00
4.09139305e-01 3.25915277e-01 3.92120123e-01 -1.57639086e-01
7.73606002e-02 6.40439868e-01 2.50714511e-01 -8.37856770e-01
-6.58588469e-01 -1.14652246e-01 -7.06400037e-01 -3.79076600e-01
5.34077227e-01 2.06377089e-01 -7.52215624e-01 1.74198794e+00
3.85248095e-01 1.59234539e-01 3.21939886e-01 1.00541270e+00
8.92325878e-01 1.14706600e+00 6.18829906e-01 -2.27993131e-01
1.46677411e+00 -7.13524282e-01 -5.88101268e-01 6.85344562e-02
9.13708687e-01 -7.18587220e-01 1.12005460e+00 1.68210402e-01
-6.45877182e-01 -3.13250780e-01 -9.54265296e-01 -1.23928726e-01
-2.91663229e-01 3.05099756e-01 5.50937831e-01 3.84140790e-01
-4.62789387e-01 2.60078639e-01 -8.86550963e-01 -4.81695741e-01
1.65710509e-01 5.15084505e-01 -5.01386940e-01 -3.77805293e-01
-1.48288524e+00 8.46197724e-01 6.91447437e-01 1.23425111e-01
-6.40058041e-01 -5.47289073e-01 -4.06413376e-01 2.28565037e-01
9.12922248e-03 -2.16166466e-01 1.08354652e+00 -2.07447350e-01
-1.28758764e+00 8.84494126e-01 -4.46168095e-01 3.24705429e-03
-2.38386720e-01 -1.84560403e-01 -2.47403041e-01 -2.87091643e-01
-8.79021827e-03 3.44495714e-01 -2.31677771e-01 -8.24423313e-01
-5.14781296e-01 -4.39299375e-01 -3.35861683e-01 1.90674663e-01
-1.06156617e-01 1.59112588e-01 -6.23674870e-01 -3.30365032e-01
4.63533290e-02 -1.08597898e+00 -2.85522938e-01 -1.65221319e-01
5.65962344e-02 -5.60079455e-01 4.37148958e-01 -7.15493679e-01
1.21863675e+00 -2.18000364e+00 2.13706240e-01 5.08318007e-01
-1.32494017e-01 8.42102766e-01 -2.05412045e-01 1.01625562e+00
-2.87846476e-01 -4.24329154e-02 -1.41665429e-01 6.96465790e-01
-2.10741222e-01 -6.12691045e-03 5.95225208e-02 4.46570665e-01
6.89710677e-02 9.54468727e-01 -8.04742515e-01 -7.24267736e-02
-1.14014102e-02 6.37013972e-01 -3.77367735e-01 2.98837811e-01
-4.59997654e-01 5.36874294e-01 -7.21184611e-01 7.10619330e-01
8.14557672e-01 -5.09461284e-01 8.03763032e-01 -1.87677309e-01
4.51099267e-03 1.43854290e-01 -7.08032072e-01 1.43053675e+00
1.27691831e-02 2.20663562e-01 -3.40808421e-01 -8.07116032e-01
1.29920173e+00 3.34565431e-01 4.91647124e-01 -6.84245706e-01
-2.15583434e-03 4.67046529e-01 2.95703948e-01 -5.62662959e-01
-2.84535050e-01 -3.02619487e-01 1.59118682e-01 1.65817961e-01
-3.74291331e-01 4.79184419e-01 2.12106138e-01 6.23628870e-02
9.21895325e-01 1.68173403e-01 5.39130509e-01 -3.18601131e-01
9.51041698e-01 -1.21392366e-02 8.13639700e-01 2.81096160e-01
-1.17701948e-01 1.69894248e-01 4.70576406e-01 -3.75553846e-01
-8.39213192e-01 -7.68676341e-01 -3.67759377e-01 1.24789548e+00
6.28547370e-02 -6.48150086e-01 -7.95793951e-01 -4.50873375e-01
2.98800189e-02 4.87744749e-01 -3.59436795e-02 -3.10221046e-01
-3.67440283e-01 -1.29038453e+00 6.34993851e-01 4.23812181e-01
2.61982948e-01 -8.75463128e-01 1.09889179e-01 2.91835070e-01
-4.36413437e-01 -5.27672231e-01 -5.68226218e-01 2.53616899e-01
-4.86939907e-01 -1.25576103e+00 -7.08863258e-01 -1.05435586e+00
3.63946378e-01 2.72539854e-01 5.89397907e-01 1.94902077e-01
-2.85046428e-01 -3.57923031e-01 -1.92477137e-01 -4.77062464e-01
-3.03104252e-01 5.93024418e-02 -2.57882178e-02 -2.90127546e-01
1.09465337e+00 -4.56527412e-01 -7.16726601e-01 6.61494970e-01
-9.11525309e-01 7.05303326e-02 7.91125834e-01 1.07260156e+00
6.98935986e-01 -4.75835294e-01 8.93439114e-01 -8.93703222e-01
5.68079650e-01 -6.34137273e-01 -6.65455818e-01 3.47948432e-01
-5.91362417e-01 7.26651549e-02 3.20638090e-01 -1.38694197e-01
-1.17723882e+00 4.85216171e-01 -8.11049879e-01 4.28975344e-01
-3.20742130e-02 9.08803523e-01 -6.35166943e-01 1.11427628e-01
6.88362002e-01 3.86032730e-01 5.24980649e-02 -5.23144364e-01
-1.31200209e-01 9.87359345e-01 2.37372503e-01 -6.96695924e-01
1.02473721e-01 -1.80081334e-02 1.48986056e-01 -7.55405545e-01
-7.42587388e-01 -8.05112422e-01 -4.24355805e-01 1.66129749e-02
7.06864834e-01 -8.80845428e-01 -1.10430157e+00 6.18839085e-01
-1.08995461e+00 -3.60349789e-02 8.94790530e-01 5.81308246e-01
-2.45847255e-01 7.36269712e-01 -8.35803032e-01 -5.33282936e-01
-4.39879954e-01 -1.19226742e+00 8.35827827e-01 2.57993639e-01
-1.67569533e-01 -7.15689957e-01 5.29347301e-01 5.08388519e-01
6.71892762e-02 -1.89607311e-02 1.25110972e+00 -9.32715416e-01
-4.75305408e-01 -1.92130148e-01 -2.54851401e-01 2.19600704e-02
1.79037273e-01 4.14356887e-02 -6.90738738e-01 -2.40139663e-01
-5.07439911e-01 -3.88258964e-01 7.91052878e-01 2.77583957e-01
1.02718723e+00 6.54535592e-02 -7.41931617e-01 2.49101996e-01
1.30881071e+00 6.89820111e-01 9.97449279e-01 3.48156184e-01
3.79283220e-01 4.17904854e-01 1.08077598e+00 2.04355925e-01
-8.69591981e-02 9.50954318e-01 1.84751451e-01 -2.06781417e-01
4.47441757e-01 -2.38400266e-01 2.50487328e-01 2.49675870e-01
-1.34888768e-01 -5.83320558e-01 -1.43474245e+00 2.08556786e-01
-2.10274243e+00 -8.27942014e-01 -4.86352086e-01 1.92497134e+00
1.17842507e+00 -7.83944800e-02 1.20611884e-01 -7.92572126e-02
7.59564519e-01 -4.26302880e-01 -6.87397957e-01 -3.93323451e-01
-2.55756974e-01 1.99060306e-01 1.74204901e-01 4.39821482e-01
-8.58618319e-01 9.60880637e-01 6.29299736e+00 1.01075959e+00
-1.01595259e+00 -1.82711840e-01 5.60237885e-01 2.72580922e-01
-7.89563060e-02 -8.73493031e-02 -1.04434526e+00 5.63006043e-01
1.02187181e+00 -6.86630905e-02 -1.18515499e-01 6.28792942e-01
3.99195969e-01 1.00419074e-01 -8.57401788e-01 6.41421795e-01
-3.30875784e-01 -1.46522224e+00 -1.40662566e-01 9.38009396e-02
4.28042024e-01 4.55339178e-02 -2.50395477e-01 3.35315429e-02
1.27998948e-01 -1.12043142e+00 -2.79754311e-01 5.53224266e-01
5.29212415e-01 -7.33460069e-01 9.81027842e-01 5.95554948e-01
-9.30524766e-01 1.95415601e-01 -4.70317274e-01 1.00709654e-01
-1.07422046e-01 7.07140326e-01 -1.12202239e+00 5.00681341e-01
5.62782526e-01 6.41462088e-01 2.43654847e-02 9.56008971e-01
-1.71655521e-01 6.87565804e-01 -1.69988140e-01 -2.91450322e-01
2.52699610e-02 -4.49753165e-01 -4.00341600e-02 1.55749094e+00
1.30739421e-01 5.02869070e-01 1.65786833e-01 4.66530234e-01
2.56836554e-03 6.07921541e-01 -1.94008425e-01 -2.59694278e-01
3.62606913e-01 1.09849238e+00 -4.55437452e-01 2.06361748e-02
-6.53427124e-01 5.43110013e-01 4.85184729e-01 3.16210955e-01
-1.02922070e+00 -4.10679489e-01 9.44755018e-01 3.86044271e-02
1.33405596e-01 -9.70699191e-02 1.64162204e-01 -8.23574305e-01
-1.90410197e-01 -1.00194931e+00 7.17507303e-01 -5.91381133e-01
-1.19530761e+00 -1.15703698e-02 -2.92965204e-01 -8.37558627e-01
1.54577434e-01 -8.54511499e-01 -1.49354801e-01 1.14407992e+00
-1.31277287e+00 -9.76198018e-01 -5.21312766e-02 9.54324454e-02
3.02183181e-01 3.37313823e-02 9.87404168e-01 3.34362537e-01
-9.19619620e-01 4.18347090e-01 7.03682303e-01 -1.11931428e-01
9.49933112e-01 -6.32151961e-01 1.52510703e-01 4.60231215e-01
-2.39536688e-01 9.80926454e-01 5.79854250e-01 -6.56717241e-01
-1.29640532e+00 -9.60238516e-01 1.12190783e+00 -1.80781811e-01
3.05511475e-01 -3.54993492e-01 -1.36337125e+00 5.42550087e-01
-8.76715332e-02 -4.39628571e-01 1.36680222e+00 -1.53174652e-02
-2.60945559e-01 6.31378591e-02 -9.06443417e-01 4.97302145e-01
6.95590138e-01 -3.16230386e-01 -2.96034813e-01 4.93440628e-01
2.10322052e-01 -4.33001935e-01 -1.07492518e+00 6.30574226e-01
8.88420641e-01 -7.96856105e-01 1.10920525e+00 -8.47030401e-01
2.58558899e-01 -6.11978769e-01 -1.44352928e-01 -7.97570765e-01
-7.30467856e-01 2.07383651e-02 1.69059291e-01 1.04820645e+00
7.01224923e-01 -5.75584769e-01 7.11092889e-01 5.63787460e-01
-8.60284790e-02 -9.85648394e-01 -6.65608764e-01 -6.21784151e-01
1.17021233e-01 1.36328369e-01 4.11160767e-01 7.13173926e-01
4.12669539e-01 6.18544340e-01 -3.06110471e-01 1.10246465e-01
1.05478384e-01 1.41504064e-01 7.69114912e-01 -1.02079797e+00
-2.52102017e-01 -1.82836816e-01 -2.95031756e-01 -1.17935264e+00
2.78891027e-01 -1.24522412e+00 5.98360635e-02 -1.54970276e+00
8.42628956e-01 -3.31283331e-01 -6.46811962e-01 7.73599386e-01
-5.32854438e-01 9.86707732e-02 -3.36479515e-01 3.84354711e-01
-2.45086536e-01 3.72127175e-01 9.39198315e-01 -5.65983094e-02
-1.84974074e-01 4.89338636e-02 -6.27116859e-01 2.47032046e-01
1.14067245e+00 -4.94403750e-01 -2.78831393e-01 1.05085149e-01
6.19873554e-02 -1.07577085e-01 5.03061293e-03 -4.76064652e-01
-4.02459875e-03 -6.75544679e-01 4.20549810e-01 -5.39841712e-01
2.53086537e-01 -4.90035892e-01 7.41885900e-01 9.09617603e-01
-3.24522913e-01 -2.89531231e-01 3.78215015e-01 5.16877174e-01
-1.40678762e-02 -1.87238231e-01 7.84796536e-01 1.88398674e-01
-6.34895265e-01 8.85075256e-02 -6.47281289e-01 -5.25337458e-01
1.25136530e+00 -2.54163034e-02 -3.36893380e-01 4.86781448e-02
-7.41773129e-01 1.51296422e-01 4.65777099e-01 2.43236706e-01
4.60660100e-01 -1.02272177e+00 -6.76479995e-01 2.24174261e-01
5.51726580e-01 -3.89792264e-01 2.86187381e-01 6.67594373e-01
-7.81077921e-01 1.00571227e+00 -1.72659725e-01 -7.70320058e-01
-1.75910592e+00 6.02659881e-01 3.89669478e-01 -1.67535886e-01
-7.44024590e-02 6.11934483e-01 5.32072961e-01 -5.38441002e-01
3.65204737e-02 8.67438689e-02 -3.18405628e-01 -5.42369783e-01
5.20931065e-01 1.36297747e-01 2.23535284e-01 -5.04624724e-01
-7.31267691e-01 5.07763147e-01 -3.07166159e-01 5.32853365e-01
1.33199704e+00 1.84498638e-01 -3.72808188e-01 1.09312251e-01
1.04116321e+00 -1.30769968e-01 -8.39007676e-01 -2.78062135e-01
4.48184222e-01 -3.55828464e-01 -3.74146760e-01 -9.38908219e-01
-5.37980676e-01 6.06596947e-01 6.85527563e-01 -5.23015141e-01
1.04696059e+00 1.74148798e-01 6.08455122e-01 6.06431186e-01
2.78283656e-01 -7.03977406e-01 -8.37339088e-02 4.76455510e-01
6.54932380e-01 -9.94436383e-01 -7.24567175e-02 -7.06381440e-01
-5.62965214e-01 9.13808346e-01 7.53388584e-01 4.65471029e-01
4.81370747e-01 2.01422781e-01 2.10933208e-01 -6.27354458e-02
-1.00591302e+00 2.02721834e-01 2.74539143e-01 3.92513961e-01
1.19047356e+00 -8.88528600e-02 -1.24129331e+00 8.15616190e-01
4.16008413e-01 3.26355308e-01 -2.45017316e-02 1.03005624e+00
-6.25867486e-01 -1.88751411e+00 -2.35915616e-01 1.90463364e-01
-5.17440915e-01 1.19070709e-02 -6.63085341e-01 4.54192370e-01
-2.69157469e-01 8.35356712e-01 -5.52096725e-01 -2.47860178e-01
1.66847289e-01 4.47652608e-01 1.31806895e-01 -5.42127609e-01
-2.24640012e-01 3.69443357e-01 3.48676592e-02 -3.18952799e-01
-3.74649405e-01 -5.72725117e-01 -1.87716055e+00 -2.73039609e-01
-5.39214492e-01 6.23695314e-01 5.02604008e-01 6.93537354e-01
7.98683524e-01 1.27530009e-01 4.88086253e-01 -2.43170574e-01
-2.31486708e-01 -8.11155617e-01 -4.89941090e-01 2.64528841e-01
-2.40159824e-01 -5.96860647e-01 2.31489643e-01 1.09756999e-02]
|
[4.759156227111816, 5.588006973266602]
|
ec96ecf9-3740-4013-8de8-c590ff1191c4
|
deep-predictive-motion-tracking-in-magnetic
|
1909.11625
| null |
https://arxiv.org/abs/1909.11625v3
|
https://arxiv.org/pdf/1909.11625v3.pdf
|
Deep Predictive Motion Tracking in Magnetic Resonance Imaging: Application to Fetal Imaging
|
Fetal magnetic resonance imaging (MRI) is challenged by uncontrollable, large, and irregular fetal movements. It is, therefore, performed through visual monitoring of fetal motion and repeated acquisitions to ensure diagnostic-quality images are acquired. Nevertheless, visual monitoring of fetal motion based on displayed slices, and navigation at the level of stacks-of-slices is inefficient. The current process is highly operator-dependent, increases scanner usage and cost, and significantly increases the length of fetal MRI scans which makes them hard to tolerate for pregnant women. To help build automatic MRI motion tracking and navigation systems to overcome the limitations of the current process and improve fetal imaging, we have developed a new real time image-based motion tracking method based on deep learning that learns to predict fetal motion directly from acquired images. Our method is based on a recurrent neural network, composed of spatial and temporal encoder-decoders, that infers motion parameters from anatomical features extracted from sequences of acquired slices. We compared our trained network on held out test sets (including data with different characteristics, e.g. different fetuses scanned at different ages, and motion trajectories recorded from volunteer subjects) with networks designed for estimation as well as methods adopted to make predictions. The results show that our method outperformed alternative techniques, and achieved real-time performance with average errors of 3.5 and 8 degrees for the estimation and prediction tasks, respectively. Our real-time deep predictive motion tracking technique can be used to assess fetal movements, to guide slice acquisitions, and to build navigation systems for fetal MRI.
|
['Ali Gholipour', 'Seyed Sadegh Mohseni Salehi', 'Ayush Singh']
|
2019-09-25
| null | null | null | null |
['3d-object-reconstruction']
|
['computer-vision']
|
[ 2.77164996e-01 1.28894866e-01 -2.47459654e-02 -5.16159296e-01
-5.32453239e-01 -7.01458335e-01 2.80117512e-01 -1.83300048e-01
-2.77588457e-01 1.41512468e-01 6.36915043e-02 -4.15032893e-01
-1.85654312e-01 -4.15895432e-01 -7.23724544e-01 -6.61682725e-01
-6.32754385e-01 5.44865727e-01 4.40338343e-01 3.45886976e-01
1.25439018e-01 7.13465869e-01 -8.30846071e-01 3.15787941e-01
4.69686329e-01 7.57246673e-01 4.92703706e-01 9.09910560e-01
1.16240546e-01 1.05900514e+00 -2.37169936e-01 -6.16541691e-02
1.44154327e-02 -6.01475716e-01 -7.51723588e-01 -1.56574458e-01
3.57631117e-01 -6.33253276e-01 -5.12186348e-01 5.71532071e-01
8.63188803e-01 -4.35171239e-02 5.59596837e-01 -6.14427567e-01
-4.05737758e-01 5.56423187e-01 -5.37141383e-01 6.98302746e-01
1.87812209e-01 1.24951638e-01 -1.78376421e-01 -7.64097750e-01
9.99961257e-01 4.79996353e-01 9.24655616e-01 1.01053369e+00
-1.01720083e+00 -7.13448286e-01 -3.56928676e-01 1.47982433e-01
-9.13498342e-01 -7.20310152e-01 5.90292752e-01 -9.38117623e-01
7.50431180e-01 1.92373604e-01 9.26347613e-01 1.11221159e+00
7.70067990e-01 3.40427071e-01 5.32047510e-01 -1.68371484e-01
7.41002476e-03 -3.23195368e-01 -5.65697849e-01 1.12067938e+00
7.50921550e-04 1.68655351e-01 -3.85015041e-01 1.14008591e-01
1.22388303e+00 -1.92870691e-01 -6.27868056e-01 -6.86940491e-01
-1.15183830e+00 5.16651928e-01 1.86641723e-01 7.31598020e-01
-4.06077206e-01 3.17904264e-01 3.68668437e-01 1.92009304e-02
2.23091364e-01 2.36477152e-01 -2.13901728e-01 -5.33349574e-01
-1.51119709e+00 -2.87440002e-01 5.87432504e-01 5.14111876e-01
3.18668894e-02 2.35434204e-01 -2.09177643e-01 5.79399109e-01
4.89369094e-01 5.21924853e-01 9.53126848e-01 -1.25628924e+00
4.89649773e-01 1.30744185e-03 -1.93467066e-02 -1.16702986e+00
-1.17899311e+00 -1.75242215e-01 -9.26682651e-01 4.08568792e-02
4.07765567e-01 -3.17622870e-01 -1.27227092e+00 1.73428428e+00
3.67257953e-01 2.64963329e-01 -4.06464547e-01 1.13032949e+00
8.32226157e-01 4.20482904e-01 -1.28481733e-02 -6.36997640e-01
7.78547168e-01 -8.21480572e-01 -8.69158685e-01 -1.31663457e-01
1.00767553e+00 -4.69816744e-01 5.44265509e-01 2.63521791e-01
-1.55053926e+00 -2.47816861e-01 -9.38308954e-01 3.28657389e-01
3.98791701e-01 8.68646950e-02 4.43775922e-01 8.37348461e-01
-1.37316799e+00 1.04914033e+00 -1.85266817e+00 1.23777971e-01
4.53920275e-01 5.53466856e-01 -6.01476371e-01 1.80361301e-01
-9.18481648e-01 1.22515845e+00 5.22476435e-03 6.01385355e-01
-1.04401231e+00 -1.09581697e+00 -9.81396854e-01 -7.07679540e-02
1.83441862e-03 -4.95697409e-01 1.35509372e+00 -7.63840437e-01
-1.49616265e+00 7.36156344e-01 -7.62248710e-02 -2.16473892e-01
9.11688089e-01 -1.46040589e-01 -5.95112503e-01 6.64504528e-01
6.41112924e-02 4.50033367e-01 8.55415165e-01 -7.69612730e-01
8.19322467e-02 -1.88685939e-01 -6.53370857e-01 -1.43464789e-01
-3.15501057e-02 1.11758366e-01 -5.50917447e-01 -5.51621675e-01
5.78403771e-01 -1.15724361e+00 -3.63826066e-01 2.77104110e-01
1.32583737e-01 7.40744710e-01 5.87680161e-01 -1.27392566e+00
1.28521967e+00 -1.82427073e+00 -1.44616723e-01 3.59010130e-01
3.92099947e-01 5.41543365e-01 4.37471084e-02 -3.03187314e-03
-3.94015908e-01 -4.35945243e-02 -2.57264912e-01 2.16807425e-01
-7.62643218e-01 -1.08602025e-01 3.95281196e-01 7.36929297e-01
-2.72548944e-01 1.03755522e+00 -1.03770339e+00 -5.71441948e-01
3.39008778e-01 5.30429840e-01 -4.51026440e-01 4.72361624e-01
4.26382363e-01 1.16474342e+00 -2.45305032e-01 3.86586756e-01
4.28833395e-01 -3.17529380e-01 6.08605862e-01 -2.35188738e-01
-6.88354373e-02 2.15202853e-01 -7.38366902e-01 2.26414514e+00
-5.47371805e-01 8.11137378e-01 -2.21052095e-02 -9.21075761e-01
3.92915159e-01 9.36862946e-01 9.54111397e-01 -9.14658844e-01
9.89069343e-02 2.75264531e-01 3.84299874e-01 -1.32815623e+00
-1.73514426e-01 -1.17564075e-01 4.72991168e-01 6.10494077e-01
5.14026023e-02 4.30997200e-02 4.98994961e-02 -1.07975192e-01
1.34339058e+00 4.05191988e-01 -5.59196100e-02 -1.19404502e-01
2.73571372e-01 -4.31581050e-01 5.10703683e-01 6.27353966e-01
-1.59457728e-01 1.08700550e+00 3.57600451e-01 -8.45302284e-01
-1.03213143e+00 -1.05077684e+00 -1.06065251e-01 5.91012120e-01
-5.71004488e-02 1.10057056e-01 -6.11297190e-01 -7.32200086e-01
-4.37056601e-01 4.61207747e-01 -8.26985955e-01 -1.48953214e-01
-1.26245260e+00 -6.04322910e-01 5.71640253e-01 8.27670395e-01
-1.71189487e-01 -1.17874348e+00 -1.33463621e+00 6.90436780e-01
-2.40477875e-01 -8.78717065e-01 -6.72762334e-01 -2.66559273e-02
-9.84560728e-01 -9.59250808e-01 -1.04575646e+00 -5.85740089e-01
7.97141731e-01 -1.41929388e-01 8.88970375e-01 -1.42586976e-02
-3.79604965e-01 1.47017524e-01 -1.40610129e-01 6.11549206e-02
-7.54209995e-01 -2.00746253e-01 7.00087622e-02 -3.17584485e-01
-7.17808545e-01 -9.07174945e-01 -1.03853345e+00 2.03467250e-01
-7.84971356e-01 3.81885111e-01 6.30238950e-01 8.62599194e-01
1.43743426e-01 -8.43744040e-01 3.51329833e-01 -9.91379142e-01
4.86949421e-02 -5.27469933e-01 -6.48931623e-01 4.33201790e-01
-6.34243786e-01 2.30896145e-01 2.86049843e-01 -6.13425910e-01
-1.09642065e+00 -3.55018936e-02 -2.42747590e-01 -7.64805257e-01
2.15307325e-01 4.51446742e-01 4.57969785e-01 -5.04069328e-01
5.81708729e-01 2.06401020e-01 9.31661054e-02 -1.99886307e-01
-2.09466685e-02 1.77324876e-01 7.56864309e-01 7.66547993e-02
3.17855895e-01 4.23934191e-01 2.39709198e-01 -4.06390876e-01
-2.76223689e-01 6.41521886e-02 -7.07119286e-01 -7.40819991e-01
8.43791604e-01 -3.45342755e-01 -5.48815370e-01 2.07076982e-01
-1.18454003e+00 -5.57024419e-01 2.58842260e-01 9.28234994e-01
-5.27353585e-01 4.14930195e-01 -1.05086231e+00 -5.18655837e-01
-5.96168041e-01 -1.40204990e+00 5.30119538e-01 1.32131577e-01
-3.95218790e-01 -1.17795825e+00 1.29792094e-01 2.47950509e-01
7.50581920e-01 5.66198349e-01 7.99647629e-01 -2.45586529e-01
-4.66145128e-01 -1.65938392e-01 1.15048304e-01 -9.73725766e-02
3.18208218e-01 3.55626419e-02 -8.08842599e-01 -3.88909042e-01
1.64637923e-01 -4.92878072e-02 2.85670042e-01 9.36873555e-01
1.26970351e+00 -1.61847368e-01 -3.73425871e-01 9.98776376e-01
1.10578537e+00 7.78241396e-01 3.19241941e-01 2.39272341e-02
8.22974861e-01 5.58277190e-01 4.35037017e-01 3.57661217e-01
2.28561088e-01 5.69211125e-01 4.20020789e-01 3.59462127e-02
-1.54084608e-01 3.70677114e-02 -5.29632457e-02 1.12974429e+00
-4.28184539e-01 5.92246763e-02 -1.44751120e+00 5.89828730e-01
-1.49037814e+00 -8.15655887e-01 -5.11000082e-02 2.24649453e+00
8.26297283e-01 -6.06814921e-02 -2.27045089e-01 -3.11554253e-01
5.53748608e-01 6.93884939e-02 -3.72080773e-01 -3.98854047e-01
6.09195411e-01 1.30991846e-01 4.51154709e-01 2.53707945e-01
-8.83441865e-01 4.43836808e-01 6.39506435e+00 1.69401661e-01
-1.85367739e+00 4.22350913e-01 6.10633492e-01 -3.31046849e-01
-1.60394311e-01 -3.64145249e-01 8.28522295e-02 5.97886324e-01
1.16567469e+00 2.52259791e-01 1.77880704e-01 4.91157353e-01
4.67833400e-01 2.13355105e-02 -1.25204277e+00 8.30942035e-01
-4.14101295e-02 -1.69849110e+00 -5.05223334e-01 -3.69915724e-01
7.78598487e-01 1.90879166e-01 -7.45596364e-02 -2.20865533e-01
-3.14327329e-01 -1.23689926e+00 6.91416800e-01 7.86243796e-01
1.15146780e+00 -4.76483703e-01 6.21903777e-01 6.01480544e-01
-8.65958154e-01 2.68652827e-01 1.03544697e-01 4.53860670e-01
2.83944815e-01 1.12923279e-01 -1.09503806e+00 5.25604010e-01
7.11234152e-01 3.84312332e-01 -1.09212428e-01 1.04924035e+00
4.51688096e-02 6.10644698e-01 -1.32087916e-01 3.25052649e-01
1.94294378e-01 8.79778937e-02 4.42599386e-01 1.29743588e+00
5.77548087e-01 7.68771768e-02 -4.08833534e-01 5.02587676e-01
2.50556380e-01 -9.95376036e-02 -6.38745070e-01 2.04289764e-01
3.21042806e-01 1.35173619e+00 -9.48580503e-01 -1.50502250e-01
-3.76065016e-01 7.10027099e-01 2.97155231e-01 2.00969025e-01
-9.28608000e-01 1.54943513e-02 -1.33378550e-01 4.72633421e-01
1.97738603e-01 -2.66474575e-01 -9.34886038e-02 -9.99918580e-01
1.44190475e-01 -4.44672614e-01 2.40168185e-03 -7.20145345e-01
-3.49492580e-01 7.19423294e-01 -2.05937624e-01 -1.38129902e+00
-1.00854242e+00 -1.73437878e-01 -7.33387053e-01 6.98480725e-01
-1.00717056e+00 -7.21445441e-01 -2.76526630e-01 1.49596781e-01
2.54807472e-01 1.02494121e-01 7.46120155e-01 6.51838243e-01
-2.51653671e-01 5.66402197e-01 -9.94827524e-02 3.24056804e-01
4.46291029e-01 -9.20886397e-01 2.67600209e-01 8.49051595e-01
4.89594415e-02 5.61479568e-01 5.63129485e-01 -7.03016639e-01
-1.31674778e+00 -9.49710369e-01 5.68605959e-01 -3.37235898e-01
4.33106750e-01 5.99252433e-03 -1.12451267e+00 7.40666091e-01
-1.01920344e-01 5.71073353e-01 5.67952931e-01 -6.89347625e-01
4.13072348e-01 1.32212996e-01 -1.01515341e+00 3.91445041e-01
1.08533168e+00 -4.72739451e-02 -2.69694149e-01 1.23505913e-01
2.48656422e-01 -1.37060106e+00 -1.00775230e+00 7.81438708e-01
1.05884945e+00 -9.16633606e-01 6.61669493e-01 -4.89080459e-01
5.71809113e-01 1.31114095e-01 7.00131655e-01 -1.14040244e+00
-3.74101400e-01 -6.41560078e-01 -3.55166286e-01 6.22800887e-01
6.18696034e-01 -2.93961316e-01 1.05796850e+00 9.01087701e-01
-2.37200573e-01 -9.14514184e-01 -1.12696302e+00 -2.87448198e-01
-2.78645664e-01 -6.07514143e-01 9.86593142e-02 1.12908816e+00
-9.82131213e-02 -2.58345932e-01 -5.20476878e-01 3.82853568e-01
2.28488788e-01 1.65982559e-01 2.51309350e-02 -4.99242157e-01
-2.96974957e-01 -3.18655759e-01 -6.15628481e-01 -6.27047479e-01
-1.96554735e-01 -8.33698213e-01 1.27512574e-01 -1.47599351e+00
-7.00695589e-02 -4.28568363e-01 -8.81974772e-02 3.91273439e-01
4.47760932e-02 2.84720600e-01 6.07088953e-02 1.94943666e-01
-1.91639215e-01 5.76063693e-02 1.69992113e+00 2.22735286e-01
-1.22450456e-01 3.30853492e-01 1.18562743e-01 9.70695496e-01
3.34179640e-01 -6.38700843e-01 -4.32410032e-01 -8.72756720e-01
2.20879242e-01 1.28309596e+00 9.00535062e-02 -1.09599638e+00
4.23036814e-01 7.84035623e-02 9.42930281e-01 -5.38520336e-01
1.63557678e-01 -7.53665864e-01 4.07199085e-01 8.94259751e-01
-3.36962104e-01 2.30920330e-01 1.03717409e-01 1.58954915e-02
-1.71688963e-02 -2.38261282e-01 8.25763464e-01 -1.34809077e-01
-2.84948647e-01 6.18925750e-01 -5.82394123e-01 -1.21341031e-02
9.04459059e-01 -2.76721090e-01 2.06668526e-01 -5.20636678e-01
-1.37664342e+00 1.73147202e-01 2.66084950e-02 2.70082772e-01
8.60512674e-01 -1.11861384e+00 -6.11097515e-01 2.70335495e-01
-4.03297901e-01 3.25097591e-02 6.54326856e-01 1.43397343e+00
-1.35655987e+00 3.77002865e-01 -3.11274230e-01 -8.88756275e-01
-1.02114820e+00 3.60423863e-01 6.85262978e-01 -2.99787760e-01
-9.23773170e-01 8.86268139e-01 1.02225475e-01 -2.60818690e-01
-7.15187564e-02 -4.31244731e-01 -3.62814784e-01 -2.93707758e-01
4.70054299e-01 1.97444260e-01 3.52104485e-01 -4.34491426e-01
-4.24372047e-01 6.71649098e-01 5.57596758e-02 -2.30832085e-01
1.34845912e+00 -1.78143755e-01 1.92546532e-01 3.66105437e-01
1.04011559e+00 -4.03182879e-02 -1.46151590e+00 1.40069827e-01
1.77574173e-01 -4.53357130e-01 -1.99335814e-01 -8.01392555e-01
-1.61628067e+00 1.09352994e+00 9.50333893e-01 2.79849619e-02
1.15183723e+00 -2.56749392e-01 6.50215864e-01 -9.10157114e-02
3.48271132e-01 -6.60024881e-01 -9.47809964e-02 2.01844387e-02
8.60315442e-01 -1.13279808e+00 -2.21861660e-01 -1.20396532e-01
-6.16806090e-01 1.43826878e+00 7.01452374e-01 1.84079573e-01
4.88330066e-01 6.57338083e-01 4.32234108e-01 -1.68886811e-01
-7.14804649e-01 7.60540366e-01 4.92447346e-01 3.94239902e-01
7.10066915e-01 -1.68156475e-01 -1.88747078e-01 2.48759016e-01
2.54531065e-03 2.64770895e-01 5.29791415e-01 1.02684236e+00
-6.18900955e-02 -7.23770082e-01 -1.43276393e-01 5.67817152e-01
-9.40618217e-01 9.35487524e-02 5.96729517e-01 7.03223825e-01
5.27906083e-02 3.51463407e-01 -1.67747706e-01 -4.45850268e-02
-4.18361537e-02 -7.45200962e-02 8.52347732e-01 -4.99371976e-01
-5.31853974e-01 8.41344520e-02 8.41940194e-02 -9.80674148e-01
-2.47421801e-01 -7.70883560e-01 -1.44400799e+00 1.29980847e-01
-1.12566218e-01 -1.31722689e-02 8.92187059e-01 1.00790882e+00
1.36340380e-01 7.62169003e-01 7.04625666e-01 -9.93652225e-01
-1.46090955e-01 -9.91487145e-01 -3.80908728e-01 9.12616178e-02
5.46404362e-01 -4.70004350e-01 9.41092819e-02 1.90840304e-01]
|
[13.969100952148438, -2.433586597442627]
|
137c1b95-b1e6-4cc3-8b24-ff8c8e6ea268
|
tpmil-trainable-prototype-enhanced-multiple
|
2305.00696
| null |
https://arxiv.org/abs/2305.00696v1
|
https://arxiv.org/pdf/2305.00696v1.pdf
|
TPMIL: Trainable Prototype Enhanced Multiple Instance Learning for Whole Slide Image Classification
|
Digital pathology based on whole slide images (WSIs) plays a key role in cancer diagnosis and clinical practice. Due to the high resolution of the WSI and the unavailability of patch-level annotations, WSI classification is usually formulated as a weakly supervised problem, which relies on multiple instance learning (MIL) based on patches of a WSI. In this paper, we aim to learn an optimal patch-level feature space by integrating prototype learning with MIL. To this end, we develop a Trainable Prototype enhanced deep MIL (TPMIL) framework for weakly supervised WSI classification. In contrast to the conventional methods which rely on a certain number of selected patches for feature space refinement, we softly cluster all the instances by allocating them to their corresponding prototypes. Additionally, our method is able to reveal the correlations between different tumor subtypes through distances between corresponding trained prototypes. More importantly, TPMIL also enables to provide a more accurate interpretability based on the distance of the instances from the trained prototypes which serves as an alternative to the conventional attention score-based interpretability. We test our method on two WSI datasets and it achieves a new SOTA. GitHub repository: https://github.com/LitaoYang-Jet/TPMIL
|
['ZongYuan Ge', 'Antonio Di Ieva', 'Dwarikanath Mahapatra', 'Sidong Liu', 'Deval Mehta', 'Litao Yang']
|
2023-05-01
| null | null | null | null |
['whole-slide-images', 'multiple-instance-learning']
|
['computer-vision', 'methodology']
|
[ 1.87651619e-01 3.45672965e-01 -3.19340467e-01 -3.33927810e-01
-1.16025746e+00 -3.09532285e-01 4.90139991e-01 7.40781546e-01
-2.75036365e-01 5.81254959e-01 2.52180099e-01 -5.18851839e-02
-5.14843822e-01 -5.76458693e-01 -6.07056499e-01 -1.08707607e+00
1.93723515e-01 5.86415648e-01 8.68875682e-02 -5.34778051e-02
5.34980074e-02 3.99768293e-01 -1.31089818e+00 5.27592659e-01
9.91462350e-01 1.26610005e+00 2.58556843e-01 3.91464055e-01
-1.49972633e-01 4.00546551e-01 -3.90063435e-01 -1.66069075e-01
-1.82740930e-02 -1.86912492e-01 -8.00550520e-01 1.04213366e-02
3.46718907e-01 2.22001955e-01 -1.30816633e-02 1.05803919e+00
1.78634584e-01 -1.55215681e-01 6.95197821e-01 -1.04176521e+00
-3.37025136e-01 3.69899541e-01 -5.41014791e-01 6.63537607e-02
5.59970587e-02 9.15995017e-02 1.38402653e+00 -7.47772455e-01
6.90190732e-01 6.92555249e-01 6.67699099e-01 3.81575406e-01
-1.24997306e+00 -3.38988364e-01 1.06918611e-01 4.75126326e-01
-1.58156133e+00 -2.57051051e-01 8.14733565e-01 -2.55732745e-01
4.58453536e-01 6.65431976e-01 7.14003682e-01 7.95887053e-01
6.42826557e-02 8.99200082e-01 8.92522275e-01 -4.42736000e-01
1.98114350e-01 2.68811136e-01 3.35379988e-01 6.06340528e-01
1.11634061e-01 -3.63483459e-01 -2.83124268e-01 -4.19244021e-02
5.06652355e-01 4.39635336e-01 -5.26628673e-01 -2.41080970e-01
-1.45117640e+00 6.03756011e-01 9.74332094e-01 6.06038392e-01
-1.14641070e-01 -1.08514175e-01 3.68448079e-01 -6.52929321e-02
5.65764248e-01 6.13116026e-01 -3.32142830e-01 3.67015243e-01
-8.47008109e-01 -1.52904630e-01 3.18577737e-01 4.49046761e-01
6.76461518e-01 -6.15474403e-01 -5.32797277e-01 7.22233295e-01
1.94306105e-01 -2.06546988e-02 6.97518945e-01 -3.76939863e-01
1.48643166e-01 1.12764418e+00 -1.70296878e-01 -8.10957074e-01
-5.53240001e-01 -8.75975370e-01 -1.17709541e+00 4.13056575e-02
3.70345533e-01 5.10622382e-01 -7.25731015e-01 1.46370292e+00
5.07207036e-01 4.69410747e-01 -1.06261864e-01 8.93315136e-01
8.19975615e-01 3.15187365e-01 2.16486696e-02 -2.41667833e-02
1.47029293e+00 -8.27154815e-01 -4.94516313e-01 2.05254897e-01
9.48614955e-01 -3.21954042e-01 1.36266136e+00 2.85738587e-01
-6.34560168e-01 -3.74422908e-01 -9.29476857e-01 -1.59812063e-01
-4.77907687e-01 5.44441700e-01 4.53950226e-01 1.34696350e-01
-1.02236378e+00 4.54501987e-01 -9.44708526e-01 -3.20320964e-01
8.41022134e-01 4.26934779e-01 -6.23685718e-01 -3.69528681e-02
-7.79309750e-01 6.42793775e-01 3.22033077e-01 2.37470880e-01
-4.40564305e-01 -9.77148712e-01 -8.05318415e-01 1.88367099e-01
2.22834781e-01 -4.93941665e-01 6.90855503e-01 -1.05857503e+00
-1.36595094e+00 9.48710322e-01 -3.49104464e-01 -3.15493882e-01
4.43998873e-01 2.45984048e-01 -5.32208681e-02 2.66200036e-01
7.38965273e-02 4.33571190e-01 5.87222338e-01 -1.07257748e+00
-4.72524434e-01 -5.31114042e-01 -1.65741872e-02 1.34241402e-01
-7.59603024e-01 -4.94068474e-01 -3.75869870e-01 -4.86402303e-01
2.26705179e-01 -8.31969559e-01 -1.59854293e-01 2.76551992e-01
-6.91092432e-01 -5.09257376e-01 5.57187557e-01 -5.96727073e-01
1.21723878e+00 -2.30008984e+00 3.94304007e-01 3.32554817e-01
5.77788115e-01 3.50222647e-01 -1.08242564e-01 2.25515179e-02
-6.69789240e-02 8.91343206e-02 -2.63122678e-01 -5.25648117e-01
-6.16886169e-02 5.12119457e-02 1.92483500e-01 6.10354185e-01
5.30347466e-01 9.51340199e-01 -9.46933568e-01 -5.47700167e-01
1.42968327e-01 3.40789288e-01 -5.15518785e-01 1.62265807e-01
-2.03112498e-01 7.25878358e-01 -3.32627237e-01 6.42986000e-01
5.03027856e-01 -5.84352612e-01 2.84156315e-02 -4.39063817e-01
1.26139671e-01 -3.58209126e-02 -8.75967145e-01 1.72986710e+00
-3.63151968e-01 4.52781618e-01 -1.75481904e-02 -1.35643637e+00
7.30341554e-01 3.07413965e-01 5.55063844e-01 -3.77045721e-01
-9.47412029e-02 3.64522308e-01 -4.61948514e-02 -4.57684427e-01
-2.25191619e-02 -7.71517456e-02 2.60436591e-02 7.44383186e-02
1.28186479e-01 1.87307492e-01 9.03921649e-02 1.10681504e-01
1.13320363e+00 -1.59113154e-01 5.04213929e-01 -3.60194623e-01
7.53676772e-01 -1.00094907e-01 6.94246471e-01 5.71238041e-01
-1.53740183e-01 7.81192183e-01 5.70034623e-01 -4.87425506e-01
-8.59201193e-01 -9.07672346e-01 -6.36518598e-01 7.02475905e-01
2.23197356e-01 -3.11644614e-01 -4.59122092e-01 -1.03151548e+00
-1.05023034e-01 3.18971336e-01 -1.03680038e+00 -1.32550001e-01
-3.01247537e-01 -8.31528485e-01 2.84833193e-01 4.38608259e-01
1.55107051e-01 -7.05624402e-01 -1.67961597e-01 -2.14267485e-02
-2.63200223e-01 -7.55968034e-01 -4.33514893e-01 2.23679841e-01
-6.26247168e-01 -1.15816736e+00 -6.79683089e-01 -6.50570035e-01
1.21124625e+00 1.37838617e-01 9.14035201e-01 4.10961300e-01
-6.68316185e-01 4.70002694e-03 -5.23124635e-01 -3.95361722e-01
-3.20112050e-01 3.30714375e-01 -3.32834512e-01 4.55690086e-01
2.45000497e-01 -3.45032007e-01 -8.67191672e-01 1.09239191e-01
-1.13764739e+00 3.64610493e-01 8.82020473e-01 1.20412862e+00
1.01950490e+00 -1.79145724e-01 4.48004693e-01 -9.75265682e-01
1.84658885e-01 -5.56103349e-01 -3.23003948e-01 5.44473946e-01
-2.55982697e-01 2.17311203e-01 6.98358595e-01 -2.97815859e-01
-8.24350059e-01 -5.07984776e-03 -2.90250421e-01 -1.87969178e-01
-3.48150492e-01 7.25097299e-01 -1.95557669e-01 -6.36881441e-02
6.75541222e-01 1.84353679e-01 1.91662222e-01 -1.70256525e-01
-2.62308512e-02 6.33164167e-01 1.13585770e-01 -2.59021610e-01
6.15712523e-01 5.69643736e-01 1.65100485e-01 -7.32673883e-01
-1.03186500e+00 -6.51440382e-01 -7.14522839e-01 -6.63959458e-02
6.30662262e-01 -5.54221153e-01 -7.86609828e-01 1.44032449e-01
-7.76376128e-01 -4.08059448e-01 -4.73786771e-01 3.20598722e-01
-3.20493430e-01 3.69126678e-01 -6.36850595e-01 -3.18945885e-01
-4.76586461e-01 -1.27138364e+00 1.36549819e+00 1.50438949e-01
-2.42765754e-01 -1.07683933e+00 1.34505942e-01 3.60541254e-01
3.80346864e-01 4.53220367e-01 1.05837440e+00 -8.52317154e-01
-3.93981218e-01 -5.65361798e-01 -3.62654507e-01 1.05937086e-01
2.98447251e-01 8.10966268e-02 -8.68256688e-01 -3.58320415e-01
-2.93887764e-01 -9.05551910e-02 1.02926993e+00 4.66585547e-01
1.50067449e+00 -3.26446712e-01 -6.66189253e-01 8.24057162e-01
1.49255300e+00 -2.94062078e-01 4.18900132e-01 3.18934232e-01
6.63649082e-01 4.78429258e-01 6.77406371e-01 4.05976653e-01
2.72631288e-01 8.24439943e-01 5.20891130e-01 -3.44816715e-01
-1.45781502e-01 8.35495163e-03 -1.94169745e-01 4.88091826e-01
-8.26435685e-02 -1.42137989e-01 -9.66938078e-01 5.59109926e-01
-1.91501915e+00 -6.44249141e-01 -5.69238514e-02 2.33283758e+00
8.45474362e-01 5.26723228e-02 -2.34339952e-01 2.93508112e-01
5.78985214e-01 -7.56101161e-02 -5.28651536e-01 1.88893482e-01
-1.37881124e-02 8.69562328e-02 8.55527744e-02 4.34556961e-01
-1.08964097e+00 5.38623095e-01 4.43182039e+00 1.12803376e+00
-1.20658171e+00 2.11404458e-01 1.05736840e+00 -2.40461394e-01
-3.35577548e-01 -3.31875116e-01 -8.34997475e-01 5.47878802e-01
5.33469439e-01 7.04431906e-02 -4.75383848e-02 6.28462672e-01
1.76943809e-01 -3.35309282e-02 -1.10942686e+00 7.43425846e-01
-1.71254873e-02 -1.52996373e+00 8.57497230e-02 2.13167772e-01
5.31244755e-01 -1.68798372e-01 1.24669835e-01 1.55646220e-01
-2.51720935e-01 -9.59165215e-01 3.36415410e-01 6.23312533e-01
8.22082102e-01 -5.39007008e-01 1.14114201e+00 2.59485781e-01
-9.96184587e-01 -1.37848973e-01 -3.99342328e-01 2.78816581e-01
-3.72526139e-01 9.04643714e-01 -1.22642124e+00 6.43164456e-01
5.55718839e-01 8.18221092e-01 -6.94699764e-01 1.15375018e+00
-1.82774529e-01 5.12861788e-01 -3.85667443e-01 -2.09929626e-02
1.15955174e-02 -3.49411392e-03 4.09483433e-01 1.17342436e+00
4.41422731e-01 -3.05900015e-02 4.00822656e-03 7.64195383e-01
4.73727770e-02 2.09812149e-01 -1.74633592e-01 3.02842051e-01
2.59438366e-01 1.66389835e+00 -8.10356200e-01 -2.62109160e-01
-1.95102155e-01 8.89504254e-01 6.14945889e-01 1.86220542e-01
-6.26813114e-01 -2.45500877e-01 5.90775251e-01 2.55907059e-01
9.34824422e-02 2.13526651e-01 -3.24071556e-01 -1.25840688e+00
8.61541107e-02 -5.12904108e-01 6.40903771e-01 -3.89310360e-01
-1.35301208e+00 7.79819071e-01 -3.60171586e-01 -1.55975103e+00
2.39820465e-01 -6.36605442e-01 -8.36514354e-01 5.75170577e-01
-1.78049159e+00 -1.50168133e+00 -7.52060890e-01 4.70123529e-01
3.33893180e-01 1.46499038e-01 1.04603684e+00 4.53107879e-02
-8.23913693e-01 8.76493573e-01 2.76824236e-01 1.62590936e-01
7.94470012e-01 -1.47657216e+00 -3.31356972e-01 3.49387854e-01
9.02205259e-02 4.32337672e-01 4.91575539e-01 -2.54496396e-01
-1.00026000e+00 -1.24160028e+00 6.41544104e-01 -3.03022176e-01
6.38006032e-01 -2.04416275e-01 -1.10552597e+00 5.16665936e-01
-9.55455974e-02 5.35898745e-01 1.22931015e+00 1.39180988e-01
-1.46259889e-01 -5.49183667e-01 -1.07397103e+00 4.34339166e-01
7.87732244e-01 -2.07719043e-01 -7.72613138e-02 5.39090574e-01
4.95918930e-01 -2.60532767e-01 -1.18067491e+00 4.45260555e-01
3.69932950e-01 -6.83131397e-01 8.40600491e-01 -4.02228296e-01
3.93910319e-01 -4.67340171e-01 2.70320028e-01 -1.44475496e+00
-5.06664574e-01 3.88111807e-02 1.42400041e-01 1.11786664e+00
5.31491756e-01 -7.67578244e-01 7.69791305e-01 5.66188335e-01
-4.80045885e-01 -1.16053712e+00 -1.18348002e+00 -3.42847675e-01
-1.42033547e-01 -6.68391809e-02 5.56032181e-01 8.41269493e-01
3.70265454e-01 -1.38742194e-01 1.67575404e-01 3.62685829e-01
6.54247582e-01 2.96889722e-01 3.63465697e-01 -1.33655655e+00
-6.48213863e-01 -7.62319744e-01 -7.35729218e-01 -4.68446732e-01
1.41348153e-01 -1.23122907e+00 1.99002512e-02 -1.50424111e+00
4.39728618e-01 -7.54390717e-01 -7.83173859e-01 6.96616232e-01
-4.96067911e-01 4.80950505e-01 -1.94714487e-01 3.87738973e-01
-9.62596714e-01 4.85041231e-01 1.15035963e+00 -4.49488550e-01
-1.10884033e-01 1.01679482e-01 -6.95134699e-01 4.39075828e-01
8.14191639e-01 -2.29526907e-01 -4.90894392e-02 -1.99679360e-01
1.71864673e-01 -6.44972473e-02 3.99713516e-01 -8.65789473e-01
3.43150169e-01 -2.28960551e-02 4.06337410e-01 -4.06676292e-01
2.48289302e-01 -6.29240155e-01 1.23932488e-01 4.68303889e-01
-5.46387434e-01 -7.17840254e-01 7.07194358e-02 5.77416003e-01
-4.58718538e-01 -2.85495430e-01 7.15490997e-01 -8.09748322e-02
-4.01953131e-01 4.88234937e-01 -1.49217829e-01 -4.02260244e-01
1.13708723e+00 -2.08228320e-01 -4.02484477e-01 2.14701921e-01
-8.49348724e-01 2.71092117e-01 3.46088856e-01 6.25125766e-02
5.94186127e-01 -1.20082986e+00 -7.86564767e-01 3.07035565e-01
6.85484052e-01 4.37737733e-01 5.69408894e-01 1.36185360e+00
-3.67865771e-01 4.73674089e-01 5.71637861e-02 -8.12834382e-01
-1.28360605e+00 3.78874838e-01 3.79829943e-01 -7.10467875e-01
-4.87676859e-01 9.06921804e-01 5.62865376e-01 -3.47467780e-01
1.20712630e-01 -4.01171654e-01 -3.27051967e-01 -8.13319013e-02
5.85870385e-01 -9.42013636e-02 3.05205643e-01 -3.86271536e-01
-2.73648590e-01 3.91535014e-01 -2.89272904e-01 4.86717224e-01
1.56287658e+00 1.09539352e-01 -2.83562988e-01 4.19922769e-01
1.33907890e+00 -1.44720241e-01 -1.24677038e+00 -4.02188569e-01
1.21706218e-01 -3.51607025e-01 6.25356287e-02 -6.42009079e-01
-1.05545425e+00 7.81360686e-01 5.84319115e-01 3.85920107e-02
1.19200099e+00 3.07297796e-01 5.44414341e-01 9.30287018e-02
6.08123392e-02 -7.09453285e-01 2.35515326e-01 -9.14774686e-02
9.05449927e-01 -1.53397357e+00 -1.84898540e-01 -4.39371496e-01
-3.21989417e-01 1.23572016e+00 4.16576117e-01 -3.22151482e-02
5.20104349e-01 1.77027181e-01 -9.81497765e-02 -1.25410765e-01
-5.88278234e-01 -2.38684863e-01 5.00850856e-01 3.10773760e-01
4.03455794e-01 3.26735854e-01 -3.27637166e-01 9.03626800e-01
1.53326526e-01 -7.12544024e-02 1.91784650e-01 4.69486386e-01
-3.16778481e-01 -1.06643176e+00 -2.06546634e-01 7.56838560e-01
-3.59178156e-01 5.48394956e-02 -1.28226221e-01 5.49186409e-01
8.33849311e-02 5.41859925e-01 7.92669430e-02 -2.35611543e-01
1.46581069e-01 -8.47257003e-02 2.43260309e-01 -7.10271180e-01
-4.70976859e-01 2.40412802e-02 -3.67942750e-01 -3.33130360e-01
-1.98055491e-01 -5.89958727e-01 -1.16280913e+00 2.70962030e-01
-5.07968366e-01 1.91808105e-01 4.64206338e-01 1.02422476e+00
4.90175635e-01 6.25663102e-01 6.54525518e-01 -7.98616111e-01
-2.79020727e-01 -8.57204914e-01 -5.62283337e-01 6.72519982e-01
4.25631702e-01 -5.49156487e-01 -4.53647405e-01 -1.65869996e-01]
|
[15.083977699279785, -2.8334624767303467]
|
0dbdb894-3c7f-49c9-9e88-b3dbe97035c8
|
the-right-spin-learning-object-motion-from
|
2203.00115
| null |
https://arxiv.org/abs/2203.00115v1
|
https://arxiv.org/pdf/2203.00115v1.pdf
|
The Right Spin: Learning Object Motion from Rotation-Compensated Flow Fields
|
Both a good understanding of geometrical concepts and a broad familiarity with objects lead to our excellent perception of moving objects. The human ability to detect and segment moving objects works in the presence of multiple objects, complex background geometry, motion of the observer and even camouflage. How humans perceive moving objects so reliably is a longstanding research question in computer vision and borrows findings from related areas such as psychology, cognitive science and physics. One approach to the problem is to teach a deep network to model all of these effects. This contrasts with the strategy used by human vision, where cognitive processes and body design are tightly coupled and each is responsible for certain aspects of correctly identifying moving objects. Similarly from the computer vision perspective, there is evidence that classical, geometry-based techniques are better suited to the "motion-based" parts of the problem, while deep networks are more suitable for modeling appearance. In this work, we argue that the coupling of camera rotation and camera translation can create complex motion fields that are difficult for a deep network to untangle directly. We present a novel probabilistic model to estimate the camera's rotation given the motion field. We then rectify the flow field to obtain a rotation-compensated motion field for subsequent segmentation. This strategy of first estimating camera motion, and then allowing a network to learn the remaining parts of the problem, yields improved results on the widely used DAVIS benchmark as well as the recently published motion segmentation data set MoCA (Moving Camouflaged Animals).
|
['Karteek Alahari', 'Cordelia Schmid', 'Erik Learned-Miller', 'Pia Bideau']
|
2022-02-28
| null | null | null | null |
['motion-segmentation']
|
['computer-vision']
|
[ 1.96845427e-01 -2.81376064e-01 -1.11958692e-02 -1.72937736e-01
5.27826846e-02 -8.95678461e-01 6.20933950e-01 -2.27042824e-01
-5.34413397e-01 2.95211375e-01 -1.76455483e-01 -3.15541148e-01
-7.18606710e-02 -5.98016858e-01 -6.90156579e-01 -8.31689000e-01
1.21707864e-01 3.24001342e-01 4.80119616e-01 -2.39319026e-01
4.23972428e-01 6.94720268e-01 -1.36554265e+00 -1.33526444e-01
5.57428896e-01 5.06087065e-01 2.86810666e-01 9.86279845e-01
5.99727705e-02 6.51836693e-01 -4.33649004e-01 -1.40919924e-01
3.04896951e-01 -4.59636688e-01 -1.05674124e+00 3.60252351e-01
9.75570738e-01 -2.40782216e-01 -2.98337549e-01 9.93837476e-01
1.08573064e-01 4.94660348e-01 6.66982889e-01 -1.03295636e+00
-7.57630408e-01 1.53585628e-01 -7.39159048e-01 4.55993056e-01
2.97067344e-01 2.13206947e-01 6.70682430e-01 -4.98394340e-01
7.55100012e-01 1.42127848e+00 6.34893298e-01 6.35590136e-01
-1.28150105e+00 -5.66859208e-02 3.50517094e-01 1.89557076e-01
-1.06202447e+00 -1.04782358e-01 7.67904520e-01 -8.26465011e-01
6.56715810e-01 2.46917725e-01 8.81330192e-01 9.86245215e-01
3.75674069e-01 6.63596094e-01 8.21406543e-01 -4.50656205e-01
1.41475186e-01 -2.41536368e-02 1.65554792e-01 5.26995659e-01
3.35695028e-01 2.65145391e-01 -4.28797165e-03 2.13544622e-01
1.19679117e+00 1.29367441e-01 -4.11911666e-01 -7.68773079e-01
-1.29210496e+00 7.60425270e-01 5.00360608e-01 5.38049102e-01
-1.76751792e-01 4.22148526e-01 7.09633678e-02 3.35301198e-02
1.35761306e-01 4.99764532e-01 -4.88729388e-01 1.23869339e-02
-1.09644067e+00 3.81085217e-01 8.56410861e-01 5.96552253e-01
5.62639713e-01 2.02041492e-01 1.81830838e-01 4.52536851e-01
4.10575509e-01 4.18152034e-01 3.17615569e-01 -1.32854378e+00
6.42399415e-02 2.51755208e-01 1.43559143e-01 -1.33052135e+00
-5.87014616e-01 -3.32810223e-01 -8.93386602e-01 6.43163860e-01
1.07799745e+00 1.29791666e-02 -1.19077146e+00 1.91972494e+00
4.00680751e-01 1.98718384e-01 -2.02552244e-01 1.19371080e+00
4.01304334e-01 4.38846916e-01 1.86995775e-01 1.79932803e-01
1.39370215e+00 -9.25110281e-01 -4.02303934e-01 -5.72181106e-01
4.59226191e-01 -8.45418632e-01 7.49275386e-01 5.07609427e-01
-1.02775013e+00 -6.86571419e-01 -9.91494298e-01 -2.41318077e-01
-4.67741102e-01 1.90365314e-02 7.37053275e-01 6.46271765e-01
-1.23529387e+00 8.35249722e-01 -9.86181498e-01 -5.74460924e-01
6.00788713e-01 4.38423991e-01 -2.82813638e-01 -1.80834010e-01
-8.24483871e-01 1.32921791e+00 3.27966690e-01 2.51466632e-01
-6.92855656e-01 -4.16817665e-01 -9.19867516e-01 -1.01789668e-01
3.08717430e-01 -1.27259624e+00 1.12469947e+00 -1.41550946e+00
-1.31202352e+00 8.78212154e-01 -1.17016219e-01 -2.62129724e-01
8.09919178e-01 -3.11812282e-01 -6.48407191e-02 2.99169153e-01
-2.67077327e-01 9.81494606e-01 1.07060885e+00 -1.32849634e+00
-4.71642941e-01 -3.93988162e-01 2.19844699e-01 1.70904011e-01
3.22067976e-01 -2.14636520e-01 -3.91328603e-01 -7.33394444e-01
1.79981291e-01 -1.29083216e+00 -4.96844918e-01 2.11043760e-01
-1.51766405e-01 2.76695192e-02 9.85635459e-01 -7.64241338e-01
6.66358232e-01 -1.87652361e+00 5.22675395e-01 -4.64145392e-02
3.75894278e-01 3.98017347e-01 -1.18914440e-01 -7.96036199e-02
-3.10562223e-01 1.17933564e-02 -3.42673719e-01 -1.33097664e-01
-1.55375674e-01 2.56132245e-01 7.60312891e-03 8.56868804e-01
2.54210711e-01 1.06077147e+00 -8.88501167e-01 -1.55389532e-01
5.95977485e-01 6.56592786e-01 -5.85822523e-01 -2.39478841e-01
-8.24364722e-02 6.40289903e-01 -1.53047040e-01 3.51098537e-01
7.64178276e-01 -1.93887278e-01 4.26635519e-02 -4.99865040e-02
2.43944880e-02 -1.73552632e-01 -1.27087080e+00 1.65701544e+00
-1.95159018e-01 1.08375752e+00 1.13924816e-01 -1.08973980e+00
4.40441608e-01 1.96201682e-01 3.25754464e-01 -3.42067212e-01
3.99409622e-01 -1.72630847e-01 3.74854654e-01 -6.61111712e-01
4.48833078e-01 -2.91403294e-01 2.50577390e-01 2.60954797e-01
-1.98844355e-02 -3.90476912e-01 3.67018096e-02 8.17664564e-02
8.87571812e-01 4.76886332e-01 1.78726777e-01 -3.02673578e-01
4.83376890e-01 1.90224707e-01 4.00305241e-01 6.46206141e-01
-3.64040911e-01 7.67929554e-01 1.18098848e-01 -7.61585295e-01
-1.07256162e+00 -1.03663146e+00 1.72176585e-01 9.30868566e-01
4.15859550e-01 2.24555582e-01 -9.43510771e-01 -4.87152398e-01
-1.84202820e-01 5.70035219e-01 -6.80319250e-01 -9.64645594e-02
-7.92756021e-01 -6.96506798e-01 1.74039289e-01 6.50723040e-01
3.35385889e-01 -1.12743831e+00 -1.05037177e+00 1.52316123e-01
-2.24048480e-01 -1.15150452e+00 -4.27661896e-01 -1.07500732e-01
-8.35351586e-01 -1.10180974e+00 -8.83197963e-01 -8.01822066e-01
7.05802917e-01 5.71741223e-01 1.16258800e+00 2.74312258e-01
-5.58862209e-01 5.30617416e-01 -1.33099094e-01 -2.32427701e-01
-3.45824987e-01 -2.19695330e-01 -4.98976372e-02 -2.78095901e-01
2.59416461e-01 -4.39342827e-01 -7.68793464e-01 2.76420325e-01
-1.21698403e+00 1.48041829e-01 5.48966467e-01 4.69859421e-01
1.18029043e-01 8.93200338e-02 3.51364026e-03 -6.39122725e-01
1.16419367e-01 -2.37514034e-01 -5.75031877e-01 1.00478463e-01
-2.46450286e-02 -6.59021884e-02 3.07992905e-01 -7.76725173e-01
-8.93767178e-01 3.95506591e-01 5.20746522e-02 -3.99222910e-01
-7.05864072e-01 7.12427944e-02 -1.02907196e-01 -3.09211135e-01
5.44317305e-01 -3.07996385e-02 -7.19006732e-02 -2.53723532e-01
6.76382959e-01 -1.31582513e-01 7.24005401e-01 -4.34778452e-01
9.77270603e-01 9.60567951e-01 1.99687108e-01 -1.08875835e+00
-5.90273440e-01 -4.82267350e-01 -1.13331199e+00 -2.03932390e-01
1.41512620e+00 -5.32141745e-01 -6.87680423e-01 7.04892516e-01
-1.46375263e+00 -3.36382031e-01 -2.06276834e-01 6.20797813e-01
-5.85375309e-01 4.84022677e-01 -4.62166429e-01 -7.18160212e-01
2.65733421e-01 -1.29775798e+00 9.67482746e-01 4.31236178e-01
-3.83074969e-01 -1.44318497e+00 9.55384411e-03 3.60588014e-01
3.76815736e-01 4.44259495e-01 8.92278314e-01 -2.13040575e-01
-8.10668528e-01 -7.73720220e-02 -1.23557426e-01 3.66987467e-01
7.36859292e-02 1.99471831e-01 -8.87615800e-01 -2.14964692e-02
2.78193742e-01 1.64753720e-01 9.80633795e-01 8.96733999e-01
8.49292099e-01 1.45696965e-03 -3.88131171e-01 6.09906495e-01
1.51285756e+00 2.67625481e-01 6.64909601e-01 1.98236182e-01
1.18788862e+00 1.00530803e+00 2.43462086e-01 -1.50807679e-01
4.32342947e-01 7.76661813e-01 3.99346977e-01 -4.09963399e-01
-6.22682683e-02 1.38199940e-01 2.15152830e-01 5.64501107e-01
-3.31813633e-01 -2.26835176e-01 -1.02982783e+00 4.09193665e-01
-1.71855044e+00 -1.16018200e+00 -4.78597939e-01 2.05512190e+00
3.72537136e-01 5.54251745e-02 1.73619464e-01 1.81524992e-01
7.07307458e-01 -1.33587625e-02 -4.48497534e-01 -4.42911446e-01
-8.28296989e-02 9.12628993e-02 6.24212742e-01 6.40368879e-01
-1.34563577e+00 1.14522791e+00 6.90612841e+00 4.47815210e-01
-1.19734514e+00 -7.56777227e-02 6.16262674e-01 2.36580223e-01
-1.07250482e-01 1.59480453e-01 -5.54110289e-01 2.24830493e-01
5.89985967e-01 2.27831468e-01 4.91744787e-01 5.68521738e-01
3.75149876e-01 -4.94663268e-01 -1.28983450e+00 8.85717154e-01
2.59731621e-01 -1.15612197e+00 -4.73992229e-02 2.07381994e-01
6.77421093e-01 -1.29757553e-01 2.00163990e-01 -6.57314807e-02
4.32513863e-01 -1.31465197e+00 8.21418703e-01 6.34795725e-01
5.81353530e-02 -2.81480402e-01 4.59497690e-01 4.74522650e-01
-1.01869130e+00 7.33405426e-02 -2.70673782e-01 -3.22723091e-01
3.12983066e-01 2.39364073e-01 -4.78531450e-01 3.75998050e-01
5.72375536e-01 6.25647843e-01 -7.14954674e-01 1.24984038e+00
-1.76773831e-01 2.30301112e-01 -3.04877847e-01 2.45758459e-01
3.39324683e-01 -4.01280344e-01 5.45101166e-01 1.02565753e+00
1.28143951e-01 4.76862490e-03 5.61805926e-02 1.07619941e+00
3.23672682e-01 -2.86639988e-01 -5.57832301e-01 1.47696957e-01
-6.76810294e-02 1.29005325e+00 -1.22542357e+00 -2.90983558e-01
-3.81667644e-01 1.01279545e+00 8.58044252e-02 6.49079204e-01
-7.54648805e-01 6.00003861e-02 5.80606580e-01 1.27704173e-01
5.63296080e-01 -7.04859912e-01 -1.62189871e-01 -1.15993357e+00
-2.01152876e-01 -7.13575304e-01 9.33994204e-02 -8.62566590e-01
-1.00530887e+00 2.38807097e-01 1.13835029e-01 -1.01531637e+00
-2.19197214e-01 -9.25752759e-01 -6.79681957e-01 7.68667519e-01
-1.29368556e+00 -1.07574725e+00 -3.03071678e-01 3.81599665e-01
6.51254356e-01 4.54121292e-01 4.65304315e-01 1.71443239e-01
-3.81855458e-01 -1.01108722e-01 -2.28767574e-01 2.50364363e-01
4.58141148e-01 -1.46733785e+00 5.08389592e-01 1.06281412e+00
5.57348907e-01 6.63446784e-01 9.84763622e-01 -3.79651994e-01
-1.25108886e+00 -7.63597548e-01 4.97636169e-01 -1.04297912e+00
4.41859305e-01 -2.77911544e-01 -1.10611439e+00 5.67580938e-01
1.58399612e-01 1.82429940e-01 3.19477320e-01 -3.55559379e-01
-1.74495578e-01 4.00417209e-01 -9.67219472e-01 5.79069078e-01
9.39766228e-01 -2.87258387e-01 -8.45863342e-01 4.21930850e-02
4.00969535e-01 -4.74500626e-01 -4.40005481e-01 9.51110572e-02
7.68454850e-01 -1.02308810e+00 1.22146761e+00 -6.74848795e-01
2.93931365e-01 -4.23383772e-01 8.84979740e-02 -1.35715306e+00
-3.48441899e-01 -4.89772260e-01 1.56096444e-01 8.34598303e-01
5.51761985e-02 -2.84588456e-01 8.49518418e-01 6.06839836e-01
1.02333032e-01 -3.48219037e-01 -6.62134826e-01 -5.84805846e-01
4.81433153e-01 -5.40345073e-01 6.00048080e-02 1.18266010e+00
-5.49941778e-01 2.30409563e-01 -1.88627884e-01 1.97060883e-01
5.43037415e-01 2.68210284e-03 7.72922933e-01 -1.48208117e+00
-4.59950298e-01 -7.84656405e-01 -6.96271300e-01 -1.14951420e+00
1.21408433e-01 -6.10801280e-01 2.02840775e-01 -1.62312448e+00
-3.52003612e-02 -1.83278997e-03 1.87596753e-01 -3.76295149e-02
-1.72395840e-01 2.39447832e-01 3.81425679e-01 5.75802065e-02
-4.64905441e-01 6.27784804e-03 1.59610629e+00 -7.06751272e-02
-2.98111260e-01 2.85666250e-03 -5.50109684e-01 1.32912660e+00
5.66795826e-01 -3.49418372e-01 -3.43715429e-01 -6.55440092e-01
1.24763846e-01 -4.77027632e-02 8.57997477e-01 -9.97411489e-01
2.72273064e-01 -2.68313229e-01 7.13533700e-01 -4.28466380e-01
3.40440303e-01 -9.46151555e-01 1.14561781e-01 5.93752861e-01
-9.31041539e-02 1.23597912e-01 2.65745163e-01 5.71555197e-01
1.42455974e-03 -1.23417825e-01 1.07215273e+00 -3.40634346e-01
-9.99086380e-01 2.29617774e-01 -7.36297250e-01 1.03295051e-01
9.73847747e-01 -5.17604709e-01 -3.01365137e-01 -4.81676787e-01
-1.07220614e+00 -2.18811762e-02 6.82802856e-01 6.08764231e-01
5.01681685e-01 -9.93175924e-01 -3.56384993e-01 6.23246133e-02
-3.08442771e-01 1.68782681e-01 3.47408116e-01 9.96322274e-01
-9.29639459e-01 3.19397479e-01 -3.82182658e-01 -8.47536325e-01
-1.16421819e+00 5.89926481e-01 6.08575404e-01 1.95504978e-01
-5.52895606e-01 9.16462958e-01 6.33261144e-01 -9.07041728e-02
-2.86887344e-02 -8.09360862e-01 -2.44325906e-01 -1.52761132e-01
3.66140127e-01 3.69688332e-01 -2.68390298e-01 -1.07739758e+00
-4.17644501e-01 1.09853816e+00 1.84623912e-01 -1.24033958e-01
1.10962629e+00 -3.68447930e-01 -8.17289401e-04 3.71127903e-01
1.04227507e+00 -1.91193193e-01 -1.44741118e+00 4.38094437e-02
8.66938755e-02 -4.35185313e-01 1.75065473e-02 -6.62293494e-01
-1.20893812e+00 1.27914202e+00 6.10797346e-01 3.48502368e-01
8.98876131e-01 -2.79378798e-02 5.48518956e-01 1.10832490e-01
7.23711848e-02 -9.42619681e-01 1.12919196e-01 4.47463542e-01
5.31346440e-01 -1.28695965e+00 -2.39230506e-02 -5.11818111e-01
-6.75236166e-01 1.11404383e+00 5.96746981e-01 -4.45094675e-01
6.15810633e-01 1.42807856e-01 1.68123409e-01 -2.17442870e-01
-4.45662796e-01 -3.13054651e-01 6.49422169e-01 7.75822461e-01
3.75178128e-01 -1.26681149e-01 9.88907889e-02 5.38439527e-02
-1.49410382e-01 -9.18170735e-02 5.87185860e-01 8.18628490e-01
-4.60881025e-01 -9.30874467e-01 -5.91052234e-01 -5.23219109e-02
-5.18011570e-01 2.37103656e-01 -5.34007490e-01 1.04011273e+00
4.94330317e-01 7.27173865e-01 2.63678402e-01 -2.79417261e-02
1.05445050e-01 -1.85961965e-02 8.61731470e-01 -5.84071279e-01
-3.24944019e-01 2.70655066e-01 -3.47965062e-01 -6.06047690e-01
-8.63577783e-01 -6.95559680e-01 -1.13813353e+00 -2.51459837e-01
-1.58812970e-01 -2.45677233e-01 7.49172390e-01 1.26527762e+00
-1.20382123e-01 6.14860952e-01 2.73067150e-02 -1.31677651e+00
-1.55090585e-01 -5.83819330e-01 -4.13248479e-01 6.16803825e-01
5.64784169e-01 -7.74717510e-01 -4.57207978e-01 3.79789889e-01]
|
[8.985719680786133, -0.3763781487941742]
|
4b0c7f78-ef4e-4022-9069-f4c30b672c7b
|
open-vocabulary-attribute-detection
|
2211.12914
| null |
https://arxiv.org/abs/2211.12914v2
|
https://arxiv.org/pdf/2211.12914v2.pdf
|
Open-vocabulary Attribute Detection
|
Vision-language modeling has enabled open-vocabulary tasks where predictions can be queried using any text prompt in a zero-shot manner. Existing open-vocabulary tasks focus on object classes, whereas research on object attributes is limited due to the lack of a reliable attribute-focused evaluation benchmark. This paper introduces the Open-Vocabulary Attribute Detection (OVAD) task and the corresponding OVAD benchmark. The objective of the novel task and benchmark is to probe object-level attribute information learned by vision-language models. To this end, we created a clean and densely annotated test set covering 117 attribute classes on the 80 object classes of MS COCO. It includes positive and negative annotations, which enables open-vocabulary evaluation. Overall, the benchmark consists of 1.4 million annotations. For reference, we provide a first baseline method for open-vocabulary attribute detection. Moreover, we demonstrate the benchmark's value by studying the attribute detection performance of several foundation models. Project page https://ovad-benchmark.github.io
|
['Thomas Brox', 'Simon Ging', 'Sudhanshu Mittal', 'María A. Bravo']
|
2022-11-23
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Bravo_Open-Vocabulary_Attribute_Detection_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Bravo_Open-Vocabulary_Attribute_Detection_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['open-vocabulary-object-detection', 'open-vocabulary-attribute-detection']
|
['computer-vision', 'computer-vision']
|
[-2.99056899e-03 4.80144583e-02 -2.98902541e-01 -5.72424948e-01
-1.17256963e+00 -6.16550684e-01 9.66120660e-01 4.78253514e-01
-4.54815924e-01 4.65846032e-01 3.45879614e-01 1.72430843e-01
4.40899581e-01 -5.36980987e-01 -6.51154280e-01 -3.53173852e-01
1.31035671e-01 9.57034707e-01 1.60377808e-02 -6.93668500e-02
-1.10348858e-01 -5.85138761e-02 -2.00905704e+00 4.99003470e-01
4.49963540e-01 1.45423949e+00 2.18734950e-01 4.41663742e-01
-3.68149966e-01 5.03634989e-01 -4.38847333e-01 -6.78706527e-01
1.80893317e-01 2.76592433e-01 -7.67142951e-01 -2.46584326e-01
1.26621950e+00 -3.10792923e-01 -3.11506897e-01 9.78926003e-01
5.55462122e-01 -1.58997864e-01 1.03710532e+00 -1.59295797e+00
-1.16758871e+00 3.48855853e-01 -1.60107613e-01 1.00153886e-01
2.76004612e-01 4.00609314e-01 1.81567252e+00 -1.35855246e+00
8.56656611e-01 1.12982368e+00 5.93707085e-01 7.76084185e-01
-1.36087775e+00 -7.24956810e-01 2.10507572e-01 3.42244804e-01
-1.75061536e+00 -4.77841258e-01 2.75341094e-01 -7.05030024e-01
1.28803432e+00 1.84725493e-01 6.53735816e-01 1.70509088e+00
-1.29230052e-01 1.06522059e+00 8.86166036e-01 -2.90224940e-01
8.38637501e-02 3.19169253e-01 5.46777368e-01 5.60195029e-01
4.08373475e-01 -3.37852188e-03 -7.03787863e-01 -2.29603976e-01
1.72611699e-01 -2.53609627e-01 -8.89224857e-02 -7.55495071e-01
-1.59635842e+00 8.55833769e-01 1.96821511e-01 -2.17342198e-01
-1.45107731e-01 9.36832875e-02 7.85072923e-01 4.25183713e-01
8.58469844e-01 5.80577135e-01 -6.94219470e-01 -2.16997843e-02
-3.68137300e-01 3.77809554e-01 1.02753806e+00 1.41513801e+00
7.95756876e-01 -2.28710547e-01 -6.48344457e-01 1.11703324e+00
4.39370900e-01 7.79068589e-01 1.08873792e-01 -6.98022962e-01
3.41402858e-01 5.96767366e-01 -1.08034939e-01 -4.67190742e-01
-2.15244442e-01 -1.84032947e-01 -4.34458673e-01 8.72048885e-02
3.49016041e-01 1.89826995e-01 -1.02395988e+00 1.56834328e+00
4.98134673e-01 5.44871166e-02 1.64200515e-01 7.93931842e-01
1.62979329e+00 4.02460128e-01 6.44720197e-01 2.14358404e-01
1.70382524e+00 -9.22269225e-01 -7.43655264e-01 -3.47777784e-01
8.98963273e-01 -8.29925537e-01 1.58948874e+00 3.73516120e-02
-3.53728443e-01 -4.73439753e-01 -1.03420973e+00 -3.79722893e-01
-7.91469634e-01 -1.14469841e-01 6.43502891e-01 2.88045108e-01
-8.19013298e-01 -2.30703518e-01 -2.03408271e-01 -5.47852039e-01
7.60344267e-01 -6.07412532e-02 -6.24348879e-01 -3.59707735e-02
-1.26566875e+00 9.87888753e-01 3.26437145e-01 -6.60610259e-01
-1.39773917e+00 -1.19534290e+00 -1.05289602e+00 -1.31175756e-01
5.64054847e-01 -7.96523690e-01 1.57961857e+00 -8.18089724e-01
-5.65723836e-01 1.70035756e+00 5.87523989e-02 -4.98201668e-01
3.55927974e-01 -4.10778075e-01 -3.72012615e-01 -1.75860263e-02
4.82017875e-01 1.13358676e+00 8.57054055e-01 -9.50973928e-01
-7.33061254e-01 -5.52009344e-01 1.65169731e-01 1.23081855e-01
-5.99862397e-01 1.09888976e-02 -5.08744359e-01 -8.36235642e-01
-4.80537117e-01 -9.31150734e-01 1.00122377e-01 2.37621069e-01
-4.46054101e-01 -8.05627882e-01 5.01643717e-01 -2.29310766e-01
9.20860231e-01 -2.36431718e+00 -3.53836715e-01 -5.07425666e-01
5.54489076e-01 7.15671405e-02 -4.05580223e-01 1.73224598e-01
6.28862903e-02 4.72862162e-02 9.24032032e-02 -4.97364342e-01
3.35920781e-01 1.38779014e-01 -5.06489515e-01 2.60335326e-01
3.16343248e-01 1.21072018e+00 -7.35013068e-01 -8.65945101e-01
6.53294027e-02 3.49466175e-01 -3.89760971e-01 2.14653119e-01
-7.59644091e-01 -6.31998405e-02 -5.59707880e-01 1.16792810e+00
2.75413334e-01 -3.28700066e-01 -5.42270243e-01 -4.05874282e-01
5.92822954e-02 1.28948033e-01 -7.53787100e-01 1.85810387e+00
-3.46059799e-01 9.92071390e-01 -1.24485172e-01 -5.00269651e-01
1.02874875e+00 3.22282434e-01 3.87573183e-01 -6.40830696e-01
3.44818756e-02 1.01728402e-01 -5.17729640e-01 -4.05168980e-01
5.61768532e-01 9.77470577e-02 -4.74306554e-01 2.40727201e-01
5.15283465e-01 -2.23253101e-01 3.57665569e-02 4.53195125e-01
9.57978547e-01 1.15528747e-01 4.20576453e-01 -3.60627115e-01
3.66406500e-01 4.14496571e-01 2.24569932e-01 1.00651264e+00
-5.33317864e-01 6.18614316e-01 2.37856001e-01 -8.33194435e-01
-1.30334258e+00 -1.03538823e+00 -5.58504820e-01 1.67226434e+00
8.98225829e-02 -7.50153244e-01 -5.31383216e-01 -7.48680115e-01
4.11198884e-01 8.51637185e-01 -9.84880447e-01 -1.25130087e-01
2.71776468e-01 -4.45297390e-01 7.60851026e-01 5.45786858e-01
2.01164320e-01 -1.00417149e+00 -2.23089546e-01 -1.32893890e-01
-4.54389632e-01 -1.45756912e+00 -5.78955531e-01 1.82571501e-01
-2.64593810e-01 -1.02936244e+00 -4.18849736e-01 -9.74717200e-01
7.78388306e-02 2.58907944e-01 1.86851382e+00 -2.97843516e-01
-6.86345160e-01 1.03191841e+00 -3.43634069e-01 -1.26577628e+00
-3.56435120e-01 9.68132913e-02 2.06046656e-01 9.96151865e-02
1.36171842e+00 2.91497000e-02 -3.43546659e-01 2.59944588e-01
-2.84179360e-01 -6.44603819e-02 5.02780378e-01 6.98248208e-01
1.12964261e+00 -1.18029535e+00 5.81036866e-01 -5.96211791e-01
4.50151384e-01 -4.87113953e-01 -6.35096014e-01 2.77996004e-01
-7.84993708e-01 2.84715705e-02 -1.72998875e-01 -6.14895940e-01
-8.84183586e-01 2.82373428e-01 2.99916547e-02 -6.41500592e-01
-5.36082149e-01 1.49086565e-01 -3.01466495e-01 1.13851972e-01
7.64980733e-01 1.21663809e-01 8.56890678e-02 -5.57236910e-01
7.27091551e-01 9.31795537e-01 5.50324559e-01 -4.92189467e-01
5.89311898e-01 4.76310581e-01 -2.79245317e-01 -9.97669756e-01
-1.48361504e+00 -8.01265776e-01 -6.26198769e-01 -1.26088768e-01
1.25925505e+00 -1.62573040e+00 -6.46146834e-01 3.87306213e-01
-1.09104586e+00 -2.42815435e-01 -5.14737248e-01 3.33541811e-01
-7.65824854e-01 -3.79354693e-02 -2.87705243e-01 -4.01968509e-01
-5.79265654e-01 -8.07047963e-01 1.41788340e+00 -2.45221362e-01
-4.86588776e-01 -6.59808815e-01 2.05736145e-01 6.43926919e-01
1.25542939e-01 -6.15523793e-02 6.51650012e-01 -1.45977199e+00
-5.56853712e-01 -3.38168204e-01 -3.31796318e-01 3.22853893e-01
-3.27123821e-01 -1.89228565e-01 -1.52415907e+00 4.60966974e-02
-4.16155010e-01 -1.07005453e+00 1.07271314e+00 2.68593341e-01
1.10503483e+00 5.63156791e-02 -4.18169379e-01 7.13730097e-01
1.18142843e+00 -4.97439981e-01 2.24993959e-01 6.52010381e-01
8.32402766e-01 4.95198250e-01 1.23532856e+00 4.81890738e-01
6.25894666e-01 7.27645576e-01 6.94899201e-01 5.97028583e-02
-4.34920162e-01 -1.87174112e-01 3.95825272e-03 2.66121745e-01
7.64812753e-02 -1.42281622e-01 -1.13114917e+00 7.81682134e-01
-1.56900048e+00 -8.37254524e-01 -3.77739102e-01 2.07593560e+00
9.92990255e-01 6.59433082e-02 -1.43403515e-01 -4.94811028e-01
6.02068186e-01 2.52277404e-01 -5.28965950e-01 -7.83529282e-02
-2.93778211e-01 -1.61141261e-01 3.09827656e-01 1.96105108e-01
-1.73465586e+00 1.01847434e+00 5.99278259e+00 9.96686518e-01
-5.06528616e-01 3.59405518e-01 4.88591641e-01 -4.94872451e-01
-2.91647762e-01 -2.16968685e-01 -1.40095699e+00 1.25294507e-01
7.95056880e-01 -1.73390359e-01 -2.04315886e-01 1.26076698e+00
-4.59848464e-01 1.01339273e-01 -1.64720905e+00 1.26250076e+00
3.83373797e-01 -1.25005853e+00 3.95314395e-01 2.27821879e-02
3.90238971e-01 5.68239152e-01 1.22013092e-02 9.30265546e-01
1.73723713e-01 -1.00219107e+00 7.05623329e-01 3.82234514e-01
1.24476504e+00 -3.70005786e-01 5.67460477e-01 9.08952113e-03
-1.12393463e+00 4.05045971e-02 -4.83786792e-01 3.62957902e-02
-2.11387083e-01 2.76047051e-01 -9.95076835e-01 -1.46971136e-01
1.05701649e+00 7.63921440e-01 -7.94138670e-01 8.67084205e-01
-8.84004012e-02 5.56273222e-01 -1.51337817e-01 -2.87199587e-01
2.78049141e-01 1.03016198e-03 7.71436930e-01 1.34028983e+00
-2.54680365e-01 -1.55534791e-02 3.02120388e-01 7.80184209e-01
-3.32124650e-01 3.09742928e-01 -1.06153333e+00 -3.97908650e-02
6.02974594e-01 1.34667969e+00 -1.02353450e-02 -4.37475324e-01
-1.08836222e+00 7.04489470e-01 4.87540454e-01 1.44581825e-01
-8.34431052e-01 -2.18236119e-01 1.38006294e+00 -4.41226028e-02
3.56996387e-01 1.10440962e-01 -2.41800249e-01 -1.18243611e+00
-5.61759062e-02 -8.58407140e-01 7.29601145e-01 -7.44033098e-01
-1.74467921e+00 3.44100535e-01 -9.45311114e-02 -1.22099125e+00
5.43919615e-02 -6.29215956e-01 3.77468728e-02 7.59439766e-01
-1.49872017e+00 -1.66119599e+00 -7.96682239e-01 7.99982131e-01
1.09710002e+00 -5.49572825e-01 1.23906267e+00 4.09818977e-01
-1.71443090e-01 7.67661691e-01 -7.76342228e-02 3.35388333e-01
1.42181635e+00 -1.32870889e+00 6.51541710e-01 1.59186438e-01
5.06722748e-01 2.06699386e-01 8.11092257e-01 -7.02525258e-01
-1.32504034e+00 -1.32740259e+00 9.96172249e-01 -1.35445213e+00
1.07684946e+00 -7.87891686e-01 -1.01962101e+00 9.59249914e-01
1.42954722e-01 6.14643455e-01 8.87725115e-01 4.62934792e-01
-9.17454362e-01 -9.59428400e-02 -6.59619272e-01 3.57241839e-01
1.22924185e+00 -7.65940666e-01 -8.14416289e-01 5.39302766e-01
1.19938004e+00 -1.19027779e-01 -1.05309260e+00 4.86171424e-01
6.35598600e-01 -3.12326163e-01 1.37376428e+00 -1.06772780e+00
4.41689998e-01 6.04856610e-02 -5.76889753e-01 -1.06029105e+00
-3.06445867e-01 6.53503835e-02 -3.04961652e-01 1.54865789e+00
7.24178612e-01 -3.76565576e-01 3.44729543e-01 6.25683129e-01
-2.85665244e-01 -5.87370038e-01 -9.16211665e-01 -8.83804202e-01
2.09790301e-02 -6.52664185e-01 4.41612303e-01 1.00802529e+00
-4.13513392e-01 8.39125991e-01 -1.92919806e-01 2.15942552e-03
9.30475354e-01 1.63635030e-01 9.90904868e-01 -1.60602367e+00
4.89573069e-02 -2.29847208e-01 -7.11847901e-01 -7.63899922e-01
3.79866749e-01 -1.05956542e+00 -2.82835625e-02 -1.31014752e+00
5.88944674e-01 -5.84787726e-01 -2.32501656e-01 6.81604445e-01
-2.74671853e-01 5.99791110e-01 1.46408021e-01 3.45124096e-01
-1.09948969e+00 8.30049515e-01 9.24068689e-01 -7.30154872e-01
2.97721058e-01 -1.78130448e-01 -6.61747396e-01 6.49290085e-01
6.79572046e-01 -4.53542858e-01 -3.55835646e-01 -4.47833955e-01
1.92135081e-01 -5.93465269e-01 6.44785762e-01 -7.21495032e-01
-4.81004156e-02 7.72056952e-02 2.27424994e-01 -7.03555465e-01
5.69337547e-01 -6.38680339e-01 -3.56465131e-01 2.73918994e-02
-6.33593321e-01 -3.30660492e-01 1.94468588e-01 9.61198092e-01
-2.22677454e-01 -6.17814474e-02 5.41521370e-01 -6.97980598e-02
-1.38315415e+00 6.60231650e-01 -1.50574431e-01 7.44648874e-01
1.34392917e+00 -1.04037173e-01 -5.26218295e-01 -1.88015208e-01
-7.99187422e-01 5.64811707e-01 4.42912519e-01 9.89513695e-01
5.79963386e-01 -1.45942187e+00 -1.06561375e+00 -9.00786817e-02
1.37179863e+00 -9.91890728e-02 -1.16843775e-01 5.96821129e-01
-1.01124354e-01 5.41707039e-01 -2.44511276e-01 -1.00378001e+00
-1.60668588e+00 1.01839328e+00 1.10212520e-01 1.78856403e-01
-7.00337946e-01 1.03528929e+00 6.39207304e-01 -5.12452483e-01
6.22008860e-01 1.08393639e-01 -4.18704778e-01 4.76807922e-01
7.57242501e-01 -2.91312728e-02 1.97464041e-02 -6.28905237e-01
-3.97379220e-01 3.64635676e-01 -2.00303182e-01 1.99803755e-01
1.18913639e+00 -2.31011540e-01 6.87918812e-02 7.37179637e-01
1.09387624e+00 -4.06783581e-01 -1.07092690e+00 -5.84869683e-01
-1.86000858e-02 -3.85285228e-01 1.38477497e-02 -7.20463216e-01
-5.46086371e-01 8.81248057e-01 8.71255159e-01 4.12124246e-02
6.09640718e-01 7.36460745e-01 3.81142467e-01 6.52951896e-01
3.89611840e-01 -1.21712422e+00 1.72958538e-01 7.55500078e-01
9.64745760e-01 -1.95769167e+00 -2.35053360e-01 -3.86647373e-01
-1.02648926e+00 7.33362079e-01 1.09895217e+00 4.06415522e-01
5.97487986e-01 1.53025746e-01 3.00827682e-01 -4.93166775e-01
-1.07777452e+00 -5.53476214e-01 4.75710422e-01 8.73358428e-01
4.14847106e-01 2.49535233e-01 5.10648265e-02 7.59091437e-01
3.48215476e-02 -3.73919278e-01 2.49542713e-01 4.88063067e-01
-5.50678670e-01 -6.30440652e-01 -1.49344862e-01 7.23295152e-01
-3.27190876e-01 -5.82279444e-01 -6.18692100e-01 7.47677207e-01
-6.22999333e-02 8.20789039e-01 2.68978447e-01 -1.59693941e-01
3.29796344e-01 4.54285413e-01 1.82411432e-01 -8.10035408e-01
-2.45442897e-01 -4.18356538e-01 6.53707623e-01 -8.12457681e-01
-1.57016858e-01 -7.23953485e-01 -7.57340372e-01 9.60859954e-02
-2.36981049e-01 -1.72673196e-01 6.20317519e-01 6.43234372e-01
5.06916285e-01 2.26056173e-01 -7.99818430e-03 -4.09426481e-01
-7.54945695e-01 -1.08466029e+00 -3.37387443e-01 6.71549559e-01
4.29379523e-01 -9.96224940e-01 -2.09941238e-01 1.27587378e-01]
|
[9.929677963256836, 1.7223104238510132]
|
738e661a-a650-40ad-9183-4653e11fe9a7
|
natural-evolution-strategies-and-quantum
|
2005.04447
| null |
https://arxiv.org/abs/2005.04447v2
|
https://arxiv.org/pdf/2005.04447v2.pdf
|
Natural evolution strategies and variational Monte Carlo
|
A notion of quantum natural evolution strategies is introduced, which provides a geometric synthesis of a number of known quantum/classical algorithms for performing classical black-box optimization. Recent work of Gomes et al. [2019] on heuristic combinatorial optimization using neural quantum states is pedagogically reviewed in this context, emphasizing the connection with natural evolution strategies. The algorithmic framework is illustrated for approximate combinatorial optimization problems, and a systematic strategy is found for improving the approximation ratios. In particular it is found that natural evolution strategies can achieve approximation ratios competitive with widely used heuristic algorithms for Max-Cut, at the expense of increased computation time.
|
['Shravan Veerapaneni', 'Tianchen Zhao', 'James Stokes', 'Giuseppe Carleo']
|
2020-05-09
| null | null | null | null |
['variational-monte-carlo']
|
['miscellaneous']
|
[ 1.56831771e-01 3.22186351e-01 -3.73981625e-01 -1.30446389e-01
-5.72421432e-01 -4.58052725e-01 4.05819148e-01 1.95044369e-01
-5.75415790e-01 1.16646707e+00 -1.91437036e-01 -1.02708630e-01
-4.55280036e-01 -1.18637896e+00 -5.48772454e-01 -9.35523331e-01
1.15992486e-01 6.69762731e-01 -2.74223089e-01 -8.25413465e-01
1.02761495e+00 6.71688855e-01 -1.80296218e+00 -3.65833491e-01
1.08015382e+00 7.52639115e-01 -2.95524985e-01 1.03781569e+00
-4.38386530e-01 3.67981821e-01 -6.18865967e-01 -1.03643501e+00
4.59737837e-01 -1.01467609e+00 -1.24136031e+00 -3.09201241e-01
2.89991200e-01 1.40764713e-01 -7.87349164e-01 1.68173623e+00
5.99549174e-01 7.09191024e-01 6.22703850e-01 -9.32789266e-01
-8.23418617e-01 7.29891598e-01 1.30371839e-01 2.86101490e-01
3.61803919e-01 1.70281738e-01 1.35113668e+00 -3.84960651e-01
7.99344838e-01 1.01530039e+00 4.62161273e-01 9.96696353e-01
-1.29388523e+00 -8.67909268e-02 -7.15681911e-01 9.31245685e-01
-1.76934981e+00 -1.43249750e-01 5.23393393e-01 4.10264760e-01
1.41281617e+00 6.00783110e-01 1.34876180e+00 3.54178518e-01
5.69084287e-01 7.89357543e-01 9.50073719e-01 -9.04801011e-01
8.31713617e-01 2.31069718e-02 3.22437197e-01 1.12501335e+00
1.88627630e-03 7.49185681e-01 -8.82967889e-01 -8.88735801e-02
3.54308277e-01 -5.95889211e-01 4.11763452e-02 -5.37261248e-01
-6.53115213e-01 1.20242894e+00 4.47106868e-01 2.56244659e-01
-2.67652392e-01 3.98883700e-01 3.93158525e-01 5.46712399e-01
-2.61021983e-02 1.26764143e+00 -1.17368802e-01 -5.65369248e-01
-9.70038354e-01 5.50094724e-01 1.06462586e+00 1.09610260e+00
9.96390581e-01 3.33805442e-01 1.18633267e-03 3.61464113e-01
-1.13441542e-01 4.17684913e-01 1.78101793e-01 -1.49178922e+00
-2.82413870e-01 5.39319701e-02 -2.03575343e-02 -3.98650765e-01
-4.48373556e-01 -1.57974944e-01 -5.13456702e-01 2.45179355e-01
1.44693583e-01 -4.77677286e-02 -7.59720325e-01 1.37233388e+00
2.90427119e-01 -1.30417079e-01 1.37508482e-01 5.57410896e-01
4.67407465e-01 9.15174127e-01 -4.05179083e-01 -7.62459457e-01
8.27979863e-01 -1.19995677e+00 -1.21080029e+00 4.56885427e-01
6.42587781e-01 -8.03067982e-01 5.25199771e-01 5.99398017e-01
-1.56118751e+00 -2.20752761e-01 -1.15451491e+00 -3.17095071e-02
-7.47911572e-01 -4.95762646e-01 1.18829668e+00 1.47479117e+00
-1.19887424e+00 1.35642767e+00 -6.29989386e-01 -3.18901241e-01
1.24696299e-01 5.79328537e-01 1.31107703e-01 2.68304795e-01
-1.26817381e+00 1.46783471e+00 7.47567236e-01 1.89829677e-01
-3.65550607e-01 -4.58539367e-01 -4.60924059e-01 5.25206625e-02
4.26855177e-01 -7.62711883e-01 1.45026982e+00 -6.38488531e-01
-2.58184576e+00 5.99329233e-01 -1.49773806e-01 -5.67823172e-01
9.06479359e-02 3.30662221e-01 -1.87792107e-01 3.61972451e-01
-3.71766537e-01 5.71544409e-01 4.72774446e-01 -3.16535860e-01
-4.89243805e-01 -1.34194076e-01 3.17453235e-01 5.66944361e-01
-2.15207592e-01 -1.73995510e-01 -6.18655384e-02 -2.25390568e-01
2.82373548e-01 -9.77361083e-01 -8.31229568e-01 -4.90745604e-01
-8.73663202e-02 -4.73596632e-01 -1.44033074e-01 -1.28692433e-01
1.49345207e+00 -1.30601192e+00 8.92806172e-01 5.16786337e-01
-4.66302037e-02 4.22721393e-02 8.40732083e-02 6.72152102e-01
4.11346890e-02 1.00708850e-01 -3.16203952e-01 2.33173519e-01
3.62687677e-01 3.55840445e-01 -1.42425358e-01 4.25278038e-01
-2.11322725e-01 1.32298505e+00 -9.35531795e-01 -5.32893658e-01
2.09468946e-01 2.94179918e-05 -8.93735170e-01 -2.48138577e-01
-2.77984679e-01 -9.42994654e-02 -2.47387499e-01 6.75813973e-01
3.75668466e-01 2.50586987e-01 -3.96930538e-02 1.35245278e-01
-4.42437947e-01 3.43280852e-01 -1.26136112e+00 1.93302083e+00
-3.60337347e-01 6.62587523e-01 -9.73462164e-02 -1.14887595e+00
4.58090872e-01 1.28000095e-01 4.12182182e-01 -7.80015111e-01
3.63547921e-01 4.08758312e-01 7.74436817e-02 -1.70252562e-01
1.08172703e+00 -3.20999622e-01 -2.02989936e-01 4.74483341e-01
5.09955287e-01 -1.16141772e+00 8.65630209e-01 1.85513273e-02
8.00222397e-01 2.27548510e-01 5.17160058e-01 -5.67480028e-01
7.48200297e-01 5.17431021e-01 2.44020954e-01 1.23108590e+00
-4.87325132e-01 2.22579166e-01 2.39769399e-01 -7.50599146e-01
-1.15229917e+00 -1.15260375e+00 -3.94582540e-01 1.07554960e+00
3.64900023e-01 -8.85598123e-01 -9.77383316e-01 1.55653745e-01
-5.41424394e-01 1.19697249e+00 -7.08683968e-01 -3.95424843e-01
-5.52657783e-01 -1.00407600e+00 4.83438015e-01 -1.63511395e-01
3.08727801e-01 -1.21400225e+00 -7.51048684e-01 3.48385125e-01
5.68631664e-02 -5.90119898e-01 9.51533485e-03 5.62398553e-01
-1.16591501e+00 -7.52364635e-01 -3.46181661e-01 -4.38041389e-01
1.07028887e-01 -1.02113344e-01 8.96155417e-01 1.20227464e-01
-6.90030873e-01 3.38669747e-01 -2.38898441e-01 -1.54698789e-01
-6.49592578e-01 8.81059617e-02 1.73989832e-01 -7.36959815e-01
5.63172042e-01 -4.11148250e-01 -8.28737915e-02 -3.56345415e-01
-7.71675169e-01 -1.47331759e-01 3.89092684e-01 1.35602295e+00
7.47069120e-01 3.74891996e-01 -7.36875683e-02 -3.49705577e-01
5.17921865e-01 2.09564358e-01 -9.30958271e-01 4.17210072e-01
-9.39921379e-01 7.34370828e-01 6.66846216e-01 9.72348005e-02
-8.05000544e-01 -2.23847166e-01 -3.97306919e-01 6.93214238e-02
6.48309290e-02 2.82850206e-01 2.15658382e-01 -1.04076290e+00
8.64844620e-01 5.95314085e-01 -2.03365549e-01 1.67000756e-01
7.37688363e-01 1.83205351e-01 3.03081453e-01 -7.40955174e-01
6.70568824e-01 5.77423386e-02 7.38178194e-01 -1.30070388e+00
-8.07902217e-01 5.74140251e-02 -5.35745800e-01 -1.13245755e-01
1.04040968e+00 7.35884979e-02 -9.76702571e-01 2.18286425e-01
-9.09261167e-01 1.83454547e-02 -7.44061589e-01 5.37449718e-01
-1.38758719e+00 4.09163833e-01 -7.74687231e-01 -1.06054604e+00
-3.25228333e-01 -1.46210325e+00 4.12622184e-01 7.79345870e-01
8.58257189e-02 -9.77862179e-01 3.21088016e-01 2.31323540e-01
4.15139109e-01 -2.86925226e-01 9.06745493e-01 -3.40662062e-01
-6.99464738e-01 -2.39395201e-01 3.37710530e-01 -4.75542620e-02
-5.66744745e-01 3.93099666e-01 -5.87344468e-01 2.35753581e-01
9.84573141e-02 -2.19178185e-01 5.12573600e-01 4.47601587e-01
9.90568161e-01 -6.47704676e-02 1.51079819e-01 8.21523070e-01
1.68774068e+00 3.98229629e-01 5.98045409e-01 4.70758170e-01
2.19788447e-01 2.58526236e-01 3.68932664e-01 1.65479630e-01
-3.04987967e-01 4.63461071e-01 2.73188710e-01 7.85584331e-01
3.05484623e-01 2.20700353e-01 1.15702882e-01 1.15128613e+00
-4.32845145e-01 1.83595806e-01 -6.82910383e-01 1.18547469e-01
-1.63344467e+00 -1.36811113e+00 1.95821282e-03 1.90763164e+00
8.35033000e-01 1.12861842e-01 7.91726708e-02 1.36425197e-01
7.70399392e-01 -1.58422336e-01 -4.31404531e-01 -1.44559932e+00
-3.02580535e-01 1.16343081e+00 6.92324638e-01 7.94196784e-01
-7.66470134e-01 1.36876476e+00 7.88184786e+00 1.42297328e+00
-4.38725501e-01 -7.82289430e-02 3.19574833e-01 -5.38946509e-01
-3.15399766e-01 6.06729835e-02 -5.63557088e-01 9.38440785e-02
1.49707639e+00 -8.20209324e-01 1.41228247e+00 6.18647814e-01
-3.27406734e-01 -1.56466499e-01 -7.64165640e-01 1.04653919e+00
-1.20131537e-01 -1.90335298e+00 4.80656177e-02 -9.88051966e-02
1.32005954e+00 -1.19704761e-01 -7.72029087e-02 5.03857791e-01
1.08204514e-01 -1.08681238e+00 6.56563997e-01 4.53515381e-01
7.40553960e-02 -1.69841647e+00 6.60859764e-01 1.01173826e-01
-8.39432955e-01 -1.66136369e-01 -5.21928608e-01 -2.69287586e-01
1.80994362e-01 -1.48765624e-01 -5.20845652e-02 7.07507789e-01
4.20078993e-01 1.03344873e-01 -2.56957054e-01 1.38836014e+00
-2.83314079e-01 2.13729382e-01 -5.79090238e-01 -1.11161447e+00
7.51875341e-01 -1.06933868e+00 7.12066889e-01 8.05767894e-01
1.41648740e-01 5.46570957e-01 -3.75384897e-01 1.09839451e+00
1.61614999e-01 4.62196350e-01 -5.01416445e-01 -5.18201828e-01
2.63124257e-01 1.08974624e+00 -9.09048975e-01 -2.91345805e-01
-3.21828848e-04 9.87636745e-01 9.64462757e-02 1.21793993e-01
-9.38546658e-01 -1.13355589e+00 4.58247006e-01 -8.48857939e-01
1.73843712e-01 -2.09153876e-01 -7.52181530e-01 -9.97604370e-01
-7.44582474e-01 -7.14840472e-01 1.25378892e-01 -4.13817436e-01
-8.37002456e-01 3.28646868e-01 1.58952328e-03 -5.46479225e-01
-2.92133629e-01 -7.73358762e-01 -5.16019583e-01 7.04707563e-01
-7.24535584e-01 -1.27432406e-01 4.26074475e-01 3.40534374e-02
3.75408024e-01 -2.81263292e-01 1.00307858e+00 -1.49071410e-01
-8.18975508e-01 6.44617140e-01 5.99757195e-01 -6.82942927e-01
-2.14987412e-01 -1.47786951e+00 2.93896377e-01 8.51298451e-01
5.11948764e-01 7.10742295e-01 1.24463165e+00 -1.72196716e-01
-2.09664941e+00 -9.02043581e-02 7.80988693e-01 -1.08993925e-01
9.74042952e-01 9.03654024e-02 -4.80803698e-01 -7.37764314e-02
5.70341527e-01 -7.05776751e-01 7.30964303e-01 1.69815093e-01
3.94182831e-01 2.79984772e-01 -1.20099878e+00 1.18191016e+00
1.10989189e+00 -6.37124240e-01 -6.37186229e-01 6.91273928e-01
4.62738395e-01 -6.51912451e-01 -7.49681056e-01 -6.73171505e-02
5.86885095e-01 -1.12129605e+00 9.52159405e-01 -6.69873893e-01
2.66803682e-01 1.08227164e-01 -1.41934648e-01 -1.14861894e+00
-2.06039310e-01 -1.60650289e+00 -3.88645023e-01 1.73952788e-01
1.82935178e-01 -4.96872365e-01 1.04642475e+00 6.04099154e-01
-1.44754142e-01 -1.01868832e+00 -1.46006203e+00 -9.62704122e-01
6.61382616e-01 -3.90136719e-01 3.74747366e-01 5.38562894e-01
5.09026229e-01 2.75340915e-01 5.59733948e-03 -3.29667479e-01
7.17246592e-01 3.43711525e-01 2.65216857e-01 -5.97076297e-01
-2.38981053e-01 -1.27017117e+00 -6.21346295e-01 -7.35623837e-01
3.56719732e-01 -1.02419496e+00 -1.24723772e-02 -1.15073180e+00
-4.57235947e-02 2.57222921e-01 -2.10208759e-01 -2.12408110e-01
4.56346273e-02 3.38699043e-01 5.99395409e-02 -4.36204225e-01
-6.06210768e-01 1.01483357e+00 1.25171769e+00 2.84050945e-02
-3.54610562e-01 -8.22694674e-02 -2.52786249e-01 5.60669720e-01
1.01235223e+00 -5.30896664e-01 -7.05986619e-02 1.37445316e-01
7.46970773e-01 1.32563248e-01 -2.31969804e-01 -1.17404175e+00
3.05866987e-01 -4.74619925e-01 -1.98234648e-01 -5.86882353e-01
5.15712023e-01 -2.52399117e-01 -6.93687126e-02 9.40832078e-01
-4.07881260e-01 3.86248410e-01 7.81886727e-02 3.03564310e-01
-6.07555881e-02 -1.29924047e+00 1.06160176e+00 -3.17949682e-01
-8.00986588e-01 3.21961492e-02 -7.01597095e-01 9.90561619e-02
9.83822584e-01 -3.80627930e-01 4.15283209e-03 -2.37954110e-01
-7.94259846e-01 -4.40125838e-02 3.11207324e-01 -2.16142982e-01
3.63777757e-01 -1.23238063e+00 -1.81085780e-01 -1.61281586e-01
-2.28201330e-01 -6.74654186e-01 2.02756375e-01 8.05397034e-01
-1.34752035e+00 7.44415283e-01 -5.88847280e-01 -5.76007739e-02
-8.43672514e-01 7.00675666e-01 5.97714961e-01 9.54163820e-03
-2.53508896e-01 1.24244773e+00 -8.12359035e-01 -4.54911500e-01
1.97004646e-01 5.60735129e-02 3.50930095e-01 -5.43351881e-02
4.16379064e-01 9.96908426e-01 -6.71596453e-02 -3.74470860e-01
-6.05702139e-02 5.61537862e-01 3.56196374e-01 -5.12744844e-01
1.22567654e+00 -1.83804017e-02 -6.15659952e-01 4.03025895e-01
1.20285428e+00 -2.49445811e-01 -3.42600524e-01 -8.45561828e-03
-5.37171103e-02 4.38065007e-02 3.46512824e-01 -4.25307304e-01
-6.91688061e-01 9.23794746e-01 5.36262333e-01 3.13674450e-01
1.10446203e+00 -5.29640973e-01 8.33351314e-01 1.41177809e+00
7.56351292e-01 -1.75085437e+00 -2.56522775e-01 9.10978854e-01
3.30210567e-01 -8.24701846e-01 3.49165291e-01 3.72369327e-02
-2.46896088e-01 1.87125921e+00 3.44729334e-01 -3.85370195e-01
2.69170851e-01 -1.08446993e-01 -5.38223922e-01 -2.01503813e-01
-6.28948510e-01 -5.17063737e-01 2.66714007e-01 1.28905430e-01
9.18641761e-02 5.60602918e-02 -9.03663576e-01 -1.00332595e-01
-7.95958281e-01 -3.93753976e-01 8.93141747e-01 1.12419188e+00
-6.92001998e-01 -1.21323287e+00 -2.67977446e-01 1.69303834e-01
-3.50615680e-01 -4.07098860e-01 -4.59832579e-01 6.64095819e-01
-1.09875455e-01 6.07063949e-01 -2.23600060e-01 -2.87842274e-01
-8.82928520e-02 4.15656090e-01 1.44674981e+00 -3.43773574e-01
-1.02089977e+00 -3.94058973e-01 4.48658392e-02 -8.30954373e-01
-2.24315211e-01 -7.07457900e-01 -1.40252006e+00 -7.61762202e-01
-5.79358757e-01 7.66364515e-01 7.88921654e-01 9.60005224e-01
-1.22521818e-01 3.34398448e-01 3.46698314e-01 -8.94096196e-01
-9.09591556e-01 -3.48507494e-01 -4.77402091e-01 -3.42406750e-01
-6.98637171e-03 -6.42688870e-01 -3.63913625e-01 -5.95906496e-01]
|
[5.575254917144775, 4.935143947601318]
|
fab2f8cf-929c-4b1c-bf5d-4ea4f3512106
|
merge-double-thompson-sampling-for-large
|
1812.04412
| null |
https://arxiv.org/abs/1812.04412v2
|
https://arxiv.org/pdf/1812.04412v2.pdf
|
MergeDTS: A Method for Effective Large-Scale Online Ranker Evaluation
|
Online ranker evaluation is one of the key challenges in information retrieval. While the preferences of rankers can be inferred by interleaving methods, the problem of how to effectively choose the ranker pair that generates the interleaved list without degrading the user experience too much is still challenging. On the one hand, if two rankers have not been compared enough, the inferred preference can be noisy and inaccurate. On the other, if two rankers are compared too many times, the interleaving process inevitably hurts the user experience too much. This dilemma is known as the exploration versus exploitation tradeoff. It is captured by the $K$-armed dueling bandit problem, which is a variant of the $K$-armed bandit problem, where the feedback comes in the form of pairwise preferences. Today's deployed search systems can evaluate a large number of rankers concurrently, and scaling effectively in the presence of numerous rankers is a critical aspect of $K$-armed dueling bandit problems. In this paper, we focus on solving the large-scale online ranker evaluation problem under the so-called Condorcet assumption, where there exists an optimal ranker that is preferred to all other rankers. We propose Merge Double Thompson Sampling (MergeDTS), which first utilizes a divide-and-conquer strategy that localizes the comparisons carried out by the algorithm to small batches of rankers, and then employs Thompson Sampling (TS) to reduce the comparisons between suboptimal rankers inside these small batches. The effectiveness (regret) and efficiency (time complexity) of MergeDTS are extensively evaluated using examples from the domain of online evaluation for web search. Our main finding is that for large-scale Condorcet ranker evaluation problems, MergeDTS outperforms the state-of-the-art dueling bandit algorithms.
|
['Masrour Zoghi', 'Ilya Markov', 'Maarten de Rijke', 'Chang Li']
|
2018-12-11
| null | null | null | null |
['online-ranker-evaluation']
|
['miscellaneous']
|
[ 1.27869755e-01 -3.49225521e-01 -6.39377832e-01 -2.83108801e-01
-1.55842209e+00 -1.11081636e+00 5.61558567e-02 1.18955508e-01
-5.59383392e-01 7.65520632e-01 5.89539558e-02 -5.78540921e-01
-1.12184656e+00 -4.43617582e-01 -6.04244530e-01 -8.13506007e-01
-4.69993800e-01 1.14879668e+00 -8.95596668e-02 -2.26357341e-01
3.55174571e-01 1.33603126e-01 -1.59651411e+00 3.30674559e-01
1.11185336e+00 1.36465549e+00 9.28864852e-02 6.70524061e-01
-1.21069714e-01 3.51674676e-01 -5.20624816e-01 -4.44476187e-01
7.71787882e-01 -4.89259571e-01 -8.66652966e-01 -2.41007403e-01
4.21974093e-01 -6.45477414e-01 -6.98118657e-02 1.21697068e+00
4.13523287e-01 3.97024900e-01 1.85779616e-01 -1.03713882e+00
-4.29746173e-02 1.02434206e+00 -7.47080982e-01 9.11767334e-02
4.26108122e-01 -2.68919021e-01 1.73231554e+00 -2.92623937e-01
4.89533871e-01 1.29265130e+00 1.12888724e-01 6.39089569e-03
-1.34271193e+00 -7.17027545e-01 5.91798425e-01 4.17040676e-01
-1.21758342e+00 -1.76859617e-01 6.31096482e-01 2.41819881e-02
8.78806561e-02 1.09352601e+00 8.37310135e-01 4.09031212e-01
-3.90706301e-01 1.40852213e+00 1.30554736e+00 -3.17154944e-01
4.86114502e-01 4.78608385e-02 4.08167183e-01 3.13075721e-01
3.07606071e-01 2.42477700e-01 -7.65367746e-01 -9.59141254e-01
3.02012563e-01 1.27147734e-01 -4.13511544e-01 -5.12457013e-01
-8.65791261e-01 8.49360168e-01 3.08724582e-01 -6.55862316e-02
-4.52044278e-01 2.63079166e-01 1.11786604e-01 7.79219270e-01
4.32041705e-01 6.98995173e-01 -5.33285081e-01 -3.46968114e-01
-1.33540440e+00 5.46722233e-01 9.16236877e-01 6.76919341e-01
7.66774416e-01 -8.53297651e-01 -5.65115511e-01 9.69272017e-01
8.48099813e-02 5.10919809e-01 2.87804216e-01 -1.00492859e+00
6.81470633e-01 4.53466266e-01 6.44789398e-01 -7.13668168e-01
1.12759724e-01 -5.33944607e-01 -4.20510232e-01 7.22136423e-02
6.03173018e-01 -2.29383670e-02 -7.05426633e-01 1.67250848e+00
3.83490235e-01 -4.52955604e-01 -5.21851838e-01 1.31342638e+00
-1.41514670e-02 5.32191932e-01 -4.52161402e-01 -6.76437557e-01
1.28117979e+00 -9.89145100e-01 -6.50091588e-01 -3.77754718e-01
5.00236332e-01 -6.54102266e-01 9.40791607e-01 8.81935775e-01
-1.32573056e+00 1.49383873e-01 -9.31463122e-01 1.85615048e-01
-5.52969389e-02 -2.62100846e-01 9.53129590e-01 7.62304008e-01
-8.82446706e-01 6.71124995e-01 -3.76021028e-01 2.50108480e-01
1.86956137e-01 6.22453690e-01 -1.01143979e-02 -5.30063093e-01
-1.31580448e+00 4.83498394e-01 5.18497964e-03 3.37068737e-01
-7.40275979e-01 -4.96574312e-01 -1.55280819e-02 2.71095634e-01
1.17153800e+00 -3.79602671e-01 1.54186058e+00 -1.08212101e+00
-1.21649694e+00 3.61845821e-01 -2.91931391e-01 -1.65357038e-01
1.04319847e+00 -4.04250026e-01 1.07640833e-01 -1.67619362e-01
4.50960994e-02 -3.30645964e-02 4.43842292e-01 -1.10051322e+00
-9.14618552e-01 -6.92436814e-01 2.63901919e-01 6.89931154e-01
-2.27503687e-01 -1.45237625e-01 -7.27139354e-01 -3.45859557e-01
4.76533085e-01 -1.05769563e+00 -4.53114331e-01 -1.89062268e-01
-3.76184225e-01 -2.92999893e-01 2.11443424e-01 -3.72540712e-01
1.64692438e+00 -1.89813316e+00 1.24821477e-01 8.99007320e-01
2.53795809e-03 -3.83356325e-02 -2.40117356e-01 5.36604583e-01
3.51897180e-01 2.26256177e-01 1.70835286e-01 -5.02436645e-02
3.02413672e-01 1.10662565e-01 -4.17905509e-01 3.19873452e-01
-9.16202366e-01 6.69958949e-01 -1.02867687e+00 -2.95107991e-01
-4.30935770e-01 -4.24319595e-01 -7.06908166e-01 2.13502899e-01
-4.25035566e-01 -8.15245137e-02 -7.63963521e-01 6.49948299e-01
6.37911081e-01 -3.52504283e-01 4.90586162e-01 1.84023287e-02
3.45198177e-02 2.94729382e-01 -1.64805567e+00 1.31837142e+00
-3.55347514e-01 2.00381622e-01 3.08049768e-01 -9.19019282e-01
3.54813784e-01 7.02841580e-02 4.59848821e-01 -6.93350375e-01
-1.03644751e-01 5.63009143e-01 -1.08678883e-03 -1.10472903e-01
5.64088941e-01 1.28859594e-01 -2.93339789e-01 9.84625041e-01
-5.72720110e-01 1.15862332e-01 6.19660437e-01 3.87305558e-01
1.17539465e+00 -3.57393891e-01 2.25804240e-01 -1.31540835e-01
-2.46115699e-02 1.36262715e-01 6.04248285e-01 1.57774365e+00
-2.47654971e-02 1.50589496e-01 6.94269180e-01 -5.42889416e-01
-5.10879040e-01 -9.10321116e-01 6.13715164e-02 1.61908126e+00
5.43772459e-01 -2.44270444e-01 -4.17963266e-01 -7.52550840e-01
3.57457221e-01 3.93448979e-01 -6.23494446e-01 1.20626077e-01
-1.32539496e-01 -6.89362884e-01 4.02216911e-02 2.70639509e-02
4.31363583e-01 -5.28433323e-01 -5.68491340e-01 3.14156204e-01
-4.64439064e-01 -3.94970804e-01 -9.81679261e-01 3.49079996e-01
-8.34132552e-01 -1.22443748e+00 -6.65122688e-01 -8.83248970e-02
6.29621923e-01 5.08158863e-01 1.09993076e+00 1.44561455e-02
-4.41845395e-02 8.71607587e-02 -4.04663175e-01 -1.95881426e-01
3.77009362e-01 -3.92466923e-03 -3.72909680e-02 7.95181245e-02
2.31106356e-01 -2.79408187e-01 -1.13666892e+00 7.75389612e-01
-8.87078047e-01 -1.98739007e-01 8.16576719e-01 1.07692194e+00
6.84225440e-01 1.97570518e-01 1.87533349e-01 -1.22657502e+00
8.87850702e-01 -2.89756417e-01 -9.73748505e-01 6.03190839e-01
-8.67098153e-01 4.62709725e-01 2.90569991e-01 -8.08618009e-01
-7.25885868e-01 -1.63830593e-01 5.66722691e-01 -1.60796806e-01
7.02900648e-01 8.27920914e-01 9.04001947e-03 6.43664822e-02
4.07138795e-01 9.57054198e-02 -2.59627014e-01 -5.58340549e-01
3.80279571e-01 7.89073348e-01 5.50754443e-02 -8.82493377e-01
5.53840935e-01 3.56416345e-01 -1.92111313e-01 -2.44093820e-01
-9.87505198e-01 -7.81996667e-01 3.64286840e-01 -1.32901803e-01
-5.94791695e-02 -5.57178497e-01 -1.07064915e+00 3.15581001e-02
-7.04907596e-01 -3.39361936e-01 -1.84510410e-01 2.64519185e-01
-4.28093702e-01 2.25509375e-01 -3.31165075e-01 -1.31336272e+00
-3.58672768e-01 -1.27493405e+00 7.52509832e-01 1.44605651e-01
-2.82403558e-01 -3.25978279e-01 1.08944207e-01 7.52339661e-01
3.56077343e-01 -3.08463842e-01 8.20129573e-01 -8.28525901e-01
-1.05417907e+00 -5.57996750e-01 -1.64071128e-01 -1.70502529e-01
-2.67556310e-01 -5.67500114e-01 -4.34666663e-01 -6.04883313e-01
-2.14952677e-01 -3.58587027e-01 8.24077666e-01 5.53213477e-01
1.27583981e+00 -7.32321978e-01 -3.80328834e-01 3.66854429e-01
1.12104201e+00 4.60024595e-01 2.74753332e-01 3.63753051e-01
6.02521524e-02 5.85942328e-01 9.98352408e-01 5.79343140e-01
-5.79785109e-02 8.62800658e-01 3.64896625e-01 1.43153548e-01
7.09340990e-01 -1.41941771e-01 1.31366536e-01 2.48934135e-01
-4.70013265e-03 -3.52087706e-01 -4.33202714e-01 4.43740934e-01
-2.16660857e+00 -7.43848264e-01 4.27803606e-01 2.71879721e+00
1.13451612e+00 -3.69715434e-03 1.08099930e-01 2.05365390e-01
5.97883821e-01 7.44634718e-02 -9.08318639e-01 -4.14974958e-01
1.37341350e-01 -7.61639103e-02 8.12485158e-01 6.25180662e-01
-7.36211479e-01 4.65438932e-01 5.53645754e+00 1.28868437e+00
-7.48583615e-01 -2.94529665e-02 1.00632679e+00 -9.22884047e-01
-5.32006145e-01 3.39575291e-01 -7.12013245e-01 5.49790502e-01
4.41582620e-01 -3.71258467e-01 1.20480871e+00 9.26838040e-01
8.08565915e-02 -5.88472664e-01 -1.31626844e+00 1.04575729e+00
-2.64286727e-01 -9.94079590e-01 -1.65707797e-01 4.21891809e-01
8.74328792e-01 -7.25906268e-02 2.41917029e-01 1.71671689e-01
8.38115335e-01 -7.42294729e-01 7.32348859e-01 3.36536430e-02
7.26660371e-01 -8.73916268e-01 5.78180373e-01 5.51980138e-01
-9.09651399e-01 -4.33366269e-01 -1.15164123e-01 9.36826617e-02
1.73408598e-01 8.68706822e-01 -4.12257969e-01 4.43857044e-01
8.40773284e-01 -1.75187469e-01 8.81418306e-03 1.31940544e+00
7.32493261e-03 3.07024062e-01 -7.45814741e-01 -4.87200916e-01
3.99191588e-01 -5.10706186e-01 5.57515562e-01 5.06989360e-01
4.53522384e-01 2.95283258e-01 2.68747717e-01 3.67737889e-01
-2.32356608e-01 1.59282088e-01 -6.10377192e-02 -1.33728266e-01
8.09697807e-01 1.04832017e+00 -3.92433912e-01 -5.11552870e-01
1.53404057e-01 7.77105331e-01 3.02759171e-01 5.27804136e-01
-3.43418151e-01 -2.22459137e-01 6.61848605e-01 2.17284501e-01
2.90187210e-01 2.60469526e-01 -1.21021062e-01 -8.10600340e-01
3.29944253e-01 -1.14483678e+00 9.52465892e-01 -4.28010374e-01
-1.21464205e+00 3.12348247e-01 1.42202765e-01 -1.02799940e+00
-1.67901799e-01 -2.38250405e-01 -2.48546317e-01 7.22968578e-01
-1.28175366e+00 -4.43334013e-01 6.77245483e-02 3.61087769e-01
4.65554833e-01 2.96397001e-01 4.58454996e-01 1.20846756e-01
-2.28892565e-01 8.19745123e-01 6.91925704e-01 -4.35660750e-01
6.80372477e-01 -1.16926348e+00 -4.01002496e-01 4.26693857e-01
2.16956496e-01 8.42890918e-01 8.04567575e-01 -4.05461222e-01
-1.84144711e+00 -4.62313205e-01 8.04912627e-01 9.82140307e-04
5.16001046e-01 -2.26574868e-01 -4.78331774e-01 2.44624436e-01
-2.91463375e-01 -8.15635100e-02 7.68417537e-01 7.84151375e-01
-4.19227779e-01 -5.39520085e-01 -1.01450503e+00 7.80915976e-01
1.02669525e+00 -2.98442662e-01 -2.04110950e-01 4.23013002e-01
2.76693076e-01 -3.86985213e-01 -4.92992967e-01 3.12102407e-01
1.04411268e+00 -1.03902996e+00 8.44983995e-01 -4.35646623e-01
5.09985611e-02 -1.74535856e-01 -1.99363276e-01 -1.31636953e+00
-9.58683491e-02 -1.18743682e+00 -1.23293787e-01 6.71090901e-01
5.79304457e-01 -6.32574975e-01 9.77424443e-01 1.10855639e+00
5.34405768e-01 -9.70070302e-01 -1.16410005e+00 -8.58456373e-01
-2.69536585e-01 -1.76368386e-01 7.42665052e-01 5.54199040e-01
2.42375895e-01 -5.00729457e-02 -6.03940845e-01 -1.31566688e-01
7.30526328e-01 7.40540385e-01 6.89799309e-01 -1.12351573e+00
-9.06934679e-01 -6.36359632e-01 4.83687580e-01 -1.55866921e+00
-5.33351123e-01 -4.45022702e-01 2.67074674e-01 -1.34298003e+00
6.30661786e-01 -8.09499621e-01 -6.85535133e-01 3.60098630e-01
-4.37119424e-01 -1.06224678e-01 2.22952366e-01 3.57008010e-01
-1.11585498e+00 1.69667318e-01 1.25442863e+00 -2.63215989e-01
-4.79131639e-01 4.17997211e-01 -1.09306610e+00 2.21134931e-01
4.08667773e-01 -4.81361717e-01 -4.36800569e-01 -1.68488801e-01
8.49335611e-01 6.42672002e-01 -1.73102900e-01 -2.76845038e-01
6.00889444e-01 -4.12438840e-01 -9.14425496e-03 -8.58592212e-01
3.26736629e-01 -7.55374193e-01 3.93798172e-01 4.84086514e-01
-7.16607153e-01 -9.86121502e-03 -1.87595412e-01 8.51579130e-01
-1.67193741e-01 -1.73397392e-01 3.29593599e-01 -1.93257481e-01
-1.37749895e-01 5.43938577e-01 -1.40605733e-01 5.94768114e-02
7.19437003e-01 -1.00669667e-01 -2.71559656e-01 -7.99008608e-01
-4.49271142e-01 7.29702830e-01 2.34069332e-01 -6.38730079e-02
2.82044739e-01 -1.25100410e+00 -5.39560735e-01 4.52202968e-02
1.28454000e-01 1.51460757e-02 2.48525292e-01 8.15422714e-01
7.66106769e-02 3.46724153e-01 5.09796500e-01 -3.18411499e-01
-1.50961077e+00 4.00067151e-01 -4.55400497e-02 -1.04869127e+00
1.92330495e-01 9.80122566e-01 1.05333678e-01 -6.14739656e-02
4.96539891e-01 1.22844145e-01 2.64709055e-01 4.12791252e-01
6.67824447e-01 5.50116181e-01 1.40224740e-01 3.69436532e-01
-1.78711638e-01 2.56191820e-01 -5.79196811e-01 -5.41512609e-01
1.13336349e+00 -1.88992843e-01 -3.75392228e-01 1.88506633e-01
9.71605599e-01 5.69951907e-02 -1.04222977e+00 -5.25871456e-01
1.53598562e-01 -9.98981953e-01 3.08653414e-01 -1.11462259e+00
-8.68059874e-01 2.47251078e-01 4.52257574e-01 4.80676621e-01
1.17713499e+00 -1.65110882e-02 7.33481288e-01 6.27578378e-01
6.84674084e-01 -1.43581927e+00 -1.53380543e-01 3.32385927e-01
6.82324708e-01 -1.20791936e+00 1.63444251e-01 -3.98465544e-02
-4.83766466e-01 5.56349993e-01 1.89991415e-01 2.48250857e-01
4.49530602e-01 -1.47495970e-01 -1.88519862e-02 -1.62717223e-01
-1.02482617e+00 -4.34913337e-02 3.76126021e-01 -3.11188966e-01
1.01205334e-01 3.98834199e-01 -7.42408931e-01 6.99873090e-01
-1.65032744e-01 -5.85633554e-02 -9.66312215e-02 9.24675882e-01
-5.17393112e-01 -1.46295631e+00 -6.30677998e-01 9.66599166e-01
-5.42766273e-01 -2.24601954e-01 -4.53166276e-01 2.24500641e-01
-3.96088600e-01 1.16949606e+00 -1.62889794e-01 -1.83506295e-01
1.45227581e-01 -2.17840999e-01 3.43373686e-01 -1.94948375e-01
-6.45676911e-01 4.07138616e-01 2.58840919e-01 -9.45626020e-01
2.47123502e-02 -6.06885672e-01 -5.28941274e-01 -2.95804292e-01
-6.31409585e-01 9.02564764e-01 6.44103229e-01 9.08363998e-01
3.48381668e-01 9.93151739e-02 9.71119940e-01 -5.99111617e-01
-1.33856499e+00 -7.04992175e-01 -8.97077560e-01 4.41070348e-01
3.59922796e-01 -5.82441509e-01 -5.70088804e-01 -6.12472057e-01]
|
[4.6884541511535645, 3.3872735500335693]
|
886c4a20-2573-4d7b-a1fb-aed181182a80
|
scenario-based-cost-optimization-of-water
|
2307.00845
| null |
https://arxiv.org/abs/2307.00845v1
|
https://arxiv.org/pdf/2307.00845v1.pdf
|
Scenario Based Cost Optimization of Water Distribution Networks Powered by Grid-Connected Photovoltaic Systems
|
The paper presents a predictive control method for the water distribution networks (WDNs) powered by photovoltaics (PVs) and the electrical grid. This builds on the controller introduced in a previous study and is designed to reduce the economic costs associated with operating the WDN. To account for the uncertainty of the system, the problem is solved in a scenario optimization framework, where multiple scenarios are sampled from the uncertain variables related to PV power production. To accomplish this, a day-ahead PV power prediction method with a stochastic model is employed. The method is tested on a high-fidelity model of a WDN of a Danish town and the results demonstrate a substantial reduction in electrical costs through the integration of PVs, with PVs supplying $66.95\%$ of the required energy. The study also compares the effectiveness of the stochastic optimization method with a deterministic optimization approach.
|
['John Leth', 'Jan Dimon Bendtsen', 'Carsten Kallesøe', 'Mirhan Ürkmez']
|
2023-07-03
| null | null | null | null |
['stochastic-optimization']
|
['methodology']
|
[ 1.10853903e-01 3.30511153e-01 3.00311297e-01 8.11078995e-02
5.25995679e-02 -5.36817491e-01 6.88434780e-01 2.05492809e-01
2.27504537e-01 1.32961726e+00 1.33268148e-01 -2.23784335e-02
-7.36532927e-01 -9.61855352e-01 -1.56116217e-01 -1.12179863e+00
-8.35581496e-02 3.19640517e-01 -1.92802325e-01 -1.79310098e-01
-6.75766021e-02 5.75192809e-01 -1.45392394e+00 -3.63167375e-01
1.15816331e+00 9.52117920e-01 6.70587838e-01 3.17643881e-01
2.20767871e-01 2.33953446e-01 -6.61041498e-01 2.81449258e-01
4.28258091e-01 -2.04610720e-01 -7.89772943e-02 4.66714315e-02
-6.00547552e-01 -4.39678639e-01 6.08230792e-02 6.69489622e-01
7.13275433e-01 3.63547206e-01 8.95931542e-01 -1.55137992e+00
-1.89114420e-03 2.72087455e-01 -2.16572627e-01 -2.39394501e-01
9.42400247e-02 3.95006001e-01 7.20073521e-01 -3.80532444e-01
3.14259708e-01 1.03471828e+00 3.65272909e-01 -8.03061873e-02
-1.55555391e+00 -2.25708008e-01 -1.58176646e-01 -4.39820252e-02
-1.08754921e+00 -2.84494460e-02 5.64527154e-01 -4.78316188e-01
1.52292025e+00 1.82466760e-01 1.45690906e+00 2.74809152e-01
4.63038594e-01 1.43401921e-01 1.35197425e+00 -5.28971553e-01
7.48211026e-01 -2.92640626e-02 -4.66871023e-01 -2.08067119e-01
5.50715208e-01 5.32426238e-01 6.47417083e-02 -9.43631530e-02
1.84362635e-01 -3.03723991e-01 -3.26202035e-01 -4.90739375e-01
-4.55906332e-01 8.27938437e-01 3.98211032e-01 4.92598116e-01
-6.38943791e-01 1.13501616e-01 -7.28159845e-02 -5.51182590e-02
4.41021681e-01 2.46336401e-01 -6.16806686e-01 2.85303313e-02
-1.02261508e+00 1.76640466e-01 1.22717512e+00 8.50495219e-01
2.08392993e-01 5.83491743e-01 -3.75149220e-01 3.67241889e-01
8.33055079e-01 1.06317031e+00 9.85580236e-02 -1.30539668e+00
1.50737748e-01 4.64514315e-01 6.94770873e-01 -2.33134583e-01
-4.07064050e-01 -1.40298218e-01 -8.39852691e-01 8.57982993e-01
1.98983684e-01 -9.69974816e-01 -8.56021702e-01 1.25961494e+00
2.50127882e-01 -2.69234478e-01 2.47243077e-01 1.75353408e-01
-1.69359878e-01 1.25438190e+00 8.72571543e-02 -7.88336992e-01
8.84921670e-01 -5.04560113e-01 -9.59532142e-01 4.23605710e-01
-2.09133513e-03 -5.31345963e-01 1.90009981e-01 2.39844635e-01
-1.17966497e+00 -5.87924123e-02 -8.93732250e-01 5.66199124e-01
-5.73800564e-01 -3.23218070e-02 -1.94217667e-01 6.98810756e-01
-1.17209542e+00 8.43844116e-01 -7.80858815e-01 -4.67892021e-01
1.96539491e-01 1.94462553e-01 3.74384046e-01 4.32540387e-01
-1.05973589e+00 1.61886334e+00 6.58310175e-01 5.36873400e-01
-5.81958890e-01 -1.01178336e+00 -7.02140689e-01 7.36636162e-01
-1.10497242e-02 -7.41370618e-01 1.29453838e+00 -6.06635809e-01
-2.09661555e+00 -3.62441987e-01 -3.86939384e-02 -5.35728693e-01
6.67609870e-01 3.75378698e-01 -7.28848651e-02 3.30537111e-02
-1.88122034e-01 1.12090632e-01 3.98482084e-01 -1.27572668e+00
-8.95997465e-01 5.81811443e-02 -5.34737945e-01 2.53366172e-01
6.91259727e-02 -5.24402440e-01 7.77503729e-01 -4.03430223e-01
-4.04118598e-01 -7.79785872e-01 -6.26830637e-01 -2.24807918e-01
-1.91299051e-01 -3.50386888e-01 8.39791596e-01 -8.55955720e-01
8.01154673e-01 -1.72332466e+00 1.48644775e-01 9.25851703e-01
-6.65991545e-01 8.42082277e-02 2.69332677e-01 1.18614459e+00
1.68101583e-02 -1.63995117e-01 -6.33172691e-01 3.06846444e-02
5.54026484e-01 9.12602663e-01 -1.04308739e-01 8.30909312e-02
-2.41612531e-02 5.94172478e-01 -8.18360746e-01 2.37715065e-01
7.81682909e-01 3.84375870e-01 1.16283543e-01 1.26686208e-02
-5.10492206e-01 1.89929724e-01 -3.62871110e-01 3.07177603e-01
7.65290916e-01 2.84375280e-01 6.24248564e-01 7.57630765e-02
-7.01215386e-01 -4.10342127e-01 -1.45157599e+00 1.22031534e+00
-7.85013318e-01 4.87112701e-01 4.56927747e-01 -1.14886117e+00
8.83174717e-01 4.50442940e-01 7.34942138e-01 -5.65683722e-01
-2.34858647e-01 5.74164808e-01 1.04077853e-01 -6.47167623e-01
1.40361220e-01 -2.89183736e-01 5.45533419e-01 7.72896856e-02
1.76581219e-01 -8.42973173e-01 6.03397846e-01 -2.95834512e-01
8.42083097e-01 4.06995088e-01 4.94778603e-01 -1.00034165e+00
5.41319728e-01 1.89328361e-02 8.46290350e-01 -1.44228935e-01
2.23267719e-01 -2.39003375e-01 6.14300489e-01 1.49881229e-01
-1.08246851e+00 -9.49886262e-01 -2.38573372e-01 -3.28215063e-02
2.08753366e-02 4.80046600e-01 -4.08673376e-01 -6.86070845e-02
5.78573167e-01 1.64338779e+00 -2.12503642e-01 4.06774312e-01
-2.73707330e-01 -9.95769978e-01 -4.13890988e-01 2.12884203e-01
4.22360450e-01 -6.90805376e-01 -1.00417793e+00 6.97507024e-01
3.80850673e-01 -5.77587306e-01 2.94850141e-01 4.21508193e-01
-8.25210989e-01 -8.83579314e-01 -9.27115142e-01 -5.26124179e-01
7.70614088e-01 -4.56266493e-01 7.96584547e-01 -2.83011645e-01
-2.01142386e-01 4.66779619e-01 1.18972220e-01 -7.92333066e-01
-4.17764872e-01 -1.83778316e-01 -7.90709779e-02 -4.67163175e-01
-2.38112837e-01 -8.39514375e-01 -6.01807952e-01 -1.22731123e-02
-5.06740212e-01 -1.12559088e-01 3.30221146e-01 8.65473866e-01
2.85344720e-01 8.02198589e-01 9.47100282e-01 -3.70885551e-01
7.44277716e-01 -5.49932659e-01 -1.52575719e+00 4.26619798e-01
-1.13876951e+00 1.45866245e-01 5.10539711e-01 -7.25429058e-02
-1.59406209e+00 5.16749561e-01 2.86093324e-01 1.21052153e-01
-1.66519210e-01 3.93521845e-01 -4.13764834e-01 -2.39204839e-01
-2.72497147e-01 -2.90709455e-02 3.93062457e-02 -4.92205828e-01
8.53086635e-02 4.23455179e-01 -3.99841666e-02 -3.25217068e-01
1.20207512e+00 4.24288698e-02 7.46703386e-01 -7.03547478e-01
2.99369931e-01 -7.91938975e-02 -2.90401697e-01 -4.16701198e-01
4.42517310e-01 -7.86078095e-01 -9.25678670e-01 6.80316567e-01
-9.49805021e-01 -7.74266481e-01 -8.94322991e-01 3.99670541e-01
-6.25705361e-01 1.27142251e-01 9.79108959e-02 -1.24942446e+00
-3.63978505e-01 -8.52249146e-01 2.21277222e-01 6.27862811e-01
1.43096834e-01 -1.11100245e+00 3.07534993e-01 -4.52138424e-01
8.31987977e-01 8.89185846e-01 9.88608122e-01 2.18484662e-02
-6.02456927e-01 3.71352553e-01 8.42077360e-02 5.83426118e-01
2.85428017e-01 4.15604383e-01 -6.72318697e-01 -4.24945205e-01
5.67760039e-03 4.13838625e-01 -1.38747413e-02 5.23671806e-01
4.66110617e-01 -5.65064013e-01 -4.33775187e-01 8.12897179e-03
2.54280949e+00 8.55974257e-01 4.14375961e-01 3.64686340e-01
-1.95030123e-01 6.69524491e-01 6.44822478e-01 6.71329200e-01
3.55844468e-01 5.03133297e-01 8.56264293e-01 1.82416663e-01
3.17898571e-01 9.53001752e-02 1.90255865e-01 1.07988581e-01
-2.25431189e-01 -5.92170238e-01 -8.29483449e-01 9.91868556e-01
-1.92882097e+00 -8.18818331e-01 -2.25111082e-01 2.07421494e+00
3.39339167e-01 -2.48797804e-01 -2.17031226e-01 1.50984228e-01
6.10990465e-01 -1.54085651e-01 -3.91777426e-01 -9.18893158e-01
-1.17323242e-01 3.69673908e-01 9.74759698e-01 5.27253985e-01
-2.19076142e-01 -4.09700900e-01 5.74374151e+00 6.50889874e-01
-5.98629117e-01 -3.10712725e-01 4.20184731e-01 -3.01980469e-02
-2.79134631e-01 1.06940903e-01 -4.48740900e-01 1.11374068e+00
1.26681554e+00 -8.50535512e-01 6.90307260e-01 4.97120321e-01
1.15395880e+00 -1.00940108e+00 -8.46905291e-01 1.36462957e-01
-5.33820331e-01 -1.07448173e+00 -4.83580410e-01 3.58787537e-01
1.32978952e+00 1.29590377e-01 -6.13642097e-01 3.31743136e-02
5.92659295e-01 -5.86562932e-01 4.78883207e-01 1.00630355e+00
1.70862656e-02 -9.82049346e-01 1.10197949e+00 5.73516965e-01
-1.12282753e+00 -6.99313343e-01 8.47553834e-02 -2.04324394e-01
9.08345282e-01 7.24991560e-01 -6.67871237e-01 9.19298351e-01
8.88011754e-01 2.90571958e-01 1.91004783e-01 1.34189236e+00
-5.13792634e-01 2.81442940e-01 -9.42227662e-01 -3.75383914e-01
7.29167014e-02 -8.69558573e-01 5.80799222e-01 7.99078107e-01
1.06068611e+00 3.47749174e-01 -2.41616160e-01 6.72813773e-01
3.98566693e-01 -8.33781585e-02 -5.82077742e-01 4.70503896e-01
6.26456618e-01 1.21213734e+00 -1.66772678e-01 -1.01448923e-01
-1.43425554e-01 3.55370283e-01 -6.49025381e-01 6.81528270e-01
-4.15077627e-01 -5.86316049e-01 4.74697053e-01 -1.53758869e-01
5.59912443e-01 -1.48364352e-02 -5.37895739e-01 -4.25736666e-01
1.77808151e-01 1.28766503e-02 -7.48070469e-03 -1.14594233e+00
-1.46227992e+00 -7.55688846e-02 4.88760650e-01 -1.17144120e+00
-7.85684943e-01 -3.72168362e-01 -1.06574869e+00 1.47170722e+00
-1.95718420e+00 -7.78201699e-01 -1.71770185e-01 4.90376316e-02
3.69769037e-01 -5.96516281e-02 8.06562066e-01 -4.57768105e-02
-6.31252289e-01 -5.35071790e-01 1.24824202e+00 -5.95545948e-01
-1.35819778e-01 -1.60050762e+00 -3.68823469e-01 8.39435995e-01
-9.63360786e-01 -2.62495935e-01 8.98118556e-01 -4.58257943e-01
-1.12521410e+00 -8.57715905e-01 1.06555688e+00 5.34538150e-01
8.28656912e-01 1.40940160e-01 -4.82643425e-01 2.56669462e-01
1.35656691e+00 -3.99079114e-01 2.17222571e-01 -8.32273483e-01
8.06770563e-01 -3.70579123e-01 -2.03218007e+00 2.36038804e-01
2.24177942e-01 1.86764434e-01 -5.37557423e-01 1.17760070e-01
1.94234684e-01 -2.39642754e-01 -1.40344787e+00 4.73788321e-01
5.72759151e-01 -2.37539649e-01 4.97922301e-01 3.34420472e-01
-1.81655705e-01 -4.88000393e-01 4.05039079e-02 -2.26067019e+00
7.77270971e-03 -8.26947987e-01 -4.39296752e-01 1.55549598e+00
3.27654570e-01 -1.26316047e+00 2.53022194e-01 9.99219477e-01
7.01889321e-02 -5.60293078e-01 -1.72627926e+00 -8.80064726e-01
1.25046507e-01 5.01410425e-01 9.07118797e-01 5.60030580e-01
1.53393507e-01 -1.24711789e-01 2.48884648e-01 5.26644111e-01
9.74371791e-01 1.11193508e-01 1.59899637e-01 -1.30551291e+00
1.15851901e-01 -3.42379421e-01 8.72861817e-02 1.38914585e-02
-2.15099916e-01 -2.57634193e-01 2.19654709e-01 -2.25737166e+00
-4.29142118e-01 -2.24189505e-01 -4.56352495e-02 4.96614933e-01
3.97544116e-01 -3.79659444e-01 5.22766650e-01 -1.97093040e-01
8.25393319e-01 1.09279346e+00 7.63163447e-01 -2.23694846e-01
-3.03836286e-01 4.46389407e-01 1.14664279e-01 5.28996170e-01
1.17581570e+00 -2.72000164e-01 -5.65389037e-01 -5.00735268e-02
1.30397184e-02 3.39164764e-01 2.22910687e-01 -9.90468264e-01
2.18100473e-03 -4.33509082e-01 2.22258180e-01 -9.50761378e-01
7.45276064e-02 -1.81732798e+00 1.02268314e+00 1.15469110e+00
3.29312861e-01 -1.68621168e-01 2.71097809e-01 7.06722736e-01
-1.06869586e-01 -4.95428145e-01 7.05892682e-01 2.11696044e-01
-5.31148195e-01 -3.15562874e-01 -7.67712593e-01 -7.95689106e-01
1.58111107e+00 -1.99504375e-01 -5.61785877e-01 -1.42023578e-01
-9.45401490e-01 1.07420123e+00 2.85638720e-01 -1.58136964e-01
-8.08439553e-02 -9.61273372e-01 -6.44181907e-01 -1.86229020e-01
-6.43529713e-01 -8.15625954e-03 4.37517576e-02 4.07889545e-01
-6.40560746e-01 3.01697969e-01 -3.24919760e-01 -3.86758149e-01
-8.60599995e-01 2.24994943e-01 7.65070856e-01 -4.45917428e-01
-1.92109406e-01 -2.49887541e-01 -7.40225315e-01 -1.77199170e-01
-1.68399245e-01 -6.30745411e-01 -1.94776326e-01 5.74537218e-01
-2.16726899e-01 1.15117514e+00 8.64962116e-02 -1.71455786e-01
-1.35556725e-03 4.00704682e-01 1.18733275e+00 -2.22255126e-01
1.91797495e+00 -4.96523201e-01 -1.38980389e-01 2.93995827e-01
6.32867992e-01 -3.42262864e-01 -1.64997339e+00 1.79703996e-01
5.85642867e-02 -7.03671426e-02 8.21579322e-02 -1.32528520e+00
-1.12447703e+00 2.64297754e-01 8.23156953e-01 7.32887208e-01
1.40571308e+00 -1.07015872e+00 2.19948843e-01 3.40332150e-01
4.29046273e-01 -1.52795982e+00 -8.73963416e-01 2.69483566e-01
1.02489805e+00 -7.37982631e-01 3.01431388e-01 -1.35823026e-01
-2.70142764e-01 1.26121783e+00 1.79695040e-01 -3.00949216e-01
1.01665759e+00 4.68975604e-01 -5.83295166e-01 2.54544884e-01
-6.47115767e-01 -1.54677466e-01 -3.85292023e-01 5.70513606e-01
-3.45217764e-01 4.13770586e-01 -8.14796686e-01 1.01112753e-01
2.83585191e-01 4.39896494e-01 7.58438349e-01 1.03379560e+00
-1.77281424e-01 -1.13807702e+00 -3.39875221e-01 3.77597958e-01
-4.22153473e-02 1.41775385e-01 4.83698010e-01 9.60811555e-01
4.31050450e-01 1.05358994e+00 7.29962364e-02 6.84015155e-01
8.71637881e-01 5.40228039e-02 5.59958108e-02 -3.19178104e-01
-7.47804046e-01 9.79818255e-02 2.49121457e-01 -5.56770749e-02
-6.34820998e-01 -1.18933833e+00 -1.09774232e+00 -2.38776818e-01
-3.43171775e-01 4.66262162e-01 1.24091852e+00 9.69307601e-01
-2.35563354e-03 6.35949492e-01 1.24933815e+00 -1.03233421e+00
-7.58633614e-01 -8.98264050e-01 -9.96261001e-01 -5.97813129e-01
1.41914496e-02 -4.43928778e-01 -6.84338868e-01 -2.49326766e-01]
|
[5.676445007324219, 2.5356333255767822]
|
165c7cca-f9b8-42f6-82d2-98c78fbbb775
|
query-efficient-decision-based-black-box
|
2307.00477
| null |
https://arxiv.org/abs/2307.00477v1
|
https://arxiv.org/pdf/2307.00477v1.pdf
|
Query-Efficient Decision-based Black-Box Patch Attack
|
Deep neural networks (DNNs) have been showed to be highly vulnerable to imperceptible adversarial perturbations. As a complementary type of adversary, patch attacks that introduce perceptible perturbations to the images have attracted the interest of researchers. Existing patch attacks rely on the architecture of the model or the probabilities of predictions and perform poorly in the decision-based setting, which can still construct a perturbation with the minimal information exposed -- the top-1 predicted label. In this work, we first explore the decision-based patch attack. To enhance the attack efficiency, we model the patches using paired key-points and use targeted images as the initialization of patches, and parameter optimizations are all performed on the integer domain. Then, we propose a differential evolutionary algorithm named DevoPatch for query-efficient decision-based patch attacks. Experiments demonstrate that DevoPatch outperforms the state-of-the-art black-box patch attacks in terms of patch area and attack success rate within a given query budget on image classification and face verification. Additionally, we conduct the vulnerability evaluation of ViT and MLP on image classification in the decision-based patch attack setting for the first time. Using DevoPatch, we can evaluate the robustness of models to black-box patch attacks. We believe this method could inspire the design and deployment of robust vision models based on various DNN architectures in the future.
|
['Wenqiang Zhang', 'Shouhong Ding', 'Shuang Wu', 'Bo Li', 'Zhaoyu Chen']
|
2023-07-02
| null | null | null | null |
['face-verification']
|
['computer-vision']
|
[ 3.16113263e-01 6.51430711e-02 -8.32577795e-02 -1.58297256e-01
-6.08996511e-01 -9.16396618e-01 4.29870665e-01 -2.17396170e-01
-3.25785875e-01 3.59790713e-01 -4.84988093e-01 -5.79834878e-01
3.30514982e-02 -9.37582254e-01 -1.21883821e+00 -8.96502435e-01
-1.30824581e-01 3.09803393e-02 2.84692079e-01 -3.11675847e-01
1.38655707e-01 7.19247043e-01 -1.30911148e+00 2.13451505e-01
5.51133573e-01 1.12287188e+00 -3.51468056e-01 4.85237271e-01
4.44384605e-01 4.64249492e-01 -1.14479733e+00 -8.81170511e-01
8.46281230e-01 -2.76402920e-01 -3.24902564e-01 -5.25819659e-01
6.50600612e-01 -4.46468920e-01 -6.20387554e-01 1.54367936e+00
9.50939119e-01 -2.38682732e-01 2.73959756e-01 -1.64131272e+00
-5.89930356e-01 7.33068049e-01 -4.44717288e-01 1.82943672e-01
1.00729056e-01 6.19607985e-01 5.50681889e-01 -5.91344774e-01
3.32133710e-01 1.46934569e+00 6.64021075e-01 9.37273383e-01
-1.27019751e+00 -1.31259584e+00 8.90011340e-02 5.76499879e-01
-1.64021719e+00 -4.51430380e-01 9.07031238e-01 -1.13584965e-01
7.75291085e-01 4.51370746e-01 2.62605339e-01 1.53534603e+00
3.09183061e-01 6.39779806e-01 1.11049616e+00 -3.70962113e-01
4.22108471e-01 2.41327643e-01 -1.84373707e-01 5.05955994e-01
2.72028267e-01 9.17952776e-01 -2.80993223e-01 -5.22128046e-01
3.32959294e-01 -1.72478929e-01 -4.84325111e-01 -9.79119092e-02
-6.33588374e-01 9.37210560e-01 5.71817160e-01 -1.00571968e-01
-2.75921136e-01 2.59391308e-01 3.59730124e-01 5.51414371e-01
1.77107424e-01 6.32972062e-01 -4.21669573e-01 3.32363397e-01
-9.07114625e-01 3.00484985e-01 7.97396898e-01 5.19708574e-01
5.88890791e-01 3.03527594e-01 -2.99704373e-01 4.76290077e-01
2.11344063e-01 7.98199832e-01 1.55516893e-01 -6.59343600e-01
3.96564513e-01 1.81919381e-01 -2.99795806e-01 -1.34067392e+00
-4.19500321e-02 -5.08447170e-01 -6.83627725e-01 5.36568046e-01
2.26753160e-01 -5.94846427e-01 -1.08745146e+00 2.02339983e+00
4.53503579e-01 3.95400137e-01 3.33500266e-01 7.11360276e-01
6.13747001e-01 5.99354982e-01 -2.02101186e-01 6.66181184e-03
1.26793361e+00 -6.50272369e-01 -3.67414117e-01 -1.32401183e-01
3.48045975e-01 -5.70408821e-01 7.25531399e-01 3.66153061e-01
-7.02864826e-01 -5.27579069e-01 -1.44066775e+00 6.76028430e-01
-3.72228324e-01 -1.10503227e-01 2.83015389e-02 1.35690176e+00
-1.01615024e+00 3.96839082e-01 -7.65447021e-01 9.51168463e-02
7.17279613e-01 7.38918185e-01 -2.99215198e-01 -5.11569381e-02
-1.43425202e+00 7.46803880e-01 1.63878754e-01 1.44202471e-01
-1.37196863e+00 -8.49985540e-01 -6.31109416e-01 1.99879967e-02
3.57597530e-01 -2.41020456e-01 9.11105812e-01 -1.03564084e+00
-1.43099046e+00 6.23079717e-01 3.33957762e-01 -8.84590268e-01
5.07661879e-01 3.00150126e-01 -5.68961322e-01 3.48469764e-01
-4.24536496e-01 8.77128065e-01 1.32460153e+00 -1.25016487e+00
-2.22377867e-01 -3.20554137e-01 3.35809737e-01 -3.21124882e-01
-6.30248010e-01 1.87354162e-01 -1.14581764e-01 -9.14054215e-01
-3.71578872e-01 -1.20210886e+00 -3.13193917e-01 1.46654293e-01
-6.96688056e-01 1.67014480e-01 1.19354141e+00 -4.84889686e-01
9.59488213e-01 -2.36345482e+00 -8.80967304e-02 4.16214257e-01
5.37022986e-02 7.93927729e-01 -4.36803311e-01 2.14149043e-01
-2.15379551e-01 3.91148895e-01 -1.93546697e-01 -1.58848450e-01
4.71861698e-02 1.07087880e-01 -8.00515831e-01 5.93415618e-01
2.90356219e-01 9.05806303e-01 -1.64810717e-01 -1.76852614e-01
-2.36990929e-01 5.54057479e-01 -8.00004661e-01 1.19446486e-01
-3.43679935e-01 1.97866291e-01 -3.25756609e-01 7.34475076e-01
1.05228221e+00 3.24075013e-01 -1.96780428e-01 -2.49157950e-01
4.62937444e-01 -1.93396047e-01 -9.16532576e-01 8.11037660e-01
3.92306224e-02 7.15985835e-01 8.43694732e-02 -9.52628195e-01
9.46841538e-01 2.67349273e-01 1.03263117e-01 -6.47573173e-01
3.92909437e-01 -8.04483965e-02 3.68850678e-01 -2.48901501e-01
1.19703680e-01 1.53150231e-01 -1.23835057e-01 4.54142451e-01
-1.39990717e-01 1.61566287e-01 -3.52657586e-01 -7.52706751e-02
1.22588885e+00 -4.72502798e-01 -1.60537302e-01 -6.36704862e-02
4.34878588e-01 -2.38734588e-01 6.37260795e-01 9.86090541e-01
-3.14599037e-01 4.42886055e-01 6.55560136e-01 -4.19780999e-01
-9.14964318e-01 -8.28821063e-01 -2.15977788e-01 6.45649672e-01
3.12128305e-01 -2.32132837e-01 -1.23330665e+00 -9.63944972e-01
7.82642365e-02 5.97053289e-01 -7.78990507e-01 -6.40784681e-01
-4.06391382e-01 -6.63618088e-01 1.43700540e+00 3.03545922e-01
7.67801464e-01 -1.10442770e+00 -5.67887306e-01 -1.16281062e-01
2.52651840e-01 -1.14101028e+00 -4.16554660e-01 9.98616740e-02
-3.71658236e-01 -9.78711188e-01 -5.51914334e-01 -7.33891547e-01
8.14269364e-01 -1.53101578e-01 5.12454391e-01 1.80539012e-01
-2.98346519e-01 3.11789334e-01 -3.08788896e-01 -6.69981480e-01
-6.48090780e-01 -2.07365483e-01 2.61570096e-01 2.46942684e-01
2.04953581e-01 -5.97157121e-01 -6.37963951e-01 6.70080781e-01
-1.15541840e+00 -6.02658212e-01 5.67746639e-01 9.41711843e-01
4.97097224e-01 4.58435893e-01 3.22553307e-01 -5.26667774e-01
6.87617362e-01 -2.07229555e-01 -9.31855679e-01 4.35424596e-01
-4.41353232e-01 5.19654043e-02 7.49564230e-01 -1.18129301e+00
-5.52623034e-01 1.03623339e-03 -4.04917806e-01 -9.13948417e-01
1.67703969e-04 2.99529195e-01 -6.24308228e-01 -9.04186308e-01
1.00671637e+00 2.83139378e-01 5.07724024e-02 -2.01225385e-01
2.33724177e-01 5.50484180e-01 4.99128819e-01 -6.27103329e-01
1.30041134e+00 3.35473776e-01 1.09158956e-01 -7.43803382e-01
-9.70516503e-02 4.37804341e-01 1.58090785e-01 -1.66999117e-01
5.28828740e-01 -7.75783062e-01 -8.64469528e-01 9.38252568e-01
-1.13740659e+00 -2.73096830e-01 -9.56670940e-02 6.54422939e-02
-1.93117902e-01 4.90669608e-01 -4.98990089e-01 -6.59896672e-01
-5.52467108e-01 -1.61634421e+00 6.44004047e-01 2.87685484e-01
2.45328739e-01 -3.92524362e-01 -1.05816916e-01 2.24944651e-01
5.55036962e-01 3.02082717e-01 9.55783010e-01 -9.73134696e-01
-7.39462197e-01 -5.22824764e-01 8.90794396e-02 7.38452435e-01
-2.82747477e-01 6.65451214e-02 -1.01530957e+00 -7.69046247e-01
2.94658214e-01 -3.44371587e-01 7.57054925e-01 2.76934654e-01
1.30982995e+00 -8.45169783e-01 -4.31162953e-01 1.06925786e+00
1.27960920e+00 5.73092163e-01 9.56309557e-01 3.62048507e-01
4.60973263e-01 4.24187183e-01 3.44166130e-01 3.36171955e-01
-1.53561041e-01 6.85196042e-01 9.21510458e-01 -7.03764930e-02
2.73946345e-01 -3.03776085e-01 5.50535440e-01 1.04976706e-02
5.65641880e-01 -5.15689015e-01 -6.88837826e-01 2.27977261e-01
-1.34235060e+00 -1.06878889e+00 6.44419134e-01 2.06307054e+00
7.88766146e-01 3.46005619e-01 -1.22358045e-02 2.75022566e-01
9.10636663e-01 1.53481483e-01 -7.50539005e-01 -4.92552727e-01
-2.92560548e-01 4.47694898e-01 7.85232186e-01 2.59680986e-01
-1.28139877e+00 1.10902619e+00 5.84662485e+00 1.06379700e+00
-1.59749889e+00 5.80100939e-02 8.72857928e-01 -1.53235599e-01
-7.22066239e-02 -4.74027321e-02 -1.06362712e+00 6.06340110e-01
8.91648710e-01 -5.58853894e-02 5.70048392e-01 9.63254631e-01
-2.37368464e-01 4.09867048e-01 -9.89826143e-01 8.24562013e-01
2.31932878e-01 -1.27486885e+00 8.40723291e-02 3.72591615e-01
6.63475573e-01 -5.11608236e-02 6.99333310e-01 2.03417942e-01
2.72584975e-01 -1.13637316e+00 7.05921173e-01 1.08742714e-01
7.75722027e-01 -9.92034137e-01 5.95201075e-01 2.26930112e-01
-6.56765819e-01 -3.90136063e-01 -6.59585178e-01 3.97259295e-01
-2.19443291e-01 1.61087230e-01 -7.98026025e-01 8.66854116e-02
8.23127925e-01 3.79796438e-02 -7.24301577e-01 1.10964119e+00
-3.72305661e-01 9.56268430e-01 -5.40004969e-01 -1.41368613e-01
4.60752193e-03 3.26597273e-01 8.46981049e-01 7.95940220e-01
2.77329683e-01 -1.88768748e-02 2.94476282e-02 9.41040039e-01
-2.44024768e-01 -2.49226689e-01 -6.04315996e-01 5.71923330e-02
8.88668478e-01 8.51068556e-01 -3.75945807e-01 5.70141524e-02
5.09293452e-02 7.92992234e-01 -3.48566696e-02 3.60728800e-01
-1.08244860e+00 -5.00214219e-01 1.02636588e+00 -2.36863002e-01
7.78047264e-01 1.59977078e-01 -1.35805547e-01 -8.24419379e-01
1.89762488e-01 -1.66647005e+00 2.15959415e-01 -3.65781456e-01
-1.31175673e+00 1.09299660e+00 -7.62738287e-03 -1.10383511e+00
-8.39017257e-02 -6.66315138e-01 -6.47386551e-01 6.35834575e-01
-1.21719873e+00 -1.04793942e+00 1.48472637e-01 9.05999124e-01
-1.08640552e-01 -6.04019344e-01 7.65712380e-01 1.56215414e-01
-8.41741800e-01 1.65346944e+00 -5.48146181e-02 4.43344295e-01
4.99379754e-01 -5.06871819e-01 6.32754564e-01 1.16332304e+00
3.80286336e-01 5.40140986e-01 8.14726889e-01 -4.98371780e-01
-1.39427829e+00 -1.16028380e+00 8.61803535e-03 -1.10309862e-01
3.56578588e-01 -4.74777877e-01 -7.86929429e-01 3.91655922e-01
1.92234188e-01 1.70154288e-01 6.15824938e-01 -5.65135002e-01
-7.96079218e-01 -4.59991574e-01 -1.70956743e+00 9.67833042e-01
6.54214084e-01 -7.05412447e-01 -2.93895245e-01 1.67800382e-01
1.08767521e+00 -3.68176699e-01 -5.56967676e-01 6.45815432e-01
4.93477017e-01 -7.94448018e-01 1.20379508e+00 -5.29757261e-01
7.80770257e-02 -1.97353125e-01 -3.62889558e-01 -1.15127110e+00
2.64145918e-02 -9.70914781e-01 -4.20622863e-02 1.18337524e+00
6.02821946e-01 -8.80535066e-01 1.00523436e+00 4.80399072e-01
3.16693097e-01 -7.67159641e-01 -1.30919206e+00 -9.83387291e-01
2.52740741e-01 -4.89524484e-01 1.05496621e+00 6.99294209e-01
-4.35327858e-01 -3.57777596e-01 -4.73624289e-01 8.50538552e-01
6.81469202e-01 -2.87467808e-01 8.88997257e-01 -7.48904109e-01
-6.78986609e-01 -3.96240741e-01 -8.48514020e-01 -7.04581738e-01
3.58892411e-01 -4.85504687e-01 1.78318296e-05 -4.30355698e-01
-1.84494302e-01 -3.15488845e-01 -3.73485565e-01 8.31869781e-01
-1.36078402e-01 4.58969712e-01 3.53438497e-01 -1.85047779e-02
-1.20213702e-02 4.44071889e-01 8.52907479e-01 -6.77191257e-01
-5.26544303e-02 1.69699058e-01 -7.77717829e-01 2.61370540e-01
8.84367883e-01 -8.51804495e-01 -4.45860773e-01 -2.25500435e-01
-1.21080868e-01 -1.66969210e-01 5.50821602e-01 -1.14046967e+00
3.69915217e-01 -1.97162125e-02 1.17679514e-01 -3.13518852e-01
5.08621812e-01 -9.87681389e-01 2.18319237e-01 7.36989975e-01
-3.74530435e-01 1.55117765e-01 4.84209329e-01 6.03204310e-01
-8.36407244e-02 -1.97220102e-01 1.01975763e+00 2.75044471e-01
-3.28086853e-01 5.09059429e-01 -2.10103408e-01 -1.37244686e-01
1.36164665e+00 -2.64474332e-01 -6.44769430e-01 -3.69958490e-01
-2.61053532e-01 -8.70765969e-02 5.02480865e-01 3.49741220e-01
8.34198892e-01 -1.11126244e+00 -6.44655287e-01 5.84842384e-01
-8.96925330e-02 -3.80507171e-01 1.56366885e-01 2.93185204e-01
-5.77613294e-01 1.41071692e-01 -3.27591151e-01 -5.47896206e-01
-1.61081052e+00 1.03326678e+00 7.33618975e-01 -3.83574516e-02
-2.36593574e-01 1.08852684e+00 2.31513709e-01 -3.43059152e-01
6.17196441e-01 7.10021704e-02 7.27674663e-02 -2.70961970e-01
5.32997191e-01 -1.58079006e-02 9.06136539e-03 -6.38430715e-01
-4.37753797e-01 4.24489945e-01 -2.11534172e-01 -9.97190326e-02
1.01939058e+00 4.49949622e-01 -1.25153258e-01 -6.87690377e-01
1.33107042e+00 -2.22927015e-02 -1.24503446e+00 -1.90968543e-01
-3.53808939e-01 -4.11773026e-01 5.02144173e-02 -9.42471862e-01
-1.60754395e+00 7.61689484e-01 9.95686054e-01 9.57169086e-02
1.44985938e+00 -3.56784284e-01 7.91478097e-01 5.06380498e-01
3.42129409e-01 -4.28015649e-01 1.23955920e-01 1.06508046e-01
8.72604012e-01 -7.96089649e-01 -1.69656456e-01 -2.73365021e-01
-4.11139816e-01 7.05784321e-01 7.76063383e-01 -1.93736419e-01
9.07729745e-01 6.10522866e-01 2.85198748e-01 9.37852487e-02
-8.36640596e-01 3.20504814e-01 2.15876222e-01 8.60294044e-01
-6.53458476e-01 -5.28555661e-02 -5.34613729e-02 6.07678175e-01
-3.68253201e-01 -5.13822138e-01 3.16906512e-01 8.04410517e-01
4.99462290e-03 -1.36824262e+00 -7.79410660e-01 7.58591294e-02
-7.54078507e-01 -1.06706001e-01 -5.76347530e-01 5.17718494e-01
2.36211777e-01 1.15410459e+00 -2.95070678e-01 -1.11963749e+00
2.47073188e-01 -2.70639062e-01 3.75298887e-01 -6.32857606e-02
-8.71543229e-01 -3.52076918e-01 -2.32347280e-01 -4.97740179e-01
9.89854783e-02 -3.66585910e-01 -6.22456312e-01 -5.09272873e-01
-5.63419819e-01 3.45903933e-02 7.67875969e-01 7.67608404e-01
6.27242982e-01 1.64209157e-01 1.06265688e+00 -7.42079139e-01
-1.12488341e+00 -6.57335401e-01 -1.78940833e-01 6.37631770e-03
4.16512161e-01 -4.36226308e-01 -5.75757563e-01 -3.90655339e-01]
|
[5.513742446899414, 7.887500762939453]
|
711542b7-6493-4270-9c87-cf6fea679c06
|
netherlands-dataset-a-new-public-dataset-for
|
1904.00770
| null |
http://arxiv.org/abs/1904.00770v1
|
http://arxiv.org/pdf/1904.00770v1.pdf
|
Netherlands Dataset: A New Public Dataset for Machine Learning in Seismic Interpretation
|
Machine learning and, more specifically, deep learning algorithms have seen
remarkable growth in their popularity and usefulness in the last years. This is
arguably due to three main factors: powerful computers, new techniques to train
deeper networks and larger datasets. Although the first two are readily
available in modern computers and ML libraries, the last one remains a
challenge for many domains. It is a fact that big data is a reality in almost
all fields nowadays, and geosciences are not an exception. However, to achieve
the success of general-purpose applications such as ImageNet - for which there
are +14 million labeled images for 1000 target classes - we not only need more
data, we need more high-quality labeled data. When it comes to the Oil&Gas
industry, confidentiality issues hamper even more the sharing of datasets. In
this work, we present the Netherlands interpretation dataset, a contribution to
the development of machine learning in seismic interpretation. The Netherlands
F3 dataset acquisition was carried out in the North Sea, Netherlands offshore.
The data is publicly available and contains pos-stack data, 8 horizons and well
logs of 4 wells. For the purposes of our machine learning tasks, the original
dataset was reinterpreted, generating 9 horizons separating different seismic
facies intervals. The interpreted horizons were used to generate approximatelly
190,000 labeled images for inlines and crosslines. Finally, we present two deep
learning applications in which the proposed dataset was employed and produced
compelling results.
|
['Emilio Vital Brazil', 'Lais Baroni', 'Reinaldo Mozart Silva', 'Rodrigo S. Ferreira', 'Daniel Civitarese', 'Daniela Szwarcman']
|
2019-03-26
| null | null | null | null |
['seismic-interpretation']
|
['miscellaneous']
|
[-6.97120875e-02 2.05083311e-01 5.42178079e-02 -3.16629827e-01
-9.14942741e-01 -5.22029638e-01 5.49034595e-01 2.05711365e-01
-7.62259364e-01 8.52724612e-01 2.64621645e-01 -3.38601232e-01
-3.54379326e-01 -1.13754582e+00 -6.39486849e-01 -9.05196846e-01
-4.88775402e-01 6.33306324e-01 2.50660479e-01 -4.58585143e-01
3.67556155e-01 8.20439339e-01 -1.27551997e+00 4.07721668e-01
4.35664624e-01 1.09099507e+00 1.78796262e-01 3.69360954e-01
-1.27763167e-04 7.70085871e-01 -5.59703708e-01 -4.44185734e-01
3.85803163e-01 1.16146374e-02 -9.36355889e-01 -5.52852713e-02
7.01871887e-02 -2.42613554e-01 -1.09313771e-01 8.13552856e-01
5.35110593e-01 -5.96187487e-02 5.57084501e-01 -7.39346743e-01
-1.43022627e-01 7.95191765e-01 -6.68787360e-01 3.43713313e-01
-2.42627144e-01 -9.17375684e-02 1.17062998e+00 -9.23128068e-01
3.69125664e-01 9.93714213e-01 6.09357595e-01 8.24202225e-02
-6.61954641e-01 -4.78919089e-01 -4.65208560e-01 2.67878413e-01
-1.12571490e+00 -2.68051296e-01 6.13531530e-01 -6.76008224e-01
8.58442664e-01 1.83230698e-01 8.58021975e-01 9.51275587e-01
4.14760299e-02 5.57746649e-01 1.08980954e+00 -4.57398713e-01
4.80639338e-01 -1.20174184e-01 -5.97636737e-02 3.22079360e-01
2.70369828e-01 1.65743634e-01 -3.49190116e-01 -3.86461578e-02
8.28772724e-01 -2.74607856e-02 -2.06156537e-01 2.58901656e-01
-1.23818636e+00 1.02530265e+00 5.22270143e-01 6.48120463e-01
-3.10222059e-01 -6.99986592e-02 6.09182239e-01 3.13983679e-01
6.80527031e-01 4.61601049e-01 -4.73939866e-01 -3.06841701e-01
-9.36748505e-01 4.77379531e-01 7.85959423e-01 4.71561611e-01
8.52730811e-01 3.83755535e-01 7.34615982e-01 9.04104531e-01
1.06924415e-01 4.64715958e-01 6.61692083e-01 -6.95620179e-01
6.52123570e-01 4.89809215e-01 -7.81795606e-02 -1.00726306e+00
-7.33286977e-01 -3.90973240e-01 -1.09462976e+00 3.63343686e-01
5.67986548e-01 -1.98501021e-01 -8.20561707e-01 1.16100097e+00
-1.16655156e-01 -7.41514117e-02 2.52878308e-01 8.75496507e-01
7.57797718e-01 7.31195986e-01 -1.66235149e-01 2.45071232e-01
1.34465051e+00 -5.65495789e-01 -3.94172668e-01 -5.25324285e-01
6.93604350e-01 -6.09678209e-01 7.81054020e-01 5.85665464e-01
-9.37994182e-01 -3.30494225e-01 -1.08180952e+00 1.73330307e-01
-5.83961546e-01 -2.56524205e-01 7.74187267e-01 4.10639137e-01
-8.81145895e-01 6.51480198e-01 -7.87356436e-01 -1.26949862e-01
5.60748160e-01 2.58091211e-01 -6.36607289e-01 1.53552994e-01
-1.37309694e+00 8.53989005e-01 6.74099863e-01 5.05084574e-01
-1.00900769e+00 -3.83099586e-01 -8.28900397e-01 1.33111835e-01
2.04473749e-01 -2.91689765e-02 1.19031310e+00 -8.69570494e-01
-1.00995910e+00 9.52202022e-01 4.34200168e-01 -6.87105954e-01
6.72056735e-01 -2.95142949e-01 -4.01023984e-01 1.46701466e-02
-4.19440642e-02 3.84930104e-01 4.08818841e-01 -1.18011916e+00
-9.03837562e-01 -3.67027611e-01 -4.08781320e-02 -2.98363239e-01
-2.50911117e-01 -3.87098528e-02 -5.40664457e-02 -8.32483590e-01
4.54915799e-02 -6.64358616e-01 -3.72733116e-01 -4.93553221e-01
-2.95221835e-01 -2.38165576e-02 5.90552449e-01 -8.17463994e-01
8.60957921e-01 -2.26510143e+00 1.00030020e-01 3.64978522e-01
1.77286297e-01 2.43098095e-01 1.48683622e-01 5.53536832e-01
-3.04360539e-01 1.88157082e-01 -7.84080386e-01 -8.22261795e-02
-2.26243988e-01 3.94414961e-01 -3.93828452e-01 6.13049150e-01
2.67133027e-01 5.74156284e-01 -6.47816539e-01 -2.19587058e-01
1.38516515e-01 1.94703430e-01 -2.37669542e-01 -3.00183110e-02
-1.26177594e-01 4.21991169e-01 -3.98689955e-01 6.11703336e-01
6.58688962e-01 1.78441331e-02 -1.78517401e-01 -6.53309599e-02
-4.13002610e-01 -9.60866660e-02 -1.00303614e+00 1.41391408e+00
-2.95139015e-01 7.60601044e-01 -2.50186890e-01 -1.49186838e+00
1.16499448e+00 5.39509177e-01 4.61087912e-01 -8.77021968e-01
1.06445290e-01 8.68859351e-01 2.10355714e-01 -8.67343903e-01
5.73405266e-01 -4.56716120e-01 -5.83683178e-02 3.14271629e-01
-2.00152323e-01 -3.24478000e-01 3.85264635e-01 -1.80867061e-01
7.88115799e-01 -1.77533463e-01 1.55976117e-01 -4.41428304e-01
3.47588658e-01 2.35402092e-01 4.52090681e-01 3.24062526e-01
2.74117857e-01 8.80426288e-01 7.31912017e-01 -9.56842244e-01
-1.55536616e+00 -6.90600693e-01 -3.94766837e-01 5.74782789e-01
-4.23917592e-01 5.39214276e-02 -5.17852664e-01 -1.95781544e-01
-2.05631182e-01 2.69296288e-01 -6.52741075e-01 2.11456060e-01
-7.66128302e-01 -9.44171906e-01 7.63475716e-01 5.86223781e-01
6.84611738e-01 -1.23118043e+00 -8.17018032e-01 3.04302365e-01
1.02760643e-01 -9.41895545e-01 5.26591718e-01 4.24863428e-01
-1.04706097e+00 -1.07391465e+00 -1.03261316e+00 -7.99398839e-01
3.35456580e-01 -1.73647493e-01 1.09886205e+00 7.04110339e-02
-1.75114021e-01 -2.79565096e-01 -3.91204923e-01 -6.10678613e-01
-4.14669812e-01 3.48561049e-01 -4.06646997e-01 -1.10313289e-01
1.91313475e-01 -5.08736253e-01 -4.18272108e-01 1.01954669e-01
-1.18899786e+00 -7.18706995e-02 6.01258516e-01 9.56610203e-01
2.19233766e-01 2.86321521e-01 8.32541823e-01 -9.01774287e-01
2.42022455e-01 -6.76122308e-01 -6.85095251e-01 -1.10723920e-01
-1.35230064e-01 -1.37309775e-01 5.70832670e-01 2.55503863e-01
-1.04285324e+00 -2.63764828e-01 -6.53018951e-01 1.01147354e-01
-2.58805573e-01 1.09196031e+00 1.60881892e-01 2.25880474e-01
7.92167127e-01 -1.62855938e-01 -1.00224949e-01 -5.96851051e-01
4.05898696e-04 9.01806474e-01 6.17742598e-01 -5.19457638e-01
6.90451026e-01 6.93169236e-01 5.03027104e-02 -1.23180306e+00
-8.22641790e-01 -3.03841114e-01 -6.09767497e-01 -2.09237397e-01
1.01559734e+00 -9.30825710e-01 -2.78860301e-01 7.57081091e-01
-9.61833298e-01 -3.87676150e-01 -3.10695708e-01 5.37106693e-01
-1.44336447e-01 2.04959452e-01 -6.15940452e-01 -6.39932513e-01
-2.23461181e-01 -1.34871709e+00 9.16404963e-01 2.07943693e-01
5.79482950e-02 -1.27837634e+00 9.12130326e-02 1.99729979e-01
4.67074484e-01 6.14236474e-01 9.18729186e-01 -7.92824864e-01
-3.97030145e-01 -3.83719236e-01 -1.91716179e-01 5.93361080e-01
1.36230424e-01 1.85269434e-02 -1.36607957e+00 -2.92583525e-01
1.11056790e-01 -3.79118621e-01 9.81962562e-01 3.26068968e-01
1.11384428e+00 -2.08303723e-02 1.53826140e-02 4.40833062e-01
1.46432078e+00 3.63154739e-01 9.07022476e-01 1.01199293e+00
6.34123206e-01 9.42734480e-01 3.60271960e-01 7.21531689e-01
1.72676742e-01 1.44992098e-01 6.88948452e-01 -5.40902495e-01
2.94184625e-01 3.29488367e-01 -9.77368653e-02 9.61421847e-01
-7.33031392e-01 -7.89351463e-02 -1.52748764e+00 8.02671969e-01
-1.58410597e+00 -1.04326749e+00 -3.62517625e-01 2.23025918e+00
6.54516876e-01 3.91569167e-01 -6.25906736e-02 5.53009868e-01
3.30316663e-01 3.29031140e-01 -1.28037751e-01 -8.90123397e-02
-4.76282030e-01 4.33449775e-01 6.41408503e-01 3.29570204e-01
-1.13037384e+00 5.97260475e-01 5.62474394e+00 7.03404248e-01
-1.39614952e+00 2.47164723e-02 8.81092012e-01 2.64533460e-01
-4.24057424e-01 -8.01077858e-03 -5.37760019e-01 2.40010574e-01
9.41893995e-01 2.43861601e-02 6.18822277e-02 7.84493804e-01
3.75909865e-01 -2.27070972e-01 -9.06471491e-01 7.67044365e-01
-2.10414842e-01 -1.69518983e+00 -2.99227387e-01 3.07812631e-01
5.91042161e-01 4.74131048e-01 -1.55713663e-01 4.28836234e-02
-8.24542250e-03 -1.06508219e+00 1.00851810e+00 3.75023514e-01
5.02738237e-01 -9.97525990e-01 1.14943123e+00 1.90088958e-01
-9.40163970e-01 -2.15578258e-01 -4.74154115e-01 -2.43961573e-01
2.71442920e-01 8.57894301e-01 -6.39175832e-01 7.17390776e-01
1.02856851e+00 7.75302708e-01 -5.42728186e-01 1.16117275e+00
-1.16368972e-01 7.49926746e-01 -2.99339771e-01 2.33023137e-01
6.58419013e-01 -3.22661459e-01 1.82239279e-01 1.13219440e+00
5.31193972e-01 -1.80918381e-01 -1.26480967e-01 3.96462321e-01
1.68400723e-03 -5.99171370e-02 -6.82498515e-01 -6.46138713e-02
3.16092014e-01 1.25090230e+00 -8.54633033e-01 -2.96078950e-01
-6.33674145e-01 2.38581985e-01 1.45597100e-01 2.77197748e-01
-5.60727358e-01 -3.48866194e-01 2.27194652e-01 1.40098885e-01
-3.08039039e-02 -2.72384733e-01 -3.75536650e-01 -9.04160082e-01
-9.11346003e-02 -1.03652048e+00 4.04792100e-01 -5.97814262e-01
-1.23032701e+00 5.88450193e-01 1.75179541e-01 -1.18029678e+00
-3.15284371e-01 -7.97507226e-01 -5.38851857e-01 7.63102710e-01
-1.65721631e+00 -9.76017416e-01 -4.01428491e-01 2.37597197e-01
6.45291269e-01 -4.50546473e-01 7.69043803e-01 7.77391493e-01
-3.13263625e-01 -7.13663772e-02 4.58410978e-01 3.84052098e-01
4.48208600e-01 -1.20736635e+00 6.25729024e-01 5.74908435e-01
1.69225574e-01 1.50141269e-01 6.35780096e-01 -3.89619738e-01
-1.02602756e+00 -7.77437866e-01 7.48932719e-01 6.57818615e-02
9.99952853e-01 -2.11246476e-01 -1.22264135e+00 8.31239760e-01
2.17209995e-01 1.61374696e-02 5.57510674e-01 -3.26090455e-01
1.14259124e-01 -1.80934682e-01 -8.59983265e-01 3.29640865e-01
2.83856601e-01 -4.00179237e-01 -6.66621923e-01 4.19059396e-01
1.69297591e-01 -4.83853906e-01 -1.06715786e+00 3.31307501e-01
3.19132656e-01 -1.18222857e+00 8.23801577e-01 -2.66009569e-01
8.20994020e-01 -1.59404784e-01 -1.71897233e-01 -1.22802377e+00
3.80711406e-02 -2.02869579e-01 5.11322677e-01 1.08306444e+00
5.96818149e-01 -7.70831287e-01 8.70789707e-01 2.72106737e-01
-5.57831883e-01 -6.93211555e-01 -1.12368357e+00 -6.34378672e-01
4.61582094e-01 -5.91945410e-01 8.06909502e-01 9.65117276e-01
-2.50982314e-01 -4.48795259e-02 -1.83559954e-01 1.59750134e-01
6.01204395e-01 -8.41439217e-02 4.32965755e-01 -1.30868113e+00
-2.45145991e-01 -3.80492598e-01 -4.44640815e-01 -4.07854289e-01
-5.18145636e-02 -6.92060590e-01 -1.08796358e-01 -1.48411119e+00
-1.36214972e-01 -6.57587767e-01 -1.31278196e-02 7.17376709e-01
3.07765216e-01 3.38001549e-01 -1.23408996e-01 4.19399559e-01
3.86968523e-01 2.61025578e-01 1.00777352e+00 -2.30407089e-01
1.63143605e-01 -1.14622831e-01 -1.09916009e-01 9.09519732e-01
9.72064257e-01 -4.08105254e-01 -5.72079122e-02 -8.05002928e-01
6.94296002e-01 -4.55601066e-02 4.02386010e-01 -1.12086260e+00
-8.26019645e-02 5.77590466e-02 3.96107614e-01 -6.60648942e-01
1.56036034e-01 -8.84421408e-01 3.93999785e-01 5.91809094e-01
-3.29066291e-02 1.79752260e-01 1.41250774e-01 1.36609465e-01
-7.92155147e-01 -7.02614784e-01 8.50035191e-01 -5.93912840e-01
-1.10559034e+00 1.69968605e-01 -3.40308577e-01 -9.65538342e-03
1.00588262e+00 -6.19660579e-02 -4.00547981e-01 -9.88165215e-02
-5.64232469e-01 -4.79633808e-02 3.07187557e-01 2.60179371e-01
5.65285027e-01 -9.64165092e-01 -1.01075959e+00 1.56165540e-01
-1.45786218e-02 6.75558269e-01 2.87287176e-01 6.84544921e-01
-1.15705669e+00 1.43078908e-01 -2.23642156e-01 -5.81866920e-01
-6.56282365e-01 5.87074906e-02 3.49487156e-01 -2.03984991e-01
-8.61184478e-01 8.24708462e-01 8.90859365e-02 -1.44152656e-01
-3.77957034e-03 -2.72964895e-01 -4.72867697e-01 4.42642897e-01
6.27336025e-01 3.94076526e-01 4.48027790e-01 -5.25200665e-01
-5.90484887e-02 3.97540033e-01 -7.13139190e-04 -2.07797334e-01
1.98061371e+00 3.80771548e-01 -2.36390144e-01 6.10144556e-01
1.05525780e+00 -1.57963604e-01 -1.06194401e+00 1.14062093e-01
4.80935663e-01 -3.70280683e-01 9.59793702e-02 -3.74667406e-01
-1.47387671e+00 1.09217012e+00 5.42511523e-01 6.00949764e-01
8.50540519e-01 1.44716352e-01 6.95361078e-01 4.45364922e-01
3.80393088e-01 -9.71538424e-01 -8.73822421e-02 6.76218867e-01
1.06236291e+00 -1.32021248e+00 1.10430218e-01 1.78189635e-01
-4.66298014e-01 1.53623807e+00 1.93725437e-01 -1.14935920e-01
5.26108325e-01 5.17962992e-01 1.72127917e-01 -5.60445666e-01
-1.66584596e-01 1.59434259e-01 -1.69534877e-01 4.06479388e-01
6.61215663e-01 -2.55804837e-01 -1.54093862e-01 3.39037895e-01
-4.63136107e-01 -1.36107616e-02 6.47417247e-01 1.24610960e+00
-6.49236858e-01 -1.14932942e+00 -6.25874102e-01 5.59440315e-01
-6.49756491e-01 -4.13187891e-02 1.73903212e-01 1.02266955e+00
1.79407552e-01 5.38897514e-01 2.22767830e-01 -6.18462861e-02
3.19517285e-01 -3.89338493e-01 2.31473777e-03 -4.56373513e-01
-6.71325266e-01 1.02739215e-01 3.22789550e-01 4.26218919e-02
-6.12004578e-01 -5.79852164e-01 -1.43944621e+00 -3.99816513e-01
-2.70636797e-01 3.55675429e-01 8.32837343e-01 1.06407714e+00
-2.98646122e-01 4.10306871e-01 5.60629070e-01 -1.35385406e+00
-4.63455975e-01 -1.12141728e+00 -1.01511955e+00 2.08761215e-01
3.06872159e-01 -5.27969658e-01 -6.32254481e-01 2.26106495e-01]
|
[7.068394184112549, 2.324829339981079]
|
6323c906-2fa2-4a6a-9d9d-25b67d9b162d
|
several-refinements-of-modulation-spectrum
| null | null |
https://aclanthology.org/O15-1010
|
https://aclanthology.org/O15-1010.pdf
|
調變頻譜分解之改良於強健性語音辨識(Several Refinements of Modulation Spectrum Factorization for Robust Speech Recognition) [In Chinese]
| null |
['Hsin-Min Wang', 'Kuan-Yu Chen', 'Hsiao-Tsung Hung', 'Ting-Hao Chang', 'Berlin Chen']
|
2015-10-01
|
several-refinements-of-modulation-spectrum-1
|
https://aclanthology.org/O15-1010
|
https://aclanthology.org/O15-1010.pdf
|
roclingijclclp-2015-10
|
['robust-speech-recognition']
|
['speech']
|
[-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]
|
[-7.1816020011901855, 3.8985209465026855]
|
26e9d7b8-2293-4bc0-a0a0-58151f1a4b5a
|
multi-label-few-shot-learning-for-aspect
|
2105.14174
| null |
https://arxiv.org/abs/2105.14174v1
|
https://arxiv.org/pdf/2105.14174v1.pdf
|
Multi-Label Few-Shot Learning for Aspect Category Detection
|
Aspect category detection (ACD) in sentiment analysis aims to identify the aspect categories mentioned in a sentence. In this paper, we formulate ACD in the few-shot learning scenario. However, existing few-shot learning approaches mainly focus on single-label predictions. These methods can not work well for the ACD task since a sentence may contain multiple aspect categories. Therefore, we propose a multi-label few-shot learning method based on the prototypical network. To alleviate the noise, we design two effective attention mechanisms. The support-set attention aims to extract better prototypes by removing irrelevant aspects. The query-set attention computes multiple prototype-specific representations for each query instance, which are then used to compute accurate distances with the corresponding prototypes. To achieve multi-label inference, we further learn a dynamic threshold per instance by a policy network. Extensive experimental results on three datasets demonstrate that the proposed method significantly outperforms strong baselines.
|
['Zhong Su', 'Renhong Cheng', 'Tiegang Gao', 'Hang Gao', 'Chao Xue', 'Honglei Guo', 'Shiwan Zhao', 'Mengting Hu']
|
2021-05-29
| null |
https://aclanthology.org/2021.acl-long.495
|
https://aclanthology.org/2021.acl-long.495.pdf
|
acl-2021-5
|
['aspect-category-detection']
|
['natural-language-processing']
|
[ 6.87896907e-02 1.02316357e-01 -5.32691240e-01 -6.97247326e-01
-1.19379508e+00 -1.90862179e-01 5.61604202e-01 2.99978077e-01
-3.67970943e-01 3.64244401e-01 3.77458513e-01 7.92526603e-02
9.85320956e-02 -7.97719955e-01 -3.44425738e-01 -6.35664880e-01
4.81434494e-01 5.77751458e-01 1.94083095e-01 -1.34433746e-01
4.05089051e-01 -5.02580628e-02 -1.49099064e+00 3.78608376e-01
6.71621740e-01 9.38558280e-01 9.49311405e-02 3.27263087e-01
-6.39237523e-01 6.35470331e-01 -7.09194958e-01 -6.24696136e-01
-9.76732839e-03 -3.73883694e-01 -6.15819156e-01 3.23179305e-01
1.28814623e-01 -2.56909132e-01 2.62616932e-01 1.18415439e+00
6.67585194e-01 5.33281863e-01 8.16487491e-01 -1.17103505e+00
-6.05788052e-01 5.63424706e-01 -7.74898648e-01 1.60356417e-01
2.06771612e-01 5.21990806e-02 1.62467766e+00 -1.26048529e+00
4.00503784e-01 1.28277647e+00 6.11454844e-01 6.11853123e-01
-9.35941815e-01 -4.43471104e-01 6.38396502e-01 3.03694099e-01
-1.12159574e+00 -2.68705934e-01 9.57947493e-01 -2.27405488e-01
1.00857925e+00 5.52510954e-02 6.74097955e-01 9.02278423e-01
-1.43497065e-01 1.05571270e+00 8.02387059e-01 -4.51603979e-01
6.06147051e-01 4.13613081e-01 7.51815617e-01 5.19088984e-01
2.27888599e-01 -5.18272042e-01 -4.28558260e-01 -3.28955770e-01
1.41750500e-01 3.99897009e-01 -7.93676749e-02 -2.75615841e-01
-8.32009315e-01 1.23727405e+00 2.69417286e-01 2.60323495e-01
-3.39043021e-01 -1.62514582e-01 6.23087287e-01 2.39255026e-01
8.67921531e-01 4.70614195e-01 -4.97914135e-01 -2.21570283e-02
-6.77304447e-01 8.41417760e-02 7.41492271e-01 1.05146146e+00
1.11259067e+00 -3.21295977e-01 -7.92845607e-01 1.16856277e+00
4.29227710e-01 1.45998821e-01 6.28988922e-01 -5.89414537e-01
4.30216879e-01 8.86318684e-01 7.77680725e-02 -9.18359995e-01
-3.76839787e-01 -3.35974246e-01 -5.83559096e-01 -8.05172324e-02
-1.23836398e-01 -2.49500275e-01 -9.72094476e-01 1.30054438e+00
6.18178606e-01 3.37303609e-01 1.17244475e-01 1.00983024e+00
9.10937250e-01 7.60766506e-01 2.77889729e-01 -3.85515481e-01
1.62830222e+00 -1.38419294e+00 -8.60436320e-01 -4.77071524e-01
7.43849635e-01 -6.49736345e-01 1.35106778e+00 -3.45889181e-02
-5.63693404e-01 -4.22858357e-01 -1.06434131e+00 1.02255335e-02
-5.26689470e-01 -1.32881552e-01 6.09253764e-01 4.54328030e-01
-6.07697666e-01 3.73997927e-01 -5.61370313e-01 -2.57769465e-01
5.01032293e-01 1.25233963e-01 7.63022229e-02 -2.48269364e-01
-1.31183994e+00 5.64227581e-01 3.50372642e-01 -2.37039313e-01
-7.24695385e-01 -5.17803550e-01 -1.13489127e+00 2.73481995e-01
4.93662179e-01 -6.06435895e-01 1.39621472e+00 -9.71859336e-01
-1.25913203e+00 8.53087008e-01 -4.15201813e-01 -3.17744553e-01
-5.14736325e-02 -1.53366297e-01 -3.84683549e-01 2.41502166e-01
4.96170491e-01 6.53122246e-01 8.92903507e-01 -1.23559117e+00
-8.18232596e-01 -2.99897075e-01 2.51649201e-01 5.67839682e-01
-5.87610781e-01 1.44523993e-01 -7.18677163e-01 -6.19284153e-01
-1.21434614e-01 -6.62261009e-01 -3.94127786e-01 -1.71314508e-01
-4.41452831e-01 -5.90193629e-01 8.22239518e-01 -1.06147803e-01
1.22681665e+00 -2.17180753e+00 -2.17657387e-01 -1.42761096e-01
1.91718563e-01 3.05968523e-01 -2.28482530e-01 1.23573102e-01
2.28825197e-01 -2.00340912e-01 -2.77783155e-01 -7.20111847e-01
8.31406787e-02 6.99071214e-02 -2.71156460e-01 1.70826659e-01
3.79809916e-01 9.66402888e-01 -1.04825854e+00 -7.14256465e-01
-3.05006634e-02 3.98691356e-01 -4.80695784e-01 4.10368949e-01
-3.60698909e-01 -1.04679987e-01 -6.02928042e-01 7.46101677e-01
5.23128808e-01 -3.44093353e-01 -1.37729540e-01 -2.30152965e-01
1.72443286e-01 1.36060268e-01 -9.60633159e-01 1.56727946e+00
-4.67166811e-01 3.36877137e-01 -3.74327391e-01 -9.95750606e-01
9.86922801e-01 2.19613582e-01 3.53753150e-01 -4.77616161e-01
2.21848607e-01 -5.13831228e-02 -2.38044962e-01 -5.24106324e-01
5.77333868e-01 -6.40580833e-01 -3.83881748e-01 7.49701023e-01
2.56120294e-01 8.82486105e-02 2.14797899e-01 2.10284039e-01
8.41679633e-01 -1.96975499e-01 5.08992672e-01 -1.10188015e-01
4.58145142e-01 -3.34506854e-02 9.00244176e-01 8.17383647e-01
-6.10553265e-01 7.78433323e-01 6.98026538e-01 -4.72157806e-01
-7.83764541e-01 -5.99288940e-01 1.23670809e-01 1.40828550e+00
3.94546926e-01 -4.17816252e-01 -6.98506415e-01 -1.03421855e+00
-2.74061412e-01 1.03279948e+00 -7.91926742e-01 -5.18321574e-01
-1.07478529e-01 -8.58725548e-01 -5.10751344e-02 5.13744235e-01
3.14976841e-01 -1.20353389e+00 -5.67092061e-01 2.78303266e-01
-2.03565165e-01 -8.06513548e-01 -6.85009241e-01 1.63237005e-01
-5.79275310e-01 -9.68425512e-01 -7.74325907e-01 -1.02342379e+00
5.76370001e-01 6.01781964e-01 1.14021862e+00 1.25504553e-03
-1.27535596e-01 2.60442466e-01 -5.37014484e-01 -5.89031219e-01
9.82595608e-02 1.13402292e-01 -1.72891647e-01 3.34661543e-01
1.20779824e+00 -1.78425252e-01 -7.09788322e-01 1.24766760e-01
-6.57537818e-01 -2.55738139e-01 5.91596186e-01 9.37471747e-01
9.57739353e-01 -9.82633084e-02 7.52898932e-01 -1.32274342e+00
8.46732676e-01 -6.51069582e-01 -2.74982631e-01 3.41026574e-01
-7.68090606e-01 -5.88170625e-02 5.45536935e-01 -4.88974035e-01
-1.18709087e+00 -9.09321010e-03 -1.56791642e-01 -4.87869352e-01
-2.48318017e-01 5.22241712e-01 -2.90604621e-01 4.41838533e-01
2.20895499e-01 1.28432706e-01 -2.45941624e-01 -3.89678627e-01
3.70037436e-01 8.47571313e-01 2.38308962e-02 -1.92524508e-01
3.69156420e-01 3.89954388e-01 -4.35699940e-01 -6.95356607e-01
-1.75719702e+00 -1.15560842e+00 -5.43353677e-01 -1.58499375e-01
8.95474553e-01 -1.04325581e+00 -2.48613879e-01 2.02643365e-01
-1.10722744e+00 5.57429902e-02 -4.59290028e-01 4.35047060e-01
-5.04979551e-01 1.73042595e-01 -6.09835565e-01 -8.77831042e-01
-6.90615714e-01 -1.11716783e+00 1.53356063e+00 4.63862181e-01
-2.33865082e-01 -9.16808128e-01 4.06198412e-01 9.61278677e-02
1.43464461e-01 -1.84049502e-01 9.21326160e-01 -1.09500659e+00
-2.18153730e-01 -4.36902761e-01 -1.84884012e-01 7.15963542e-02
4.53979336e-02 -3.16370457e-01 -1.23920918e+00 -1.28696516e-01
1.37903988e-01 -3.56089145e-01 1.13306844e+00 3.39390010e-01
9.34779406e-01 -3.22200209e-01 -3.13587099e-01 3.13302875e-01
1.39577889e+00 5.51074855e-02 4.32089984e-01 3.12701136e-01
7.43070424e-01 5.83942652e-01 1.23562288e+00 5.03453076e-01
3.70273411e-01 5.12766600e-01 2.16647759e-01 6.79140538e-02
1.55889526e-01 -1.53249040e-01 1.26044244e-01 1.02164841e+00
3.30220133e-01 -1.10557213e-01 -6.20314658e-01 6.58406734e-01
-1.96843874e+00 -9.56557274e-01 2.17514932e-01 1.96694958e+00
7.15052187e-01 4.97238278e-01 1.38020515e-01 -1.20096497e-01
1.00984931e+00 4.42970425e-01 -6.59698427e-01 -3.51490438e-01
5.92152476e-02 -2.62123883e-01 -2.10593089e-01 3.91038835e-01
-1.36245644e+00 1.01084566e+00 5.73008633e+00 8.24101985e-01
-7.35955775e-01 4.15218562e-01 7.46468604e-01 -2.47154757e-01
-5.30125856e-01 3.20882536e-02 -1.17556393e+00 3.98080081e-01
7.46298075e-01 -3.39067340e-01 -1.64396718e-01 1.28769696e+00
7.28431949e-03 9.53485295e-02 -7.67688870e-01 8.94684672e-01
4.63686019e-01 -1.03190935e+00 1.48321927e-01 -2.67217338e-01
8.65941584e-01 -1.14083648e-01 -1.47747308e-01 9.27779734e-01
1.60914529e-02 -3.60500693e-01 2.48580575e-01 5.36484897e-01
4.79735732e-01 -1.10222435e+00 8.95419717e-01 3.31309289e-01
-1.15925181e+00 -1.87935606e-01 -9.83506560e-01 4.12504040e-02
2.99567461e-01 8.03582728e-01 -6.25402331e-01 1.02672659e-01
5.46712518e-01 9.69647586e-01 -4.98615891e-01 1.17443860e+00
-2.81705230e-01 5.84384739e-01 2.19576195e-01 -5.27532101e-01
4.65117067e-01 -2.08052501e-01 4.52582151e-01 1.11165690e+00
9.33239758e-02 8.02854747e-02 2.37069070e-01 6.33824885e-01
-2.06341118e-01 4.06037420e-01 -5.75528443e-01 1.39577359e-01
4.08114880e-01 1.70087481e+00 -8.80066156e-01 -6.68307364e-01
-8.38339806e-01 9.19242740e-01 6.74199998e-01 2.43280023e-01
-7.00060308e-01 -7.85222113e-01 8.01222563e-01 -2.87731946e-01
6.06628895e-01 3.97808731e-01 -2.47623384e-01 -1.29600418e+00
-2.21577838e-01 -5.69916844e-01 6.50520205e-01 -6.37343168e-01
-1.58951032e+00 6.22160077e-01 -2.26147160e-01 -1.41896307e+00
-7.72115216e-02 -3.21523786e-01 -1.20081913e+00 5.25948524e-01
-1.60313725e+00 -1.01215780e+00 -1.69385776e-01 2.84785509e-01
1.11094677e+00 -1.49003461e-01 8.84235680e-01 1.70017987e-01
-6.64195657e-01 6.00574553e-01 9.66876559e-03 6.71446994e-02
7.67288148e-01 -1.31736171e+00 3.90495747e-01 6.10050738e-01
9.81949419e-02 5.73860288e-01 6.01652741e-01 -7.08378851e-01
-8.94128859e-01 -1.46037912e+00 1.02751029e+00 -1.89393267e-01
5.92353523e-01 -2.49455258e-01 -1.10053706e+00 6.00061953e-01
2.21682921e-01 2.19941344e-02 1.12133539e+00 4.42586541e-01
-4.37862217e-01 -1.46200210e-01 -9.44190443e-01 5.19745827e-01
5.85044146e-01 -5.53289890e-01 -8.89122307e-01 5.05173445e-01
1.20069587e+00 2.42898211e-01 -6.23070478e-01 2.25707084e-01
2.11382613e-01 -7.25914717e-01 7.32984245e-01 -7.87990570e-01
3.98501128e-01 -1.71981886e-01 -2.65240148e-02 -1.50286412e+00
-4.76597130e-01 -8.12928975e-02 -3.95499200e-01 1.57534325e+00
5.12273610e-01 -3.30081224e-01 6.48911417e-01 6.04329526e-01
-2.17768073e-01 -9.99631822e-01 -7.36715674e-01 -5.02847075e-01
-2.29387760e-01 -2.88294196e-01 7.83437550e-01 8.59211564e-01
2.25893378e-01 1.02418530e+00 -4.24441397e-01 1.33807296e-02
7.47416496e-01 6.21445894e-01 3.95574272e-01 -1.23601007e+00
-4.72151875e-01 -4.47910786e-01 -2.91760832e-01 -7.23036528e-01
3.68713647e-01 -5.68000615e-01 4.76885885e-01 -1.60709727e+00
7.11797178e-01 -1.56001493e-01 -6.39021397e-01 4.75906134e-01
-9.09378946e-01 2.36458585e-01 -1.30001709e-01 2.09944591e-01
-1.39258003e+00 8.61543715e-01 1.02290809e+00 -4.42530960e-01
-6.97664469e-02 2.98132867e-01 -1.00655532e+00 7.84103155e-01
9.19227421e-01 -6.64257586e-01 -4.79035556e-01 -1.12649262e-01
1.67346686e-01 -3.16173077e-01 -2.45842159e-01 -7.82216847e-01
1.66379943e-01 -2.90202890e-02 1.77309245e-01 -7.08756864e-01
4.18849319e-01 -6.22330546e-01 -6.02886438e-01 3.47452790e-01
-5.81938505e-01 -3.60581249e-01 -3.05745840e-01 9.52592611e-01
-3.72692049e-01 -6.44243479e-01 8.31692815e-01 -2.85615474e-01
-9.82847035e-01 5.12611806e-01 -2.46753797e-01 3.23554188e-01
1.12316298e+00 2.23266989e-01 -2.95375049e-01 -3.34207118e-01
-5.03801823e-01 4.29189533e-01 2.94249356e-01 6.21070206e-01
7.05627620e-01 -1.51582360e+00 -3.61958593e-01 4.07808982e-02
8.07474792e-01 -3.36401686e-02 3.23403180e-01 6.79088295e-01
1.00157201e-01 2.62513369e-01 1.83665559e-01 -3.51034462e-01
-1.13320339e+00 1.04030943e+00 1.41035050e-01 -2.64101416e-01
-5.59051812e-01 7.70917475e-01 2.87121415e-01 -6.14619076e-01
2.72763848e-01 2.65375137e-01 -8.42356265e-01 4.98998195e-01
1.00498652e+00 1.09499842e-01 -2.12160647e-02 -6.27435505e-01
-2.90933400e-01 6.02831542e-01 -4.69988734e-01 9.92046446e-02
1.32563591e+00 -3.97089869e-01 8.41119438e-02 8.96681190e-01
1.45296979e+00 -3.68416160e-01 -1.20539820e+00 -4.80163366e-01
1.18019365e-01 -3.45826000e-01 -3.84569392e-02 -3.78261417e-01
-9.59635913e-01 1.00773048e+00 3.65689725e-01 2.82628328e-01
9.93619859e-01 2.03124940e-01 8.00758779e-01 5.21141589e-01
6.93458244e-02 -1.31028736e+00 3.00522417e-01 5.52126110e-01
3.15198064e-01 -1.53230131e+00 -1.58854023e-01 -3.20166320e-01
-1.02949238e+00 7.96834648e-01 1.03582716e+00 -1.29110471e-03
7.22256541e-01 -8.18626434e-02 2.56209165e-01 -4.19989705e-01
-9.86154556e-01 -5.42075515e-01 2.84828722e-01 5.19891560e-01
3.01407099e-01 -4.28635329e-02 -4.03592050e-01 9.14014161e-01
3.37532938e-01 -3.92065704e-01 4.04807568e-01 9.38368082e-01
-9.74043727e-01 -8.04284155e-01 -4.29101810e-02 8.61445665e-01
-4.03953463e-01 -2.01733425e-01 -2.49105245e-01 2.45603383e-01
-1.51054516e-01 9.82178688e-01 1.32851258e-01 -3.51410568e-01
4.26734000e-01 2.78453290e-01 -2.23515734e-01 -1.05412579e+00
-4.54426289e-01 2.98957199e-01 3.93707398e-03 -3.16860497e-01
-4.67016965e-01 -5.39909482e-01 -1.03702021e+00 2.38392740e-01
-5.74981153e-01 5.07092893e-01 3.73794317e-01 1.07012343e+00
3.38941544e-01 5.25323153e-01 8.78926933e-01 -5.77241302e-01
-5.69392383e-01 -1.02061164e+00 -7.57740974e-01 5.38347065e-01
1.45110101e-01 -7.33184397e-01 -4.21882421e-01 -2.42965341e-01]
|
[11.272027969360352, 6.599706649780273]
|
4b05acac-ebfe-4452-a093-17f9b21ee91b
|
non-adaptive-adaptive-sampling-on-turnstile
|
2004.10969
| null |
https://arxiv.org/abs/2004.10969v1
|
https://arxiv.org/pdf/2004.10969v1.pdf
|
Non-Adaptive Adaptive Sampling on Turnstile Streams
|
Adaptive sampling is a useful algorithmic tool for data summarization problems in the classical centralized setting, where the entire dataset is available to the single processor performing the computation. Adaptive sampling repeatedly selects rows of an underlying matrix $\mathbf{A}\in\mathbb{R}^{n\times d}$, where $n\gg d$, with probabilities proportional to their distances to the subspace of the previously selected rows. Intuitively, adaptive sampling seems to be limited to trivial multi-pass algorithms in the streaming model of computation due to its inherently sequential nature of assigning sampling probabilities to each row only after the previous iteration is completed. Surprisingly, we show this is not the case by giving the first one-pass algorithms for adaptive sampling on turnstile streams and using space $\text{poly}(d,k,\log n)$, where $k$ is the number of adaptive sampling rounds to be performed. Our adaptive sampling procedure has a number of applications to various data summarization problems that either improve state-of-the-art or have only been previously studied in the more relaxed row-arrival model. We give the first relative-error algorithms for column subset selection, subspace approximation, projective clustering, and volume maximization on turnstile streams that use space sublinear in $n$. We complement our volume maximization algorithmic results with lower bounds that are tight up to lower order terms, even for multi-pass algorithms. By a similar construction, we also obtain lower bounds for volume maximization in the row-arrival model, which we match with competitive upper bounds. See paper for full abstract.
|
['Samson Zhou', 'Ilya Razenshteyn', 'David P. Woodruff', 'Sepideh Mahabadi']
|
2020-04-23
| null | null | null | null |
['data-summarization']
|
['miscellaneous']
|
[ 5.08633256e-01 7.20895082e-02 -1.61982045e-01 7.83440918e-02
-1.07253063e+00 -7.51496792e-01 7.72034302e-02 9.17781413e-01
-5.71975350e-01 6.71327710e-01 1.42674670e-01 -1.69470385e-01
-4.78321671e-01 -9.18734133e-01 -8.88707638e-01 -8.88661563e-01
-5.82352221e-01 1.07364833e+00 2.89108515e-01 -1.34751871e-01
4.16739702e-01 5.37952781e-01 -1.46869874e+00 2.39292026e-01
6.31506681e-01 1.07125628e+00 -8.20530280e-02 1.11272383e+00
-2.74015605e-01 3.60198587e-01 -4.44865167e-01 -2.42973655e-01
5.57447493e-01 -6.57117963e-01 -9.22584116e-01 2.55431235e-01
1.91676438e-01 -2.56731421e-01 -1.44357622e-01 9.41675842e-01
3.92712414e-01 2.29547739e-01 5.26828349e-01 -9.49374437e-01
3.22322637e-01 1.08115304e+00 -1.01614690e+00 3.31843227e-01
3.02474737e-01 -3.24954271e-01 8.76763642e-01 -5.93167365e-01
6.60551369e-01 8.47146034e-01 2.88752973e-01 3.96737903e-02
-1.38845193e+00 -4.46542203e-01 2.00333819e-01 -1.26224563e-01
-1.30818355e+00 -4.61066693e-01 4.00976390e-01 -1.27496406e-01
7.69157946e-01 9.11612511e-01 6.48994923e-01 -2.66613048e-02
-3.45272928e-01 9.23157275e-01 6.48952723e-01 -5.24675131e-01
5.39695203e-01 -1.28950417e-01 3.75885278e-01 6.01833105e-01
8.01338255e-01 -6.59385681e-01 -7.01099813e-01 -6.33058310e-01
3.38200092e-01 3.70998867e-02 -3.02902579e-01 -5.54232359e-01
-1.27234566e+00 7.84064710e-01 -9.07117873e-02 1.19209900e-01
-3.69636804e-01 3.23397338e-01 6.99434996e-01 4.10224169e-01
5.15839398e-01 2.82496154e-01 -4.98463631e-01 -1.72694072e-01
-1.41681457e+00 4.27139014e-01 1.16646445e+00 1.28955448e+00
6.94938838e-01 -1.80164129e-01 -1.10482417e-01 6.52610838e-01
-4.79815632e-01 7.48354971e-01 2.53953822e-02 -1.16426802e+00
8.08411956e-01 4.97945130e-01 1.06525503e-01 -7.01897085e-01
-4.19532180e-01 -4.26001340e-01 -1.15949154e+00 -3.26912820e-01
6.85106397e-01 -1.65834606e-01 -2.61896580e-01 1.50761068e+00
4.64756489e-01 -3.50199312e-01 -9.89179984e-02 4.62592602e-01
-3.83893331e-03 8.80501032e-01 -5.52342176e-01 -1.20051563e+00
1.23426592e+00 -5.00647426e-01 -2.54296541e-01 1.68043718e-01
7.84209371e-01 -5.35067797e-01 7.08448172e-01 8.05170000e-01
-1.58231020e+00 7.02098086e-02 -8.05340350e-01 2.87434101e-01
8.71638060e-02 -6.93028942e-02 4.97497559e-01 7.35333204e-01
-9.77344275e-01 5.91754019e-01 -8.85097384e-01 -4.37283665e-01
4.62012023e-01 4.61730629e-01 -3.18560936e-02 -1.96147352e-01
-3.48012596e-01 -6.54348806e-02 2.74919331e-01 -2.70153373e-01
-5.18054903e-01 -6.58083498e-01 -4.31825191e-01 2.32843429e-01
8.13363850e-01 -6.97246850e-01 1.00673187e+00 -5.99539816e-01
-1.05307460e+00 5.40659249e-01 -6.06036067e-01 -6.76550567e-01
5.49578726e-01 -6.65609315e-02 5.41307688e-01 5.04306495e-01
-5.82894161e-02 -1.18965708e-01 4.35722917e-01 -1.04533887e+00
-6.27121508e-01 -8.29916298e-01 -2.80675795e-02 4.33752149e-01
-2.97345787e-01 -3.00525539e-02 -4.91158783e-01 -3.79510462e-01
3.77505034e-01 -9.60291743e-01 -6.90867960e-01 -4.57881451e-01
-4.96135831e-01 -1.00008257e-01 1.83806211e-01 -8.58539492e-02
1.54533696e+00 -2.05464935e+00 3.57839525e-01 6.68237388e-01
3.32767427e-01 -2.14744527e-02 3.74524772e-01 9.74802613e-01
2.49553889e-01 5.79019785e-02 -5.22325575e-01 -2.48271033e-01
3.74961011e-02 -1.19629458e-01 -5.25632024e-01 7.86855519e-01
-7.10044324e-01 2.41870686e-01 -7.86944687e-01 -4.74153012e-01
-1.92070290e-01 -2.55879760e-01 -8.68134081e-01 -1.39987245e-01
-2.92043865e-01 -2.94719160e-01 -2.56949872e-01 1.95842475e-01
9.08070683e-01 -3.57509702e-01 5.33354580e-01 2.56159399e-02
-8.67891535e-02 -2.26366837e-02 -1.81928432e+00 1.51934159e+00
-3.01139057e-01 3.14028233e-01 5.62514961e-01 -1.19961262e+00
2.88905740e-01 6.43445626e-02 8.90279412e-01 -9.12846327e-02
8.71740654e-02 5.41800082e-01 -4.58416522e-01 -1.05986096e-01
8.42047393e-01 -9.79601368e-02 -4.16370451e-01 8.78577232e-01
-3.69942248e-01 -9.58595425e-02 6.86057508e-01 7.67143488e-01
1.40940559e+00 -5.41993797e-01 3.81353140e-01 -3.82421255e-01
3.02663356e-01 1.93293184e-01 4.00250256e-01 1.13957071e+00
2.31101424e-01 6.11436188e-01 1.03300714e+00 -1.68406174e-01
-1.06426632e+00 -7.95740068e-01 -4.25560866e-03 1.15347600e+00
2.54686147e-01 -7.28082180e-01 -1.02994287e+00 -3.13005894e-01
-9.15918350e-02 5.57009161e-01 -5.12103856e-01 2.76391685e-01
-7.07979798e-01 -9.52746272e-01 2.49841452e-01 2.71905631e-01
2.04242408e-01 -6.18873835e-01 -9.68767583e-01 3.46833140e-01
6.81538060e-02 -8.43309641e-01 -6.78313076e-01 3.79928917e-01
-1.29965389e+00 -1.16048670e+00 -6.53571844e-01 -3.41085702e-01
8.56813788e-01 4.48458999e-01 8.05389583e-01 -4.71671112e-02
-1.08010471e-01 4.25886691e-01 -3.45792949e-01 -3.83065134e-01
-2.73208618e-02 3.20984840e-01 8.97859484e-02 -1.51558369e-01
7.46022165e-02 -5.20567834e-01 -8.28801334e-01 -1.09919747e-02
-1.31410682e+00 -1.33948341e-01 2.99584180e-01 7.05614209e-01
8.58377755e-01 1.30908892e-01 2.47004122e-01 -1.40213072e+00
4.73585993e-01 -4.89507794e-01 -6.35026932e-01 6.41523600e-02
-3.17724824e-01 1.89636528e-01 9.32363391e-01 -5.80887347e-02
-4.38983560e-01 2.75704801e-01 1.62944347e-01 -1.70810670e-01
1.87128544e-01 4.27962273e-01 -4.27331449e-03 4.67736691e-01
5.07331073e-01 5.60783088e-01 1.38952017e-01 -3.08396816e-01
4.22703087e-01 4.51907486e-01 2.63274491e-01 -5.87286651e-01
5.45517862e-01 8.10164452e-01 4.36773717e-01 -1.11677170e+00
-4.86781031e-01 -6.84269905e-01 -4.17482018e-01 1.15323894e-01
1.28622681e-01 -7.41254568e-01 -1.04450715e+00 1.10970482e-01
-8.19817841e-01 -2.75202602e-01 -8.09630394e-01 2.82670110e-01
-7.82673836e-01 6.34721100e-01 -3.89863819e-01 -1.07446849e+00
-4.86371845e-01 -8.70238900e-01 8.34226251e-01 -1.55250371e-01
-6.54222742e-02 -4.27033395e-01 -6.09600320e-02 1.46553442e-01
2.68272251e-01 1.94599003e-01 1.01173222e+00 -9.30964947e-01
-7.45601773e-01 -3.62339795e-01 -9.60875023e-03 -3.67279164e-02
-3.09185028e-01 -3.61785054e-01 -2.98150271e-01 -5.26190996e-01
-3.15538831e-02 1.09230861e-01 1.03514600e+00 5.72367847e-01
1.50703835e+00 -7.64389455e-01 -6.26857758e-01 4.78597254e-01
1.62762845e+00 -1.29429653e-01 4.10823047e-01 -2.01966260e-02
3.37557316e-01 3.41467112e-01 4.97751445e-01 1.13730192e+00
6.42040670e-02 4.48024482e-01 2.69370198e-01 2.48846889e-01
4.07889515e-01 1.96975753e-01 2.16579929e-01 7.79075205e-01
9.32700112e-02 -4.61314797e-01 -5.80304146e-01 9.30098355e-01
-1.82402873e+00 -1.02317429e+00 -3.66048962e-01 2.85135531e+00
8.30039382e-01 2.34887645e-01 7.19945550e-01 4.74634647e-01
5.72172403e-01 7.56363049e-02 -4.07501787e-01 -6.91688955e-01
-9.50416997e-02 5.75018227e-01 1.15351343e+00 5.32583892e-01
-7.67142951e-01 4.64498222e-01 5.46184444e+00 1.10434449e+00
-5.80576241e-01 9.66641679e-03 6.98257983e-01 -9.06402469e-01
-6.05589211e-01 1.13451019e-01 -6.91101134e-01 4.88656789e-01
1.05271339e+00 -3.76186371e-01 5.65561235e-01 8.39803100e-01
1.74967974e-01 -7.28204846e-01 -1.26667154e+00 9.79256094e-01
1.09781429e-01 -1.52285969e+00 -8.55333731e-02 4.18962032e-01
8.55477750e-01 -2.65033245e-01 -1.67722836e-01 -1.59274101e-01
3.29887897e-01 -5.25121331e-01 4.96012688e-01 1.03508279e-01
7.69188225e-01 -1.12372637e+00 4.83473361e-01 6.11397743e-01
-1.25895309e+00 -1.83400795e-01 -3.29796910e-01 4.69319038e-02
2.93606400e-01 9.39933717e-01 -6.77897811e-01 7.94479072e-01
5.71872711e-01 -1.52287215e-01 -1.36300186e-02 9.98118043e-01
8.06156635e-01 7.93485224e-01 -1.14179659e+00 -3.04591179e-01
7.42096230e-02 -2.95649588e-01 8.23999345e-01 1.26538515e+00
5.03951252e-01 5.87757349e-01 1.59638360e-01 -8.05912074e-03
-3.46849084e-01 5.70770204e-01 -6.79650784e-01 2.98609525e-01
6.73942268e-01 7.80848563e-01 -1.16287386e+00 -5.07159173e-01
-1.98313221e-01 8.32658291e-01 1.08939409e-01 -2.33870931e-02
-4.13287669e-01 -9.24487829e-01 3.71584713e-01 6.83103442e-01
7.05179572e-01 -5.25011897e-01 -6.79742992e-01 -8.89051199e-01
2.96556264e-01 -6.37208223e-01 6.98089778e-01 1.80067960e-02
-7.87366033e-01 3.21597695e-01 4.31334287e-01 -1.11057627e+00
-2.16546282e-01 -9.87700820e-02 -1.40381441e-01 4.97026354e-01
-8.35121393e-01 -2.90064424e-01 1.31077677e-01 4.08879429e-01
5.66742301e-01 -4.78310976e-03 5.85226476e-01 -4.52199019e-02
-3.78285199e-01 7.43136346e-01 7.55718172e-01 -3.18100810e-01
2.49173164e-01 -1.18995428e+00 -1.36396065e-01 1.02691424e+00
-5.38841672e-02 5.45744538e-01 9.48282063e-01 -3.99937481e-01
-1.93594944e+00 -7.43538201e-01 8.04722786e-01 -1.10853910e-01
2.66625285e-01 -3.69924337e-01 -5.73736548e-01 3.71731281e-01
1.78466097e-01 -2.44143084e-01 8.15992594e-01 6.04756065e-02
7.77955875e-02 -5.65664411e-01 -1.12720096e+00 4.08669114e-01
1.07363391e+00 2.90950704e-02 3.33900936e-02 4.78188545e-01
4.68660802e-01 -4.30192947e-01 -7.27963865e-01 1.16438940e-01
3.53092134e-01 -8.56265604e-01 7.49250054e-01 -2.20734254e-01
1.95146635e-01 -2.69580990e-01 -4.06883985e-01 -8.98972511e-01
1.43169433e-01 -1.03220546e+00 -2.00592175e-01 9.93594110e-01
4.98092383e-01 -3.63537431e-01 1.00673151e+00 4.49576616e-01
1.21948503e-01 -8.88136029e-01 -1.00109756e+00 -4.90322560e-01
3.78047153e-02 -5.10361910e-01 5.59184253e-01 4.64320451e-01
3.58255506e-01 1.56032652e-01 -1.22876242e-01 1.74663458e-02
8.45762968e-01 4.61318225e-01 1.03695786e+00 -9.42206085e-01
-5.54948688e-01 -4.63229120e-01 2.21376717e-02 -1.43311989e+00
-5.58872223e-01 -8.29143882e-01 -1.78910375e-01 -1.47204196e+00
5.54885685e-01 -6.25670671e-01 1.21685684e-01 6.02655523e-02
4.91214171e-02 -6.11710697e-02 2.28704393e-01 2.83013433e-01
-1.03162432e+00 2.23487064e-01 7.56936312e-01 1.24839552e-01
-5.77907324e-01 2.57438272e-01 -7.42282033e-01 4.81322765e-01
5.94739318e-01 -4.89187777e-01 -4.25425768e-01 -2.86814064e-01
7.05929756e-01 6.12790585e-01 -2.70380408e-01 -9.65632141e-01
4.71165091e-01 -2.41927326e-01 3.23023647e-02 -1.04018867e+00
1.40455142e-01 -6.85760915e-01 1.78639978e-01 5.49129009e-01
-5.27893603e-01 1.67213067e-01 1.45430202e-02 9.66006577e-01
1.07143708e-01 -3.19122374e-01 7.44654775e-01 -2.20811203e-01
1.51052490e-01 5.03799736e-01 -5.29178202e-01 2.90274590e-01
1.17616594e+00 -4.29376513e-01 -2.39661522e-02 -5.54485500e-01
-5.90044200e-01 3.39535356e-01 5.60582757e-01 -5.91522813e-01
3.74138147e-01 -8.52832079e-01 -8.16979825e-01 -5.52866645e-02
-7.03740194e-02 6.25911176e-01 5.89250088e-01 1.03745139e+00
-8.19702923e-01 3.08569551e-01 3.77908379e-01 -6.70141697e-01
-1.32805860e+00 5.53350151e-01 -1.98854044e-01 -4.61875409e-01
-3.75549972e-01 8.48155856e-01 -4.65392973e-03 3.10134470e-01
2.43211433e-01 -3.30028176e-01 3.98324877e-01 2.48803571e-01
4.70933259e-01 8.14624786e-01 2.55153179e-01 -1.64716616e-01
-3.58538002e-01 2.51215190e-01 -2.32848659e-01 -3.77350569e-01
1.48828053e+00 -3.60266745e-01 -3.17491323e-01 3.73342931e-01
1.15485823e+00 3.70380521e-01 -8.56765866e-01 -4.11076546e-01
-3.02932978e-01 -4.30769712e-01 -4.67570901e-01 -1.59510612e-01
-1.00354600e+00 5.52304506e-01 2.36677691e-01 7.34386623e-01
1.45576775e+00 1.47124723e-01 7.37752140e-01 4.29995626e-01
6.60803735e-01 -1.24718475e+00 -2.27785781e-01 3.96067768e-01
5.72496057e-01 -5.90210319e-01 4.59428847e-01 -3.92156571e-01
-4.81708735e-01 1.13362932e+00 4.75212820e-02 -3.36194873e-01
6.53168619e-01 3.46272111e-01 -7.83057809e-01 -8.98339823e-02
-8.97604942e-01 -7.56704286e-02 -4.10591364e-01 -2.96367169e-03
2.55615085e-01 2.33850256e-01 -8.50235105e-01 5.62984586e-01
-3.09919149e-01 -1.67879388e-01 8.44934404e-01 1.01110768e+00
-7.44273126e-01 -1.25715959e+00 -4.38586384e-01 1.05056477e+00
-6.77738965e-01 -1.00235887e-01 -8.47039670e-02 5.54957688e-01
-3.24588478e-01 7.56159186e-01 2.01398000e-01 -3.03429943e-02
3.33289355e-01 1.40249148e-01 5.95840156e-01 -4.66102749e-01
-6.28193796e-01 3.17896843e-01 6.92108050e-02 -3.39603454e-01
-7.92928264e-02 -1.21550691e+00 -1.32794487e+00 -6.59962952e-01
-1.66255966e-01 5.13138115e-01 5.35052896e-01 6.51954412e-01
4.87350643e-01 6.67086020e-02 9.32284892e-01 -6.44359112e-01
-6.27019286e-01 -7.84862995e-01 -8.83987248e-01 6.48154989e-02
1.49756193e-01 9.14283618e-02 -2.79371679e-01 4.31316160e-03]
|
[6.594390869140625, 4.870911598205566]
|
e6a25784-22f6-45b8-80d3-e8a1f657fa9b
|
multi-oriented-text-detection-and
|
1707.07150
| null |
http://arxiv.org/abs/1707.07150v2
|
http://arxiv.org/pdf/1707.07150v2.pdf
|
Multi-Oriented Text Detection and Verification in Video Frames and Scene Images
|
In this paper, we bring forth a novel approach of video text detection using
Fourier-Laplacian filtering in the frequency domain that includes a
verification technique using Hidden Markov Model (HMM). The proposed approach
deals with the text region appearing not only in horizontal or vertical
directions, but also in any other oblique or curved orientation in the image.
Until now only a few methods have been proposed that look into curved text
detection in video frames, wherein lies our novelty. In our approach, we first
apply Fourier-Laplacian transform on the image followed by an ideal
Laplacian-Gaussian filtering. Thereafter K-means clustering is employed to
obtain the asserted text areas depending on a maximum difference map. Next, the
obtained connected components (CC) are skeletonized to distinguish various text
strings. Complex components are disintegrated into simpler ones according to a
junction removal algorithm followed by a concatenation performed on possible
combination of the disjoint skeletons to obtain the corresponding text area.
Finally these text hypotheses are verified using HMM-based text/non-text
classification system. False positives are thus eliminated giving us a robust
text detection performance. We have tested our framework in multi-oriented text
lines in four scripts, namely, English, Chinese, Devanagari and Bengali, in
video frames and scene texts. The results obtained show that proposed approach
surpasses existing methods on text detection.
|
['Umapada Pal', 'Ayan Kumar Bhunia', 'Aneeshan Sain', 'Partha Pratim Roy']
|
2017-07-22
| null | null | null | null |
['curved-text-detection']
|
['computer-vision']
|
[ 6.55870259e-01 -2.86828339e-01 1.81242824e-01 2.85840444e-02
-3.68794411e-01 -6.28156304e-01 7.85107434e-01 1.97922736e-01
-4.04063612e-01 4.97515231e-01 -8.78632739e-02 -1.92175075e-01
-6.33700490e-02 -6.34189606e-01 -4.12350714e-01 -7.59955347e-01
2.99710274e-01 4.82016027e-01 6.71571493e-01 1.48985060e-02
6.88323975e-01 5.87728322e-01 -1.51286149e+00 2.23879308e-01
7.16544092e-01 2.81355560e-01 6.11826144e-02 1.17876363e+00
-4.18964088e-01 4.79182273e-01 -5.99627376e-01 -3.34829986e-01
1.73721462e-02 -7.06601799e-01 -7.28003085e-01 7.26024330e-01
2.19124943e-01 -2.90831059e-01 9.99431163e-02 1.08475411e+00
3.35693151e-01 -5.84626978e-04 1.17398965e+00 -8.37222397e-01
-8.00067559e-02 5.64926803e-01 -1.03687012e+00 2.53346443e-01
7.18559086e-01 -5.29481508e-02 6.25720918e-01 -9.43873882e-01
7.71907687e-01 1.08387053e+00 5.37921846e-01 1.12277627e-01
-1.13302922e+00 -2.42070213e-01 -1.86119109e-01 2.51545459e-01
-1.39561975e+00 -4.23727721e-01 7.67437398e-01 -5.80286562e-01
8.87409270e-01 3.20211500e-01 5.89070082e-01 7.19446599e-01
2.88125694e-01 7.20014393e-01 9.72871542e-01 -9.99952793e-01
2.27367003e-02 1.79169253e-01 3.20619196e-01 7.44837999e-01
3.12053710e-01 -6.79997563e-01 -3.26558977e-01 4.31945838e-04
4.21689928e-01 -9.33994725e-02 -2.42181987e-01 -3.46607231e-02
-1.34352887e+00 4.87669706e-01 -4.83607620e-01 1.00798821e+00
-2.85724699e-01 -3.57753277e-01 4.54251349e-01 -6.37389719e-02
1.20902270e-01 -3.48593801e-01 4.75753881e-02 -1.72165811e-01
-1.55517864e+00 -1.69393010e-02 7.65602410e-01 8.08099627e-01
6.56568348e-01 4.06085700e-02 3.62353832e-01 4.79321927e-01
3.32753628e-01 4.95158970e-01 5.78176379e-01 -8.46716985e-02
6.54902697e-01 6.52751029e-01 -8.95260200e-02 -1.21835625e+00
-2.65529990e-01 -1.20491385e-02 -7.15465724e-01 2.81264871e-01
4.88932371e-01 -2.91151375e-01 -7.73936570e-01 8.39687467e-01
4.83943433e-01 -3.30756530e-02 1.45322368e-01 5.28076470e-01
4.92557436e-01 8.21073890e-01 -2.59992212e-01 -4.54551548e-01
1.45451307e+00 -6.19722724e-01 -1.10001659e+00 4.63170528e-01
4.74508703e-01 -1.47634006e+00 7.80145288e-01 8.49242389e-01
-9.69364285e-01 -4.43836182e-01 -1.05193532e+00 2.19491705e-01
-5.06490946e-01 7.23939002e-01 -2.09516689e-01 8.74068797e-01
-7.61308610e-01 3.61427397e-01 -8.99739027e-01 -9.22933519e-01
-2.43725941e-01 2.82204151e-01 -2.65203506e-01 4.32940125e-01
-6.04992390e-01 5.16102612e-01 4.66853648e-01 2.88405746e-01
-2.22496703e-01 4.61650848e-01 -4.65276212e-01 -9.30287242e-02
2.71998167e-01 -2.80980200e-01 5.93020141e-01 -1.28155625e+00
-1.54039109e+00 7.60056138e-01 -3.55853111e-01 -2.16917366e-01
8.66824210e-01 1.08702883e-01 -6.46632135e-01 6.43468976e-01
-5.35619371e-02 1.66675925e-01 1.34377265e+00 -1.02685797e+00
-6.94523096e-01 -3.12847346e-01 -5.54292142e-01 5.00859380e-01
-3.62011671e-01 2.28490025e-01 -6.66612923e-01 -6.67500079e-01
5.62786341e-01 -6.54312313e-01 3.40863526e-01 -3.87124121e-01
-7.56110668e-01 -1.91898748e-01 1.30060184e+00 -9.93186355e-01
1.31145000e+00 -2.05914187e+00 1.12006828e-01 4.56523389e-01
-6.77441806e-02 1.40332609e-01 4.88847077e-01 6.79466784e-01
5.41610494e-02 7.64962733e-02 -2.44059294e-01 -1.85259402e-01
-2.66409516e-01 -6.25163391e-02 7.25012049e-02 7.44712055e-01
1.99598130e-02 4.85899113e-02 -2.63900608e-01 -1.31989610e+00
4.76433724e-01 4.49290901e-01 -7.03719631e-02 -2.62953460e-01
6.85157403e-02 3.00630629e-01 -4.94485319e-01 6.81926906e-01
8.63381147e-01 3.52643847e-01 1.70921892e-01 -2.88776040e-01
-5.26455760e-01 -4.21055198e-01 -1.74976909e+00 1.02966011e+00
7.05633387e-02 8.10665309e-01 6.23514503e-02 -1.14651573e+00
1.02330625e+00 5.29174089e-01 4.37673748e-01 8.25677738e-02
3.82241786e-01 1.71522692e-01 -2.28363484e-01 -7.84768999e-01
6.65352225e-01 2.34610036e-01 4.59578156e-01 3.30992520e-01
-7.12133422e-02 -4.75924537e-02 6.41334236e-01 3.43276076e-02
5.00533879e-01 2.76556283e-01 3.57167274e-01 -1.37587115e-01
1.18323457e+00 5.56533486e-02 -7.39909932e-02 5.33165216e-01
-1.23517543e-01 5.17106533e-01 6.44060433e-01 5.02041541e-02
-1.02584779e+00 -6.22837305e-01 -3.57338101e-01 6.54481232e-01
-6.69457614e-02 -1.63683072e-01 -1.14196897e+00 -5.37009299e-01
-3.70217532e-01 3.44440699e-01 -3.42002243e-01 4.87952381e-01
-6.61097348e-01 -7.64448881e-01 6.20772600e-01 -1.75158918e-01
8.13321888e-01 -1.01017332e+00 -6.20109141e-01 5.78119792e-02
-1.75502509e-01 -1.06780088e+00 -2.64790744e-01 -9.28620622e-03
-1.06522083e+00 -1.03097367e+00 -1.02061284e+00 -9.81417596e-01
7.96532750e-01 2.17490777e-01 3.29247802e-01 -1.19781597e-02
-5.10281920e-01 4.96565640e-01 -5.20678163e-01 6.74122795e-02
-8.05757523e-01 -1.14085324e-01 -2.07569167e-01 4.29491907e-01
3.31221461e-01 -2.34234810e-01 -2.51421481e-01 3.91642451e-01
-8.86777699e-01 -1.71186894e-01 5.93779027e-01 4.26980197e-01
2.84205705e-01 6.21883154e-01 1.28094688e-01 -6.37679040e-01
5.43485403e-01 -1.28018156e-01 -7.08497584e-01 3.19349229e-01
-3.21402282e-01 -1.50507152e-01 7.46606469e-01 -4.10973877e-01
-1.34587455e+00 3.78873020e-01 -1.10701174e-01 4.31117453e-02
-5.57481408e-01 2.52841800e-01 -5.25658168e-02 -9.39113572e-02
5.58213055e-01 7.15325415e-01 -2.18525290e-01 -3.10232580e-01
1.07759051e-01 1.00317049e+00 4.26088154e-01 -1.60522848e-01
7.95173645e-01 5.03991485e-01 7.96289220e-02 -1.66267872e+00
3.43073934e-01 -7.22962320e-01 -1.01208639e+00 -5.62662125e-01
1.31367290e+00 -4.30928558e-01 -4.27251488e-01 7.35896766e-01
-1.17379701e+00 3.55481863e-01 2.99200982e-01 7.07210422e-01
-4.22953755e-01 1.28685749e+00 -6.47478938e-01 -1.16035020e+00
-3.25264961e-01 -1.04814529e+00 1.01431489e+00 1.80055693e-01
-1.82131343e-02 -9.02006388e-01 -8.52568150e-02 3.16084892e-01
-2.07385883e-01 5.31602502e-02 8.86636376e-01 -7.12470412e-01
-3.55690897e-01 -4.69834805e-01 -1.84741453e-03 2.39316180e-01
3.10669411e-02 8.05346489e-01 -6.58408582e-01 4.45592329e-02
1.31736668e-02 2.97715932e-01 6.83232486e-01 3.30289155e-01
4.92445499e-01 -3.23740952e-02 -4.14408505e-01 2.56401956e-01
1.47560251e+00 4.73647505e-01 5.89193642e-01 3.66305083e-01
6.01596832e-01 3.19487780e-01 6.02718771e-01 5.46456039e-01
-2.64567286e-01 4.79363412e-01 -2.77222563e-02 -2.19282985e-01
3.61377858e-02 1.23675838e-01 4.79767621e-01 8.29299390e-01
-2.43866056e-01 -5.84589958e-01 -8.75344157e-01 3.51402551e-01
-1.56026912e+00 -1.04113591e+00 -9.09873307e-01 2.12636065e+00
4.35829878e-01 4.00630027e-01 2.59788096e-01 8.89671743e-01
1.30984461e+00 -1.12044923e-01 6.38372004e-02 -4.91247594e-01
-3.02431852e-01 3.82687040e-02 5.52996874e-01 4.99609470e-01
-1.31726563e+00 8.56084049e-01 5.28634024e+00 9.58403766e-01
-1.18798018e+00 -1.02450371e-01 1.58745989e-01 3.70505303e-01
2.59192646e-01 7.00944588e-02 -8.46903980e-01 4.00002927e-01
5.14162064e-01 6.34459332e-02 1.28001219e-03 4.97870535e-01
4.68852490e-01 -8.11488509e-01 -4.75887299e-01 8.55554700e-01
3.86829436e-01 -6.80646360e-01 3.20614547e-01 -6.86199963e-02
5.85798562e-01 -5.93108237e-01 -1.57656595e-01 -2.88254410e-01
-4.77763057e-01 -4.27649885e-01 6.40662491e-01 4.29613680e-01
5.38898051e-01 -7.32897639e-01 7.63487875e-01 4.16852444e-01
-1.43929040e+00 2.05764100e-01 -2.06797853e-01 2.58472949e-01
9.42291990e-02 3.24603319e-01 -1.01473892e+00 7.05129504e-01
2.40119874e-01 5.54798722e-01 -5.68466902e-01 9.71424043e-01
9.48702097e-02 7.21490502e-01 -4.59457278e-01 -3.85138333e-01
3.08035403e-01 -6.47389412e-01 8.31490397e-01 1.64601386e+00
4.34910566e-01 -1.56117931e-01 -1.21420719e-01 4.95385528e-01
3.58166575e-01 8.73232722e-01 -6.06953502e-01 -6.18859529e-02
-8.89303759e-02 1.29725242e+00 -1.78470170e+00 -6.82537019e-01
-5.19443274e-01 1.35673857e+00 -5.16146302e-01 3.40982616e-01
-7.13949800e-01 -8.02940249e-01 -4.39154148e-01 6.18480369e-02
4.04954404e-01 -3.52026403e-01 -7.77152646e-03 -1.23314321e+00
4.87831570e-02 -9.13428843e-01 3.49384189e-01 -6.28669381e-01
-4.95096356e-01 5.26539207e-01 -1.03784474e-02 -1.26200593e+00
-7.78567716e-02 -5.27007520e-01 -7.65917540e-01 7.33370423e-01
-8.03964198e-01 -1.19163120e+00 -1.28778204e-01 8.84739757e-01
1.02432132e+00 -1.60013586e-01 4.69921827e-01 2.32206032e-01
-8.80614877e-01 2.17686921e-01 4.41110432e-01 1.27922446e-01
7.21859396e-01 -1.11866879e+00 -5.12344821e-04 1.38017833e+00
1.49562269e-01 4.02761132e-01 9.15089428e-01 -1.04896283e+00
-1.04581070e+00 -4.40288246e-01 1.03011894e+00 6.12774454e-02
5.99884570e-01 -2.74329275e-01 -7.19005942e-01 6.37036502e-01
5.20496666e-01 -6.04899228e-01 1.22271545e-01 -5.29707015e-01
3.48438382e-01 3.77041280e-01 -9.97266948e-01 6.04437649e-01
2.53097206e-01 -3.01793724e-01 -6.76525056e-01 4.05503184e-01
-1.43016428e-01 -7.31213167e-02 -5.76035619e-01 -1.76195297e-02
5.33466995e-01 -1.32180643e+00 5.14110506e-01 2.64006909e-02
1.02044329e-01 -5.65661311e-01 2.75031179e-01 -4.36555952e-01
2.13494182e-01 -7.50808358e-01 6.10191643e-01 1.35478365e+00
3.10178459e-01 -3.76076043e-01 8.82743120e-01 -2.66193390e-01
1.62583575e-01 -1.29329041e-01 -7.79140234e-01 -2.88849086e-01
-3.16058457e-01 -1.62217841e-01 -1.86133236e-01 1.07497621e+00
2.86260307e-01 2.09079430e-01 -2.40480483e-01 1.33933634e-01
5.70164800e-01 -1.34809151e-01 6.55774951e-01 -1.11039424e+00
-2.21855804e-01 -6.07831597e-01 -6.34754777e-01 -7.90680587e-01
-1.79558620e-01 -5.34756601e-01 -1.85594231e-01 -1.43664503e+00
1.14617661e-01 1.50509700e-01 4.03664857e-01 -3.76364067e-02
1.69037264e-02 2.57853389e-01 8.17658007e-03 3.01824391e-01
-3.40891987e-01 -1.00251317e-01 7.97514677e-01 1.71723679e-01
-2.69816488e-01 2.14929715e-01 3.06100816e-01 1.04358137e+00
6.63458705e-01 -2.35281706e-01 -1.60837799e-01 1.06156126e-01
5.23345219e-03 2.94032872e-01 2.12862454e-02 -1.30830991e+00
3.24235767e-01 1.28749698e-01 6.57331109e-01 -1.25328231e+00
8.70711133e-02 -9.59756315e-01 1.41512707e-01 5.94344378e-01
-1.57776494e-02 2.52511531e-01 -3.42632234e-02 5.54014087e-01
-8.42464194e-02 -8.13117921e-01 7.63433933e-01 -4.76737432e-02
-4.25350368e-01 -5.00103712e-01 -1.26457632e+00 -5.87171555e-01
1.25229836e+00 -9.11816239e-01 5.07538505e-02 -3.01136315e-01
-9.55344319e-01 -2.94617772e-01 4.32254285e-01 -1.67527422e-01
4.86674786e-01 -5.98670661e-01 -5.33043563e-01 1.89987853e-01
-1.98024094e-01 -4.77038175e-01 3.29226077e-01 1.24068046e+00
-1.08570731e+00 3.62364262e-01 -3.02989781e-01 -7.97733009e-01
-2.01808071e+00 5.46263516e-01 1.66594252e-01 -1.18552193e-01
-6.14032865e-01 1.62620381e-01 -3.30281407e-01 1.74249098e-01
3.11341971e-01 -3.89003515e-01 -7.64408708e-01 3.17413300e-01
1.52150363e-01 6.52386189e-01 9.12760347e-02 -1.02449322e+00
-1.58912897e-01 1.31663632e+00 2.67190039e-01 -4.60905224e-01
6.65663004e-01 -3.55251729e-01 -1.00098647e-01 4.32354420e-01
1.03846335e+00 6.92242622e-01 -6.68479443e-01 1.45775467e-01
1.54120445e-01 -2.20811844e-01 -1.83681026e-01 -3.94828081e-01
-4.84469086e-01 8.85055363e-01 6.56173110e-01 5.75410306e-01
1.17794168e+00 -4.04222667e-01 3.57519805e-01 4.42451894e-01
-1.83232009e-01 -1.30766296e+00 -1.28032312e-01 2.09664926e-01
4.01149422e-01 -7.76377618e-01 1.96478143e-01 -5.32871664e-01
-5.76543093e-01 1.92729521e+00 2.24181622e-01 9.76688042e-02
4.30524260e-01 3.58863026e-01 -1.58301502e-01 6.30008951e-02
-2.38377273e-01 -4.91316289e-01 2.42314991e-02 2.85156816e-01
6.11932874e-01 -3.07334453e-01 -9.10802245e-01 -4.81591895e-02
4.82479744e-02 -1.73545167e-01 9.77982283e-01 1.09483755e+00
-7.72485673e-01 -9.01941478e-01 -1.11217773e+00 2.42212161e-01
-7.63024271e-01 -6.34656772e-02 -6.06697679e-01 1.08286166e+00
3.74721944e-01 1.05593920e+00 8.79786834e-02 -1.95580482e-01
2.11934015e-01 5.10529816e-01 4.56708729e-01 -1.33319810e-01
-5.80887020e-01 9.51472759e-01 1.07162111e-01 1.16699420e-01
-5.72009981e-01 -9.93160665e-01 -1.46328354e+00 -1.19924441e-01
-6.97925270e-01 2.73641765e-01 1.08643341e+00 8.38866651e-01
-3.56118798e-01 1.99479088e-01 4.39779669e-01 -6.57270133e-01
-1.09715402e-01 -9.72356379e-01 -6.13965333e-01 2.81887501e-01
1.66538447e-01 -3.44541878e-01 -7.06624508e-01 7.50278115e-01]
|
[11.909997940063477, 2.47222638130188]
|
2b71094a-bb3b-446e-ac31-b044beee658c
|
fusqa-fetal-ultrasound-segmentation-quality
|
2303.04418
| null |
https://arxiv.org/abs/2303.04418v1
|
https://arxiv.org/pdf/2303.04418v1.pdf
|
FUSQA: Fetal Ultrasound Segmentation Quality Assessment
|
Deep learning models have been effective for various fetal ultrasound segmentation tasks. However, generalization to new unseen data has raised questions about their effectiveness for clinical adoption. Normally, a transition to new unseen data requires time-consuming and costly quality assurance processes to validate the segmentation performance post-transition. Segmentation quality assessment efforts have focused on natural images, where the problem has been typically formulated as a dice score regression task. In this paper, we propose a simplified Fetal Ultrasound Segmentation Quality Assessment (FUSQA) model to tackle the segmentation quality assessment when no masks exist to compare with. We formulate the segmentation quality assessment process as an automated classification task to distinguish between good and poor-quality segmentation masks for more accurate gestational age estimation. We validate the performance of our proposed approach on two datasets we collect from two hospitals using different ultrasound machines. We compare different architectures, with our best-performing architecture achieving over 90% classification accuracy on distinguishing between good and poor-quality segmentation masks from an unseen dataset. Additionally, there was only a 1.45-day difference between the gestational age reported by doctors and estimated based on CRL measurements using well-segmented masks. On the other hand, this difference increased and reached up to 7.73 days when we calculated CRL from the poorly segmented masks. As a result, AI-based approaches can potentially aid fetal ultrasound segmentation quality assessment and might detect poor segmentation in real-time screening in the future.
|
['Mohammad Yaqub', 'Ibrahim Almakk', 'Sevim Cengiz']
|
2023-03-08
| null | null | null | null |
['age-estimation', 'age-estimation']
|
['computer-vision', 'miscellaneous']
|
[ 2.88320452e-01 6.37787938e-01 1.99282721e-01 -7.23859370e-01
-9.63721335e-01 -6.70348108e-01 2.66436208e-02 5.44442236e-01
-3.42527002e-01 3.89149845e-01 -2.30925247e-01 -4.76211905e-01
-1.84910953e-01 -7.94818044e-01 -7.94046223e-01 -5.55419683e-01
-2.44632229e-01 8.50338042e-01 1.46986455e-01 2.81697780e-01
-1.06072828e-01 4.06721026e-01 -1.01426876e+00 5.83918929e-01
1.28417230e+00 1.05854392e+00 -1.41127199e-01 1.03802443e+00
1.01038243e-03 1.22536078e-01 -5.96071780e-01 -6.55761719e-01
5.24525225e-01 -7.28115022e-01 -1.09638274e+00 -2.94050723e-01
5.26762664e-01 -4.89869058e-01 1.18967839e-01 8.22690129e-01
7.85295546e-01 -2.77472168e-01 7.30962932e-01 -7.14126527e-01
-2.00548738e-01 7.08268821e-01 -2.83594400e-01 4.91902232e-01
2.53983080e-01 -3.23970541e-02 4.12529260e-01 -3.85453105e-01
4.32207376e-01 6.95581913e-01 9.20747936e-01 6.67351663e-01
-9.26127076e-01 -6.09220743e-01 -5.93223274e-01 -2.49703214e-01
-1.07985532e+00 -3.47558886e-01 2.77133107e-01 -7.69054115e-01
3.22700381e-01 4.49332148e-01 9.78572011e-01 6.88919544e-01
2.52832234e-01 7.48230040e-01 8.23959768e-01 -2.29784369e-01
1.50973976e-01 -5.86325116e-02 -3.25406253e-01 9.38842118e-01
6.24569416e-01 -1.07151389e-01 6.56183213e-02 2.65800692e-02
8.21621060e-01 -3.04363430e-01 -2.13959724e-01 -2.99878418e-01
-8.81878078e-01 6.82472169e-01 3.64783108e-01 8.04693222e-01
-4.64578599e-01 -1.68063149e-01 4.43070561e-01 2.19470188e-01
5.04741549e-01 9.24131095e-01 -4.99231666e-01 -4.75566626e-01
-1.66438353e+00 -1.56087965e-01 8.34410489e-01 7.14085698e-01
5.33431470e-01 -2.12796822e-01 -3.58473718e-01 5.52485585e-01
9.43048447e-02 3.64637554e-01 5.09208739e-01 -1.09978640e+00
3.80611748e-01 5.76825321e-01 1.56093210e-01 -1.07807982e+00
-9.28395808e-01 -6.52619541e-01 -8.18242133e-01 -2.43162051e-01
6.36593521e-01 -6.75687611e-01 -1.23016500e+00 1.24858284e+00
1.96390942e-01 3.56960952e-01 -4.83055077e-02 9.60334003e-01
1.21336830e+00 2.79538125e-01 -1.28041163e-01 -3.53900999e-01
1.00345135e+00 -5.75397611e-01 -6.05039060e-01 7.61082992e-02
1.00205302e+00 -5.41496634e-01 5.77579558e-01 5.74467540e-01
-1.59340858e+00 -4.81884062e-01 -1.00454521e+00 5.70388317e-01
5.92040457e-03 1.45262480e-01 5.01453400e-01 1.23648620e+00
-1.35980248e+00 9.24943864e-01 -1.37279153e+00 -3.07507694e-01
7.29194045e-01 4.81411934e-01 -1.99485064e-01 -1.43234758e-02
-1.11096740e+00 9.06800985e-01 4.19757873e-01 1.77698627e-01
-7.65609264e-01 -9.36562419e-01 -9.44846809e-01 2.54222184e-01
9.33324993e-02 -3.38470221e-01 1.38981891e+00 -8.64750803e-01
-1.25566292e+00 1.09985793e+00 1.62384346e-01 -5.83491862e-01
8.88599634e-01 -1.31510824e-01 -3.31569642e-01 5.52805483e-01
8.18345919e-02 5.77333212e-01 5.31153262e-01 -1.24167848e+00
-4.91338283e-01 -1.96151733e-01 -1.32042989e-01 -5.98881282e-02
-4.22738306e-02 -5.97846881e-02 -5.81993759e-01 -3.87650758e-01
6.07869744e-01 -8.59744310e-01 -2.06477433e-01 1.70982094e-03
3.49158458e-02 3.60242546e-01 2.47102991e-01 -9.37532127e-01
1.30576336e+00 -1.82632494e+00 -4.03207093e-01 4.48246032e-01
5.42818487e-01 8.25433969e-01 -4.55227345e-02 -2.19741642e-01
8.29999615e-03 5.66051304e-01 -4.04480249e-01 4.78780009e-02
-5.05577981e-01 1.24041893e-01 7.23325193e-01 4.44987267e-01
4.43967611e-01 9.94757116e-01 -1.27023399e+00 -7.65741050e-01
1.82165265e-01 2.53729522e-01 -7.30342805e-01 5.86660087e-01
3.60265434e-01 8.55338454e-01 -3.47456664e-01 6.30107760e-01
8.07964087e-01 -1.46765381e-01 2.38533944e-01 -2.40700632e-01
9.79520679e-02 -1.16156057e-01 -8.81383538e-01 1.85776353e+00
-4.33314353e-01 6.56073570e-01 5.42091131e-02 -1.13047981e+00
8.49163413e-01 7.11284161e-01 8.67482424e-01 -7.11775541e-01
5.43734014e-01 4.82112586e-01 6.54064655e-01 -9.76397216e-01
1.15110986e-01 -2.66173095e-01 2.00962514e-01 1.95259854e-01
9.35333520e-02 -4.48924333e-01 3.74013841e-01 -1.56737134e-01
1.25025678e+00 -4.72984388e-02 1.38905525e-01 -3.92802030e-01
3.12695861e-01 -2.83214808e-01 5.65807700e-01 1.15835059e+00
-5.37221432e-01 1.26759946e+00 6.06042027e-01 -5.68495989e-01
-7.60269701e-01 -8.33207786e-01 -4.33652908e-01 5.74221492e-01
3.98280285e-02 1.27934352e-01 -1.04960537e+00 -9.54066694e-01
-3.18475753e-01 3.90188187e-01 -8.62666130e-01 -7.26832896e-02
-7.15393662e-01 -8.39220762e-01 9.07016873e-01 4.50601548e-01
3.27107996e-01 -9.29382622e-01 -1.07345164e+00 5.08857191e-01
-1.56244397e-01 -1.00225043e+00 -1.08082322e-02 -1.70881361e-01
-8.37538540e-01 -1.15981972e+00 -1.29961860e+00 -5.67722917e-01
6.39707685e-01 -5.35614848e-01 1.20704865e+00 4.38695788e-01
-4.13265347e-01 1.72097996e-01 -6.21683538e-01 -4.76571679e-01
-9.95683610e-01 3.90604064e-02 -1.68545455e-01 -6.77662194e-02
-1.03588931e-01 -2.17009902e-01 -1.19098604e+00 3.30493391e-01
-8.85679662e-01 -2.61933580e-02 6.64353132e-01 6.81471109e-01
1.80021942e-01 -5.46364367e-01 5.71369469e-01 -1.08614826e+00
3.15227091e-01 -6.53164983e-01 -4.51025993e-01 3.45905274e-01
-6.40697777e-01 -1.41076460e-01 2.95222878e-01 -1.92167282e-01
-9.93561208e-01 -3.16795230e-01 -4.52385068e-01 -3.21346849e-01
-1.70260102e-01 5.20248890e-01 4.74735260e-01 -1.10718779e-01
7.81407058e-01 -9.57354456e-02 -7.27063715e-02 -3.30213726e-01
-1.00158125e-01 7.17420638e-01 3.27715963e-01 -5.68699896e-01
2.61655867e-01 3.10833037e-01 1.82615463e-02 -4.02258843e-01
-8.49031508e-01 -4.56760764e-01 -6.93753242e-01 -3.55981201e-01
1.06563437e+00 -3.30755025e-01 -6.27244174e-01 1.60233557e-01
-1.07722890e+00 -3.06789160e-01 -1.65330932e-01 6.14260137e-01
-2.96652973e-01 5.01253426e-01 -6.51072562e-01 -6.78689897e-01
-6.26736522e-01 -1.50211048e+00 9.53834832e-01 3.81764382e-01
-1.69287503e-01 -1.07644641e+00 2.33547091e-02 4.36237752e-01
6.43158674e-01 7.16954648e-01 5.93304157e-01 -7.87186682e-01
-5.72557822e-02 -4.74020034e-01 -3.92516971e-01 4.19319570e-01
2.77165562e-01 7.33625516e-02 -9.40559566e-01 -4.42100853e-01
2.86407210e-02 -3.31806540e-02 5.07086754e-01 6.88029110e-01
1.32074976e+00 9.93974507e-03 -1.49166852e-01 7.89366543e-01
1.11535358e+00 6.21298969e-01 3.54148895e-01 -1.00098304e-01
5.38057208e-01 4.17833209e-01 8.25717270e-01 4.03449565e-01
2.17703640e-01 8.99869204e-02 4.40671623e-01 -5.81461608e-01
1.11123109e-02 2.90630698e-01 -5.59353530e-01 9.32427824e-01
-3.80034834e-01 -3.83117974e-01 -1.64012468e+00 6.78945720e-01
-1.42271388e+00 -3.72615844e-01 -8.73331428e-02 2.01052880e+00
8.80632102e-01 2.55665898e-01 -2.26833880e-01 1.54493526e-02
5.71879864e-01 -3.08135867e-01 -2.75270790e-01 -8.67113292e-01
4.19290274e-01 5.27884245e-01 4.09124881e-01 1.14541471e-01
-1.08875632e+00 4.84800130e-01 6.34826231e+00 3.89221042e-01
-1.34387815e+00 2.25016505e-01 1.16727495e+00 2.32016772e-01
-7.52274543e-02 -4.04852778e-01 -3.03544477e-02 5.15768945e-01
9.11532998e-01 1.86911330e-01 -2.25013494e-01 4.24705505e-01
1.60706952e-01 -3.04193437e-01 -1.32890582e+00 8.86245251e-01
-7.48180076e-02 -1.41824567e+00 -1.56125277e-01 -4.34738070e-01
1.04233527e+00 -6.90428680e-03 -1.85475141e-01 1.51506752e-01
-2.71995842e-01 -1.22706139e+00 2.25108534e-01 6.27477288e-01
1.24613380e+00 -3.02495301e-01 1.28180861e+00 2.86569506e-01
-7.38490820e-01 3.12698722e-01 -3.09731904e-02 1.75514251e-01
-1.36341244e-01 4.15153742e-01 -1.53083837e+00 7.51978219e-01
8.85271847e-01 3.26447368e-01 -4.69386846e-01 1.43607724e+00
2.93609977e-01 9.63594317e-01 -2.50528395e-01 2.99188942e-01
4.30353671e-01 -1.48077890e-01 2.90602744e-01 1.55713892e+00
5.62073648e-01 3.29328716e-01 -2.63990343e-01 6.73242033e-01
2.23539248e-02 2.37761006e-01 -2.10491329e-01 1.16569109e-01
3.92444395e-02 1.17313242e+00 -1.06568336e+00 -4.83871132e-01
-4.55746770e-01 6.16575956e-01 -1.62950516e-01 1.38895050e-01
-8.16217959e-01 -2.25696340e-01 -4.31463216e-03 1.95984229e-01
5.46699055e-02 4.31649327e-01 -5.44902682e-01 -8.02290976e-01
-1.37982532e-01 -6.04154050e-01 4.60737973e-01 -3.43564242e-01
-7.79159069e-01 7.39033759e-01 -4.28281277e-02 -1.29544306e+00
-1.93838656e-01 -3.27197701e-01 -6.31546557e-01 7.15976179e-01
-1.37732184e+00 -5.93142927e-01 -5.80715001e-01 -2.44866997e-01
5.05893826e-01 1.17860578e-01 7.74519801e-01 6.48905933e-01
-3.42914999e-01 9.63623881e-01 -1.00440130e-01 4.75284785e-01
3.97409141e-01 -1.36079371e+00 2.66664296e-01 8.19280803e-01
-3.89023095e-01 3.81504267e-01 6.20495021e-01 -7.21524298e-01
-7.26009727e-01 -1.07057810e+00 3.86605352e-01 -3.61540020e-01
7.63116330e-02 1.79026887e-01 -1.17240894e+00 3.25027019e-01
1.39235109e-02 2.62616545e-01 7.62011230e-01 -3.82718146e-01
3.59661341e-01 4.92942259e-02 -1.64215231e+00 1.96334660e-01
8.36879551e-01 2.99125671e-01 -4.06849831e-01 5.55252694e-02
4.60982949e-01 -1.01392066e+00 -1.39804757e+00 1.13467252e+00
8.22577119e-01 -1.12939775e+00 4.54208225e-01 -5.62421203e-01
5.05139470e-01 2.04051346e-01 4.74529654e-01 -1.06664252e+00
4.42321822e-02 -6.95883274e-01 1.04264162e-01 7.76845932e-01
7.42560804e-01 -3.94380480e-01 1.11646855e+00 8.65527749e-01
-2.55829751e-01 -1.15663958e+00 -8.75082910e-01 -2.88011014e-01
2.43077293e-01 -5.73294640e-01 6.92699552e-01 1.15576196e+00
-2.93200523e-01 -3.23105097e-01 -2.84587834e-02 2.34721094e-01
2.23677427e-01 5.27331457e-02 2.02807114e-01 -1.11957145e+00
-1.19359478e-01 -4.57352012e-01 -6.25207186e-01 -6.22406960e-01
-5.00246525e-01 -8.69810224e-01 1.88905939e-01 -1.77024996e+00
1.03151672e-01 -8.10082853e-01 -3.90301913e-01 2.95428336e-01
-3.73088449e-01 3.86083424e-01 -2.92522609e-02 -3.58406186e-01
-5.75178742e-01 -9.40467045e-02 1.68813109e+00 1.52846901e-02
-3.00730228e-01 6.30708456e-01 -3.83623332e-01 8.62130344e-01
9.00602221e-01 -5.58951616e-01 -1.43743262e-01 -7.40999341e-01
3.58644962e-01 7.04418242e-01 -3.37610781e-01 -1.08879697e+00
-1.42835394e-01 7.20893145e-02 4.64139313e-01 -4.65979040e-01
-7.50463754e-02 -7.54446745e-01 -6.00219369e-02 7.54041970e-01
-2.58434743e-01 -2.57998288e-01 2.90510356e-01 -5.44842333e-02
-1.55752808e-01 -5.18150389e-01 8.47426891e-01 -2.52706468e-01
-1.23511605e-01 4.37786788e-01 -3.39119315e-01 3.83519262e-01
8.70631754e-01 -5.41211843e-01 2.63535976e-01 -3.07294548e-01
-1.23712242e+00 3.91954839e-01 1.68582618e-01 2.66835511e-01
7.87455440e-01 -7.18022525e-01 -1.15465295e+00 2.61102498e-01
-5.12692146e-02 4.17110622e-01 3.60789686e-01 1.10382926e+00
-1.36616278e+00 1.36983588e-01 -8.58277455e-02 -1.02930129e+00
-1.17287850e+00 7.98589587e-02 6.37590766e-01 -2.02236518e-01
-1.85320258e-01 1.11080122e+00 3.46428715e-02 -4.01430637e-01
8.85677114e-02 -7.69342124e-01 -3.49586338e-01 1.16496786e-01
1.84656724e-01 5.35139382e-01 5.57805657e-01 -4.58317786e-01
-2.26337448e-01 6.10263109e-01 1.24122553e-01 3.13533306e-01
1.11718035e+00 -7.72508048e-03 8.56147856e-02 1.14884503e-01
1.20097315e+00 -3.38476419e-01 -9.56822991e-01 -1.13391383e-02
1.88682660e-01 -5.34910321e-01 -1.78620085e-01 -8.61078918e-01
-1.38706136e+00 1.02982092e+00 1.19928443e+00 6.08544290e-01
1.15587771e+00 9.47132558e-02 7.76670575e-01 1.21893950e-01
1.60402078e-02 -7.29305863e-01 -9.01549011e-02 1.13106817e-01
7.10536063e-01 -1.62743986e+00 -2.20828786e-01 -2.70385623e-01
-5.76353490e-01 1.07397199e+00 6.72479987e-01 7.30592459e-02
4.53237474e-01 2.43629083e-01 4.89640385e-01 -2.64472455e-01
-1.31266415e-01 3.20281424e-02 7.11888075e-01 4.76395279e-01
5.58725655e-01 3.58125269e-01 -4.55439895e-01 6.56290531e-01
-1.00857615e-01 -4.03336808e-02 6.18362665e-01 8.05944324e-01
-2.79794574e-01 -7.44120955e-01 -1.09907873e-01 1.15920842e+00
-1.08214831e+00 9.14663728e-03 1.98810011e-01 6.31203115e-01
6.06795371e-01 9.56580997e-01 -3.20265405e-02 -3.16736177e-02
3.64881247e-01 -9.99919325e-02 4.61568773e-01 -1.06275511e+00
-9.67107534e-01 2.62818523e-02 -2.45862342e-02 -6.54674053e-01
-3.86565655e-01 -6.37491643e-01 -1.39722526e+00 3.56426567e-01
-4.39953327e-01 2.87170440e-01 7.68144190e-01 8.89192879e-01
2.99219554e-03 7.89276123e-01 5.26853383e-01 -6.59483135e-01
-9.50897485e-02 -9.11145270e-01 -2.37596720e-01 4.44572806e-01
5.54388463e-01 -3.52353841e-01 -1.95086915e-02 -1.70746930e-02]
|
[14.195574760437012, -2.381269931793213]
|
8d41327f-a7b1-45c6-8e53-9a7573b6fe13
|
nlp-analytics-in-finance-with-dore-a-french
| null | null |
https://aclanthology.org/2020.lrec-1.275
|
https://aclanthology.org/2020.lrec-1.275.pdf
|
NLP Analytics in Finance with DoRe: A French 250M Tokens Corpus of Corporate Annual Reports
|
Recent advances in neural computing and word embeddings for semantic processing open many new applications areas which had been left unaddressed so far because of inadequate language understanding capacity. But this new kind of approaches rely even more on training data to be operational. Corpora for financial applications exists, but most of them concern stock market prediction and are in English. To address this need for the French language and regulation oriented applications which require a deeper understanding of the text content, we hereby present {``}DoRe{''}, a French and dialectal French Corpus for NLP analytics in Finance, Regulation and Investment. This corpus is composed of: (a) 1769 Annual Reports from 336 companies among the most capitalized companies in: France (Euronext Paris) {\&} Belgium (Euronext Brussels), covering a time frame from 2009 to 2019, and (b) related MetaData containing information for each company about its ISIN code, capitalization and sector. This corpus is designed to be as modular as possible in order to allow for maximum reuse in different tasks pertaining to Economics, Finance and Regulation. After presenting existing resources, we relate the construction of the DoRe corpus and the rationale behind our choices, concluding on the spectrum of possible uses of this new resource for NLP applications.
|
['Corentin Masson', 'Patrick Paroubek']
|
2020-05-01
| null | null | null |
lrec-2020-5
|
['stock-market-prediction']
|
['time-series']
|
[-4.86858189e-01 1.54117957e-01 -2.97434896e-01 -3.04205000e-01
-3.80118877e-01 -9.88968372e-01 9.55430925e-01 5.60585678e-01
-8.21452141e-01 7.80886114e-01 5.45088947e-01 -7.56629109e-01
-3.31023693e-01 -1.03009772e+00 -2.76566535e-01 -4.07305390e-01
2.49563158e-02 5.59799492e-01 -1.24002337e-01 -5.33956766e-01
5.99473894e-01 6.97399259e-01 -1.34737575e+00 4.24683601e-01
3.43779981e-01 1.34320295e+00 2.57646680e-01 1.24549776e-01
-7.29970157e-01 7.91960239e-01 -5.75687528e-01 -8.16871047e-01
4.26405579e-01 4.54868609e-03 -8.21342349e-01 -1.86764747e-01
-2.89926976e-01 1.38405800e-01 -2.12749451e-01 1.03660774e+00
2.73445934e-01 9.93872955e-02 5.53295135e-01 -5.86756051e-01
-9.57088768e-01 8.78185332e-01 -2.45623380e-01 5.56344926e-01
1.88725501e-01 -2.30380088e-01 1.58867073e+00 -1.11075783e+00
9.15667653e-01 7.91615248e-01 3.76307964e-01 3.60517025e-01
-7.28270650e-01 -4.94987369e-01 2.01919034e-01 1.28610030e-01
-9.27071095e-01 -2.31475830e-01 5.13234675e-01 -6.71678245e-01
1.36350179e+00 -1.22112349e-01 6.85580909e-01 1.00655055e+00
9.72583517e-02 3.76068622e-01 9.88203347e-01 -6.44606590e-01
2.50015259e-01 7.01687694e-01 1.91645384e-01 1.28701061e-01
6.48572981e-01 -9.23913196e-02 -4.62598592e-01 -2.61194892e-02
8.04003239e-01 -3.23069483e-01 -2.10388899e-01 -7.55322054e-02
-1.24810231e+00 1.35289097e+00 1.50494605e-01 1.12624907e+00
-5.19865096e-01 -8.49415734e-02 7.19402075e-01 5.44994712e-01
7.38534987e-01 6.69564605e-01 -1.06177759e+00 -4.26579148e-01
-8.26114655e-01 4.49482381e-01 1.24264729e+00 5.78910351e-01
3.92988145e-01 2.81398445e-01 4.57578897e-01 8.78136277e-01
3.56806129e-01 -2.96109226e-02 7.78125286e-01 -4.79478806e-01
8.72445405e-01 6.36614680e-01 2.76310176e-01 -9.46589887e-01
-4.26556677e-01 -4.42271203e-01 -4.21104819e-01 9.44332555e-02
5.34255564e-01 -2.00781763e-01 -2.90142268e-01 1.18234444e+00
-1.34682968e-01 -4.87960577e-01 3.82184863e-01 6.08551741e-01
7.52035797e-01 5.66832721e-01 -1.21446177e-01 -2.51850605e-01
1.78107750e+00 -5.39986014e-01 -7.56716907e-01 -9.06255469e-02
5.80063343e-01 -9.39710796e-01 7.20875919e-01 5.99827647e-01
-1.07504463e+00 -3.25276285e-01 -8.87359738e-01 8.36904123e-02
-1.06067216e+00 -1.40229180e-01 8.29682112e-01 7.72248805e-01
-7.16237426e-01 6.42135382e-01 -3.61717045e-01 -1.78344667e-01
5.68268657e-01 1.55242682e-01 -4.33616370e-01 2.23690152e-01
-1.59045112e+00 1.28882062e+00 7.85621047e-01 -1.80115774e-02
4.64165956e-02 -5.52728951e-01 -9.01282668e-01 1.89496115e-01
3.38862419e-01 -1.45286873e-01 1.10085344e+00 -9.07247961e-01
-1.21338224e+00 1.07807302e+00 3.26296151e-01 -8.04167390e-01
3.74917507e-01 -1.64907202e-01 -8.37976158e-01 3.31045352e-02
3.96925136e-02 1.84307858e-01 1.77275226e-01 -5.74387312e-01
-7.58078575e-01 -3.13231409e-01 7.95961246e-02 -1.72277540e-01
-4.77701604e-01 6.27312899e-01 -1.75986186e-01 -1.21492541e+00
-9.26300064e-02 -3.89265537e-01 -1.45217419e-01 -7.17435539e-01
1.29958183e-01 -4.53104556e-01 2.02577412e-01 -8.83769035e-01
1.38886929e+00 -1.92236233e+00 -2.38864377e-01 1.46773174e-01
1.72437206e-02 2.56595045e-01 9.26996469e-02 8.97226691e-01
-5.72295249e-01 5.05143940e-01 -9.07508358e-02 9.02132988e-02
4.06970650e-01 1.75991997e-01 -6.40362978e-01 5.36951840e-01
2.75686771e-01 1.01138115e+00 -5.76500952e-01 -1.84819862e-01
2.55992889e-01 3.07353288e-01 -2.88540125e-01 -3.38635564e-01
-1.19649075e-01 1.08368047e-01 -5.29377341e-01 8.66024017e-01
4.37024981e-01 7.49362186e-02 1.62845150e-01 7.94368535e-02
-5.60346544e-01 6.21528804e-01 -1.17650247e+00 1.26338768e+00
-4.77255911e-01 6.43014133e-01 -7.30883554e-02 -1.29990935e+00
1.28750360e+00 5.00422716e-01 8.56745183e-01 -5.53660214e-01
4.88466889e-01 6.54371202e-01 -6.06846362e-02 -4.07533497e-01
7.72179127e-01 -5.94304800e-01 -1.88438639e-01 3.30077112e-01
-1.07086105e-02 -2.86244042e-02 6.61786199e-01 -1.91756681e-01
9.96397674e-01 4.07919250e-02 6.16868794e-01 -5.37891269e-01
7.93196440e-01 -5.77324517e-02 6.35519385e-01 1.45457417e-01
-2.48304650e-01 2.70519257e-01 8.18256497e-01 -8.60752046e-01
-1.05565894e+00 -7.70584226e-01 -6.06105804e-01 6.73760951e-01
-7.30339408e-01 -2.41166264e-01 -4.00756776e-01 -5.51674008e-01
2.08071083e-01 9.52078164e-01 -6.82019770e-01 6.49503469e-01
-5.33146977e-01 -7.87664831e-01 2.30412692e-01 3.66385311e-01
4.31636006e-01 -1.39083135e+00 -6.03424191e-01 5.77505410e-01
2.17040047e-01 -1.13235664e+00 2.17841297e-01 3.13186228e-01
-8.56456876e-01 -1.38345456e+00 -8.25101197e-01 -5.61754048e-01
8.35049674e-02 -5.84489882e-01 1.34246826e+00 -5.01605034e-01
-8.07577819e-02 1.80270076e-01 -5.33179760e-01 -7.47630775e-01
-3.59783232e-01 2.27524266e-01 -9.40921232e-02 -1.05020396e-01
9.30431306e-01 -1.40711740e-01 -8.28937441e-02 -6.12507015e-02
-9.52757537e-01 -6.62791371e-01 4.11085814e-01 7.05100536e-01
2.64429390e-01 1.08466446e-01 1.00025833e+00 -1.09831059e+00
9.07600284e-01 -6.00874126e-01 -8.50990415e-01 1.11827239e-01
-7.68823028e-01 -3.26523989e-01 4.50617045e-01 1.70658946e-01
-9.51661050e-01 -4.50285435e-01 -3.89967650e-01 1.40916690e-01
-2.17426121e-01 9.39968109e-01 1.09643098e-02 3.60154688e-01
3.68874878e-01 1.45882845e-01 -5.25139794e-02 -7.54663408e-01
3.32339436e-01 7.52037525e-01 3.88999462e-01 -4.06506747e-01
5.38878143e-01 2.97126591e-01 -2.53737986e-01 -6.88475907e-01
-5.62748373e-01 -4.96795654e-01 -7.28901565e-01 -4.07883525e-02
9.73639965e-01 -7.69774675e-01 -6.44511044e-01 1.70586377e-01
-1.15649176e+00 2.81954169e-01 -6.76399112e-01 7.96274960e-01
-4.95773882e-01 3.70354652e-02 -6.74990594e-01 -9.02199268e-01
-7.33611211e-02 -7.48806715e-01 3.48941833e-01 -8.85759667e-02
-3.18491966e-01 -1.53784788e+00 1.18091784e-01 2.85797000e-01
4.72593755e-01 3.89655352e-01 1.06608045e+00 -1.14269209e+00
-2.02025622e-01 -4.99441713e-01 -2.15797335e-01 6.49424195e-01
6.25301003e-02 -9.34681967e-02 -8.57842743e-01 -1.84495836e-01
1.86611012e-01 -1.73737049e-01 1.00627899e+00 2.07196057e-01
8.23672116e-01 -2.11257294e-01 2.31523693e-01 1.24402136e-01
1.79152834e+00 4.27195817e-01 5.78393161e-01 1.06324124e+00
5.21381311e-02 1.00715101e+00 6.05366766e-01 7.09495127e-01
1.18082136e-01 4.06414360e-01 2.11637437e-01 4.30125743e-01
5.61170638e-01 8.92872214e-02 5.46585321e-01 1.02931929e+00
-3.89639378e-01 -1.52222723e-01 -1.03644705e+00 8.58772516e-01
-1.56522453e+00 -9.90715504e-01 -4.88602668e-02 2.13515687e+00
6.62491083e-01 3.34202439e-01 5.97355999e-02 1.24044590e-01
4.70131397e-01 5.24753273e-01 -6.98626488e-02 -9.03154016e-01
-4.71396953e-01 6.38181925e-01 7.05188036e-01 1.35566533e-01
-9.99595165e-01 8.36755455e-01 5.92956161e+00 8.46343219e-01
-8.09816539e-01 1.57664776e-01 5.46928942e-01 -9.93586406e-02
-5.63834310e-01 -3.13542830e-03 -9.55139816e-01 4.73672897e-01
1.23839760e+00 -3.71677250e-01 1.72343388e-01 7.48837650e-01
3.55677247e-01 4.48593386e-02 -7.85679698e-01 8.27086866e-01
-4.63756993e-02 -1.81821609e+00 -1.76104218e-01 3.23052257e-01
7.03325868e-01 2.38275006e-01 -2.03933064e-02 5.27693391e-01
1.68716744e-01 -9.89747941e-01 9.56486702e-01 4.47172314e-01
6.10153317e-01 -9.83453572e-01 1.12275541e+00 4.68775854e-02
-1.04657471e+00 -3.62672776e-01 -5.82622826e-01 -1.86989114e-01
3.97826642e-01 9.76189375e-01 -2.67957062e-01 9.81775820e-01
7.17671573e-01 8.38683724e-01 -2.73047775e-01 4.64466512e-01
-2.08116561e-01 4.65814561e-01 -3.65571752e-02 -2.80985385e-01
6.70536935e-01 -5.09074986e-01 2.31976360e-01 1.11573470e+00
5.54633141e-01 -1.92541137e-01 -2.77673155e-01 8.03664744e-01
-2.11292237e-01 6.79505646e-01 -6.96454227e-01 -6.56022549e-01
3.23777467e-01 9.33547974e-01 -9.37712669e-01 -1.19320817e-01
-1.01857436e+00 3.91549557e-01 2.01022461e-01 1.98187709e-01
-5.84988475e-01 -6.55209661e-01 6.81184530e-01 3.69545341e-01
5.37640810e-01 -2.91280985e-01 -4.95345891e-01 -1.13938344e+00
1.10439211e-01 -7.23274529e-01 5.00836670e-01 -3.20454597e-01
-1.17557228e+00 6.10642731e-01 -1.17174238e-02 -8.83433342e-01
-3.77867669e-01 -1.26445699e+00 -1.65239558e-01 9.61115718e-01
-1.59042287e+00 -7.20415711e-01 5.45397341e-01 8.63794684e-02
5.33115149e-01 -1.00571585e+00 8.24405670e-01 4.19530690e-01
-4.57184643e-01 -1.04117565e-01 2.38669366e-01 3.77698779e-01
4.10740882e-01 -1.41342437e+00 4.76432174e-01 4.40692693e-01
5.10143757e-01 6.33729637e-01 3.81518155e-01 -4.87139791e-01
-7.21220672e-01 -7.91095853e-01 1.70749545e+00 -4.75585759e-01
1.41115129e+00 -2.72562951e-01 -7.22143412e-01 6.24499083e-01
5.03896773e-01 -3.35532337e-01 7.30873883e-01 6.56400770e-02
-2.96966046e-01 -7.81404227e-02 -1.03424442e+00 2.58003861e-01
2.79706717e-01 -4.37139213e-01 -1.04659677e+00 4.06861395e-01
4.25563186e-01 -3.50936619e-03 -1.20894599e+00 -2.48785131e-02
3.23449999e-01 -1.12662518e+00 7.73140848e-01 -5.67106843e-01
6.09705806e-01 9.03234184e-02 -1.56228051e-01 -8.07485878e-01
-1.78350434e-01 -5.43396831e-01 2.39310339e-01 1.29518199e+00
5.78905880e-01 -1.25116265e+00 6.65551066e-01 4.59629178e-01
-1.57949239e-01 -9.66971159e-01 -1.38361633e+00 -8.52577567e-01
6.94231749e-01 -9.22651112e-01 7.87139893e-01 1.09207892e+00
-3.11789359e-03 -1.21720722e-02 2.36989379e-01 -5.08310318e-01
-1.10423997e-01 8.81129503e-02 2.13716015e-01 -1.44144344e+00
-9.89259556e-02 -8.50798845e-01 -3.84411097e-01 -6.57932520e-01
3.79504681e-01 -9.15001214e-01 -6.83796763e-01 -1.56591260e+00
-3.04883122e-01 -1.68992072e-01 -4.45564389e-01 2.57045239e-01
6.58375263e-01 2.83184629e-02 2.56876945e-01 1.72237769e-01
9.77207273e-02 2.11090162e-01 7.00816393e-01 6.05389476e-02
1.56949788e-01 2.70690843e-02 -1.06584215e+00 7.81939507e-01
7.24400401e-01 -2.89829046e-01 4.71781306e-02 -1.38178617e-01
7.71670699e-01 -3.11865330e-01 1.69807494e-01 -5.14098883e-01
-4.49384227e-02 -9.53698158e-02 2.63814658e-01 -7.23898828e-01
9.42608938e-02 -8.43863189e-01 -1.31183654e-01 3.54145616e-01
-2.30007127e-01 4.67608005e-01 1.87200919e-01 4.52628374e-01
-8.69434655e-01 -9.44276214e-01 6.22146726e-01 -5.06030023e-01
-8.24472785e-01 1.44124761e-01 -6.30246043e-01 1.53179318e-01
1.25923204e+00 -3.15080702e-01 -8.55436623e-02 -1.52500361e-01
-7.35509813e-01 -2.72852443e-02 2.51212209e-01 4.90402907e-01
3.58738601e-01 -1.26676846e+00 -8.33908021e-01 1.46552429e-01
1.11557215e-01 -2.74499863e-01 7.47219920e-02 5.75661004e-01
-9.28700984e-01 1.02514935e+00 -6.22469597e-02 4.62133169e-01
-6.42338753e-01 6.27370656e-01 -1.95862576e-02 -6.94105566e-01
-5.90173066e-01 7.88043857e-01 -2.36658871e-01 -1.34485200e-01
-8.89503434e-02 -6.51656032e-01 -8.94619048e-01 8.62823248e-01
4.15301234e-01 2.32736290e-01 2.30373487e-01 -7.11653531e-01
-3.27538878e-01 2.87469387e-01 7.02842511e-03 -2.83572167e-01
1.82043386e+00 1.75979398e-02 -3.94990742e-01 7.78620243e-01
1.18952572e+00 2.03834444e-01 -5.89416742e-01 -2.04243883e-02
6.69590950e-01 -3.90423030e-01 8.18053558e-02 -6.44293785e-01
-1.29130077e+00 7.04343021e-01 1.00417770e-01 6.82862282e-01
8.58066022e-01 -7.33650923e-02 6.27517104e-01 2.77295440e-01
2.91160822e-01 -1.66133261e+00 -4.15706098e-01 7.00215995e-01
1.15158987e+00 -8.13077033e-01 7.27991462e-02 -1.38357937e-01
-6.36930168e-01 1.68611908e+00 -3.64301056e-01 -1.67482927e-01
1.09611070e+00 2.14541536e-02 2.77439263e-02 -3.86743248e-01
-5.54453671e-01 -1.93492472e-01 4.49157320e-02 5.01325130e-01
7.83345103e-01 -2.43173853e-01 -8.08654428e-01 1.17560387e+00
-4.68594402e-01 5.06158508e-02 6.42795384e-01 8.30801964e-01
-3.59368116e-01 -1.50431252e+00 -3.49872172e-01 5.52021742e-01
-1.19510531e+00 -2.18498603e-01 -4.26843435e-01 1.37263024e+00
3.35467160e-01 6.10411525e-01 3.67265999e-01 1.60997391e-01
5.17297208e-01 2.56640911e-01 7.72769973e-02 -7.77737498e-01
-1.00315487e+00 1.19351819e-01 3.97883117e-01 -9.10564065e-02
-5.65525293e-01 -1.08105576e+00 -1.08574581e+00 -4.05739516e-01
-4.33732085e-02 2.72713006e-01 8.76679838e-01 8.27712059e-01
-1.14720702e-01 6.07977986e-01 3.77486438e-01 -5.31036019e-01
-7.98729956e-01 -9.84211802e-01 -1.36307335e+00 1.31484538e-01
-1.95370063e-01 -6.62898004e-01 -3.92065495e-01 -1.50834501e-01]
|
[11.08462142944336, 7.126943588256836]
|
ba09dd94-9847-4750-9849-ca5f679f3cec
|
assessing-post-deletion-in-sina-weibo-multi
|
1906.10861
| null |
https://arxiv.org/abs/1906.10861v2
|
https://arxiv.org/pdf/1906.10861v2.pdf
|
Assessing Post Deletion in Sina Weibo: Multi-modal Classification of Hot Topics
|
Widespread Chinese social media applications such as Weibo are widely known for monitoring and deleting posts to conform to Chinese government requirements. In this paper, we focus on analyzing a dataset of censored and uncensored posts in Weibo. Despite previous work that only considers text content of posts, we take a multi-modal approach that takes into account both text and image content. We categorize this dataset into 14 categories that have the potential to be censored on Weibo, and seek to quantify censorship by topic. Specifically, we investigate how different factors interact to affect censorship. We also investigate how consistently and how quickly different topics are censored. To this end, we have assembled an image dataset with 18,966 images, as well as a text dataset with 994 posts from 14 categories. We then utilized deep learning, CNN localization, and NLP techniques to analyze the target dataset and extract categories, for further analysis to better understand censorship mechanisms in Weibo. We found that sentiment is the only indicator of censorship that is consistent across the variety of topics we identified. Our finding matches with recently leaked logs from Sina Weibo. We also discovered that most categories like those related to anti-government actions (e.g. protest) or categories related to politicians (e.g. Xi Jinping) are often censored, whereas some categories such as crisis-related categories (e.g. rainstorm) are less frequently censored. We also found that censored posts across all categories are deleted in three hours on average.
|
['King-wa Fu', 'Jedidiah R. Crandall', 'Rajkumar Pandi', 'Michael Carl Tschantz', 'Dahlia Qiu Shi', 'Meisam Navaki Arefi', 'Miao Sha']
|
2019-06-26
|
assessing-post-deletion-in-sina-weibo-multi-1
|
https://aclanthology.org/D19-5001
|
https://aclanthology.org/D19-5001.pdf
|
ws-2019-11
|
['multi-modal-classification']
|
['miscellaneous']
|
[-4.38036770e-01 -1.15452014e-01 -1.82532415e-01 -5.29830337e-01
-1.00424552e+00 -1.34015501e+00 9.39777136e-01 3.65581483e-01
-1.99489787e-01 6.13634765e-01 6.92603350e-01 -7.22720981e-01
5.75501770e-02 -8.50809038e-01 -8.49416137e-01 -5.33014715e-01
7.16162249e-02 2.99628347e-01 -2.82493860e-01 6.91854581e-02
3.56306940e-01 5.73614955e-01 -8.07214081e-01 4.41800952e-02
7.32021451e-01 7.96882749e-01 -3.77520412e-01 2.20787078e-01
-2.19416931e-01 7.87565529e-01 -8.16399813e-01 -4.75150257e-01
-1.54686123e-01 -1.49905905e-01 -4.51382607e-01 1.39701575e-01
5.71398318e-01 -5.01769900e-01 -6.79407656e-01 1.01001513e+00
3.27594459e-01 -1.58418581e-01 8.23784709e-01 -8.09072196e-01
-1.08060265e+00 8.21186304e-01 -6.82452679e-01 3.75292569e-01
1.61571786e-01 -2.27063358e-01 7.94116199e-01 -1.08463120e+00
6.66471183e-01 1.08020604e+00 6.85598135e-01 2.71127611e-01
-1.04169464e+00 -6.49663508e-01 6.44443110e-02 -1.78917363e-01
-1.24297822e+00 -4.91693974e-01 4.94986385e-01 -6.17406726e-01
3.23999912e-01 1.88176379e-01 -4.10405314e-03 1.44876075e+00
7.91549087e-01 5.77891111e-01 1.25524986e+00 -2.04950422e-01
3.13718766e-01 7.33800828e-02 4.11002487e-01 3.93825233e-01
5.39007366e-01 -1.47293925e-01 -3.33202183e-01 -4.72276956e-01
2.21243888e-01 5.61785877e-01 -9.47351754e-02 3.57120693e-01
-1.11172783e+00 1.21691179e+00 3.58253241e-01 3.30750823e-01
-1.59040332e-01 9.18061957e-02 1.54283330e-01 1.74943343e-01
7.87054241e-01 9.05707926e-02 -6.89482629e-01 1.84174001e-01
-1.05767286e+00 -5.65697402e-02 1.13637149e+00 1.08769345e+00
7.31807232e-01 -2.25453779e-01 -8.47331211e-02 7.49992549e-01
1.51423573e-01 5.88002384e-01 3.36919129e-01 -6.26366794e-01
4.36380953e-01 5.06852925e-01 3.71519566e-01 -1.04262388e+00
-6.58596337e-01 -3.99759620e-01 -7.47496009e-01 -4.16751690e-02
8.13724875e-01 -5.64193070e-01 -1.14158881e+00 1.42200172e+00
3.63606326e-02 -4.79274869e-01 -1.90300480e-01 6.90995753e-01
7.81175435e-01 5.36266446e-01 -1.18300371e-01 -4.36785854e-02
1.64039338e+00 -5.44932306e-01 -8.06964159e-01 -2.27045521e-01
3.03773880e-01 -6.99403465e-01 9.08484995e-01 2.15031385e-01
-4.71087635e-01 1.22717604e-01 -4.32351947e-01 -9.79004148e-03
-8.06730688e-01 -6.02084696e-02 5.06850839e-01 5.99312603e-01
-6.93425417e-01 5.53638458e-01 -6.75887465e-01 -8.78932953e-01
6.48573995e-01 -1.18700929e-01 -1.34719282e-01 -1.44442683e-02
-1.11050117e+00 4.90205228e-01 -1.81757942e-01 -1.21045887e-01
-1.38769412e+00 -1.73040301e-01 -4.87252623e-01 1.92475036e-01
3.76571357e-01 -1.00029349e-01 1.01329613e+00 -9.52958405e-01
-8.14014912e-01 7.06151843e-01 -3.27605009e-01 -2.98832655e-01
5.55141091e-01 -2.29948625e-01 -6.44753337e-01 1.04776174e-02
3.81345510e-01 1.59523472e-01 8.37614834e-01 -1.16888964e+00
-6.73008144e-01 -6.77189410e-01 -3.78809758e-02 -4.31545198e-01
-7.21307337e-01 3.50507259e-01 -2.97330856e-01 -9.33968306e-01
3.97406876e-01 -9.95955646e-01 1.53151795e-01 -4.77723271e-01
-6.88054621e-01 -3.82990569e-01 8.17813158e-01 -8.65817249e-01
1.24034977e+00 -1.93036652e+00 -5.42994797e-01 -1.78808525e-01
4.21734631e-01 -7.13613927e-01 2.18024209e-01 8.17424536e-01
1.91505313e-01 7.11339355e-01 -5.38545325e-02 -4.97961164e-01
4.54337848e-03 1.89773977e-01 -7.52110720e-01 8.31006944e-01
-2.81644404e-01 8.19327593e-01 -7.50674069e-01 -3.49535227e-01
-4.47756171e-01 2.35405773e-01 3.53137180e-02 -2.12624922e-01
-1.50926769e-01 4.63572532e-01 -3.73638600e-01 1.38736618e+00
7.60175407e-01 -3.31401467e-01 -4.24715653e-02 3.09024125e-01
-3.76023293e-01 2.01266363e-01 -1.15841739e-01 1.00821042e+00
-1.98957264e-01 1.06411648e+00 6.56478107e-02 -4.15635765e-01
1.11245656e+00 3.81633759e-01 2.90559918e-01 -1.78918406e-01
4.32131976e-01 2.50733316e-01 -6.68752611e-01 -3.70284766e-01
7.22698033e-01 -1.26065359e-01 -5.58361590e-01 5.56964576e-01
-2.61876702e-01 3.05062532e-01 -3.16955507e-01 3.84613216e-01
1.30770111e+00 -3.23413372e-01 3.26208882e-02 -3.19264084e-01
-2.47702777e-01 2.69914746e-01 4.45241243e-01 1.25315106e+00
-4.95796025e-01 1.06062078e+00 1.00679088e+00 -4.56672817e-01
-9.54597354e-01 -9.28237259e-01 -4.00581032e-01 1.15707481e+00
-1.34654254e-01 1.81465864e-01 -6.31727636e-01 -1.19911540e+00
-1.10230530e-02 9.88037467e-01 -7.76722491e-01 3.65064383e-01
-1.69721693e-01 -8.99812698e-01 6.12979949e-01 2.10221574e-01
3.87736738e-01 -1.04921257e+00 1.34159163e-01 1.47881180e-01
-3.42241943e-01 -9.98155475e-01 -8.10652435e-01 3.70091021e-01
-8.39569092e-01 -9.27425146e-01 -7.23632812e-01 -5.79663038e-01
8.58825684e-01 2.87961513e-01 9.52997625e-01 -2.03575388e-01
4.08917874e-01 4.83813465e-01 -5.92499614e-01 -5.31119823e-01
-2.30559617e-01 4.04364735e-01 -5.36879785e-02 3.38044822e-01
6.55440807e-01 -5.21784365e-01 -5.17523706e-01 5.36607921e-01
-1.01070404e+00 -6.99527621e-01 4.22544777e-01 5.51539004e-01
9.62094516e-02 1.03551641e-01 8.42997491e-01 -1.09931362e+00
3.97010744e-01 -1.33550727e+00 -6.74640894e-01 -7.36491233e-02
-7.57058024e-01 -6.53901637e-01 4.65448320e-01 -4.46610928e-01
-9.51066494e-01 5.22959977e-02 4.47128296e-01 -1.91627547e-01
-5.16593039e-01 9.27006721e-01 -4.92342785e-02 3.77732843e-01
7.13846922e-01 7.76516348e-02 -3.70015979e-01 -6.48988605e-01
-3.40331346e-02 1.13530540e+00 4.78717744e-01 -2.40379289e-01
7.13756323e-01 1.08547068e+00 -6.15228236e-01 -8.80597770e-01
-1.18978965e+00 -4.90664661e-01 -4.49677229e-01 -1.58184603e-01
9.49065208e-01 -9.31693017e-01 -5.53713858e-01 7.20835328e-01
-1.25282359e+00 -9.94330496e-02 2.84862190e-01 4.26349342e-01
1.75249100e-01 3.44688952e-01 -7.00036585e-01 -8.33757460e-01
-2.09026411e-01 -3.32569003e-01 7.85742700e-01 2.36050844e-01
-1.25305699e-02 -1.15509260e+00 -5.50782047e-02 5.70162416e-01
4.73469228e-01 6.64860725e-01 6.20894253e-01 -1.06464529e+00
-6.26945853e-01 -7.37960279e-01 -3.35700840e-01 1.02861300e-01
2.73604214e-01 -2.43197326e-02 -8.68418455e-01 -4.88770008e-01
1.87243000e-01 1.19012386e-01 1.01872003e+00 4.87739652e-01
9.68862295e-01 -6.14143193e-01 -3.73722672e-01 4.09427822e-01
1.27176023e+00 1.03904754e-02 5.66710711e-01 6.50820792e-01
6.48901403e-01 5.71665227e-01 3.08741421e-01 6.45255744e-01
6.34824395e-01 1.67114332e-01 7.51187980e-01 5.64433858e-02
4.05187041e-01 -2.81464100e-01 7.16789365e-01 5.08761764e-01
3.08765382e-01 -7.51502872e-01 -1.23857999e+00 1.09844327e+00
-1.62533057e+00 -8.06186736e-01 -5.97157001e-01 2.05558515e+00
5.77911973e-01 9.43040177e-02 -1.20904319e-01 -4.76848811e-01
1.11737311e+00 3.65649760e-01 -4.73351389e-01 -9.87665728e-02
-2.89227307e-01 -4.23813015e-01 1.09258068e+00 4.53761876e-01
-1.25875556e+00 8.73911023e-01 6.16730452e+00 3.76755565e-01
-1.04944980e+00 2.60152310e-01 8.34471881e-01 -1.74425006e-01
-4.71180320e-01 4.08982843e-01 -8.81675601e-01 7.48188436e-01
9.06961977e-01 1.45827204e-01 4.27497119e-01 8.47558260e-01
5.55475295e-01 -2.01124728e-01 -7.01229572e-01 3.74053717e-01
1.51914388e-01 -1.18277419e+00 -3.19136232e-01 5.90678513e-01
1.11714506e+00 5.28534710e-01 -1.52101278e-01 1.68988675e-01
5.50108552e-01 -1.03764319e+00 1.14192593e+00 8.11775923e-01
6.80393338e-01 -5.92091441e-01 9.04676735e-01 4.34794217e-01
-4.58644807e-01 -2.07397863e-01 -4.08469737e-01 -4.01987843e-02
1.66559294e-01 1.24601185e+00 -6.60495996e-01 2.87765682e-01
1.07513750e+00 7.29837000e-01 -8.29282761e-01 5.98348081e-01
-4.73464668e-01 1.23277032e+00 -3.61926168e-01 -1.65518031e-01
3.03168118e-01 5.63881807e-02 5.65764368e-01 9.88301158e-01
5.09814501e-01 1.37232572e-01 -1.51976377e-01 6.51036263e-01
-3.36450607e-01 -3.09654683e-01 -6.28666103e-01 -1.91484675e-01
7.40346014e-01 1.32419503e+00 -9.60074186e-01 -2.96351194e-01
-7.47763097e-01 9.03246284e-01 1.23363711e-01 6.66577578e-01
-1.11240804e+00 -5.40937901e-01 1.03301667e-01 4.69271928e-01
3.68154645e-01 -7.80695140e-01 -7.44809747e-01 -1.54639065e+00
-2.31527537e-01 -4.17439252e-01 4.64078248e-01 -6.97828054e-01
-1.63130081e+00 4.14654642e-01 -1.93189055e-01 -8.19107354e-01
4.76003289e-01 -1.18816249e-01 -9.30310369e-01 6.09234333e-01
-1.50803471e+00 -1.21642768e+00 -5.75293541e-01 6.30432606e-01
4.10153121e-01 -2.04041135e-02 4.10023749e-01 4.09736484e-01
-7.64472127e-01 3.29567492e-01 6.26962483e-01 4.51193482e-01
9.40167904e-01 -1.35112298e+00 6.01407707e-01 7.28434086e-01
-3.40118594e-02 7.70432353e-01 5.69206953e-01 -1.01294553e+00
-1.17830527e+00 -1.29631698e+00 1.55301106e+00 -1.10776711e+00
1.11471438e+00 -4.61975992e-01 -8.73549581e-01 1.07156205e+00
2.40076017e-02 6.81228936e-02 4.56817001e-01 1.13738306e-01
-1.96427077e-01 6.94168955e-02 -1.06001484e+00 3.09344560e-01
7.59330690e-01 -5.39082468e-01 -6.39504552e-01 4.75698471e-01
5.20960629e-01 -6.93215579e-02 -6.03783190e-01 -2.30272040e-01
5.08756816e-01 -7.64380455e-01 2.59364158e-01 -5.11080325e-01
5.98689854e-01 -1.03255704e-01 -1.89970002e-01 -8.69787753e-01
-3.86546344e-01 -7.84418643e-01 1.35758175e-02 1.88464105e+00
5.62773645e-01 -9.49552357e-01 9.14328992e-01 6.78535342e-01
-9.48543400e-02 -2.98583925e-01 -1.08636796e+00 -7.96923995e-01
5.23128569e-01 -2.43464530e-01 5.80148280e-01 1.25913119e+00
-3.32105219e-01 1.64423719e-01 -4.97297734e-01 4.71878350e-01
6.84634864e-01 7.63113275e-02 7.21745491e-01 -1.22018397e+00
4.16349888e-01 1.28541412e-02 2.43945152e-01 -9.82475519e-01
-9.34790224e-02 -7.37559080e-01 -1.94320530e-02 -1.51865709e+00
2.79229641e-01 -4.26067889e-01 -9.62951034e-02 7.37218738e-01
2.33575925e-01 3.36597413e-01 -2.83861756e-01 8.63437235e-01
-2.72859901e-01 3.58383119e-01 7.44474053e-01 -9.73732546e-02
-8.36366341e-02 1.23623110e-01 -9.86511588e-01 8.51578057e-01
8.85582507e-01 -9.59703743e-01 6.11690044e-01 -5.74120283e-01
4.23561037e-01 7.44524449e-02 6.56821489e-01 -4.46199834e-01
3.26676965e-01 -2.39831209e-01 4.53687787e-01 -1.11197209e+00
-3.55076551e-01 -9.40973997e-01 -9.10130441e-02 7.90296420e-02
3.97863239e-02 -1.39737532e-01 -1.95246935e-01 9.13151503e-01
-1.41595930e-01 -2.05163449e-01 6.64080560e-01 -7.17509016e-02
-1.31571189e-01 2.07479730e-01 -9.28292215e-01 -3.12987156e-02
8.30586672e-01 -6.75235093e-02 -7.24400580e-01 -8.54669809e-01
-9.49000955e-01 2.46220082e-01 6.27950609e-01 4.95325118e-01
2.71272600e-01 -1.22280490e+00 -8.27718139e-01 -2.27686107e-01
-3.27635407e-02 -3.06090295e-01 -1.31751616e-02 9.85590041e-01
-3.14943552e-01 3.21257114e-01 4.76857632e-01 -3.22873265e-01
-7.51401126e-01 4.48189825e-01 1.68767899e-01 2.72143520e-02
-4.51046407e-01 1.77898347e-01 4.54105586e-02 -4.69050825e-01
2.87976742e-01 -2.90958822e-01 -2.44120821e-01 8.51124823e-01
2.46349707e-01 5.90322912e-01 -7.02776462e-02 -5.33522129e-01
-4.88577813e-01 2.99961627e-01 -1.71031691e-02 1.64034627e-02
1.45832992e+00 -5.56631804e-01 -4.56922650e-01 4.69206810e-01
1.26972663e+00 5.08493245e-01 -1.28173876e+00 -5.46517447e-02
3.74857694e-01 -3.68823111e-01 2.78288145e-02 -1.00787294e+00
-1.06347811e+00 3.51429015e-01 2.31236622e-01 9.25464213e-01
7.50681818e-01 5.45989633e-01 9.16677892e-01 1.57635361e-01
3.69312584e-01 -8.71236861e-01 -8.36563706e-02 9.54651237e-01
9.64222014e-01 -1.40864754e+00 1.39942735e-01 8.58024135e-02
-3.68458152e-01 1.24492192e+00 6.19103797e-02 -5.81273958e-02
8.85477364e-01 -2.46094793e-01 1.29453182e-01 -6.84561133e-01
-4.39750284e-01 1.24366194e-01 -1.33573383e-01 1.81354880e-01
2.01268807e-01 1.94442570e-01 -4.92570668e-01 1.05766976e+00
-1.78856656e-01 -2.17457592e-01 1.20775485e+00 7.54784226e-01
-5.96336424e-01 -5.28875232e-01 -9.19485509e-01 6.38562500e-01
-1.08959162e+00 -1.96557939e-01 -7.86761761e-01 5.56250513e-01
-1.11776195e-01 1.31110692e+00 6.39137626e-02 -1.62977159e-01
9.08044502e-02 -7.08579049e-02 -4.55878317e-01 -6.83773160e-01
-6.37745619e-01 4.69391681e-02 1.86783403e-01 -1.99814606e-02
1.07156150e-01 -9.93578970e-01 -9.83169854e-01 -7.52500057e-01
-7.49444783e-01 4.05220948e-02 8.92452180e-01 7.15238094e-01
3.32084239e-01 9.69062224e-02 1.01397133e+00 -6.70294583e-01
-4.93151218e-01 -1.18440104e+00 -8.69419098e-01 1.60837471e-01
5.95468879e-01 -2.19392762e-01 -1.18013072e+00 1.41047090e-01]
|
[8.178330421447754, 10.25284194946289]
|
71918b96-330d-4a4f-a44c-837810c3317c
|
sgl-speaking-the-graph-languages-of-semantic
| null | null |
https://aclanthology.org/2021.naacl-main.30
|
https://aclanthology.org/2021.naacl-main.30.pdf
|
SGL: Speaking the Graph Languages of Semantic Parsing via Multilingual Translation
|
Graph-based semantic parsing aims to represent textual meaning through directed graphs. As one of the most promising general-purpose meaning representations, these structures and their parsing have gained a significant interest momentum during recent years, with several diverse formalisms being proposed. Yet, owing to this very heterogeneity, most of the research effort has focused mainly on solutions specific to a given formalism. In this work, instead, we reframe semantic parsing towards multiple formalisms as Multilingual Neural Machine Translation (MNMT), and propose SGL, a many-to-many seq2seq architecture trained with an MNMT objective. Backed by several experiments, we show that this framework is indeed effective once the learning procedure is enhanced with large parallel corpora coming from Machine Translation: we report competitive performances on AMR and UCCA parsing, especially once paired with pre-trained architectures. Furthermore, we find that models trained under this configuration scale remarkably well to tasks such as cross-lingual AMR parsing: SGL outperforms all its competitors by a large margin without even explicitly seeing non-English to AMR examples at training time and, once these examples are included as well, sets an unprecedented state of the art in this task. We release our code and our models for research purposes at https://github.com/SapienzaNLP/sgl.
|
['Roberto Navigli', 'Rocco Tripodi', 'Luigi Procopio']
|
2021-06-01
| null | null | null |
naacl-2021-4
|
['ucca-parsing']
|
['natural-language-processing']
|
[ 4.49545264e-01 4.68696564e-01 -2.06367806e-01 -4.72210556e-01
-1.32865584e+00 -8.37308347e-01 7.68743813e-01 1.53533980e-01
-3.15852851e-01 8.58084738e-01 4.06684607e-01 -5.05083919e-01
1.95392564e-01 -7.60111690e-01 -8.47546220e-01 -4.22025949e-01
2.69980520e-01 7.28192925e-01 -1.30274847e-01 -4.94204134e-01
-3.99829559e-02 1.43108487e-01 -9.65021908e-01 4.58834857e-01
6.95966661e-01 6.44132614e-01 2.13360399e-01 6.47675931e-01
-5.58892906e-01 7.68428504e-01 -4.47448701e-01 -1.03366721e+00
6.42111897e-02 -5.48159778e-01 -1.26646781e+00 -1.90564305e-01
4.15155560e-01 3.48271012e-01 2.11140458e-02 1.13873339e+00
3.43352199e-01 -1.18940711e-01 2.43785292e-01 -7.68608570e-01
-9.15746808e-01 9.62993026e-01 -1.72184959e-01 -4.63646911e-02
4.72065866e-01 7.42964968e-02 1.48019421e+00 -8.43347430e-01
8.53665888e-01 1.52984321e+00 5.43900788e-01 8.24213862e-01
-1.28639913e+00 -1.89372778e-01 3.62058222e-01 -5.03031164e-02
-8.39536965e-01 -3.73119265e-01 6.90308332e-01 9.58560109e-02
1.41114533e+00 1.13737501e-01 3.12848270e-01 1.59202695e+00
9.07640681e-02 9.63354886e-01 1.10086691e+00 -6.98308408e-01
6.77168816e-02 -1.87088683e-01 1.88426852e-01 7.08260655e-01
1.78040162e-01 -3.74900967e-01 -3.85028660e-01 -6.24923706e-02
5.01365304e-01 -4.32892084e-01 -1.02998525e-01 -1.36337474e-01
-1.40459561e+00 1.03610826e+00 3.04950714e-01 5.93685389e-01
-1.65672243e-01 4.88395780e-01 5.74671865e-01 5.29929578e-01
7.82769024e-01 5.13407826e-01 -7.82729030e-01 -1.73976168e-01
-3.84960890e-01 8.23488161e-02 9.23284411e-01 9.43574786e-01
6.32798195e-01 -2.15368066e-02 1.50704220e-01 1.03580809e+00
1.06610814e-02 5.31104386e-01 4.51459765e-01 -8.39013934e-01
8.45165431e-01 5.05865872e-01 -3.70094717e-01 -5.61444581e-01
-4.50614959e-01 -5.15225053e-01 -7.99930334e-01 -2.64626324e-01
5.94564378e-01 -2.79519796e-01 -7.30243206e-01 2.05179191e+00
1.43573716e-01 -1.26198053e-01 5.99873841e-01 7.51108170e-01
6.36667550e-01 7.09124148e-01 3.92105281e-01 1.37101844e-01
1.63995850e+00 -1.00695467e+00 -5.72532654e-01 -7.20726609e-01
1.02603471e+00 -6.68794990e-01 1.26461709e+00 3.01616251e-01
-1.16378677e+00 -3.02525878e-01 -7.56453097e-01 -3.45643997e-01
-6.08333468e-01 6.03016503e-02 1.03733325e+00 4.60366219e-01
-1.34390628e+00 6.98464632e-01 -8.67037356e-01 -8.30177307e-01
3.22595835e-01 1.35625809e-01 -4.61242914e-01 -2.82584220e-01
-1.36498058e+00 9.84963417e-01 4.45810944e-01 1.61891162e-01
-4.33588356e-01 -3.39178562e-01 -1.02247000e+00 -4.99720313e-02
4.32254434e-01 -1.07597339e+00 1.35238695e+00 -1.29050183e+00
-1.53650486e+00 1.25736177e+00 -1.82179958e-01 -6.74118578e-01
2.49057844e-01 -2.34837368e-01 -2.74019837e-01 2.39087511e-02
1.68916479e-01 7.25883484e-01 4.09362674e-01 -8.99341762e-01
-3.43032062e-01 -5.33251762e-01 3.07517976e-01 1.67673379e-01
-8.00883695e-02 3.79080802e-01 -2.31913164e-01 -6.71445727e-01
-6.63900003e-02 -9.07582939e-01 -4.28348571e-01 -5.48268974e-01
-3.30974698e-01 -4.62651968e-01 2.07105637e-01 -6.41453207e-01
8.36449802e-01 -1.85474885e+00 4.26166564e-01 -3.07004809e-01
-1.17364287e-01 1.70917079e-01 -5.02815783e-01 6.91823125e-01
-9.73819941e-02 3.56537074e-01 -8.12641740e-01 -5.55937827e-01
1.93055898e-01 5.83803058e-01 -2.35673785e-01 1.01712137e-01
7.06959546e-01 1.43589115e+00 -1.12620664e+00 -2.48793542e-01
2.42009349e-02 4.21863586e-01 -4.50159788e-01 8.63244012e-02
-6.25540495e-01 6.11678660e-01 -5.97295642e-01 5.06130993e-01
3.60760480e-01 -4.46300238e-01 6.20893717e-01 2.45060980e-01
1.87960133e-01 7.61971831e-01 -5.29808879e-01 2.41413569e+00
-7.37254322e-01 2.95210719e-01 2.54846662e-02 -1.40499532e+00
8.58839154e-01 3.82354617e-01 9.56168249e-02 -7.71360517e-01
6.97292462e-02 5.73554933e-01 -8.09773430e-02 -1.40109837e-01
4.79107767e-01 -4.18945819e-01 -4.37478870e-01 2.46654749e-01
4.11661714e-01 -1.52337372e-01 2.25456938e-01 1.58559263e-01
1.19651258e+00 5.42215884e-01 5.78997135e-01 -2.94273436e-01
5.49498439e-01 1.84034795e-01 2.58562446e-01 5.80412149e-01
1.93846285e-01 6.99456930e-01 6.86686099e-01 -4.48761195e-01
-9.77882266e-01 -9.36489880e-01 8.81130993e-03 1.17276311e+00
-2.31642917e-01 -4.60654706e-01 -8.77687454e-01 -1.01194751e+00
-3.10074985e-01 9.39010561e-01 -5.07496238e-01 2.03728393e-01
-9.20360208e-01 -9.30029809e-01 6.09704733e-01 4.73625392e-01
2.97989935e-01 -1.39326918e+00 -1.75583661e-01 4.63642240e-01
-1.92107350e-01 -1.44665062e+00 4.99043465e-02 3.74329835e-01
-9.37638938e-01 -8.00888598e-01 -6.36941075e-01 -8.38201523e-01
2.51019210e-01 -6.12747483e-02 1.71240258e+00 8.23226497e-02
1.00076430e-01 3.73827130e-01 -6.51878357e-01 -2.45685056e-01
-9.06547487e-01 5.49147964e-01 -5.27709186e-01 -1.90225542e-01
3.43660980e-01 -6.52868152e-01 -2.52522439e-01 -2.93085009e-01
-7.92639077e-01 3.00277561e-01 7.04254925e-01 7.74104774e-01
5.53311050e-01 -6.29704058e-01 7.27616429e-01 -1.40794909e+00
6.18432105e-01 -5.13443530e-01 -4.52835441e-01 2.96487600e-01
-2.51612276e-01 2.87168294e-01 9.52865124e-01 1.36942714e-01
-1.06862962e+00 -1.53876007e-01 -6.17374837e-01 4.66456488e-02
-5.19810617e-01 6.32292330e-01 -3.17809820e-01 4.11198556e-01
4.90826905e-01 1.28776774e-01 -1.81048661e-01 -6.13163114e-01
7.95546055e-01 4.20246392e-01 3.76871556e-01 -9.37591016e-01
5.00173390e-01 1.82679534e-01 1.33285999e-01 -6.34451270e-01
-1.25967383e+00 -1.56179190e-01 -6.00472331e-01 3.68678391e-01
1.31209254e+00 -8.66583645e-01 -2.73382962e-01 2.03233019e-01
-1.50458574e+00 -6.38980210e-01 -2.11406633e-01 2.43070781e-01
-6.16097927e-01 2.80967206e-01 -8.15091550e-01 -6.07438982e-01
-5.51811397e-01 -8.90915751e-01 1.29936755e+00 -2.14751258e-01
-1.98284805e-01 -1.51936233e+00 1.76849738e-02 4.08203900e-01
4.59444761e-01 3.47817689e-01 1.17877483e+00 -9.74802792e-01
-5.72206736e-01 -4.45345715e-02 -2.77188927e-01 3.79113913e-01
8.34428594e-02 -5.24878383e-01 -1.03243124e+00 -1.49609655e-01
-9.32849199e-02 -5.52491784e-01 9.01925623e-01 1.24334395e-01
9.34430540e-01 -2.96280861e-01 -3.98672409e-02 4.43712294e-01
1.45151281e+00 -2.71308631e-01 4.19224381e-01 4.35332626e-01
8.76419425e-01 6.62651181e-01 3.40022504e-01 -1.79493383e-01
6.01923108e-01 5.64785838e-01 5.71868360e-01 -1.68843091e-01
-3.51510435e-01 -2.98845947e-01 5.32941759e-01 1.13619018e+00
-8.41737166e-02 -4.58245188e-01 -9.69870090e-01 5.01748741e-01
-1.88505578e+00 -5.95977545e-01 -2.70581484e-01 2.07607722e+00
7.58126080e-01 1.29865870e-01 -1.42604545e-01 -3.98216933e-01
4.54794854e-01 4.33316618e-01 -2.56654859e-01 -7.92712033e-01
-4.68883038e-01 7.66212523e-01 3.82635206e-01 5.78494906e-01
-9.77967739e-01 1.56349230e+00 5.51486206e+00 8.20185483e-01
-9.79426801e-01 3.55446219e-01 6.79698765e-01 3.62768441e-01
-5.80238461e-01 2.14578509e-01 -6.15613341e-01 1.21965639e-01
1.28997195e+00 8.37472305e-02 6.12319887e-01 6.64053917e-01
-1.12438299e-01 2.99282819e-01 -1.21365261e+00 7.00554192e-01
-5.68640535e-04 -1.17278647e+00 3.22054565e-01 -1.46844998e-01
3.95670503e-01 4.88023013e-01 -1.85863212e-01 5.34424722e-01
6.98792636e-01 -1.10315394e+00 6.57534599e-01 -7.83087090e-02
6.50553942e-01 -5.12904227e-01 6.85573876e-01 2.08374113e-01
-1.17747545e+00 2.21631482e-01 -5.99004745e-01 -2.11645797e-01
3.48914862e-01 3.97098988e-01 -4.95839000e-01 1.15308368e+00
1.57515272e-01 9.39342916e-01 -6.51149988e-01 2.20331609e-01
-7.85095155e-01 8.59191120e-01 -1.10303499e-01 -8.38509127e-02
5.95390320e-01 -3.38277608e-01 4.99151945e-01 1.54779327e+00
3.08066010e-01 -1.14118628e-01 1.60031661e-01 8.72141838e-01
-3.85154486e-01 4.81055975e-01 -7.52051711e-01 -2.97779322e-01
-1.97192691e-02 1.40287113e+00 -7.01119363e-01 -4.60148245e-01
-8.04715276e-01 1.14266872e+00 9.22440886e-01 3.19273204e-01
-7.03039587e-01 -7.78080150e-02 6.20074034e-01 -2.25487858e-01
9.22759175e-02 -3.28509390e-01 -2.71393150e-01 -1.53188634e+00
9.82555151e-02 -9.71931100e-01 7.00356007e-01 -8.61627400e-01
-1.54296291e+00 8.52791727e-01 -1.59925193e-01 -8.08402061e-01
-4.59400564e-01 -1.02370226e+00 -3.96947116e-01 9.06932414e-01
-1.76899278e+00 -1.49442160e+00 2.81233579e-01 2.32959896e-01
8.89795244e-01 -2.05423031e-03 1.25438774e+00 1.05627418e-01
-3.65143359e-01 4.73302364e-01 -2.24559814e-01 3.37274641e-01
5.81528008e-01 -1.50145018e+00 1.07534015e+00 9.51412857e-01
7.10646629e-01 5.94603419e-01 4.83752698e-01 -3.91410112e-01
-1.66201341e+00 -1.10957778e+00 1.23940384e+00 -7.14817703e-01
1.17091501e+00 -7.30717838e-01 -1.07838118e+00 1.07891285e+00
5.63656330e-01 -4.83867154e-02 3.87670606e-01 5.02385199e-01
-4.71150935e-01 2.79801667e-01 -7.07486212e-01 5.99704444e-01
1.45550156e+00 -5.11965275e-01 -6.87401175e-01 5.83989084e-01
1.10223770e+00 -4.56440508e-01 -8.44719827e-01 2.64845759e-01
1.63926452e-01 -7.55873144e-01 6.87428892e-01 -9.54409659e-01
7.43474543e-01 2.16619894e-01 -4.04971749e-01 -1.41190612e+00
-7.82035962e-02 -6.32813394e-01 3.04141015e-01 1.29664218e+00
8.26633692e-01 -9.94849265e-01 6.20663166e-01 2.49006003e-01
-4.54177171e-01 -7.14623213e-01 -9.16861176e-01 -8.49130154e-01
7.13741064e-01 -6.55685604e-01 4.36570108e-01 1.05060768e+00
-6.83825985e-02 9.84004557e-01 -1.49732515e-01 -1.56035530e-03
5.20344257e-01 2.88906246e-01 6.86200500e-01 -1.11821055e+00
-5.54754555e-01 -5.20946264e-01 -1.41326115e-01 -1.09555173e+00
7.20124543e-01 -1.68210518e+00 -2.55695790e-01 -1.82001424e+00
1.83371812e-01 -2.81797677e-01 -2.26889446e-01 8.26381266e-01
-3.41639251e-01 2.56635725e-01 2.52834767e-01 -2.69745737e-02
-7.11571097e-01 4.16611522e-01 1.19740319e+00 7.30289072e-02
2.41346672e-01 -1.77953050e-01 -8.45332980e-01 6.71620369e-01
9.20717418e-01 -5.45012593e-01 -1.26516908e-01 -8.99611592e-01
4.87092078e-01 1.81600347e-01 4.04797375e-01 -5.91395974e-01
-3.39788288e-01 1.14106219e-02 -7.66934529e-02 4.14871164e-02
1.83099374e-01 -5.41554570e-01 -1.26231117e-02 2.33694285e-01
-4.33192164e-01 2.00125411e-01 2.68700033e-01 3.22134286e-01
-2.83897400e-01 -2.21352369e-01 5.11009693e-01 -4.97377962e-01
-6.89739883e-01 1.11960009e-01 -1.77637875e-01 4.87905651e-01
3.17180604e-01 2.07683548e-01 -4.15471554e-01 -2.63062209e-01
-7.61308193e-01 1.71436649e-02 3.82295370e-01 6.16014183e-01
8.62402022e-02 -1.02982497e+00 -9.03252423e-01 -1.36287063e-01
1.89200982e-01 2.57647671e-02 -3.79421823e-02 7.44822979e-01
-2.25840122e-01 7.06166387e-01 9.12103876e-02 -4.23658371e-01
-9.33546484e-01 5.49375832e-01 1.90122113e-01 -6.59211457e-01
-7.50088334e-01 6.79513454e-01 3.88061404e-01 -7.99458265e-01
-3.43691498e-01 -4.80072916e-01 -1.17320968e-02 -2.95173556e-01
1.69496551e-01 -1.69024408e-01 2.12180942e-01 -5.77851236e-01
-2.93410569e-01 5.77809453e-01 1.16672292e-01 -1.96048766e-02
1.39770484e+00 -5.14145382e-02 -2.52612829e-01 4.48223352e-01
1.15709686e+00 7.36557096e-02 -8.62983584e-01 -2.76570857e-01
4.30777758e-01 9.87152830e-02 -3.94632369e-01 -1.01118922e+00
-9.74497259e-01 1.04670012e+00 -2.00582184e-02 2.90818602e-01
1.00805771e+00 3.97916079e-01 8.57869148e-01 4.59773183e-01
6.00552797e-01 -6.56094611e-01 -1.20455340e-01 9.05584395e-01
8.09850991e-01 -1.32048285e+00 -4.38492924e-01 -4.90933716e-01
-6.48871362e-01 1.10260522e+00 9.44046080e-02 -2.71854222e-01
9.02380720e-02 1.66203320e-01 2.13783830e-01 -1.87381074e-01
-8.28213871e-01 -5.02414227e-01 9.06270519e-02 5.29216588e-01
8.82549584e-01 2.20273808e-01 -5.00412703e-01 6.94995821e-01
-4.11298752e-01 -1.84772357e-01 3.49180609e-01 7.21772969e-01
-2.72901893e-01 -1.73909748e+00 1.25423640e-01 6.65741190e-02
-9.51570511e-01 -5.72552145e-01 -5.56680739e-01 1.03187513e+00
-4.05399561e-01 8.71300876e-01 -2.07259670e-01 1.47337243e-01
3.93472642e-01 5.00659883e-01 7.37603962e-01 -9.31714535e-01
-7.86976993e-01 -9.46928039e-02 5.73555350e-01 -6.29361331e-01
-4.50882196e-01 -5.91869414e-01 -1.49692726e+00 -8.79612342e-02
9.86640751e-02 1.50209039e-01 7.58943677e-01 1.06426036e+00
3.47808689e-01 5.13588846e-01 1.04163162e-01 -6.30462050e-01
-6.39065444e-01 -1.00989056e+00 -2.32997686e-01 5.47420204e-01
-6.11554384e-02 -1.38563260e-01 -2.85890609e-01 -5.14842272e-02]
|
[10.548314094543457, 9.3319673538208]
|
4f94ba46-348d-4ddd-ba8a-1e4f89dd1404
|
a-force-sensing-surgical-drill-for-real-time
|
2304.02583
| null |
https://arxiv.org/abs/2304.02583v1
|
https://arxiv.org/pdf/2304.02583v1.pdf
|
A force-sensing surgical drill for real-time force feedback in robotic mastoidectomy
|
Purpose: Robotic assistance in otologic surgery can reduce the task load of operating surgeons during the removal of bone around the critical structures in the lateral skull base. However, safe deployment into the anatomical passageways necessitates the development of advanced sensing capabilities to actively limit the interaction forces between the surgical tools and critical anatomy. Methods: We introduce a surgical drill equipped with a force sensor that is capable of measuring accurate tool-tissue interaction forces to enable force control and feedback to surgeons. The design, calibration and validation of the force-sensing surgical drill mounted on a cooperatively controlled surgical robot are described in this work. Results: The force measurements on the tip of the surgical drill are validated with raw-egg drilling experiments, where a force sensor mounted below the egg serves as ground truth. The average root mean square error (RMSE) for points and path drilling experiments are 41.7 (pm 12.2) mN and 48.3 (pm 13.7) mN respectively. Conclusions: The force-sensing prototype measures forces with sub-millinewton resolution and the results demonstrate that the calibrated force-sensing drill generates accurate force measurements with minimal error compared to the measured drill forces. The development of such sensing capabilities is crucial for the safe use of robotic systems in a clinical context.
|
['Deepa Galaiya', 'Russell Taylor', 'Francis Creighton', 'Katherina Sapozhnikov', 'Harsha Mohan', 'Seena Vafaee', 'Aditi Kishore', 'Manish Sahu', 'Anna Goodridge', 'Yuxin Chen']
|
2023-04-05
| null | null | null | null |
['anatomy']
|
['miscellaneous']
|
[-2.92776138e-01 6.57655835e-01 1.98286459e-01 4.11554784e-01
-3.27603787e-01 -3.90973181e-01 -2.87466913e-01 -7.71796778e-02
-7.26048648e-01 2.70492196e-01 2.27979958e-01 -1.68129817e-01
-4.55053091e-01 1.76803079e-02 -6.05635226e-01 -5.06487310e-01
-4.57980782e-01 2.53971249e-01 3.96277398e-01 -4.82123196e-01
7.04149306e-01 6.34616852e-01 -1.21175897e+00 -2.84286231e-01
6.79784715e-01 9.99272406e-01 8.23109806e-01 3.62748951e-01
6.19776607e-01 4.77846354e-01 -2.99136221e-01 1.28345966e-01
6.54330850e-01 4.78159040e-01 -5.48461974e-01 -6.36283815e-01
-9.36856940e-02 -5.24509847e-01 9.37055275e-02 7.96823978e-01
7.49495864e-01 -1.97213754e-01 3.50768298e-01 -6.02218091e-01
-2.33198069e-02 6.82746589e-01 -6.27907634e-01 -2.96681374e-01
4.50923145e-01 -1.08658768e-01 1.81752294e-01 -1.18965459e+00
1.11868131e+00 4.94684875e-01 1.10587597e+00 7.71451414e-01
-7.27308035e-01 -8.75729382e-01 -7.02985823e-01 -4.46951866e-01
-1.21239102e+00 -4.65026736e-01 4.49849248e-01 -9.78602648e-01
8.73031378e-01 -2.95278560e-02 1.22308850e+00 3.77735019e-01
1.77120185e+00 -6.18874170e-02 6.94525659e-01 -5.60886085e-01
7.44601712e-02 -4.03775275e-02 -5.41534126e-01 6.40812695e-01
6.49295390e-01 5.11148453e-01 -3.75105530e-01 -8.43017325e-02
1.41454518e+00 -6.67083040e-02 -5.74434638e-01 -4.74738866e-01
-1.18203139e+00 2.20598057e-01 7.38988161e-01 5.08133709e-01
-7.58899093e-01 2.84105033e-01 2.75927335e-01 5.26097417e-02
-2.37542450e-01 9.28389728e-01 -1.88063368e-01 -7.52517760e-01
-3.81004065e-01 -1.04375422e-01 9.13689911e-01 9.16013956e-01
-4.00398433e-01 -2.48708770e-01 3.28272223e-01 7.70315886e-01
4.62305427e-01 4.72615153e-01 7.60644555e-01 -9.40616429e-01
3.55063140e-01 5.18400192e-01 4.13570940e-01 -8.06913733e-01
-1.06111479e+00 -2.82450616e-01 -2.83662617e-01 2.25976184e-01
1.72757238e-01 -5.99961817e-01 -9.04505432e-01 1.03535461e+00
2.40987405e-01 -6.27496779e-01 -1.73836321e-01 1.11633098e+00
4.62299079e-01 -5.41865647e-01 -3.92752022e-01 -3.04140866e-01
1.11352491e+00 -4.12093103e-01 -5.35565436e-01 -4.57402706e-01
6.48738503e-01 -9.06507909e-01 1.07068717e+00 4.57078964e-01
-1.20450783e+00 -2.36480325e-01 -9.54631925e-01 2.16736197e-01
4.36174840e-01 3.04397136e-01 4.46917921e-01 2.41514385e-01
-7.59911001e-01 5.24567962e-01 -1.25290775e+00 -1.74604058e-01
-2.35896870e-01 7.17821538e-01 -5.20527720e-01 5.55320621e-01
-6.25123143e-01 1.50596583e+00 -1.84219554e-01 4.44231480e-01
-2.38414407e-01 -6.63515270e-01 -4.64823782e-01 -3.17522347e-01
-2.12571457e-01 -4.73032832e-01 1.51828980e+00 4.34283763e-01
-1.97906899e+00 8.64008844e-01 7.97386348e-01 -3.91180158e-01
5.72138250e-01 -7.48182297e-01 7.40296245e-02 3.68986905e-01
7.80380145e-02 3.82147223e-01 4.39309895e-01 -6.27191782e-01
-7.44105726e-02 -4.19371545e-01 -3.16772401e-01 -9.20916572e-02
-2.52869070e-01 -2.51230925e-01 2.93188065e-01 -6.62935436e-01
9.21592236e-01 -1.51689303e+00 -3.67611498e-01 5.28032482e-01
-1.03478357e-01 2.79283911e-01 2.86816031e-01 -4.76106912e-01
9.57884908e-01 -2.25559545e+00 2.73259394e-02 1.78739399e-01
1.05618328e-01 3.23292278e-02 6.31427109e-01 8.23475242e-01
1.95289299e-01 -4.65309173e-01 1.87003329e-01 5.01717031e-01
-6.16265833e-01 -5.45031205e-02 1.66015968e-01 7.97793806e-01
-7.59559631e-01 4.78906751e-01 -7.13478565e-01 -3.99489075e-01
-2.31396377e-01 1.21633865e-01 -7.93603122e-01 8.76949579e-02
3.94974828e-01 6.87083125e-01 -2.22206339e-01 1.01236868e+00
4.38858867e-01 2.94469148e-01 -2.99468189e-02 -2.23397404e-01
-2.37728983e-01 2.11085662e-01 -7.93005586e-01 2.13435411e+00
-8.17406714e-01 3.03162020e-02 7.73429036e-01 1.77609965e-01
1.44838595e+00 4.82668340e-01 8.88222098e-01 -4.48843539e-01
5.47120452e-01 1.05748665e+00 5.64050496e-01 -7.92989731e-01
2.16660082e-01 -6.27478540e-01 -1.78438034e-02 4.02677983e-01
7.09971879e-03 -1.17343736e+00 -6.02501452e-01 -9.56282318e-02
1.01658630e+00 -1.23466484e-01 3.10433477e-01 -7.20693231e-01
3.52060013e-02 1.86704248e-02 2.80332446e-01 2.46694297e-01
-1.52052298e-01 7.69799575e-02 4.56480170e-03 -3.44408721e-01
-7.51594901e-01 -1.07059264e+00 -3.86936396e-01 3.94424319e-01
3.33499759e-01 2.13647429e-02 -5.26746869e-01 5.81271127e-02
7.26129353e-01 -6.82952031e-02 -4.39747751e-01 -3.02273989e-01
-3.46354008e-01 3.25645238e-01 4.11912531e-01 6.21670008e-01
-3.64596695e-01 -8.38116527e-01 -1.40134013e+00 4.53833640e-01
7.86595270e-02 -1.06161606e+00 -2.27556869e-01 2.19807744e-01
-1.06326306e+00 -1.18884730e+00 -3.64713132e-01 -6.51587546e-01
7.13558435e-01 -1.02017760e-01 2.81483412e-01 -8.94017592e-02
-5.03017366e-01 5.48087001e-01 -4.36823487e-01 -1.14237022e+00
-3.39696676e-01 -1.97761375e-02 5.38497388e-01 -9.57571507e-01
-2.22757503e-01 -5.73413372e-01 -9.57335830e-01 5.74541807e-01
-3.33888590e-01 -6.02132082e-02 9.80728090e-01 5.32819629e-01
2.93966055e-01 -8.88180673e-01 4.62982893e-01 -1.97569534e-01
8.91587257e-01 -3.07736155e-02 -3.59796345e-01 -4.91952598e-01
-4.12477463e-01 -1.36876553e-01 4.20356423e-01 -5.84874570e-01
-7.14122772e-01 1.62698068e-02 6.65460015e-03 -2.88058013e-01
5.44318497e-01 7.47871518e-01 8.70338678e-01 -5.90255976e-01
1.46579647e+00 -5.23233891e-01 9.08402145e-01 -2.38210365e-01
-2.55393714e-01 7.63446093e-01 7.59293020e-01 -5.47248185e-01
3.10339779e-01 3.15840632e-01 3.65374774e-01 -4.68213767e-01
-1.20113745e-01 -4.56464082e-01 -6.34923518e-01 -6.20598555e-01
3.74006957e-01 -5.05139887e-01 -1.08629084e+00 8.22381005e-02
-7.11277902e-01 -3.88528466e-01 -2.48455152e-01 1.32924163e+00
-7.84469604e-01 7.01513663e-02 -9.63630199e-01 -6.97461665e-01
-1.04338074e+00 -1.26079106e+00 8.90704095e-01 -3.76596265e-02
-6.00649893e-01 -4.74725485e-01 1.49770200e-01 -7.45318457e-02
5.20509541e-01 6.35858536e-01 1.34230211e-01 1.47957757e-01
-6.65668622e-02 -1.14875245e+00 6.04805827e-01 6.11781552e-02
2.34553382e-01 -2.70843431e-02 -4.71692413e-01 -5.55398703e-01
4.53059614e-01 -1.99488819e-01 -8.88530612e-02 5.21029890e-01
7.28000224e-01 -3.47916514e-01 -5.47370493e-01 3.45107734e-01
1.42149806e+00 1.22386709e-01 3.96241814e-01 1.86489210e-01
5.38240820e-02 6.01553380e-01 1.03524888e+00 3.34731698e-01
-2.82073766e-01 6.83097899e-01 4.60774302e-01 6.10291839e-01
8.08576569e-02 -1.01927020e-01 1.68325678e-01 1.17679250e+00
-5.06687582e-01 7.70153046e-01 -1.22899890e+00 4.39177781e-01
-1.07830107e+00 -3.04750234e-01 2.03766972e-02 2.53886914e+00
9.22798991e-01 2.32718706e-01 -5.13472021e-01 1.48874894e-01
8.65045488e-02 -7.66990006e-01 -4.74409968e-01 -7.03004539e-01
1.03651810e+00 2.73690760e-01 1.05995917e+00 4.62223470e-01
-2.05548644e-01 5.01273274e-01 6.39775896e+00 -1.35018021e-01
-1.69628787e+00 -1.30610958e-01 -9.00583744e-01 -3.32538843e-01
2.36522377e-01 -4.89808410e-01 -5.02983630e-01 4.82737601e-01
5.81475616e-01 -3.24577898e-01 -1.88713834e-01 1.01311433e+00
2.26212427e-01 -5.24419367e-01 -9.35075223e-01 7.73105323e-01
-2.84873813e-01 -1.28360081e+00 -4.92700994e-01 2.35421788e-02
3.28036457e-01 7.74139240e-02 -6.11581653e-02 -8.87495652e-02
-3.34912926e-01 -5.01220345e-01 7.31981456e-01 1.08655047e+00
1.22072077e+00 -2.20471263e-01 7.52425313e-01 6.40398622e-01
-1.00530136e+00 -5.77912390e-01 -4.39584762e-01 -5.94821870e-01
3.30609828e-01 7.04285502e-01 -1.69135463e+00 1.46077126e-01
6.35677695e-01 8.75386894e-02 1.99174523e-01 1.08555126e+00
-1.71209738e-01 -1.74494296e-01 -5.22087574e-01 -2.09645137e-01
-3.00310820e-01 2.52964407e-01 6.80512846e-01 4.15039659e-01
5.43890953e-01 2.92554289e-01 -3.64612550e-01 4.20416683e-01
1.93880707e-01 2.50057548e-01 -4.28091854e-01 2.32055977e-01
6.43532515e-01 1.07518828e+00 -4.51787353e-01 6.89642429e-01
2.19844520e-01 2.33226761e-01 -1.33709878e-01 -3.56676787e-01
-3.54092628e-01 -7.21979320e-01 6.12865567e-01 6.59758568e-01
-3.81926417e-01 -5.10981858e-01 -6.42642021e-01 -6.62268877e-01
5.87905526e-01 1.10028580e-01 -3.99489492e-01 -1.10898995e+00
-4.94043291e-01 2.45921105e-01 -2.58368105e-01 -1.72367513e+00
-5.03413796e-01 -8.79885733e-01 -4.30804640e-01 6.01994455e-01
-5.78484237e-01 -6.84508920e-01 -6.04552150e-01 2.59771854e-01
-8.53330866e-02 2.26418167e-01 7.98071444e-01 -4.07864809e-01
1.70981050e-01 4.35946137e-01 -3.93198788e-01 -1.71019197e-01
9.72078323e-01 -7.51076519e-01 -1.93261251e-01 3.99971187e-01
-1.05990970e+00 1.19468617e+00 8.44054341e-01 -8.36910307e-01
-1.77601635e+00 -3.24427575e-01 5.57697058e-01 -1.51498795e-01
6.61868393e-01 -2.54455388e-01 -5.41606188e-01 6.64051950e-01
-5.71065187e-01 3.40379700e-02 7.51547694e-01 -1.26470059e-01
1.85654610e-01 -1.33886606e-01 -1.58476067e+00 5.16750991e-01
1.04771364e+00 -4.34621871e-01 -8.42378914e-01 3.02697659e-01
2.53288597e-01 -1.45157087e+00 -1.45933461e+00 8.18495572e-01
1.68437541e+00 -5.32504976e-01 8.20377171e-01 1.08406276e-01
4.07391816e-01 2.35587079e-02 3.69292974e-01 -1.17566061e+00
-1.45312414e-01 -5.76106787e-01 1.21111438e-01 1.02532491e-01
3.88356864e-01 -1.06003940e+00 7.51172841e-01 6.59991860e-01
-6.07757568e-01 -1.33717358e+00 -1.26539469e+00 -5.68364918e-01
5.95939830e-02 -1.23402581e-01 -9.07260925e-02 1.59965217e-01
1.36355877e+00 -3.30773890e-01 4.67349112e-01 2.89112031e-01
5.74918911e-02 -2.44036734e-01 7.06485569e-01 -1.52563381e+00
-4.88555841e-02 -2.42361188e-01 -8.70321810e-01 -4.00859594e-01
-4.83598202e-01 -5.72532952e-01 3.69898528e-01 -1.74376214e+00
-4.45564330e-01 -6.41347229e-01 3.04345220e-01 1.82230130e-01
3.26049149e-01 -1.95395306e-01 1.05716735e-01 6.13486469e-01
6.91332281e-01 2.95576882e-02 1.79777741e+00 6.99095488e-01
-5.67259908e-01 2.22535506e-01 -2.38795742e-01 7.75996447e-01
6.51839316e-01 -4.00428027e-01 -1.94623262e-01 -1.96466967e-01
4.51366574e-01 5.13059914e-01 2.11736280e-02 -1.40326047e+00
9.89852250e-01 5.50693041e-03 1.53838322e-01 -1.84303463e-01
1.60070390e-01 -1.30570197e+00 3.99647564e-01 1.34742415e+00
1.25924200e-01 -6.13627136e-02 2.86888093e-01 1.62611440e-01
4.68927696e-02 -2.78844610e-02 9.03752089e-01 -9.44302306e-02
7.14377463e-02 -2.41500914e-01 -3.65158707e-01 -3.95394802e-01
1.15670013e+00 -5.64904571e-01 -1.22320414e-01 -5.05905971e-02
-1.05448413e+00 -3.32517087e-01 4.26927418e-01 1.07187852e-01
8.29622209e-01 -7.04313993e-01 -2.72933543e-01 4.56779510e-01
1.58251539e-01 3.61092746e-01 4.37861949e-01 1.67837846e+00
-1.20095766e+00 3.25135916e-01 -7.09982395e-01 -7.57120192e-01
-8.36952031e-01 1.42919034e-01 1.00936067e+00 3.93579602e-01
-6.06239259e-01 1.19699037e+00 -2.54722685e-01 -3.35912824e-01
7.69945234e-02 -9.72220302e-01 2.00027317e-01 -3.36122572e-01
1.57106757e-01 2.74729341e-01 3.94900590e-01 1.40091479e-01
-3.98578823e-01 9.34967041e-01 1.56805307e-01 -8.42750892e-02
1.37023461e+00 -3.99770867e-03 -1.72175616e-02 3.57565612e-01
4.39167559e-01 6.79659963e-01 -1.02372777e+00 2.80821770e-01
-2.45233536e-01 -7.27874041e-01 -2.26304904e-01 -9.95001614e-01
-9.09854829e-01 5.37081778e-01 3.69080216e-01 -1.93988994e-01
9.21519518e-01 1.04616202e-01 5.69207966e-01 2.72808164e-01
1.15572512e+00 -1.24168539e+00 -1.78948585e-02 1.96601540e-01
1.79659748e+00 -3.74365956e-01 -1.08827658e-01 -9.00582373e-01
-1.80294469e-01 1.10389626e+00 5.38450837e-01 -4.23370659e-01
1.14759457e+00 9.48361099e-01 4.48049992e-01 -5.29891670e-01
-1.74880370e-01 4.74055558e-01 1.96045160e-01 2.51063526e-01
7.89434850e-01 3.69533122e-01 -1.15238667e+00 4.29020375e-01
-1.00885177e+00 6.92912698e-01 7.82366335e-01 1.65310001e+00
-6.24055147e-01 -2.35614404e-01 -3.09376121e-01 7.82801449e-01
-6.43065333e-01 3.92804205e-01 1.47811323e-01 7.74348438e-01
-1.46709576e-01 8.26046288e-01 -1.56511649e-01 -3.41375530e-01
1.04570496e+00 -2.72554904e-01 7.00695634e-01 -7.19182611e-01
-9.23031092e-01 1.65844098e-01 -1.95471630e-01 -8.52968693e-01
7.44294673e-02 -3.49681497e-01 -1.72145486e+00 1.05482273e-01
-6.08324349e-01 1.04440033e-01 1.38960624e+00 3.34494233e-01
5.64006269e-01 5.45362830e-01 2.78181911e-01 -1.06537879e+00
-8.46261144e-01 -1.42300487e+00 -1.02553582e+00 -6.00775331e-02
1.97113723e-01 -1.38422251e+00 -3.73375475e-01 -4.55089241e-01]
|
[13.766202926635742, -3.014909505844116]
|
86a3c5b6-12db-44b1-818d-4a82c2bc688f
|
speaker-verification-across-ages
|
2306.07501
| null |
https://arxiv.org/abs/2306.07501v1
|
https://arxiv.org/pdf/2306.07501v1.pdf
|
Speaker Verification Across Ages: Investigating Deep Speaker Embedding Sensitivity to Age Mismatch in Enrollment and Test Speech
|
In this paper, we study the impact of the ageing on modern deep speaker embedding based automatic speaker verification (ASV) systems. We have selected two different datasets to examine ageing on the state-of-the-art ECAPA-TDNN system. The first dataset, used for addressing short-term ageing (up to 10 years time difference between enrollment and test) under uncontrolled conditions, is VoxCeleb. The second dataset, used for addressing long-term ageing effect (up to 40 years difference) of Finnish speakers under a more controlled setup, is Longitudinal Corpus of Finnish Spoken in Helsinki (LCFSH). Our study provides new insights into the impact of speaker ageing on modern ASV systems. Specifically, we establish a quantitative measure between ageing and ASV scores. Further, our research indicates that ageing affects female English speakers to a greater degree than male English speakers, while in the case of Finnish, it has a greater impact on male speakers than female speakers.
|
['Tomi Kinnunen', 'Md Sahidullah', 'Vishwanath Pratap Singh']
|
2023-06-13
| null | null | null | null |
['speaker-verification']
|
['speech']
|
[-1.29140422e-01 1.27074793e-01 1.51241124e-01 -3.66176367e-01
-6.94004297e-01 -3.42363238e-01 7.78425336e-01 1.45942867e-01
-9.07672703e-01 4.63664174e-01 6.83313072e-01 -6.91072941e-01
1.78491637e-01 -3.48724246e-01 -4.40254897e-01 -6.52516961e-01
-4.53186184e-02 1.16903894e-01 -1.38563037e-01 -3.29176515e-01
-1.47590086e-01 2.69843042e-01 -1.57725906e+00 -2.03135580e-01
6.24229133e-01 4.01863217e-01 -1.26129523e-01 7.15798616e-01
1.31439060e-01 9.31580663e-02 -9.63989198e-01 -6.47014022e-01
-5.59555478e-02 -2.04771712e-01 -6.52441323e-01 -4.54303056e-01
9.93002713e-01 -2.93723255e-01 -6.62154913e-01 6.84396327e-01
1.22886467e+00 -1.13459960e-01 5.92766225e-01 -1.12090147e+00
-6.21254563e-01 1.20473766e+00 -1.07349001e-01 4.62651253e-01
1.82013512e-01 2.63246000e-01 7.22765625e-01 -8.11176121e-01
4.18293923e-01 1.55673099e+00 8.47203493e-01 9.93907928e-01
-1.16206229e+00 -1.11743391e+00 1.51330322e-01 2.36985117e-01
-1.42488241e+00 -1.08587682e+00 5.80745935e-01 -5.81315458e-01
6.77730262e-01 2.74863243e-01 5.52450359e-01 1.58564043e+00
3.05547770e-02 5.07237017e-01 1.01789320e+00 -5.63556194e-01
1.45995438e-01 5.48321664e-01 5.65907955e-01 3.88884135e-02
1.41674682e-01 4.52932239e-01 -8.35927129e-01 -1.37057289e-01
-5.47564179e-02 -6.55192435e-01 -4.46973383e-01 2.72032678e-01
-1.05496931e+00 8.84672821e-01 -1.79977849e-01 7.65170038e-01
-1.77899823e-01 7.66859278e-02 6.73703790e-01 6.16447747e-01
4.57582355e-01 -2.30891835e-02 -6.46035135e-01 -4.46565390e-01
-1.09955156e+00 4.95668203e-02 6.39180005e-01 2.37237260e-01
2.68770993e-01 4.63853657e-01 -2.62086421e-01 9.46309805e-01
5.49646974e-01 6.37187064e-01 9.95888948e-01 -3.41561496e-01
4.91345644e-01 2.04218738e-02 -3.32838953e-01 -5.41138232e-01
-1.64104477e-01 -5.36450922e-01 -5.77844203e-01 3.96556258e-01
6.54104292e-01 -2.16282159e-01 -7.73999155e-01 2.38383436e+00
1.33448035e-01 3.48327979e-02 2.92899787e-01 4.39344615e-01
9.18944538e-01 3.58618796e-01 3.13120663e-01 -1.90713078e-01
1.55544221e+00 -4.07403231e-01 -6.52408957e-01 -3.62112612e-01
7.30830133e-01 -7.07235515e-01 1.28934240e+00 1.68252826e-01
-8.71540964e-01 -5.30651689e-01 -1.07837105e+00 2.09746882e-01
-5.79545617e-01 2.30447166e-02 9.45348218e-02 1.55296087e+00
-1.48883426e+00 3.39785427e-01 -3.86059523e-01 -5.56970894e-01
1.47245228e-01 3.21711570e-01 -4.96498466e-01 7.93951303e-02
-1.75846851e+00 1.02911329e+00 -1.81743249e-01 2.64128689e-02
-7.27238834e-01 -1.00836313e+00 -1.05727935e+00 -6.93084002e-02
-3.14598858e-01 -1.49272740e-01 1.19917083e+00 -6.29160047e-01
-1.38081181e+00 1.09325242e+00 -3.37404668e-01 -4.99045759e-01
5.73289573e-01 -1.97944045e-01 -8.60346675e-01 -4.92500842e-01
-1.14879899e-01 4.28485453e-01 8.02700162e-01 -9.03493822e-01
-2.17094287e-01 -8.83506656e-01 -5.27929723e-01 -1.33797685e-02
-6.11031413e-01 3.65538090e-01 1.33874774e-01 -6.29833221e-01
-4.08524245e-01 -9.66269732e-01 3.76955032e-01 -2.98498720e-01
-1.38732150e-01 -4.15453672e-01 9.64742720e-01 -1.32197285e+00
1.45368254e+00 -2.38453078e+00 -1.15813173e-01 -3.61583307e-02
1.85554296e-01 5.18947601e-01 -1.19867086e-01 2.73196071e-01
-2.66788214e-01 3.35018992e-01 -1.33678079e-01 -6.20813966e-01
2.25735039e-01 -8.70850682e-02 3.86327282e-02 4.57374990e-01
6.75482601e-02 4.38399076e-01 -3.52945328e-01 -3.66758227e-01
-1.14876181e-01 8.79212141e-01 -1.34676799e-01 -3.11765730e-01
5.06105006e-01 1.72947943e-01 2.03700647e-01 3.57144117e-01
8.22457790e-01 8.98776114e-01 -1.08705744e-01 3.30947608e-01
-4.59834129e-01 4.99846369e-01 -9.38630521e-01 1.17755842e+00
-5.81374884e-01 1.21096551e+00 3.66127193e-01 -4.97000366e-01
8.60721350e-01 6.65826797e-01 -1.20601922e-01 -7.64821947e-01
1.75090417e-01 3.05161178e-01 5.11028826e-01 -1.69584736e-01
5.81188738e-01 -2.77589917e-01 2.06692312e-02 3.28376800e-01
-8.99903327e-02 2.22763106e-01 3.59137431e-02 1.84502050e-01
9.92825270e-01 -6.00651562e-01 -1.57639444e-01 -4.45077270e-01
6.72444940e-01 -7.10731924e-01 6.22151434e-01 4.65368539e-01
-8.82891655e-01 3.27806592e-01 5.55936217e-01 2.55044222e-01
-8.15858781e-01 -1.12859583e+00 -4.81896371e-01 9.60221052e-01
-6.67015731e-01 -2.41020963e-01 -8.64345908e-01 -4.97288495e-01
2.81836897e-01 9.55141008e-01 -6.64085746e-01 -4.51920092e-01
-5.71073174e-01 -6.42830789e-01 1.19134974e+00 5.14535427e-01
3.29817981e-01 -1.11660767e+00 -1.34245321e-01 -6.50758669e-02
1.50422454e-01 -9.62230384e-01 -8.27173829e-01 -1.69994414e-01
-5.33316910e-01 -6.34682417e-01 -1.14971960e+00 -9.01980340e-01
-8.63462910e-02 -3.82701576e-01 1.02489543e+00 -1.48095429e-01
-7.14762807e-02 5.66661954e-01 3.14495303e-02 -7.98029304e-01
-8.98997843e-01 2.90919185e-01 6.72463953e-01 -9.08027142e-02
5.50673604e-01 -2.51109749e-01 -2.44957149e-01 2.90972948e-01
-6.45100057e-01 -7.59785056e-01 4.63353723e-01 5.46608031e-01
-2.72792280e-01 -1.25125185e-01 1.01708066e+00 -5.27703226e-01
6.90352678e-01 -1.00842416e-01 -1.81442499e-01 2.08725050e-01
-8.49720895e-01 -9.23582464e-02 -9.01291370e-02 -7.28830159e-01
-9.71673667e-01 -3.50248843e-01 -4.48660135e-01 -2.34216597e-04
-2.22674608e-01 3.83122236e-01 -5.51375210e-01 2.16106549e-01
4.30078328e-01 1.43938407e-01 1.24508590e-01 -7.06189156e-01
6.49531838e-03 1.40073979e+00 4.91739482e-01 -1.89086467e-01
7.22904265e-01 -5.63614257e-02 -7.36296356e-01 -1.46844935e+00
1.95175946e-01 -2.64858365e-01 -6.05682135e-01 -2.35433891e-01
8.00525725e-01 -1.11653972e+00 -5.55236101e-01 1.11899722e+00
-8.70629728e-01 -5.39826453e-01 -1.33206487e-01 6.81085646e-01
1.45094573e-01 1.73361719e-01 -5.63834071e-01 -1.15008271e+00
-6.08872056e-01 -1.03734791e+00 7.42823243e-01 -8.36899281e-02
-6.51380360e-01 -1.05384231e+00 3.58866006e-01 4.58327770e-01
5.23832262e-01 -8.68637040e-02 9.07078445e-01 -6.73109174e-01
3.22747886e-01 -1.71971902e-01 1.71477944e-01 6.64847732e-01
1.49109541e-02 1.09110974e-01 -1.39611876e+00 -5.67913294e-01
-1.39862061e-01 1.39136940e-01 8.64860177e-01 3.57355058e-01
3.59307379e-01 8.76021280e-04 -4.61809002e-02 -4.09181416e-02
9.11156416e-01 1.35954767e-01 8.08829188e-01 2.42283821e-01
5.07128477e-01 8.36210012e-01 1.82679057e-01 -8.29118565e-02
3.84653687e-01 7.41114259e-01 -2.35808983e-01 7.04657361e-02
-5.39336324e-01 -1.42240211e-01 1.07852578e+00 1.18044508e+00
3.86571586e-01 -8.18297789e-02 -1.12383413e+00 9.72109854e-01
-9.38101053e-01 -6.96894467e-01 -2.30724767e-01 2.59410095e+00
8.60358059e-01 -2.27951109e-02 5.52365482e-01 6.62353039e-01
1.11524427e+00 1.71844050e-01 -2.85961121e-01 -8.55050802e-01
-3.33804876e-01 1.54999465e-01 2.55361736e-01 4.93389994e-01
-6.17639244e-01 5.72073638e-01 6.43291950e+00 7.06596792e-01
-1.47775996e+00 2.57482022e-01 5.57970822e-01 -2.33344793e-01
-2.01837569e-01 -3.56319994e-01 -9.20941353e-01 6.97331488e-01
1.73428857e+00 -2.56056279e-01 4.14418019e-02 3.79482746e-01
2.97391802e-01 1.61831796e-01 -1.15406990e+00 9.43326354e-01
1.60109758e-01 -4.96272594e-01 -3.26179683e-01 5.54327726e-01
1.99129984e-01 2.01203614e-01 4.43881959e-01 6.02439523e-01
-2.15449914e-01 -9.01902616e-01 1.04126871e+00 1.23676829e-01
1.21909654e+00 -7.75752485e-01 9.19212699e-01 -4.11756150e-03
-8.24768007e-01 -1.14717126e-01 1.10648498e-01 5.71854822e-02
7.32710883e-02 7.08579600e-01 -8.02212775e-01 -2.66098063e-02
6.54906631e-01 1.51723951e-01 -8.82499576e-01 6.37218118e-01
5.70197999e-02 1.12340224e+00 -1.32584184e-01 1.50707677e-01
-1.07652053e-01 1.84347376e-01 6.60142899e-01 1.24666572e+00
4.20632422e-01 -5.70698380e-01 -8.32469583e-01 3.10050100e-01
-1.26182899e-01 1.77129284e-01 -5.30664206e-01 -3.93198133e-01
6.30489111e-01 8.64445090e-01 -1.43897422e-02 -2.22020149e-01
-2.88619906e-01 7.59681523e-01 8.46224576e-02 3.85242999e-01
-5.33684731e-01 -4.89418179e-01 1.07307208e+00 3.44106913e-01
1.32360518e-01 -1.14025086e-01 -3.22717547e-01 -7.56004214e-01
-3.18796337e-02 -1.07814217e+00 2.34601825e-01 -1.81484655e-01
-1.00819993e+00 4.56071198e-01 -2.81865925e-01 -5.76194942e-01
-2.35731915e-01 -4.63217914e-01 -8.60969365e-01 1.31223774e+00
-1.21334016e+00 -9.49300647e-01 1.46982700e-01 3.48100930e-01
3.28032285e-01 -3.90017241e-01 8.51383746e-01 7.37810075e-01
-9.83311355e-01 1.24025226e+00 2.41299629e-01 2.17807785e-01
9.55685914e-01 -1.09972644e+00 6.96835279e-01 6.82264924e-01
-7.47287795e-02 5.82747936e-01 6.71998560e-01 -4.31367218e-01
-9.23888564e-01 -7.84328878e-01 1.57530463e+00 -6.23709500e-01
3.81107122e-01 -6.52124882e-01 -9.49175239e-01 4.71095562e-01
3.24878991e-01 -6.79047406e-01 8.46793711e-01 5.19046366e-01
-5.91756344e-01 -2.21054927e-01 -1.33288765e+00 5.78391552e-01
8.33009183e-01 -9.37410712e-01 -5.64186156e-01 -4.89355803e-01
8.54761839e-01 1.99475825e-01 -8.90742123e-01 2.18695626e-01
9.23559427e-01 -9.06782269e-01 7.88961828e-01 -3.29810351e-01
4.75370623e-02 1.80848256e-01 -1.04935028e-01 -1.50312209e+00
-8.60038996e-02 -4.95836049e-01 5.28906994e-02 1.85854030e+00
6.82332277e-01 -1.14235163e+00 5.10366619e-01 8.12594593e-01
-9.59424451e-02 -4.40740138e-01 -1.40634012e+00 -9.72287118e-01
7.04428911e-01 -5.02797186e-01 7.36744821e-01 1.01440036e+00
-1.83866784e-01 3.29752505e-01 9.08139423e-02 2.26592124e-01
3.29692811e-01 -7.78958142e-01 6.44600987e-01 -1.33337367e+00
4.71243188e-02 -6.53003752e-01 -7.63456166e-01 -1.39418095e-02
4.41539615e-01 -5.63769102e-01 -1.02962971e-01 -1.14225292e+00
-7.49868751e-02 -1.67108178e-01 -3.91687065e-01 1.72942057e-01
-1.51942253e-01 8.32950994e-02 3.38050276e-01 -3.78077447e-01
3.20669502e-01 4.95757818e-01 5.68441868e-01 -3.03652972e-01
-2.93862522e-01 1.26385465e-01 -7.87632287e-01 2.19725445e-01
8.55211079e-01 -3.53841633e-01 -1.54760823e-01 -1.98560685e-01
-3.37910831e-01 -2.41487771e-01 1.30145296e-01 -1.01928580e+00
-2.04406172e-01 5.18420637e-01 1.13084048e-01 -3.28899026e-01
3.61807495e-01 -2.72063673e-01 -4.99069132e-02 7.13759363e-01
-3.47027570e-01 1.83529034e-01 6.57529414e-01 1.34815335e-01
-1.20415740e-01 -2.16803491e-01 7.29702234e-01 4.90738064e-01
-2.20455453e-01 1.34066399e-02 -8.77501309e-01 1.57381862e-01
4.49072212e-01 -1.68772951e-01 -2.15076968e-01 -5.29883385e-01
-6.31848156e-01 -5.59256645e-03 3.50743324e-01 6.83486998e-01
3.11250240e-01 -1.28521419e+00 -1.11196089e+00 3.56754273e-01
1.28154814e-01 -7.35041559e-01 3.85192901e-01 7.38603830e-01
1.94675941e-02 4.05189127e-01 1.44532815e-01 -1.10877216e-01
-1.90082085e+00 1.70807779e-01 2.47769520e-01 5.15288375e-02
-3.32283437e-01 1.10240471e+00 1.75289854e-01 -5.82138002e-01
5.71882367e-01 -5.46361767e-02 -1.82563439e-01 5.14237046e-01
6.22507811e-01 7.27337897e-01 3.14672798e-01 -1.13710320e+00
-6.04324818e-01 2.76229858e-01 -3.16099495e-01 -7.68572628e-01
1.05874908e+00 -2.25315541e-01 1.37561351e-01 8.98090720e-01
1.32489181e+00 5.04556179e-01 -5.47453880e-01 -1.36215195e-01
-1.61471456e-01 -1.46630570e-01 2.97534704e-01 -8.67297947e-01
-8.85964155e-01 1.11818182e+00 1.28468609e+00 1.01486981e-01
7.06142902e-01 -1.12020858e-01 9.44926798e-01 -1.55406132e-01
-1.76263496e-01 -1.13305104e+00 -4.97082114e-01 4.60327178e-01
9.25955772e-01 -1.17409027e+00 -4.82913107e-01 2.95462348e-02
-6.30771458e-01 5.44830024e-01 4.14283633e-01 7.75310755e-01
7.91660845e-01 -5.73920421e-02 5.95871150e-01 7.68312141e-02
-5.07066846e-01 -7.80723616e-02 2.39980906e-01 8.04967582e-01
7.87417412e-01 3.13435942e-01 -4.41647947e-01 6.12292588e-01
-8.77314329e-01 -4.43997890e-01 5.28918326e-01 4.74302649e-01
1.33292064e-01 -1.21116734e+00 -5.95643163e-01 3.91501099e-01
-5.51161528e-01 -2.54964888e-01 -7.33147085e-01 8.61461818e-01
1.89171433e-01 1.03183794e+00 2.67743975e-01 -5.07904112e-01
4.26690191e-01 7.37456918e-01 1.66377768e-01 -3.99286479e-01
-6.54584944e-01 -3.79408389e-01 3.69735539e-01 1.50337249e-01
-3.08728274e-02 -1.32572067e+00 -7.51955748e-01 -5.24540305e-01
-4.42569405e-01 7.68608153e-02 1.23999596e+00 7.30706334e-01
2.56908298e-01 5.88425756e-01 4.27925080e-01 -5.51689565e-01
-6.88391149e-01 -1.52374947e+00 -9.14013088e-01 1.45152837e-01
6.90047503e-01 -4.10251647e-01 -9.19816554e-01 -3.05185169e-01]
|
[14.307567596435547, 6.154906749725342]
|
9b5d8732-ad2f-4bef-9f4e-4f307c47d8cd
|
martingale-posterior-neural-processes
|
2304.09431
| null |
https://arxiv.org/abs/2304.09431v1
|
https://arxiv.org/pdf/2304.09431v1.pdf
|
Martingale Posterior Neural Processes
|
A Neural Process (NP) estimates a stochastic process implicitly defined with neural networks given a stream of data, rather than pre-specifying priors already known, such as Gaussian processes. An ideal NP would learn everything from data without any inductive biases, but in practice, we often restrict the class of stochastic processes for the ease of estimation. One such restriction is the use of a finite-dimensional latent variable accounting for the uncertainty in the functions drawn from NPs. Some recent works show that this can be improved with more "data-driven" source of uncertainty such as bootstrapping. In this work, we take a different approach based on the martingale posterior, a recently developed alternative to Bayesian inference. For the martingale posterior, instead of specifying prior-likelihood pairs, a predictive distribution for future data is specified. Under specific conditions on the predictive distribution, it can be shown that the uncertainty in the generated future data actually corresponds to the uncertainty of the implicitly defined Bayesian posteriors. Based on this result, instead of assuming any form of the latent variables, we equip a NP with a predictive distribution implicitly defined with neural networks and use the corresponding martingale posteriors as the source of uncertainty. The resulting model, which we name as Martingale Posterior Neural Process (MPNP), is demonstrated to outperform baselines on various tasks.
|
['Juho Lee', 'Edwin Fong', 'Giung Nam', 'Eunggu Yun', 'Hyungi Lee']
|
2023-04-19
| null | null | null | null |
['bayesian-inference']
|
['methodology']
|
[ 1.69989452e-01 3.62520754e-01 5.84585313e-03 -4.79530513e-01
-5.96018195e-01 -5.79998136e-01 1.03969073e+00 -6.17029630e-02
-4.52952445e-01 1.00432944e+00 1.91603854e-01 -2.77090073e-01
-1.49320811e-01 -1.18863475e+00 -1.19753039e+00 -8.62931192e-01
-2.83229095e-03 9.20490265e-01 -5.19797988e-02 4.31770682e-01
2.06594050e-01 3.53397727e-01 -1.42778695e+00 -1.85484141e-01
6.61566257e-01 9.33576882e-01 1.14210285e-01 5.89962900e-01
-4.66129154e-01 5.68809628e-01 -6.03866458e-01 -4.99351442e-01
1.46453395e-01 -2.15795279e-01 -5.38629174e-01 -2.25965962e-01
-1.42729774e-01 -3.94498408e-01 5.76256774e-02 1.19514263e+00
1.48990154e-01 2.32295185e-01 1.24565613e+00 -1.20288992e+00
-4.63237733e-01 8.00528586e-01 -2.52550215e-01 -9.63367745e-02
4.27445732e-02 1.43450513e-01 9.48667645e-01 -6.94896758e-01
3.18993032e-01 1.52734423e+00 8.55859578e-01 4.07592028e-01
-1.70187581e+00 -6.37394071e-01 1.91861302e-01 -3.63866508e-01
-1.32014263e+00 -2.48016581e-01 5.40450335e-01 -5.75091839e-01
4.10840243e-01 -2.34089270e-01 4.78557318e-01 1.60999620e+00
3.78703117e-01 8.64094377e-01 1.12814689e+00 -3.13703686e-01
8.95960927e-01 1.95321187e-01 2.60114938e-01 1.94683924e-01
3.37247103e-01 3.34283620e-01 -4.53108847e-01 -4.32189256e-01
9.38433170e-01 4.23095264e-02 -2.78899401e-01 -4.01884526e-01
-9.06239390e-01 9.92387891e-01 -2.33082294e-01 -1.43614843e-01
-6.08846605e-01 6.04501784e-01 2.94543028e-01 -1.55113474e-01
6.20288014e-01 2.10863695e-01 -5.56729078e-01 -4.05434102e-01
-1.01282477e+00 2.70790339e-01 1.41703582e+00 9.49505389e-01
7.34673381e-01 -3.83248515e-02 -3.95755231e-01 5.25697172e-01
5.83612144e-01 7.05294132e-01 2.03660250e-01 -1.08049822e+00
3.40068996e-01 -2.63321608e-01 6.56616032e-01 -4.95924532e-01
7.61349052e-02 -3.99553180e-01 -7.75050879e-01 1.57440126e-01
8.10669184e-01 -5.87332547e-01 -1.19041443e+00 2.20952654e+00
-1.18325511e-02 4.88225579e-01 1.82893917e-01 3.58751833e-01
-2.22180896e-02 8.98783684e-01 2.53118843e-01 -2.01888546e-01
1.04436648e+00 -3.81185085e-01 -7.78569579e-01 -3.81353796e-02
-1.04586095e-01 -7.35146254e-02 9.46331918e-01 6.15389585e-01
-9.91913736e-01 -3.58491391e-01 -8.99481297e-01 3.32598269e-01
-1.82709754e-01 -2.57831454e-01 5.92423022e-01 8.89828801e-01
-1.13182676e+00 8.49862337e-01 -1.28020024e+00 -2.57323831e-01
3.35978806e-01 1.97744727e-01 1.27642509e-02 2.27512017e-01
-1.28053546e+00 8.58865917e-01 7.59796143e-01 4.65877876e-02
-1.45442498e+00 -6.61389172e-01 -5.41276038e-01 2.98027724e-01
3.35186362e-01 -8.25701356e-01 1.50427103e+00 -8.66866827e-01
-1.95193231e+00 4.20018941e-01 -2.95072645e-01 -9.00106132e-01
7.81388402e-01 -4.37733978e-01 7.92456344e-02 2.39957068e-02
-2.10666612e-01 5.88944316e-01 1.32916617e+00 -1.32893503e+00
-4.77793366e-01 1.69287261e-03 5.96261472e-02 2.19105165e-02
8.12734757e-03 -2.70254880e-01 -3.91959339e-01 -3.05618048e-01
2.58125346e-02 -9.70991135e-01 -1.78897232e-01 -8.51848200e-02
-3.85493964e-01 -3.30388129e-01 3.70161325e-01 -4.93572831e-01
7.36687124e-01 -2.01468492e+00 -7.26182014e-02 3.51022989e-01
9.89282727e-02 -2.71913260e-01 2.67823696e-01 3.56402516e-01
6.81379810e-02 3.91486943e-01 -6.26015782e-01 -7.19284177e-01
3.75023603e-01 6.09145701e-01 -8.34296942e-01 5.10654151e-01
3.61129016e-01 6.59937739e-01 -9.93730545e-01 -7.85993412e-02
7.97665641e-02 5.62714875e-01 -2.53480375e-01 1.58875644e-01
-6.69197381e-01 4.74769413e-01 -3.54354531e-01 -9.13207605e-02
6.63432240e-01 -2.12105691e-01 -4.31842692e-02 4.02979255e-01
9.88648087e-02 2.74432778e-01 -1.42125535e+00 1.32357991e+00
-4.91286904e-01 4.10222262e-01 -8.23468119e-02 -8.29134881e-01
9.40028608e-01 4.91304874e-01 1.26128078e-01 3.57735962e-01
-6.54530525e-02 4.69220951e-02 -2.21306205e-01 -2.61570197e-02
2.41205886e-01 -6.71292543e-01 8.93400460e-02 6.56906247e-01
3.74738336e-01 -3.44839543e-01 7.75237102e-03 -2.43043695e-02
9.41635191e-01 6.47638500e-01 1.09205179e-01 -3.49460304e-01
2.14140832e-01 -4.77401972e-01 5.45774877e-01 1.39316928e+00
2.22400371e-02 5.95238268e-01 9.23099875e-01 -4.49481681e-02
-1.10345364e+00 -1.67080760e+00 -4.12080109e-01 7.31598318e-01
-3.72936666e-01 -1.11974508e-01 -7.15463936e-01 -4.80520904e-01
2.46783514e-02 1.32224452e+00 -7.29548037e-01 2.63893064e-02
-1.12293527e-01 -9.56288755e-01 5.97017646e-01 6.47564709e-01
2.89593041e-01 -9.85754192e-01 -4.27765161e-01 3.76594275e-01
7.42390975e-02 -6.89483464e-01 7.17044156e-03 5.46128273e-01
-9.61169899e-01 -5.17679453e-01 -8.36695910e-01 2.94361729e-02
3.86547416e-01 -6.39773905e-01 1.18001580e+00 -9.04684365e-01
5.78111112e-01 5.94383955e-01 1.72304660e-01 -9.11085367e-01
-5.05461097e-01 -1.78473890e-01 1.60491198e-01 4.66830060e-02
4.38000947e-01 -9.34733927e-01 -2.04198092e-01 -2.25215942e-01
-9.72311616e-01 -3.48234884e-02 4.64897543e-01 7.95419395e-01
5.37809670e-01 4.09238756e-01 4.67377812e-01 -8.81391406e-01
7.73660898e-01 -7.81838417e-01 -7.42044389e-01 7.70449191e-02
-5.37167549e-01 7.48185575e-01 4.89262223e-01 -6.29911423e-01
-1.50784802e+00 -3.07060301e-01 5.43131121e-02 -5.14751077e-01
-4.51081902e-01 6.04683518e-01 -2.82996237e-01 7.85351813e-01
4.06015605e-01 1.87773526e-01 -6.49596602e-02 -4.48656797e-01
3.92937928e-01 4.87784952e-01 5.74785352e-01 -1.10094655e+00
4.99561787e-01 6.75394893e-01 1.79916695e-01 -5.09603918e-01
-1.14658749e+00 -8.24765787e-02 -4.38982993e-01 1.23975687e-01
9.43211854e-01 -8.58488679e-01 -6.65417492e-01 5.34721732e-01
-1.38755107e+00 -4.35939074e-01 -7.25141525e-01 7.57306814e-01
-9.96726036e-01 9.75401327e-02 -6.43413544e-01 -1.48963785e+00
-1.12307586e-01 -8.53415430e-01 9.36435997e-01 4.09907512e-02
-3.09331656e-01 -1.35635698e+00 1.06636390e-01 -3.75729144e-01
3.16556603e-01 1.30464509e-01 9.16723073e-01 -9.58828390e-01
-4.92303461e-01 -2.49472916e-01 -8.04992691e-02 7.61358202e-01
-1.41585737e-01 9.14116129e-02 -1.12390101e+00 1.34383515e-01
6.40343904e-01 3.02180182e-02 8.71165812e-01 8.33692968e-01
1.16270280e+00 -2.43995726e-01 -2.27917790e-01 4.78572577e-01
1.47242916e+00 1.61828950e-01 6.16155386e-01 5.16808555e-02
4.33676124e-01 7.75468588e-01 6.55270889e-02 5.81602275e-01
1.78177148e-01 -1.45404960e-03 3.37060869e-01 6.88908339e-01
5.12980640e-01 -7.46323049e-01 5.38244247e-01 3.49644065e-01
-1.23607323e-01 -3.72437149e-01 -1.05792236e+00 4.61479425e-01
-1.84916818e+00 -1.08837306e+00 3.32320342e-03 2.57318377e+00
1.08947372e+00 4.46201354e-01 -1.42916784e-01 -2.40178883e-01
9.51987505e-01 -8.01760480e-02 -7.53206193e-01 -1.67165741e-01
-2.81880107e-02 3.38452220e-01 6.48854256e-01 6.94162905e-01
-9.96926069e-01 5.12491763e-01 6.68662024e+00 8.21917474e-01
-7.05899775e-01 1.59618914e-01 6.05560601e-01 5.12260981e-02
-4.70065475e-01 2.28186488e-01 -1.27735734e+00 8.09675813e-01
1.45680082e+00 -2.79335439e-01 3.03513944e-01 7.92774677e-01
2.24473283e-01 -2.46621087e-01 -1.61981750e+00 7.21848011e-01
-3.96392822e-01 -8.59094262e-01 8.44549853e-03 2.74951547e-01
6.81304574e-01 4.09798734e-02 9.84433070e-02 5.26947558e-01
1.17401147e+00 -1.15473628e+00 1.05424583e+00 1.11416376e+00
5.37564695e-01 -8.09236228e-01 8.29304039e-01 7.44338214e-01
-4.72191542e-01 3.21391672e-02 -4.71861422e-01 -1.86865777e-01
3.94477099e-01 1.04365647e+00 -7.71411359e-01 1.75764084e-01
3.69672239e-01 5.11787474e-01 6.89234138e-02 8.84514511e-01
-6.82932675e-01 1.19880128e+00 -7.65239120e-01 4.15412560e-02
1.81877390e-01 -5.16940892e-01 6.81734145e-01 1.21207428e+00
5.48407316e-01 -5.08625627e-01 -9.54754502e-02 1.40887046e+00
5.39127551e-02 -3.51273835e-01 -5.27064741e-01 -1.12448141e-01
5.66699982e-01 8.06752741e-01 -6.44356132e-01 -5.63926160e-01
-1.97411180e-01 6.93076313e-01 1.83141246e-01 8.03934634e-01
-7.93518066e-01 3.27907689e-02 4.32487875e-01 -2.12330058e-01
4.39789444e-01 -2.30014741e-01 -4.53762829e-01 -9.71916735e-01
-6.46471679e-02 -3.50302339e-01 1.28135592e-01 -8.63918185e-01
-1.79084277e+00 5.20820692e-02 6.69423521e-01 -6.87843442e-01
-7.59926260e-01 -5.99597275e-01 -6.03877604e-01 1.58249283e+00
-1.19314742e+00 -7.34427273e-01 3.08089137e-01 2.48273551e-01
3.46044570e-01 1.89713314e-01 8.08797121e-01 -6.06250763e-01
-2.68728137e-01 -9.49320868e-02 4.78941292e-01 -7.84444734e-02
4.89629775e-01 -1.67869711e+00 4.24636990e-01 6.71711564e-01
-5.82681075e-02 9.50552821e-01 1.18867302e+00 -8.54631960e-01
-9.90500867e-01 -9.71471012e-01 3.99900556e-01 -7.50022173e-01
8.86801660e-01 -4.63721752e-01 -1.01650608e+00 9.81395364e-01
2.28215754e-02 -2.05047920e-01 5.43977320e-01 4.19913679e-01
-2.15266675e-01 1.57617241e-01 -1.10959363e+00 5.39817035e-01
6.51867449e-01 -5.50207376e-01 -7.92083502e-01 2.12979347e-01
8.25410962e-01 -7.02451244e-02 -7.70519912e-01 2.64411390e-01
5.13577700e-01 -7.97628343e-01 6.66740239e-01 -5.31924009e-01
4.72568244e-01 -3.36477667e-01 -2.34247729e-01 -1.43375885e+00
-9.76929963e-02 -6.81085527e-01 -4.89080369e-01 1.40666318e+00
3.86816978e-01 -9.03594196e-01 7.41939723e-01 1.10963988e+00
1.66256681e-01 -4.82812285e-01 -8.53848517e-01 -9.19542015e-01
5.06402612e-01 -1.06584394e+00 6.91107988e-01 4.59948719e-01
-5.62061429e-01 1.81182846e-01 -2.82510668e-01 3.81348997e-01
7.68898427e-01 -2.84119159e-01 5.90497851e-01 -1.62928450e+00
-7.51382649e-01 -2.47877955e-01 2.20813639e-02 -1.22585475e+00
4.51574951e-01 -5.09037614e-01 5.17061293e-01 -1.43746233e+00
1.96617350e-01 -3.73431385e-01 -2.32251570e-01 6.15318455e-02
3.16616287e-03 -4.50717449e-01 5.76836877e-02 2.80121118e-01
-1.93239644e-01 7.23976672e-01 8.63325238e-01 2.77189851e-01
-1.66029409e-01 4.21845526e-01 -5.29133737e-01 1.14328623e+00
8.22246671e-01 -6.07737124e-01 -6.08964503e-01 -1.18343696e-01
3.63811761e-01 3.06661308e-01 4.65007514e-01 -9.34142351e-01
1.61868423e-01 -1.72675923e-01 4.77040440e-01 -6.40773475e-01
5.18905580e-01 -6.55980170e-01 4.50822294e-01 9.84988213e-02
-5.70901930e-01 -4.17118043e-01 3.53148468e-02 1.13536286e+00
-8.94812867e-03 -8.39132071e-01 4.25017178e-01 -3.69212300e-01
-1.42185062e-01 2.46130198e-01 -7.11396098e-01 -5.00651784e-02
6.94234192e-01 -7.62473345e-02 1.40101671e-01 -8.28993022e-01
-9.13709879e-01 2.33926531e-02 3.28068376e-01 -1.08682945e-01
2.80239463e-01 -9.49453294e-01 -5.01649141e-01 -2.48129833e-02
-3.23325843e-01 4.17897940e-01 -8.92237350e-02 6.03542328e-01
4.40140478e-02 2.55224496e-01 1.71523631e-01 -7.79830754e-01
-2.65103698e-01 5.48864245e-01 3.01048249e-01 -4.44682986e-01
-4.36981380e-01 6.90618515e-01 4.52465087e-01 -4.18209225e-01
2.34387308e-01 -4.39239502e-01 -3.67501080e-02 -2.36839638e-03
4.18359399e-01 2.73289084e-01 -4.61853206e-01 -4.69852164e-02
1.79488271e-01 -4.15376462e-02 1.31338388e-01 -9.80195701e-01
1.25830698e+00 -3.03171277e-01 -2.13391945e-01 1.33802140e+00
7.10308850e-01 -1.15365401e-01 -2.06393814e+00 -2.49469921e-01
2.58396685e-01 -7.23014027e-02 -6.72605410e-02 -7.28662014e-01
-4.85771984e-01 1.22383630e+00 3.44257891e-01 2.21411869e-01
7.11982548e-01 -1.06288671e-01 1.75773650e-01 3.68628621e-01
5.18859208e-01 -9.00843382e-01 -4.02551413e-01 8.15538883e-01
7.00470686e-01 -9.29520547e-01 -3.77251327e-01 -8.77214372e-02
-5.47852337e-01 1.13873291e+00 1.33977786e-01 -2.27643296e-01
8.77706766e-01 4.26532298e-01 -4.36699659e-01 1.42744839e-01
-8.49341810e-01 3.78501639e-02 4.49376367e-02 8.04879785e-01
2.18413085e-01 1.22443900e-01 7.50850961e-02 8.46946478e-01
-3.10527951e-01 2.94007659e-01 5.71087301e-01 7.42113292e-01
-3.86766970e-01 -8.90863538e-01 -4.80570942e-01 6.16486967e-01
-7.31308460e-01 -3.98567259e-01 3.02205116e-01 4.85454589e-01
-3.68283838e-02 7.36436069e-01 4.19614285e-01 3.05637240e-01
-1.64484158e-01 6.05278850e-01 3.22689772e-01 -8.02914977e-01
1.10863045e-01 -4.54853997e-02 1.27199113e-01 -6.96377084e-02
-2.66726881e-01 -9.50816035e-01 -1.00260603e+00 -1.69937715e-01
-1.52679414e-01 1.67546108e-01 9.51491058e-01 1.13242376e+00
-2.27398518e-02 2.03803569e-01 1.66151430e-02 -9.33958054e-01
-1.17676044e+00 -1.27392101e+00 -8.04891407e-01 2.47319505e-01
1.55462891e-01 -7.17589498e-01 -6.74383521e-01 2.39158377e-01]
|
[7.043673515319824, 3.84466290473938]
|
e77704bb-f8f0-4ccc-b101-80e172383fbc
|
matt-a-manifold-attention-network-for-eeg
|
2210.01986
| null |
https://arxiv.org/abs/2210.01986v1
|
https://arxiv.org/pdf/2210.01986v1.pdf
|
MAtt: A Manifold Attention Network for EEG Decoding
|
Recognition of electroencephalographic (EEG) signals highly affect the efficiency of non-invasive brain-computer interfaces (BCIs). While recent advances of deep-learning (DL)-based EEG decoders offer improved performances, the development of geometric learning (GL) has attracted much attention for offering exceptional robustness in decoding noisy EEG data. However, there is a lack of studies on the merged use of deep neural networks (DNNs) and geometric learning for EEG decoding. We herein propose a manifold attention network (mAtt), a novel geometric deep learning (GDL)-based model, featuring a manifold attention mechanism that characterizes spatiotemporal representations of EEG data fully on a Riemannian symmetric positive definite (SPD) manifold. The evaluation of the proposed MAtt on both time-synchronous and -asyncronous EEG datasets suggests its superiority over other leading DL methods for general EEG decoding. Furthermore, analysis of model interpretation reveals the capability of MAtt in capturing informative EEG features and handling the non-stationarity of brain dynamics.
|
['Chun-Shu Wei', 'Jing-Lun Chou', 'Yue-Ting Pan']
|
2022-10-05
| null | null | null | null |
['eeg-decoding', 'eeg-decoding']
|
['medical', 'time-series']
|
[ 1.53596401e-02 -5.23916371e-02 5.56767285e-01 -3.23954731e-01
-5.11724949e-01 -1.57818004e-01 4.93819147e-01 -2.17600465e-01
-1.37899846e-01 5.71589351e-01 1.33906931e-01 -2.64053494e-01
-8.23947728e-01 -1.55575365e-01 -7.19621718e-01 -8.74303758e-01
-7.80128419e-01 2.40892783e-01 -6.05795801e-01 -1.22152850e-01
3.19951594e-01 7.21953154e-01 -1.19153392e+00 1.07519530e-01
1.03774917e+00 1.16337430e+00 2.76924223e-01 4.53625292e-01
2.48127997e-01 4.43390846e-01 -5.94317377e-01 -3.62325698e-01
-3.60033326e-02 -4.46516693e-01 -5.72592080e-01 -1.77976087e-01
-1.39071653e-02 1.88309118e-01 -6.03583097e-01 1.18294132e+00
6.00156844e-01 -6.17704540e-02 1.02956545e+00 -1.32837808e+00
-1.01227999e+00 4.20787752e-01 -4.55743074e-01 8.39002848e-01
2.15239406e-01 -4.10807878e-02 7.33870864e-01 -1.04718840e+00
7.19412491e-02 8.82894337e-01 5.04021227e-01 3.26652497e-01
-1.15011060e+00 -7.46509790e-01 4.38951403e-02 7.50920534e-01
-1.74870348e+00 -1.75384149e-01 8.57281506e-01 -5.73699415e-01
9.29842293e-01 2.24325791e-01 1.02160811e+00 1.37362981e+00
9.46600735e-01 6.22339606e-01 1.12123930e+00 5.87851666e-02
2.38430575e-01 -1.46122992e-01 2.81929970e-01 2.91247785e-01
1.71243921e-01 -1.15392804e-01 -8.82848620e-01 2.17272609e-01
8.34816635e-01 -1.06809966e-01 -6.06893122e-01 -1.93428367e-01
-1.25148427e+00 4.57309812e-01 4.70887780e-01 5.91104746e-01
-7.13212252e-01 1.77455142e-01 2.99035728e-01 4.30968791e-01
8.29206169e-01 5.11670887e-01 -3.01355541e-01 -4.18275684e-01
-8.76562119e-01 -9.71608981e-02 6.29886508e-01 1.01502025e+00
4.34633195e-01 3.53203386e-01 -8.13885927e-02 4.27435994e-01
2.02005118e-01 4.58327889e-01 5.35509646e-01 -3.43893737e-01
5.50919652e-01 5.77827811e-01 -3.14783573e-01 -1.29933345e+00
-9.10961330e-01 -7.73804009e-01 -1.30860376e+00 -3.43151301e-01
-3.91552821e-02 -2.56715268e-01 -2.36644804e-01 1.69835544e+00
-4.68736500e-01 6.25136375e-01 -1.69388771e-01 1.21156406e+00
5.01359880e-01 5.11389434e-01 1.72814131e-02 -1.76704109e-01
9.69862223e-01 -1.74424559e-01 -1.02469456e+00 2.68431395e-01
5.35360575e-01 -9.66801643e-02 8.02158058e-01 6.67660236e-01
-1.07149374e+00 -4.08166498e-01 -1.08976042e+00 6.59994408e-02
-4.45901424e-01 1.95081607e-01 4.89999771e-01 6.16946399e-01
-1.28714716e+00 6.38338864e-01 -1.00108802e+00 -1.78894907e-01
5.42017758e-01 9.29578185e-01 -4.17984307e-01 2.93745100e-01
-9.89330590e-01 1.05604541e+00 1.61597654e-01 5.72261989e-01
-9.17956293e-01 -7.45914638e-01 -5.57140052e-01 1.26254201e-01
-1.97159484e-01 -4.46818560e-01 3.40619445e-01 -9.87717390e-01
-1.69412410e+00 3.67468357e-01 -1.01482570e-02 -5.74629724e-01
2.75074989e-01 -4.43923175e-01 -5.33065200e-01 3.22577000e-01
-2.63717294e-01 6.00700498e-01 1.08323896e+00 -7.73665547e-01
1.01420470e-01 -6.39056683e-01 -3.69103700e-01 1.47844523e-01
-6.10918283e-01 -1.90448537e-01 1.38827100e-01 -7.20081270e-01
2.65666366e-01 -8.15528452e-01 3.07815850e-01 -4.82825100e-01
-5.28418422e-01 -4.18322772e-01 6.31488085e-01 -8.37485790e-01
1.19867623e+00 -2.14870763e+00 8.55014324e-01 2.70375133e-01
2.75882453e-01 6.51476830e-02 -2.59490639e-01 3.48388076e-01
-4.29635823e-01 -1.36216924e-01 -3.76794994e-01 -3.56628239e-01
3.64992619e-02 -6.53249919e-02 -2.94626117e-01 9.88662958e-01
6.41892433e-01 1.10067940e+00 -7.87217557e-01 1.15714334e-02
2.76944250e-01 7.97680974e-01 -4.89680707e-01 1.80605918e-01
2.16730535e-01 1.00240862e+00 -2.95433462e-01 3.42194945e-01
6.32029653e-01 -2.06853777e-01 -2.97223419e-01 -3.01892847e-01
-1.04574546e-01 1.57172278e-01 -8.75032008e-01 1.91078138e+00
-2.59357572e-01 1.04419434e+00 -2.05929413e-01 -1.33891726e+00
7.84656227e-01 6.00550652e-01 8.50222647e-01 -7.94208765e-01
4.87415850e-01 1.07827023e-01 2.36713201e-01 -8.74866605e-01
-9.20190662e-02 1.65240154e-01 4.24919397e-01 4.83793437e-01
5.89823604e-01 2.37553507e-01 -2.60949701e-01 -6.63992688e-02
9.64632213e-01 1.37075230e-01 -1.95562407e-01 -1.02017236e+00
6.43456638e-01 -6.73217237e-01 1.05419524e-01 4.07093853e-01
2.86883060e-02 6.42948985e-01 6.14421844e-01 -2.88192630e-01
-8.18274796e-01 -9.83539581e-01 -6.45392060e-01 6.97439492e-01
1.50972009e-01 -3.84221762e-01 -1.17842460e+00 -9.81792659e-02
-2.76938379e-01 6.93993449e-01 -5.13565779e-01 -4.89341766e-01
-3.23222071e-01 -1.19623888e+00 5.76128483e-01 3.89705330e-01
4.03969735e-01 -8.72087419e-01 -5.23169160e-01 2.66992509e-01
-2.10405529e-01 -1.05921459e+00 -2.63711095e-01 1.79622114e-01
-9.43990171e-01 -1.06031656e+00 -8.66334260e-01 -6.79674804e-01
6.91250980e-01 -1.69559498e-03 6.24530911e-01 -2.78187960e-01
-2.73331493e-01 7.88688302e-01 -3.29150558e-01 -5.46730042e-01
1.83488309e-01 2.86113918e-01 4.44532484e-01 6.61028922e-01
6.22954011e-01 -9.49824989e-01 -6.46239936e-01 1.65863961e-01
-8.00624371e-01 2.15882227e-01 5.33769727e-01 6.56669736e-01
2.18424350e-01 -4.91051562e-02 1.09474242e+00 -1.89187467e-01
1.05341721e+00 -8.57999265e-01 -3.91115159e-01 1.85961068e-01
-3.92862886e-01 1.57246083e-01 5.83560348e-01 -3.48489732e-01
-7.27662802e-01 -4.32152063e-01 -5.18776067e-02 -4.60324526e-01
1.50835020e-02 4.87130165e-01 3.69135737e-02 -4.02694106e-01
4.36482787e-01 4.88950938e-01 -3.42086852e-02 -5.27833641e-01
7.32200779e-03 7.40003228e-01 4.19378757e-01 -2.79809266e-01
4.14319754e-01 2.84037977e-01 1.39143154e-01 -1.22135878e+00
-3.94608796e-01 -1.65903583e-01 -9.66019630e-01 -2.73475975e-01
1.10245538e+00 -8.34309757e-01 -9.50962543e-01 5.27970314e-01
-1.27778721e+00 -7.59616345e-02 2.30863035e-01 9.03313696e-01
-7.94141889e-01 2.53885120e-01 -5.95869482e-01 -9.24243450e-01
-4.48831975e-01 -1.36454296e+00 1.11621428e+00 -2.11575866e-01
-1.98586449e-01 -1.15610063e+00 -1.62270993e-01 -7.06148371e-02
3.81200671e-01 1.24096692e-01 1.01026678e+00 -5.55683911e-01
-6.85237646e-01 -1.81561008e-01 -2.39690617e-01 3.77472192e-01
-9.14926752e-02 -5.08289456e-01 -1.10222685e+00 -4.56996977e-01
5.13921976e-01 1.35441497e-01 4.60850716e-01 5.22940099e-01
1.50815368e+00 -1.62509009e-01 -1.43449074e-02 9.74853098e-01
1.36151803e+00 1.54979065e-01 6.31988823e-01 1.43845811e-01
7.88168490e-01 5.20400345e-01 -2.92809635e-01 5.79680443e-01
3.85449052e-01 4.69517499e-01 5.83285391e-01 9.40909609e-02
2.44914293e-01 2.07149744e-01 4.24119115e-01 1.46386075e+00
-5.28107822e-01 7.60570318e-02 -9.02097762e-01 3.57665390e-01
-1.58111906e+00 -4.80886817e-01 -2.83827841e-01 2.19776106e+00
2.79086262e-01 -1.31320000e-01 -1.56866193e-01 2.51588106e-01
5.54628372e-01 -3.43775630e-01 -6.11812651e-01 -2.48670936e-01
-4.73371655e-01 4.00927842e-01 2.90781111e-01 1.78115219e-02
-8.89342427e-01 4.28124309e-01 6.07266760e+00 5.84392428e-01
-1.26182580e+00 3.67707402e-01 4.01349783e-01 -5.81994690e-02
-1.22361720e-01 -6.29547715e-01 -2.63571113e-01 6.31633401e-01
1.36335993e+00 -2.61463732e-01 9.02496278e-01 3.52708459e-01
5.11470973e-01 2.08215028e-01 -1.38835430e+00 1.68247914e+00
2.91220963e-01 -1.16129434e+00 2.02267334e-01 1.65686458e-01
5.74803174e-01 1.82812274e-01 3.67188543e-01 3.77090760e-02
-7.27026463e-01 -1.24960029e+00 6.91189170e-01 9.58990574e-01
7.42304564e-01 -8.73961985e-01 8.89369190e-01 3.46959174e-01
-9.82312977e-01 -2.93011338e-01 -4.88992929e-01 -1.02710292e-01
-6.70645833e-02 3.33801210e-01 -2.74497420e-01 6.44727170e-01
6.67852521e-01 1.40494478e+00 -8.56550336e-01 1.19230604e+00
-3.73949762e-03 5.88381946e-01 -1.91060062e-02 -1.59630626e-01
3.10779363e-01 -6.98146641e-01 9.04752433e-01 1.27080762e+00
5.17788410e-01 1.56905755e-01 -6.27331376e-01 1.28717899e+00
1.16706736e-01 1.50268376e-01 -7.42910624e-01 -9.54129696e-02
2.79357210e-02 1.04759037e+00 -6.45642400e-01 1.38807744e-01
-3.57393950e-01 9.66933787e-01 3.34130615e-01 6.13043725e-01
-7.90998101e-01 -2.50326693e-01 7.58790731e-01 -9.56323519e-02
-1.78692654e-01 -6.18302703e-01 -6.32827103e-01 -1.30986226e+00
1.48822412e-01 -5.40408313e-01 -1.72737688e-02 -1.00632918e+00
-1.36950362e+00 8.55051160e-01 1.77828789e-01 -1.31246388e+00
1.39544144e-01 -8.38971317e-01 -6.08710766e-01 8.92143071e-01
-1.46739042e+00 -7.40650058e-01 -1.97830483e-01 1.03741741e+00
3.15441847e-01 -4.02464539e-01 8.14735591e-01 5.40450335e-01
-7.50989258e-01 4.84458476e-01 2.42427573e-01 -1.08234130e-01
1.19784653e-01 -1.27378762e+00 6.88227117e-02 6.13753200e-01
1.42584831e-01 7.55669653e-01 4.39977109e-01 -2.39907309e-01
-1.93756914e+00 -1.12112296e+00 4.41956222e-01 -4.34735417e-01
8.46956491e-01 -6.14750743e-01 -1.04124594e+00 6.72125399e-01
4.29344386e-01 -4.99324352e-01 6.53990209e-01 -1.37372404e-01
1.67570710e-01 -3.06029677e-01 -5.99562109e-01 5.94985247e-01
1.05194640e+00 -7.45872676e-01 -6.24541938e-01 5.28928578e-01
2.00672314e-01 3.32015604e-02 -1.00547874e+00 2.59196997e-01
1.92413703e-01 -8.18755805e-01 7.24847674e-01 -4.89859909e-01
7.42479786e-02 5.49090952e-02 6.32860139e-02 -1.62205935e+00
-4.04577494e-01 -9.94752526e-01 -1.54506281e-01 6.06620848e-01
1.01380177e-01 -9.06575382e-01 2.66891658e-01 5.00397921e-01
-5.22895396e-01 -7.78679132e-01 -1.30159843e+00 -6.99853599e-01
7.24221095e-02 -7.08621979e-01 4.77336019e-01 5.91821730e-01
6.20454967e-01 2.82124817e-01 -3.27282310e-01 3.80599171e-01
6.03294373e-01 -4.21526968e-01 8.03659409e-02 -1.32223964e+00
2.20831320e-01 -5.94246030e-01 -1.07489979e+00 -8.46036613e-01
3.66144866e-01 -1.25048745e+00 -3.37758303e-01 -1.23790622e+00
-4.27246541e-02 -4.13326733e-02 -7.23548412e-01 -5.27399257e-02
1.70596406e-01 1.52622253e-01 -4.07641707e-03 2.08792731e-01
-5.28514981e-01 9.08019900e-01 1.08110631e+00 -9.10883676e-03
-1.68170020e-01 -1.48642659e-01 -3.11561018e-01 4.26847816e-01
7.90736973e-01 -3.10453653e-01 -4.85947937e-01 -5.88350952e-01
3.29494029e-01 1.02642395e-01 4.21461433e-01 -1.41418004e+00
4.47589964e-01 6.36951447e-01 5.04616797e-01 -4.10897672e-01
2.53048480e-01 -7.92028010e-01 1.47229284e-01 1.71681568e-01
-2.78292269e-01 4.40259397e-01 1.93133205e-01 7.06492424e-01
-2.75788158e-01 1.62285030e-01 5.63615859e-01 1.66849434e-01
-2.78202206e-01 6.55483067e-01 -7.19709337e-01 1.71417799e-02
8.55763674e-01 -3.14144194e-01 1.01454847e-01 -2.64976144e-01
-8.89595747e-01 -9.95304659e-02 -3.28066885e-01 5.21799743e-01
9.02728379e-01 -1.32686925e+00 -7.23456919e-01 8.45305383e-01
-6.92781880e-02 -3.73987019e-01 2.75538892e-01 1.59945631e+00
-3.70770484e-01 7.43910193e-01 -5.57813644e-01 -8.97488117e-01
-4.42810118e-01 3.64620864e-01 5.36555290e-01 3.83083016e-01
-1.01695168e+00 7.52634108e-01 4.66983140e-01 1.26705766e-01
4.89281923e-01 -3.77208531e-01 -5.89830339e-01 9.88526922e-03
4.19383317e-01 5.09040058e-01 4.85455245e-01 -8.65846038e-01
-2.89901525e-01 5.21374822e-01 3.18087399e-01 3.06334756e-02
1.62210774e+00 -5.22939973e-02 -5.11287391e-01 6.79419339e-01
1.42733693e+00 -8.21204305e-01 -1.36808014e+00 1.20927043e-01
3.64355028e-01 -3.63099622e-03 2.05372542e-01 -3.51771414e-01
-1.01536417e+00 1.27640462e+00 7.31142819e-01 2.91330963e-01
1.23498988e+00 -3.51344854e-01 4.21833485e-01 4.46245253e-01
7.43590117e-01 -9.28346038e-01 1.81990907e-01 4.56110775e-01
1.26408124e+00 -8.31942081e-01 -3.56562078e-01 7.91762173e-02
-4.96713817e-01 1.47404706e+00 2.19118536e-01 -4.81898934e-01
1.09788430e+00 3.62239704e-02 -4.44537222e-01 -5.42461812e-01
-4.16737169e-01 1.38855264e-01 6.59938514e-01 6.31891489e-01
2.75211304e-01 1.36560097e-01 -1.71424687e-01 8.94129515e-01
-3.71244967e-01 -1.23501748e-01 4.11665887e-01 4.46004540e-01
2.14507151e-02 -4.26458389e-01 -1.90943643e-01 5.93544185e-01
-4.05025870e-01 -2.58342147e-01 -9.06189978e-02 6.25832498e-01
-8.38329792e-02 9.12070692e-01 1.46506041e-01 -5.19160628e-01
1.38250723e-01 1.72416434e-01 7.09278047e-01 -4.40699250e-01
-4.67545956e-01 -3.74945030e-02 -4.64447767e-01 -5.08540869e-01
-4.59694684e-01 -6.67414844e-01 -9.17944849e-01 9.87690315e-03
-3.23951781e-01 1.25423104e-01 8.66452992e-01 1.31651008e+00
7.74304688e-01 7.24956572e-01 4.86513615e-01 -1.22553158e+00
-2.55096555e-01 -1.17306614e+00 -1.05578470e+00 4.22579825e-01
5.52714050e-01 -9.51889336e-01 -3.27009976e-01 -1.62994452e-02]
|
[13.041927337646484, 3.4791946411132812]
|
4511808b-71d4-4722-84ee-bd3fe9c5e018
|
automated-discovery-of-mathematical-1
|
2011.04521
| null |
https://arxiv.org/abs/2011.04521v1
|
https://arxiv.org/pdf/2011.04521v1.pdf
|
Automated Discovery of Mathematical Definitions in Text with Deep Neural Networks
|
Automatic definition extraction from texts is an important task that has numerous applications in several natural language processing fields such as summarization, analysis of scientific texts, automatic taxonomy generation, ontology generation, concept identification, and question answering. For definitions that are contained within a single sentence, this problem can be viewed as a binary classification of sentences into definitions and non-definitions. In this paper, we focus on automatic detection of one-sentence definitions in mathematical texts, which are difficult to separate from surrounding text. We experiment with several data representations, which include sentence syntactic structure and word embeddings, and apply deep learning methods such as the Convolutional Neural Network (CNN) and the Long Short-Term Memory network (LSTM), in order to identify mathematical definitions. Our experiments demonstrate the superiority of CNN and its combination with LSTM, when applied on the syntactically-enriched input representation. We also present a new dataset for definition extraction from mathematical texts. We demonstrate that this dataset is beneficial for training supervised models aimed at extraction of mathematical definitions. Our experiments with different domains demonstrate that mathematical definitions require special treatment, and that using cross-domain learning is inefficient for that task.
|
['Lior Reznik', 'Sergey Shevchuk', 'Marina Litvak', 'Natalia Vanetik']
|
2020-11-09
| null | null | null | null |
['definition-extraction']
|
['natural-language-processing']
|
[ 7.41041541e-01 7.52195567e-02 -1.23641774e-01 -3.53544950e-01
-4.54747975e-01 -6.75187707e-01 7.83363938e-01 1.05418348e+00
-7.45107114e-01 8.06870282e-01 2.33437896e-01 -6.22919917e-01
-3.50159049e-01 -1.14566684e+00 -3.80154580e-01 -4.35707092e-01
6.40719160e-02 2.10997120e-01 -1.50291696e-01 -2.78704464e-01
4.92725760e-01 5.51043928e-01 -1.37927854e+00 1.70143291e-01
8.82527471e-01 8.79320800e-01 3.71495068e-01 7.18194783e-01
-7.83990264e-01 7.68765569e-01 -1.13225949e+00 -3.38676989e-01
-3.21562558e-01 -3.25436354e-01 -1.17211390e+00 -1.36437222e-01
4.48669851e-01 4.18554200e-03 -2.68968362e-02 1.14877152e+00
3.49211186e-01 3.63840818e-01 6.68662012e-01 -7.16000140e-01
-7.53533483e-01 8.19273531e-01 -6.70780167e-02 2.48230249e-01
4.68593657e-01 -3.70264262e-01 9.75733697e-01 -6.63459659e-01
6.43231332e-01 1.19552612e+00 6.40010297e-01 6.20568216e-01
-9.32534695e-01 -3.82074028e-01 9.86513570e-02 1.77380234e-01
-1.23599064e+00 -2.54603654e-01 7.85748422e-01 -4.50930953e-01
1.30078614e+00 3.46373320e-01 3.67797881e-01 9.47274148e-01
3.93438905e-01 8.30909669e-01 7.00300932e-01 -8.35182190e-01
2.94688165e-01 -4.32778373e-02 7.65743256e-01 7.20303476e-01
2.24486485e-01 -4.85584527e-01 -2.74208426e-01 -2.00098231e-01
6.89916670e-01 -2.70201653e-01 -5.33948958e-01 1.52707487e-01
-1.25991607e+00 1.19496155e+00 1.38037786e-01 8.21691334e-01
-2.74448663e-01 1.39902607e-01 7.86662340e-01 3.20528537e-01
6.39474273e-01 1.16967905e+00 -6.94662452e-01 -5.51770292e-02
-8.96870911e-01 2.95295179e-01 9.68428552e-01 1.05834472e+00
4.64003593e-01 -2.23628804e-01 -1.40559092e-01 9.88167822e-01
-2.30481774e-01 2.59636015e-01 9.01900172e-01 -4.86312747e-01
3.97353679e-01 7.89048672e-01 -1.61547616e-01 -1.14086568e+00
-7.67544031e-01 -3.40541303e-01 -1.06337810e+00 -3.03896904e-01
3.86957467e-01 -2.95550466e-01 -7.89104342e-01 1.67739820e+00
2.95402594e-02 -1.70995012e-01 3.05987865e-01 6.82473123e-01
1.69548512e+00 7.35111475e-01 -3.91591638e-02 -2.00182796e-01
1.59814191e+00 -6.02265775e-01 -1.05124712e+00 -9.01002884e-02
9.38272297e-01 -5.37302613e-01 7.82462001e-01 2.38307372e-01
-7.74751544e-01 -5.11980712e-01 -1.20315444e+00 -3.96170408e-01
-1.12860489e+00 2.42660254e-01 4.96829718e-01 2.40508378e-01
-6.32070422e-01 7.75975645e-01 -4.34850156e-01 -5.01223266e-01
3.48772794e-01 4.02283788e-01 -4.58901972e-01 2.80117303e-01
-1.57187366e+00 1.15959728e+00 7.85849869e-01 -5.92150772e-03
-1.97037995e-01 -5.00905633e-01 -1.46917963e+00 3.85652930e-01
3.67034137e-01 -8.08156788e-01 1.07294416e+00 -9.50532734e-01
-1.10647273e+00 1.04450905e+00 -2.19896227e-01 -7.34250546e-01
-2.31907874e-01 -1.50107563e-01 -2.02083409e-01 3.21467966e-01
2.80279279e-01 3.60984981e-01 6.26863301e-01 -1.03772140e+00
-4.51888949e-01 -3.47283989e-01 3.92215461e-01 -4.91222925e-02
-5.47072232e-01 1.60063446e-01 6.75910860e-02 -7.09961176e-01
-6.17922433e-02 -3.20568085e-01 -2.24357069e-01 -1.95971876e-01
-4.56742257e-01 -9.40622449e-01 4.15177107e-01 -7.93624997e-01
1.39639270e+00 -1.71301734e+00 3.09011072e-01 -3.73419095e-03
5.33546567e-01 4.81262982e-01 -1.91251457e-01 4.45746541e-01
-1.25370666e-01 4.50283170e-01 -4.93330657e-01 -1.57918334e-01
2.59512812e-01 2.79987723e-01 -2.28036880e-01 1.96581081e-01
6.00847840e-01 8.33699167e-01 -1.17068350e+00 -6.00502014e-01
3.03034365e-01 2.95346618e-01 2.51065999e-01 2.15862170e-01
-6.19127989e-01 -1.04072504e-01 -3.80277306e-01 7.32216164e-02
6.11226141e-01 -2.18132526e-01 2.67650574e-01 8.69578868e-02
-8.13578516e-02 7.29487956e-01 -8.13447118e-01 1.72363925e+00
-7.49141455e-01 9.90347624e-01 -1.83306903e-01 -1.48262894e+00
1.10730159e+00 3.16489637e-01 1.31877199e-01 -5.61315894e-01
3.21906447e-01 1.54131502e-01 -1.19804136e-01 -8.09972048e-01
6.40766680e-01 -1.66081741e-01 -2.51198798e-01 4.16477412e-01
3.33006114e-01 -4.35515374e-01 7.96970487e-01 1.60334200e-01
1.14967966e+00 -2.46878102e-01 6.90617144e-01 -2.91590691e-01
7.97559679e-01 1.07164942e-02 4.64888752e-01 8.25427592e-01
1.25567421e-01 4.82223719e-01 7.14894772e-01 -5.71510911e-01
-7.00485229e-01 -8.77978623e-01 -4.69476193e-01 8.61047566e-01
-1.84910763e-02 -7.62823462e-01 -6.13094330e-01 -6.67773426e-01
-1.58881947e-01 8.44122469e-01 -5.24352312e-01 -1.74724888e-02
-6.96392894e-01 -5.21662414e-01 5.93519032e-01 3.88596356e-01
7.54556298e-01 -1.20595157e+00 -8.63763571e-01 3.42005759e-01
-6.79343101e-03 -1.23903751e+00 -7.64621571e-02 4.59847689e-01
-5.99589527e-01 -1.27026749e+00 -6.82949483e-01 -1.29789627e+00
4.78626996e-01 -1.69353768e-01 1.41478872e+00 2.21542820e-01
-2.53823549e-01 4.07017738e-01 -5.07176459e-01 -5.96372247e-01
-5.33816099e-01 3.82713586e-01 -1.18132167e-01 -3.67437631e-01
6.53224289e-01 -3.02809387e-01 -1.45949107e-02 -5.20606935e-01
-1.25160217e+00 4.01817299e-02 2.37449259e-01 1.09325194e+00
9.03161913e-02 2.13560522e-01 8.46465886e-01 -9.62486863e-01
1.21330893e+00 -3.67120445e-01 -3.14086467e-01 4.62194175e-01
-2.78240412e-01 3.52833241e-01 8.70258868e-01 -5.36074817e-01
-8.68488908e-01 -1.31542951e-01 -4.09488916e-01 3.83260809e-02
-3.70613992e-01 9.76043582e-01 -4.14199606e-02 3.21694911e-01
5.85367024e-01 4.83567834e-01 -7.28706196e-02 -4.12131667e-01
2.96474278e-01 8.87574792e-01 3.55092406e-01 -3.88442010e-01
4.97202843e-01 -3.43902521e-02 -4.83531207e-02 -1.41113424e+00
-1.10088456e+00 -5.44982314e-01 -9.68203425e-01 6.73568249e-02
9.98837113e-01 -4.35828954e-01 -4.62807089e-01 1.10001400e-01
-1.87861264e+00 -3.64522561e-02 -1.16826281e-01 3.31078023e-01
-1.66486382e-01 6.38657689e-01 -5.49102247e-01 -7.17424989e-01
-7.86630630e-01 -6.34776235e-01 1.08600581e+00 4.99289095e-01
-4.94143397e-01 -1.58714330e+00 -9.99532342e-02 1.88838348e-01
3.81615072e-01 4.53396380e-01 1.39882135e+00 -1.05620170e+00
3.53445997e-03 -1.11874893e-01 -3.32075566e-01 5.38343251e-01
2.92661130e-01 -2.03904614e-01 -8.03883851e-01 1.17270157e-01
4.82941483e-04 -3.85213256e-01 1.28041685e+00 2.07222730e-01
1.14692867e+00 -4.54785526e-01 -8.05331990e-02 3.57852936e-01
1.20069861e+00 9.75730922e-03 2.78594643e-01 3.60745400e-01
5.48165977e-01 7.47455120e-01 3.49303663e-01 2.94413388e-01
9.05874744e-02 9.08023342e-02 1.67506218e-01 -3.03115547e-01
1.29903495e-01 2.86028743e-01 3.65081914e-02 8.65329206e-01
4.04490501e-01 -2.99472034e-01 -1.00455010e+00 7.85314202e-01
-1.85420728e+00 -9.60159302e-01 -1.53586879e-01 1.91908002e+00
1.18316376e+00 8.09108168e-02 -3.00947309e-01 3.14352632e-01
7.65370488e-01 2.68942773e-01 -1.96326450e-01 -1.00595760e+00
-2.37954676e-01 9.85652149e-01 1.14662267e-01 3.29925418e-01
-1.51361978e+00 9.54460144e-01 5.65150881e+00 9.64719534e-01
-1.26322782e+00 -9.01503563e-02 3.11900467e-01 4.19018716e-01
-2.43438378e-01 -3.83474618e-01 -6.19055927e-01 3.13477427e-01
8.90560687e-01 -3.65398467e-01 -2.49452055e-01 6.37799978e-01
2.55510733e-02 -1.83414131e-01 -1.23413372e+00 1.02051651e+00
4.06428248e-01 -1.63209832e+00 2.74535835e-01 -4.23282892e-01
3.92562270e-01 -3.37098747e-01 -4.98408347e-01 3.59138429e-01
-2.39022553e-01 -1.27080762e+00 1.40123725e-01 3.40807617e-01
5.55198133e-01 -7.38377810e-01 1.21037972e+00 3.50475639e-01
-1.07447922e+00 2.54609972e-01 -6.17483079e-01 -5.65744400e-01
-1.21067926e-01 8.34515810e-01 -9.18845415e-01 6.08928621e-01
2.19098348e-02 8.63309145e-01 -5.63860059e-01 7.92822003e-01
-4.97034281e-01 3.17215085e-01 -2.78287232e-01 -8.93536031e-01
2.25076482e-01 -1.92326367e-01 3.26137751e-01 1.76575649e+00
-8.63368958e-02 1.21896446e-01 9.62835178e-02 1.20955181e+00
-3.06219727e-01 3.63018453e-01 -8.66988719e-01 -3.46124589e-01
1.83435261e-01 1.18269348e+00 -7.31181562e-01 -6.49853885e-01
-2.19157651e-01 7.49717236e-01 4.04100209e-01 2.35914499e-01
-6.53169096e-01 -1.11354840e+00 5.00875115e-01 -3.38913292e-01
1.45394132e-01 -5.78897834e-01 -4.67685163e-01 -1.39564204e+00
8.16442221e-02 -4.75182772e-01 2.92274863e-01 -4.95153487e-01
-1.32171047e+00 5.29069364e-01 8.36966261e-02 -8.22258115e-01
-2.50658751e-01 -1.05571318e+00 -8.30728531e-01 7.46634126e-01
-1.54484951e+00 -9.02768612e-01 -1.18168920e-01 2.04638496e-01
7.24207282e-01 4.17830870e-02 1.32845795e+00 1.24422044e-01
-5.41565180e-01 4.01607245e-01 5.73691651e-02 7.58211195e-01
3.54648978e-01 -1.54682505e+00 1.32604882e-01 6.89379573e-01
1.90589905e-01 9.58944142e-01 5.43149233e-01 -2.96241164e-01
-1.20925069e+00 -1.12282383e+00 1.54605556e+00 -1.07129980e-02
6.94153488e-01 -6.05363011e-01 -9.39296126e-01 1.61748528e-01
5.47834814e-01 -4.82682645e-01 9.84722912e-01 3.37562948e-01
-1.20638497e-01 2.02065274e-01 -9.17007565e-01 6.25858545e-01
7.81866789e-01 -5.83072782e-01 -1.24968457e+00 7.27109492e-01
9.79224265e-01 -2.41004512e-01 -1.08823013e+00 3.43895376e-01
3.66196126e-01 -5.20859897e-01 8.04565787e-01 -8.23683918e-01
8.47992539e-01 -3.15155648e-03 2.03730941e-01 -1.15744448e+00
-1.95322316e-02 -3.84053409e-01 3.27494144e-02 1.40970397e+00
5.50597072e-01 -4.61623549e-01 3.67517591e-01 3.96879733e-01
1.09689608e-02 -8.22804868e-01 -1.01496816e+00 -6.11615360e-01
5.94235003e-01 -4.35237676e-01 3.51049751e-01 1.22728550e+00
1.58320293e-01 1.20214832e+00 2.42947608e-01 -1.58044547e-01
2.24632308e-01 4.84871835e-01 6.17529035e-01 -1.15224051e+00
1.54608905e-01 -6.15216553e-01 -6.08950257e-01 -1.33658600e+00
6.67332351e-01 -1.07395148e+00 5.30299246e-02 -2.08401132e+00
-9.93121299e-04 -9.27842557e-02 -9.93823931e-02 4.16108698e-01
-2.19916448e-01 -3.70279312e-01 1.04053482e-01 -4.90775228e-01
-6.35499358e-01 7.71691144e-01 1.00989830e+00 -7.08284259e-01
-3.93829793e-02 -2.50769496e-01 -4.61085409e-01 7.05647588e-01
8.95081341e-01 -3.53409052e-01 -3.06082685e-02 -5.27055979e-01
1.38653979e-01 -1.62674084e-01 1.65713057e-01 -7.79420972e-01
1.18989624e-01 -6.62877113e-02 3.67569685e-01 -5.71293950e-01
1.96212471e-01 -3.68820846e-01 -7.95817614e-01 4.43244994e-01
-5.15322268e-01 2.31107920e-02 2.67857254e-01 9.19093415e-02
-3.46081495e-01 -8.85945559e-01 5.17337382e-01 -3.03560525e-01
-6.47808850e-01 -1.34426177e-01 -7.05830812e-01 1.86110511e-01
4.64825451e-01 1.38442397e-01 -3.18903357e-01 -2.43778959e-01
-5.72166443e-01 4.48999286e-01 -2.65563056e-02 3.99851620e-01
1.02708304e+00 -1.06541061e+00 -8.42073858e-01 -1.04216985e-01
1.10098355e-01 3.91798019e-01 -3.36233079e-01 3.60324353e-01
-6.89476430e-01 5.51414967e-01 2.40559354e-01 -5.71397662e-01
-1.48287868e+00 5.78832150e-01 2.75080651e-01 -5.27038574e-01
-5.71694016e-01 8.90004694e-01 1.55642405e-01 -6.76349878e-01
3.45125288e-01 -8.70430231e-01 -7.37171888e-01 1.27780497e-01
6.26657963e-01 -1.11567052e-02 -1.38760671e-01 -3.18256825e-01
-3.51575524e-01 6.40094638e-01 3.45670395e-02 3.28445762e-01
1.51097918e+00 2.26968944e-01 -6.05132520e-01 6.03559732e-01
1.52125204e+00 -3.08550894e-01 -2.15485185e-01 -3.79335731e-01
5.87170482e-01 1.03649115e-02 8.40219930e-02 -4.51262712e-01
-4.57637072e-01 9.77347136e-01 5.32152876e-02 6.89347506e-01
1.14499319e+00 -4.85725068e-02 9.52341676e-01 8.48845124e-01
-3.01222000e-02 -1.20926499e+00 -1.51495459e-02 1.15716195e+00
1.13661778e+00 -1.32073486e+00 6.40977174e-02 -3.37292701e-01
7.90571645e-02 2.02034259e+00 3.53747159e-01 -2.54778504e-01
5.45595706e-01 1.91830799e-01 -1.23516969e-01 -4.13356692e-01
-6.88160598e-01 -6.08157992e-01 2.57009268e-01 3.70602578e-01
6.67348683e-01 -5.73803559e-02 -7.75062561e-01 8.05554867e-01
-3.36288393e-01 -1.84456766e-01 6.76770806e-01 9.96087253e-01
-4.65536207e-01 -1.12919509e+00 -2.34042957e-01 6.94934726e-01
-5.67445338e-01 -5.06815970e-01 -8.42938900e-01 7.88171411e-01
4.33892220e-01 1.18933129e+00 2.20568001e-01 -7.44573846e-02
6.88938200e-02 1.76403686e-01 6.65741682e-01 -1.12599945e+00
-8.27832460e-01 -3.16924721e-01 4.92953479e-01 1.57845333e-01
-6.44165397e-01 -2.46650964e-01 -1.58542120e+00 -1.15183040e-01
-5.68494916e-01 3.84131342e-01 6.85132444e-01 1.39646864e+00
2.10775882e-02 8.01772058e-01 3.87819022e-01 -4.77060854e-01
-5.59448898e-01 -1.31014800e+00 -3.69495362e-01 7.06403852e-01
4.71910745e-01 -4.38655943e-01 -4.92554665e-01 1.36433631e-01]
|
[10.162980079650879, 8.822569847106934]
|
cc0cb748-5970-4cd2-9069-5da24176b3a2
|
sentence-representations-via-gaussian
|
2305.12990
| null |
https://arxiv.org/abs/2305.12990v1
|
https://arxiv.org/pdf/2305.12990v1.pdf
|
Sentence Representations via Gaussian Embedding
|
Recent progress in sentence embedding, which represents the meaning of a sentence as a point in a vector space, has achieved high performance on tasks such as a semantic textual similarity (STS) task. However, sentence representations as a point in a vector space can express only a part of the diverse information that sentences have, such as asymmetrical relationships between sentences. This paper proposes GaussCSE, a Gaussian distribution-based contrastive learning framework for sentence embedding that can handle asymmetric relationships between sentences, along with a similarity measure for identifying inclusion relations. Our experiments show that GaussCSE achieves the same performance as previous methods in natural language inference tasks, and is able to estimate the direction of entailment relations, which is difficult with point representations.
|
['Koichi Takeda', 'Ryohei Sasano', 'Hayato Tsukagoshi', 'Shohei Yoda']
|
2023-05-22
| null | null | null | null |
['semantic-textual-similarity']
|
['natural-language-processing']
|
[ 2.78300583e-01 1.08275078e-01 -2.72823870e-01 -9.15821016e-01
-2.89815158e-01 -5.94416797e-01 1.01593578e+00 9.57476139e-01
-3.42025369e-01 4.01919156e-01 7.02422082e-01 -5.73489845e-01
-4.90085147e-02 -9.15211380e-01 -4.05887097e-01 -4.11341846e-01
1.09379388e-01 5.36057353e-01 6.98636426e-03 -7.76700556e-01
5.30883670e-01 2.87864327e-01 -1.27637029e+00 6.06373549e-01
4.90797281e-01 7.79571831e-01 -5.41079603e-03 6.58313870e-01
-7.83374012e-01 7.71420836e-01 -8.78108740e-01 -8.33715856e-01
-3.42256814e-01 -4.52779174e-01 -1.23166811e+00 -3.75222921e-01
7.16198683e-01 9.76322070e-02 -4.82822955e-01 1.11691749e+00
1.94940150e-01 9.11547393e-02 1.19123721e+00 -1.44168055e+00
-1.13185823e+00 7.57324636e-01 -1.26980767e-01 5.00209451e-01
8.53445530e-01 -5.21646440e-01 1.70405900e+00 -1.13190460e+00
4.58895057e-01 1.79912007e+00 8.69798303e-01 3.53667736e-01
-1.41650259e+00 3.83493267e-02 -3.61618958e-02 5.43566406e-01
-9.58782792e-01 -1.12204805e-01 9.75037634e-01 -3.87338430e-01
1.41597271e+00 6.49310291e-01 4.10122007e-01 1.07128131e+00
5.49224973e-01 7.20503032e-01 7.21530497e-01 -4.64104950e-01
1.04411155e-01 2.21991241e-01 5.91308594e-01 2.37998500e-01
1.57365084e-01 -5.71669400e-01 -3.51342648e-01 -3.61302733e-01
2.70227551e-01 8.08206871e-02 -1.73671991e-01 -2.71011651e-01
-1.16390312e+00 1.18647611e+00 7.05087483e-01 7.12223828e-01
-2.80638300e-02 2.33144149e-01 7.58461416e-01 6.55955613e-01
7.67118633e-01 7.44376540e-01 -4.24474955e-01 -1.57168344e-01
-4.86472666e-01 5.11691034e-01 8.30320835e-01 8.10964465e-01
5.74753821e-01 -3.22969466e-01 -4.35451269e-01 8.82434070e-01
2.26269990e-01 3.49763811e-01 6.14904642e-01 -6.60216689e-01
6.80869699e-01 7.82860279e-01 -3.13282102e-01 -1.58572173e+00
-1.57903716e-01 -3.16998474e-02 -6.68761492e-01 -2.49566063e-01
-3.97340916e-02 2.75983989e-01 -1.28800899e-01 1.71359670e+00
3.36588081e-03 -1.02084987e-01 3.74588639e-01 5.35401762e-01
1.09332919e+00 6.90209329e-01 -1.97543710e-01 7.38652200e-02
1.57501233e+00 -6.45506144e-01 -8.31108332e-01 -4.25264001e-01
1.03716290e+00 -6.53425872e-01 1.24221444e+00 -5.00173569e-01
-1.07285559e+00 -4.56044108e-01 -1.05530334e+00 -4.98245239e-01
-6.39586031e-01 -3.43695700e-01 8.08016539e-01 3.57603073e-01
-9.79804635e-01 8.64489496e-01 -3.82100374e-01 -4.19582605e-01
3.25900167e-01 8.40073824e-02 -4.85621303e-01 4.37461119e-03
-1.55426025e+00 1.28704858e+00 5.13155282e-01 -1.10010654e-01
3.05931509e-01 -8.87572408e-01 -1.68222213e+00 4.00645703e-01
2.87942104e-02 -8.86999428e-01 9.25943196e-01 -5.35757542e-01
-8.92836690e-01 1.02163744e+00 -4.92721766e-01 -6.07237339e-01
-8.32970515e-02 -2.32065879e-02 -3.66159439e-01 -4.23915917e-03
1.83533385e-01 4.35246617e-01 6.13279223e-01 -8.37549746e-01
9.65227559e-03 -4.85469997e-01 4.77958202e-01 3.87024760e-01
-5.53853154e-01 2.34840348e-01 4.66252804e-01 -6.94186628e-01
2.63152063e-01 -5.97585559e-01 -8.93566459e-02 3.01089525e-01
-3.54048252e-01 -1.01375031e+00 8.57888877e-01 -4.51968402e-01
1.24463654e+00 -2.14274502e+00 3.36012691e-01 -2.19583064e-01
2.29973853e-01 1.46000743e-01 -9.89399776e-02 9.39025462e-01
-3.65154088e-01 2.53142208e-01 -5.79209208e-01 -4.78657931e-01
4.22680140e-01 5.05866826e-01 -5.42925835e-01 3.91184717e-01
6.37407720e-01 1.06960356e+00 -1.20777464e+00 -6.03307605e-01
8.48783255e-02 3.18572760e-01 -4.51276600e-01 9.79212448e-02
-6.20380230e-03 -3.50813031e-01 -9.29147080e-02 -1.52916402e-01
5.59802532e-01 -1.31115079e-01 3.88332456e-01 -4.11627889e-01
3.61326575e-01 7.92932749e-01 -8.63148153e-01 1.80260551e+00
-5.55309713e-01 1.13011467e+00 -4.88703310e-01 -1.43920898e+00
1.08383858e+00 2.27076471e-01 4.19609901e-03 -3.16152871e-01
7.36990124e-02 -1.79233462e-01 8.75272378e-02 -5.19301355e-01
8.85822713e-01 -4.29803401e-01 -4.01969016e-01 5.51389813e-01
8.81489664e-02 -7.29678094e-01 2.81554043e-01 6.00942791e-01
1.06945324e+00 -3.56264502e-01 4.20192659e-01 -4.83732164e-01
6.13766134e-01 -4.21359539e-01 1.93113908e-02 6.07293189e-01
-5.76442666e-02 5.23217618e-01 9.41112101e-01 -3.53971332e-01
-1.12007356e+00 -1.33330250e+00 -4.86615866e-01 7.75967836e-01
1.77036479e-01 -8.76775503e-01 -1.93901315e-01 -8.88570547e-01
3.59412581e-01 1.13447869e+00 -7.10850477e-01 -5.65652311e-01
-3.80112946e-01 -4.23097521e-01 4.16639984e-01 9.65293229e-01
-1.25469804e-01 -7.91376829e-01 -3.96645814e-02 -1.68096602e-01
-1.81735307e-01 -1.10831130e+00 -5.78101099e-01 1.70962840e-01
-6.87610865e-01 -9.73706305e-01 -1.89067066e-01 -1.01452351e+00
6.40174329e-01 1.97380513e-01 1.37134373e+00 -1.53361142e-01
-1.43681377e-01 2.29719564e-01 -4.00522441e-01 -2.10785747e-01
-6.19445801e-01 -5.13821363e-01 1.27492949e-01 -1.96249858e-01
6.96239471e-01 -4.49952722e-01 -2.88708031e-01 -3.38301480e-01
-1.03850591e+00 -3.88453692e-01 -6.17844462e-02 1.23253381e+00
3.42448056e-03 -8.53821114e-02 4.95915622e-01 -9.42534864e-01
1.31266212e+00 -5.78012884e-01 3.59466255e-01 3.24115068e-01
-5.09570658e-01 4.76474315e-01 6.53409004e-01 -3.55554193e-01
-7.21216917e-01 -7.06076145e-01 -3.44146997e-01 -2.07514599e-01
6.43243566e-02 7.61474907e-01 7.20397830e-02 3.04155022e-01
7.24996567e-01 3.10152054e-01 2.59454519e-01 -2.51886159e-01
7.13302493e-01 8.27982068e-01 3.76839161e-01 -4.43356305e-01
5.35224617e-01 2.31903329e-01 2.32490391e-01 -9.80026841e-01
-1.14010763e+00 -6.69784844e-01 -8.81842136e-01 4.85785127e-01
6.56543851e-01 -4.63915467e-01 -5.56613624e-01 -1.13773413e-01
-1.63984132e+00 4.49501455e-01 -5.59522331e-01 5.28521180e-01
-4.58730519e-01 8.84615958e-01 -7.31885731e-01 -4.84617680e-01
-3.07420373e-01 -6.25891626e-01 1.10809708e+00 -1.82387754e-01
-9.23301458e-01 -1.74260068e+00 2.24124312e-01 1.94710627e-01
5.45077682e-01 5.75392731e-02 1.37474239e+00 -1.04340219e+00
2.80418664e-01 -4.20028001e-01 -1.06636427e-01 5.88654160e-01
2.81649143e-01 -8.96872953e-02 -6.18791103e-01 -1.90869197e-01
4.38982606e-01 -3.17352831e-01 9.28221107e-01 8.94407257e-02
1.06914330e+00 -4.12097752e-01 -2.34253347e-01 2.31554687e-01
1.14484203e+00 -2.26146311e-01 5.61412573e-01 5.26586734e-02
5.91092110e-01 6.51383758e-01 3.60185772e-01 1.66543275e-01
4.69129711e-01 6.97751045e-01 2.99921095e-01 3.91658396e-02
-4.50203381e-02 -2.84317583e-01 2.01755837e-01 1.07736588e+00
5.36823034e-01 -1.28173515e-01 -5.47418356e-01 5.11751115e-01
-1.81814575e+00 -1.21081448e+00 -2.78937608e-01 1.74181449e+00
1.04065824e+00 2.09797904e-01 -3.15183878e-01 4.63429660e-01
6.38956130e-01 6.20128155e-01 -1.21215299e-01 -1.10698700e+00
-3.12372804e-01 3.00301701e-01 -1.89327404e-01 4.94120181e-01
-1.00296164e+00 7.82497644e-01 6.89994907e+00 6.84666097e-01
-8.53539526e-01 -5.44265807e-02 2.35233307e-01 3.77908111e-01
-9.54102576e-01 5.06843030e-02 -4.06482369e-01 4.44520503e-01
7.13749588e-01 -5.20994842e-01 -3.08748454e-01 7.60585785e-01
-3.26743096e-01 3.04441810e-01 -1.67806721e+00 1.09816873e+00
7.02506781e-01 -1.52008152e+00 3.62646610e-01 -4.80803400e-01
3.89841676e-01 -3.68630916e-01 1.30545804e-02 2.91423082e-01
-1.44856155e-01 -1.11694145e+00 2.96867818e-01 3.46732080e-01
5.51036716e-01 -6.80254757e-01 1.16464186e+00 2.76653320e-01
-1.06036174e+00 1.22800894e-01 -8.03820133e-01 -5.40279388e-01
2.14184225e-01 4.30163234e-01 -9.83290195e-01 5.71903527e-01
2.82990485e-01 1.14071143e+00 -6.45015478e-01 5.21071017e-01
-4.08984780e-01 2.89202124e-01 5.52616380e-02 -6.19810343e-01
1.47900119e-01 -2.37723291e-01 6.49436891e-01 1.43423533e+00
-9.68171004e-03 -2.65942842e-01 -2.72771679e-02 1.15648830e+00
3.49393524e-02 1.13901012e-01 -1.09456670e+00 -2.67558634e-01
4.69305634e-01 9.49606359e-01 -3.21626782e-01 -3.83297056e-01
-4.79699701e-01 1.14351153e+00 5.54649234e-01 8.70233029e-03
-5.86850643e-01 -6.85997903e-01 9.41421092e-01 -1.78060547e-01
6.71151578e-02 -3.11185777e-01 -2.47980475e-01 -1.21426105e+00
3.35761964e-01 -2.93235391e-01 1.88514933e-01 -9.09114897e-01
-1.80957532e+00 6.45594060e-01 1.58002943e-01 -1.01449955e+00
-3.71809751e-01 -8.55695307e-01 -9.92596865e-01 9.06995475e-01
-1.12280631e+00 -8.94603491e-01 2.92815447e-01 3.53331387e-01
5.80473840e-01 -4.59387414e-02 1.23726344e+00 -7.50917122e-02
-1.15908138e-01 6.80259407e-01 7.45715424e-02 2.91783482e-01
4.85671788e-01 -1.72976696e+00 7.48491287e-01 3.49538058e-01
6.31108820e-01 1.01571953e+00 9.40156698e-01 -2.15411007e-01
-1.26834774e+00 -7.21478999e-01 1.72569311e+00 -7.43605018e-01
1.06343639e+00 -5.42923868e-01 -1.06479597e+00 5.98752916e-01
3.74500990e-01 -4.66256104e-02 9.64232564e-01 6.16043091e-01
-7.94086516e-01 6.41492829e-02 -9.88906443e-01 8.34572613e-01
9.24272060e-01 -1.25073302e+00 -1.54841173e+00 5.70418715e-01
1.11349797e+00 -1.07406124e-01 -9.80268717e-01 2.47553945e-01
1.42028317e-01 -6.28539324e-01 1.17739320e+00 -1.27463877e+00
8.68669271e-01 1.59925390e-02 -4.63353574e-01 -1.62009025e+00
-3.79991740e-01 -5.33506684e-02 -1.79418787e-01 1.17057443e+00
3.88765126e-01 -6.08079970e-01 5.86703956e-01 4.38597172e-01
-4.08511423e-02 -1.03642738e+00 -8.84317815e-01 -9.16544199e-01
3.90532047e-01 -3.06936800e-01 5.84740639e-01 1.20079565e+00
6.13688529e-01 1.13500440e+00 1.01485260e-01 -1.97249413e-01
3.25126261e-01 2.49066487e-01 2.67632633e-01 -1.38658428e+00
-1.31035835e-01 -6.39985383e-01 -1.30742729e+00 -1.00590026e+00
8.75950396e-01 -1.49240792e+00 -3.49406689e-01 -1.94244182e+00
3.64084065e-01 5.33254780e-02 -5.47390711e-03 5.31642921e-02
-4.39916402e-01 -1.19489126e-01 2.25973666e-01 -1.75005674e-01
-4.74542320e-01 1.03120410e+00 1.00857401e+00 -6.70677364e-01
5.07100105e-01 -1.12981081e-01 -6.69192374e-01 6.58169389e-01
6.18772984e-01 -4.34113324e-01 -6.02933705e-01 -4.15268242e-01
4.38484490e-01 6.69200625e-03 3.10659796e-01 -3.57603818e-01
1.29195973e-01 9.63872895e-02 1.61806092e-01 -6.03255510e-01
7.16372967e-01 -6.47198319e-01 -4.33528870e-01 5.06192386e-01
-8.88454735e-01 3.71984839e-01 -1.46573827e-01 5.46730280e-01
-6.19621456e-01 -6.36148274e-01 1.76644176e-01 4.51309793e-02
-5.18765271e-01 -9.38856825e-02 -1.31771937e-01 3.86734903e-01
6.53124928e-01 -1.65613458e-01 -5.48861325e-01 -3.99353176e-01
-5.43173909e-01 1.22743383e-01 1.88984469e-01 7.32330322e-01
1.01788211e+00 -1.68700302e+00 -9.58687067e-01 5.12970239e-02
3.83580714e-01 -2.94364363e-01 6.13944642e-02 5.11796892e-01
-1.90566719e-01 3.87990117e-01 1.65674152e-04 -6.67923152e-01
-1.54067874e+00 7.42037952e-01 1.12854823e-01 -4.51812632e-02
-6.61508918e-01 1.00204372e+00 1.03237681e-01 -6.79915130e-01
-1.38652653e-01 -4.19134438e-01 -4.63792145e-01 1.54868916e-01
5.61620653e-01 -8.07066076e-03 -2.29834672e-02 -7.65951455e-01
-5.37203133e-01 5.03172874e-01 -1.99507907e-01 9.98762101e-02
1.11093974e+00 1.12852016e-02 -7.49764681e-01 1.05861616e+00
1.89560330e+00 -1.99102595e-01 -3.72576654e-01 -4.70963299e-01
2.56529659e-01 -6.49616897e-01 -2.22798660e-01 -9.72043872e-02
-2.46218875e-01 1.10002518e+00 -8.76227245e-02 6.91291630e-01
4.34567183e-01 4.93508011e-01 7.26747155e-01 7.04061627e-01
-5.39402477e-02 -7.63098836e-01 4.05196071e-01 8.47005606e-01
1.19801259e+00 -1.28748655e+00 5.79608604e-03 -5.49953043e-01
-7.49852180e-01 1.42070162e+00 2.05826625e-01 -2.41533235e-01
8.34930718e-01 1.38025641e-01 -2.86173880e-01 -3.74883562e-01
-8.36625934e-01 5.62081113e-02 5.43271542e-01 6.03685677e-01
8.39161634e-01 2.19429046e-01 -6.56168699e-01 4.17875111e-01
-5.69096327e-01 -8.14311147e-01 4.67108995e-01 8.78341675e-01
-2.67998040e-01 -1.05990076e+00 1.13676995e-01 6.38603270e-01
-3.19219500e-01 -4.94190037e-01 -5.94506979e-01 3.68506491e-01
-3.76397431e-01 9.31563377e-01 5.57661831e-01 -3.43279362e-01
2.58348584e-01 1.54985175e-01 6.15660906e-01 -9.26861227e-01
-3.83109987e-01 -7.54138589e-01 3.45979452e-01 -2.83913076e-01
-3.24914634e-01 -4.09947306e-01 -1.22654986e+00 -4.00247931e-01
-2.62893677e-01 4.65684325e-01 6.28127992e-01 1.12775147e+00
2.18619078e-01 5.26544631e-01 7.48660445e-01 -4.89178419e-01
-1.08924150e+00 -1.04103744e+00 -6.71910524e-01 1.08649695e+00
2.43862256e-01 -4.71088022e-01 -6.46559119e-01 -3.78861517e-01]
|
[10.892816543579102, 8.866333961486816]
|
2fdded08-9182-4561-b712-56ee3f9a9969
|
learning-a-single-convolutional-layer-model
|
2305.14039
| null |
https://arxiv.org/abs/2305.14039v1
|
https://arxiv.org/pdf/2305.14039v1.pdf
|
Learning a Single Convolutional Layer Model for Low Light Image Enhancement
|
Low-light image enhancement (LLIE) aims to improve the illuminance of images due to insufficient light exposure. Recently, various lightweight learning-based LLIE methods have been proposed to handle the challenges of unfavorable prevailing low contrast, low brightness, etc. In this paper, we have streamlined the architecture of the network to the utmost degree. By utilizing the effective structural re-parameterization technique, a single convolutional layer model (SCLM) is proposed that provides global low-light enhancement as the coarsely enhanced results. In addition, we introduce a local adaptation module that learns a set of shared parameters to accomplish local illumination correction to address the issue of varied exposure levels in different image regions. Experimental results demonstrate that the proposed method performs favorably against the state-of-the-art LLIE methods in both objective metrics and subjective visual effects. Additionally, our method has fewer parameters and lower inference complexity compared to other learning-based schemes.
|
['Wenpeng Ding', 'Gang Li', 'Haichuan Ma', 'Zhenzhong Chen', 'Daiqin Yang', 'Baoxin Teng', 'Yuantong Zhang']
|
2023-05-23
| null | null | null | null |
['image-enhancement', 'low-light-image-enhancement']
|
['computer-vision', 'computer-vision']
|
[ 3.41107249e-01 -5.51138163e-01 1.06017761e-01 -4.03837472e-01
-4.73773211e-01 1.76489186e-02 2.31973827e-01 -2.86806643e-01
-5.50800383e-01 7.12321341e-01 -2.44015940e-02 -1.04666486e-01
-5.44537511e-03 -5.77291846e-01 -6.73898697e-01 -1.06617510e+00
2.47522160e-01 -9.15089309e-01 3.07610512e-01 -1.44699022e-01
3.81220579e-01 5.16109467e-01 -1.64655781e+00 5.47130927e-02
1.10651636e+00 9.42616343e-01 4.94956493e-01 6.25778019e-01
1.79815114e-01 8.98652434e-01 -2.91637510e-01 -1.34536698e-01
2.97297984e-01 -3.40375751e-01 -2.91371852e-01 8.39860439e-02
8.75311613e-01 -6.52239323e-01 -3.08375746e-01 1.10925138e+00
9.95535731e-01 2.28594244e-01 2.56607205e-01 -8.45499158e-01
-5.69452643e-01 -2.79626600e-03 -9.35648203e-01 3.35645378e-01
-8.69458020e-02 3.80705297e-01 4.19733107e-01 -1.15316594e+00
1.45161033e-01 9.70314682e-01 7.06514657e-01 3.43097240e-01
-1.01753032e+00 -7.93827832e-01 2.27474719e-01 4.31721807e-01
-1.29295969e+00 -5.13185501e-01 8.91244650e-01 9.73678455e-02
7.41687715e-01 4.44511622e-02 3.61514837e-01 5.78634441e-01
5.09061992e-01 4.71730173e-01 1.56652701e+00 -6.42407894e-01
5.02851233e-02 6.90879822e-02 -2.12345570e-01 9.30078864e-01
2.47731060e-01 2.43188411e-01 -4.42176700e-01 2.14279607e-01
7.91382849e-01 -5.34372330e-02 -3.29145551e-01 -1.12094700e-01
-7.61262894e-01 3.60887587e-01 6.76366985e-01 1.38657957e-01
-2.37442255e-01 3.22021812e-01 2.76260078e-01 -1.28020808e-01
6.09370708e-01 2.82554954e-01 -5.47607720e-01 3.55671883e-01
-6.62346840e-01 -3.78321931e-02 1.71855450e-01 5.83620310e-01
1.00723100e+00 1.18080713e-01 -3.78054976e-01 9.61153805e-01
3.19976062e-01 5.32687843e-01 1.00154437e-01 -9.55903053e-01
4.22712713e-01 3.51140022e-01 3.01581949e-01 -1.15382898e+00
-4.38614964e-01 -5.86585224e-01 -1.11519635e+00 5.94835162e-01
-5.62027954e-02 -1.41727582e-01 -7.96587586e-01 1.68139398e+00
3.57051313e-01 4.38000977e-01 2.45334860e-02 1.00589931e+00
7.97863305e-01 7.38340557e-01 2.16313586e-01 -4.56852078e-01
1.12083721e+00 -1.22188330e+00 -1.05491316e+00 -1.57007158e-01
1.36044845e-01 -1.04571342e+00 1.30227900e+00 3.52727592e-01
-1.36345541e+00 -9.88934875e-01 -1.27919853e+00 -2.02101946e-01
-2.19673753e-01 4.72657442e-01 6.26247823e-01 8.43300104e-01
-1.10750735e+00 4.25617814e-01 -6.14356756e-01 8.02074224e-02
4.64349419e-01 3.04665864e-01 1.71057001e-01 -2.92453319e-01
-1.04524052e+00 8.37083101e-01 1.63300633e-01 5.46678424e-01
-8.00343752e-01 -8.50731790e-01 -7.15142310e-01 -4.25346494e-02
3.74733180e-01 -5.34981668e-01 9.57352459e-01 -8.21711898e-01
-1.90354085e+00 5.34244001e-01 -2.74833202e-01 5.46135008e-02
2.88952917e-01 -3.56619596e-01 -4.90745515e-01 8.12517256e-02
-4.21664983e-01 4.17349875e-01 1.01915371e+00 -1.57653081e+00
-6.53663933e-01 1.18928328e-02 5.49570173e-02 4.03505385e-01
-4.61176306e-01 2.24337786e-01 -6.77930832e-01 -4.71894264e-01
-1.56237468e-01 -6.35339320e-01 -2.83909976e-01 2.61127204e-01
1.82160903e-02 4.87533361e-02 9.72794652e-01 -5.37148952e-01
1.42928100e+00 -2.05495906e+00 -3.76990020e-01 1.33180129e-03
3.01219642e-01 7.86197007e-01 -1.88481465e-01 1.03413790e-01
-1.47559913e-02 -1.13230594e-01 -2.41595313e-01 -3.39073569e-01
-2.29419589e-01 2.53655147e-02 5.99809140e-02 5.94304681e-01
1.24121852e-01 7.67899513e-01 -8.21097374e-01 -7.24672794e-01
6.58296168e-01 9.00524974e-01 -5.49537838e-01 3.91353965e-01
1.40146390e-01 4.24503297e-01 -1.73811197e-01 5.49630225e-01
1.26914716e+00 -6.03066683e-02 -8.04784074e-02 -8.47008526e-01
-6.08744383e-01 -3.04180652e-01 -1.22658300e+00 1.60960042e+00
-8.77963603e-01 5.18119633e-01 1.31696805e-01 -4.03834641e-01
7.74929166e-01 3.52999568e-02 2.79260784e-01 -9.80319023e-01
4.19644773e-01 1.19098499e-01 -3.05151373e-01 -8.02660108e-01
3.73198211e-01 2.30231527e-02 5.64775169e-01 3.21025491e-01
-2.47490719e-01 9.07376111e-02 -1.39919750e-03 -1.86392158e-01
5.76828301e-01 3.68012190e-01 2.32302040e-01 -7.65134692e-02
8.86828303e-01 -7.95858264e-01 6.30367458e-01 7.48679101e-01
-4.28284794e-01 3.82772624e-01 -2.82238156e-01 -4.22120780e-01
-9.50707495e-01 -8.88836563e-01 -2.68934727e-01 1.07525277e+00
6.17361724e-01 -1.06476262e-01 -8.18874061e-01 -2.01690987e-01
-5.94657600e-01 3.89611065e-01 -2.66199261e-01 -1.85387507e-01
-7.12515116e-01 -1.11819148e+00 1.85904399e-01 3.23900878e-01
1.35192037e+00 -8.07632029e-01 -7.71988213e-01 3.29616629e-02
-2.57821023e-01 -1.20368719e+00 -5.65665305e-01 -5.07254247e-03
-6.53212249e-01 -7.82427609e-01 -5.19901991e-01 -9.34682786e-01
7.28473723e-01 6.54381335e-01 8.63817930e-01 1.04862183e-01
-5.21273613e-01 5.87295443e-02 -1.98695540e-01 -4.71908242e-01
-3.24521326e-02 -2.52607226e-01 -1.29307449e-01 2.73075938e-01
5.92236705e-02 -4.82068717e-01 -1.30383039e+00 3.94853979e-01
-1.06549859e+00 2.05214649e-01 8.49288225e-01 8.63247752e-01
7.07328618e-01 5.26936173e-01 3.74272525e-01 -6.01436913e-01
6.37328029e-01 1.13826886e-01 -7.25187778e-01 4.83575165e-01
-9.77129221e-01 -1.78722695e-01 6.40326798e-01 -4.22162175e-01
-1.90923250e+00 -1.29104361e-01 -1.44533841e-02 -1.28808975e-01
-1.38078809e-01 1.82652190e-01 -3.20452541e-01 -8.17078590e-01
4.95282203e-01 1.70932963e-01 -1.87736675e-01 -4.63665903e-01
3.11291724e-01 5.89041770e-01 6.42022133e-01 -4.35500473e-01
7.47488141e-01 5.27852833e-01 2.86464244e-01 -6.30899429e-01
-9.94857609e-01 -2.78820127e-01 -4.84293640e-01 -5.26012123e-01
7.30735540e-01 -1.08192039e+00 -9.38795447e-01 9.62171912e-01
-9.62245643e-01 -4.66544628e-01 1.17366247e-01 3.85786444e-01
-3.00860912e-01 5.66119969e-01 -6.51538432e-01 -7.36427009e-01
-5.72342098e-01 -1.12466514e+00 7.99344718e-01 7.72602499e-01
6.83981776e-01 -9.71234024e-01 -5.69391735e-02 2.54953712e-01
8.61623645e-01 1.88822016e-01 6.97827160e-01 6.01887882e-01
-7.43774176e-01 2.08344445e-01 -7.55945444e-01 7.26992905e-01
3.48945409e-01 -8.96835104e-02 -1.26336980e+00 -5.68352044e-01
1.08629316e-01 -2.38016024e-01 7.72622824e-01 6.99993551e-01
1.67064047e+00 -1.41844466e-01 -5.32630198e-02 1.14489627e+00
1.93940437e+00 1.80150017e-01 9.36761260e-01 5.01042843e-01
6.44666076e-01 3.40410918e-01 6.93124592e-01 5.27737975e-01
2.13366777e-01 5.24466753e-01 5.80292225e-01 -1.06592238e+00
-5.60734749e-01 9.05419141e-03 2.34893128e-01 6.53321266e-01
-1.69782996e-01 -2.58523166e-01 -3.24165851e-01 2.33518153e-01
-1.64718974e+00 -7.91671276e-01 -1.09203495e-01 1.95319831e+00
1.02603137e+00 -6.50119632e-02 -4.13484514e-01 -5.79306707e-02
7.34782755e-01 3.62494677e-01 -6.73589110e-01 -2.97282815e-01
-3.21299016e-01 3.00330371e-01 8.08129668e-01 5.73401928e-01
-1.08793974e+00 8.54042113e-01 6.79826546e+00 8.68528605e-01
-1.20719063e+00 1.85747400e-01 7.12438703e-01 -2.52464842e-02
4.21865396e-02 -1.65978834e-01 -8.17739248e-01 5.14281929e-01
5.48086822e-01 2.73531049e-01 5.42575777e-01 4.18537885e-01
7.51845837e-01 -3.32477838e-01 -5.39736390e-01 1.11394095e+00
3.65799546e-01 -1.03804159e+00 -2.39111945e-01 -2.22146884e-01
1.05718231e+00 -8.33750665e-02 3.94968987e-01 -8.40571597e-02
-6.19845726e-02 -7.78942168e-01 3.12570512e-01 7.14605749e-01
1.04856169e+00 -8.73201966e-01 7.81270921e-01 -1.40853096e-02
-1.22436190e+00 -2.80726165e-01 -4.98476118e-01 2.67876056e-03
5.51405437e-02 5.57397127e-01 -2.84460396e-01 4.97723520e-01
9.67545033e-01 6.07254922e-01 -7.62322485e-01 1.12993288e+00
-4.40472394e-01 5.48807502e-01 -6.07301444e-02 1.75903648e-01
1.40674442e-01 -2.23829210e-01 1.46916226e-01 1.28026843e+00
7.66991749e-02 2.71048754e-01 5.24637587e-02 7.22559750e-01
1.89395379e-02 4.69311997e-02 -1.22967571e-01 6.99815810e-01
1.27813995e-01 1.50378871e+00 -2.94374287e-01 -6.35961965e-02
-6.16353691e-01 9.70695257e-01 2.36875191e-02 6.72013521e-01
-9.74252582e-01 -6.68392360e-01 4.35634702e-01 3.64445746e-02
1.74836457e-01 -1.47932380e-01 -2.15277702e-01 -9.31401253e-01
4.92414534e-02 -6.69988751e-01 7.60562941e-02 -1.25583637e+00
-1.04312468e+00 5.54509997e-01 -8.13676268e-02 -1.18801463e+00
5.29805779e-01 -6.40091479e-01 -6.21403337e-01 9.42350268e-01
-2.55839849e+00 -1.27190340e+00 -8.57943714e-01 8.65652442e-01
4.96714592e-01 2.85633989e-02 3.42650741e-01 5.73643744e-01
-8.29829633e-01 6.75387442e-01 1.68313354e-01 -3.04384172e-01
1.16448116e+00 -8.14569533e-01 -1.81769580e-01 1.33317828e+00
-5.36674738e-01 5.26312590e-01 6.02818608e-01 -3.10735017e-01
-1.09533465e+00 -1.24980056e+00 6.07312083e-01 1.66536465e-01
3.48704398e-01 -5.04585356e-02 -7.88013875e-01 1.81369156e-01
5.39243937e-01 2.95220673e-01 5.14396429e-01 -2.91279554e-01
-1.62069052e-01 -7.50051141e-01 -1.09002984e+00 7.61098027e-01
8.35128784e-01 -3.37345988e-01 1.18596433e-02 1.51255846e-01
6.34037852e-01 -5.55338204e-01 -7.33750105e-01 5.44100404e-01
4.74396825e-01 -1.10797453e+00 1.15303171e+00 3.91037297e-03
4.61069882e-01 -7.11258113e-01 -1.14803553e-01 -1.22216988e+00
-5.78154147e-01 -7.55308092e-01 3.87455560e-02 1.27063143e+00
2.41804365e-02 -6.68020964e-01 4.23012644e-01 4.51813370e-01
-2.62297004e-01 -8.29796553e-01 -4.43424344e-01 -4.20242399e-01
-4.58626270e-01 -1.05411313e-01 4.72464293e-01 5.46769261e-01
-5.95596731e-01 1.03057750e-01 -8.08501720e-01 5.14930427e-01
1.03065979e+00 5.06552532e-02 6.19570315e-01 -6.84105456e-01
-7.71027431e-02 -1.06632449e-01 3.52927521e-02 -9.83631551e-01
2.97336355e-02 -4.04289722e-01 2.25688457e-01 -1.37824273e+00
5.26659310e-01 -4.39753264e-01 -7.93208539e-01 4.86418813e-01
-6.03433490e-01 5.12230039e-01 -6.97183982e-02 7.93741737e-03
-7.44813681e-01 5.85690856e-01 1.54405689e+00 -1.16656376e-02
-2.84785658e-01 -2.40138069e-01 -7.50106037e-01 7.30650902e-01
8.76749516e-01 -1.88600406e-01 -5.11534095e-01 -7.58870244e-01
2.46872142e-01 -4.68223482e-01 5.52879751e-01 -1.07508886e+00
4.64323342e-01 -3.08408320e-01 6.00940943e-01 -5.86468279e-01
1.51471421e-01 -9.91283774e-01 -2.47412264e-01 2.98228621e-01
-3.53637218e-01 -1.89605758e-01 2.86257982e-01 4.41639066e-01
-1.41566008e-01 2.27538589e-03 1.39996946e+00 6.78604096e-02
-9.45462227e-01 3.21572989e-01 -2.02122837e-01 -2.94740349e-01
8.96328270e-01 -2.77235508e-01 -5.38176835e-01 -7.74139315e-02
-4.40466926e-02 8.40194672e-02 2.66444415e-01 1.00584053e-01
8.84763002e-01 -1.16956162e+00 -5.43017566e-01 3.40727150e-01
1.30390361e-01 -2.17418849e-01 7.54825711e-01 9.71128404e-01
-5.50270677e-01 1.39373034e-01 -3.78780603e-01 -3.46434683e-01
-1.43678236e+00 4.50421900e-01 6.05129123e-01 -1.67890549e-01
-5.24153888e-01 7.68109500e-01 3.16914201e-01 -9.55171883e-02
2.61742026e-01 -1.77590773e-01 -9.23215598e-02 -6.57011449e-01
8.23572993e-01 5.26843727e-01 1.53492361e-01 -3.16138595e-01
-5.67096137e-02 8.72530460e-01 -1.17229283e-01 4.19594139e-01
1.42067659e+00 -7.80662477e-01 -1.45172998e-01 -7.91589171e-02
1.08534443e+00 4.96997237e-02 -1.79179192e+00 -4.08934534e-01
-6.92936540e-01 -8.92146945e-01 7.82422483e-01 -1.08924615e+00
-1.22420144e+00 8.11642349e-01 1.28272712e+00 -4.50866163e-01
1.84682631e+00 -5.93043745e-01 8.79226685e-01 2.25094780e-01
-4.41911183e-02 -1.41079748e+00 3.88173878e-01 2.22073033e-01
5.74111819e-01 -1.45827615e+00 2.66558349e-01 -4.15690929e-01
-2.96563625e-01 9.83444989e-01 8.78536761e-01 1.06941476e-01
5.83782971e-01 5.31473696e-01 2.74453044e-01 -1.70067381e-02
-5.82397938e-01 -1.44972414e-01 1.93780482e-01 6.41995966e-01
3.81831139e-01 -4.01681364e-01 -3.56251508e-01 3.12974043e-02
5.76930821e-01 1.93621963e-01 3.65679473e-01 8.05354297e-01
-6.18548214e-01 -9.18936789e-01 -3.94283742e-01 7.74050271e-03
-6.03578091e-01 -3.46694618e-01 2.41029844e-01 6.36883855e-01
4.96544272e-01 1.15226173e+00 -2.63570309e-01 -1.71296686e-01
3.05757940e-01 -5.08693278e-01 5.25096297e-01 -4.34195325e-02
-4.48884398e-01 3.10849547e-01 -3.83309424e-01 -7.93166637e-01
-9.87903297e-01 -1.32716060e-01 -8.81430089e-01 -1.82119235e-01
-4.99929041e-01 -4.30306673e-01 8.23137522e-01 6.46964729e-01
3.39760900e-01 7.84150124e-01 1.00516641e+00 -1.06094515e+00
-2.45126948e-01 -6.44085824e-01 -6.36668742e-01 2.13348269e-01
5.31044841e-01 -5.53281486e-01 -4.21938598e-01 1.82404533e-01]
|
[10.752089500427246, -2.4559710025787354]
|
28f6a275-51a1-4099-9a31-5acd31e40887
|
improving-back-translation-with-uncertainty
|
1909.00157
| null |
https://arxiv.org/abs/1909.00157v1
|
https://arxiv.org/pdf/1909.00157v1.pdf
|
Improving Back-Translation with Uncertainty-based Confidence Estimation
|
While back-translation is simple and effective in exploiting abundant monolingual corpora to improve low-resource neural machine translation (NMT), the synthetic bilingual corpora generated by NMT models trained on limited authentic bilingual data are inevitably noisy. In this work, we propose to quantify the confidence of NMT model predictions based on model uncertainty. With word- and sentence-level confidence measures based on uncertainty, it is possible for back-translation to better cope with noise in synthetic bilingual corpora. Experiments on Chinese-English and English-German translation tasks show that uncertainty-based confidence estimation significantly improves the performance of back-translation.
|
['Maosong Sun', 'Chao Wang', 'Yang Liu', 'Shuo Wang', 'Huanbo Luan']
|
2019-08-31
|
improving-back-translation-with-uncertainty-1
|
https://aclanthology.org/D19-1073
|
https://aclanthology.org/D19-1073.pdf
|
ijcnlp-2019-11
|
['low-resource-neural-machine-translation']
|
['natural-language-processing']
|
[ 1.30698606e-01 2.69891769e-01 -4.08922553e-01 -5.01120031e-01
-1.64431465e+00 -7.12988019e-01 6.91260517e-01 -2.48691335e-01
-4.74706441e-01 1.45783651e+00 1.80691898e-01 -7.26044774e-01
6.31608248e-01 -4.37323660e-01 -1.10122168e+00 -2.64064111e-02
6.00991070e-01 9.33091640e-01 -5.07589757e-01 -2.46515676e-01
-1.71101943e-01 -9.13832784e-02 -7.28044450e-01 5.17921031e-01
1.48274338e+00 5.82504153e-01 1.79940179e-01 2.52712250e-01
-1.20167084e-01 1.35338202e-01 -5.39088786e-01 -1.08673072e+00
5.20112395e-01 -7.11276233e-01 -5.54886222e-01 -3.33979428e-01
2.65471160e-01 -3.86924745e-04 -9.94204581e-02 1.41349173e+00
4.35172439e-01 -3.46175730e-01 6.65707767e-01 -7.29083061e-01
-8.39640677e-01 1.35178578e+00 -2.08867103e-01 5.27100861e-02
2.35710919e-01 2.19706416e-01 9.76007402e-01 -1.63615501e+00
7.58988261e-01 1.24233842e+00 6.87005222e-01 8.00506949e-01
-1.45543981e+00 -7.73241103e-01 -9.70315561e-02 -6.35397807e-02
-1.38307047e+00 -6.56344175e-01 2.44396120e-01 -1.58374801e-01
1.05261910e+00 1.53441012e-01 3.90110463e-01 1.62666094e+00
5.30941606e-01 8.28536808e-01 1.46373463e+00 -6.54087543e-01
1.56048927e-02 3.98742229e-01 -4.49924797e-01 3.39894652e-01
3.32580209e-01 4.14074868e-01 -8.34927857e-01 -6.74620131e-03
4.23505217e-01 -7.66220450e-01 -1.90939054e-01 1.31985202e-01
-1.65361285e+00 7.56448984e-01 -1.24562442e-01 2.54473448e-01
-3.82606745e-01 -2.75090244e-02 4.69951838e-01 7.16167331e-01
9.45921481e-01 6.96978152e-01 -7.29858100e-01 -4.66740102e-01
-1.33143806e+00 -1.01825424e-01 7.25742042e-01 1.38648522e+00
5.31884253e-01 3.17505270e-01 -3.42043430e-01 9.56368864e-01
3.41872685e-02 1.08469534e+00 6.41299009e-01 -5.79280317e-01
9.28511202e-01 6.96435496e-02 1.57149893e-03 -3.58012319e-01
2.33487666e-01 -7.05433249e-01 -8.27587783e-01 -2.94523150e-01
2.48097092e-01 -4.59099114e-01 -8.93871844e-01 1.96340966e+00
-2.15455815e-01 -8.75491500e-02 5.00055790e-01 7.79754043e-01
2.48446375e-01 7.44524956e-01 -1.54431015e-01 -6.16309822e-01
7.76439965e-01 -1.01205945e+00 -1.00035036e+00 -6.06674910e-01
6.35141134e-01 -1.13915503e+00 1.13234746e+00 7.44901448e-02
-1.34462476e+00 -4.73922610e-01 -9.68528330e-01 1.95744053e-01
-8.46229419e-02 3.11277092e-01 2.48599395e-01 5.84021688e-01
-7.59198725e-01 7.42147326e-01 -7.45093405e-01 -3.51085477e-02
2.25228980e-01 1.83894783e-01 -5.53564668e-01 -2.67647088e-01
-1.72536111e+00 1.42823696e+00 5.76123238e-01 2.53493756e-01
-7.68876314e-01 -5.34424305e-01 -7.37746358e-01 -3.53942364e-01
5.01496866e-02 -7.54585862e-01 1.47035170e+00 -1.09998631e+00
-1.77936399e+00 7.68012226e-01 -2.33592659e-01 -7.13678241e-01
1.09310794e+00 -2.58708000e-01 -3.23105007e-01 -3.30800116e-01
3.47160041e-01 7.04017043e-01 6.13538563e-01 -9.41167712e-01
-4.33050066e-01 -8.94372985e-02 -7.20521390e-01 4.46257830e-01
-6.40873536e-02 1.95316672e-01 -4.03883308e-01 -8.43121111e-01
1.43316567e-01 -1.12355149e+00 -3.58864874e-01 -6.00939631e-01
-4.13342834e-01 1.53218478e-01 1.08335204e-01 -1.04694641e+00
8.94148827e-01 -1.64549875e+00 3.17978561e-01 4.48432006e-02
-5.07890701e-01 3.30687165e-01 -2.97003716e-01 3.51219237e-01
3.44969839e-01 4.05289978e-01 -3.58338565e-01 -4.08126354e-01
-4.65335026e-02 3.77490610e-01 -4.87799525e-01 8.90709460e-02
6.03515744e-01 1.27073491e+00 -8.64469051e-01 -6.12134457e-01
-1.48543743e-02 2.75507778e-01 -2.73847789e-01 7.50015816e-03
-4.60179031e-01 7.54776597e-01 -2.03024998e-01 6.90388322e-01
4.28877980e-01 1.25086203e-01 3.39838058e-01 1.69120893e-01
1.29747525e-01 6.39522731e-01 -4.64822501e-01 1.84524524e+00
-8.58754158e-01 7.76338398e-01 -4.14540291e-01 -3.93318534e-01
1.02990437e+00 5.22564590e-01 -1.60347924e-01 -6.30319953e-01
1.63845494e-01 9.40914512e-01 1.82193264e-01 8.46131444e-02
6.49771869e-01 -4.42855090e-01 -1.62654728e-01 3.41504097e-01
3.16438466e-01 -4.99453902e-01 2.00581282e-01 -1.12063199e-01
5.41393459e-01 4.37476367e-01 9.46716890e-02 -4.04439121e-01
1.60710603e-01 2.97065496e-01 8.30150485e-01 5.09504914e-01
-1.24419983e-02 7.55926609e-01 1.75140686e-02 -1.48414522e-01
-1.47228527e+00 -1.19495773e+00 -2.68980414e-01 4.22130615e-01
-3.20414990e-01 -4.20333594e-01 -9.09452796e-01 -7.10048139e-01
-2.71916509e-01 1.22353613e+00 -2.33523384e-01 -8.77015516e-02
-7.04077065e-01 -9.25664306e-01 8.01657021e-01 4.31635618e-01
2.69805521e-01 -5.98890007e-01 3.67837697e-01 5.64724028e-01
-8.95600855e-01 -1.37815261e+00 -7.83814192e-01 1.79808840e-01
-1.11991501e+00 -3.53961706e-01 -7.95977354e-01 -6.23361111e-01
5.37714005e-01 -1.25982434e-01 1.39441645e+00 -5.06537378e-01
4.35106188e-01 -4.31853950e-01 -1.95829138e-01 -4.51809734e-01
-1.22478199e+00 2.02967405e-01 7.44257629e-01 -3.88313532e-01
5.32440901e-01 -4.29494053e-01 5.84794097e-02 4.80532199e-01
-4.76202726e-01 4.47250009e-01 8.85233819e-01 1.35541666e+00
9.60068822e-01 -5.15133202e-01 6.86897933e-01 -6.52869582e-01
9.34484839e-01 -2.49246702e-01 -4.83031482e-01 6.71433151e-01
-1.09028220e+00 3.82252663e-01 6.25624418e-01 -8.83109510e-01
-1.07161665e+00 -3.03034842e-01 -8.49615261e-02 -6.00974858e-01
4.98123020e-01 7.34397590e-01 -7.61368722e-02 1.76459521e-01
9.88995433e-01 4.39856797e-01 -2.76420057e-01 -3.61956000e-01
4.49650019e-01 8.44769180e-01 5.05433857e-01 -9.41561580e-01
7.23856032e-01 -2.54204720e-01 -3.17549974e-01 -2.06160873e-01
-7.11038351e-01 3.15736502e-01 -6.37896836e-01 -1.41748646e-02
4.83135551e-01 -1.29158843e+00 1.99228823e-01 1.51388198e-01
-1.36872363e+00 -2.32681826e-01 -1.60448119e-01 1.01810920e+00
-6.49857640e-01 8.44696760e-02 -8.77320528e-01 -6.25637233e-01
-7.47128785e-01 -1.43884408e+00 9.32766914e-01 -2.16037422e-01
-4.32013869e-01 -8.58235359e-01 1.15192272e-01 4.42002565e-01
5.03957510e-01 -2.14501858e-01 7.96385884e-01 -7.52905130e-01
-4.41078693e-01 -3.67705375e-01 6.61086142e-02 8.29563498e-01
1.14522632e-02 -3.02026272e-02 -7.10879922e-01 -2.44027346e-01
1.08706333e-01 -3.77589405e-01 6.80845320e-01 1.50805056e-01
2.35698089e-01 -3.78036678e-01 -1.12360537e-01 4.53308851e-01
1.15772390e+00 -2.37108484e-01 3.63417536e-01 1.31899446e-01
4.43144530e-01 4.72865850e-01 9.34352517e-01 -1.21011585e-01
1.76450759e-01 6.11760974e-01 -2.69939631e-01 2.25497097e-01
-1.99700207e-01 -7.80642807e-01 7.73846924e-01 1.60695422e+00
8.55925679e-02 -1.07495889e-01 -1.06094575e+00 4.11619753e-01
-1.86995125e+00 -6.71827555e-01 9.34898388e-04 2.13463044e+00
1.54570949e+00 4.41568464e-01 -5.07869124e-01 -3.76846373e-01
9.20895398e-01 -3.48373801e-01 -4.31930095e-01 -5.20036399e-01
-6.57472908e-01 1.01751834e-01 7.01127470e-01 6.04517221e-01
-4.81193781e-01 1.51941669e+00 7.00581217e+00 1.19769418e+00
-9.63376701e-01 4.44678545e-01 8.86998653e-01 -5.76169789e-02
-6.86108768e-01 1.25732243e-01 -8.67917418e-01 4.23471570e-01
1.31092501e+00 -5.62612116e-01 6.00799680e-01 6.87329948e-01
1.85775235e-01 2.53412694e-01 -1.31993997e+00 7.64577508e-01
-1.21299349e-01 -1.32447934e+00 2.46181190e-01 4.37227227e-02
1.34281337e+00 4.78909820e-01 1.17766142e-01 5.51243305e-01
4.11484063e-01 -9.47352171e-01 1.01232111e+00 3.84926498e-01
1.26426435e+00 -7.90702641e-01 9.42309678e-01 6.95551097e-01
-4.75063682e-01 5.20084739e-01 -6.16286278e-01 1.69846192e-01
1.89506680e-01 7.90275037e-01 -1.28656137e+00 6.02281272e-01
1.62597626e-01 3.52553159e-01 -2.35442519e-01 3.46916080e-01
-6.27467275e-01 8.14020872e-01 -3.35390061e-01 3.55618447e-02
2.08244786e-01 -4.82791603e-01 6.10141575e-01 1.31457520e+00
7.36680567e-01 -4.64405894e-01 -2.11665004e-01 1.10107780e+00
-5.33760607e-01 3.45859945e-01 -7.05090821e-01 -3.88755918e-01
4.74377513e-01 7.26786137e-01 -9.65808704e-02 -4.62540030e-01
-1.99834391e-01 1.28596783e+00 4.23695445e-01 3.32745314e-01
-7.26754665e-01 1.05233841e-01 6.18463397e-01 -2.81676263e-01
-2.83700615e-01 -3.78484130e-01 -6.27280653e-01 -1.68514776e+00
2.87491679e-01 -1.39308894e+00 -3.03709179e-01 -6.91318691e-01
-1.37704921e+00 1.22480321e+00 -2.59800434e-01 -1.41013908e+00
-9.10949469e-01 -5.23934782e-01 -9.10918191e-02 1.40842974e+00
-1.11013472e+00 -1.38766897e+00 8.71306181e-01 9.81674790e-02
8.85662258e-01 -4.71700490e-01 1.08410347e+00 1.04449920e-01
-3.87169719e-01 9.52441812e-01 5.38880348e-01 1.02754794e-01
9.33838248e-01 -9.69776690e-01 1.09051895e+00 1.13724899e+00
5.13942420e-01 7.26877630e-01 9.46247458e-01 -1.07150924e+00
-1.16781497e+00 -1.15334868e+00 1.44277561e+00 -7.38133430e-01
6.73625886e-01 -4.45208937e-01 -6.87382221e-01 6.53205276e-01
2.47105151e-01 -1.41247541e-01 6.21582866e-01 1.28701121e-01
-4.84467149e-01 2.54537612e-01 -1.08509028e+00 9.33489382e-01
9.35218513e-01 -9.56311166e-01 -7.89701045e-01 5.81551373e-01
9.42859113e-01 -4.95268911e-01 -9.61874187e-01 6.27733111e-01
4.69075382e-01 -2.53292382e-01 5.36600590e-01 -7.91340709e-01
4.03174818e-01 -9.94333476e-02 -5.79478264e-01 -1.96716213e+00
2.29201093e-01 -8.57992589e-01 8.46835598e-02 1.01428139e+00
1.25096011e+00 -4.93191838e-01 4.62058783e-01 5.59417844e-01
-1.94511607e-01 -5.03587723e-01 -1.33618581e+00 -1.16805458e+00
7.50208855e-01 -6.90772414e-01 5.68931103e-01 9.33406174e-01
2.29015723e-01 5.57369649e-01 -6.51571214e-01 -3.80914919e-02
6.55505121e-01 -1.22672468e-01 4.55228746e-01 -9.12724555e-01
-4.01152402e-01 -2.65188664e-01 7.95646682e-02 -8.37574840e-01
4.96972203e-01 -1.16306496e+00 2.66064912e-01 -1.29945827e+00
1.70286566e-01 -1.91179901e-01 -4.66558076e-02 2.28539780e-01
-3.42568904e-01 4.75169063e-01 8.05915669e-02 5.10229111e-01
-1.11495636e-01 6.89781964e-01 1.34977901e+00 -2.23531559e-01
3.91794741e-02 6.88105449e-02 -3.29864144e-01 4.13851112e-01
9.66134012e-01 -8.00447762e-01 -5.14931642e-02 -7.70356894e-01
4.06161189e-01 3.49991828e-01 -3.56858790e-01 -7.17681110e-01
7.75782391e-03 -2.51929462e-01 2.63209641e-01 -3.86666626e-01
2.84217358e-01 -4.70901012e-01 3.17755580e-01 3.26258361e-01
-6.19327247e-01 2.65206784e-01 3.28232348e-01 4.73719716e-01
-4.03889537e-01 -1.81014568e-01 6.28472149e-01 -3.13950717e-01
8.66893157e-02 4.56348062e-02 -3.68182063e-01 1.07429855e-01
4.30234194e-01 -2.12218845e-03 -4.07128148e-02 -3.64506692e-01
-4.25527215e-01 1.55339062e-01 6.61709547e-01 6.96965516e-01
5.35295665e-01 -1.67897558e+00 -1.40145946e+00 1.78501353e-01
2.68324256e-01 -3.57539356e-01 -3.03568155e-01 7.18313575e-01
-2.18281537e-01 6.75481737e-01 -2.23774865e-01 -5.06348073e-01
-8.87831867e-01 2.47591868e-01 4.31641668e-01 -3.86815280e-01
-3.91585454e-02 7.19508469e-01 -3.89193088e-01 -8.20620775e-01
-1.15501605e-01 -1.57279193e-01 5.96237302e-01 -5.19128263e-01
2.74349481e-01 -8.53784382e-03 2.10202023e-01 -7.66004980e-01
-9.11119357e-02 1.33875519e-01 -1.07717983e-01 -9.62300897e-01
8.00902128e-01 -2.68672019e-01 2.30051540e-02 5.51033795e-01
8.78098905e-01 1.77847475e-01 -9.27017868e-01 -6.64739072e-01
2.53355861e-01 -4.45814818e-01 -9.07503814e-02 -1.22214115e+00
-4.93362308e-01 1.07031918e+00 3.09082329e-01 -6.10413432e-01
6.69674158e-01 -4.59385514e-01 8.65619481e-01 7.10108399e-01
9.48254883e-01 -1.36845994e+00 -4.13749367e-01 8.04231703e-01
9.15375590e-01 -1.60827768e+00 -3.51025850e-01 -1.24705583e-01
-7.73192286e-01 1.05802321e+00 4.63835090e-01 4.52045649e-01
1.43475369e-01 2.88122565e-01 3.63367438e-01 6.19683862e-01
-1.02056599e+00 1.55733824e-01 4.56412494e-01 4.87727046e-01
5.66844046e-01 4.39005733e-01 -5.95453262e-01 6.20818496e-01
-3.54134947e-01 -1.20272674e-02 3.50054353e-01 4.11072284e-01
-1.98764071e-01 -1.59548640e+00 -2.98341781e-01 2.70663559e-01
-7.20908940e-01 -8.29967558e-01 -5.62738299e-01 4.24460143e-01
-2.17349544e-01 1.09210181e+00 -3.38152826e-01 -6.47146583e-01
9.02129337e-02 4.05935079e-01 6.00339234e-01 -6.50412321e-01
-5.91526747e-01 3.21071297e-02 5.70511281e-01 -1.84546784e-01
-3.72738764e-02 -6.18551254e-01 -9.43008304e-01 -2.56368876e-01
-6.14680648e-01 5.46601295e-01 1.14744639e+00 9.86289263e-01
2.57552415e-01 6.08937852e-02 5.13988256e-01 -4.14102465e-01
-1.13048983e+00 -1.48010242e+00 2.24634707e-02 1.01270534e-01
-1.60564691e-01 -8.23576599e-02 -2.85150141e-01 -1.02387279e-01]
|
[11.625262260437012, 10.326958656311035]
|
a38cedf7-b247-43dd-977a-129aa1c889d6
|
efficient-learning-of-high-level-plans-from
|
2303.09628
| null |
https://arxiv.org/abs/2303.09628v1
|
https://arxiv.org/pdf/2303.09628v1.pdf
|
Efficient Learning of High Level Plans from Play
|
Real-world robotic manipulation tasks remain an elusive challenge, since they involve both fine-grained environment interaction, as well as the ability to plan for long-horizon goals. Although deep reinforcement learning (RL) methods have shown encouraging results when planning end-to-end in high-dimensional environments, they remain fundamentally limited by poor sample efficiency due to inefficient exploration, and by the complexity of credit assignment over long horizons. In this work, we present Efficient Learning of High-Level Plans from Play (ELF-P), a framework for robotic learning that bridges motion planning and deep RL to achieve long-horizon complex manipulation tasks. We leverage task-agnostic play data to learn a discrete behavioral prior over object-centric primitives, modeling their feasibility given the current context. We then design a high-level goal-conditioned policy which (1) uses primitives as building blocks to scaffold complex long-horizon tasks and (2) leverages the behavioral prior to accelerate learning. We demonstrate that ELF-P has significantly better sample efficiency than relevant baselines over multiple realistic manipulation tasks and learns policies that can be easily transferred to physical hardware.
|
['Stelian Coros', 'Georg Martius', 'Otmar Hilliges', 'Marco Bagatella', 'Núria Armengol Urpí']
|
2023-03-16
| null | null | null | null |
['motion-planning']
|
['robots']
|
[-5.27376309e-03 1.78536788e-01 -4.50842381e-01 -1.22109741e-01
-1.05699551e+00 -5.42603970e-01 7.44122148e-01 1.17279992e-01
-5.55310905e-01 8.36997330e-01 4.89155173e-01 -3.24343652e-01
-3.08767349e-01 -6.61191463e-01 -1.13203216e+00 -3.16802323e-01
-7.30388284e-01 7.98844755e-01 5.20805240e-01 -3.46154153e-01
3.79796147e-01 5.79174578e-01 -1.51306009e+00 2.05693036e-01
6.48206949e-01 7.80535936e-01 6.45373285e-01 7.87134349e-01
3.60478431e-01 1.12896776e+00 -1.30989984e-01 5.72542131e-01
4.79469478e-01 4.26755063e-02 -9.83174324e-01 -1.19468823e-01
1.45300478e-01 -8.78844202e-01 -6.90803826e-01 5.23663163e-01
4.10366952e-01 6.84508562e-01 5.79021633e-01 -1.21963847e+00
-2.15221685e-03 6.36130035e-01 -3.55982840e-01 -4.02349979e-02
2.74172455e-01 9.62943673e-01 9.62155104e-01 -5.09650826e-01
7.78979540e-01 1.65039277e+00 3.73692721e-01 4.30030853e-01
-1.13610637e+00 -4.00797188e-01 4.82357115e-01 6.17001429e-02
-6.35548532e-01 -2.20311284e-01 4.24720526e-01 -5.10493517e-01
1.27524078e+00 -3.79203826e-01 7.90527225e-01 1.12531018e+00
5.07626355e-01 1.12666285e+00 9.64042783e-01 2.47866232e-02
5.06297827e-01 -9.35624957e-01 -2.44661719e-01 9.54155087e-01
-9.15960371e-02 7.48906672e-01 -4.26979631e-01 -7.97881559e-02
1.11416006e+00 -3.52940634e-02 2.26188134e-02 -7.51958549e-01
-1.49523687e+00 6.76792264e-01 7.11456537e-01 -2.98773170e-01
-6.51474714e-01 1.00089014e+00 3.35289568e-01 8.61728713e-02
-3.19270223e-01 9.66853440e-01 -6.14835143e-01 -5.28858364e-01
-4.54894066e-01 9.16983187e-01 7.71927059e-01 1.21173859e+00
7.93299794e-01 8.19713399e-02 -4.92159933e-01 3.99536103e-01
1.60072953e-01 3.51426691e-01 1.01274163e-01 -1.55322874e+00
6.04496062e-01 1.21327318e-01 5.63839853e-01 -4.58152473e-01
-6.91238761e-01 -2.38453731e-01 -2.21545771e-01 8.61120701e-01
3.55716139e-01 -2.65781105e-01 -1.12329972e+00 1.88374949e+00
3.88133556e-01 3.94839570e-02 -2.84249093e-02 8.24910283e-01
1.82578534e-01 6.50564611e-01 2.71065116e-01 3.30585361e-01
9.95332420e-01 -1.48671257e+00 -7.29860812e-02 -6.16611302e-01
5.68013787e-01 -2.57386506e-01 1.39755285e+00 4.22114819e-01
-1.33560812e+00 -4.55281258e-01 -1.08542705e+00 -2.83866048e-01
7.82989413e-02 -3.67731526e-02 1.02896869e+00 -1.37300476e-01
-9.31248665e-01 1.00630331e+00 -1.44602799e+00 -7.40471408e-02
7.59135604e-01 5.07223129e-01 -2.23819062e-01 -1.70055673e-01
-5.50894439e-01 1.08882070e+00 5.58529198e-01 -1.84044614e-01
-1.98614705e+00 -5.64627290e-01 -9.27242279e-01 2.18253165e-01
9.67436790e-01 -7.36320317e-01 1.82141590e+00 -7.56381229e-02
-1.91206253e+00 1.93196833e-01 3.58576238e-01 -5.82701325e-01
5.05136013e-01 -6.35677338e-01 5.50973475e-01 9.77999195e-02
2.26370871e-01 1.09967613e+00 6.48118436e-01 -1.16185260e+00
-8.39889050e-01 3.40419821e-02 7.02862740e-01 4.51787233e-01
3.03011209e-01 -5.39617956e-01 -3.80107522e-01 -5.02236784e-01
9.30916169e-04 -1.21471035e+00 -7.77402937e-01 2.92718410e-01
-1.24173254e-01 -3.57437104e-01 5.60401022e-01 -3.81605446e-01
5.89246869e-01 -1.84416306e+00 4.92219061e-01 -6.37163222e-02
-4.17739861e-02 -1.11438684e-01 -4.64713871e-01 6.56565070e-01
5.14452279e-01 -3.35303694e-01 -1.50189981e-01 -2.42020383e-01
3.40627402e-01 5.61515272e-01 -4.78956729e-01 2.75785595e-01
2.13590294e-01 1.21663332e+00 -1.49842680e+00 -2.70866036e-01
2.65429139e-01 -1.19580470e-01 -1.05276299e+00 3.43764067e-01
-1.06550539e+00 6.23964906e-01 -7.50104368e-01 5.85544586e-01
3.79284322e-02 -1.03291452e-01 1.44174919e-01 1.71066925e-01
-1.68837577e-01 5.34453332e-01 -9.31907892e-01 2.32575035e+00
-6.25465214e-01 2.57319897e-01 2.64076471e-01 -7.49581158e-01
5.51562965e-01 -5.47736771e-02 5.94596028e-01 -5.63869774e-01
6.62211608e-03 2.08262026e-01 -9.03008506e-02 -4.85865474e-01
7.23479450e-01 3.78317274e-02 -4.69550610e-01 4.58042324e-01
1.08646475e-01 -1.00324464e+00 1.79531187e-01 -1.23517243e-02
1.51331902e+00 9.11670089e-01 3.27561408e-01 -2.54659712e-01
-6.70556054e-02 4.31669623e-01 5.82471430e-01 1.04467237e+00
-1.84433654e-01 1.16859384e-01 5.64844549e-01 -3.96853209e-01
-1.16380572e+00 -1.25795972e+00 5.28052688e-01 1.25201797e+00
3.86641830e-01 -1.96673408e-01 -2.78349042e-01 -4.59676683e-01
2.34624296e-01 6.93483829e-01 -4.02930588e-01 -2.24938661e-01
-1.05138838e+00 -9.96256545e-02 3.77400517e-01 7.48181045e-01
3.40755403e-01 -1.41671371e+00 -1.18443322e+00 5.59798837e-01
8.47192332e-02 -1.07783556e+00 -4.26748574e-01 4.22108889e-01
-9.68821347e-01 -1.21358204e+00 -3.71819645e-01 -8.44158769e-01
3.42931032e-01 2.48844102e-01 1.10005260e+00 -9.14076269e-02
-3.29678953e-01 6.04347289e-01 -2.43324459e-01 -3.98784429e-01
-3.68253797e-01 7.85110742e-02 7.75815919e-02 -9.28794682e-01
-4.79860842e-01 -7.31441438e-01 -6.81755304e-01 9.38297436e-02
-5.51791847e-01 2.18419686e-01 9.59511876e-01 1.02163434e+00
6.66247904e-01 -3.38901766e-02 4.09684688e-01 -4.06186283e-01
5.81800461e-01 -3.43660027e-01 -9.82871354e-01 -5.08700348e-02
-3.05681795e-01 5.24736226e-01 3.73956650e-01 -5.69071710e-01
-9.62220848e-01 2.37939581e-01 -1.79860368e-02 -3.13748181e-01
4.96554673e-02 5.24320185e-01 2.11630151e-01 -3.94603685e-02
7.43572414e-01 -1.51727021e-01 -2.17131469e-02 -2.83358127e-01
5.88259459e-01 -2.81840395e-02 7.80429423e-01 -1.40319240e+00
6.23226523e-01 4.17805403e-01 2.66185671e-01 -4.19975817e-01
-7.14725494e-01 -1.98139653e-01 -3.81834298e-01 -1.26946852e-01
8.30309689e-01 -1.05484128e+00 -1.12389135e+00 3.33550096e-01
-1.02598655e+00 -1.43566668e+00 -6.00342035e-01 6.92846000e-01
-1.45222008e+00 9.81833786e-02 -7.81472266e-01 -7.28607595e-01
-5.54511398e-02 -1.38711405e+00 1.28007865e+00 3.56625230e-03
-1.61577702e-01 -4.50246215e-01 3.74889299e-02 -9.69910473e-02
2.15747938e-01 4.92203593e-01 1.00521612e+00 1.03928164e-01
-1.23813605e+00 1.35007352e-01 1.51095241e-02 -1.11842602e-01
-2.33797476e-01 -4.79986131e-01 -4.19681638e-01 -5.53892732e-01
-3.64648849e-01 -1.06756294e+00 7.73918688e-01 3.39352041e-01
1.26384795e+00 -2.25803331e-01 -4.45093364e-01 4.95852530e-01
1.16658866e+00 1.96403358e-02 4.18298095e-01 4.39697444e-01
4.08311456e-01 4.50948507e-01 1.15881670e+00 6.47860110e-01
4.52521861e-01 7.46300995e-01 8.28971028e-01 4.10716116e-01
-7.31783882e-02 -6.94534242e-01 4.22181100e-01 2.06287384e-01
-2.10921280e-02 2.58211922e-02 -9.54014063e-01 5.83162785e-01
-2.16573071e+00 -9.37293053e-01 4.44653660e-01 1.86084962e+00
8.32790375e-01 5.61494410e-01 2.75286347e-01 -4.01975274e-01
-6.69426993e-02 1.98880166e-01 -1.04783475e+00 -6.45776540e-02
3.50414604e-01 3.18749279e-01 6.05140209e-01 7.72597253e-01
-1.09111667e+00 1.36213970e+00 6.28014088e+00 8.02402616e-01
-9.35333252e-01 3.09261754e-02 2.54141204e-02 -3.93219113e-01
-9.09445360e-02 2.04612315e-01 -5.79727888e-01 -9.02294666e-02
3.88666928e-01 -7.17299804e-02 7.94775367e-01 1.23149908e+00
2.16554686e-01 -4.48522747e-01 -1.55626202e+00 6.49800301e-01
-6.47577882e-01 -1.43591917e+00 -2.74741709e-01 1.66597679e-01
7.01157928e-01 5.39712071e-01 -2.52725221e-02 8.85830045e-01
1.14136457e+00 -1.09443319e+00 1.03136253e+00 3.23432684e-01
6.01261139e-01 -5.74995816e-01 5.18772677e-02 8.23284984e-01
-1.16528523e+00 -6.09441757e-01 -4.08660531e-01 -4.84613627e-01
4.03987676e-01 -7.57496729e-02 -9.73737776e-01 2.08467066e-01
4.26521719e-01 5.93351960e-01 1.71757415e-01 9.87314999e-01
-4.66002941e-01 1.59844264e-01 -5.39770782e-01 -2.39509121e-01
8.15790474e-01 1.27844557e-01 4.44758892e-01 7.13361025e-01
1.35166198e-01 2.53315210e-01 8.82010996e-01 7.72712588e-01
1.64336562e-01 -4.99316186e-01 -5.13382137e-01 -2.33159453e-01
5.34741461e-01 8.38800311e-01 -6.06056511e-01 -1.89426228e-01
-1.61415767e-02 6.95385933e-01 8.24797153e-01 2.97829092e-01
-7.13442326e-01 1.04083434e-01 9.37511325e-01 -5.14955260e-02
3.82705152e-01 -9.58499134e-01 -1.98538944e-01 -8.49250674e-01
4.62408476e-02 -7.31800437e-01 1.09204851e-01 -6.88607752e-01
-8.22556674e-01 1.18224367e-01 1.80927321e-01 -1.01458037e+00
-4.56788450e-01 -6.76366091e-01 -3.83368284e-01 5.67315280e-01
-1.57231152e+00 -1.15517914e+00 -2.95819372e-01 4.57305968e-01
8.05687606e-01 -2.19026580e-02 7.76092410e-01 -2.26860031e-01
4.18648645e-02 1.73420370e-01 -1.61346272e-01 -2.05751657e-01
3.45169961e-01 -1.22352636e+00 5.80223739e-01 5.00823736e-01
-1.98427990e-01 2.53740728e-01 6.94381237e-01 -7.59310544e-01
-1.75918317e+00 -8.93475235e-01 2.08618632e-03 -5.51167965e-01
6.10427439e-01 -3.47538143e-01 -4.88760561e-01 8.37407231e-01
-2.01636389e-01 3.11347656e-02 -9.00281742e-02 2.14486197e-01
-2.08878681e-01 2.26777703e-01 -9.81040776e-01 1.03653026e+00
1.35424459e+00 -3.46588671e-01 -7.66487837e-01 4.96065199e-01
9.72733855e-01 -9.63739395e-01 -7.60047674e-01 5.66487610e-01
7.91301250e-01 -6.56275451e-01 1.10284472e+00 -8.67352962e-01
6.54708862e-01 -1.82569072e-01 -2.04127818e-01 -1.16828561e+00
-4.54170823e-01 -9.03126240e-01 -3.63609225e-01 2.08112538e-01
1.44260213e-01 -1.25009418e-01 1.00277174e+00 4.21765625e-01
-5.46074390e-01 -1.16624892e+00 -7.28543878e-01 -9.23574865e-01
2.18547806e-01 -5.07869780e-01 5.70227444e-01 2.95663625e-01
1.90002888e-01 2.28662565e-01 -4.35284168e-01 2.72132814e-01
5.03334761e-01 2.54853338e-01 1.21862257e+00 -8.01615655e-01
-7.61635363e-01 -6.25402451e-01 -3.51967849e-02 -1.71004343e+00
2.60094076e-01 -6.60697043e-01 8.31227660e-01 -1.70802593e+00
-1.58974856e-01 -9.29242432e-01 5.26621938e-02 5.49789071e-01
9.26191360e-03 -5.54065764e-01 4.58985835e-01 2.43810445e-01
-7.30332196e-01 9.82051313e-01 1.71097791e+00 -7.48835430e-02
-4.14305329e-01 4.91752662e-03 -2.16664165e-01 7.20457673e-01
6.81715429e-01 -2.57791162e-01 -6.76680923e-01 -6.65551901e-01
8.04821327e-02 6.70286059e-01 3.34858119e-01 -1.27343380e+00
2.77027011e-01 -7.44635642e-01 1.60718933e-02 -6.49285793e-01
6.92891955e-01 -5.75464725e-01 -3.74876946e-01 8.98519933e-01
-5.57585835e-01 5.23445234e-02 2.60901213e-01 9.74243701e-01
1.99880347e-01 -4.16094549e-02 6.28565788e-01 -4.57160324e-01
-1.03935397e+00 7.54163086e-01 -4.63287771e-01 2.18636587e-01
1.12947977e+00 9.13769305e-02 -1.60923302e-01 -3.10643524e-01
-7.25995004e-01 8.20838571e-01 5.43942332e-01 4.78008479e-01
5.06305456e-01 -1.12419081e+00 -4.19466913e-01 -2.59442985e-01
2.59623155e-02 6.22727275e-01 9.66378227e-02 5.02253175e-01
-6.83410525e-01 2.94597685e-01 -5.39549649e-01 -4.96062696e-01
-6.87533319e-01 6.81363702e-01 1.50639609e-01 -5.91803670e-01
-1.04226327e+00 9.79397595e-01 3.25867414e-01 -6.15167081e-01
5.56403220e-01 -6.00393534e-01 2.20903143e-01 -6.40742362e-01
2.05899522e-01 3.05245191e-01 -5.62757134e-01 1.50020704e-01
-7.09827468e-02 2.22085387e-01 1.18464297e-02 -4.89309520e-01
1.47497511e+00 2.93582559e-01 2.47940034e-01 2.12720603e-01
6.56866908e-01 -4.42226261e-01 -2.33257985e+00 -2.07565337e-01
1.41803563e-01 -4.16043818e-01 -1.23948120e-01 -7.51338422e-01
-3.37535769e-01 8.84382546e-01 2.66795009e-02 -5.86430252e-01
5.56628227e-01 -2.15742532e-02 9.58752334e-01 1.09458208e+00
1.14320719e+00 -1.36893559e+00 8.02726567e-01 1.04230475e+00
1.11507130e+00 -1.15526414e+00 6.28440455e-02 -2.05019891e-01
-4.77208644e-01 1.04749751e+00 8.70930374e-01 -5.49395263e-01
4.80512947e-01 3.68402034e-01 -5.04433393e-01 -1.63217381e-01
-1.02276480e+00 -3.91321301e-01 -9.96266603e-02 7.21656859e-01
-2.19053835e-01 1.10673897e-01 7.18273818e-02 3.11449349e-01
-2.80876070e-01 2.41423994e-01 4.22349989e-01 1.66030622e+00
-8.86887729e-01 -9.68642473e-01 -1.15996553e-02 4.42437649e-01
1.00709781e-01 3.13254416e-01 1.51917323e-01 8.04754436e-01
-1.46661505e-01 6.04732096e-01 -1.25577435e-01 -2.26355806e-01
3.47027659e-01 -3.13743830e-01 1.04559064e+00 -1.02184081e+00
-3.01896662e-01 -1.40114695e-01 3.15264612e-01 -1.22524214e+00
5.02030253e-02 -7.92508900e-01 -1.62946582e+00 -1.84483245e-01
2.66335785e-01 -1.67056724e-01 6.20738626e-01 1.07680106e+00
4.37542647e-01 7.10386813e-01 1.65870696e-01 -1.62942147e+00
-1.20080304e+00 -7.61350453e-01 -2.19564006e-01 1.58877447e-01
4.74253595e-01 -1.10199845e+00 2.67225862e-01 -4.14965153e-01]
|
[4.51918363571167, 0.9849057793617249]
|
080e8e56-7bf2-4458-8a84-5d544c7ea062
|
shared-logistic-normal-distributions-for-soft
| null | null |
https://aclanthology.org/N09-1009
|
https://aclanthology.org/N09-1009.pdf
|
Shared Logistic Normal Distributions for Soft Parameter Tying in Unsupervised Grammar Induction
|
We present a family of priors over probabilistic grammar weights, called the shared logistic normal distribution. This family extends the partitioned logistic normal distribution, enabling factored covariance between the probabilities of different derivation events in the probabilistic grammar, providing a new way to encode prior knowledge about an unknown grammar. We describe a variational EM algorithm for learning a probabilistic grammar based on this family of priors. We then experiment with unsupervised dependency grammar induction and show significant improvements using our model for both monolingual learning and bilingual learning with a non-parallel, multilingual corpus.
|
['Noah A. Smith', 'Shay Cohen']
|
2009-06-01
| null | null | null | null |
['dependency-grammar-induction', 'unsupervised-dependency-parsing']
|
['natural-language-processing', 'natural-language-processing']
|
[-1.27439380e-01 4.80653912e-01 -8.09957981e-02 -7.28809178e-01
-1.10528231e+00 -6.39980614e-01 7.57457197e-01 -1.59850836e-01
-5.68418980e-01 8.28724980e-01 3.49895239e-01 -5.70455968e-01
9.77219455e-03 -7.06160128e-01 -8.21972549e-01 -8.52532685e-01
-2.13994548e-01 1.20375943e+00 9.20274034e-02 -7.82573447e-02
-1.01993106e-01 5.59388883e-02 -1.01576924e+00 -3.85092199e-02
8.42746139e-01 2.36998014e-02 7.57133603e-01 5.99309385e-01
-8.50131214e-02 2.86084741e-01 -3.32289070e-01 -4.39040989e-01
-1.71268329e-01 -2.05917463e-01 -6.89252377e-01 -2.37572446e-01
-2.61745974e-02 2.56208889e-02 -1.02756031e-01 1.16045880e+00
7.09108487e-02 2.22476453e-01 1.06163013e+00 -1.02231705e+00
-6.79944932e-01 1.43892050e+00 -3.70052457e-01 1.91503644e-01
2.25094453e-01 -4.59674358e-01 1.44928145e+00 -7.12990224e-01
7.81239212e-01 1.68761647e+00 4.22247648e-01 2.80008882e-01
-1.74755466e+00 -6.33722007e-01 3.55180472e-01 1.41226545e-01
-1.52783823e+00 -2.11117312e-01 6.65397406e-01 -3.68248820e-01
1.10985708e+00 -3.94197166e-01 2.55101442e-01 1.48140502e+00
4.57460582e-01 8.58303726e-01 1.18582404e+00 -8.07314098e-01
6.43387139e-02 7.38695040e-02 3.14255595e-01 6.56483054e-01
1.69945300e-01 3.92085165e-01 -5.62709451e-01 -3.85993898e-01
6.28910720e-01 -7.01815009e-01 -6.06246255e-02 -4.05941427e-01
-1.06722510e+00 1.22559726e+00 -4.65908617e-01 2.19750926e-01
-4.30884883e-02 2.63308078e-01 3.03058922e-02 1.94125623e-01
4.68140244e-01 -2.32692301e-01 -8.31094325e-01 -4.14420925e-02
-6.46410882e-01 2.46769935e-01 1.07177949e+00 1.09259403e+00
9.99958634e-01 -3.01219784e-02 1.15013476e-02 9.73312140e-01
8.60831380e-01 1.22047448e+00 3.14705409e-02 -8.23672771e-01
2.74209410e-01 -2.82865882e-01 -7.60040730e-02 -2.86688179e-01
-5.78902483e-01 -1.01477653e-01 -4.28273976e-01 -5.27078807e-02
5.97825408e-01 -4.38077211e-01 -9.11893070e-01 2.35146141e+00
2.11765617e-01 -1.95350610e-02 1.86900780e-01 3.82851422e-01
3.26919675e-01 6.61399901e-01 3.15299064e-01 -3.07688296e-01
1.26670563e+00 -2.72493422e-01 -7.26991117e-01 -2.98233423e-02
4.30085808e-01 -8.72448087e-01 9.43760753e-01 5.76026082e-01
-8.09372902e-01 -2.75393993e-01 -6.53815806e-01 -1.39190644e-01
-1.43847272e-01 -2.05059022e-01 8.09233069e-01 8.29070508e-01
-1.17989969e+00 4.06615734e-01 -1.20895970e+00 -2.63382286e-01
-2.02981949e-01 4.85975981e-01 -3.90230656e-01 -1.38225824e-01
-1.25190413e+00 9.03670430e-01 8.16923738e-01 -2.84258664e-01
-8.16706717e-01 -4.83596861e-01 -1.21749902e+00 -8.25290531e-02
3.87830324e-02 -4.48722482e-01 1.37577689e+00 -3.57389718e-01
-1.54247952e+00 8.32212448e-01 -3.45003277e-01 -2.32121989e-01
-1.53525189e-01 -1.73696466e-02 -3.46166283e-01 -3.55442286e-01
1.88344270e-01 8.28738570e-01 7.70744383e-01 -1.19081604e+00
-6.16189122e-01 -1.99326470e-01 -2.52844483e-01 2.24050820e-01
2.55419284e-01 2.76129425e-01 -3.83067876e-01 -8.10320735e-01
2.84509838e-01 -1.10014272e+00 3.66677009e-02 -1.02968371e+00
-2.97910571e-01 -7.13205695e-01 2.01385438e-01 -8.90513539e-01
8.06990027e-01 -2.02645826e+00 4.88756239e-01 5.33170164e-01
-2.07602859e-01 -7.51819551e-01 -1.35576174e-01 3.31426769e-01
1.79269277e-02 8.99371058e-02 -5.85807681e-01 -4.15785611e-01
4.38218087e-01 9.24606681e-01 -2.69744277e-01 3.22399318e-01
1.60420567e-01 5.69747031e-01 -1.05149698e+00 -5.20419002e-01
-1.20292179e-01 7.19855547e-01 -6.34526134e-01 1.69041768e-01
-4.26575989e-01 5.82386732e-01 -2.61042535e-01 2.67115176e-01
6.33319139e-01 2.64076680e-01 9.88561511e-01 2.97231823e-01
1.66813359e-01 3.44628423e-01 -1.32361650e+00 1.83167100e+00
-4.20579463e-01 2.88186312e-01 1.14054747e-01 -7.11380541e-01
8.33574891e-01 3.84361684e-01 8.80835392e-03 -1.40223354e-01
1.53352275e-01 3.42464000e-02 1.48065105e-01 -2.51133531e-01
2.61882722e-01 -6.25668406e-01 -5.32653570e-01 7.30203271e-01
9.22820270e-01 -2.10554734e-01 3.82362694e-01 2.89993733e-01
6.44492745e-01 7.46789098e-01 1.16854861e-01 -8.32913816e-01
2.05951869e-01 -5.18823206e-01 8.53798211e-01 1.00615716e+00
2.35564947e-01 3.46151501e-01 7.68628895e-01 -1.36955947e-01
-8.60734344e-01 -1.80544269e+00 -6.28341198e-01 1.57975876e+00
-4.30264294e-01 -4.37872916e-01 -5.80911934e-01 -7.00895309e-01
-1.70531660e-01 1.17021739e+00 -5.01131594e-01 2.70658135e-01
-6.88620865e-01 -1.31906497e+00 3.89072448e-01 5.68815231e-01
-2.09832534e-01 -1.11254954e+00 1.71478987e-01 3.79973948e-01
-4.30103064e-01 -1.13388073e+00 -6.11962557e-01 6.51476324e-01
-7.93453038e-01 -1.00425148e+00 -1.09869450e-01 -9.64375794e-01
4.42571223e-01 -6.73812985e-01 1.47792697e+00 -4.99034315e-01
1.86974153e-01 4.54257429e-01 -1.22479707e-01 -3.78491819e-01
-8.55107427e-01 2.81634368e-02 2.74000108e-01 -2.69544154e-01
5.92590451e-01 -7.75729120e-01 2.43010536e-01 -1.41015217e-01
-7.43898630e-01 -1.86770558e-01 4.05455709e-01 1.02264905e+00
6.86892748e-01 -1.94001839e-01 2.13344783e-01 -1.04095280e+00
5.01274645e-01 -6.46128714e-01 -9.17441964e-01 3.71475607e-01
-3.72013241e-01 1.01157558e+00 5.87877184e-02 -4.28262144e-01
-1.60737717e+00 6.04199432e-02 -3.37879628e-01 2.36508995e-01
-3.88019025e-01 5.43165386e-01 -5.03123283e-01 3.25184077e-01
2.50253260e-01 -3.59204486e-02 -3.71514976e-01 -5.79629600e-01
9.02990639e-01 3.36869329e-01 5.98874271e-01 -1.31585324e+00
4.88946557e-01 1.77567616e-01 5.88765889e-02 -6.22049391e-01
-5.43948054e-01 -9.20647532e-02 -9.41657364e-01 1.37104869e-01
1.06221843e+00 -9.56003308e-01 -4.04296935e-01 2.40003005e-01
-1.19533467e+00 -5.31586051e-01 -1.04702167e-01 9.74032104e-01
-7.76804328e-01 4.31015164e-01 -7.30918586e-01 -7.35469282e-01
8.94783884e-02 -1.16078663e+00 1.04835677e+00 -1.32412180e-01
-3.16011101e-01 -1.60637045e+00 7.10964739e-01 -1.54430926e-01
-9.05186906e-02 -9.08516720e-02 1.44941843e+00 -7.39853442e-01
-2.95951694e-01 3.96983087e-01 1.23133443e-01 2.59828120e-01
-5.24188653e-02 7.32064545e-02 -6.93745494e-01 -2.77907342e-01
-6.64991736e-02 -6.18585981e-02 8.68922591e-01 6.12378061e-01
4.57642645e-01 1.38023838e-01 -4.19850886e-01 5.95142305e-01
1.37470543e+00 8.03111792e-02 1.47485331e-01 -1.55291110e-01
6.06199145e-01 5.01873314e-01 1.59305289e-01 1.89928070e-01
8.20320547e-01 3.96782964e-01 -9.61363316e-02 4.01193112e-01
1.99498162e-01 -4.04179156e-01 6.91285074e-01 1.20653176e+00
-4.17347923e-02 -1.52944982e-01 -1.09246969e+00 8.02662432e-01
-1.65167809e+00 -5.82807720e-01 -8.07624906e-02 2.13082695e+00
1.17655551e+00 6.28496930e-02 4.14731614e-02 -3.97329628e-01
8.66344571e-01 -8.40233937e-02 9.48365405e-02 -6.60919726e-01
-4.41716790e-01 7.74077713e-01 5.04842639e-01 1.28169918e+00
-1.09676063e+00 1.31950533e+00 8.28044701e+00 7.33702779e-01
-2.98238903e-01 5.72832286e-01 2.01713890e-02 2.39414528e-01
-7.80318916e-01 4.72163022e-01 -1.12337065e+00 2.34395251e-01
1.27400470e+00 5.49790561e-02 5.89277446e-01 3.18487644e-01
-1.85688302e-01 -1.52192786e-01 -1.11221206e+00 6.54886901e-01
7.86500871e-02 -6.90193474e-01 -3.61240283e-02 1.75602123e-01
7.45429337e-01 5.02470970e-01 -1.62672833e-01 6.24262810e-01
1.54078054e+00 -7.47527182e-01 6.18465185e-01 6.07090235e-01
6.06485367e-01 -1.00969958e+00 5.42954624e-01 3.87666404e-01
-8.82363141e-01 3.00655603e-01 -4.21829104e-01 1.30209804e-01
5.03994644e-01 5.28404891e-01 -5.30005336e-01 3.07466239e-01
4.93282348e-01 2.89546013e-01 -1.20385505e-01 2.54275411e-01
-9.15326655e-01 9.54149008e-01 -6.60761654e-01 2.80045033e-01
-3.35907709e-04 -6.00045443e-01 6.72470748e-01 1.42350221e+00
2.52018780e-01 2.11180672e-02 2.86662996e-01 8.86741638e-01
-1.68146968e-01 1.69529393e-01 -4.92368937e-01 2.09195629e-01
2.68681049e-01 9.46786284e-01 -8.12035501e-01 -1.82540402e-01
-6.21040106e-01 6.26327455e-01 6.95155442e-01 4.74313855e-01
-6.92029953e-01 -1.36979222e-01 4.90641683e-01 -4.57603812e-01
5.25842786e-01 -6.51361704e-01 -1.71522237e-02 -1.28160810e+00
-2.52770841e-01 -6.21064186e-01 4.96977597e-01 -4.26312685e-01
-1.50319386e+00 4.85643148e-01 7.42872119e-01 -2.55054206e-01
-6.30071640e-01 -8.24291885e-01 -3.34960699e-01 1.25707746e+00
-1.33150268e+00 -1.10855234e+00 6.47524059e-01 4.70568120e-01
1.75409988e-01 -1.82370678e-01 1.20957994e+00 -1.83540776e-01
-2.15585411e-01 4.86880481e-01 2.74266094e-01 2.81848479e-02
8.85064781e-01 -1.93459284e+00 6.50185525e-01 7.79370368e-01
7.48642683e-01 6.27595127e-01 7.60783195e-01 -8.33640158e-01
-1.01099133e+00 -7.24248469e-01 1.15726328e+00 -8.54349077e-01
7.84321785e-01 -7.81646192e-01 -8.59514296e-01 1.32451916e+00
5.22356093e-01 -3.75323385e-01 9.20357525e-01 8.14502239e-01
-5.75229108e-01 3.30530912e-01 -9.81132030e-01 3.85323495e-01
8.57326627e-01 -5.73542297e-01 -1.01750338e+00 4.73098755e-01
6.18226647e-01 -2.59405047e-01 -1.16446149e+00 2.60367036e-01
4.88470197e-01 -6.37229204e-01 5.41300893e-01 -4.72853124e-01
-3.02435428e-01 -9.30847079e-02 -5.58436573e-01 -1.59051454e+00
-4.71543282e-01 -8.16746473e-01 -8.39001779e-03 1.13750553e+00
6.87859476e-01 -5.60417831e-01 4.38009948e-01 3.66663605e-01
-1.12773985e-01 -1.14555530e-01 -9.39586580e-01 -7.90191114e-01
8.09864700e-01 -8.90209675e-01 4.37075615e-01 8.13197911e-01
9.01242346e-03 4.77345884e-01 -2.71264136e-01 5.15190601e-01
1.06538439e+00 -1.79396212e-01 1.65498897e-01 -1.39219022e+00
-6.66071773e-01 -2.03228801e-01 -1.46229625e-01 -1.01504660e+00
8.95505548e-01 -1.35579634e+00 4.41243649e-01 -1.21372652e+00
5.11918843e-01 -1.84308037e-01 -3.10391486e-01 5.19832551e-01
-3.01496804e-01 -6.09962381e-02 -2.60316670e-01 -2.55491734e-01
-3.80919993e-01 5.27860701e-01 6.27468288e-01 1.49201751e-01
-6.22143969e-02 -3.35517138e-01 -4.83647972e-01 7.11524069e-01
7.08978057e-01 -1.01774108e+00 -2.95714557e-01 -5.67741275e-01
3.93476158e-01 -1.50045902e-02 -6.15672618e-02 -3.62603009e-01
-7.54591227e-02 -2.28288233e-01 9.16138142e-02 -6.44669950e-01
-9.29052904e-02 -2.86166757e-01 1.19570673e-01 -1.80399105e-01
3.67257209e-03 1.61282822e-01 1.62528932e-01 6.18336022e-01
-1.24293491e-01 -1.78984895e-01 6.35722041e-01 4.69916277e-02
-5.21741331e-01 3.21122371e-02 -7.76550353e-01 5.02920926e-01
4.24721032e-01 4.06962693e-01 3.06873202e-01 -2.36778885e-01
-1.51726174e+00 2.21013308e-01 1.61286384e-01 4.83962953e-01
2.10751012e-01 -1.18607664e+00 -1.05070662e+00 5.08437276e-01
-7.01467320e-02 -2.67304242e-01 -8.32978785e-02 5.50789773e-01
-2.02722952e-01 3.85614097e-01 1.38936983e-02 -7.37382114e-01
-8.73107374e-01 3.97880226e-01 1.61009580e-01 -3.74976844e-01
-4.82744694e-01 9.61169183e-01 2.75471568e-01 -1.06142163e+00
5.35890125e-02 -3.92983705e-01 1.09395720e-02 -2.69358724e-01
1.41075239e-01 -1.78792160e-02 -1.75896227e-01 -7.97022939e-01
-2.55203664e-01 4.78584230e-01 -6.96114153e-02 -8.12596440e-01
1.05293036e+00 -2.74249673e-01 -4.50314641e-01 1.02468240e+00
7.04796731e-01 2.05029532e-01 -1.33015096e+00 -3.72162104e-01
4.25657928e-01 1.05567545e-01 2.70724874e-02 -7.31159091e-01
-6.23020470e-01 6.57824516e-01 3.32672238e-01 -1.75603956e-01
6.57410443e-01 4.28994685e-01 3.32270414e-01 4.54931706e-01
4.84631449e-01 -1.07191086e+00 -7.55830109e-01 1.08741903e+00
5.47749698e-01 -1.14427221e+00 -3.30868036e-01 -3.51492643e-01
-5.95717013e-01 1.04628456e+00 -1.06725115e-02 -1.89339697e-01
1.20007098e+00 6.73153818e-01 1.54486284e-01 5.65206856e-02
-7.57716596e-01 -3.84091169e-01 3.38916600e-01 9.55793202e-01
7.28159189e-01 5.36948800e-01 -4.31993634e-01 9.30732787e-01
-5.75774312e-01 -6.23902082e-01 4.96553719e-01 6.57273591e-01
-1.55294508e-01 -1.88420594e+00 -3.14306766e-01 6.39172420e-02
-6.41277492e-01 -5.07107079e-01 1.26255611e-02 8.88362050e-01
3.21212798e-01 9.56958175e-01 2.87021697e-01 1.63631633e-01
-7.85783976e-02 7.77771533e-01 1.15562427e+00 -9.83661115e-01
1.18046552e-02 4.86573398e-01 1.60295501e-01 -1.97196901e-01
-3.53401452e-01 -1.17177594e+00 -1.30486763e+00 8.10075030e-02
-3.21617693e-01 4.30359513e-01 7.47859716e-01 1.18867230e+00
-1.74623370e-01 1.71572164e-01 1.95887133e-01 -6.72596812e-01
-5.53183973e-01 -1.27324855e+00 -9.47022021e-01 2.10185334e-01
-5.94255365e-02 -6.91623569e-01 -4.38083053e-01 2.87280887e-01]
|
[10.385549545288086, 9.699658393859863]
|
fac71cd8-8123-4a47-aa98-4e5f6c972f8b
|
occlusion-guided-compact-template-learning
|
1903.04752
| null |
http://arxiv.org/abs/1903.04752v2
|
http://arxiv.org/pdf/1903.04752v2.pdf
|
Occlusion-guided compact template learning for ensemble deep network-based pose-invariant face recognition
|
Concatenation of the deep network representations extracted from different
facial patches helps to improve face recognition performance. However, the
concatenated facial template increases in size and contains redundant
information. Previous solutions aim to reduce the dimensionality of the facial
template without considering the occlusion pattern of the facial patches. In
this paper, we propose an occlusion-guided compact template learning (OGCTL)
approach that only uses the information from visible patches to construct the
compact template. The compact face representation is not sensitive to the
number of patches that are used to construct the facial template and is more
suitable for incorporating the information from different view angles for
image-set based face recognition. Instead of using occlusion masks in face
matching (e.g., DPRFS [38]), the proposed method uses occlusion masks in
template construction and achieves significantly better image-set based face
verification performance on a challenging database with a template size that is
an order-of-magnitude smaller than DPRFS.
|
['Ioannis A. Kakadiaris', 'Yuhang Wu']
|
2019-03-12
| null | null | null | null |
['robust-face-recognition']
|
['computer-vision']
|
[ 2.91022539e-01 1.48154125e-02 1.01510763e-01 -5.10452509e-01
-5.13721526e-01 -4.15805697e-01 3.71308655e-01 -4.47163701e-01
-1.10334225e-01 2.54636765e-01 -1.12884738e-01 -5.04525239e-03
-2.40692660e-01 -7.76122510e-01 -6.39018774e-01 -1.05232179e+00
5.83806634e-01 3.72816361e-02 -2.52199531e-01 2.54931394e-02
8.46010223e-02 1.06088233e+00 -1.98507404e+00 3.38230163e-01
5.06937206e-01 1.43869913e+00 7.11867586e-02 -1.85974121e-01
-3.15586887e-02 7.29561225e-02 -5.64670146e-01 -4.22684431e-01
7.28260517e-01 -2.98283547e-01 -4.70304824e-02 3.99485677e-01
1.16720450e+00 -3.15948397e-01 -4.44932252e-01 7.78394043e-01
6.31250143e-01 -8.08240939e-03 5.42635024e-01 -1.10690248e+00
-1.74103692e-01 -1.52076229e-01 -5.55068135e-01 -2.21485138e-01
5.46018630e-02 -7.37057626e-02 4.79764372e-01 -1.22417581e+00
6.26846731e-01 1.29095316e+00 5.73267460e-01 5.89668095e-01
-1.15215540e+00 -1.00969076e+00 -4.12929468e-02 2.86384195e-01
-1.81361532e+00 -1.04558182e+00 1.02735424e+00 -3.36980611e-01
7.25499094e-01 3.90909284e-01 5.65289497e-01 6.60513580e-01
-3.90880257e-02 1.41412988e-01 1.02397037e+00 -6.13983452e-01
-1.10986330e-01 3.94064635e-02 -8.91278014e-02 7.04547226e-01
4.72816974e-01 1.55194476e-01 -4.70201105e-01 -2.50910491e-01
1.04942882e+00 2.44750991e-01 -3.96143198e-01 -4.24221426e-01
-6.48041666e-01 7.26406276e-01 2.92033225e-01 4.41350788e-01
-4.02734518e-01 -1.56267449e-01 9.05419812e-02 1.92697212e-01
2.70796806e-01 1.19936578e-01 -1.89227775e-01 5.22538662e-01
-1.20945668e+00 3.10483333e-02 4.24959749e-01 7.00857341e-01
1.09530938e+00 3.16897064e-01 -1.62081704e-01 1.05044115e+00
4.38808769e-01 8.41939867e-01 1.05644941e-01 -9.09473598e-01
4.57717419e-01 7.42804229e-01 -8.59646425e-02 -1.37405837e+00
-3.47180516e-01 -3.59340996e-01 -9.91892040e-01 1.81902364e-01
4.29703355e-01 7.89596662e-02 -9.74377155e-01 1.58403528e+00
4.66376632e-01 -2.15942115e-02 -1.55700460e-01 6.89074695e-01
1.02079666e+00 3.00529033e-01 -4.77909476e-01 -2.20836416e-01
1.44018698e+00 -5.89165151e-01 -6.47521138e-01 -9.65025425e-02
3.31457943e-01 -9.50823784e-01 4.21382785e-01 2.38045290e-01
-8.57415378e-01 -7.40294933e-01 -1.28985119e+00 1.76204629e-02
-6.59401566e-02 5.35902679e-01 2.20973328e-01 9.03482974e-01
-1.16556597e+00 3.26593906e-01 -4.34376866e-01 -2.11778563e-02
6.44832075e-01 8.01442444e-01 -9.01578605e-01 -4.97762471e-01
-4.75806683e-01 5.90746224e-01 3.25325993e-03 3.85602862e-01
-4.89272445e-01 -4.25825715e-01 -1.02868164e+00 3.58414054e-01
1.91426635e-01 -1.87686563e-01 4.51985925e-01 -9.17208850e-01
-1.43288267e+00 7.38765895e-01 -6.81826413e-01 8.07338879e-02
1.27996057e-01 2.97815055e-01 -2.85492212e-01 3.06147754e-01
-3.11535746e-01 6.98385477e-01 1.44739628e+00 -1.05003011e+00
3.23754326e-02 -8.18263292e-01 -2.27567151e-01 -9.70328078e-02
-3.64198297e-01 1.14033617e-01 -5.55248201e-01 -6.29046619e-01
5.55079818e-01 -9.71651018e-01 -4.61839996e-02 3.50763947e-01
8.89377967e-02 -1.29682168e-01 1.21596777e+00 -6.97845757e-01
8.77894223e-01 -2.46105075e+00 -1.06666796e-01 5.19320786e-01
1.50052503e-01 6.12349987e-01 -5.46671569e-01 2.25300744e-01
-2.04993963e-01 -1.77373692e-01 8.89411047e-02 -4.19126183e-01
-1.86114952e-01 2.51589328e-01 7.48640224e-02 6.22397840e-01
2.03826472e-01 7.07100332e-01 -8.74688625e-02 -5.10748923e-01
2.00396717e-01 8.86626720e-01 -5.86979806e-01 1.06984384e-01
2.19017074e-01 3.17339420e-01 -1.54611781e-01 6.83481872e-01
1.37404597e+00 1.02721520e-01 3.98775011e-01 -5.68342447e-01
1.00535117e-01 3.19365636e-02 -1.26684606e+00 1.36692560e+00
-5.11830509e-01 5.89839339e-01 2.91506797e-01 -8.77830327e-01
1.41218805e+00 6.30801737e-01 5.67169130e-01 -8.74084651e-01
2.48620406e-01 2.39866197e-01 5.24468422e-02 -1.56275749e-01
-1.46425352e-01 1.45994559e-01 4.64701176e-01 4.59378004e-01
2.86698610e-01 2.51439452e-01 7.55872056e-02 -6.17667496e-01
6.53277397e-01 -1.66165277e-01 2.06948563e-01 -2.22329453e-01
9.61643636e-01 -8.19923699e-01 8.92025948e-01 2.34651804e-01
2.59105321e-02 6.39108181e-01 2.08019152e-01 -8.07103813e-01
-9.59482729e-01 -6.53284013e-01 -3.64521056e-01 3.52830499e-01
-2.95758005e-02 -4.82135385e-01 -8.47843349e-01 -5.66897511e-01
2.32539494e-02 -3.06510050e-02 -5.56393206e-01 1.10381478e-02
-9.15934861e-01 -4.13365990e-01 4.08645034e-01 2.52676696e-01
6.21205807e-01 -8.58169734e-01 -4.60113287e-01 -2.70899776e-02
-1.54101267e-01 -1.07820225e+00 -6.47801936e-01 -2.39176437e-01
-8.44493091e-01 -1.28317511e+00 -6.54308736e-01 -7.73452997e-01
1.25185061e+00 5.42767882e-01 5.36537230e-01 3.31086665e-01
-4.13844824e-01 1.60628855e-01 -2.32321844e-02 -2.90291846e-01
-2.53604297e-02 -3.39810610e-01 2.35815886e-02 7.03355610e-01
3.78009528e-01 -6.04289770e-01 -6.59985483e-01 6.41236842e-01
-7.22018659e-01 -2.02562213e-01 6.32753551e-01 1.09688413e+00
6.47152841e-01 -1.22355260e-01 3.94581944e-01 -4.33884948e-01
6.73276260e-02 1.05927952e-01 -7.53054917e-01 4.03680176e-01
-3.37184757e-01 -2.43261401e-02 5.43499947e-01 -4.58730787e-01
-1.04939854e+00 4.32802767e-01 -1.17601775e-01 -6.37488604e-01
-6.00794293e-02 -4.65895273e-02 -5.60577452e-01 -8.16611588e-01
4.07595962e-01 2.77158022e-01 4.00061965e-01 -6.59735978e-01
-1.23413801e-01 5.52702487e-01 1.07379101e-01 -2.89033562e-01
7.67018855e-01 4.97838587e-01 3.78604501e-01 -9.05999005e-01
-2.83396125e-01 -3.59612852e-01 -9.71810639e-01 -3.12887698e-01
4.34341639e-01 -8.65891933e-01 -8.03476632e-01 4.53731656e-01
-1.24579799e+00 3.52524906e-01 6.99174702e-02 3.69725317e-01
-2.33733639e-01 4.02548701e-01 -1.84257984e-01 -6.94112599e-01
-6.12603128e-01 -1.33987796e+00 1.20908570e+00 2.22827598e-01
2.16485307e-01 -4.38301951e-01 -3.31246942e-01 3.53094250e-01
4.06035334e-01 1.38470098e-01 7.52639472e-01 -2.50189215e-01
-7.40907848e-01 -4.65021104e-01 -3.57700825e-01 4.35458094e-01
5.08609533e-01 -1.43198386e-01 -1.25390828e+00 -5.53227544e-01
4.48339850e-01 1.21143408e-01 7.25757778e-01 2.20570102e-01
1.05912304e+00 -5.48271358e-01 -2.54443944e-01 6.90156698e-01
1.50772464e+00 4.20066983e-01 1.00863993e+00 -6.07076436e-02
6.19847536e-01 9.89171505e-01 4.11785275e-01 4.56526011e-01
-4.56403494e-02 1.14402544e+00 4.72992659e-02 -1.40701652e-01
-3.88109446e-01 5.97918816e-02 2.27075174e-01 3.81470054e-01
-7.55907968e-02 2.92144194e-02 -5.07775545e-01 1.56671748e-01
-1.47551632e+00 -1.04221714e+00 3.27732921e-01 2.34321690e+00
4.08323973e-01 -5.04339933e-01 -1.43668771e-01 3.63046497e-01
7.91845977e-01 1.56148627e-01 -3.82390410e-01 -2.75985092e-01
-2.85742432e-01 5.69073200e-01 1.08582817e-01 3.00805956e-01
-7.76158154e-01 6.18126392e-01 5.72455597e+00 9.35794175e-01
-1.35961723e+00 8.73204172e-02 5.24267912e-01 -1.11342944e-01
-3.76942120e-02 -1.78283691e-01 -1.05441308e+00 1.73662141e-01
5.14283359e-01 3.11220556e-01 3.69936854e-01 6.02281153e-01
4.63020168e-02 4.07719277e-02 -9.02840316e-01 1.46254492e+00
4.62376267e-01 -1.30089235e+00 1.49096444e-01 5.03221571e-01
5.85532963e-01 -3.20489526e-01 1.22286953e-01 -3.15422684e-01
-6.94140255e-01 -1.09066856e+00 4.58267659e-01 4.04871821e-01
9.31187689e-01 -7.71381378e-01 7.91786015e-01 8.46045613e-02
-1.27452242e+00 -6.56906888e-02 -6.67666256e-01 2.45302394e-01
-2.74568826e-01 5.02874553e-01 -7.01343834e-01 4.96650577e-01
6.31541908e-01 4.84226942e-01 -5.48863649e-01 8.88595760e-01
1.10973991e-01 1.48123980e-01 -6.20799601e-01 3.69529516e-01
-1.22271121e-01 -3.38954955e-01 2.54452467e-01 7.25760937e-01
6.12724483e-01 -7.24766627e-02 -4.24514003e-02 7.72171319e-01
-1.64249256e-01 2.98277676e-01 -7.40835369e-01 1.77341297e-01
5.54696143e-01 1.43889892e+00 -5.01680791e-01 -8.16786215e-02
-5.04708648e-01 5.57112157e-01 2.71827459e-01 3.09910029e-01
-4.42152619e-01 -3.11229706e-01 7.54450142e-01 2.70788938e-01
8.41918647e-01 -2.79369295e-01 -1.32180089e-02 -8.12144279e-01
5.54702520e-01 -1.06801641e+00 9.70530231e-03 -2.02243388e-01
-8.73305500e-01 9.22940016e-01 -2.45216355e-01 -1.36941898e+00
-1.08931355e-01 -6.67736769e-01 -5.07214725e-01 1.14627314e+00
-1.40269268e+00 -1.25650132e+00 -4.77380067e-01 7.21281469e-01
1.25269517e-01 -2.25747064e-01 9.13611114e-01 4.88277107e-01
-6.36895120e-01 1.08561337e+00 4.15918091e-03 2.36762449e-01
6.86371028e-01 -2.33943298e-01 2.41434053e-01 6.34245634e-01
2.72191882e-01 7.67434657e-01 1.28008023e-01 -3.57782036e-01
-1.43444681e+00 -1.02758825e+00 8.66887271e-01 9.26908944e-03
-3.93398911e-01 -4.94623065e-01 -7.97986567e-01 2.13273123e-01
-1.26635730e-02 2.75866270e-01 8.54036629e-01 -1.92334309e-01
-8.43700409e-01 -7.36620903e-01 -1.50010777e+00 3.67754370e-01
1.02213633e+00 -5.08361220e-01 -2.62399707e-02 5.08299880e-02
2.72026006e-02 -1.65647358e-01 -8.92701626e-01 6.58086479e-01
1.01362813e+00 -9.35732365e-01 9.00020301e-01 -2.04818440e-03
-2.55018085e-01 -4.55145001e-01 -2.53673375e-01 -7.93496907e-01
-3.33300114e-01 -4.92438108e-01 2.51151353e-01 1.17847455e+00
2.04105154e-01 -9.64975536e-01 9.15113926e-01 4.11917448e-01
1.60893574e-01 -7.94223607e-01 -1.26337969e+00 -8.70373249e-01
-4.25996512e-01 1.66037053e-01 9.29801524e-01 5.95529556e-01
-2.80747086e-01 -1.16056792e-01 -2.97813654e-01 1.34710506e-01
7.48486042e-01 3.23647380e-01 8.32623780e-01 -1.42303514e+00
-1.06815090e-02 -2.18023494e-01 -7.13130653e-01 -7.89876759e-01
2.13350579e-01 -7.49425709e-01 -1.17999427e-01 -1.11395025e+00
-6.20985515e-02 -6.51791096e-01 -8.39562938e-02 8.24691296e-01
2.96627402e-01 8.11941147e-01 4.85478073e-01 1.42344907e-01
-1.36514157e-02 6.49852216e-01 1.35788822e+00 -2.45335266e-01
2.88018640e-02 -8.99631083e-02 -4.89634275e-01 4.06826764e-01
7.92754650e-01 -3.70956779e-01 -2.51900345e-01 -4.04358745e-01
-3.98801297e-01 -5.56173958e-02 1.84627771e-01 -1.02114069e+00
1.97711766e-01 8.93735513e-02 8.11830521e-01 -6.43534362e-01
8.32185805e-01 -1.02456427e+00 6.35313928e-01 3.62753510e-01
2.54450530e-01 8.11546296e-02 3.93986702e-01 2.12533548e-01
-4.04554963e-01 -3.44657332e-01 8.97812366e-01 -4.11240160e-02
-8.63800272e-02 4.96929079e-01 -9.69179869e-02 -7.21727848e-01
8.23363602e-01 -5.94000995e-01 -2.31007755e-01 -8.63896236e-02
-4.26643223e-01 -3.38352680e-01 4.61788863e-01 4.22983140e-01
8.56709063e-01 -1.53605509e+00 -6.43456876e-01 9.12213147e-01
-3.14851515e-02 -2.27647260e-01 4.60634708e-01 7.66151428e-01
-4.10840005e-01 8.24408352e-01 -5.33781111e-01 -6.81519926e-01
-1.84568083e+00 5.16722381e-01 4.63036329e-01 7.10213976e-03
-4.97445494e-01 6.16712153e-01 4.51062173e-01 -1.85723141e-01
1.87441900e-01 1.44131616e-01 -3.42686266e-01 1.06093667e-01
8.10407639e-01 2.28074834e-01 4.58781391e-01 -1.12444675e+00
-2.60793388e-01 1.28035533e+00 -2.22519636e-01 2.37107113e-01
1.37252557e+00 1.01050109e-01 -3.18204850e-01 -3.21724176e-01
1.44919503e+00 1.19359322e-01 -1.15771210e+00 -4.22087401e-01
-2.39040181e-01 -8.41715515e-01 1.99535880e-02 -3.44817311e-01
-1.60888040e+00 9.68800962e-01 8.23544323e-01 -4.85087365e-01
1.37722528e+00 -3.83986712e-01 5.99681377e-01 3.02987218e-01
5.76246083e-01 -7.48560905e-01 -1.99124664e-02 1.77350625e-01
1.34134233e+00 -9.76463020e-01 5.82801215e-02 -8.00221980e-01
1.22260392e-01 1.36674976e+00 6.63153052e-01 5.39629348e-02
8.96013558e-01 8.13174248e-02 -1.91327389e-02 -2.62929112e-01
-3.80513787e-01 -1.12695657e-01 7.28924274e-01 6.53313577e-01
1.53661311e-01 -2.15916276e-01 -2.66809791e-01 3.08429211e-01
5.16195036e-02 -3.02562565e-01 -8.04586932e-02 6.32468820e-01
-1.98630407e-01 -1.59926021e+00 -6.70238316e-01 4.40032333e-01
-6.86717406e-02 7.87361711e-03 -3.27953100e-01 4.93846208e-01
4.76476669e-01 9.60344136e-01 1.15597121e-01 -3.67788166e-01
3.17804903e-01 1.57835767e-01 1.12564600e+00 -4.21300024e-01
-4.71611261e-01 2.53612041e-01 -2.20650420e-01 -6.47509515e-01
-5.06370962e-01 -5.94176352e-01 -6.53423309e-01 -2.61787504e-01
-5.36490321e-01 -1.11905269e-01 6.84755027e-01 7.80991316e-01
8.09652627e-01 -1.23083688e-01 9.10552680e-01 -1.07538450e+00
-1.56325832e-01 -8.39061737e-01 -4.97192860e-01 1.53062269e-01
2.14952588e-01 -9.44336116e-01 -3.08339179e-01 -2.64647245e-01]
|
[13.158958435058594, 0.4359027147293091]
|
d8859631-f363-4857-a496-ed3b75cdb88c
|
gedi-generative-discriminator-guided-sequence
|
2009.06367
| null |
https://arxiv.org/abs/2009.06367v2
|
https://arxiv.org/pdf/2009.06367v2.pdf
|
GeDi: Generative Discriminator Guided Sequence Generation
|
While large-scale language models (LMs) are able to imitate the distribution of natural language well enough to generate realistic text, it is difficult to control which regions of the distribution they generate. This is especially problematic because datasets used for training large LMs usually contain significant toxicity, hate, bias, and negativity. We propose GeDi as an efficient method for using smaller LMs as generative discriminators to guide generation from large LMs to make them safer and more controllable. GeDi guides generation at each step by computing classification probabilities for all possible next tokens via Bayes rule by normalizing over two class-conditional distributions; one conditioned on the desired attribute, or control code, and another conditioned on the undesired attribute, or anti control code. We find that GeDi gives stronger controllability than the state of the art method while also achieving generation speeds more than 30 times faster. Additionally, training GeDi on only four topics allows us to controllably generate new topics zero-shot from just a keyword, unlocking a new capability that previous controllable generation methods do not have. Lastly, we show that GeDi can make GPT-2 (1.5B parameters) significantly less toxic without sacrificing linguistic quality, making it by far the most practical existing method for detoxifying large language models while maintaining a fast generation speed.
|
['Nazneen Fatema Rajani', 'Richard Socher', 'Akhilesh Deepak Gotmare', 'Shafiq Joty', 'Bryan McCann', 'Ben Krause', 'Nitish Shirish Keskar']
|
2020-09-14
| null |
https://aclanthology.org/2021.findings-emnlp.424
|
https://aclanthology.org/2021.findings-emnlp.424.pdf
|
findings-emnlp-2021-11
|
['linguistic-acceptability']
|
['natural-language-processing']
|
[ 2.42562622e-01 2.03047141e-01 -2.74019957e-01 2.24811539e-01
-7.30858147e-01 -1.15793705e+00 8.73794019e-01 3.99006605e-01
-2.56198555e-01 1.27599454e+00 1.81202784e-01 -5.07106960e-01
2.13697329e-01 -1.22950852e+00 -6.76822722e-01 -7.72264004e-01
1.39040574e-01 7.31862247e-01 -5.86170703e-02 -4.18877691e-01
2.92992562e-01 5.12787879e-01 -1.33949220e+00 2.59022117e-01
1.22175562e+00 1.52225912e-01 1.57289803e-01 6.51701152e-01
-3.75160836e-02 6.10012889e-01 -1.10273528e+00 -3.41088831e-01
-5.79823069e-02 -5.82320452e-01 -4.39643502e-01 -3.32281768e-01
2.92481989e-01 -4.83358540e-02 3.39066163e-02 8.17044795e-01
6.41240537e-01 6.25239834e-02 1.12714446e+00 -1.19896078e+00
-1.04594457e+00 9.59124029e-01 -3.56555998e-01 -1.05606467e-01
4.34357375e-01 3.08375269e-01 9.20292199e-01 -4.74707097e-01
7.23779023e-01 1.37229633e+00 3.64349186e-01 1.12351525e+00
-1.65320730e+00 -1.00735235e+00 1.03499666e-01 -5.17906070e-01
-1.45927107e+00 -3.09270293e-01 2.62511700e-01 -3.91785026e-01
1.09160352e+00 4.48007435e-01 5.75980067e-01 1.50902104e+00
5.20828247e-01 5.92277169e-01 9.75958765e-01 -2.70562738e-01
5.03838480e-01 4.78273928e-01 -2.51203388e-01 5.37029862e-01
6.42373204e-01 2.36358494e-02 -4.41893458e-01 -4.53929991e-01
5.04125476e-01 -3.11350852e-01 -4.01493967e-01 -3.02801095e-02
-1.21091783e+00 1.15677977e+00 9.95461568e-02 3.71104985e-01
-1.35159895e-01 2.17210874e-01 -8.92649889e-02 4.05905806e-02
3.19621950e-01 1.20807707e+00 -3.44792277e-01 -5.93886264e-02
-7.57874668e-01 6.33381665e-01 9.33577299e-01 1.01826024e+00
5.83463192e-01 2.84182191e-01 -6.29755497e-01 5.50780535e-01
-5.41911460e-02 9.26813602e-01 6.59832001e-01 -4.76484716e-01
3.86613131e-01 4.06630278e-01 2.51129270e-01 -7.06631243e-01
-1.73576429e-01 -4.16744858e-01 -7.14527786e-01 4.91278917e-02
5.69638848e-01 -5.49923360e-01 -1.05245793e+00 2.29801917e+00
-3.06321196e-02 -1.78464308e-01 1.31564572e-01 1.55973941e-01
2.53399014e-01 1.09510756e+00 4.36940372e-01 -1.22713394e-01
1.10151947e+00 -4.85459566e-01 -3.97059351e-01 -2.09073618e-01
9.11425173e-01 -5.93648732e-01 1.33248770e+00 4.53047335e-01
-9.57630098e-01 -3.27889808e-02 -1.01943433e+00 1.84196874e-01
-6.72827542e-01 -7.99959749e-02 9.34471905e-01 1.01738107e+00
-1.10755706e+00 4.87372696e-01 -4.43255633e-01 -1.77115113e-01
2.68553257e-01 5.06092370e-01 -2.99999177e-01 1.39062986e-01
-1.50491560e+00 9.28328574e-01 3.67137134e-01 -5.20612955e-01
-1.14907277e+00 -1.00454295e+00 -8.09648633e-01 2.19141215e-01
1.07682794e-02 -7.55094171e-01 9.27085757e-01 -3.62924516e-01
-1.58726847e+00 3.77748877e-01 -3.95426042e-02 -3.61467928e-01
3.44575882e-01 1.29929647e-01 -1.96236655e-01 -2.42733762e-01
1.94836751e-01 1.13677108e+00 7.42199957e-01 -1.08776414e+00
-2.65544415e-01 -7.31641054e-03 9.37022120e-02 1.87790006e-01
-8.15682471e-01 -2.72528887e-01 -2.48542838e-02 -7.69700646e-01
-6.44446492e-01 -1.04948246e+00 -2.67938226e-01 -3.50164056e-01
-8.29322278e-01 -3.15063715e-01 4.76183504e-01 -1.38453647e-01
1.31128931e+00 -1.79137385e+00 1.32588193e-01 1.98072031e-01
2.68519610e-01 3.81856412e-01 -5.46935081e-01 6.40588403e-01
-8.38826075e-02 8.20009172e-01 -1.94166545e-02 -2.45068204e-02
1.14634715e-01 -1.04372829e-01 -6.47260785e-01 6.43542632e-02
2.82803297e-01 7.99906075e-01 -1.04240620e+00 -1.85548440e-01
-1.50870159e-01 5.95744550e-01 -8.42028201e-01 -2.02335715e-02
-6.07590914e-01 2.54460186e-01 -4.10264045e-01 2.40516096e-01
3.77118349e-01 -1.35609373e-01 1.27948716e-01 2.83959657e-01
-5.94512336e-02 3.32467556e-01 -9.13158536e-01 1.22803617e+00
-6.46754563e-01 3.89347434e-01 -4.35154349e-01 -3.35456729e-01
8.98645222e-01 3.20429385e-01 1.14380576e-01 -3.89443159e-01
7.88340345e-02 1.30048901e-01 6.14223629e-02 4.48374376e-02
5.24900079e-01 -4.90560919e-01 -4.40107435e-01 7.62937188e-01
-6.82835281e-02 -7.69924283e-01 5.93030035e-01 5.35718799e-01
1.04781461e+00 -3.98608953e-01 4.28419299e-02 -4.99105841e-01
1.71351179e-01 -1.14417560e-01 4.30312902e-01 1.00400662e+00
3.11972916e-01 3.81988317e-01 8.57887149e-01 7.25816414e-02
-1.14436769e+00 -1.18144381e+00 1.29366979e-01 9.21653867e-01
-1.83073640e-01 -7.29878306e-01 -9.92542982e-01 -4.57736343e-01
6.24615103e-02 1.46493816e+00 -7.02378690e-01 -5.46220779e-01
-2.30586797e-01 -9.39996898e-01 9.41740632e-01 3.94534081e-01
4.00322787e-02 -8.35858405e-01 -1.08750470e-01 1.17033981e-01
-1.55099675e-01 -5.52601576e-01 -8.11565161e-01 3.43382031e-01
-5.67115247e-01 -4.29019779e-01 -7.25749969e-01 -5.42225659e-01
9.39561367e-01 -3.38633806e-02 1.13411701e+00 -3.99428159e-02
-1.56671464e-01 -1.28201649e-01 3.11676338e-02 -6.19615972e-01
-8.15736532e-01 4.42927510e-01 8.04109052e-02 -5.59455335e-01
7.85456300e-02 -5.34657240e-01 -2.86227733e-01 1.85131207e-01
-1.14775741e+00 1.95121154e-01 4.94638562e-01 7.71293819e-01
3.14116806e-01 2.77289450e-01 7.31209874e-01 -1.20135939e+00
1.12797582e+00 -5.18318772e-01 -5.02621412e-01 3.01082790e-01
-7.20394194e-01 4.79846746e-01 1.13437653e+00 -8.58916104e-01
-1.00964308e+00 -1.29782960e-01 -7.35875517e-02 1.92924500e-01
-9.20585841e-02 3.10588270e-01 -4.33658898e-01 2.66806513e-01
1.13837886e+00 3.29413265e-01 -2.68414617e-01 4.26334254e-02
7.08314478e-01 4.48717207e-01 1.71012074e-01 -8.57251585e-01
1.02097213e+00 -1.03142560e-02 -1.25434637e-01 -6.66132629e-01
-5.41870415e-01 2.00187653e-01 -2.39694729e-01 2.85500050e-01
9.38887358e-01 -9.21672344e-01 -7.15652168e-01 4.07999903e-01
-1.01119542e+00 -7.40693986e-01 -2.07363144e-01 1.48572221e-01
-2.76415229e-01 -2.13008164e-03 -7.63993025e-01 -5.99643767e-01
-4.61771101e-01 -1.11493528e+00 9.16034698e-01 2.87678152e-01
-6.98546052e-01 -1.11550665e+00 2.05188796e-01 8.41980428e-02
6.00371182e-01 2.42395863e-01 1.44585776e+00 -7.96444595e-01
-4.86580044e-01 -3.06309581e-01 2.46973738e-01 7.00859576e-02
2.74232060e-01 1.84620351e-01 -6.66224837e-01 -3.69038224e-01
-3.37263882e-01 -4.11210537e-01 7.62988210e-01 7.37476861e-04
8.78165007e-01 -6.09880149e-01 -5.36599159e-01 2.99112380e-01
1.18284202e+00 3.29810739e-01 6.64323986e-01 1.43193053e-02
6.22992516e-01 1.53409973e-01 2.30207846e-01 4.19428974e-01
1.38881862e-01 3.72636497e-01 -8.77892897e-02 -1.12342071e-02
4.45825867e-02 -7.48266518e-01 6.35219753e-01 3.24304372e-01
3.30135345e-01 -1.01854074e+00 -7.96564758e-01 4.60075140e-01
-1.26688588e+00 -1.05369782e+00 1.76512361e-01 2.38838720e+00
1.47998822e+00 2.80867964e-01 -5.29308394e-02 -2.63324287e-02
6.07894421e-01 -1.32152423e-01 -5.68825662e-01 -5.26820123e-01
-2.04033971e-01 5.60048759e-01 5.27795553e-01 7.14190066e-01
-7.58714855e-01 1.24645221e+00 6.81156445e+00 1.30952549e+00
-1.30764377e+00 -3.02958727e-01 7.95220971e-01 -3.25839847e-01
-1.07211041e+00 -7.26660565e-02 -1.11480021e+00 5.38090408e-01
1.08293498e+00 -6.21300638e-01 5.63269317e-01 7.24626839e-01
2.75436640e-01 -2.79019754e-02 -1.18328607e+00 5.64094126e-01
3.65787894e-02 -1.39548993e+00 7.52293646e-01 2.46801943e-01
1.03800428e+00 -5.31741381e-01 3.95970374e-01 3.36584836e-01
8.13807249e-01 -1.50174904e+00 7.74744928e-01 1.98574513e-01
9.05176520e-01 -1.08096218e+00 2.01822177e-01 5.42906940e-01
-6.85906470e-01 3.04846000e-02 -5.29775262e-01 2.47668903e-02
-7.32024908e-02 6.18928730e-01 -1.08022559e+00 -1.34573281e-01
-6.51690587e-02 1.72374800e-01 -5.57366192e-01 5.24937332e-01
-5.85105896e-01 7.01918542e-01 -2.95883417e-01 -6.20553792e-01
2.09905788e-01 -1.15250237e-01 5.00075459e-01 1.12423718e+00
5.68593740e-01 -3.36506441e-02 -3.27155814e-02 1.18859696e+00
-3.26123357e-01 4.17504311e-02 -9.85523820e-01 -5.39297402e-01
8.06940377e-01 1.20897186e+00 -8.10595334e-01 -5.03398061e-01
2.34426379e-01 9.19976592e-01 2.31745273e-01 2.98225641e-01
-8.75121355e-01 -6.34260595e-01 7.01241195e-01 1.21553928e-01
1.07387230e-01 -2.29269758e-01 -4.41948533e-01 -1.15466523e+00
-5.75034499e-01 -1.21492076e+00 1.41234607e-01 -8.07664633e-01
-1.06633830e+00 6.19119823e-01 -6.49465397e-02 -6.57254398e-01
-4.15141344e-01 -5.33140361e-01 -6.35443687e-01 1.04165053e+00
-8.99986863e-01 -1.02666342e+00 2.47294173e-01 4.51530576e-01
3.84332955e-01 -1.08643837e-01 1.18024361e+00 9.34941620e-02
-6.67256534e-01 9.68312025e-01 2.80769020e-02 -1.60419032e-01
9.29339051e-01 -1.40005720e+00 5.58572590e-01 6.58153236e-01
1.12626247e-01 1.21629155e+00 9.21846449e-01 -8.40953946e-01
-1.26751566e+00 -1.14679039e+00 1.14359236e+00 -7.75873065e-01
5.47779620e-01 -8.58339548e-01 -5.49600065e-01 4.30293620e-01
9.08734724e-02 -8.93539131e-01 8.47418725e-01 2.32831948e-02
-4.83105898e-01 1.17863625e-01 -1.13939333e+00 1.13435376e+00
8.15075278e-01 -4.14844990e-01 -1.06740616e-01 5.77003598e-01
8.16651046e-01 -3.00646722e-01 -6.25589550e-01 -3.09432093e-02
2.40493178e-01 -4.86521989e-01 9.27258551e-01 -6.95910871e-01
5.98618269e-01 -3.06542158e-01 4.06324044e-02 -1.75180316e+00
-4.99835879e-01 -9.57592130e-01 3.42529744e-01 1.44045794e+00
1.15609729e+00 -7.21123874e-01 7.06231773e-01 8.89999449e-01
8.15872476e-02 -5.41681170e-01 -3.18437845e-01 -8.16030920e-01
7.27998853e-01 -1.21391609e-01 8.25918794e-01 8.80112767e-01
3.50805581e-01 6.81034446e-01 -4.70337600e-01 -6.12512007e-02
3.05403858e-01 -1.55637965e-01 6.89954460e-01 -9.14390326e-01
-2.87590921e-01 -5.80171168e-01 -5.26346229e-02 -9.22821820e-01
2.75825828e-01 -1.15567386e+00 7.69841224e-02 -1.26832652e+00
4.61456984e-01 -5.05637527e-01 7.61498883e-02 7.84304261e-01
-3.72925550e-01 2.46204540e-01 1.63122892e-01 -1.74139559e-01
-1.56789452e-01 4.67563838e-01 1.33534455e+00 -3.49534571e-01
-3.15229625e-01 -1.91504285e-01 -1.42254996e+00 3.14430416e-01
1.04925954e+00 -5.70023119e-01 -1.05826628e+00 -3.19410115e-01
4.53328967e-01 -1.63531512e-01 -1.50294587e-01 -9.44963694e-01
8.25371221e-02 -4.43928361e-01 4.89169389e-01 -1.44970417e-01
1.34879947e-01 -3.11154453e-03 2.59630442e-01 4.22290981e-01
-6.76886737e-01 -3.82153876e-02 4.00896400e-01 4.19504493e-01
2.58317292e-01 -2.44874015e-01 7.26012588e-01 -1.44718528e-01
4.68900427e-02 3.91008884e-01 -1.18382847e+00 2.79996365e-01
8.80112767e-01 1.43056124e-01 -5.46282053e-01 -5.58349788e-01
-3.26036543e-01 2.74113975e-02 7.34729528e-01 2.80593127e-01
3.98912162e-01 -1.18431878e+00 -6.92936838e-01 1.93331257e-01
-1.17941229e-02 -2.92682827e-01 -2.66510434e-02 1.27261147e-01
-4.52052861e-01 5.38962245e-01 -9.20858458e-02 -1.58323392e-01
-9.25629020e-01 7.38185942e-01 5.16269431e-02 -4.58308369e-01
-6.91338927e-02 1.04273438e+00 5.75258493e-01 -3.37891847e-01
-6.30873367e-02 -1.78201079e-01 -6.14881776e-02 2.61760857e-02
6.02526963e-01 3.88002843e-02 -1.12155564e-01 -9.37128812e-02
-1.40952975e-01 2.18813911e-01 -2.65882641e-01 -4.40793693e-01
1.03286052e+00 3.93987656e-01 -1.42231911e-01 8.91133621e-02
7.98982799e-01 5.33104241e-01 -9.39429641e-01 4.71697062e-01
-4.06543761e-01 -2.38915607e-01 -1.98363826e-01 -1.09688902e+00
-7.41249979e-01 9.13689196e-01 3.36475042e-03 2.12374002e-01
6.53713703e-01 -2.41638809e-01 8.00016582e-01 5.37990034e-01
3.07678968e-01 -9.34600651e-01 2.80614853e-01 4.45500106e-01
7.38789558e-01 -4.94729608e-01 -1.25957146e-01 -2.72213131e-01
-8.32748115e-01 7.80908525e-01 6.79697871e-01 2.20642835e-01
2.05878377e-01 5.76693416e-01 -3.56128290e-02 1.92523241e-01
-1.11635923e+00 1.94739237e-01 -2.67036166e-02 8.43002200e-01
6.21510148e-01 2.25021929e-01 -3.43033910e-01 4.25180197e-01
-5.55461586e-01 -3.57328892e-01 8.65176499e-01 5.56964517e-01
-4.26261872e-01 -1.47159183e+00 -3.14185739e-01 4.45254475e-01
-4.15845901e-01 -5.70699394e-01 -6.69457257e-01 6.06099308e-01
8.55503902e-02 1.20046341e+00 -1.22169696e-01 -3.79807919e-01
-5.81907555e-02 8.57491344e-02 4.21593338e-01 -9.33367908e-01
-4.60073918e-01 -9.01594236e-02 1.36376306e-01 -6.46764487e-02
5.82522094e-01 -4.21636015e-01 -1.26666403e+00 -6.80773079e-01
-5.09000599e-01 5.18387794e-01 3.38622659e-01 6.09001577e-01
4.18166339e-01 2.01359779e-01 5.93373239e-01 -5.99098921e-01
-5.55131674e-01 -7.76601136e-01 -4.72528964e-01 3.17641437e-01
-6.73205033e-02 -5.32747507e-01 -3.92594427e-01 1.04354739e-01]
|
[11.666237831115723, 9.120566368103027]
|
30df1ccf-6f6f-4273-a049-0646980402a8
|
assessing-word-importance-using-models
|
2305.19689
| null |
https://arxiv.org/abs/2305.19689v1
|
https://arxiv.org/pdf/2305.19689v1.pdf
|
Assessing Word Importance Using Models Trained for Semantic Tasks
|
Many NLP tasks require to automatically identify the most significant words in a text. In this work, we derive word significance from models trained to solve semantic task: Natural Language Inference and Paraphrase Identification. Using an attribution method aimed to explain the predictions of these models, we derive importance scores for each input token. We evaluate their relevance using a so-called cross-task evaluation: Analyzing the performance of one model on an input masked according to the other model's weight, we show that our method is robust with respect to the choice of the initial task. Additionally, we investigate the scores from the syntax point of view and observe interesting patterns, e.g. words closer to the root of a syntactic tree receive higher importance scores. Altogether, these observations suggest that our method can be used to identify important words in sentences without any explicit word importance labeling in training.
|
['François Yvon', 'Ondřej Bojar', 'Dávid Javorský']
|
2023-05-31
| null | null | null | null |
['paraphrase-identification']
|
['natural-language-processing']
|
[ 3.41493577e-01 4.21697050e-01 -1.63025945e-01 -5.45954645e-01
-5.77554643e-01 -6.03428781e-01 8.09877574e-01 9.49128926e-01
-7.78443396e-01 6.89330816e-01 6.26230776e-01 -3.61914605e-01
-1.25332043e-01 -5.85541070e-01 -6.28911078e-01 -4.71752346e-01
4.42470551e-01 5.01785696e-01 1.90557465e-01 -2.96222568e-01
7.92715490e-01 3.73278320e-01 -1.43524253e+00 5.14649808e-01
8.56480837e-01 6.38449192e-01 3.48930955e-01 4.14637446e-01
-4.85610276e-01 7.22949505e-01 -7.79339731e-01 -6.69778407e-01
-2.81152546e-01 -4.58028525e-01 -1.30986285e+00 -2.46757731e-01
2.94755667e-01 2.97947019e-01 4.42795753e-01 1.18597376e+00
1.52070373e-01 1.22015020e-02 9.05744970e-01 -7.68097639e-01
-5.26763260e-01 1.03370547e+00 -7.91105106e-02 4.89031106e-01
5.81935823e-01 -1.16919428e-01 1.74174452e+00 -1.13821638e+00
6.44876540e-01 1.14790773e+00 4.57232654e-01 3.18128765e-01
-1.41502225e+00 -1.07253142e-01 2.01070458e-01 4.89042878e-01
-9.90734637e-01 -3.36098343e-01 7.08393335e-01 -6.01546943e-01
9.74486351e-01 2.02493668e-01 2.61263818e-01 1.00908160e+00
2.28027726e-04 5.36439478e-01 1.20213366e+00 -8.31819177e-01
3.17354113e-01 3.68955225e-01 7.31659353e-01 4.95642692e-01
3.26457143e-01 -2.51782298e-01 -6.15078390e-01 -1.02696441e-01
-2.91464608e-02 -6.04110360e-01 -1.84191048e-01 -1.35229066e-01
-1.07523811e+00 9.29627955e-01 2.77120084e-01 8.68178546e-01
-3.93553376e-01 1.24003343e-01 4.53219771e-01 4.24298495e-01
5.34831643e-01 8.24044824e-01 -6.43306434e-01 -6.38967454e-02
-6.96182787e-01 1.65236816e-01 7.77201474e-01 5.09665310e-01
8.50861490e-01 -5.03151655e-01 -5.76331735e-01 8.57818663e-01
2.15479791e-01 -2.71291416e-02 6.68430090e-01 -7.41418660e-01
4.34159786e-01 5.40484250e-01 5.95766939e-02 -7.75473654e-01
-3.21003795e-01 -4.85787570e-01 -1.67641580e-01 -3.13661527e-03
6.14842474e-01 1.74204782e-01 -3.92826378e-01 1.95266342e+00
6.31507710e-02 -1.99802414e-01 2.14454085e-01 5.79386711e-01
3.87703300e-01 3.76375794e-01 4.15323466e-01 -2.62677163e-01
1.68882132e+00 -6.95756912e-01 -5.80331624e-01 -3.19855630e-01
7.67297208e-01 -8.37006807e-01 1.62762272e+00 1.74184516e-01
-9.46244478e-01 -6.02113426e-01 -8.02550077e-01 -2.01434195e-01
-3.65232110e-01 1.13825448e-01 1.02898985e-01 4.01936650e-01
-8.74663293e-01 9.38328683e-01 -2.52943337e-01 -4.45022076e-01
-1.37786210e-01 -4.02032472e-02 -1.57769904e-01 3.57944310e-01
-1.37601340e+00 1.41275203e+00 5.30104220e-01 -2.99273133e-01
-3.97388488e-01 -4.73732173e-01 -6.60499811e-01 4.69997913e-01
9.69964564e-02 -6.78060472e-01 1.21752024e+00 -1.17540288e+00
-1.16634822e+00 1.34341753e+00 -5.37267745e-01 -4.30517286e-01
2.99519032e-01 -2.08172604e-01 -5.58319464e-02 1.91662028e-01
4.51851398e-01 2.44227797e-01 7.85119116e-01 -1.00148666e+00
-4.29186314e-01 -3.49826425e-01 2.16166839e-01 -4.16886471e-02
-4.38989997e-01 3.28814834e-01 -1.17331557e-01 -8.66864264e-01
-1.12850450e-01 -7.41208851e-01 -4.35875393e-02 -1.54360309e-01
-2.90630549e-01 -6.90791309e-01 1.22893834e-02 -6.50960565e-01
1.28605378e+00 -2.02917171e+00 3.79236013e-01 3.52845639e-01
-2.63641495e-03 1.86499525e-02 -1.05661981e-01 3.74357581e-01
-3.93632412e-01 3.25068653e-01 -2.90265143e-01 -4.32714820e-01
2.59731948e-01 6.98135719e-02 -2.63596803e-01 1.32121267e-02
3.55048567e-01 8.57571363e-01 -8.70559096e-01 -5.65754294e-01
-2.33622473e-02 1.55406356e-01 -5.22036314e-01 1.63926736e-01
-2.16435760e-01 1.19496636e-01 -3.63555223e-01 8.10679048e-02
7.55393505e-02 -2.78434485e-01 2.49014989e-01 -2.64050096e-01
-2.32374892e-02 9.17074978e-01 -7.00364470e-01 1.35698426e+00
-7.80850351e-01 6.19363904e-01 -3.12206775e-01 -1.09152627e+00
8.21692705e-01 1.52311936e-01 -1.84069589e-01 -6.47234678e-01
1.99440513e-02 4.15068537e-01 2.03539908e-01 -6.64822817e-01
5.58194518e-01 -4.56762105e-01 -2.23309606e-01 5.25208771e-01
5.38385026e-02 1.09367944e-01 3.35039109e-01 1.43046319e-01
8.67338359e-01 -2.06588402e-01 6.44773841e-01 -6.57276988e-01
9.92720664e-01 -1.80198342e-01 -1.86605267e-02 6.94102466e-01
8.84165093e-02 3.53778183e-01 8.83521676e-01 -2.51246899e-01
-1.02616978e+00 -8.88147533e-01 -2.69966960e-01 1.33097899e+00
3.52488854e-03 -4.52783108e-01 -7.88800299e-01 -7.76529491e-01
-3.41237197e-03 1.31647921e+00 -9.41679955e-01 -2.90387452e-01
-5.80487490e-01 -5.07305443e-01 4.76390719e-01 4.42244977e-01
-1.79456636e-01 -1.30600166e+00 -5.25072694e-01 8.80898908e-02
-4.33116734e-01 -1.08399844e+00 -1.57499731e-01 3.03602219e-01
-7.12848246e-01 -8.43048453e-01 -4.02812779e-01 -7.46516228e-01
7.13462770e-01 -1.05306476e-01 1.34039164e+00 4.05864537e-01
8.06177557e-02 9.99672711e-02 -5.82621515e-01 -2.22083092e-01
-8.04293215e-01 3.71369720e-01 -1.66502416e-01 -7.99155533e-02
5.80613077e-01 -3.93107325e-01 -2.24072352e-01 3.74202803e-02
-7.57774174e-01 -1.50951087e-01 5.53110719e-01 6.52773380e-01
1.55033648e-01 -4.16756541e-01 5.35647929e-01 -8.31357181e-01
1.14758503e+00 -4.42988366e-01 -3.16076010e-01 4.80075449e-01
-7.97192037e-01 8.90157700e-01 7.88937330e-01 -1.38577208e-01
-7.44261682e-01 -2.48844668e-01 -4.10253614e-01 1.79979950e-01
-1.67304978e-01 6.60080850e-01 -5.89074008e-02 2.92984754e-01
7.01414466e-01 1.42068088e-01 -1.89988136e-01 -5.91811717e-01
2.88461804e-01 5.18245935e-01 2.22378254e-01 -8.40696514e-01
7.32203662e-01 8.49429816e-02 -1.89054504e-01 -6.36166513e-01
-1.25831270e+00 -4.54574674e-01 -5.70862770e-01 -1.18940540e-01
8.35702360e-01 -5.16312778e-01 -6.42379045e-01 -2.53027026e-02
-1.65239108e+00 -6.47625029e-02 -3.16009492e-01 3.27060997e-01
-5.01340687e-01 6.49551511e-01 -3.23089004e-01 -5.72636843e-01
-2.71289468e-01 -8.90437186e-01 1.12448871e+00 -1.45790905e-01
-9.07375872e-01 -1.19863462e+00 1.23082094e-01 3.71012509e-01
2.23227635e-01 -3.49468112e-01 1.49909937e+00 -1.00527787e+00
-1.26466289e-01 5.74439131e-02 -2.31073320e-01 4.14210826e-01
4.40680236e-02 -1.21508643e-01 -9.84593570e-01 1.94986299e-01
1.68115482e-01 -1.92301974e-01 1.06602824e+00 1.75208434e-01
1.12144458e+00 -4.25208449e-01 -6.29089177e-02 2.27833111e-02
1.26561558e+00 -3.80739123e-01 5.11066973e-01 5.02236724e-01
3.98186505e-01 1.11254084e+00 6.24150693e-01 1.87848136e-01
2.14609399e-01 8.31711829e-01 1.35789782e-01 2.54452974e-01
4.28031646e-02 -2.48370498e-01 5.15140712e-01 5.24529576e-01
5.21937720e-02 -1.38554752e-01 -7.98269629e-01 5.55797100e-01
-1.65541613e+00 -9.46520090e-01 -2.55268753e-01 2.33125591e+00
1.04895854e+00 4.60934639e-01 -4.85684983e-02 1.84186712e-01
6.87704027e-01 1.62213624e-01 3.06086894e-02 -7.75051653e-01
-9.79501978e-02 4.69842553e-01 1.42811850e-01 9.28226054e-01
-6.70890808e-01 9.86079752e-01 6.48096895e+00 8.07866991e-01
-8.59452665e-01 1.57943189e-01 3.81661683e-01 2.98323244e-01
-7.91109264e-01 1.78288564e-01 -7.05646574e-01 6.19833052e-01
9.98910904e-01 -3.63546193e-01 1.42852515e-01 7.63764620e-01
3.54940891e-01 -1.58117265e-01 -1.34205306e+00 3.86787683e-01
1.32797807e-01 -1.09237587e+00 3.99646252e-01 -1.81148350e-01
1.55851007e-01 -2.66883194e-01 -2.69505233e-01 1.11789465e-01
2.43737325e-02 -9.46947098e-01 8.46019208e-01 3.68386775e-01
2.31542110e-01 -5.03715336e-01 7.77985215e-01 3.69036943e-01
-8.61378431e-01 1.55813888e-01 -3.61985356e-01 -3.03715348e-01
1.20461516e-01 6.97821558e-01 -9.70311582e-01 1.79936126e-01
1.80934206e-01 2.99598128e-01 -8.72895002e-01 6.04945540e-01
-8.75449955e-01 6.20084584e-01 8.26159641e-02 -5.27024388e-01
1.27603754e-01 -5.98284863e-02 6.64055288e-01 1.43712175e+00
1.97134107e-01 -3.03016186e-01 -2.77383596e-01 1.09668696e+00
-9.67585966e-02 4.30196643e-01 -4.66210455e-01 1.38100967e-01
3.78981143e-01 1.12868297e+00 -6.21005833e-01 -5.45713961e-01
-2.70504862e-01 1.09765565e+00 8.04106772e-01 6.65150583e-02
-7.37794876e-01 -4.90367413e-01 6.22729063e-01 5.97752705e-02
2.91059792e-01 -1.31830588e-01 -4.73384082e-01 -1.18877566e+00
2.70302624e-01 -4.88666177e-01 1.82420924e-01 -9.31817412e-01
-1.31454420e+00 5.52836657e-01 2.00225203e-03 -7.48605371e-01
-4.53886777e-01 -8.36704791e-01 -7.08531618e-01 1.06827366e+00
-1.77456701e+00 -5.94432831e-01 -6.88980818e-02 2.11635813e-01
6.51204467e-01 1.69328868e-01 8.05867434e-01 -4.94461693e-03
-3.76657665e-01 4.75664049e-01 -1.93714708e-01 1.36056840e-01
5.66533327e-01 -1.41846013e+00 5.42179823e-01 9.75012183e-01
4.47097600e-01 8.88056517e-01 1.20108211e+00 -3.43918264e-01
-5.71167588e-01 -7.41492689e-01 1.96860194e+00 -6.24980032e-01
9.08210397e-01 -1.91979453e-01 -1.03234828e+00 3.85463327e-01
2.81636924e-01 -5.44162512e-01 7.66021729e-01 5.12461722e-01
-6.27882242e-01 1.30890474e-01 -7.32518137e-01 4.91922557e-01
8.97768199e-01 -7.59053290e-01 -1.37855613e+00 3.41700882e-01
6.61854684e-01 1.98901281e-01 -5.59975207e-01 3.78035940e-02
3.85312438e-01 -9.04236615e-01 8.88704300e-01 -9.16641772e-01
8.62246752e-01 -9.98692065e-02 -6.50076345e-02 -1.41594422e+00
-4.02890414e-01 -1.23673538e-02 2.56474823e-01 1.17159081e+00
8.81415725e-01 -5.32465696e-01 3.71764928e-01 5.81803203e-01
-5.93107380e-02 -3.93270314e-01 -8.98658395e-01 -6.69359386e-01
1.47927508e-01 -3.70996386e-01 2.62227982e-01 9.07599866e-01
3.68152529e-01 9.29509878e-01 -1.23609034e-02 8.49051103e-02
3.98642898e-01 1.46611392e-01 3.26002687e-01 -1.42040586e+00
-4.55487460e-01 -6.39918506e-01 -1.34118229e-01 -6.62744105e-01
8.03384244e-01 -1.12801313e+00 8.40650871e-02 -1.35172355e+00
2.73000449e-01 -1.36745796e-01 -3.33474815e-01 4.71206099e-01
-5.97046494e-01 1.43710077e-01 3.17856044e-01 1.71991408e-01
-3.56907636e-01 3.66198361e-01 6.63033307e-01 -2.18995977e-02
1.58660024e-01 -5.54230548e-02 -9.32701111e-01 9.07640457e-01
8.52877975e-01 -6.54092371e-01 -1.07528947e-01 -4.94986266e-01
6.74071848e-01 -3.73797327e-01 5.38588881e-01 -6.30690753e-01
-6.71439469e-02 -2.28312835e-01 -2.62440573e-02 6.84839040e-02
4.59237956e-02 -7.66541421e-01 -2.95641541e-01 6.01135314e-01
-9.30964112e-01 1.57084495e-01 -5.25335521e-02 2.38420025e-01
-2.20584497e-01 -1.06695509e+00 7.69532442e-01 -1.15173690e-01
-7.28766024e-01 -4.92735475e-01 -4.85283405e-01 2.17860743e-01
7.78513968e-01 -1.55432403e-01 -1.83472112e-01 -1.65914491e-01
-8.00509691e-01 -7.26700798e-02 3.82551640e-01 4.94298965e-01
4.07687008e-01 -1.07445836e+00 -7.65457511e-01 -1.39588594e-01
4.96546447e-01 -6.77832484e-01 -2.16754258e-01 7.29131758e-01
-2.74262309e-01 4.06363964e-01 -3.31364609e-02 -2.28391692e-01
-1.48636699e+00 5.89104116e-01 8.20372999e-02 -5.43737173e-01
-1.80936828e-01 6.23806894e-01 8.67833421e-02 -1.74109280e-01
1.68083590e-02 -6.44511998e-01 -5.19572854e-01 4.60509211e-01
4.43963706e-01 9.72234085e-02 2.31875598e-01 -6.50582671e-01
-4.99013633e-01 6.65266216e-01 -1.34503515e-02 -2.98152566e-01
1.11150646e+00 -1.21626504e-01 -3.23786855e-01 6.28007710e-01
1.37020791e+00 3.50388050e-01 -4.47951496e-01 -2.39227861e-01
6.20018005e-01 -3.23157459e-01 -2.49829113e-01 -6.05669439e-01
-5.12679875e-01 9.20529127e-01 -2.78559625e-02 3.06197941e-01
7.15033293e-01 2.90795118e-01 3.62813354e-01 5.60934484e-01
7.46692419e-02 -1.19617200e+00 1.21618032e-01 7.33333528e-01
9.03708398e-01 -1.05066490e+00 -1.39546812e-01 -4.29684907e-01
-6.29888594e-01 1.24387372e+00 2.11428925e-01 2.46181265e-02
1.51023835e-01 -1.76522076e-01 -1.33094788e-01 -2.43505955e-01
-7.61535823e-01 -3.66715878e-01 3.35386246e-01 1.61566123e-01
7.71303773e-01 -9.63015333e-02 -9.99780416e-01 4.59778696e-01
-4.67338115e-01 -3.10754091e-01 5.04560530e-01 4.75134492e-01
-7.04551637e-01 -1.26229906e+00 -1.93122864e-01 2.56094933e-01
-4.95286167e-01 -5.72674990e-01 -8.02933276e-01 4.22022045e-01
-1.71722904e-01 7.81669438e-01 4.80049253e-02 -9.04479995e-02
3.30407977e-01 5.03103673e-01 3.42948228e-01 -8.00837457e-01
-7.85069883e-01 -5.58803022e-01 4.28566068e-01 -2.62312263e-01
-5.01410484e-01 -5.77212334e-01 -1.20690012e+00 -6.41829148e-02
-1.90918460e-01 6.21382296e-01 6.53075695e-01 1.33707273e+00
2.12825194e-01 3.55660230e-01 4.72333312e-01 -3.99217039e-01
-7.90991187e-01 -9.91302490e-01 -2.76927292e-01 7.78636217e-01
2.14135110e-01 -4.56390202e-01 -5.76877177e-01 5.55285700e-02]
|
[10.921977043151855, 9.000479698181152]
|
99731909-de5a-4d88-ab44-a2e65369aefc
|
learned-harmonic-mean-estimation-of-the
|
2307.00048
| null |
https://arxiv.org/abs/2307.00048v1
|
https://arxiv.org/pdf/2307.00048v1.pdf
|
Learned harmonic mean estimation of the marginal likelihood with normalizing flows
|
Computing the marginal likelihood (also called the Bayesian model evidence) is an important task in Bayesian model selection, providing a principled quantitative way to compare models. The learned harmonic mean estimator solves the exploding variance problem of the original harmonic mean estimation of the marginal likelihood. The learned harmonic mean estimator learns an importance sampling target distribution that approximates the optimal distribution. While the approximation need not be highly accurate, it is critical that the probability mass of the learned distribution is contained within the posterior in order to avoid the exploding variance problem. In previous work a bespoke optimization problem is introduced when training models in order to ensure this property is satisfied. In the current article we introduce the use of normalizing flows to represent the importance sampling target distribution. A flow-based model is trained on samples from the posterior by maximum likelihood estimation. Then, the probability density of the flow is concentrated by lowering the variance of the base distribution, i.e. by lowering its "temperature", ensuring its probability mass is contained within the posterior. This approach avoids the need for a bespoke optimisation problem and careful fine tuning of parameters, resulting in a more robust method. Moreover, the use of normalizing flows has the potential to scale to high dimensional settings. We present preliminary experiments demonstrating the effectiveness of the use of flows for the learned harmonic mean estimator. The harmonic code implementing the learned harmonic mean, which is publicly available, has been updated to now support normalizing flows.
|
['Jason D. McEwen', 'Alessio Spurio Mancini', 'Matthew A. Price', 'Alicja Polanska']
|
2023-06-30
| null | null | null | null |
['model-selection']
|
['methodology']
|
[ 8.70956481e-02 1.53388456e-01 -2.60464787e-01 -2.91572601e-01
-9.67106044e-01 -5.08170426e-01 6.15724564e-01 1.44973591e-01
-5.08581221e-01 8.19166660e-01 -2.93912981e-02 -2.93883622e-01
-5.59078515e-01 -7.48707116e-01 -6.50882959e-01 -1.01847839e+00
-7.41436109e-02 6.50577843e-01 4.83460158e-01 4.66086537e-01
5.87573707e-01 5.64953685e-01 -1.63100743e+00 -3.68009031e-01
7.30699122e-01 9.61057484e-01 2.60258198e-01 8.94895494e-01
-8.45406204e-02 1.52761787e-01 -6.43597364e-01 -2.96161115e-01
3.21679145e-01 -6.83188796e-01 -5.33495069e-01 -2.02290267e-01
5.47271013e-01 -3.66805464e-01 3.87849987e-01 1.09056723e+00
3.71818990e-01 5.23343086e-01 1.22958863e+00 -1.19002652e+00
1.07515216e-01 2.46023804e-01 -4.84517783e-01 3.20650846e-01
1.14454841e-03 -1.74644534e-02 1.02061963e+00 -6.09690130e-01
4.48720396e-01 1.27632141e+00 5.83643854e-01 3.41317534e-01
-1.53702354e+00 -5.31495333e-01 -5.71621768e-02 -1.19317412e-01
-1.66619897e+00 -4.63831902e-01 4.19760495e-01 -5.61913550e-01
7.29075611e-01 1.55647874e-01 7.43064880e-01 5.56197226e-01
4.00970012e-01 4.53612536e-01 7.53434956e-01 -4.34165686e-01
6.65309608e-01 5.76332033e-01 -1.73457131e-01 6.49327874e-01
4.33404773e-01 1.89943612e-01 -4.65189189e-01 -4.94855225e-01
8.06250155e-01 -5.49915433e-01 -3.70751143e-01 -6.77930415e-01
-6.14888787e-01 1.04023826e+00 -3.60862985e-02 1.50548249e-01
-7.79086798e-02 4.13534105e-01 1.98315829e-01 -2.40898788e-01
3.63696337e-01 4.42402333e-01 -3.10965866e-01 -3.98719609e-01
-1.25351167e+00 6.06203377e-01 1.04706943e+00 6.55001402e-01
7.84752488e-01 -1.35470539e-01 5.69981448e-02 6.29217207e-01
9.07368124e-01 6.62548125e-01 4.96300757e-02 -1.44317150e+00
2.46938303e-01 1.70525193e-01 4.16644275e-01 -8.44940245e-01
-5.93765751e-02 -4.72217470e-01 -3.86316270e-01 4.38459247e-01
1.02140284e+00 -9.74878669e-02 -5.62410593e-01 1.84285355e+00
5.39650440e-01 1.90163270e-01 -1.98897973e-01 7.40767419e-01
-9.54191312e-02 7.20843375e-01 1.66910604e-01 -3.45445037e-01
1.10906756e+00 -2.44236261e-01 -5.14728725e-01 -1.09791704e-01
2.37637937e-01 -7.96325386e-01 8.62410426e-01 5.26273131e-01
-1.00500405e+00 -1.80808395e-01 -1.08694863e+00 3.04725975e-01
-5.69169484e-02 -2.49530837e-01 4.65984851e-01 9.35083747e-01
-6.23942077e-01 8.07752669e-01 -1.06520188e+00 -3.41524124e-01
2.22882301e-01 2.75277972e-01 -1.73371702e-01 3.21634203e-01
-9.51586545e-01 1.04962969e+00 4.39142466e-01 2.59247981e-02
-7.95443058e-01 -8.91593575e-01 -8.87439191e-01 3.17611486e-01
1.06700622e-01 -5.28946102e-01 1.23360801e+00 -6.78233683e-01
-1.69033420e+00 2.69717813e-01 -2.63921112e-01 -3.86160344e-01
7.41045773e-01 -1.52359931e-02 2.51933575e-01 2.19462886e-01
-8.74540284e-02 5.80560744e-01 1.10576451e+00 -1.07585359e+00
-5.66769540e-01 -3.82015333e-02 -2.73716629e-01 3.04833055e-02
-1.76615104e-01 -3.39521646e-01 -3.58102649e-01 -3.36450636e-01
4.26658541e-02 -8.28127384e-01 -1.59241483e-01 9.49315578e-02
-8.50395486e-02 -1.74060181e-01 6.16765738e-01 -4.78825212e-01
1.36744845e+00 -2.06960106e+00 -1.28656611e-01 6.85742974e-01
-3.14687863e-02 -8.74333680e-02 3.41761649e-01 1.67517021e-01
2.45212361e-01 1.47357807e-01 -7.22474277e-01 -3.11240554e-01
1.32994890e-01 1.30206242e-01 -3.23423862e-01 8.87038708e-01
2.52511412e-01 1.28113732e-01 -8.99996102e-01 -5.80008566e-01
3.58495235e-01 6.72599792e-01 -7.99575627e-01 2.47200742e-01
-1.65594578e-01 2.79343128e-01 -3.38342607e-01 -7.45064914e-02
7.75654733e-01 -9.07691289e-03 1.17732689e-01 -7.31967688e-02
-1.99839801e-01 2.96685964e-01 -1.86253452e+00 1.14614499e+00
-4.55833822e-01 5.50842166e-01 1.55262500e-01 -8.41375470e-01
1.03575873e+00 1.31510600e-01 6.19939506e-01 -8.40685591e-02
5.70679642e-02 3.43461812e-01 -6.53198408e-03 -2.19125569e-01
4.29266751e-01 -6.71761274e-01 1.62641317e-01 6.09097898e-01
1.45558342e-01 -5.91960132e-01 3.13593239e-01 2.59249449e-01
4.68027383e-01 4.23514128e-01 3.34284991e-01 -6.84643328e-01
4.45242643e-01 -1.90619767e-01 5.52708268e-01 8.78295243e-01
-1.35718018e-01 5.70348144e-01 8.49349737e-01 7.77232274e-02
-1.07896078e+00 -1.30811274e+00 -6.85168982e-01 5.33917844e-01
-3.18538025e-02 -4.47061449e-01 -8.09036434e-01 -6.39735997e-01
5.91939688e-02 9.74106491e-01 -4.59394306e-01 -1.90377504e-01
-4.20913219e-01 -7.88008928e-01 2.10229531e-01 3.67037714e-01
1.94658622e-01 -5.23987830e-01 -8.38051260e-01 2.70683467e-01
-2.53634099e-02 -5.71475565e-01 -3.54674965e-01 3.10118675e-01
-8.76899242e-01 -1.06845999e+00 -6.60971403e-01 -9.98986885e-02
5.72349310e-01 -3.53749633e-01 9.26192343e-01 -2.12783471e-01
-1.08958855e-01 4.11565483e-01 8.82946700e-02 -4.89120424e-01
-4.76964206e-01 -1.70452911e-02 -2.14297306e-02 -1.68921977e-01
2.61338741e-01 -3.09615731e-01 -4.99794126e-01 3.73969406e-01
-8.67942274e-01 -3.54210645e-01 1.01364091e-01 7.30119169e-01
5.12397707e-01 3.27993959e-01 5.49534321e-01 -5.11139154e-01
4.34530973e-01 -3.20945561e-01 -1.11419010e+00 -7.16184005e-02
-6.38768375e-01 5.28050244e-01 4.30575401e-01 -3.73392940e-01
-1.20541286e+00 1.16698928e-01 -7.46950731e-02 -1.82526097e-01
-4.08874378e-02 3.64738166e-01 -5.62818609e-02 2.33518273e-01
5.71478248e-01 -3.82964820e-01 1.74061179e-01 -4.18167263e-01
3.38065684e-01 5.47062099e-01 4.10717249e-01 -6.63456261e-01
6.11644804e-01 5.03700614e-01 3.84633809e-01 -8.48099113e-01
-7.44636655e-01 -3.85895312e-01 -4.84674931e-01 -3.54927093e-01
7.26543725e-01 -2.61309266e-01 -7.86369085e-01 1.02881223e-01
-8.04637671e-01 -3.75067234e-01 -4.57711667e-01 9.27045345e-01
-9.06486988e-01 3.54368955e-01 -1.51137277e-01 -1.32256496e+00
3.37668099e-02 -1.14404535e+00 9.36714113e-01 3.62134367e-01
-5.47978163e-01 -1.23315132e+00 2.60045201e-01 -1.17223196e-01
3.36240053e-01 5.13456501e-02 9.36106145e-01 -5.49013555e-01
-5.56751072e-01 -3.59987140e-01 7.09688067e-02 4.06954020e-01
-3.24587859e-02 5.79332113e-01 -9.56276715e-01 -2.13907227e-01
1.14451967e-01 8.36314410e-02 9.40416276e-01 8.86587322e-01
7.76010811e-01 -8.13515186e-02 -2.91171879e-01 2.32713416e-01
1.31958997e+00 2.61026230e-02 3.79764825e-01 2.28109166e-01
1.70418501e-01 8.23990643e-01 6.23606086e-01 6.21630073e-01
5.75856604e-02 5.53943098e-01 1.13225825e-01 4.37940508e-01
6.19815812e-02 -2.62275994e-01 3.22344273e-01 2.97950834e-01
3.33904207e-01 -8.16530883e-02 -8.74050498e-01 4.42680031e-01
-1.61457133e+00 -1.06499064e+00 1.59312319e-03 2.75843382e+00
8.02784383e-01 4.76653218e-01 2.68507332e-01 1.75204635e-01
5.69263041e-01 -8.20787027e-02 -3.81176233e-01 -5.32430410e-01
4.27015275e-01 9.03166384e-02 4.74246323e-01 1.13131893e+00
-9.87760544e-01 4.89240289e-01 6.48048782e+00 9.37161148e-01
-8.83411109e-01 -1.66430548e-01 3.69860202e-01 -6.51221275e-02
-3.56634647e-01 4.04904336e-01 -1.21505094e+00 6.32178843e-01
1.17089915e+00 -2.31692418e-01 2.27890193e-01 8.04351270e-01
5.00262856e-01 -9.10216212e-01 -9.89118099e-01 5.46268702e-01
-2.48005420e-01 -9.22466874e-01 -3.41217250e-01 2.57820278e-01
3.14134717e-01 -3.54906976e-01 -7.68625811e-02 8.42870697e-02
8.98629576e-02 -7.60492027e-01 8.69571805e-01 7.94664145e-01
4.98571396e-01 -1.08049488e+00 6.90786839e-01 4.00851309e-01
-1.16696072e+00 1.15164690e-01 -4.47890878e-01 1.83220744e-01
5.47618747e-01 1.01650953e+00 -1.04358554e+00 -2.71417480e-02
5.05795717e-01 3.35979015e-01 -1.61691979e-01 1.30371404e+00
-1.63081825e-01 6.92245126e-01 -9.14906085e-01 -1.14083670e-01
-4.71927691e-03 -4.02217388e-01 8.30285013e-01 1.28867090e+00
3.59040409e-01 -4.90236163e-01 4.31167707e-02 1.08857167e+00
4.03621346e-01 6.70458376e-02 -3.54735821e-01 -4.15355638e-02
5.21551728e-01 1.05904448e+00 -1.02260232e+00 -3.35149348e-01
-6.03137389e-02 2.44749978e-01 3.93365994e-02 2.76404262e-01
-7.17096508e-01 -5.88652909e-01 5.81779361e-01 9.44736525e-02
4.26297247e-01 -1.57460004e-01 -2.90242553e-01 -7.80600011e-01
-1.44759610e-01 -5.16550183e-01 4.88817573e-01 -3.61560404e-01
-1.21155477e+00 1.55049548e-01 6.99096918e-01 -9.72131789e-01
-7.37239361e-01 -5.97646415e-01 -5.19082844e-01 1.14412308e+00
-1.15371299e+00 -2.52622098e-01 1.24921091e-02 1.84686646e-01
1.59517348e-01 1.65278926e-01 6.21907234e-01 1.03281610e-01
-4.45776969e-01 3.43314499e-01 1.28675461e-01 -3.81316602e-01
7.71647334e-01 -1.39799917e+00 -1.19265579e-01 7.13861167e-01
-1.59623668e-01 7.96081245e-01 1.21151757e+00 -8.14374208e-01
-8.75471115e-01 -5.85704625e-01 7.55737543e-01 -3.92896891e-01
6.53458476e-01 -3.53036314e-01 -1.06239080e+00 3.89455438e-01
-2.02335492e-01 -2.78596431e-01 5.73231399e-01 -1.05292760e-01
-1.08903185e-01 -1.83255672e-01 -1.31652355e+00 4.29412037e-01
2.06100151e-01 -2.56922692e-01 -3.69717568e-01 1.10101394e-01
1.64646551e-01 -1.56693861e-01 -9.14718151e-01 4.33846056e-01
8.32371652e-01 -9.47856903e-01 6.99127436e-01 -2.04923704e-01
-4.69548889e-02 -5.59480846e-01 -1.29714459e-01 -1.11065924e+00
-4.10143174e-02 -5.74375093e-01 -3.42088878e-01 1.25965440e+00
4.48191941e-01 -6.55967712e-01 7.98074961e-01 7.82188416e-01
2.62828887e-01 -7.85649359e-01 -1.12354982e+00 -7.92357326e-01
1.31777033e-01 -6.74968064e-01 2.72087008e-01 3.47748548e-01
-1.07911773e-01 1.04625992e-01 -9.96329412e-02 1.65361807e-01
9.35187459e-01 -1.89275965e-01 7.23314226e-01 -1.30387533e+00
-6.16715670e-01 -6.43174767e-01 -3.07262272e-01 -1.06685829e+00
9.26612988e-02 -7.06198335e-01 3.54330778e-01 -1.37324882e+00
1.40194997e-01 -3.67536515e-01 5.23330346e-02 -5.21427952e-02
-9.57851112e-02 -7.10402802e-02 1.25808731e-01 -6.83096750e-03
-1.91002339e-01 4.92473602e-01 1.05806291e+00 2.15327159e-01
-3.12715203e-01 1.88749284e-01 -3.52347553e-01 7.95889854e-01
7.94205368e-01 -8.25655043e-01 -4.99593526e-01 2.60289252e-01
2.87272811e-01 -1.84387311e-01 2.73329377e-01 -7.80174434e-01
-2.15969626e-02 -1.40851289e-01 4.07325834e-01 -5.13205051e-01
4.02986795e-01 -7.92987227e-01 1.68020248e-01 4.55308527e-01
-3.42744648e-01 -2.78813362e-01 1.65666804e-01 5.88069201e-01
1.72042921e-01 -1.00418937e+00 1.24174047e+00 8.31190031e-03
-8.85350630e-02 -7.88462609e-02 -6.87314510e-01 3.27098258e-02
8.40460896e-01 -1.52334973e-01 1.05239153e-01 -6.11371636e-01
-3.81326407e-01 3.24000530e-02 3.71046931e-01 -6.61728084e-02
3.30719501e-01 -9.01356936e-01 -4.74526733e-01 1.90426812e-01
-2.20889434e-01 -1.62512250e-03 -2.48687845e-02 9.43314910e-01
-4.30963159e-01 1.90359607e-01 2.95691103e-01 -7.84360707e-01
-8.03022027e-01 2.55258054e-01 6.72880709e-01 -1.50749266e-01
-4.73728448e-01 6.74532533e-01 -5.12383655e-02 -2.96298504e-01
2.56807685e-01 -1.31222889e-01 3.96684334e-02 9.73909274e-02
6.68744147e-01 6.85685813e-01 -1.12109967e-01 -2.84152269e-01
-4.78899866e-01 5.10821879e-01 8.44972357e-02 -7.16351569e-01
1.16827142e+00 -1.07179686e-01 -1.02319337e-01 6.18838906e-01
1.25303197e+00 1.78709954e-01 -1.58710134e+00 2.73269087e-01
1.31095991e-01 -6.16054893e-01 2.16864362e-01 -5.00120401e-01
-7.08890200e-01 8.81867409e-01 5.28589249e-01 2.65047044e-01
7.28682756e-01 -2.49888208e-02 3.27615649e-01 9.34209898e-02
1.11024395e-01 -1.14522994e+00 -2.08240733e-01 4.24400538e-01
6.10614657e-01 -8.92044425e-01 3.46968025e-01 -2.11076513e-01
-4.78998572e-01 1.20281589e+00 4.50718313e-01 -1.36715293e-01
9.10592914e-01 5.11319697e-01 -1.09633520e-01 7.51178265e-02
-4.47568804e-01 -2.06098035e-02 4.04572606e-01 4.96569544e-01
4.13518935e-01 -1.69302821e-01 -3.99877161e-01 1.96895167e-01
-3.55118752e-01 -3.39247614e-01 4.24168736e-01 8.12782288e-01
-7.01807141e-01 -1.04563987e+00 -5.21954417e-01 4.29713696e-01
-4.64243889e-01 1.31869450e-01 3.92183010e-03 6.90869927e-01
-5.85451163e-02 9.89959002e-01 4.16738331e-01 2.31038883e-01
1.29452094e-01 2.43683338e-01 5.14353931e-01 -2.92030841e-01
-9.73444805e-02 4.98347968e-01 1.17444284e-02 -4.05530483e-01
-3.01195413e-01 -7.74572968e-01 -1.26718044e+00 -2.37798363e-01
-5.57648361e-01 5.04170477e-01 8.95126939e-01 9.87274885e-01
-7.85469487e-02 1.29325435e-01 2.95807809e-01 -8.85975182e-01
-7.77205348e-01 -8.66583586e-01 -7.31411994e-01 8.90662372e-02
3.42335194e-01 -9.29093301e-01 -8.79611075e-01 -8.95712301e-02]
|
[6.697727203369141, 3.8521478176116943]
|
c97eb3e4-9aa0-403a-a5d2-345c81b142c9
|
learning-through-structure-towards-deep
|
2109.10376
| null |
https://arxiv.org/abs/2109.10376v1
|
https://arxiv.org/pdf/2109.10376v1.pdf
|
Learning through structure: towards deep neuromorphic knowledge graph embeddings
|
Computing latent representations for graph-structured data is an ubiquitous learning task in many industrial and academic applications ranging from molecule synthetization to social network analysis and recommender systems. Knowledge graphs are among the most popular and widely used data representations related to the Semantic Web. Next to structuring factual knowledge in a machine-readable format, knowledge graphs serve as the backbone of many artificial intelligence applications and allow the ingestion of context information into various learning algorithms. Graph neural networks attempt to encode graph structures in low-dimensional vector spaces via a message passing heuristic between neighboring nodes. Over the recent years, a multitude of different graph neural network architectures demonstrated ground-breaking performances in many learning tasks. In this work, we propose a strategy to map deep graph learning architectures for knowledge graph reasoning to neuromorphic architectures. Based on the insight that randomly initialized and untrained (i.e., frozen) graph neural networks are able to preserve local graph structures, we compose a frozen neural network with shallow knowledge graph embedding models. We experimentally show that already on conventional computing hardware, this leads to a significant speedup and memory reduction while maintaining a competitive performance level. Moreover, we extend the frozen architecture to spiking neural networks, introducing a novel, event-based and highly sparse knowledge graph embedding algorithm that is suitable for implementation in neuromorphic hardware.
|
['Dominik Dold', 'Thomas Runkler', 'Marcel Hildebrandt', 'Victor Caceres Chian']
|
2021-09-21
| null | null | null | null |
['knowledge-graph-embeddings', 'knowledge-graph-embeddings']
|
['graphs', 'methodology']
|
[ 1.68406740e-01 4.26648319e-01 -8.40370283e-02 -9.65796337e-02
2.97098964e-01 -5.31263351e-01 5.85445583e-01 6.68312430e-01
-4.01090950e-01 6.03899419e-01 -5.91792166e-02 -2.07358241e-01
-4.38783586e-01 -1.39075029e+00 -6.97298586e-01 -7.98869789e-01
-2.01546580e-01 5.40010631e-01 3.98297846e-01 -3.16715211e-01
3.49127986e-02 7.21663296e-01 -1.77986813e+00 1.97390214e-01
4.15724725e-01 7.45406449e-01 1.80411458e-01 4.58416939e-01
-4.04472619e-01 8.81221175e-01 -2.86058158e-01 -5.11084080e-01
5.67846335e-02 -2.87373960e-01 -6.53164506e-01 -2.80570894e-01
3.09146136e-01 3.84423733e-01 -8.79053533e-01 1.24066925e+00
1.44944236e-01 2.80500531e-01 4.13124472e-01 -1.22233880e+00
-8.13997567e-01 9.10950541e-01 1.89185917e-01 1.71175763e-01
5.95450588e-02 -1.36380300e-01 9.83159065e-01 -4.70652610e-01
1.00530875e+00 9.83816862e-01 4.94430602e-01 4.61614609e-01
-1.43052995e+00 -3.51104498e-01 1.99787989e-01 5.48702955e-01
-1.20891941e+00 -1.17335737e-01 8.19471717e-01 -3.77896190e-01
1.31074882e+00 8.74554664e-02 1.24861395e+00 9.49119151e-01
6.06191337e-01 1.71682060e-01 9.68944609e-01 -3.73230964e-01
7.07656682e-01 -1.43616404e-02 5.56858838e-01 1.11150992e+00
7.43895531e-01 -1.58008039e-01 -7.82556772e-01 -1.56294644e-01
6.03264511e-01 3.34631681e-01 -2.50873983e-01 -7.59676814e-01
-1.07722235e+00 9.60656643e-01 9.21720028e-01 6.25463188e-01
-3.69042099e-01 6.17537498e-01 3.73787731e-01 3.93294066e-01
1.85668662e-01 3.91445488e-01 6.41787890e-03 1.86998129e-01
-4.67725784e-01 -9.39163193e-02 1.11448920e+00 5.90864718e-01
1.06637561e+00 3.52732450e-01 1.74398839e-01 2.79707879e-01
3.32852125e-01 2.87007838e-01 8.09113503e-01 -4.57003534e-01
-1.83366075e-01 1.11132312e+00 -6.87609792e-01 -1.37400424e+00
-5.03177762e-01 -5.15458584e-01 -9.42929268e-01 1.97947949e-01
3.60123180e-02 4.49363261e-01 -7.80161083e-01 1.63944054e+00
1.95148915e-01 2.88524926e-01 2.24358857e-01 6.55013263e-01
6.21782601e-01 4.60636258e-01 -3.39583568e-02 -8.38602185e-02
1.23607469e+00 -4.90692705e-01 -5.85852206e-01 -1.06314145e-01
6.43576145e-01 3.91430892e-02 6.82898462e-01 5.84255755e-01
-7.05998719e-01 -2.06485108e-01 -1.16585088e+00 -1.67325616e-01
-1.03230774e+00 -5.03303349e-01 1.17486751e+00 6.87281847e-01
-1.56341410e+00 9.39139783e-01 -8.85241866e-01 -6.94958806e-01
5.24988115e-01 6.24624491e-01 -7.85953939e-01 4.01949603e-03
-1.00721395e+00 6.47049665e-01 8.51748288e-01 1.05915159e-01
-5.47853172e-01 -1.85045421e-01 -8.80840480e-01 3.64839315e-01
2.20891073e-01 -8.11517298e-01 4.11216408e-01 -8.71772707e-01
-1.58993745e+00 9.26343083e-01 1.58954442e-01 -9.41182494e-01
-2.50666052e-01 3.65036428e-01 -2.89854348e-01 2.58027852e-01
-5.91974378e-01 4.10098583e-01 9.10617173e-01 -7.17063785e-01
1.33363914e-03 -6.61265194e-01 7.66910762e-02 -2.53311723e-01
-7.32017696e-01 -5.38537920e-01 -1.82113394e-01 -3.78373593e-01
4.34330553e-01 -9.56173718e-01 -2.34019116e-01 4.96183708e-02
5.70861483e-03 -8.57178569e-02 7.96430707e-01 -1.49738878e-01
9.44912910e-01 -2.03112626e+00 9.12095666e-01 5.40991724e-01
1.03450108e+00 3.91041160e-01 -1.20209074e-02 5.48239112e-01
-7.81568214e-02 -6.78536817e-02 -1.88624337e-01 1.63072154e-01
1.19492479e-01 5.30035853e-01 -2.99857140e-01 4.57309157e-01
-4.06626537e-02 1.34750986e+00 -1.04392600e+00 -1.67020038e-02
2.61413395e-01 6.91733658e-01 -6.23567283e-01 -1.02227733e-01
-5.04130721e-01 -1.38326868e-01 -3.46649855e-01 5.08656576e-02
1.10871702e-01 -5.36831200e-01 7.00085700e-01 -1.16622359e-01
1.22479156e-01 8.18257779e-02 -9.56929147e-01 1.96829021e+00
-2.96039253e-01 7.86751509e-01 -2.03316331e-01 -1.35244095e+00
1.15385985e+00 -3.75663824e-02 3.35966289e-01 -6.15983307e-01
2.69697100e-01 2.03057751e-01 2.09981743e-02 7.34162405e-02
6.24056399e-01 -1.95485428e-02 9.72848013e-02 4.05437469e-01
5.41219473e-01 4.38920371e-02 1.54662654e-01 4.38335031e-01
1.52399445e+00 -1.08744919e-01 3.93872827e-01 -3.54126662e-01
3.99758756e-01 -1.03178442e-01 9.83154178e-02 4.80796576e-01
4.23892550e-02 1.98290106e-02 6.70276701e-01 -7.22142756e-01
-6.09959722e-01 -1.18693757e+00 3.29744071e-01 9.17972267e-01
-9.60851759e-02 -7.30635583e-01 -7.99997747e-01 -2.71415591e-01
-1.09114610e-02 2.87857682e-01 -7.06425786e-01 -7.43645608e-01
-2.86914140e-01 -3.75144690e-01 4.71522450e-01 2.48907253e-01
1.98779494e-01 -1.08831418e+00 -6.18892670e-01 3.93579751e-01
5.54446816e-01 -9.26890731e-01 2.75979966e-01 7.50735700e-01
-1.06132090e+00 -1.05543172e+00 -3.05697709e-01 -9.18896794e-01
6.99298859e-01 1.63340166e-01 9.71699536e-01 2.57920235e-01
-5.66954494e-01 6.57554805e-01 -3.69668365e-01 -1.51778445e-01
-4.24647063e-01 3.11957747e-01 2.03294396e-01 1.26562193e-01
3.22888553e-01 -1.07764304e+00 -3.52599025e-01 -2.71187901e-01
-1.15520430e+00 5.02383858e-02 4.19443995e-01 7.88116097e-01
9.12316501e-01 -8.55848715e-02 4.01348740e-01 -1.07603145e+00
5.17966807e-01 -4.98649567e-01 -9.44399297e-01 3.04849595e-01
-7.01004744e-01 7.12952912e-01 8.12132299e-01 -3.92810315e-01
-3.26741517e-01 1.78462267e-01 7.29426369e-02 -5.41564107e-01
9.74746495e-02 7.31441677e-01 -1.83579996e-02 -5.64032495e-01
8.10963392e-01 4.36376840e-01 1.38535619e-01 -1.02397740e-01
6.61437035e-01 1.13931812e-01 3.54280442e-01 -2.69861013e-01
6.85896158e-01 5.21707296e-01 4.69076037e-01 -1.03749692e+00
-3.96014184e-01 -2.48390347e-01 -5.74425519e-01 -2.17537761e-01
8.89669120e-01 -4.45144951e-01 -9.90832686e-01 2.39134803e-01
-9.90093052e-01 -3.64054799e-01 -5.90152502e-01 3.57639313e-01
-6.25101209e-01 2.79329062e-01 -5.64065039e-01 -4.63327765e-01
-3.17987531e-01 -8.15181971e-01 7.54562438e-01 2.27606565e-01
1.31739244e-01 -1.31144881e+00 3.12542707e-01 -1.92826331e-01
5.41114867e-01 3.31209004e-01 1.36451352e+00 -6.84319139e-01
-9.36119974e-01 -4.30455655e-02 -4.23055179e-02 -7.59249106e-02
-4.12133634e-02 -3.14759493e-01 -9.40564871e-01 -3.01526099e-01
-2.06698611e-01 -2.01297745e-01 1.20407259e+00 -7.96936974e-02
9.87140715e-01 -2.19928548e-01 -4.18603569e-01 7.09376454e-01
1.56789064e+00 -2.10325241e-01 6.65163457e-01 3.94448265e-02
9.07206714e-01 4.31283563e-01 -5.25161922e-01 1.74895227e-01
2.21282139e-01 5.55774450e-01 6.26464128e-01 4.87426400e-01
-1.86761618e-01 -3.05494696e-01 3.74299586e-01 1.40620172e+00
-1.08859554e-01 -3.03684354e-01 -9.96450186e-01 3.21563601e-01
-1.97389221e+00 -8.85649800e-01 -6.59245998e-02 2.24720716e+00
4.90490943e-01 2.39704058e-01 -3.07670891e-01 1.92433551e-01
6.11932516e-01 1.74585521e-01 -5.28144300e-01 -5.66543341e-01
-2.47243375e-01 6.77486718e-01 4.34628397e-01 2.73597777e-01
-6.10865831e-01 1.05777872e+00 5.14497423e+00 4.00039166e-01
-1.32278645e+00 1.60575345e-01 8.76373202e-02 1.88652650e-01
-5.75007498e-01 1.30749598e-01 -4.14350778e-01 -4.42774827e-03
1.47033107e+00 -4.77880031e-01 9.41796601e-01 6.89807117e-01
-5.37344992e-01 2.22844169e-01 -1.13889241e+00 1.29447055e+00
8.67196918e-02 -1.86669874e+00 5.42198479e-01 1.52333945e-01
4.62165922e-01 2.40855783e-01 -1.49750933e-01 3.03747207e-01
2.14576989e-01 -1.11543715e+00 5.86599290e-01 6.93415225e-01
4.87582594e-01 -6.46044135e-01 4.72133577e-01 1.24185480e-01
-1.24763203e+00 -2.67114360e-02 -7.59226441e-01 -2.90310234e-01
-2.13414699e-01 6.29332125e-01 -5.78039706e-01 6.03634775e-01
3.10241640e-01 8.39362025e-01 -6.62953913e-01 7.92873383e-01
-2.31451094e-01 4.56673205e-01 -2.55555362e-01 -3.90135974e-01
1.41611367e-01 -3.27986330e-01 4.73884851e-01 7.66396880e-01
1.89841554e-01 -1.00861229e-02 -1.19169064e-01 9.59386110e-01
-4.81154472e-01 8.74867439e-02 -1.17640150e+00 -6.73816681e-01
1.01181194e-01 1.33608508e+00 -1.39277565e+00 -1.02523677e-01
-1.80670619e-01 1.11250401e+00 8.77741337e-01 2.88135946e-01
-3.50820810e-01 -5.46594977e-01 5.96516073e-01 4.36200872e-02
3.43066037e-01 -5.89920819e-01 1.95239305e-01 -1.13615918e+00
-2.23584533e-01 -3.54204357e-01 2.69266218e-01 -5.71528137e-01
-9.61827636e-01 7.99287617e-01 -2.70877749e-01 -7.75235832e-01
-2.45569438e-01 -1.04011714e+00 -2.23598316e-01 3.86716247e-01
-1.44853020e+00 -9.59956646e-01 -3.96036446e-01 7.96674252e-01
-1.93120077e-01 -3.22370291e-01 1.26773274e+00 -3.35458070e-02
-2.75453389e-01 3.32772523e-01 2.62484491e-01 1.25988051e-01
1.13952257e-01 -1.28081572e+00 5.06587625e-01 6.30924940e-01
1.08870029e+00 9.26586449e-01 3.14346880e-01 -5.38631976e-01
-2.29718304e+00 -1.16915882e+00 6.27828300e-01 -1.26273766e-01
9.55802500e-01 -7.53078997e-01 -1.21803463e+00 6.65226936e-01
1.14239762e-02 4.50512856e-01 5.42716920e-01 1.41070232e-01
-6.72555387e-01 -2.20531553e-01 -8.36174309e-01 6.21576011e-01
1.21295059e+00 -9.39451993e-01 -5.20330310e-01 4.00342196e-01
8.46015394e-01 -2.56444305e-01 -9.13117647e-01 -2.11437345e-01
4.49586302e-01 -8.17708194e-01 6.49084508e-01 -7.54454494e-01
-3.48489694e-02 -1.60822496e-01 -2.03258812e-01 -1.23903847e+00
-3.09582531e-01 -8.46015036e-01 -5.46636879e-01 6.20683610e-01
6.07931092e-02 -1.06890655e+00 1.03177726e+00 2.36941382e-01
-5.40014058e-02 -6.03689432e-01 -9.90951717e-01 -6.96159005e-01
-1.83789730e-01 -2.62606233e-01 3.34660351e-01 8.75378072e-01
1.61531433e-01 3.95140707e-01 1.99198127e-01 1.69955008e-02
6.31451905e-01 2.67979234e-01 2.61544704e-01 -1.74849641e+00
-4.52033162e-01 -6.50716364e-01 -1.53247416e+00 -4.16948676e-01
5.65446675e-01 -1.70838583e+00 -3.33185643e-01 -1.65423465e+00
3.07161099e-04 -6.90087005e-02 -7.24864304e-01 6.47073746e-01
3.87885720e-01 3.35366488e-01 8.71298239e-02 -1.83013901e-01
-7.69480526e-01 4.91872102e-01 5.96278310e-01 -2.04287305e-01
-8.53989199e-02 -6.51751935e-01 -3.36967349e-01 4.69571501e-01
7.57326901e-01 -6.53042316e-01 -6.58667624e-01 -4.46102887e-01
8.59554529e-01 -2.71768451e-01 5.62588751e-01 -1.18864512e+00
6.28834307e-01 2.36082569e-01 -2.65744105e-02 1.52228281e-01
2.69537926e-01 -9.17844951e-01 4.59367752e-01 7.58631945e-01
-2.16893703e-01 -5.08694127e-02 2.89783657e-01 1.00650263e+00
-2.03712806e-01 -1.14968769e-01 4.98602003e-01 -9.54389647e-02
-9.31493402e-01 4.54077601e-01 -4.19406623e-01 -1.34052292e-01
9.15309548e-01 -2.77136773e-01 -6.05295897e-01 -9.75188911e-02
-7.76627362e-01 -3.28980833e-01 5.90289891e-01 3.80336344e-01
8.36944997e-01 -1.16540575e+00 -1.56631932e-01 3.76307994e-01
2.97443300e-01 -4.14237738e-01 7.29851648e-02 4.72027928e-01
-6.29239619e-01 5.29514253e-01 -5.44279635e-01 -5.19528568e-01
-9.50630784e-01 6.28593624e-01 2.03435466e-01 -2.20837578e-01
-7.76275873e-01 7.31710017e-01 -8.91780481e-02 -1.95119336e-01
1.07466422e-01 -6.18091166e-01 -1.35804430e-01 2.60070749e-02
3.41001928e-01 1.83453843e-01 3.81880671e-01 -4.84717071e-01
-3.31899136e-01 2.99912095e-01 6.24472611e-02 2.52439678e-01
1.51173472e+00 4.29398686e-01 -5.47107160e-01 7.12681174e-01
1.14136255e+00 -2.60298789e-01 -6.40667439e-01 -2.49844685e-01
3.15227151e-01 1.28164858e-01 2.80026406e-01 -1.62530214e-01
-1.10085118e+00 9.21870887e-01 5.52434325e-01 4.02454376e-01
9.12198007e-01 3.34315561e-02 4.83678907e-01 1.18300998e+00
9.56775129e-01 -7.37293959e-01 1.03792302e-01 5.61661541e-01
6.50095403e-01 -6.68813407e-01 4.92548794e-02 -1.98152378e-01
-3.99821252e-02 1.36315846e+00 9.20986161e-02 -3.52605969e-01
8.17358971e-01 9.78014916e-02 -5.39868593e-01 -6.37128234e-01
-8.89804959e-01 -2.75207400e-01 1.93646938e-01 6.99592590e-01
8.32366869e-02 1.23936206e-01 -7.45284632e-02 5.81952870e-01
-8.91868696e-02 -2.78641377e-02 5.27297795e-01 8.30122352e-01
-5.69927573e-01 -1.27718031e+00 2.28259802e-01 7.03583360e-01
2.98491549e-02 -1.44265920e-01 -4.94040489e-01 6.04024827e-01
-1.99863076e-01 3.93618256e-01 -1.59716263e-01 -5.81764579e-01
1.38000190e-01 3.62812042e-01 6.87922597e-01 -8.82471740e-01
-7.25012660e-01 -6.99751258e-01 -2.81026572e-01 -8.77860367e-01
-3.69911790e-01 -1.53830141e-01 -1.62175047e+00 -2.51605779e-01
-1.34402081e-01 2.25007355e-01 8.59347701e-01 8.17342758e-01
6.82247698e-01 6.92444801e-01 1.01977110e-01 -7.37114131e-01
-8.43026768e-03 -3.59164208e-01 -8.50291252e-01 2.92479038e-01
-3.41781452e-02 -7.47778118e-01 -1.16499469e-01 4.77516316e-02]
|
[6.964616298675537, 6.2605485916137695]
|
02fe9c28-6670-48ec-babb-6933c9495e7e
|
spectral-reconstruction-and-disparity-from
|
2103.10179
| null |
https://arxiv.org/abs/2103.10179v2
|
https://arxiv.org/pdf/2103.10179v2.pdf
|
Spectral Reconstruction and Disparity from Spatio-Spectrally Coded Light Fields via Multi-Task Deep Learning
|
We present a novel method to reconstruct a spectral central view and its aligned disparity map from spatio-spectrally coded light fields. Since we do not reconstruct an intermediate full light field from the coded measurement, we refer to this as principal reconstruction. The coded light fields correspond to those captured by a light field camera in the unfocused design with a spectrally coded microlens array. In this application, the spectrally coded light field camera can be interpreted as a single-shot spectral depth camera. We investigate several multi-task deep learning methods and propose a new auxiliary loss-based training strategy to enhance the reconstruction performance. The results are evaluated using a synthetic as well as a new real-world spectral light field dataset that we captured using a custom-built camera. The results are compared to state-of-the art compressed sensing reconstruction and disparity estimation. We achieve a high reconstruction quality for both synthetic and real-world coded light fields. The disparity estimation quality is on par with or even outperforms state-of-the-art disparity estimation from uncoded RGB light fields.
|
['Michael Heizmann', 'Jiayang Shi', 'Maximilian Schambach']
|
2021-03-18
| null | null | null | null |
['spectral-reconstruction']
|
['computer-vision']
|
[ 9.10459042e-01 -3.39624017e-01 4.35789466e-01 -3.90847623e-01
-7.86270797e-01 -2.37758234e-01 3.75221997e-01 -6.07745886e-01
-7.18092322e-01 8.05097282e-01 3.74374576e-02 1.43031403e-01
1.25609279e-01 -6.28173530e-01 -8.77008677e-01 -8.73729348e-01
6.35800242e-01 5.69986641e-01 3.18554223e-01 9.54861268e-02
4.05838579e-01 2.14106545e-01 -2.01361609e+00 2.76863605e-01
7.04223931e-01 1.04531729e+00 8.74656975e-01 6.77015007e-01
9.82859060e-02 8.64746630e-01 7.27740526e-02 -2.12863341e-01
6.34939432e-01 -3.11142534e-01 -3.90916109e-01 1.90022334e-01
1.10244417e+00 -8.83989811e-01 -3.58064502e-01 1.30287981e+00
5.13041556e-01 -8.52951035e-02 2.35579178e-01 -5.82837105e-01
-1.21591732e-01 -2.47874469e-01 -5.18883467e-01 -6.17389604e-02
7.04113543e-01 1.93157479e-01 6.45387232e-01 -9.43449497e-01
7.86446214e-01 7.06995547e-01 4.37612712e-01 4.81531978e-01
-1.27393806e+00 -4.39029187e-01 -5.91594577e-01 1.88464433e-01
-1.12800932e+00 -7.65164077e-01 9.49265420e-01 -6.83338106e-01
7.23853767e-01 -3.05576503e-01 7.96027839e-01 6.50072813e-01
4.83833075e-01 3.04079503e-01 1.67237365e+00 -6.99806511e-01
3.49809468e-01 -2.49373898e-01 -3.77107322e-01 9.51122463e-01
1.90714389e-01 6.43397450e-01 -7.88603187e-01 1.77282184e-01
7.83731341e-01 2.35111296e-01 -8.55655074e-01 -4.45754230e-01
-1.34237254e+00 5.69315851e-01 1.34079412e-01 9.17442217e-02
-2.47914523e-01 2.85220802e-01 -3.04519087e-01 8.50688294e-02
6.36358738e-01 3.17975581e-01 -1.26437962e-01 -4.14195955e-02
-1.01722527e+00 -3.58961262e-02 6.08171403e-01 4.67862099e-01
1.33367074e+00 3.64441872e-02 2.85827190e-01 8.05422008e-01
3.71703684e-01 1.14031255e+00 2.85387605e-01 -1.48596096e+00
1.25654146e-01 1.69796571e-01 3.33318204e-01 -3.98641586e-01
-2.04748869e-01 -3.40969533e-01 -4.76377964e-01 7.23032653e-01
4.75853115e-01 1.04636930e-01 -9.58681583e-01 1.44405043e+00
5.52720800e-02 4.96510178e-01 8.98558944e-02 1.14099407e+00
8.00278485e-01 6.38159454e-01 -9.32235658e-01 -5.09992599e-01
8.79339874e-01 -7.52501190e-01 -6.44606233e-01 -4.86066461e-01
-1.46464244e-01 -1.00883627e+00 8.01037610e-01 8.24061751e-01
-1.38637388e+00 -4.46766943e-01 -1.02730095e+00 -3.20388287e-01
2.27088600e-01 1.38594329e-01 5.32191217e-01 4.52523708e-01
-1.03944564e+00 3.61796916e-01 -8.72861087e-01 -2.56031975e-02
1.96178839e-01 -4.08352911e-02 -2.90920496e-01 -8.39282155e-01
-6.94875121e-01 8.13297212e-01 7.72334337e-02 -4.38325465e-01
-1.10816836e+00 -8.37410927e-01 -6.75096631e-01 -3.30504239e-01
2.64867336e-01 -9.90437984e-01 1.05218101e+00 -5.37502825e-01
-1.77723825e+00 1.33257127e+00 -2.78570294e-01 -1.93353966e-01
2.67851710e-01 -1.92442667e-02 -1.30589768e-01 5.94746113e-01
2.75378764e-01 4.82540309e-01 9.54886854e-01 -1.41411710e+00
-7.31982231e-01 -3.61563057e-01 -1.62573643e-02 1.29974186e-01
3.44597697e-01 -2.69180685e-01 -2.48566151e-01 -7.52706360e-03
3.71478677e-01 -8.82392704e-01 7.02935606e-02 3.32125872e-01
-1.80923671e-01 6.44863486e-01 6.80647373e-01 -3.28400373e-01
5.02002656e-01 -1.82646406e+00 1.80809960e-01 -5.74904561e-01
2.38635749e-01 1.52067706e-01 5.78932315e-02 3.07943851e-01
1.01801224e-01 -8.57127368e-01 -6.07437849e-01 -8.18042397e-01
-5.55703580e-01 1.96552381e-01 -3.05239975e-01 9.53942239e-01
-6.23409867e-01 4.41084325e-01 -9.41593587e-01 -2.56531388e-01
8.98927689e-01 5.46767473e-01 -7.99904943e-01 3.88219535e-01
-4.79174912e-01 7.31367350e-01 -2.07417198e-02 8.51798773e-01
1.10521698e+00 -2.32616141e-01 -1.61387622e-01 -3.48851085e-01
-4.60341483e-01 -1.46683544e-01 -9.79074836e-01 2.43040752e+00
-8.77045810e-01 9.78080213e-01 4.59006011e-01 -6.16163909e-01
7.78138280e-01 -4.75411750e-02 5.82918704e-01 -9.28146124e-01
-2.22790346e-01 5.74507236e-01 -4.42154646e-01 -7.15560019e-01
6.48706794e-01 -7.56820440e-01 4.99085516e-01 4.57312226e-01
5.98632544e-02 -9.23287272e-01 -4.24601436e-02 -8.78342241e-02
1.02161348e+00 2.02863947e-01 2.38290697e-01 -2.49010548e-01
5.04641414e-01 -3.93065900e-01 4.34894264e-01 4.90485489e-01
1.50383353e-01 9.65430081e-01 -2.71931916e-01 -4.88465756e-01
-1.18762624e+00 -1.24774933e+00 -4.41052973e-01 4.77739155e-01
4.75232691e-01 -7.43485913e-02 -3.12918603e-01 2.69441009e-01
-2.89691705e-02 5.08260310e-01 -1.75463155e-01 4.10071969e-01
-2.75178879e-01 -5.30686080e-01 -3.70102897e-02 -2.06342280e-01
8.16510260e-01 -6.73140287e-01 -1.18334568e+00 3.98871936e-02
-5.12922287e-01 -1.58130443e+00 -1.49737364e-02 7.10773766e-02
-4.81395304e-01 -1.35093331e+00 -6.87843025e-01 -6.01091087e-01
2.32763499e-01 5.75456619e-01 1.23521698e+00 -4.53070611e-01
-4.60649997e-01 6.22470140e-01 -7.22235516e-02 -1.12671532e-01
-3.78808230e-01 -7.60142326e-01 6.20165281e-02 3.25168788e-01
1.12649769e-01 -8.62967789e-01 -1.01652884e+00 1.97758526e-01
-1.03745592e+00 4.99797255e-01 4.04392958e-01 7.77141571e-01
7.57424891e-01 -9.98039544e-02 -1.28704026e-01 -4.96327698e-01
-3.98455672e-02 -1.84291899e-02 -1.18838656e+00 -3.55655514e-02
-6.85003698e-01 7.79002206e-03 6.56715214e-01 1.11566605e-02
-1.24465680e+00 3.78900200e-01 -8.02612752e-02 -7.20013440e-01
-1.15219854e-01 1.18642837e-01 2.51932710e-01 -6.24273777e-01
7.04381943e-01 5.42060614e-01 1.31692857e-01 -3.36063236e-01
1.05382334e-02 6.40285671e-01 8.82999122e-01 -3.14104438e-01
3.94663453e-01 1.30301929e+00 5.75764060e-01 -9.57625866e-01
-7.36564457e-01 -6.97233140e-01 -2.90666699e-01 -5.92358291e-01
9.53293502e-01 -1.12989914e+00 -9.79514778e-01 8.11100543e-01
-1.22460890e+00 -4.66913313e-01 -3.25823277e-01 7.90696323e-01
-1.00099707e+00 5.80888748e-01 -6.27660513e-01 -6.33382976e-01
1.14362776e-01 -1.36701846e+00 1.56421506e+00 -8.53215605e-02
8.31000805e-01 -8.95221353e-01 5.28185964e-01 6.29437923e-01
4.61323798e-01 2.47107282e-01 4.48413700e-01 7.48948216e-01
-1.23325813e+00 3.03982824e-01 -2.03975096e-01 4.54396755e-01
-1.68994721e-02 -2.55823076e-01 -1.36240649e+00 -4.13119346e-01
4.72804487e-01 -5.97433627e-01 1.33576238e+00 8.62933457e-01
8.05605114e-01 3.55393857e-01 -1.76643595e-01 1.32643235e+00
2.29189610e+00 2.44079590e-01 7.53795445e-01 9.20713395e-02
5.67927957e-01 5.34728527e-01 4.57169712e-01 5.86327791e-01
2.66145885e-01 9.40681636e-01 8.45779061e-01 4.01699357e-02
-5.17628610e-01 -1.13518447e-01 2.55102754e-01 4.54375178e-01
-8.11576992e-02 -3.01831156e-01 -8.88636708e-01 3.65209073e-01
-1.47003627e+00 -1.11477482e+00 -2.07491562e-01 2.30080152e+00
7.07873285e-01 -3.98926437e-01 -5.42109549e-01 7.08809197e-02
4.34725285e-01 3.37476104e-01 -5.01892984e-01 2.59948552e-01
-4.36253309e-01 3.97532940e-01 6.26222134e-01 1.28080893e+00
-6.53258383e-01 8.19812238e-01 6.18985796e+00 5.76411009e-01
-1.49429476e+00 2.83547282e-01 2.01338962e-01 -4.70264494e-01
-6.02363884e-01 2.97577549e-02 -5.45310318e-01 6.55737221e-01
3.61090630e-01 1.30703062e-01 9.16126370e-01 3.09159547e-01
1.61811933e-01 -9.09831524e-01 -1.15586984e+00 1.85635591e+00
3.09350014e-01 -1.49850786e+00 -2.98166305e-01 3.01882565e-01
1.05038714e+00 5.46174824e-01 1.07700244e-01 -5.41367710e-01
7.25513995e-02 -7.27504492e-01 5.45073271e-01 9.01996553e-01
1.47712135e+00 -1.88853905e-01 3.45873564e-01 4.44854021e-01
-7.96809673e-01 5.30352257e-02 -4.79943097e-01 -1.98760435e-01
4.85320896e-01 1.12145734e+00 -4.42244649e-01 3.83381128e-01
6.80502713e-01 1.13080668e+00 -3.10251445e-01 8.70659769e-01
2.16626916e-02 2.36470535e-01 -3.03181380e-01 4.62370723e-01
-1.11674264e-01 -6.29535735e-01 6.21693015e-01 7.99854636e-01
5.91828525e-01 2.98440635e-01 2.70808563e-02 1.03889024e+00
1.52499691e-01 -5.91301322e-01 -8.86684835e-01 4.98093665e-01
1.02709392e-02 1.00738084e+00 -2.86174268e-01 -2.59763040e-02
-6.83858097e-01 9.65948880e-01 -1.85432538e-01 4.36023623e-01
-3.69016379e-01 -4.86715846e-02 6.16602182e-01 3.75276387e-01
1.81034103e-01 -1.50537938e-01 -2.28198282e-02 -1.64033067e+00
7.07467273e-02 -1.48927897e-01 -2.18370035e-01 -1.61403370e+00
-1.02960324e+00 5.27389169e-01 3.72073278e-02 -1.41968715e+00
-2.81866729e-01 -7.15292752e-01 -1.54186741e-01 8.88505101e-01
-2.18676114e+00 -9.01544392e-01 -7.38615036e-01 1.00077522e+00
2.01058894e-01 -2.48685092e-01 5.40311515e-01 3.12326252e-01
1.91506490e-01 -3.61473858e-01 4.87184227e-01 -5.35250366e-01
7.52333403e-01 -9.53355968e-01 -3.19872022e-01 1.15682268e+00
-2.19258428e-01 1.97069421e-02 7.67670870e-01 -3.59327972e-01
-1.53340673e+00 -7.65431821e-01 4.42196548e-01 1.21655772e-02
2.06152618e-01 -2.19239965e-01 -4.81170714e-01 5.89468002e-01
3.09456468e-01 5.07651329e-01 3.72707069e-01 -6.38980687e-01
-5.45990691e-02 -4.84120101e-01 -1.40805209e+00 -5.12261316e-02
1.12702012e+00 -9.57916260e-01 -4.30069387e-01 4.46122795e-01
4.41866219e-01 -5.24656594e-01 -3.02449346e-01 4.74658459e-01
6.83556497e-01 -1.92364359e+00 1.05778229e+00 4.49635178e-01
9.78831470e-01 -3.21624577e-01 -6.51855230e-01 -1.41190231e+00
-7.96298608e-02 -6.19384885e-01 1.17631018e-01 4.93359536e-01
-2.48229653e-01 -8.51417780e-01 1.09221804e+00 6.16059750e-02
-4.94182914e-01 -3.78107727e-01 -1.03108394e+00 -5.91452479e-01
-3.35807472e-01 -4.82149184e-01 8.15005675e-02 8.87009740e-01
-4.14495289e-01 2.59905517e-01 -4.01265293e-01 6.92659095e-02
1.38832974e+00 5.38033664e-01 5.52315235e-01 -1.21228778e+00
-5.47460496e-01 -1.03424594e-01 -3.78354192e-01 -1.35469699e+00
3.19318920e-01 -8.94011557e-01 1.11461952e-01 -1.48716998e+00
2.92468935e-01 -3.93493235e-01 2.70620167e-01 -1.89128399e-01
3.58206183e-01 3.74664724e-01 1.03883006e-01 4.13350940e-01
-3.19947839e-01 5.41658401e-01 1.32881665e+00 7.54570542e-03
6.85243681e-02 -4.79408391e-02 -1.44657195e-01 6.05721593e-01
2.85504311e-01 -1.85589612e-01 -4.53675419e-01 -6.98473036e-01
7.35473752e-01 7.32068419e-01 6.31813586e-01 -1.31777561e+00
5.87990940e-01 -1.97769061e-01 1.03602096e-01 -4.24506962e-01
9.29679155e-01 -9.28673804e-01 3.57119828e-01 4.69284266e-01
-1.03890136e-01 -5.95656931e-01 -2.11750090e-01 5.82085907e-01
-3.58094960e-01 -2.62208879e-01 1.33326340e+00 -4.27807897e-01
-7.76998043e-01 2.51489908e-01 -1.06681190e-01 8.62468779e-02
7.95450389e-01 -3.42921525e-01 -6.27258599e-01 -5.75287163e-01
-3.37428927e-01 -3.99244010e-01 1.09845650e+00 -3.39185357e-01
1.15925777e+00 -1.03675508e+00 -7.49594569e-01 7.35110402e-01
3.17218065e-01 5.05224578e-02 1.67969659e-01 5.61819255e-01
-1.15726364e+00 5.03643811e-01 -3.89344037e-01 -1.08189416e+00
-1.11039507e+00 4.02035981e-01 7.59775639e-01 1.12835072e-01
-5.89736104e-01 8.36510181e-01 2.82501578e-01 -3.06060851e-01
-1.41484663e-01 -5.25381505e-01 3.60210478e-01 -5.81779659e-01
4.38162744e-01 4.87653732e-01 2.24334434e-01 -4.70348775e-01
-2.15726584e-01 1.26715541e+00 6.46327496e-01 -3.26094925e-01
1.47439671e+00 -3.61876607e-01 -2.50193596e-01 4.92674947e-01
1.31684756e+00 3.42245489e-01 -1.65827668e+00 -5.72937548e-01
-1.14704931e+00 -1.19103134e+00 5.50461948e-01 -7.22917914e-01
-1.31178296e+00 1.10934055e+00 8.69351149e-01 -7.92254284e-02
1.36472547e+00 9.08045936e-03 6.71076715e-01 3.00399214e-01
9.65235770e-01 -6.33938432e-01 1.65758595e-01 3.77064049e-01
5.14699399e-01 -1.58902633e+00 1.57643601e-01 -4.45886701e-01
-2.33754590e-01 1.21809828e+00 6.00035302e-02 -2.18348131e-02
7.66026378e-01 4.20856357e-01 -1.46427050e-01 -1.67514622e-01
-6.52091622e-01 -2.77968496e-01 -1.29198581e-01 7.37371743e-01
1.53937533e-01 -1.05235524e-01 2.01662898e-01 -6.29840612e-01
-1.98210832e-02 4.98528898e-01 1.02276075e+00 5.80786765e-01
-6.11654818e-01 -8.49975348e-01 -3.49639326e-01 1.63338870e-01
-2.70040691e-01 -2.39184827e-01 2.11837143e-01 2.57483304e-01
3.23726892e-01 1.00182569e+00 1.02561176e-01 -2.15070277e-01
7.10575581e-02 -2.73250014e-01 1.09955919e+00 -6.39654040e-01
1.29914731e-01 6.65320782e-04 -3.29839587e-01 -8.99963379e-01
-9.82245266e-01 -6.35932207e-01 -8.24056387e-01 -2.00159550e-01
2.89339256e-02 -4.03202474e-01 8.91855240e-01 6.53896511e-01
2.11832169e-02 -1.98939107e-02 7.05514371e-01 -1.22424078e+00
1.69051901e-01 -6.21138215e-01 -9.55918074e-01 2.96120673e-01
1.06539106e+00 -6.82914197e-01 -7.49391377e-01 1.56695038e-01]
|
[9.650556564331055, -2.650047540664673]
|
30d30cb9-64b4-4cb7-9e0d-f2160e6303e6
|
dependency-parsing-for-chinese-long-sentence
| null | null |
https://aclanthology.org/Y15-2039
|
https://aclanthology.org/Y15-2039.pdf
|
Dependency parsing for Chinese long sentence: A second-stage main structure parsing method
| null |
['Weiguang Qu', 'Bo Li', 'Yunfei Long']
|
2015-10-01
|
dependency-parsing-for-chinese-long-sentence-1
|
https://aclanthology.org/Y15-2039
|
https://aclanthology.org/Y15-2039.pdf
|
paclic-2015-10
|
['transition-based-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]
|
[-7.409140110015869, 3.593579053878784]
|
22295380-c62e-4061-8c17-1bf90ebef534
|
collaborative-video-object-segmentation-by
|
2003.08333
| null |
https://arxiv.org/abs/2003.08333v2
|
https://arxiv.org/pdf/2003.08333v2.pdf
|
Collaborative Video Object Segmentation by Foreground-Background Integration
|
This paper investigates the principles of embedding learning to tackle the challenging semi-supervised video object segmentation. Different from previous practices that only explore the embedding learning using pixels from foreground object (s), we consider background should be equally treated and thus propose Collaborative video object segmentation by Foreground-Background Integration (CFBI) approach. Our CFBI implicitly imposes the feature embedding from the target foreground object and its corresponding background to be contrastive, promoting the segmentation results accordingly. With the feature embedding from both foreground and background, our CFBI performs the matching process between the reference and the predicted sequence from both pixel and instance levels, making the CFBI be robust to various object scales. We conduct extensive experiments on three popular benchmarks, i.e., DAVIS 2016, DAVIS 2017, and YouTube-VOS. Our CFBI achieves the performance (J$F) of 89.4%, 81.9%, and 81.4%, respectively, outperforming all the other state-of-the-art methods. Code: https://github.com/z-x-yang/CFBI.
|
['Yunchao Wei', 'Zongxin Yang', 'Yi Yang']
|
2020-03-18
| null |
https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/3385_ECCV_2020_paper.php
|
https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123500324.pdf
|
eccv-2020-8
|
['one-shot-visual-object-segmentation']
|
['computer-vision']
|
[ 1.71144769e-01 -1.54644594e-01 -3.55095357e-01 -1.60044864e-01
-6.40232563e-01 -4.51692969e-01 3.79461408e-01 -1.76902294e-01
-3.27824056e-01 4.17673379e-01 -7.54731819e-02 -4.02144603e-02
3.57994795e-01 -5.45969367e-01 -6.89388335e-01 -9.64711845e-01
3.30797173e-02 2.91147530e-02 8.24971557e-01 3.67515862e-01
9.07293111e-02 1.76858574e-01 -1.21499562e+00 2.96255678e-01
8.21335316e-01 1.20963025e+00 2.52920032e-01 6.90026224e-01
-1.73184797e-01 8.00245762e-01 -6.01898491e-01 -4.69169796e-01
5.51044762e-01 -3.70523185e-01 -6.55094922e-01 3.86755258e-01
6.79449379e-01 -5.74441254e-01 -6.78916633e-01 1.17461622e+00
3.32816988e-01 7.28418827e-02 5.83926857e-01 -1.38771546e+00
-8.47473502e-01 3.09164017e-01 -1.04743087e+00 5.36905587e-01
3.11590615e-03 4.22006011e-01 8.91036868e-01 -1.00267041e+00
6.38119042e-01 1.14965522e+00 3.55243295e-01 6.84935033e-01
-1.14319861e+00 -7.19473958e-01 6.15063310e-01 3.71870399e-01
-1.35196972e+00 -2.79174477e-01 8.13206315e-01 -6.25825346e-01
4.90631878e-01 2.38677338e-01 7.31057882e-01 8.41245949e-01
-1.89703569e-01 1.44357848e+00 9.87086236e-01 -1.49050087e-01
1.83977769e-03 2.07869988e-02 2.43059874e-01 6.45876825e-01
2.46513784e-01 -1.44550890e-01 -2.78181553e-01 2.09543426e-02
7.22232819e-01 1.87440470e-01 -5.01988888e-01 -3.17790657e-01
-1.37485206e+00 5.16557515e-01 4.41960573e-01 4.58697677e-02
-2.13990673e-01 4.22081441e-01 5.08727193e-01 -1.88087091e-01
4.22961384e-01 -2.61639625e-01 -4.43319738e-01 7.35251456e-02
-1.12711370e+00 6.44486547e-02 6.50099277e-01 1.15207648e+00
7.50283778e-01 2.14910984e-01 -3.49398315e-01 6.35422111e-01
6.60704672e-01 4.31856871e-01 2.51164436e-01 -1.17717421e+00
4.42912936e-01 5.27206659e-01 1.75286114e-01 -9.96273816e-01
3.12979937e-01 -2.10281163e-01 -6.48776948e-01 3.07117235e-02
4.09157515e-01 -7.51446486e-02 -1.10206556e+00 1.32081378e+00
7.93329120e-01 7.98230171e-01 1.10025732e-02 1.04816747e+00
1.06063771e+00 1.11135876e+00 1.29852071e-01 -1.80645704e-01
1.03999722e+00 -1.60134280e+00 -7.35171914e-01 -3.03194225e-01
3.31125349e-01 -6.83594167e-01 9.32316959e-01 3.04991931e-01
-1.06881654e+00 -8.16702425e-01 -9.68939781e-01 -1.21036079e-02
2.07602531e-02 3.11087877e-01 4.05318111e-01 4.93173689e-01
-8.17369342e-01 4.90018487e-01 -9.62314665e-01 -1.41856577e-02
7.98461616e-01 2.14441359e-01 -6.75291047e-02 -5.27553968e-02
-8.85731459e-01 3.50147188e-01 4.62735832e-01 1.59789875e-01
-1.18297505e+00 -6.87528551e-01 -5.24755776e-01 -3.09315603e-02
6.47447586e-01 -4.17223871e-01 1.00340617e+00 -1.28353977e+00
-1.35528886e+00 9.28366661e-01 -1.64467961e-01 -2.81388432e-01
7.41514742e-01 -5.01372099e-01 -6.15447611e-02 4.41473126e-01
1.80476129e-01 9.27392781e-01 9.46059763e-01 -1.50445151e+00
-8.28206062e-01 -4.12597992e-02 7.58725256e-02 -5.68060088e-04
-3.46473336e-01 9.30053815e-02 -1.07882476e+00 -6.68913424e-01
4.08319337e-03 -8.12256157e-01 -1.19739398e-01 4.53055769e-01
-4.37849820e-01 -3.16358179e-01 1.21803105e+00 -1.00572050e+00
1.33241224e+00 -2.48735428e+00 1.16538666e-01 -2.70748824e-01
4.01382774e-01 5.08573890e-01 -7.62444064e-02 4.25178045e-03
1.46542132e-01 3.09507430e-01 -2.94595450e-01 -2.48933256e-01
-1.48874866e-02 2.69688934e-01 -6.80148154e-02 6.92318797e-01
3.60834002e-01 9.66221333e-01 -9.20473337e-01 -9.59812343e-01
2.99673617e-01 4.54983175e-01 -5.69479942e-01 3.17054302e-01
-2.83737123e-01 3.26169521e-01 -4.66122270e-01 9.58447695e-01
7.21663296e-01 -4.92299139e-01 -3.81781161e-02 -2.47357845e-01
4.17658985e-02 -2.57372167e-02 -1.30262363e+00 1.37797046e+00
2.25452557e-01 6.73501253e-01 2.59595484e-01 -8.26651812e-01
7.49860168e-01 1.70126498e-01 7.05923975e-01 -3.76247972e-01
1.11711817e-03 1.62540212e-01 8.12826604e-02 -4.57047880e-01
1.99183255e-01 3.51171613e-01 3.37855548e-01 1.89635694e-01
6.44230992e-02 2.70520419e-01 4.84726101e-01 4.38118786e-01
7.82621801e-01 4.73798990e-01 -4.79267351e-03 -2.69029379e-01
5.58777571e-01 -2.44777992e-01 1.05572987e+00 5.66110730e-01
-9.30082321e-01 6.57182574e-01 5.83337605e-01 -1.79513842e-01
-8.45766306e-01 -1.20327568e+00 1.11262705e-02 9.51703250e-01
7.52773583e-01 -2.73956388e-01 -8.23473573e-01 -9.53769743e-01
3.38677652e-02 4.13177431e-01 -5.90312243e-01 8.53307247e-02
-6.72695458e-01 -5.36888242e-01 2.54177719e-01 6.92858934e-01
7.54508793e-01 -8.92191470e-01 -4.29673731e-01 1.39871195e-01
-2.38566026e-01 -1.26820254e+00 -8.86879146e-01 -1.91012651e-01
-9.21489537e-01 -1.03620839e+00 -9.99383688e-01 -9.18813586e-01
5.11885226e-01 4.65236455e-01 9.12898779e-01 2.58352697e-01
-2.07455814e-01 4.40645516e-01 -4.01066989e-01 -8.11078846e-02
-2.33818233e-01 -2.29980469e-01 -1.85881525e-01 3.07768434e-01
1.30858764e-01 -1.79088205e-01 -9.21335757e-01 4.63615179e-01
-8.72831106e-01 2.27795571e-01 4.65995729e-01 7.21930683e-01
8.75594676e-01 -1.63146347e-01 3.65587085e-01 -6.86823428e-01
-1.63728282e-01 -4.99576330e-01 -5.88723838e-01 2.84819961e-01
-4.43086922e-01 -5.84723711e-01 2.39151791e-01 -7.48796642e-01
-8.35975826e-01 7.33492225e-02 1.05201511e-03 -8.15948486e-01
-1.23388201e-01 -7.89418817e-02 -4.37542528e-01 2.22058897e-03
1.07575491e-01 3.91215444e-01 -2.02472761e-01 -3.85392904e-01
2.73272127e-01 6.73405588e-01 4.37530309e-01 -3.88692766e-01
6.70418382e-01 5.37849784e-01 -5.12808025e-01 -7.13995218e-01
-7.94418514e-01 -7.67491162e-01 -6.58432126e-01 -4.98273194e-01
1.13214660e+00 -9.28246021e-01 -3.65377307e-01 4.40234333e-01
-1.27385592e+00 -5.59885085e-01 -2.18394279e-01 3.25480849e-01
-4.06929255e-01 6.94798529e-01 -8.46899152e-01 -7.21288562e-01
-3.18190992e-01 -1.34611368e+00 1.05471194e+00 4.64826345e-01
7.65563697e-02 -7.95186043e-01 -2.18257397e-01 5.45460999e-01
7.94847310e-02 4.05044913e-01 4.70732033e-01 -5.50125003e-01
-1.03093302e+00 -1.00812744e-02 -5.72682679e-01 6.52712524e-01
1.25123635e-01 5.56634367e-01 -1.00011182e+00 -2.79043674e-01
-1.24350891e-01 -1.50256250e-02 1.08153749e+00 4.06717777e-01
1.11349285e+00 -3.36068988e-01 -3.89978617e-01 6.69358194e-01
1.49477446e+00 3.48052502e-01 7.07086504e-01 3.90679538e-01
9.83583450e-01 4.32506353e-01 7.90464520e-01 2.90311009e-01
2.90609926e-01 5.07373333e-01 5.45918763e-01 -1.44010797e-01
-3.54127079e-01 3.58349681e-02 6.42188787e-01 8.46837401e-01
4.18827310e-02 -3.79596770e-01 -8.47075045e-01 7.93835342e-01
-1.97481179e+00 -8.63466382e-01 -2.56277502e-01 1.76957417e+00
9.04062688e-01 2.91139543e-01 2.25973085e-01 -3.42004970e-02
9.54149485e-01 5.18346548e-01 -6.37408018e-01 3.74202393e-02
-4.47443202e-02 -1.71463415e-01 5.64274788e-01 2.71827996e-01
-1.48023343e+00 9.97655571e-01 5.33304119e+00 8.69095862e-01
-1.04216576e+00 4.07699108e-01 9.51656699e-01 -1.81047991e-01
2.71002892e-02 -5.25541557e-03 -8.19784403e-01 8.83113146e-01
6.30254447e-01 -3.20634618e-02 2.59482920e-01 9.17722523e-01
9.52120572e-02 -3.21344249e-02 -9.49038982e-01 9.65241849e-01
-6.36106506e-02 -1.31201923e+00 -3.31364945e-02 -1.26550570e-01
1.13842785e+00 5.01950793e-02 -6.60735590e-04 2.70005375e-01
-1.25860512e-01 -7.33220458e-01 9.34565067e-01 3.53455812e-01
4.57646966e-01 -4.71042067e-01 5.63633025e-01 1.77090451e-01
-1.38326395e+00 2.83534676e-02 -2.03822419e-01 4.26564276e-01
2.23081052e-01 4.56109643e-01 -2.44735107e-01 4.66318488e-01
1.03079367e+00 9.92927372e-01 -5.14960229e-01 1.07066357e+00
-1.87384263e-01 9.84117389e-01 -1.78900838e-01 1.48792088e-01
4.01738048e-01 -2.60093451e-01 5.89499235e-01 1.44281566e+00
-4.70983870e-02 -2.22168062e-02 4.93740201e-01 8.88769925e-01
-1.35004103e-01 1.93867072e-01 -6.28812015e-02 -2.99920768e-01
2.29205057e-01 1.18832278e+00 -9.91766870e-01 -6.70450628e-01
-6.36420965e-01 1.15264952e+00 -7.67811015e-02 5.65532506e-01
-1.26068306e+00 -2.14082524e-01 7.48327196e-01 1.13885656e-01
8.72912526e-01 -2.17060477e-01 -3.54620740e-02 -1.20786798e+00
1.24145217e-01 -7.36337781e-01 2.37528563e-01 -6.43652439e-01
-1.14844620e+00 2.82536238e-01 -2.78485697e-02 -1.20663691e+00
4.28162634e-01 -6.95648849e-01 -7.40453959e-01 6.09246075e-01
-1.57759666e+00 -1.08943081e+00 -4.09602910e-01 2.69631445e-01
7.50173092e-01 1.26995474e-01 2.52325181e-02 6.58543766e-01
-8.47132564e-01 4.45964098e-01 1.77999482e-01 5.11616170e-01
5.90543032e-01 -1.19918001e+00 3.20274264e-01 1.09245539e+00
2.00472161e-01 3.82216066e-01 3.92747819e-01 -7.35787928e-01
-1.30009663e+00 -1.34747803e+00 4.11918372e-01 -3.00366789e-01
7.13330448e-01 -2.23671973e-01 -1.11772156e+00 5.83874881e-01
3.07557791e-01 5.05857348e-01 6.10719383e-01 -6.17864251e-01
-1.29639894e-01 -1.73433676e-01 -9.26809907e-01 6.98846757e-01
9.95263457e-01 -2.78416663e-01 -2.83311516e-01 2.95103341e-01
9.24432516e-01 -4.51208889e-01 -8.77196670e-01 4.78515714e-01
4.86820430e-01 -9.19142544e-01 1.06118429e+00 -4.25304323e-01
5.35507143e-01 -6.66633666e-01 -2.98357189e-01 -5.59261620e-01
-2.69312531e-01 -5.24006307e-01 -7.11067080e-01 1.44959486e+00
5.36703914e-02 -4.10720617e-01 8.48581791e-01 3.17843050e-01
-1.39020249e-01 -1.15722060e+00 -8.36949646e-01 -8.20842385e-01
-1.13012642e-01 -2.88534552e-01 2.43984789e-01 7.21371830e-01
-7.92631984e-01 -1.02200948e-01 -4.13248003e-01 3.01147789e-01
7.06456900e-01 3.53404701e-01 8.87550533e-01 -7.71777153e-01
-3.98722708e-01 -5.14404714e-01 -4.81997281e-01 -1.46847045e+00
7.93313086e-02 -8.69485855e-01 9.83244553e-03 -1.64720762e+00
4.08805549e-01 -2.06452757e-01 -4.75145459e-01 3.28197271e-01
-4.74529833e-01 4.17550176e-01 5.00481308e-01 4.31188852e-01
-9.65762794e-01 6.53559744e-01 1.57878923e+00 -3.82771790e-01
-3.14636268e-02 -1.13157615e-01 -5.27528822e-01 8.59949708e-01
8.59631658e-01 -6.09750092e-01 -4.88423184e-02 -4.30431604e-01
-3.99696320e-01 -2.58235097e-01 4.80595499e-01 -7.28397191e-01
4.14712802e-02 -4.03636098e-01 4.13185894e-01 -6.34512126e-01
1.79796934e-01 -6.95609629e-01 -2.19238806e-03 6.25158489e-01
-1.28787801e-01 -2.42947891e-01 2.62685269e-01 8.98285568e-01
-1.96313486e-01 -1.95992887e-01 8.71824563e-01 -1.15894666e-02
-1.07111061e+00 6.30711913e-01 -2.39754826e-01 2.84825087e-01
1.38913321e+00 -5.53129077e-01 -2.80483931e-01 -1.15362734e-01
-6.13472044e-01 4.52185154e-01 5.23682177e-01 3.51819366e-01
7.66922951e-01 -1.31459832e+00 -6.96777880e-01 -2.62503196e-02
1.96655840e-03 2.07244471e-01 2.48922408e-01 1.16739404e+00
-7.17779875e-01 5.15307933e-02 6.83988854e-02 -8.98893178e-01
-1.39643919e+00 6.22852862e-01 2.35299230e-01 -2.32459456e-02
-6.28805816e-01 9.22848761e-01 5.86560249e-01 -4.53934707e-02
3.83048117e-01 -2.64480561e-01 1.15089184e-02 -1.83592801e-05
4.75971907e-01 6.34446740e-01 -5.96722841e-01 -7.81060576e-01
-4.46073264e-01 5.61412871e-01 -1.11035958e-01 1.41542882e-01
1.02690184e+00 -1.61505774e-01 -2.46310700e-02 4.91300404e-01
1.32157028e+00 2.93240994e-02 -1.80749595e+00 -4.02743399e-01
1.15240626e-02 -7.62281537e-01 -1.07594579e-01 -4.45146471e-01
-1.53626037e+00 8.94111931e-01 7.33531415e-01 2.92050410e-02
1.10039306e+00 1.20094039e-01 1.00746310e+00 -3.06322966e-02
7.85890818e-02 -1.01364136e+00 4.35961545e-01 2.48989359e-01
6.18407309e-01 -1.39993191e+00 2.23559991e-01 -4.65951353e-01
-8.20546031e-01 9.26653326e-01 7.89982438e-01 -3.59407187e-01
5.49340963e-01 1.42237693e-01 1.13047123e-01 1.41201004e-01
-6.51502430e-01 -2.56708801e-01 5.30225933e-01 2.77292281e-01
2.73019344e-01 -6.30564094e-02 -2.02756405e-01 4.88558471e-01
6.44653320e-01 -1.72051489e-01 3.12720418e-01 9.78952229e-01
-4.51863319e-01 -7.69267201e-01 -2.51489997e-01 4.98876691e-01
-6.70938194e-01 -1.26595395e-02 -4.04399782e-01 8.85206163e-01
2.99004614e-01 7.10620284e-01 -4.40447628e-02 -2.54003763e-01
-7.21214414e-02 -9.11877677e-02 3.17018360e-01 -6.41384602e-01
-4.60899353e-01 5.35053134e-01 -4.03295606e-01 -6.52170539e-01
-6.77469075e-01 -7.66513884e-01 -1.45976853e+00 -6.14970699e-02
-6.34374321e-01 -2.03540578e-01 2.91277826e-01 5.31036735e-01
2.39283621e-01 6.37798667e-01 5.06148458e-01 -1.00829029e+00
-2.65416354e-01 -7.00696230e-01 -4.27901357e-01 3.55777949e-01
3.29792976e-01 -6.19787753e-01 -4.55714107e-01 3.58273536e-01]
|
[9.254261016845703, -0.1406562626361847]
|
fb448f23-2a1a-47a3-a808-2cb10df7c0e1
|
expected-scalarised-returns-dominance-a-new
|
2106.01048
| null |
https://arxiv.org/abs/2106.01048v3
|
https://arxiv.org/pdf/2106.01048v3.pdf
|
Expected Scalarised Returns Dominance: A New Solution Concept for Multi-Objective Decision Making
|
In many real-world scenarios, the utility of a user is derived from the single execution of a policy. In this case, to apply multi-objective reinforcement learning, the expected utility of the returns must be optimised. Various scenarios exist where a user's preferences over objectives (also known as the utility function) are unknown or difficult to specify. In such scenarios, a set of optimal policies must be learned. However, settings where the expected utility must be maximised have been largely overlooked by the multi-objective reinforcement learning community and, as a consequence, a set of optimal solutions has yet to be defined. In this paper we address this challenge by proposing first-order stochastic dominance as a criterion to build solution sets to maximise expected utility. We also propose a new dominance criterion, known as expected scalarised returns (ESR) dominance, that extends first-order stochastic dominance to allow a set of optimal policies to be learned in practice. We then define a new solution concept called the ESR set, which is a set of policies that are ESR dominant. Finally, we define a new multi-objective distributional tabular reinforcement learning (MOT-DRL) algorithm to learn the ESR set in a multi-objective multi-armed bandit setting.
|
['Patrick Mannion', 'Enda Howley', 'Diederik M. Roijers', 'Timothy Verstraeten', 'Conor F. Hayes']
|
2021-06-02
| null | null | null | null |
['multi-objective-reinforcement-learning']
|
['methodology']
|
[ 8.81573036e-02 -1.04307979e-02 -4.94670093e-01 -3.52500618e-01
-9.66290534e-01 -7.93684304e-01 3.16853106e-01 1.73791498e-01
-5.79941809e-01 1.28170955e+00 9.82162133e-02 -3.22169036e-01
-9.99584198e-01 -8.85519445e-01 -6.54201806e-01 -9.10712898e-01
-2.06912030e-02 6.75295413e-01 -1.71090379e-01 -1.58780292e-01
2.68002540e-01 3.10885340e-01 -1.37631249e+00 5.25983702e-03
1.18959522e+00 1.14305508e+00 3.21109265e-01 5.29338479e-01
3.67699028e-03 5.51585019e-01 -6.80945456e-01 -2.59003162e-01
5.17243326e-01 -4.27785486e-01 -7.91890025e-01 1.16966173e-01
-2.59781271e-01 -3.55065495e-01 4.02217120e-01 1.09298515e+00
5.13384342e-01 4.88394231e-01 7.67135501e-01 -1.34979177e+00
-2.20531315e-01 7.16771185e-01 -5.97705960e-01 1.09988488e-01
2.12126672e-01 8.90483782e-02 1.59450364e+00 -3.69040929e-02
3.19294691e-01 1.30844510e+00 -3.06021888e-02 3.10213119e-01
-1.47571361e+00 -2.78841883e-01 4.73237306e-01 1.32386699e-01
-9.11088109e-01 -9.93757993e-02 6.83310270e-01 -1.39456436e-01
4.05390352e-01 5.20597756e-01 5.48659623e-01 7.50109673e-01
1.23765536e-01 1.03778923e+00 1.35619283e+00 -4.18954730e-01
7.16287374e-01 1.88279644e-01 -2.55777836e-01 2.63433158e-01
2.99621344e-01 2.99483597e-01 -1.61645934e-01 -4.60021526e-01
4.57334310e-01 -1.18642978e-01 -2.93882757e-01 -7.62874186e-01
-7.76679218e-01 1.33294749e+00 1.30289331e-01 -1.37072208e-03
-8.02650034e-01 9.87999886e-02 3.08945805e-01 4.36651409e-01
3.98030370e-01 6.69264019e-01 -4.27339286e-01 -1.94463462e-01
-6.84141636e-01 4.79558706e-01 7.83310175e-01 3.77234429e-01
7.88213074e-01 2.62972880e-02 -5.44175744e-01 9.27246213e-01
2.49933958e-01 3.82802188e-01 2.34707162e-01 -1.06242025e+00
5.80486774e-01 3.51949453e-01 7.81885326e-01 -6.68498039e-01
-1.63049728e-01 -5.73671103e-01 -3.78470153e-01 4.52499181e-01
4.27013159e-01 -4.54732925e-01 -5.68482637e-01 1.97418928e+00
5.19544303e-01 -1.37825564e-01 2.38255411e-01 1.06159735e+00
-1.08588509e-01 6.94091558e-01 -1.90597922e-01 -6.55481458e-01
7.87721097e-01 -5.35265803e-01 -5.47479868e-01 -1.74418211e-01
3.78380209e-01 -4.72050130e-01 1.05141842e+00 5.35322249e-01
-1.09850216e+00 9.32054147e-02 -8.61701846e-01 8.37396324e-01
-1.13337539e-01 -4.08292353e-01 5.60335040e-01 9.11942661e-01
-7.53296793e-01 5.66902995e-01 -3.84462714e-01 1.21313013e-01
2.94607937e-01 5.27843416e-01 1.76960930e-01 -2.74984539e-02
-1.21477246e+00 9.48550463e-01 7.18425274e-01 -1.88753992e-01
-1.00931156e+00 -5.17227590e-01 -5.82687080e-01 2.57363826e-01
1.14017296e+00 -6.60525322e-01 1.56410205e+00 -1.32224452e+00
-1.61344826e+00 2.18415484e-01 1.44219860e-01 -4.02960569e-01
6.87381506e-01 1.41767776e-02 -2.52333954e-02 3.05139590e-02
1.71595830e-02 5.11887185e-02 9.59358573e-01 -1.47486866e+00
-1.09779561e+00 -2.02867404e-01 8.08482766e-01 5.35763383e-01
-3.50144386e-01 -2.75072120e-02 3.86386253e-02 -5.38958013e-01
-6.17190182e-01 -8.43068659e-01 -5.42650938e-01 -6.72869086e-01
-1.23028263e-01 -3.26662034e-01 4.14687276e-01 -3.05869073e-01
1.42647576e+00 -1.81070340e+00 2.83279002e-01 6.61640406e-01
-7.67753050e-02 2.28648726e-02 -2.98679709e-01 3.47978383e-01
1.76632494e-01 1.52181506e-01 -3.86372626e-01 5.86196373e-04
3.43727022e-01 5.41711867e-01 -2.27023199e-01 3.94532233e-01
3.54511999e-02 5.59093833e-01 -1.05855167e+00 -1.36939272e-01
-1.03261461e-02 -1.94448635e-01 -8.68412614e-01 3.35990191e-01
-5.69098413e-01 2.84778953e-01 -9.13935721e-01 4.43904281e-01
4.65390056e-01 -1.45395681e-01 2.74828166e-01 4.55236226e-01
-1.47143990e-01 -8.29372182e-02 -1.65177524e+00 1.03899813e+00
-6.54122055e-01 -2.29175121e-01 8.39329511e-02 -1.48132563e+00
6.24461055e-01 2.08344683e-01 9.42398608e-01 -4.73887980e-01
1.16456926e-01 2.71552056e-01 7.83113018e-02 -2.40213275e-01
4.78095621e-01 -5.10802865e-01 -1.42633304e-01 8.47325027e-01
-2.39492640e-01 3.72493900e-02 4.48877454e-01 -1.35144964e-01
8.99160147e-01 4.72354069e-02 5.53415060e-01 -4.02065605e-01
4.71098423e-01 -2.10346803e-01 7.87015557e-01 9.37113762e-01
4.47160974e-02 2.51686513e-01 8.37220669e-01 -1.92668661e-01
-7.69332707e-01 -8.95493925e-01 1.24486871e-01 1.17359352e+00
-7.85267949e-02 7.69106001e-02 -5.32199442e-01 -8.31590474e-01
2.72139907e-01 7.25055218e-01 -4.79450166e-01 -1.13297187e-01
-2.62615770e-01 -1.00594079e+00 -5.22583760e-02 3.85114178e-02
1.38705745e-01 -9.48281050e-01 -1.12765515e+00 4.89990473e-01
-1.96955040e-01 -6.78799987e-01 -6.44472718e-01 3.98955107e-01
-5.05605102e-01 -1.02221823e+00 -8.84274244e-01 2.55806930e-03
4.80226308e-01 -7.21148029e-02 1.08752823e+00 -4.00841743e-01
2.07505241e-01 6.38463676e-01 -3.96795869e-01 -6.00036025e-01
-1.75179735e-01 -2.89862473e-02 1.72233954e-01 4.50839669e-01
-1.07666254e-01 -3.83039862e-01 -5.76492131e-01 2.12154359e-01
-1.26848495e+00 -4.34871316e-01 5.72958231e-01 9.99921858e-01
7.01169133e-01 4.59024966e-01 1.04042077e+00 -7.24817812e-01
1.14135122e+00 -6.57648563e-01 -1.02631533e+00 5.83373427e-01
-6.09578550e-01 6.25964165e-01 8.57634604e-01 -6.64513707e-01
-1.00229502e+00 -2.32878670e-01 1.57429576e-01 -3.22766304e-01
1.62428081e-01 9.10306990e-01 -3.00120533e-01 1.91744894e-01
2.22152829e-01 9.02331993e-02 -5.91374524e-02 -2.32059285e-01
3.52063715e-01 7.10797966e-01 6.18868843e-02 -1.16651046e+00
4.74340260e-01 1.24345370e-01 5.43444119e-02 -5.35524070e-01
-1.01562369e+00 -4.53586400e-01 -1.08481729e-02 -2.71614403e-01
3.64570618e-01 -4.60475266e-01 -9.06671882e-01 -1.29200637e-01
-8.53953898e-01 -3.42968374e-01 -5.78003109e-01 4.81689930e-01
-1.18756688e+00 2.60319352e-01 3.37585211e-01 -1.43292964e+00
-1.84755608e-01 -1.30973899e+00 4.85429049e-01 2.28361577e-01
-1.64835140e-01 -1.05738747e+00 5.27655743e-02 2.04055488e-01
2.58539230e-01 4.38580543e-01 1.03651392e+00 -4.90068167e-01
-4.37456071e-01 2.49238044e-01 9.34088528e-02 3.66095424e-01
2.21316859e-01 -2.85930574e-01 -4.16561693e-01 -5.56573093e-01
-2.99228001e-02 -4.56664145e-01 5.79377770e-01 7.70615458e-01
1.19127023e+00 -8.68236423e-01 8.31295848e-02 3.82138997e-01
1.55631983e+00 6.54584289e-01 2.10997149e-01 6.06746316e-01
-4.81744632e-02 6.95554376e-01 8.81908596e-01 1.02817988e+00
1.87320128e-01 6.90524638e-01 7.38340557e-01 3.35667133e-01
8.51346195e-01 -1.14540718e-01 3.49810511e-01 1.42487109e-01
-1.92454055e-01 -4.61256534e-01 -7.15091050e-01 4.70640749e-01
-2.15173435e+00 -1.08631265e+00 5.79537749e-01 2.57978821e+00
9.13215220e-01 1.83222964e-01 6.16787493e-01 2.20273450e-01
5.84777117e-01 9.08912122e-02 -1.03104925e+00 -8.50754797e-01
-4.67136921e-03 8.50224867e-02 5.58973968e-01 6.77709818e-01
-9.94603097e-01 4.57519442e-01 5.81249619e+00 1.05911100e+00
-1.07394445e+00 -1.27642930e-01 8.20701241e-01 -1.93027943e-01
-7.17491746e-01 2.90776566e-02 -7.08184958e-01 5.68728924e-01
9.53678131e-01 -5.07031798e-01 8.29808772e-01 8.45230222e-01
5.64878106e-01 -2.39617422e-01 -8.75233054e-01 8.52725804e-01
-4.72920477e-01 -8.64493668e-01 -1.29349068e-01 5.02823472e-01
8.92281234e-01 -2.83672720e-01 1.61697775e-01 3.78509402e-01
9.06357050e-01 -1.08324766e+00 7.12618291e-01 5.46257436e-01
7.12923944e-01 -1.46130848e+00 4.75731999e-01 6.96837127e-01
-8.65219653e-01 -6.87996209e-01 -3.03848743e-01 1.47627657e-02
1.13294832e-01 7.35201359e-01 -6.24482214e-01 7.20287025e-01
3.40394944e-01 2.29474485e-01 1.45144209e-01 1.24159884e+00
-1.41545102e-01 4.41904008e-01 -3.81296188e-01 -3.93109590e-01
8.31881344e-01 -5.12288630e-01 5.17739058e-01 7.72291481e-01
5.55840552e-01 1.30038604e-01 4.75026011e-01 7.83436358e-01
8.29481035e-02 2.35398456e-01 -4.27473843e-01 -6.26948476e-02
2.72394240e-01 1.12021697e+00 -5.15053988e-01 1.37129694e-01
-2.35191777e-01 5.74739695e-01 2.84032583e-01 4.85372573e-01
-8.17057252e-01 -1.71896592e-01 9.69044209e-01 -2.72139877e-01
3.66104782e-01 9.25702602e-02 -2.79374756e-02 -8.84550989e-01
3.00689694e-02 -1.08247244e+00 7.56508589e-01 -1.35266453e-01
-1.34996450e+00 1.39988661e-01 3.28180462e-01 -1.14463818e+00
-5.84027052e-01 -4.62553233e-01 -4.66833651e-01 8.17637622e-01
-1.73533332e+00 -4.69577849e-01 4.50948358e-01 5.38574100e-01
4.22252953e-01 -2.04916313e-01 6.33565545e-01 -6.33005891e-03
-4.36360389e-01 3.66453558e-01 6.50932789e-01 -5.39458096e-01
3.93167764e-01 -1.57279015e+00 -6.16543114e-01 4.66393381e-01
-1.01905011e-01 1.83158994e-01 8.43729854e-01 -2.94543535e-01
-1.35673463e+00 -9.24003780e-01 3.29211771e-01 5.43126166e-02
6.96148515e-01 6.72355667e-02 -4.89214927e-01 3.89055848e-01
3.34338695e-02 -8.49871635e-02 6.25320375e-01 1.63512617e-01
1.05086766e-01 -3.75947922e-01 -1.24283671e+00 6.56801999e-01
5.57513833e-01 -3.13837342e-02 -3.24601620e-01 6.68392777e-02
4.74805146e-01 -1.63763568e-01 -8.69790792e-01 3.72465730e-01
5.63328743e-01 -7.14736938e-01 8.48921537e-01 -7.98277855e-01
3.64738911e-01 -6.39829859e-02 -4.25066710e-01 -1.94099760e+00
-2.70384341e-01 -8.82580996e-01 -3.38579059e-01 9.53225851e-01
3.97292733e-01 -7.54403234e-01 6.26148999e-01 6.32687747e-01
1.81694165e-01 -1.17113078e+00 -1.18246925e+00 -1.14822042e+00
2.44073018e-01 -2.56085217e-01 8.95042956e-01 7.39984691e-01
-1.31468222e-01 9.83602926e-02 -5.96313715e-01 3.47113386e-02
7.19974756e-01 4.41799611e-01 4.56829816e-01 -1.05710506e+00
-6.90591455e-01 -6.58403456e-01 1.87644169e-01 -1.04532051e+00
1.34234115e-01 -6.42067790e-01 -1.99739207e-02 -1.51568437e+00
1.84374228e-01 -6.60909534e-01 -7.53269613e-01 2.70853221e-01
-1.43594161e-01 -6.00439012e-01 3.66985351e-01 -1.87639222e-01
-7.13724613e-01 8.32278430e-01 1.30485082e+00 -1.86597809e-01
-3.96305382e-01 5.68876266e-01 -1.04805851e+00 5.06736279e-01
1.03239644e+00 -6.23376191e-01 -7.23541439e-01 -4.85940315e-02
4.19344097e-01 5.57548404e-01 -1.41163975e-01 -4.54814672e-01
-2.38931596e-01 -1.14929116e+00 -3.70251983e-02 -4.46747392e-01
1.79926157e-01 -8.80574942e-01 8.38358402e-02 3.68554860e-01
-4.16069657e-01 3.98198627e-02 -1.67915553e-01 5.60550690e-01
-1.63031295e-01 -6.40394986e-01 7.75018692e-01 -3.04672420e-01
-3.97139519e-01 4.57122564e-01 -3.29233348e-01 2.92923838e-01
1.28854156e+00 -3.29587832e-02 1.35733902e-01 -6.30059183e-01
-4.67237562e-01 7.09344566e-01 2.59635854e-03 3.11531097e-01
6.54837132e-01 -1.36834002e+00 -8.84465098e-01 -3.32047909e-01
2.10467447e-03 -1.32507339e-01 -2.68752240e-02 6.54475510e-01
1.36769369e-01 4.51979458e-01 -9.88083482e-02 -2.00982407e-01
-1.00932670e+00 4.57731634e-01 3.55380952e-01 -9.18237269e-01
-1.00288957e-01 4.22266781e-01 1.89105883e-01 -4.01285052e-01
1.19939297e-01 -1.20596766e-01 -2.53243119e-01 4.42390889e-01
3.97613287e-01 3.66011590e-01 -1.11902364e-01 -2.89813310e-01
3.56697366e-02 1.73455074e-01 9.93266478e-02 -4.54194158e-01
1.56621349e+00 -2.23034814e-01 1.16372876e-01 3.16337973e-01
8.74793828e-01 -1.34272918e-01 -1.43514061e+00 -3.16884398e-01
2.55302012e-01 -5.82049131e-01 1.67569607e-01 -9.72918093e-01
-9.19984460e-01 3.44852805e-01 3.09351206e-01 4.95053947e-01
1.34359848e+00 -4.02056962e-01 4.33853298e-01 4.55069453e-01
4.96635407e-01 -1.48788846e+00 6.62629902e-02 5.71648419e-01
8.75349224e-01 -1.17926788e+00 -2.14676842e-01 2.79648602e-01
-8.84132862e-01 9.79735613e-01 2.71741480e-01 1.65543601e-01
4.90170836e-01 1.01029105e-01 -2.59170115e-01 3.05052727e-01
-7.35608280e-01 -6.00368440e-01 3.78847569e-01 4.56945449e-01
8.16425383e-02 3.88614208e-01 -6.16690397e-01 7.05489933e-01
-8.51572379e-02 -1.01990275e-01 3.67923915e-01 1.01892471e+00
-6.04693592e-01 -1.42189479e+00 -6.95396841e-01 7.14030743e-01
-6.88851714e-01 2.27769881e-01 -4.11505401e-02 3.75184506e-01
-7.13778809e-02 1.15474749e+00 -3.01782548e-01 3.37126516e-02
3.28623444e-01 -1.28141716e-01 4.04675275e-01 -5.28066516e-01
-3.82555395e-01 1.44471779e-01 1.08003438e-01 -2.35428378e-01
-4.63232040e-01 -5.93727529e-01 -8.92412603e-01 -1.32965505e-01
-2.84615606e-01 5.30800819e-01 4.59994823e-01 1.12624502e+00
-5.53320087e-02 5.27940810e-01 1.04605544e+00 -4.40177977e-01
-1.32228160e+00 -5.04584551e-01 -8.27149034e-01 2.69256294e-01
4.65980709e-01 -8.18457901e-01 -1.44611880e-01 -5.09918034e-01]
|
[4.475583553314209, 2.5765879154205322]
|
ade0b5fa-da17-4eb2-8afd-9bd5ea3681d8
|
an-erudite-fine-grained-visual-classification
| null | null |
http://openaccess.thecvf.com//content/CVPR2023/html/Chang_An_Erudite_Fine-Grained_Visual_Classification_Model_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Chang_An_Erudite_Fine-Grained_Visual_Classification_Model_CVPR_2023_paper.pdf
|
An Erudite Fine-Grained Visual Classification Model
|
Current fine-grained visual classification (FGVC) models are isolated. In practice, we first need to identify the coarse-grained label of an object, then select the corresponding FGVC model for recognition. This hinders the application of the FGVC algorithm in real-life scenarios. In this paper, we propose an erudite FGVC model jointly trained by several different datasets, which can efficiently and accurately predict an object's fine-grained label across the combined label space. We found through a pilot study that positive and negative transfers co-occur when different datasets are mixed for training, i.e., the knowledge from other datasets is not always useful. Therefore, we first propose a feature disentanglement module and a feature re-fusion module to reduce negative transfer and boost positive transfer between different datasets. In detail, we reduce negative transfer by decoupling the deep features through many dataset-specific feature extractors. Subsequently, these are channel-wise re-fused to facilitate positive transfer. Finally, we propose a meta-learning based dataset-agnostic spatial attention layer to take full advantage of the multi-dataset training data, given that localisation is dataset-agnostic between different datasets. Experimental results across 11 different mixed-datasets built on four different FGVC datasets demonstrate the effectiveness of the proposed method. Furthermore, the proposed method can be easily combined with existing FGVC methods to obtain state-of-the-art results.
|
['Zhanyu Ma', 'Yi-Zhe Song', 'Timothy Hospedales', 'Ruoyi Du', 'Yujun Tong', 'Dongliang Chang']
|
2023-01-01
| null | null | null |
cvpr-2023-1
|
['fine-grained-image-classification']
|
['computer-vision']
|
[ 1.97959319e-01 -4.44319695e-01 -1.16706237e-01 -5.77892482e-01
-7.90352643e-01 -7.32680380e-01 6.96714103e-01 -6.83581531e-02
-4.57052737e-01 6.70705616e-01 -5.14476188e-02 1.12913713e-01
-7.88210034e-02 -6.23768449e-01 -7.56244957e-01 -8.60882103e-01
3.52114052e-01 1.87962428e-01 2.38971114e-01 2.32962266e-01
2.52641380e-01 7.29720056e-01 -1.80676806e+00 7.91677773e-01
8.21141958e-01 1.49569249e+00 6.02086604e-01 3.28834623e-01
-2.22969681e-01 8.05664778e-01 -5.68088353e-01 -9.48024318e-02
1.97821021e-01 -2.42108151e-01 -7.30012357e-01 1.21287286e-01
7.28039384e-01 -1.63016513e-01 1.92142185e-03 9.51804757e-01
5.05348980e-01 -2.03232076e-02 8.18233907e-01 -1.37497246e+00
-6.78881884e-01 1.76290467e-01 -7.11170197e-01 1.57876909e-01
-1.93797484e-01 2.53015727e-01 8.52875590e-01 -1.10074723e+00
5.24266243e-01 1.15083528e+00 4.85592186e-01 4.31908846e-01
-1.19494939e+00 -9.60784376e-01 6.13325477e-01 2.36874178e-01
-1.68198097e+00 -3.15363854e-01 8.77303958e-01 -5.33042252e-01
7.41475642e-01 3.36854935e-01 5.45958340e-01 1.02730954e+00
3.66655029e-02 8.66063118e-01 1.39063907e+00 -3.02324086e-01
1.39399946e-01 3.07529092e-01 4.59331200e-02 5.52325428e-01
8.51123333e-02 1.52238429e-01 -4.91492957e-01 1.79376096e-01
6.50465071e-01 2.11546823e-01 -3.00070733e-01 -5.69084108e-01
-1.27452123e+00 7.22218990e-01 9.69720006e-01 4.30978566e-01
-3.35060626e-01 1.70112893e-01 4.05251503e-01 2.23431438e-01
3.46591204e-01 3.93215925e-01 -4.57237720e-01 2.48902872e-01
-9.91593957e-01 5.43858409e-02 3.19756627e-01 9.87323046e-01
9.75365460e-01 -2.06755131e-01 -6.07751369e-01 8.64969015e-01
4.10319418e-01 4.62386310e-01 5.89543223e-01 -5.60769200e-01
5.27163446e-01 8.00283313e-01 -1.08422115e-02 -1.17309630e+00
-4.87157226e-01 -8.29007745e-01 -8.32933247e-01 1.73546106e-01
2.37107202e-01 1.50926217e-01 -9.79734480e-01 1.78018951e+00
2.42491841e-01 4.07303363e-01 -1.59800172e-01 1.12197971e+00
9.89220679e-01 3.73680025e-01 2.42757782e-01 4.65068407e-02
1.30519962e+00 -1.10245037e+00 -4.17097658e-01 -4.63895828e-01
6.31545484e-01 -4.71932441e-01 1.09835672e+00 2.22255364e-01
-5.20714819e-01 -8.27890992e-01 -1.20752370e+00 -7.31843710e-02
-6.30634725e-01 5.43527007e-01 6.65632367e-01 4.35819298e-01
-7.70217001e-01 4.00745511e-01 -6.06375992e-01 -8.82885605e-02
8.35061133e-01 5.05797982e-01 -5.12343943e-01 -2.25861117e-01
-9.72187698e-01 7.02859282e-01 5.41431308e-01 3.79669905e-01
-8.75410557e-01 -6.39708281e-01 -5.87476909e-01 5.83972670e-02
1.24694876e-01 -4.57300246e-01 8.19889724e-01 -1.27870905e+00
-9.74415481e-01 8.22061360e-01 -1.01182468e-01 -4.82393103e-03
4.73276138e-01 2.04312000e-02 -4.20528859e-01 7.08522648e-02
2.23959222e-01 8.38574886e-01 8.77305627e-01 -1.50530922e+00
-9.54533994e-01 -4.40473586e-01 4.44320366e-02 2.96270490e-01
-4.49568570e-01 -3.45412761e-01 -5.94426394e-01 -6.48814023e-01
-1.27121374e-01 -7.54434168e-01 1.36903703e-01 1.80199757e-01
-2.61600703e-01 -1.39976546e-01 9.05258894e-01 -3.03051710e-01
1.06128657e+00 -2.38577414e+00 1.66340247e-01 2.21212402e-01
1.81306347e-01 3.31941873e-01 -3.50598752e-01 1.35003895e-01
-4.32083569e-02 -6.81964234e-02 -1.81298777e-02 -2.49794662e-01
-1.71642721e-01 1.11347320e-03 -1.71051279e-01 5.61997116e-01
5.77327728e-01 1.13159239e+00 -7.31568396e-01 -4.63505238e-01
3.24148864e-01 4.68948454e-01 -3.41797501e-01 1.44884378e-01
-2.35550608e-02 4.94440198e-01 -6.30238533e-01 6.92465246e-01
9.44030464e-01 -5.71948528e-01 6.54914603e-02 -7.29867041e-01
8.90213437e-03 -2.61732489e-01 -1.17153823e+00 1.69831538e+00
-6.55268669e-01 4.97527599e-01 -5.12582175e-02 -9.18082297e-01
8.78688693e-01 -3.77819799e-02 1.08473182e-01 -9.87551033e-01
3.17450732e-01 2.17693284e-01 -1.12515211e-01 -2.54413366e-01
2.67076701e-01 -1.24816671e-01 -1.21698610e-01 4.15100753e-01
3.20722282e-01 3.01073581e-01 -5.26595414e-02 -8.63244534e-02
7.47133434e-01 5.28844818e-02 3.12527478e-01 -2.53561586e-01
5.09476840e-01 -9.36515927e-02 4.37909484e-01 6.75121605e-01
-3.63462657e-01 7.01030374e-01 8.41645971e-02 -1.48828909e-01
-4.64441091e-01 -7.93393672e-01 -2.08090439e-01 1.31593788e+00
5.46589851e-01 -5.01253605e-02 -5.47073960e-01 -1.17211986e+00
2.25010589e-01 3.72521937e-01 -9.20937657e-01 -3.56696278e-01
-3.28839719e-01 -7.09625840e-01 4.56676513e-01 7.95023084e-01
6.86966896e-01 -1.03745520e+00 -6.58486247e-01 3.60760130e-02
-7.13209063e-02 -9.72123325e-01 -3.08417201e-01 4.60626096e-01
-5.23433208e-01 -9.80743825e-01 -6.02405131e-01 -8.83486271e-01
6.55005753e-01 5.45757949e-01 9.04083431e-01 8.41751024e-02
-2.23601937e-01 7.43732676e-02 -3.28352481e-01 -1.84369281e-01
1.72033221e-01 1.69608623e-01 -2.65186578e-01 5.28395772e-01
3.45204145e-01 -3.05712968e-01 -7.93350399e-01 3.61099511e-01
-7.20004499e-01 1.07592031e-01 7.14932203e-01 1.11789227e+00
8.20892751e-01 -6.04831614e-03 8.71728182e-01 -9.71527636e-01
3.06245506e-01 -3.94249052e-01 -4.28925395e-01 5.27448773e-01
-3.05451006e-01 -4.03002426e-02 7.52712309e-01 -5.67763150e-01
-1.11894262e+00 1.88676387e-01 6.29081503e-02 -6.83771491e-01
-2.85119176e-01 2.64764786e-01 -4.58936423e-01 -3.46036941e-01
5.35595119e-01 1.92920119e-01 -3.42619061e-01 -5.70891321e-01
5.45478582e-01 8.83836627e-01 4.07697827e-01 -3.34980339e-01
5.98796487e-01 5.19298792e-01 -2.75329381e-01 -2.69533992e-01
-8.66645753e-01 -4.78925526e-01 -8.28663707e-01 -2.28459761e-01
7.21489608e-01 -1.20657098e+00 -5.98527312e-01 4.90472645e-01
-8.89096022e-01 -2.89254338e-01 -2.02448040e-01 3.37533921e-01
-2.71878481e-01 -1.13262385e-01 -2.15834171e-01 -3.77669364e-01
-1.58840850e-01 -1.20153868e+00 1.38604772e+00 3.77419323e-01
4.93679270e-02 -9.18978214e-01 -2.52368063e-01 1.51003510e-01
4.23963934e-01 -5.80635406e-02 6.79686785e-01 -6.57005548e-01
-6.14191055e-01 -1.09853469e-01 -8.74738634e-01 2.29783967e-01
3.64524573e-01 -3.43525171e-01 -1.40991735e+00 -5.07472515e-01
-2.87013501e-01 -6.55156076e-01 1.19775522e+00 1.27249971e-01
1.44309902e+00 -1.19416684e-01 -6.83595955e-01 7.79066563e-01
1.59995520e+00 6.97834976e-03 2.23037094e-01 3.24242949e-01
1.15602767e+00 5.20525873e-01 8.53601515e-01 3.05680841e-01
4.94686216e-01 8.15651596e-01 5.22593439e-01 -2.78498292e-01
-6.12185240e-01 -1.66273162e-01 3.16686444e-02 4.38996285e-01
1.56793907e-01 -2.49840409e-01 -6.92523539e-01 5.50650537e-01
-1.78116405e+00 -6.64212823e-01 1.12369329e-01 1.88078964e+00
6.77171588e-01 -2.27207035e-01 -1.48414120e-01 1.40243610e-02
6.26991749e-01 1.39922589e-01 -6.54140234e-01 -4.51803319e-02
-1.52228460e-01 1.15169190e-01 5.29227436e-01 7.70984292e-02
-1.44635928e+00 1.01707697e+00 5.31642103e+00 1.21036291e+00
-1.75500119e+00 3.05615783e-01 6.27715945e-01 -2.85359651e-01
-2.43153676e-01 -3.71883988e-01 -8.72859836e-01 5.07969260e-01
4.85467553e-01 2.41134435e-01 2.62067258e-01 7.43605912e-01
-3.13145250e-01 3.26643363e-02 -1.18829238e+00 1.14642191e+00
1.40520811e-01 -1.21171045e+00 9.36364639e-04 -7.33201429e-02
7.74603069e-01 1.02620916e-02 2.46988386e-01 3.38202924e-01
4.52160239e-02 -1.06692314e+00 9.02778566e-01 4.16754156e-01
1.11758113e+00 -7.00743735e-01 7.55107462e-01 2.71587163e-01
-1.36536181e+00 -3.36307287e-01 -4.62902099e-01 1.55507505e-01
-1.81639567e-01 5.65968037e-01 -6.32762551e-01 7.47071028e-01
7.27156878e-01 8.39372873e-01 -1.09982610e+00 9.82245624e-01
1.50107965e-02 2.95000285e-01 -1.37482345e-01 7.95649737e-02
2.41371110e-01 3.74291360e-01 -1.10548496e-01 1.26200294e+00
2.27551416e-01 -1.28693014e-01 1.44843802e-01 7.28417873e-01
-9.13580284e-02 7.22124651e-02 -4.04239148e-01 1.51692227e-01
4.46623087e-01 1.48331022e+00 -8.16550374e-01 -1.53185248e-01
-5.46488762e-01 1.15764117e+00 7.65958250e-01 5.52438617e-01
-9.20637250e-01 -2.53019571e-01 4.37532395e-01 -2.03894794e-01
7.06611931e-01 3.21295738e-01 -4.52177703e-01 -1.28024650e+00
-9.54129845e-02 -7.10816443e-01 3.88240367e-01 -5.81899941e-01
-1.55447161e+00 8.19074333e-01 -3.14742237e-01 -1.42457521e+00
-1.66140161e-02 -6.30510390e-01 -3.28747690e-01 9.47905481e-01
-1.69575846e+00 -1.64332759e+00 -5.75640082e-01 7.34584868e-01
3.52565169e-01 -1.77000403e-01 7.77484953e-01 4.68940675e-01
-5.90078831e-01 9.30047274e-01 2.38308720e-02 5.40188514e-02
7.81101882e-01 -9.45037007e-01 9.02477466e-03 6.93489015e-01
1.84910670e-01 4.13804501e-01 9.69921649e-02 -5.01223326e-01
-1.19994473e+00 -1.76249564e+00 5.45762897e-01 -2.92546958e-01
2.39303023e-01 -5.28928399e-01 -1.00012481e+00 5.20148933e-01
-1.01339899e-01 7.22165167e-01 5.72146952e-01 3.91246900e-02
-6.24849260e-01 -3.89458686e-01 -1.18303454e+00 2.79013276e-01
1.18888009e+00 -7.32828140e-01 -3.16436231e-01 5.24795279e-02
4.52055216e-01 -1.38337418e-01 -6.70606732e-01 6.37457192e-01
6.01377964e-01 -8.75555336e-01 9.52227414e-01 -4.02245551e-01
1.61512986e-01 -5.53024113e-01 -4.35385823e-01 -1.41294611e+00
-7.20959127e-01 4.04783130e-01 2.50553880e-02 1.45750499e+00
2.41697043e-01 -6.07929349e-01 4.79169518e-01 2.50315875e-01
-3.85327451e-02 -7.94323742e-01 -9.75046813e-01 -6.38284147e-01
-1.76125877e-02 -3.55213076e-01 7.55670488e-01 1.04436839e+00
-3.49348992e-01 3.63720238e-01 -2.85382330e-01 1.92462981e-01
3.69026691e-01 6.71401680e-01 5.47527611e-01 -1.19530332e+00
-2.81038076e-01 -4.08034235e-01 -5.37786961e-01 -9.15787280e-01
2.20712364e-01 -1.18725085e+00 1.24801688e-01 -1.41836667e+00
4.66567487e-01 -7.98013091e-01 -7.16045141e-01 7.27636397e-01
-2.60307610e-01 7.34818041e-01 4.49055463e-01 1.86500967e-01
-7.42046058e-01 6.81372821e-01 1.40331328e+00 -3.42065036e-01
-3.67758088e-02 -2.74136841e-01 -9.30808544e-01 4.11781669e-01
5.85937381e-01 -3.56554240e-01 -5.76272547e-01 -3.37994933e-01
-1.71398818e-01 -3.41738343e-01 5.96549511e-01 -9.94668126e-01
2.05932066e-01 -1.51214555e-01 1.00471401e+00 -5.33524811e-01
3.02308738e-01 -1.06691074e+00 1.05138451e-01 1.75059125e-01
-3.07995439e-01 -4.79459018e-01 4.01894927e-01 6.13200665e-01
-4.46102232e-01 9.52420011e-02 8.42065096e-01 7.92920738e-02
-1.11294699e+00 3.38083655e-01 1.55043080e-01 1.21516567e-02
1.24704790e+00 -1.72892392e-01 -5.92568099e-01 8.94591287e-02
-4.73319322e-01 2.27531999e-01 5.03522575e-01 5.32081366e-01
6.28526986e-01 -1.59756112e+00 -5.21364033e-01 5.71536601e-01
6.04305029e-01 -9.90146622e-02 7.10422575e-01 7.86148906e-01
-8.50050990e-03 4.93948489e-01 -4.78700817e-01 -8.38230550e-01
-1.13660669e+00 6.86045289e-01 3.12598407e-01 -2.41398811e-01
-3.14693481e-01 1.11197925e+00 8.47833216e-01 -5.77250421e-01
1.05583645e-01 -2.95289189e-01 -3.55450928e-01 4.21259403e-01
6.62247419e-01 1.47193633e-02 2.94026732e-01 -8.40054929e-01
-7.58072734e-01 8.65108252e-01 -2.18534306e-01 1.88821554e-01
1.17930782e+00 -2.83394873e-01 2.32796595e-01 4.88045722e-01
1.61896002e+00 -2.66088039e-01 -1.55583072e+00 -2.88888991e-01
-3.27208579e-01 -6.25418544e-01 1.17968537e-01 -1.04075289e+00
-1.53247786e+00 1.07281852e+00 1.02186549e+00 -1.93793595e-01
1.50336492e+00 1.31867304e-01 2.43437946e-01 1.48641691e-02
4.19372618e-01 -8.75799060e-01 -3.22900936e-02 2.42778808e-01
9.68318999e-01 -1.38180852e+00 -1.77842349e-01 -4.04649168e-01
-8.39372635e-01 7.36837268e-01 9.10766006e-01 -5.98023944e-02
7.13142991e-01 1.50983781e-01 1.34328589e-01 -1.95826277e-01
-8.05997550e-01 -3.32745343e-01 5.93911648e-01 6.51177764e-01
2.67675072e-01 2.56663382e-01 2.73103714e-02 9.53148067e-01
1.86392173e-01 2.42058650e-01 -2.15433184e-02 7.96939075e-01
-3.19025218e-01 -8.87396753e-01 -1.95369422e-01 6.38201296e-01
-1.04909264e-01 -9.84786898e-02 -3.71249765e-01 6.96797073e-01
4.69331652e-01 8.54585826e-01 2.60680705e-01 -5.98917663e-01
3.63206655e-01 5.12630157e-02 6.27887964e-01 -5.35369337e-01
-7.61991739e-01 1.25056207e-02 -1.42170832e-01 -7.65927792e-01
-5.12505591e-01 -3.97408068e-01 -1.18197978e+00 -4.94912192e-02
-4.27095175e-01 -2.49625847e-01 3.83909047e-01 9.01846111e-01
6.81237876e-01 7.76249468e-01 7.17032611e-01 -9.72864687e-01
-2.74484068e-01 -9.03689742e-01 -7.11918354e-01 4.40135449e-01
4.52707887e-01 -1.13608098e+00 -2.47214645e-01 -1.95955187e-01]
|
[9.700075149536133, 2.07536244392395]
|
cb7be7c3-aead-4468-80a3-3485a48c88b7
|
semantic-role-labeling-in-conversational-chat
| null | null |
https://aclanthology.org/Y18-1064
|
https://aclanthology.org/Y18-1064.pdf
|
Semantic Role Labeling in Conversational Chat using Deep Bi-Directional Long Short-Term Memory Networks with Attention Mechanism
| null |
['Fariz Ikhwantri', 'Ahmad Rizqi Meydiarso', 'Alfan Farizki Wicaksono', 'Rahmad Mahendra', 'Valdi Rachman']
| null | null | null | null |
paclic-2018-12
|
['semantic-role-labeling']
|
['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]
|
[-7.3783860206604, 3.729959726333618]
|
0193f076-edb7-4e55-a5e5-d16ee6461488
|
optimised-preprocessing-for-automatic-mouth
| null | null |
https://aclanthology.org/2020.signlang-1.5
|
https://aclanthology.org/2020.signlang-1.5.pdf
|
Optimised Preprocessing for Automatic Mouth Gesture Classification
|
Mouth gestures are facial expressions in sign language, that do not refer to lip patterns of a spoken language. Research on this topic has been limited so far. The aim of this work is to automatically classify mouth gestures from video material by training a neural network. This could render time-consuming manual annotation unnecessary and help advance the field of automatic sign language translation. However, it is a challenging task due to the little data available as training material and the similarity of different mouth gesture classes. In this paper we focus on the preprocessing of the data, such as finding the area of the face important for mouth gesture recognition. Furthermore we analyse the duration of mouth gestures and determine the optimal length of video clips for classification. Our experiments show, that this can improve the classification results significantly and helps to reach a near human accuracy.
|
['Rolf-Rainer Grigat', 'Maren Brumm']
|
2020-05-01
| null | null | null |
lrec-2020-5
|
['sign-language-translation']
|
['computer-vision']
|
[ 3.92088771e-01 -2.01266736e-01 -3.44385445e-01 -5.66128314e-01
-3.25047106e-01 -3.25360239e-01 5.98242342e-01 -4.61406708e-01
-5.88627577e-01 5.74779630e-01 1.74648046e-01 -2.04767570e-01
1.07611194e-01 -2.16360703e-01 -1.45288631e-01 -9.66013014e-01
1.38654262e-01 3.10406834e-01 2.03948736e-01 9.47890878e-02
4.66686666e-01 9.00613010e-01 -1.90265691e+00 3.15412670e-01
1.95745766e-01 7.34682798e-01 -9.59335268e-03 5.79917610e-01
-4.45076734e-01 4.30223882e-01 -6.71801984e-01 -3.27853449e-02
1.26024410e-01 -8.40151548e-01 -7.50425100e-01 2.47402146e-01
4.57433641e-01 -4.64801133e-01 5.69656417e-02 8.66576135e-01
6.52365148e-01 1.49921030e-01 7.69328535e-01 -1.18402863e+00
-1.09169744e-02 2.41381779e-01 -2.87273347e-01 7.89794922e-02
2.57897049e-01 8.79998505e-03 4.63729262e-01 -6.72326803e-01
9.18306887e-01 1.00767326e+00 3.81958127e-01 1.08840537e+00
-7.72574723e-01 -5.63151360e-01 -6.87390268e-02 3.99251461e-01
-1.30801451e+00 -8.83749604e-01 9.31370437e-01 -3.09252352e-01
8.08894396e-01 2.22009614e-01 6.59584820e-01 1.01293755e+00
-3.65175724e-01 8.71413171e-01 1.15448439e+00 -9.66074407e-01
2.66142279e-01 -3.90765034e-02 1.47770852e-01 5.88353634e-01
-1.14915490e-01 -1.62461743e-01 -4.69360828e-01 3.54026467e-01
7.09856927e-01 -1.92624182e-01 -1.68982625e-01 -2.64039308e-01
-8.81932437e-01 6.89288735e-01 5.58071658e-02 1.03583133e+00
-5.11338890e-01 -1.37476921e-02 3.88074338e-01 4.15175080e-01
1.35346711e-01 7.13069830e-03 -3.73274207e-01 -4.82147545e-01
-1.27072406e+00 -7.78108165e-02 1.04818201e+00 4.79808867e-01
2.43951246e-01 -8.38413090e-02 2.82326967e-01 7.38650620e-01
5.26696444e-01 4.55757618e-01 5.53413510e-01 -8.61067891e-01
1.65815219e-01 6.29995584e-01 -1.44134983e-01 -5.96567631e-01
-4.32747066e-01 6.35102630e-01 -3.50757509e-01 5.94005108e-01
1.16378725e+00 -2.97848552e-01 -1.17542410e+00 1.36169994e+00
3.42491835e-01 -3.14250857e-01 -1.92942452e-02 9.91820157e-01
7.77425826e-01 4.20943469e-01 8.40800703e-02 -4.50709462e-01
1.38784707e+00 -6.80035651e-01 -1.04621816e+00 2.37697223e-03
6.31920338e-01 -1.02750194e+00 1.02299094e+00 4.69380647e-01
-6.92475021e-01 -1.27593160e-01 -7.51146317e-01 6.21324964e-02
-5.68334103e-01 6.03071630e-01 5.78850210e-01 1.01924908e+00
-8.74107540e-01 3.60510558e-01 -8.56447935e-01 -8.05264533e-01
3.47871184e-01 6.52277589e-01 -5.49063683e-01 9.70560089e-02
-7.21786857e-01 8.91209483e-01 1.78817570e-01 3.22014600e-01
-5.06823920e-02 5.70239685e-02 -5.74253261e-01 -3.30515921e-01
3.10222566e-01 1.26146659e-01 1.33825409e+00 -1.38351321e+00
-1.96218169e+00 1.01519918e+00 -5.22349656e-01 -1.45213768e-01
6.47929847e-01 1.02851853e-01 -2.60550052e-01 3.56126010e-01
-4.57343727e-01 9.17499006e-01 1.01429105e+00 -1.05896151e+00
-6.64418817e-01 -5.32120287e-01 -3.54758054e-01 -8.68732557e-02
-1.77689239e-01 5.95254779e-01 -5.61968446e-01 -5.04514635e-01
1.55875206e-01 -1.03108180e+00 1.65524155e-01 1.70850173e-01
1.47064790e-01 -5.58754086e-01 1.18898904e+00 -1.01512015e+00
9.55970228e-01 -2.11281133e+00 -1.04138516e-01 3.52614254e-01
-3.11351627e-01 5.70764661e-01 -9.42022726e-02 2.94655234e-01
2.85724755e-02 9.98863368e-04 -1.16576657e-01 -2.95107126e-01
-1.11132585e-01 6.23383880e-01 -1.03197120e-01 5.66860914e-01
1.26237825e-01 6.36587143e-01 -4.02675539e-01 -7.58891106e-01
4.13356215e-01 7.36867130e-01 -1.33514881e-01 1.53671131e-01
8.17046873e-03 4.89194185e-01 -3.84841442e-01 9.15127277e-01
3.64849776e-01 4.87755477e-01 1.21427909e-01 -9.62558836e-02
-1.88133985e-01 1.86728448e-01 -1.22345543e+00 1.23803103e+00
-3.83577913e-01 1.12979031e+00 2.07882017e-01 -1.05904973e+00
9.73671615e-01 5.76828122e-01 6.07276738e-01 -5.72620749e-01
6.96800709e-01 4.86025542e-01 2.48755038e-01 -1.00364685e+00
1.08567819e-01 -1.76411286e-01 4.94975954e-01 6.07815206e-01
-1.59411550e-01 -4.25584540e-02 5.18624961e-01 -7.44239748e-01
4.58014518e-01 2.33945504e-01 2.90255010e-01 4.20125872e-02
7.88270891e-01 -1.65689752e-01 5.52569702e-02 1.87801972e-01
-3.71701598e-01 4.49164003e-01 3.95553410e-01 -5.50987840e-01
-7.96926260e-01 -3.33782107e-01 1.47097977e-03 9.65648770e-01
-2.10641280e-01 9.60382819e-02 -1.24679589e+00 -6.67415619e-01
-3.82794499e-01 4.54916477e-01 -5.32313764e-01 3.11867237e-01
-1.00263643e+00 -6.14578068e-01 6.30855918e-01 4.79010433e-01
4.12668794e-01 -1.59175038e+00 -1.07333910e+00 4.11661193e-02
-1.20001420e-01 -1.10139394e+00 -3.06702256e-01 1.14290841e-01
-1.08401763e+00 -1.15399444e+00 -1.01353800e+00 -9.51267064e-01
9.17202473e-01 -1.29691839e-01 4.06513780e-01 2.64017999e-01
-4.43151325e-01 2.68784463e-01 -6.98301911e-01 -5.50365746e-01
-6.02971137e-01 4.78303544e-02 6.22208044e-02 1.10043354e-01
1.04167998e+00 -3.06788027e-01 -1.93714544e-01 6.06548727e-01
-8.43348444e-01 -4.03306544e-01 6.27403736e-01 6.03184700e-01
2.27481142e-01 -3.29013675e-01 2.14476794e-01 -1.79689035e-01
6.51212931e-01 2.25530326e-01 -5.58685482e-01 2.39907846e-01
-3.97250712e-01 1.43436328e-01 1.94861084e-01 -7.05054522e-01
-7.91831970e-01 5.00241637e-01 -3.58013064e-01 -8.81247371e-02
-6.99602544e-01 -8.85853730e-03 -2.28654310e-01 -3.81450623e-01
3.64256322e-01 2.00043425e-01 5.36048651e-01 -6.43990695e-01
1.00177199e-01 1.19531357e+00 1.82931751e-01 -6.57886416e-02
3.32415044e-01 4.80077654e-01 7.70173967e-02 -1.52664423e+00
-2.50390451e-03 -7.09361017e-01 -1.16952276e+00 -5.44870675e-01
9.58874941e-01 -8.08678791e-02 -8.76473784e-01 5.66453993e-01
-1.17391121e+00 -3.36848646e-01 -3.71903926e-02 6.89582050e-01
-5.80631375e-01 5.65709949e-01 -1.84819639e-01 -1.14868987e+00
-2.74100453e-01 -1.18079233e+00 1.05730498e+00 2.27413923e-01
-4.62843537e-01 -7.46262491e-01 -1.35071337e-01 3.39391649e-01
2.43716076e-01 -2.40723789e-02 4.71081108e-01 -6.42130613e-01
-4.15567875e-01 -5.40798366e-01 -1.03108175e-01 4.47309077e-01
3.58573556e-01 4.11741853e-01 -1.02890885e+00 9.90013704e-02
-5.62743098e-02 -2.12370947e-01 7.46330380e-01 5.56117117e-01
7.14154422e-01 -4.03658338e-02 -3.14142942e-01 2.75599182e-01
9.36308324e-01 7.09657371e-01 7.05764890e-01 2.35169485e-01
2.55453527e-01 1.11015403e+00 6.86479807e-01 1.49714082e-01
-1.85731143e-01 8.94324601e-01 9.28451642e-02 -3.67962159e-02
-3.02513868e-01 1.04907952e-01 4.45927590e-01 5.34305274e-01
-6.17814600e-01 -1.58204257e-01 -9.01921153e-01 5.04408419e-01
-1.44726253e+00 -9.40493882e-01 -1.07597440e-01 1.98169243e+00
5.69145501e-01 -2.81054974e-01 4.52077210e-01 6.87648833e-01
6.51357353e-01 -2.08868697e-01 -1.52060896e-01 -6.29746616e-01
9.01776403e-02 3.21164191e-01 4.17801738e-01 6.93311751e-01
-1.27851295e+00 1.01741850e+00 6.12584591e+00 4.64180827e-01
-1.66196036e+00 -1.40677318e-01 1.30332455e-01 8.11925977e-02
4.60234493e-01 -3.24039102e-01 -8.38873863e-01 3.47322106e-01
7.77852118e-01 3.52630317e-01 4.51160610e-01 7.30340719e-01
4.84849781e-01 -2.49887049e-01 -1.16469634e+00 1.23518753e+00
4.80536729e-01 -6.88852072e-01 -1.19555704e-01 2.16386184e-01
2.05293149e-01 -1.45805880e-01 -4.11278933e-01 -8.23916495e-02
-4.88292128e-01 -1.03614831e+00 4.06992316e-01 3.93697143e-01
6.01081312e-01 -4.54246044e-01 9.20561075e-01 3.09752315e-01
-1.02964556e+00 3.27658355e-02 -6.08346658e-03 -1.12928472e-01
2.89333045e-01 -1.66414589e-01 -1.14144647e+00 -1.52624413e-01
4.93824601e-01 1.62209913e-01 -3.26917231e-01 1.22267747e+00
-3.37367833e-01 6.29926503e-01 -7.49573886e-01 -7.39418566e-01
2.64695555e-01 -1.07757032e-01 4.01566297e-01 1.17955315e+00
3.85158449e-01 8.10455009e-02 -1.81313843e-01 4.14820135e-01
1.98306918e-01 6.02161109e-01 -5.68665683e-01 -3.39447290e-01
2.62680138e-03 8.54542434e-01 -1.26632071e+00 -1.01355910e-01
-3.75962973e-01 1.01030707e+00 -4.82407629e-01 2.39841282e-01
-3.24835986e-01 -4.12087828e-01 5.22513986e-01 1.07161298e-01
3.57560009e-01 -2.90297210e-01 -1.43947154e-01 -8.83751214e-01
2.67213464e-01 -8.36469948e-01 1.13001637e-01 -3.45243245e-01
-7.57201195e-01 4.57215577e-01 2.46492047e-02 -1.24781644e+00
-8.05919349e-01 -1.12605548e+00 -4.81738120e-01 6.80079579e-01
-1.32494330e+00 -1.12303960e+00 -2.70527482e-01 4.56146508e-01
6.86535716e-01 -1.20873265e-01 9.19607222e-01 3.66607010e-01
-1.58192664e-01 6.66462779e-01 -5.78725114e-02 4.40925598e-01
6.63208127e-01 -8.07557106e-01 5.84278814e-02 6.43336535e-01
2.81908363e-01 4.65672642e-01 7.81752586e-01 -4.52891558e-01
-1.06894481e+00 -2.55619138e-01 1.34752989e+00 -3.15770090e-01
4.09033298e-01 3.61407697e-02 -7.23746777e-01 1.46408290e-01
3.85623164e-02 -2.38078400e-01 6.46661520e-01 -2.39072263e-01
3.47569324e-02 7.94427767e-02 -1.11548913e+00 6.33468270e-01
7.91858375e-01 -4.28600460e-01 -7.74818540e-01 1.59011304e-01
-2.22169504e-01 -1.07480377e-01 -4.74623233e-01 1.11928508e-01
1.02903175e+00 -6.82571411e-01 6.72325850e-01 -5.15031278e-01
4.27690707e-02 -1.23383336e-01 6.18132725e-02 -7.80550778e-01
6.51409447e-01 -3.80638808e-01 2.55737931e-01 1.18921542e+00
3.32160562e-01 -4.48475242e-01 1.22477746e+00 8.10114801e-01
3.69580925e-01 -3.89116019e-01 -1.25970566e+00 -7.68457949e-01
-2.05176145e-01 -3.47664267e-01 2.23384902e-01 6.80078268e-01
-3.33690923e-03 -5.32732951e-03 -2.52916694e-01 -2.70118803e-01
4.68602985e-01 9.30133238e-02 8.44847441e-01 -1.46073508e+00
4.00182962e-01 -7.86945641e-01 -6.13654435e-01 -8.62693787e-01
1.73231989e-01 -3.98779154e-01 2.85321504e-01 -1.38230968e+00
-3.83639097e-01 -7.41925389e-02 2.59029239e-01 7.78640211e-01
3.13020766e-01 3.92106652e-01 2.94847548e-01 1.32956520e-01
6.32755682e-02 1.03863038e-01 1.04729903e+00 -5.41254096e-02
-4.77955043e-01 2.44482458e-01 1.39222175e-01 9.35236692e-01
9.28926885e-01 -3.68490279e-01 -1.33264869e-01 -5.99803589e-02
-4.07642305e-01 -7.52330646e-02 1.63818106e-01 -7.26280510e-01
1.60792321e-01 -4.39009033e-02 2.12737590e-01 -6.43620372e-01
4.20325547e-01 -1.25148833e+00 -2.45504677e-01 6.48677588e-01
-2.07144707e-01 -1.75905764e-01 1.11727072e-02 5.46065625e-03
-4.53545272e-01 -7.72076786e-01 7.28743076e-01 -8.97159651e-02
-9.24247742e-01 -1.91977501e-01 -6.03539228e-01 -4.36157644e-01
9.98707831e-01 -6.71864510e-01 3.12600374e-01 -3.84046108e-01
-9.18499470e-01 -2.53030509e-01 4.51459080e-01 5.05578935e-01
4.92227316e-01 -9.57314491e-01 -3.54304582e-01 4.18747962e-01
-8.24489966e-02 -4.35367107e-01 -9.85527262e-02 9.23440158e-01
-8.39620054e-01 4.55673158e-01 -5.35073161e-01 -5.01177609e-01
-2.26379895e+00 2.89222866e-01 3.51243556e-01 2.81327933e-01
-5.08167207e-01 5.84684670e-01 -6.48922920e-01 -1.10176101e-01
6.73516333e-01 -6.65615439e-01 -8.06530833e-01 4.06691760e-01
7.63701737e-01 3.12897265e-01 -1.40922638e-02 -1.05886114e+00
-4.64998603e-01 9.61474001e-01 2.54112631e-01 -2.48649046e-01
1.20256841e+00 -1.73573438e-02 -2.24232629e-01 3.30839306e-01
1.27350092e+00 -6.45386130e-02 -9.44358110e-01 2.21098736e-01
2.09787712e-01 -5.10750771e-01 5.17679984e-03 -6.44705892e-01
-8.70705962e-01 1.16469359e+00 1.00298166e+00 -3.06469202e-03
1.23579371e+00 -8.82611722e-02 6.67003512e-01 8.14129889e-01
1.91496804e-01 -1.36572707e+00 -3.21230024e-01 3.88027519e-01
8.76477659e-01 -1.38418782e+00 -2.55096048e-01 -3.49733025e-01
-3.83438230e-01 1.58701861e+00 1.43358216e-01 2.88566232e-01
6.84524834e-01 2.38004401e-01 5.89305997e-01 6.52564839e-02
-3.23318802e-02 -5.33577800e-01 3.83464754e-01 6.53209805e-01
5.87948561e-01 -1.66477650e-01 -9.48554993e-01 1.31846830e-01
-7.32299984e-02 6.85487211e-01 4.24047530e-01 1.11844909e+00
-4.95106667e-01 -1.54085076e+00 -6.71272159e-01 2.05207735e-01
-5.81072807e-01 2.07668424e-01 -8.25173378e-01 1.08643913e+00
1.67241752e-01 8.82411122e-01 -4.37546708e-02 -6.90844357e-02
3.14910352e-01 7.00338900e-01 7.34678507e-01 -9.27752703e-02
-1.80028453e-01 2.18855247e-01 1.60505742e-01 -2.85027444e-01
-1.00866926e+00 -9.68834043e-01 -1.32392776e+00 5.84341250e-02
-2.53143311e-01 1.58310160e-02 1.28155112e+00 1.11258173e+00
-2.89627761e-01 -3.83880176e-02 3.04055303e-01 -1.15951812e+00
-5.74542522e-01 -1.23241663e+00 -3.72673959e-01 4.68205333e-01
3.32027167e-01 -6.73899114e-01 -4.21449721e-01 5.70458829e-01]
|
[9.073851585388184, -6.339036464691162]
|
8c0756b7-7164-40d8-a9c0-f2624af03ba3
|
multi-scale-single-image-dehazing-using
|
2111.05700
| null |
https://arxiv.org/abs/2111.05700v2
|
https://arxiv.org/pdf/2111.05700v2.pdf
|
Multi-Scale Single Image Dehazing Using Laplacian and Gaussian Pyramids
|
Model driven single image dehazing was widely studied on top of different priors due to its extensive applications. Ambiguity between object radiance and haze and noise amplification in sky regions are two inherent problems of model driven single image dehazing. In this paper, a dark direct attenuation prior (DDAP) is proposed to address the former problem. A novel haze line averaging is proposed to reduce the morphological artifacts caused by the DDAP which enables a weighted guided image filter with a smaller radius to further reduce the morphological artifacts while preserve the fine structure in the image. A multi-scale dehazing algorithm is then proposed to address the latter problem by adopting Laplacian and Guassian pyramids to decompose the hazy image into different levels and applying different haze removal and noise reduction approaches to restore the scene radiance at different levels of the pyramid. The resultant pyramid is collapsed to restore a haze-free image. Experiment results demonstrate that the proposed algorithm outperforms state of the art dehazing algorithms and the noise is indeed prevented from being amplified in the sky region.
|
['Chaobing Zheng', 'Haiyan Shu', 'Zhengguo Li']
|
2021-11-10
| null | null | null | null |
['image-dehazing']
|
['computer-vision']
|
[ 5.63164532e-01 -3.40537459e-01 7.84595191e-01 -1.99843780e-03
-3.79065156e-01 -1.23247892e-01 3.43604088e-01 -2.36623541e-01
-2.05088794e-01 6.37690246e-01 4.04888421e-01 5.84274717e-02
-2.56872028e-01 -8.80450368e-01 -4.02720064e-01 -1.37361526e+00
4.39012945e-01 -5.12701988e-01 6.69731975e-01 -4.58724350e-01
4.53254819e-01 2.93691278e-01 -1.78499913e+00 5.74123412e-02
1.39065492e+00 6.53591573e-01 5.40521502e-01 7.29552209e-01
1.11746177e-01 8.53143036e-01 -6.35328233e-01 3.94411087e-02
7.22655118e-01 -5.16846716e-01 -2.55993158e-02 5.84046900e-01
6.35385156e-01 -4.91386205e-01 -2.73992091e-01 1.67498124e+00
4.13902432e-01 3.46174091e-01 5.84579527e-01 -8.69115770e-01
-7.51952410e-01 -1.85681552e-01 -1.12292719e+00 4.57044572e-01
-2.35577226e-01 9.61447582e-02 2.14625672e-01 -1.05744469e+00
-3.01184878e-02 1.19672191e+00 3.85952532e-01 -1.22989714e-02
-9.80264246e-01 -5.13896465e-01 -9.86146033e-02 3.96953225e-01
-1.63834524e+00 -2.16955379e-01 9.76071537e-01 -2.97262371e-01
4.41692233e-01 3.08516294e-01 4.91409659e-01 -7.14678690e-02
5.73589206e-01 2.51835763e-01 1.49938595e+00 -7.00904071e-01
1.61908343e-01 5.70364930e-02 -4.23498340e-02 5.97914815e-01
7.25201070e-01 -1.05311433e-02 -3.96808624e-01 8.39473754e-02
5.88408649e-01 3.37088168e-01 -6.16002738e-01 -7.19821602e-02
-7.82016039e-01 5.12452662e-01 5.04198134e-01 2.90240616e-01
-6.07495368e-01 -1.08512387e-01 -4.22037393e-01 9.22745690e-02
5.96615672e-01 1.85697645e-01 4.44605276e-02 6.95470691e-01
-1.14913094e+00 3.66869122e-01 2.25968763e-01 6.01342440e-01
1.15403843e+00 4.36059445e-01 -6.04913617e-03 8.09131324e-01
6.01488471e-01 7.95781016e-01 2.28005156e-01 -9.34315026e-01
1.65496960e-01 5.53792536e-01 3.78062040e-01 -1.17207277e+00
2.33793661e-01 -4.25169826e-01 -1.13189876e+00 8.20000052e-01
5.45847937e-02 -2.37076706e-03 -1.47255373e+00 1.03196537e+00
4.55079645e-01 5.24430215e-01 2.33961612e-01 1.03792882e+00
5.36876261e-01 1.19650698e+00 -2.50749856e-01 -1.24589741e-01
1.31439912e+00 -8.26840341e-01 -1.08954239e+00 -3.63821834e-01
1.13029815e-01 -1.11295211e+00 4.90387976e-01 5.84282935e-01
-1.10325718e+00 -6.73970282e-01 -1.36549771e+00 -8.33752081e-02
-3.85845542e-01 -1.89419657e-01 -3.37642170e-02 7.31925488e-01
-1.13251698e+00 1.30324528e-01 -5.38089812e-01 -2.25904416e-02
4.83000539e-02 5.18471450e-02 3.69226970e-02 -6.00874007e-01
-1.11992657e+00 1.17573309e+00 4.49531555e-01 4.27282184e-01
-8.30464125e-01 -7.36271739e-01 -9.20459867e-01 2.44081821e-02
2.42231861e-01 -6.25033855e-01 4.31188017e-01 -8.95701468e-01
-1.26492620e+00 4.04237956e-01 -2.09159389e-01 -4.74673808e-01
7.34299421e-02 -3.87515247e-01 -5.63744128e-01 2.58374423e-01
-8.32595304e-02 2.63006806e-01 1.53760982e+00 -1.61835396e+00
-7.75872827e-01 -3.36822540e-01 -3.72546881e-01 6.74668491e-01
6.36816919e-02 -1.68736249e-01 -2.87898868e-01 -9.08652842e-01
3.78057837e-01 -5.58153391e-01 -2.43679538e-01 -1.37619138e-01
-1.45574167e-01 4.47331697e-01 1.17013681e+00 -9.62606788e-01
1.27603436e+00 -2.45442295e+00 4.80546989e-02 1.73732191e-01
2.36108810e-01 4.87046838e-01 -7.46468529e-02 2.81646758e-01
3.39467376e-02 -2.54365176e-01 -8.54723334e-01 7.45036014e-05
-5.67321002e-01 1.53502390e-01 -1.10792570e-01 8.23462367e-01
1.21433988e-01 2.06522956e-01 -4.88498122e-01 -4.03521180e-01
6.20904803e-01 9.51396465e-01 -2.63601989e-01 3.60281348e-01
2.09262833e-01 3.96251053e-01 -1.73644274e-01 4.12797928e-01
1.57532930e+00 4.10944194e-01 -6.32208347e-01 -3.00356358e-01
-5.20508647e-01 -3.76330703e-01 -1.38391793e+00 9.82362509e-01
-2.59171367e-01 6.24861896e-01 6.97281897e-01 -6.22727334e-01
1.08032048e+00 1.55464008e-01 1.89080369e-02 -4.38517600e-01
2.95000393e-02 1.78995490e-01 -4.93473671e-02 -6.27280176e-01
5.83275199e-01 -5.17014861e-01 6.61521137e-01 -2.57648140e-01
-4.50688392e-01 -6.45251393e-01 -9.83695909e-02 6.75438568e-02
6.76767588e-01 -1.11580089e-01 2.03402162e-01 -5.44425249e-01
9.81306970e-01 -2.02279612e-02 6.00995123e-01 6.67314827e-01
-2.37606689e-01 9.69975650e-01 -5.03891885e-01 -3.36815894e-01
-8.93268108e-01 -9.65385556e-01 -1.09666651e-02 7.85994411e-01
5.80866933e-01 1.63760647e-01 -9.04034019e-01 2.12627828e-01
-2.41349488e-01 9.81291592e-01 -5.45441329e-01 -3.06793809e-01
-5.20297766e-01 -1.10700309e+00 -7.65228346e-02 -2.92947829e-01
1.29424739e+00 -5.74601114e-01 -5.40515959e-01 8.83418843e-02
-3.29894871e-01 -9.06706512e-01 -2.63334215e-01 -6.35721833e-02
-7.74672627e-01 -8.32978010e-01 -9.48905945e-01 -8.61782193e-01
7.69671679e-01 1.11336875e+00 5.52823365e-01 1.85448542e-01
-2.70127982e-01 1.09186083e-01 -3.90972495e-01 -7.45493293e-01
-3.85328680e-01 -7.46564865e-01 -2.66032904e-01 6.68514550e-01
2.31897265e-01 -6.28073215e-01 -1.00527525e+00 1.35424927e-01
-1.43208611e+00 1.70983315e-01 8.33570421e-01 6.15535617e-01
3.58331889e-01 1.08824933e+00 -2.16608327e-02 -4.82613623e-01
2.36662686e-01 -4.30936456e-01 -9.02301729e-01 -2.29376592e-02
-6.05923831e-01 -1.79139867e-01 2.86669970e-01 -1.43954068e-01
-1.70972192e+00 -2.11280406e-01 3.79803836e-01 -3.21041137e-01
-3.20332944e-01 2.77720541e-01 -1.85432151e-01 -4.11131978e-01
4.62679476e-01 7.59786248e-01 2.35871933e-02 -5.96589506e-01
2.31851444e-01 6.59843683e-01 7.70801485e-01 1.53407663e-01
1.44263434e+00 9.23591256e-01 2.29603097e-01 -1.33852708e+00
-7.60918021e-01 -6.55707240e-01 -4.13863361e-01 -2.69237339e-01
1.25264943e+00 -1.23807442e+00 3.42516825e-02 7.94283390e-01
-9.43944871e-01 2.97485455e-03 1.43610507e-01 4.81242359e-01
1.11916609e-01 6.50771797e-01 -3.49533588e-01 -1.13004303e+00
-3.39751273e-01 -9.56194997e-01 6.83695257e-01 5.61658323e-01
6.98312461e-01 -7.56445825e-01 -3.21082845e-02 4.44593012e-01
7.53702700e-01 3.05436939e-01 8.79278660e-01 2.34201789e-01
-1.00527740e+00 8.17793701e-03 -3.92436624e-01 7.36454546e-01
4.93813157e-01 -1.85115501e-01 -8.40840101e-01 -2.71233976e-01
6.95001841e-01 4.83106464e-01 9.89984274e-01 6.36348605e-01
4.38781798e-01 -1.96026489e-01 1.59928501e-01 7.91525602e-01
1.96449077e+00 1.13266021e-01 1.08479249e+00 8.05625618e-01
6.57890737e-01 6.19142532e-01 6.16494417e-01 2.47096732e-01
-7.28882924e-02 3.73829573e-01 7.37240553e-01 -4.71193641e-01
-4.98269111e-01 3.24915558e-01 2.15630323e-01 6.70674086e-01
-8.89274254e-02 -2.88907588e-01 -5.12920499e-01 8.66235614e-01
-1.44020975e+00 -6.68151438e-01 -5.42611599e-01 2.10230613e+00
5.64918280e-01 -1.27741337e-01 -6.23444736e-01 2.07463577e-01
8.85629952e-01 3.67400497e-01 -1.02712251e-01 -9.47611257e-02
-4.65498477e-01 1.07101671e-01 8.68589818e-01 8.03127825e-01
-9.18191969e-01 7.20942736e-01 5.22689581e+00 8.26483905e-01
-7.99701989e-01 5.67853786e-02 2.06702694e-01 2.23649904e-01
-1.17015734e-01 1.30336881e-01 -5.51222920e-01 6.07573450e-01
5.78767240e-01 -1.72859013e-01 4.67685580e-01 3.03045541e-01
6.13301396e-01 -7.49264121e-01 -3.35651934e-02 9.63016748e-01
2.63590991e-01 -7.75118947e-01 2.83658624e-01 -2.39191744e-02
1.01299047e+00 -1.58004954e-01 2.24168658e-01 -2.65017480e-01
1.65549785e-01 -6.92270517e-01 5.11115909e-01 6.85567796e-01
2.27926672e-01 -7.39767194e-01 9.30803180e-01 3.44793320e-01
-1.03355682e+00 -3.73135246e-02 -6.11828387e-01 -2.38110363e-01
2.31093124e-01 8.20317924e-01 -4.64106381e-01 6.60538018e-01
9.44090426e-01 3.40572864e-01 -6.95975542e-01 1.21061540e+00
-2.20477387e-01 6.57553375e-01 -2.94544995e-01 6.52808309e-01
2.82082349e-01 -7.11515784e-01 8.66662085e-01 1.06303298e+00
5.20925462e-01 6.68864191e-01 -1.68979079e-01 6.38143957e-01
3.22099537e-01 -2.16917232e-01 -5.29506564e-01 6.13229275e-01
2.04092771e-01 1.01017320e+00 -4.43455160e-01 -3.41966420e-01
-6.42182469e-01 1.06778049e+00 -5.78829885e-01 7.11994171e-01
-5.89496672e-01 -6.31094754e-01 5.75113297e-01 2.57789046e-01
6.17081165e-01 -3.16945314e-01 -1.88359395e-01 -9.13934290e-01
-1.69966921e-01 -7.86396563e-01 1.31195962e-01 -1.20053661e+00
-8.53115082e-01 2.71702141e-01 1.06860645e-01 -1.29335117e+00
5.76116204e-01 -4.14388895e-01 -7.44819522e-01 1.33897376e+00
-2.13572717e+00 -9.58073795e-01 -7.75943696e-01 7.50925541e-01
7.51628458e-01 1.53050438e-01 2.32421368e-01 2.46555865e-01
-4.55372363e-01 -2.32746243e-01 4.32177007e-01 -3.38798910e-01
6.96359694e-01 -1.07362986e+00 -1.64272621e-01 1.79460979e+00
-4.45136309e-01 4.80341882e-01 1.21202517e+00 -8.14469814e-01
-1.14883113e+00 -1.25124919e+00 4.13519084e-01 -3.63233350e-02
3.11471134e-01 2.87083630e-02 -1.40755415e+00 3.49305898e-01
6.38835371e-01 -6.16622493e-02 3.25710058e-01 -1.01082361e+00
-1.39964178e-01 -3.41514051e-01 -1.22235107e+00 5.75292706e-01
1.11915365e-01 -1.28007904e-01 -8.74548852e-01 -1.46991946e-02
6.50128961e-01 -1.66957483e-01 -4.85103667e-01 5.14910281e-01
5.50038628e-02 -1.37429893e+00 9.31063473e-01 2.73956567e-01
8.78496096e-02 -1.14177787e+00 -3.42999786e-01 -1.31033301e+00
-6.49982035e-01 -6.90598547e-01 -6.22901658e-04 1.16337895e+00
2.15407126e-02 -6.05224192e-01 3.96139354e-01 2.41333291e-01
-1.83724672e-01 -9.06947479e-02 -5.92672586e-01 -4.27144021e-01
-2.50341207e-01 2.06819221e-01 3.23328018e-01 6.39498949e-01
-8.20648491e-01 6.38217255e-02 -6.88777626e-01 1.07726157e+00
1.29337645e+00 -4.14717458e-02 6.86661124e-01 -9.21754003e-01
1.54359370e-01 3.19666713e-02 -3.36105138e-01 -6.46660984e-01
-5.06027699e-01 -1.35857195e-01 4.35538411e-01 -1.66467822e+00
5.37904352e-02 3.45744014e-01 -3.70748490e-01 -1.82317361e-01
-7.45246589e-01 6.03658736e-01 1.15182787e-01 3.55590045e-01
-1.37307063e-01 7.53278434e-01 1.17891848e+00 -2.19424993e-01
-2.60182470e-01 -2.12134853e-01 -5.97183287e-01 8.24898481e-01
7.51931250e-01 -5.23755193e-01 -5.08655548e-01 -6.07418478e-01
-1.60514653e-01 -2.89303631e-01 4.11887169e-01 -1.33185124e+00
3.97933453e-01 -1.15510479e-01 4.50753987e-01 -7.65124202e-01
3.68898451e-01 -1.01190031e+00 2.13888735e-01 3.26248556e-01
2.26702690e-01 -2.82039970e-01 1.60716832e-01 9.41291571e-01
-5.12146592e-01 -1.46356806e-01 1.41356146e+00 -1.98731318e-01
-7.92337716e-01 -7.76238516e-02 -7.77088463e-01 -3.77606124e-01
8.63238752e-01 -6.16021872e-01 -2.73846269e-01 -4.53842938e-01
-3.95522565e-01 1.24119986e-02 5.28791726e-01 6.09668754e-02
8.49726856e-01 -6.78003252e-01 -1.06552851e+00 4.19755816e-01
-1.79582275e-02 8.90134051e-02 6.78027570e-01 7.80384541e-01
-9.02466357e-01 -8.00588429e-02 -1.08744077e-01 -3.05038422e-01
-1.47629464e+00 6.36692584e-01 4.06513900e-01 5.14860153e-02
-8.23913515e-01 8.25001180e-01 6.46180093e-01 2.29778603e-01
2.02922598e-02 -6.06403798e-02 -2.20758438e-01 -4.91828471e-01
9.00152981e-01 6.74509168e-01 1.92348268e-02 -8.64304781e-01
-1.22955572e-02 9.51893628e-01 4.47392883e-03 -2.68529374e-02
1.33918059e+00 -9.47383344e-01 -5.51447630e-01 3.35177742e-02
8.10021162e-01 3.33334565e-01 -1.43698514e+00 -3.71600032e-01
-2.69546300e-01 -9.29493248e-01 6.97072089e-01 -5.80467045e-01
-1.00188315e+00 8.61737013e-01 8.84623766e-01 2.49873802e-01
1.72601986e+00 -6.50827467e-01 7.01939762e-01 2.02716254e-02
-5.18015400e-02 -8.12749028e-01 -5.29665649e-02 1.60837144e-01
7.07460284e-01 -1.00169063e+00 4.95121509e-01 -6.21244311e-01
-4.06387508e-01 8.83811712e-01 3.61509681e-01 -3.01924169e-01
6.71287298e-01 9.58701447e-02 3.22465688e-01 -2.55525410e-01
-1.64891452e-01 -4.59638000e-01 2.15703309e-01 6.22642815e-01
-6.49305210e-02 -4.23722088e-01 -1.47514343e-01 -6.86851144e-02
1.92680821e-01 -1.93395257e-01 8.70114088e-01 9.74597514e-01
-1.18047881e+00 -4.42234308e-01 -1.24328530e+00 3.96350399e-02
-6.11368597e-01 -3.56511295e-01 9.48351920e-02 4.90229934e-01
4.53100771e-01 1.52959871e+00 -2.92801291e-01 7.15107610e-03
2.60876656e-01 -8.86763036e-02 2.98635274e-01 -4.46188718e-01
-1.44933969e-01 4.54984784e-01 -5.21149695e-01 -1.59228966e-01
-7.12569892e-01 -1.92433849e-01 -9.09104526e-01 -1.41687185e-01
-4.60963756e-01 2.80128688e-01 6.14609301e-01 6.61355436e-01
2.29618236e-01 5.18636167e-01 6.15276992e-01 -9.81038570e-01
-8.26392174e-02 -1.00707674e+00 -1.17614043e+00 3.07410121e-01
9.72685099e-01 -6.00720882e-01 -9.35307980e-01 3.97741467e-01]
|
[10.857128143310547, -3.1557719707489014]
|
b6f4561c-a2d8-4701-8ad5-495bbef497fd
|
deepfake-detection-using-biological-features
|
2301.05819
| null |
https://arxiv.org/abs/2301.05819v1
|
https://arxiv.org/pdf/2301.05819v1.pdf
|
Deepfake Detection using Biological Features: A Survey
|
Deepfake is a deep learning-based technique that makes it easy to change or modify images and videos. In investigations and court, visual evidence is commonly employed, but these pieces of evidence may now be suspect due to technological advancements in deepfake. Deepfakes have been used to blackmail individuals, plan terrorist attacks, disseminate false information, defame individuals, and foment political turmoil. This study describes the history of deepfake, its development and detection, and the challenges based on physiological measurements such as eyebrow recognition, eye blinking detection, eye movement detection, ear and mouth detection, and heartbeat detection. The study also proposes a scope in this field and compares the different biological features and their classifiers. Deepfakes are created using the generative adversarial network (GANs) model, and were once easy to detect by humans due to visible artifacts. However, as technology has advanced, deepfakes have become highly indistinguishable from natural images, making it important to review detection methods.
|
['Abhishek Gulhane', 'Jaivanti Dhokey', 'Shrushti Kale', 'Kundan Patil']
|
2023-01-14
| null | null | null | null |
['face-swapping']
|
['computer-vision']
|
[ 1.29427671e-01 4.11710190e-03 9.76905692e-03 5.24506904e-03
-1.26033887e-01 -8.63202333e-01 6.41978025e-01 -8.52667242e-02
-4.32925284e-01 9.67712045e-01 -3.00735980e-02 -1.16507158e-01
4.02851284e-01 -4.98142362e-01 -3.52126062e-01 -7.81397939e-01
1.14173487e-01 -2.86357850e-01 -1.45232558e-01 1.48084387e-01
3.65860999e-01 8.49743664e-01 -1.31202590e+00 3.08431000e-01
5.57745039e-01 7.18526959e-01 -6.59327209e-01 1.04447615e+00
1.48033425e-01 6.18171930e-01 -1.40074348e+00 -9.25734758e-01
1.79160058e-01 -8.80906761e-01 -3.70203167e-01 -6.29778206e-02
7.62363672e-01 -7.27593362e-01 -4.52077746e-01 1.06777513e+00
9.77235794e-01 -4.13231999e-01 3.58829677e-01 -1.55656111e+00
-1.06251788e+00 4.68735211e-02 -6.49414122e-01 5.89623213e-01
3.67134303e-01 6.49374187e-01 6.63536638e-02 -5.20758688e-01
4.77317661e-01 1.36908424e+00 8.59941840e-01 1.22551882e+00
-1.28916550e+00 -1.08123958e+00 -4.43684578e-01 2.64661998e-01
-1.37066936e+00 -6.80295408e-01 5.63654900e-01 -4.87807333e-01
6.55744433e-01 3.34635198e-01 1.13734651e+00 1.53698587e+00
5.72700441e-01 6.15575135e-01 1.51639616e+00 -3.01375449e-01
-5.57466364e-03 2.44349703e-01 -3.12578768e-01 7.01198876e-01
4.95459646e-01 4.13982570e-01 -6.67736948e-01 -2.80457884e-01
7.74685979e-01 -2.06127495e-01 -3.23296696e-01 1.92106336e-01
-7.71418393e-01 7.48233020e-01 1.11008346e-01 1.10275641e-01
-2.71491826e-01 1.82526901e-01 3.88346434e-01 2.11859241e-01
1.80089876e-01 4.27120328e-01 2.32707843e-01 -3.48794878e-01
-9.62801397e-01 1.83843542e-02 4.14768606e-01 8.00879076e-02
-3.91418813e-03 4.33741242e-01 -1.89227805e-01 5.90769172e-01
1.32271245e-01 8.49374831e-01 6.46088243e-01 -8.85993063e-01
-1.93956628e-01 2.85574168e-01 4.30767164e-02 -1.28317523e+00
-2.83101827e-01 2.47509778e-02 -7.05139518e-01 7.24643826e-01
4.85636681e-01 -3.52997422e-01 -1.14002931e+00 1.40003550e+00
3.07609260e-01 2.66838670e-01 -1.36782706e-01 8.10063124e-01
1.08272851e+00 3.31818998e-01 1.13670966e-02 -4.01756018e-02
1.31880343e+00 -3.28967869e-01 -1.00139022e+00 2.56462093e-03
-5.04949130e-02 -7.04973400e-01 7.97594607e-01 6.72443509e-01
-1.01029658e+00 -3.83291036e-01 -1.09063971e+00 6.17142254e-03
-5.46405256e-01 -8.47261176e-02 4.61312383e-01 1.60876787e+00
-9.82366681e-01 3.83128107e-01 -5.13077796e-01 -4.72237438e-01
8.86716843e-01 3.33551288e-01 -4.19805616e-01 2.68096149e-01
-9.52581704e-01 9.95302498e-01 -3.79617848e-02 3.04532051e-01
-8.73017132e-01 -3.13095212e-01 -6.05011404e-01 -2.99025595e-01
-5.46803325e-02 -6.66148961e-01 9.59334314e-01 -8.75682890e-01
-1.63710940e+00 1.56798935e+00 -4.42946963e-02 -7.52649128e-01
5.79363883e-01 -4.64894623e-02 -9.17724073e-01 3.95355701e-01
-2.75468290e-01 7.30969429e-01 1.53745496e+00 -7.46114254e-01
-2.26176023e-01 -2.67079264e-01 -1.27479315e-01 -4.23359424e-01
-4.94615138e-02 3.53422105e-01 2.32160255e-01 -7.82147408e-01
-6.42115772e-01 -8.94256294e-01 5.44195771e-01 4.17918712e-01
-6.65609598e-01 5.93976304e-02 1.05499029e+00 -8.36247623e-01
9.06301200e-01 -2.28725052e+00 -5.45688212e-01 -4.73048016e-02
7.74241686e-01 1.03900146e+00 -1.17411233e-01 2.29307815e-01
5.67577630e-02 6.13494992e-01 1.94784686e-01 6.53413162e-02
-6.47983029e-02 -1.52185753e-01 -1.60145223e-01 8.29135060e-01
4.57365662e-02 1.23788786e+00 -7.62075067e-01 -2.51860291e-01
3.71682316e-01 7.05529630e-01 -1.77039415e-01 -1.40326962e-01
4.32172388e-01 6.02201939e-01 -2.43185218e-02 1.05392253e+00
7.90663064e-01 2.32087029e-03 -2.50485510e-01 -1.80109739e-01
5.70275821e-02 1.63884684e-02 -4.63708341e-01 9.08514678e-01
-4.28882763e-02 1.33059454e+00 3.96456942e-02 -5.56540489e-01
9.11685705e-01 2.94639230e-01 1.35844976e-01 -7.42993176e-01
4.99732137e-01 1.28394309e-02 3.10803890e-01 -7.47459292e-01
2.37904012e-01 -2.71175534e-01 2.48680189e-01 3.63986075e-01
-3.11919570e-01 -2.74445772e-01 -1.33167073e-01 6.53674500e-03
9.98240530e-01 -2.83585817e-01 2.57751912e-01 1.99861914e-01
2.97104001e-01 -4.05574173e-01 3.46547991e-01 6.16904140e-01
-6.64619327e-01 3.73325318e-01 3.26504230e-01 -6.59036458e-01
-7.29073286e-01 -1.10680580e+00 -1.03550233e-01 4.00807053e-01
2.84755398e-02 -2.02544287e-01 -1.01862967e+00 -6.11839652e-01
1.56870708e-01 5.90385437e-01 -8.65635753e-01 -6.45510256e-01
-1.07352920e-01 -7.77037978e-01 1.23195744e+00 1.71973452e-01
7.47071087e-01 -1.15177977e+00 -9.33046043e-01 8.21421295e-02
7.67060891e-02 -8.73379469e-01 -3.59422982e-01 -3.70909542e-01
-4.26442087e-01 -1.34224081e+00 -1.00844419e+00 -2.30706766e-01
5.04757047e-01 -1.27708495e-01 7.68715978e-01 1.20873041e-01
-8.81740034e-01 6.08524323e-01 -1.05731525e-01 -9.35621262e-01
-7.31370330e-01 -6.41523719e-01 2.02430427e-01 2.07666308e-01
7.78831720e-01 -3.70594800e-01 -7.81704366e-01 7.96107054e-02
-7.46875167e-01 -4.62153047e-01 2.94806719e-01 7.32983768e-01
7.55819157e-02 -1.14414304e-01 6.05374694e-01 -6.67775571e-01
1.10429919e+00 6.18544109e-02 -3.45886081e-01 9.77332816e-02
-5.55379033e-01 -4.97720808e-01 5.88135503e-04 -6.96654558e-01
-6.80933356e-01 -7.45719910e-01 6.28813058e-02 -4.06371295e-01
-2.33544454e-01 -1.29502937e-01 1.85659394e-01 -4.90057886e-01
7.84827530e-01 1.57480717e-01 1.65162548e-01 -1.70014352e-01
1.35034963e-01 8.41069818e-01 8.99258256e-01 1.61933571e-01
7.70010531e-01 6.57130957e-01 -9.21057910e-02 -1.10302198e+00
-4.65413660e-01 1.39094248e-01 -3.28611791e-01 -5.45247376e-01
8.12959969e-01 -4.67057407e-01 -1.23685539e+00 1.18645930e+00
-1.15285170e+00 -1.07203074e-01 -2.98720747e-01 3.48932832e-01
1.02254778e-01 4.77340102e-01 -7.45926619e-01 -8.78177762e-01
-4.56754655e-01 -7.20190167e-01 6.69664681e-01 6.09629810e-01
-5.34709990e-01 -9.48710501e-01 1.41453877e-01 4.02430832e-01
4.37388271e-01 7.86672533e-01 6.26452029e-01 -3.65883648e-01
-2.27961391e-01 -5.35316050e-01 -7.05800802e-02 6.08372688e-01
4.60112005e-01 3.58783126e-01 -1.37868154e+00 -1.82789281e-01
-1.14703916e-01 -2.59545892e-01 5.72360456e-01 5.60259700e-01
1.15086174e+00 -4.39551294e-01 -3.92850786e-01 6.88092172e-01
6.25852227e-01 7.71723449e-01 1.19145620e+00 3.55083108e-01
3.93687576e-01 4.20497239e-01 -5.55623099e-02 2.76123226e-01
-1.08874708e-01 4.49752033e-01 3.77228379e-01 -3.51553887e-01
-4.97547984e-01 -3.00778121e-01 5.02260029e-01 3.11057240e-01
-1.14346802e-01 -5.33933520e-01 -4.55352992e-01 2.26935178e-01
-9.47201669e-01 -1.19390213e+00 -2.00166762e-01 2.09055471e+00
6.21475816e-01 6.27916213e-03 3.42114598e-01 2.25005299e-01
1.11456501e+00 -1.17659375e-01 -8.98754001e-01 -9.61053491e-01
-2.85412908e-01 4.57614541e-01 3.58801782e-01 -6.27371594e-02
-9.03898418e-01 8.28445971e-01 7.85071898e+00 4.61017072e-01
-1.81269550e+00 1.39666557e-01 6.01323307e-01 -2.08269790e-01
4.82479371e-02 -6.50766373e-01 -5.75739264e-01 9.36136544e-01
8.42013717e-01 -2.08600402e-01 3.15667212e-01 3.28444988e-01
3.13033193e-01 -2.39707142e-01 -8.83952677e-01 1.54438782e+00
4.83967841e-01 -1.42120874e+00 -1.73256606e-01 2.81676680e-01
3.05390775e-01 -3.03812087e-01 5.58922470e-01 -5.35123385e-02
1.08464003e-01 -1.43704832e+00 4.55516428e-01 5.44807911e-01
1.15219367e+00 -6.57774270e-01 6.46651030e-01 -1.48952350e-01
-4.72074956e-01 -2.05245917e-03 -7.19741657e-02 6.97023496e-02
4.25750278e-02 3.80793095e-01 -9.61008608e-01 -4.32346821e-01
8.45249891e-01 3.79841536e-01 -5.68965673e-01 1.22776926e+00
-5.77958345e-01 6.57795548e-01 -1.47946686e-01 -2.04763070e-01
-1.18948564e-01 1.16890833e-01 8.04747343e-01 1.02602088e+00
2.06049711e-01 -1.92614034e-01 -4.63291824e-01 1.09994817e+00
-3.33747894e-01 -2.73909330e-01 -5.78906059e-01 -5.23826718e-01
4.29612994e-01 1.19177139e+00 -8.09524000e-01 -2.28554994e-01
-2.09253371e-01 1.19998479e+00 -4.95058417e-01 2.10222587e-01
-9.25844371e-01 -6.25342965e-01 7.76255846e-01 2.84231603e-01
-1.08062655e-01 1.63119018e-01 -2.85877325e-02 -1.05791390e+00
-1.85135245e-01 -1.09475696e+00 2.16019943e-01 -1.08576822e+00
-1.14580643e+00 5.04802585e-01 -2.73888677e-01 -1.10669017e+00
-1.76193133e-01 -5.29758990e-01 -6.69611454e-01 7.95648098e-01
-1.18010354e+00 -1.01228774e+00 -2.73777068e-01 5.85474074e-01
2.17977166e-01 -3.62355292e-01 8.17197442e-01 1.69188708e-01
-5.96317410e-01 9.41605508e-01 -2.12074909e-02 3.71006399e-01
9.25318539e-01 -8.59287441e-01 5.55122733e-01 9.59941149e-01
6.21094666e-02 6.54782295e-01 5.04879832e-01 -6.17587328e-01
-9.20111239e-01 -5.57243705e-01 5.52064180e-01 -4.32661742e-01
4.48251039e-01 -3.90328825e-01 -6.39575183e-01 3.82542938e-01
3.68489176e-01 6.52455688e-02 8.73320639e-01 -2.84582525e-01
-4.10881996e-01 -1.26244828e-01 -1.75908291e+00 5.55591643e-01
5.93049824e-01 -8.68568361e-01 -5.94193935e-01 1.19018391e-01
-8.82092044e-02 -3.23493004e-01 -4.23943073e-01 6.91094920e-02
9.22245562e-01 -1.06046128e+00 8.19125950e-01 -4.99992490e-01
-4.65711243e-02 -1.65839300e-01 7.30115891e-01 -1.34487832e+00
-1.14939153e-01 -1.11974847e+00 -1.92921519e-01 1.33610451e+00
-5.24562374e-02 -1.00302124e+00 8.03144336e-01 6.40797794e-01
3.27005267e-01 5.30322548e-03 -1.03379667e+00 -8.02076578e-01
-7.24733770e-02 -2.30479240e-03 5.37558854e-01 1.05431795e+00
2.22391961e-03 1.50415316e-01 -7.31214643e-01 -6.23003347e-04
5.33251941e-01 -2.12315693e-01 8.56143475e-01 -1.25238562e+00
-9.52621698e-02 -4.01414484e-01 -1.00660348e+00 -5.57132542e-01
-2.20501110e-01 -5.74481845e-01 -4.90119636e-01 -9.30051386e-01
-9.71871521e-03 1.18535779e-01 -1.17476329e-01 6.66277945e-01
6.94037825e-02 8.81391704e-01 2.78860152e-01 4.66076173e-02
2.49643713e-01 5.48899472e-02 1.23836696e+00 -4.02951419e-01
-9.48097408e-02 -1.49708882e-01 -8.25410247e-01 8.72436225e-01
9.98975694e-01 -5.06583214e-01 -1.40468523e-01 -3.83765884e-02
2.20262915e-01 -4.28990543e-01 1.01619709e+00 -9.01081979e-01
-4.80419248e-02 1.25690475e-01 9.64864969e-01 -2.56480247e-01
4.75107789e-01 -5.06220222e-01 4.84191999e-02 8.09881151e-01
-1.10694066e-01 -5.96027225e-02 6.47455871e-01 3.36448401e-01
-1.09204222e-02 -2.30093196e-01 1.18876719e+00 -5.70471175e-02
-6.69254243e-01 -4.06629033e-02 -9.14883733e-01 -2.98223216e-02
1.11232877e+00 -6.81954980e-01 -7.61468053e-01 -6.41784728e-01
-7.89385676e-01 -2.86372364e-01 4.87978518e-01 4.47511137e-01
7.96892524e-01 -1.02317822e+00 -5.63122630e-01 4.65966642e-01
-1.47617668e-01 -6.97995424e-01 3.59059334e-01 8.94816220e-01
-7.20059395e-01 6.77330121e-02 -6.63958907e-01 -4.28571999e-01
-1.72722483e+00 5.95196545e-01 5.78299880e-01 5.14877439e-01
-8.26674223e-01 1.02067661e+00 1.49226204e-01 2.74923563e-01
2.71993922e-03 -2.06974968e-01 -1.90851554e-01 -9.91054922e-02
9.54484284e-01 5.77281058e-01 -1.60203159e-01 -5.49722552e-01
-4.28515762e-01 3.42147380e-01 -3.60576659e-02 1.65171638e-01
8.94433141e-01 -1.71270639e-01 1.92983150e-02 1.55244440e-01
6.82828188e-01 5.40803790e-01 -1.03872144e+00 4.74646360e-01
-5.61010540e-01 -7.19147682e-01 9.90462378e-02 -1.16954422e+00
-1.19016850e+00 1.13842094e+00 9.94088769e-01 4.42771018e-01
1.02310562e+00 -2.57833332e-01 8.79732966e-01 -9.58855730e-03
9.81064141e-02 -8.97906780e-01 1.62541628e-01 -1.76512659e-01
7.21654356e-01 -8.63993883e-01 -6.78607374e-02 -1.69492856e-01
-3.05601716e-01 1.19267607e+00 3.77758473e-01 2.36787707e-01
2.98136294e-01 4.52322900e-01 2.93711752e-01 -2.47022554e-01
-2.61925936e-01 -1.33678429e-02 1.76330313e-01 1.26755142e+00
1.71844840e-01 -1.58190206e-02 -8.87403041e-02 1.72329858e-01
-2.12296128e-01 2.80111551e-01 7.22242653e-01 6.52580917e-01
-1.41465049e-02 -9.59920168e-01 -7.52775013e-01 4.82381552e-01
-9.08992708e-01 -1.03165761e-01 -1.05305219e+00 9.43008006e-01
5.41954100e-01 1.09025693e+00 1.77042261e-02 -3.41684312e-01
1.46933913e-03 3.28342952e-02 6.83241367e-01 -2.58403003e-01
-4.79918718e-01 -1.10709198e-01 -4.59221983e-03 -4.26225811e-01
-6.32704794e-01 -5.09198308e-01 -7.40380704e-01 -6.99185133e-01
-2.30922207e-01 -4.28069383e-02 6.53849423e-01 6.86462820e-01
4.35843974e-01 2.94537634e-01 2.19602987e-01 -5.40343523e-01
1.22628631e-02 -8.47694933e-01 -8.16540480e-01 2.47351408e-01
5.37605464e-01 -5.69501996e-01 -5.06984890e-01 3.51443350e-01]
|
[12.656061172485352, 1.0557646751403809]
|
ac4e68e9-9f27-427a-a70a-df72a989a55a
|
iterative-spectral-clustering-for
|
1706.09719
| null |
http://arxiv.org/abs/1706.09719v1
|
http://arxiv.org/pdf/1706.09719v1.pdf
|
Iterative Spectral Clustering for Unsupervised Object Localization
|
This paper addresses the problem of unsupervised object localization in an
image. Unlike previous supervised and weakly supervised algorithms that require
bounding box or image level annotations for training classifiers in order to
learn features representing the object, we propose a simple yet effective
technique for localization using iterative spectral clustering. This iterative
spectral clustering approach along with appropriate cluster selection strategy
in each iteration naturally helps in searching of object region in the image.
In order to estimate the final localization window, we group the proposals
obtained from the iterative spectral clustering step based on the perceptual
similarity, and average the coordinates of the proposals from the top scoring
groups. We benchmark our algorithm on challenging datasets like Object
Discovery and PASCAL VOC 2007, achieving an average CorLoc percentage of 51%
and 35% respectively which is comparable to various other weakly supervised
algorithms despite being completely unsupervised.
|
['Shanmuganathan Raman', 'Aditya Vora']
|
2017-06-29
| null | null | null | null |
['unsupervised-object-localization']
|
['computer-vision']
|
[ 1.91314936e-01 1.03182411e-02 -1.79646149e-01 -2.96553403e-01
-9.33593631e-01 -7.72424698e-01 6.35009944e-01 5.31761050e-01
-6.60632491e-01 3.57181698e-01 -9.61406678e-02 2.87866145e-01
-2.57784128e-01 -2.06253842e-01 -6.02073669e-01 -9.55712974e-01
-1.30893037e-01 6.78758502e-01 7.71379650e-01 4.67936426e-01
5.38610578e-01 4.39762235e-01 -1.56736434e+00 2.83460259e-01
6.39584899e-01 1.05743754e+00 5.80374718e-01 6.52456820e-01
-1.46899000e-01 5.96607149e-01 -3.84441227e-01 1.63627282e-01
3.06879610e-01 -2.94690609e-01 -9.87253845e-01 4.73756582e-01
4.16588902e-01 4.55278218e-01 3.68768126e-01 1.17980134e+00
2.29304984e-01 3.53169918e-01 1.08775997e+00 -7.69103050e-01
-2.73746774e-02 4.92295295e-01 -7.93748558e-01 8.05168599e-02
3.15823525e-01 -2.84760773e-01 1.09991956e+00 -1.25036561e+00
7.34588146e-01 9.47701693e-01 6.91948652e-01 1.94799062e-02
-1.41517484e+00 -2.28956893e-01 -3.11146434e-02 6.54597133e-02
-1.80054033e+00 -4.72146451e-01 7.77079225e-01 -6.16261065e-01
4.45181340e-01 7.89067820e-02 2.62547880e-01 2.96909928e-01
-6.01998806e-01 6.69219017e-01 1.23119819e+00 -7.65691459e-01
4.86966044e-01 5.34970760e-01 6.77896142e-02 7.87835419e-01
-1.70023739e-01 -1.59245729e-01 -4.40466940e-01 -3.68010759e-01
4.86237645e-01 -2.59309977e-01 1.82138413e-01 -9.36468422e-01
-1.49454188e+00 7.03555167e-01 7.42071509e-01 4.07754689e-01
-3.67828101e-01 1.02169983e-01 2.82681316e-01 -3.36275220e-01
3.40687573e-01 3.40608984e-01 -3.36126655e-01 2.88936466e-01
-1.30020642e+00 -1.23557322e-01 5.23335278e-01 8.75217378e-01
1.01226628e+00 -4.05290753e-01 4.54560444e-02 9.90406632e-01
3.27536553e-01 3.93281989e-02 1.86019763e-01 -1.08546484e+00
-1.21022144e-03 7.63725042e-01 1.89886287e-01 -8.00433755e-01
-3.83762419e-01 -4.55855757e-01 -3.52853715e-01 1.07415088e-01
6.60585582e-01 1.88008070e-01 -1.02758920e+00 1.13484967e+00
6.49154484e-01 3.39565724e-01 -5.62538020e-02 8.80714536e-01
5.42667866e-01 4.15838450e-01 2.58384019e-01 -2.93823987e-01
1.10543096e+00 -8.79707634e-01 -2.69296855e-01 9.35334265e-02
6.63173079e-01 -9.79839861e-01 9.06834245e-01 3.22685301e-01
-7.60261834e-01 -7.20637321e-01 -9.34194922e-01 2.58956164e-01
-4.56703603e-01 7.47219622e-01 5.85401595e-01 4.53177929e-01
-9.57081258e-01 4.56749141e-01 -5.87399423e-01 -5.89796603e-01
5.11481941e-01 5.08414626e-01 -4.91404146e-01 1.95832312e-01
-3.18254918e-01 5.34634054e-01 9.77618098e-01 -1.24900430e-01
-8.45275521e-01 -5.14587104e-01 -5.26951134e-01 -2.96478003e-01
3.64552975e-01 -8.25549066e-02 7.59332061e-01 -1.06200099e+00
-1.11923575e+00 1.07095015e+00 -2.46323526e-01 -6.66011393e-01
3.13281208e-01 1.37936071e-01 1.18984483e-01 4.75719154e-01
2.73083448e-01 1.03387558e+00 9.80420053e-01 -1.55330038e+00
-9.66833472e-01 -3.27306300e-01 -4.06532347e-01 2.96118438e-01
-1.15035966e-01 1.72887921e-01 -7.03783393e-01 -3.54940742e-01
6.29284322e-01 -1.02532077e+00 -3.98122668e-01 -1.92959264e-01
-2.29656920e-01 -5.81131756e-01 1.02630246e+00 -3.39474350e-01
7.63773143e-01 -2.46710420e+00 1.68148294e-01 5.11755407e-01
6.39804155e-02 -2.24707291e-01 1.38108313e-01 6.28178846e-03
2.55189519e-02 -2.54623592e-01 -2.62924939e-01 -3.43650341e-01
-8.32209215e-02 2.38388982e-02 -1.34724200e-01 8.41713727e-01
5.24746478e-02 6.26392663e-01 -9.54568982e-01 -1.02396178e+00
4.29251164e-01 1.78489447e-01 -4.32684451e-01 7.68657029e-02
-2.28593439e-01 4.63537961e-01 -1.77496687e-01 6.90668702e-01
4.95390475e-01 -2.47408047e-01 9.88558680e-02 -3.75020921e-01
-8.65937546e-02 -3.86222638e-02 -1.36563528e+00 1.63840473e+00
-4.22374345e-02 5.44162333e-01 1.74392387e-02 -1.25867355e+00
8.39978337e-01 2.10771799e-01 7.81542182e-01 -1.73104048e-01
-2.15322077e-02 1.98386416e-01 -7.27438405e-02 -2.25116476e-01
3.90992761e-01 9.57714021e-03 -4.68964390e-02 3.32985908e-01
5.48118293e-01 5.19425049e-02 3.22937220e-01 4.52855766e-01
7.38329411e-01 4.09042686e-01 2.96890855e-01 -5.61140656e-01
6.57061756e-01 2.29492754e-01 1.46623418e-01 8.93008769e-01
-3.81315440e-01 8.34179640e-01 1.76753908e-01 -3.13649714e-01
-9.71142173e-01 -1.09247351e+00 -3.64558220e-01 1.46369171e+00
3.41217756e-01 -2.94546455e-01 -1.02372062e+00 -9.92198706e-01
-2.02388138e-01 1.54458150e-01 -6.84899151e-01 2.17848554e-01
-3.25249076e-01 -5.87958455e-01 2.00454757e-01 4.88484025e-01
3.91916513e-01 -1.00840437e+00 -4.37691152e-01 4.07925956e-02
-1.19326353e-01 -1.23175335e+00 -3.72057498e-01 7.11433113e-01
-8.44892263e-01 -1.13972282e+00 -3.71694505e-01 -1.12352097e+00
1.21264064e+00 1.80247620e-01 7.84379601e-01 1.60314329e-02
-5.20145655e-01 4.04577672e-01 -4.28032994e-01 -1.14362471e-01
-9.68209803e-02 7.65353441e-02 2.36437559e-01 1.98007688e-01
4.78461713e-01 -1.50825381e-01 -5.14072835e-01 4.36972439e-01
-5.11233568e-01 -3.31332892e-01 6.18031204e-01 5.60228229e-01
9.41992462e-01 1.93055436e-01 3.17621291e-01 -8.98086071e-01
4.68325801e-02 -3.29951078e-01 -7.38492668e-01 1.84325352e-01
-2.77986497e-01 -2.53785439e-02 3.13733965e-01 -4.63316560e-01
-7.94190228e-01 1.15824592e+00 2.96613693e-01 -3.00969869e-01
-6.30713999e-01 1.18454590e-01 1.67906284e-01 -3.81391615e-01
9.61483181e-01 3.27106744e-01 -3.80186617e-01 -3.70532304e-01
5.14666080e-01 5.33323348e-01 7.41798520e-01 -4.79581177e-01
1.01684940e+00 6.64325535e-01 -2.37206534e-01 -8.70737255e-01
-9.17433500e-01 -1.23153400e+00 -1.29581046e+00 -3.66333187e-01
1.06702709e+00 -8.21351945e-01 -5.54790258e-01 -2.68267453e-01
-8.66110027e-01 1.30787129e-02 -2.50579476e-01 5.82764924e-01
-7.74814367e-01 4.00036871e-01 -5.46377525e-02 -1.04876149e+00
8.93998966e-02 -1.00542653e+00 1.11580312e+00 9.42066312e-02
-3.29974443e-01 -8.89791131e-01 -6.32438809e-02 5.46028197e-01
-8.98972526e-02 1.96283519e-01 5.59416711e-01 -7.16117918e-01
-5.84553421e-01 -3.61574471e-01 -3.26954991e-01 2.19376832e-01
-6.60187602e-02 -2.08183870e-01 -1.12719536e+00 -1.15316346e-01
-3.03753674e-01 -3.70770395e-01 9.03077602e-01 5.07873356e-01
1.09757364e+00 8.69460553e-02 -5.97008348e-01 3.81947815e-01
1.47457206e+00 -7.45222624e-03 2.05996647e-01 1.99703470e-01
5.72508335e-01 6.44790530e-01 9.69816029e-01 1.71049759e-01
-1.42087176e-01 6.43144548e-01 3.12914103e-01 -6.37733862e-02
-1.37888342e-01 -1.21918567e-01 2.13129014e-01 3.90390396e-01
-3.29147577e-02 3.72943968e-01 -8.78925145e-01 8.47868025e-01
-1.87936831e+00 -7.40154624e-01 -1.39964178e-01 2.12257552e+00
5.55277705e-01 2.99893558e-01 4.70949382e-01 1.70805037e-01
7.62278318e-01 -1.63987100e-01 -1.19619735e-01 1.26120355e-02
5.22392876e-02 9.34834853e-02 7.37865865e-01 4.10277188e-01
-1.57849979e+00 1.18005836e+00 6.91958952e+00 1.04965758e+00
-5.97105920e-01 2.05401838e-01 4.39812034e-01 -3.46283382e-03
4.57181036e-01 2.93015063e-01 -7.69152403e-01 3.71697515e-01
6.63868487e-01 3.95875603e-01 1.75796762e-01 1.10263216e+00
1.70468435e-01 -6.26892388e-01 -1.12725425e+00 8.65791857e-01
3.64391282e-02 -1.24806881e+00 -2.07277969e-01 5.21321222e-02
1.03729081e+00 9.12924930e-02 2.53749862e-02 -1.24292389e-01
2.77710855e-01 -9.11595821e-01 6.95031404e-01 3.76283526e-01
3.01671952e-01 -9.13544059e-01 6.44208968e-01 5.89505315e-01
-1.33714557e+00 -8.80331397e-02 -4.79511470e-01 2.81965941e-01
-2.85705656e-01 4.35799122e-01 -1.33103049e+00 1.41138911e-01
8.57176423e-01 4.26593632e-01 -8.25385988e-01 1.33121884e+00
1.37459282e-02 7.41761565e-01 -5.54802775e-01 -2.78794281e-02
4.76887792e-01 -2.53097624e-01 4.37312156e-01 1.47548783e+00
1.33129135e-02 -1.25147387e-01 5.37304282e-01 7.38224030e-01
4.15121205e-02 4.06952739e-01 -4.03071433e-01 2.31080130e-01
4.43157822e-01 1.55958569e+00 -1.54286802e+00 -3.71597707e-01
-9.77650359e-02 9.71465528e-01 4.09451395e-01 2.93756485e-01
-6.49966359e-01 -3.56944412e-01 -5.74396327e-02 1.80020198e-01
6.65106893e-01 -4.04147387e-01 -2.24454582e-01 -5.29066861e-01
-7.25370795e-02 -3.31006378e-01 5.28687954e-01 -6.50350213e-01
-8.24580491e-01 3.41418445e-01 -1.21886402e-01 -1.15620708e+00
-1.57894082e-02 -4.49653953e-01 -4.71122921e-01 4.93176222e-01
-1.01917970e+00 -1.16184497e+00 -1.89440951e-01 4.57495183e-01
5.57101369e-01 -1.77615717e-01 6.50631368e-01 2.29111034e-03
-1.40066609e-01 1.34736389e-01 1.91070691e-01 2.54123688e-01
7.41168559e-01 -1.68071699e+00 -4.19762462e-01 7.16629982e-01
7.34437644e-01 7.21466899e-01 7.69291222e-01 -4.70659375e-01
-8.98239791e-01 -1.06469727e+00 5.37881434e-01 -8.28173101e-01
6.02947474e-01 -5.71379781e-01 -8.05688798e-01 2.48729259e-01
1.92693308e-01 2.92931110e-01 5.35720885e-01 8.78330171e-02
-3.01434338e-01 -4.56064381e-02 -1.07688558e+00 2.67358005e-01
8.20080161e-01 -5.60713768e-01 -4.94282722e-01 7.23993778e-01
2.77418017e-01 -5.49999177e-02 -7.41080284e-01 4.55359250e-01
1.96683586e-01 -9.08371091e-01 9.91319180e-01 -1.66135684e-01
-1.03116862e-01 -8.91975284e-01 -2.04343006e-01 -8.60366702e-01
-3.63538533e-01 -3.27494740e-01 1.60458758e-01 1.19943476e+00
5.08335114e-01 5.68393171e-02 1.15155089e+00 -1.73831925e-01
1.51911527e-01 -4.50851828e-01 -8.73433948e-01 -5.02661049e-01
-4.48806196e-01 -5.60039401e-01 -2.61095732e-01 7.02337921e-01
-6.61185235e-02 2.00965598e-01 -8.96153376e-02 3.80092055e-01
1.19858551e+00 2.28155062e-01 7.46232092e-01 -1.26526594e+00
-8.00435096e-02 -3.29932213e-01 -7.04390049e-01 -7.60001779e-01
2.56600469e-01 -1.04423201e+00 6.04413271e-01 -1.18195081e+00
6.52275026e-01 -5.39219677e-01 -3.13338518e-01 3.61751199e-01
2.78425999e-02 8.47497284e-01 -1.29565299e-01 3.93132716e-01
-1.33556402e+00 7.01093376e-02 5.15814781e-01 -1.03619829e-01
-1.61777571e-01 -8.15046579e-02 -3.24752301e-01 1.00001180e+00
4.86830831e-01 -5.97936511e-01 -6.75656199e-02 1.65319845e-01
-1.52926758e-01 -5.07293701e-01 4.68261629e-01 -1.12847316e+00
6.07707381e-01 2.86838673e-02 7.65058994e-01 -1.03250968e+00
2.24890321e-01 -1.08503711e+00 -9.68159959e-02 3.24114233e-01
-4.62940991e-01 -6.16883337e-01 -1.45907387e-01 8.59235346e-01
-2.93267041e-01 -4.15534735e-01 1.21685946e+00 -2.05594912e-01
-1.08964050e+00 -2.05831919e-02 -4.06082720e-01 -3.61546636e-01
1.39906597e+00 -4.70385045e-01 2.81559706e-01 -1.23274878e-01
-1.21916568e+00 1.11584648e-01 5.56821942e-01 1.62189692e-01
5.39449513e-01 -1.25191844e+00 -4.92116272e-01 5.55879548e-02
4.96369958e-01 -2.94045676e-02 -2.27165967e-01 1.13443339e+00
-4.89797771e-01 3.35132539e-01 2.28720069e-01 -1.13892972e+00
-1.58004856e+00 5.98184764e-01 2.10431993e-01 2.22534463e-01
-3.74447644e-01 8.49927664e-01 1.05075136e-01 -3.40950251e-01
5.60568213e-01 -3.62135656e-02 -1.80266276e-01 1.27226397e-01
2.78602690e-01 2.52744287e-01 2.77397726e-02 -9.89893079e-01
-5.86469948e-01 8.12805533e-01 3.99676859e-02 -8.96395668e-02
1.16384971e+00 -2.09090203e-01 -2.86424965e-01 4.04077500e-01
1.32581115e+00 2.08804179e-02 -1.14264154e+00 -4.05839384e-01
8.39807689e-01 -3.21639538e-01 9.71065015e-02 -4.93128628e-01
-6.25767410e-01 5.27300298e-01 9.42326248e-01 1.58045933e-01
8.47104847e-01 6.80065572e-01 -1.48067418e-02 5.10692716e-01
3.50377232e-01 -1.44467318e+00 3.39159727e-01 2.51391947e-01
4.54759419e-01 -1.38229704e+00 7.05689937e-02 -4.88992512e-01
-5.19649923e-01 9.29299772e-01 3.77674133e-01 -4.59135473e-01
4.98245507e-01 2.35259920e-01 -2.34665871e-02 -2.14503005e-01
-3.12132984e-01 -5.95055819e-01 6.54391766e-01 5.95719337e-01
3.93722802e-01 -8.18620697e-02 -1.25920951e-01 3.41600366e-02
9.25983265e-02 -4.49306369e-01 3.14660668e-02 7.34030187e-01
-1.06958759e+00 -8.21619332e-01 -7.00301409e-01 2.85282195e-01
-4.76502031e-01 1.12952463e-01 -7.52554119e-01 4.22705561e-01
4.37203526e-01 8.97161663e-01 4.66534942e-02 -1.67114139e-01
-6.60268664e-02 2.62150675e-01 4.92746562e-01 -8.77121627e-01
-3.81040484e-01 5.10504127e-01 -1.99054599e-01 -4.49677765e-01
-8.59994054e-01 -7.43113875e-01 -1.52118242e+00 6.03595495e-01
-7.28190124e-01 5.07807076e-01 8.27743411e-01 9.55828249e-01
-4.51762266e-02 -1.84003555e-04 6.64101601e-01 -1.01156402e+00
-1.22517593e-01 -9.87668097e-01 -6.12820745e-01 5.28335094e-01
1.74913704e-01 -6.67035997e-01 -4.67279971e-01 4.75686610e-01]
|
[9.380619049072266, 0.9312050938606262]
|
0a6c6d37-1820-44a2-8494-df112656ff0c
|
parallel-algorithms-for-densest-subgraph
|
2103.00154
| null |
https://arxiv.org/abs/2103.00154v1
|
https://arxiv.org/pdf/2103.00154v1.pdf
|
Parallel Algorithms for Densest Subgraph Discovery Using Shared Memory Model
|
The problem of finding dense components of a graph is a widely explored area in data analysis, with diverse applications in fields and branches of study including community mining, spam detection, computer security and bioinformatics. This research project explores previously available algorithms in order to study them and identify potential modifications that could result in an improved version with considerable performance and efficiency leap. Furthermore, efforts were also steered towards devising a novel algorithm for the problem of densest subgraph discovery. This paper presents an improved implementation of a widely used densest subgraph discovery algorithm and a novel parallel algorithm which produces better results than a 2-approximation.
|
['Anil Vullikanti', 'Saliya Ekanayake', 'Indika Perera', 'M. D. I. Maduranga', 'Y. A. M. M. A. Ali', 'B. D. M. De Zoysa']
|
2021-02-27
| null | null | null | null |
['computer-security', 'spam-detection']
|
['miscellaneous', 'natural-language-processing']
|
[ 2.93162137e-01 1.39493302e-01 -1.10222593e-01 -1.85282424e-01
-6.16216324e-02 -3.00264776e-01 3.85530114e-01 4.85966295e-01
-3.26166660e-01 8.88079226e-01 5.02278768e-02 -5.44453025e-01
-4.91700292e-01 -1.11593556e+00 -6.79833218e-02 -5.40529490e-01
-5.21894336e-01 9.44243670e-01 6.72831357e-01 -4.27951477e-02
5.42952597e-01 8.91563416e-01 -1.32420671e+00 5.39408885e-02
6.33957624e-01 6.35003746e-02 6.84151053e-02 5.96436679e-01
-2.46189013e-01 1.50067180e-01 -4.46933061e-01 -3.83541226e-01
3.65150511e-01 -5.19696116e-01 -1.26322258e+00 5.08163154e-01
-7.22543225e-02 2.27706000e-01 -1.23422928e-01 8.30379844e-01
2.04953894e-01 1.01425357e-01 1.88403234e-01 -1.20780993e+00
1.51062503e-01 5.25772333e-01 -1.11672187e+00 6.75912082e-01
6.63527608e-01 -2.93456972e-01 7.50958323e-01 -3.19647700e-01
8.25006545e-01 1.18344903e+00 3.97396803e-01 -8.56936127e-02
-1.34750068e+00 -4.69572783e-01 -9.92682055e-02 3.25032979e-01
-1.74428439e+00 1.08132018e-02 5.29285431e-01 1.61048002e-03
1.17576551e+00 7.17271030e-01 9.22627091e-01 1.92863882e-01
9.17012841e-02 4.44129407e-01 1.35898626e+00 -4.48546499e-01
2.19579354e-01 3.21219444e-01 2.09732831e-01 7.22600341e-01
1.01726627e+00 -1.60566330e-01 -2.48329937e-01 -6.64492667e-01
4.93187934e-01 -1.15098968e-01 -8.38997215e-02 -6.35664046e-01
-8.86523783e-01 9.80137944e-01 2.07084760e-01 9.58956063e-01
-2.61121392e-01 -1.71134844e-01 1.61184281e-01 5.33367932e-01
3.80217165e-01 4.77088630e-01 -3.88087660e-01 -1.06405921e-01
-8.82318914e-01 3.40932697e-01 1.31701839e+00 7.38452673e-01
9.83566880e-01 -2.62345701e-01 4.37547177e-01 4.40881699e-01
2.14340776e-01 3.24613042e-02 2.91994624e-02 -3.97118360e-01
5.39731979e-02 1.10684514e+00 -3.47292721e-01 -1.29201627e+00
-7.60045528e-01 -5.27240038e-01 -6.34413302e-01 -1.60533324e-01
4.06034261e-01 2.82942593e-01 -5.28371632e-01 9.99452412e-01
7.25856662e-01 1.58433750e-01 -3.11329067e-01 6.14431739e-01
4.89357024e-01 4.17139113e-01 -1.77941069e-01 -5.01334906e-01
1.25002563e+00 -4.58375812e-01 -3.36121649e-01 2.26429567e-01
7.03088522e-01 -9.84342754e-01 2.35503152e-01 5.69373965e-01
-7.90912271e-01 -7.97590390e-02 -5.77866614e-01 2.73325235e-01
-4.38902229e-01 -6.33926570e-01 9.34962749e-01 1.08260167e+00
-1.25028253e+00 6.13236547e-01 -6.08841598e-01 -1.03561246e+00
1.78035840e-01 6.53367758e-01 -5.37718892e-01 -2.27312237e-01
-6.12254858e-01 8.91698539e-01 5.61388612e-01 -3.50929558e-01
-1.92613021e-01 -2.68985510e-01 -4.74162877e-01 -6.12490848e-02
6.69294894e-01 -5.89053333e-01 5.94665170e-01 -5.12623310e-01
-8.38160038e-01 9.40632463e-01 -1.83247358e-01 -5.91735721e-01
4.28903624e-02 4.23529059e-01 -4.78957176e-01 6.27544969e-02
1.43263355e-01 7.89597556e-02 2.27498382e-01 -6.12453163e-01
-7.45072961e-01 -6.80332184e-01 -1.83328658e-01 5.90010323e-02
-2.47184753e-01 2.57549167e-01 -1.77875340e-01 -3.58030885e-01
3.75687778e-01 -8.91510010e-01 -7.49098003e-01 -6.81724131e-01
-3.28541666e-01 -3.65993738e-01 7.88433909e-01 -4.04096842e-01
1.26868296e+00 -1.73101139e+00 2.36293435e-01 1.04767931e+00
4.43395346e-01 1.72903657e-01 3.53202559e-02 1.09774160e+00
4.91527282e-03 1.54796287e-01 -8.58016014e-02 4.99375015e-01
-5.22362411e-01 3.09770644e-01 1.67049199e-01 7.98735797e-01
-1.02012694e-01 2.97579944e-01 -7.74014056e-01 -4.16118592e-01
1.26101911e-01 5.66770062e-02 -4.49928969e-01 -1.94112375e-01
9.04402435e-02 2.66872406e-01 -7.26099193e-01 6.07472301e-01
7.56176233e-01 -4.34939533e-01 7.44411826e-01 3.23223531e-01
-3.65004212e-01 5.83507307e-02 -1.46094418e+00 1.09621382e+00
1.29684031e-01 4.14783984e-01 -5.92982117e-03 -1.34936154e+00
1.04181015e+00 1.68109000e-01 8.29838157e-01 -4.22142774e-01
3.00769925e-01 1.68687761e-01 1.84948400e-01 -3.59869123e-01
5.66136420e-01 -1.32076755e-01 1.67347044e-01 6.55181944e-01
-4.10888761e-01 1.92348540e-01 5.28235555e-01 5.49476445e-01
1.65343976e+00 -4.05560732e-01 6.36249065e-01 -6.59929335e-01
6.45423770e-01 4.55304056e-01 2.48208448e-01 5.11291087e-01
-1.44495159e-01 1.34599078e-02 6.45552397e-01 -7.20652461e-01
-9.08394217e-01 -6.19971812e-01 -6.53069317e-02 7.38459349e-01
2.05552444e-01 -7.28336751e-01 -5.75174272e-01 -4.13435340e-01
9.92140174e-02 1.60635054e-01 -2.49172062e-01 2.90601432e-01
-5.92505157e-01 -9.87058997e-01 7.01380596e-02 -2.75086705e-02
2.59969026e-01 -7.81472385e-01 -5.37388086e-01 4.88714159e-01
2.71681219e-01 -9.23701763e-01 -1.35357743e-02 -1.22615751e-02
-1.40033078e+00 -1.61979818e+00 -4.21442538e-01 -8.89599204e-01
8.55012000e-01 7.09995627e-01 1.06820118e+00 5.33830106e-01
-6.38659716e-01 1.36578217e-01 -6.75751507e-01 -3.31302106e-01
-3.18314224e-01 3.66120011e-01 -7.77763277e-02 -3.60228509e-01
7.71784604e-01 -6.02229834e-01 -2.39794761e-01 2.40350038e-01
-8.43947291e-01 -2.28220001e-01 6.56611085e-01 4.36423153e-01
4.26873118e-01 5.76301634e-01 5.31581283e-01 -1.42959845e+00
8.67499590e-01 -7.06571162e-01 -8.06019187e-01 8.18889216e-02
-1.06542695e+00 1.90453425e-01 2.98448026e-01 1.22475371e-01
-6.66105628e-01 2.90677875e-01 -2.74461180e-01 4.23382103e-01
-1.71367437e-01 5.26630878e-01 1.82811543e-02 -4.71367627e-01
5.81104994e-01 1.46456257e-01 1.97583571e-01 -7.06319928e-01
9.60449427e-02 5.14143944e-01 -9.93882492e-02 -2.64982671e-01
7.61703730e-01 6.35035872e-01 4.54535753e-01 -1.33481944e+00
-2.17830971e-01 -1.26796854e+00 -3.61343920e-01 -5.33633120e-02
2.22416356e-01 -4.63885725e-01 -4.59479630e-01 1.88065961e-01
-8.08569133e-01 3.94368768e-01 2.05168709e-01 3.71479154e-01
-1.70969814e-01 8.44303608e-01 -2.81239599e-01 -5.80998302e-01
-2.28448778e-01 -6.80643618e-01 6.71322882e-01 1.87338859e-01
-3.77702534e-01 -8.35537970e-01 5.86364329e-01 3.80560726e-01
3.30178857e-01 3.60217541e-01 8.30058813e-01 -8.54563653e-01
-8.58222783e-01 -2.61968315e-01 -2.46188328e-01 -3.59342068e-01
8.54468495e-02 -1.64347306e-01 -3.33350837e-01 -6.83108032e-01
-3.01587075e-01 1.95498690e-01 4.93524849e-01 2.02685699e-01
7.91768730e-01 -2.43061006e-01 -9.12608802e-01 3.76993924e-01
1.57268012e+00 2.71287411e-01 6.80847287e-01 4.74422514e-01
1.98936060e-01 5.61375141e-01 4.18646663e-01 4.49661076e-01
1.22363538e-01 3.86340678e-01 9.50621665e-02 1.26156537e-02
2.46357024e-01 1.21546775e-01 -2.87578523e-01 7.14043200e-01
-3.25582564e-01 -1.39516219e-01 -8.62055242e-01 6.69103920e-01
-1.70197070e+00 -1.10082519e+00 -8.83365870e-01 2.10708284e+00
4.70776021e-01 6.19674437e-02 5.64727783e-01 3.96789968e-01
7.62247324e-01 -2.08631247e-01 -2.09502056e-02 -8.69331598e-01
1.06645979e-01 6.94712222e-01 6.40663028e-01 3.68147999e-01
-6.02015853e-01 7.69894779e-01 7.38152790e+00 5.50184846e-01
-6.52370095e-01 -2.46534958e-01 2.42468253e-01 2.69369006e-01
-3.91470402e-01 3.95259440e-01 -7.32666373e-01 1.20412670e-01
1.03088737e+00 -5.57892978e-01 3.64302665e-01 8.03533196e-01
2.46176466e-01 -6.11588180e-01 -4.38558012e-01 5.06913781e-01
-1.20886862e-01 -1.33086658e+00 -6.58430606e-02 6.80849969e-01
8.42407405e-01 -1.66848674e-01 -6.00995302e-01 -1.77112758e-01
1.16495326e-01 -6.71410680e-01 -2.28925616e-01 1.42939657e-01
2.36258239e-01 -9.83737767e-01 7.64750779e-01 3.77792478e-01
-1.29643321e+00 1.05739972e-02 -4.47706401e-01 -5.99685431e-01
5.60459234e-02 7.56524563e-01 -1.43409717e+00 7.18354642e-01
6.30556107e-01 3.70049149e-01 -4.86422449e-01 1.65673923e+00
1.87267706e-01 5.16086042e-01 -4.45010215e-01 -5.34652174e-01
2.54110634e-01 -3.71824980e-01 8.44717264e-01 1.25768673e+00
2.14847058e-01 1.85143575e-01 3.42682600e-01 4.07890409e-01
1.42724290e-01 5.30375302e-01 -9.23545957e-01 -2.08013251e-01
2.76243269e-01 1.12937272e+00 -1.42417550e+00 -2.73507953e-01
-5.18534780e-01 5.36177874e-01 3.58692467e-01 -5.58953062e-02
-3.70463908e-01 -6.53461993e-01 5.50670326e-01 6.83910489e-01
3.77083957e-01 -3.36016119e-01 -1.20110802e-01 -8.29938710e-01
-4.38998431e-01 -1.00560260e+00 9.15657520e-01 1.71578377e-02
-9.96289134e-01 6.42668188e-01 3.30239564e-01 -6.09930336e-01
-2.10436419e-01 -4.33119237e-01 -6.50426388e-01 8.44973862e-01
-9.97868717e-01 -6.82796180e-01 -6.25768676e-02 6.44138455e-01
6.27116933e-02 -2.29699463e-01 6.56370342e-01 1.72919393e-01
-3.75720561e-01 1.04099728e-01 1.13679305e-01 -3.91613990e-01
3.13481003e-01 -9.76769626e-01 2.63172060e-01 1.13708830e+00
3.03763986e-01 6.45886838e-01 9.15845931e-01 -8.58214438e-01
-1.49553931e+00 -7.41827548e-01 1.19732320e+00 3.70688811e-02
6.67905629e-01 -1.77490748e-02 -7.79594064e-01 2.61981934e-01
2.90760964e-01 -5.63406527e-01 7.32981503e-01 4.78956044e-01
1.59263745e-01 -6.87451586e-02 -1.36391068e+00 2.48443618e-01
9.87347364e-01 8.28168318e-02 -3.42046559e-01 3.05446386e-01
1.48881733e-01 -6.14645593e-02 -7.51491845e-01 2.11157277e-01
3.40322137e-01 -1.14490676e+00 6.62168443e-01 -4.66084331e-01
-3.66692960e-01 -3.60017031e-01 4.67811018e-01 -9.20030296e-01
-6.22910678e-01 -8.96318376e-01 2.30387170e-02 6.24846816e-01
2.55852997e-01 -6.94955647e-01 1.28497195e+00 3.16340059e-01
2.60663986e-01 -8.32765281e-01 -9.21869397e-01 -6.14648640e-01
-4.32214230e-01 -3.29502560e-02 5.31237364e-01 8.14388752e-01
2.59246111e-01 5.85532308e-01 -2.77456790e-01 -5.94680458e-02
7.93827176e-01 5.31605780e-01 9.38477397e-01 -1.51170492e+00
-1.32018551e-01 -3.75303686e-01 -8.64280105e-01 -5.64624429e-01
-4.07674998e-01 -9.69399750e-01 -5.40384293e-01 -1.83220661e+00
5.64863861e-01 -4.37274158e-01 3.88046317e-02 3.63135040e-01
8.50159600e-02 3.38264853e-01 -1.24680445e-01 -6.06059060e-02
-5.77853024e-01 -2.86820322e-01 9.66606379e-01 1.78533301e-01
-4.15591627e-01 2.44944915e-01 -9.52897727e-01 3.57553363e-01
9.62276459e-01 -7.52464592e-01 -3.91597122e-01 1.09692223e-01
2.27977008e-01 2.08735503e-02 -1.27044069e-02 -7.61800230e-01
2.65365630e-01 -2.73058772e-01 1.56515792e-01 -8.86867166e-01
-1.49294347e-01 -7.98502445e-01 8.35942805e-01 1.00101185e+00
2.10989922e-01 4.41967428e-01 3.52919251e-02 7.43047178e-01
-5.20581678e-02 -4.71210241e-01 7.68728435e-01 -4.50212777e-01
-1.04756379e+00 1.86018452e-01 -5.34290493e-01 -2.50861138e-01
1.45743120e+00 -5.20020783e-01 -1.19899474e-01 -1.15213133e-01
-5.92593908e-01 2.07553431e-01 5.64262390e-01 8.64519179e-02
5.69975376e-01 -7.17568696e-01 -7.52097428e-01 1.87434196e-01
-6.46235943e-02 -4.03956056e-01 -1.19318508e-01 9.90970969e-01
-9.64492679e-01 6.81491733e-01 -3.31341088e-01 -3.98118407e-01
-1.84554386e+00 7.48406053e-01 -1.74506858e-01 -5.12744844e-01
-8.47557604e-01 6.47915006e-01 -3.78027022e-01 5.65051958e-02
5.81000512e-03 1.56385645e-01 -8.10695961e-02 -3.04908007e-01
4.83158678e-01 8.55951130e-01 9.80045572e-02 -5.14137328e-01
-6.57502472e-01 1.76914603e-01 -1.48696080e-01 4.13758814e-01
1.45966840e+00 -1.27870634e-01 -7.85242498e-01 -8.76821056e-02
1.03446293e+00 1.69209376e-01 -1.00312263e-01 -1.17331803e-01
5.59210122e-01 -8.25964034e-01 1.04731433e-02 -4.35881793e-01
-8.05812955e-01 1.08125053e-01 2.26285249e-01 8.72066557e-01
1.23884654e+00 2.96376199e-01 6.11071706e-01 5.65041780e-01
6.80953145e-01 -1.04963148e+00 -2.56048948e-01 2.96081841e-01
5.50907612e-01 -9.04751539e-01 5.37107408e-01 -6.93742394e-01
-2.62256294e-01 1.11822999e+00 3.86826158e-01 -2.57170320e-01
8.06930184e-01 4.05120224e-01 -5.25698245e-01 -4.69966084e-01
-7.43363976e-01 -5.59180200e-01 1.16438210e-01 6.17509425e-01
5.66712499e-01 5.99115789e-02 -1.25837278e+00 -1.09890535e-01
9.60758179e-02 -1.12021603e-02 7.21481919e-01 1.07675767e+00
-8.95719588e-01 -1.57583296e+00 -3.70976985e-01 9.42069829e-01
-4.88111168e-01 1.23541370e-01 -8.13361406e-01 9.69850421e-01
-1.33819589e-02 8.92974973e-01 -1.22679748e-01 -2.49205112e-01
1.34873003e-01 -3.49278986e-01 5.45688033e-01 -6.15068734e-01
-4.73425806e-01 2.03507543e-01 4.62774903e-01 -4.95069444e-01
-6.72881365e-01 -6.46491408e-01 -1.10446668e+00 -8.76995444e-01
-5.46368718e-01 8.68829906e-01 6.29360080e-01 7.65195727e-01
3.21795195e-01 4.31357734e-02 7.11840093e-01 -1.14652492e-01
-9.91656408e-02 -7.50313282e-01 -9.43529904e-01 1.07005395e-01
-2.93763250e-01 -4.33623910e-01 3.54942773e-03 -3.08261126e-01]
|
[6.954156875610352, 5.267507553100586]
|
36d443fc-cbad-48a6-94ef-acf53fb998fe
|
a-papier-macha-approach-to-learning-3d
| null | null |
http://openaccess.thecvf.com/content_cvpr_2018/html/Groueix_A_Papier-Mache_Approach_CVPR_2018_paper.html
|
http://openaccess.thecvf.com/content_cvpr_2018/papers/Groueix_A_Papier-Mache_Approach_CVPR_2018_paper.pdf
|
A Papier-Mâché Approach to Learning 3D Surface Generation
|
We introduce a method for learning to generate the surface of 3D shapes. Our approach represents a 3D shape as a collection of parametric surface elements and, in contrast to methods generating voxel grids or point clouds, naturally infers a surface representation of the shape. Beyond its novelty, our new shape generation framework, AtlasNet, comes with significant advantages, such as improved precision and generalization capabilities, and the possibility to generate a shape of arbitrary resolution without memory issues. We demonstrate these benefits and compare to strong baselines on the ShapeNet benchmark for two applications: (i) auto-encoding shapes, and (ii) single-view reconstruction from a still image. We also provide results showing its potentialfor other applications, such as morphing, parametrization, super-resolution, matching, and co-segmentation.
|
['Thibault Groueix', 'Bryan C. Russell', 'Matthew Fisher', 'Mathieu Aubry', 'Vladimir G. Kim']
|
2018-06-01
| null | null | null |
cvpr-2018-6
|
['3d-surface-generation']
|
['computer-vision']
|
[ 2.73047596e-01 3.86567920e-01 2.52119213e-01 -2.49007016e-01
-1.05537987e+00 -7.99140215e-01 1.00052619e+00 7.47845322e-02
1.21831164e-01 5.33806860e-01 -7.44589511e-03 -6.52003735e-02
3.19928899e-02 -1.09639013e+00 -7.26735532e-01 -3.52932483e-01
-9.70242952e-04 1.21992660e+00 3.77643824e-01 -3.28555740e-02
1.59686118e-01 1.35456789e+00 -1.64738202e+00 1.75046265e-01
6.67518795e-01 1.08133137e+00 -6.64063469e-02 4.15636152e-01
-4.14013982e-01 -2.94658154e-01 -1.71591237e-01 -3.96918774e-01
4.91918802e-01 8.21193680e-02 -8.43906105e-01 4.83299315e-01
8.93727660e-01 -1.11483924e-01 1.88337252e-01 8.46343517e-01
4.76799577e-01 -4.27384153e-02 9.51543868e-01 -1.03853381e+00
-3.28973711e-01 2.03888007e-02 -6.01112962e-01 -4.15919930e-01
3.84836257e-01 1.44142076e-01 6.83842659e-01 -1.27486992e+00
1.14308274e+00 1.47231913e+00 8.94778371e-01 4.78562713e-01
-1.82062626e+00 -2.77396798e-01 -1.79838315e-01 -6.19433463e-01
-1.35616231e+00 -5.88487029e-01 7.67987013e-01 -6.34336770e-01
9.26661074e-01 2.93122381e-01 7.28055060e-01 5.87214947e-01
4.23038341e-02 6.50246382e-01 1.12470388e+00 -1.33960277e-01
1.92855015e-01 -1.30898640e-01 -5.66986382e-01 6.56276882e-01
-3.35432414e-05 2.73212731e-01 -1.20891243e-01 -5.19922554e-01
1.56300747e+00 -3.92726660e-01 4.85934913e-02 -9.24636841e-01
-1.28388715e+00 5.02651393e-01 3.45779747e-01 -2.52673626e-02
-3.87344807e-01 4.42313880e-01 7.52165839e-02 -1.17878981e-01
6.96458042e-01 5.35636604e-01 -4.04648006e-01 5.79332374e-02
-1.12058496e+00 6.80492461e-01 6.31416440e-01 1.22736895e+00
9.53658938e-01 3.62533629e-01 1.47653237e-01 6.66190982e-01
2.48320445e-01 7.25612044e-01 5.24665862e-02 -1.33606577e+00
1.58338591e-01 6.84920371e-01 7.79572800e-02 -8.59153032e-01
-4.59092915e-01 -1.51638299e-01 -6.80200040e-01 7.79262364e-01
9.45420414e-02 4.89106700e-02 -1.27089405e+00 1.26331854e+00
6.08758450e-01 4.43018109e-01 -2.86093682e-01 6.59184456e-01
1.06851673e+00 3.73876154e-01 -1.57720909e-01 2.16864333e-01
1.18197775e+00 -4.00948465e-01 -9.60336700e-02 -4.99614282e-03
1.73207030e-01 -7.86161005e-01 8.40097964e-01 3.38053703e-01
-1.72789204e+00 -4.04127926e-01 -7.12058663e-01 -2.12813884e-01
-2.72983402e-01 -3.73844385e-01 7.74968863e-01 4.04501528e-01
-1.40772772e+00 8.60167921e-01 -9.56502795e-01 -1.12803131e-01
9.13055778e-01 3.94779384e-01 -5.39908707e-01 2.34251395e-01
-3.26329499e-01 5.39501190e-01 2.36532912e-01 -6.76563621e-01
-7.97296822e-01 -1.37242067e+00 -1.03719509e+00 -1.23088509e-01
-4.14700061e-02 -1.17774391e+00 1.00906098e+00 -6.31118953e-01
-1.37047982e+00 1.43618000e+00 -6.32586703e-02 -1.96665421e-01
5.23056209e-01 2.89737463e-01 4.06670235e-02 1.22795980e-02
-9.47551727e-02 1.14444959e+00 8.23441505e-01 -1.67150390e+00
-3.06225628e-01 -4.34030175e-01 -1.69048026e-01 2.00971931e-01
5.00832260e-01 -2.79268622e-01 -4.33949828e-01 -7.94469416e-01
6.23377800e-01 -7.53930211e-01 -4.60635781e-01 4.05039996e-01
-5.18889427e-01 -5.76619990e-02 7.91086912e-01 -4.85374689e-01
3.81799549e-01 -1.98346508e+00 2.49711707e-01 5.17369926e-01
2.17879638e-01 -4.31413576e-02 -1.66492254e-01 2.31403559e-01
-3.35437693e-02 5.27682066e-01 -7.99808204e-01 -6.18465841e-01
1.17357839e-02 1.61791891e-01 -3.42802793e-01 2.53297329e-01
3.99039149e-01 1.15656173e+00 -7.96984971e-01 -5.54237902e-01
3.09943885e-01 7.63673067e-01 -6.55770421e-01 6.06164224e-02
-6.07739985e-01 7.45145082e-01 -3.66017133e-01 6.66249633e-01
7.98560977e-01 -2.77744472e-01 -1.54725939e-01 -3.30374509e-01
-4.51272316e-02 2.04240426e-01 -1.36178160e+00 1.98637760e+00
-5.12015700e-01 7.80335143e-02 4.25364792e-01 -3.59260499e-01
1.03094304e+00 3.72094005e-01 7.02656448e-01 -3.69310647e-01
-6.76618591e-02 2.95326114e-01 -4.82886791e-01 1.42303765e-01
4.42393243e-01 -3.62993896e-01 1.05942294e-01 6.16761804e-01
2.45978720e-02 -1.21316957e+00 -2.06235379e-01 4.48338464e-02
5.79754710e-01 7.35117674e-01 2.20153227e-01 -4.69187140e-01
1.92803517e-01 2.95221582e-02 9.75155607e-02 1.82148010e-01
6.04824364e-01 1.23119974e+00 1.32249936e-01 -4.81052160e-01
-1.42420340e+00 -1.51218188e+00 -6.46755397e-01 3.65433902e-01
4.51240875e-02 -1.54365748e-01 -6.26388669e-01 -4.17772889e-01
5.13306320e-01 5.67940116e-01 -4.50019091e-01 4.70177531e-01
-7.68376410e-01 -3.52024764e-01 3.07579368e-01 6.18264616e-01
1.36582464e-01 -1.14137459e+00 -5.37403524e-01 9.58920643e-02
1.55118242e-01 -1.20637250e+00 -6.15428209e-01 -2.76538044e-01
-1.45198107e+00 -1.01711631e+00 -5.32010615e-01 -8.21481466e-01
8.97696316e-01 -8.81616399e-02 1.63268781e+00 2.04073027e-01
-4.19991285e-01 7.02046454e-01 3.41896296e-01 -3.04913372e-01
-6.83981836e-01 -1.08856149e-02 -1.43611670e-01 -1.82516545e-01
-3.88191968e-01 -1.18508589e+00 -4.99397933e-01 1.60699710e-01
-9.87305045e-01 2.78583139e-01 2.97706515e-01 3.89788032e-01
1.41025972e+00 -2.83233702e-01 4.14879322e-01 -1.04266083e+00
3.50867510e-01 -1.96307927e-01 -8.22192609e-01 2.32460070e-02
-3.95345032e-01 1.42207578e-01 3.02456886e-01 -1.11617394e-01
-1.03016734e+00 2.34307498e-01 -3.00176769e-01 -4.73292381e-01
-6.51996493e-01 7.34633952e-02 -2.13879764e-01 -3.24740142e-01
7.45634377e-01 1.59037098e-01 4.70597237e-01 -7.70691574e-01
8.28805327e-01 -1.91582832e-02 6.82041824e-01 -9.73289669e-01
1.10482907e+00 9.51713026e-01 4.15552944e-01 -9.17860270e-01
-2.19218969e-01 -5.95318638e-02 -1.03673697e+00 2.30604783e-01
6.55565739e-01 -7.44864285e-01 -4.22296047e-01 2.94248641e-01
-1.38246822e+00 -4.44522202e-01 -6.70789599e-01 -1.25208780e-01
-1.15418303e+00 2.61585563e-01 -4.43732232e-01 -4.95844960e-01
-4.81249660e-01 -9.54240918e-01 1.55797970e+00 -5.16402088e-02
-3.35388482e-01 -1.17863631e+00 -1.08304396e-01 8.67632926e-02
4.26777065e-01 9.14535701e-01 1.25622964e+00 -2.12452471e-01
-9.34318900e-01 3.36949602e-02 -2.21365348e-01 3.12077273e-02
1.37572691e-01 2.29779389e-02 -9.29305911e-01 -1.60468802e-01
-5.27393281e-01 -2.80135691e-01 3.58311594e-01 3.84577185e-01
1.32967150e+00 -1.32660836e-01 -5.71194410e-01 1.09083307e+00
1.49687219e+00 -8.50440785e-02 7.74254084e-01 -2.23627791e-01
6.73441231e-01 5.94028592e-01 1.37334168e-01 3.45595628e-01
3.39480937e-01 9.36263084e-01 5.59981525e-01 -2.76373029e-01
-6.47657990e-01 -3.70479405e-01 -3.44296157e-01 5.75208545e-01
-2.80401051e-01 2.91082531e-01 -1.03226376e+00 6.22023582e-01
-1.28425920e+00 -6.28419161e-01 -1.63763463e-01 2.36651349e+00
7.98416674e-01 -9.01872739e-02 3.06443274e-01 -2.26382658e-01
3.70242804e-01 -3.92412245e-02 -6.50125802e-01 -2.94548452e-01
-2.07719013e-01 7.65644729e-01 2.52414793e-01 7.77781248e-01
-9.74597812e-01 1.04621446e+00 7.41108513e+00 7.86920428e-01
-1.08485103e+00 -1.32573828e-01 7.88048804e-01 2.15116411e-01
-9.93009686e-01 -1.50113940e-01 -5.72612226e-01 5.11841848e-02
4.10359263e-01 -1.97195709e-01 4.68166918e-01 6.41351402e-01
-1.09954916e-01 2.59580702e-01 -1.23638082e+00 1.01802719e+00
-8.70396569e-03 -1.77524602e+00 4.11089599e-01 1.86983526e-01
8.94570112e-01 1.58665836e-01 -7.13127032e-02 -1.35821059e-01
3.94684583e-01 -1.47060013e+00 7.88855851e-01 5.28416634e-01
1.44437170e+00 -6.68104708e-01 1.36528626e-01 3.86842310e-01
-1.17065108e+00 7.94230759e-01 -3.35143745e-01 4.06864822e-01
3.93776298e-01 5.56482017e-01 -1.00115120e+00 6.73931956e-01
3.76142651e-01 5.17071486e-01 -3.83468211e-01 1.17204452e+00
-1.27410479e-02 2.28481501e-01 -5.67740500e-01 6.18047178e-01
-1.46252289e-01 -4.91647780e-01 7.46919930e-01 1.09465873e+00
3.52025062e-01 2.32939273e-01 2.19316244e-01 1.46716475e+00
-8.58504605e-03 1.56665638e-01 -9.33766782e-01 1.66571215e-01
6.90041065e-01 1.31216288e+00 -9.35611904e-01 -2.61277795e-01
-4.80413772e-02 7.34090090e-01 2.92416751e-01 2.62986481e-01
-4.42561924e-01 -7.14815035e-02 6.78913236e-01 7.05952883e-01
4.13665593e-01 -5.37947059e-01 -9.21092510e-01 -7.25246489e-01
-1.32271647e-01 -6.58736408e-01 -7.10272118e-02 -8.76176775e-01
-1.19039750e+00 6.88100874e-01 2.80102372e-01 -1.13550758e+00
-2.82076150e-01 -4.17624623e-01 -4.60536689e-01 9.60150361e-01
-1.46373045e+00 -1.42836702e+00 -2.73655444e-01 2.70708889e-01
3.54031742e-01 5.80457412e-03 1.06873190e+00 1.37254409e-03
3.63823384e-01 2.55339712e-01 -4.10762638e-01 -1.25477418e-01
1.93734393e-01 -1.41459441e+00 1.10847080e+00 3.34521443e-01
3.51787210e-01 3.13533098e-01 2.97626972e-01 -7.41837919e-01
-1.30339670e+00 -1.04739881e+00 3.76582593e-01 -8.16241086e-01
1.00379296e-01 -3.65698546e-01 -1.02886045e+00 6.94956601e-01
-1.59120902e-01 2.39455715e-01 2.84469306e-01 -1.08415440e-01
-2.68129289e-01 2.59808570e-01 -1.62732029e+00 5.27689755e-01
1.47776687e+00 -1.95942670e-01 -3.91857088e-01 2.57423341e-01
5.88566124e-01 -9.45960045e-01 -1.29855490e+00 5.33114195e-01
5.82539916e-01 -9.75614786e-01 1.59004450e+00 -6.56468868e-01
4.57646221e-01 -2.03338459e-01 -8.06422606e-02 -1.44130313e+00
-3.84828925e-01 -7.81428099e-01 -1.43156677e-01 1.00556970e+00
3.03353667e-01 -5.31948507e-01 1.16510665e+00 6.41566277e-01
-4.48797375e-01 -9.90525782e-01 -1.05521572e+00 -7.94283986e-01
4.65988666e-01 -4.90691453e-01 1.22192109e+00 7.84550011e-01
-6.63464427e-01 -5.24692191e-03 2.13393301e-01 1.45688131e-01
9.27534044e-01 4.91023183e-01 9.06054437e-01 -1.62040544e+00
-1.35924131e-01 -6.92248464e-01 -5.52515388e-01 -1.22735178e+00
1.26500338e-01 -1.32324564e+00 -2.02530608e-01 -1.79584777e+00
-3.25458288e-01 -1.00870216e+00 3.42361629e-01 4.06888217e-01
3.05382103e-01 5.63273728e-01 -2.02370975e-02 3.34096816e-03
-3.04365996e-02 3.76286209e-01 1.77004206e+00 1.10687837e-01
-1.82559639e-01 -1.26699626e-01 -4.83174175e-01 8.97210658e-01
5.44559181e-01 -2.11995199e-01 -3.08022022e-01 -6.07391536e-01
-3.10793221e-02 3.68106037e-01 5.12803137e-01 -7.51332343e-01
-9.82648656e-02 -2.62479782e-01 4.38842863e-01 -8.05430353e-01
7.43436933e-01 -7.82391608e-01 5.86347103e-01 1.00165099e-01
1.13242663e-01 1.73433125e-01 4.73096907e-01 2.44517997e-01
1.34494409e-01 2.10774317e-02 9.63599622e-01 -4.00164068e-01
-3.74361336e-01 7.42789865e-01 1.86694980e-01 3.58829498e-01
8.67570102e-01 -7.60094643e-01 3.28490213e-02 -1.88237369e-01
-8.93820584e-01 1.08305179e-02 1.07126057e+00 3.32278550e-01
1.03348410e+00 -1.75826585e+00 -9.54164207e-01 5.66619575e-01
-6.95191622e-02 6.22697949e-01 6.35033697e-02 4.62009758e-01
-6.39459491e-01 9.96541977e-02 -7.72802904e-02 -8.89943540e-01
-9.27808642e-01 1.69926077e-01 6.31099403e-01 1.08989693e-01
-1.04772949e+00 6.45346463e-01 4.08877850e-01 -8.05313289e-01
-1.43831804e-01 -2.65948504e-01 1.89304486e-01 -2.61969924e-01
2.02896953e-01 2.97878206e-01 3.58101875e-01 -7.94220328e-01
-3.31259638e-01 1.02919078e+00 5.55560410e-01 -1.73076391e-01
1.44255018e+00 2.97264367e-01 -3.12748611e-01 2.81443119e-01
8.60185683e-01 2.43120402e-01 -1.37043440e+00 -2.19365045e-01
3.78426947e-02 -6.44687831e-01 -8.24091211e-02 -8.13493848e-01
-1.16831136e+00 7.46065736e-01 2.70592421e-01 7.93628320e-02
5.60534656e-01 2.66029060e-01 9.54091966e-01 -4.52523977e-02
8.11650276e-01 -5.86594105e-01 -1.00980820e-02 4.42523897e-01
1.34470773e+00 -9.16186452e-01 1.59095615e-01 -8.50360513e-01
-3.11700165e-01 1.21350801e+00 3.16677451e-01 -5.46435654e-01
9.05760229e-01 7.27151155e-01 4.16420214e-02 -5.29024839e-01
-5.30934036e-01 -4.79587689e-02 8.47827613e-01 1.04732931e+00
4.65518385e-01 2.19683617e-01 1.63733616e-01 1.80471659e-01
-5.42699933e-01 -3.02898884e-01 3.86874199e-01 5.95249474e-01
-2.58517385e-01 -1.25002432e+00 -3.75701368e-01 5.79367459e-01
-1.49748385e-01 8.91790986e-02 -2.81355053e-01 8.44467044e-01
-3.90954735e-03 2.03604668e-01 4.59412932e-01 9.08022933e-03
5.83780527e-01 1.46178693e-01 9.02004957e-01 -1.05399132e+00
-3.99230063e-01 1.03461727e-01 2.59845369e-02 -7.60856628e-01
-2.40944356e-01 -8.36351097e-01 -1.53967905e+00 -7.73321092e-02
9.87330973e-02 -1.55469120e-01 7.43625343e-01 5.42078555e-01
8.20943534e-01 8.94131660e-02 3.44678938e-01 -1.47885931e+00
-3.54949862e-01 -5.11761725e-01 -4.63711590e-01 8.02224815e-01
4.70587723e-02 -7.02343464e-01 -1.17767274e-01 1.58512607e-01]
|
[8.776239395141602, -3.6381020545959473]
|
0f01ab96-df9d-414a-b174-27a075322f22
|
time-series-segmentation-applied-to-a-new
| null | null |
https://ceur-ws.org/Vol-3379/DARLI-AP_2023_2.pdf
|
https://ceur-ws.org/Vol-3379/DARLI-AP_2023_2.pdf
|
Time Series Segmentation Applied to a New Data Set for Mobile Sensing of Human Activities
|
Human activity recognition (HAR) systems implement workflows that automatically detect activities from motion data,
captured e.g. by wearable devices such as smartphones. These devices contain multiple sensors that record human motion as
acceleration, rotation and orientation in long time series (TS) data. As a first step, HAR methods typically partition such
recordings into smaller subsequences before applying feature extraction and classification. In this study, we evaluate the
performance of 6 classical and recently published TS segmentation (TSS) algorithms on a new large HAR benchmark of 126
TS with up to 13 different activities, called MOSAD, recorded with 6 participants using ordinary smartphone sensors. Our
results show that the ClaSP algorithm achieves significantly more accurate results compared to the other methods, scoring
the best segmentations in 57 out of 126 TS. The FLOSS algorithm also shows promising results, particularly for long TS with
many segments. MOSAD is freely available at https://github.com/ermshaua/mobile-sensing-human-activity-data-set.
|
['Ulf Leser', 'Sunita Singh', 'Arik Ermshaus']
|
2023-03-28
| null | null | null |
data-analytics-solutions-for-real-life
|
['activity-recognition', 'human-activity-recognition', 'change-point-detection', 'time-series', 'human-activity-recognition']
|
['computer-vision', 'computer-vision', 'time-series', 'time-series', 'time-series']
|
[ 2.15393379e-01 -4.06937450e-01 -3.15469563e-01 -5.07584177e-02
-7.13997483e-01 -5.72079897e-01 5.14250159e-01 1.93277687e-01
-5.06236970e-01 5.92407584e-01 4.13309902e-01 1.82171687e-02
4.94090915e-02 -4.77080792e-01 -3.62905711e-01 -5.27724802e-01
-3.38709503e-01 2.64217407e-01 4.33998942e-01 2.13943467e-01
8.84225518e-02 3.31809253e-01 -1.56514370e+00 8.14660266e-02
5.61978638e-01 1.05424583e+00 -1.16267696e-01 8.94997716e-01
2.52592295e-01 6.14813566e-01 -7.61663854e-01 4.29046378e-02
1.01580195e-01 -5.82300961e-01 -6.89401627e-01 1.35672495e-01
1.31286830e-01 8.46026093e-03 -5.96615002e-02 1.92667454e-01
4.97708023e-01 1.74645171e-01 2.84156144e-01 -1.22703385e+00
2.42734253e-01 1.75072938e-01 -4.92535293e-01 3.94055873e-01
1.18269920e+00 8.22533071e-02 5.08047223e-01 -8.75556946e-01
4.54884171e-01 6.34530723e-01 1.11392605e+00 2.54257143e-01
-1.17802298e+00 -3.17879319e-01 -4.08362180e-01 2.88878471e-01
-1.67288947e+00 -4.87537444e-01 4.63961631e-01 -5.91911554e-01
1.23546743e+00 6.83403432e-01 1.21082962e+00 1.49771142e+00
1.41008645e-01 1.02882564e+00 8.28867495e-01 -1.54311270e-01
4.62098300e-01 -4.60672349e-01 4.24488276e-01 3.26341480e-01
6.63962126e-01 -5.45072138e-01 -7.80190766e-01 -3.73862475e-01
5.78094363e-01 3.72576118e-01 -1.97144479e-01 -1.77524954e-01
-1.78480065e+00 1.05681077e-01 -2.42683813e-01 6.93471253e-01
-7.45846808e-01 -1.04843341e-01 3.98120999e-01 2.13821214e-02
2.60618210e-01 1.30028278e-01 -4.06740665e-01 -1.06048059e+00
-1.15021336e+00 3.64009529e-01 8.63058507e-01 7.33093560e-01
4.91276413e-01 -3.79763335e-01 -3.67021769e-01 7.55131125e-01
-1.53659738e-03 4.93896186e-01 7.51724958e-01 -1.03171647e+00
5.31367362e-01 7.62990713e-01 4.52010185e-01 -8.98804724e-01
-8.23043942e-01 -9.60463732e-02 -9.13939416e-01 -5.42715609e-01
7.93184161e-01 -2.79411256e-01 -4.51024890e-01 1.26901329e+00
5.24036765e-01 6.57501221e-01 -2.90132165e-01 7.40207374e-01
4.68120217e-01 4.34812546e-01 1.38630331e-01 -3.93929273e-01
1.45922267e+00 -8.19228113e-01 -8.05456758e-01 -1.74275845e-01
9.00146127e-01 -4.78627145e-01 1.10542059e+00 6.82282984e-01
-1.01033974e+00 -6.29638255e-01 -1.11875558e+00 1.39591485e-01
-2.92125225e-01 3.11201453e-01 3.60565394e-01 9.82945979e-01
-6.49242997e-01 6.28440738e-01 -1.50105286e+00 -7.48893678e-01
4.85428870e-01 3.37949276e-01 -3.74560088e-01 3.03684771e-01
-8.58666301e-01 3.53111267e-01 1.87868595e-01 -6.06497340e-02
-5.30549765e-01 -4.79350597e-01 -7.89557219e-01 -3.79260719e-01
3.73837948e-01 -4.40835446e-01 1.44217277e+00 -6.19603992e-01
-1.57268083e+00 8.21085632e-01 -5.24612963e-01 -5.89551628e-01
6.89530790e-01 -6.19113207e-01 -5.99334300e-01 1.56410351e-01
4.26350012e-02 -3.14409807e-02 5.42584062e-01 -4.13801789e-01
-6.05223298e-01 -4.62661088e-01 -4.81551677e-01 4.76774983e-02
-2.58390814e-01 -3.78067680e-02 -4.62557852e-01 -5.35822213e-01
-8.38581771e-02 -1.23240304e+00 -5.14265262e-02 -7.35240579e-01
-4.84814852e-01 -2.97956675e-01 6.49594843e-01 -8.51789594e-01
1.79145133e+00 -1.99094093e+00 8.39348696e-03 3.12317163e-01
1.46703497e-01 3.21678966e-01 4.14537579e-01 5.45273125e-01
2.49125555e-01 -1.87058628e-01 -3.22787315e-01 -5.05018592e-01
-7.21634477e-02 3.08920711e-01 2.99876153e-01 6.90162301e-01
-3.53421897e-01 9.99658942e-01 -8.87639582e-01 -4.45775002e-01
5.26249111e-01 4.11747038e-01 -2.06937101e-02 1.53101280e-01
3.11278135e-01 8.21999311e-01 -3.32967103e-01 6.60068750e-01
3.53425294e-01 -3.42501491e-01 3.44089001e-01 4.34610844e-02
-2.69053638e-01 3.40673983e-01 -1.39351785e+00 1.94525647e+00
-1.12715520e-01 5.78777075e-01 -4.14864749e-01 -8.83487403e-01
7.06559122e-01 3.99851948e-01 1.20515013e+00 -4.73624021e-01
1.16722517e-01 2.44030952e-01 -4.70061600e-01 -6.84847891e-01
3.95879388e-01 6.57328486e-01 -4.12793607e-01 3.76054943e-01
-9.19069722e-02 4.90663856e-01 6.32743239e-01 -1.23355098e-01
1.61360788e+00 2.08258554e-01 7.91462183e-01 4.38602827e-02
6.93847358e-01 -2.23452877e-02 6.37356997e-01 7.04512239e-01
-5.18797159e-01 7.60556877e-01 2.49759451e-01 -4.03383702e-01
-6.52968764e-01 -1.03740871e+00 2.15233147e-01 9.42181945e-01
1.55073088e-02 -1.03110492e+00 -1.08519065e+00 -3.89957428e-01
-1.97826937e-01 2.61649728e-01 -5.73560596e-01 1.95922673e-01
-7.03951299e-01 -7.57466674e-01 1.01404464e+00 6.42604232e-01
8.37953150e-01 -9.50223684e-01 -1.26037467e+00 3.34004998e-01
-5.47919214e-01 -1.27217185e+00 -6.13352418e-01 -1.13045655e-01
-9.27489996e-01 -1.21625006e+00 -8.59075665e-01 -3.80909413e-01
2.15397328e-01 3.11098218e-01 9.14658308e-01 -2.31027871e-01
-2.08016768e-01 6.19726062e-01 -4.91700590e-01 -1.98118791e-01
1.83491975e-01 3.54160458e-01 1.56502470e-01 3.96307319e-01
7.59424269e-01 -6.07873857e-01 -9.46803689e-01 6.34537339e-01
-4.79541481e-01 -1.63374960e-01 4.11786437e-01 8.72252136e-02
7.62850702e-01 -4.11302865e-01 1.06119484e-01 -3.76824826e-01
6.90911353e-01 -6.79731965e-01 -2.10293636e-01 8.51244181e-02
-4.71306145e-01 -4.31472301e-01 2.88669765e-01 -5.42520761e-01
-7.04293191e-01 3.80610198e-01 -1.80354282e-01 -1.43867403e-01
-6.89069211e-01 3.29687476e-01 -8.95845741e-02 3.43908191e-01
7.85377443e-01 3.12300384e-01 -2.13030264e-01 -7.18043625e-01
-2.97381461e-01 8.30197692e-01 5.59832692e-01 -2.90358573e-01
1.73482925e-01 5.40107369e-01 -4.52884510e-02 -1.35017598e+00
-3.50382477e-01 -1.05152833e+00 -7.94435740e-01 -4.04472381e-01
9.42394972e-01 -7.49941826e-01 -8.73069644e-01 9.14152026e-01
-7.59197474e-01 -5.49129844e-01 -3.47790211e-01 6.32299662e-01
-6.59371376e-01 4.53042209e-01 -4.92666066e-01 -9.72658694e-01
-3.65261137e-01 -5.98442972e-01 1.32426095e+00 2.63096869e-01
-1.11004460e+00 -6.66571677e-01 5.34683764e-01 5.66604972e-01
3.59604061e-02 7.38833308e-01 -2.71665215e-01 -6.63819909e-01
9.12127122e-02 -3.05765778e-01 5.93207479e-01 8.38918611e-02
3.63214284e-01 -9.64481011e-02 -6.62564218e-01 -1.14843160e-01
-1.86205164e-01 1.35546774e-01 3.75102699e-01 6.68874443e-01
9.98505235e-01 -1.62167475e-01 -4.99298364e-01 4.18185443e-01
8.41893256e-01 4.07062829e-01 9.73459482e-01 3.71870816e-01
7.01770663e-01 2.82323658e-01 7.87713766e-01 6.97226405e-01
3.33668560e-01 9.43111658e-01 -1.83960423e-01 2.57639825e-01
2.13755190e-01 -1.33877650e-01 8.21248889e-01 6.78218961e-01
-7.87769854e-01 -7.98401088e-02 -1.17481303e+00 5.30985236e-01
-2.14586759e+00 -1.24529243e+00 -6.48847282e-01 2.55103803e+00
3.39200974e-01 3.89289632e-02 1.04611754e+00 8.39191556e-01
5.26319623e-01 1.96269408e-01 -3.13326031e-01 -1.44601017e-01
4.11893101e-03 1.11145072e-01 4.39787507e-01 1.84267491e-01
-1.22182202e+00 3.52127373e-01 5.69320393e+00 5.72344840e-01
-7.12413192e-01 3.24243814e-01 2.91410238e-01 -4.42406684e-01
5.01904488e-01 -4.16089267e-01 -6.71527267e-01 8.48086417e-01
1.56892645e+00 1.20420136e-01 1.81413665e-01 6.55503690e-01
7.97778487e-01 -6.55854821e-01 -8.47104013e-01 1.43033516e+00
-1.73604786e-01 -1.07079840e+00 -5.51848471e-01 2.89562106e-01
5.43829322e-01 1.21324696e-01 -4.81258214e-01 1.08951330e-01
-4.36075658e-01 -7.60324240e-01 6.79344296e-01 6.98579550e-01
5.08984506e-01 -5.15552282e-01 5.35203278e-01 4.32396561e-01
-1.66290390e+00 -1.12023775e-03 3.19673508e-01 -3.29008698e-01
3.61210823e-01 7.12785721e-01 -7.53150284e-01 5.13046384e-01
9.72367346e-01 1.18912542e+00 -7.80768752e-01 1.15843689e+00
1.89906582e-02 9.75952029e-01 -6.06150568e-01 -1.09587915e-01
-8.53090212e-02 -3.90007079e-01 5.96221685e-01 1.60112643e+00
6.56478345e-01 2.90204003e-03 5.33069782e-02 1.16221189e-01
2.55166978e-01 -2.39286330e-02 -6.14103496e-01 -1.32004738e-01
3.93920004e-01 1.09672952e+00 -8.87474477e-01 -4.38558072e-01
-3.73406202e-01 1.07609367e+00 -3.38659346e-01 1.95903048e-01
-1.05234706e+00 -3.40533853e-01 7.19436347e-01 5.31841874e-01
8.98044929e-02 -8.00011039e-01 -2.85421729e-01 -1.25709140e+00
3.15359592e-01 -8.67355466e-01 7.45024741e-01 -5.17949224e-01
-6.37760043e-01 1.10174946e-01 1.94789588e-01 -1.58779824e+00
-5.06963968e-01 -2.76067853e-01 -4.63555276e-01 2.65604079e-01
-4.19623733e-01 -7.61627495e-01 -8.02175939e-01 1.06327128e+00
8.38171601e-01 3.21254045e-01 7.03559041e-01 5.12537420e-01
-8.72062802e-01 2.61089593e-01 -5.07997423e-02 1.72876909e-01
4.43924099e-01 -1.11750460e+00 6.26619339e-01 8.64535749e-01
3.44460487e-01 6.51419759e-01 6.00999236e-01 -7.15617418e-01
-1.38203037e+00 -8.20347905e-01 1.03159976e+00 -7.61224329e-01
3.68006110e-01 -4.38271523e-01 -7.77332604e-01 8.89428496e-01
-1.41219245e-02 1.78780803e-03 1.08271348e+00 -1.28615797e-02
2.37046614e-01 -9.33631733e-02 -9.99020576e-01 3.87816638e-01
1.48241365e+00 -2.81974286e-01 -6.58321083e-01 -3.26964855e-02
-9.17109773e-02 -4.58163589e-01 -1.23423529e+00 3.02446455e-01
1.00567508e+00 -1.13419318e+00 1.04099488e+00 -8.76669139e-02
-2.92237014e-01 -4.72116321e-01 7.36273453e-02 -9.07737494e-01
-2.95746233e-02 -1.02877021e+00 -8.66901815e-01 9.04087424e-01
3.10442857e-02 -7.23122716e-01 8.54760230e-01 3.15326899e-01
1.59468688e-02 -4.63331014e-01 -1.01966047e+00 -9.36258912e-01
-8.67888749e-01 -8.97313952e-01 6.54132962e-01 7.88229704e-01
3.20900112e-01 2.06726998e-01 -4.83130246e-01 -2.42985561e-01
5.20625830e-01 -2.58660734e-01 1.01489246e+00 -1.21425641e+00
-1.29020959e-01 -1.30170509e-01 -6.13862216e-01 -1.06190693e+00
-4.52074856e-01 -2.95889437e-01 -2.95310706e-01 -1.39041889e+00
-1.90271601e-01 2.04962522e-01 -2.74945140e-01 5.32989621e-01
2.24281698e-01 5.01430452e-01 -8.37926567e-02 3.42455238e-01
-1.15985692e+00 9.77929309e-02 7.09459245e-01 1.24603510e-01
-7.21360505e-01 4.82451081e-01 -1.34955704e-01 8.25994194e-01
1.14885843e+00 -2.98087627e-01 -3.04656535e-01 6.83579892e-02
3.40043567e-02 4.03886801e-03 4.03450668e-01 -1.73036075e+00
1.70867831e-01 -1.18464656e-01 5.59570670e-01 -6.87279105e-01
3.77564281e-01 -6.57036543e-01 7.89517045e-01 7.19966471e-01
-3.03345937e-02 1.26239598e-01 -8.15579444e-02 4.48336154e-01
-6.66892827e-02 4.00981545e-01 2.45344460e-01 6.23281002e-02
-8.40886474e-01 1.68545216e-01 -9.18065488e-01 4.64023910e-02
1.10496759e+00 -8.05927396e-01 -7.73117179e-03 -2.63604522e-01
-1.06707954e+00 -1.18225804e-02 2.48884127e-01 6.74926877e-01
4.06271815e-01 -1.41879344e+00 -3.19542944e-01 2.48351261e-01
3.33784401e-01 -1.20676570e-01 1.49283648e-01 1.60723674e+00
-4.83973324e-01 5.68917632e-01 -2.83775359e-01 -8.39065552e-01
-1.52210021e+00 -4.30553220e-03 2.27827817e-01 -3.14096898e-01
-6.40910685e-01 1.47276133e-01 -7.42049754e-01 4.10423093e-02
9.64629129e-02 -6.88300192e-01 -1.78529352e-01 2.93690801e-01
8.10592532e-01 1.33884227e+00 3.32093835e-01 -8.70333791e-01
-8.51611972e-01 6.45760536e-01 5.77718139e-01 -1.71865240e-01
1.03962374e+00 -3.08942676e-01 3.53230268e-01 7.84373164e-01
1.02074587e+00 9.14193317e-02 -1.16356587e+00 2.49715358e-01
5.77207923e-01 -4.94797647e-01 -5.08043289e-01 -3.03381443e-01
-6.43053472e-01 5.18307626e-01 8.44224632e-01 4.80424523e-01
1.10337007e+00 -1.43687546e-01 1.13502586e+00 2.59217590e-01
6.85615599e-01 -1.20502710e+00 2.52148602e-02 4.25236911e-01
4.60298181e-01 -8.52962434e-01 -1.98634699e-01 -8.46075565e-02
-7.77718663e-01 7.41538703e-01 1.90089092e-01 1.29917497e-03
3.80088866e-01 5.88220395e-02 -2.42218286e-01 -8.60021263e-02
-2.74705350e-01 -3.86832178e-01 2.55988926e-01 6.16369247e-01
4.38594371e-01 2.13401854e-01 -6.77120626e-01 6.34733558e-01
-1.27657071e-01 5.00340581e-01 2.76314974e-01 1.26548564e+00
-2.27165833e-01 -9.05656517e-01 -6.06176496e-01 4.59568620e-01
-5.37884831e-01 6.02345288e-01 -4.67190236e-01 5.87796926e-01
2.64701128e-01 1.31164873e+00 -1.76592742e-03 -6.22236311e-01
7.56166339e-01 2.37759367e-01 4.94475722e-01 -2.76334912e-01
-7.36217380e-01 -5.06071299e-02 3.91424477e-01 -1.17646527e+00
-6.14448428e-01 -1.37264550e+00 -1.26517105e+00 -3.62900525e-01
4.85064834e-01 1.66431367e-01 4.70288873e-01 8.04751158e-01
6.54594362e-01 2.30476603e-01 3.23354274e-01 -1.16867125e+00
8.11512098e-02 -9.53799009e-01 -5.80517709e-01 7.24007666e-01
3.56302857e-01 -5.84950209e-01 -1.83239609e-01 4.11138475e-01]
|
[7.402915954589844, 0.6345326900482178]
|
9efcac34-10fc-480f-9885-65c4fe15081d
|
discovering-bayesian-market-views-for
|
1802.09911
| null |
http://arxiv.org/abs/1802.09911v2
|
http://arxiv.org/pdf/1802.09911v2.pdf
|
Discovering Bayesian Market Views for Intelligent Asset Allocation
|
Along with the advance of opinion mining techniques, public mood has been
found to be a key element for stock market prediction. However, how market
participants' behavior is affected by public mood has been rarely discussed.
Consequently, there has been little progress in leveraging public mood for the
asset allocation problem, which is preferred in a trusted and interpretable
way. In order to address the issue of incorporating public mood analyzed from
social media, we propose to formalize public mood into market views, because
market views can be integrated into the modern portfolio theory. In our
framework, the optimal market views will maximize returns in each period with a
Bayesian asset allocation model. We train two neural models to generate the
market views, and benchmark the model performance on other popular asset
allocation strategies. Our experimental results suggest that the formalization
of market views significantly increases the profitability (5% to 10% annually)
of the simulated portfolio at a given risk level.
|
['Carlo Vercellis', 'Lorenzo Malandri', 'Frank Z. Xing', 'Erik Cambria']
|
2018-02-27
| null | null | null | null |
['stock-market-prediction']
|
['time-series']
|
[-5.61113775e-01 1.06782056e-01 -4.17606443e-01 -3.46762091e-01
-1.66979089e-01 -4.81485695e-01 5.04722714e-01 4.49950993e-02
-9.12758335e-02 5.39890051e-01 1.75052494e-01 -6.19950235e-01
4.85746451e-02 -1.43632829e+00 -3.04463148e-01 -3.44664395e-01
1.71043724e-01 2.65224427e-01 1.75294474e-01 -3.69276553e-01
4.48925525e-01 -6.12392984e-02 -1.37942767e+00 2.77296841e-01
6.40820265e-01 1.47683406e+00 -3.23323995e-01 4.34272066e-02
-2.02604368e-01 9.29784179e-01 -5.55604875e-01 -1.12824762e+00
7.98030078e-01 -2.77920306e-01 -2.12664112e-01 -1.58584535e-01
-3.80431324e-01 -5.55614054e-01 2.34886497e-01 1.08434784e+00
3.21780473e-01 -3.27188134e-01 5.89111269e-01 -1.23450661e+00
-6.51598334e-01 9.94648993e-01 -8.52599919e-01 2.51422077e-01
-3.14613059e-02 -8.06796551e-02 1.66191471e+00 -6.33669198e-01
2.54050106e-01 9.90189254e-01 3.57779473e-01 4.55841273e-02
-8.21593106e-01 -1.03449047e+00 5.58165371e-01 6.25301078e-02
-9.69561338e-01 2.01903991e-02 9.40770030e-01 -3.28823060e-01
6.41664863e-01 1.94726825e-01 1.21771801e+00 6.12130284e-01
7.03279376e-01 8.90845776e-01 1.16518223e+00 -2.89394408e-01
4.98078853e-01 4.86751527e-01 2.36800492e-01 1.10966869e-01
8.59729886e-01 1.43154472e-01 -7.75187433e-01 -4.30301249e-01
4.66667205e-01 3.20985019e-01 -7.28963315e-02 -2.55670726e-01
-7.38522708e-01 1.21524584e+00 4.57538553e-02 -2.58704066e-01
-6.63775921e-01 4.23428081e-02 6.30552247e-02 6.95510447e-01
1.12821376e+00 2.84260511e-01 -5.68136275e-01 -4.33010422e-02
-6.60434246e-01 3.84219587e-01 1.16214824e+00 5.47268927e-01
5.19208610e-01 1.13784343e-01 2.12522984e-01 2.38226742e-01
9.53693926e-01 8.27410519e-01 2.95264423e-01 -8.60338986e-01
4.22033221e-01 9.01064813e-01 3.45701545e-01 -1.27157247e+00
7.26071279e-03 -5.65382361e-01 -4.54544157e-01 3.96989226e-01
1.47515073e-01 -4.43173468e-01 -1.50140971e-01 1.41132700e+00
2.41318747e-01 -3.45366001e-02 1.29032224e-01 5.04929900e-01
1.93167329e-01 6.25686705e-01 -3.83033603e-01 -5.46586215e-01
1.12989831e+00 -6.57031894e-01 -8.12230051e-01 -1.98611990e-01
2.50923634e-01 -5.76815188e-01 4.36434329e-01 6.48987591e-01
-1.16112113e+00 4.61743250e-02 -1.11884964e+00 9.29192066e-01
-1.82562605e-01 -5.94468117e-01 9.07068491e-01 1.11059785e+00
-7.63974428e-01 4.51925188e-01 -6.36599839e-01 2.90120929e-01
5.81280887e-01 1.96395084e-01 2.04107106e-01 7.21143603e-01
-1.47182953e+00 9.04580116e-01 1.93656549e-01 3.33407968e-02
-5.27578831e-01 -5.12833834e-01 -2.69188851e-01 9.56056938e-02
7.36227930e-01 -7.09435344e-01 1.51208723e+00 -1.25906599e+00
-1.67950273e+00 3.80136818e-01 2.79500782e-01 -9.39135373e-01
6.28249109e-01 -3.51612329e-01 -3.89283627e-01 -4.55153473e-02
-2.02394783e-01 -6.34894073e-02 9.60076094e-01 -9.90806043e-01
-7.97404468e-01 -2.60384679e-01 3.00505549e-01 1.12248488e-01
-6.35663390e-01 1.92881003e-01 1.34520933e-01 -7.74262786e-01
8.06850940e-03 -8.51046979e-01 -3.64096820e-01 -3.02092433e-01
-3.77174746e-03 -8.11453909e-02 3.04445535e-01 -3.73361677e-01
1.42504621e+00 -1.71558356e+00 -2.69604445e-01 6.86765134e-01
1.81513369e-01 -1.76133543e-01 6.12675905e-01 4.92869079e-01
-9.71225724e-02 4.97720003e-01 9.67895612e-02 5.33514842e-02
4.57952738e-01 -2.15490416e-01 -9.99722719e-01 2.98292160e-01
-2.37159356e-01 8.19415748e-01 -4.44492847e-01 7.69616291e-02
-2.10615560e-01 -1.48964018e-01 -8.76633763e-01 2.53168166e-01
-3.27775627e-01 -2.95255810e-01 -7.95696914e-01 6.71983182e-01
5.15826523e-01 -7.31034935e-01 3.20405245e-01 2.16265425e-01
1.10138929e-03 1.93712607e-01 -1.39139318e+00 5.97066283e-01
-9.14060026e-02 1.85885295e-01 -1.18018933e-01 -6.50533855e-01
1.04141641e+00 3.71281058e-01 4.55749720e-01 -5.44811010e-01
1.68220162e-01 2.37192139e-01 2.04858541e-01 1.42600179e-01
4.11328524e-01 -5.11698127e-01 -1.39511400e-03 1.47318745e+00
-6.84171021e-01 1.63209409e-01 -5.06487340e-02 2.18431316e-02
6.45944715e-01 -1.50759175e-01 6.57863259e-01 -2.87664026e-01
1.51732475e-01 -3.83384734e-01 6.91037595e-01 5.73387742e-01
-1.99739598e-02 2.53584623e-01 7.53517926e-01 -6.43046200e-01
-5.30890703e-01 -7.34205127e-01 1.70673564e-01 8.61000597e-01
-4.49886844e-02 -4.20667589e-01 -5.81960976e-01 -8.27743828e-01
2.52633780e-01 8.23360085e-01 -3.88906807e-01 -1.27977869e-02
-7.63649046e-02 -1.09176219e+00 -1.46967337e-01 5.23523390e-01
5.43466926e-01 -9.70085859e-01 -1.01138270e+00 1.44956544e-01
2.49342442e-01 -4.62193727e-01 -2.66055614e-01 4.59244475e-02
-5.64128041e-01 -1.17035890e+00 -7.46801019e-01 2.41712704e-01
4.78699028e-01 2.56945312e-01 1.16797650e+00 -2.03625523e-02
5.52156687e-01 4.02676582e-01 -4.64632660e-01 -1.19954216e+00
-2.21157029e-01 5.37118763e-02 2.37752736e-01 3.41538996e-01
5.25016725e-01 -6.24643207e-01 -8.64889383e-01 4.80555356e-01
-7.61567414e-01 1.00058533e-01 3.45535696e-01 4.45532203e-01
1.40824869e-01 1.40502334e-01 1.13732636e+00 -1.13518941e+00
8.79956722e-01 -6.57618821e-01 -1.07077730e+00 2.40187883e-01
-1.30513072e+00 -1.73991367e-01 7.76360556e-02 -1.20737135e-01
-1.21703231e+00 -5.63401759e-01 2.28300348e-01 -5.54428510e-02
5.59912145e-01 1.16628623e+00 -1.73299700e-01 2.29298085e-01
1.26790345e-01 -6.69735819e-02 1.97094187e-01 -1.28177494e-01
2.43796319e-01 6.34522557e-01 -4.42619532e-01 -4.94462967e-01
8.85723233e-01 6.01344228e-01 -3.53855610e-01 -1.17554918e-01
-1.06894159e+00 1.34965599e-01 -3.59516591e-02 -2.42231831e-01
5.38520575e-01 -9.26680267e-01 -8.11079383e-01 5.40279210e-01
-8.33500147e-01 6.20493293e-02 -1.89309433e-01 4.84155148e-01
-5.17881453e-01 2.40493760e-01 -3.66781384e-01 -1.34531736e+00
-5.62492847e-01 -9.83153880e-01 2.58228481e-01 2.94850111e-01
-4.34434652e-01 -9.99541938e-01 1.86226591e-01 4.61005479e-01
6.38273895e-01 1.63534824e-02 8.47085476e-01 -1.23133373e+00
-1.04394305e+00 -4.81093019e-01 1.77668512e-01 4.57613528e-01
1.94944739e-01 3.75718740e-03 -1.07801628e+00 -4.47623730e-02
5.34426093e-01 1.31377727e-02 6.03284299e-01 3.35394591e-01
7.37839162e-01 -3.88420522e-01 -4.38105920e-03 1.64070636e-01
9.18346524e-01 6.38591230e-01 5.26015282e-01 7.60232687e-01
-5.36381304e-02 8.70106757e-01 7.46507287e-01 9.38843369e-01
3.94559532e-01 1.77483454e-01 4.69941527e-01 3.46204489e-01
9.24767613e-01 -1.42229170e-01 6.94864213e-01 9.25816655e-01
-1.83284134e-01 -3.64554107e-01 -8.64252985e-01 5.72381355e-02
-2.07175899e+00 -1.15661478e+00 4.72090364e-01 2.17062807e+00
6.44880891e-01 7.11696863e-01 3.10069740e-01 -3.48661765e-02
5.95647633e-01 5.09605944e-01 -6.17824554e-01 -6.97131595e-03
-1.20258801e-01 -3.75928800e-03 5.28834939e-01 7.42752850e-02
-8.53288352e-01 4.60296631e-01 7.06714916e+00 3.76497865e-01
-9.59191442e-01 -1.88868806e-01 1.14815116e+00 -2.52314329e-01
-8.58003855e-01 2.77008802e-01 -9.88821447e-01 5.94162524e-01
9.53658342e-01 -1.01937377e+00 -2.33714767e-02 1.02495944e+00
1.70486823e-01 3.48497778e-02 -1.09916258e+00 5.39369166e-01
6.54390408e-03 -1.58119178e+00 1.09144956e-01 5.99766552e-01
6.92443013e-01 -3.16180259e-01 3.26797068e-01 2.26740301e-01
4.21779215e-01 -7.93129325e-01 9.58309233e-01 7.46701419e-01
-1.58776268e-01 -1.09409821e+00 1.06529844e+00 4.59704638e-01
-9.18588579e-01 -1.39361784e-01 -2.10044414e-01 -3.92184019e-01
4.11794901e-01 6.00892484e-01 -4.43756104e-01 5.36657751e-01
6.81122720e-01 7.32913494e-01 -2.81630278e-01 7.06210077e-01
-1.46863595e-01 7.70157099e-01 -3.25622588e-01 -2.81002671e-01
-2.73735840e-02 -4.38084930e-01 2.60956556e-01 2.26001412e-01
5.78221202e-01 1.30378261e-01 1.47964045e-01 8.66565228e-01
-1.07827425e-01 2.75234312e-01 -6.74433589e-01 -2.99967974e-01
2.40233228e-01 8.98842454e-01 -8.62051725e-01 -1.50772274e-01
-7.86107600e-01 2.57683158e-01 -4.29145783e-01 1.85265347e-01
-7.44616151e-01 -5.04852645e-02 5.18762648e-01 4.57063884e-01
4.80201483e-01 3.46863240e-01 -3.23688060e-01 -1.15523756e+00
1.82563946e-01 -1.12684441e+00 4.55611378e-01 -4.60188359e-01
-1.67278922e+00 3.11466366e-01 9.77777094e-02 -1.27268147e+00
-4.22560990e-01 -6.79972410e-01 -9.56626177e-01 8.33551228e-01
-1.52891648e+00 -6.61807597e-01 4.85822499e-01 1.74565725e-02
1.83155939e-01 -7.97738314e-01 6.59215569e-01 -1.34324387e-01
-5.02065480e-01 3.51524919e-01 -1.19518712e-02 -1.36746485e-02
5.33698976e-01 -1.16335499e+00 5.04397571e-01 5.47322750e-01
1.20706938e-01 7.92847991e-01 5.76509714e-01 -9.18097854e-01
-1.17516482e+00 -6.34930849e-01 3.87793422e-01 -6.34049356e-01
1.22131312e+00 3.77904773e-02 -7.19884217e-01 7.20103323e-01
6.51783586e-01 -3.49718750e-01 1.39052474e+00 2.80257493e-01
-1.72203735e-01 -4.38138574e-01 -8.43179047e-01 6.59617841e-01
4.18083310e-01 -2.67457455e-01 -7.70917118e-01 -5.63932136e-02
7.71520436e-01 8.81998837e-02 -7.96693385e-01 2.60869861e-01
6.16309345e-01 -1.40732443e+00 6.75321102e-01 -3.56250852e-01
2.63905019e-01 -1.96039766e-01 -2.65854329e-01 -1.22922206e+00
1.42535374e-01 -8.67480099e-01 -3.42266232e-01 1.13303041e+00
7.49563277e-01 -1.36975658e+00 1.08439136e+00 1.01840734e+00
5.44802308e-01 -8.51846039e-01 -8.15819323e-01 -3.66273731e-01
1.82533279e-01 -4.89983618e-01 9.50912178e-01 6.16392016e-01
1.23458207e-01 1.45068362e-01 -5.00997186e-01 -1.72608465e-01
5.40926516e-01 7.74293244e-01 8.61102283e-01 -1.51579452e+00
-5.94137192e-01 -5.29043615e-01 -4.20598164e-02 -7.66322017e-01
2.34924853e-01 -5.74861109e-01 -5.91604114e-01 -1.08390701e+00
4.15419638e-01 -1.58334315e-01 -8.10761333e-01 8.52109194e-02
-5.70113026e-02 -2.06823558e-01 4.26394999e-01 1.99515641e-01
-5.52028894e-01 6.54930055e-01 9.21240151e-01 -1.25465482e-01
3.54477279e-02 4.64967251e-01 -1.41456020e+00 1.17405951e+00
8.11318398e-01 -2.94336557e-01 -5.65188468e-01 1.44830033e-01
1.32524824e+00 1.33962214e-01 -1.34765536e-01 -3.37842882e-01
2.19273180e-01 -5.99181414e-01 1.40915155e-01 -9.94157434e-01
7.52702504e-02 -8.67344558e-01 3.45669955e-01 4.15266246e-01
-1.64990485e-01 5.42233586e-01 -2.71399677e-01 8.04297388e-01
-3.35390389e-01 -3.18329334e-01 2.33108476e-01 -2.97087699e-01
-2.61351198e-01 2.07316130e-01 -5.15143752e-01 -5.55161275e-02
1.27502275e+00 -1.98406368e-01 -3.06688726e-01 -9.48832452e-01
-2.96873808e-01 4.29916501e-01 4.52674091e-01 2.97484994e-01
4.96159673e-01 -1.20713246e+00 -5.86409986e-01 9.94276851e-02
-1.24297991e-01 -2.37517133e-01 1.79487374e-03 5.66894054e-01
-3.51708889e-01 1.97577581e-01 7.69970194e-02 -9.42799598e-02
-9.06074166e-01 3.05421442e-01 3.21545899e-01 -3.42309952e-01
-3.33433598e-01 4.90374714e-01 3.15185606e-01 7.56848312e-04
1.64956078e-01 -2.01009944e-01 -4.86526340e-01 6.23490334e-01
7.95662701e-01 3.15957099e-01 -2.12595791e-01 -2.04438180e-01
-5.42581938e-02 2.38028377e-01 -3.18317056e-01 -3.36333781e-01
1.66665661e+00 -7.95777068e-02 -1.51407212e-01 7.98150063e-01
3.61228019e-01 3.49728167e-01 -1.05472124e+00 -7.59332106e-02
4.20696795e-01 -4.38432634e-01 3.62722576e-02 -5.53135157e-01
-1.38427699e+00 6.03937507e-01 3.64320248e-01 8.35982025e-01
7.68536091e-01 -3.88023078e-01 7.11089432e-01 5.67311227e-01
7.47982264e-01 -1.19938946e+00 3.21183391e-02 2.37473026e-01
6.59373045e-01 -1.14496481e+00 3.42625022e-01 -1.82050586e-01
-8.77460420e-01 9.01515424e-01 4.11628276e-01 -2.00491846e-01
1.32987380e+00 2.37525940e-01 3.61211687e-01 -2.22961798e-01
-1.14524949e+00 3.48050863e-01 1.07687786e-02 -3.00803930e-02
2.69846231e-01 1.62134677e-01 -1.31131366e-01 1.40016580e+00
-3.42874467e-01 -6.26917034e-02 7.03925073e-01 1.14884388e+00
-6.98031366e-01 -1.29229248e+00 -3.17419499e-01 7.20972300e-01
-1.05373180e+00 -1.09617859e-01 -4.60056067e-01 5.53764105e-01
-3.39321047e-01 6.96551621e-01 9.21163484e-02 -3.12300533e-01
2.82589763e-01 1.22331306e-01 -2.17662200e-01 -8.21508408e-01
-7.72884786e-01 4.53078002e-01 5.07282577e-02 -3.83875161e-01
-6.12003922e-01 -7.98123300e-01 -9.46080685e-01 -4.25589323e-01
-6.12299204e-01 4.45659220e-01 1.91493109e-01 8.76620829e-01
2.31628180e-01 1.65770754e-01 9.17716146e-01 -2.68880188e-01
-1.17487943e+00 -5.89893758e-01 -8.50145221e-01 -7.40547478e-02
-1.72172695e-01 -7.21433938e-01 -6.19763196e-01 -2.74001181e-01]
|
[4.535121917724609, 4.169627666473389]
|
8cd09695-4fb5-4d6a-bb41-cee849395e40
|
viplo-vision-transformer-based-pose
|
2304.08114
| null |
https://arxiv.org/abs/2304.08114v1
|
https://arxiv.org/pdf/2304.08114v1.pdf
|
ViPLO: Vision Transformer based Pose-Conditioned Self-Loop Graph for Human-Object Interaction Detection
|
Human-Object Interaction (HOI) detection, which localizes and infers relationships between human and objects, plays an important role in scene understanding. Although two-stage HOI detectors have advantages of high efficiency in training and inference, they suffer from lower performance than one-stage methods due to the old backbone networks and the lack of considerations for the HOI perception process of humans in the interaction classifiers. In this paper, we propose Vision Transformer based Pose-Conditioned Self-Loop Graph (ViPLO) to resolve these problems. First, we propose a novel feature extraction method suitable for the Vision Transformer backbone, called masking with overlapped area (MOA) module. The MOA module utilizes the overlapped area between each patch and the given region in the attention function, which addresses the quantization problem when using the Vision Transformer backbone. In addition, we design a graph with a pose-conditioned self-loop structure, which updates the human node encoding with local features of human joints. This allows the classifier to focus on specific human joints to effectively identify the type of interaction, which is motivated by the human perception process for HOI. As a result, ViPLO achieves the state-of-the-art results on two public benchmarks, especially obtaining a +2.07 mAP performance gain on the HICO-DET dataset. The source codes are available at https://github.com/Jeeseung-Park/ViPLO.
|
['Jong-Seok Lee', 'Jin-Woo Park', 'Jeeseung Park']
|
2023-04-17
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Park_ViPLO_Vision_Transformer_Based_Pose-Conditioned_Self-Loop_Graph_for_Human-Object_Interaction_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Park_ViPLO_Vision_Transformer_Based_Pose-Conditioned_Self-Loop_Graph_for_Human-Object_Interaction_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['human-object-interaction-detection']
|
['computer-vision']
|
[-9.79040377e-03 -3.21832821e-02 -2.53981445e-02 -1.89590424e-01
-2.14965701e-01 4.10729684e-02 3.92176121e-01 -2.80087471e-01
-3.81604999e-01 2.45481193e-01 2.57982194e-01 3.44840705e-01
2.86856052e-02 -5.71759701e-01 -8.20468068e-01 -7.12264538e-01
3.60790454e-02 3.80605757e-01 6.00768387e-01 -1.09054707e-01
1.53244389e-02 2.69597590e-01 -1.72542334e+00 3.80179971e-01
6.52903676e-01 1.27865529e+00 5.07305801e-01 3.04354399e-01
3.33540440e-01 8.14671457e-01 -5.85460782e-01 -8.03145096e-02
1.87508296e-02 -5.17213464e-01 -5.82930148e-01 8.27735066e-02
4.25324261e-01 -2.11012214e-01 -6.11951351e-01 1.00302649e+00
6.70881569e-01 1.78155154e-01 6.71035826e-01 -1.67629683e+00
-4.38850105e-01 3.09909880e-01 -7.52532363e-01 2.01202080e-01
4.44325715e-01 2.97300637e-01 9.72402692e-01 -1.17074394e+00
5.79810679e-01 1.54610884e+00 4.87039417e-01 4.11805630e-01
-7.60314524e-01 -7.54033864e-01 2.45457813e-02 7.38073349e-01
-1.73877656e+00 -1.91013187e-01 8.26000273e-01 -4.74284112e-01
1.08508909e+00 1.47865355e-01 9.57190871e-01 1.00661325e+00
4.74023551e-01 1.27677894e+00 6.92870557e-01 -2.48171493e-01
-9.84601453e-02 -5.50250076e-02 2.76536673e-01 1.03970301e+00
1.78428650e-01 1.49707943e-01 -8.20022345e-01 3.02644551e-01
9.18211520e-01 8.20737612e-03 -4.28597897e-01 -4.94770437e-01
-1.36144412e+00 6.62696660e-01 1.00830746e+00 7.76452944e-02
-4.26121622e-01 2.50447601e-01 2.46462137e-01 -1.95362687e-01
-2.50380561e-02 1.39425606e-01 2.07299720e-02 4.40984040e-01
-4.61441159e-01 2.02740535e-01 4.95158494e-01 1.01022208e+00
6.79395199e-01 -3.33407313e-01 -6.19374931e-01 6.76399350e-01
6.26904905e-01 4.13856387e-01 2.70422906e-01 -9.01485085e-01
4.50491786e-01 7.82870531e-01 -1.67585164e-01 -1.09235394e+00
-6.43934190e-01 -6.80068314e-01 -1.02790666e+00 2.54353672e-01
3.69330257e-01 1.96875915e-01 -9.44717884e-01 1.67048693e+00
6.17807746e-01 2.44803224e-02 -2.33782798e-01 1.29752374e+00
1.03164268e+00 5.72099209e-01 -1.68540236e-02 2.45973960e-01
1.88521659e+00 -1.46905422e+00 -6.80478573e-01 -5.16312242e-01
4.54218060e-01 -4.74898964e-01 1.00757229e+00 2.99263477e-01
-7.60001779e-01 -1.04725277e+00 -1.12139237e+00 -4.16749895e-01
-3.95602733e-01 4.28518802e-01 5.63790262e-01 4.69047613e-02
-7.63836384e-01 6.71369433e-02 -6.36343360e-01 -5.80915928e-01
3.30078572e-01 2.90510148e-01 -4.21132982e-01 -8.64644423e-02
-1.38952804e+00 8.85357201e-01 3.51724327e-01 5.86295545e-01
-8.80065799e-01 -2.94487685e-01 -1.03169096e+00 1.76755086e-01
6.80077851e-01 -9.62833583e-01 9.56469238e-01 -5.56936324e-01
-1.10071743e+00 7.14318395e-01 -1.73469171e-01 -4.16262925e-01
4.27191198e-01 -4.39381748e-01 -1.50199771e-01 2.54916489e-01
2.31125474e-01 9.87623692e-01 1.03608227e+00 -1.16149151e+00
-9.00712788e-01 -5.49084067e-01 4.60386761e-02 4.05815691e-01
-5.48506565e-02 -7.61122853e-02 -1.02267671e+00 -5.30101597e-01
1.34265885e-01 -1.06899691e+00 1.48068257e-02 3.13604295e-01
-6.33307278e-01 -6.10458612e-01 8.02774429e-01 -6.82224929e-01
1.28540218e+00 -2.28721642e+00 3.73362392e-01 8.72748941e-02
4.48977441e-01 1.53904885e-01 -2.94939615e-02 2.23413974e-01
1.03396006e-01 -3.90446246e-01 -1.48386642e-01 -3.03670377e-01
-1.12505332e-01 1.66484639e-01 1.38409480e-01 5.05024493e-01
2.52033025e-01 1.02779150e+00 -7.71493495e-01 -8.26248884e-01
3.74609500e-01 5.34082353e-01 -5.72869718e-01 4.88339812e-01
6.73016161e-02 5.78820288e-01 -3.06696057e-01 6.80781126e-01
4.12397921e-01 -3.20430756e-01 -9.97395068e-02 -6.88726902e-01
-1.21256508e-01 -6.85429666e-03 -1.21106970e+00 1.78687608e+00
-4.27031852e-02 4.24115807e-01 1.46131933e-01 -8.41930985e-01
7.62017548e-01 2.24082395e-01 3.44244570e-01 -6.82933986e-01
3.38458180e-01 -5.94622232e-02 2.52774477e-01 -4.70400989e-01
1.68287735e-02 5.03880382e-01 -1.17468804e-01 -3.31927001e-01
2.60556340e-01 2.50178456e-01 2.14054942e-01 1.66661188e-01
9.39346671e-01 1.55560404e-01 3.44817728e-01 -1.84568167e-01
7.03935385e-01 -4.58759628e-02 7.59303570e-01 6.95402086e-01
-4.63676333e-01 6.35329843e-01 2.84421116e-01 -2.94046283e-01
-6.08220160e-01 -1.10754311e+00 8.91865864e-02 9.17449713e-01
5.90128779e-01 -3.48085225e-01 -8.23748410e-01 -6.11643612e-01
-1.33933892e-04 4.60789949e-01 -7.36222208e-01 -4.93827313e-01
-6.46449983e-01 -4.16437149e-01 1.76989153e-01 8.69832873e-01
1.01720583e+00 -1.37441278e+00 -9.17809665e-01 1.09807014e-01
-4.83729094e-01 -1.08349311e+00 -6.65230095e-01 1.19059049e-01
-2.43209898e-01 -1.20398605e+00 -7.31903493e-01 -1.01543057e+00
5.85413337e-01 4.61942226e-01 8.75448763e-01 2.52418332e-02
-6.12556756e-01 3.24494511e-01 -2.37250566e-01 -3.30291212e-01
2.38242075e-01 6.56924918e-02 -1.97695903e-02 7.63806552e-02
4.10822809e-01 -1.16124392e-01 -8.90298545e-01 6.53764784e-01
-5.42693853e-01 2.68555999e-01 8.24619889e-01 1.04108763e+00
5.10021925e-01 1.28708407e-01 1.20881461e-01 -3.73922408e-01
1.44499481e-01 -2.81765759e-01 -3.17245483e-01 3.23841542e-01
-2.65711606e-01 -5.90962283e-02 2.91027874e-01 -3.08621496e-01
-1.01266026e+00 4.17700052e-01 3.83101739e-02 -5.54984987e-01
-2.05844313e-01 2.53937036e-01 -5.65761089e-01 -2.74668653e-02
3.86279732e-01 8.03680718e-03 -1.56058399e-02 -3.72861356e-01
2.26621941e-01 4.97393936e-01 7.43487656e-01 -1.94234103e-01
7.41935015e-01 3.66719514e-01 4.97914404e-02 -9.02594805e-01
-1.05079997e+00 -6.79611087e-01 -8.02920163e-01 -5.57446659e-01
1.40101314e+00 -9.06540513e-01 -9.04198527e-01 6.57644272e-01
-1.27187228e+00 -1.70898512e-01 2.32509281e-02 6.57826543e-01
-4.67316717e-01 4.04203802e-01 -5.88579237e-01 -7.61389434e-01
-3.49977046e-01 -1.20165801e+00 1.18163443e+00 3.87218207e-01
-1.88194826e-01 -3.84807676e-01 -3.21521401e-01 6.21916056e-01
-2.03577094e-02 1.40233114e-01 7.38967419e-01 -2.18340680e-01
-6.82746828e-01 9.93223977e-04 -4.43562210e-01 3.55777830e-01
-1.84657782e-01 -3.22756469e-01 -9.18044031e-01 -3.46341193e-01
-7.91446045e-02 -3.16234499e-01 9.62302268e-01 3.64036560e-01
1.07918310e+00 1.47921106e-04 -5.15614748e-01 5.60852408e-01
1.06413388e+00 2.91387916e-01 7.64693022e-01 1.94412857e-01
1.11015296e+00 8.05523753e-01 8.92133057e-01 3.84804159e-01
5.09335101e-01 9.47539151e-01 5.50028920e-01 -2.54069656e-01
-5.70714653e-01 -3.21617991e-01 4.02246416e-01 6.67425334e-01
-1.52680159e-01 -3.34551513e-01 -9.24498022e-01 4.90309834e-01
-2.19554710e+00 -7.74273634e-01 -3.11915636e-01 2.10320258e+00
4.94502336e-01 3.31549793e-01 1.72292605e-01 1.20638736e-01
8.19626868e-01 -1.61117048e-03 -5.76720476e-01 2.06980616e-01
1.10004254e-01 -7.33151659e-02 3.62663239e-01 4.22458619e-01
-1.24681449e+00 9.47725356e-01 5.04539490e+00 8.79192233e-01
-6.68891013e-01 1.83860362e-01 8.22549313e-02 1.28532439e-01
4.98243302e-01 -1.98615953e-01 -1.15465558e+00 4.33887333e-01
3.09074789e-01 2.74383754e-01 3.32501560e-01 7.64524579e-01
7.49623701e-02 -3.46575499e-01 -1.20122099e+00 1.15819263e+00
1.16042413e-01 -8.35233390e-01 -6.78181276e-02 -1.45060644e-02
8.37371647e-02 -1.10729270e-01 -5.90524450e-02 3.74655247e-01
-2.55754650e-01 -8.49661708e-01 8.61160696e-01 6.51401460e-01
5.79060018e-01 -6.29541218e-01 8.30213606e-01 3.55380386e-01
-1.78015637e+00 -2.20764637e-01 -3.30833137e-01 -1.59975681e-02
5.73408566e-02 4.71766621e-01 -5.43067455e-01 4.94922966e-01
1.07982159e+00 6.15346909e-01 -7.40490139e-01 1.20306313e+00
-5.64583361e-01 4.11265194e-01 -4.06431526e-01 -5.31212911e-02
1.55975521e-01 -2.34702695e-03 5.17410338e-01 1.10640776e+00
1.13217801e-01 -5.63109107e-02 4.84293044e-01 9.09648240e-01
2.56883323e-01 -2.15611309e-01 -5.47166765e-01 2.27880210e-01
4.19763178e-01 1.15708435e+00 -6.64704978e-01 -3.47009301e-01
-5.14107764e-01 1.31796861e+00 4.07272249e-01 3.59111845e-01
-1.13267243e+00 -6.88556671e-01 4.75911766e-01 2.00383559e-01
3.73737484e-01 -3.53869438e-01 -1.19473776e-02 -9.87655818e-01
1.68108001e-01 -7.12181211e-01 4.82797474e-01 -8.52456033e-01
-1.05648601e+00 4.90804136e-01 2.02195451e-01 -1.12960196e+00
-1.39593752e-02 -7.64077008e-01 -4.92691189e-01 6.39259458e-01
-1.15083277e+00 -1.32457054e+00 -8.81209433e-01 8.36375892e-01
6.18413329e-01 1.05446316e-01 4.40032423e-01 4.12288874e-01
-8.47592413e-01 6.58602059e-01 -4.82770115e-01 3.28878552e-01
5.86234927e-01 -9.93223727e-01 3.21680456e-01 7.11146414e-01
1.24366574e-01 4.75847542e-01 4.66342479e-01 -6.69655740e-01
-1.37237072e+00 -9.67734635e-01 7.01877654e-01 -3.44005376e-01
2.99322426e-01 -6.19775832e-01 -8.29958558e-01 4.99174654e-01
6.20848648e-02 4.72935550e-02 2.23992810e-01 -1.31257996e-01
-2.24759132e-01 -1.61814317e-01 -8.67183745e-01 5.49355686e-01
1.47478426e+00 -4.47172910e-01 -5.90286553e-01 3.08405966e-01
6.67024493e-01 -2.76107401e-01 -7.29663670e-01 5.48743784e-01
6.61358178e-01 -8.86278927e-01 1.12994492e+00 -1.02178596e-01
1.72988266e-01 -7.19581187e-01 -1.61259845e-02 -9.33832645e-01
-8.55957568e-01 -1.40499681e-01 -3.68591845e-01 1.04879391e+00
1.02237947e-01 -4.96605426e-01 5.72543859e-01 1.37567848e-01
-1.21915258e-01 -8.77198279e-01 -8.43626797e-01 -8.18515360e-01
-6.28142238e-01 -3.00419014e-02 1.41302153e-01 5.05928814e-01
-2.28675872e-01 8.48870039e-01 -4.82965320e-01 3.78680140e-01
9.29974437e-01 -1.12601407e-01 8.10571492e-01 -1.18740642e+00
-4.46405351e-01 -2.13032305e-01 -6.00773215e-01 -1.33279133e+00
-4.42529209e-02 -8.04672062e-01 3.52206200e-01 -1.71289861e+00
3.35154474e-01 7.38692470e-03 -3.11917752e-01 6.71922326e-01
-1.90776333e-01 3.60255748e-01 3.32371294e-01 3.72707039e-01
-6.59764469e-01 6.91026509e-01 1.38193929e+00 -2.92649329e-01
-4.68320027e-03 -1.81118816e-01 -2.03676447e-01 7.44908392e-01
7.79090405e-01 -2.34553859e-01 -4.22375858e-01 -4.12449628e-01
-3.14959377e-01 -1.46295667e-01 8.79150450e-01 -1.62792659e+00
5.66690266e-01 1.36270091e-01 6.09086692e-01 -1.05849075e+00
5.30839205e-01 -7.23713458e-01 4.77822758e-02 8.33684862e-01
-1.88399881e-01 -1.64677024e-01 -1.46616325e-01 5.67365587e-01
-2.50551105e-01 -3.45552526e-02 8.87513697e-01 -9.53684971e-02
-1.27559757e+00 3.03431928e-01 -2.80319899e-01 -6.58141226e-02
9.92640257e-01 -1.66763335e-01 -1.88050136e-01 -2.08807647e-01
-5.96336782e-01 5.87640285e-01 7.60509297e-02 5.80252409e-01
8.68376076e-01 -1.35454404e+00 -4.80896622e-01 3.47186446e-01
3.21651816e-01 1.99560910e-01 3.18832487e-01 1.01799846e+00
-3.36751610e-01 3.82734686e-01 -4.35442209e-01 -7.96565711e-01
-1.20734251e+00 6.42093420e-01 2.74699301e-01 -1.48515493e-01
-8.39126408e-01 1.01090741e+00 8.33193123e-01 -2.03313693e-01
6.86535895e-01 -2.82156199e-01 -2.68970132e-01 -1.12462416e-01
3.85201365e-01 4.50641006e-01 -2.62621284e-01 -8.18283141e-01
-7.11968184e-01 5.66063821e-01 1.10458478e-01 7.97578469e-02
6.27241254e-01 -7.33681545e-02 -8.53208676e-02 4.28966492e-01
1.30180681e+00 -5.04266560e-01 -1.20699620e+00 -2.64868438e-01
-1.83850333e-01 -3.13503623e-01 3.25878859e-02 -6.95298612e-01
-1.11063933e+00 1.08306313e+00 8.70983899e-01 -2.19517678e-01
9.93226767e-01 2.89270878e-01 1.00237036e+00 3.52037519e-01
5.53406835e-01 -1.11653769e+00 3.15464437e-01 4.99831527e-01
1.23329377e+00 -1.16795444e+00 -1.53440043e-01 -7.70918250e-01
-6.69802725e-01 7.32588708e-01 1.17887247e+00 -1.61171660e-01
6.26950145e-01 -4.61914912e-02 -2.06276700e-01 -4.73933160e-01
-4.50492829e-01 -5.79677820e-01 5.77569366e-01 6.46215856e-01
2.90477157e-01 5.02811186e-02 -3.92445475e-01 6.01829708e-01
-4.78349403e-02 -2.25682318e-01 -1.99352965e-01 7.99467504e-01
-6.68449879e-01 -5.48721433e-01 -4.86501276e-01 2.84595698e-01
2.03190550e-01 1.79894641e-01 -4.28105742e-01 9.34359729e-01
5.35368621e-01 9.88093734e-01 2.92425063e-02 -7.07524657e-01
5.53079724e-01 -1.27275586e-01 5.40162623e-01 -5.24648488e-01
-4.09630001e-01 1.92874029e-01 1.95598807e-02 -1.09746158e+00
-2.31426775e-01 -3.57929349e-01 -1.45934308e+00 1.87509358e-01
-3.57448637e-01 -7.76526183e-02 3.35072994e-01 7.85874844e-01
3.91547382e-01 8.51654530e-01 3.60190541e-01 -1.13789284e+00
-3.68181258e-01 -9.98176932e-01 -6.25007033e-01 4.39069182e-01
2.88424417e-02 -1.27721500e+00 -1.80918798e-01 -2.97827035e-01]
|
[9.535074234008789, 1.332589030265808]
|
20d8844d-a171-4822-a8ad-19c5ed02fd46
|
steering-prototype-with-prompt-tuning-for
|
2303.09447
| null |
https://arxiv.org/abs/2303.09447v1
|
https://arxiv.org/pdf/2303.09447v1.pdf
|
Steering Prototype with Prompt-tuning for Rehearsal-free Continual Learning
|
Prototype, as a representation of class embeddings, has been explored to reduce memory footprint or mitigate forgetting for continual learning scenarios. However, prototype-based methods still suffer from abrupt performance deterioration due to semantic drift and prototype interference. In this study, we propose Contrastive Prototypical Prompt (CPP) and show that task-specific prompt-tuning, when optimized over a contrastive learning objective, can effectively address both obstacles and significantly improve the potency of prototypes. Our experiments demonstrate that CPP excels in four challenging class-incremental learning benchmarks, resulting in 4% to 6% absolute improvements over state-of-the-art methods. Moreover, CPP does not require a rehearsal buffer and it largely bridges the performance gap between continual learning and offline joint-learning, showcasing a promising design scheme for continual learning systems under a Transformer architecture.
|
['Dimitris N. Metaxas', 'Ting Liu', 'Di Liu', 'Han Zhang', 'Zizhao Zhang', 'Long Zhao', 'Zhuowei Li']
|
2023-03-16
| null | null | null | null |
['class-incremental-learning']
|
['computer-vision']
|
[ 1.08475186e-01 -7.66015872e-02 -2.71672487e-01 -3.08816701e-01
-6.32468641e-01 -4.29198444e-01 7.38964677e-01 6.33711576e-01
-8.25709581e-01 5.59428155e-01 1.91940032e-02 -5.09076178e-01
-2.38366440e-01 -5.05006373e-01 -7.64296412e-01 -4.56382543e-01
-1.46032525e-02 2.58326918e-01 5.90108693e-01 -3.15013766e-01
2.77326226e-01 3.46842498e-01 -2.00838900e+00 2.83712864e-01
8.71633232e-01 8.95829201e-01 2.58372277e-01 3.92091244e-01
-3.83316815e-01 7.70858586e-01 -6.86570764e-01 -2.37478942e-01
1.24221973e-01 9.37775224e-02 -6.42692029e-01 -2.59317309e-01
9.71216798e-01 -3.38241488e-01 -3.90461773e-01 5.70433676e-01
7.96261847e-01 3.04916263e-01 3.54566365e-01 -1.21824992e+00
-8.99122298e-01 6.77502990e-01 -3.48993391e-01 4.21021998e-01
-5.62511990e-03 3.95590067e-01 9.95822251e-01 -1.39197624e+00
2.71764368e-01 9.76347983e-01 9.91770864e-01 7.83875227e-01
-1.37146318e+00 -6.99217677e-01 4.96597379e-01 3.23790967e-01
-1.43688190e+00 -6.78744316e-01 6.10920787e-01 -2.24444360e-01
1.35735679e+00 1.66124493e-01 7.11795032e-01 1.20583403e+00
-1.49618581e-01 1.28778303e+00 9.93349791e-01 -6.13606632e-01
3.73039722e-01 4.13615167e-01 8.09986472e-01 6.30864024e-01
3.68341058e-01 2.51228631e-01 -1.06770408e+00 2.75175227e-03
2.27273449e-01 1.51965961e-01 -3.51305097e-01 -5.68213940e-01
-6.86953187e-01 5.27027249e-01 4.71441746e-01 3.46914828e-02
-2.50227243e-01 1.34978503e-01 4.50302899e-01 3.99175197e-01
1.90664187e-01 6.26586854e-01 -5.20517647e-01 -3.70432049e-01
-1.11512578e+00 2.58393764e-01 3.62525433e-01 8.60325575e-01
8.42513978e-01 2.54321188e-01 -4.42538530e-01 8.83065164e-01
2.57890802e-02 4.60947663e-01 9.00140584e-01 -4.27053005e-01
2.92168915e-01 7.45722055e-01 -4.80754972e-02 -4.53855336e-01
-4.29676175e-01 -7.08437920e-01 -2.71306604e-01 5.78702986e-02
1.19876750e-01 1.60937026e-01 -1.07102561e+00 1.79710376e+00
3.04056913e-01 4.14557070e-01 -4.88553196e-02 5.04200816e-01
6.95830643e-01 4.22839254e-01 4.05022919e-01 -2.40450531e-01
9.77305412e-01 -1.23810494e+00 -5.66733479e-01 -5.47244072e-01
7.60984778e-01 -4.24247533e-01 1.85733187e+00 5.35527945e-01
-9.63132501e-01 -6.77911460e-01 -1.38230526e+00 -1.13853104e-01
-3.23336393e-01 -2.23460257e-01 7.39596725e-01 8.56348157e-01
-1.11981320e+00 7.35089481e-01 -9.08426583e-01 -2.70377427e-01
5.79234540e-01 3.51028323e-01 -7.34994859e-02 -1.46351159e-01
-8.99067223e-01 8.34219038e-01 5.12884676e-01 -2.24521697e-01
-9.91684914e-01 -1.48783934e+00 -4.68680769e-01 2.04475269e-01
2.45989397e-01 -5.45840681e-01 1.47702217e+00 -5.98826706e-01
-1.41205823e+00 5.01660109e-01 1.44215807e-01 -7.18304157e-01
4.82492238e-01 -8.45035076e-01 -4.13431585e-01 -2.93428749e-01
-4.65090156e-01 8.75866592e-01 9.61954892e-01 -1.00805902e+00
-6.95159674e-01 -1.00191019e-01 -3.88283208e-02 2.60962427e-01
-1.43030286e+00 -5.36345363e-01 -4.50411379e-01 -4.92764890e-01
-2.43306667e-01 -8.51789534e-01 1.12047251e-02 -3.77336377e-03
1.70310199e-01 -4.90873575e-01 1.22539127e+00 -1.41524851e-01
1.75529468e+00 -2.30040812e+00 -2.50015378e-01 -3.12773734e-01
6.12488538e-02 8.95118058e-01 -5.52504361e-01 2.44389772e-01
9.37615633e-02 -8.50474685e-02 -1.32314414e-02 -5.29749513e-01
-2.74795499e-02 2.07810238e-01 -5.94764769e-01 1.72893003e-01
2.31795788e-01 1.05051351e+00 -9.46167767e-01 -3.50709707e-02
3.86038683e-02 4.89682406e-01 -7.31568158e-01 3.25055808e-01
-2.87884802e-01 -2.90913999e-01 1.12304688e-01 5.60669303e-01
5.54028332e-01 -1.29438072e-01 1.78572491e-01 2.50881258e-03
-3.15677896e-02 4.53528702e-01 -9.69417393e-01 1.80097008e+00
-6.25936210e-01 6.03596568e-01 -3.71660382e-01 -8.09745908e-01
1.00497472e+00 -8.36196020e-02 1.91345215e-02 -1.47932971e+00
-3.54871571e-01 2.62907684e-01 -2.64276415e-01 -2.08537430e-01
9.35814321e-01 8.09470713e-02 1.34009153e-01 4.01800752e-01
2.14451730e-01 2.35187590e-01 4.68953326e-02 1.81866691e-01
1.30840802e+00 -9.88207571e-03 -1.25248199e-02 -2.81588912e-01
1.18919417e-01 -9.44107473e-02 5.33957839e-01 1.01970220e+00
-3.20002586e-01 3.09358180e-01 3.41237560e-02 -5.93395889e-01
-7.48763680e-01 -1.25346887e+00 -8.50040745e-03 1.53759277e+00
4.66804281e-02 -6.17159605e-01 -3.59034866e-01 -8.56813431e-01
4.21604007e-01 1.11675155e+00 -4.75027084e-01 -8.05971682e-01
-8.00549686e-01 -8.22645724e-01 3.99790138e-01 8.33704472e-01
3.35148484e-01 -6.76430225e-01 -9.52740550e-01 3.50214928e-01
3.49460959e-01 -6.39682055e-01 -5.02972066e-01 5.72185397e-01
-1.19097960e+00 -8.91945302e-01 -4.92764801e-01 -5.23204982e-01
5.35935640e-01 7.78021932e-01 1.34230840e+00 1.82780623e-01
-4.09679562e-01 6.16611481e-01 -2.19013870e-01 -3.47248584e-01
3.16026695e-02 4.14614171e-01 4.26466674e-01 -3.51810247e-01
4.28408206e-01 -7.50860572e-01 -8.74407351e-01 5.98646998e-02
-7.58648694e-01 -1.80148497e-01 7.57974684e-01 1.15085161e+00
3.97739708e-01 -2.30944648e-01 8.34537983e-01 -1.03444910e+00
8.39148402e-01 -3.08746427e-01 -4.69902068e-01 4.79903162e-01
-1.63685262e+00 2.47371778e-01 4.83525753e-01 -8.07966948e-01
-1.00330138e+00 -2.30999187e-01 1.34551764e-01 -6.76815331e-01
2.63914108e-01 3.21229786e-01 1.60022914e-01 -1.50420601e-02
1.09048760e+00 3.77943546e-01 -1.15501449e-01 -6.05529070e-01
6.42475963e-01 6.79932356e-01 6.46725178e-01 -6.46808267e-01
4.69831884e-01 -4.05892096e-02 -4.57650632e-01 -7.03972697e-01
-9.43295360e-01 -5.36113620e-01 -3.88470620e-01 -6.33295402e-02
-5.98819517e-02 -1.26867080e+00 -4.98065323e-01 4.27891761e-01
-8.02903533e-01 -8.58361542e-01 -6.27360165e-01 1.93721354e-01
-2.36852616e-01 -5.73885627e-02 -5.28052211e-01 -6.78190351e-01
-6.84649229e-01 -8.20582092e-01 6.25846565e-01 4.74923432e-01
-2.61554986e-01 -8.83124232e-01 6.89002573e-02 6.93364814e-02
8.52411211e-01 -4.00941849e-01 1.07403350e+00 -8.53806317e-01
-6.28414392e-01 5.93267567e-03 -1.19818300e-01 4.84750658e-01
-1.81251541e-01 -2.71164477e-01 -1.22183192e+00 -8.57941866e-01
-3.85036170e-01 -6.46009445e-01 9.78442550e-01 -2.58643568e-01
9.85460281e-01 -4.58431095e-01 -3.28774989e-01 5.54417372e-01
1.44376957e+00 1.05971768e-01 3.06305975e-01 5.05759239e-01
5.93458712e-01 1.19733028e-01 6.54130399e-01 5.21501720e-01
3.26912522e-01 6.90121949e-01 1.85088843e-01 3.65548581e-01
-7.87819147e-01 -5.75699985e-01 4.52251762e-01 8.82696867e-01
5.93786418e-01 -1.72297470e-02 -1.18814528e+00 8.60812664e-01
-1.84180701e+00 -5.47986567e-01 1.73623934e-01 2.47450709e+00
1.20277762e+00 4.02831852e-01 3.49478647e-02 1.94200963e-01
2.04440266e-01 1.10431843e-01 -9.26436603e-01 -4.42911953e-01
-3.26070935e-02 4.15866792e-01 3.89972836e-01 2.27980539e-01
-9.40585792e-01 1.11916494e+00 7.09866476e+00 9.35823262e-01
-1.39213657e+00 4.15850490e-01 1.98598027e-01 -5.43786645e-01
-3.67826611e-01 -1.05722375e-01 -1.38488948e+00 3.39563727e-01
1.41454577e+00 -2.70253718e-01 3.07744443e-01 1.25110126e+00
-3.01894277e-01 6.92254305e-02 -1.35927033e+00 1.04673243e+00
9.80087966e-02 -1.47164726e+00 1.15187615e-01 -3.42107922e-01
8.55136395e-01 1.28692538e-01 5.78917563e-01 1.07089615e+00
1.36097863e-01 -6.96446478e-01 9.63734746e-01 1.97725087e-01
7.25609481e-01 -8.31556022e-01 2.07308084e-01 2.91806340e-01
-9.07525718e-01 -5.14275849e-01 -2.65070707e-01 -9.50968713e-02
-1.06597073e-01 6.50029898e-01 -1.06237626e+00 -4.50643227e-02
6.69565797e-01 4.40797955e-01 -1.25456679e+00 1.25425839e+00
-2.56718576e-01 8.98394585e-01 -1.86771855e-01 -1.81816578e-01
1.92352861e-01 4.70566839e-01 2.74227113e-01 1.36173248e+00
1.75072655e-01 -2.80523002e-01 5.41826300e-02 4.98418719e-01
-1.75453320e-01 8.39780271e-03 -3.60150784e-01 -6.83903918e-02
1.09944057e+00 8.95604134e-01 -3.19620967e-01 -4.11739886e-01
-2.65005767e-01 9.18074489e-01 9.52068627e-01 2.19087005e-01
-7.55435348e-01 -3.77355486e-01 7.54423857e-01 1.98047012e-01
4.44475025e-01 -2.99518347e-01 -3.61748904e-01 -8.17066550e-01
9.76927951e-02 -8.51915896e-01 2.70470977e-01 -3.29421133e-01
-9.21970725e-01 3.81001949e-01 -9.23241749e-02 -8.40506434e-01
2.42997408e-02 -4.59744841e-01 -6.49259269e-01 3.05197656e-01
-1.66122448e+00 -9.83882904e-01 -3.11747640e-01 4.32475835e-01
6.44205630e-01 -1.53141186e-01 9.47244763e-01 4.74816203e-01
-7.56209373e-01 1.33508790e+00 2.75152385e-01 -5.89837015e-01
8.88347864e-01 -1.33064866e+00 5.06088197e-01 7.47042775e-01
2.59706289e-01 9.28029180e-01 6.80527925e-01 -3.65502119e-01
-1.86712205e+00 -1.33886635e+00 7.34439015e-01 -3.94860983e-01
6.00670278e-01 -5.38426340e-01 -1.11614084e+00 5.03094673e-01
-9.62167454e-04 3.15527201e-01 7.49604225e-01 5.99887431e-01
-9.27621126e-01 -4.62568283e-01 -8.03818524e-01 6.90191627e-01
1.35596049e+00 -6.39924765e-01 -6.67261779e-01 2.72782266e-01
1.10056269e+00 -2.51637608e-01 -6.20517671e-01 3.77114654e-01
5.24452448e-01 -9.77243245e-01 9.87456679e-01 -5.33931375e-01
-4.35013026e-02 1.29609136e-02 -6.54924661e-03 -1.36097646e+00
-3.29676777e-01 -5.99439085e-01 -8.08399141e-01 1.50789940e+00
3.62450004e-01 -4.74623501e-01 8.97807777e-01 6.27189100e-01
-2.82384157e-01 -8.85736585e-01 -8.17984939e-01 -1.32211983e+00
1.81718662e-01 -4.23150152e-01 5.83799183e-01 7.36406744e-01
-1.22165889e-01 4.78168190e-01 -1.17380656e-01 -1.46846354e-01
5.70483923e-01 -1.50131047e-01 7.56476760e-01 -1.16218936e+00
-2.88916618e-01 -5.37427664e-01 -1.12881713e-01 -1.16577935e+00
8.76992717e-02 -8.37668121e-01 6.41576888e-04 -1.06976569e+00
9.70385298e-02 -8.99088860e-01 -8.76619160e-01 8.78654957e-01
-4.40197408e-01 2.81179268e-02 3.61459076e-01 1.52499571e-01
-9.50831115e-01 8.44331741e-01 5.64241529e-01 -3.30980211e-01
-5.32561362e-01 -3.52565765e-01 -7.90923715e-01 2.33073860e-01
5.16047597e-01 -4.21183288e-01 -9.04983580e-01 -6.79285705e-01
2.37440035e-01 -4.32211578e-01 5.22198156e-02 -1.38999689e+00
6.36545181e-01 8.72680470e-02 4.81164940e-02 -4.79719847e-01
2.01817840e-01 -5.77925622e-01 -1.40546694e-01 5.18647611e-01
-5.08924842e-01 9.60088000e-02 6.57865167e-01 9.02441919e-01
3.41610089e-02 -1.89138725e-01 6.92727983e-01 2.93843657e-01
-1.12075651e+00 2.02328026e-01 6.14529755e-03 2.71338969e-01
9.42710042e-01 -9.39057469e-02 -6.76166236e-01 3.83013576e-01
-3.25616568e-01 2.88742244e-01 3.02606344e-01 7.54303336e-01
7.33885169e-01 -1.38528943e+00 -3.00245523e-01 2.57621318e-01
4.56361860e-01 -1.46000892e-01 3.39044660e-01 7.16943324e-01
-7.90225044e-02 4.55482364e-01 -6.98725432e-02 -6.56256437e-01
-1.22853291e+00 7.97403395e-01 -6.82442710e-02 -2.22019345e-01
-6.69751287e-01 1.29245567e+00 -2.08172515e-01 -3.99637967e-01
7.87643194e-01 -4.27202731e-02 1.61755323e-01 2.96184599e-01
8.44229996e-01 4.48906064e-01 5.23505926e-01 2.11712316e-01
-4.04362798e-01 9.14377272e-02 -7.77125418e-01 1.34121507e-01
1.37104535e+00 -4.33383137e-02 4.55731541e-01 5.80668032e-01
1.20283484e+00 -3.99960160e-01 -1.62298548e+00 -5.97148240e-01
3.10602427e-01 -2.80582041e-01 3.20342720e-01 -9.57269490e-01
-8.56908143e-01 8.88090968e-01 1.06225395e+00 -2.10698500e-01
1.07264173e+00 -4.92969841e-01 9.57228720e-01 8.07191133e-01
5.09632826e-01 -1.29552484e+00 5.25478661e-01 6.04875207e-01
5.94842196e-01 -1.20559442e+00 9.00817513e-02 2.91464627e-01
-3.50151151e-01 9.07890975e-01 9.93946016e-01 1.20557219e-01
5.73467672e-01 2.77926624e-01 -3.44230235e-02 6.05785474e-02
-1.33633184e+00 6.37870133e-02 7.39581808e-02 5.53504288e-01
3.76961440e-01 -5.93742318e-02 -2.19817072e-01 6.30000591e-01
-1.27136847e-02 4.53113541e-02 1.52602479e-01 1.27765882e+00
-6.78491414e-01 -1.10917795e+00 3.08890808e-02 1.31360203e-01
1.25972390e-01 -3.02818656e-01 1.03034433e-02 7.96434939e-01
4.89209816e-02 6.69277787e-01 2.79478222e-01 -5.97127497e-01
7.19359756e-01 4.63876903e-01 5.22559524e-01 -5.75601041e-01
-7.86138237e-01 -4.32126760e-01 -1.04701802e-01 -6.07497573e-01
1.60670117e-01 -4.61397529e-01 -1.10813367e+00 -1.63713396e-01
-4.71427143e-01 6.42587543e-02 5.65474570e-01 5.98753989e-01
7.26490498e-01 5.15603364e-01 4.85922456e-01 -4.53308105e-01
-1.07463360e+00 -8.09008777e-01 -2.19040498e-01 2.52405077e-01
2.73583949e-01 -7.70087481e-01 -3.46157759e-01 -3.11747432e-01]
|
[9.816043853759766, 3.390683174133301]
|
d04127dc-eb03-4364-a6ba-6ea741707327
|
deep-learning-for-text-attribute-transfer-a
|
2011.00416
| null |
https://arxiv.org/abs/2011.00416v5
|
https://arxiv.org/pdf/2011.00416v5.pdf
|
Deep Learning for Text Style Transfer: A Survey
|
Text style transfer is an important task in natural language generation, which aims to control certain attributes in the generated text, such as politeness, emotion, humor, and many others. It has a long history in the field of natural language processing, and recently has re-gained significant attention thanks to the promising performance brought by deep neural models. In this paper, we present a systematic survey of the research on neural text style transfer, spanning over 100 representative articles since the first neural text style transfer work in 2017. We discuss the task formulation, existing datasets and subtasks, evaluation, as well as the rich methodologies in the presence of parallel and non-parallel data. We also provide discussions on a variety of important topics regarding the future development of this task. Our curated paper list is at https://github.com/zhijing-jin/Text_Style_Transfer_Survey
|
['Olga Vechtomova', 'Zhiting Hu', 'Rada Mihalcea', 'Zhijing Jin', 'Di Jin']
|
2020-11-01
|
deep-learning-for-text-attribute-transfer-a-1
|
https://aclanthology.org/2022.cl-1.6
|
https://aclanthology.org/2022.cl-1.6.pdf
|
cl-acl-2022-3
|
['text-attribute-transfer']
|
['natural-language-processing']
|
[ 1.87447414e-01 2.85165727e-01 -1.35636538e-01 -5.59270501e-01
-5.39250135e-01 -5.74342608e-01 1.06380081e+00 -1.81796983e-01
-2.49169186e-01 1.07604361e+00 7.99284637e-01 -7.71165267e-02
3.31085593e-01 -6.14775419e-01 -3.18897635e-01 -4.29451615e-01
6.08259261e-01 6.69811726e-01 -4.96915817e-01 -7.75842547e-01
5.09382963e-01 7.06654936e-02 -1.12542176e+00 5.71476042e-01
8.84034276e-01 6.50826752e-01 -9.53138173e-02 6.05260074e-01
-4.83332366e-01 8.53527784e-01 -9.09426987e-01 -7.28058040e-01
-3.11725527e-01 -8.10557067e-01 -1.07666755e+00 -3.61612380e-01
3.94290030e-01 -2.38108531e-01 -4.02695775e-01 9.22782123e-01
9.18259084e-01 2.50931978e-01 8.15853655e-01 -1.23642588e+00
-1.01263440e+00 7.51868248e-01 -1.48410857e-01 -1.18875533e-01
3.46475393e-01 1.77938029e-01 1.06423581e+00 -9.80103552e-01
8.42226207e-01 1.49137807e+00 2.32647166e-01 1.27073812e+00
-9.54231143e-01 -6.51365638e-01 4.84262370e-02 4.34375182e-02
-6.10579193e-01 -2.68155426e-01 1.01458979e+00 -2.87428468e-01
7.81528413e-01 2.03827798e-01 4.75567520e-01 1.75490463e+00
4.87584442e-01 1.34905601e+00 1.19061685e+00 -4.89718467e-01
-1.06015116e-01 2.19826788e-01 7.05266893e-02 2.77615458e-01
-8.00613165e-02 -1.16816409e-01 -6.97287023e-01 -4.25342694e-02
4.80280071e-01 -4.45830703e-01 -2.37361267e-01 8.83403644e-02
-1.40275848e+00 1.05780733e+00 3.57505500e-01 4.27519381e-01
-3.39738131e-02 -4.60301712e-02 8.76134515e-01 6.21972978e-01
9.26392913e-01 7.33056545e-01 -5.06930351e-01 -3.51597607e-01
-6.59351170e-01 6.16010427e-01 1.03117490e+00 1.14270687e+00
3.28844905e-01 2.87967831e-01 -8.08532059e-01 1.06926978e+00
-2.58114915e-02 4.15902674e-01 5.60755491e-01 -6.08453810e-01
5.63185036e-01 4.29115742e-01 7.51191452e-02 -7.65016735e-01
-3.51238489e-01 -1.21896021e-01 -1.20021379e+00 -9.21356529e-02
2.68293530e-01 -7.38538146e-01 -5.68593144e-01 1.83501267e+00
3.50638181e-02 -6.10696435e-01 -1.18715473e-01 7.43531108e-01
1.30008650e+00 8.99706841e-01 2.09415436e-01 -5.82468994e-02
1.44104004e+00 -1.18737638e+00 -1.08299291e+00 -2.33380422e-01
5.01028955e-01 -1.04407859e+00 1.38994193e+00 2.59486109e-01
-1.15271235e+00 -4.84250933e-01 -7.08636343e-01 -6.56499624e-01
-6.25039041e-01 1.83922350e-01 6.36266470e-01 2.74255753e-01
-9.77024436e-01 4.89592999e-01 -2.58751839e-01 -6.87869489e-01
3.97072911e-01 -6.39821077e-03 2.07858328e-02 2.24306986e-01
-1.70297611e+00 9.74343061e-01 1.05758555e-01 1.15475068e-02
-4.51313943e-01 -6.42723024e-01 -5.69832861e-01 -1.40020519e-01
2.91623995e-02 -1.05509818e+00 1.59415913e+00 -1.13572180e+00
-1.98919058e+00 1.26504803e+00 -2.49922827e-01 -2.15791717e-01
6.77337289e-01 -5.56497633e-01 -3.33312929e-01 -3.20800811e-01
-1.04419805e-01 9.79695618e-01 8.00248384e-01 -8.03550839e-01
-5.25794148e-01 -1.65982351e-01 4.89015952e-02 4.20203120e-01
-5.08135736e-01 3.36116463e-01 -2.46592499e-02 -1.14902353e+00
-7.65447974e-01 -1.04528069e+00 8.11048597e-02 -1.59560427e-01
-4.40531671e-01 -8.50370467e-01 6.70012176e-01 -5.07014871e-01
1.26196885e+00 -1.95779467e+00 4.34625149e-01 -5.09001255e-01
1.40174806e-01 2.75565982e-01 -1.73315957e-01 9.56549048e-01
5.61658144e-02 1.88814878e-01 -3.08384836e-01 -3.47973108e-01
2.40647316e-01 -1.36330426e-01 -6.37011230e-01 9.71097127e-03
4.08187777e-01 1.20722151e+00 -9.50288355e-01 -1.27215877e-01
-1.07755765e-01 4.07705009e-01 -3.71495485e-01 5.15126586e-01
-5.32690942e-01 5.35460413e-01 -5.48688114e-01 1.91850007e-01
3.70245606e-01 -1.07748024e-01 -1.24225803e-01 7.28611350e-02
-1.15848511e-01 8.08460534e-01 -4.64418560e-01 1.68355000e+00
-5.41031599e-01 1.05384064e+00 1.84473380e-01 -6.19123340e-01
1.12785625e+00 6.12618387e-01 2.05958530e-01 -5.09376824e-01
4.29495215e-01 1.18406139e-01 -1.47037283e-01 -2.46713296e-01
7.61613011e-01 -2.48172909e-01 -4.14483786e-01 8.53774190e-01
7.17406645e-02 -6.21677816e-01 3.59791309e-01 3.44758332e-01
4.95029539e-01 1.97249442e-01 3.75485033e-01 -5.42978585e-01
6.90267205e-01 -1.88099265e-01 2.65397877e-01 5.38583159e-01
-2.31864288e-01 4.98729348e-01 7.35876381e-01 -5.74715555e-01
-1.06596434e+00 -6.18394375e-01 5.64020202e-02 1.40907168e+00
-3.26811045e-01 -4.25596386e-01 -9.11158979e-01 -6.59851849e-01
-2.48202547e-01 1.04128790e+00 -1.06998801e+00 -2.51473844e-01
-6.98588490e-01 -5.66781104e-01 5.74771166e-01 3.92293841e-01
5.63025594e-01 -1.92572451e+00 -1.23124234e-01 8.34895298e-02
-5.59362054e-01 -7.49853730e-01 -6.70698643e-01 -2.66807258e-01
-8.06821585e-01 -5.79661787e-01 -1.00062823e+00 -9.12396371e-01
3.89664769e-01 -4.86865640e-02 1.49190795e+00 -5.82592487e-02
1.32344633e-01 1.70740589e-01 -4.60651100e-01 -9.89260137e-01
-6.77807152e-01 5.59136391e-01 -2.28332847e-01 -1.89573944e-01
6.26591265e-01 -2.16434047e-01 -4.08142179e-01 -1.26466915e-01
-8.89823198e-01 3.94526422e-01 3.11456054e-01 1.17746699e+00
8.87776958e-04 -6.95868850e-01 8.90618145e-01 -1.18370855e+00
1.39608061e+00 -3.42700034e-01 -7.65801817e-02 -1.75753795e-02
-4.56884235e-01 -2.56521739e-02 6.52057827e-01 -1.54144332e-01
-1.42173278e+00 -4.98140335e-01 -1.56624168e-01 2.75141865e-01
-3.97998005e-01 4.56101537e-01 -8.88636261e-02 3.93412620e-01
7.00298011e-01 -4.35039774e-03 3.81840840e-02 -3.51169020e-01
5.35677791e-01 7.88164794e-01 3.04409619e-02 -8.68017673e-01
5.73597193e-01 2.44689569e-01 -1.80903643e-01 -8.56107593e-01
-1.12600720e+00 -3.15148234e-02 -5.00603855e-01 -1.88208282e-01
7.96409667e-01 -6.94779694e-01 -4.11936373e-01 7.10955203e-01
-1.55528915e+00 -6.93860352e-01 -2.10659876e-01 1.75983936e-01
-6.36948049e-01 -6.61926437e-03 -1.02139485e+00 -3.71710092e-01
-1.04327929e+00 -9.10668433e-01 9.55184758e-01 3.67600918e-01
-7.96542704e-01 -1.23827863e+00 1.43016696e-01 3.94934356e-01
6.72191501e-01 1.94969267e-01 1.00493813e+00 -5.50942242e-01
8.36880058e-02 1.82173550e-02 -1.57486200e-01 2.77771145e-01
1.69188261e-01 2.50441842e-02 -9.78289664e-01 -2.27284953e-01
-2.15959046e-02 -6.85122430e-01 7.92153418e-01 2.50526249e-01
1.00166774e+00 -3.19044352e-01 4.79319021e-02 2.64074117e-01
7.16589868e-01 5.25727542e-03 6.04378819e-01 3.26312363e-01
8.15157533e-01 9.46701288e-01 6.01828754e-01 3.73041302e-01
3.93011868e-01 4.34927851e-01 -2.00574808e-02 -1.90941647e-01
-2.79296786e-01 -2.12422818e-01 4.92284894e-01 9.80092525e-01
-4.53125797e-02 -6.59319460e-01 -6.48766875e-01 3.82046968e-01
-1.66538322e+00 -9.87609863e-01 -2.32356593e-01 1.71001363e+00
1.21457732e+00 2.95912381e-02 2.24838212e-01 -3.18127751e-01
7.12948322e-01 4.76817578e-01 -4.29868847e-01 -1.02717137e+00
-2.73644716e-01 2.27156639e-01 -1.50036633e-01 4.81890082e-01
-9.68596101e-01 1.33588445e+00 6.14981842e+00 7.18003929e-01
-1.31384730e+00 -1.38811871e-01 8.74593437e-01 -1.56983867e-01
-4.48503017e-01 -2.67407060e-01 -6.51177883e-01 3.02594304e-01
8.02615285e-01 -5.96498489e-01 4.67914104e-01 6.94211006e-01
4.40760225e-01 3.41083735e-01 -1.20927119e+00 6.88134968e-01
3.29696953e-01 -1.09083712e+00 3.41657460e-01 -3.33700866e-01
1.08678091e+00 1.51813492e-01 2.27259040e-01 6.25848770e-01
2.12081909e-01 -9.20927286e-01 4.96618390e-01 3.38912576e-01
8.56074154e-01 -8.03118944e-01 6.41328156e-01 1.27905384e-01
-5.75575173e-01 2.72811055e-01 -1.11304715e-01 -3.46728176e-01
2.31120914e-01 5.60204327e-01 -4.77396309e-01 2.66257107e-01
4.38152969e-01 1.02406740e+00 -3.05344224e-01 4.77645546e-01
-7.89957523e-01 8.22336972e-01 2.82767862e-01 -6.23245835e-01
2.94557542e-01 -2.24317461e-01 6.88920081e-01 1.54405475e+00
-7.40410015e-03 3.23240571e-02 -2.71583408e-01 9.51300800e-01
-4.93730724e-01 4.04850781e-01 -5.74660957e-01 -2.56511331e-01
1.92064270e-01 1.37862003e+00 -3.13347459e-01 -5.22107840e-01
-3.00097555e-01 9.53173220e-01 4.07863349e-01 5.16991258e-01
-5.74138343e-01 -6.51913404e-01 6.87195301e-01 -2.24189565e-01
-2.49731675e-01 -1.80878475e-01 -7.35546172e-01 -1.30919814e+00
-1.17295228e-01 -1.14306045e+00 3.00802022e-01 -8.58157277e-01
-1.64620984e+00 6.34689867e-01 -1.42972037e-01 -1.07581329e+00
-3.79395247e-01 -7.11612701e-01 -8.97554994e-01 9.85643089e-01
-1.19170153e+00 -1.16939509e+00 -1.96522161e-01 2.65843362e-01
1.04962945e+00 -3.11979294e-01 9.70110476e-01 2.94633750e-02
-5.10851264e-01 6.96167946e-01 1.39009550e-01 1.54351220e-01
1.47984910e+00 -1.29491651e+00 7.91877925e-01 3.41773450e-01
-3.29685807e-01 5.17888725e-01 7.54059255e-01 -5.20574450e-01
-1.02241778e+00 -9.73703682e-01 1.45448446e+00 -6.09578133e-01
7.72239506e-01 -6.53714478e-01 -8.35641801e-01 6.36498272e-01
1.06144881e+00 -7.65392721e-01 6.25088215e-01 1.76032707e-01
-9.06794369e-02 2.50950977e-02 -9.20552075e-01 1.08377767e+00
7.84537256e-01 -4.20985729e-01 -5.68799734e-01 3.90655845e-01
7.29468942e-01 -4.82898206e-01 -6.74663603e-01 1.73294470e-01
5.27384341e-01 -7.62943625e-01 3.56045663e-01 -8.92524779e-01
1.14895928e+00 2.26096377e-01 3.94494951e-01 -1.65783119e+00
-6.35896564e-01 -9.40729260e-01 4.29619998e-02 1.51844096e+00
4.53327656e-01 -5.73566735e-01 6.84890866e-01 5.97307622e-01
-1.86532930e-01 -7.13124752e-01 -4.22504276e-01 -1.77447170e-01
8.51842999e-01 1.06721260e-01 3.53293151e-01 1.11634016e+00
2.25581437e-01 1.07793212e+00 -5.92120528e-01 -8.57171714e-01
3.64439666e-01 1.68430433e-01 9.11212623e-01 -1.13068545e+00
3.32797691e-02 -9.10967052e-01 3.76901418e-01 -1.04504800e+00
4.73990470e-01 -1.06832469e+00 -6.29711971e-02 -1.68152678e+00
1.01259865e-01 1.35043979e-01 -5.93915284e-02 4.01753753e-01
-4.19627011e-01 1.94294721e-01 2.73462713e-01 -3.86875160e-02
-3.58396173e-01 9.16071296e-01 1.83164966e+00 -2.37826452e-01
-1.44438654e-01 6.09482937e-02 -1.06003737e+00 3.95654112e-01
1.32173944e+00 -2.53858536e-01 -3.21206987e-01 -6.34262860e-01
4.21615690e-01 -1.66753992e-01 -5.70172258e-02 -3.24068069e-01
-1.60941303e-01 -3.12928140e-01 3.35561693e-01 -4.15175885e-01
1.88211560e-01 -3.43090951e-01 -5.08034468e-01 2.55479485e-01
-9.57763255e-01 1.68751433e-01 3.60760719e-01 8.41244012e-02
-2.28937075e-01 2.38033924e-02 7.91321039e-01 -1.81102708e-01
-3.07619780e-01 2.33251974e-01 -7.20718563e-01 3.82868022e-01
5.96345842e-01 4.18592691e-01 -3.73110741e-01 -7.90948927e-01
-3.90860379e-01 3.17212969e-01 2.19056219e-01 1.07339406e+00
2.89006680e-01 -1.38619542e+00 -1.27909255e+00 -1.37244955e-01
2.24738032e-01 -1.27689511e-01 1.86880261e-01 5.58432579e-01
-1.69543445e-01 6.02384150e-01 -3.92805219e-01 -1.37260497e-01
-9.98676717e-01 2.94114798e-01 3.77321355e-02 -3.43604237e-01
-3.58653098e-01 6.64135456e-01 4.00148749e-01 -5.93342662e-01
3.16219451e-03 -8.57730675e-03 -3.69834781e-01 5.84631003e-02
5.94428837e-01 4.82383430e-01 -2.98591889e-02 -4.63156432e-01
-2.73345094e-02 2.28338063e-01 -3.67126912e-01 -2.43861720e-01
1.08222687e+00 -8.92582536e-02 -4.84647691e-01 6.82059348e-01
1.02944863e+00 -8.90211910e-02 -8.10646594e-01 -2.36582607e-01
-9.29126516e-02 8.12951103e-02 -2.40951180e-01 -1.04441845e+00
-7.75632441e-01 1.11489069e+00 2.08323379e-03 1.13444284e-01
9.25016224e-01 -1.35496140e-01 9.60142016e-01 4.06870157e-01
-9.32815522e-02 -1.29368615e+00 1.84900746e-01 1.31877339e+00
1.51357043e+00 -1.20444322e+00 -2.04943717e-01 -5.00240102e-02
-9.21092868e-01 1.16571867e+00 9.32632267e-01 -2.20879242e-01
4.62547094e-01 5.75709008e-02 3.28883260e-01 2.59890198e-03
-1.05641937e+00 3.26394171e-01 3.41450930e-01 4.66703445e-01
1.40499377e+00 1.99076161e-01 -8.28189373e-01 4.45447415e-01
-6.57467723e-01 2.60245297e-02 5.61018705e-01 6.61122918e-01
-2.42215663e-01 -1.35758042e+00 -1.43511772e-01 4.92588818e-01
-5.05461097e-01 -3.54458153e-01 -1.09390724e+00 5.27046919e-01
-3.65687728e-01 8.98256719e-01 -3.96972144e-04 -1.27215356e-01
3.84996086e-01 1.99769020e-01 4.08731669e-01 -6.84375286e-01
-9.37792480e-01 -1.14480011e-01 4.28358734e-01 -1.69842839e-01
-2.95055509e-01 -5.78721464e-01 -9.81718123e-01 -5.77166140e-01
2.05845773e-01 2.18186066e-01 5.12224555e-01 8.63145530e-01
4.13260490e-01 4.50399518e-01 5.41544020e-01 -1.00155210e+00
-5.32375038e-01 -1.45702696e+00 -3.06277394e-01 5.04371464e-01
1.64282814e-01 -2.62176931e-01 -2.65789062e-01 1.42142445e-01]
|
[11.724002838134766, 9.414905548095703]
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.