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
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
77865b2e-d071-484d-a801-42e8af5bdda9
|
on-hyperspectral-unmixing
|
2106.14177
| null |
https://arxiv.org/abs/2106.14177v1
|
https://arxiv.org/pdf/2106.14177v1.pdf
|
On Hyperspectral Unmixing
|
In this article the author reviews Jos\'e Bioucas-Dias' key contributions to hyperspectral unmixing (HU), in memory of him as an influential scholar and for his many beautiful ideas introduced to the hyperspectral community. Our story will start with vertex component analysis (VCA) -- one of the most celebrated HU algorithms, with more than 2,000 Google Scholar citations. VCA was pioneering, invented at a time when HU research just began to emerge, and it shows sharp insights on a then less-understood subject. Then we will turn to SISAL, another widely-used algorithm. SISAL is not only a highly successful algorithm, it is also a demonstration of its inventor's ingenuity on applied optimization and on smart formulation for practical noisy cases. Our tour will end with dependent component analysis (DECA), perhaps a less well-known contribution. DECA adopts a statistical inference framework, and the author's latest research indicates that such framework has great potential for further development, e.g., there are hidden connections between SISAL and DECA. The development of DECA shows foresight years ahead, in that regard.
|
['Wing-Kin Ma']
|
2021-06-27
| null | null | null | null |
['hyperspectral-unmixing']
|
['computer-vision']
|
[ 4.91064489e-01 -1.24542803e-01 -2.23532513e-01 5.26459850e-02
-6.10922694e-01 -5.76009214e-01 4.03690487e-01 -2.99739033e-01
-1.52154798e-02 5.25256634e-01 2.20500529e-01 -5.09279370e-01
-5.21041155e-01 -7.50308275e-01 -2.14013517e-01 -1.14351642e+00
-2.93851167e-01 2.74462372e-01 -3.00728172e-01 -3.85864943e-01
-4.71736640e-02 4.84291106e-01 -1.57127821e+00 -3.23275745e-01
8.69775534e-01 9.95479703e-01 2.18854442e-01 7.14325488e-01
7.48148188e-02 7.64152467e-01 -1.77482814e-01 -4.82152253e-01
3.51612985e-01 -8.57309163e-01 -5.36434054e-01 2.04740256e-01
1.18498569e-02 2.13798985e-01 -2.26184502e-01 1.17004514e+00
3.81954074e-01 -1.20391004e-01 4.31199729e-01 -1.32285881e+00
-6.15281999e-01 4.55404878e-01 -1.03102911e+00 -2.40142853e-03
-1.27208680e-01 5.04508615e-02 1.15466285e+00 -6.54886663e-01
3.24756652e-01 7.63171852e-01 9.64813352e-01 7.44242296e-02
-9.89203990e-01 -4.92627680e-01 -1.20900795e-01 3.89931798e-01
-1.41089010e+00 -3.01149428e-01 7.83275127e-01 -5.76045334e-01
6.25498891e-01 6.45522058e-01 9.10884440e-01 7.32346296e-01
-1.27634436e-01 1.02015603e+00 1.43220186e+00 -4.98931259e-01
3.58712107e-01 -5.59880473e-02 1.19979948e-01 6.30331457e-01
2.75598943e-01 1.19624190e-01 -3.50510895e-01 -1.41265437e-01
3.12732548e-01 -8.43302682e-02 -7.16918945e-01 -8.77671987e-02
-1.11489975e+00 1.01567805e+00 1.82069197e-01 3.93762797e-01
-2.97287434e-01 -2.52070755e-01 6.69063702e-02 4.31388289e-01
5.82917511e-01 4.81633186e-01 -1.92627177e-01 -2.50405073e-01
-1.29510629e+00 -7.57142482e-03 9.13325727e-01 6.81531250e-01
6.00736141e-01 2.93584615e-01 3.38460177e-01 9.00920808e-01
3.02665591e-01 8.64342928e-01 3.29326212e-01 -7.04887092e-01
-1.48999363e-01 2.10562363e-01 -3.74545418e-02 -8.69687855e-01
-4.80635047e-01 -8.75301063e-01 -1.12311935e+00 1.64721847e-01
2.63125539e-01 -3.05063128e-01 -6.73661649e-01 1.37897062e+00
-1.51552498e-01 1.77341506e-01 -6.77147061e-02 9.32877481e-01
5.64228237e-01 7.16519058e-01 -2.77601600e-01 -6.46115482e-01
1.25434470e+00 -8.53554428e-01 -7.25214660e-01 -3.12519222e-01
-8.92517790e-02 -9.59781826e-01 5.02089977e-01 8.53831470e-01
-8.83939266e-01 -5.69362640e-02 -1.22466612e+00 3.34076583e-01
-4.20589983e-01 -1.08970225e-01 1.04336691e+00 9.61312175e-01
-1.04862869e+00 4.79676604e-01 -7.19716072e-01 -8.07860970e-01
4.87964034e-01 -6.13078550e-02 -8.41480643e-02 -2.91276395e-01
-9.07059312e-01 8.43182802e-01 -5.19820191e-02 2.87100822e-01
-4.19354707e-01 -6.09827936e-01 -6.39445841e-01 -1.98060766e-01
7.05199718e-01 -6.95054948e-01 1.07018077e+00 -8.89845073e-01
-1.65211594e+00 7.25069344e-01 -4.71392870e-01 -2.54848808e-01
1.93859816e-01 2.05401719e-01 -8.93033683e-01 2.73480922e-01
-1.35209104e-02 -4.06689197e-03 9.37400281e-01 -1.10609865e+00
-5.00196576e-01 -6.00669026e-01 -3.24851364e-01 8.64516944e-02
-3.05236280e-01 1.00122765e-01 -2.43353784e-01 -8.60039949e-01
7.02124178e-01 -9.14995730e-01 -3.08017462e-01 -6.39600083e-02
-6.98490024e-01 -1.44361127e-02 6.80332780e-01 -6.35335803e-01
1.28338695e+00 -1.98816562e+00 1.88709900e-01 4.83734995e-01
5.03953218e-01 2.63316780e-01 -3.20422947e-02 8.31862569e-01
-4.35862601e-01 1.97390065e-01 -9.38417256e-01 -1.04390897e-01
9.39022657e-03 -6.00632764e-02 -3.81356716e-01 8.22892845e-01
-1.34249359e-01 9.65429544e-01 -1.07324028e+00 3.56714010e-01
3.99362803e-01 3.11805218e-01 -6.21948019e-02 -8.35722461e-02
9.43021849e-02 1.74885973e-01 -2.88755819e-03 1.05099237e+00
8.11635435e-01 -4.13851619e-01 6.32856637e-02 -3.15123767e-01
-6.02659285e-01 -2.73204297e-01 -1.09464967e+00 1.30877995e+00
1.93849858e-03 9.30195689e-01 4.63096857e-01 -1.18991220e+00
6.42070830e-01 3.76714766e-01 8.30110013e-01 -3.46327394e-01
1.28448740e-01 2.90725231e-01 9.16908830e-02 -6.07159019e-01
4.19769049e-01 -5.26366293e-01 5.48166096e-01 4.96785611e-01
-3.54009509e-01 -5.62081933e-01 2.45213166e-01 1.51512906e-01
7.88043141e-01 -9.64149460e-02 7.64687955e-01 -4.13626999e-01
2.46241540e-01 3.19775462e-01 1.73776045e-01 7.78840899e-01
-3.90465856e-01 6.73318744e-01 2.12484807e-01 -6.07269630e-02
-8.73829901e-01 -8.52009714e-01 -6.05409443e-01 7.06450045e-01
-9.26776528e-02 -4.92749691e-01 -4.07453984e-01 1.91317126e-01
-6.62456155e-02 2.72815704e-01 -5.63729525e-01 2.93735087e-01
6.43809885e-02 -1.50275803e+00 3.50933909e-01 1.52568221e-01
6.42984152e-01 -6.21121764e-01 -1.84432045e-01 1.51133403e-01
-1.12274155e-01 -9.60214376e-01 1.71753988e-02 4.27777857e-01
-8.18052769e-01 -1.24087942e+00 -1.15964532e+00 -4.19167608e-01
2.26563394e-01 8.33924294e-01 8.81558955e-01 -3.69455487e-01
-2.88529307e-01 5.06543159e-01 -4.44602281e-01 -6.74067199e-01
-1.36546001e-01 -2.56657779e-01 3.19143906e-02 6.87005892e-02
5.70055604e-01 -9.48869646e-01 -4.59803313e-01 1.99767455e-01
-1.00224710e+00 -3.89051661e-02 3.91857147e-01 5.66158473e-01
4.81524408e-01 2.82524884e-01 2.33660400e-01 -8.05037737e-01
4.81838256e-01 -6.94585204e-01 -5.73133171e-01 2.33574569e-01
-9.19184506e-01 -5.11311054e-01 3.36805284e-01 1.81215689e-01
-6.73427284e-01 -3.34620267e-01 -1.79921016e-01 -1.72395229e-01
-1.06438957e-01 9.81803238e-01 3.89302373e-02 -2.34415442e-01
7.07262516e-01 2.55960971e-01 6.46942407e-02 -4.26999569e-01
3.88957322e-01 8.44837427e-01 6.77590609e-01 -3.32042843e-01
1.02412891e+00 8.48922789e-01 2.89487273e-01 -1.71982765e+00
-7.87121475e-01 -8.67067158e-01 -2.35489473e-01 -1.23014830e-01
7.09850252e-01 -8.72488856e-01 -6.06760442e-01 7.83100486e-01
-7.73681462e-01 -2.24263221e-01 -2.40784824e-01 6.29919112e-01
-1.92966431e-01 4.92583364e-01 -2.34407037e-01 -1.10053122e+00
-2.74328124e-02 -1.07952607e+00 7.31826723e-01 3.93393666e-01
1.46496400e-01 -1.06106520e+00 2.63213515e-01 3.81610543e-01
6.25175059e-01 3.14657301e-01 6.42568707e-01 -1.70827791e-01
-3.26906651e-01 -2.07935080e-01 -4.83082771e-01 5.01574278e-01
9.35929939e-02 2.11114392e-01 -1.41743588e+00 -2.17661440e-01
3.14809591e-01 1.14060037e-01 9.08928990e-01 8.14293265e-01
1.19261158e+00 4.66754325e-02 -2.55340666e-01 1.11933315e+00
1.73918557e+00 3.81063998e-01 7.93962181e-01 2.49223396e-01
6.75197423e-01 4.22976911e-01 -8.29462335e-02 4.51757908e-01
3.58550578e-01 3.99248660e-01 6.17590487e-01 -3.79176676e-01
-1.10337071e-01 6.61938712e-02 2.40181118e-01 1.20771527e+00
-7.63679564e-01 -3.57844770e-01 -7.45620370e-01 2.94354200e-01
-1.69847596e+00 -1.32117438e+00 -7.08406091e-01 2.16796231e+00
3.88002604e-01 -4.59691256e-01 2.89597064e-01 3.33949685e-01
5.60640395e-01 5.96682787e-01 -5.99920392e-01 2.10158844e-02
-8.80654454e-01 5.87594509e-01 7.00557649e-01 5.23056388e-01
-1.28427029e+00 6.25445724e-01 6.96894264e+00 7.08297253e-01
-1.13764906e+00 2.57257015e-01 4.91548091e-01 1.39814436e-01
-4.42651570e-01 1.01116039e-01 -1.57616079e-01 2.30791524e-01
5.99290371e-01 -3.99765849e-01 8.43384147e-01 6.84740901e-01
4.40981746e-01 -5.12426496e-01 -4.84630436e-01 1.31841767e+00
3.82304430e-01 -1.24750817e+00 -5.56774080e-01 2.56892353e-01
8.83491695e-01 5.49991250e-01 1.01706557e-01 -1.38723329e-01
2.43424013e-01 -1.11961043e+00 3.45827013e-01 4.14218992e-01
6.13792837e-01 -6.76265001e-01 6.15835547e-01 1.32039353e-01
-1.17819560e+00 -4.72199395e-02 -5.35812676e-01 -3.22163045e-01
1.88543856e-01 1.24468923e+00 -2.41823837e-01 7.64230371e-01
4.59348559e-01 1.03243184e+00 -2.48115569e-01 1.25743985e+00
-4.03671056e-01 9.58460987e-01 -2.56195605e-01 3.11594307e-02
3.55698943e-01 -8.22738767e-01 9.82950211e-01 1.40030372e+00
4.85394031e-01 5.66364586e-01 4.41989973e-02 7.09943831e-01
3.16666901e-01 1.12660021e-01 -6.92460001e-01 -6.53336823e-01
3.66065353e-01 1.43248868e+00 -8.79138291e-01 -1.17676713e-01
-5.53812623e-01 9.21816826e-01 -3.91565055e-01 6.72398448e-01
-6.96693182e-01 -4.14197594e-01 8.31352711e-01 -4.17825952e-02
9.77136567e-02 -2.47541353e-01 -4.04844254e-01 -1.32078564e+00
-2.68474132e-01 -1.00468767e+00 2.76292533e-01 -9.89674270e-01
-1.39512599e+00 4.77361411e-01 -2.64873534e-01 -1.20001769e+00
2.17855662e-01 -8.37543845e-01 -6.00210428e-01 1.08495104e+00
-1.71117723e+00 -8.47359717e-01 -3.43079686e-01 4.95247900e-01
2.58371860e-01 -2.29819238e-01 9.18620169e-01 2.13536024e-01
-8.58158290e-01 1.11323923e-01 5.28365970e-01 -1.97787493e-01
4.61111814e-01 -1.13291419e+00 4.76625450e-02 1.12653732e+00
3.64724815e-01 4.90733325e-01 8.23056579e-01 -3.10203314e-01
-1.72707486e+00 -6.84765458e-01 6.52383447e-01 -1.80859804e-01
1.10421610e+00 1.55513927e-01 -6.30157590e-01 5.76609433e-01
3.24063689e-01 -3.29356313e-01 1.03985178e+00 1.51100829e-01
-2.70218998e-01 -5.57575151e-02 -8.00646663e-01 4.58470345e-01
9.08396900e-01 -6.85379386e-01 -1.93436533e-01 6.83051646e-01
1.65406838e-01 -1.43339217e-01 -6.78576112e-01 2.92976886e-01
6.64140821e-01 -1.30756891e+00 8.43851388e-01 -3.09021443e-01
3.61380041e-01 -4.67707425e-01 -5.64004064e-01 -1.32487857e+00
-5.15196562e-01 -1.03245318e+00 1.66932419e-01 6.84073985e-01
4.52224612e-01 -8.85107160e-01 5.59952021e-01 1.87949911e-01
-4.00935292e-01 -7.88468897e-01 -6.50260448e-01 -9.00873184e-01
-1.46232903e-01 -8.98711979e-01 4.93780345e-01 1.29760718e+00
3.32659841e-01 9.59271938e-02 -5.39002419e-01 2.97042727e-01
8.96750569e-01 2.39579514e-01 5.87225378e-01 -1.46765649e+00
-4.07335788e-01 -8.65110874e-01 -2.88829714e-01 -8.98841202e-01
-3.64475518e-01 -1.02426875e+00 -1.93348583e-02 -1.43953526e+00
2.92284429e-01 -1.27115414e-01 -2.03302681e-01 3.92265469e-01
-1.63203165e-01 7.17976272e-01 2.55811512e-01 3.86637419e-01
7.40900412e-02 7.39699900e-02 1.28949547e+00 -3.71386379e-01
-2.46977434e-01 1.42107323e-01 -1.22201037e+00 6.75311863e-01
8.86392355e-01 -1.17145352e-01 -5.03889859e-01 -1.35477021e-01
5.13172209e-01 -3.93714979e-02 3.36690396e-01 -1.01991260e+00
8.07268694e-02 -2.62564391e-01 1.50071666e-01 -4.46303695e-01
4.05495495e-01 -6.71662509e-01 4.53018457e-01 2.57287681e-01
3.79690200e-01 -1.90567911e-01 1.55426040e-01 2.95356244e-01
-2.51523525e-01 -3.05150479e-01 8.49950314e-01 -2.25516558e-01
-8.94383073e-01 5.59782207e-01 -3.12194377e-01 -8.99102315e-02
8.97090375e-01 -3.01424533e-01 -4.77194697e-01 -7.55482435e-01
-8.48426878e-01 -8.98735300e-02 4.14052814e-01 -1.37114450e-01
2.67910928e-01 -8.82145166e-01 -8.23482215e-01 1.07000507e-01
1.08207159e-01 -4.19015050e-01 3.35895985e-01 1.35890579e+00
-4.71910357e-01 3.04113567e-01 2.26958901e-01 -6.87265158e-01
-9.86834168e-01 3.43494207e-01 2.11824849e-01 1.14315294e-01
-7.41159201e-01 1.09350765e+00 -3.70437764e-02 3.22948769e-02
-5.16226925e-02 -7.79859498e-02 5.67536466e-02 3.18222791e-01
8.05619776e-01 7.56649733e-01 1.84314121e-02 -4.95450348e-01
-3.31083149e-01 6.71628892e-01 5.44641912e-01 -2.66958416e-01
1.70466352e+00 -7.00406209e-02 -7.25409269e-01 5.98135293e-01
9.93861377e-01 3.29667538e-01 -8.44209254e-01 -1.00936577e-01
-3.45948935e-01 -1.28462702e-01 4.28881109e-01 -9.85731423e-01
-1.07963455e+00 8.79169047e-01 4.29595083e-01 7.02127695e-01
1.35164678e+00 -1.72626749e-01 5.63250542e-01 3.01656216e-01
2.66132265e-01 -8.34176540e-01 -4.15876120e-01 5.08245766e-01
8.14325750e-01 -1.33085763e+00 4.61294830e-01 -3.65743011e-01
-4.62105095e-01 1.18181944e+00 -2.07116082e-01 3.61966133e-01
9.46187556e-01 1.96289569e-01 1.78267300e-01 -4.28185433e-01
-4.28990163e-02 -7.26815104e-01 1.74819618e-01 8.45490396e-01
5.94801843e-01 4.73461837e-01 -1.71201169e-01 3.63806158e-01
-3.77950609e-01 -3.06689236e-02 5.46644032e-01 6.42996073e-01
-7.93302953e-01 -8.75131369e-01 -6.57494426e-01 6.63126290e-01
5.27358241e-03 -5.18552899e-01 -2.96296537e-01 8.95437002e-01
1.52764887e-01 1.19557989e+00 -3.95620078e-01 -2.30502427e-01
-3.36530549e-03 1.83219358e-01 5.41538298e-01 -2.86830336e-01
3.99715453e-03 3.45443100e-01 -2.06830442e-01 -4.97486442e-01
-7.56632686e-01 -7.57711768e-01 -5.51924765e-01 -7.57653654e-01
-2.90359855e-01 3.44936401e-01 9.66626048e-01 9.99529302e-01
-1.50857821e-01 3.57992530e-01 7.25195527e-01 -8.68740737e-01
-1.31083056e-01 -8.58869970e-01 -1.18804920e+00 -3.16539705e-01
4.42531645e-01 -2.22726315e-01 -6.45191252e-01 -1.19564541e-01]
|
[10.032906532287598, -1.9924603700637817]
|
3f9f8185-7df6-4262-a475-57ae9b2941d1
|
dynamic-aggregated-network-for-gait
| null | null |
http://openaccess.thecvf.com//content/CVPR2023/html/Ma_Dynamic_Aggregated_Network_for_Gait_Recognition_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Ma_Dynamic_Aggregated_Network_for_Gait_Recognition_CVPR_2023_paper.pdf
|
Dynamic Aggregated Network for Gait Recognition
|
Gait recognition is beneficial for a variety of applications, including video surveillance, crime scene investigation, and social security, to mention a few. However, gait recognition often suffers from multiple exterior factors in real scenes, such as carrying conditions, wearing overcoats, and diverse viewing angles. Recently, various deep learning-based gait recognition methods have achieved promising results, but they tend to extract one of the salient features using fixed-weighted convolutional networks, do not well consider the relationship within gait features in key regions, and ignore the aggregation of complete motion patterns. In this paper, we propose a new perspective that actual gait features include global motion patterns in multiple key regions, and each global motion pattern is composed of a series of local motion patterns. To this end, we propose a Dynamic Aggregation Network (DANet) to learn more discriminative gait features. Specifically, we create a dynamic attention mechanism between the features of neighboring pixels that not only adaptively focuses on key regions but also generates more expressive local motion patterns. In addition, we develop a self-attention mechanism to select representative local motion patterns and further learn robust global motion patterns. Extensive experiments on three popular public gait datasets, i.e., CASIA-B, OUMVLP, and Gait3D, demonstrate that the proposed method can provide substantial improvements over the current state-of-the-art methods.
|
['Yongzhen Huang', 'Xuecai Hu', 'Chunshui Cao', 'Dezhi Zheng', 'Ying Fu', 'Kang Ma']
|
2023-01-01
| null | null | null |
cvpr-2023-1
|
['gait-recognition']
|
['computer-vision']
|
[-2.47998595e-01 -1.03982460e+00 -2.57349819e-01 -2.07436100e-01
-2.35463336e-01 2.66718380e-02 1.15934238e-01 -1.94867343e-01
-3.18879336e-01 4.84974951e-01 4.84015912e-01 3.60973001e-01
-1.23512879e-01 -9.06773269e-01 -2.96448201e-01 -1.01868296e+00
-3.49703163e-01 -1.43750012e-01 5.67825735e-01 -3.12939942e-01
4.00162637e-02 3.10702801e-01 -1.39497054e+00 -1.59247696e-01
7.43195713e-01 8.44668925e-01 1.24705531e-01 2.74065107e-01
5.29518947e-02 5.55736065e-01 -4.99434024e-01 4.27794531e-02
-1.24144003e-01 -3.20889145e-01 -2.66321778e-01 3.48537534e-01
1.34482592e-01 -6.25775695e-01 -8.85747671e-01 9.26805794e-01
6.98714495e-01 2.85808146e-01 3.70809436e-01 -1.30653083e+00
-7.93412149e-01 4.38271202e-02 -1.03301942e+00 6.05393708e-01
2.84723729e-01 7.78636515e-01 8.36455166e-01 -7.55590439e-01
4.36866224e-01 1.35119081e+00 5.96995950e-01 3.37903082e-01
-7.54033804e-01 -7.18641400e-01 3.47418457e-01 7.02683747e-01
-1.34853196e+00 -2.04880178e-01 1.06755424e+00 -2.68640786e-01
7.02501833e-01 3.93412784e-02 8.52089286e-01 1.11989939e+00
3.23175251e-01 1.08846748e+00 5.72369456e-01 1.05981424e-01
7.82466307e-03 -8.38823855e-01 1.94405347e-01 8.21182013e-01
5.00662208e-01 -2.56387472e-01 -4.72513855e-01 2.43490282e-02
9.59021807e-01 3.47484559e-01 -4.89747554e-01 -3.40535969e-01
-1.34077942e+00 5.01528263e-01 4.30351466e-01 4.69384640e-01
-3.78320336e-01 2.51247585e-01 6.09002352e-01 3.04663852e-02
2.85475194e-01 -3.93868834e-01 -1.05811365e-01 -3.57973784e-01
-6.81863666e-01 3.75783890e-01 1.24036007e-01 5.26133955e-01
9.82511640e-01 3.05637985e-01 -2.58097082e-01 9.51927125e-01
4.04076099e-01 7.55104125e-01 8.33351552e-01 -7.24630117e-01
7.07526088e-01 7.01158285e-01 2.90075876e-02 -1.79501927e+00
-4.34957057e-01 -1.16800658e-01 -1.24388039e+00 -3.55743021e-01
8.47988278e-02 -1.28948197e-01 -8.60956550e-01 1.62676561e+00
2.33963400e-01 4.68674481e-01 -2.63770610e-01 1.08914351e+00
7.54537761e-01 7.99840510e-01 8.50969329e-02 -2.59113908e-01
1.10910702e+00 -8.24784338e-01 -8.50819409e-01 -3.33833158e-01
4.89986002e-01 -4.27843988e-01 1.03144646e+00 3.28769945e-02
-7.05153406e-01 -8.97435427e-01 -1.16089749e+00 2.19852760e-01
-1.86370075e-01 2.27704659e-01 5.42586565e-01 3.77073318e-01
-7.06758082e-01 6.04896426e-01 -1.08095622e+00 -4.42696631e-01
4.70426142e-01 1.95957720e-01 -2.77395874e-01 -4.06048030e-01
-1.23663259e+00 3.68505210e-01 4.73416865e-01 3.52617860e-01
-5.16512096e-01 -1.02119707e-01 -1.11116958e+00 9.07420292e-02
3.07187140e-01 -4.80681807e-01 6.16511464e-01 -6.03419006e-01
-1.15845156e+00 4.08497065e-01 -2.61509806e-01 -7.32726902e-02
5.28608322e-01 -2.88021803e-01 -8.85309219e-01 1.05584063e-01
3.68477941e-01 3.21858287e-01 7.67667711e-01 -8.02900791e-01
-5.33472300e-01 -3.30095202e-01 -3.05918813e-01 9.10956562e-02
-6.31567001e-01 -4.45954241e-02 -8.96550655e-01 -9.39288855e-01
9.92001519e-02 -6.47844613e-01 -2.50518590e-01 -4.44632843e-02
-2.46893808e-01 -2.60456711e-01 1.31686580e+00 -6.82234168e-01
1.68837547e+00 -2.30070066e+00 3.22854742e-02 2.17520133e-01
1.84492216e-01 4.83637780e-01 -1.56265050e-01 2.22502187e-01
1.52382955e-01 1.79391857e-02 -3.92501384e-01 -1.05677806e-01
-3.47689837e-02 3.61461222e-01 2.69287899e-02 3.88437480e-01
3.53831291e-01 8.84704709e-01 -1.10937297e+00 -6.78825080e-01
5.20180166e-01 2.92812884e-01 -4.23442245e-01 6.70654839e-03
2.45746523e-01 3.39419127e-01 -8.75969350e-01 7.91072309e-01
8.63290370e-01 -1.73332751e-01 -9.34707075e-02 -3.77821803e-01
-7.46583417e-02 -3.00992668e-01 -1.22881460e+00 1.70453393e+00
-3.18074925e-03 3.34722936e-01 -2.58236200e-01 -1.15251732e+00
9.48982656e-01 -8.63789860e-03 9.18032825e-01 -7.74749517e-01
1.55036882e-01 2.03095451e-01 -9.51490328e-02 -9.49815273e-01
5.28879821e-01 2.97988325e-01 -2.09257364e-01 3.63862932e-01
2.64135450e-02 6.14602685e-01 4.60280567e-01 -1.19422011e-01
1.09423006e+00 1.51602954e-01 2.45277762e-01 -3.51550095e-02
7.11580932e-01 -2.58079857e-01 1.26787841e+00 3.62188756e-01
-7.31966257e-01 7.24078417e-01 1.81963310e-01 -7.79953599e-01
-7.67700911e-01 -9.35705841e-01 2.72703975e-01 6.84719741e-01
7.00370848e-01 -4.74788576e-01 -4.52041239e-01 -5.99289060e-01
-1.45181969e-01 -2.93880031e-02 -5.04578173e-01 -5.75516641e-01
-1.12481344e+00 -1.19498312e+00 4.56144482e-01 8.88666332e-01
1.27353203e+00 -1.20386493e+00 -6.29198730e-01 5.69269598e-01
-5.16248941e-01 -1.12914634e+00 -9.74824011e-01 -5.36539555e-01
-7.14959025e-01 -1.03658819e+00 -1.04444587e+00 -9.82887864e-01
3.69078666e-01 5.66318572e-01 7.28815317e-01 3.54813069e-01
-3.43535215e-01 1.31637469e-01 -4.64487672e-01 1.64365098e-01
4.25002068e-01 -5.24791814e-02 2.10853830e-01 4.28424001e-01
5.12684166e-01 -6.70068145e-01 -8.56139839e-01 4.04310346e-01
-7.51482427e-01 -1.21800669e-01 5.77985525e-01 1.05448163e+00
5.11496544e-01 2.42539674e-01 3.43466818e-01 -1.35140046e-01
6.22931957e-01 -6.32443428e-01 1.04524583e-01 2.45161161e-01
7.40597695e-02 -1.55713081e-01 6.70739591e-01 -4.40283567e-01
-8.36775303e-01 -1.97443470e-01 -2.06268862e-01 -6.48616672e-01
-2.09521100e-01 6.80889964e-01 -5.53291500e-01 -1.70697421e-02
1.08201385e-01 7.21986413e-01 -1.52086943e-01 -3.27101767e-01
-1.51874736e-01 3.47464502e-01 5.33053637e-01 -6.77469969e-01
9.63935792e-01 3.99929285e-01 -2.10436247e-02 -8.15787017e-01
-1.83051482e-01 -4.11389470e-01 -5.24069488e-01 -4.78880942e-01
8.53820920e-01 -6.97392881e-01 -5.54454386e-01 1.09312546e+00
-8.13099205e-01 -1.32525906e-01 1.26688018e-01 4.57478404e-01
-4.14680541e-01 8.94951701e-01 -7.14738667e-01 -4.67208296e-01
-2.97971338e-01 -1.16457140e+00 1.02014375e+00 7.38038361e-01
-1.50231691e-02 -8.25855315e-01 -3.19315940e-02 1.68379825e-02
3.21158469e-01 5.30102611e-01 7.74597824e-01 -1.22367209e-02
-4.87927407e-01 -8.55748355e-02 -2.71456033e-01 5.65066040e-02
7.14417160e-01 2.50279218e-01 -4.67336714e-01 -3.00772637e-01
-4.74088252e-01 -2.14679942e-01 9.86242235e-01 4.29990619e-01
1.18392766e+00 -2.07367271e-01 -5.42972565e-01 6.41674280e-01
1.14597476e+00 4.00070965e-01 6.81741118e-01 4.23619241e-01
9.70973670e-01 3.14983994e-01 7.94194102e-01 5.51721513e-01
4.49330300e-01 7.02254474e-01 2.31875360e-01 -2.66141575e-02
4.79237698e-02 -1.98683634e-01 3.88696253e-01 1.13160491e+00
-3.50217640e-01 -3.38799767e-02 -8.29083383e-01 7.44265079e-01
-2.23473001e+00 -1.18732572e+00 1.46935567e-01 1.95015967e+00
2.84353614e-01 2.29865551e-01 3.67738396e-01 1.72723547e-01
9.10129786e-01 6.99554026e-01 -8.23290408e-01 1.86687469e-01
-2.88098305e-01 4.32855338e-02 2.49250039e-01 -1.18863210e-01
-1.45537281e+00 7.13585973e-01 5.51880407e+00 9.51893151e-01
-1.17550230e+00 -1.01349287e-01 6.17495656e-01 1.00304373e-01
7.30531514e-02 -3.40493619e-01 -4.04698193e-01 1.01745617e+00
1.80082887e-01 -2.07310185e-01 -4.56478149e-02 8.11974525e-01
4.54345733e-01 2.23557755e-01 -3.59478801e-01 1.20442688e+00
4.05868776e-02 -1.09427476e+00 3.06103081e-01 2.32607778e-03
4.72167134e-01 -3.30604821e-01 4.52069007e-02 3.18417251e-01
1.46940216e-01 -7.29099274e-01 3.13342839e-01 4.65243489e-01
4.96206582e-01 -9.85168040e-01 7.75389612e-01 7.78277069e-02
-2.02513218e+00 -1.68275908e-01 -4.39590245e-01 -1.16471099e-02
2.34958455e-01 5.34729660e-01 1.60470083e-01 9.11642015e-01
9.99513388e-01 1.43436766e+00 -5.35160422e-01 1.27059579e+00
2.37415712e-02 6.29481077e-01 -2.65460163e-01 -9.50868055e-02
5.84726930e-01 -1.72139555e-01 4.97440487e-01 1.17733514e+00
6.38706207e-01 2.94845194e-01 5.17066240e-01 5.98996758e-01
1.34604514e-01 1.57472298e-01 -4.84658897e-01 1.25698969e-01
3.95256162e-01 1.04881918e+00 -7.75301218e-01 -2.30850175e-01
-4.82698917e-01 9.66030836e-01 6.25083596e-02 3.88562113e-01
-7.84174681e-01 -4.78186548e-01 1.09390962e+00 -1.11184806e-01
4.11401510e-01 -6.12299800e-01 2.08074197e-01 -1.57949460e+00
4.17250097e-01 -6.52955532e-01 4.97144163e-01 -5.75067222e-01
-1.29495263e+00 1.35173187e-01 -1.03856184e-01 -1.71752214e+00
2.58631036e-02 -4.92294192e-01 -1.33261955e+00 4.36529517e-01
-1.25311410e+00 -1.08734417e+00 -6.90370321e-01 8.34576488e-01
6.87701464e-01 -4.75874506e-02 3.37625921e-01 7.04138458e-01
-1.08084953e+00 6.10965133e-01 9.05193985e-02 7.02043414e-01
6.26910329e-01 -7.87007213e-01 6.54652596e-01 1.13546324e+00
-2.27505639e-01 5.15470028e-01 4.03997481e-01 -7.32127607e-01
-1.41806519e+00 -1.29639113e+00 4.11173522e-01 2.21882567e-01
6.33670628e-01 7.87846446e-02 -1.08174706e+00 4.16728288e-01
-1.61784589e-01 3.62161398e-01 3.33122343e-01 -3.00107360e-01
2.33572740e-02 -2.95324117e-01 -8.65667939e-01 7.60900676e-01
1.41600955e+00 -3.95301767e-02 -3.55012178e-01 -4.50593196e-02
3.58248115e-01 -2.20579624e-01 -6.71212494e-01 6.82637393e-01
5.69259465e-01 -9.50107992e-01 1.06467903e+00 -3.76434892e-01
3.66885781e-01 -7.00325310e-01 -5.50381541e-02 -1.29923546e+00
-6.07630610e-01 -3.26059550e-01 -2.64106929e-01 1.28652847e+00
-4.92831260e-01 -6.94698155e-01 8.43333542e-01 1.35686830e-01
-2.30842695e-01 -7.44617462e-01 -9.54981804e-01 -7.60835052e-01
-2.26314202e-01 -1.58664420e-01 7.65134394e-01 1.01735711e+00
-8.83328542e-02 -9.68017057e-02 -8.07902098e-01 6.02306873e-02
6.16565108e-01 2.55498499e-01 7.46661425e-01 -9.36247528e-01
-1.13404378e-01 -5.78455806e-01 -1.08529484e+00 -1.15313387e+00
4.21486869e-02 -2.60003418e-01 4.33983654e-02 -1.48160851e+00
1.51912227e-01 -3.24858457e-01 -6.15302622e-01 5.57015300e-01
-7.21712768e-01 1.55285522e-01 5.81570417e-02 3.68900269e-01
-7.22964525e-01 8.17805469e-01 1.43930387e+00 -5.32597840e-01
-9.07546654e-02 -2.85923034e-01 -3.95084292e-01 7.87421525e-01
7.48093069e-01 -5.63834459e-02 -2.91986436e-01 -5.02721071e-01
-3.50998312e-01 -9.10117179e-02 3.28178078e-01 -1.50224423e+00
2.36187816e-01 -3.70314211e-01 6.84881508e-01 -6.23855114e-01
2.66391844e-01 -4.59647387e-01 -5.81704080e-02 7.25900829e-01
2.50547916e-01 2.35358030e-01 8.13629776e-02 6.91908360e-01
-4.98683006e-01 3.42774242e-01 4.81209904e-01 -1.53613910e-01
-1.50907862e+00 8.59164596e-01 -4.91041988e-01 9.31608975e-02
8.65154624e-01 -4.71104354e-01 -2.25580558e-01 -2.00263321e-01
-5.36234200e-01 6.04951859e-01 3.99012208e-01 6.72671139e-01
8.59608233e-01 -1.93041813e+00 -6.83821857e-01 3.14906329e-01
2.57832199e-01 -4.66581807e-02 7.32022464e-01 8.00117016e-01
-5.04074991e-01 3.76859605e-02 -6.11092865e-01 -6.83902919e-01
-1.14325881e+00 2.56932646e-01 2.50500292e-01 -3.24198186e-01
-8.16404164e-01 4.25636023e-01 8.52140486e-02 -9.74488109e-02
-1.64576709e-01 -3.47923070e-01 -4.57325280e-01 -7.00395182e-02
6.38826132e-01 4.11530435e-01 -2.73230851e-01 -1.02564728e+00
-5.29144704e-01 1.20541549e+00 -5.11265583e-02 5.05818844e-01
1.14390481e+00 -1.68884143e-01 1.54389292e-01 2.38787442e-01
1.21430981e+00 -3.62260580e-01 -1.39825332e+00 -2.60962814e-01
-1.60642371e-01 -7.53894925e-01 -3.55134070e-01 5.62879210e-03
-1.55681551e+00 9.10331070e-01 5.97093582e-01 -3.27104740e-02
1.38712358e+00 -4.83645588e-01 1.34406281e+00 4.72261757e-01
5.94203234e-01 -1.16495955e+00 4.88463283e-01 4.31224048e-01
4.08218324e-01 -1.14791620e+00 -6.58936501e-02 -2.58327186e-01
-5.78889608e-01 1.13512194e+00 9.74864900e-01 -4.34689462e-01
5.92037141e-01 -2.68211961e-02 -1.22336395e-01 -2.98729707e-02
-4.09727395e-01 -2.97743261e-01 9.20079648e-02 6.32457852e-01
1.90481171e-01 3.70524898e-02 -3.61937374e-01 5.79018474e-01
2.90496737e-01 1.34967491e-01 1.66551217e-01 1.16066575e+00
-5.88013053e-01 -8.25029373e-01 -3.60151559e-01 5.62024236e-01
-2.01413721e-01 2.17255443e-01 -4.06100117e-02 7.77692497e-01
3.83166581e-01 8.00488353e-01 1.15001172e-01 -8.22833240e-01
3.67798865e-01 -2.27652490e-01 1.55935407e-01 -7.52239227e-02
-8.23775306e-02 -1.62007343e-02 -1.32415578e-01 -7.12923288e-01
-6.01368189e-01 -7.45532632e-01 -1.15650463e+00 -4.85786706e-01
1.22403108e-01 -4.96492945e-02 -2.27337614e-01 1.05724716e+00
3.48819643e-01 5.60852051e-01 6.94733381e-01 -1.10867023e+00
-6.11780956e-02 -7.98325837e-01 -6.37943268e-01 7.37038076e-01
2.49946535e-01 -1.07871056e+00 -5.42492755e-02 6.17750697e-02]
|
[14.295323371887207, 1.4185725450515747]
|
c60c4aaf-bcad-4f92-9f75-52d470fb1183
|
splinecnn-fast-geometric-deep-learning-with
|
1711.0892
| null |
http://arxiv.org/abs/1711.08920v2
|
http://arxiv.org/pdf/1711.08920v2.pdf
|
SplineCNN: Fast Geometric Deep Learning with Continuous B-Spline Kernels
|
We present Spline-based Convolutional Neural Networks (SplineCNNs), a variant
of deep neural networks for irregular structured and geometric input, e.g.,
graphs or meshes. Our main contribution is a novel convolution operator based
on B-splines, that makes the computation time independent from the kernel size
due to the local support property of the B-spline basis functions. As a result,
we obtain a generalization of the traditional CNN convolution operator by using
continuous kernel functions parametrized by a fixed number of trainable
weights. In contrast to related approaches that filter in the spectral domain,
the proposed method aggregates features purely in the spatial domain. In
addition, SplineCNN allows entire end-to-end training of deep architectures,
using only the geometric structure as input, instead of handcrafted feature
descriptors. For validation, we apply our method on tasks from the fields of
image graph classification, shape correspondence and graph node classification,
and show that it outperforms or pars state-of-the-art approaches while being
significantly faster and having favorable properties like domain-independence.
|
['Heinrich Müller', 'Matthias Fey', 'Jan Eric Lenssen', 'Frank Weichert']
|
2017-11-24
|
splinecnn-fast-geometric-deep-learning-with-1
|
http://openaccess.thecvf.com/content_cvpr_2018/html/Fey_SplineCNN_Fast_Geometric_CVPR_2018_paper.html
|
http://openaccess.thecvf.com/content_cvpr_2018/papers/Fey_SplineCNN_Fast_Geometric_CVPR_2018_paper.pdf
|
cvpr-2018-6
|
['superpixel-image-classification']
|
['computer-vision']
|
[ 5.40474616e-03 3.44403803e-01 3.29939388e-02 -3.64474058e-01
-1.51895806e-01 -4.27037776e-01 7.08803952e-01 4.08154607e-01
-5.32969356e-01 4.38270628e-01 -1.95218682e-01 -3.98416877e-01
-2.25522146e-01 -1.20085227e+00 -1.12954259e+00 -5.44597089e-01
-5.35076201e-01 3.35297734e-01 2.81670690e-01 -2.59025842e-01
-2.74330527e-02 8.37975264e-01 -1.22820210e+00 4.67838086e-02
8.35121512e-01 1.37133813e+00 -2.28095070e-01 4.90389496e-01
-1.42773926e-01 4.05287981e-01 -6.90364689e-02 -3.72452497e-01
2.83325970e-01 1.21789105e-01 -7.80166924e-01 3.38625256e-03
9.92317319e-01 -2.20658202e-02 -5.00930071e-01 9.92365718e-01
1.99355260e-01 4.08056408e-01 7.06529498e-01 -9.81903374e-01
-1.20488834e+00 3.92861038e-01 -1.96600094e-01 -2.87120223e-01
-1.69004984e-02 -9.67667401e-02 1.00705028e+00 -1.07634914e+00
6.29143059e-01 1.25687504e+00 1.29955685e+00 2.80162841e-01
-1.85690784e+00 -2.58016616e-01 1.07093148e-01 -1.14864267e-01
-1.34480083e+00 -4.36412580e-02 9.48150277e-01 -4.53469127e-01
7.57537127e-01 9.98244956e-02 6.10786736e-01 9.32206869e-01
4.82187793e-02 4.57212657e-01 7.12228358e-01 -3.42738032e-01
2.63310462e-01 -1.82992369e-01 5.65329567e-02 1.04067230e+00
1.11644879e-01 1.79560278e-02 -5.34279607e-02 -3.17522883e-01
1.34984612e+00 -5.75562455e-02 -2.94800788e-01 -9.44756329e-01
-1.16353166e+00 1.09192288e+00 1.07589793e+00 2.49606773e-01
-3.29273015e-01 7.27076352e-01 5.54879606e-01 3.69981200e-01
7.67973006e-01 2.02060953e-01 -4.03163701e-01 3.27789366e-01
-8.48376811e-01 1.77651212e-01 8.69535744e-01 9.07847941e-01
9.97689664e-01 2.42960751e-01 -1.46905988e-01 6.57725751e-01
7.88069591e-02 1.67459175e-01 -3.56775708e-02 -6.45629764e-01
5.53054027e-02 6.69136047e-01 -1.81136504e-01 -1.00882924e+00
-5.79614341e-01 -4.31090176e-01 -1.11875474e+00 4.67385530e-01
6.80149853e-01 6.81258664e-02 -1.02620590e+00 1.59889042e+00
1.71523079e-01 2.96551883e-01 -2.97215194e-01 7.83667386e-01
1.07562625e+00 2.74724096e-01 -4.08249944e-02 4.49060977e-01
1.19853234e+00 -7.96586514e-01 -3.41542125e-01 3.01262289e-01
5.51556528e-01 -5.26872575e-01 1.19351518e+00 2.09305689e-01
-9.81721044e-01 -6.39052808e-01 -8.89078379e-01 -3.51747543e-01
-6.59436047e-01 3.41709763e-01 8.74788582e-01 4.29721266e-01
-1.60823345e+00 1.18958890e+00 -8.15202653e-01 -3.13914806e-01
6.63363397e-01 4.58861709e-01 -5.08609056e-01 1.97023898e-01
-9.70416486e-01 6.33631408e-01 2.87483126e-01 5.72246388e-02
-4.51189250e-01 -9.25764143e-01 -1.11836493e+00 3.42306525e-01
7.52936676e-02 -5.53975284e-01 1.01531446e+00 -1.11838591e+00
-1.49808931e+00 7.65126288e-01 2.10457727e-01 -6.78569496e-01
5.61890721e-01 3.94044966e-02 -1.03259653e-01 9.66318473e-02
-3.52090657e-01 5.45922875e-01 8.81225049e-01 -9.71861780e-01
3.08765918e-02 -5.81976473e-02 2.16477528e-01 -3.78329962e-01
-4.69127804e-01 -2.53025562e-01 -1.00705236e-01 -8.45291257e-01
-1.27535552e-01 -6.91499174e-01 -4.26565528e-01 6.16143346e-01
-2.45418727e-01 -5.13738692e-01 8.58935058e-01 -5.22099733e-01
7.62611508e-01 -2.24389243e+00 2.02944487e-01 6.50621653e-01
5.31899273e-01 1.94656968e-01 -3.22123647e-01 5.36110401e-01
-2.98430890e-01 3.23149860e-02 -4.50530082e-01 -2.85555720e-01
1.05908237e-01 9.37793106e-02 2.50656605e-02 7.57459164e-01
6.34838581e-01 1.12835073e+00 -8.96712780e-01 -2.02435285e-01
3.01249236e-01 8.37907910e-01 -4.01223987e-01 -6.28085285e-02
-4.33294266e-01 3.92863035e-01 -3.18215787e-01 3.80598575e-01
8.63598585e-01 -4.79171067e-01 -5.49562313e-02 -2.76458710e-01
-6.32012859e-02 1.41978428e-01 -1.09335148e+00 1.94757187e+00
-7.05312788e-01 7.22486854e-01 3.67090166e-01 -1.49963105e+00
1.11741543e+00 2.93909103e-01 3.72339457e-01 -4.19388562e-01
8.17658603e-02 4.59531426e-01 -1.85905367e-01 -5.71310967e-02
1.28749952e-01 2.38487661e-01 2.02550352e-01 2.52266396e-02
3.05890888e-01 6.97662905e-02 5.33240102e-02 -4.62687798e-02
1.08871889e+00 1.98561102e-01 1.62703916e-01 -7.87931383e-01
4.97561514e-01 -3.33024532e-01 4.97591496e-02 4.44165677e-01
3.48237842e-01 5.00042796e-01 6.42840803e-01 -1.05533767e+00
-1.11681843e+00 -1.18408036e+00 -4.19684172e-01 1.00039279e+00
-4.33159545e-02 -3.16110104e-01 -6.96499407e-01 -5.68130910e-01
3.64463955e-01 3.03614080e-01 -1.01241481e+00 1.36846617e-01
-8.59175861e-01 -1.51894659e-01 6.00807667e-01 8.30045462e-01
5.08599937e-01 -1.06280291e+00 -4.34273362e-01 3.05421621e-01
6.26480997e-01 -1.11868143e+00 -5.85252762e-01 1.93279520e-01
-1.04303694e+00 -1.09247363e+00 -8.47710907e-01 -1.05410469e+00
6.92097425e-01 -1.07601129e-01 1.44257259e+00 2.18543723e-01
-2.27098435e-01 3.24263126e-01 -1.12875655e-01 4.61480580e-02
-1.28052399e-01 1.59346893e-01 -4.12105381e-01 1.48342252e-01
-1.20410033e-01 -1.02856123e+00 -4.87508148e-01 -1.03768528e-01
-8.79101455e-01 5.30206738e-03 5.02204299e-01 1.06428921e+00
5.03568530e-01 -3.01975489e-01 4.09140646e-01 -8.94084632e-01
7.69722998e-01 -2.55403399e-01 -8.60842049e-01 1.63494885e-01
-3.02912474e-01 3.71930301e-01 1.00235116e+00 -5.53966045e-01
-4.09421861e-01 2.50126123e-01 -1.59290120e-01 -6.59114838e-01
-3.30919951e-01 7.35812306e-01 2.62045562e-01 -6.86222494e-01
8.99575830e-01 5.13646342e-02 2.29058012e-01 -5.83204329e-01
6.15325391e-01 1.26573965e-02 5.56951880e-01 -6.66435778e-01
8.66439402e-01 5.56161582e-01 5.07901251e-01 -1.07006896e+00
-2.38361359e-01 -2.48668686e-01 -6.88140154e-01 -4.58092876e-02
8.16347122e-01 -5.83132803e-01 -8.79730344e-01 4.15950328e-01
-1.42576742e+00 -5.95080674e-01 -4.23427731e-01 3.16671729e-01
-6.92258120e-01 3.21241826e-01 -7.00120032e-01 -6.11894846e-01
-3.60267580e-01 -9.15657759e-01 1.15854251e+00 -4.93535213e-02
6.18378557e-02 -1.49623096e+00 -2.05849901e-01 -6.31326437e-01
7.16469109e-01 8.85006368e-01 1.14780855e+00 -6.59043789e-01
-5.88152349e-01 -2.97322810e-01 -5.65644562e-01 4.39736217e-01
-5.81670702e-02 1.87624786e-02 -7.80622065e-01 -3.36909264e-01
-3.51672590e-01 -2.08650500e-01 1.13386321e+00 5.16145468e-01
1.64884782e+00 -4.23384488e-01 -1.25109449e-01 1.03077209e+00
1.57644725e+00 -5.36103547e-01 5.41693687e-01 1.10624237e-02
8.89034986e-01 4.24921393e-01 -2.00053394e-01 1.46864459e-01
2.49130473e-01 6.64653897e-01 7.67379999e-01 -7.56133735e-01
-1.80001661e-01 -9.38395690e-03 -4.15333509e-02 2.95258164e-01
-5.32818198e-01 1.64777771e-01 -8.00018847e-01 4.99932975e-01
-1.99024355e+00 -6.32359207e-01 -4.10660326e-01 2.22904372e+00
6.13284767e-01 -6.00787671e-03 3.39415580e-01 1.42615400e-02
5.96790612e-01 1.01041615e-01 -3.10278714e-01 -5.56217492e-01
-1.80715751e-02 1.00395072e+00 8.96485865e-01 4.75323141e-01
-1.43954098e+00 7.73973048e-01 5.71452904e+00 9.01596367e-01
-1.35411370e+00 -5.13315834e-02 3.96616906e-01 3.68292183e-01
-3.38749647e-01 -2.67183959e-01 -7.01260492e-02 1.23850681e-01
6.59130096e-01 1.87054187e-01 7.42454529e-01 9.96654153e-01
-2.03371406e-01 4.66043741e-01 -1.30889213e+00 7.35699773e-01
-3.12164515e-01 -1.75629961e+00 -1.64744839e-01 6.48266673e-02
5.24870038e-01 1.55868888e-01 1.04298480e-01 1.28058329e-01
3.30481410e-01 -1.50201309e+00 8.11563432e-01 5.08688986e-01
1.10343599e+00 -7.50245750e-01 4.61329460e-01 6.43071234e-02
-1.52815640e+00 2.47304097e-01 -3.96950215e-01 -8.81384760e-02
-2.57122666e-01 6.33225858e-01 -6.65655911e-01 7.05570102e-01
6.33842349e-01 7.20055342e-01 -4.36554074e-01 1.01737118e+00
-4.73269932e-02 4.95779544e-01 -4.43593651e-01 -3.67031842e-02
4.89505559e-01 -2.96943426e-01 3.23029399e-01 1.43677437e+00
1.67530417e-01 -2.92306572e-01 5.00182867e-01 1.35469007e+00
-2.94108391e-01 1.76684871e-01 -7.10401893e-01 8.38026404e-02
9.36131328e-02 1.39760220e+00 -6.96312964e-01 -6.98368549e-02
-6.19130254e-01 7.74598420e-01 6.85628653e-01 6.39621615e-01
-5.82637131e-01 -8.27161551e-01 3.94059360e-01 3.19310933e-01
7.29261637e-01 -6.02956533e-01 -4.85987693e-01 -8.48077714e-01
3.07694584e-01 -4.18420225e-01 -1.06221177e-02 -2.28120342e-01
-1.57485056e+00 7.71324277e-01 -2.21458659e-01 -9.98009443e-01
7.39508197e-02 -1.13446832e+00 -9.41298246e-01 1.19486606e+00
-1.66900706e+00 -1.58045435e+00 -4.20808733e-01 6.82762146e-01
1.98372379e-02 5.99026456e-02 1.06247950e+00 2.06792042e-01
5.56104593e-02 6.51540220e-01 1.61929488e-01 5.96486211e-01
3.56935084e-01 -1.54605091e+00 9.68961179e-01 4.50519919e-01
9.66525748e-02 5.85429907e-01 4.17225450e-01 -3.95156592e-01
-1.25882363e+00 -1.22781062e+00 5.88799179e-01 5.18780202e-02
8.48601580e-01 -7.64330447e-01 -1.11108613e+00 4.87194657e-01
1.03527561e-01 7.49458075e-01 2.57367104e-01 2.91809112e-01
-6.23766422e-01 9.49885398e-02 -1.11237991e+00 4.06568706e-01
1.18604481e+00 -4.74017590e-01 -1.04617309e-02 3.80885124e-01
7.72151053e-01 -5.25092781e-01 -1.11841369e+00 5.33311546e-01
4.61167097e-01 -8.32206488e-01 1.16130352e+00 -8.03609908e-01
3.55585277e-01 -1.39551014e-01 3.30676064e-02 -1.37293696e+00
-5.49888134e-01 -6.08735561e-01 -1.70560941e-01 5.71625471e-01
3.64469647e-01 -7.79917419e-01 8.50973964e-01 2.80832350e-01
-4.93613183e-01 -1.14067960e+00 -1.14986098e+00 -1.09646142e+00
2.90969431e-01 -2.31648371e-01 5.66403806e-01 9.18927372e-01
-3.29193741e-01 2.45649219e-02 -7.95220733e-02 -4.93265875e-03
6.61467791e-01 1.31954834e-01 6.61013365e-01 -1.77632356e+00
-3.24315220e-01 -6.79283798e-01 -6.99943841e-01 -1.01291025e+00
3.97802204e-01 -1.23196864e+00 -1.96049020e-01 -1.47391307e+00
-5.16576231e-01 -6.97882056e-01 -7.59060830e-02 6.61124170e-01
2.25294501e-01 2.87075013e-01 -1.04249112e-01 -1.67742237e-01
-2.35343799e-01 6.68520987e-01 1.40638304e+00 -1.32256448e-01
-1.70485377e-01 5.83326295e-02 -5.95291071e-02 8.29523146e-01
7.08114326e-01 2.52262969e-03 -1.89481124e-01 -4.60243225e-01
2.09614918e-01 -1.76068798e-01 9.99617100e-01 -9.13371325e-01
1.33983195e-01 1.80633701e-02 3.95216346e-01 -2.70919025e-01
2.58773059e-01 -8.85870755e-01 3.97050232e-02 3.44300538e-01
-3.39880764e-01 5.86628588e-03 3.19009274e-01 6.50756896e-01
-3.38466503e-02 -6.45031929e-02 8.70042443e-01 -1.17435884e-02
-4.49965924e-01 6.17961466e-01 -1.32718598e-02 -1.03653498e-01
7.28726149e-01 -3.86717200e-01 -1.17469482e-01 -2.56630093e-01
-8.56791973e-01 -1.56227142e-01 5.60273051e-01 2.88393535e-02
5.78356981e-01 -1.68049324e+00 -8.35339665e-01 2.45854244e-01
4.19383571e-02 3.77204478e-01 -1.25415819e-02 8.98995161e-01
-7.62297750e-01 2.58893192e-01 -2.14477345e-01 -6.59867764e-01
-8.48963976e-01 6.33153617e-01 3.94387990e-01 -2.60499507e-01
-9.50544298e-01 8.32364678e-01 1.62677020e-01 -6.31640851e-01
4.15159017e-01 -7.32392073e-01 1.22604579e-01 -2.86318839e-01
3.24244983e-02 1.78301871e-01 4.08858687e-01 -3.39884937e-01
-2.68077642e-01 4.36150432e-01 4.39403743e-01 3.57170612e-01
1.60037851e+00 6.16316199e-01 -3.94623429e-01 2.01378062e-01
1.29873860e+00 -6.51998669e-02 -1.30324352e+00 -5.07543087e-01
1.03038207e-01 -2.90144712e-01 6.32474720e-02 -4.48209018e-01
-1.17917955e+00 8.72077227e-01 3.43042642e-01 5.49504161e-01
9.14930701e-01 -2.20398288e-02 6.44591272e-01 5.67064822e-01
1.50798067e-01 -7.71685183e-01 5.28852902e-02 7.17504263e-01
1.15067279e+00 -9.64630544e-01 -2.86423326e-01 -6.01308763e-01
3.99334170e-02 1.71490932e+00 2.72611558e-01 -9.16694105e-01
1.06930101e+00 1.68054312e-01 -3.24785024e-01 -3.41770947e-01
-3.36482942e-01 -2.57430881e-01 7.77786195e-01 7.78848112e-01
7.66912580e-01 1.43444002e-01 -2.26110145e-01 2.55008310e-01
-6.95951357e-02 -1.78989381e-01 1.97149530e-01 5.51145613e-01
-2.14085892e-01 -1.11159670e+00 -9.46472660e-02 6.13447964e-01
-2.24941939e-01 -3.27390522e-01 -2.12382317e-01 8.64008307e-01
1.08787930e-02 3.23809236e-01 1.90503046e-01 -1.05589710e-01
3.86071533e-01 -4.15055566e-02 5.64068615e-01 -5.55094481e-01
-6.69601440e-01 -2.04336047e-01 4.43686247e-02 -5.43862581e-01
-3.76879513e-01 -2.42344007e-01 -1.15479875e+00 -4.05540913e-01
-3.34432364e-01 -7.20956400e-02 6.76412642e-01 5.88737488e-01
3.60516042e-01 6.04879856e-01 3.39999318e-01 -1.24413669e+00
-7.83724070e-01 -9.98856604e-01 -5.33543706e-01 5.65845728e-01
5.31064749e-01 -7.13425696e-01 -3.86704616e-02 -3.74948420e-02]
|
[8.27674388885498, -3.668534994125366]
|
95981c75-1fa5-4afd-b11b-d2679f354b0e
|
discrete-opinion-tree-induction-for-aspect
| null | null |
https://aclanthology.org/2022.acl-long.145
|
https://aclanthology.org/2022.acl-long.145.pdf
|
Discrete Opinion Tree Induction for Aspect-based Sentiment Analysis
|
Dependency trees have been intensively used with graph neural networks for aspect-based sentiment classification. Though being effective, such methods rely on external dependency parsers, which can be unavailable for low-resource languages or perform worse in low-resource domains. In addition, dependency trees are also not optimized for aspect-based sentiment classification. In this paper, we propose an aspect-specific and language-agnostic discrete latent opinion tree model as an alternative structure to explicit dependency trees. To ease the learning of complicated structured latent variables, we build a connection between aspect-to-context attention scores and syntactic distances, inducing trees from the attention scores. Results on six English benchmarks and one Chinese dataset show that our model can achieve competitive performance and interpretability.
|
['Yue Zhang', 'Zhongqing Wang', 'Zhiyang Teng', 'Chenhua Chen']
| null | null | null | null |
acl-2022-5
|
['aspect-based-sentiment-analysis']
|
['natural-language-processing']
|
[-6.95747510e-02 1.29210249e-01 -6.52737856e-01 -7.33047187e-01
-5.48007369e-01 -5.19237757e-01 4.26044792e-01 1.73311815e-01
-3.17451119e-01 4.48024869e-01 3.91459435e-01 -6.51548326e-01
1.45211861e-01 -9.31265891e-01 -1.72151312e-01 -5.10912538e-01
1.84427455e-01 6.59686267e-01 2.66787186e-02 -2.73416609e-01
5.91877811e-02 7.50926584e-02 -1.01897752e+00 3.80883664e-01
1.05752301e+00 8.45860958e-01 -7.89046735e-02 4.95785952e-01
-6.96723282e-01 7.12548077e-01 -5.08695424e-01 -8.88597965e-01
-6.90759718e-02 -3.34245622e-01 -6.79064453e-01 -2.83268802e-02
1.08063385e-01 2.10291550e-01 1.04469031e-01 9.15739536e-01
2.54967600e-01 -1.80919021e-01 6.56599402e-01 -1.06593132e+00
-9.52915370e-01 9.24266458e-01 -5.91778874e-01 1.55334145e-01
1.02337472e-01 -1.22338161e-01 1.69018316e+00 -7.70982444e-01
4.60887462e-01 1.18616474e+00 6.32341087e-01 3.27162325e-01
-1.02586842e+00 -4.10228193e-01 6.98120296e-01 1.84653267e-01
-8.19598496e-01 -1.45737380e-01 1.23232114e+00 -1.55116066e-01
1.40614617e+00 7.59626403e-02 6.87601686e-01 1.25453126e+00
3.45665097e-01 7.56926954e-01 1.07220280e+00 -3.31801713e-01
2.10103482e-01 2.93397903e-01 5.49975395e-01 8.42713177e-01
4.84939277e-01 -4.87231255e-01 -5.84073186e-01 -1.13865964e-01
3.51467937e-01 -6.59209117e-02 8.71321782e-02 -4.75550801e-01
-6.03394151e-01 1.34542942e+00 5.31649768e-01 3.57956558e-01
-1.77515894e-01 -6.02085143e-02 4.97906178e-01 3.30696553e-01
1.06625259e+00 4.64768022e-01 -9.58353877e-01 -4.59258631e-02
-3.99177760e-01 -4.55759555e-01 1.08942306e+00 9.03957069e-01
8.27953339e-01 2.91729301e-01 1.11814868e-02 8.68076682e-01
4.80679810e-01 4.33734983e-01 5.41079044e-01 -1.78007931e-01
7.68330514e-01 1.13070512e+00 -5.99541068e-01 -1.02075541e+00
-6.71268046e-01 -6.73297405e-01 -8.91967475e-01 -1.82861120e-01
1.25137895e-01 -1.61103368e-01 -1.06947064e+00 1.47113323e+00
1.87613845e-01 -3.05711627e-01 7.90045410e-02 5.95014393e-01
7.73249865e-01 4.46582764e-01 2.07760125e-01 -1.36305869e-01
1.45748854e+00 -1.36986303e+00 -6.24648154e-01 -7.10634351e-01
1.04471791e+00 -5.21171212e-01 1.41196799e+00 1.82718143e-01
-5.03040373e-01 -2.40826145e-01 -8.67070019e-01 -1.66172177e-01
-7.46800959e-01 7.56601095e-02 1.31910217e+00 9.93079662e-01
-1.00397921e+00 2.84714013e-01 -1.01034522e+00 -2.71572560e-01
5.03250182e-01 4.10010964e-01 -4.35859114e-01 1.14320777e-01
-1.05296218e+00 8.28151584e-01 1.61026433e-01 1.83714107e-01
-2.69243985e-01 -3.36792290e-01 -1.11764288e+00 1.61687165e-01
4.50517893e-01 -8.55124176e-01 9.30989921e-01 -1.10975385e+00
-1.68706310e+00 7.27019370e-01 -2.52211720e-01 -3.22179347e-01
-2.61242595e-02 -3.05639595e-01 -2.69700259e-01 -1.26766801e-01
1.75824806e-01 2.41985217e-01 9.51953650e-01 -9.70564663e-01
-3.93970579e-01 -5.99031806e-01 5.19564629e-01 3.51477802e-01
-8.53063464e-01 7.27749914e-02 -6.62883282e-01 -6.43989146e-01
1.17220536e-01 -9.82206643e-01 -6.45873725e-01 -5.91727257e-01
-5.24866223e-01 -4.07059789e-01 8.09858203e-01 -5.73380172e-01
1.22771990e+00 -1.90660477e+00 2.07509860e-01 2.02103257e-02
2.38232762e-01 1.43550828e-01 -2.20821157e-01 1.54550388e-01
-1.72696151e-02 3.60545933e-01 -4.59048122e-01 -5.75040996e-01
-6.77042305e-02 4.13845956e-01 -2.77092513e-02 8.96305367e-02
4.38836634e-01 1.18660533e+00 -6.65863752e-01 -5.01632512e-01
-1.64585665e-01 5.20740509e-01 -7.85975039e-01 1.15874680e-02
-3.57575804e-01 2.67473668e-01 -7.23002315e-01 8.84107530e-01
3.63752753e-01 -3.66754264e-01 5.01496255e-01 -2.36535762e-02
2.75371253e-01 7.39356160e-01 -5.56150138e-01 1.60867739e+00
-1.08343732e+00 4.79479790e-01 -2.00729236e-01 -1.05470228e+00
9.32658851e-01 9.54697132e-02 1.27940834e-01 -6.02683127e-01
1.56514287e-01 -7.30914697e-02 6.32012933e-02 -2.31703937e-01
5.17712712e-01 -2.74603516e-01 -4.51379240e-01 4.59479153e-01
1.37864813e-01 -1.22811362e-01 3.76154840e-01 2.10577846e-01
9.68461215e-01 1.33757770e-01 3.54057759e-01 -2.50884920e-01
6.81775689e-01 -1.78029239e-01 6.62823439e-01 4.15424258e-01
7.86302239e-02 5.20739138e-01 1.01998758e+00 -5.97443163e-01
-6.04004085e-01 -7.12662458e-01 5.83424531e-02 1.37665248e+00
-2.19611734e-01 -9.33215678e-01 -4.54105258e-01 -1.53882527e+00
-3.28721106e-01 7.13733435e-01 -5.16894579e-01 -1.24445483e-01
-6.46852732e-01 -1.10837162e+00 2.06736028e-01 8.32826257e-01
2.51386791e-01 -1.04791868e+00 -1.54502660e-01 1.67564362e-01
1.08465943e-02 -1.39410543e+00 -3.48296493e-01 6.60410523e-01
-1.14029562e+00 -8.29903543e-01 -2.48840377e-01 -7.57750750e-01
7.78334439e-01 5.12713846e-03 1.53566027e+00 4.59410064e-02
2.22129077e-01 1.43441707e-01 -5.50114393e-01 -5.82527697e-01
-7.44959861e-02 8.17334712e-01 -1.15184195e-01 -3.79989929e-02
7.28150308e-01 -6.64153755e-01 -3.10124189e-01 1.24541901e-01
-7.62029886e-01 -9.41150561e-02 6.74531758e-01 9.09506977e-01
4.68749434e-01 -1.24126688e-01 3.57197344e-01 -1.56520116e+00
6.46456182e-01 -2.61788964e-01 -6.32343829e-01 1.07073598e-01
-1.01315701e+00 3.26665640e-01 9.94636476e-01 -1.27983108e-01
-1.00413013e+00 -8.23303312e-02 -3.07019949e-01 4.83149923e-02
-9.43632871e-02 9.28220809e-01 -5.04328966e-01 1.66450933e-01
2.24384993e-01 -9.32375342e-02 -5.13847888e-01 -4.76640433e-01
5.17955482e-01 5.18033624e-01 -3.11460793e-01 -3.44467044e-01
6.91572428e-01 4.09142405e-01 -9.49950889e-02 -7.74835348e-01
-1.32646787e+00 -3.24397117e-01 -9.23694611e-01 3.94726932e-01
1.03372765e+00 -7.86201537e-01 -7.07372651e-02 3.32623839e-01
-1.13103950e+00 -1.24570630e-01 -1.17187731e-01 3.99354577e-01
-1.81668356e-01 3.76199275e-01 -7.41464078e-01 -5.94652593e-01
-5.47062278e-01 -1.22814751e+00 1.20154548e+00 1.08705893e-01
-2.01808855e-01 -1.55314040e+00 1.09194748e-01 5.06720006e-01
5.79395473e-01 -1.52392671e-01 1.26710498e+00 -8.32506299e-01
-3.86916906e-01 -3.77937973e-01 -1.43926680e-01 3.66391420e-01
2.42126673e-01 -1.54056117e-01 -1.00891864e+00 -9.07027572e-02
-3.24029475e-02 -9.28882286e-02 1.07558858e+00 2.37133563e-01
9.51015890e-01 -3.01082522e-01 -3.08023214e-01 8.08315217e-01
1.30662727e+00 -1.31353229e-01 3.54780763e-01 4.15199906e-01
1.24934292e+00 6.55191898e-01 2.61004657e-01 7.69726858e-02
7.92937100e-01 5.15925050e-01 2.47251630e-01 -2.01588422e-01
-3.74941938e-02 -2.11382464e-01 5.51590502e-01 1.61258125e+00
-1.22130886e-01 -3.86372626e-01 -9.15182412e-01 5.09944618e-01
-1.56963134e+00 -3.85769010e-01 -4.52774495e-01 1.78809571e+00
6.26210690e-01 5.62589169e-01 -7.13861212e-02 -4.79903817e-02
3.03105503e-01 5.74481130e-01 -3.86084080e-01 -8.57013166e-01
-2.73248106e-01 4.86411482e-01 3.41543525e-01 4.92227942e-01
-1.26322973e+00 1.43570924e+00 5.90050459e+00 5.48282027e-01
-1.09042776e+00 3.95450085e-01 7.26849735e-01 9.13561657e-02
-6.48273349e-01 3.40432525e-01 -9.81258094e-01 1.51369542e-01
9.28558767e-01 4.97964621e-02 -2.40910172e-01 1.15422928e+00
-2.22382158e-01 1.69523641e-01 -1.03181064e+00 6.72299504e-01
2.52990663e-01 -8.89225602e-01 3.38001579e-01 -1.13537339e-02
6.85708582e-01 1.61823407e-01 1.21922001e-01 5.86856127e-01
4.14176404e-01 -9.13277864e-01 1.88688159e-01 -1.78364404e-02
5.07123232e-01 -8.10934305e-01 1.00101995e+00 3.94047946e-02
-1.32113588e+00 1.03149004e-01 -6.04314268e-01 -1.77735329e-01
1.82859465e-01 7.66256154e-01 -5.19754112e-01 7.23162472e-01
6.89754128e-01 9.48394179e-01 -9.11687672e-01 4.85028863e-01
-8.54698539e-01 8.47756684e-01 -1.88835606e-01 -3.55239391e-01
4.42106366e-01 -4.77757812e-01 4.16846871e-01 1.17116070e+00
3.72441113e-02 -3.61423939e-01 1.30197763e-01 4.84554410e-01
-8.68413076e-02 5.10210097e-01 -9.03951168e-01 -2.14170322e-01
-2.32065156e-01 1.63674819e+00 -1.07398391e+00 -2.12363616e-01
-8.26130509e-01 1.00755608e+00 6.22588813e-01 3.48181814e-01
-7.61556566e-01 -3.00855488e-01 6.58627212e-01 -1.77924350e-01
5.94666183e-01 -3.33668023e-01 -6.32012069e-01 -1.61751485e+00
1.43247023e-01 -8.29067230e-01 6.54096246e-01 -4.55412596e-01
-1.43400443e+00 1.00105321e+00 -3.15154076e-01 -1.00123811e+00
-2.46808633e-01 -9.07705128e-01 -6.96712434e-01 6.63442850e-01
-1.89642441e+00 -1.51564610e+00 5.93844950e-02 3.95045966e-01
6.83136463e-01 -2.91932970e-01 1.11553276e+00 1.94376498e-01
-7.31290162e-01 7.64665544e-01 -2.21243992e-01 2.76952565e-01
4.10351485e-01 -1.54347324e+00 8.08673263e-01 8.07099700e-01
6.65197372e-01 7.59322584e-01 4.08345789e-01 -5.21330297e-01
-1.50859678e+00 -9.96687412e-01 1.09378147e+00 -9.32895303e-01
1.00044513e+00 -6.79145515e-01 -8.35472882e-01 7.42384255e-01
4.39634234e-01 -1.90994486e-01 1.03400910e+00 9.00698185e-01
-7.04178512e-01 -1.04499355e-01 -5.00441790e-01 6.73969865e-01
1.04475737e+00 -6.79966271e-01 -5.31223118e-01 4.05878156e-01
9.41833317e-01 -3.68115343e-02 -6.22817218e-01 4.22583967e-01
3.11004490e-01 -8.60079288e-01 6.76190495e-01 -8.96031499e-01
5.19797087e-01 -6.45236596e-02 -3.63377295e-02 -1.39791334e+00
-3.72745752e-01 -2.63718039e-01 -7.66139701e-02 1.30247450e+00
1.13087165e+00 -8.64745915e-01 1.18127894e+00 4.41837400e-01
-2.23858058e-02 -9.11651433e-01 -8.16359401e-01 -6.70871019e-01
2.92829692e-01 -7.03356683e-01 4.73268658e-01 1.01375508e+00
-3.83488946e-02 1.24682701e+00 -1.65245369e-01 1.25080153e-01
3.29870313e-01 4.56668228e-01 5.98512292e-01 -1.31304967e+00
-5.30555069e-01 -5.56126893e-01 -3.59247953e-01 -9.36934292e-01
5.80770373e-01 -1.14618409e+00 -2.92770803e-01 -1.85472000e+00
1.25881076e-01 -5.45142353e-01 -4.21760529e-01 7.14062154e-01
-4.63415980e-01 1.62681565e-01 -5.57105541e-02 -6.71209544e-02
-7.38862753e-01 8.21425378e-01 1.05340886e+00 -4.18489546e-01
-2.55172998e-01 2.00929776e-01 -9.67228651e-01 1.05061126e+00
1.01853359e+00 -8.40563655e-01 -5.19165158e-01 -7.01798320e-01
7.19470680e-01 -2.65416592e-01 -3.26008916e-01 -5.39929390e-01
9.54218730e-02 8.44728723e-02 1.85952540e-02 -5.94187677e-01
2.79230684e-01 -8.03683519e-01 -4.18224066e-01 2.07523853e-01
-4.48347554e-02 3.76921892e-01 -8.24858546e-02 6.30579233e-01
-4.76276308e-01 -2.52730489e-01 1.88323870e-01 -2.02495173e-01
-5.08629739e-01 4.48773742e-01 -3.45447809e-01 1.26706168e-01
6.35049284e-01 3.89847718e-02 -2.18869895e-01 -3.85157317e-01
-3.71186644e-01 1.00736298e-01 3.10838163e-01 7.00075209e-01
4.88724560e-01 -1.02479720e+00 -4.98812467e-01 3.54058683e-01
3.54325920e-01 -3.98488156e-02 -1.86505795e-01 8.21164310e-01
-2.99678802e-01 5.74559808e-01 1.92289308e-01 -5.92882633e-01
-1.24804974e+00 5.75342596e-01 1.47635043e-02 -8.26546609e-01
-5.27346253e-01 9.45843577e-01 4.38216299e-01 -9.48550642e-01
-8.35921168e-02 -4.30546910e-01 -6.41529143e-01 3.06405485e-01
1.06542759e-01 -2.48363703e-01 2.57357985e-01 -6.12781882e-01
-4.52879846e-01 7.83256352e-01 -2.37379491e-01 1.28073514e-01
1.47829318e+00 1.26840279e-03 -3.71525347e-01 5.71305752e-01
1.13822174e+00 2.34087616e-01 -7.57280469e-01 -9.07713249e-02
5.00302196e-01 -1.58490047e-01 6.76662698e-02 -3.81793588e-01
-1.40849495e+00 1.02855480e+00 4.94721681e-02 4.50670063e-01
1.06820357e+00 9.16138217e-02 7.35465825e-01 6.26399100e-01
2.52073824e-01 -1.01142359e+00 -1.62112080e-02 9.72657502e-01
4.67648715e-01 -1.29510987e+00 1.56622119e-02 -5.67338169e-01
-8.43700111e-01 8.77923548e-01 7.50594974e-01 6.34609014e-02
8.19480002e-01 3.11703116e-01 4.27418828e-01 -3.34381372e-01
-9.04571116e-01 -5.02363324e-01 3.86087567e-01 6.83667302e-01
8.32524002e-01 6.72338754e-02 -3.99249911e-01 7.43388116e-01
-4.01502222e-01 -5.31109810e-01 3.79351050e-01 7.67229259e-01
-2.71474458e-02 -1.47050583e+00 2.16231555e-01 6.67632997e-01
-7.25734890e-01 -6.27237856e-01 -5.27829468e-01 6.24665797e-01
-3.41900676e-01 8.88173759e-01 -1.95173070e-01 -2.60447919e-01
3.71143639e-01 2.78447360e-01 1.23994760e-01 -9.29134309e-01
-6.89789712e-01 -7.66359717e-02 4.36908811e-01 -5.33340394e-01
-4.57209647e-01 -3.71364057e-01 -9.48361516e-01 1.89948857e-01
-6.51469767e-01 2.65358627e-01 8.50571930e-01 9.66575682e-01
4.04855132e-01 6.40263677e-01 5.88884830e-01 -3.25616509e-01
-1.76345944e-01 -9.80171025e-01 -3.96057427e-01 1.51850119e-01
-1.01458780e-01 -3.63257229e-01 -3.69119018e-01 -2.36974940e-01]
|
[11.440799713134766, 6.68006706237793]
|
270fcbcc-26a5-46ff-bc44-cdb0bca68e7a
|
wasserstein-auto-encoders-of-merge-trees-and
|
2307.02509
| null |
https://arxiv.org/abs/2307.02509v1
|
https://arxiv.org/pdf/2307.02509v1.pdf
|
Wasserstein Auto-Encoders of Merge Trees (and Persistence Diagrams)
|
This paper presents a computational framework for the Wasserstein auto-encoding of merge trees (MT-WAE), a novel extension of the classical auto-encoder neural network architecture to the Wasserstein metric space of merge trees. In contrast to traditional auto-encoders which operate on vectorized data, our formulation explicitly manipulates merge trees on their associated metric space at each layer of the network, resulting in superior accuracy and interpretability. Our novel neural network approach can be interpreted as a non-linear generalization of previous linear attempts [65] at merge tree encoding. It also trivially extends to persistence diagrams. Extensive experiments on public ensembles demonstrate the efficiency of our algorithms, with MT-WAE computations in the orders of minutes on average. We show the utility of our contributions in two applications adapted from previous work on merge tree encoding [65]. First, we apply MT-WAE to data reduction and reliably compress merge trees by concisely representing them with their coordinates in the final layer of our auto-encoder. Second, we document an application to dimensionality reduction, by exploiting the latent space of our auto-encoder, for the visual analysis of ensemble data. We illustrate the versatility of our framework by introducing two penalty terms, to help preserve in the latent space both the Wasserstein distances between merge trees, as well as their clusters. In both applications, quantitative experiments assess the relevance of our framework. Finally, we provide a C++ implementation that can be used for reproducibility.
|
['Julien Tierny', 'Mahieu Pont']
|
2023-07-05
| null | null | null | null |
['dimensionality-reduction']
|
['methodology']
|
[ 4.41543132e-01 1.98522970e-01 3.50530297e-01 -3.61168295e-01
-4.96727079e-01 -7.43825734e-01 6.50187373e-01 3.26783866e-01
-2.79492944e-01 2.59562820e-01 4.21594977e-01 -5.27995050e-01
-6.62247479e-01 -1.03611112e+00 -7.11485565e-01 -7.18703747e-01
-5.57343066e-01 4.69850272e-01 -2.75546432e-01 -1.64642930e-01
5.17189875e-02 5.48574269e-01 -1.57745790e+00 2.66298801e-01
6.56708837e-01 9.14405823e-01 -1.12678632e-01 1.07282460e+00
5.92680313e-02 6.22032642e-01 -4.26971346e-01 -5.45834839e-01
4.20380056e-01 -1.51979372e-01 -8.26590657e-01 3.85304191e-03
6.79536879e-01 -3.65754694e-01 -6.16984487e-01 8.63275647e-01
1.86508834e-01 5.83122037e-02 8.81338060e-01 -1.32148230e+00
-1.04426885e+00 5.56601465e-01 -2.25135773e-01 1.21311255e-01
-3.68404090e-02 1.80793628e-02 1.41731894e+00 -7.48678505e-01
5.58838665e-01 1.32631779e+00 1.01122880e+00 1.07299469e-01
-1.88171363e+00 -3.16024423e-01 -1.05895594e-01 3.23528320e-01
-1.51297641e+00 -3.67047966e-01 5.96964955e-01 -7.22616136e-01
1.11768699e+00 4.66656446e-01 6.26217425e-01 7.18212843e-01
2.95173317e-01 6.22969925e-01 8.89586389e-01 -5.32776177e-01
3.49957168e-01 -2.47497126e-01 3.15052390e-01 8.69365394e-01
1.99136019e-01 2.37662077e-01 -3.80211741e-01 -1.81040093e-01
6.83469117e-01 1.60607263e-01 5.61038367e-02 -1.16371930e+00
-1.18925381e+00 1.01702857e+00 6.37211740e-01 3.47916573e-01
-1.66383624e-01 5.79345822e-01 3.76428664e-01 3.49224299e-01
6.53160393e-01 2.95542330e-01 -5.17997555e-02 -2.60425210e-01
-8.93491685e-01 1.80454567e-01 7.43669271e-01 9.45429146e-01
7.51216233e-01 -7.76098594e-02 -1.99524611e-01 4.78767931e-01
1.25690892e-01 2.22969741e-01 3.47862512e-01 -1.24011195e+00
2.91099995e-01 6.38173997e-01 -3.08140129e-01 -1.03263581e+00
-2.50999719e-01 -3.16538692e-01 -1.16956329e+00 2.86039114e-01
9.74645168e-02 2.35013321e-01 -7.38601506e-01 1.80494916e+00
8.78694654e-02 2.70412955e-02 1.42012417e-01 2.70337671e-01
3.49126816e-01 5.02046525e-01 -2.91879773e-01 -2.79296115e-02
1.25069463e+00 -5.42980671e-01 -6.54802620e-01 1.65029719e-01
9.81130362e-01 -3.72094601e-01 9.32497144e-01 2.92106807e-01
-1.05119789e+00 -2.88048238e-01 -1.29040980e+00 -3.31321090e-01
-7.68271148e-01 -6.02755658e-02 8.65657628e-01 7.76601553e-01
-1.54217494e+00 1.16421115e+00 -1.06823659e+00 -5.33772230e-01
5.81577957e-01 5.53103745e-01 -5.35051644e-01 2.47617275e-01
-8.95737648e-01 7.94750452e-01 5.89523077e-01 7.20697641e-02
-4.78899598e-01 -3.77221048e-01 -1.01851380e+00 3.90380561e-01
-6.81593418e-02 -8.34411621e-01 1.24185896e+00 -4.70360637e-01
-1.10549295e+00 5.42651713e-01 -1.93204880e-01 -8.82056832e-01
2.54816532e-01 -1.78426102e-01 -9.04021934e-02 -1.45692557e-01
-1.05231084e-01 7.86643267e-01 5.17165065e-01 -9.10130918e-01
-5.27677119e-01 -3.57556373e-01 1.19917803e-01 4.41655517e-02
-7.02515244e-01 -4.57900137e-01 -4.05291468e-03 -8.64651263e-01
3.55299674e-02 -8.51125836e-01 -2.65153319e-01 1.49837479e-01
-2.10863709e-01 -3.99662733e-01 1.04488087e+00 -3.75040174e-01
1.53855789e+00 -2.22557449e+00 5.42536736e-01 1.28526852e-01
7.16090918e-01 -3.07629444e-02 -2.07948998e-01 6.83889508e-01
-5.58982730e-01 3.35466087e-01 -8.25430989e-01 -5.35780430e-01
3.88436913e-01 6.47419095e-01 -5.44624865e-01 4.82983857e-01
2.97163069e-01 1.02577007e+00 -8.16746831e-01 -3.43341798e-01
3.76297504e-01 5.85533261e-01 -6.65568531e-01 -3.07833795e-02
1.34908348e-01 -2.89454371e-01 1.47007659e-01 2.16521934e-01
5.02018154e-01 -2.07944959e-01 1.80085987e-01 -1.34366453e-01
-5.06642014e-02 2.61973619e-01 -1.15788746e+00 1.67989135e+00
-4.40420806e-01 1.05890262e+00 -1.62504986e-01 -9.56405818e-01
8.02384496e-01 1.24721773e-01 6.72427356e-01 -2.80086428e-01
-8.93572643e-02 1.50792738e-02 -8.15512985e-02 -8.97255838e-02
7.32930422e-01 -3.58322002e-02 -1.60955220e-01 8.22206438e-01
2.55455971e-01 -1.35777280e-01 4.57508802e-01 5.18565953e-01
1.20462716e+00 1.16615348e-01 4.29020286e-01 -2.81836718e-01
2.18481928e-01 -4.79300827e-01 2.35880148e-02 7.52488256e-01
-1.42655913e-02 4.91561443e-01 6.39819384e-01 -7.81100929e-01
-1.44375455e+00 -1.31861806e+00 -3.67087722e-01 9.65393186e-01
-4.47867721e-01 -1.07290471e+00 -8.67715299e-01 -5.78820348e-01
1.02799600e-02 8.73176217e-01 -8.98531795e-01 -2.78272748e-01
-5.16679823e-01 -8.29833329e-01 6.90969288e-01 7.28531539e-01
1.71797484e-01 -7.59789169e-01 -7.95932531e-01 1.90191746e-01
4.67408076e-02 -6.89175487e-01 -2.65208632e-01 6.76996350e-01
-1.04615200e+00 -9.52267826e-01 -3.12757999e-01 -4.76879209e-01
4.02811140e-01 1.09900713e-01 1.05632186e+00 -2.25924432e-01
-3.21300954e-01 4.60640490e-01 -1.21775456e-01 -2.37862214e-01
-5.61046660e-01 2.66426355e-01 2.40404621e-01 -1.90935433e-01
7.00516939e-01 -1.08738601e+00 -3.89366448e-01 -6.10813908e-02
-1.21227729e+00 2.48861790e-01 3.89972329e-01 8.57817709e-01
3.72983962e-01 2.52620578e-01 -8.55862275e-02 -7.52523839e-01
9.25675690e-01 -2.93581545e-01 -6.34656072e-01 1.19426049e-01
-8.39534521e-01 5.90246618e-01 3.72231692e-01 3.34119946e-02
-2.56580085e-01 2.02445090e-01 8.82115513e-02 -5.25782049e-01
8.48410279e-02 2.77783960e-01 1.50395274e-01 5.68891428e-02
5.47940612e-01 1.17420182e-01 1.62884802e-01 -4.44724411e-01
9.42015350e-01 6.70030832e-01 6.71569884e-01 -4.36421990e-01
9.48194444e-01 5.67090571e-01 2.64095336e-01 -3.88537467e-01
-5.59523046e-01 -1.68144673e-01 -1.14033151e+00 1.89474732e-01
8.10923040e-01 -5.66910386e-01 -9.17860150e-01 7.17927963e-02
-1.22765887e+00 -6.27629459e-02 -7.73381770e-01 6.72632530e-02
-8.47328544e-01 5.14055252e-01 -4.69871491e-01 -7.38623321e-01
-3.01994294e-01 -1.00170445e+00 1.36344230e+00 -3.77811790e-01
-5.07098913e-01 -1.30398607e+00 3.41356963e-01 -1.76941052e-01
3.54436159e-01 4.07992661e-01 1.25773692e+00 -5.89672387e-01
-6.01807058e-01 -5.57123981e-02 -1.37572899e-01 3.64060074e-01
8.17464218e-02 8.55504870e-02 -9.83700752e-01 -4.65292543e-01
-1.75993860e-01 -2.93999799e-02 1.31288147e+00 3.42088610e-01
1.28778613e+00 -5.68757355e-01 -3.44366252e-01 7.51604259e-01
1.23775613e+00 -8.31759796e-02 4.55543756e-01 3.40579271e-01
8.08243990e-01 6.37462676e-01 2.63521895e-02 4.42276984e-01
4.89547580e-01 6.65111899e-01 4.58752781e-01 -2.88331956e-02
6.79322612e-03 -3.12497467e-01 4.36686218e-01 1.21984661e+00
-6.14533573e-02 -9.94412974e-02 -8.58139217e-01 6.05669558e-01
-1.96298110e+00 -9.34268415e-01 -6.95872381e-02 2.26607013e+00
4.81955230e-01 -6.45103282e-04 1.38242692e-01 5.04101813e-01
4.31281298e-01 3.32390875e-01 -3.37055326e-01 -7.64030039e-01
-1.96965579e-02 5.14817655e-01 6.57235026e-01 6.89404666e-01
-1.37404895e+00 6.18319392e-01 7.40083647e+00 7.38331139e-01
-5.57624042e-01 1.90845151e-02 3.51900041e-01 -4.05870348e-01
-5.07133007e-01 2.64797449e-01 -4.12852854e-01 1.77646443e-01
1.21824932e+00 -1.99804619e-01 6.92364633e-01 7.58942783e-01
-2.72464186e-01 3.51236135e-01 -1.72231209e+00 8.29522371e-01
-4.90798317e-02 -1.37100101e+00 7.83400238e-02 4.58780438e-01
3.05537641e-01 -1.14316806e-01 2.89787740e-01 1.80920914e-01
8.70360076e-01 -1.24475503e+00 5.77239096e-01 4.26583499e-01
1.02786934e+00 -9.51145768e-01 5.08482456e-01 2.95315366e-02
-1.48587823e+00 -2.19708547e-01 -4.17729467e-01 -2.47106135e-01
1.40447497e-01 5.67587256e-01 -9.48207617e-01 5.44185102e-01
7.24639773e-01 9.77808535e-01 -8.57529938e-01 5.00274181e-01
1.17788007e-02 3.37470323e-01 -4.52168763e-01 1.24510325e-01
2.65912533e-01 -3.36184889e-01 7.41942823e-01 1.25880218e+00
4.70243126e-01 -1.99570015e-01 -2.43937716e-01 1.00207436e+00
1.64006595e-02 -2.85504282e-01 -1.17293525e+00 -2.74728626e-01
4.81748283e-01 1.07450783e+00 -3.68630737e-01 -4.46109682e-01
-2.16629412e-02 9.62990224e-01 6.51829958e-01 1.64361790e-01
-4.14380103e-01 -7.23057687e-01 6.26509190e-01 -1.19954824e-01
6.88492894e-01 -5.28376162e-01 -4.20822412e-01 -9.90234554e-01
4.17372920e-02 -7.06114173e-01 5.20316362e-01 -6.81967378e-01
-1.15641928e+00 6.01873398e-01 2.05189526e-01 -1.17314875e+00
-6.42509341e-01 -8.19159627e-01 -1.63779259e-01 6.52723789e-01
-9.24755514e-01 -1.03561974e+00 -5.21720015e-02 2.04245940e-01
2.63713419e-01 -1.59866557e-01 1.25385129e+00 1.00306749e-01
-5.56551516e-01 5.74794948e-01 6.43623292e-01 -1.97814386e-02
1.85642689e-01 -1.77641988e+00 1.07013738e+00 8.28131557e-01
7.05706656e-01 8.35971594e-01 6.67088985e-01 -3.05146247e-01
-1.23267210e+00 -1.06930864e+00 9.45828259e-01 -7.87229002e-01
9.14953232e-01 -8.55329514e-01 -9.62522566e-01 1.09772706e+00
3.21258605e-01 -2.18950018e-01 8.42675030e-01 4.66778398e-01
-4.76251513e-01 -5.48273548e-02 -7.57272124e-01 7.15893269e-01
1.32516813e+00 -7.35099077e-01 -6.07091665e-01 3.32291156e-01
8.70123029e-01 -1.97444960e-01 -1.19203269e+00 3.07911128e-01
6.44190073e-01 -1.23109376e+00 1.01897645e+00 -6.02528155e-01
4.77057219e-01 -4.40745384e-01 -4.98908907e-01 -1.41336846e+00
-7.70822763e-01 -6.34706855e-01 -5.23880959e-01 1.17723083e+00
-7.81720057e-02 -4.41625088e-01 5.64582467e-01 2.57369787e-01
1.14401340e-01 -9.41082180e-01 -1.16011584e+00 -6.23857260e-01
2.51678526e-01 -6.76213980e-01 8.51981580e-01 7.20024884e-01
-1.76423341e-02 3.26891840e-01 -3.40613842e-01 -7.99916778e-03
6.36111081e-01 -2.58679446e-02 7.53913939e-01 -1.40401828e+00
-3.96270216e-01 -7.76248097e-01 -8.20823967e-01 -1.13685155e+00
1.70851517e-02 -1.29576087e+00 -2.07736880e-01 -1.49641550e+00
2.08987996e-01 -2.34283403e-01 -4.21381116e-01 5.13191283e-01
1.27135053e-01 2.83741891e-01 2.96257406e-01 3.31836820e-01
-4.23041314e-01 8.27887177e-01 6.03550017e-01 -1.38166636e-01
1.73054431e-02 -3.63132596e-01 -7.58805931e-01 5.85989654e-01
3.63407761e-01 -6.07432127e-01 -3.52648109e-01 -4.78507459e-01
2.21381918e-01 -3.78933489e-01 5.92774987e-01 -1.09033692e+00
1.52071968e-01 3.52268100e-01 2.46891499e-01 -7.58732080e-01
2.89806902e-01 -9.19695079e-01 2.22578123e-01 4.99960393e-01
-4.60515469e-01 5.27733266e-01 1.02693401e-01 7.34909534e-01
-6.54629841e-02 -1.25547782e-01 4.14254516e-01 1.92297980e-01
-2.80507505e-01 1.80951715e-01 -4.30340528e-01 -3.05176049e-01
9.12006080e-01 -4.19397593e-01 -1.26094013e-01 -5.30295908e-01
-6.86907470e-01 1.94692075e-01 4.81385916e-01 3.26460570e-01
4.04155046e-01 -1.65587485e+00 -5.06670058e-01 3.72130543e-01
1.19097874e-01 1.17517047e-01 -1.06280539e-02 6.43809378e-01
-5.35983026e-01 5.99635482e-01 -1.55657113e-01 -7.30040491e-01
-1.31868434e+00 7.88882792e-01 2.95577884e-01 -4.35043484e-01
-7.11771607e-01 5.85634232e-01 3.94836366e-01 -4.33597982e-01
3.16536218e-01 -5.75541615e-01 4.64162044e-02 4.92960066e-02
4.16604549e-01 4.03331906e-01 6.87895948e-03 -6.22560322e-01
-2.84633279e-01 6.12212539e-01 -1.54382274e-01 -3.10577840e-01
1.53163397e+00 -2.13110577e-02 -3.14929456e-01 7.48899639e-01
1.33880520e+00 -2.40116924e-01 -1.28331554e+00 -2.78958887e-01
1.70504242e-01 -2.24135563e-01 -2.06088219e-02 -2.35154882e-01
-6.39840961e-01 1.27690351e+00 8.83243799e-01 4.95431602e-01
1.15753150e+00 -8.69921446e-02 4.73541200e-01 6.90491736e-01
-1.16707206e-01 -7.85018146e-01 -1.05946235e-01 5.29852092e-01
8.11615705e-01 -7.62079477e-01 7.00685233e-02 -1.10099450e-01
-2.94792980e-01 1.30588031e+00 4.88940440e-02 -1.30136326e-01
5.60178518e-01 3.94571602e-01 -1.69953436e-01 -3.73662084e-01
-9.78449464e-01 9.89825930e-03 1.91392958e-01 5.70239961e-01
4.51047063e-01 2.38443837e-01 4.42843921e-02 1.14086255e-01
-8.70789826e-01 -1.18723825e-01 5.30046523e-01 1.00210238e+00
-3.23936343e-01 -1.17449486e+00 -2.32254103e-01 4.60185468e-01
-7.28578167e-03 -3.36935043e-01 -4.51195359e-01 8.14668357e-01
1.15418524e-01 5.47538817e-01 6.01843894e-01 -5.94065607e-01
2.80033797e-01 3.25414091e-01 4.40423608e-01 -3.75679851e-01
-3.26572061e-01 -2.51472712e-01 -8.83156732e-02 -4.07804847e-01
-3.05823326e-01 -7.67756641e-01 -7.39937007e-01 -6.49748862e-01
-1.47144556e-01 6.15496039e-02 5.74323952e-01 6.33567989e-01
5.87660491e-01 6.62537456e-01 5.68741322e-01 -6.59274697e-01
-6.20844007e-01 -1.02876794e+00 -5.19430697e-01 4.65037048e-01
6.21484041e-01 -6.68271482e-01 -1.83818489e-01 1.27301961e-01]
|
[8.117996215820312, 4.049621105194092]
|
953b1846-3ae1-4dfc-ad39-8b7958e3cce0
|
the-uncertainty-based-retrieval-framework-for
| null | null |
https://aclanthology.org/2022.lt4hala-1.25
|
https://aclanthology.org/2022.lt4hala-1.25.pdf
|
The Uncertainty-based Retrieval Framework for Ancient Chinese CWS and POS
|
Automatic analysis for modern Chinese has greatly improved the accuracy of text mining in related fields, but the study of ancient Chinese is still relatively rare. Ancient text division and lexical annotation are important parts of classical literature comprehension, and previous studies have tried to construct auxiliary dictionary and other fused knowledge to improve the performance. In this paper, we propose a framework for ancient Chinese Word Segmentation and Part-of-Speech Tagging that makes a twofold effort: on the one hand, we try to capture the wordhood semantics; on the other hand, we re-predict the uncertain samples of baseline model by introducing external knowledge. The performance of our architecture outperforms pre-trained BERT with CRF and existing tools such as Jiayan.
|
['Zhichen Ren', 'Pengyu Wang']
| null | null | null | null |
lt4hala-lrec-2022-6
|
['chinese-word-segmentation']
|
['natural-language-processing']
|
[ 2.92471759e-02 -9.30153802e-02 -4.84083951e-01 -3.88403088e-01
-6.95049345e-01 -4.83018577e-01 3.92338663e-01 1.39915673e-02
-1.02037752e+00 1.11771011e+00 3.15337658e-01 -6.58167660e-01
3.96187395e-01 -7.39712536e-01 -2.95894951e-01 -5.08635461e-01
3.87522638e-01 8.40632200e-01 3.80772114e-01 -2.33192772e-01
5.70939422e-01 -1.55485077e-02 -9.05216694e-01 8.10752809e-02
1.39309967e+00 5.60534596e-01 6.68502271e-01 2.41716221e-01
-6.61478043e-01 6.29030347e-01 -7.91601896e-01 -7.29653060e-01
-2.04319030e-01 -4.91314143e-01 -9.91500199e-01 -7.35926181e-02
-4.89387184e-01 -1.17100276e-01 -8.96448568e-02 1.27320087e+00
4.64823872e-01 2.42824461e-02 5.63192368e-01 -2.99369603e-01
-8.55875790e-01 1.47788942e+00 -6.12082005e-01 3.63404721e-01
7.93034956e-03 -3.81648719e-01 1.10426843e+00 -8.12053561e-01
6.51051819e-01 9.77068961e-01 3.85825634e-01 6.08375251e-01
-6.43477380e-01 -7.76270926e-01 4.90680307e-01 4.54730839e-01
-1.48652267e+00 -2.69469947e-01 7.07222819e-01 -8.41476172e-02
8.15257907e-01 -4.29694876e-02 7.66330361e-01 1.14955080e+00
9.25691798e-02 1.30201375e+00 1.01388764e+00 -7.44668961e-01
2.00473949e-01 1.57699659e-01 3.53079736e-01 2.05437332e-01
1.42805219e-01 -5.08497953e-01 -1.35453075e-01 2.84237295e-01
6.18290663e-01 -1.76493019e-01 -1.56026006e-01 7.57562816e-01
-1.00979030e+00 8.35207820e-01 -2.73168921e-01 9.51588035e-01
1.01290625e-02 -3.07922393e-01 3.56729537e-01 -1.60442188e-01
6.37109876e-01 4.96998101e-01 -7.83647776e-01 -3.92444879e-01
-1.17011392e+00 4.68871370e-02 8.57437730e-01 1.09288526e+00
5.69146395e-01 -8.17984119e-02 -3.10872365e-02 9.37122524e-01
1.03024155e-01 3.78826350e-01 1.04761350e+00 -3.80334437e-01
5.07601261e-01 4.63059813e-01 -2.93336362e-01 -4.91095483e-01
-2.95568854e-01 -3.79327714e-01 -5.60734093e-01 -6.98867559e-01
2.47657776e-01 -3.32450449e-01 -1.34910488e+00 1.09215486e+00
-3.17202918e-02 -4.75470200e-02 -3.84390093e-02 6.95293903e-01
6.27256572e-01 6.35271430e-01 3.99531752e-01 -3.63815427e-01
1.48719704e+00 -1.01436555e+00 -1.26135099e+00 -6.07525051e-01
6.10263884e-01 -1.06602991e+00 9.71759856e-01 4.76961553e-01
-7.96619415e-01 -4.69612122e-01 -9.34610724e-01 -2.90453643e-01
-4.42583531e-01 4.74323809e-01 7.26572037e-01 8.87802124e-01
-3.14679682e-01 6.15843892e-01 -1.05109632e+00 -1.15122506e-02
5.07452309e-01 1.21088594e-01 1.22312225e-01 -5.94229773e-02
-1.51749635e+00 9.92249191e-01 1.04115033e+00 1.70251027e-01
-4.67040151e-01 -3.93660635e-01 -6.74149096e-01 -8.50282088e-02
7.23552883e-01 -4.55197766e-02 1.18051565e+00 -9.48930740e-01
-1.64373076e+00 7.24286437e-01 -2.37947971e-01 -2.96030402e-01
3.29174697e-01 -6.97274327e-01 -6.30221367e-01 -1.76984787e-01
3.44567001e-01 2.65779734e-01 3.62667233e-01 -9.27483678e-01
-8.53479743e-01 -5.05250394e-01 -4.08573300e-01 1.02938071e-01
-3.44582170e-01 2.09906548e-01 -1.07627046e+00 -1.04467452e+00
3.61857228e-02 -6.65910184e-01 -5.53628564e-01 -7.40983129e-01
-3.70541751e-01 -5.09867787e-01 6.27026737e-01 -9.81960237e-01
1.67838013e+00 -1.77525795e+00 -1.84987321e-01 2.35025838e-01
-1.51346892e-01 5.13876498e-01 8.46136287e-02 1.37819156e-01
3.29942197e-01 6.29419804e-01 -4.18953389e-01 -2.53473639e-01
-2.99173653e-01 6.50351167e-01 -2.22563118e-01 6.41633272e-02
2.17499003e-01 1.03662217e+00 -9.36164320e-01 -1.05433714e+00
-2.56425273e-02 3.20555344e-02 -2.38845721e-01 -6.49177432e-02
-5.66962898e-01 4.55014348e-01 -9.42593455e-01 9.05963004e-01
6.15208924e-01 -4.73840721e-02 5.58885455e-01 2.66531497e-01
-2.61196107e-01 7.53840506e-01 -1.07470667e+00 1.89921939e+00
-2.40426943e-01 2.70653725e-01 -4.20299143e-01 -1.12437141e+00
6.76011324e-01 2.76114315e-01 2.48366266e-01 -8.64969492e-01
6.72023475e-01 4.67597544e-01 2.32722536e-01 -5.13373792e-01
9.49134052e-01 -1.61728978e-01 -1.57151699e-01 1.89390481e-01
9.28885639e-02 1.70818530e-02 4.23326820e-01 1.61694974e-01
6.79985583e-01 7.22903371e-01 6.43613875e-01 -2.22260043e-01
6.08440697e-01 3.78865093e-01 8.94959927e-01 5.49258351e-01
2.28499487e-01 5.64033270e-01 3.02503169e-01 -1.03796326e-01
-8.59555542e-01 -6.42140865e-01 -4.34942216e-01 1.04121351e+00
2.02434570e-01 -7.26932049e-01 -7.79873848e-01 -1.03440952e+00
-4.67089117e-01 1.07814848e+00 -4.60621566e-01 3.08341384e-01
-1.14305699e+00 -1.16165566e+00 8.50535750e-01 8.14627051e-01
6.05789840e-01 -1.08914757e+00 -2.09957734e-01 6.56171978e-01
-5.46294570e-01 -1.32162452e+00 -3.22411269e-01 4.26837325e-01
-7.24184573e-01 -8.86335254e-01 -7.55390644e-01 -8.64524543e-01
3.02684635e-01 -1.40768111e-01 1.08183742e+00 2.27825195e-01
-3.00087058e-03 -4.97636110e-01 -9.56587911e-01 -7.58261144e-01
-1.52433529e-01 4.97995406e-01 -3.43982905e-01 -4.85492110e-01
8.58750999e-01 -3.31472486e-01 -1.80578858e-01 1.32219747e-01
-8.54399264e-01 -1.60487160e-01 8.05777013e-01 6.64952815e-01
6.79812193e-01 3.59002911e-02 6.45525277e-01 -1.39750731e+00
4.44569319e-01 -4.01201338e-01 -3.53491187e-01 5.03417432e-01
-6.53942943e-01 1.29081026e-01 6.21310532e-01 -5.09239197e-01
-1.62930322e+00 4.69399877e-02 -8.90004218e-01 3.45383197e-01
-8.51556361e-02 8.73988569e-01 -6.27491832e-01 4.77422804e-01
2.69773066e-01 3.64215583e-01 -7.74004519e-01 -8.58309746e-01
4.97867107e-01 9.98127580e-01 5.04847586e-01 -9.77997482e-01
4.61974978e-01 1.28265679e-01 -4.14395183e-01 -1.02728999e+00
-1.18648434e+00 -4.93660301e-01 -1.06724477e+00 2.78370619e-01
9.73489463e-01 -9.45286930e-01 -2.16043741e-01 2.90313661e-01
-1.08883941e+00 -1.33517340e-01 -6.86203465e-02 6.48914635e-01
1.46778852e-01 8.43159616e-01 -6.17570877e-01 -8.71296823e-01
-3.75735432e-01 -7.27384031e-01 9.27689910e-01 4.61284429e-01
-2.37145066e-01 -9.25659597e-01 4.80637662e-02 2.70470142e-01
6.06184863e-02 -3.11682016e-01 9.46209669e-01 -1.11506605e+00
-3.22823435e-01 6.56793118e-02 -6.20150566e-02 3.20906103e-01
2.05849800e-02 -1.05152279e-01 -8.39054286e-01 2.81570226e-01
5.95526099e-02 -1.64742991e-01 1.09257913e+00 2.70097882e-01
1.15324879e+00 -4.47099144e-03 -4.91426617e-01 3.03967506e-01
1.13512564e+00 4.99117047e-01 9.23736453e-01 6.45067692e-01
6.36689425e-01 4.26939994e-01 8.99182260e-01 4.26303267e-01
4.69437510e-01 2.34905571e-01 -1.98686138e-01 1.14573628e-01
7.68576711e-02 -1.49862722e-01 1.64416716e-01 1.19891047e+00
-3.02100569e-01 -2.89255589e-01 -9.91099417e-01 8.58349860e-01
-1.88804781e+00 -5.77756226e-01 -1.77327067e-01 1.81331408e+00
1.47407162e+00 5.04198074e-01 -1.79604098e-01 -7.43047940e-03
5.33635020e-01 -5.79253323e-02 -1.42832130e-01 1.44859239e-01
-4.55310136e-01 7.36340702e-01 6.16613984e-01 2.79315740e-01
-9.85929787e-01 1.64440060e+00 5.60924053e+00 1.37180924e+00
-9.83026206e-01 2.28554100e-01 4.26239669e-01 3.38884383e-01
-4.59559649e-01 4.86916453e-01 -1.21864796e+00 5.62704384e-01
7.63871849e-01 5.61823063e-02 -1.50198087e-01 8.87516856e-01
-2.92706728e-01 -2.82371521e-01 -7.74737298e-01 5.42465806e-01
2.23108858e-01 -1.06422567e+00 5.04604056e-02 1.12884548e-02
7.42848158e-01 5.04613444e-02 -5.51120579e-01 5.51062584e-01
3.36436182e-01 -9.83272731e-01 7.51263440e-01 3.83842647e-01
5.06828845e-01 -6.65663540e-01 1.12498093e+00 5.83971918e-01
-1.08884919e+00 4.42846775e-01 -4.47422713e-01 -4.53754468e-03
5.49685180e-01 6.39307559e-01 -6.74225152e-01 8.52858007e-01
5.55818737e-01 6.66764975e-01 -7.13373303e-01 8.48308861e-01
-7.80711651e-01 1.20770133e+00 -2.14776531e-01 -3.25396299e-01
2.89110422e-01 -3.67267370e-01 2.20688835e-01 1.46110010e+00
2.20556960e-01 2.77331233e-01 4.07773018e-01 6.90485001e-01
-1.63553178e-01 5.98378897e-01 1.45673081e-01 -4.15230900e-01
4.60616857e-01 1.01763678e+00 -1.18322456e+00 -5.32822490e-01
-5.71365058e-01 1.01651132e+00 3.93497318e-01 1.31397486e-01
-7.57692695e-01 -6.25910342e-01 5.71374297e-02 -1.08015224e-01
4.47684646e-01 -4.76433128e-01 -8.01891208e-01 -1.53233683e+00
-5.77024817e-02 -7.68712163e-01 4.82342422e-01 -1.07584104e-01
-1.31980789e+00 5.38065076e-01 -3.50158244e-01 -6.52822793e-01
-4.46370952e-02 -6.40692592e-01 -5.03660738e-01 9.26125944e-01
-1.67079031e+00 -1.23444653e+00 2.97437102e-01 3.81064981e-01
9.68746245e-01 -3.09568383e-02 8.24216306e-01 4.51028645e-01
-7.49852479e-01 4.04927492e-01 1.71126753e-01 6.01990342e-01
7.41368353e-01 -1.24364352e+00 4.88355398e-01 1.15164125e+00
3.41603637e-01 8.22665870e-01 4.09286022e-01 -1.24603534e+00
-8.05262744e-01 -8.46550226e-01 1.26047790e+00 -2.98614770e-01
7.24744976e-01 -3.16266924e-01 -1.24644816e+00 8.07980835e-01
2.72863984e-01 -6.55495524e-01 8.50383222e-01 4.80778277e-01
1.13918088e-01 3.56766135e-01 -4.79456574e-01 6.86135650e-01
7.64801741e-01 -3.05581033e-01 -1.07964897e+00 8.15308392e-02
5.61272025e-01 -2.33200908e-01 -7.48221397e-01 2.87211090e-01
5.31891823e-01 -2.03040317e-01 4.83755410e-01 -4.84115928e-01
3.68193895e-01 -2.33390540e-01 1.64645948e-02 -1.00680161e+00
-2.14455396e-01 -4.29996282e-01 2.48144224e-01 1.61648774e+00
6.97198451e-01 -1.99971512e-01 8.79559219e-01 3.78841251e-01
-4.59975064e-01 -4.82492149e-01 -7.20730424e-01 -6.22220993e-01
4.59376365e-01 -6.51539326e-01 4.45402086e-01 9.12147522e-01
1.92498162e-01 8.66217077e-01 -4.98769611e-01 -1.66902646e-01
2.36528367e-01 4.38161977e-02 2.12934762e-01 -1.24446797e+00
-5.26267886e-02 -4.12014782e-01 2.30552509e-01 -1.53542495e+00
4.18746084e-01 -7.88605571e-01 3.01649809e-01 -1.51724625e+00
1.37741208e-01 -6.12947166e-01 -6.45673275e-02 5.63264668e-01
-6.56674325e-01 6.90537542e-02 -7.72488043e-02 1.85458437e-01
-5.73106349e-01 7.11561918e-01 1.23090208e+00 1.36366144e-01
-4.10698205e-01 7.16008395e-02 -8.26268792e-01 8.68651509e-01
8.27135623e-01 -7.20715940e-01 1.23200566e-01 -5.61002910e-01
3.54844838e-01 -2.52786487e-01 -3.83136272e-01 -5.35243213e-01
3.02238286e-01 -2.80826032e-01 6.55475497e-01 -9.55380499e-01
-8.91953632e-02 -6.35395467e-01 -4.25501227e-01 1.55911282e-01
-4.09918949e-02 -7.83284083e-02 3.19044925e-02 4.26688701e-01
-2.42357388e-01 -8.16583633e-01 4.95867610e-01 -5.53436756e-01
-1.03521276e+00 -6.83917627e-02 -6.72067344e-01 4.29685444e-01
7.53851473e-01 -1.85487568e-02 -1.50816157e-01 1.29413158e-01
-3.98270309e-01 3.74753058e-01 1.33690059e-01 4.69401091e-01
3.39158118e-01 -7.63675213e-01 -6.22495890e-01 -7.97941983e-02
-2.06160247e-01 1.85206428e-01 -8.83994177e-02 6.13003790e-01
-5.12750089e-01 3.39969575e-01 -1.02146737e-01 -8.75388384e-02
-8.95277023e-01 3.77253503e-01 -1.79700121e-01 -5.62067688e-01
-5.19650936e-01 6.83564842e-01 -1.56792387e-01 -3.40625554e-01
2.18160942e-01 -1.19705386e-01 -7.16638327e-01 1.48154482e-01
4.00326997e-01 2.86090463e-01 4.49638143e-02 -5.47158182e-01
-4.22992378e-01 4.43552911e-01 -4.97536898e-01 -1.12999797e-01
1.28994727e+00 -2.12615654e-01 -2.84188718e-01 3.58613551e-01
5.80441058e-01 4.90697950e-01 -4.94933963e-01 -3.69816333e-01
6.56556606e-01 -3.28789711e-01 1.21804520e-01 -9.43318963e-01
-6.73427939e-01 9.34314132e-01 -6.99985698e-02 -3.24333876e-01
9.05244648e-01 2.22956296e-02 1.12078381e+00 3.31562787e-01
1.47176474e-01 -1.69452178e+00 -2.69861668e-01 1.07790816e+00
1.98340535e-01 -1.11896813e+00 3.17624599e-01 -4.72667009e-01
-1.02196252e+00 1.24075592e+00 6.86119378e-01 1.88812405e-01
8.08727980e-01 4.24719185e-01 1.25643551e-01 -1.37834633e-02
-2.38851875e-01 -6.46984458e-01 3.73060405e-01 3.44061524e-01
8.32692206e-01 -6.40563592e-02 -1.14906585e+00 1.36427331e+00
-4.28090751e-01 -8.94268826e-02 3.45363528e-01 1.08076370e+00
-6.58833683e-01 -1.65725577e+00 -3.36501971e-02 3.42529446e-01
-1.20456362e+00 -6.13143921e-01 -3.65972370e-01 8.80064070e-01
4.39970374e-01 8.61389160e-01 -4.68229465e-02 -1.39738962e-01
8.66444781e-02 2.00912401e-01 3.42593819e-01 -8.36134672e-01
-5.58010578e-01 7.66512275e-01 2.22706348e-01 4.19425219e-02
-5.55921972e-01 -7.13854790e-01 -1.51944685e+00 1.49448216e-01
-7.96320140e-01 5.76712191e-01 5.29079437e-01 1.40923774e+00
-2.65477955e-01 6.32077873e-01 9.09322426e-02 -4.00315464e-01
-2.56251097e-01 -1.38809872e+00 -6.85320377e-01 -6.66223690e-02
-3.02559644e-01 -2.83374250e-01 4.09682430e-02 1.93979472e-01]
|
[10.016508102416992, 10.108893394470215]
|
1e1b9f40-405c-4625-b576-c5a09aa7e63f
|
entity-agnostic-representation-learning-for
|
2302.01849
| null |
https://arxiv.org/abs/2302.01849v1
|
https://arxiv.org/pdf/2302.01849v1.pdf
|
Entity-Agnostic Representation Learning for Parameter-Efficient Knowledge Graph Embedding
|
We propose an entity-agnostic representation learning method for handling the problem of inefficient parameter storage costs brought by embedding knowledge graphs. Conventional knowledge graph embedding methods map elements in a knowledge graph, including entities and relations, into continuous vector spaces by assigning them one or multiple specific embeddings (i.e., vector representations). Thus the number of embedding parameters increases linearly as the growth of knowledge graphs. In our proposed model, Entity-Agnostic Representation Learning (EARL), we only learn the embeddings for a small set of entities and refer to them as reserved entities. To obtain the embeddings for the full set of entities, we encode their distinguishable information from their connected relations, k-nearest reserved entities, and multi-hop neighbors. We learn universal and entity-agnostic encoders for transforming distinguishable information into entity embeddings. This approach allows our proposed EARL to have a static, efficient, and lower parameter count than conventional knowledge graph embedding methods. Experimental results show that EARL uses fewer parameters and performs better on link prediction tasks than baselines, reflecting its parameter efficiency.
|
['Huajun Chen', 'Jeff Z. Pan', 'Yang Gao', 'Yushan Zhu', 'Zhen Yao', 'Wen Zhang', 'Mingyang Chen']
|
2023-02-03
| null | null | null | null |
['knowledge-graph-embedding', 'entity-embeddings']
|
['graphs', 'methodology']
|
[-4.00553077e-01 5.42187393e-01 -7.83734500e-01 -1.76978260e-01
-3.05049658e-01 -7.40235448e-01 2.82569170e-01 4.66884732e-01
-3.92421067e-01 5.43639660e-01 2.78653115e-01 -3.08323741e-01
-3.19811881e-01 -1.27892792e+00 -6.67016566e-01 -3.23093653e-01
-4.41314012e-01 5.54133654e-01 2.99162537e-01 -9.92817711e-03
-1.27508149e-01 4.70738530e-01 -1.13724935e+00 -1.42786905e-01
7.30301738e-01 8.07062030e-01 -1.82131097e-01 4.75927472e-01
-3.73744071e-01 8.02433908e-01 -4.09111202e-01 -7.46024370e-01
8.67943987e-02 1.72566567e-02 -1.06825089e+00 -3.91211480e-01
1.83301985e-01 -4.15082693e-01 -1.17126966e+00 9.04634118e-01
2.51617044e-01 2.65046388e-01 7.13924229e-01 -1.53190160e+00
-1.82343102e+00 1.01631188e+00 -4.09061402e-01 2.59504825e-01
2.11127222e-01 -5.00081241e-01 1.56718588e+00 -9.10918772e-01
7.41006434e-01 1.01552165e+00 5.78495145e-01 3.10838491e-01
-1.27558172e+00 -4.39756483e-01 2.42668509e-01 5.99978805e-01
-1.79633343e+00 -2.71890044e-01 7.08061993e-01 -4.91440892e-01
1.27081859e+00 4.49511036e-02 5.33633530e-01 6.27168715e-01
-3.15060705e-01 4.18978542e-01 1.30725607e-01 -4.22062784e-01
6.86052367e-02 4.19431448e-01 5.77250540e-01 8.85053992e-01
8.67413819e-01 -2.79513925e-01 -3.67374361e-01 -3.65882874e-01
7.03926146e-01 1.00966178e-01 -4.31525201e-01 -8.67038250e-01
-1.10794413e+00 1.00743079e+00 9.23604548e-01 2.47897178e-01
-2.75017202e-01 6.81546092e-01 4.13816065e-01 4.97334629e-01
1.39735073e-01 4.27965373e-01 -5.72287917e-01 1.54727548e-01
-8.95408988e-02 -3.13099504e-01 9.13589478e-01 1.30322897e+00
1.12895203e+00 -1.46941856e-01 -8.35067220e-03 7.32921064e-01
3.33005577e-01 2.94610620e-01 4.32203323e-01 -5.65784931e-01
5.45772135e-01 1.06292486e+00 4.55738865e-02 -1.45500481e+00
-2.57380694e-01 -1.96502626e-01 -5.04429817e-01 -5.25311172e-01
-1.06288381e-01 -7.37125278e-02 -8.06012452e-01 1.81970942e+00
3.79738420e-01 2.65608221e-01 3.81099671e-01 5.57324767e-01
9.90444422e-01 7.40047872e-01 3.04716416e-02 1.23255789e-01
1.44489479e+00 -1.08247030e+00 -7.15361059e-01 -5.60742579e-02
1.07306647e+00 -9.20256972e-03 6.80232525e-01 -5.88282704e-01
-6.84380293e-01 -2.62456983e-01 -1.03522480e+00 -3.43849689e-01
-1.12337017e+00 1.26671731e-01 1.03130007e+00 6.32467389e-01
-1.10814118e+00 4.53957766e-01 -8.14974189e-01 -2.41937548e-01
3.33669066e-01 5.27466595e-01 -8.09031129e-01 8.60146061e-02
-1.53034389e+00 8.41713667e-01 8.42133164e-01 -1.22917593e-01
-2.73656636e-01 -6.63506925e-01 -1.45268738e+00 5.39191723e-01
3.95289153e-01 -6.69748724e-01 5.60525775e-01 -2.86414683e-01
-1.01489532e+00 5.12392998e-01 -9.23857987e-02 -3.47704321e-01
-4.14124757e-01 -4.55860347e-02 -7.36727834e-01 2.17838570e-01
-1.16567291e-01 2.89567918e-01 5.46580136e-01 -1.19978499e+00
-3.49329054e-01 -1.56377032e-01 5.20061851e-01 1.88544225e-02
-1.01845610e+00 -2.29158387e-01 -7.60644794e-01 -3.44137490e-01
1.02082692e-01 -8.46198082e-01 4.87004519e-02 1.51069537e-01
-4.96548355e-01 -4.11314487e-01 7.98216760e-01 -6.28919542e-01
1.58985376e+00 -2.28311443e+00 4.05409992e-01 4.52191859e-01
7.06040144e-01 3.70040387e-01 -2.85711229e-01 7.89625347e-01
-1.29342109e-01 4.00183469e-01 9.28951427e-02 -7.98373669e-03
3.27237844e-01 5.44643223e-01 -1.83697820e-01 3.36918354e-01
3.19389403e-01 1.42292857e+00 -1.13218999e+00 -6.13085747e-01
2.28682104e-02 7.19618440e-01 -5.28289020e-01 9.71641578e-03
1.12533145e-01 -3.46129149e-01 -6.46657526e-01 5.23948550e-01
4.67317343e-01 -7.52506793e-01 6.16891801e-01 -4.77877915e-01
4.68813807e-01 2.22864151e-01 -1.34383821e+00 1.34202933e+00
-5.79083323e-01 6.75114572e-01 -4.33458894e-01 -1.07151127e+00
9.07489538e-01 3.28255922e-01 1.95041806e-01 -3.36610466e-01
-2.08099857e-01 9.97147039e-02 -2.25813255e-01 -2.94973344e-01
7.76109934e-01 3.99658561e-01 -1.42071858e-01 5.04548192e-01
3.76465142e-01 7.77509451e-01 7.75801614e-02 6.27240062e-01
1.31371474e+00 -3.17258328e-01 3.88095111e-01 2.54653603e-01
3.52189988e-01 -3.83733958e-01 5.49625516e-01 4.12732244e-01
-6.04714453e-02 -9.06341746e-02 7.09308922e-01 -3.72354239e-01
-9.11049485e-01 -1.20959055e+00 5.92227466e-02 1.07005596e+00
3.50114256e-01 -9.09187436e-01 -2.00340942e-01 -8.33674133e-01
6.21491969e-01 4.64110106e-01 -8.47917199e-01 -6.67556107e-01
-5.17273366e-01 -1.90892085e-01 4.49992180e-01 9.33193028e-01
-4.86355536e-02 -6.47428572e-01 3.83334123e-02 1.07235901e-01
1.45552218e-01 -1.07500589e+00 -5.52019119e-01 2.11080626e-01
-6.15456343e-01 -1.12724745e+00 -4.92153555e-01 -1.18670559e+00
1.02160656e+00 2.13348880e-01 9.22254205e-01 1.42037645e-01
-1.63575038e-01 5.55496633e-01 -6.29398823e-01 2.03279078e-01
-5.70752705e-03 1.73225284e-01 1.04217850e-01 -9.23120230e-02
5.37462115e-01 -6.48146749e-01 -5.74500203e-01 9.94961262e-02
-7.51407146e-01 -4.85336721e-01 5.35183549e-01 1.01122880e+00
6.66090488e-01 3.24290134e-02 7.05379903e-01 -1.00160635e+00
7.08997548e-01 -7.65568078e-01 -3.97623688e-01 6.48194790e-01
-7.62617707e-01 5.22909224e-01 6.67381644e-01 -4.12823319e-01
-4.41912651e-01 -1.96468711e-01 5.64668238e-01 -6.67000413e-01
4.09645557e-01 9.37708199e-01 -3.10082920e-02 -2.76732624e-01
4.30203944e-01 1.80963889e-01 -3.54679585e-01 -4.03258860e-01
1.13399708e+00 6.75921202e-01 4.09348577e-01 -4.26485866e-01
1.10245252e+00 8.71440917e-02 -1.76861167e-01 -6.38181806e-01
-4.86749589e-01 -5.47452748e-01 -7.77488053e-01 3.61057460e-01
6.04174733e-01 -1.02508152e+00 -6.55378520e-01 -2.56928086e-01
-1.12952542e+00 1.02763705e-01 -5.83703160e-01 5.03534555e-01
-1.13899745e-01 5.06017148e-01 -6.39815271e-01 -5.92428565e-01
-2.46923640e-01 -6.70738995e-01 7.60711312e-01 1.11272283e-01
-7.90769309e-02 -1.34998095e+00 1.41707107e-01 5.58709614e-02
3.48098397e-01 1.06058024e-01 1.44233906e+00 -1.02718079e+00
-6.54744029e-01 -5.49009562e-01 -5.78663647e-01 1.22173756e-01
2.85016060e-01 -8.76786858e-02 -5.35124242e-01 -2.73968667e-01
-1.03310001e+00 -2.06166878e-01 8.30447733e-01 -2.58039117e-01
1.12786233e+00 -6.06269360e-01 -8.57037961e-01 8.01801026e-01
1.49583495e+00 -8.55962634e-02 3.99732918e-01 1.73589155e-01
1.03854966e+00 1.68968201e-01 1.65169001e-01 3.29913527e-01
8.27889144e-01 6.26497924e-01 2.17314780e-01 8.80684778e-02
-1.68109044e-01 -6.65976286e-01 1.12375543e-01 1.19887912e+00
2.10081935e-02 -3.75661194e-01 -7.25298464e-01 1.02466023e+00
-1.94022894e+00 -8.37259531e-01 1.96563438e-01 2.04883909e+00
9.18296337e-01 -2.37003863e-01 -3.65154371e-02 -9.28722881e-03
7.88188100e-01 1.62395149e-01 -6.04802489e-01 -4.40822989e-01
6.11193068e-02 2.05200016e-01 7.54791975e-01 4.35740680e-01
-9.04411972e-01 1.03134930e+00 6.11545134e+00 3.44184577e-01
-6.75018728e-01 1.14023402e-01 -1.77976489e-01 7.10960925e-02
-7.71467984e-01 1.09029591e-01 -5.88573575e-01 4.49226737e-01
1.06286490e+00 -7.20520675e-01 5.20283759e-01 1.02077174e+00
-7.09639370e-01 6.63009644e-01 -1.35032308e+00 1.00076449e+00
-4.77512889e-02 -1.58059180e+00 3.20792168e-01 5.87166436e-02
5.59246719e-01 -5.85183837e-02 -1.27586335e-01 5.89459717e-01
4.81098413e-01 -9.82631385e-01 1.80981785e-01 4.68668461e-01
9.52199936e-01 -7.95515537e-01 8.28311265e-01 -2.91492313e-01
-1.60831082e+00 -2.16522917e-01 -6.63323820e-01 3.16762984e-01
1.34267896e-01 3.34581971e-01 -6.96525931e-01 8.02833855e-01
4.67124999e-01 7.53818572e-01 -4.84948844e-01 8.11837912e-01
-3.55297774e-01 3.01849365e-01 -2.65797645e-01 -1.29344538e-01
-1.82873327e-02 -1.48441136e-01 1.83061212e-01 1.12625492e+00
1.78782761e-01 1.71074331e-01 1.94864720e-02 7.71536648e-01
-7.10062981e-01 8.19787979e-02 -7.82375932e-01 -7.11044908e-01
1.30934274e+00 1.15321326e+00 -3.56704563e-01 -3.58191788e-01
-6.75956786e-01 1.12328792e+00 1.02997541e+00 5.03764808e-01
-7.69864321e-01 -1.12321687e+00 8.68657351e-01 -1.70987234e-01
7.83230364e-01 -3.88511240e-01 2.72711873e-01 -1.21413970e+00
2.13953644e-01 -7.04186633e-02 6.27224267e-01 -4.81507331e-01
-1.35142899e+00 5.05665243e-01 -3.70359346e-02 -7.54199326e-01
2.13792566e-02 -6.44711494e-01 -4.00761992e-01 7.15780973e-01
-1.77653372e+00 -1.23364627e+00 -1.28864646e-01 5.93502343e-01
-4.73149955e-01 -1.63474694e-01 1.16335118e+00 4.84454155e-01
-6.91980660e-01 1.18276513e+00 4.06860560e-01 7.21839249e-01
4.50496644e-01 -1.37810361e+00 4.99364704e-01 3.96769404e-01
4.85875487e-01 9.23290908e-01 7.38730207e-02 -4.41362590e-01
-1.82426393e+00 -1.18935764e+00 1.16031301e+00 -4.86677438e-01
1.13914490e+00 -3.60360861e-01 -1.07234383e+00 1.32056725e+00
-1.60147592e-01 6.21835589e-01 1.16701245e+00 5.45804799e-01
-8.88889015e-01 -4.02449518e-02 -9.35691416e-01 5.81158638e-01
1.19709194e+00 -1.03479218e+00 -7.01448083e-01 3.36605012e-01
1.33797276e+00 -5.27809821e-02 -1.50342238e+00 3.51579785e-02
3.70648742e-01 -2.07807660e-01 1.19315684e+00 -1.04265249e+00
8.61554891e-02 -3.66110831e-01 -2.01633096e-01 -1.31496906e+00
-7.17034817e-01 -3.73978317e-01 -1.19954777e+00 1.20151711e+00
6.23260617e-01 -1.04976511e+00 7.42028058e-01 5.39675653e-01
1.76060796e-01 -8.31353009e-01 -9.04790640e-01 -9.80160356e-01
-1.57507211e-01 1.99647531e-01 8.97920132e-01 1.46271348e+00
5.12835503e-01 5.13742566e-01 -2.27351859e-01 6.52377546e-01
5.02605021e-01 4.61089253e-01 5.27646005e-01 -1.36426723e+00
-2.00884387e-01 -1.72011793e-01 -1.14699507e+00 -9.18889284e-01
3.95313442e-01 -1.47450280e+00 -6.20820820e-01 -1.83994818e+00
1.97959647e-01 -7.03712940e-01 -7.27501392e-01 9.31612015e-01
-1.84467956e-01 -1.46392226e-01 -7.93684274e-02 1.36187017e-01
-7.25617111e-01 7.45986342e-01 7.42266715e-01 -2.32808873e-01
-1.23665780e-01 -6.77855015e-01 -9.01493669e-01 2.89037019e-01
5.22127151e-01 -3.75132859e-01 -6.12285018e-01 -6.79070592e-01
4.21072811e-01 -2.05526143e-01 1.83138654e-01 -5.07303238e-01
4.23311263e-01 -1.14316558e-02 1.80347160e-01 -1.70340329e-01
4.40449506e-01 -9.14995432e-01 1.46794260e-01 1.40334010e-01
-3.61977696e-01 -1.00408360e-01 -9.70101799e-04 1.07041562e+00
-2.37578318e-01 -1.21313639e-01 2.53331155e-01 3.61848533e-01
-1.05893135e+00 6.43158078e-01 3.11238497e-01 6.07092120e-02
1.25962102e+00 -2.59013474e-01 -7.33582973e-01 -2.42496207e-01
-9.19366002e-01 4.65345949e-01 4.50134099e-01 6.05744779e-01
7.98300207e-01 -1.79036081e+00 -2.22708002e-01 2.36760583e-02
6.70816302e-01 -2.98581310e-02 3.38038281e-02 4.51362401e-01
-3.09694171e-01 4.90733594e-01 -4.86834033e-04 6.06738292e-02
-9.41294611e-01 9.34760749e-01 1.76355168e-01 -1.49677813e-01
-8.45343471e-01 9.48669076e-01 -4.24106158e-02 -6.22913241e-01
2.45255977e-01 -1.78057268e-01 -2.87006617e-01 1.39454737e-01
3.76372904e-01 4.59361702e-01 -1.53856441e-01 -5.43851614e-01
-5.79270065e-01 5.59291005e-01 -2.44247690e-01 5.41606307e-01
1.34026039e+00 6.43506879e-03 -2.32062653e-01 3.59123200e-01
1.58816648e+00 7.59472772e-02 -6.60489857e-01 -6.56640828e-01
1.10312432e-01 -5.76514781e-01 1.45960331e-01 -3.59375507e-01
-1.22233057e+00 4.57298636e-01 2.41571620e-01 3.34397107e-01
5.64699173e-01 5.17475605e-01 8.44493389e-01 8.06851923e-01
4.79008555e-01 -9.36248481e-01 -1.40060544e-01 3.21737498e-01
4.87902999e-01 -9.55049574e-01 -4.76892032e-02 -6.04960203e-01
-4.61788595e-01 1.05515313e+00 5.37813008e-01 1.36749642e-02
8.78777742e-01 -6.77957535e-02 -4.15904492e-01 -5.26875854e-01
-8.02474380e-01 -2.95607924e-01 3.16032439e-01 9.56101060e-01
2.32628658e-01 3.21729571e-01 -1.09228775e-01 7.69659162e-01
3.79502811e-02 -2.49436617e-01 4.47032034e-01 9.72088277e-01
-4.34275657e-01 -1.08955443e+00 3.02090526e-01 6.65348589e-01
3.65815312e-02 -1.71769649e-01 -4.58831131e-01 7.89805353e-01
-1.93976894e-01 6.84034407e-01 2.84945369e-01 -6.45162821e-01
4.19258505e-01 2.27456063e-01 2.35320345e-01 -8.16515505e-01
6.66047931e-02 -9.73336935e-01 1.87273815e-01 -4.19859171e-01
8.10966548e-03 -9.37043726e-02 -1.48164237e+00 -4.79047388e-01
-8.10338318e-01 4.64878559e-01 4.85805899e-01 4.34389502e-01
9.94953275e-01 6.03489220e-01 6.08153641e-01 -1.96512952e-01
-5.09500444e-01 -5.01663268e-01 -8.61981392e-01 3.83318096e-01
2.28025451e-01 -9.22298491e-01 -3.05116266e-01 -2.06221461e-01]
|
[8.754498481750488, 7.9013566970825195]
|
dd360a10-fed2-4661-91a0-0cdf62546cee
|
quantitative-argument-summarization-and
|
2010.05369
| null |
https://arxiv.org/abs/2010.05369v1
|
https://arxiv.org/pdf/2010.05369v1.pdf
|
Quantitative Argument Summarization and Beyond: Cross-Domain Key Point Analysis
|
When summarizing a collection of views, arguments or opinions on some topic, it is often desirable not only to extract the most salient points, but also to quantify their prevalence. Work on multi-document summarization has traditionally focused on creating textual summaries, which lack this quantitative aspect. Recent work has proposed to summarize arguments by mapping them to a small set of expert-generated key points, where the salience of each key point corresponds to the number of its matching arguments. The current work advances key point analysis in two important respects: first, we develop a method for automatic extraction of key points, which enables fully automatic analysis, and is shown to achieve performance comparable to a human expert. Second, we demonstrate that the applicability of key point analysis goes well beyond argumentation data. Using models trained on publicly available argumentation datasets, we achieve promising results in two additional domains: municipal surveys and user reviews. An additional contribution is an in-depth evaluation of argument-to-key point matching models, where we substantially outperform previous results.
|
['Noam Slonim', 'Dan Lahav', 'Roni Friedman', 'Lilach Eden', 'Yoav Kantor', 'Roy Bar-Haim']
|
2020-10-11
| null |
https://aclanthology.org/2020.emnlp-main.3
|
https://aclanthology.org/2020.emnlp-main.3.pdf
|
emnlp-2020-11
|
['key-point-matching']
|
['natural-language-processing']
|
[ 4.05993283e-01 6.36721909e-01 -7.05286503e-01 -1.71380192e-01
-1.48031139e+00 -8.49677563e-01 1.23072445e+00 1.27994967e+00
-2.30422974e-01 7.79997349e-01 1.25383115e+00 -3.68345529e-01
-2.84740418e-01 -6.15482628e-01 -4.17339474e-01 -1.38105184e-01
3.57636809e-01 4.88192379e-01 1.52488440e-01 -4.92794007e-01
1.03718090e+00 3.79942134e-02 -1.47688222e+00 8.07579339e-01
1.18049169e+00 5.97474635e-01 -3.82370561e-01 7.33065724e-01
-3.72001886e-01 8.13659668e-01 -1.16446149e+00 -9.23773229e-01
-3.12946409e-01 -3.05543274e-01 -1.00278902e+00 -3.22163612e-01
7.73559213e-01 -1.74146861e-01 5.76971710e-01 9.43436861e-01
4.34655875e-01 -2.50271499e-01 1.10094512e+00 -1.10111010e+00
-4.46964979e-01 9.87541020e-01 -6.28380597e-01 2.39680484e-01
7.13493824e-01 -6.41938746e-01 1.71359611e+00 -5.39690077e-01
9.34463739e-01 1.28653133e+00 7.11877704e-01 1.30655959e-01
-1.16918659e+00 -1.46316379e-01 4.20904279e-01 -1.44533768e-01
-4.24244076e-01 -4.78769213e-01 1.00583720e+00 -4.45368409e-01
8.67065251e-01 4.67261851e-01 7.77034104e-01 8.90206158e-01
6.74750730e-02 1.02026093e+00 1.10331130e+00 -5.78490496e-01
2.50415564e-01 6.37220442e-02 5.89546978e-01 2.07531586e-01
6.24518812e-01 -8.12246978e-01 -4.80004191e-01 -9.00675297e-01
2.21133884e-02 -2.76627183e-01 -9.09296647e-02 -2.52056092e-01
-1.27141345e+00 1.09883749e+00 7.85349980e-02 4.43396181e-01
-6.28573000e-01 -7.74431601e-02 7.58129478e-01 1.20934054e-01
9.36495781e-01 7.43419170e-01 -3.69754642e-01 -4.29447293e-01
-1.22914362e+00 8.40571821e-01 1.19636524e+00 5.31985402e-01
3.87112051e-01 -6.17807746e-01 -4.61088330e-01 1.01326394e+00
3.98692898e-02 5.19470870e-01 1.66184410e-01 -1.28054917e+00
1.04433966e+00 1.03335726e+00 4.64735806e-01 -1.58638096e+00
-2.19494462e-01 -2.89525002e-01 -4.80491102e-01 -8.36996213e-02
1.36146799e-01 -2.48759702e-01 -1.97233945e-01 1.47442198e+00
2.30570063e-01 -8.11281919e-01 3.30354244e-01 3.04696888e-01
1.19558680e+00 4.60493863e-01 -4.13684174e-02 -4.32345152e-01
1.58706331e+00 -6.43780291e-01 -8.26098263e-01 -4.17731851e-02
7.78197765e-01 -9.82544482e-01 8.61813247e-01 3.18656832e-01
-1.46157026e+00 1.00574911e-01 -1.00872326e+00 -1.67898595e-01
-4.42675769e-01 3.01526129e-01 5.57548881e-01 4.28112179e-01
-8.12987626e-01 4.46406037e-01 -2.37345934e-01 -4.54580426e-01
4.22930956e-01 -1.42406449e-01 -1.90639853e-01 4.88263756e-01
-1.08271480e+00 1.05586791e+00 1.06177464e-01 -4.24716502e-01
1.29958466e-02 -7.95293331e-01 -8.60674918e-01 1.35337472e-01
4.71177638e-01 -7.82375097e-01 1.48727810e+00 -5.50146639e-01
-1.00736129e+00 9.00618315e-01 -5.04900992e-01 -5.50311208e-01
5.27177513e-01 -3.89074236e-01 -2.92210609e-01 3.49573702e-01
8.26332569e-01 4.87906218e-01 3.87235373e-01 -1.37075746e+00
-9.03720021e-01 -2.46647179e-01 4.09048408e-01 2.92090535e-01
-6.09290004e-01 3.05826247e-01 1.13217838e-01 -7.12476492e-01
-6.29891157e-02 -4.07039404e-01 -1.06176525e-01 -5.38748920e-01
-4.83533144e-01 -7.33017564e-01 6.79434657e-01 -5.58359742e-01
1.61115491e+00 -1.49250615e+00 1.66369244e-01 -7.87291769e-03
4.15495247e-01 1.89648345e-01 1.84441775e-01 1.08081543e+00
1.95229068e-01 5.16535640e-01 -2.75231242e-01 -1.70719564e-01
1.51080072e-01 -1.15413800e-01 -8.24857831e-01 1.14402119e-02
7.74786919e-02 1.02755904e+00 -1.05929947e+00 -7.77570963e-01
-6.75612018e-02 6.35148063e-02 -5.60326517e-01 -3.56318086e-01
-2.46910095e-01 -2.17098892e-01 -6.50602877e-01 5.41638017e-01
2.57457465e-01 -3.55066180e-01 1.25831664e-01 -4.63294804e-01
-3.58678550e-01 7.13232875e-01 -9.12399292e-01 1.22373772e+00
-3.03804994e-01 5.67980230e-01 9.49700549e-02 -9.78000343e-01
7.79466689e-01 4.13038433e-01 6.31215334e-01 -5.15992641e-01
5.43211261e-03 4.59876478e-01 -4.29421037e-01 -2.55622745e-01
1.00994623e+00 2.52576768e-01 -5.76250732e-01 9.39980030e-01
-4.05115604e-01 -6.38268411e-01 7.81287372e-01 8.71237159e-01
9.94745612e-01 -3.82668585e-01 7.80760646e-01 -3.96672577e-01
5.23483038e-01 4.54905987e-01 1.17220394e-01 8.74018788e-01
1.26451820e-01 5.05828798e-01 1.17399907e+00 -2.66540825e-01
-1.18592346e+00 -6.26535892e-01 -7.04669133e-02 7.84433782e-01
-1.07532091e-01 -1.04725480e+00 -7.58360922e-01 -9.03945088e-01
1.86626807e-01 9.22892392e-01 -9.10216749e-01 4.34939176e-01
-6.69284046e-01 -5.71946263e-01 3.41818571e-01 3.42624217e-01
3.08956683e-01 -7.93628931e-01 -9.50418890e-01 4.12986755e-01
-7.38364279e-01 -9.11387205e-01 -6.74868003e-02 -1.74650058e-01
-8.50131333e-01 -1.38776791e+00 -8.68271530e-01 -2.75960356e-01
5.52609444e-01 4.65158165e-01 1.34163070e+00 3.09204403e-02
2.33382300e-01 4.81417775e-01 -4.90543216e-01 -8.23805869e-01
-6.01587057e-01 4.27153498e-01 -3.69693160e-01 -5.35472572e-01
2.37191856e-01 -4.64667886e-01 -3.58525336e-01 -5.84145337e-02
-7.61811376e-01 4.51786146e-02 3.41338873e-01 8.04945469e-01
1.99087873e-01 -4.78557259e-01 9.17479277e-01 -1.17401326e+00
1.63597262e+00 -3.99775565e-01 -4.88210954e-02 4.62107569e-01
-5.10864496e-01 1.24441087e-01 3.94961327e-01 4.98732068e-02
-1.03939354e+00 -7.77195334e-01 -1.86422274e-01 6.72029614e-01
1.08702600e-01 9.82304156e-01 3.87847900e-01 6.41885400e-01
8.22515965e-01 -4.27741557e-01 -2.25176811e-02 -2.29398355e-01
8.39665234e-01 7.75017262e-01 4.11751390e-01 -9.53526974e-01
5.11097431e-01 7.03850567e-01 -1.57169059e-01 -8.52764606e-01
-1.36435032e+00 -6.36624634e-01 -5.16103625e-01 -9.84763652e-02
3.72657925e-01 -6.32365406e-01 -5.56597829e-01 -1.69572785e-01
-1.41048884e+00 1.56609580e-01 -4.30820227e-01 2.17258945e-01
-4.87949908e-01 6.66886628e-01 -2.73397774e-01 -7.50946343e-01
-7.07933605e-01 -6.26728952e-01 1.31281233e+00 1.63834408e-01
-1.06637883e+00 -1.16783047e+00 3.94628227e-01 5.79325378e-01
3.36021513e-01 5.03763437e-01 1.07878244e+00 -9.05613184e-01
2.15658650e-01 -2.98082948e-01 -2.81390041e-01 -1.23727787e-02
2.97561884e-01 3.37485075e-01 -4.79742378e-01 -7.26451874e-02
-6.96157366e-02 -4.26757425e-01 1.00733435e+00 6.92143619e-01
5.34797728e-01 -5.93651593e-01 -6.01080656e-01 -2.87534624e-01
8.98760021e-01 -2.24843711e-01 2.82298416e-01 7.59197474e-01
2.83617675e-01 1.02978349e+00 7.33438075e-01 4.06553596e-01
7.07346022e-01 5.03003538e-01 2.88107634e-01 -5.96791096e-02
2.20172741e-02 -2.28370473e-01 2.08622398e-04 8.76799464e-01
-2.12562636e-01 -3.38756382e-01 -7.80951858e-01 1.04144812e+00
-2.27671599e+00 -1.44426656e+00 -3.25140983e-01 1.74734902e+00
6.90872014e-01 3.15751612e-01 2.74750769e-01 3.84764731e-01
5.34719110e-01 5.85053325e-01 -3.14766951e-02 -6.76998496e-01
-3.80218208e-01 1.54651077e-02 7.36320913e-02 5.53665221e-01
-9.86615479e-01 6.04960918e-01 6.83467054e+00 5.80809593e-01
-5.87363720e-01 -1.91155970e-01 5.22115946e-01 1.16233058e-01
-8.60980749e-01 2.62252420e-01 -8.12295675e-01 1.52812362e-01
6.17141128e-01 -5.05986691e-01 -4.35993940e-01 7.43225694e-01
3.03339869e-01 -3.72459114e-01 -8.95752847e-01 4.50767964e-01
4.04377282e-01 -1.84717250e+00 3.39658767e-01 1.34162247e-01
1.00801694e+00 -3.97903174e-01 -5.25237285e-02 -5.20551875e-02
6.50153577e-01 -6.97222233e-01 7.37643659e-01 2.13598296e-01
3.89615595e-01 -6.76713884e-01 8.55792165e-01 2.51959831e-01
-9.05686021e-01 -4.03208360e-02 -7.09554031e-02 -2.14533344e-01
5.13972342e-01 6.14249349e-01 -8.36473525e-01 8.28638911e-01
4.27746385e-01 1.01351130e+00 -5.71881175e-01 7.12889850e-01
-5.31238496e-01 6.11205697e-01 -1.80333942e-01 -4.45339292e-01
4.42956448e-01 5.24386875e-02 7.51962245e-01 1.32013655e+00
3.58206123e-01 1.70367077e-01 8.96963198e-03 4.25658435e-01
-1.50489822e-01 5.22466123e-01 -7.33540058e-01 -1.14041775e-01
5.63744664e-01 1.40151882e+00 -7.75524795e-01 -7.70987988e-01
-2.87513614e-01 2.72178054e-01 3.27703446e-01 9.42700729e-02
-3.34445059e-01 -4.38179195e-01 1.68913558e-01 7.09711239e-02
1.96399465e-01 9.15840361e-03 -5.62487483e-01 -1.17114866e+00
3.03847402e-01 -1.06181693e+00 6.39554620e-01 -6.80835903e-01
-1.21109080e+00 2.36609995e-01 3.94883186e-01 -1.02792537e+00
-5.96953094e-01 -2.19115764e-01 -8.98906469e-01 6.85042143e-01
-1.36171055e+00 -1.29220521e+00 3.96813676e-02 -1.37651786e-01
5.42353511e-01 3.81182544e-02 7.86507547e-01 -1.81345567e-01
-9.43545923e-02 1.02618851e-01 6.24333881e-03 -4.80458215e-02
7.60296941e-01 -1.39245427e+00 5.81670284e-01 5.98866463e-01
1.64591774e-01 8.86473656e-01 1.03119600e+00 -8.09456468e-01
-6.20771766e-01 -3.48789304e-01 1.47383106e+00 -8.51997197e-01
8.55428338e-01 8.54264013e-03 -7.32698798e-01 3.69825333e-01
6.73895895e-01 -8.64111364e-01 7.92496502e-01 6.58268154e-01
-3.46182942e-01 2.13626564e-01 -9.08941805e-01 7.47967303e-01
6.72032058e-01 -3.20597231e-01 -1.44148016e+00 2.81962126e-01
3.77328902e-01 -1.91261858e-01 -8.32639813e-01 3.62889260e-01
6.98122859e-01 -9.50641036e-01 9.29064155e-01 -6.74529552e-01
1.02890229e+00 -8.40578005e-02 -1.31321073e-01 -1.48619914e+00
2.39280276e-02 -5.69085956e-01 -1.56544343e-01 1.48098850e+00
6.34918511e-01 -4.51080650e-01 6.60879552e-01 3.99429321e-01
-2.24436913e-02 -7.61153638e-01 -5.97921133e-01 -1.63033605e-01
2.89974600e-01 -1.76500622e-02 6.39075339e-01 9.51970577e-01
5.20923734e-01 6.63659632e-01 -1.36161059e-01 -3.08521152e-01
5.39306581e-01 6.27955854e-01 1.00074017e+00 -1.84587932e+00
3.42267126e-01 -9.80817318e-01 1.51329726e-01 -8.49177778e-01
1.13115482e-01 -5.50478518e-01 -4.27256703e-01 -2.51329160e+00
5.98533154e-01 -1.62681416e-01 -4.36941795e-02 1.28615573e-01
-4.50240225e-01 -1.75246850e-01 1.77547988e-02 3.98537993e-01
-7.57391512e-01 2.47214988e-01 9.80658829e-01 -3.01589280e-01
-3.17468077e-01 -2.34522261e-02 -1.29422235e+00 9.21417713e-01
6.37404501e-01 -3.63415807e-01 -3.25242758e-01 -2.21508175e-01
9.08491969e-01 3.76559757e-02 3.07177812e-01 -4.39641476e-01
3.35233361e-01 -1.90354586e-01 2.74933372e-02 -1.19904053e+00
8.41248259e-02 -2.09654227e-01 -3.19262266e-01 2.70894468e-01
-8.27716887e-01 3.69836241e-01 6.86444044e-02 4.58833545e-01
-3.38443160e-01 -5.64177513e-01 6.01287074e-02 -1.95098743e-01
-1.23088427e-01 -4.22117829e-01 -2.26208106e-01 4.96764839e-01
5.40750861e-01 -9.06253830e-02 -9.82062757e-01 -6.54687464e-01
-2.34603435e-01 2.85981297e-01 4.69902277e-01 3.08419526e-01
3.52467716e-01 -1.16532958e+00 -1.16557777e+00 -6.81554019e-01
3.64210099e-01 -3.28006357e-01 -6.98603168e-02 7.87357926e-01
-1.53842375e-01 6.49354815e-01 9.70070343e-03 -3.81838471e-01
-1.50032055e+00 1.70782968e-01 -3.94358754e-01 -6.70182049e-01
-6.16927803e-01 2.73059934e-01 -3.07956129e-01 -2.63244003e-01
-1.19042039e-01 -4.72500980e-01 -8.91433120e-01 9.88914490e-01
6.17895067e-01 4.88669962e-01 3.51037681e-02 -7.12523580e-01
-2.67854214e-01 7.24475801e-01 -4.29869890e-01 -4.65825409e-01
1.59356034e+00 -7.08941147e-02 -3.54558915e-01 4.80740786e-01
8.51893604e-01 7.52093136e-01 -7.25134015e-01 -1.28251225e-01
3.65533412e-01 -1.44329756e-01 -2.62051195e-01 -7.93974459e-01
-5.04376665e-02 5.53304315e-01 -5.06079376e-01 1.00513101e+00
5.56045771e-01 2.12521911e-01 5.80576539e-01 4.93197352e-01
-8.54792297e-02 -1.35606956e+00 2.50738919e-01 4.49089766e-01
1.04914379e+00 -1.26513183e+00 5.95584810e-01 -3.41633141e-01
-7.08632648e-01 1.15300441e+00 -1.28393292e-01 7.56532848e-02
1.72235534e-01 8.86030793e-02 1.77022576e-01 -9.08207595e-01
-8.67976010e-01 7.85515010e-02 4.82485235e-01 2.57409155e-01
7.36837029e-01 -2.51860134e-02 -9.98874426e-01 4.36031729e-01
-5.50999939e-01 -3.81647646e-01 8.06326032e-01 1.02714682e+00
-7.04301655e-01 -1.12589681e+00 -3.21478784e-01 8.40415955e-01
-8.34072769e-01 -9.20470059e-02 -1.03232837e+00 8.03402841e-01
-6.17091298e-01 1.35702980e+00 -1.53196380e-01 3.41284066e-01
6.33506000e-01 -1.22854903e-01 2.22163856e-01 -7.31530964e-01
-8.71338367e-01 -6.88591227e-02 8.44568431e-01 -1.17114998e-01
-9.55818236e-01 -9.31054652e-01 -9.21040952e-01 -1.95748344e-01
-4.39354926e-01 8.09505284e-01 6.93020523e-01 9.24915135e-01
3.94669801e-01 5.18888891e-01 2.75704294e-01 -5.92038095e-01
-5.45226276e-01 -9.89926934e-01 3.69631052e-02 6.40361428e-01
3.83845687e-01 -5.78502715e-01 -3.40055794e-01 -1.05354935e-01]
|
[12.151368141174316, 9.560551643371582]
|
dfbbbe9b-1641-4b23-836f-d8fdef0a0c49
|
a-semi-supervised-object-detection-algorithm
|
2306.04834
| null |
https://arxiv.org/abs/2306.04834v1
|
https://arxiv.org/pdf/2306.04834v1.pdf
|
A Semi-supervised Object Detection Algorithm for Underwater Imagery
|
Detection of artificial objects from underwater imagery gathered by Autonomous Underwater Vehicles (AUVs) is a key requirement for many subsea applications. Real-world AUV image datasets tend to be very large and unlabelled. Furthermore, such datasets are typically imbalanced, containing few instances of objects of interest, particularly when searching for unusual objects in a scene. It is therefore, difficult to fit models capable of reliably detecting these objects. Given these factors, we propose to treat artificial objects as anomalies and detect them through a semi-supervised framework based on Variational Autoencoders (VAEs). We develop a method which clusters image data in a learned low-dimensional latent space and extracts images that are likely to contain anomalous features. We also devise an anomaly score based on extracting poorly reconstructed regions of an image. We demonstrate that by applying both methods on large image datasets, human operators can be shown candidate anomalous samples with a low false positive rate to identify objects of interest. We apply our approach to real seafloor imagery gathered by an AUV and evaluate its sensitivity to the dimensionality of the latent representation used by the VAE. We evaluate the precision-recall tradeoff and demonstrate that by choosing an appropriate latent dimensionality and threshold, we are able to achieve an average precision of 0.64 on unlabelled datasets.
|
['Stefan B. Williams', 'Oscar Pizarro', 'Suraj Bijjahalli']
|
2023-06-07
| null | null | null | null |
['semi-supervised-object-detection']
|
['computer-vision']
|
[ 2.92133957e-01 1.08924761e-01 6.68378413e-01 -4.14638907e-01
-1.02345657e+00 -4.78627235e-01 4.02287960e-01 2.81318843e-01
-7.24417686e-01 4.71229583e-01 -2.83482343e-01 1.79805905e-01
-2.29825988e-01 -9.14842010e-01 -9.80261743e-01 -1.06095123e+00
-4.99158561e-01 4.92980599e-01 4.76392448e-01 -4.69579361e-02
4.38928455e-01 6.02626443e-01 -2.26336408e+00 1.34453058e-01
8.68328750e-01 1.00958633e+00 1.71157241e-01 8.86893451e-01
1.20606340e-01 4.52919573e-01 -8.43031526e-01 -1.94180220e-01
5.92438757e-01 -3.18616539e-01 -4.01969671e-01 4.45222706e-01
7.06202507e-01 -4.58421826e-01 8.20732415e-02 1.45405185e+00
3.47958058e-01 1.13424867e-01 9.25896108e-01 -1.21399605e+00
1.60666481e-01 2.29054559e-02 -3.22444260e-01 5.26840270e-01
-6.50605038e-02 2.53177851e-01 9.90885258e-01 -9.23693597e-01
5.92156231e-01 9.06641185e-01 5.96025705e-01 6.01645887e-01
-1.31327713e+00 -3.25640291e-01 9.09052510e-03 1.37858436e-01
-1.17343748e+00 -6.21390164e-01 6.53465807e-01 -6.19603097e-01
7.54640639e-01 1.80258974e-01 5.68467140e-01 9.34019208e-01
2.09340006e-01 7.15333223e-01 5.85666060e-01 -2.31572181e-01
6.02812350e-01 1.29485190e-01 -2.07901821e-01 5.35561025e-01
5.64392447e-01 4.39633541e-02 -4.27332044e-01 -4.78831917e-01
2.38102362e-01 1.84509128e-01 -2.39588410e-01 -5.16792357e-01
-8.20575058e-01 1.10961843e+00 8.70194808e-02 -2.04193056e-01
-6.88289642e-01 -8.62099305e-02 4.68057334e-01 5.30507982e-01
5.74582577e-01 5.00129580e-01 -3.05579841e-01 1.61669385e-02
-7.80216873e-01 3.96543205e-01 6.42653048e-01 6.50478721e-01
7.48854101e-01 1.67417109e-01 3.19531977e-01 8.15605819e-01
4.02891099e-01 5.40338159e-01 4.50422585e-01 -9.45236087e-01
2.32603438e-02 7.78447449e-01 2.56238490e-01 -8.15852225e-01
-3.06271732e-01 -1.00180149e-01 -3.81562382e-01 7.55995512e-01
1.84607238e-01 -2.06527919e-01 -9.62314963e-01 1.22637963e+00
3.46843928e-01 -3.54537717e-03 7.02276111e-01 8.66050065e-01
5.50624251e-01 7.46592283e-01 -2.49649808e-01 -3.65968287e-01
1.09518647e+00 -2.59539157e-01 -4.93181825e-01 -3.86744380e-01
8.60141218e-01 -2.19655186e-01 8.67993653e-01 6.79423988e-01
-7.87363350e-01 -1.62679836e-01 -1.14836526e+00 4.12890106e-01
-1.81736246e-01 -3.96460705e-02 1.99247718e-01 4.38899368e-01
-7.13553190e-01 6.61288798e-01 -1.13960099e+00 -2.80616075e-01
5.25613189e-01 2.65834361e-01 -5.02893925e-01 8.51263255e-02
-6.71866417e-01 5.68349123e-01 4.92480904e-01 2.54274815e-01
-1.36227930e+00 -2.23307967e-01 -1.11023808e+00 -1.89612657e-01
2.47316897e-01 1.86883032e-01 1.01456058e+00 -1.07018495e+00
-6.88960075e-01 9.15691435e-01 5.12738489e-02 -8.13914061e-01
7.12124646e-01 -3.29574227e-01 -4.01087523e-01 3.32671702e-01
2.24581242e-01 3.15728575e-01 1.20915258e+00 -1.62222433e+00
-1.08718669e+00 -5.37930548e-01 -3.09258401e-01 7.68143982e-02
-3.50484192e-01 -1.25025630e-01 -7.13147083e-03 -3.22209537e-01
5.96498907e-01 -9.03113663e-01 -3.27100962e-01 2.20016852e-01
-5.94040044e-02 -1.52722284e-01 1.07225835e+00 -3.53705674e-01
6.54653788e-01 -2.41357970e+00 -2.19580228e-03 3.01394582e-01
1.73110560e-01 2.11744204e-01 4.39438522e-02 2.53735721e-01
3.71659666e-01 -1.84091255e-01 -7.67422140e-01 -3.05707961e-01
-2.95636624e-01 8.30493152e-01 -4.41901863e-01 8.72248113e-01
4.19589281e-01 1.52725637e-01 -1.05721188e+00 -4.12389457e-01
1.15136802e-02 3.17645483e-02 -6.24141574e-01 5.26833475e-01
-1.34616390e-01 1.38925448e-01 -2.56457776e-01 8.78778279e-01
6.31911874e-01 2.01768339e-01 1.76061895e-02 1.31057858e-01
-1.00825422e-01 -2.51861393e-01 -1.23256731e+00 1.05397701e+00
-1.80239975e-02 7.97449708e-01 1.60510883e-01 -1.15000391e+00
1.18751538e+00 1.35481834e-01 3.25679004e-01 -3.16006362e-01
-4.09502015e-02 3.61582339e-01 -1.12681441e-01 -1.01472294e+00
7.75810838e-01 -2.95765817e-01 2.13714629e-01 1.29096940e-01
5.64575680e-02 4.28443737e-02 1.56284273e-01 2.15186983e-01
1.22848344e+00 -2.54420459e-01 2.30301898e-02 -3.01381648e-01
1.68624505e-01 2.09062234e-01 8.33344162e-01 9.86995995e-01
-3.41184437e-01 6.40801907e-01 7.21064746e-01 -7.40148842e-01
-1.18157125e+00 -1.01854706e+00 -4.99394536e-01 7.28964746e-01
2.81581283e-01 6.56618029e-02 -3.27408075e-01 -7.16864884e-01
6.94664642e-02 5.24834931e-01 -7.73981333e-01 -2.68129945e-01
-1.99049965e-01 -1.09101760e+00 5.48534036e-01 3.87108147e-01
2.02105492e-02 -1.30791771e+00 -1.27088547e+00 2.69130677e-01
-1.19695058e-02 -9.99114871e-01 5.63303649e-01 4.65759754e-01
-1.16274512e+00 -1.10993218e+00 -3.31311911e-01 -5.48474610e-01
1.06892395e+00 7.66590536e-02 1.01553023e+00 2.68930048e-01
-4.72748786e-01 3.96796137e-01 -7.77697146e-01 -5.47854602e-01
-7.02666759e-01 -6.17067635e-01 6.19200110e-01 2.36356959e-01
6.04618967e-01 -2.74479508e-01 -4.89237905e-01 3.62349659e-01
-1.19279885e+00 -7.72245705e-01 5.54801702e-01 1.08143580e+00
6.01237655e-01 2.06684783e-01 3.35897356e-01 -7.26452470e-01
1.73294470e-01 -7.86267102e-01 -9.49398816e-01 -1.35859534e-01
-4.75846529e-01 7.62649700e-02 3.18620563e-01 -3.74424636e-01
-5.02069116e-01 3.06066871e-01 -4.08385508e-02 -6.82937860e-01
-3.96363497e-01 2.95952559e-01 2.10281849e-01 -9.31254700e-02
9.30202663e-01 3.90278488e-01 2.61778831e-01 -2.10629821e-01
-4.37244415e-01 8.20189178e-01 4.88191605e-01 -2.70841300e-01
6.85776830e-01 8.33844125e-01 -1.82576209e-01 -1.40173876e+00
-6.35214090e-01 -5.84967673e-01 -4.49481755e-01 -5.12610137e-01
6.94376349e-01 -1.07355130e+00 -4.31339711e-01 3.42558175e-01
-6.52387261e-01 -1.64196223e-01 -3.23340476e-01 5.80542266e-01
-3.73065412e-01 5.45357049e-01 -2.38986656e-01 -1.39537156e+00
-1.37584433e-01 -1.27336943e+00 1.19209194e+00 -4.93115075e-02
1.01015419e-01 -4.48858202e-01 -2.46839356e-02 6.85193837e-02
3.67646269e-03 4.57720816e-01 3.05135459e-01 -9.84684587e-01
-3.68618429e-01 -4.70522165e-01 4.99582104e-02 6.03570342e-01
6.36456767e-03 2.81472027e-01 -1.20013595e+00 -5.38702130e-01
2.13497385e-01 -4.99226928e-01 1.13477755e+00 2.59488106e-01
8.68352473e-01 -3.30858350e-01 -1.17556959e-01 4.82702881e-01
1.39505899e+00 1.43622652e-01 6.04022503e-01 5.71811855e-01
4.15171683e-01 9.56130505e-01 8.90750766e-01 7.22771943e-01
9.56998859e-03 1.13125764e-01 1.02924657e+00 2.39375398e-01
9.00404274e-01 2.45828822e-01 4.77414072e-01 2.33678102e-01
-1.83402225e-01 -2.29783073e-01 -8.75625730e-01 1.07262695e+00
-1.82178414e+00 -8.87447417e-01 -2.55674750e-01 2.32673621e+00
2.46029451e-01 2.76695460e-01 -1.04875274e-01 4.81871277e-01
4.90550429e-01 -4.25933301e-02 -6.49662256e-01 -5.33072539e-02
-1.58422828e-01 -4.52803165e-01 8.00072789e-01 1.66918024e-01
-1.35205007e+00 5.91000021e-01 5.79764700e+00 1.31493673e-01
-7.72103131e-01 -3.20472449e-01 3.42366189e-01 -2.29927916e-02
-1.23352736e-01 -2.08824471e-01 -6.84326828e-01 4.30553108e-01
8.39311898e-01 4.21568394e-01 -1.36659175e-01 1.19612324e+00
5.02167605e-02 -2.50228018e-01 -7.66051352e-01 8.73039722e-01
2.59523779e-01 -8.92354667e-01 6.35489523e-02 2.26008967e-01
5.74259520e-01 2.61332393e-01 -2.34436348e-01 3.65039147e-02
1.40922964e-01 -7.56293714e-01 6.55403316e-01 4.58901525e-01
3.78795683e-01 -8.54306638e-01 1.09993196e+00 3.67684275e-01
-6.20576560e-01 -4.55745190e-01 -7.78046429e-01 -1.01772800e-01
-1.63910434e-01 4.67207819e-01 -8.50614488e-01 -8.09522122e-02
1.25317550e+00 5.85502446e-01 -4.09351975e-01 1.17206478e+00
8.46610144e-02 7.49690533e-01 -6.14873648e-01 -5.52964322e-02
3.83730650e-01 -2.35889956e-01 9.62271392e-01 8.74191761e-01
5.06302774e-01 1.64457068e-01 1.78961344e-02 5.89324951e-01
6.83939010e-02 4.29326482e-02 -9.86527741e-01 1.30716041e-01
3.33491623e-01 1.00749540e+00 -8.44425023e-01 -3.08954895e-01
-2.54112422e-01 9.72071648e-01 2.53969757e-03 1.51730806e-01
-2.73665220e-01 -3.58009517e-01 9.01225448e-01 -1.83918059e-01
5.62112093e-01 -4.62864265e-02 9.73575115e-02 -1.12513399e+00
3.07583600e-01 -7.35974073e-01 5.35097480e-01 -4.25888270e-01
-1.33083844e+00 8.07631075e-01 -3.22532833e-01 -1.96784127e+00
-2.05228105e-01 -7.50810206e-01 -5.03372967e-01 2.61756986e-01
-1.34624910e+00 -5.43229043e-01 -5.92665434e-01 2.97456533e-01
7.57204056e-01 -2.52021492e-01 7.74718642e-01 1.13403928e-02
-2.15402737e-01 2.91264832e-01 4.07058865e-01 2.71264583e-01
3.57747316e-01 -1.44011176e+00 2.13506892e-01 1.21009207e+00
1.33616120e-01 5.32418638e-02 1.20350873e+00 -7.64344096e-01
-1.38259542e+00 -1.11411738e+00 2.39849404e-01 -5.77023208e-01
5.95899642e-01 -2.70457447e-01 -1.41033673e+00 7.41111457e-01
-5.31227291e-01 4.25521046e-01 6.21754050e-01 -3.61555725e-01
4.31577414e-02 6.78798184e-02 -1.37623274e+00 1.67664528e-01
6.69684112e-01 -2.29333982e-01 -7.43520141e-01 3.97952944e-01
2.56807357e-01 -2.43627846e-01 -7.43256450e-01 6.62056208e-01
4.59377319e-01 -1.11636245e+00 6.48852110e-01 -5.30265689e-01
3.65953863e-01 -4.78978157e-01 -4.34579283e-01 -1.23278022e+00
2.75737196e-01 -1.22145489e-01 8.11612085e-02 7.09050417e-01
4.11407769e-01 -5.44450819e-01 8.81551862e-01 4.27806228e-01
-1.57721519e-01 -2.25879669e-01 -1.04511631e+00 -7.79474974e-01
-3.70565802e-01 -4.56503272e-01 2.86270678e-02 8.29215586e-01
-3.83972347e-01 -2.46954352e-01 -6.10440314e-01 1.05475748e+00
1.07708406e+00 7.00614303e-02 9.36683238e-01 -1.76004624e+00
-1.74166933e-02 5.62727898e-02 -1.07936323e+00 -5.55150330e-01
-1.47521541e-01 -2.80467153e-01 6.84283078e-01 -1.17023492e+00
-2.63420790e-01 -3.53864163e-01 -7.86995888e-02 3.95806223e-01
-5.01514897e-02 5.42434812e-01 -4.69124466e-01 4.08041269e-01
-4.22847956e-01 6.81582987e-01 3.26559573e-01 -8.21261629e-02
-3.15056175e-01 5.85380346e-02 -1.03607196e-02 1.02158475e+00
5.95329583e-01 -8.22486222e-01 4.00581509e-02 -3.56356680e-01
2.98060298e-01 -5.91616891e-02 4.06710029e-01 -1.17311347e+00
8.16849172e-02 8.70249048e-03 3.93957347e-01 -5.81084490e-01
3.25641423e-01 -9.39741313e-01 -3.14614594e-01 6.29159808e-01
-1.20772213e-01 -1.09844215e-01 1.35096043e-01 9.40976560e-01
-3.63754511e-01 -6.11432672e-01 9.17535186e-01 -1.91843092e-01
-1.11444426e+00 4.21741568e-02 -6.32070839e-01 -2.57493407e-01
1.19156778e+00 -2.16608942e-01 9.36446115e-02 -2.09050491e-01
-5.72910428e-01 4.67234135e-01 6.05791271e-01 2.97444701e-01
1.04925239e+00 -6.00939214e-01 -9.73167300e-01 6.36532843e-01
8.14751446e-01 3.40833426e-01 2.87454695e-01 5.77037454e-01
-1.08739424e+00 -6.36526108e-01 -1.86424866e-01 -1.06844163e+00
-1.45773590e+00 3.25567871e-01 3.79476577e-01 6.26762092e-01
-1.06791997e+00 9.44805682e-01 -7.92436525e-02 -3.95655870e-01
2.74833590e-01 -1.94170535e-01 -4.14137185e-01 4.02010649e-01
8.55113089e-01 2.06534311e-01 1.29336685e-01 -7.45940924e-01
-2.09926113e-01 2.92594194e-01 -1.36110038e-02 9.73868147e-02
1.63703763e+00 7.10249096e-02 -5.23733981e-02 5.31893671e-01
8.99887502e-01 -1.36282653e-01 -1.68686748e+00 -1.34406850e-01
8.97458494e-02 -6.38073325e-01 4.47999090e-02 -6.79798611e-03
-9.15496767e-01 7.51376927e-01 1.08207059e+00 6.24052703e-01
9.57096159e-01 1.83291405e-01 4.14428294e-01 8.82900596e-01
2.03744903e-01 -1.09467340e+00 -2.72887610e-02 1.73622116e-01
6.34927154e-01 -1.88434386e+00 1.95571668e-02 2.56802171e-01
-8.56328726e-01 1.15037739e+00 4.84185189e-01 -3.98060173e-01
4.07994449e-01 2.29183570e-01 6.02553785e-01 -5.58796763e-01
-6.59249365e-01 -2.09875956e-01 -7.54136667e-02 3.15619081e-01
-5.22048354e-01 -2.41157487e-02 2.32447423e-02 -1.50943277e-02
-6.31824136e-02 -5.05067766e-01 1.05252862e+00 1.25836432e+00
-9.60036933e-01 -3.56266171e-01 -5.29038727e-01 8.99526238e-01
-7.24291742e-01 2.38760278e-01 -3.59021649e-02 4.13275659e-01
-8.64489377e-02 7.60217607e-01 5.57992876e-01 -4.15995032e-01
2.69621938e-01 3.23256105e-03 -2.87600428e-01 -3.86448145e-01
3.49004753e-02 -1.72424242e-02 -5.21211699e-02 -4.19688553e-01
-5.85368991e-01 -9.15892363e-01 -1.18198752e+00 4.92980361e-01
-4.11776692e-01 4.69382018e-01 8.74178946e-01 8.54015887e-01
1.70279518e-02 2.23095402e-01 8.09967041e-01 -1.14890134e+00
-5.74451447e-01 -1.02249742e+00 -9.18659270e-01 4.75489378e-01
9.14151847e-01 -7.80380726e-01 -1.10630333e+00 9.41986144e-02]
|
[7.698256492614746, 2.230764389038086]
|
08c6f90a-37f0-43db-bdb3-a4d89b7eb49b
|
least-square-value-iteration-is-robust-under
|
2306.10694
| null |
https://arxiv.org/abs/2306.10694v1
|
https://arxiv.org/pdf/2306.10694v1.pdf
|
Least Square Value Iteration is Robust Under Locally Bounded Misspecification Error
|
The success of reinforcement learning heavily relies on the function approximation of policy, value or models, where misspecification (a mismatch between the ground-truth and best function approximators) naturally occurs especially when the ground-truth is complex. As misspecification error does not vanish even with infinite number of samples, designing algorithms that are robust under misspecification is of paramount importance. Recently, it is shown that policy-based approaches can be robust even when the policy function approximation is under a large locally-bounded misspecification error, with which the function class may have $\Omega(1)$ approximation error in certain states and actions but is only small on average under a policy-induced state-distribution; whereas it is only known that value-based approach can effectively learn under globally-bounded misspecification error, i.e., the approximation errors to value functions have a uniform upper bound on all state-actions. Yet it remains an open question whether similar robustness can be achieved with value-based approaches. In this paper, we answer this question affirmatively by showing that the algorithm, Least-Square-Value-Iteration [Jin et al, 2020], with carefully designed exploration bonus can achieve robustness under local misspecification error bound. In particular, we show that algorithm achieves a regret bound of $\widetilde{O}\left(\sqrt{d^3KH^4} + dKH^2\zeta \right)$, where $d$ is the dimension of linear features, $H$ is the length of the episode, $K$ is the total number of episodes, and $\zeta$ is the local bound of the misspecification error. Moreover, we show that the algorithm can achieve the same regret bound without knowing $\zeta$ and can be used as robust policy evaluation oracle that can be applied to improve sample complexity in policy-based approaches.
|
['Lin Yang', 'Yunfan Li']
|
2023-06-19
| null | null | null | null |
['open-question']
|
['natural-language-processing']
|
[ 1.50060020e-02 4.95583057e-01 -5.39198875e-01 8.16806853e-02
-1.12065673e+00 -7.51456797e-01 1.11636370e-01 1.95566714e-01
-7.54476070e-01 1.33621407e+00 -2.78756112e-01 -5.55324852e-01
-6.21862650e-01 -8.74196708e-01 -1.06664371e+00 -9.00524855e-01
-3.19558531e-01 4.46383893e-01 -5.41834533e-03 -1.88053221e-01
3.38697284e-01 2.22356424e-01 -1.34123647e+00 -4.10088956e-01
9.33588922e-01 1.29059482e+00 -1.51759803e-01 6.35577261e-01
2.21455991e-01 5.48861623e-01 -7.14322090e-01 -2.09572852e-01
6.17050886e-01 -5.67612112e-01 -6.48292542e-01 -1.39173582e-01
1.11144699e-01 -3.96025866e-01 -1.65008605e-01 1.35059941e+00
4.89029914e-01 3.78942698e-01 3.02165926e-01 -1.12274587e+00
-2.73286030e-02 6.56303525e-01 -2.70712376e-01 -8.66290405e-02
7.68865421e-02 5.49924731e-01 8.55198681e-01 -3.32503356e-02
3.45712066e-01 1.03151453e+00 4.74458218e-01 5.90847075e-01
-1.26701558e+00 -7.26287127e-01 3.99589568e-01 -1.66588947e-01
-1.11528480e+00 -4.73908074e-02 2.29103342e-01 -4.08587247e-01
7.49348223e-01 3.52654099e-01 6.43190086e-01 7.07499385e-01
1.08052187e-01 5.77907920e-01 1.24985051e+00 -2.66645849e-01
7.79709697e-01 2.53630370e-01 -1.97519794e-01 8.18046093e-01
5.04224479e-01 8.09361160e-01 -2.03673672e-02 -1.25606403e-01
8.39613795e-01 -2.05242798e-01 -4.47419852e-01 -5.10670066e-01
-9.21387494e-01 9.79797602e-01 2.95043319e-01 1.01226494e-01
-3.83143067e-01 5.55907190e-01 3.25994492e-01 7.40593255e-01
1.06493175e-01 9.02047575e-01 -6.88863158e-01 -4.88666981e-01
-6.76935315e-01 5.03567696e-01 7.76233256e-01 6.96147263e-01
7.18209743e-01 4.23976719e-01 -2.12688237e-01 2.51507193e-01
-2.22156659e-01 7.59329975e-01 4.10640270e-01 -1.35120130e+00
5.34078717e-01 5.54374158e-01 7.46170998e-01 -5.85854113e-01
-2.37550840e-01 -8.39952946e-01 -6.41472220e-01 6.63502753e-01
8.59617710e-01 -6.18946910e-01 -4.68748748e-01 2.18629861e+00
2.95329928e-01 -1.45172268e-01 8.28054100e-02 7.52330601e-01
-3.15648228e-01 3.22393328e-01 -3.98647308e-01 -7.88643837e-01
8.47864628e-01 -5.06156743e-01 -4.86401886e-01 -3.05914640e-01
6.79915249e-01 -1.09570161e-01 1.26547384e+00 4.19347107e-01
-1.24871862e+00 -6.30083755e-02 -1.03375602e+00 8.55656087e-01
-1.56035712e-02 -3.00159514e-01 4.69505668e-01 8.37137699e-01
-5.87349296e-01 8.28920484e-01 -6.93197370e-01 1.45932883e-01
2.54452318e-01 4.83585447e-01 -1.68701947e-01 1.74532592e-01
-1.08223259e+00 9.79546189e-01 5.91211438e-01 4.78555039e-02
-1.20498109e+00 -7.17841625e-01 -4.84227628e-01 1.34678945e-01
1.10217631e+00 -3.18700045e-01 1.27590835e+00 -1.00612247e+00
-1.64212310e+00 1.88361928e-01 3.10137153e-01 -8.30620706e-01
7.90761828e-01 1.42517492e-01 -3.77527922e-02 -8.15338865e-02
-1.21001437e-01 -1.80810858e-02 9.81955290e-01 -8.60732436e-01
-7.84531176e-01 -4.82733905e-01 4.49430138e-01 1.67538851e-01
-8.47862810e-02 -4.29534912e-01 2.59860873e-01 -2.34847277e-01
-6.73327595e-02 -8.73748839e-01 -5.32487035e-01 -9.33541134e-02
-1.04971007e-01 1.38290584e-01 2.17439517e-01 -3.53090405e-01
1.44999301e+00 -1.94719815e+00 3.74522060e-02 4.53901827e-01
-3.41241241e-01 1.92589998e-01 1.99579708e-02 3.88388544e-01
-1.73327457e-02 3.20191979e-01 -4.13421333e-01 2.87444651e-01
3.80873889e-01 3.11407000e-01 -4.48953211e-01 6.29880607e-01
-1.39256090e-01 6.27596676e-01 -8.58757794e-01 1.20922491e-01
8.86411145e-02 -7.94257000e-02 -6.42305076e-01 -4.63149920e-02
-4.68852103e-01 2.62340963e-01 -6.22209251e-01 2.46713191e-01
2.90201306e-01 -4.79371846e-02 2.82911032e-01 2.42575213e-01
-1.35427818e-01 2.85868021e-03 -1.71089876e+00 1.15407121e+00
-3.59008163e-01 3.74215022e-02 2.83061296e-01 -1.53198493e+00
5.37240744e-01 2.79700786e-01 4.78553712e-01 -5.82083583e-01
1.49918854e-01 3.51168394e-01 -7.24725425e-02 -1.89741284e-01
-4.61080261e-02 -5.00005126e-01 -2.87424266e-01 3.66650909e-01
-1.98574960e-01 -2.58903861e-01 1.44424841e-01 -3.36174071e-01
1.28391778e+00 -6.28762366e-03 3.88406515e-01 -3.38868439e-01
3.94848824e-01 -3.91870067e-02 7.65304506e-01 1.04182565e+00
-3.89445692e-01 -1.13442792e-02 1.05979109e+00 -3.24486315e-01
-9.15921807e-01 -8.05169761e-01 -1.83088809e-01 8.46970022e-01
2.15227678e-02 7.18636764e-03 -7.69730747e-01 -9.67964411e-01
3.19896519e-01 8.80457759e-01 -9.00657117e-01 -4.03037250e-01
-3.97418708e-01 -7.41143048e-01 4.65001911e-01 4.02268440e-01
5.67930281e-01 -9.11167443e-01 -9.32596743e-01 3.10526401e-01
3.38314086e-01 -5.46178818e-01 -3.86563987e-01 4.44451213e-01
-7.40328789e-01 -1.16233146e+00 -4.99134153e-01 -1.16885781e-01
6.53998256e-01 -4.32227671e-01 7.51952291e-01 -2.58430392e-01
-3.22289243e-02 5.32572031e-01 6.33073822e-02 -4.23047751e-01
-2.08908230e-01 -1.54893652e-01 1.50032908e-01 -2.20575392e-01
-2.97409862e-01 -3.93097550e-01 -7.26114810e-01 1.86553270e-01
-7.38824606e-01 -4.57721442e-01 3.67853165e-01 1.03043497e+00
5.67040026e-01 4.12326157e-01 7.37728179e-01 -5.18836856e-01
6.58352375e-01 -1.83610573e-01 -1.21908987e+00 2.76079208e-01
-9.22034681e-01 4.95220989e-01 7.76053965e-01 -5.62873363e-01
-6.22076988e-01 -1.16714299e-01 2.02984124e-01 -4.23572749e-01
2.72672147e-01 4.64867443e-01 4.02237140e-02 -3.53393927e-02
7.54138231e-01 3.44946653e-01 2.91292518e-01 -2.97426254e-01
2.51566082e-01 1.48378044e-01 2.69343108e-01 -9.70495224e-01
6.08530939e-01 2.13116348e-01 2.62681276e-01 -3.53054047e-01
-8.36730003e-01 1.30415902e-01 1.90422311e-01 -1.97333284e-02
2.93425977e-01 -5.98737419e-01 -1.27097154e+00 -8.31464604e-02
-3.63140762e-01 -7.52877533e-01 -9.53241825e-01 5.22037804e-01
-1.09146130e+00 1.45547315e-01 -3.95086110e-02 -1.40810478e+00
-1.03427343e-01 -1.08374584e+00 3.64858568e-01 1.88747823e-01
1.70793727e-01 -7.27894843e-01 -5.94425993e-03 2.37812218e-03
4.23137575e-01 4.48112845e-01 8.83758307e-01 -4.90968347e-01
-5.01962245e-01 -2.39341602e-01 1.32533431e-01 6.83188200e-01
-1.30970374e-01 -3.86782557e-01 -4.34507161e-01 -7.67950714e-01
1.51419953e-01 -4.63312864e-01 5.16948581e-01 4.67851639e-01
1.15312696e+00 -9.65315640e-01 2.70093232e-02 3.00125271e-01
1.52671683e+00 5.32855272e-01 3.19405019e-01 5.14484823e-01
-1.10688150e-01 3.34382385e-01 8.19022775e-01 8.76853526e-01
-3.01737594e-03 4.76996779e-01 7.05296099e-01 4.30453002e-01
6.08948529e-01 -1.28186718e-01 5.33241928e-01 -2.05249917e-02
3.40825059e-02 2.33373344e-01 -6.13992035e-01 5.42125881e-01
-1.86950219e+00 -1.10846364e+00 5.57594895e-01 2.96714354e+00
1.12797463e+00 5.15483797e-01 4.15491194e-01 1.57056853e-01
4.77681667e-01 -2.38621607e-01 -1.03472781e+00 -6.58319056e-01
9.22305807e-02 3.46689433e-01 9.17108476e-01 7.84904718e-01
-6.54484272e-01 5.42579770e-01 5.66103792e+00 9.24016833e-01
-1.07603467e+00 -2.47443002e-02 5.43223143e-01 -3.58808905e-01
-1.34723589e-01 1.48054078e-01 -6.54412448e-01 7.94853210e-01
1.00480175e+00 -4.35414076e-01 8.20193529e-01 1.03318083e+00
2.82019377e-01 -3.66674155e-01 -9.33471620e-01 5.18023312e-01
-5.33466041e-01 -9.68024433e-01 -5.64750731e-01 2.78769284e-01
8.64459097e-01 -4.40830052e-01 2.58191526e-01 7.50227451e-01
5.00354648e-01 -9.65127051e-01 6.97724998e-01 3.66167456e-01
7.75220454e-01 -1.15383697e+00 6.93653941e-01 8.07088375e-01
-8.66818190e-01 -6.74026191e-01 -3.00492406e-01 -1.16382003e-01
-2.31150582e-01 4.66277212e-01 -7.07734346e-01 4.02702063e-01
4.38508421e-01 -2.20866948e-01 1.02752060e-01 9.51497972e-01
-9.27956551e-02 4.88393098e-01 -6.20133638e-01 -2.24842504e-01
4.95832831e-01 -2.89248049e-01 5.50533533e-01 5.85510790e-01
3.16662848e-01 7.20505267e-02 3.38776499e-01 6.56487763e-01
3.28633308e-01 -8.29519778e-02 -3.62386614e-01 -2.84081161e-01
4.40010041e-01 6.59227788e-01 -3.21521580e-01 -1.93373859e-01
9.90529824e-03 5.68976700e-01 3.21349412e-01 4.84357446e-01
-8.73383880e-01 -4.32187498e-01 7.98149645e-01 7.18379915e-02
4.83664989e-01 -1.22167267e-01 -1.08835906e-01 -1.03173709e+00
1.61011532e-01 -1.02082980e+00 5.44216275e-01 -9.83781964e-02
-8.66027832e-01 -4.33588028e-03 3.45629044e-02 -9.77889597e-01
-6.11932814e-01 -4.66634303e-01 -2.31833979e-01 6.93253517e-01
-1.02938962e+00 -3.93206984e-01 4.57966983e-01 5.56477368e-01
1.74698263e-01 -6.85886592e-02 7.09085464e-01 -1.49402469e-01
-7.06718743e-01 8.28722954e-01 5.94241023e-01 -2.44682759e-01
2.56200016e-01 -1.39162982e+00 -5.05722165e-01 5.60913026e-01
-3.74428600e-01 4.19027299e-01 1.05237257e+00 -4.07009959e-01
-1.50669372e+00 -9.47083533e-01 2.50141531e-01 -1.76524848e-01
6.88178241e-01 -6.37802929e-02 -6.02423966e-01 6.59045458e-01
-2.66673476e-01 2.86351651e-01 8.98504034e-02 -8.23546052e-02
-1.25878081e-01 -4.57916796e-01 -1.52835846e+00 6.16745889e-01
7.92576909e-01 -2.90494919e-01 -2.25663409e-01 1.89314455e-01
5.64708292e-01 -4.35773849e-01 -1.12946355e+00 6.04850292e-01
5.01448870e-01 -9.00624335e-01 6.36789858e-01 -8.64565670e-01
-2.32269019e-01 -1.59886569e-01 -2.89491415e-01 -1.17786825e+00
1.05097875e-01 -9.73179340e-01 -5.43392420e-01 6.79391861e-01
3.89653057e-01 -9.04962957e-01 7.42286801e-01 6.83205843e-01
2.29533166e-01 -1.14157021e+00 -1.40595078e+00 -1.18212676e+00
4.83791739e-01 -4.32332397e-01 5.50886095e-01 6.05988443e-01
2.21403122e-01 -2.43681774e-01 -1.96428388e-01 3.48905511e-02
6.94174767e-01 1.96996197e-01 5.32735705e-01 -7.07838237e-01
-8.61551821e-01 -6.99517190e-01 -1.07119590e-01 -8.01914990e-01
8.59044567e-02 -3.41077328e-01 -3.50601450e-02 -9.93822396e-01
-1.37704790e-01 -5.67500174e-01 -5.06720185e-01 5.42714238e-01
-1.82138346e-02 -4.73121107e-01 2.38798425e-01 -3.82494122e-01
-5.02942860e-01 6.31301284e-01 1.39686882e+00 4.25616354e-02
-3.81787747e-01 3.05757552e-01 -6.34050548e-01 6.32348239e-01
8.68972957e-01 -3.63089472e-01 -5.72749197e-01 1.27256811e-01
4.56694692e-01 6.69132113e-01 3.09071660e-01 -9.68162537e-01
-1.16671138e-01 -6.40697181e-01 -4.22607511e-02 -8.82475451e-02
6.80930987e-02 -7.51039863e-01 1.30746722e-01 8.51181626e-01
-6.21328652e-01 -1.22439139e-01 3.58052403e-02 8.29398155e-01
2.02885658e-01 -4.55092102e-01 8.92243207e-01 -2.18297958e-01
-1.03567295e-01 2.62402058e-01 -2.79246390e-01 2.46665061e-01
1.09065759e+00 -5.73408604e-02 -1.58699304e-01 -5.95854223e-01
-6.73392713e-01 3.27831596e-01 3.29277843e-01 -1.73033610e-01
2.43319944e-01 -1.13823605e+00 -3.78055125e-01 7.54995644e-02
-3.28893721e-01 -1.08656250e-01 1.57875896e-01 8.52386236e-01
6.69147670e-02 3.48042369e-01 2.34940514e-01 -2.43264243e-01
-5.85373521e-01 7.27555513e-01 7.20240057e-01 -4.94530410e-01
-2.24389881e-01 6.33620024e-01 -6.47702366e-02 -1.69582710e-01
4.91399854e-01 -4.23025608e-01 4.81090575e-01 -1.35609359e-01
4.33894306e-01 5.15714407e-01 -9.59060565e-02 1.22161962e-01
-1.30081579e-01 3.16438079e-01 7.56084397e-02 -4.47753996e-01
1.05714238e+00 -1.18344426e-02 2.94131845e-01 2.82969296e-01
1.02808988e+00 -3.32523108e-01 -1.79270661e+00 -1.27450064e-01
-3.66554707e-02 -3.22937548e-01 6.55608475e-02 -1.07887399e+00
-9.27704692e-01 6.83149815e-01 8.04334998e-01 4.11015421e-01
1.06850886e+00 -3.77941966e-01 2.79033929e-01 6.39039397e-01
6.82799339e-01 -1.38345313e+00 -6.93050995e-02 3.80072832e-01
9.13170397e-01 -1.04402947e+00 -7.80063048e-02 3.61645073e-01
-5.52174151e-01 7.99697459e-01 7.25217044e-01 -3.31790775e-01
3.13720465e-01 8.77464116e-02 -5.60789049e-01 2.58337468e-01
-8.57552767e-01 -3.41689795e-01 -3.55332042e-03 1.73228577e-01
-1.75999790e-01 2.33844891e-01 -6.49337411e-01 6.85020208e-01
-1.41949162e-01 -1.01048812e-01 2.33625129e-01 1.13030910e+00
-8.38575125e-01 -9.89906251e-01 -3.32249254e-01 3.77114683e-01
-5.67725778e-01 2.27442726e-01 1.76683605e-01 1.05116856e+00
-4.47040275e-02 7.27355957e-01 -1.45572707e-01 5.55387251e-02
4.62568760e-01 5.60080782e-02 7.27426231e-01 -1.68019131e-01
-5.07135272e-01 -4.51385453e-02 5.60862608e-02 -9.51299250e-01
8.67074803e-02 -4.29141372e-01 -1.31991506e+00 -2.39713803e-01
-2.43282109e-01 4.82827514e-01 5.06526053e-01 1.00784838e+00
2.10062131e-01 1.60660803e-01 9.01467264e-01 -3.22374076e-01
-1.59191930e+00 -6.88266158e-01 -6.99852526e-01 7.24322498e-02
5.61615944e-01 -6.75077021e-01 -7.54282296e-01 -6.88977003e-01]
|
[4.299725532531738, 2.847247362136841]
|
e4c7da36-50ee-406c-85d3-f5e73058ba66
|
an-underwater-image-enhancement-benchmark
|
1901.05495
| null |
https://arxiv.org/abs/1901.05495v2
|
https://arxiv.org/pdf/1901.05495v2.pdf
|
An Underwater Image Enhancement Benchmark Dataset and Beyond
|
Underwater image enhancement has been attracting much attention due to its significance in marine engineering and aquatic robotics. Numerous underwater image enhancement algorithms have been proposed in the last few years. However, these algorithms are mainly evaluated using either synthetic datasets or few selected real-world images. It is thus unclear how these algorithms would perform on images acquired in the wild and how we could gauge the progress in the field. To bridge this gap, we present the first comprehensive perceptual study and analysis of underwater image enhancement using large-scale real-world images. In this paper, we construct an Underwater Image Enhancement Benchmark (UIEB) including 950 real-world underwater images, 890 of which have the corresponding reference images. We treat the rest 60 underwater images which cannot obtain satisfactory reference images as challenging data. Using this dataset, we conduct a comprehensive study of the state-of-the-art underwater image enhancement algorithms qualitatively and quantitatively. In addition, we propose an underwater image enhancement network (called Water-Net) trained on this benchmark as a baseline, which indicates the generalization of the proposed UIEB for training Convolutional Neural Networks (CNNs). The benchmark evaluations and the proposed Water-Net demonstrate the performance and limitations of state-of-the-art algorithms, which shed light on future research in underwater image enhancement. The dataset and code are available at https://li-chongyi.github.io/proj_benchmark.html.
|
['DaCheng Tao', 'Sam Kwong', 'Runmin Cong', 'Chunle Guo', 'Wenqi Ren', 'Junhui Hou', 'Chongyi Li']
|
2019-01-11
| null | null | null | null |
['underwater-image-restoration']
|
['computer-vision']
|
[ 4.05594438e-01 -6.53394386e-02 9.04238760e-01 -4.27741647e-01
-4.46894884e-01 -1.16439581e-01 2.12294802e-01 -7.69104287e-02
-1.28507376e+00 6.54468536e-01 1.92958727e-01 2.55032703e-02
-1.42873982e-02 -1.01760721e+00 -8.28198850e-01 -8.87663722e-01
-4.73815709e-01 -5.84233224e-01 3.37390661e-01 -6.73533559e-01
3.15747231e-01 1.81166574e-01 -1.79494357e+00 -5.53083755e-02
1.06334376e+00 8.61582696e-01 6.31669760e-01 7.54634142e-01
2.75088429e-01 5.01693070e-01 -6.90595746e-01 -5.62015176e-01
4.77631927e-01 -4.51904714e-01 -3.13600600e-01 -1.27346337e-01
4.28251714e-01 -8.61218691e-01 -6.86805248e-01 1.43172359e+00
1.14712703e+00 3.59438986e-01 2.08453223e-01 -9.76080120e-01
-7.93334723e-01 5.37574291e-01 -3.16071033e-01 2.46383145e-01
-1.57372072e-01 8.12757462e-02 6.21457696e-01 -9.41910923e-01
4.53876674e-01 8.85609567e-01 7.95180678e-01 6.00686133e-01
-3.50908041e-01 -6.99452877e-01 -1.82531267e-01 2.31543809e-01
-9.64290023e-01 -4.49761122e-01 5.98709226e-01 -9.58917439e-02
4.42742079e-01 6.21402748e-02 8.18351388e-01 5.18181741e-01
3.44641268e-01 8.00436139e-01 1.26607203e+00 -2.70042479e-01
3.10566932e-01 -2.08299920e-01 -7.00109005e-02 4.93609190e-01
3.56372058e-01 4.00734156e-01 -3.71209830e-01 3.25521857e-01
5.80491126e-01 1.31755382e-01 -7.50238538e-01 2.02387452e-01
-5.53828537e-01 7.08511412e-01 7.40110934e-01 -6.89282492e-02
-2.00203493e-01 -1.31769836e-01 4.12867546e-01 6.19191706e-01
4.73869383e-01 3.23624134e-01 -4.35956806e-01 2.80702319e-02
-7.41613686e-01 2.98022032e-01 5.86087584e-01 7.72065043e-01
9.94915783e-01 3.08946699e-01 3.50520313e-01 9.98349726e-01
4.91899163e-01 6.82828784e-01 3.98345739e-01 -9.98502731e-01
3.36422235e-01 1.50214255e-01 7.47156888e-02 -8.06429863e-01
-5.05144238e-01 -1.09221160e-01 -7.94761896e-01 7.11670458e-01
1.26336247e-01 -5.11802435e-01 -8.67660403e-01 1.35941589e+00
-9.61344037e-03 -2.68694456e-03 8.33062232e-01 1.07401717e+00
1.40990281e+00 9.30948377e-01 -1.02720879e-01 -1.31022915e-01
1.17219555e+00 -1.01408899e+00 -8.09475660e-01 -2.21540123e-01
2.60088533e-01 -6.73619449e-01 7.54235029e-01 2.89878756e-01
-1.04651594e+00 -5.87385654e-01 -1.33408892e+00 5.48776761e-02
-4.92530465e-01 -1.66848153e-02 3.68809044e-01 5.91749191e-01
-1.25053990e+00 6.67691469e-01 -8.32689464e-01 -2.70043135e-01
1.57373279e-01 5.78256287e-02 -5.41437626e-01 -3.02065134e-01
-1.52562439e+00 8.13469589e-01 3.92364353e-01 6.24487281e-01
-1.22132242e+00 -5.47344983e-01 -1.31435037e+00 -1.08290814e-01
-3.45297642e-02 4.99334857e-02 1.29044795e+00 -6.56891525e-01
-1.38558984e+00 5.22663951e-01 2.98955679e-01 -1.80068150e-01
3.87766033e-01 -3.30643386e-01 -4.31523442e-01 3.72876614e-01
-7.56228939e-02 6.67355776e-01 3.62791449e-01 -1.65504253e+00
-8.47849250e-01 -2.74046659e-01 3.25760663e-01 3.45355660e-01
-5.62157035e-01 2.72208631e-01 -5.49855947e-01 -6.50544226e-01
1.08769551e-01 -7.60106564e-01 -4.25699443e-01 4.72570896e-01
-6.63077161e-02 3.55829626e-01 9.84719515e-01 -7.53731787e-01
8.44181240e-01 -2.23687243e+00 -1.89318597e-01 -2.90584177e-01
-1.35933399e-01 5.05431890e-01 -5.13932228e-01 6.52121127e-01
2.23417953e-01 3.58320139e-02 -6.64244592e-01 -4.93399471e-01
-1.82592168e-01 4.36347514e-01 8.31337869e-02 7.29838014e-01
-8.54276866e-02 5.10686696e-01 -1.07338929e+00 -5.14005661e-01
2.53363937e-01 6.34750128e-01 -3.27195168e-01 7.02404380e-01
5.08020282e-01 3.43698204e-01 -1.40044272e-01 9.44860816e-01
1.20010185e+00 5.67490458e-01 -1.39432922e-01 -4.78245795e-01
-6.40094221e-01 -5.66472590e-01 -1.11849725e+00 1.34891284e+00
-3.78862262e-01 9.88759458e-01 5.14859557e-01 -7.58902013e-01
1.03010321e+00 1.93535581e-01 1.70362383e-01 -8.29897761e-01
2.61439055e-01 3.16275328e-01 -2.51825769e-02 -9.69640672e-01
9.21186268e-01 -3.56749058e-01 3.39140266e-01 8.18722043e-03
2.67084222e-03 -1.15748256e-01 5.59407473e-01 6.59349933e-02
7.54565358e-01 -6.38568634e-03 2.89176777e-02 -3.95612031e-01
1.75264075e-01 -3.03368032e-01 5.77855825e-01 7.66179800e-01
-5.03437936e-01 8.82357240e-01 -1.41324714e-01 -3.20913166e-01
-1.06911767e+00 -9.98188019e-01 -5.43182075e-01 1.02917480e+00
8.99303138e-01 1.74286868e-02 -8.04055274e-01 -1.31577011e-02
-4.55023974e-01 3.68507765e-02 -9.52854872e-01 -3.23383585e-02
-3.70074302e-01 -1.23694646e+00 8.24431539e-01 6.00428402e-01
1.11137521e+00 -1.56155145e+00 -8.83802772e-01 2.82075614e-01
-8.28382894e-02 -1.05495512e+00 -1.18844844e-01 1.68213263e-01
-9.32620466e-01 -1.00014758e+00 -1.08842695e+00 -1.00881016e+00
7.41920114e-01 5.73909044e-01 8.96589816e-01 3.24321359e-01
1.07865420e-03 -4.58306298e-02 -9.09608185e-01 -6.53016627e-01
-4.38734859e-01 -5.87189794e-01 1.16255283e-02 -1.93439409e-01
-1.89743582e-02 -3.83854389e-01 -1.00663662e+00 4.55868632e-01
-1.58295226e+00 -3.73054594e-01 8.23316514e-01 9.84459162e-01
2.17429847e-01 3.55990021e-03 2.79659152e-01 -3.97367865e-01
4.11416054e-01 -3.79802585e-01 -5.84631741e-01 1.01471320e-01
-3.15391004e-01 -2.37375721e-01 3.52850974e-01 -1.17652878e-01
-1.21234381e+00 -2.29152128e-01 -8.10571969e-01 1.42338246e-01
-3.88559289e-02 8.98050845e-01 -1.44076541e-01 -4.91843909e-01
4.46351528e-01 4.76156414e-01 1.18856236e-01 -6.36043131e-01
-2.03928992e-01 9.67676222e-01 7.30609357e-01 -2.91277736e-01
7.10742056e-01 7.54162669e-01 -3.83695126e-01 -1.26043260e+00
-7.57648647e-01 -5.00164270e-01 -2.15443492e-01 -5.59943080e-01
8.35974276e-01 -1.21356344e+00 -4.56130147e-01 1.10701799e+00
-9.03918266e-01 -6.23281777e-01 1.39988318e-01 4.62411880e-01
-6.83805794e-02 7.58703709e-01 -1.05269694e+00 -8.20005298e-01
-6.91291988e-01 -1.48212922e+00 9.47617948e-01 8.33054245e-01
7.12539077e-01 -7.99480438e-01 2.35522151e-01 -2.39879694e-02
7.97495544e-01 2.77121335e-01 -1.53361201e-01 1.83193572e-02
-1.45128503e-01 -1.86545793e-02 -4.89567727e-01 8.06500375e-01
7.08154216e-02 1.56623393e-01 -1.04477668e+00 -6.38627172e-01
-2.48063654e-01 -5.11427283e-01 1.18583357e+00 4.11716521e-01
8.61021698e-01 1.26037449e-01 2.64678478e-01 1.06268883e+00
1.72893786e+00 3.33034217e-01 1.33402860e+00 8.79487872e-01
2.84643769e-01 7.59381235e-01 7.14835942e-01 5.85920036e-01
3.20647031e-01 2.39262134e-01 9.66116846e-01 -5.69008648e-01
1.67994738e-01 -7.77002275e-02 4.81092930e-01 7.43959546e-01
-6.78627491e-01 -3.77414048e-01 -4.69231069e-01 7.07154989e-01
-1.25819623e+00 -8.20081949e-01 -1.22613050e-01 1.75414538e+00
6.82327032e-01 -3.01365644e-01 -3.68729264e-01 1.94938317e-01
7.27033019e-01 1.89500242e-01 -3.64781082e-01 -3.52177396e-02
-5.20721674e-01 4.86777797e-02 7.37270176e-01 2.84744948e-01
-1.44111705e+00 5.34213066e-01 5.68377781e+00 3.87176663e-01
-1.01367879e+00 -5.64706000e-03 4.18818712e-01 5.83947241e-01
1.03259392e-01 -3.44854832e-01 -6.92705572e-01 3.03349078e-01
7.06350327e-01 2.05074415e-01 1.25382049e-02 8.26477051e-01
5.20604014e-01 5.99673092e-02 -3.11651826e-01 7.07482994e-01
1.48085371e-01 -1.07306564e+00 -7.97762871e-02 1.24251796e-03
9.44239855e-01 4.35534120e-01 -2.57795185e-01 3.23822021e-01
2.17003524e-01 -6.89031959e-01 7.04645276e-01 3.13581854e-01
8.50298166e-01 -4.54072565e-01 1.47376966e+00 -3.87199596e-02
-1.22235870e+00 -1.08915105e-01 -9.81040597e-01 -2.42812485e-01
3.81113082e-01 1.87355623e-01 7.60402754e-02 5.56394458e-01
1.45137942e+00 7.62583733e-01 -4.73957628e-01 1.56742597e+00
-4.51540679e-01 6.44756794e-01 -2.40727648e-01 5.71439154e-02
4.52294439e-01 -4.87848103e-01 3.89678806e-01 1.46036255e+00
6.89731061e-01 4.25712436e-01 -2.22011060e-01 3.47390473e-01
-2.39981294e-01 1.58576086e-01 -4.48698431e-01 2.07314372e-01
1.85814828e-01 1.37703300e+00 -4.38103139e-01 -2.84328580e-01
-6.58429980e-01 7.50548244e-01 -2.02993274e-01 3.99496853e-01
-5.72059393e-01 -8.90414119e-01 8.13875616e-01 -4.07222867e-01
1.02636009e-01 -1.01920508e-01 9.12886262e-02 -1.11136746e+00
-2.24717557e-01 -7.70473003e-01 2.43823141e-01 -1.03327978e+00
-1.28742957e+00 1.01614964e+00 -1.33378923e-01 -1.63848054e+00
5.10093927e-01 -8.60695481e-01 -7.86840498e-01 6.23887360e-01
-2.33626151e+00 -8.09615254e-01 -1.21661925e+00 1.82461396e-01
6.14062369e-01 1.43909439e-01 5.74278235e-01 5.33411503e-01
-4.13824350e-01 5.51295996e-01 6.13544226e-01 6.42561972e-01
7.20234334e-01 -1.20944726e+00 1.80236220e-01 1.16980720e+00
-5.08035898e-01 4.36661303e-01 1.04864633e+00 -3.69710952e-01
-1.30403256e+00 -1.17151451e+00 2.46229663e-01 3.06310654e-01
6.72239244e-01 1.59573227e-01 -1.24887085e+00 3.56974930e-01
6.14560127e-01 1.76809207e-01 5.69763899e-01 -8.19174170e-01
2.62254715e-01 -1.93360165e-01 -1.12039769e+00 5.67892909e-01
7.61874199e-01 1.56110361e-01 -4.04926419e-01 -1.81510419e-01
6.29960358e-01 -4.73145425e-01 -1.15481162e+00 8.62436831e-01
6.89495564e-01 -1.23157310e+00 8.54724407e-01 -2.92614978e-02
9.98843789e-01 -6.23543024e-01 -3.90135258e-01 -1.56608152e+00
2.48923779e-01 -2.40839571e-01 6.49099290e-01 1.00833404e+00
4.64099854e-01 -6.07264638e-01 4.97506291e-01 2.10444808e-01
-8.46804142e-01 -6.21493638e-01 -5.83276868e-01 -5.81674516e-01
1.67411521e-01 -2.66864479e-01 3.01506728e-01 6.50139868e-01
-1.85856923e-01 -2.89943725e-01 -7.77422905e-01 5.84704518e-01
9.53545272e-01 7.71971419e-03 7.60581732e-01 -8.86921644e-01
-1.52645394e-01 -8.37721303e-02 -5.30588686e-01 -1.09105861e+00
-2.54468441e-01 -1.01775520e-01 8.05576682e-01 -1.88104331e+00
3.32402498e-01 -9.12436619e-02 -2.41142169e-01 5.27928233e-01
-5.21675825e-01 9.91565466e-01 1.57204464e-01 2.79909815e-03
-4.17719245e-01 1.10579979e+00 1.44388318e+00 -2.09070802e-01
3.41105163e-02 -2.12947056e-01 -5.94199657e-01 6.33627951e-01
9.64387894e-01 -1.77341640e-01 6.93659037e-02 -6.89973474e-01
8.01041573e-02 -2.32652761e-02 1.34370998e-01 -1.09269297e+00
3.45557600e-01 7.18235373e-02 7.71383345e-02 -2.35135198e-01
2.66125858e-01 -6.47634387e-01 -1.02227010e-01 6.04390562e-01
-1.67867109e-01 1.06367737e-01 2.62393981e-01 4.27047193e-01
-8.23622823e-01 -7.31026053e-01 1.19585955e+00 -2.27500334e-01
-1.44840336e+00 3.68937671e-01 -4.30097908e-01 -2.56758258e-02
7.12240815e-01 -3.76653790e-01 -6.65529132e-01 -4.75681424e-01
-3.92591536e-01 3.68057132e-01 6.05432451e-01 1.97135985e-01
1.21090376e+00 -5.70572317e-01 -1.24178445e+00 -3.19605507e-02
4.00879860e-01 -1.72652733e-02 6.09114110e-01 5.08247554e-01
-1.31726205e+00 -5.98670781e-01 -5.45352042e-01 -1.79554611e-01
-1.27134693e+00 8.31258148e-02 6.20288074e-01 1.92653164e-01
-7.16542542e-01 7.60113120e-01 2.73739159e-01 -5.76766849e-01
5.50634861e-02 -8.85164291e-02 -6.23960376e-01 -2.49868557e-01
1.04884565e+00 3.82165909e-01 -1.66009888e-01 -6.42014980e-01
1.42734293e-02 7.22375572e-01 4.32938576e-01 1.04897767e-01
1.90162349e+00 -3.73635411e-01 -2.51240432e-01 -6.15554266e-02
1.21916473e+00 -2.26745814e-01 -1.76249135e+00 -3.10858730e-02
-5.45901120e-01 -5.65008223e-01 1.38806239e-01 -4.58346248e-01
-1.56007862e+00 9.67263043e-01 1.00614715e+00 1.50009200e-01
1.56990004e+00 -2.24194229e-01 8.75109076e-01 4.43649411e-01
2.60380864e-01 -9.10084903e-01 -2.44044811e-02 5.88954926e-01
1.00177753e+00 -1.69801450e+00 4.97702099e-02 -3.10434878e-01
-6.35273933e-01 1.31223714e+00 6.41149640e-01 -3.61217484e-02
6.57077014e-01 7.54582465e-01 8.38128269e-01 -1.10877044e-01
-2.97108553e-02 -4.71638322e-01 -3.40021372e-01 6.40487015e-01
2.41049543e-01 -2.27937266e-01 -4.74908382e-01 4.60069656e-01
-1.29970118e-01 -1.13555081e-01 1.04879546e+00 1.10379303e+00
-8.49352181e-01 -7.75905609e-01 -4.41606730e-01 1.55535281e-01
-7.80215859e-01 -3.89853805e-01 3.35603416e-01 8.14056635e-01
3.94070297e-02 1.11658895e+00 -2.53179848e-01 -5.06481826e-01
6.06631517e-01 -8.20832849e-01 -5.72244190e-02 -8.11258405e-02
-2.87432551e-01 -1.07161082e-01 4.10735719e-02 -2.74927691e-02
-1.05652535e+00 -4.36098635e-01 -1.17516685e+00 -1.06346468e-02
-2.56785095e-01 5.59717834e-01 8.26161683e-01 6.28979623e-01
-2.63995081e-01 5.15914679e-01 6.86354935e-01 -1.59561896e+00
-4.05255556e-01 -1.50593746e+00 -9.92592990e-01 4.51438963e-01
2.71823883e-01 -5.68185747e-01 -8.16410959e-01 3.29457998e-01]
|
[10.687390327453613, -3.5251107215881348]
|
80d1ceef-27ae-446a-9bf1-0b38649f82f5
|
leveraging-procedural-generation-to-benchmark
|
1912.01588
| null |
https://arxiv.org/abs/1912.01588v2
|
https://arxiv.org/pdf/1912.01588v2.pdf
|
Leveraging Procedural Generation to Benchmark Reinforcement Learning
|
We introduce Procgen Benchmark, a suite of 16 procedurally generated game-like environments designed to benchmark both sample efficiency and generalization in reinforcement learning. We believe that the community will benefit from increased access to high quality training environments, and we provide detailed experimental protocols for using this benchmark. We empirically demonstrate that diverse environment distributions are essential to adequately train and evaluate RL agents, thereby motivating the extensive use of procedural content generation. We then use this benchmark to investigate the effects of scaling model size, finding that larger models significantly improve both sample efficiency and generalization.
|
['Karl Cobbe', 'John Schulman', 'Jacob Hilton', 'Christopher Hesse']
|
2019-12-03
| null |
https://proceedings.icml.cc/static/paper_files/icml/2020/2971-Paper.pdf
|
https://proceedings.icml.cc/static/paper_files/icml/2020/2971-Paper.pdf
|
icml-2020-1
|
['procgen-hard-100m']
|
['playing-games']
|
[ 6.50041178e-03 -1.02482274e-01 -7.77327269e-03 -2.34715790e-02
-8.65863025e-01 -8.48874629e-01 5.87975860e-01 1.15930391e-02
-9.23452020e-01 1.06597972e+00 1.68369964e-01 -3.82254869e-01
-3.20632085e-02 -9.30615544e-01 -7.45201886e-01 -3.57670873e-01
-6.42550647e-01 4.59763467e-01 3.43626082e-01 -2.46585310e-01
2.92621106e-01 3.38188022e-01 -1.87027478e+00 -3.67884077e-02
8.91187847e-01 3.16522449e-01 1.65306270e-01 9.22672510e-01
5.05465686e-01 1.02858269e+00 -8.19933772e-01 -1.09543256e-01
5.70876956e-01 -6.87684476e-01 -7.38424003e-01 -1.90614164e-01
-1.80431485e-01 -6.46070540e-01 -1.73306987e-01 6.03750944e-01
9.14813042e-01 7.68790126e-01 4.53918904e-01 -1.21961796e+00
-2.59181529e-01 6.53453588e-01 -1.43696100e-01 3.37260723e-01
3.60583395e-01 7.18232989e-01 8.77492249e-01 -3.05090308e-01
7.38041341e-01 9.24527526e-01 4.62770730e-01 6.33400083e-01
-1.31502342e+00 -7.38665462e-01 1.70356277e-02 -2.68245161e-01
-1.07492876e+00 -5.94587982e-01 3.93580109e-01 -1.53506771e-01
1.08970118e+00 3.45290452e-02 8.43446791e-01 1.23927343e+00
3.32779624e-02 6.74439907e-01 1.27304888e+00 -2.19887987e-01
7.53591597e-01 -1.78811148e-01 -3.33205372e-01 7.10328639e-01
3.58989686e-01 9.23570156e-01 -5.81263959e-01 -2.19446272e-01
1.02500892e+00 -5.96916735e-01 -1.79877996e-01 -7.57066488e-01
-9.53614175e-01 8.69481206e-01 2.64982969e-01 -1.72344670e-01
-3.82008463e-01 6.62144661e-01 4.43033755e-01 3.57918292e-01
-8.93005133e-02 1.26218498e+00 -1.84470922e-01 -9.97895122e-01
-5.34647286e-01 9.89634275e-01 8.77126217e-01 1.02108598e+00
4.99421477e-01 3.30030590e-01 -7.42475763e-02 9.01509881e-01
-6.63143173e-02 3.98697704e-01 4.85322326e-01 -1.81540382e+00
3.61691862e-01 2.49619320e-01 5.62001228e-01 -5.49033940e-01
-4.46986914e-01 -5.27913332e-01 -9.81532317e-03 5.15327334e-01
3.15965056e-01 -6.11948907e-01 -5.79016328e-01 2.08329391e+00
1.24927513e-01 1.01887532e-01 1.86301053e-01 6.15225255e-01
5.41751683e-01 2.47344658e-01 4.24792111e-01 -2.08039097e-02
7.85011709e-01 -1.03805041e+00 -1.42801806e-01 -4.84639794e-01
7.37502694e-01 -2.58877754e-01 1.54179728e+00 2.48861298e-01
-1.49362814e+00 -3.88472229e-01 -1.05458260e+00 4.53695178e-01
-2.35853568e-02 -4.74033266e-01 1.09151030e+00 8.12929749e-01
-1.03427124e+00 8.06388438e-01 -9.72420752e-01 -4.32256877e-01
4.91608560e-01 3.10711861e-01 -1.32779047e-01 -5.74720502e-02
-8.26959550e-01 8.29034746e-01 5.15179932e-01 -3.77852142e-01
-1.44228804e+00 -5.32240212e-01 -8.41372490e-01 4.72236425e-02
4.42616165e-01 -7.02425301e-01 1.69003510e+00 -4.42548722e-01
-1.72997582e+00 5.66784620e-01 3.27439755e-01 -5.39150715e-01
4.77016091e-01 -8.95510092e-02 1.28923461e-01 -1.02999277e-01
-8.53767022e-02 9.17383254e-01 1.75930426e-01 -1.44294369e+00
-6.39636219e-01 4.12710719e-02 3.94064009e-01 6.46067381e-01
-1.11057619e-02 -1.21803239e-01 -1.88630700e-01 -4.80520666e-01
-3.17625016e-01 -1.13235140e+00 -5.18419206e-01 -7.37602711e-01
1.79287508e-01 1.73279226e-01 1.57719761e-01 -7.71583468e-02
7.16922820e-01 -1.87452960e+00 -8.81329030e-02 1.03299737e-01
3.62188220e-02 -2.01161131e-01 -5.13756514e-01 4.63826478e-01
3.18295032e-01 1.66631892e-01 -1.57680944e-01 -1.77140146e-01
1.43971294e-01 3.76966536e-01 -1.90681487e-01 5.58589883e-02
-5.39761372e-02 1.06818628e+00 -1.11933577e+00 -2.99301416e-01
-3.81116681e-02 -7.93195739e-02 -1.26794648e+00 3.28272432e-01
-4.18909788e-01 6.14999771e-01 -5.27535200e-01 3.59646112e-01
2.21067578e-01 9.19304439e-04 2.27833271e-01 5.79012513e-01
4.45067976e-03 5.87048113e-01 -1.02263904e+00 1.67698658e+00
-5.13309836e-01 2.60675579e-01 -8.30477923e-02 -4.12605703e-01
7.28123248e-01 -2.56047547e-02 3.32927644e-01 -1.07013440e+00
2.01244697e-01 8.11862200e-02 4.04557675e-01 -2.34166875e-01
7.71166563e-01 -1.19025409e-01 -1.73855871e-01 7.79896498e-01
-4.77185249e-02 -7.00671613e-01 5.07358611e-01 9.62373689e-02
1.47347903e+00 5.15744448e-01 1.58760190e-01 -3.40798676e-01
-3.07161361e-01 2.37783387e-01 4.20329154e-01 1.38931775e+00
-4.38577324e-01 1.36857018e-01 4.98081386e-01 -2.56448835e-01
-1.04852295e+00 -1.33323324e+00 1.65751636e-01 1.50287390e+00
1.01617984e-01 -4.79340345e-01 -6.55671060e-01 -5.73589206e-01
1.93617470e-03 9.57806408e-01 -6.40688658e-01 -1.58718213e-01
-5.69921672e-01 -7.82189906e-01 7.56432176e-01 7.16187894e-01
5.29833317e-01 -1.69968796e+00 -1.26046741e+00 2.02874362e-01
3.14290561e-02 -7.28385627e-01 -1.34862825e-01 5.78996003e-01
-8.40289116e-01 -1.12111342e+00 -2.22437575e-01 -5.59244394e-01
2.46539801e-01 1.87936828e-01 1.66309941e+00 5.70215583e-01
-1.71177045e-01 7.69161940e-01 -4.95566845e-01 -1.97275758e-01
-6.40828133e-01 3.43417376e-01 1.50292208e-02 -1.16030002e+00
-6.77840486e-02 -9.60265815e-01 -6.28134191e-01 3.27813148e-01
-7.24744439e-01 6.36233315e-02 4.30993825e-01 8.63882959e-01
2.83157766e-01 -3.46033499e-02 7.09287226e-01 -9.29439485e-01
1.30167758e+00 -4.05848861e-01 -8.05622399e-01 -4.71979678e-02
-5.28118253e-01 2.53717393e-01 4.83482152e-01 -2.31263474e-01
-8.53393674e-01 -3.05322677e-01 -2.18353510e-01 1.82986453e-01
-1.31629735e-01 4.83377814e-01 1.17764466e-01 -2.89129317e-01
1.12548566e+00 2.67564774e-01 -8.01397264e-02 8.99549052e-02
3.74160498e-01 1.19127661e-01 2.71664321e-01 -1.38989878e+00
6.48782670e-01 -1.05477065e-01 -1.75492540e-01 -3.88267308e-01
-2.83423990e-01 1.01249091e-01 1.68365061e-01 -1.18330531e-01
3.37210298e-01 -9.11644995e-01 -9.91553247e-01 1.49107277e-01
-3.95362586e-01 -1.41212726e+00 -5.54817379e-01 4.84282017e-01
-1.23244464e+00 -7.16309622e-02 -8.36153865e-01 -6.50756001e-01
5.04898280e-02 -1.41569531e+00 7.42901444e-01 2.90878505e-01
-3.29873949e-01 -7.17870712e-01 6.89878643e-01 8.28967839e-02
5.28464377e-01 4.98833088e-03 7.69935012e-01 -5.71626961e-01
-5.46364903e-01 3.84115875e-01 1.50751069e-01 -1.84918687e-01
-3.30987908e-02 -1.96086630e-01 -8.80649805e-01 -5.64013124e-01
-2.24913135e-01 -1.04838407e+00 5.58157682e-01 2.41738528e-01
1.25751150e+00 3.18210348e-02 -9.21790153e-02 6.21975183e-01
1.36959040e+00 3.82716626e-01 6.93022490e-01 7.29579508e-01
1.99131653e-01 2.49366239e-01 5.62589347e-01 6.28501415e-01
4.61741328e-01 5.65862954e-01 4.13917333e-01 2.29180932e-01
1.63627639e-01 -5.90885997e-01 2.41645232e-01 4.62385774e-01
-1.37747258e-01 -2.98128158e-01 -9.35019851e-01 4.86984462e-01
-1.61895669e+00 -1.01667213e+00 6.86215937e-01 2.26249623e+00
1.06847513e+00 4.23454732e-01 5.35586119e-01 -2.16036186e-01
1.17505111e-01 1.00326255e-01 -7.48222947e-01 -4.25336897e-01
2.91180581e-01 6.36756778e-01 4.69846010e-01 5.87345839e-01
-7.62543976e-01 1.31833518e+00 8.20053577e+00 6.23547792e-01
-7.26959109e-01 -1.52022764e-01 6.43074214e-01 -3.75721872e-01
-3.93633842e-01 -2.22739935e-01 -4.78493333e-01 2.99473107e-01
1.16467214e+00 -2.73302585e-01 1.08058369e+00 1.10692537e+00
2.50758320e-01 -3.97635281e-01 -1.04643440e+00 4.45454687e-01
-4.64659005e-01 -1.24187076e+00 -2.89737970e-01 1.74212769e-01
9.41201448e-01 3.02343398e-01 2.04566956e-01 1.05581307e+00
1.39400280e+00 -1.32964432e+00 6.19049072e-01 -7.31397569e-02
5.26924431e-01 -9.34629679e-01 3.42193782e-01 3.32926810e-01
-8.98035526e-01 -9.44570228e-02 -3.99425238e-01 -3.23446333e-01
-2.19665378e-01 -2.50165313e-01 -8.66675258e-01 1.51311442e-01
5.82586706e-01 1.93039313e-01 -6.65372431e-01 1.35791099e+00
-3.54297191e-01 7.37166643e-01 -3.37955177e-01 -2.25794300e-01
2.46525347e-01 -9.50628668e-02 1.01241857e-01 7.95787454e-01
1.57013938e-01 1.04826592e-01 2.55714953e-01 1.01094413e+00
-1.23016693e-01 -5.54941632e-02 -7.66189158e-01 -1.83420926e-01
9.48150992e-01 8.98241878e-01 -8.30510378e-01 1.96230654e-02
-5.40003814e-02 6.98067427e-01 7.81895518e-01 4.55795765e-01
-1.16563654e+00 -1.16729915e-01 7.56807923e-01 -2.46177360e-01
2.94964880e-01 -5.66299617e-01 -3.84425163e-01 -8.48271489e-01
-4.26581800e-01 -1.40768170e+00 1.80049747e-01 -7.79894471e-01
-9.32111740e-01 5.30106306e-01 -6.79762587e-02 -8.40713203e-01
-5.24576604e-01 -2.86415368e-01 -4.57815945e-01 7.02698469e-01
-1.25784516e+00 -4.50935841e-01 -4.52800304e-01 3.48862320e-01
2.83852965e-01 -1.21078670e-01 1.05300260e+00 -2.17635840e-01
-5.72532058e-01 7.40120888e-01 3.82166840e-02 -2.41769150e-01
5.69177628e-01 -1.30235100e+00 7.10639834e-01 4.84494209e-01
5.35823070e-02 7.63512909e-01 8.08088958e-01 -6.16977572e-01
-1.26570964e+00 -8.78721654e-01 -3.06487232e-01 -7.03282952e-01
4.01420891e-01 -4.22419429e-01 -3.94599199e-01 7.67959177e-01
1.71393290e-01 -1.98235750e-01 6.93818510e-01 4.38512266e-01
-8.61273259e-02 3.01019400e-01 -1.18168676e+00 1.08175278e+00
1.45294452e+00 -2.71601528e-01 -4.02226686e-01 -1.23524897e-01
7.06254125e-01 -7.04265416e-01 -9.00162816e-01 4.04177636e-01
5.03539324e-01 -1.31221211e+00 8.58987510e-01 -5.69591701e-01
5.79166591e-01 -5.17465249e-02 -1.15997672e-01 -1.80418932e+00
-2.57182181e-01 -5.59288502e-01 1.44300252e-01 7.85995185e-01
5.60516477e-01 -6.21968448e-01 1.25049126e+00 5.17532527e-01
2.12112367e-02 -6.95979476e-01 -4.54926461e-01 -7.87132800e-01
4.88977104e-01 -4.82481956e-01 7.75313914e-01 5.98509908e-01
1.62225291e-01 1.24240868e-01 -1.69056475e-01 -1.49607241e-01
3.38229269e-01 8.98247492e-03 1.16899228e+00 -6.37105346e-01
-8.45269561e-01 -6.34916902e-01 1.05274856e-01 -1.14230204e+00
5.53898551e-02 -4.79119241e-01 4.62490261e-01 -1.24001658e+00
3.18268001e-01 -9.14784551e-01 -1.37184963e-01 3.08277786e-01
-3.87475252e-01 1.47008464e-01 3.03591132e-01 1.15515087e-02
-1.06803155e+00 9.07740414e-01 1.53022802e+00 4.11261141e-01
-4.09680545e-01 -1.81995079e-01 -9.01552200e-01 4.20213491e-01
1.09265339e+00 -2.94718623e-01 -9.88249362e-01 -4.23200995e-01
2.92162359e-01 8.50083977e-02 9.28344876e-02 -1.51213777e+00
-1.03385471e-01 -6.15075588e-01 3.94567758e-01 3.12705524e-02
2.64790416e-01 -4.63569134e-01 -4.76603508e-02 4.03988838e-01
-7.04795361e-01 3.82983893e-01 6.21878803e-01 4.15482879e-01
2.36478150e-01 -2.48880759e-01 7.15406537e-01 -4.84362811e-01
-6.19692147e-01 1.51960999e-01 -6.98690355e-01 7.34440446e-01
1.08135390e+00 -2.31078729e-01 -3.70570034e-01 -6.84958220e-01
-2.45496199e-01 3.01288217e-01 9.64435339e-01 1.49379745e-01
3.15884352e-01 -1.04466367e+00 -3.47453982e-01 1.82458684e-01
1.50942862e-01 -9.93603319e-02 -7.03990534e-02 2.20072702e-01
-8.11869860e-01 1.06749386e-01 -5.07147491e-01 -1.72602981e-01
-8.56148303e-01 3.30573559e-01 7.22833991e-01 -3.62852365e-01
-3.91543627e-01 9.70330596e-01 1.69034246e-02 -7.73220837e-01
3.18182319e-01 -2.47256562e-01 8.85910839e-02 -7.22019613e-01
3.38357806e-01 3.22794527e-01 -1.85532629e-01 1.19052552e-01
-1.36726111e-01 -2.12573975e-01 2.42389068e-02 -5.60075700e-01
1.25739682e+00 2.77601629e-01 4.48487282e-01 5.48986383e-02
4.63166654e-01 1.10211946e-01 -1.80729795e+00 1.42680928e-01
-1.10283159e-01 -3.83730799e-01 -1.03643037e-01 -9.49184120e-01
-7.57400453e-01 2.65511155e-01 5.40751219e-01 1.84641302e-01
8.66644323e-01 -3.69345337e-01 4.27416950e-01 7.24419534e-01
1.08789158e+00 -1.03879714e+00 3.16977382e-01 7.37428665e-01
6.67728305e-01 -9.77207422e-01 -3.04638501e-03 1.72131695e-02
-7.74469137e-01 4.69978809e-01 1.10712171e+00 -4.19151276e-01
1.03102863e-01 5.79948843e-01 -1.17555149e-01 -2.01312795e-01
-9.99042392e-01 -3.04233640e-01 -5.06030440e-01 9.45484519e-01
4.17406589e-01 1.67428344e-01 -1.08619705e-01 3.87956440e-01
-8.40186298e-01 1.59650922e-01 6.03951991e-01 1.36693382e+00
-5.59178054e-01 -1.35564363e+00 -8.72545689e-02 4.62147266e-01
-1.66421428e-01 -2.77553801e-03 -1.98798552e-01 1.07581294e+00
-1.99625090e-01 9.05837238e-01 -2.49898061e-02 -4.46260065e-01
3.34717512e-01 -7.53850043e-02 1.03698492e+00 -8.03989172e-01
-7.34222710e-01 -4.17314261e-01 4.14463341e-01 -8.52485120e-01
-5.51934205e-02 -5.46001673e-01 -1.20342386e+00 -6.50923014e-01
4.39134389e-02 3.55818659e-01 2.82372832e-01 6.67135537e-01
3.07012081e-01 6.41721427e-01 3.39631498e-01 -6.42346919e-01
-7.51612067e-01 -8.85410726e-01 -5.03286898e-01 3.58576328e-01
-9.81998071e-02 -9.48797405e-01 -3.00473988e-01 -3.65637988e-01]
|
[3.929730176925659, 1.5652248859405518]
|
93d7467d-a572-4a1f-8388-23866a20bea1
|
reference-guided-image-inpainting-using
|
2301.08044
| null |
https://arxiv.org/abs/2301.08044v1
|
https://arxiv.org/pdf/2301.08044v1.pdf
|
Reference Guided Image Inpainting using Facial Attributes
|
Image inpainting is a technique of completing missing pixels such as occluded region restoration, distracting objects removal, and facial completion. Among these inpainting tasks, facial completion algorithm performs face inpainting according to the user direction. Existing approaches require delicate and well controlled input by the user, thus it is difficult for an average user to provide the guidance sufficiently accurate for the algorithm to generate desired results. To overcome this limitation, we propose an alternative user-guided inpainting architecture that manipulates facial attributes using a single reference image as the guide. Our end-to-end model consists of attribute extractors for accurate reference image attribute transfer and an inpainting model to map the attributes realistically and accurately to generated images. We customize MS-SSIM loss and learnable bidirectional attention maps in which importance structures remain intact even with irregular shaped masks. Based on our evaluation using the publicly available dataset CelebA-HQ, we demonstrate that the proposed method delivers superior performance compared to some state-of-the-art methods specialized in inpainting tasks.
|
['Hanseok Ko', 'Youngsaeng Jin', 'David Han', 'Yuanming Li', 'Jeonggi Kwak', 'Dongsik Yoon']
|
2023-01-19
| null | null | null | null |
['facial-inpainting', 'image-inpainting', 'ms-ssim']
|
['computer-vision', 'computer-vision', 'computer-vision']
|
[ 5.06935298e-01 1.59723371e-01 5.68703935e-02 -7.03425527e-01
-9.39259887e-01 -2.86012292e-01 3.76750112e-01 -4.15807456e-01
-2.21004814e-01 9.23212707e-01 8.27415213e-02 2.02726275e-01
2.98349500e-01 -5.92124760e-01 -1.01024878e+00 -5.60179412e-01
4.80694950e-01 5.28220654e-01 -2.97859639e-01 -2.00317562e-01
1.99489266e-01 6.96362317e-01 -1.73125374e+00 4.33741480e-01
1.19168174e+00 1.14921165e+00 2.50145108e-01 4.57835734e-01
-1.47507608e-01 6.18966103e-01 -4.38170820e-01 -6.80791080e-01
4.76511508e-01 -5.40802062e-01 -5.23627102e-01 4.67461020e-01
9.81882095e-01 -6.48350358e-01 -1.39726341e-01 9.79050815e-01
5.44300258e-01 9.85801592e-02 7.68101871e-01 -1.19754875e+00
-1.07064497e+00 1.17869012e-01 -9.05941427e-01 -3.18913370e-01
5.26758373e-01 1.77303135e-01 6.77988350e-01 -1.18087590e+00
8.78857493e-01 1.51016927e+00 5.22340059e-01 8.73216331e-01
-1.47958207e+00 -9.06178355e-01 3.95034403e-02 6.65606791e-03
-1.39623785e+00 -9.29796338e-01 1.01930225e+00 -2.42736295e-01
3.73516917e-01 3.81592214e-01 5.24217665e-01 1.08122325e+00
1.47231251e-01 6.80526316e-01 1.11822820e+00 -4.17338997e-01
1.29399523e-02 1.50347948e-01 -5.52226424e-01 7.37259328e-01
-3.04530352e-01 1.50335237e-01 -5.87732494e-01 -2.44315967e-01
9.85195994e-01 1.30466847e-02 -4.44901168e-01 -3.85651618e-01
-9.41337645e-01 5.39337218e-01 3.07922870e-01 -3.85748595e-01
-6.22726977e-01 2.45652452e-01 7.07251430e-02 4.44192916e-01
8.24740410e-01 2.34054565e-01 -2.45219529e-01 3.27281028e-01
-1.36462450e+00 3.82215470e-01 2.31264859e-01 1.17088151e+00
8.96783233e-01 1.47811696e-01 -4.76771891e-01 1.06481612e+00
7.26717487e-02 4.88052636e-01 3.17565501e-02 -1.31089258e+00
3.38869154e-01 3.23550254e-01 4.50426012e-01 -8.97939801e-01
1.45386562e-01 -3.01563174e-01 -7.92171359e-01 6.62960351e-01
2.39838272e-01 1.19643390e-01 -1.15670550e+00 1.91901112e+00
5.44250548e-01 1.91296101e-01 -2.77054816e-01 1.06052268e+00
7.03832805e-01 6.02615654e-01 1.75565168e-01 -1.84892222e-01
1.36889708e+00 -1.13999474e+00 -9.74767923e-01 -1.38817251e-01
-1.85788088e-02 -1.03284729e+00 1.50978196e+00 3.98030072e-01
-1.46697628e+00 -6.55903876e-01 -7.66239882e-01 -5.75036168e-01
1.30135417e-02 4.92887795e-01 4.97534782e-01 3.20464313e-01
-1.24643743e+00 6.15323901e-01 -5.70613027e-01 -2.40681261e-01
8.71188819e-01 3.68136555e-01 -8.14137816e-01 -1.42531276e-01
-8.97037745e-01 7.42380023e-01 -8.90802816e-02 1.54549003e-01
-1.14570498e+00 -1.02290201e+00 -8.37943912e-01 4.41945717e-02
4.69681202e-03 -1.00892782e+00 1.09533370e+00 -1.52874184e+00
-1.54066670e+00 1.35225952e+00 -3.32919508e-01 -1.36173621e-01
8.42071235e-01 -4.00241941e-01 -2.37039104e-03 2.71947891e-01
2.23149300e-01 1.19753563e+00 1.50942230e+00 -1.44909024e+00
-4.20309484e-01 -4.89868581e-01 -1.14179894e-01 3.44361693e-01
-3.53170902e-01 1.45140186e-01 -7.16719270e-01 -1.01893485e+00
-5.00699133e-02 -6.46003127e-01 6.13346808e-02 8.32071185e-01
-2.63647586e-01 2.07125261e-01 1.08206081e+00 -9.87281203e-01
8.97594512e-01 -2.18732357e+00 2.27728143e-01 -9.63655040e-02
1.02551132e-01 3.53282504e-02 -4.07986492e-01 1.31876796e-01
-2.90455073e-01 -2.94792980e-01 -3.67008656e-01 -1.07389903e+00
-2.67369837e-01 2.38039106e-01 -5.10806680e-01 5.75272262e-01
4.43660051e-01 6.32352054e-01 -7.25901246e-01 -5.99267960e-01
6.11077845e-02 8.94211590e-01 -7.02521384e-01 7.49027014e-01
-2.54216433e-01 8.60230803e-01 -2.59096026e-01 9.86774087e-01
1.06673634e+00 1.50233567e-01 -4.22419190e-01 -3.93945664e-01
1.79402262e-01 -1.62909374e-01 -8.48979771e-01 2.11819363e+00
-6.40975296e-01 5.77810168e-01 6.65590882e-01 -5.26750863e-01
9.44168806e-01 3.04198712e-01 2.74707764e-01 -4.75187629e-01
1.78951696e-02 1.51567400e-01 -5.15747190e-01 -5.15322030e-01
4.34925258e-01 -6.89330995e-02 4.43162143e-01 4.57969099e-01
8.09870213e-02 -2.62004822e-01 -7.44858310e-02 1.22845434e-01
6.42589808e-01 5.82984030e-01 -4.93857861e-02 -1.52206346e-01
6.17103875e-01 -4.12827730e-01 4.99300778e-01 2.53566563e-01
9.13339257e-02 1.31167650e+00 4.07133311e-01 -5.97562790e-01
-1.26547229e+00 -1.03695071e+00 -7.08150715e-02 1.26440644e+00
-3.96314310e-04 -1.18735842e-01 -1.18610036e+00 -5.18582582e-01
-1.09041296e-01 7.08339870e-01 -9.15283740e-01 9.63206291e-02
-4.72502053e-01 -3.17395896e-01 3.06695580e-01 4.09343183e-01
6.60378516e-01 -1.31768787e+00 -3.89970124e-01 4.77687642e-02
-4.41497952e-01 -1.08279920e+00 -1.05473888e+00 -4.21040326e-01
-8.45442951e-01 -8.60525966e-01 -9.09471631e-01 -8.24341714e-01
1.35225749e+00 1.35085762e-01 1.21798098e+00 2.62594789e-01
-3.88746619e-01 1.91691458e-01 -9.10161436e-02 -2.32714355e-01
-4.17420059e-01 -5.72337918e-02 -2.11021945e-01 5.43240130e-01
-2.64976740e-01 -8.43736649e-01 -9.59710121e-01 2.69120187e-01
-1.08061671e+00 3.77750278e-01 5.90555966e-01 9.34411228e-01
7.83785462e-01 -4.91729498e-01 3.76540780e-01 -9.93422747e-01
5.52090108e-01 -1.95920989e-01 -5.16175330e-01 1.86511800e-01
-4.45580900e-01 -2.18155924e-02 6.45321727e-01 -5.28183401e-01
-1.25204396e+00 3.84105057e-01 -3.00147474e-01 -8.75053585e-01
-5.67206256e-02 -2.38655835e-01 -4.69227672e-01 -2.56640464e-01
5.42366862e-01 3.20339382e-01 3.39258075e-01 -4.48861808e-01
4.96166885e-01 6.27204716e-01 8.14192533e-01 -8.30817282e-01
7.53350139e-01 7.56228685e-01 -8.78568590e-02 -5.56296945e-01
-7.12044239e-01 1.25205383e-01 -5.75307846e-01 -3.19347322e-01
6.58163309e-01 -9.32175517e-01 -5.37530899e-01 4.62802708e-01
-1.36773753e+00 -3.17559689e-01 -2.34315366e-01 -8.01375806e-02
-9.19167101e-01 1.31323114e-01 -5.28870106e-01 -6.14370406e-01
-7.30257392e-01 -1.28367341e+00 1.51359069e+00 9.92872268e-02
-2.09965169e-01 -5.51880717e-01 -3.56124818e-01 4.25564677e-01
5.93546927e-01 5.24424911e-01 7.48925865e-01 1.44061685e-01
-7.95011520e-01 4.36752252e-02 -2.94433743e-01 3.36773038e-01
3.14782530e-01 6.52696118e-02 -1.20707810e+00 -2.65019625e-01
1.58095546e-02 -4.47421223e-01 8.01295340e-01 2.66790658e-01
1.49803782e+00 -4.32963490e-01 -2.29177624e-01 1.09576166e+00
1.11363780e+00 -8.83615837e-02 1.11884260e+00 1.02273598e-01
7.02494562e-01 7.89824188e-01 8.44314992e-01 4.48816597e-01
1.18596427e-01 7.04242468e-01 5.37354767e-01 -4.86130923e-01
-3.88158917e-01 -4.53770339e-01 1.79978549e-01 -2.28420715e-03
-1.81132585e-01 4.06632992e-03 -2.50106782e-01 4.99750108e-01
-1.66735530e+00 -9.63115990e-01 2.85876483e-01 2.13815308e+00
1.27758193e+00 -3.50367188e-01 -1.82594553e-01 -1.20775796e-01
6.76715076e-01 1.32909361e-02 -4.79304254e-01 -2.29637355e-01
2.20786240e-02 2.43371069e-01 7.98885599e-02 7.86520839e-01
-8.87399435e-01 1.16841662e+00 5.76695442e+00 1.07746756e+00
-1.15793228e+00 2.44182035e-01 1.15910316e+00 -1.21179737e-01
-5.17394006e-01 -7.13004321e-02 -5.83068609e-01 5.43717444e-01
3.15480173e-01 3.22661221e-01 5.97027183e-01 6.95737302e-01
4.88273591e-01 -7.09698945e-02 -1.15046251e+00 1.16770136e+00
3.24113280e-01 -1.24229455e+00 3.37266594e-01 -2.41717622e-01
6.90966904e-01 -6.55634284e-01 4.19610769e-01 -1.38139993e-01
-1.25790626e-01 -1.25894451e+00 8.65620434e-01 8.08422446e-01
1.35209525e+00 -8.40372384e-01 1.38191000e-01 -1.24017335e-02
-7.37977087e-01 8.95825773e-02 -4.33655292e-01 2.16144755e-01
8.83545578e-02 4.80459034e-01 -6.53761208e-01 1.78484961e-01
5.52200258e-01 4.27175254e-01 -4.13608789e-01 7.24308252e-01
-3.31357330e-01 2.53985256e-01 -1.26895696e-01 7.05755413e-01
-1.19969301e-01 -4.61644173e-01 3.83867949e-01 8.40196192e-01
5.28565109e-01 2.62556702e-01 -3.01059067e-01 1.13283646e+00
-3.60940456e-01 1.48221955e-01 -5.49442053e-01 3.16281706e-01
3.71726781e-01 1.46605349e+00 -3.55208546e-01 -1.14031516e-01
-2.84261912e-01 1.56969452e+00 4.44434732e-01 5.02624094e-01
-1.00375783e+00 -2.85067946e-01 8.66486371e-01 5.83898187e-01
2.38254547e-01 5.50147779e-02 -3.03343534e-01 -8.99702489e-01
1.16245396e-01 -8.37984979e-01 1.47261262e-01 -1.30925810e+00
-1.10757780e+00 9.78924334e-01 -1.38275042e-01 -1.19580615e+00
-2.15963513e-01 -2.08950624e-01 -5.35754561e-01 1.12800741e+00
-1.68168461e+00 -1.61373997e+00 -7.01791108e-01 7.95712352e-01
6.89330637e-01 -1.41877040e-01 7.64685690e-01 5.37620068e-01
-3.08507353e-01 9.52995539e-01 -3.26989859e-01 -1.41391620e-01
1.14986682e+00 -8.20047259e-01 2.96828866e-01 6.80372417e-01
-1.51793331e-01 4.89688784e-01 1.00285733e+00 -5.67894161e-01
-1.24084580e+00 -1.31274343e+00 6.94458365e-01 -3.22901487e-01
-1.20628532e-02 -4.36656386e-01 -9.11872506e-01 7.35488594e-01
4.79833096e-01 4.70784068e-01 2.66911209e-01 -6.19299054e-01
-3.30078334e-01 -5.16461790e-01 -1.51873839e+00 6.97989464e-01
1.15906644e+00 -2.39073515e-01 -8.80411044e-02 3.59499842e-01
5.13152599e-01 -6.34559155e-01 -5.12790442e-01 2.48306379e-01
5.14282107e-01 -1.04757071e+00 9.71333921e-01 -4.56366748e-01
7.21977532e-01 -3.87767166e-01 6.71015233e-02 -1.28921509e+00
-6.87909350e-02 -1.12137794e+00 1.15458332e-01 1.45289934e+00
1.81892887e-01 -3.23384762e-01 8.98564100e-01 8.11068475e-01
-1.75631791e-01 -8.87732148e-01 -6.63586378e-01 -2.07812443e-01
-2.86045134e-01 -2.43671257e-02 8.36182773e-01 7.80457795e-01
-6.80526853e-01 1.94256082e-02 -6.79892242e-01 5.80708981e-02
8.76953840e-01 2.26664215e-01 9.38875735e-01 -9.50614750e-01
-3.55180688e-02 -9.14989263e-02 -1.14501230e-01 -9.66445684e-01
5.07749557e-01 -5.30810177e-01 -7.14375600e-02 -1.26175177e+00
1.15395561e-01 -5.15556395e-01 2.37992272e-01 6.10554099e-01
-2.25453526e-01 6.67841315e-01 -8.16726163e-02 2.03444585e-01
-1.71755776e-01 8.96686018e-01 1.72963560e+00 -7.97118545e-02
-1.11209869e-01 -5.90121299e-02 -8.19360137e-01 6.18736744e-01
5.65755725e-01 -4.35122579e-01 -6.32249594e-01 -6.47913933e-01
-1.26701016e-02 2.21032038e-01 4.08787817e-01 -7.90407538e-01
3.58851738e-02 -2.91188419e-01 7.77275383e-01 -4.55622554e-01
7.13711083e-01 -8.72440457e-01 3.45323950e-01 3.39266732e-02
-4.55475211e-01 1.77282676e-01 1.52959794e-01 3.12902063e-01
-2.08687991e-01 -1.79858673e-02 1.10239601e+00 -3.69901881e-02
-4.22931999e-01 7.15058804e-01 1.95403501e-01 -4.37372848e-02
1.06868744e+00 -3.32571208e-01 8.62256438e-02 -7.97331333e-01
-6.81415379e-01 6.85914885e-03 7.26882100e-01 1.89747080e-01
1.08050847e+00 -1.44653368e+00 -1.03612626e+00 6.50301576e-01
-8.64040386e-03 -4.75283898e-02 2.81427681e-01 6.61949337e-01
-7.67862201e-01 -1.55555800e-01 -6.66212380e-01 -4.13589925e-01
-1.37537825e+00 4.67735469e-01 2.96878189e-01 2.00825125e-01
-5.03823757e-01 8.14259589e-01 4.69961047e-01 -2.54709899e-01
3.26531380e-01 9.30374265e-02 6.02363907e-02 -9.71435010e-02
6.55185103e-01 2.49214564e-02 1.02986634e-01 -6.08685613e-01
1.30701419e-02 7.08723366e-01 -3.43500912e-01 -2.25620136e-01
1.18930292e+00 -3.57204527e-01 -3.66161168e-01 -2.89021373e-01
9.88116562e-01 8.43308866e-02 -1.81693709e+00 5.44885285e-02
-8.10819685e-01 -9.68996346e-01 -1.16516486e-01 -7.14953780e-01
-1.49375188e+00 9.09371793e-01 5.40884972e-01 -4.43811506e-01
1.65264213e+00 -2.88371474e-01 8.52155924e-01 -1.09273553e-01
3.08582455e-01 -8.65648687e-01 1.31522521e-01 -1.24939522e-02
1.50133860e+00 -1.11085141e+00 1.84545654e-03 -7.60155439e-01
-6.55634403e-01 9.29795027e-01 8.61574113e-01 -1.71314046e-01
4.17169750e-01 2.86526442e-01 2.43025824e-01 1.07376985e-01
-6.00975871e-01 1.75487265e-01 2.17847690e-01 7.95253932e-01
4.18735266e-01 -2.65222192e-01 -2.96301663e-01 2.74796367e-01
-1.12208568e-01 5.79798315e-03 3.11202794e-01 4.72232282e-01
-4.14223634e-02 -1.18753314e+00 -6.10950410e-01 2.33234137e-01
-5.61650157e-01 -1.42472029e-01 -1.54744655e-01 6.39540017e-01
6.93983957e-02 5.56173980e-01 1.56293049e-01 4.84455489e-02
2.10754916e-01 -1.50704741e-01 6.75729275e-01 -5.59869587e-01
-4.02090490e-01 4.98645231e-02 -7.59677961e-02 -7.95597017e-01
-2.32208684e-01 -3.71818662e-01 -1.05006528e+00 -2.53055394e-01
6.36764802e-03 -1.88330933e-01 4.93085355e-01 5.25260985e-01
7.76263535e-01 2.48150870e-01 7.11116552e-01 -1.12795913e+00
-3.49305332e-01 -1.02430618e+00 -4.15968329e-01 7.39377916e-01
3.14943403e-01 -7.44476855e-01 -1.26272172e-01 5.13343275e-01]
|
[12.599967002868652, -0.2276587039232254]
|
8133c4af-62a5-4d12-b814-867b84c8cfdd
|
the-guide-and-the-explorer-smart-agents-for
| null | null |
https://openreview.net/forum?id=G9JXCpShpni
|
https://openreview.net/pdf?id=G9JXCpShpni
|
The guide and the explorer: smart agents for resource-limited iterated batch reinforcement learning
|
Iterated batch reinforcement learning (RL) is a growing subfield fueled by the demand from systems engineers for intelligent control solutions that they can apply within their technical and organizational constraints. Model-based RL (MBRL) suits this scenario well for its sample efficiency and modularity. Recent MBRL techniques combine efficient neural system models with classical planning (like model predictive control; MPC). In this paper we add two components to this classical setup. The first is a Dyna-style policy learned on the system model using model-free techniques. We call it the guide since it guides the planner. The second component is the explorer, a strategy to expand the limited knowledge of the guide during planning. Through a rigorous ablation study we show that exploration is crucial for optimal performance. We apply this approach with a DQN guide and a heating explorer to improve the state of the art of the resource-limited Acrobot benchmark system by about 10%.
|
['Gabriel Hurtado', 'Othman Gaizi', 'Balázs Kégl', 'Albert Thomas']
|
2021-09-29
| null | null | null | null |
['acrobot']
|
['playing-games']
|
[ 1.50164083e-01 4.44305927e-01 -5.82454443e-01 2.26072341e-01
-4.40064490e-01 -7.41269946e-01 6.78696632e-01 -8.58422294e-02
-5.23514926e-01 9.65268970e-01 -5.02351895e-02 -5.17963350e-01
-3.89727026e-01 -6.54251397e-01 -8.20627630e-01 -8.14091742e-01
-2.61808008e-01 5.99693477e-01 1.52883276e-01 -5.44920921e-01
3.16055298e-01 5.87705135e-01 -1.26609325e+00 -1.14374444e-01
7.76678085e-01 1.04282475e+00 5.73419988e-01 6.83205545e-01
1.89370826e-01 1.07034838e+00 -1.84360981e-01 4.88063186e-01
6.56839371e-01 -4.76476252e-01 -9.35832918e-01 -1.27008468e-01
-5.06950140e-01 -2.90857166e-01 -8.32187086e-02 5.54947615e-01
4.74407047e-01 4.92549449e-01 1.07685231e-01 -1.17063713e+00
9.91747007e-02 7.63136327e-01 -4.40323204e-01 -9.78712961e-02
7.20480178e-03 6.24346733e-01 8.97400558e-01 -1.77531108e-01
8.75494003e-01 1.03766513e+00 3.00751060e-01 7.12881267e-01
-1.38456297e+00 -2.73261666e-01 4.03886855e-01 1.46671697e-01
-1.14426422e+00 -3.39316577e-01 6.14787281e-01 -2.47821271e-01
1.47553945e+00 2.35292748e-01 1.01271343e+00 8.74062002e-01
4.46421415e-01 7.34262705e-01 1.23795247e+00 -6.52742386e-01
9.18001890e-01 -8.82304925e-03 -2.66054571e-01 7.32473969e-01
-1.38420507e-01 8.80398512e-01 -1.11517161e-01 9.04623419e-02
8.07091594e-01 -3.13127577e-01 -8.60033706e-02 -8.02255034e-01
-9.82011497e-01 9.47924197e-01 6.73874557e-01 1.22322388e-01
-3.95056129e-01 6.32474422e-01 3.00899684e-01 5.32640398e-01
-9.44139287e-02 1.41589200e+00 -5.68586588e-01 -3.37070882e-01
-8.38044822e-01 5.06381631e-01 1.01853180e+00 1.01208901e+00
5.96947849e-01 1.94813564e-01 -1.00513518e-01 4.98260230e-01
1.49062097e-01 6.23171329e-02 4.76155818e-01 -1.37975407e+00
2.03317836e-01 6.27590537e-01 4.36902255e-01 -3.47474098e-01
-6.84785962e-01 -5.26794672e-01 -5.20122528e-01 6.27502859e-01
1.80290505e-01 -4.19851035e-01 -9.26924706e-01 1.67209136e+00
4.08394217e-01 -1.73544124e-01 4.66441996e-02 7.07665145e-01
-2.38506854e-01 8.56978178e-01 -1.34894595e-01 -4.26635891e-01
8.76139402e-01 -1.45290518e+00 -3.20681632e-01 -3.70514870e-01
6.67357147e-01 -1.42222285e-01 8.85654092e-01 7.75007308e-01
-1.15945029e+00 -3.01023543e-01 -1.26997602e+00 4.20226663e-01
-3.52755338e-01 -1.32464105e-02 6.56237006e-01 4.88572866e-01
-1.15269315e+00 9.86474395e-01 -1.21486521e+00 -3.74410212e-01
6.09954782e-02 6.70942247e-01 1.85518578e-01 1.76240832e-01
-8.87181103e-01 1.37403989e+00 7.13829219e-01 4.38224757e-03
-1.38832247e+00 -4.29272115e-01 -5.82250714e-01 8.10428783e-02
1.28855956e+00 -4.89314228e-01 1.71930325e+00 -6.43180609e-01
-2.29399014e+00 1.65175110e-01 4.16433901e-01 -8.66718650e-01
6.14345968e-01 -2.12903634e-01 1.75373435e-01 6.67594671e-02
-5.12260318e-01 6.13618314e-01 9.61876273e-01 -1.17806435e+00
-7.59437382e-01 2.30117500e-01 4.51886296e-01 2.12000757e-01
1.86649621e-01 -4.35065418e-01 -1.45641133e-01 -1.75568029e-01
-2.46142864e-01 -1.28334200e+00 -8.65465462e-01 -4.12054539e-01
-4.14209306e-01 -1.60561696e-01 5.16282797e-01 -3.29422444e-01
1.20594096e+00 -1.58887649e+00 6.18994772e-01 3.51458371e-01
-7.89101422e-02 1.87441617e-01 -1.68150380e-01 8.17435324e-01
1.04912415e-01 6.57342002e-03 -4.52128723e-02 -4.56554666e-02
4.20566469e-01 4.63169515e-01 -4.55124378e-01 3.70795071e-01
1.69447884e-01 1.01079273e+00 -9.40179050e-01 -2.67571002e-01
2.65084326e-01 -1.10232100e-01 -7.14397371e-01 3.11874777e-01
-9.79112089e-01 5.68438888e-01 -6.32145166e-01 4.74518836e-01
1.24150723e-01 -4.42261063e-02 4.34112400e-01 1.26900926e-01
-6.16621971e-01 -6.94111362e-02 -1.02961671e+00 1.80766845e+00
-6.57824576e-01 2.05800533e-01 4.36833292e-01 -1.03305578e+00
8.52330029e-01 7.77302161e-02 5.03013015e-01 -9.52398598e-01
1.64289430e-01 1.98591799e-01 5.78626171e-02 -4.27238017e-01
5.46275795e-01 -8.75969455e-02 -5.21840453e-02 3.64152670e-01
1.39684118e-02 -6.15849495e-01 3.73848915e-01 -9.43841487e-02
1.33457100e+00 7.48457849e-01 6.94092453e-01 -5.95800698e-01
4.31805074e-01 2.96624869e-01 6.23363853e-01 9.71874833e-01
-6.21353276e-03 -4.60621081e-02 7.32050359e-01 -6.04187548e-01
-1.22200286e+00 -6.83510482e-01 3.29290450e-01 1.18200111e+00
-4.43745442e-02 -3.09331834e-01 -6.85984969e-01 -5.45859218e-01
-1.96420297e-01 1.06116259e+00 -6.11768246e-01 -2.21260935e-01
-9.11881030e-01 -3.70132357e-01 9.24962759e-02 4.65561539e-01
2.09635824e-01 -1.28724062e+00 -1.56844187e+00 5.13949215e-01
4.70405459e-01 -6.78464890e-01 -1.09789528e-01 8.17510664e-01
-7.41004765e-01 -9.08712804e-01 -4.74788278e-01 -3.45756739e-01
3.77852529e-01 -3.80975366e-01 1.04884374e+00 3.95645164e-02
-2.24115044e-01 4.05909717e-01 -3.40971470e-01 -3.22847456e-01
-6.43849492e-01 4.35848892e-01 1.30009472e-01 -5.05000055e-01
-4.52017069e-01 -5.27209878e-01 -4.38696027e-01 2.96305984e-01
-6.72261119e-01 4.08984870e-01 8.22468162e-01 9.75997388e-01
7.68388689e-01 1.19875260e-01 5.03215969e-01 -4.00419682e-01
6.69987917e-01 -2.52628267e-01 -1.40032828e+00 3.93589407e-01
-1.01188743e+00 6.35463119e-01 8.81885886e-01 -4.57803369e-01
-8.73705149e-01 4.13473785e-01 1.03046089e-01 -2.97319829e-01
-9.30599310e-03 4.93645698e-01 6.11794228e-03 -3.11081082e-01
5.14396191e-01 1.79907918e-01 1.74536318e-01 -3.85904938e-01
4.76173192e-01 5.64871542e-02 3.26420963e-01 -9.19892967e-01
5.43825507e-01 9.05339122e-02 3.95324111e-01 -5.03895283e-01
-5.10577440e-01 -1.18644079e-02 -2.34508336e-01 -2.51821458e-01
6.78725004e-01 -4.26461339e-01 -1.06713343e+00 -1.10672694e-02
-8.38572145e-01 -1.23140335e+00 -7.74883986e-01 2.93625474e-01
-1.27186120e+00 -2.24696264e-01 -4.46633756e-01 -1.22547460e+00
-2.02838853e-01 -1.39242494e+00 6.92479193e-01 3.47825885e-01
-1.04195192e-01 -6.42746389e-01 5.40603578e-01 -3.50917757e-01
7.10908711e-01 5.63537657e-01 9.83069241e-01 -5.03201008e-01
-6.92104697e-01 5.62084541e-02 2.88551748e-01 1.61944494e-01
-3.89191180e-01 -1.72670633e-01 -6.85621202e-01 -5.98837972e-01
1.61025319e-02 -4.47506279e-01 6.51075184e-01 3.88926595e-01
9.79455292e-01 -5.81707597e-01 -5.36130071e-01 4.54865754e-01
1.73393941e+00 5.03504395e-01 4.13204521e-01 6.75309718e-01
3.28419566e-01 5.46299577e-01 7.28800714e-01 4.46313620e-01
-6.64772093e-02 8.72177064e-01 8.27467203e-01 2.46825695e-01
4.04320210e-01 -2.91180074e-01 4.26182657e-01 4.55755591e-01
-2.86336988e-01 -6.02967851e-02 -1.07322586e+00 3.60100418e-01
-2.21011496e+00 -6.36202812e-01 6.42609179e-01 2.04749799e+00
7.73892403e-01 2.53116310e-01 2.49159053e-01 -1.95270732e-01
2.36201555e-01 -8.25440809e-02 -1.04359722e+00 -7.11909652e-01
3.96396428e-01 1.84898928e-01 6.42107010e-01 7.01760888e-01
-8.08177710e-01 9.46218848e-01 6.22533083e+00 1.05072975e+00
-1.00658154e+00 -1.46203607e-01 4.33909744e-01 -4.47767675e-01
1.70429707e-01 2.97760107e-02 -7.74281085e-01 1.15110703e-01
1.33776987e+00 -3.51391613e-01 1.20842648e+00 1.28387165e+00
3.65672916e-01 -5.41033506e-01 -1.35768342e+00 4.62625772e-01
-4.92615372e-01 -1.37731254e+00 -5.79711854e-01 2.95498997e-01
8.26937318e-01 8.89076442e-02 -3.28150354e-02 8.24654937e-01
6.06380343e-01 -1.24402809e+00 8.74446273e-01 5.92463255e-01
4.17817086e-01 -1.01107073e+00 4.49881494e-01 6.09147787e-01
-9.66986358e-01 -6.51490152e-01 -2.20688820e-01 -1.87994957e-01
3.32807481e-01 2.72881631e-02 -9.41209555e-01 4.45206642e-01
2.22719893e-01 1.95594907e-01 -3.27637047e-01 1.04776645e+00
-2.69169003e-01 4.30564910e-01 -5.21898329e-01 -4.05806035e-01
7.64949799e-01 -6.58069775e-02 7.22784579e-01 7.82800555e-01
2.21560746e-01 -1.58715025e-01 4.25295115e-01 9.99546528e-01
3.99340451e-01 -3.04034799e-01 -6.12154722e-01 -2.14543477e-01
2.42679611e-01 1.28097451e+00 -8.58044624e-01 -8.81028622e-02
2.06942230e-01 4.76524621e-01 3.28034997e-01 2.27519378e-01
-7.91346908e-01 -5.45580871e-02 1.59255072e-01 -2.47209802e-01
4.14441884e-01 -4.07933295e-01 -6.53720647e-02 -4.83421236e-01
-3.36706519e-01 -1.06729448e+00 8.75380933e-02 -5.88595986e-01
-7.02312112e-01 5.33692479e-01 3.66406143e-01 -9.87556875e-01
-9.45891559e-01 -6.40160978e-01 -2.88758636e-01 6.00357234e-01
-1.47743976e+00 -8.43486905e-01 -3.24117541e-02 1.82649642e-01
5.90750515e-01 -2.41063297e-01 9.07484233e-01 -4.04017866e-01
-6.33669019e-01 1.29404575e-01 2.59371459e-01 -6.34979248e-01
1.07191510e-01 -1.49976027e+00 3.43756318e-01 6.69431090e-01
-3.56298923e-01 4.56887364e-01 1.02509332e+00 -3.67992491e-01
-1.96747398e+00 -8.39726567e-01 1.07791588e-01 -2.31637642e-01
7.89520264e-01 -2.52695918e-01 -4.92065430e-01 5.07810533e-01
4.38753664e-01 -2.81089097e-01 -3.54853682e-02 -6.75251335e-02
2.42457911e-01 -1.43325999e-01 -9.67057049e-01 7.17479944e-01
6.83943212e-01 4.41901907e-02 -3.93586159e-01 1.86111912e-01
1.03983653e+00 -6.67533517e-01 -7.42423356e-01 2.36596063e-01
5.18670857e-01 -7.51618087e-01 7.08375275e-01 -5.99468350e-01
3.43572438e-01 -3.23368043e-01 -7.57509172e-02 -1.55968857e+00
-1.77516252e-01 -1.33148468e+00 -7.43333757e-01 6.43783748e-01
4.33517456e-01 -4.36903417e-01 6.71628833e-01 6.31060123e-01
-2.72957563e-01 -1.35172248e+00 -1.05430865e+00 -1.20026445e+00
2.28151903e-01 -1.73177376e-01 6.21098816e-01 4.21699554e-01
4.20443594e-01 3.21989596e-01 -4.32929307e-01 -2.92768061e-01
3.16852182e-01 1.52475879e-01 4.73997682e-01 -6.45111859e-01
-9.37533379e-01 -6.78642631e-01 2.79864609e-01 -9.56217706e-01
5.71863167e-03 -4.47157502e-01 4.26565379e-01 -1.26675165e+00
-2.08842918e-01 -5.18212140e-01 -2.12127000e-01 4.76220250e-01
4.89901990e-01 -5.11073768e-01 4.23638463e-01 2.59996671e-02
-8.41691911e-01 7.00736403e-01 1.24307811e+00 -4.11697961e-02
-8.60653877e-01 -7.54360110e-03 -3.94828647e-01 4.93376970e-01
1.09244978e+00 -2.70531535e-01 -5.67530215e-01 -1.02451079e-01
6.81080699e-01 3.88431221e-01 2.12170795e-01 -1.20654511e+00
3.47491473e-01 -4.69441861e-01 -1.30404338e-01 -2.29166657e-01
2.67359406e-01 -9.10876572e-01 5.20845614e-02 1.02896202e+00
-6.66864395e-01 1.59823328e-01 4.48030233e-01 6.54064417e-01
2.92908102e-01 -3.92840832e-01 8.38642001e-01 -4.23636585e-01
-8.75890613e-01 -6.19618082e-03 -5.09656370e-01 -2.22402483e-01
1.18892980e+00 1.71922639e-01 -2.93044180e-01 -2.44946063e-01
-7.66254783e-01 6.73554659e-01 3.44928414e-01 1.81674272e-01
5.86619414e-02 -9.13005114e-01 -2.89784908e-01 2.83557419e-02
-2.31799409e-01 5.62384818e-03 -1.42070344e-02 9.59907234e-01
-4.16627467e-01 7.09833145e-01 -4.21029627e-01 -2.48081073e-01
-5.92012882e-01 9.44776952e-01 6.37091041e-01 -8.78354430e-01
-6.45035207e-01 5.26360571e-01 -1.88130215e-01 -3.70165586e-01
2.48285368e-01 -5.62002599e-01 3.22696492e-02 -1.56209439e-01
4.40641969e-01 5.97306907e-01 -1.39259711e-01 2.22105250e-01
-1.89276546e-01 1.77887708e-01 3.05963978e-02 -5.49198210e-01
1.56598687e+00 4.69238088e-02 5.62801659e-02 3.88182253e-01
5.07262290e-01 -5.25241375e-01 -1.78227592e+00 1.26605481e-01
2.08509415e-01 1.68175250e-01 2.42004097e-01 -1.22667181e+00
-6.92502975e-01 5.97055078e-01 4.69856739e-01 5.81745625e-01
1.14849329e+00 -3.83548707e-01 2.65951633e-01 8.54257762e-01
7.99723864e-01 -1.59167373e+00 -7.06516504e-02 7.53168583e-01
1.14860964e+00 -8.14904094e-01 1.04474992e-01 2.74928659e-01
-5.61232924e-01 1.09747255e+00 6.34013951e-01 -4.45374548e-01
2.66101629e-01 4.93665099e-01 -5.00210047e-01 1.15351304e-01
-1.13684535e+00 -2.32094720e-01 -9.92255732e-02 5.51502526e-01
-1.61270246e-01 -3.97548378e-02 -2.40335181e-01 6.54455304e-01
-9.54849720e-02 1.80836141e-01 3.69137317e-01 1.25142801e+00
-6.93796277e-01 -1.18872857e+00 -1.44791260e-01 1.76329002e-01
4.83029261e-02 2.58761406e-01 -1.76615924e-01 1.14062703e+00
-1.52904376e-01 7.04639971e-01 -2.86660969e-01 -3.98424298e-01
2.35633686e-01 -1.33223608e-01 6.78914607e-01 -4.64332074e-01
-7.05064237e-01 1.33161291e-01 2.22073734e-01 -1.19482315e+00
-5.57727627e-02 -4.19274360e-01 -1.16913784e+00 -1.03238776e-01
-2.41863146e-01 2.38800094e-01 7.45953500e-01 1.04813230e+00
3.50326031e-01 7.43560016e-01 5.31919360e-01 -1.20997608e+00
-1.10363877e+00 -7.18863070e-01 -4.37114418e-01 -5.91726661e-01
4.56669182e-01 -6.09578252e-01 9.37362108e-03 -2.04096466e-01]
|
[4.336859703063965, 2.03424072265625]
|
23437eb7-9e32-479a-a2b8-88086a92bf91
|
mixed-integer-optimal-control-via
|
2305.01461
| null |
https://arxiv.org/abs/2305.01461v1
|
https://arxiv.org/pdf/2305.01461v1.pdf
|
Mixed-Integer Optimal Control via Reinforcement Learning: A Case Study on Hybrid Vehicle Energy Management
|
Many optimal control problems require the simultaneous output of continuous and discrete control variables. Such problems are usually formulated as mixed-integer optimal control (MIOC) problems, which are challenging to solve due to the complexity of the solution space. Numerical methods such as branch-and-bound are computationally expensive and unsuitable for real-time control. This paper proposes a novel continuous-discrete reinforcement learning (CDRL) algorithm, twin delayed deep deterministic actor-Q (TD3AQ), for MIOC problems. TD3AQ combines the advantages of both actor-critic and Q-learning methods, and can handle the continuous and discrete action spaces simultaneously. The proposed algorithm is evaluated on a hybrid electric vehicle (HEV) energy management problem, where real-time control of the continuous variable engine torque and discrete variable gear ratio is essential to maximize fuel economy while satisfying driving constraints. Simulation results on different drive cycles show that TD3AQ can achieve near-optimal solutions compared to dynamic programming (DP) and outperforms the state-of-the-art discrete RL algorithm Rainbow, which is adopted for MIOC by discretizing continuous actions into a finite set of discrete values.
|
['Yuan Lin', 'Jinming Xu']
|
2023-05-02
| null | null | null | null |
['q-learning', 'energy-management']
|
['methodology', 'time-series']
|
[-2.93905944e-01 1.31694302e-01 -7.06949651e-01 1.28890231e-01
-9.28550899e-01 -4.47832137e-01 3.27419400e-01 8.58079940e-02
-4.71505553e-01 1.26822197e+00 -4.45587456e-01 -5.33185482e-01
-7.48680532e-01 -7.04853117e-01 -5.38934231e-01 -1.00869632e+00
-1.11766376e-01 6.66452289e-01 -1.53811961e-01 -3.39377999e-01
2.00584486e-01 4.71364617e-01 -1.42284918e+00 -4.01829988e-01
1.13251853e+00 9.58907902e-01 4.63972420e-01 6.21047378e-01
-7.76529387e-02 7.87031293e-01 -4.13433850e-01 1.41225964e-01
4.06953603e-01 -5.39854109e-01 -3.74914467e-01 1.97482914e-01
-4.10363525e-01 -4.01458889e-01 -1.78897023e-01 8.81451666e-01
7.33192146e-01 6.24829888e-01 4.09253240e-01 -1.90968215e+00
-3.19972724e-01 1.80934593e-01 -5.51780462e-01 1.69581518e-01
-2.00585544e-01 6.51796877e-01 8.93118262e-01 -4.24654841e-01
2.40047380e-01 1.40678835e+00 1.28114969e-01 4.83998716e-01
-1.00748515e+00 -5.52675068e-01 5.49286604e-01 5.05373359e-01
-1.07727742e+00 4.81731519e-02 7.87452817e-01 -2.26237208e-01
1.24968433e+00 1.74578354e-02 1.09572482e+00 4.58303422e-01
7.82493651e-01 8.35835993e-01 1.09877729e+00 -2.86284089e-01
6.24574065e-01 -1.59833446e-01 -4.30194885e-01 7.03183949e-01
1.49630219e-01 5.42315066e-01 2.13728651e-01 -4.45656898e-03
6.73917949e-01 7.40260482e-02 3.12572837e-01 -4.11087155e-01
-9.08174694e-01 1.20121145e+00 4.41917926e-02 -2.78710753e-01
-8.57673824e-01 3.54607850e-01 5.15565932e-01 4.71531957e-01
2.49875709e-01 6.66794479e-01 -5.64086020e-01 -2.19426140e-01
-5.11455178e-01 7.79935777e-01 5.60030460e-01 8.98006439e-01
3.94012243e-01 7.12764502e-01 -5.34137130e-01 6.32477582e-01
3.54989380e-01 7.41301060e-01 3.36754590e-01 -1.50500178e+00
5.22961676e-01 2.97046840e-01 8.04531515e-01 -7.06015110e-01
-3.32374871e-01 -1.58636898e-01 -5.39992154e-01 7.99957275e-01
1.25999004e-01 -7.20071673e-01 -4.98069316e-01 1.33650923e+00
5.08569181e-01 -1.45906180e-01 2.69759357e-01 1.00110710e+00
2.68615298e-02 1.31987786e+00 6.03006594e-02 -9.58752871e-01
9.82295394e-01 -1.16026616e+00 -1.40073061e+00 -2.06402555e-01
3.69148612e-01 -3.34796369e-01 6.08090997e-01 5.71376562e-01
-1.33345473e+00 -4.92389649e-01 -1.30702329e+00 4.15218383e-01
-3.56742501e-01 9.56358090e-02 2.89942503e-01 3.08204234e-01
-7.71533370e-01 6.35221779e-01 -7.43382931e-01 5.59004545e-01
4.95163463e-02 5.53040743e-01 3.74577850e-01 9.41308439e-02
-1.49527085e+00 1.15848815e+00 5.71447790e-01 4.46219176e-01
-1.14670205e+00 -4.84091431e-01 -8.06993663e-01 -2.30179161e-01
1.24074173e+00 -2.28077739e-01 1.65670609e+00 -8.72618079e-01
-2.16522527e+00 -8.82304274e-03 2.77913719e-01 -3.06454390e-01
5.87876201e-01 -6.82964474e-02 -3.27664822e-01 1.59503922e-01
-2.52559215e-01 4.09507662e-01 9.48103368e-01 -9.15175617e-01
-9.53684986e-01 -8.75640512e-02 3.23908716e-01 4.73493874e-01
1.28570244e-01 -3.19337487e-01 2.73657199e-02 -3.49838048e-01
-7.19780087e-01 -1.01767313e+00 -6.42903209e-01 -7.32885525e-02
1.68998450e-01 -8.05766642e-01 9.12405670e-01 -6.80644929e-01
1.37889087e+00 -1.65284324e+00 5.02275169e-01 1.85048971e-02
-2.39707157e-01 5.08154690e-01 -8.90584886e-02 5.81220031e-01
9.66498330e-02 -8.49166662e-02 6.02416173e-02 5.36174923e-02
3.66794825e-01 6.66819453e-01 1.53976500e-01 4.74591076e-01
4.67303753e-01 8.25828433e-01 -1.32780731e+00 -5.94299376e-01
5.38216650e-01 5.69685325e-02 -3.88209790e-01 4.54063892e-01
-8.37088943e-01 5.66922948e-02 -9.29547369e-01 4.13637102e-01
4.96678770e-01 4.42778289e-01 3.12235773e-01 2.52392262e-01
-6.86956882e-01 -4.50855017e-01 -1.53025138e+00 1.04107678e+00
-8.08823347e-01 3.84701133e-01 6.19960248e-01 -1.42395198e+00
9.82495904e-01 3.93096328e-01 8.50563586e-01 -1.20966244e+00
2.50240624e-01 1.76876068e-01 -1.54157668e-01 -7.28768826e-01
4.98667657e-01 -2.17875496e-01 -4.73298654e-02 3.04273605e-01
-2.77179241e-01 -5.56049109e-01 4.18270856e-01 -3.20204109e-01
7.43218482e-01 1.59194842e-01 2.90589094e-01 -2.97413498e-01
7.54698992e-01 2.99204499e-01 1.03681934e+00 3.51085812e-01
-4.49830174e-01 -4.38861817e-01 9.57218826e-01 -1.24466494e-01
-1.20933747e+00 -6.01600945e-01 3.27575564e-01 8.92259181e-01
4.17479217e-01 1.60713062e-01 -3.70973915e-01 -5.10205030e-01
4.32223231e-01 1.03636038e+00 -1.77785784e-01 -3.56968343e-01
-7.87398994e-01 -4.38197792e-01 -3.28029752e-01 3.89273047e-01
4.26074773e-01 -1.00209653e+00 -8.46148849e-01 8.19246531e-01
3.05198520e-01 -8.44818950e-01 -4.81491417e-01 2.69222975e-01
-6.44191682e-01 -1.05700719e+00 -6.95834398e-01 -5.96658707e-01
3.71616751e-01 4.47166860e-02 7.39856184e-01 -2.00260118e-01
-3.33430082e-01 3.15064400e-01 -1.91994607e-01 -6.63560212e-01
-4.98850077e-01 -2.18298152e-01 1.38915405e-01 -1.17169984e-01
-1.62179604e-01 1.87519804e-01 -4.69338387e-01 4.63463247e-01
-7.27931619e-01 -9.62634236e-02 7.06242502e-01 1.22708833e+00
1.06714034e+00 7.98664510e-01 1.05552721e+00 -4.83880967e-01
9.67764676e-01 -4.24595028e-01 -1.40333366e+00 3.11544448e-01
-9.31458414e-01 2.11471945e-01 1.19385791e+00 -4.62776333e-01
-1.07846475e+00 6.26718923e-02 6.45860583e-02 -7.44587779e-01
5.82163036e-01 3.63949537e-01 -1.36103377e-01 1.12496845e-01
-1.35262832e-01 1.62823841e-01 5.62998235e-01 -1.35836769e-02
1.59156248e-01 6.49873972e-01 2.20489562e-01 -6.23231411e-01
4.85081136e-01 -3.35921586e-01 3.38195264e-01 -3.50675255e-01
-5.00961840e-01 -2.69148469e-01 -2.94004470e-01 -6.00094616e-01
8.47696006e-01 -7.23482430e-01 -1.09051740e+00 4.58884329e-01
-8.95594776e-01 -6.73176348e-01 -2.66156673e-01 4.88332987e-01
-1.12668633e+00 -7.98666701e-02 -2.46399537e-01 -1.25158978e+00
-2.59606332e-01 -1.26421094e+00 4.96055692e-01 4.23665047e-01
5.01654804e-01 -1.00169981e+00 8.75329450e-02 1.48808002e-01
2.02656090e-01 7.54691184e-01 9.26882148e-01 1.35252908e-01
-2.94394732e-01 -9.93587822e-02 2.62538433e-01 4.78275418e-01
-2.14716569e-01 1.95116192e-01 -1.48903742e-01 -8.51114571e-01
-2.04627007e-01 -6.26310289e-01 9.72813144e-02 5.39130986e-01
1.08205712e+00 -5.45316160e-01 -1.44833416e-01 2.94838175e-02
1.85015345e+00 1.03859615e+00 2.49598831e-01 6.23923540e-01
3.18687707e-01 3.95627260e-01 1.77908158e+00 1.02314723e+00
2.94056177e-01 6.76628828e-01 9.07964110e-01 -1.47904232e-01
5.83435297e-01 -9.64488909e-02 5.91081321e-01 5.39830387e-01
1.26832157e-01 -2.47730449e-01 -7.19724715e-01 5.62580764e-01
-2.13294148e+00 -8.21245432e-01 3.98444198e-02 1.99380445e+00
8.78788829e-01 3.17642987e-01 3.39857727e-01 2.09140778e-01
7.80602694e-01 3.22489776e-02 -1.18951440e+00 -1.25704443e+00
4.14711237e-01 -1.56088071e-02 9.27263319e-01 4.99259591e-01
-7.08059549e-01 4.91136074e-01 5.64700317e+00 1.19223273e+00
-9.79222775e-01 3.71489599e-02 4.69075143e-01 -2.30739251e-01
6.50093704e-02 -3.32465827e-01 -6.95015371e-01 5.81732035e-01
1.31205571e+00 -5.67929387e-01 9.68554318e-01 8.58233869e-01
9.89326954e-01 -2.03691721e-01 -8.19747508e-01 8.06044757e-01
-6.66824818e-01 -9.84562337e-01 -6.36052251e-01 -1.03841759e-01
1.01120782e+00 -3.35562110e-01 -5.28047010e-02 9.12012458e-01
4.30002958e-01 -9.64951694e-01 7.83159733e-01 5.28620243e-01
6.81277633e-01 -1.53969014e+00 8.08382809e-01 5.72836280e-01
-1.17095459e+00 -9.08192754e-01 -4.37822014e-01 2.89050415e-02
2.06937134e-01 1.46831319e-01 -3.45988601e-01 5.32489181e-01
1.93157539e-01 5.80470681e-01 2.40179002e-01 8.89432192e-01
-1.53957099e-01 2.19788477e-01 -1.13005377e-01 -7.55772293e-01
9.29512382e-01 -5.67169607e-01 4.34936523e-01 5.61811566e-01
1.02084771e-01 3.13750833e-01 7.42326856e-01 9.36661601e-01
4.63597715e-01 -8.09862986e-02 -3.68412405e-01 -4.90789145e-01
5.15601993e-01 1.16582179e+00 -4.12151277e-01 -1.87260643e-01
-1.96528688e-01 2.80512601e-01 7.04414770e-02 2.98348695e-01
-1.28361154e+00 -9.48227644e-01 7.42593944e-01 -2.55817711e-01
4.75598961e-01 -5.23466110e-01 2.05816671e-01 -2.21452251e-01
-1.20259382e-01 -8.02875221e-01 2.96542376e-01 -5.12473702e-01
-8.59405518e-01 -7.71517903e-02 2.29714707e-01 -1.39128804e+00
-5.52195907e-01 -6.17064297e-01 -5.10664344e-01 7.77061999e-01
-2.00041699e+00 -4.84517246e-01 1.14135690e-01 5.41521430e-01
1.06710386e+00 -2.10934132e-01 3.09395552e-01 2.07785472e-01
-1.04055941e+00 2.08748519e-01 1.01409829e+00 -5.66597283e-01
1.05852447e-01 -1.34912610e+00 -2.76385725e-01 2.84395874e-01
-1.09684241e+00 -5.27981639e-01 8.44618499e-01 -4.31656539e-01
-2.23472762e+00 -1.15235221e+00 2.02534005e-01 4.49045271e-01
6.78027868e-01 1.25923762e-02 -5.10794699e-01 1.39305055e-01
3.93797040e-01 1.50543615e-01 -2.50678807e-01 -7.52740204e-01
7.82679439e-01 -5.45832634e-01 -1.22014058e+00 2.22416654e-01
3.42961848e-01 2.84038503e-02 -2.86155760e-01 3.87179375e-01
9.23130333e-01 -5.19956589e-01 -1.05641341e+00 2.50521570e-01
1.58739626e-01 -2.49432907e-01 6.77981198e-01 -7.37110198e-01
2.15508685e-01 -1.64270908e-01 3.17362368e-01 -1.93097687e+00
-3.53460878e-01 -1.00104439e+00 -4.71316785e-01 1.02245665e+00
3.80278192e-02 -5.64144254e-01 3.79146039e-01 4.70941693e-01
-2.20462263e-01 -1.36525035e+00 -1.26479304e+00 -1.13815498e+00
3.26269329e-01 -7.51478449e-02 5.64068317e-01 5.60738683e-01
-5.15403599e-02 4.82927077e-02 -4.95475411e-01 1.96643367e-01
6.33740425e-01 1.13569185e-01 3.43742341e-01 -5.35682440e-01
-2.85918832e-01 -3.52308095e-01 3.82310748e-02 -6.64392412e-01
4.51039404e-01 -4.43314791e-01 3.53799909e-01 -1.70182025e+00
-4.43071038e-01 -4.41804916e-01 -4.75363284e-01 3.99571180e-01
5.19967936e-02 -5.08324623e-01 1.92403272e-01 -3.60928148e-01
-8.17209780e-01 1.29317760e+00 1.78430915e+00 -4.85000730e-01
-5.13639808e-01 9.44612399e-02 -5.84832020e-02 2.63005883e-01
8.81633222e-01 -4.68096256e-01 -7.68841028e-01 -1.64578289e-01
1.18440747e-01 1.14544511e+00 -8.77452269e-02 -6.83489561e-01
-1.22376448e-02 -1.12783933e+00 -1.68596715e-01 -8.43097746e-01
1.83868304e-01 -9.81347680e-01 9.14294347e-02 9.75159466e-01
-3.82835656e-01 5.08874178e-01 2.61023015e-01 8.24928403e-01
-1.92968488e-01 -3.42513025e-01 1.05861700e+00 -1.18068911e-01
-1.04990637e+00 4.06749547e-01 -8.38794887e-01 1.86269775e-01
1.77981687e+00 5.64522631e-02 4.93413433e-02 -2.37834394e-01
-4.84517425e-01 1.34003055e+00 -3.31514090e-01 4.80458677e-01
6.53126717e-01 -1.44131327e+00 -6.10576034e-01 -2.47627571e-01
-4.69957411e-01 -1.56503439e-01 2.51490831e-01 6.98825479e-01
-4.43807542e-01 5.87554276e-01 -4.57736731e-01 -2.86698908e-01
-9.61597800e-01 8.60511541e-01 7.00826943e-01 -5.73586881e-01
-3.63556921e-01 1.31021738e-01 -7.06173182e-01 -3.09953004e-01
2.08248094e-01 -2.69145966e-01 -2.18805358e-01 3.30449760e-01
1.38663232e-01 9.48545098e-01 -7.79795423e-02 -5.70671111e-02
2.80265557e-03 4.59019393e-01 2.63608903e-01 -8.15133676e-02
1.38700795e+00 -1.80893973e-01 2.72515625e-01 3.06358129e-01
1.11090064e+00 -8.57174516e-01 -1.73561144e+00 -8.14315956e-03
-2.42293060e-01 -4.21872586e-01 6.91454828e-01 -7.74127722e-01
-1.15136170e+00 5.92972875e-01 7.31964648e-01 3.37182134e-01
1.00227034e+00 -8.00074100e-01 9.34698641e-01 4.26716149e-01
4.42153960e-01 -2.00423813e+00 1.63213074e-01 6.22627378e-01
8.83061886e-01 -1.08583033e+00 2.98687845e-01 2.80719310e-01
-8.95698786e-01 1.25639319e+00 9.63249385e-01 -3.51205885e-01
4.50896859e-01 2.17053518e-01 -2.59404778e-01 2.70202428e-01
-1.26893425e+00 -5.92350028e-02 -1.08914390e-01 1.02311008e-01
-1.70381799e-01 1.56304508e-01 -7.88251400e-01 3.15438539e-01
4.66708660e-01 9.52383652e-02 4.11064953e-01 1.25441790e+00
-5.29480815e-01 -9.62798774e-01 -5.09292126e-01 2.81150728e-01
-2.70016730e-01 7.60907114e-01 2.41663158e-01 1.15814853e+00
1.92986056e-01 9.92956996e-01 1.60959557e-01 9.51860324e-02
5.16434193e-01 -3.01530778e-01 4.16250676e-01 -2.18778849e-01
-3.93250167e-01 4.82431166e-02 1.25249133e-01 -6.77871943e-01
-2.71852881e-01 -5.12134612e-01 -1.73078775e+00 -2.38875061e-01
-4.33026135e-01 3.93976599e-01 7.71447301e-01 9.14084613e-01
7.35156238e-02 9.31308508e-01 1.36281919e+00 -8.82857978e-01
-1.37740600e+00 -5.25050700e-01 -8.48485589e-01 -2.33966202e-01
6.25551522e-01 -1.13610029e+00 -2.89522111e-01 -6.44091189e-01]
|
[5.282005786895752, 2.2994890213012695]
|
a4c61a36-79d8-40c1-b260-a97d19026d69
|
bilinear-cnn-models-for-fine-grained-visual
| null | null |
http://openaccess.thecvf.com/content_iccv_2015/html/Lin_Bilinear_CNN_Models_ICCV_2015_paper.html
|
http://openaccess.thecvf.com/content_iccv_2015/papers/Lin_Bilinear_CNN_Models_ICCV_2015_paper.pdf
|
Bilinear CNN Models for Fine-Grained Visual Recognition
|
We propose bilinear models, a recognition architecture that consists of two feature extractors whose outputs are multiplied using outer product at each location of the image and pooled to obtain an image descriptor. This architecture can model local pairwise feature interactions in a translationally invariant manner which is particularly useful for fine-grained categorization. It also generalizes various orderless texture descriptors such as the Fisher vector, VLAD and O2P. We present experiments with bilinear models where the feature extractors are based on convolutional neural networks. The bilinear form simplifies gradient computation and allows end-to-end training of both networks using image labels only. Using networks initialized from the ImageNet dataset followed by domain specific fine-tuning we obtain 84.1% accuracy of the CUB-200-2011 dataset requiring only category labels at training time. We present experiments and visualizations that analyze the effects of fine-tuning and the choice two networks on the speed and accuracy of the models. Results show that the architecture compares favorably to the existing state of the art on a number of fine-grained datasets while being substantially simpler and easier to train. Moreover, our most accurate model is fairly efficient running at 8 frames/sec on a NVIDIA Tesla K40 GPU. The source code for the complete system will be made available at http://vis-www.cs.umass.edu/bcnn
|
['Subhransu Maji', 'Aruni RoyChowdhury', 'Tsung-Yu Lin']
|
2015-12-01
| null | null | null |
iccv-2015-12
|
['fine-grained-visual-recognition']
|
['computer-vision']
|
[-8.80729780e-02 -5.79145253e-01 -1.20950356e-01 -7.16532588e-01
-5.67401171e-01 -7.54377246e-01 8.95105124e-01 1.37463599e-01
-6.94851160e-01 3.68924797e-01 -1.80879936e-01 -2.55338877e-01
-2.74459302e-01 -6.44374430e-01 -7.66459644e-01 -7.93204665e-01
-2.46148854e-01 3.24030429e-01 2.98950672e-01 -9.26601216e-02
4.76415396e-01 8.94260824e-01 -1.66267991e+00 4.84225720e-01
1.38987929e-01 1.77252042e+00 -1.56704873e-01 8.45740139e-01
2.26758182e-01 6.44538760e-01 -2.18056783e-01 -1.36695474e-01
5.26308835e-01 3.64086568e-01 -7.90295780e-01 -3.12956899e-01
1.21008098e+00 -2.57801235e-01 -3.69484335e-01 7.52347350e-01
5.27935505e-01 2.57676661e-01 7.64787078e-01 -1.10577083e+00
-5.67934871e-01 4.45213318e-02 -3.61463487e-01 2.88468957e-01
-1.49634048e-01 -1.13418996e-02 1.03702688e+00 -1.20109308e+00
4.79358107e-01 1.30741048e+00 8.71773362e-01 2.02515155e-01
-1.39514816e+00 -6.34373426e-01 -1.00381471e-01 4.05212581e-01
-1.54903281e+00 -4.28420305e-01 4.11885202e-01 -4.94595647e-01
1.30480576e+00 3.70661139e-01 2.22912788e-01 8.21409047e-01
4.94797081e-01 2.57275999e-01 1.28536749e+00 -3.76228243e-01
3.54502276e-02 -1.62669703e-01 5.90268373e-01 1.03292930e+00
1.23487376e-01 3.14695507e-01 -5.88691592e-01 -4.77583975e-01
9.17102814e-01 -1.31874392e-02 2.69644648e-01 -7.06183434e-01
-1.31657708e+00 9.42499399e-01 8.50176752e-01 1.96096078e-01
-1.64666295e-01 4.73285198e-01 4.58410233e-01 4.35957611e-01
4.79303718e-01 3.51933926e-01 -5.15873849e-01 -1.06122652e-02
-8.21881175e-01 2.58692145e-01 8.06271553e-01 5.38826585e-01
9.17880774e-01 -1.04008988e-01 -7.94233382e-02 9.76907432e-01
1.10172227e-01 2.88467020e-01 4.77328122e-01 -1.03489065e+00
6.03420287e-02 4.51649576e-01 -7.81335384e-02 -1.15982437e+00
-6.73301816e-01 -3.91409129e-01 -8.53618681e-01 5.18961549e-01
5.82749605e-01 1.44068077e-01 -1.03140008e+00 1.51463830e+00
1.40286326e-01 5.29476106e-02 -4.36069906e-01 9.43299651e-01
8.63647759e-01 3.91816288e-01 7.28183612e-02 6.32000268e-01
1.40145099e+00 -1.14541423e+00 -7.09381979e-03 1.34870941e-02
7.24532664e-01 -9.94407594e-01 9.88281250e-01 3.41983616e-01
-7.98914373e-01 -7.93303549e-01 -1.16266048e+00 -4.60509986e-01
-7.57293344e-01 4.59265828e-01 8.40368569e-01 5.05292475e-01
-1.36417162e+00 9.54363942e-01 -1.04956138e+00 -3.12102109e-01
4.03726190e-01 7.87670314e-01 -6.84513807e-01 1.07416086e-01
-8.19023967e-01 7.45677769e-01 2.67363250e-01 1.28930241e-01
-6.48409069e-01 -6.01900518e-01 -7.54021943e-01 1.16254054e-01
-3.37658226e-01 -6.09709680e-01 1.18707514e+00 -8.89529645e-01
-1.37525618e+00 8.98116946e-01 -1.51716068e-01 -4.40509468e-01
4.06675547e-01 1.02968387e-01 -1.11577168e-01 1.25606833e-02
-2.47718040e-02 1.00416005e+00 7.95789778e-01 -6.75516069e-01
-5.76879025e-01 -2.83036828e-01 6.11515865e-02 3.50304060e-02
-2.56820142e-01 7.27551058e-02 -3.42974812e-01 -6.02275074e-01
8.25410336e-02 -1.21510744e+00 -1.76850781e-01 2.14288697e-01
-2.16704965e-01 -3.57942104e-01 1.00953865e+00 -2.22155064e-01
6.48863912e-01 -2.18317962e+00 -1.50160342e-01 4.46831554e-01
2.20902279e-01 4.94405441e-02 -4.20695752e-01 3.43418270e-01
-3.01488549e-01 8.78045559e-02 7.93202966e-02 -2.12208614e-01
2.33287558e-01 1.02503197e-02 -1.97094157e-01 7.28669226e-01
1.29079774e-01 8.63133669e-01 -3.48673135e-01 -1.49860635e-01
4.70675915e-01 7.45009243e-01 -6.18387759e-01 3.47924419e-02
2.01668635e-01 1.18427768e-01 -3.10946673e-01 4.96695876e-01
8.40024054e-01 -4.50308591e-01 -8.31623301e-02 -6.27966702e-01
-2.87707627e-01 2.08328187e-01 -1.06965804e+00 1.38389575e+00
-5.14417350e-01 8.33267450e-01 -1.25468284e-01 -9.46127892e-01
9.76567030e-01 -1.80745214e-01 2.30023474e-01 -6.27989352e-01
2.74491936e-01 2.90025949e-01 9.71810594e-02 1.23101790e-02
4.86127138e-01 1.40838653e-01 -1.24245949e-01 4.42952275e-01
2.94522732e-01 -9.46096405e-02 2.58644253e-01 -7.04205632e-02
8.83858740e-01 -2.04952527e-02 1.14791118e-01 -6.81583405e-01
4.38622624e-01 -3.10165565e-02 4.29693311e-02 9.13195431e-01
-6.51635751e-02 5.30280709e-01 2.95734882e-01 -1.09018600e+00
-1.05747378e+00 -1.01519692e+00 -5.13004243e-01 1.54158342e+00
8.40740725e-02 -4.85715955e-01 -5.59786022e-01 -3.53554964e-01
1.55575097e-01 6.53672218e-02 -8.68187308e-01 -9.25424881e-03
-4.09848750e-01 -6.57567203e-01 6.17498457e-01 6.61011457e-01
7.21429288e-01 -9.10070479e-01 -4.12149847e-01 -6.76460341e-02
2.10831061e-01 -9.41772699e-01 -3.39875877e-01 4.69279349e-01
-7.61063933e-01 -1.14777792e+00 -4.44981784e-01 -8.47401500e-01
5.81933081e-01 2.00023681e-01 9.62390721e-01 -2.17724610e-02
-5.35236239e-01 2.26852715e-01 -7.39353821e-02 -3.13026756e-02
1.92770436e-01 3.94904584e-01 1.84391916e-01 -5.05445190e-02
5.10476887e-01 -2.40022838e-01 -6.58980966e-01 5.03705382e-01
-7.92172432e-01 -5.72028048e-02 3.39268625e-01 1.13830936e+00
6.73566282e-01 -2.98648089e-01 -1.76938444e-01 -6.28205657e-01
5.32203734e-01 -1.60107151e-01 -7.78559268e-01 -3.25661376e-02
-2.53884584e-01 2.03272954e-01 7.10634053e-01 -3.93487602e-01
-6.72475278e-01 1.70011953e-01 -1.60425171e-01 -3.40463936e-01
-3.86040986e-01 2.72783637e-01 5.35022795e-01 -9.65071082e-01
8.44333589e-01 -1.59462526e-01 -1.62014797e-01 -5.33390403e-01
2.30064750e-01 6.12027228e-01 3.79906595e-01 -7.93532014e-01
4.33445513e-01 5.35954773e-01 2.09547162e-01 -9.16178107e-01
-5.87093353e-01 -5.04010737e-01 -7.97671020e-01 1.63151801e-01
7.13850021e-01 -7.70663738e-01 -1.11408508e+00 8.00554454e-01
-8.62897456e-01 -5.27594268e-01 -4.95679164e-03 5.02695620e-01
-6.03828251e-01 3.79297584e-02 -7.43537784e-01 -6.45470545e-02
-2.53815860e-01 -1.38247013e+00 1.22308218e+00 2.46823892e-01
-1.40569732e-01 -1.08958364e+00 -1.14894442e-01 6.57982752e-02
8.26983631e-01 1.10692292e-01 8.06063235e-01 -5.27740061e-01
-5.08259833e-01 -2.59871930e-01 -5.99299788e-01 3.50967556e-01
-5.89476116e-02 8.71772394e-02 -1.06862497e+00 -4.66609776e-01
-3.24617565e-01 -6.18909299e-01 1.16646385e+00 2.72182971e-01
1.53110456e+00 -1.86829165e-01 -1.89950988e-01 9.93747413e-01
1.38307667e+00 -2.38605201e-01 2.84578592e-01 6.12494886e-01
8.26690316e-01 4.35868323e-01 2.76424795e-01 3.01703542e-01
3.66287738e-01 9.23802197e-01 2.56056637e-01 -2.51525819e-01
-1.64359108e-01 1.46152198e-01 8.39937702e-02 6.20793462e-01
-2.34557390e-01 1.15635045e-01 -1.02616894e+00 2.57449538e-01
-1.75148523e+00 -7.26681590e-01 4.81676273e-02 2.09016681e+00
5.01745403e-01 -7.44049996e-02 -5.92620634e-02 -2.10608795e-01
4.62055713e-01 2.38625064e-01 -4.55496490e-01 -8.42012465e-01
-1.11841857e-01 6.23552382e-01 9.67295706e-01 7.31120527e-01
-1.51317763e+00 1.09700501e+00 6.50509691e+00 1.03371739e+00
-1.66232634e+00 5.97166978e-02 8.95251513e-01 -1.28347501e-01
2.61009097e-01 -1.89195722e-01 -8.64539146e-01 1.28331274e-01
1.07407904e+00 1.90994516e-01 4.90221322e-01 9.19139385e-01
-3.95924039e-02 1.11281537e-01 -1.07402992e+00 1.08562016e+00
-8.60225931e-02 -1.58549774e+00 -9.41524189e-03 6.76723346e-02
6.05506599e-01 7.03665018e-01 2.88076282e-01 1.43130630e-01
3.39441419e-01 -1.19242406e+00 7.49800622e-01 4.74319994e-01
9.52229142e-01 -6.60618842e-01 6.59262717e-01 -2.31761672e-02
-1.24220002e+00 1.30043887e-02 -6.19391441e-01 -2.72547722e-01
-3.95555943e-01 3.46582264e-01 -6.52911961e-01 1.24257833e-01
9.81360793e-01 5.33333957e-01 -9.32667017e-01 8.84389877e-01
4.16650921e-02 3.91322941e-01 -6.61961615e-01 -8.90346766e-02
6.47205532e-01 -3.54560912e-02 -5.06955956e-04 1.51662171e+00
2.31902122e-01 -1.54682428e-01 2.77832687e-01 3.95354837e-01
-9.31826532e-02 1.08745284e-01 -4.11593050e-01 1.11373425e-01
1.34273887e-01 1.58297718e+00 -6.88299298e-01 -3.00645560e-01
-2.27669969e-01 7.90659189e-01 7.56421745e-01 2.77961016e-01
-5.57295740e-01 -6.70604944e-01 9.24804926e-01 -1.43780351e-01
5.50835073e-01 -5.37419021e-01 -2.54271865e-01 -1.15315437e+00
-1.55429006e-01 -8.07582617e-01 2.86974907e-01 -6.58774614e-01
-1.39068925e+00 8.33524168e-01 -1.26523823e-01 -8.80972207e-01
-3.64460170e-01 -1.43944705e+00 -3.04231435e-01 8.72643709e-01
-1.29137969e+00 -1.21745932e+00 -4.93072182e-01 6.67419374e-01
8.46971571e-02 -4.94525619e-02 1.06119514e+00 2.44142354e-01
-2.54319459e-01 7.93885171e-01 3.08521837e-01 1.74786985e-01
8.89902353e-01 -1.10568225e+00 6.79593027e-01 4.69249159e-01
2.15782270e-01 7.81855583e-01 3.30316216e-01 -1.58510029e-01
-1.28690720e+00 -1.07692015e+00 8.35628688e-01 -2.81317145e-01
8.07524323e-01 -5.70099592e-01 -6.77423835e-01 6.22165084e-01
2.79094726e-02 5.92515469e-01 6.04031801e-01 1.71055093e-01
-8.68334651e-01 -2.45897368e-01 -1.10813916e+00 4.42917049e-01
8.16202164e-01 -7.82177687e-01 -1.42083585e-01 4.36268568e-01
1.66465849e-01 -7.00068951e-01 -1.04195416e+00 3.68778437e-01
9.83503819e-01 -1.05450261e+00 1.03477097e+00 -5.68297982e-01
5.96739873e-02 -2.48966411e-01 -4.88023341e-01 -1.18629205e+00
-8.27016830e-01 -1.60053149e-01 2.69848168e-01 5.21442294e-01
2.24883735e-01 -9.77038145e-01 6.32984877e-01 4.53971922e-01
1.68502390e-01 -8.40924382e-01 -9.90825176e-01 -6.98695838e-01
1.89275131e-01 -3.71744841e-01 4.21287924e-01 8.15397620e-01
-3.37015629e-01 1.37299329e-01 -1.15477443e-01 3.03216521e-02
6.53072774e-01 2.69750476e-01 6.48247302e-01 -1.19248676e+00
-1.23354428e-01 -6.85499132e-01 -8.04570377e-01 -8.61023068e-01
2.48349831e-01 -8.99145544e-01 -2.44325429e-01 -9.58370090e-01
9.20827836e-02 -7.45094001e-01 -5.20067334e-01 9.03775871e-01
3.80122423e-01 1.05212843e+00 2.21346602e-01 2.56837636e-01
-4.69016761e-01 1.58782333e-01 8.76631796e-01 -1.28296286e-01
1.97530195e-01 -2.53265381e-01 -5.48489451e-01 8.27941835e-01
9.77311075e-01 -2.90466309e-01 9.57748219e-02 -4.86171365e-01
-1.24015503e-01 -4.80720520e-01 6.89582407e-01 -1.06286180e+00
3.29136550e-01 1.02824174e-01 7.75787830e-01 -3.08515787e-01
5.57125390e-01 -6.52213454e-01 7.59788752e-02 3.94129634e-01
-3.61605495e-01 2.82220364e-01 4.96413022e-01 1.96033381e-02
-2.90983677e-01 -6.90865517e-02 1.00048470e+00 7.58000016e-02
-9.71980035e-01 5.17875493e-01 -2.36606792e-01 -3.12684953e-01
8.34436834e-01 -8.29936936e-02 -6.40709519e-01 -1.74715862e-01
-6.37169361e-01 -2.25116298e-01 6.41910851e-01 5.06323934e-01
2.03190640e-01 -1.49340463e+00 -4.92487848e-01 5.75509906e-01
8.93511772e-02 -5.22028446e-01 2.54559964e-01 6.73685431e-01
-8.97408128e-01 8.40919435e-01 -6.07942462e-01 -8.24378371e-01
-1.31280577e+00 2.12647900e-01 5.29405534e-01 -2.28247136e-01
-1.30392268e-01 9.03459489e-01 3.57664704e-01 -6.70159101e-01
1.14561461e-01 -4.31765109e-01 3.27507108e-02 -1.17197409e-01
5.73887348e-01 3.80701423e-01 4.71225888e-01 -9.27107215e-01
-6.28697574e-01 8.67013395e-01 -3.21956903e-01 3.95074748e-02
1.31608355e+00 1.74553722e-01 -2.88690478e-01 2.04862684e-01
1.73493874e+00 -2.67031103e-01 -1.21817613e+00 -1.25373229e-01
-2.75633544e-01 -4.32488561e-01 3.23412001e-01 -7.43249834e-01
-8.56607020e-01 1.02773285e+00 1.07189703e+00 1.09488271e-01
9.00124967e-01 -1.07871488e-01 3.56025159e-01 7.91044414e-01
3.15800577e-01 -6.90769732e-01 -1.36374325e-01 9.63033438e-01
8.74106288e-01 -1.15570462e+00 -2.67469566e-02 -2.05858231e-01
-2.90084511e-01 1.36221552e+00 5.19899964e-01 -6.33383930e-01
7.69703627e-01 3.63006294e-01 2.81086057e-01 -1.27526104e-01
-6.83694005e-01 3.31941284e-02 6.51854873e-01 3.60606730e-01
6.71784878e-01 2.34505773e-01 -2.87756585e-02 6.67732209e-02
-3.80191982e-01 -1.07671201e-01 4.62377146e-02 7.80617237e-01
-2.45513201e-01 -1.23002326e+00 -1.90392122e-01 5.82480431e-01
-5.72858214e-01 -2.49570504e-01 -3.18761438e-01 7.02858329e-01
1.55679271e-01 5.25287211e-01 3.54867578e-01 -4.51898038e-01
1.49446175e-01 -3.69663388e-02 7.12770104e-01 -3.42465669e-01
-7.88277268e-01 -2.11057097e-01 3.72305773e-02 -1.00759864e+00
-2.95936853e-01 -4.23053056e-01 -9.06964958e-01 -7.29121745e-01
-1.14310600e-01 -4.61691543e-02 9.02076781e-01 7.42892146e-01
6.96029246e-01 1.24935605e-01 4.30553585e-01 -1.55429447e+00
-5.99707067e-01 -9.45425570e-01 -4.57999736e-01 3.03720117e-01
4.19028342e-01 -9.33989704e-01 -3.01063746e-01 -1.18612207e-01]
|
[9.406373023986816, 1.757855772972107]
|
57f5e2ff-9333-4ed6-a8d4-0b279ef58dc5
|
client-driven-lightweight-method-to-generate
| null | null |
https://www.scitepress.org/Link.aspx?doi=10.5220/0011335300003289
|
https://www.scitepress.org/Link.aspx?doi=10.5220/0011335300003289
|
Client-driven Lightweight Method to Generate Artistic Media for Feature-length Sports Videos
|
This paper proposes a lightweight methodology to attract users and increase views of videos through personalized artistic media i.e., static thumbnails and animated Graphics Interchange Format (GIF) images. The proposed method analyzes lightweight thumbnail containers (LTC) using computational resources of the client device to recognize personalized events from feature-length sports videos. In addition, instead of processing the entire video, small video segments are used in order to generate artistic media. This makes our approach more computationally efficient compared to existing methods that use the entire video data. Further, the proposed method retrieves and uses thumbnail containers and video segments, which reduces the required transmission bandwidth as well as the amount of locally stored data that are used during artistic media generation. After conducting experiments on the NVIDIA Jetson TX2, the computational complexity of our method was 3.78 times lower than that of the state-of-the-art method. To the best of our knowledge, this is the first technique that uses LTC to generate artistic media while providing lightweight and high-performance services on resource-constrained devices.
|
['Eun-Seok Ryu', 'Jaehyuk Choi', 'Ghulam Mujtaba']
|
2022-07-01
| null | null | null |
conference-2022-7
|
['animated-gif-generation', 'sports-analytics', 'video-generation']
|
['computer-vision', 'computer-vision', 'computer-vision']
|
[ 2.39207223e-01 -4.88643557e-01 -3.08334976e-02 -3.88986394e-02
-3.23035538e-01 -6.65695071e-01 1.75077870e-01 5.36094084e-02
-4.93537545e-01 4.38538373e-01 -4.28169258e-02 -4.13008362e-01
2.48987958e-01 -1.04198313e+00 -7.16277480e-01 -2.38648996e-01
-2.00054068e-02 1.69739619e-01 9.06755626e-01 1.55803382e-01
6.27635539e-01 5.30661583e-01 -2.41080189e+00 7.02518940e-01
6.79067612e-01 1.38686836e+00 4.62095201e-01 7.36763358e-01
-5.49348772e-01 1.00971591e+00 -8.28583777e-01 -3.54253680e-01
2.49149457e-01 -3.31998676e-01 -2.90032595e-01 1.43550396e-01
5.75708330e-01 -8.76239479e-01 -2.00804830e-01 6.92763209e-01
5.13714314e-01 1.51743993e-01 6.25886247e-02 -1.23713982e+00
-3.76443774e-01 4.57394987e-01 -6.92072630e-01 4.05408323e-01
7.11114526e-01 -6.94900677e-02 4.01612878e-01 -7.44618416e-01
8.13952923e-01 9.29517567e-01 2.96717823e-01 3.40135366e-01
-5.15492201e-01 -8.03845942e-01 -8.80645066e-02 3.19021583e-01
-1.14854705e+00 -3.62211972e-01 6.53079271e-01 -3.71007115e-01
7.33424187e-01 7.07707286e-01 9.79650974e-01 6.96926415e-01
5.17571568e-02 8.34230125e-01 8.27423751e-01 -5.40814579e-01
4.00114268e-01 1.23789050e-01 -1.11434869e-01 5.15290201e-01
2.76498407e-01 -5.29275596e-01 -7.01308548e-01 -3.14811349e-01
1.12001777e+00 1.56070977e-01 -1.75162047e-01 -5.57570085e-02
-1.16351044e+00 5.61431408e-01 -2.48724476e-01 8.70006010e-02
-5.33914506e-01 4.48899455e-02 6.14632249e-01 -7.22511560e-02
3.09598237e-01 8.05477574e-02 2.18824238e-01 -8.25968623e-01
-1.16762483e+00 1.35147959e-01 6.57618582e-01 1.12984169e+00
2.61656106e-01 3.66760582e-01 -5.90797327e-02 5.80142021e-01
7.88196400e-02 5.56016743e-01 5.71151733e-01 -1.08568752e+00
6.49020851e-01 6.20594800e-01 2.13362232e-01 -1.17138052e+00
4.16623130e-02 4.04029079e-02 -2.78066814e-01 2.43405804e-01
3.53649795e-01 -1.40388817e-01 -2.87946343e-01 8.80478442e-01
2.25099474e-01 4.30819809e-01 -2.92918354e-01 1.08791947e+00
8.45388353e-01 9.55857038e-01 -1.48823157e-01 -2.19429046e-01
1.41599000e+00 -7.78289914e-01 -5.35522521e-01 2.91510314e-01
1.13386713e-01 -1.38491964e+00 1.33868301e+00 6.67308033e-01
-1.37477183e+00 -6.83433592e-01 -1.12099040e+00 3.35153863e-02
1.35977520e-02 1.78824767e-01 7.66927779e-01 9.79628563e-01
-9.26308811e-01 3.64462584e-01 -6.96213305e-01 -2.73312956e-01
2.64093429e-02 1.10405043e-01 2.57426560e-01 2.94894725e-01
-7.13725686e-01 5.83202727e-02 1.79886103e-01 -4.38599110e-01
-3.00669849e-01 -7.09219456e-01 -2.27704510e-01 2.31035158e-01
3.99250269e-01 -2.44718000e-01 1.06375897e+00 -1.14757919e+00
-1.72076452e+00 5.05097330e-01 -1.55486763e-02 -4.07688856e-01
5.31118691e-01 -3.78038764e-01 -3.78636986e-01 7.48301983e-01
-8.93007964e-02 4.52557772e-01 9.02892232e-01 -8.28380883e-01
-1.04433620e+00 -5.67950122e-02 3.79900157e-01 2.41044790e-01
-7.90705562e-01 2.64497101e-01 -8.96161914e-01 -7.47070253e-01
-7.34938905e-02 -9.32446361e-01 2.33066127e-01 -1.13443360e-02
-1.38808623e-01 1.91671904e-02 1.07890105e+00 -4.71999198e-01
1.60109615e+00 -2.09878778e+00 -2.96971679e-01 2.08560035e-01
1.09597571e-01 5.28160632e-01 3.09208453e-01 3.25486183e-01
3.91348958e-01 5.72706610e-02 7.35092461e-01 4.56313081e-02
-2.09276602e-01 -1.11785054e-01 -4.49124634e-01 -4.12164330e-02
-4.85271364e-01 1.71722636e-01 -6.41617239e-01 -6.62106633e-01
4.52717185e-01 6.72639132e-01 -7.41417408e-01 -1.50737502e-02
1.23595260e-01 3.04263085e-02 -5.35220504e-01 7.07872510e-01
8.43515575e-01 -5.66338561e-02 1.51546195e-01 -1.70637414e-01
-7.14734614e-01 6.71747923e-02 -1.34819579e+00 1.35606611e+00
-4.87794489e-01 7.20063567e-01 -2.83291787e-01 -3.69071692e-01
8.91346455e-01 3.17410499e-01 7.54882336e-01 -8.74277711e-01
1.20839633e-01 4.09222007e-01 -6.97940528e-01 -6.69579089e-01
1.16676974e+00 4.87036109e-01 1.41888574e-01 5.86264968e-01
-6.00081801e-01 3.38765055e-01 5.06090879e-01 3.15857619e-01
9.07162964e-01 3.26073915e-01 -1.09990753e-01 -7.23948553e-02
4.31634277e-01 -2.26994641e-02 3.66772950e-01 6.19741797e-01
1.17083296e-01 3.13428283e-01 1.72687098e-01 -7.23662138e-01
-1.27231741e+00 -9.80536520e-01 1.76487893e-01 1.12101305e+00
3.85359645e-01 -7.92483926e-01 -9.66889262e-01 -6.12101443e-02
-2.34701023e-01 6.75689816e-01 2.53817469e-01 4.96243745e-01
-7.54090190e-01 -3.41016173e-01 2.67713577e-01 4.49906766e-01
5.76812088e-01 -1.01200974e+00 -1.43825316e+00 3.65540504e-01
-1.68770373e-01 -1.22987437e+00 -6.00961089e-01 -9.81748343e-01
-1.11865020e+00 -9.81218934e-01 -7.38476872e-01 -6.89141929e-01
4.84459668e-01 7.58862674e-01 9.01778877e-01 7.44405910e-02
-4.64148223e-01 3.35237950e-01 -7.43506491e-01 -2.60078281e-01
2.79836934e-02 -2.21943438e-01 6.71179965e-02 1.78184271e-01
3.61886531e-01 -5.22222817e-01 -8.29352915e-01 3.20724338e-01
-8.98317039e-01 6.43542171e-01 1.98602676e-01 1.53320774e-01
8.69841993e-01 2.43620247e-01 5.93374148e-02 -8.02061915e-01
4.19744104e-01 -3.09241712e-01 -8.30699801e-01 2.75165588e-02
-1.74291328e-01 -2.21229762e-01 9.90573227e-01 -5.63717902e-01
-1.00939214e+00 -4.84883301e-02 2.25153834e-01 -4.62162316e-01
1.66154876e-01 -6.11245222e-02 3.65183860e-01 -1.09793067e-01
2.62503505e-01 2.95257777e-01 -1.56869128e-01 -5.75865090e-01
-2.63153040e-03 1.00294363e+00 5.57273030e-01 -4.79835361e-01
3.12110692e-01 6.11651123e-01 -3.14527154e-02 -1.00820279e+00
-1.55429840e-01 -4.06438291e-01 -1.40478775e-01 -7.45009005e-01
7.47571111e-01 -7.76411951e-01 -1.27564490e+00 5.24003744e-01
-1.04242158e+00 9.55485553e-02 -2.09969468e-02 6.48870707e-01
-4.96105909e-01 4.74913061e-01 -7.21207500e-01 -9.14207280e-01
-5.44418693e-01 -1.29038155e+00 1.11837637e+00 4.71024424e-01
-5.24033755e-02 -3.26544434e-01 -4.19675350e-01 4.53433156e-01
6.58190846e-01 4.45020020e-01 3.32756639e-01 1.76546752e-01
-1.00664139e+00 -3.23077679e-01 -4.04830664e-01 -1.79472834e-01
-1.23608802e-02 4.16374505e-01 -5.24953127e-01 -1.81835651e-01
-1.48876354e-01 6.60992116e-02 2.53966659e-01 2.24203154e-01
1.55487490e+00 -3.60917360e-01 -1.35666415e-01 5.85185409e-01
1.50989187e+00 8.06134403e-01 9.38474000e-01 6.32411599e-01
6.41470432e-01 2.06427515e-01 8.11301649e-01 1.07057333e+00
3.44631106e-01 9.23378527e-01 2.15580881e-01 1.83600783e-02
-1.33538291e-01 -1.22438058e-01 4.16696936e-01 9.16833282e-01
-7.75997221e-01 -5.14059246e-01 -4.93954360e-01 8.45157057e-02
-1.53177428e+00 -1.37800682e+00 -4.33544666e-01 2.55808759e+00
3.87956679e-01 1.27330735e-01 1.71975985e-01 2.19084442e-01
8.92477870e-01 5.91145605e-02 -3.69301587e-01 -6.82475150e-01
8.16708282e-02 3.61476809e-01 6.34005904e-01 1.59672946e-01
-8.60677183e-01 5.16763330e-01 5.58246565e+00 1.02535796e+00
-1.26427758e+00 7.57246464e-02 4.23989624e-01 -8.02340567e-01
-1.84589386e-01 -1.91857234e-01 -7.71936655e-01 1.04850328e+00
1.01506364e+00 -1.18663296e-01 5.77770412e-01 1.10751617e+00
4.66788918e-01 -4.25723195e-01 -4.57037330e-01 1.46059954e+00
3.37330312e-01 -1.52413511e+00 1.58582091e-01 1.07796393e-01
5.16774297e-01 -4.12006915e-01 3.63422781e-02 -9.36475396e-02
-3.51040184e-01 -1.71224415e-01 1.07481468e+00 5.63285708e-01
9.21830297e-01 -8.98933589e-01 4.86292601e-01 -1.73340276e-01
-1.41893744e+00 1.23819724e-01 -4.05246884e-01 -1.19204938e-01
4.08936411e-01 4.91279185e-01 -4.91275609e-01 8.53725374e-02
9.14809763e-01 1.28187776e-01 -5.15840530e-01 1.09992278e+00
3.92084777e-01 6.29698217e-01 -4.12738055e-01 -3.39155763e-01
1.00990288e-01 -1.57240510e-01 5.11976540e-01 1.07526577e+00
9.98350322e-01 1.53099984e-01 2.28599951e-01 2.24418849e-01
5.48911765e-02 4.50371951e-01 -2.22218439e-01 -9.76030454e-02
5.79884291e-01 1.08067095e+00 -1.12996340e+00 -8.28644991e-01
-5.01380205e-01 1.14110112e+00 -3.07725698e-01 -2.89183799e-02
-1.19352901e+00 -6.98529661e-01 1.59548968e-01 7.31729984e-01
2.29459807e-01 -3.81863683e-01 4.57102526e-03 -1.14939857e+00
2.84797132e-01 -6.50228739e-01 7.48862401e-02 -9.04805958e-01
-7.02379167e-01 9.53492701e-01 -1.85310945e-01 -1.77206552e+00
-3.52719605e-01 -2.98168033e-01 -3.88243675e-01 4.71116841e-01
-9.43754673e-01 -7.18001127e-01 -7.77317822e-01 6.17145360e-01
8.67940962e-01 -2.25152642e-01 5.28959036e-01 6.61242366e-01
-2.03862548e-01 4.62819725e-01 3.78388315e-01 -4.34814692e-01
6.47512674e-01 -7.36865222e-01 3.38766932e-01 8.26428950e-01
-1.41311288e-01 4.84497368e-01 5.33258379e-01 -8.26961398e-01
-1.77805483e+00 -7.11197734e-01 6.50054276e-01 1.73119843e-01
4.39005226e-01 -2.96839565e-01 -4.35784250e-01 8.12528059e-02
2.09938824e-01 -3.32708418e-01 9.63452578e-01 -4.48989570e-01
-1.55614065e-02 -2.49491975e-01 -9.45426226e-01 7.44049907e-01
9.63811755e-01 -1.51155561e-01 1.82242602e-01 2.68083811e-01
4.01333004e-01 -6.03710771e-01 -7.58256495e-01 -1.44418225e-01
1.07336462e+00 -1.35626137e+00 9.34848368e-01 9.77231339e-02
3.38798612e-01 -4.14856285e-01 -2.10849330e-01 -4.92017418e-01
4.81091365e-02 -7.75437415e-01 -1.17403656e-01 1.18485546e+00
8.21909215e-03 -2.55020618e-01 8.24878037e-01 8.01990211e-01
-1.97420403e-01 -5.21692038e-01 -5.54763079e-01 -6.27870858e-01
-1.13326490e+00 -5.04732609e-01 7.72751689e-01 5.76024592e-01
1.85342990e-02 -2.66383559e-01 -6.25271022e-01 -1.49016351e-01
7.22570181e-01 3.69622350e-01 8.74347270e-01 -1.01925790e+00
-4.89147902e-01 -6.60921484e-02 -6.42013967e-01 -7.94536352e-01
-5.78126729e-01 -4.19185281e-01 -5.56139886e-01 -1.19886863e+00
1.16011715e-02 -5.60034215e-01 1.65573791e-01 9.87110510e-02
1.46198153e-01 7.89244592e-01 7.96973109e-01 5.28514802e-01
-7.58816183e-01 -8.15114379e-02 1.04570758e+00 2.99278527e-01
-5.21027982e-01 4.83573824e-02 -3.10003042e-01 7.75201797e-01
8.59389126e-01 -1.31264880e-01 -6.23737574e-01 -5.94568312e-01
2.08187371e-01 1.89108953e-01 2.10784808e-01 -1.29417336e+00
2.18348444e-01 -1.64236322e-01 3.05632591e-01 -8.18146229e-01
4.64944363e-01 -7.02530026e-01 6.61634624e-01 3.92925233e-01
5.66780530e-02 3.76300752e-01 2.06188217e-01 1.90389544e-01
-3.06136757e-02 -3.16844642e-01 3.78528506e-01 -7.67225921e-02
-8.82351637e-01 -1.60027090e-02 -8.13623428e-01 -3.93918425e-01
1.27422929e+00 -6.58955991e-01 -4.85062093e-01 -3.26621324e-01
-3.58273268e-01 -3.39586198e-01 7.90143847e-01 4.51593459e-01
7.76077747e-01 -1.51860595e+00 -2.10122645e-01 2.16705158e-01
-4.08976257e-01 -5.77756405e-01 7.30352759e-01 3.90179336e-01
-1.61883843e+00 4.64684337e-01 -6.91375315e-01 -4.70618606e-01
-1.73271978e+00 6.10589087e-01 -5.88731527e-01 1.21340103e-01
-7.44496822e-01 4.62679535e-01 -1.51313365e-01 6.86816037e-01
1.30720854e-01 -1.71182483e-01 -2.65647173e-01 1.89378276e-01
1.07905960e+00 9.32559669e-01 -2.36337595e-02 -6.01661742e-01
-2.90318042e-01 7.43790269e-01 5.35191894e-02 -2.08211347e-01
9.79799986e-01 -2.51235157e-01 6.63763210e-02 1.92472368e-01
9.70888376e-01 4.84895021e-01 -1.10666287e+00 2.91239202e-01
-4.79875207e-01 -1.05554748e+00 3.76208909e-02 -1.20050468e-01
-1.32134545e+00 6.89434350e-01 7.84008801e-01 3.01036745e-01
1.40817237e+00 -5.58275223e-01 1.42036021e+00 -2.12298736e-01
8.71076107e-01 -1.33955872e+00 3.13761719e-02 -1.11967675e-01
5.33240378e-01 -6.27326787e-01 2.80292351e-02 -6.68023884e-01
-7.30428219e-01 1.44075918e+00 6.33441508e-01 -2.11780518e-01
2.55054533e-01 4.59834129e-01 -1.91771060e-01 1.12236977e-01
-4.24643189e-01 1.71516195e-01 1.09425420e-03 3.30928028e-01
5.37094772e-01 2.63779908e-01 -7.66640067e-01 3.12111348e-01
-3.92713457e-01 3.81468445e-01 9.36764359e-01 1.13383150e+00
-5.50894320e-01 -1.11847651e+00 -7.15622604e-01 5.14816344e-01
-7.66621828e-01 -8.19740742e-02 1.03628650e-01 4.77112055e-01
1.11009717e-01 9.88489747e-01 5.30675292e-01 -4.70656931e-01
2.12046281e-01 -3.04768682e-01 3.12319726e-01 -1.59217864e-01
-6.34625971e-01 3.82627666e-01 9.54754371e-03 -8.41548026e-01
-5.20029187e-01 -5.27807236e-01 -1.16671073e+00 -7.95805693e-01
1.37404799e-01 1.47390351e-01 9.83435571e-01 1.31435767e-01
8.14837277e-01 4.85985607e-01 5.66419780e-01 -1.19488549e+00
2.33770549e-01 -3.91036659e-01 -3.86310667e-01 4.43411469e-01
-3.99981350e-01 -5.98083377e-01 7.28977993e-02 3.53053957e-01]
|
[10.595227241516113, -1.0071988105773926]
|
890da00f-80b7-46ca-af7a-f0ff95b64a33
|
multimodal-data-integration-for-oncology-in
|
2303.06471
| null |
https://arxiv.org/abs/2303.06471v1
|
https://arxiv.org/pdf/2303.06471v1.pdf
|
Multimodal Data Integration for Oncology in the Era of Deep Neural Networks: A Review
|
Cancer has relational information residing at varying scales, modalities, and resolutions of the acquired data, such as radiology, pathology, genomics, proteomics, and clinical records. Integrating diverse data types can improve the accuracy and reliability of cancer diagnosis and treatment. There can be disease-related information that is too subtle for humans or existing technological tools to discern visually. Traditional methods typically focus on partial or unimodal information about biological systems at individual scales and fail to encapsulate the complete spectrum of the heterogeneous nature of data. Deep neural networks have facilitated the development of sophisticated multimodal data fusion approaches that can extract and integrate relevant information from multiple sources. Recent deep learning frameworks such as Graph Neural Networks (GNNs) and Transformers have shown remarkable success in multimodal learning. This review article provides an in-depth analysis of the state-of-the-art in GNNs and Transformers for multimodal data fusion in oncology settings, highlighting notable research studies and their findings. We also discuss the foundations of multimodal learning, inherent challenges, and opportunities for integrative learning in oncology. By examining the current state and potential future developments of multimodal data integration in oncology, we aim to demonstrate the promising role that multimodal neural networks can play in cancer prevention, early detection, and treatment through informed oncology practices in personalized settings.
|
['Ghulam Rasool', 'Paul Stewart', 'Ravi P. Ramachandran', 'Aakash Tripathi', 'Asim Waqas']
|
2023-03-11
| null | null | null | null |
['data-integration']
|
['knowledge-base']
|
[ 3.66241515e-01 -4.66754660e-02 -6.11235261e-01 -1.86887588e-02
-9.29584086e-01 -1.95595786e-01 2.01722220e-01 9.98799264e-01
-1.60754487e-01 6.82068944e-01 4.44834888e-01 -5.25667608e-01
-6.34905457e-01 -7.16401339e-01 -2.13509753e-01 -1.00869846e+00
-8.14770833e-02 4.08489496e-01 -4.56813246e-01 -4.89398807e-01
-2.94959456e-01 8.33643198e-01 -8.66184652e-01 7.26689041e-01
7.39685774e-01 1.14440727e+00 2.64332853e-02 8.23594749e-01
-4.41252232e-01 9.14803147e-01 -4.70881313e-01 2.70017609e-02
-3.89783978e-01 -3.84524494e-01 -6.31279826e-01 -2.87524372e-01
3.22806478e-01 -2.23462246e-02 -7.90935457e-01 8.34500313e-01
7.79976785e-01 -5.16579807e-01 4.08957601e-01 -1.16378522e+00
-5.86486280e-01 4.22291428e-01 -4.26320016e-01 2.34069377e-01
1.59895614e-01 2.44308725e-01 7.03133821e-01 -6.62888944e-01
6.62310719e-01 8.87219191e-01 9.78761554e-01 6.06682479e-01
-1.16942012e+00 -2.71882623e-01 -1.11889876e-01 -4.60976847e-02
-9.53019261e-01 -2.02756301e-01 5.59385061e-01 -3.32894087e-01
8.89527023e-01 4.61169153e-01 9.91592765e-01 1.11045253e+00
1.04006696e+00 7.21261621e-01 8.01304817e-01 -3.05684716e-01
-1.43862173e-01 -1.61335677e-01 2.38109052e-01 9.43363428e-01
3.74426603e-01 1.29771512e-02 -7.52937734e-01 -2.21644208e-01
5.59865475e-01 7.53202438e-01 -3.25151771e-01 -1.71842501e-01
-1.39273190e+00 6.73640311e-01 8.26765060e-01 5.56328595e-01
-2.38810822e-01 2.10710287e-01 5.47613680e-01 1.00850970e-01
1.52012005e-01 3.34665030e-01 -2.68100232e-01 5.14613569e-01
-6.98655427e-01 -3.33588898e-01 5.69323003e-01 5.49018800e-01
4.59526122e-01 -1.21198758e-01 -1.36095524e-01 6.29629791e-01
2.89370418e-01 2.27383137e-01 6.54542923e-01 -6.66678607e-01
-2.08621547e-02 9.57731009e-01 -5.08875310e-01 -8.74460757e-01
-1.26748037e+00 -4.97868925e-01 -1.27006805e+00 2.04628319e-01
3.66560698e-01 -1.25673786e-01 -9.91409719e-01 1.27091074e+00
3.23439837e-01 -1.01292163e-01 2.49503955e-01 6.02153599e-01
1.81656170e+00 -2.26353317e-01 3.68083984e-01 -1.71822961e-02
1.64879751e+00 -6.45426512e-01 -9.41365778e-01 -1.05241574e-01
1.05794549e+00 -5.80848753e-01 2.53063470e-01 4.58084233e-02
-8.98805320e-01 -1.40410468e-01 -9.38752353e-01 -3.52990985e-01
-8.03482890e-01 1.14488006e-01 8.44750345e-01 5.22961318e-01
-1.13880205e+00 4.66630876e-01 -1.05724597e+00 -6.47873282e-01
7.63922453e-01 6.65872216e-01 -7.67522216e-01 -2.79229462e-01
-1.01891851e+00 1.05357540e+00 3.36984962e-01 9.32439566e-02
-8.57680559e-01 -1.44506431e+00 -8.25332165e-01 -2.29316410e-02
1.98729336e-01 -1.24218917e+00 9.71419930e-01 -6.07621133e-01
-9.25848544e-01 6.73983932e-01 -1.24128526e-02 -1.64696440e-01
2.05103502e-01 4.85019386e-01 -3.11038584e-01 4.30398911e-01
-3.37803215e-01 7.39453435e-01 -6.78409487e-02 -9.70507860e-01
-6.59067094e-01 -7.26581633e-01 -3.03393334e-01 3.29983026e-01
-4.24947888e-01 -2.60667264e-01 -2.24933363e-02 -1.81197464e-01
1.61329702e-01 -4.77318704e-01 -4.64067340e-01 4.34404999e-01
-2.52662063e-01 6.38147369e-02 9.40791786e-01 -6.07749462e-01
8.79053533e-01 -1.84795153e+00 2.59441137e-01 8.47451538e-02
1.08459711e+00 1.29195720e-01 -1.10090025e-01 6.21089101e-01
-1.89601183e-01 2.75327176e-01 9.66179147e-02 -3.38162512e-01
-3.79068822e-01 3.05349588e-01 4.43155468e-01 5.32679379e-01
1.12918720e-01 1.58080184e+00 -9.92530525e-01 -6.27240479e-01
4.31156665e-01 8.04724276e-01 1.76458016e-01 -1.63620427e-01
-1.03529971e-02 4.62257564e-01 -2.08717510e-01 1.27379751e+00
2.77107060e-01 -6.89327121e-01 5.39682269e-01 -6.36759996e-01
2.65486658e-01 -1.09587409e-01 -2.98260540e-01 1.70284855e+00
-2.32459996e-02 7.84850955e-01 4.26480025e-01 -1.10444868e+00
5.83525717e-01 5.38424253e-01 1.23789430e+00 -7.12090552e-01
3.49148422e-01 2.88482010e-02 1.99118033e-01 -4.78598654e-01
7.67835081e-02 -4.20078814e-01 2.05001384e-01 7.73398355e-02
2.03996316e-01 -5.37075587e-02 -8.65623653e-02 1.81037381e-01
1.44536996e+00 -6.18438005e-01 5.06105542e-01 1.47670805e-01
2.74584442e-01 2.36046001e-01 2.25040868e-01 6.56046629e-01
-4.54839230e-01 4.44727004e-01 7.63775468e-01 -4.48854595e-01
-6.79974794e-01 -8.07557344e-01 -2.12833226e-01 8.68968606e-01
-1.01480082e-01 -2.86898226e-01 -1.50375217e-01 -7.00804472e-01
2.06212118e-01 -1.76229075e-01 -1.02555478e+00 -2.38775536e-01
-1.51154920e-01 -1.34865975e+00 8.72436106e-01 6.88947976e-01
-1.57288164e-01 -7.10404277e-01 -2.59125412e-01 1.59575880e-01
1.18134152e-02 -1.13565230e+00 1.16866842e-01 5.25808692e-01
-1.05979311e+00 -1.53526735e+00 -5.51195741e-01 -7.45733082e-01
7.65562654e-01 4.41361457e-01 9.75275457e-01 2.29469240e-01
-7.66583085e-01 5.87157428e-01 -4.30193692e-02 -7.12006569e-01
-6.29232228e-01 1.17872916e-02 -2.96561569e-01 -4.32925671e-01
2.04737738e-01 -1.14143476e-01 -6.35009527e-01 3.91923171e-03
-1.14197350e+00 8.62053707e-02 9.06767964e-01 1.30025613e+00
8.17650735e-01 -3.00716199e-02 5.10421455e-01 -7.26449847e-01
8.07976425e-01 -6.29915476e-01 3.76060940e-02 6.71537042e-01
-3.50663275e-01 -3.35258663e-01 3.83887619e-01 -2.25495860e-01
-6.02408946e-01 -4.74436618e-02 1.39769828e-02 -2.35120192e-01
-2.15693891e-01 1.17486966e+00 1.18830681e-01 -7.95536697e-01
7.59983003e-01 -2.50826985e-01 7.07797408e-01 -2.24044155e-02
1.23984247e-01 3.39978427e-01 4.32040840e-01 -2.48350829e-01
1.35623693e-01 7.02814996e-01 7.71832883e-01 -8.28553081e-01
-6.48332298e-01 -3.76791030e-01 -5.86984694e-01 -3.07241470e-01
8.11584711e-01 -6.27390325e-01 -9.22829211e-01 4.02030110e-01
-7.11308718e-01 -1.92118511e-01 -3.46107513e-01 4.53543127e-01
-9.36903805e-02 2.89884359e-01 -7.65045881e-01 -2.77635217e-01
-6.92805946e-01 -1.32988679e+00 1.08744180e+00 4.32587594e-01
-1.93566784e-01 -1.64798677e+00 1.59084290e-01 4.17593092e-01
6.08356118e-01 8.36137593e-01 1.22809315e+00 -7.55074382e-01
-4.16656226e-01 -6.30070508e-01 -3.85727584e-01 -2.23855510e-01
6.31745875e-01 1.85807809e-01 -8.68136644e-01 -1.21494502e-01
-4.35474962e-01 -4.31138337e-01 8.51492286e-01 6.04478180e-01
9.78231966e-01 2.24124834e-01 -1.05763233e+00 6.33072257e-01
1.31620681e+00 1.65196389e-01 3.02656382e-01 -9.78704989e-02
9.37012970e-01 5.93126118e-01 -1.02630220e-02 5.81132993e-02
4.40465897e-01 5.44165671e-02 7.15371311e-01 -8.14503133e-01
-3.65797013e-01 2.36170515e-01 -2.29373485e-01 4.83376950e-01
2.23739743e-01 -2.32608065e-01 -1.25211215e+00 1.80824950e-01
-1.76638377e+00 -7.75041223e-01 -5.64896017e-02 1.81567597e+00
6.31698728e-01 -1.31739601e-01 -2.51036495e-01 -8.95287246e-02
4.92828131e-01 -2.63527393e-01 -7.12941170e-01 -5.57906441e-02
-3.87803197e-01 1.64163575e-01 3.36023360e-01 2.56236464e-01
-9.50703919e-01 3.79789144e-01 7.65886736e+00 6.24919474e-01
-1.39888382e+00 -1.12328283e-01 8.51349831e-01 -1.26485512e-01
-3.80221933e-01 -4.24312234e-01 -4.47356999e-01 -1.37668729e-01
1.07744026e+00 4.41594012e-02 -2.00223327e-02 1.38241291e-01
1.46176741e-01 -2.34785989e-01 -1.36370838e+00 8.22643936e-01
2.35992018e-02 -1.99372041e+00 -9.34573859e-02 3.86803567e-01
4.99325752e-01 4.46002901e-01 1.84019789e-01 -6.69897348e-02
2.85744309e-01 -1.49984956e+00 -2.93185294e-01 1.02087593e+00
7.71789908e-01 -3.35326076e-01 1.03023648e+00 2.30688937e-02
-1.03341579e+00 -1.74828887e-01 -7.41424644e-03 4.23601478e-01
-1.93476886e-01 4.01963323e-01 -9.97889519e-01 1.03438461e+00
3.19916755e-01 9.56548035e-01 -6.86927199e-01 1.14896214e+00
5.60818911e-01 -1.27239704e-01 -2.66286522e-01 -3.36648934e-02
7.34162033e-02 1.67058527e-01 1.43882811e-01 1.21065104e+00
3.97176474e-01 -1.50851682e-02 1.52506784e-01 3.83213222e-01
-6.52756095e-02 -4.64289971e-02 -5.51145077e-01 -5.72077870e-01
1.09226286e-01 1.62704766e+00 -7.60821640e-01 -1.76847935e-01
-8.10950577e-01 1.00239955e-01 3.08417350e-01 2.92108685e-01
-4.44664598e-01 1.83897130e-02 5.56974649e-01 -9.03638005e-02
-2.79672652e-01 9.18929949e-02 -6.31191313e-01 -7.23221183e-01
-5.92292666e-01 -1.02355230e+00 1.02174318e+00 -7.17279673e-01
-1.50350344e+00 2.62255102e-01 -2.11687639e-01 -9.17930305e-01
1.96537212e-01 -1.10833585e+00 -5.85409820e-01 7.01794744e-01
-1.36035669e+00 -1.47780919e+00 -6.32903576e-01 4.82852578e-01
3.51662515e-03 -3.61792028e-01 1.12261260e+00 -9.00069252e-03
-7.89824605e-01 4.86196190e-01 4.14997488e-01 1.06405482e-01
6.51151180e-01 -1.19718957e+00 -4.54039395e-01 -3.49918641e-02
-5.69595277e-01 4.42253292e-01 2.95430958e-01 -6.06482804e-01
-1.87457681e+00 -6.98767126e-01 3.96684080e-01 -5.26971698e-01
8.77519131e-01 4.47821766e-02 -7.88935542e-01 5.35056353e-01
3.30233604e-01 3.54140192e-01 1.52974355e+00 6.24837726e-02
-1.34106681e-01 -3.67442489e-01 -1.26310706e+00 5.29775262e-01
4.91749942e-01 -4.75825995e-01 4.50601466e-02 5.04960299e-01
4.14612412e-01 -6.39039993e-01 -1.46048820e+00 7.51957595e-01
8.06248426e-01 -7.71229327e-01 1.12782478e+00 -8.41249824e-01
2.87406236e-01 -4.34944071e-02 -3.74091789e-02 -1.18797147e+00
-3.02027315e-01 -1.47219136e-01 -1.62926257e-01 5.27231932e-01
4.09479856e-01 -5.04288554e-01 8.05280566e-01 8.60874414e-01
-3.96712124e-01 -1.33807039e+00 -9.44269836e-01 9.06640291e-02
2.82614291e-01 -1.69938117e-01 4.22170699e-01 8.05897832e-01
4.50246364e-01 7.10534379e-02 2.36239105e-01 -3.71697135e-02
3.65877777e-01 -2.35986695e-01 3.56982619e-01 -1.32489669e+00
1.18022740e-01 -9.31194901e-01 -6.58407629e-01 7.98278823e-02
-1.05679937e-01 -1.11326373e+00 -5.81849456e-01 -2.03323865e+00
3.34795624e-01 -8.37146789e-02 -8.11360538e-01 7.56542504e-01
-6.86504990e-02 2.22632140e-01 -5.71230873e-02 1.00144818e-01
-5.67073286e-01 2.76866645e-01 1.61814249e+00 -6.79712713e-01
-9.05604661e-02 -5.14975429e-01 -1.12083733e+00 4.72327709e-01
7.10229456e-01 -1.26749262e-01 -1.56523079e-01 -2.29023769e-01
1.58109412e-01 5.52989304e-01 4.56837475e-01 -9.04631555e-01
6.88149333e-01 -2.63372123e-01 1.03601146e+00 -6.53235316e-01
4.28400904e-01 -9.32263911e-01 3.89484257e-01 6.84660196e-01
-3.04684430e-01 -5.16353808e-02 5.31363904e-01 4.80025470e-01
-2.13373616e-01 8.23241770e-02 4.66185987e-01 -3.14604163e-01
-2.54620284e-01 4.59452271e-01 -4.97629315e-01 -4.46204126e-01
9.67412114e-01 -2.19062120e-01 -1.04829526e+00 -2.30082214e-01
-1.04373431e+00 5.46357274e-01 1.11841470e-01 1.15641415e-01
7.93336272e-01 -1.28281021e+00 -5.61763287e-01 -4.44971733e-02
2.24399522e-01 -7.29290918e-02 6.98443115e-01 1.47553933e+00
-4.19411421e-01 5.23360372e-01 -3.60647619e-01 -7.13214576e-01
-1.42048776e+00 5.47129214e-01 9.95439172e-01 -6.17791712e-01
-2.69178122e-01 4.64585721e-01 2.06078961e-01 -4.13004845e-01
2.62628019e-01 -1.73205048e-01 -4.31835324e-01 2.59824693e-01
4.19243872e-01 2.73412794e-01 3.67241144e-01 -4.26556021e-01
-4.71783608e-01 3.25738490e-01 -3.28362346e-01 3.02421808e-01
1.14981461e+00 5.63331880e-02 -4.98692453e-01 2.58218944e-01
1.09242570e+00 -5.57947874e-01 -7.02914894e-01 -2.85134077e-01
-3.43525857e-01 1.93278417e-01 3.23443741e-01 -1.09346569e+00
-1.21207702e+00 9.01844800e-01 6.77657843e-01 3.40802699e-01
1.27344608e+00 1.78778976e-01 5.50441325e-01 2.12739602e-01
-1.75186098e-01 -7.79432535e-01 2.68636853e-01 2.70327777e-01
6.99666142e-01 -1.45084453e+00 3.37045550e-01 -2.19877779e-01
-3.32225591e-01 1.63840270e+00 5.66221714e-01 5.47061205e-01
8.44358146e-01 5.74104249e-01 3.78598005e-01 -6.34496987e-01
-8.34330738e-01 -2.53240228e-01 6.09227717e-01 7.37154901e-01
7.52964318e-01 7.99647495e-02 6.22397959e-02 4.60064799e-01
2.05327168e-01 9.67274234e-02 3.05760294e-01 1.17105210e+00
-3.48873526e-01 -1.09168482e+00 -4.03527856e-01 7.40636885e-01
-4.55420375e-01 -1.01612590e-01 -7.36694455e-01 9.11897540e-01
7.53953084e-02 7.70279765e-01 -1.44176632e-01 -3.31761718e-01
5.56697622e-02 5.12475856e-02 5.25030911e-01 -2.02759683e-01
-5.82310379e-01 3.99769485e-01 -1.68590143e-01 -3.50747168e-01
-4.75046188e-01 -4.64361876e-01 -1.27310824e+00 -2.61714846e-01
-5.40160760e-02 -3.11762422e-01 8.14276755e-01 1.06903458e+00
5.16776621e-01 1.23305237e+00 9.21610072e-02 -6.47265911e-01
8.35788772e-02 -7.27093875e-01 -2.13838831e-01 -1.05943590e-01
6.69576406e-01 -4.25364971e-01 1.56630967e-02 -3.44747514e-01]
|
[15.162378311157227, -2.782594919204712]
|
918c57c9-abcc-4a7d-96f9-ddcadef8a67e
|
kernel-spectral-clustering-and-applications
|
1505.00477
| null |
http://arxiv.org/abs/1505.00477v1
|
http://arxiv.org/pdf/1505.00477v1.pdf
|
Kernel Spectral Clustering and applications
|
In this chapter we review the main literature related to kernel spectral
clustering (KSC), an approach to clustering cast within a kernel-based
optimization setting. KSC represents a least-squares support vector machine
based formulation of spectral clustering described by a weighted kernel PCA
objective. Just as in the classifier case, the binary clustering model is
expressed by a hyperplane in a high dimensional space induced by a kernel. In
addition, the multi-way clustering can be obtained by combining a set of binary
decision functions via an Error Correcting Output Codes (ECOC) encoding scheme.
Because of its model-based nature, the KSC method encompasses three main steps:
training, validation, testing. In the validation stage model selection is
performed to obtain tuning parameters, like the number of clusters present in
the data. This is a major advantage compared to classical spectral clustering
where the determination of the clustering parameters is unclear and relies on
heuristics. Once a KSC model is trained on a small subset of the entire data,
it is able to generalize well to unseen test points. Beyond the basic
formulation, sparse KSC algorithms based on the Incomplete Cholesky
Decomposition (ICD) and $L_0$, $L_1, L_0 + L_1$, Group Lasso regularization are
reviewed. In that respect, we show how it is possible to handle large scale
data. Also, two possible ways to perform hierarchical clustering and a soft
clustering method are presented. Finally, real-world applications such as image
segmentation, power load time-series clustering, document clustering and big
data learning are considered.
|
['Johan A. K. Suykens', 'Carlos Alzate', 'Rocco Langone', 'Raghvendra Mall']
|
2015-05-03
| null | null | null | null |
['time-series-clustering']
|
['time-series']
|
[ 5.60300723e-02 -2.27820277e-01 -3.50054115e-01 -2.28969604e-01
-6.95370138e-01 -5.80286205e-01 2.35383600e-01 2.14482948e-01
-2.03547671e-01 4.65624511e-01 -2.26141378e-01 -1.32654980e-01
-8.00435066e-01 -3.66286397e-01 -4.56561506e-01 -1.26546037e+00
-3.86585534e-01 6.17013633e-01 -8.59367624e-02 3.28552157e-01
2.57184565e-01 7.01967061e-01 -1.58136439e+00 2.63374746e-01
1.11168694e+00 9.11951542e-01 6.38566911e-02 4.61882204e-01
-1.16020761e-01 3.50093901e-01 -3.60871136e-01 1.55839249e-01
1.55597612e-01 -3.98767442e-01 -5.91358602e-01 5.74896514e-01
-1.11505724e-01 6.51275098e-01 4.23914015e-01 9.87259209e-01
7.39125162e-02 2.71208882e-01 1.10621524e+00 -1.46512151e+00
-3.26819509e-01 4.29643929e-01 -5.30085087e-01 -4.55359876e-01
1.91546947e-01 -2.78075546e-01 5.71907043e-01 -8.20781410e-01
3.88796508e-01 9.20586944e-01 6.67880177e-01 1.86194465e-01
-1.82392693e+00 -2.91586071e-01 -1.06751099e-01 8.77802819e-02
-1.62950671e+00 -1.20674990e-01 9.12224412e-01 -8.91384304e-01
5.93403876e-01 4.54908222e-01 5.93127191e-01 6.89243436e-01
-2.87025273e-01 6.27626419e-01 1.38954425e+00 -7.19912648e-01
9.22670245e-01 4.37864125e-01 5.42962968e-01 5.06282210e-01
7.37718865e-02 -1.28590733e-01 -1.21339060e-01 -5.22619367e-01
2.38779187e-01 -1.25009164e-01 -3.61245811e-01 -9.42987800e-01
-9.46949363e-01 1.11461771e+00 1.71989948e-01 5.12143493e-01
-2.45218128e-01 -1.20448209e-01 5.24815679e-01 4.40870672e-02
3.23905408e-01 4.29848343e-01 -3.17638308e-01 1.34175777e-01
-1.46537697e+00 -2.03592345e-01 9.02107596e-01 6.72031879e-01
9.32551086e-01 1.47106931e-01 2.15082362e-01 9.88801301e-01
2.19554931e-01 4.28579807e-01 7.04459071e-01 -1.09262311e+00
6.14143796e-02 7.70985007e-01 -1.35369584e-01 -1.01177764e+00
-4.93433088e-01 -3.98243397e-01 -1.15654373e+00 2.27318376e-01
3.31181675e-01 3.05205584e-04 -6.68464065e-01 1.31630826e+00
3.09148163e-01 4.01534796e-01 1.80708244e-01 8.41301918e-01
2.14038059e-01 5.38639069e-01 -7.93583132e-03 -7.33777046e-01
1.07487297e+00 -6.14250183e-01 -6.56327248e-01 1.39468998e-01
6.94503963e-01 -7.05317855e-01 8.41616154e-01 6.08579397e-01
-5.74719787e-01 -4.57804769e-01 -8.91522527e-01 5.02977192e-01
-7.52428412e-01 4.75141525e-01 6.68198943e-01 8.92552376e-01
-1.02762961e+00 5.63812315e-01 -9.32532132e-01 -4.48613822e-01
4.93245348e-02 5.62184155e-01 -5.14833093e-01 2.37217546e-02
-7.31047988e-01 5.78284442e-01 6.33456945e-01 7.27806166e-02
-2.32216835e-01 -4.04445589e-01 -8.74870479e-01 3.34145427e-02
2.74484456e-01 -5.41409738e-02 2.97339976e-01 -1.05932736e+00
-1.31685674e+00 8.86893749e-01 -2.49632806e-01 -4.49825197e-01
8.96078497e-02 2.43564367e-01 -4.23115969e-01 2.94486821e-01
1.30650729e-01 2.02283412e-01 1.20837057e+00 -1.54509604e+00
-2.10701317e-01 -4.95819271e-01 -8.27640533e-01 -1.70969948e-01
-2.05057502e-01 -4.68111690e-03 -5.03513038e-01 -7.39668787e-01
3.82463992e-01 -1.13094985e+00 -5.20377278e-01 -5.61713517e-01
-5.67455351e-01 -1.62341505e-01 1.09392965e+00 -5.17411351e-01
1.47872853e+00 -2.50603914e+00 4.62603420e-01 9.51851726e-01
-1.33592919e-01 7.27719590e-02 1.65566176e-01 4.82085228e-01
-4.67276573e-01 -2.38907970e-02 -6.74529672e-01 -2.29042485e-01
-2.27089971e-02 1.97445571e-01 -2.85186879e-02 7.72768319e-01
-4.71322238e-02 3.95762593e-01 -5.47226548e-01 -5.58763266e-01
4.33908761e-01 2.87553191e-01 -2.46683642e-01 -9.55613703e-02
5.89466281e-02 1.51849821e-01 -2.91843265e-01 6.30483270e-01
7.22287476e-01 -4.98265088e-01 2.74734676e-01 -1.54776037e-01
-1.73261002e-01 -8.42627525e-01 -1.68658447e+00 1.40187430e+00
-1.38058782e-01 3.16743523e-01 5.27969539e-01 -1.78624272e+00
9.76142704e-01 3.85473639e-01 8.82434368e-01 8.21496621e-02
4.29258086e-02 3.85602891e-01 -4.00003910e-01 -2.88570166e-01
1.60157427e-01 4.44103889e-02 -1.11365423e-01 2.43411496e-01
6.83065057e-02 -8.42873566e-03 3.46923172e-01 1.99357912e-01
7.85298228e-01 -2.30299383e-01 4.01535273e-01 -6.05860233e-01
9.20943260e-01 3.98970783e-01 4.67503011e-01 4.38006312e-01
-6.44838735e-02 6.20119274e-01 5.77665687e-01 -7.17899650e-02
-7.74496794e-01 -8.36375594e-01 -5.30122995e-01 6.17481947e-01
-1.91842377e-01 -4.06354904e-01 -9.91130829e-01 -4.68948036e-01
1.12021521e-01 5.46122551e-01 -6.25226676e-01 -1.76125407e-01
-2.27985546e-01 -1.04284334e+00 1.99192971e-01 2.13757515e-01
2.16017842e-01 -7.59923160e-01 -3.01651716e-01 1.14820704e-01
6.75246194e-02 -9.43146110e-01 -2.58283943e-01 6.88626170e-01
-1.02253211e+00 -1.27143526e+00 -5.27632356e-01 -8.49909425e-01
8.60398293e-01 6.69829696e-02 6.32629037e-01 -1.13412336e-01
-5.21617889e-01 6.35967672e-01 -5.51174641e-01 1.55755490e-01
-2.59012640e-01 -1.17944274e-02 2.25244567e-01 5.36781371e-01
5.31826377e-01 -4.99662697e-01 -1.98770851e-01 3.57776344e-01
-8.97644639e-01 -2.74132431e-01 4.49858665e-01 8.79438102e-01
8.26663971e-01 5.11210501e-01 4.43842560e-01 -8.51916134e-01
5.96987545e-01 -5.57496548e-01 -7.72997856e-01 4.42032993e-01
-7.59537160e-01 7.00942278e-02 8.86466622e-01 -5.91016114e-01
-6.15532041e-01 8.26269507e-01 3.75571102e-01 -8.58569443e-01
-4.68099594e-01 5.94821513e-01 -6.14952222e-02 -1.58141881e-01
8.18765342e-01 4.59580898e-01 3.17834258e-01 -5.04793584e-01
4.50498104e-01 7.57261753e-01 3.52578223e-01 -6.34169281e-01
8.01820576e-01 6.21249974e-01 1.34050876e-01 -1.07774580e+00
-4.16830957e-01 -1.01315355e+00 -1.12678409e+00 -2.39108324e-01
1.20600414e+00 -7.08264709e-01 -7.84634054e-01 1.28559291e-01
-6.31355584e-01 -2.92437017e-01 -3.61632764e-01 6.87273681e-01
-7.60210693e-01 4.64787871e-01 -3.61974418e-01 -9.69243884e-01
-5.15556931e-02 -1.24907160e+00 8.96031082e-01 -8.29397142e-03
-1.64782614e-01 -1.17685723e+00 -5.53799681e-02 3.26872647e-01
8.30477253e-02 4.87097412e-01 1.11483133e+00 -7.54070461e-01
-1.70274392e-01 -4.01027501e-01 1.52793258e-01 6.55262172e-01
-4.07338291e-02 1.20785356e-01 -6.49631500e-01 -5.03580749e-01
6.45870194e-02 -1.32075235e-01 7.67933488e-01 6.27611697e-01
1.27872562e+00 -2.26842120e-01 -4.69350189e-01 6.29450500e-01
1.62861633e+00 3.01497638e-01 3.59062433e-01 1.48965642e-01
5.26381671e-01 6.99938893e-01 4.27113235e-01 3.52485716e-01
-1.91436857e-01 5.54056108e-01 5.72076254e-02 -1.84694722e-01
3.21020991e-01 2.15959027e-01 2.34332711e-01 7.82848239e-01
2.85520311e-02 4.07213211e-01 -9.46650624e-01 3.13959956e-01
-2.01239610e+00 -1.05257106e+00 -5.34773648e-01 2.54741669e+00
5.03094792e-01 -1.78211063e-01 2.02774569e-01 5.90388477e-01
9.67666745e-01 -2.50414848e-01 -3.46082419e-01 -3.38432193e-01
-3.56357992e-01 1.40792876e-01 6.00184321e-01 5.43144524e-01
-1.41799366e+00 7.44888902e-01 6.32792711e+00 1.27534962e+00
-1.09706116e+00 -1.54381007e-01 6.14981174e-01 1.09089479e-01
1.87526539e-01 1.97494462e-01 -4.96603251e-01 7.24613667e-01
8.27705562e-01 1.60930067e-01 6.50130749e-01 1.08024204e+00
4.00253773e-01 -4.91524875e-01 -8.39987934e-01 1.25245571e+00
9.52651203e-02 -1.15493882e+00 -3.43464285e-01 2.94018030e-01
9.09616113e-01 -4.32462305e-01 -8.69676936e-03 1.57092184e-01
1.02982603e-01 -1.01812518e+00 3.02938819e-01 4.17588472e-01
6.98083758e-01 -7.08653748e-01 4.61057723e-01 4.62450445e-01
-1.15624499e+00 -2.29891449e-01 -3.28098774e-01 2.53223300e-01
-4.55900766e-02 1.03213143e+00 -5.37164569e-01 6.03296161e-01
6.46655381e-01 6.66053414e-01 -7.24176049e-01 1.01233077e+00
1.49162754e-01 7.74906039e-01 -4.02632803e-01 1.11476481e-01
2.27783054e-01 -9.74611640e-01 3.87580872e-01 1.24824536e+00
3.33775997e-01 2.00553373e-01 5.85818410e-01 6.83022380e-01
5.36918521e-01 4.58998114e-01 -4.82641757e-01 7.01237097e-02
4.34744507e-01 1.36526895e+00 -1.20548689e+00 -3.06234926e-01
-3.39303493e-01 1.09625852e+00 2.19311088e-01 6.73788905e-01
-4.19934273e-01 -3.21933985e-01 2.74354845e-01 2.49637198e-02
2.61284024e-01 -2.96061307e-01 -3.98223341e-01 -1.16311359e+00
-1.24331355e-01 -7.58414865e-01 6.27866864e-01 -4.45277542e-01
-1.26598012e+00 2.81709969e-01 4.49523926e-02 -1.29404163e+00
-2.36420125e-01 -8.09047699e-01 -5.14781415e-01 7.17221975e-01
-9.05912519e-01 -8.04403126e-01 3.08817867e-02 9.31140900e-01
2.21822858e-01 -3.78082365e-01 9.89140332e-01 3.16023268e-02
-7.65887022e-01 7.11559802e-02 9.22502339e-01 1.20587386e-01
5.48967242e-01 -1.53787708e+00 -7.77257681e-01 6.71911180e-01
-8.10270086e-02 7.36440957e-01 6.28219903e-01 -5.49798369e-01
-1.23708129e+00 -9.95893121e-01 5.67010999e-01 4.61975113e-03
6.86418533e-01 -3.61250490e-01 -1.07164991e+00 3.58216852e-01
-1.31382272e-01 1.12773940e-01 1.22203195e+00 2.87119806e-01
-2.17896029e-01 -1.87537506e-01 -1.13184512e+00 1.39387339e-01
1.87985897e-01 -4.73580420e-01 -2.84768999e-01 4.83410180e-01
1.16255380e-01 2.17331603e-01 -1.12532830e+00 2.18922332e-01
1.94737241e-01 -9.63169098e-01 1.07416844e+00 -4.71624762e-01
-1.54627249e-01 -5.96989155e-01 -2.73334324e-01 -1.18453681e+00
-5.03173470e-01 -4.09471512e-01 -7.01237693e-02 1.09255683e+00
3.04693460e-01 -4.17728513e-01 9.83542264e-01 4.43989515e-01
-2.31777765e-02 -7.33686090e-01 -1.05867338e+00 -9.94903207e-01
3.39841172e-02 -4.71216023e-01 1.10376060e-01 1.33675873e+00
3.62720788e-01 1.57408208e-01 -2.45112330e-01 2.89839327e-01
1.08700788e+00 3.61013830e-01 4.26049054e-01 -1.53381622e+00
-2.80290842e-01 -4.80516613e-01 -4.06968296e-01 -3.10273468e-01
5.59666932e-01 -9.91219699e-01 -1.42508164e-01 -1.13161635e+00
7.86742545e-04 -7.35692203e-01 -1.22705877e-01 1.77935869e-01
5.15718460e-02 1.28919214e-01 8.80730599e-02 5.66962600e-01
-5.69887638e-01 2.14799643e-01 5.81014931e-01 -6.65383562e-02
-5.48797965e-01 2.22519681e-01 -1.40096694e-01 6.13344014e-01
8.58857155e-01 -2.84012645e-01 -2.91706026e-01 3.82291943e-01
-1.57034844e-01 2.85581321e-01 2.88460106e-01 -9.75277126e-01
5.03369272e-01 -2.03217119e-01 4.22861874e-01 -4.91559744e-01
3.51295173e-01 -1.11079824e+00 4.35918033e-01 3.59001189e-01
-1.79513901e-01 -3.11228335e-01 -2.29697704e-01 7.13813424e-01
-4.32496160e-01 -4.52561140e-01 1.05320871e+00 -7.54415989e-02
-6.78039849e-01 -8.61120075e-02 -7.08389163e-01 -3.28414798e-01
1.42288625e+00 -5.60034931e-01 2.85337269e-01 2.75678877e-02
-1.36577439e+00 3.59705389e-01 4.79777992e-01 -6.89950138e-02
4.69603598e-01 -1.38899946e+00 -3.06264341e-01 2.32943267e-01
1.15230061e-01 -1.98419929e-01 2.66766131e-01 1.20306110e+00
-3.15396845e-01 5.30062556e-01 2.71091938e-01 -8.80153537e-01
-1.29021287e+00 1.04211617e+00 2.42183119e-01 -8.41164216e-02
-2.53411978e-01 2.67944723e-01 -1.77311689e-01 -4.61546600e-01
2.52576590e-01 7.03251958e-02 -2.38772348e-01 2.77318299e-01
2.10218504e-01 6.40917659e-01 8.81524831e-02 -8.47402573e-01
-3.63734663e-01 8.87030005e-01 5.25856137e-01 -1.91782564e-02
1.09885669e+00 -6.73786849e-02 -5.85403800e-01 6.13613367e-01
1.56137240e+00 -1.88520700e-01 -8.43961716e-01 -1.95074119e-02
4.63646591e-01 -1.67050049e-01 -1.04278706e-01 -5.01630843e-01
-8.44718754e-01 6.35663867e-01 6.03358626e-01 4.47682619e-01
1.31135881e+00 1.67374417e-01 -1.54505938e-01 2.57046223e-01
1.88298374e-01 -1.57647872e+00 -1.53021544e-01 1.93824321e-01
6.44814610e-01 -1.21826768e+00 -1.30807593e-01 -6.40559673e-01
-5.72745621e-01 1.34373546e+00 2.07974046e-01 -1.62824810e-01
1.11583209e+00 2.01945379e-01 -7.80452043e-02 -2.22012848e-01
-3.17498744e-01 -3.18745434e-01 3.64896804e-01 6.63610756e-01
2.81171560e-01 2.84960806e-01 -5.15699983e-01 5.60022354e-01
-1.12333454e-01 -3.17417800e-01 2.87993491e-01 6.75144196e-01
-5.24320543e-01 -1.04193234e+00 -9.83475506e-01 4.82191950e-01
-1.92230433e-01 2.64298975e-01 -2.54108369e-01 6.87533855e-01
1.52660877e-01 1.06048083e+00 -1.35914981e-01 -1.64750263e-01
8.92407522e-02 5.09685695e-01 -3.81376222e-02 -5.47039151e-01
-2.36141607e-01 5.08389354e-01 -3.61648381e-01 -4.66518879e-01
-6.05194986e-01 -8.87071788e-01 -1.33286786e+00 -2.17042528e-02
-4.67813551e-01 7.94230580e-01 7.83651829e-01 7.57438242e-01
1.28329724e-01 9.23313424e-02 8.10425401e-01 -6.58092856e-01
-3.89807373e-01 -6.06533170e-01 -1.26226676e+00 5.30106366e-01
1.38758808e-01 -6.14324272e-01 -6.83336258e-01 4.06907648e-01]
|
[7.742588520050049, 4.351927757263184]
|
98491218-ae57-4091-9539-9f36ab5338c0
|
meta-tuning-loss-functions-and-data
|
2304.12161
| null |
https://arxiv.org/abs/2304.12161v1
|
https://arxiv.org/pdf/2304.12161v1.pdf
|
Meta-tuning Loss Functions and Data Augmentation for Few-shot Object Detection
|
Few-shot object detection, the problem of modelling novel object detection categories with few training instances, is an emerging topic in the area of few-shot learning and object detection. Contemporary techniques can be divided into two groups: fine-tuning based and meta-learning based approaches. While meta-learning approaches aim to learn dedicated meta-models for mapping samples to novel class models, fine-tuning approaches tackle few-shot detection in a simpler manner, by adapting the detection model to novel classes through gradient based optimization. Despite their simplicity, fine-tuning based approaches typically yield competitive detection results. Based on this observation, we focus on the role of loss functions and augmentations as the force driving the fine-tuning process, and propose to tune their dynamics through meta-learning principles. The proposed training scheme, therefore, allows learning inductive biases that can boost few-shot detection, while keeping the advantages of fine-tuning based approaches. In addition, the proposed approach yields interpretable loss functions, as opposed to highly parametric and complex few-shot meta-models. The experimental results highlight the merits of the proposed scheme, with significant improvements over the strong fine-tuning based few-shot detection baselines on benchmark Pascal VOC and MS-COCO datasets, in terms of both standard and generalized few-shot performance metrics.
|
['Ramazan Gokberk Cinbis', 'Orhun Buğra Baran', 'Berkan Demirel']
|
2023-04-24
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Demirel_Meta-Tuning_Loss_Functions_and_Data_Augmentation_for_Few-Shot_Object_Detection_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Demirel_Meta-Tuning_Loss_Functions_and_Data_Augmentation_for_Few-Shot_Object_Detection_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['few-shot-object-detection']
|
['computer-vision']
|
[ 2.18508944e-01 -3.95193091e-03 -4.35576439e-01 -2.91962355e-01
-8.18862736e-01 3.97443138e-02 9.22448754e-01 3.46655160e-01
-5.83458900e-01 4.94991213e-01 -5.35215922e-02 3.63071114e-01
-1.55234948e-01 -9.54386294e-01 -7.27698207e-01 -7.51082599e-01
2.24355415e-01 2.99997807e-01 7.58007526e-01 -2.71181434e-01
3.64786893e-01 2.03366384e-01 -2.01226616e+00 2.20192716e-01
7.25076556e-01 9.15385783e-01 2.81151325e-01 7.99629271e-01
-2.70635873e-01 7.01820374e-01 -5.04256189e-01 -4.96331900e-01
1.20910212e-01 -4.00083721e-01 -4.37207222e-01 2.28491914e-03
4.44687426e-01 -2.04513848e-01 -1.35902137e-01 1.10427678e+00
6.71549618e-01 5.81770539e-01 9.17700529e-01 -1.07172990e+00
-6.68778002e-01 4.06045973e-01 -7.42672443e-01 5.80851316e-01
7.81654438e-04 4.18303221e-01 1.08354151e+00 -1.22839773e+00
5.78353703e-01 1.26007712e+00 8.43242049e-01 9.24859822e-01
-1.24650526e+00 -4.32375163e-01 2.09080458e-01 4.18587863e-01
-1.33848274e+00 -5.07589757e-01 6.86346889e-01 -5.65177083e-01
9.16636348e-01 8.04889277e-02 4.71700311e-01 1.04436612e+00
-1.86095387e-02 9.45215940e-01 9.17387545e-01 -7.04582155e-01
5.27679622e-01 6.13572001e-01 3.06249142e-01 6.39824033e-01
3.18436474e-01 2.51841366e-01 -4.23239529e-01 -1.60040453e-01
3.87477189e-01 3.98221284e-01 6.19278252e-02 -7.13422060e-01
-8.26174974e-01 1.08978879e+00 5.98185837e-01 3.64804029e-01
-2.88474023e-01 1.51944458e-01 6.30983531e-01 -5.35971634e-02
5.30439794e-01 4.94216114e-01 -6.70843720e-02 1.49564922e-01
-1.07254481e+00 2.88040280e-01 4.83013809e-01 9.84631896e-01
9.47270930e-01 2.07611978e-01 -9.22860444e-01 1.05251420e+00
1.81473270e-01 8.51628333e-02 8.16152096e-01 -4.26327527e-01
1.95122108e-01 5.09057760e-01 1.41679078e-01 -5.81462801e-01
-2.46207014e-01 -6.79911256e-01 -3.79547209e-01 3.55957776e-01
6.91324398e-02 -1.62171759e-02 -1.15198648e+00 1.57146680e+00
5.94821990e-01 3.60549241e-01 -2.43165910e-01 7.16806173e-01
9.59541798e-01 5.37068367e-01 3.16596836e-01 -3.25603455e-01
1.29173708e+00 -1.29317188e+00 -5.26145756e-01 -1.95896685e-01
5.44775724e-01 -4.01218116e-01 1.19934595e+00 -1.38032541e-01
-1.04081023e+00 -7.59884715e-01 -1.21745884e+00 2.42295578e-01
-7.09297240e-01 -1.76215336e-01 4.71769512e-01 7.95427501e-01
-6.08278096e-01 6.91413224e-01 -6.95068717e-01 -5.78292429e-01
8.58787835e-01 2.46835203e-04 3.15528393e-01 -5.85513711e-02
-9.56515729e-01 9.57783639e-01 8.14871728e-01 -3.56779963e-01
-1.17432535e+00 -9.95657623e-01 -8.01377475e-01 2.70581156e-01
6.23704255e-01 -8.31551850e-01 1.34779584e+00 -5.78345060e-01
-1.26435888e+00 9.30541456e-01 3.69668975e-02 -7.81272650e-01
6.32164001e-01 -1.30507529e-01 -2.76891559e-01 4.40561352e-03
6.92749694e-02 7.49407947e-01 1.23554969e+00 -1.18508399e+00
-1.12190580e+00 -8.13108683e-02 1.98177129e-01 3.23384553e-02
-5.95944047e-01 3.14652771e-02 -3.34139138e-01 -6.62465394e-01
-4.74944502e-01 -5.45742035e-01 -2.46348664e-01 1.83505908e-01
-7.96531811e-02 -3.58450115e-01 8.14117372e-01 1.62015498e-01
1.21800005e+00 -2.00484276e+00 8.87027662e-03 -4.27370161e-01
2.75996238e-01 7.63506174e-01 -1.36812240e-01 1.98077172e-01
2.59754002e-01 -2.72351712e-01 -1.89533427e-01 -4.63737607e-01
1.52066007e-01 4.81772646e-02 -2.76109904e-01 2.89131880e-01
4.82749254e-01 1.26568329e+00 -1.08426881e+00 -5.58760524e-01
6.21640623e-01 3.75647902e-01 -4.44285870e-01 2.61434078e-01
-3.48197430e-01 -1.10278837e-01 -2.26742983e-01 7.99702704e-01
3.10598582e-01 -1.57380968e-01 -3.58732909e-01 -5.23151681e-02
-1.16005979e-01 -3.06338549e-01 -1.14830732e+00 1.55720222e+00
-5.09382546e-01 3.90696228e-01 -3.12065393e-01 -1.11048484e+00
9.22177851e-01 3.68462540e-02 1.52901486e-01 -5.99771678e-01
2.80918747e-01 1.85902402e-01 -1.73959538e-01 -3.56212795e-01
6.27420485e-01 -5.65400004e-01 1.60534173e-01 2.89913237e-01
5.26236355e-01 4.13102061e-02 4.47884262e-01 7.47595727e-02
7.03535855e-01 1.95644900e-01 7.57491291e-01 5.67851216e-02
2.97424853e-01 -2.49473900e-02 5.05368114e-01 1.28191519e+00
-5.81669271e-01 5.02954364e-01 -1.18534587e-01 -3.42398137e-01
-1.00196290e+00 -9.06108439e-01 -2.35161006e-01 1.72839332e+00
2.19970390e-01 -3.33791554e-01 -6.95423007e-01 -5.77621758e-01
9.21835974e-02 8.64830732e-01 -1.02407408e+00 -5.99727273e-01
-3.41021329e-01 -1.22294927e+00 3.38334978e-01 7.27937639e-01
2.25005850e-01 -1.10170901e+00 -1.03988838e+00 4.86747086e-01
2.56118089e-01 -7.73962140e-01 -2.09234402e-01 4.89044935e-01
-9.66227651e-01 -9.98072207e-01 -1.14067447e+00 -6.89488530e-01
3.48006099e-01 7.02708125e-01 9.02111471e-01 -5.66868819e-02
-1.05215716e+00 5.40474415e-01 -4.89229769e-01 -8.09685409e-01
-2.85706878e-01 -1.51345069e-02 -1.67361982e-02 1.59643456e-01
6.16954446e-01 -4.16563481e-01 -6.79061711e-01 2.08820209e-01
-7.71468818e-01 -3.94237816e-01 6.22430682e-01 1.21190667e+00
6.47786796e-01 -2.66463697e-01 6.59104526e-01 -1.07955384e+00
3.72533977e-01 -5.34659266e-01 -4.56123680e-01 4.10466850e-01
-8.13417614e-01 -8.29345956e-02 3.54867607e-01 -6.79101467e-01
-1.20746291e+00 -8.50383714e-02 2.30773687e-01 -7.27827609e-01
-9.93328542e-02 -3.60003337e-02 1.68049932e-01 -4.13088918e-01
1.09269261e+00 3.25354785e-01 -3.84104729e-01 -5.14637053e-01
7.18649924e-01 4.08862263e-01 4.14076298e-01 -3.81182045e-01
8.48867655e-01 6.59795284e-01 -2.39812151e-01 -9.10687506e-01
-1.18339050e+00 -1.11304891e+00 -7.52431512e-01 -2.61043310e-01
7.59702861e-01 -8.61198246e-01 -2.23936349e-01 3.11983913e-01
-8.18385839e-01 -6.14116900e-02 -8.97961259e-01 2.07032904e-01
-7.90118814e-01 1.17437042e-01 -3.82249773e-01 -1.02627397e+00
-4.31208551e-01 -7.70193338e-01 1.10779834e+00 4.65139449e-01
-5.02496213e-02 -1.07859766e+00 2.76405543e-01 1.29185572e-01
4.29749519e-01 3.22693318e-01 7.52711594e-01 -6.04542315e-01
-4.45986271e-01 -4.69700485e-01 -3.04794133e-01 1.44283533e-01
-5.72307669e-02 -1.23590983e-01 -1.41928935e+00 -5.14626086e-01
-1.17194071e-01 -6.01970971e-01 1.41875303e+00 4.63433146e-01
9.13297236e-01 -2.33860500e-02 -5.08656025e-01 6.99973762e-01
1.79473138e+00 -2.80297156e-02 3.43758643e-01 4.94915187e-01
5.18730283e-01 3.02069366e-01 9.24878955e-01 6.58064544e-01
8.70375112e-02 8.33501995e-01 4.94506478e-01 9.03924555e-02
-4.18175161e-01 -2.68985480e-01 -1.22188292e-01 2.99166858e-01
-1.73319981e-01 1.64574966e-01 -6.17234230e-01 7.10222304e-01
-2.15271592e+00 -1.35556686e+00 3.89740825e-01 2.19428182e+00
5.56470156e-01 5.07966220e-01 5.26373446e-01 6.50304109e-02
9.13750291e-01 3.17056149e-01 -7.54581392e-01 -3.20529372e-01
-4.69805226e-02 1.02205358e-01 2.24992603e-01 9.39563438e-02
-1.31283104e+00 9.66667533e-01 6.02494097e+00 1.06510055e+00
-1.03472805e+00 6.77250266e-01 2.33252794e-01 -4.68816638e-01
9.11731049e-02 -1.79455772e-01 -1.39923704e+00 3.15335125e-01
6.62466764e-01 -3.61239105e-01 5.48373759e-02 1.16664135e+00
-4.61017303e-02 9.73104611e-02 -1.17006373e+00 9.44995224e-01
3.32751185e-01 -1.57486916e+00 1.63586110e-01 -2.83640385e-01
1.04803836e+00 7.10353032e-02 1.43979460e-01 9.18279827e-01
3.17128450e-02 -5.59613168e-01 8.61330688e-01 4.71036047e-01
5.45516312e-01 -5.75464308e-01 4.42117900e-01 5.37282705e-01
-1.19885325e+00 -6.75854027e-01 -9.17582273e-01 -1.45670220e-01
3.79175730e-02 3.70653301e-01 -8.37645769e-01 2.55795717e-01
7.10027933e-01 5.80562234e-01 -7.04998910e-01 1.66646719e+00
-1.02169132e-02 4.83788311e-01 1.42779097e-01 -3.33755940e-01
3.72543454e-01 2.71253973e-01 7.54745722e-01 1.47284126e+00
-3.82434763e-02 -8.29916745e-02 1.95295110e-01 1.07879615e+00
-2.19931807e-02 7.81056881e-02 -3.76936316e-01 8.71394500e-02
4.14847583e-01 1.44000006e+00 -7.95229495e-01 -6.18476808e-01
-4.40334320e-01 6.67153656e-01 5.83035886e-01 9.82753858e-02
-8.03881645e-01 -6.14863575e-01 3.76346976e-01 1.61114603e-01
8.05660188e-01 3.50377589e-01 -8.45002383e-02 -1.05914950e+00
-1.66734695e-01 -4.41475183e-01 6.65072560e-01 -3.26932400e-01
-1.31671178e+00 2.59080350e-01 1.98560923e-01 -1.33039773e+00
-2.75595278e-01 -5.36768317e-01 -9.73698199e-01 4.33077216e-01
-1.79206502e+00 -1.38001323e+00 -3.65763545e-01 2.66249955e-01
1.06243813e+00 -1.83598965e-01 7.31562197e-01 2.95350105e-01
-6.11682355e-01 8.55902135e-01 3.56702864e-01 -2.90912569e-01
5.78586340e-01 -1.29363072e+00 3.16111565e-01 7.43705332e-01
2.93236196e-01 5.77604949e-01 7.71663547e-01 -3.05616826e-01
-9.80664074e-01 -1.44224536e+00 3.61743629e-01 -4.66230333e-01
5.84300458e-01 -3.40531379e-01 -1.05351007e+00 2.31208146e-01
-2.34511465e-01 4.61547434e-01 6.03346825e-01 2.23333165e-01
-5.56812525e-01 2.30523385e-02 -1.18172228e+00 3.48760247e-01
9.17706013e-01 -4.24100190e-01 -8.79999459e-01 4.37934399e-01
6.83939338e-01 -8.04741755e-02 -5.09389579e-01 3.78662646e-01
5.79679072e-01 -1.02626431e+00 1.10796869e+00 -9.24821615e-01
1.95477813e-01 -1.66490078e-01 -1.51302874e-01 -1.28682983e+00
-7.91917741e-01 -2.46724024e-01 -7.78035283e-01 1.20647705e+00
1.55018270e-01 -3.01458567e-01 8.08937967e-01 1.43137723e-01
-1.26889333e-01 -8.70535195e-01 -7.04436660e-01 -1.10875630e+00
-2.59428304e-02 -1.49338260e-01 3.41262400e-01 6.97662592e-01
-2.12121785e-01 3.19851309e-01 -5.15123248e-01 -1.76077813e-01
1.05365562e+00 2.14780420e-01 8.15128624e-01 -1.23711669e+00
-6.30273640e-01 -6.89919531e-01 -6.38845623e-01 -4.60179567e-01
-9.48045030e-02 -7.50392556e-01 3.33787113e-01 -1.37155581e+00
6.77958071e-01 -1.19456612e-01 -7.38610744e-01 4.08896506e-01
-6.00049496e-01 5.82188666e-01 4.25143570e-01 2.43122727e-01
-1.01267552e+00 7.13470340e-01 9.07350957e-01 -4.66135561e-01
-3.43585730e-01 2.39475965e-01 -5.57244658e-01 8.32819521e-01
4.46426779e-01 -5.31700253e-01 -3.07927012e-01 1.33898463e-02
-1.92914754e-01 -4.80687410e-01 5.45966685e-01 -1.17070365e+00
3.19156915e-01 -1.30303666e-01 3.89520496e-01 -4.10667270e-01
5.38012207e-01 -3.93312156e-01 -5.24785221e-01 7.89408326e-01
-2.53009498e-01 -8.01899076e-01 1.05637811e-01 1.11426198e+00
-8.93660039e-02 -6.00652933e-01 1.48216641e+00 -4.29454595e-01
-1.38537502e+00 4.20413107e-01 2.28888281e-02 3.18816006e-01
1.30652702e+00 -7.60777116e-01 -1.37744874e-01 1.66877225e-01
-7.77914047e-01 7.84125850e-02 2.33693749e-01 7.02138782e-01
6.19976223e-01 -1.23763680e+00 -5.93730211e-01 -8.89394209e-02
8.63063395e-01 -3.37806731e-01 3.25271577e-01 8.03126514e-01
9.10615921e-02 3.41550380e-01 -2.91309655e-01 -7.36334085e-01
-1.12665033e+00 9.46322918e-01 3.32175434e-01 -1.34493094e-02
-7.23866701e-01 1.13429141e+00 4.91916180e-01 -3.33840638e-01
5.62241971e-01 5.44359088e-02 -2.97908127e-01 3.92477065e-01
9.22308385e-01 8.61943901e-01 -9.93458368e-03 -4.04486179e-01
-1.74821898e-01 4.89052147e-01 -3.66908759e-01 3.38927150e-01
1.33379209e+00 -1.96295321e-01 5.21068513e-01 7.87265003e-01
9.69742477e-01 -6.42540574e-01 -1.35045016e+00 -6.28447711e-01
9.49687287e-02 -5.59875369e-01 9.53660086e-02 -6.36284828e-01
-4.87094969e-01 1.04254544e+00 1.00170207e+00 -8.45243782e-02
7.34796405e-01 9.78811234e-02 5.87216675e-01 5.04277945e-01
4.00604844e-01 -1.43105423e+00 4.67093229e-01 3.60912442e-01
3.95444542e-01 -1.71063364e+00 -9.70233008e-02 -1.13301821e-01
-3.23405445e-01 8.70654285e-01 8.56581569e-01 -1.39965072e-01
4.65181231e-01 -1.66767418e-01 -2.60265946e-01 -2.11370885e-01
-7.92563319e-01 -5.60393155e-01 5.01218736e-01 7.48381972e-01
1.47852957e-01 -4.43105288e-02 -1.29349276e-01 5.19429088e-01
4.37757999e-01 1.14387773e-01 2.24365443e-01 1.02568483e+00
-1.15443134e+00 -5.70465565e-01 -3.47264767e-01 6.96044922e-01
-1.77621856e-01 -7.28655048e-03 -1.70894742e-01 7.00635016e-01
2.78787881e-01 6.98068321e-01 -1.26337409e-01 -2.63547134e-02
5.59111416e-01 1.36312708e-01 6.21615887e-01 -1.21996093e+00
-5.81154525e-01 -1.90344423e-01 -3.87001514e-01 -2.88951904e-01
-3.71967405e-01 -4.74400938e-01 -8.05042326e-01 1.80984557e-01
-8.95341039e-01 1.39541915e-02 4.59033370e-01 1.03884733e+00
8.41599256e-02 6.20648861e-01 6.68925881e-01 -1.08629107e+00
-1.13532925e+00 -1.01834273e+00 -6.41852736e-01 3.65201205e-01
2.98200548e-01 -1.03109586e+00 -2.92146772e-01 -1.16478384e-01]
|
[9.848987579345703, 2.4686851501464844]
|
85ca2069-832d-476a-93df-7688707c1089
|
the-yule-frisch-waugh-lovell-theorem
|
2307.00369
| null |
https://arxiv.org/abs/2307.00369v1
|
https://arxiv.org/pdf/2307.00369v1.pdf
|
The Yule-Frisch-Waugh-Lovell Theorem
|
This paper traces the historical and analytical development of what is known in the econometrics literature as the Frisch-Waugh-Lovell theorem. This theorem demonstrates that the coefficients on any subset of covariates in a multiple regression is equal to the coefficients in a regression of the residualized outcome variable on the residualized subset of covariates, where residualization uses the complement of the subset of covariates of interest. In this paper, I suggest that the theorem should be renamed as the Yule-Frisch-Waugh-Lovell (YFWL) theorem to recognize the pioneering contribution of the statistician G. Udny Yule in its development. Second, I highlight recent work by the statistician, P. Ding, which has extended the YFWL theorem to a comparison of estimated covariance matrices of coefficients from multiple and partial, i.e. residualized regressions. Third, I show that, in cases where Ding's results do not apply, one can still resort to a computational method to conduct statistical inference about coefficients in multiple regressions using information from partial regressions.
|
['Deepankar Basu']
|
2023-07-01
| null | null | null | null |
['econometrics']
|
['miscellaneous']
|
[-3.64447534e-02 7.83637837e-02 -5.21783710e-01 -4.34355289e-01
-6.09542489e-01 -3.89361203e-01 3.22085798e-01 -1.88864186e-01
-3.56109560e-01 9.27069485e-01 2.15044469e-01 -8.96323562e-01
-6.33444786e-01 -3.43848795e-01 -6.30325556e-01 -7.07791388e-01
-5.74388914e-02 -9.99064893e-02 -2.57435024e-01 -1.55404717e-01
2.96030462e-01 4.31712806e-01 -1.10842562e+00 -4.73757595e-01
8.84218752e-01 4.87992406e-01 -9.19350088e-02 6.45433068e-01
4.55710590e-01 7.54113853e-01 -5.10802805e-01 -2.21724764e-01
3.48563462e-01 -5.56728184e-01 -3.44763964e-01 -1.97466388e-01
2.91998029e-01 -2.16061175e-01 -1.44102842e-01 7.24812686e-01
-6.82508498e-02 2.01356515e-01 9.93684113e-01 -1.50890219e+00
-5.90235829e-01 4.39484894e-01 -1.05622888e+00 1.97223932e-01
2.30248764e-01 -3.68155271e-01 1.11174548e+00 -8.98064673e-01
3.81047070e-01 1.26855218e+00 8.41439664e-01 1.04090840e-01
-1.53135478e+00 -8.97421658e-01 1.70074642e-01 -3.04456174e-01
-1.37171865e+00 -3.61855984e-01 4.23498571e-01 -7.67617166e-01
5.69298148e-01 6.08669519e-01 9.69491825e-02 6.52514040e-01
4.62186068e-01 4.05454278e-01 1.41482139e+00 -4.90392804e-01
-2.21374318e-01 1.68614611e-01 6.16199076e-01 3.53357047e-01
8.08229625e-01 3.39606285e-01 -2.05609471e-01 -3.06917161e-01
1.12805855e+00 -5.39136305e-02 -7.11067617e-02 -3.17452282e-01
-9.36323583e-01 1.24575126e+00 -9.77175459e-02 3.63465965e-01
-4.70268577e-01 1.57720521e-01 6.39403909e-02 6.08865380e-01
5.09754956e-01 1.82776436e-01 -6.86513126e-01 2.96191573e-01
-1.02392387e+00 4.78046775e-01 8.66782844e-01 8.12801778e-01
5.10936081e-01 -1.39116839e-01 2.08184466e-01 5.04776716e-01
4.33801055e-01 7.54100084e-01 9.22164246e-02 -8.98883402e-01
4.97540206e-01 3.35122555e-01 3.14790010e-01 -1.11009657e+00
-7.41466522e-01 -2.53429800e-01 -7.48100042e-01 2.52017617e-01
7.91405022e-01 -5.01144886e-01 -2.48166427e-01 1.83866870e+00
9.66498330e-02 -2.57256687e-01 -6.79525882e-02 5.16214192e-01
4.64340597e-01 3.43372911e-01 2.41784036e-01 -6.83155537e-01
9.99357224e-01 -4.32199210e-01 -7.69976735e-01 2.84112934e-02
6.73401117e-01 -5.98338485e-01 5.81850946e-01 5.43826640e-01
-1.15078604e+00 -5.16883433e-01 -6.86948359e-01 3.42662819e-02
-7.14338273e-02 1.13366969e-01 9.04359162e-01 7.75344610e-01
-1.01269042e+00 5.83723485e-01 -6.83295846e-01 -3.77974689e-01
-6.94158524e-02 7.07607806e-01 -6.10976458e-01 1.69168502e-01
-9.01292682e-01 8.97224665e-01 -3.45301718e-01 3.57928462e-02
-2.41044596e-01 -8.80310297e-01 -7.45046794e-01 1.34493038e-01
4.21215236e-01 -4.70556349e-01 1.09209764e+00 -9.99172211e-01
-7.96553254e-01 7.90069461e-01 -4.78108734e-01 9.78116393e-02
4.67280596e-01 -1.49891123e-01 -4.22879219e-01 -1.83470264e-01
5.00299573e-01 -3.12882245e-01 4.21260804e-01 -1.13935471e+00
-7.32613921e-01 -6.56379163e-01 -1.12339072e-01 -2.15155229e-01
1.96804225e-01 5.60913205e-01 1.05756901e-01 -7.13059127e-01
3.39217246e-01 -8.91578972e-01 -3.40492427e-01 -6.22797310e-01
-3.80100489e-01 -4.98769134e-01 8.95858034e-02 -8.30254972e-01
1.60633469e+00 -2.18073106e+00 5.71709275e-02 6.01050794e-01
4.49442536e-01 -3.72346431e-01 5.08680046e-02 6.51185751e-01
-7.32891142e-01 3.20603460e-01 -9.84547213e-02 -1.09089017e-01
-3.50284353e-02 -6.78159371e-02 -2.15871558e-01 9.51170087e-01
1.24319997e-02 7.15571582e-01 -6.04655087e-01 -1.82447240e-01
1.04183808e-01 9.67042968e-02 -4.38126147e-01 -2.28452668e-01
7.95546651e-01 3.46955210e-01 -4.89960074e-01 4.09491837e-01
8.11751604e-01 -1.37228712e-01 3.48437995e-01 2.68759340e-01
-7.19613850e-01 1.52603224e-01 -1.25022995e+00 8.92996609e-01
-1.91310719e-01 6.84638083e-01 3.28857839e-01 -1.34803259e+00
9.31270003e-01 3.19917709e-01 6.38234019e-01 -4.51971531e-01
4.08632383e-02 4.32749838e-01 1.00822568e-01 -3.60444129e-01
3.12275469e-01 -5.13549030e-01 -3.72070938e-01 2.27245525e-01
-3.65186244e-01 7.82065690e-02 2.43530154e-01 1.09231994e-01
1.13049054e+00 1.15477003e-01 9.46763277e-01 -8.07030380e-01
4.32019740e-01 1.89067945e-02 6.51735604e-01 9.73584056e-01
2.18984783e-02 1.95247740e-01 9.58227694e-01 -6.05090596e-02
-9.78799343e-01 -1.11201513e+00 -7.82547832e-01 1.05884552e+00
-3.88399869e-01 2.93172374e-02 -3.94708663e-01 -1.48445621e-01
5.05871892e-01 5.92097342e-01 -9.53822017e-01 2.03118980e-01
-5.22463560e-01 -8.51816535e-01 3.18674296e-01 3.60409468e-01
-3.56429517e-02 -3.79388541e-01 -4.93347645e-01 1.04373500e-01
1.52531564e-01 -4.91366029e-01 -1.13047875e-01 3.90085638e-01
-1.02250326e+00 -1.35732567e+00 -7.10926354e-01 -5.07788301e-01
6.17602348e-01 3.84425879e-01 9.33465660e-01 3.88891920e-02
2.01178700e-01 3.97554845e-01 -3.37015659e-01 -4.27360386e-01
-3.03289324e-01 -2.22003594e-01 1.38706759e-01 -4.36716586e-01
7.21739531e-01 -7.09726870e-01 -3.32778543e-01 4.38614011e-01
-6.24386072e-01 -4.11137193e-01 5.20214498e-01 9.48127806e-01
9.27353948e-02 -4.92235906e-02 8.28535676e-01 -1.02537477e+00
5.29989421e-01 -7.04171538e-01 -8.30620348e-01 1.72275841e-01
-6.89550340e-01 4.41955111e-04 4.01107728e-01 -3.80579561e-01
-7.48036444e-01 -4.69119638e-01 3.42146218e-01 1.07231103e-01
2.66998634e-02 8.83499265e-01 2.40018182e-02 1.25137672e-01
4.61400390e-01 -4.86117721e-01 -9.51063931e-02 -6.38254642e-01
-6.91596791e-03 7.02196538e-01 3.38824064e-01 -3.60320985e-01
8.50778401e-01 2.71706879e-01 5.80984712e-01 -7.73121536e-01
-6.44903123e-01 -6.89654410e-01 -7.19577730e-01 1.11025954e-02
6.26117706e-01 -8.35943401e-01 -9.82478917e-01 1.14005014e-01
-7.63015389e-01 -2.46625304e-01 -1.39551923e-01 1.14472938e+00
-6.35751545e-01 3.76401916e-02 -1.58683017e-01 -1.17575884e+00
4.04736161e-01 -8.50241303e-01 5.70606947e-01 2.02779993e-02
-6.05211973e-01 -1.25188351e+00 2.28020236e-01 1.78434744e-01
-8.87238160e-02 5.09139836e-01 1.07330787e+00 -6.79434121e-01
1.19692221e-01 -4.31552351e-01 -2.08201557e-01 6.25188202e-02
9.78045464e-02 2.50790626e-01 -6.61647439e-01 -1.74161643e-01
3.12923610e-01 1.04142010e-01 7.17840612e-01 1.04664516e+00
5.61639786e-01 -3.08870494e-01 -3.30062866e-01 6.07524455e-01
1.64568424e+00 4.44729984e-01 5.24031043e-01 3.52589846e-01
3.62018108e-01 9.08783495e-01 5.99110484e-01 5.49182475e-01
4.34544891e-01 4.96026784e-01 -1.58572823e-01 -2.72990644e-01
3.90777141e-01 -2.50815094e-01 3.55442762e-01 5.64644277e-01
-2.11762786e-01 3.66273895e-02 -5.62737703e-01 5.45491517e-01
-1.86738288e+00 -1.03229475e+00 -9.18022931e-01 2.36151767e+00
5.71342587e-01 -1.75595596e-01 4.04322475e-01 8.54355618e-02
7.25853205e-01 -9.12028626e-02 -2.86198765e-01 -6.88845694e-01
-1.86336786e-01 3.26888770e-01 1.10061955e+00 7.95941651e-01
-8.44749570e-01 5.19364059e-01 7.38456345e+00 3.90984952e-01
-3.13469172e-01 -1.15017302e-01 5.30903578e-01 7.59578571e-02
-4.06191051e-01 2.95573533e-01 -7.79900789e-01 1.10164158e-01
1.15362775e+00 -6.38056934e-01 3.26141566e-01 6.37659192e-01
8.83941591e-01 -8.15858662e-01 -1.14286304e+00 3.79330277e-01
-1.07923813e-01 -3.59602898e-01 -7.71858931e-01 5.62140465e-01
8.40739071e-01 -4.76546168e-01 3.34092617e-01 3.54847193e-01
7.10798442e-01 -1.29663253e+00 3.41316879e-01 5.53944528e-01
7.53329992e-01 -7.05466270e-01 5.94609916e-01 4.10065025e-01
-1.03892648e+00 -2.24623919e-01 -5.33228397e-01 -7.26145685e-01
-9.17170793e-02 7.79235840e-01 -4.05313998e-01 6.21121764e-01
3.43523234e-01 6.70709014e-01 -2.22021759e-01 6.98020637e-01
-6.15251176e-02 6.63973689e-01 -2.14711711e-01 2.84845442e-01
1.25318468e-01 -7.01942265e-01 1.95538774e-01 8.41308653e-01
3.95905644e-01 4.44428504e-01 -4.15707231e-01 6.97637141e-01
3.07489544e-01 3.13674331e-01 -7.55608201e-01 1.08404262e-02
2.42408827e-01 8.55269134e-01 -5.11569560e-01 -1.32677376e-01
-1.13450968e+00 2.03857869e-01 2.69315168e-02 5.06314397e-01
-4.82207894e-01 -3.19815993e-01 6.83820665e-01 2.27003381e-01
1.09671704e-01 -1.64925769e-01 -8.40805352e-01 -1.02178884e+00
-1.18888550e-01 -7.20007479e-01 3.66342783e-01 -6.14574492e-01
-1.08132994e+00 -1.78047985e-01 5.40766776e-01 -7.85120666e-01
-3.69182736e-01 -7.09166646e-01 -4.63191837e-01 1.57002926e+00
-8.94437611e-01 -7.83950150e-01 2.68089712e-01 4.45596367e-01
8.98202509e-02 3.26186530e-02 4.90403831e-01 1.17524236e-01
-7.13638961e-01 5.62452912e-01 5.33359110e-01 1.59154087e-03
6.59393191e-01 -1.21918488e+00 1.07300691e-01 8.59966516e-01
-3.25515926e-01 1.02298725e+00 1.06991875e+00 -8.75152826e-01
-1.20171118e+00 -3.80105913e-01 1.25466990e+00 -5.77550173e-01
1.08439243e+00 4.20574546e-02 -5.53153872e-01 1.24844766e+00
-7.94049874e-02 -5.38287997e-01 7.72268176e-01 6.40623629e-01
-1.35760382e-01 -2.29295883e-02 -9.98320580e-01 5.11570930e-01
5.79947531e-01 -2.63687640e-01 -6.23377740e-01 -1.78758595e-02
3.66477162e-01 1.56396136e-01 -1.20444274e+00 4.47659522e-01
1.09955955e+00 -9.81158137e-01 7.63694286e-01 -7.86127388e-01
3.91630381e-01 1.05413623e-01 -3.02426279e-01 -9.71198797e-01
-5.76649964e-01 -4.78909761e-01 7.13802278e-01 1.08850479e+00
5.97497940e-01 -8.47489953e-01 4.40805018e-01 9.96827841e-01
3.53780799e-02 -5.62033117e-01 -9.98035729e-01 -8.90951931e-01
8.00465226e-01 -3.26557279e-01 3.32709730e-01 1.15341675e+00
3.72853488e-01 1.10964097e-01 -4.87280220e-01 -7.56307924e-03
6.88322186e-01 -1.31523997e-01 1.02171063e+00 -1.74994898e+00
-2.54919440e-01 -3.12752932e-01 -2.25047618e-01 -8.58969092e-01
2.65423238e-01 -6.45006835e-01 -2.83986926e-01 -1.33440638e+00
4.63628978e-01 -4.74906117e-01 -1.61004871e-01 1.77704185e-01
-1.95596561e-01 -2.19081819e-01 1.03457637e-01 3.87705833e-01
2.62134343e-01 -1.20794021e-01 1.05507076e+00 4.66769934e-01
-4.61689770e-01 4.03492838e-01 -1.29398644e+00 7.12955534e-01
7.22619116e-01 -7.97413051e-01 -1.77704558e-01 9.72788110e-02
4.68345821e-01 7.86375463e-01 4.85082090e-01 -2.76350021e-01
-3.72478832e-03 -7.30590403e-01 3.81375462e-01 -5.24996519e-01
-5.74507937e-02 -8.57011974e-01 4.76530731e-01 3.82023036e-01
-3.78126204e-01 5.47878921e-01 7.56092891e-02 2.01462656e-01
6.73023239e-02 -3.86411905e-01 5.48318565e-01 -1.19630955e-02
-8.01105574e-02 -2.20129952e-01 -4.55759972e-01 6.24585450e-02
1.06488478e+00 -1.42899245e-01 -2.82204896e-01 -4.74040508e-01
-7.02863157e-01 1.87095091e-01 4.30826128e-01 -8.54779258e-02
2.65186787e-01 -1.15283370e+00 -7.95099735e-01 2.35502124e-01
-2.84368068e-01 -6.69068336e-01 9.98614728e-02 1.55422318e+00
-1.47074103e-01 9.62245703e-01 1.14124874e-03 1.93627421e-02
-1.30220485e+00 6.88295364e-01 4.14330661e-02 -2.89642394e-01
-5.25152624e-01 6.16510928e-01 7.43236125e-01 3.11735924e-02
-3.49370152e-01 -3.33103120e-01 -1.25876173e-01 2.00687915e-01
4.22770292e-01 9.05568600e-01 -4.16915476e-01 -9.31435108e-01
-4.22141194e-01 6.17068231e-01 1.97250187e-01 -4.73688751e-01
1.44795549e+00 -4.53453958e-01 -3.43434662e-01 1.07607067e+00
1.26621044e+00 6.81479335e-01 -7.77131021e-01 -1.77384000e-02
1.07308358e-01 -4.45409030e-01 -5.79834059e-02 -5.61244667e-01
-6.42789960e-01 4.46229219e-01 3.57936919e-02 4.08008635e-01
1.13516831e+00 -5.15206670e-03 -2.66685098e-01 6.76231235e-02
1.85286209e-01 -8.49406421e-01 -8.00040603e-01 4.11880836e-02
8.17735910e-01 -1.03168523e+00 4.63492244e-01 -4.02700305e-01
-7.29112685e-01 9.22613025e-01 -2.93799695e-02 -4.77469385e-01
9.61423457e-01 2.04227224e-01 -1.88264810e-02 -1.18324734e-01
-6.78895533e-01 -1.13681421e-01 1.86171576e-01 4.58391368e-01
7.95164227e-01 3.41291904e-01 -1.07973409e+00 7.50257492e-01
-4.50415492e-01 -1.00955248e-01 8.58701646e-01 5.23213208e-01
-3.60611111e-01 -1.16276670e+00 -7.62251019e-01 8.69441211e-01
-6.27174556e-01 -1.40947834e-01 -2.09046587e-01 1.64307249e+00
-1.58268616e-01 1.42265308e+00 7.83248097e-02 -1.45185545e-01
3.31674397e-01 6.85492754e-02 3.65988463e-01 -4.57446545e-01
-5.39532125e-01 5.32584012e-01 -4.04936895e-02 -3.70641142e-01
-6.58976912e-01 -1.35640156e+00 -7.12389469e-01 -7.76076972e-01
-4.75648910e-01 5.65542758e-01 4.15258884e-01 1.05806327e+00
-5.52734852e-01 3.11002970e-01 9.38822150e-01 -2.06093028e-01
-6.79020584e-01 -1.16943622e+00 -1.52283049e+00 2.85952934e-03
6.76790714e-01 -8.35194707e-01 -9.36290145e-01 -2.05338672e-01]
|
[7.816800594329834, 5.083799839019775]
|
e34566fb-c7aa-4685-a137-de82a7ec01a8
|
dude-dual-decoder-multilingual-asr-for-indian
|
2210.16739
| null |
https://arxiv.org/abs/2210.16739v1
|
https://arxiv.org/pdf/2210.16739v1.pdf
|
DuDe: Dual-Decoder Multilingual ASR for Indian Languages using Common Label Set
|
In a multilingual country like India, multilingual Automatic Speech Recognition (ASR) systems have much scope. Multilingual ASR systems exhibit many advantages like scalability, maintainability, and improved performance over the monolingual ASR systems. However, building multilingual systems for Indian languages is challenging since different languages use different scripts for writing. On the other hand, Indian languages share a lot of common sounds. Common Label Set (CLS) exploits this idea and maps graphemes of various languages with similar sounds to common labels. Since Indian languages are mostly phonetic, building a parser to convert from native script to CLS is easy. In this paper, we explore various approaches to build multilingual ASR models. We also propose a novel architecture called Encoder-Decoder-Decoder for building multilingual systems that use both CLS and native script labels. We also analyzed the effectiveness of CLS-based multilingual systems combined with machine transliteration.
|
['Umesh S', 'Mudit Batra', 'Arunkumar A']
|
2022-10-30
| null | null | null | null |
['transliteration']
|
['natural-language-processing']
|
[-3.00655216e-01 -4.57604825e-01 -2.17556283e-01 -5.16780257e-01
-1.16486669e+00 -8.88646960e-01 5.51216364e-01 -1.75816178e-01
-3.27066302e-01 6.56913280e-01 3.01976621e-01 -7.85666585e-01
5.05448997e-01 -4.66502160e-01 -6.03514135e-01 -2.07549453e-01
6.50839567e-01 4.45112884e-01 1.88488260e-01 -4.90114689e-01
-1.91043806e-03 4.10405904e-01 -8.89217854e-01 3.24571073e-01
1.08435035e+00 1.89512998e-01 8.00875843e-01 7.11539805e-01
-6.23499453e-01 1.04003143e+00 -7.32230604e-01 -4.81703103e-01
1.96042910e-01 -5.32356083e-01 -6.50878072e-01 -1.30321935e-01
1.52829915e-01 1.81510746e-02 -5.45311570e-01 1.28630948e+00
5.17412841e-01 -8.16477016e-02 3.72016191e-01 -7.11593091e-01
-1.14387918e+00 1.28842902e+00 -3.88454109e-01 -6.90265233e-03
6.46922410e-01 -2.87233531e-01 8.89726877e-01 -7.60721982e-01
5.39648354e-01 1.45754206e+00 4.26096976e-01 5.40262163e-01
-9.16064024e-01 -6.80688262e-01 1.19875565e-01 1.25308737e-01
-1.83479702e+00 -7.13082969e-01 6.20211661e-01 -2.94024229e-01
1.09698427e+00 3.32602590e-01 1.97165936e-01 8.69979322e-01
-3.32998000e-02 7.34184206e-01 1.54777896e+00 -7.62766719e-01
-2.47718275e-01 6.77074015e-01 9.77756754e-02 6.19855642e-01
-1.57643974e-01 -3.27079773e-01 -4.03867781e-01 3.35968584e-01
6.09473526e-01 -1.83212683e-01 -8.37588161e-02 4.74486262e-01
-1.29748356e+00 6.37933791e-01 -1.58280149e-01 7.26600051e-01
-6.04578033e-02 -2.92506844e-01 5.96763551e-01 5.16095757e-01
4.82266173e-02 2.97876090e-01 -4.65161055e-01 -1.87561974e-01
-8.34609509e-01 -5.47087491e-01 7.09527433e-01 1.34013629e+00
7.38606691e-01 4.97981697e-01 3.62996012e-01 1.74581444e+00
4.93421346e-01 1.14994991e+00 9.10617352e-01 -6.11383617e-01
6.16048098e-01 3.17461729e-01 -4.05183405e-01 -5.36472559e-01
3.22059505e-02 -1.79510340e-01 -4.82550412e-01 -3.52961183e-01
-8.43012631e-02 -3.13741639e-02 -1.00089681e+00 1.47271383e+00
-1.34868518e-01 -3.48124474e-01 6.54902458e-01 5.59036255e-01
8.43488336e-01 1.16182804e+00 -3.36987078e-02 -1.22388810e-01
1.29824412e+00 -1.19159913e+00 -9.49058414e-01 -1.84610352e-01
6.68979526e-01 -1.40432167e+00 1.42617941e+00 7.46073052e-02
-8.83279920e-01 -6.29498839e-01 -8.75901937e-01 -3.41979176e-01
-5.64218760e-01 5.77834547e-01 2.93611258e-01 9.37598288e-01
-1.30815744e+00 -1.04266122e-01 -5.96232057e-01 -6.44483566e-01
-5.84060371e-01 3.71126123e-02 -4.17305708e-01 -9.17954668e-02
-1.35478008e+00 1.06260514e+00 4.17754740e-01 -9.28204507e-02
-9.22612667e-01 -9.99093205e-02 -1.00012445e+00 -3.44927043e-01
3.62512879e-02 2.43192837e-01 1.34930182e+00 -9.84372854e-01
-1.95042205e+00 1.04593682e+00 -3.67710143e-01 -2.50761537e-03
6.15404621e-02 -4.23648879e-02 -1.11444223e+00 -3.20755512e-01
1.73401594e-01 3.24590385e-01 2.80286551e-01 -1.08719063e+00
-9.04847443e-01 -2.86173344e-01 -1.04917385e-01 3.34212065e-01
-2.14464352e-01 7.76168585e-01 -7.36366332e-01 -7.80999959e-01
2.28642344e-01 -1.05423939e+00 4.86510396e-02 -9.19659972e-01
-4.51104999e-01 -1.41913176e-01 5.34813404e-01 -1.33104026e+00
1.51184404e+00 -2.26876473e+00 -1.32975772e-01 1.24291785e-01
-5.43843627e-01 4.48431969e-01 -1.81336805e-01 5.77761114e-01
1.32673636e-01 2.19265088e-01 -1.15840510e-01 -1.12698991e-02
3.58631909e-02 5.79378784e-01 -3.33036780e-01 5.60468622e-02
-7.90616050e-02 6.24142706e-01 -7.95799851e-01 -5.34263670e-01
3.95894647e-01 5.36290109e-01 -7.26521946e-03 2.01829404e-01
1.99319705e-01 5.52999258e-01 -1.37436807e-01 8.73596966e-01
4.18859243e-01 4.33256149e-01 5.45646966e-01 4.23191227e-02
-6.00458503e-01 8.92891467e-01 -1.17862463e+00 1.59694064e+00
-1.05968690e+00 5.79135597e-01 3.31897624e-02 -7.31363237e-01
1.38506508e+00 7.19650805e-01 -4.30876762e-02 -8.00776541e-01
-1.05311029e-01 9.48347270e-01 8.26701447e-02 -4.45621222e-01
7.34572232e-01 -1.02602571e-01 -2.40257740e-01 2.88300484e-01
2.63344765e-01 -3.61993670e-01 2.69846082e-01 3.70770395e-02
6.62466168e-01 1.14025772e-01 6.37817860e-01 -2.65472561e-01
8.82298708e-01 -4.83576283e-02 6.14377618e-01 5.32766521e-01
-6.44077286e-02 3.91930372e-01 -1.16959587e-01 -8.96916017e-02
-1.07937074e+00 -1.12819278e+00 -1.73257366e-01 1.25693870e+00
-2.55657583e-01 -5.05810499e-01 -6.18281305e-01 -4.11976248e-01
-4.24737632e-01 8.26843083e-01 2.38850832e-01 3.47475767e-01
-9.21076894e-01 -2.91738391e-01 1.13164604e+00 3.86100799e-01
3.13754290e-01 -1.17147076e+00 5.97685635e-01 4.57052618e-01
-3.59699547e-01 -1.47918785e+00 -8.07374775e-01 1.89113364e-01
-3.48324120e-01 -6.20751202e-01 -7.39997029e-01 -1.35587728e+00
3.77383977e-01 4.50479716e-01 8.40767741e-01 -3.37853760e-01
2.37575307e-01 1.84739158e-01 -5.59209585e-01 -2.49870121e-01
-1.18525124e+00 1.96309835e-01 2.96299815e-01 -1.43440410e-01
5.54223001e-01 -3.28055561e-01 3.31838399e-01 3.26341718e-01
-7.09989727e-01 -1.45532474e-01 4.73475277e-01 2.90997088e-01
7.12186337e-01 -2.32111588e-01 6.32041991e-01 -8.64743948e-01
5.02721906e-01 -3.43374044e-01 -6.31998658e-01 7.60259569e-01
-2.69906253e-01 1.42327026e-01 1.11246943e+00 -3.20354968e-01
-1.20881331e+00 1.11319594e-01 -7.45732903e-01 -1.14127941e-01
-2.50129461e-01 6.42917991e-01 -5.29254436e-01 1.78946145e-02
4.45473641e-01 6.00393951e-01 -4.48900312e-01 -8.08456123e-01
5.03101766e-01 1.52101338e+00 6.61844134e-01 -7.72224128e-01
5.34890652e-01 -2.18829885e-01 -6.87248290e-01 -1.27045631e+00
-4.72305626e-01 -6.22787237e-01 -6.04833424e-01 -1.80048525e-01
7.81073451e-01 -1.19383395e+00 -2.47763753e-01 5.31473398e-01
-1.32019448e+00 -1.20208398e-01 -5.25905862e-02 8.04891706e-01
-2.83513010e-01 3.62586260e-01 -1.00335324e+00 -6.56563222e-01
-1.76459506e-01 -1.55468798e+00 8.26096773e-01 1.06780753e-01
-1.48130476e-01 -1.13025951e+00 2.73368448e-01 3.48962009e-01
3.82773161e-01 -5.69613636e-01 7.86910772e-01 -6.87744379e-01
-2.41334930e-01 8.59769732e-02 -1.10545650e-01 7.07251668e-01
6.53705835e-01 1.90306902e-01 -8.53037834e-01 -1.16059467e-01
-1.60380259e-01 -2.21332088e-01 2.71883667e-01 1.49936333e-01
5.23494005e-01 -2.02466682e-01 3.93011011e-02 5.47664821e-01
1.32676566e+00 7.47831643e-01 5.00155210e-01 4.13754523e-01
1.06986451e+00 5.00677884e-01 3.48847628e-01 -8.51245672e-02
6.76012754e-01 7.24357307e-01 -5.60167015e-01 -1.15161780e-02
-4.31530684e-01 -4.67398345e-01 1.18896973e+00 2.26639962e+00
1.91966355e-01 3.01157758e-02 -1.14133167e+00 6.82634950e-01
-1.42457318e+00 -5.41917443e-01 -2.79677093e-01 2.28091669e+00
1.40338314e+00 -3.49314749e-01 -1.01182736e-01 -1.27809569e-01
1.10213566e+00 6.32941797e-02 6.69321269e-02 -9.54378068e-01
-6.31619275e-01 1.65298209e-01 5.95626712e-01 9.65160429e-01
-7.02985823e-01 1.65343893e+00 6.52034664e+00 1.06097567e+00
-1.40366542e+00 3.80254984e-01 1.56878173e-01 5.19925833e-01
-4.83627468e-01 2.01379806e-01 -1.20988965e+00 4.53319877e-01
1.31316686e+00 -1.54685408e-01 7.35521436e-01 7.76608706e-01
2.07727656e-01 3.42822492e-01 -8.12117338e-01 1.14217544e+00
2.62367964e-01 -8.07773769e-01 2.92282492e-01 -4.32215869e-01
9.43126261e-01 4.79837716e-01 -2.84780681e-01 3.66339803e-01
8.91656935e-01 -8.18454444e-01 9.42644060e-01 8.02782923e-02
1.16487050e+00 -6.51602745e-01 5.12157321e-01 1.84526250e-01
-1.58861661e+00 3.79739851e-01 -5.83152652e-01 3.05164576e-01
3.33645225e-01 2.19493166e-01 -8.03226948e-01 7.11061001e-01
3.16830337e-01 6.84356511e-01 -5.48308015e-01 6.55146003e-01
-5.78932762e-01 9.43051934e-01 -2.68283665e-01 1.54793665e-01
3.38302702e-01 -4.94333565e-01 4.99790817e-01 1.83010817e+00
6.43429279e-01 -4.51175183e-01 5.77615976e-01 1.34379342e-01
-2.13767290e-02 8.48619044e-01 -8.25458884e-01 -3.34993482e-01
7.48654842e-01 8.56039405e-01 -5.59213698e-01 -4.93907481e-01
-7.64518619e-01 9.50202584e-01 2.25678071e-01 3.38323504e-01
-5.79007149e-01 -4.37310517e-01 4.02758628e-01 -1.89452723e-01
-2.36695170e-01 -5.26087165e-01 -6.45205677e-02 -1.25659382e+00
-9.50444415e-02 -1.44611633e+00 1.55550495e-01 -6.56953692e-01
-1.09354055e+00 1.07279718e+00 -5.58361411e-01 -1.10617423e+00
-1.20657474e-01 -5.28409123e-01 -1.23155408e-01 9.53446150e-01
-1.70660412e+00 -1.40384471e+00 2.91678190e-01 7.79874206e-01
9.42219377e-01 -6.50652051e-01 1.04385364e+00 7.67496467e-01
-5.29810250e-01 6.82777226e-01 4.00151968e-01 4.28400517e-01
9.93759692e-01 -1.26548481e+00 4.04347688e-01 1.11056316e+00
5.69523335e-01 7.67831504e-01 2.48432457e-01 -7.41566300e-01
-1.25102162e+00 -1.17282748e+00 1.35395551e+00 -1.49697825e-01
1.04086149e+00 -2.90159523e-01 -8.95007253e-01 1.07884240e+00
6.30972087e-01 -4.84233439e-01 7.45860219e-01 2.08360761e-01
-5.68021178e-01 -3.23297083e-01 -5.14913201e-01 6.64687157e-01
6.78270042e-01 -1.26403141e+00 -4.94054973e-01 3.83312196e-01
8.39154005e-01 -1.74193576e-01 -9.96750116e-01 2.82033347e-02
4.55884933e-01 -4.28146809e-01 5.13013124e-01 -1.29276678e-01
3.22070643e-02 -7.20209360e-01 -7.91918278e-01 -1.56492651e+00
-1.96488068e-01 -7.60811448e-01 5.92976093e-01 1.64078188e+00
6.69075429e-01 -7.79900372e-01 -2.40399629e-01 -6.96377158e-02
-4.33736026e-01 1.92565084e-01 -6.66430950e-01 -1.21672189e+00
-1.86217595e-02 -5.69409370e-01 6.36699021e-01 1.34632468e+00
7.41982460e-02 5.94605923e-01 -5.89220285e-01 4.68704671e-01
1.98054463e-01 -2.68123478e-01 4.32898760e-01 -7.90316999e-01
-2.12997451e-01 -2.04692304e-01 -2.76592165e-01 -8.45811605e-01
3.59626293e-01 -1.37700844e+00 1.99683130e-01 -1.39089203e+00
9.56204683e-02 -7.82153964e-01 -4.21539515e-01 6.87140644e-01
-8.26549828e-02 1.25799492e-01 3.42168570e-01 2.55038381e-01
-5.04908383e-01 2.38144934e-01 9.18345213e-01 -1.35477737e-01
-3.61619532e-01 -3.19048107e-01 -4.08956647e-01 5.78166366e-01
1.01519167e+00 -5.72036803e-01 -1.66769341e-01 -9.06736672e-01
6.93995273e-03 2.98766792e-01 -5.67232490e-01 -1.02234030e+00
2.98480570e-01 -2.38556385e-01 -3.10705602e-01 -3.70767951e-01
2.75243260e-02 -6.63347185e-01 3.66935700e-01 1.43578157e-01
-2.11235851e-01 4.37560439e-01 1.22741409e-01 -2.90187627e-01
-7.09344566e-01 -3.87875825e-01 9.31677043e-01 -2.10036576e-01
-1.03330731e+00 -1.17071383e-01 -8.54471922e-01 1.03428088e-01
7.92727649e-01 3.16111334e-02 -1.61418065e-01 -2.55941421e-01
-4.81305420e-01 -1.20265849e-01 3.89555603e-01 8.07342112e-01
4.13204402e-01 -1.29162848e+00 -8.53885531e-01 4.27424729e-01
1.89245731e-01 -6.75785899e-01 -8.95261541e-02 4.02672082e-01
-8.72826636e-01 8.01217437e-01 -2.72900671e-01 -3.15065026e-01
-1.30655861e+00 3.72354090e-01 2.47451380e-01 -1.26030326e-01
-2.69332319e-01 6.30600870e-01 1.08673275e-01 -1.31733525e+00
1.66992232e-01 2.58499607e-02 -3.32805783e-01 -2.23572850e-01
3.79138261e-01 3.56475025e-01 9.42278877e-02 -1.22440696e+00
-4.74168271e-01 6.58963084e-01 -1.77773505e-01 -5.21991730e-01
8.64312470e-01 -4.78137344e-01 -4.70844537e-01 8.59604895e-01
9.97422397e-01 8.51899862e-01 -2.68017441e-01 -4.13935959e-01
2.26361305e-01 -2.14461640e-01 -1.60174191e-01 -7.50070453e-01
-8.67548823e-01 7.84056127e-01 3.52088809e-01 -3.34788077e-02
8.17481458e-01 -5.03804460e-02 8.23782265e-01 6.26493454e-01
5.91577768e-01 -1.51554763e+00 -5.18706739e-01 1.14614630e+00
6.67932153e-01 -1.23076296e+00 -6.39082253e-01 -2.20014036e-01
-9.22682464e-01 1.04325104e+00 4.75343883e-01 2.42896438e-01
5.92663944e-01 5.46453118e-01 8.27991724e-01 2.42319912e-01
-4.07243192e-01 -2.40970314e-01 5.88277653e-02 5.66969872e-01
9.41172421e-01 7.03913033e-01 -6.55514538e-01 4.09678876e-01
-6.03904307e-01 -3.59203994e-01 7.11242020e-01 7.67188907e-01
-5.24940014e-01 -1.71920502e+00 -5.91835976e-01 -7.42409676e-02
-7.09889770e-01 -4.47658241e-01 -2.38817006e-01 5.28390229e-01
2.18331926e-02 1.24382663e+00 -3.22480768e-01 -4.54758883e-01
1.56776071e-01 3.65563840e-01 1.53927505e-01 -8.83111298e-01
-4.15414065e-01 3.46791178e-01 1.97475806e-01 -1.93504333e-01
-3.35262328e-01 -5.29975593e-01 -1.18293631e+00 -2.58717090e-01
-2.51193106e-01 3.64011824e-01 1.02835405e+00 9.13971722e-01
-1.22524105e-01 4.64475155e-01 8.94415975e-01 -5.43646254e-02
-1.92775682e-01 -9.27086711e-01 -7.93907464e-01 -5.92815876e-02
-4.32286076e-02 5.21292649e-02 -4.80020829e-02 2.99815804e-01]
|
[14.336803436279297, 7.033246994018555]
|
7757203a-35c2-4622-8c78-00ce6297d58b
|
bridging-distributional-and-risk-sensitive
|
2210.14051
| null |
https://arxiv.org/abs/2210.14051v2
|
https://arxiv.org/pdf/2210.14051v2.pdf
|
Bridging Distributional and Risk-sensitive Reinforcement Learning with Provable Regret Bounds
|
We study the regret guarantee for risk-sensitive reinforcement learning (RSRL) via distributional reinforcement learning (DRL) methods. In particular, we consider finite episodic Markov decision processes whose objective is the entropic risk measure (EntRM) of return. We identify a key property of the EntRM, the monotonicity-preserving property, which enables the risk-sensitive distributional dynamic programming framework. We then propose two novel DRL algorithms that implement optimism through two different schemes, including a model-free one and a model-based one. We prove that both of them attain $\tilde{\mathcal{O}}(\frac{\exp(|\beta| H)-1}{|\beta|H}H\sqrt{HS^2AT})$ regret upper bound, where $S$ is the number of states, $A$ the number of states, $H$ the time horizon and $T$ the number of total time steps. It matches RSVI2 proposed in \cite{fei2021exponential} with a much simpler regret analysis. To the best of our knowledge, this is the first regret analysis of DRL, which bridges DRL and RSRL in terms of sample complexity. Finally, we improve the existing lower bound by proving a tighter bound of $\Omega(\frac{\exp(\beta H/6)-1}{\beta H}H\sqrt{SAT})$ for $\beta>0$ case, which recovers the tight lower bound $\Omega(H\sqrt{SAT})$ in the risk-neutral setting.
|
['Zhi-Quan Luo', 'Hao Liang']
|
2022-10-25
| null | null | null | null |
['distributional-reinforcement-learning']
|
['methodology']
|
[ 1.42230064e-01 3.68379027e-01 -1.89358994e-01 -2.09874973e-01
-1.15100658e+00 -5.74147105e-01 -1.04448915e-01 2.91847676e-01
-1.04002130e+00 1.15406692e+00 -2.56055534e-01 -6.38677001e-01
-7.81431377e-01 -9.45390284e-01 -9.45301414e-01 -1.12450039e+00
-5.68546355e-01 2.30820671e-01 9.11225379e-02 -2.87419558e-01
2.80370533e-01 1.83227539e-01 -1.50540066e+00 -4.80181575e-01
8.46478701e-01 1.65351617e+00 -1.04618132e-01 4.82995272e-01
2.94599563e-01 7.59827733e-01 -3.79944980e-01 -4.86341685e-01
4.32464153e-01 -6.81509197e-01 -7.26436496e-01 -5.56099892e-01
-4.05792862e-01 -2.72276133e-01 -4.29235548e-01 1.38557255e+00
5.89869082e-01 4.40915227e-01 5.69461644e-01 -1.07552278e+00
-3.29719484e-02 6.43631220e-01 -7.80536175e-01 3.74959946e-01
4.60154526e-02 1.05758654e-02 1.02007604e+00 7.46093988e-02
2.73931742e-01 9.86688316e-01 3.95185798e-01 4.84114289e-01
-9.95668113e-01 -9.53567266e-01 3.37150633e-01 -2.15830937e-01
-1.01806748e+00 -2.57585477e-02 3.48316371e-01 -3.36207062e-01
7.56922901e-01 4.32074040e-01 4.78963196e-01 6.42842829e-01
4.62371558e-01 7.53533781e-01 1.44852126e+00 -4.97250110e-01
8.09367895e-01 -1.39697149e-01 5.08677308e-03 7.23136365e-01
2.95149356e-01 6.12659156e-01 -2.28722811e-01 -4.73143943e-02
7.45100200e-01 8.92810524e-02 -1.81672201e-01 -1.37983367e-01
-6.92310750e-01 1.08283436e+00 -8.34373087e-02 -1.68538722e-03
-2.84631938e-01 3.77863288e-01 2.33667418e-01 5.67099810e-01
2.18135014e-01 1.55940950e-01 -4.62335408e-01 -4.98181075e-01
-7.19830990e-01 3.93658489e-01 8.33423793e-01 1.13497925e+00
5.21836817e-01 4.12388779e-02 -3.20223540e-01 3.44064474e-01
1.76084325e-01 9.36265349e-01 1.52657434e-01 -1.16178775e+00
7.25454211e-01 1.01896137e-01 5.03468156e-01 -3.61904800e-01
-4.15750980e-01 -4.98019844e-01 -7.11713433e-01 1.82062268e-01
5.32279611e-01 -5.39732277e-01 -3.63647670e-01 2.20539713e+00
-9.36627574e-03 -4.50464368e-01 -5.21750562e-02 5.29841602e-01
-2.98745155e-01 6.03515685e-01 7.59930629e-03 -1.00277269e+00
9.05488968e-01 -2.71908343e-01 -3.92121077e-01 1.09644150e-02
4.07422811e-01 -3.02992165e-01 1.16740727e+00 4.43182409e-01
-1.39310336e+00 9.94350836e-02 -1.08661997e+00 7.78502703e-01
1.24770962e-01 -6.03675902e-01 6.33346319e-01 1.08823359e+00
-7.00183392e-01 8.18879783e-01 -8.10941517e-01 2.77181596e-01
2.72568703e-01 4.26079988e-01 1.25027269e-01 1.07566208e-01
-1.29669106e+00 3.93767238e-01 2.70341098e-01 -1.31050527e-01
-1.26899016e+00 -5.56847990e-01 -5.75329006e-01 1.20092005e-01
8.48021626e-01 -9.34044272e-02 1.30604875e+00 -4.64683861e-01
-1.66366982e+00 4.53400642e-01 3.14907670e-01 -8.17056358e-01
8.17045927e-01 1.31383138e-02 -1.71918496e-01 1.85566247e-01
-1.19227273e-02 3.05361487e-03 5.70461690e-01 -7.80652165e-01
-7.62803078e-01 -7.34576225e-01 4.06926811e-01 1.15024671e-01
-6.23067878e-02 -4.54616211e-02 1.76841915e-01 -4.72188324e-01
-1.39449343e-01 -9.56841707e-01 -4.12780702e-01 -6.08109534e-01
-1.00730538e-01 9.07961801e-02 -9.80728492e-02 -3.59511912e-01
1.48872876e+00 -2.05493355e+00 1.58229560e-01 4.94624883e-01
-2.02421576e-01 -2.74910443e-02 4.42279726e-01 4.80412424e-01
1.38887867e-01 2.06873089e-01 -4.58808482e-01 -6.48351014e-02
5.13639569e-01 8.25063661e-02 -4.37347502e-01 5.78344882e-01
-4.27537620e-01 4.98427659e-01 -8.22983742e-01 -1.47380561e-01
-8.42947811e-02 5.67608848e-02 -4.33721811e-01 8.48329589e-02
-3.00709844e-01 1.61183193e-01 -6.36501968e-01 4.17676449e-01
5.70324957e-01 3.33379321e-02 2.41866261e-01 3.67479354e-01
-3.74674022e-01 2.58069634e-02 -1.50528944e+00 1.39741802e+00
-1.73276424e-01 -3.48248959e-01 1.29674554e-01 -1.13991261e+00
6.37848258e-01 6.18012175e-02 6.61228955e-01 -1.00172460e+00
3.75751853e-01 1.54337823e-01 -3.61536860e-01 -1.04147084e-01
1.04097113e-01 -9.06395793e-01 -6.03793979e-01 7.09941983e-01
-1.31968871e-01 1.23440795e-01 9.68090743e-02 -1.28820464e-01
1.29544008e+00 2.59943038e-01 2.18763158e-01 -6.02885246e-01
1.51348710e-01 -5.76780856e-01 9.03455615e-01 1.08027995e+00
-6.64827347e-01 -1.14727952e-01 1.26477015e+00 -7.49830008e-02
-8.71891856e-01 -1.30237198e+00 -1.25288889e-01 1.17098844e+00
3.80805194e-01 -7.74351060e-02 -7.83144772e-01 -6.06673002e-01
1.01170607e-01 1.08933008e+00 -8.67732406e-01 -4.65976119e-01
-1.78618670e-01 -1.05394292e+00 6.22716963e-01 5.62217116e-01
6.37928605e-01 -1.04454148e+00 -1.17402267e+00 -2.34901980e-02
1.49352565e-01 -3.76115948e-01 -3.93796951e-01 8.03090751e-01
-7.29474843e-01 -6.42676950e-01 -6.03503525e-01 -7.83631802e-02
3.52395862e-01 -3.68344337e-01 5.84806561e-01 -6.70169234e-01
-9.81976911e-02 4.47335154e-01 -2.05738857e-01 -6.48150921e-01
4.23876010e-02 -1.90737143e-01 2.94045269e-01 -2.39293963e-01
1.06543109e-01 -5.18522799e-01 -9.91407216e-01 1.60653576e-01
-9.05456722e-01 -4.39088404e-01 5.91678858e-01 7.23796666e-01
1.03263175e+00 3.34788829e-01 5.82081914e-01 -7.21832633e-01
3.91364366e-01 -2.73281544e-01 -1.24445546e+00 3.20366770e-01
-9.33936238e-01 5.42707384e-01 5.96354127e-01 -7.57617429e-02
-1.02353430e+00 -2.42978469e-01 -1.34635884e-02 -1.93012938e-01
2.29544654e-01 2.84755439e-01 -1.61026791e-01 2.37258434e-01
3.65112275e-01 4.68130469e-01 -7.23231584e-02 -3.71444076e-01
5.11690974e-02 1.94194540e-01 2.02965319e-01 -9.40080285e-01
4.92321908e-01 3.60521555e-01 3.82317156e-01 -2.31123358e-01
-1.01135015e+00 8.27887431e-02 7.23297819e-02 -3.27687599e-02
6.44981325e-01 -5.89048684e-01 -1.71893585e+00 -5.94165288e-02
-1.67793155e-01 -6.07631028e-01 -6.54101849e-01 6.42670214e-01
-1.28994632e+00 3.14629376e-01 -5.91832697e-01 -1.67126930e+00
-3.21274370e-01 -9.60089266e-01 5.36293864e-01 2.35345125e-01
4.05314833e-01 -6.05774879e-01 4.66541089e-02 3.49525005e-01
3.12362820e-01 4.72542495e-01 1.00017309e+00 -4.96179134e-01
-4.10498559e-01 -1.05838716e-01 6.75645620e-02 5.53749621e-01
-3.39133650e-01 -6.89069569e-01 -5.64053237e-01 -7.10233450e-01
3.57377410e-01 -3.53801697e-01 8.57380807e-01 3.39255154e-01
1.24688363e+00 -7.58913934e-01 1.18294522e-01 4.20026213e-01
1.67105520e+00 8.07880282e-01 8.24764132e-01 4.41992164e-01
-4.87677455e-02 2.08739340e-01 8.11579287e-01 1.18301833e+00
1.56425178e-01 2.50941694e-01 5.95425308e-01 4.90414172e-01
7.25821137e-01 -1.69095814e-01 7.65319109e-01 3.26859593e-01
-1.70128897e-01 -1.17413208e-01 -6.05305135e-01 4.03628588e-01
-1.73393106e+00 -1.05851030e+00 5.74101150e-01 3.06411672e+00
1.04794562e+00 5.15639484e-01 4.30459917e-01 9.94016677e-02
6.09641910e-01 -1.00872166e-01 -6.86289847e-01 -7.09596455e-01
1.21982060e-02 7.56489515e-01 1.01601744e+00 6.98990643e-01
-8.83202851e-01 6.40931070e-01 4.82583952e+00 1.15456605e+00
-7.01085091e-01 1.66910701e-02 7.13774502e-01 -6.59773409e-01
-2.32822612e-01 2.06985742e-01 -8.14969838e-01 8.87530565e-01
1.26489925e+00 -2.97760665e-01 6.62811399e-01 9.30260062e-01
-2.92147137e-02 -4.79481131e-01 -1.03733647e+00 7.97004580e-01
-2.89039344e-01 -8.70025158e-01 -3.83047700e-01 3.63789529e-01
6.12394631e-01 -4.15752858e-01 4.30821300e-01 5.47807276e-01
5.95582068e-01 -1.08552372e+00 8.19995940e-01 4.63200003e-01
8.69764507e-01 -1.50608706e+00 6.31113291e-01 4.34784085e-01
-9.67586577e-01 -5.70603728e-01 -2.67124623e-01 -8.37881789e-02
2.02983879e-02 5.90810835e-01 -1.32939756e-01 6.97102308e-01
9.94940341e-01 -1.22659050e-01 2.24144101e-01 7.09718406e-01
-6.54594377e-02 3.93946320e-01 -3.25991094e-01 -1.38915896e-01
3.80846322e-01 -2.57972956e-01 5.87271392e-01 8.27191949e-01
2.83608347e-01 3.66225451e-01 2.58766741e-01 5.84573746e-01
9.78673324e-02 5.79607449e-02 -2.22189650e-01 -5.14534069e-03
4.17370558e-01 6.91158116e-01 -5.76953769e-01 1.45103529e-01
8.35658684e-02 7.48592734e-01 1.97755679e-01 1.52605176e-01
-1.20630348e+00 -9.24892843e-01 5.65747321e-01 6.73757913e-03
6.18468225e-01 -6.61492646e-02 -1.05297148e-01 -8.17186832e-01
2.94426590e-01 -4.68174726e-01 6.99937940e-01 -7.15916082e-02
-1.10720932e+00 1.85115308e-01 2.10275024e-01 -9.98683631e-01
-2.62820482e-01 -3.59300137e-01 9.51723084e-02 6.62375271e-01
-1.32382524e+00 -4.05719519e-01 5.19172251e-01 7.47336030e-01
-8.82213265e-02 -1.71930388e-01 7.18482673e-01 1.33026853e-01
-5.94670057e-01 9.34545875e-01 6.07297719e-01 -2.72392273e-01
1.91082731e-01 -1.28710794e+00 -4.38062638e-01 6.41492784e-01
-7.39233196e-01 4.13338840e-01 8.21580648e-01 -5.36485493e-01
-1.44705653e+00 -9.28660035e-01 4.35069770e-01 -1.61418810e-01
6.63128316e-01 -2.93518484e-01 -3.38116229e-01 6.53457284e-01
-2.52617717e-01 -3.23354900e-01 7.38164127e-01 -1.68221131e-01
-2.50102460e-01 -4.16661948e-01 -1.55274570e+00 6.27382219e-01
9.92338538e-01 -2.82124400e-01 -4.19106543e-01 5.56174181e-02
7.97518730e-01 -2.98216432e-01 -1.41887760e+00 4.72365886e-01
5.49705863e-01 -1.20944667e+00 5.90832591e-01 -5.07198572e-01
6.67792261e-02 6.62467405e-02 -6.49729788e-01 -7.10101485e-01
6.03946336e-02 -1.15263939e+00 -2.46308833e-01 8.62001956e-01
4.75997150e-01 -8.00247848e-01 5.59804976e-01 9.02187288e-01
-3.43447062e-03 -1.04620743e+00 -1.57817292e+00 -1.12523746e+00
6.55602396e-01 -3.15046191e-01 5.04709899e-01 4.75686848e-01
3.06695193e-01 -3.74492496e-01 -4.32053119e-01 -1.82054237e-01
9.41745102e-01 1.26972899e-01 -7.33765662e-02 -8.46035480e-01
-8.96008968e-01 -4.54291105e-01 -5.74390590e-02 -6.18608236e-01
-1.73664734e-01 -5.83180904e-01 4.95044980e-03 -8.87146354e-01
4.01043802e-01 -7.12154806e-01 -9.22737718e-01 5.26007771e-01
2.42616534e-01 -6.23400569e-01 1.69004977e-01 -2.73222417e-01
-1.04044807e+00 8.22759807e-01 9.51307356e-01 1.98974520e-01
-2.78283358e-01 3.26646745e-01 -7.96963751e-01 5.04004359e-01
8.38378370e-01 -5.33409655e-01 -6.34276688e-01 -8.63789469e-02
5.90618074e-01 6.88873589e-01 1.05987944e-01 -8.63538921e-01
-1.64327577e-01 -5.94930410e-01 5.86085059e-02 -3.83002043e-01
1.37836248e-01 -4.89497066e-01 7.52861351e-02 7.79394090e-01
-6.66662514e-01 1.01273738e-01 1.11828022e-01 1.03655624e+00
3.67515445e-01 -1.53460905e-01 9.82396722e-01 -2.23628163e-01
-4.47171964e-02 2.47855619e-01 -2.72604436e-01 2.96235532e-01
1.33831465e+00 1.21576190e-01 -2.89572030e-01 -4.93529677e-01
-6.05101585e-01 3.90805542e-01 1.90794125e-01 -1.15844794e-01
2.38996059e-01 -9.76168931e-01 -3.26529086e-01 -8.38097632e-02
-1.50582954e-01 1.37312906e-02 6.58650994e-01 1.09421980e+00
-2.22066209e-01 3.65729272e-01 -1.06803201e-01 -1.29559442e-01
-4.93684739e-01 7.48525977e-01 4.83499616e-01 -6.28858924e-01
-3.64083588e-01 8.66035998e-01 2.15048306e-02 9.09199119e-02
6.21003568e-01 -1.14869237e-01 5.14908254e-01 -1.88836426e-01
4.99481976e-01 7.51014352e-01 -8.11719596e-02 -6.86798468e-02
-3.21938217e-01 8.44967291e-02 -1.97444305e-01 -6.35957539e-01
1.33468139e+00 -1.56260267e-01 -3.70725244e-02 5.08230388e-01
1.01921785e+00 -3.41518819e-01 -1.72119856e+00 2.89251916e-02
-2.34106444e-02 -2.78029770e-01 -1.56489596e-01 -9.62147295e-01
-7.82497525e-01 9.36811984e-01 1.02694011e+00 1.81710750e-01
1.14558291e+00 -1.74490064e-01 6.14996314e-01 4.01522636e-01
8.26134384e-01 -1.69576478e+00 -1.95044931e-03 5.80888748e-01
5.56897044e-01 -7.30133533e-01 -6.68143407e-02 3.80942523e-01
-8.79136622e-01 6.19096398e-01 3.52653563e-01 -5.35540208e-02
7.12116301e-01 9.28548872e-02 -6.69830024e-01 1.05779119e-01
-4.82495844e-01 -2.47249037e-01 -3.58898610e-01 9.05871615e-02
1.19209163e-01 3.23535115e-01 -5.93130767e-01 1.18400824e+00
-2.88088530e-01 -1.48937747e-01 2.80851871e-01 1.23944068e+00
-6.47944391e-01 -1.08843076e+00 2.31619142e-02 3.53530824e-01
-7.96139956e-01 2.06985581e-03 2.06179485e-01 7.51435161e-01
-7.59751722e-02 7.86732435e-01 -7.05612898e-02 -8.05530623e-02
2.96307474e-01 3.04988444e-01 7.06512451e-01 1.27423210e-02
-3.42417210e-01 1.83210641e-01 -2.24059522e-01 -5.69515228e-01
-6.23037294e-02 -6.99308157e-01 -1.48153841e+00 -5.11712134e-01
1.26749903e-01 4.28249955e-01 4.49012876e-01 9.26007509e-01
2.56608814e-01 1.49750635e-01 9.59555984e-01 -2.70267338e-01
-1.19006622e+00 -5.54696739e-01 -1.14347982e+00 1.41604140e-01
1.11576922e-01 -7.11201251e-01 -5.12547135e-01 -5.42893112e-01]
|
[4.373906135559082, 2.9331655502319336]
|
fe0ecf29-eddd-49ee-a9ca-023843d2b205
|
exploiting-topic-based-twitter-sentiment-for
| null | null |
https://aclanthology.org/P13-2005
|
https://aclanthology.org/P13-2005.pdf
|
Exploiting Topic based Twitter Sentiment for Stock Prediction
| null |
['Huayi Li', 'Xiaotie Deng', 'Bing Liu', 'Qing Li', 'Jianfeng Si', 'Arjun Mukherjee']
|
2013-08-01
| null | null | null |
acl-2013-8
|
['stock-market-prediction', 'stock-prediction']
|
['time-series', 'time-series']
|
[-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]
|
[-7.326651573181152, 3.7653591632843018]
|
6366af3c-5591-4d31-a85d-de06ca7e5f88
|
multilingual-short-text-responses-clustering
| null | null |
https://aclanthology.org/W18-3723
|
https://aclanthology.org/W18-3723.pdf
|
Multilingual Short Text Responses Clustering for Mobile Educational Activities: a Preliminary Exploration
|
Text clustering is a powerful technique to detect topics from document corpora, so as to provide information browsing, analysis, and organization. On the other hand, the Instant Response System (IRS) has been widely used in recent years to enhance student engagement in class and thus improve their learning effectiveness. However, the lack of functions to process short text responses from the IRS prevents the further application of IRS in classes. Therefore, this study aims to propose a proper short text clustering module for the IRS, and demonstrate our implemented techniques through real-world examples, so as to provide experiences and insights for further study. In particular, we have compared three clustering methods and the result shows that theoretically better methods need not lead to better results, as there are various factors that may affect the final performance.
|
['Tsung-Yen Li', 'Chun-Yen Chang', 'Yu-Ta Chien', 'Lung-Hao Lee', 'Yuen-Hsien Tseng']
|
2018-07-01
| null | null | null |
ws-2018-7
|
['text-clustering', 'short-text-clustering']
|
['natural-language-processing', 'natural-language-processing']
|
[-1.67444140e-01 -2.29402483e-01 -3.34866382e-02 -4.11629975e-01
-3.17322969e-01 -5.90822577e-01 5.53459167e-01 6.55553043e-01
-3.13149482e-01 4.49255377e-01 5.66453263e-02 -5.07728040e-01
-4.25508916e-01 -8.78519833e-01 3.26028951e-02 -6.18476927e-01
4.17710990e-01 2.20001608e-01 6.01907015e-01 -2.52722293e-01
7.46275961e-01 5.21100283e-01 -2.04409742e+00 4.63281333e-01
1.11791623e+00 2.63557225e-01 4.24702436e-01 4.36995387e-01
-7.19879806e-01 8.15846026e-01 -1.00875914e+00 -1.48484468e-01
-2.88610458e-01 -4.36593890e-01 -9.50483978e-01 1.29237056e-01
-7.20180050e-02 -3.42185795e-02 9.57725719e-02 7.88319409e-01
5.67936778e-01 6.25494778e-01 4.99609441e-01 -1.02586019e+00
-2.98904479e-01 7.56909668e-01 -3.81944627e-01 -8.52191076e-03
7.21058428e-01 -4.76031989e-01 5.01239777e-01 -3.58385861e-01
4.24691379e-01 1.04465377e+00 1.07471272e-01 4.05158997e-01
-6.51797712e-01 -7.47758329e-01 1.79661810e-01 3.65592182e-01
-1.35877407e+00 3.82862687e-02 8.43089521e-01 -3.53713125e-01
4.14480329e-01 5.37965834e-01 4.87543762e-01 6.49843931e-01
7.03029782e-02 9.47947145e-01 1.07097638e+00 -7.91712046e-01
3.37716848e-01 7.69777298e-01 7.05028415e-01 2.41782799e-01
6.28082454e-02 -7.64900684e-01 -2.21727192e-01 -9.21760052e-02
2.79242933e-01 2.69651562e-01 -1.85589343e-01 -1.93091422e-01
-6.35096133e-01 7.15455830e-01 1.32365227e-01 7.91107357e-01
-9.36018582e-03 -5.76173544e-01 2.00865999e-01 3.12447965e-01
2.42034569e-01 4.35299277e-01 -1.25263214e-01 -4.12459731e-01
-7.78400421e-01 1.31708369e-01 8.31768811e-01 8.20774853e-01
5.17172039e-01 -3.12388241e-01 7.40692616e-02 9.24906552e-01
3.44488651e-01 -7.00544491e-02 8.97026479e-01 -5.59866250e-01
4.78068590e-01 1.19771683e+00 -5.64569868e-02 -1.04503453e+00
-1.37526602e-01 -2.00085148e-01 -4.52369452e-01 -1.71592548e-01
3.56656760e-01 7.03130960e-02 -4.39862967e-01 9.91272092e-01
3.34986925e-01 2.35340698e-03 -4.70664352e-02 6.72598898e-01
1.17514038e+00 6.40315771e-01 1.24208972e-01 -2.68396646e-01
1.31181288e+00 -6.61994457e-01 -1.07877767e+00 1.35168374e-01
1.02645183e+00 -1.24141645e+00 1.19756484e+00 7.89834559e-01
-8.66372943e-01 -6.91807508e-01 -6.42676175e-01 3.80657256e-01
-6.71205282e-01 2.97766536e-01 5.00396788e-01 1.12604189e+00
-8.57420683e-01 4.39961761e-01 -7.69998431e-01 -5.98387480e-01
-2.24919409e-01 2.82095402e-01 4.45634276e-02 -9.68466923e-02
-9.53711629e-01 4.64246750e-01 2.04909384e-01 -1.99303016e-01
-3.53072695e-02 -3.30744058e-01 -3.04319203e-01 3.23253006e-01
1.82755202e-01 -1.04653254e-01 1.19743586e+00 -5.92427731e-01
-1.69116294e+00 5.49289882e-01 -7.38484710e-02 9.58226323e-02
2.91859955e-01 -2.36377373e-01 -2.57154226e-01 1.44547656e-01
-1.45961210e-01 1.26008272e-01 3.27064320e-02 -1.08529747e+00
-7.20153630e-01 -4.17375743e-01 -3.79056372e-02 4.53408301e-01
-1.09430015e+00 3.29186857e-01 -6.07547998e-01 -3.89636189e-01
4.50455189e-01 -7.67773867e-01 -1.42371610e-01 -1.01713431e+00
-1.24570191e-01 -7.18658984e-01 1.12589014e+00 -2.76157260e-01
1.62269199e+00 -2.10440898e+00 -4.48204100e-01 4.35230613e-01
-7.24847391e-02 3.27798754e-01 4.44323152e-01 6.95138991e-01
-1.42685428e-01 4.35058832e-01 3.42447400e-01 1.14908610e-02
-2.56339461e-01 -3.92111167e-02 -1.09159067e-01 4.61228676e-02
-3.91168743e-01 6.72176629e-02 -6.95795774e-01 -6.70158625e-01
4.37604874e-01 1.41688406e-01 -2.88187504e-01 3.38222951e-01
1.04715869e-01 2.24878877e-01 -8.31741631e-01 3.99549782e-01
4.40225393e-01 -5.63999303e-02 2.51901418e-01 4.12520587e-01
-5.38428068e-01 3.75751317e-01 -1.51231492e+00 1.12739372e+00
-3.39755327e-01 7.10017085e-01 -1.75619870e-01 -1.23793936e+00
1.14697576e+00 6.33071303e-01 7.06635296e-01 -7.02996731e-01
3.13329518e-01 -8.72472152e-02 -9.19769034e-02 -6.75592244e-01
8.09053004e-01 4.86666441e-01 6.99412450e-02 7.49326885e-01
-5.12408078e-01 -3.74891087e-02 3.40203345e-01 6.56573832e-01
8.06164086e-01 -1.72629371e-01 7.41063245e-03 -1.98377520e-01
6.84475720e-01 2.14661404e-01 1.34785056e-01 5.27400017e-01
-8.74599442e-02 3.21548134e-01 1.57713935e-01 -1.89721644e-01
-3.34212601e-01 -5.34738183e-01 -2.03994592e-03 1.19125617e+00
1.59167930e-01 -5.27130067e-01 -7.49574125e-01 -5.56028783e-01
-5.15674949e-01 7.53604233e-01 1.51959360e-01 5.64553440e-02
-3.01825076e-01 -5.72636187e-01 2.25773543e-01 2.27866203e-01
5.36068559e-01 -9.23995078e-01 -4.49181229e-01 2.88047850e-01
-5.19935191e-01 -8.29835355e-01 -1.46169171e-01 4.21237573e-02
-1.03148925e+00 -8.52001190e-01 -4.85949576e-01 -9.37460542e-01
9.34381783e-01 1.01569140e+00 6.68260574e-01 5.18755674e-01
-1.75428912e-01 6.65884852e-01 -7.94087529e-01 -5.64066887e-01
-3.13925236e-01 4.00247425e-01 -3.48748975e-02 -3.39645833e-01
6.73806489e-01 -3.16160828e-01 -5.30057490e-01 5.64345896e-01
-1.05100751e+00 -2.31762394e-01 2.88784683e-01 2.76025057e-01
1.56975418e-01 7.62589574e-01 6.66803956e-01 -1.18250859e+00
1.17304575e+00 -1.78930387e-01 -5.64255536e-01 4.11446065e-01
-8.04350555e-01 -1.14575841e-01 7.34684467e-01 -3.89067382e-01
-1.28748846e+00 -8.35399106e-02 -1.86888799e-02 4.53976691e-02
-6.30117416e-01 5.37507176e-01 -1.47834092e-01 -8.27203393e-02
6.80966735e-01 1.81598961e-01 -2.69629955e-01 -4.71025944e-01
-1.72636211e-01 1.24420035e+00 -5.01411855e-02 -6.66091681e-01
3.65673184e-01 -6.38291165e-02 -2.58526206e-01 -1.01847851e+00
-5.82049012e-01 -1.19270551e+00 -5.81099033e-01 -5.77711403e-01
5.78291774e-01 -8.21691334e-01 -1.19028533e+00 -6.11780174e-02
-7.91725278e-01 -2.34735042e-01 3.05714339e-01 7.18725801e-01
-4.16938290e-02 4.31001902e-01 -6.20358050e-01 -9.65818346e-01
-3.70486788e-02 -1.15849316e+00 3.51984620e-01 7.60754287e-01
-3.33763987e-01 -9.93015468e-01 -1.57863587e-01 9.12889242e-01
2.71240145e-01 -2.51170099e-01 8.82214904e-01 -8.66084099e-01
-4.54637319e-01 -2.90344447e-01 2.16442183e-01 1.43890738e-01
6.55400157e-02 6.16185606e-01 -9.38000321e-01 -2.15345934e-01
1.69500768e-01 -1.50391422e-02 3.76273811e-01 1.16072297e-01
1.31107700e+00 -1.54174238e-01 -5.31885743e-01 -1.31544560e-01
1.28272009e+00 7.71440446e-01 6.67667150e-01 7.07528472e-01
3.70973587e-01 9.14802313e-01 1.02542043e+00 3.74142885e-01
1.96254268e-01 3.85403335e-01 3.65104750e-02 1.00124724e-01
1.14563599e-01 -7.45758135e-03 3.56347769e-01 1.28396475e+00
2.04966236e-02 -2.47934043e-01 -1.15853596e+00 4.90910530e-01
-1.84338319e+00 -8.34006131e-01 -6.27671063e-01 2.17155766e+00
6.54070437e-01 9.08915550e-02 1.38402149e-01 4.73666638e-01
6.69434845e-01 -3.76491219e-01 3.11634243e-01 -6.18930459e-01
5.68277955e-01 2.35522911e-01 1.08560547e-01 2.47105211e-01
-6.31165922e-01 8.57227683e-01 6.05366373e+00 8.70919466e-01
-1.17249608e+00 -2.72700697e-01 4.62412804e-01 2.33342439e-01
-2.39243522e-01 4.93579246e-02 -1.09378290e+00 4.30059731e-01
9.51943099e-01 -4.45269942e-01 6.12668805e-02 8.85087907e-01
6.29925072e-01 -2.97508150e-01 -6.39988482e-01 5.63104689e-01
-1.76023483e-01 -9.60233867e-01 -2.57912204e-02 -5.58059383e-03
8.14487219e-01 -7.02912271e-01 -1.63319916e-01 3.75642717e-01
4.97023821e-01 -4.74902600e-01 5.79692349e-02 1.47659853e-01
-1.01333931e-01 -1.21673501e+00 9.31060195e-01 7.79049516e-01
-1.02381945e+00 -5.58600249e-03 -3.70717883e-01 -2.36770391e-01
-5.74076176e-01 4.43498164e-01 -1.26226556e+00 7.06242919e-01
7.61966407e-01 2.04453334e-01 -7.38468826e-01 1.28910089e+00
-2.45322764e-01 9.17436123e-01 -6.97026849e-02 -6.57091975e-01
1.24648049e-01 -4.88344610e-01 3.27950940e-02 1.05791950e+00
4.41557854e-01 5.26246309e-01 4.62323606e-01 4.99161839e-01
2.74218470e-01 7.62030542e-01 -5.37402570e-01 2.29014810e-02
6.70271695e-01 1.40702009e+00 -1.32611108e+00 -3.43624681e-01
-3.59748900e-01 4.18441623e-01 -1.63208470e-01 4.38060015e-01
-5.99171102e-01 -8.04001570e-01 1.56940371e-01 5.50357699e-01
-2.52558678e-01 -3.10907096e-01 -5.11031389e-01 -6.50947690e-01
-3.54279578e-01 -8.38922024e-01 4.84066904e-01 -6.10678792e-01
-7.11629272e-01 2.15250656e-01 6.55069500e-02 -1.29130638e+00
-1.15873784e-01 -2.38388911e-01 -9.34153974e-01 6.12055600e-01
-1.03634465e+00 -4.44726437e-01 -5.38525522e-01 6.61080420e-01
6.22640312e-01 -1.65226460e-01 6.92167163e-01 4.10868853e-01
-9.08916891e-01 5.01213491e-01 2.30244532e-01 -2.94928551e-02
1.05385458e+00 -1.21865368e+00 -4.42233145e-01 7.13864863e-01
1.32379085e-01 1.05428553e+00 7.01178312e-01 -4.44627106e-01
-1.09991634e+00 -6.69829130e-01 8.73923719e-01 -1.57919616e-01
2.88526416e-01 -5.49384579e-02 -9.25411522e-01 2.49645576e-01
4.77015495e-01 -8.81197989e-01 1.10921025e+00 3.05176079e-01
3.50061536e-01 -1.92044824e-01 -9.13969398e-01 7.75265992e-01
2.44189024e-01 -2.36606568e-01 -5.46030521e-01 3.69756043e-01
4.78130072e-01 -1.92510128e-01 -8.86739552e-01 2.34845094e-02
2.02707261e-01 -1.05206656e+00 6.28403962e-01 -2.17547104e-01
3.45111489e-01 -2.02666268e-01 3.60439181e-01 -1.13195825e+00
-9.75459293e-02 -3.04498672e-01 4.65027452e-01 1.83635736e+00
3.17541063e-01 -6.58289254e-01 1.14368618e+00 9.72159624e-01
-1.22880086e-01 -3.88703018e-01 -1.78870976e-01 -5.08763433e-01
-1.30975291e-01 -4.46357280e-01 5.57886720e-01 1.29472136e+00
5.19214272e-01 3.31036568e-01 1.95326179e-01 7.26476163e-02
1.89465374e-01 4.06494200e-01 1.01920342e+00 -1.37823606e+00
1.05277628e-01 -5.11840522e-01 -8.03647861e-02 -1.00670421e+00
-7.78342336e-02 -4.53223944e-01 -4.42106389e-02 -1.63186240e+00
2.21878052e-01 -5.47110260e-01 -1.37371689e-01 4.00799125e-01
-4.06862378e-01 -2.44293064e-01 2.73949206e-02 1.99372917e-01
-6.00421965e-01 9.99263898e-02 1.22646284e+00 1.27253592e-01
-6.39748573e-01 7.30215192e-01 -8.34265649e-01 6.26067460e-01
1.07461524e+00 -5.46137154e-01 -6.72179759e-01 -8.85924175e-02
-3.38195998e-04 -1.23500517e-02 -4.17353839e-01 -1.00972378e+00
8.02148521e-01 -4.77594852e-01 2.13891819e-01 -8.82917762e-01
-2.69764841e-01 -1.02518058e+00 -2.11739123e-01 3.29834193e-01
-4.83479053e-01 4.52647544e-02 1.80523500e-01 1.70282736e-01
-6.30986571e-01 -7.17397809e-01 2.85607100e-01 -1.25672087e-01
-4.63916272e-01 -3.06325614e-01 -9.27673340e-01 -4.82527524e-01
1.30299914e+00 -3.86135608e-01 -2.59748787e-01 -5.32112956e-01
-4.43109006e-01 3.22298169e-01 3.29676777e-01 4.13753778e-01
4.14362341e-01 -9.81889486e-01 -7.24518076e-02 8.64487812e-02
-9.62450579e-02 9.06437188e-02 2.43216053e-01 4.42627966e-01
-6.56479537e-01 6.35324538e-01 -1.24629587e-01 -5.78827858e-01
-2.04966784e+00 4.01222855e-01 -3.93934786e-01 -3.59656811e-01
-3.52293164e-01 4.78721529e-01 -1.73546106e-01 -4.41233933e-01
7.27397263e-01 -2.18166113e-01 -9.80639815e-01 2.24462479e-01
5.93367875e-01 5.73580384e-01 4.10595506e-01 9.98652726e-02
-1.97040569e-02 3.17023933e-01 -2.79194444e-01 -7.29096234e-02
1.07235372e+00 -3.31965655e-01 1.30674457e-02 4.43503708e-01
6.01279914e-01 3.52592975e-01 -3.20099294e-01 -2.06505395e-02
4.11106169e-01 -5.05549729e-01 1.56179843e-02 -4.16911215e-01
-7.71121562e-01 9.73714828e-01 5.60148120e-01 7.78973222e-01
1.25570941e+00 -3.48333836e-01 4.91975874e-01 5.86099625e-01
1.48082614e-01 -1.42337680e+00 4.49158490e-01 3.77348334e-01
1.46703988e-01 -1.16482687e+00 1.43252626e-01 -7.26727545e-01
-5.54133058e-01 1.29681861e+00 8.08767378e-01 4.49654549e-01
4.62673306e-01 1.34936973e-01 1.61719307e-01 -8.74534994e-02
-7.48425364e-01 -2.74892803e-02 1.69971526e-01 2.25218654e-01
1.25397182e+00 4.82959934e-02 -7.66040564e-01 5.06994069e-01
-2.39927426e-01 -2.11490899e-01 8.93947899e-01 1.18164623e+00
-8.28276515e-01 -1.52544916e+00 -8.67728174e-01 2.51362711e-01
-6.92151129e-01 3.15723151e-01 -8.43790054e-01 7.79128134e-01
-1.00688919e-01 1.50255322e+00 -2.38063052e-01 -4.03765351e-01
3.98724973e-01 3.30913931e-01 -7.13990442e-03 -9.40909982e-01
-8.89068305e-01 1.31474495e-01 -2.00122952e-01 -6.45007715e-02
-4.18576926e-01 -4.88793403e-01 -1.44153464e+00 -5.79333961e-01
-6.37460589e-01 9.06073213e-01 8.99949491e-01 7.25230396e-01
1.68879300e-01 8.15594912e-01 8.04303467e-01 -5.88630810e-02
-2.93443143e-01 -9.78237867e-01 -3.83514196e-01 2.73212910e-01
-5.35987496e-01 -3.13051879e-01 -2.66530365e-01 6.34381995e-02]
|
[10.39604377746582, 7.284470558166504]
|
c7ad3b7f-1622-434d-a74d-ebdaacaa35f4
|
1e2caecoea-e343a-c-14e2a-ae2e-acoustic-echo-1
| null | null |
https://aclanthology.org/O17-1018
|
https://aclanthology.org/O17-1018.pdf
|
改進的向量空間可適性濾波器用於聲學回聲消除 (Acoustic Echo Cancellation Using an Improved Vector-Space-Based Adaptive Filtering Algorithm) [In Chinese]
| null |
['Ying-Ren Chien', 'Jin Li-You', 'Yu Tsao']
|
2017-11-01
|
1e2caecoea-e343a-c-14e2a-ae2e-acoustic-echo
|
https://aclanthology.org/O17-3005
|
https://aclanthology.org/O17-3005.pdf
|
roclingijclclp-2017-11
|
['acoustic-echo-cancellation', 'acoustic-echo-cancellation']
|
['medical', '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]
|
[-7.233340740203857, 3.6774749755859375]
|
58f411d6-bae4-46de-87e5-4f7e8b9acaa3
|
optimizing-multi-domain-performance-with
|
2304.06277
| null |
https://arxiv.org/abs/2304.06277v1
|
https://arxiv.org/pdf/2304.06277v1.pdf
|
Optimizing Multi-Domain Performance with Active Learning-based Improvement Strategies
|
Improving performance in multiple domains is a challenging task, and often requires significant amounts of data to train and test models. Active learning techniques provide a promising solution by enabling models to select the most informative samples for labeling, thus reducing the amount of labeled data required to achieve high performance. In this paper, we present an active learning-based framework for improving performance across multiple domains. Our approach consists of two stages: first, we use an initial set of labeled data to train a base model, and then we iteratively select the most informative samples for labeling to refine the model. We evaluate our approach on several multi-domain datasets, including image classification, sentiment analysis, and object recognition. Our experiments demonstrate that our approach consistently outperforms baseline methods and achieves state-of-the-art performance on several datasets. We also show that our method is highly efficient, requiring significantly fewer labeled samples than other active learning-based methods. Overall, our approach provides a practical and effective solution for improving performance across multiple domains using active learning techniques.
|
['Rakshitha Panduranga', 'Royston Mascarenhas', 'Akshay Gulati', 'Aayush Shah', 'Anand Gokul Mahalingam']
|
2023-04-13
| null | null | null | null |
['object-recognition']
|
['computer-vision']
|
[ 2.77185738e-01 -4.65032309e-02 -8.81335676e-01 -8.02156031e-01
-1.54902732e+00 -5.55420816e-01 5.60640275e-01 4.75714177e-01
-5.89289904e-01 6.63079441e-01 3.51536907e-02 -3.12324595e-02
1.31960228e-01 -5.52274108e-01 -5.95974982e-01 -4.94020104e-01
2.03597322e-01 7.97146976e-01 5.64296246e-01 1.32242382e-01
5.56194901e-01 2.91722447e-01 -1.35687196e+00 5.12874663e-01
1.07272685e+00 9.39621031e-01 8.01392347e-02 2.73673534e-01
-3.96392375e-01 1.12626731e+00 -7.05874860e-01 -3.53763998e-01
-6.52431860e-04 -2.03841984e-01 -9.74885941e-01 6.72443211e-01
2.93054491e-01 -3.07509869e-01 4.73451197e-01 7.27996826e-01
2.80377537e-01 2.39688709e-01 6.77624166e-01 -1.09496844e+00
-5.06055892e-01 3.50548089e-01 -7.67584503e-01 2.13572308e-01
2.69724578e-01 1.01163201e-01 1.00502038e+00 -1.31613564e+00
4.97138560e-01 1.08382046e+00 4.83127773e-01 5.47597826e-01
-1.39617276e+00 -6.70616567e-01 6.94928110e-01 2.38362119e-01
-1.04366624e+00 -9.48568285e-01 7.32692063e-01 -4.60996896e-01
6.93296134e-01 -5.95935024e-02 4.63380992e-01 8.70758176e-01
-4.24578488e-01 1.55368209e+00 1.01134038e+00 -8.23404610e-01
6.48341715e-01 4.27366078e-01 6.25788629e-01 5.17614067e-01
1.92595586e-01 -2.71204531e-01 -8.33836913e-01 -6.12144232e-01
3.36928457e-01 5.82846142e-02 8.00139233e-02 -7.11059570e-01
-8.80530357e-01 1.04921937e+00 6.18135408e-02 2.12870706e-02
-4.27465439e-01 -3.02283883e-01 -3.14373374e-02 1.37910798e-01
9.67310309e-01 8.29841554e-01 -5.54046512e-01 -1.52083531e-01
-9.82829273e-01 2.49380812e-01 6.97493851e-01 6.41215384e-01
8.53035390e-01 -4.42412317e-01 8.71865824e-03 1.37506187e+00
5.20770729e-01 1.96190849e-01 1.70543462e-01 -1.21405053e+00
6.21624708e-01 1.12350965e+00 3.28341633e-01 -4.30086643e-01
3.82728651e-02 -1.04117922e-01 -1.09562248e-01 4.13676292e-01
3.33909631e-01 -1.61364287e-01 -9.42238688e-01 1.43359268e+00
4.58386898e-01 -1.47346193e-02 -1.08974114e-01 4.49443698e-01
6.31773114e-01 5.50908625e-01 3.53564143e-01 -3.54880214e-01
7.48938560e-01 -1.32607079e+00 -6.18035078e-01 -6.45916462e-01
7.74435937e-01 -7.13477850e-01 1.13177001e+00 6.24289691e-01
-1.10129702e+00 -4.32972968e-01 -9.31348264e-01 1.36149898e-01
-9.53504220e-02 9.84132662e-02 5.46803176e-01 4.88964766e-01
-7.16397226e-01 3.51310760e-01 -1.04909933e+00 -1.05288722e-01
7.57224202e-01 3.62320542e-01 -3.00242990e-01 -4.03849542e-01
-7.12466061e-01 7.62580693e-01 4.42287415e-01 -4.34389621e-01
-8.62763047e-01 -7.56582201e-01 -9.13060188e-01 -8.16913322e-02
4.49317157e-01 -8.93707424e-02 1.79430258e+00 -1.17461073e+00
-1.33366978e+00 8.46891463e-01 -3.91087353e-01 -4.03013349e-01
2.93124616e-01 -4.53044564e-01 -2.69706607e-01 8.19421187e-02
2.06533615e-02 8.56064737e-01 6.17001176e-01 -1.45029354e+00
-8.20233822e-01 -2.92833626e-01 2.21242145e-01 5.03400743e-01
-8.48878384e-01 2.56079525e-01 -6.96021259e-01 -3.67207229e-01
-1.03174798e-01 -9.80388463e-01 -4.07092184e-01 8.83216411e-02
-1.32075166e-02 -5.88207245e-01 9.47794557e-01 -1.64715841e-01
1.17326581e+00 -2.20787811e+00 -1.96353555e-01 3.15800160e-02
4.18123543e-01 4.76524770e-01 -1.90169320e-01 1.66808560e-01
8.06414038e-02 7.62452036e-02 -4.09818560e-01 -6.56416655e-01
-2.12868631e-01 1.36213467e-01 -2.15817332e-01 1.82348371e-01
6.23614609e-01 8.66110444e-01 -8.97040844e-01 -7.06383169e-01
2.39308730e-01 1.82615861e-01 -6.27203226e-01 3.94040525e-01
-3.93399507e-01 5.10028064e-01 -4.04671282e-01 5.03535211e-01
5.10667384e-01 -7.54858851e-01 4.06937897e-01 2.80481309e-01
1.05511956e-01 5.08459091e-01 -1.23456693e+00 1.63081574e+00
-3.64497185e-01 6.26147747e-01 -1.45648822e-01 -1.39671552e+00
1.12464559e+00 9.29981917e-02 7.71037579e-01 -9.60922360e-01
-1.36628792e-01 2.15433225e-01 -2.24070087e-01 -3.35584909e-01
2.94998318e-01 2.98725933e-01 -1.12759694e-01 7.31038630e-01
1.01492293e-01 9.80856344e-02 5.02074897e-01 1.68523610e-01
1.15915477e+00 -1.59829780e-01 4.45520163e-01 -7.35156462e-02
4.24194306e-01 3.05942506e-01 6.23939157e-01 6.68882787e-01
-2.75489539e-01 3.26601267e-01 4.93974715e-01 -3.34571660e-01
-8.45142663e-01 -8.36603999e-01 -4.11267132e-02 1.38996577e+00
7.09237456e-02 -2.44200960e-01 -6.21181548e-01 -1.23618448e+00
-4.06843387e-02 7.91335225e-01 -5.29723346e-01 -6.40722811e-02
-6.19010210e-01 -1.00512886e+00 -9.22252387e-02 5.69756150e-01
6.16728365e-01 -9.04333055e-01 -1.25873208e-01 8.45216438e-02
-1.15886301e-01 -9.11211193e-01 -2.17045713e-02 2.18202129e-01
-1.23043442e+00 -1.12710679e+00 -6.41213298e-01 -8.71155202e-01
9.45589542e-01 4.00686383e-01 1.47567165e+00 5.93289696e-02
6.14003055e-02 1.55166730e-01 -5.94380975e-01 -6.37833774e-01
-2.90204197e-01 3.70798856e-01 -3.03379565e-01 -1.61352038e-01
7.29643822e-01 7.61895031e-02 -4.77058440e-01 5.31819642e-01
-8.05373192e-01 -1.41474783e-01 4.80033606e-01 7.36296594e-01
7.23304331e-01 -5.36192842e-02 7.43447661e-01 -1.56860709e+00
6.61745846e-01 -6.25126958e-01 -5.93912661e-01 4.80836332e-01
-7.70896196e-01 -7.07291663e-02 4.74565744e-01 -6.17404401e-01
-1.24792397e+00 4.62085932e-01 -2.25646906e-02 5.24230413e-02
-1.93949446e-01 5.25066078e-01 -6.84454069e-02 -8.72267932e-02
1.09847057e+00 -1.93027258e-01 -7.81703591e-02 -6.81573689e-01
6.67093545e-02 1.01047814e+00 -3.36689204e-02 -4.86404419e-01
5.64371109e-01 3.18392485e-01 -7.63079703e-01 -4.79792833e-01
-1.45355308e+00 -8.03950906e-01 -8.32021296e-01 -1.69630170e-01
3.35192055e-01 -1.07825327e+00 -1.86758358e-02 4.56398487e-01
-7.85511315e-01 -5.98194838e-01 -3.59555215e-01 4.86384243e-01
-1.88320592e-01 1.77550152e-01 -3.55448842e-01 -7.17853248e-01
-1.88529685e-01 -1.27329469e+00 1.05087757e+00 3.12477171e-01
-5.24122894e-01 -1.21241546e+00 4.28966194e-01 9.14570272e-01
1.26067996e-01 -9.79171395e-02 6.12777829e-01 -1.09539938e+00
-5.75895607e-01 -3.57113212e-01 2.71330476e-02 4.53900814e-01
2.59771198e-01 -1.11042060e-01 -9.22089517e-01 -4.02131528e-01
-1.46590501e-01 -9.63904262e-01 8.72776508e-01 2.06380501e-01
1.41499782e+00 -2.02695373e-02 -4.51788694e-01 8.94686133e-02
1.12102485e+00 6.19462311e-01 5.15836120e-01 3.54515940e-01
2.60424465e-01 5.07514060e-01 1.19049990e+00 3.37834567e-01
1.84695795e-01 5.06843150e-01 -4.65348326e-02 -6.14148021e-01
3.72810960e-02 1.89330876e-01 3.88741285e-01 6.43206179e-01
3.70557994e-01 -3.10251534e-01 -1.35804617e+00 6.71729803e-01
-1.98649383e+00 -8.29684675e-01 9.11144316e-02 2.40393257e+00
1.07317197e+00 5.15587091e-01 3.11248183e-01 3.86029869e-01
5.94599485e-01 -1.53996609e-02 -9.66187477e-01 -7.20713437e-02
2.75104821e-01 4.60403234e-01 1.67356372e-01 5.00716209e-01
-1.38710356e+00 8.37259889e-01 7.55763149e+00 6.66545749e-01
-1.06431687e+00 1.36586994e-01 8.11892569e-01 -2.50806600e-01
-3.77140194e-02 -9.35754552e-02 -9.94109213e-01 4.08292145e-01
8.19365621e-01 -1.48580462e-01 -5.72960377e-02 1.21649837e+00
3.81128974e-02 -3.93565416e-01 -1.15056753e+00 7.69889593e-01
1.25802532e-01 -1.41612577e+00 -9.08449143e-02 -6.80757733e-03
1.03714716e+00 -2.71171182e-02 -4.93888333e-02 3.83027345e-01
8.65727663e-01 -6.37479603e-01 4.62065607e-01 6.32151514e-02
5.05503654e-01 -7.68930256e-01 5.50466120e-01 6.46986663e-01
-7.10318446e-01 -1.05080038e-01 -1.73476934e-01 -3.66763622e-02
-1.02530599e-01 7.90972471e-01 -1.13626289e+00 5.55904470e-02
4.38035846e-01 9.72546399e-01 -8.35977852e-01 1.29329741e+00
-1.47234380e-01 9.46351707e-01 -3.36795375e-02 -6.43816665e-02
-4.85800765e-02 2.52615422e-01 8.65287241e-03 1.13376737e+00
-1.93042621e-01 1.30284578e-01 7.46850014e-01 4.74065930e-01
-3.05469930e-01 1.33401975e-01 -4.07376707e-01 -2.72514045e-01
7.77708054e-01 1.11830580e+00 -9.29362655e-01 -6.12486899e-01
-6.02003276e-01 6.62576795e-01 7.26674736e-01 2.84327060e-01
-6.34458303e-01 -2.80607104e-01 3.40102583e-01 1.79190695e-01
9.97375548e-02 -2.80355603e-01 -3.34725231e-01 -1.26939225e+00
-1.13469549e-01 -1.08859444e+00 5.34214079e-01 -4.14423317e-01
-1.34509599e+00 2.69959688e-01 5.24368212e-02 -1.42610061e+00
-3.60916644e-01 -5.39237916e-01 -2.77627856e-01 6.96506619e-01
-1.48773754e+00 -6.37562573e-01 -4.05415088e-01 4.71614927e-01
8.76513660e-01 -3.45911801e-01 9.60693777e-01 5.06033778e-01
-7.66529679e-01 5.57695091e-01 2.59991407e-01 4.02756631e-01
1.03821540e+00 -1.18601775e+00 5.41852713e-01 8.16329896e-01
3.64588082e-01 6.11787915e-01 2.25632414e-01 -5.64831197e-01
-1.04861546e+00 -8.55728447e-01 7.65535235e-01 -5.84283292e-01
4.18633580e-01 -3.69346172e-01 -8.78658116e-01 7.50562072e-01
1.49808899e-01 -3.74080203e-02 1.33971214e+00 5.91411889e-01
-3.64818633e-01 -1.66640535e-01 -1.05186093e+00 1.39530644e-01
7.64231861e-01 -2.37980127e-01 -7.39988625e-01 4.47622150e-01
4.57737029e-01 -2.92667568e-01 -5.64913929e-01 5.03282487e-01
1.52957156e-01 -7.18199492e-01 8.93820405e-01 -6.60755217e-01
5.25801659e-01 3.36983167e-02 1.58132747e-01 -1.35681832e+00
-4.89405572e-01 -9.51546654e-02 -3.07378471e-01 1.24031591e+00
9.58141625e-01 -4.06084746e-01 1.11564124e+00 8.27919483e-01
1.48822246e-02 -8.90012562e-01 -3.39572430e-01 -5.58593869e-01
-6.27769306e-02 -4.33000594e-01 2.05372542e-01 1.19844723e+00
-1.40237346e-01 6.02288246e-01 3.60123813e-02 -3.34211916e-01
8.34540009e-01 1.40870139e-01 8.41947377e-01 -1.59565282e+00
-1.45534739e-01 -3.11155412e-02 1.47609133e-02 -1.27216053e+00
2.21589059e-01 -6.03089213e-01 9.17264894e-02 -1.78357482e+00
3.95327657e-01 -8.84394228e-01 -4.71694946e-01 8.57561648e-01
-4.68690425e-01 3.60515475e-01 -4.76524904e-02 5.81967890e-01
-1.27686763e+00 2.74768114e-01 1.00163662e+00 -2.08628029e-01
-4.36760098e-01 1.20097332e-01 -9.78232622e-01 7.70954251e-01
6.77731097e-01 -6.15666449e-01 -7.80128658e-01 -5.45378804e-01
3.81779708e-02 -2.18511030e-01 -1.22397169e-01 -7.88120747e-01
1.84919029e-01 -4.49514538e-01 7.41008937e-01 -5.88740289e-01
4.68375832e-01 -6.37124777e-01 -3.64265919e-01 2.03126431e-01
-7.44665444e-01 -2.29753330e-01 1.43856972e-01 3.47946823e-01
-4.88015860e-01 -2.72053540e-01 1.25956917e+00 -1.05994470e-01
-1.04317558e+00 2.35533908e-01 -2.93876112e-01 3.80651683e-01
1.10124493e+00 -9.98581126e-02 -3.78723294e-01 -1.52154518e-02
-7.42978811e-01 3.87469143e-01 4.49716270e-01 5.09273231e-01
4.46443200e-01 -1.36757779e+00 -5.70039988e-01 1.20204031e-01
4.51302588e-01 1.33401290e-01 -1.69367105e-01 4.23055142e-01
-3.96698862e-01 1.92071036e-01 2.82473862e-03 -8.12606514e-01
-1.59307861e+00 2.23911420e-01 3.59288305e-02 -5.80431521e-01
-2.32026741e-01 9.19434011e-01 8.27984419e-03 -5.89482963e-01
4.99222547e-01 2.20666125e-01 -4.83042568e-01 4.14049983e-01
7.67742932e-01 2.58937716e-01 3.51536810e-01 -1.18990809e-01
-4.58877861e-01 2.34307364e-01 -6.25222325e-01 -2.32831597e-01
1.52931976e+00 7.94062018e-02 2.03798935e-01 5.11159658e-01
1.27593446e+00 -1.16475113e-01 -1.24098909e+00 -6.97120905e-01
4.40165937e-01 -5.75536430e-01 1.81498930e-01 -9.48573887e-01
-8.62992167e-01 6.71930373e-01 5.26358545e-01 2.37952784e-01
1.18997312e+00 9.18537825e-02 5.14939249e-01 5.59936106e-01
3.97502691e-01 -1.33697677e+00 6.27832949e-01 4.19056445e-01
4.31725919e-01 -1.74338353e+00 2.11893737e-01 -5.32484055e-01
-7.25835800e-01 8.29314172e-01 8.85712504e-01 -1.00214332e-01
5.93037188e-01 4.04417455e-01 2.90251911e-01 -8.46977606e-02
-1.04888690e+00 -8.37098286e-02 3.27643454e-01 2.91950107e-01
7.43275106e-01 -1.42868966e-01 -1.65988520e-01 1.60837829e-01
3.52403075e-01 2.39604935e-01 9.49732363e-02 1.36761391e+00
-7.93545365e-01 -1.69655454e+00 -4.06371266e-01 4.17096257e-01
-2.91647971e-01 2.08525956e-01 -8.35214436e-01 6.43705547e-01
5.64172603e-02 1.23985958e+00 2.68428087e-01 -1.99211627e-01
4.43339974e-01 2.53087163e-01 4.30191517e-01 -1.27274239e+00
-4.33735907e-01 -4.53725661e-04 2.67983854e-01 -4.12706852e-01
-7.10111499e-01 -8.49513650e-01 -1.02787757e+00 -1.87988296e-01
-4.45993036e-01 2.98914462e-01 4.61484700e-01 9.55219388e-01
4.56039488e-01 7.99400806e-02 9.17102933e-01 -3.88451248e-01
-3.51664037e-01 -8.08953285e-01 -2.11269438e-01 6.18769050e-01
1.71721861e-01 -7.93068290e-01 -2.13636294e-01 2.41712093e-01]
|
[9.632560729980469, 4.171031475067139]
|
7f247823-cc73-45f2-b4e0-2a9291a21f90
|
unsupervised-quality-prediction-for-improved
|
2307.01464
| null |
https://arxiv.org/abs/2307.01464v1
|
https://arxiv.org/pdf/2307.01464v1.pdf
|
Unsupervised Quality Prediction for Improved Single-Frame and Weighted Sequential Visual Place Recognition
|
While substantial progress has been made in the absolute performance of localization and Visual Place Recognition (VPR) techniques, it is becoming increasingly clear from translating these systems into applications that other capabilities like integrity and predictability are just as important, especially for safety- or operationally-critical autonomous systems. In this research we present a new, training-free approach to predicting the likely quality of localization estimates, and a novel method for using these predictions to bias a sequence-matching process to produce additional performance gains beyond that of a naive sequence matching approach. Our combined system is lightweight, runs in real-time and is agnostic to the underlying VPR technique. On extensive experiments across four datasets and three VPR techniques, we demonstrate our system improves precision performance, especially at the high-precision/low-recall operating point. We also present ablation and analysis identifying the performance contributions of the prediction and weighted sequence matching components in isolation, and the relationship between the quality of the prediction system and the benefits of the weighted sequential matcher.
|
['Michael Milford', 'Jason J. Ford', 'Helen Carson']
|
2023-07-04
| null | null | null | null |
['visual-place-recognition']
|
['computer-vision']
|
[ 3.22984606e-01 -2.36666813e-01 -4.56157923e-01 -6.18400693e-01
-8.48851502e-01 -7.67884135e-01 7.88670540e-01 4.55293983e-01
-6.07444942e-01 4.58565682e-01 2.53873527e-01 -6.61512733e-01
-9.14058536e-02 -4.61396217e-01 -6.97901309e-01 -1.12644501e-01
-5.41136563e-01 2.56374866e-01 6.87688589e-01 -2.83792764e-01
7.51182020e-01 7.75353432e-01 -1.94576800e+00 1.49386629e-01
5.72507381e-01 1.04424894e+00 2.20059544e-01 8.84016216e-01
-5.17744385e-02 8.72025251e-01 -4.36378479e-01 4.58264053e-02
3.77124101e-01 1.41630948e-01 -6.08445346e-01 -4.38965172e-01
4.39288914e-01 -2.89345145e-01 -4.92968857e-02 6.94841385e-01
5.30385792e-01 1.10451639e-01 4.29534882e-01 -1.53803110e+00
-3.96381199e-01 9.60991755e-02 -5.71169913e-01 3.13458413e-01
9.03189778e-01 4.50480551e-01 9.40712512e-01 -7.82559335e-01
6.49228156e-01 7.50750959e-01 1.03651178e+00 1.94930255e-01
-1.30132020e+00 -6.22007668e-01 2.42384046e-01 1.42765895e-01
-1.57080817e+00 -8.53434801e-01 5.94903529e-03 -5.88224709e-01
1.65968227e+00 1.96753606e-01 3.74642789e-01 6.19765222e-01
4.37883139e-01 4.59497243e-01 8.04697037e-01 -3.72394323e-01
3.78283679e-01 1.26500830e-01 -7.48454779e-02 5.87361932e-01
2.67611831e-01 5.88897943e-01 -9.20426071e-01 -4.23560292e-01
1.88398391e-01 -1.33664981e-01 -2.34432533e-01 -5.91552436e-01
-1.18351042e+00 2.77403325e-01 3.30197662e-01 1.83416262e-01
-2.70506769e-01 3.20275992e-01 2.61275649e-01 1.89033449e-01
2.28073791e-01 5.27220666e-01 -5.74401736e-01 -5.28284907e-01
-1.31929302e+00 3.41021508e-01 7.16275096e-01 1.13144624e+00
8.43576372e-01 -1.83450386e-01 -1.31566525e-01 5.04450977e-01
4.77955222e-01 3.81822586e-01 6.00555003e-01 -8.53180528e-01
2.98717827e-01 5.18765211e-01 4.36021984e-01 -1.05659437e+00
-5.50113976e-01 -2.79610921e-02 4.05180678e-02 4.25751239e-01
1.36138573e-01 1.37579720e-02 -9.48918998e-01 1.54469109e+00
1.83276758e-01 2.03164935e-01 1.09825481e-03 7.80446291e-01
2.48996302e-01 5.45197070e-01 2.31687978e-01 9.97670069e-02
1.07232535e+00 -7.12448120e-01 -2.88076341e-01 -6.90312684e-01
8.43746603e-01 -8.60285342e-01 9.11788702e-01 1.33163473e-02
-6.22681439e-01 -3.31101209e-01 -1.47525370e+00 -4.33019623e-02
-3.89708906e-01 -2.57717460e-01 5.07444143e-01 5.34847438e-01
-1.19808292e+00 6.75043464e-01 -7.92121708e-01 -7.42152810e-01
-4.75864150e-02 5.18004715e-01 -5.65593064e-01 1.15689941e-01
-8.79755974e-01 1.39129829e+00 3.13688278e-01 -2.31459230e-01
-5.13392091e-01 -6.97238505e-01 -8.15201521e-01 -1.15659289e-01
-3.23333368e-02 -1.59281552e-01 1.37047243e+00 -8.44215155e-01
-1.05717099e+00 5.23043871e-01 -4.37624216e-01 -6.83781087e-01
4.75658745e-01 -3.16075757e-02 -5.60519695e-01 -2.02405140e-01
1.37900919e-01 7.12779582e-01 5.02820134e-01 -1.06100297e+00
-1.12024581e+00 -3.15721750e-01 -1.92284480e-01 2.40987092e-01
8.63057226e-02 -6.21210337e-02 -2.51277953e-01 -1.07816346e-01
1.39559776e-01 -1.01762474e+00 -2.50541866e-01 1.91614702e-01
2.06428200e-01 1.78576916e-01 5.64651310e-01 -7.56366611e-01
1.07942271e+00 -2.28905702e+00 -4.17509466e-01 4.38846201e-01
-5.62053993e-02 1.37207538e-01 -3.05519015e-01 7.50649214e-01
2.62357090e-02 1.21426776e-01 -1.09023325e-01 -6.10969146e-04
-1.25242695e-01 3.37400474e-02 -5.64638257e-01 7.36545324e-01
3.64653394e-02 6.17603242e-01 -1.10255277e+00 -3.12496692e-01
3.39440703e-01 2.22344264e-01 -4.21838999e-01 1.19833283e-01
-1.11666702e-01 9.49015096e-02 -7.50818253e-02 6.74538970e-01
2.55693883e-01 -4.63245921e-02 3.56622159e-01 1.08278535e-01
-4.06938642e-01 5.04172325e-01 -1.05981910e+00 1.55139959e+00
-5.33350110e-01 1.04615247e+00 -2.33778045e-01 -3.09902817e-01
9.73416984e-01 1.19423091e-01 3.18789184e-01 -9.84224916e-01
-2.96348989e-01 3.06975842e-01 -5.96586093e-02 -3.76666844e-01
1.19606340e+00 2.66644865e-01 -4.93411273e-02 3.36280197e-01
-1.75823748e-01 1.46719679e-01 -1.17741093e-01 -8.08619708e-03
1.38793755e+00 4.58112836e-01 6.66153610e-01 -1.56270131e-01
2.31495485e-01 4.98363823e-01 5.44411600e-01 8.03998947e-01
-5.76983988e-01 5.24517715e-01 7.71549568e-02 -3.29804361e-01
-1.11571240e+00 -7.65956879e-01 -4.90158834e-02 1.16684031e+00
5.59485078e-01 -5.72025418e-01 -2.39817679e-01 -4.72594410e-01
3.49271983e-01 1.01049411e+00 -4.19221193e-01 -3.63813818e-01
-3.18477005e-01 -2.60794401e-01 7.69776285e-01 6.85670257e-01
1.42133623e-01 -8.00508678e-01 -9.88879979e-01 2.04570532e-01
-5.84747866e-02 -1.04011750e+00 -3.16949606e-01 9.93251428e-02
-6.32071733e-01 -8.76154542e-01 -1.32189065e-01 -4.90015179e-01
3.88805509e-01 5.02381206e-01 9.38485324e-01 1.77645162e-01
-1.95582584e-01 4.07039076e-01 -3.47523153e-01 -2.50037313e-01
-2.53626257e-01 -4.27123196e-02 2.55526543e-01 -4.54331458e-01
6.26883626e-01 -4.55348849e-01 -5.13262570e-01 4.15951461e-01
-4.47570086e-01 -1.65922031e-01 5.60611546e-01 6.33098006e-01
4.91845697e-01 -4.53483433e-01 2.01413751e-01 -4.80025619e-01
6.74984992e-01 -4.60493803e-01 -7.42401242e-01 4.10726905e-01
-1.14423442e+00 3.06527823e-01 2.22091779e-01 -3.42602909e-01
-6.63379490e-01 4.06875670e-01 -3.20367575e-01 -2.32590675e-01
-1.18681818e-01 4.07860816e-01 2.62645006e-01 -5.89756012e-01
1.04068899e+00 4.46781397e-01 1.90247055e-02 -2.14586318e-01
3.92035633e-01 8.16925764e-01 5.07993042e-01 -4.47343975e-01
6.51377738e-01 3.69351238e-01 -1.97966054e-01 -7.80295193e-01
-1.97074354e-01 -6.90183759e-01 -2.85064369e-01 -1.90063924e-01
2.85083473e-01 -8.02462459e-01 -8.71928990e-01 1.11390417e-02
-1.01742983e+00 -2.80051649e-01 -1.23552889e-01 2.35621631e-01
-5.79250515e-01 3.67681891e-01 -3.90898772e-02 -9.02709723e-01
-2.63204575e-01 -1.22632682e+00 1.06772661e+00 1.58499822e-01
-5.72777450e-01 -5.20237088e-01 3.70106608e-01 5.43746948e-02
5.59754252e-01 -8.72002542e-02 5.36328077e-01 -1.01435542e+00
-7.48028994e-01 -5.08481741e-01 -3.08339179e-01 -1.87230512e-01
-1.68459296e-01 8.11259001e-02 -9.76805151e-01 -3.30911249e-01
-5.43486714e-01 -5.86785860e-02 4.29129541e-01 -2.23059524e-02
5.85199118e-01 -2.03938290e-01 -6.75264001e-01 3.45994085e-01
1.63192248e+00 3.59658182e-01 6.28285944e-01 7.60510981e-01
3.16044956e-01 5.89534760e-01 9.67739344e-01 1.62895530e-01
5.67196310e-01 9.20783639e-01 2.57447094e-01 2.87872702e-01
5.77009656e-02 -5.57487130e-01 3.97203654e-01 1.50954291e-01
1.53756142e-01 -1.23727731e-02 -1.33308876e+00 8.42071652e-01
-1.98978806e+00 -9.98527408e-01 2.55267918e-01 2.58224177e+00
3.81528467e-01 9.88360792e-02 8.29925388e-02 -3.50157358e-02
5.09694099e-01 2.60887325e-01 -5.07880807e-01 -5.34211218e-01
3.63489538e-01 -2.38726601e-01 1.14743042e+00 5.48351467e-01
-7.31785834e-01 9.23151433e-01 7.88298893e+00 5.28851926e-01
-1.34071636e+00 -7.47018456e-02 4.37816158e-02 -4.36462127e-02
-1.14046626e-01 2.36761332e-01 -8.26276898e-01 4.39156324e-01
1.13498831e+00 -1.65168628e-01 5.15198469e-01 1.18654180e+00
1.38654917e-01 -4.86588687e-01 -1.23285925e+00 9.24257457e-01
2.00106557e-02 -1.37664986e+00 -2.78711408e-01 1.47062436e-01
2.15051115e-01 4.85319346e-01 6.85022175e-02 6.71113580e-02
3.65401596e-01 -9.22902703e-01 1.07815170e+00 2.57717550e-01
6.62799358e-01 -6.80435717e-01 5.79943299e-01 4.04973388e-01
-1.19214141e+00 -1.47761732e-01 -7.78232962e-02 -2.14004576e-01
8.39372799e-02 2.39142720e-02 -1.26228559e+00 2.97640294e-01
6.80911124e-01 3.11935604e-01 -5.94149113e-01 1.28696382e+00
-2.27815717e-01 2.78212935e-01 -3.82190824e-01 -3.58165324e-01
6.23766556e-02 4.32074189e-01 5.39822340e-01 1.33376646e+00
2.35162973e-01 -1.53470337e-01 1.06939794e-02 2.63168246e-01
3.21501374e-01 1.29659651e-02 -9.12827969e-01 3.88643965e-02
9.53054011e-01 1.09878647e+00 -6.70505702e-01 -1.04546502e-01
-2.69835472e-01 6.33154750e-01 4.15086329e-01 1.64990649e-01
-7.77254105e-01 -3.73325855e-01 1.08485305e+00 1.98966086e-01
3.79766762e-01 -5.47808766e-01 -5.38799465e-01 -8.83343816e-01
1.45109713e-01 -8.22155297e-01 1.82115659e-01 -8.89840186e-01
-7.71259606e-01 5.49509227e-01 -2.28755519e-01 -1.32760096e+00
-4.77950901e-01 -2.86405027e-01 -1.51465937e-01 9.58033621e-01
-1.53948176e+00 -8.40804517e-01 -1.36080265e-01 -3.84483188e-02
2.05685049e-01 1.55094460e-01 8.84904206e-01 1.32251918e-01
-1.26097754e-01 5.92678249e-01 1.50184363e-01 -9.72948745e-02
8.28765571e-01 -8.15671444e-01 7.95338035e-01 1.15853119e+00
1.81515723e-01 8.94091070e-01 1.04222178e+00 -7.17230499e-01
-1.45694518e+00 -9.34030533e-01 1.07004809e+00 -6.47274435e-01
5.97755969e-01 -2.91751444e-01 -6.03841722e-01 8.23351681e-01
-7.38161430e-02 -1.70556262e-01 6.73859060e-01 3.47766310e-01
-6.60839081e-01 -1.53722644e-01 -1.33846688e+00 6.90217674e-01
9.74362850e-01 -7.24636376e-01 -7.95494735e-01 -5.37201501e-02
6.21852338e-01 -5.38682461e-01 -6.11839652e-01 4.77652490e-01
1.01408184e+00 -1.09632373e+00 1.00854051e+00 -3.52942318e-01
1.07784644e-01 -8.64957035e-01 -5.45338452e-01 -9.11433756e-01
-3.44222963e-01 -2.08202645e-01 2.60820575e-02 9.76833284e-01
6.96866512e-01 -6.87670887e-01 7.36730278e-01 9.51830745e-01
8.32227394e-02 -5.52429616e-01 -8.55617642e-01 -8.41127336e-01
-7.02423096e-01 -6.92961633e-01 7.08481967e-01 7.53303647e-01
1.58755228e-01 7.95584619e-02 -3.96823257e-01 6.75285220e-01
3.97715658e-01 1.56189442e-01 7.87227869e-01 -7.27316737e-01
-3.36538374e-01 -1.91525415e-01 -1.01753712e+00 -9.77208197e-01
-1.58708438e-01 -5.84358990e-01 5.52731633e-01 -1.66929698e+00
-2.25656793e-01 -5.14242470e-01 -2.08916411e-01 6.19600534e-01
1.23926245e-01 1.48851037e-01 2.13521898e-01 4.91003394e-01
-7.14054823e-01 9.22235847e-02 2.71906108e-01 7.34967133e-03
-3.59136969e-01 -1.52883440e-01 -5.25966465e-01 5.29120445e-01
6.00490272e-01 -4.91469026e-01 -5.12180090e-01 -4.59028155e-01
2.83909827e-01 2.42989234e-04 1.62174702e-01 -1.25387812e+00
7.21319377e-01 -2.57154703e-01 4.18875933e-01 -5.86554110e-01
3.76203150e-01 -9.30542529e-01 2.77035028e-01 3.93673211e-01
-4.21090633e-01 3.08136106e-01 4.71641481e-01 6.75067425e-01
1.26706958e-01 4.12876233e-02 6.44897521e-01 1.80017650e-01
-1.49368751e+00 -4.19091620e-02 -5.43162405e-01 -1.42251104e-01
1.27951646e+00 -6.14101112e-01 -4.14119214e-01 -3.46112579e-01
-2.33564988e-01 3.41586500e-01 1.06019616e+00 7.08056033e-01
6.06976330e-01 -9.00974870e-01 -1.67342365e-01 1.70104608e-01
7.07296848e-01 -5.79941571e-01 -9.48236361e-02 6.27804518e-01
-7.11774290e-01 4.13018644e-01 -2.86559373e-01 -5.74908912e-01
-1.22622585e+00 6.76251650e-01 3.68135691e-01 2.08645999e-01
-4.67354625e-01 6.59088552e-01 -4.28803235e-01 -5.18347681e-01
5.02204537e-01 -4.98663038e-02 1.42288581e-01 -2.04730988e-01
7.18254864e-01 3.18097323e-01 2.60779977e-01 -6.68858230e-01
-9.26382661e-01 2.43890345e-01 -9.05247182e-02 -4.15026188e-01
1.10616577e+00 -2.60398984e-01 5.25004752e-02 4.74836439e-01
8.24473143e-01 -5.74362231e-03 -1.21707475e+00 1.40885845e-01
5.20188928e-01 -6.62370503e-01 -2.49118879e-02 -9.08971369e-01
-3.39480221e-01 5.70956469e-01 7.94199347e-01 -4.42860425e-02
7.82407761e-01 -2.59801149e-01 5.51389396e-01 3.27185869e-01
8.77248645e-01 -8.99592102e-01 -5.92036009e-01 4.15361613e-01
4.07008410e-01 -1.06266773e+00 1.93333805e-01 -7.81524256e-02
-5.81412673e-01 7.87303329e-01 5.20920336e-01 1.50246350e-02
3.52885723e-01 5.77670753e-01 1.81756288e-01 1.71009615e-01
-8.88706505e-01 -1.65134281e-01 1.95750982e-01 8.58963966e-01
3.94200474e-01 -7.18972385e-02 -2.47299165e-01 1.87987760e-02
3.18164937e-02 1.51893646e-01 2.76138335e-01 1.44613564e+00
-7.31582761e-01 -1.05111468e+00 -1.66726530e-01 3.93562168e-01
-1.51143551e-01 -1.99711382e-01 -4.41257805e-01 6.65487587e-01
-1.14973746e-01 9.71402526e-01 1.79066546e-02 -9.42789197e-01
3.81443232e-01 1.14046551e-01 2.32458681e-01 -3.99196416e-01
-6.85697615e-01 -5.36855876e-01 4.76558417e-01 -1.13595748e+00
-1.46010548e-01 -6.64055526e-01 -1.24759209e+00 -5.24885833e-01
-2.84769207e-01 4.03287299e-02 1.08902001e+00 9.51174915e-01
9.92236853e-01 1.07573017e-01 2.76060164e-01 -8.36054087e-01
-5.90452075e-01 -4.87758070e-01 -1.70117527e-01 -5.60033135e-02
4.71485049e-01 -5.66711187e-01 -1.53230980e-01 -2.30066478e-01]
|
[7.518453121185303, -1.8674038648605347]
|
f32a8e1b-c31e-49df-9380-33164a1aa32d
|
chinese-grammatical-errors-diagnosis-system
| null | null |
https://aclanthology.org/2020.nlptea-1.14
|
https://aclanthology.org/2020.nlptea-1.14.pdf
|
Chinese Grammatical Errors Diagnosis System Based on BERT at NLPTEA-2020 CGED Shared Task
|
In the process of learning Chinese, second language learners may have various grammatical errors due to the negative transfer of native language. This paper describes our submission to the NLPTEA 2020 shared task on CGED. We present a hybrid system that utilizes both detection and correction stages. The detection stage is a sequential labelling model based on BiLSTM-CRF and BERT contextual word representation. The correction stage is a hybrid model based on the n-gram and Seq2Seq. Without adding additional features and external data, the BERT contextual word representation can effectively improve the performance metrics of Chinese grammatical error detection and correction.
|
['Yingjie Han', 'yuke wang', 'Yingjie Yan', 'Haotian Huang', 'Yangchao Han', 'Hongying Zan']
| null | null | null | null |
aacl-nlp-tea-2020-12-1
|
['grammatical-error-detection']
|
['natural-language-processing']
|
[ 5.32541703e-03 7.13350698e-02 2.99991608e-01 -6.63628221e-01
-7.60792196e-01 -1.65411934e-01 5.31930886e-02 4.12470549e-01
-1.05665803e+00 9.79613841e-01 3.23765337e-01 -5.99860966e-01
7.88503170e-01 -8.08955669e-01 -5.92399836e-01 -1.15104534e-01
2.20576525e-01 2.87784159e-01 5.32896817e-01 -6.60475075e-01
3.96181136e-01 -7.00583830e-02 -9.34225321e-01 5.21970928e-01
1.61474431e+00 3.83668542e-01 7.82380283e-01 1.01150167e+00
-5.67671716e-01 9.96582806e-01 -9.27600503e-01 -4.68285918e-01
-2.16729894e-01 -7.38172412e-01 -9.64314401e-01 -9.59487677e-01
2.64208853e-01 -2.86111683e-01 2.08867624e-01 1.46194625e+00
8.36348236e-01 3.56985360e-01 1.62135497e-01 -4.42800134e-01
-1.11565733e+00 9.33391988e-01 1.28867477e-02 3.01287830e-01
5.15136361e-01 -9.23666880e-02 7.01680541e-01 -1.17775440e+00
4.62813705e-01 1.35589790e+00 1.04059243e+00 1.07773900e+00
-6.63440764e-01 -7.44875312e-01 4.26463783e-01 2.15919927e-01
-1.00426149e+00 8.10484309e-03 1.28301993e-01 -2.29353696e-01
1.44923520e+00 -1.63923517e-01 6.22491479e-01 8.18402112e-01
5.21713257e-01 1.06300426e+00 1.01785350e+00 -1.02377212e+00
-2.35366523e-01 -1.24417506e-01 4.68945116e-01 8.12147915e-01
-4.46458012e-02 2.39599094e-01 -4.11295772e-01 5.29975414e-01
3.50808740e-01 -2.51061797e-01 -7.97543824e-02 7.45021522e-01
-5.97287297e-01 8.32171559e-01 3.94984335e-01 4.29656237e-01
1.55767798e-01 4.62451756e-01 3.85576010e-01 6.16956949e-01
4.97140348e-01 3.68165463e-01 -8.44210625e-01 -4.27999079e-01
-6.76882982e-01 -3.44031374e-03 5.79629362e-01 1.60865319e+00
5.97795963e-01 2.61501521e-01 -6.27642393e-01 8.46222043e-01
6.23694897e-01 3.57798070e-01 8.59257102e-01 -1.57193080e-01
5.87137878e-01 4.60781604e-01 -2.58222342e-01 -1.78475305e-01
-3.54246914e-01 -2.68436193e-01 -5.05424082e-01 -7.24895820e-02
3.23153615e-01 -7.56350458e-01 -1.19983590e+00 1.70129430e+00
-1.21164054e-01 1.00797825e-01 9.41418260e-02 5.00467777e-01
1.14139569e+00 5.50127387e-01 6.80887878e-01 2.60799788e-02
8.81497800e-01 -1.32044303e+00 -1.19248903e+00 -1.57459587e-01
1.44325328e+00 -1.09812903e+00 9.88627970e-01 3.03756744e-01
-1.28861535e+00 -7.88509190e-01 -9.35865045e-01 -5.64530551e-01
-5.32346606e-01 -3.42560597e-02 1.85889229e-01 6.58285558e-01
-1.27831459e+00 8.34632158e-01 -6.74497247e-01 -2.69640058e-01
6.79006130e-02 2.91864336e-01 -1.36821732e-01 -3.25253665e-01
-1.53407300e+00 1.32024992e+00 6.37222886e-01 1.49133042e-01
-5.89226246e-01 -5.74187458e-01 -9.41360950e-01 -1.08580142e-01
-3.39610547e-01 -4.44705151e-02 1.93403268e+00 -1.06950378e+00
-1.68133271e+00 6.76163852e-01 -3.01392108e-01 -2.35798895e-01
5.06417811e-01 -7.12784052e-01 -4.08690959e-01 -4.97778147e-01
3.06193549e-02 5.63892305e-01 5.73646575e-02 -5.64024985e-01
-1.04727924e+00 -1.15027763e-01 -4.82731372e-01 4.49746609e-01
1.54593050e-01 6.38534904e-01 -2.14293867e-01 -7.69488335e-01
-2.14696288e-01 -5.25328517e-01 -5.21657586e-01 -5.47941506e-01
-3.86317670e-02 -8.38618040e-01 2.63049424e-01 -1.26731634e+00
1.47616625e+00 -1.97542763e+00 -1.73830211e-01 -1.04819439e-01
-3.43832791e-01 8.54387045e-01 -3.92995685e-01 3.46320480e-01
6.55938312e-02 5.67017853e-01 -6.73363507e-02 -5.35587549e-01
-2.35602036e-01 1.63018748e-01 1.12460487e-01 1.94542073e-02
3.65550369e-01 1.02984154e+00 -1.31838715e+00 -4.64846551e-01
-9.23513472e-02 1.75812289e-01 -5.76552153e-01 6.89767241e-01
-2.47181937e-01 5.19691229e-01 -1.90227225e-01 4.19355989e-01
7.47330844e-01 5.27471423e-01 6.54027611e-02 7.81876385e-01
-5.50761461e-01 1.04175317e+00 -7.66373277e-01 1.98067403e+00
-3.91035527e-01 2.98403442e-01 -1.27606690e-01 -6.51902139e-01
1.03783154e+00 4.67128873e-01 -4.65401620e-01 -1.06539261e+00
4.77131046e-02 4.83782172e-01 2.80422181e-01 -4.29344565e-01
7.61092544e-01 -2.71709293e-01 -1.85059339e-01 3.65670443e-01
5.09021223e-01 -1.76327243e-01 -4.65816632e-03 1.80802360e-01
9.90022242e-01 5.18700898e-01 2.85158932e-01 -5.08113086e-01
4.68859643e-01 9.69637409e-02 9.06288743e-01 8.34452271e-01
-4.80064273e-01 4.21691835e-01 1.31682917e-01 -2.70456105e-01
-5.76287091e-01 -8.47075641e-01 1.21305615e-03 1.45670819e+00
-1.83907241e-01 -3.98317069e-01 -9.04154778e-01 -9.80946422e-01
-2.78927416e-01 1.06606531e+00 -5.08922517e-01 -2.46668294e-01
-9.85722661e-01 -4.68494028e-01 8.14548969e-01 8.09128165e-01
4.46704805e-01 -1.52831101e+00 -1.60994142e-01 7.08043218e-01
-3.24358679e-02 -5.43061554e-01 -8.28023612e-01 6.54361665e-01
-9.02543426e-01 -7.01789081e-01 -4.51413065e-01 -1.48798931e+00
7.13921666e-01 -1.79299533e-01 1.12077832e+00 9.31968331e-01
-5.58298640e-02 -1.53657451e-01 -8.87172341e-01 -9.56236064e-01
-6.46827281e-01 8.10557902e-02 -1.98639274e-01 -9.42156911e-01
8.26515496e-01 2.45615505e-02 -1.52737662e-01 -3.29471022e-01
-2.42341653e-01 1.62985623e-01 3.19851935e-01 9.25480783e-01
3.49978924e-01 -5.63404977e-01 6.76175833e-01 -1.25622737e+00
6.67662680e-01 -2.56948829e-01 -4.09173310e-01 5.03520131e-01
-6.89963281e-01 5.90648390e-02 6.24242425e-01 -1.58974573e-01
-1.54455662e+00 -5.19782864e-02 -9.99770522e-01 3.29460025e-01
-1.37017697e-01 6.33282006e-01 -1.14778005e-01 -2.02897355e-01
4.26192045e-01 3.33510786e-02 -3.05810213e-01 -9.17658925e-01
2.11058393e-01 7.68041015e-01 4.53594893e-01 -5.97200215e-01
1.95561349e-01 -7.63511539e-01 -6.56685650e-01 -5.04736781e-01
-1.09753060e+00 -2.89473176e-01 -1.24170613e+00 -1.55209288e-01
1.05363095e+00 -1.15234601e+00 -1.91588923e-01 9.62813199e-01
-1.53461158e+00 -7.49194205e-01 -1.26146317e-01 7.07843065e-01
3.11588459e-02 3.09531450e-01 -1.31238139e+00 -5.22223651e-01
-6.20788097e-01 -9.61771607e-01 3.92477572e-01 5.51047146e-01
8.52467492e-02 -1.20672703e+00 3.89302939e-01 -1.99752957e-01
6.05376422e-01 -2.97777653e-01 9.54904377e-01 -9.30147290e-01
-1.35523558e-01 8.75698030e-02 6.09325105e-03 7.62682796e-01
-1.21214494e-01 8.58357700e-04 -8.13089490e-01 -1.58964217e-01
-2.90681034e-01 -4.17017430e-01 9.76954341e-01 1.28320381e-01
1.04002440e+00 -1.66554838e-01 1.20401427e-01 5.58837295e-01
1.53648973e+00 4.10449743e-01 5.68049908e-01 2.27850005e-01
8.78571153e-01 5.22836089e-01 8.27050269e-01 -1.37530595e-01
5.43364346e-01 3.22408706e-01 1.49321258e-01 1.35434434e-01
-4.52997953e-01 -6.44873977e-01 8.13641548e-01 1.73754811e+00
8.65234435e-02 -2.74745315e-01 -1.26306772e+00 7.19737828e-01
-1.62490249e+00 -5.26149035e-01 -9.30445790e-01 1.97689295e+00
1.45566201e+00 -1.17650375e-01 -4.83524859e-01 -4.19761211e-01
8.83607090e-01 -5.39616048e-01 1.14738956e-01 -1.34018707e+00
-6.09845966e-02 8.22501242e-01 4.43444103e-01 9.00815487e-01
-7.13349104e-01 1.74681759e+00 6.93677044e+00 7.54679203e-01
-9.04534400e-01 5.81832767e-01 2.75670469e-01 5.43349028e-01
-4.37637895e-01 -5.04944474e-02 -1.42718339e+00 4.78137225e-01
1.32088721e+00 7.95027539e-02 9.71947424e-03 5.22709429e-01
-5.51090948e-02 -2.85726577e-01 -8.38576376e-01 1.07018158e-01
6.05388451e-03 -7.87133396e-01 -2.39804953e-01 -4.02139872e-01
8.31289113e-01 4.66885000e-01 -4.13686186e-01 9.53233302e-01
1.00698888e+00 -1.10494268e+00 9.21842217e-01 6.04280174e-01
9.40868556e-01 -8.87777925e-01 1.10570991e+00 4.86987770e-01
-9.10971880e-01 2.44457603e-01 -6.53681159e-01 -5.91891289e-01
1.75946310e-01 1.46786675e-01 -7.96127558e-01 4.45171177e-01
6.58283055e-01 7.64387906e-01 -7.59592593e-01 1.13436139e+00
-1.17766511e+00 1.09374309e+00 6.06715679e-02 -4.49187130e-01
1.70763448e-01 -1.33665234e-01 3.60212065e-02 1.88841569e+00
6.11803293e-01 2.84115344e-01 9.50941965e-02 3.84856671e-01
-1.30356893e-01 3.56097549e-01 -2.44115636e-01 3.20569813e-01
4.58136380e-01 7.39028811e-01 -1.20628938e-01 -3.31041634e-01
-7.41688669e-01 1.13920391e+00 9.83021021e-01 -4.67990488e-02
-4.05791491e-01 -6.73780978e-01 2.49615863e-01 -4.76405412e-01
2.76955992e-01 -1.56088650e-01 -5.28013110e-01 -1.05350852e+00
-4.76675570e-01 -6.40706301e-01 3.45562398e-01 -4.16457683e-01
-1.17508531e+00 5.49133182e-01 -5.60720742e-01 -7.12611735e-01
2.23059347e-03 -7.92497873e-01 -1.17296088e+00 1.43116641e+00
-1.64507782e+00 -8.85547042e-01 9.49448626e-03 5.20027697e-01
6.02542460e-01 -1.50698259e-01 1.11759126e+00 4.42487657e-01
-8.47034931e-01 1.10386288e+00 2.86214530e-01 4.69081014e-01
1.02967727e+00 -1.80196810e+00 7.72104561e-01 1.19457674e+00
-3.73188078e-01 4.11877692e-01 2.31387496e-01 -1.13981092e+00
-5.46892643e-01 -1.43807936e+00 1.95068383e+00 -2.06972092e-01
3.43453735e-01 -3.60182792e-01 -1.15413499e+00 9.85914707e-01
4.55755502e-01 -1.85268372e-01 8.02319288e-01 2.37138525e-01
1.44070417e-01 3.14273059e-01 -1.03058612e+00 3.14977169e-01
9.01776791e-01 -3.80970865e-01 -1.05323744e+00 2.18100220e-01
1.09915090e+00 -6.87060118e-01 -8.42372119e-01 3.09549987e-01
9.13219452e-02 -4.61537272e-01 1.29492268e-01 -8.48483145e-01
5.31364143e-01 -2.11355258e-02 2.05652490e-01 -1.74575377e+00
-5.36585927e-01 -3.09498578e-01 3.17828327e-01 1.52815807e+00
6.29063845e-01 -2.41786361e-01 1.74593225e-01 4.20378268e-01
-9.48796451e-01 -5.49017489e-01 -9.32852805e-01 -6.10943317e-01
7.98029602e-01 -5.84403455e-01 4.50428039e-01 9.39495146e-01
1.99497670e-01 1.78367704e-01 -2.30696991e-01 -1.39305577e-01
6.35195598e-02 -6.37508690e-01 1.03212111e-01 -1.02115810e+00
1.86240882e-01 -2.30352774e-01 1.02220669e-01 -1.07199800e+00
1.83444411e-01 -1.07424951e+00 8.81387293e-01 -1.39603364e+00
-9.42652971e-02 -7.28692353e-01 -6.18254364e-01 6.49258554e-01
-8.56048226e-01 -3.27323079e-02 2.83862144e-01 -2.45478496e-01
-7.45120168e-01 7.10505366e-01 1.26125693e+00 3.77471626e-01
-1.38138264e-01 2.78359205e-02 -4.24044490e-01 7.85206676e-01
1.02049971e+00 -7.88472950e-01 3.45824003e-01 -1.13031590e+00
2.43920624e-01 -1.27932981e-01 -3.35395932e-01 -8.20612669e-01
4.02820349e-01 3.19497734e-02 4.85214800e-01 -5.57642579e-01
-4.99506414e-01 -3.05639803e-01 -7.28567541e-01 9.39651966e-01
-4.60910231e-01 6.07329011e-01 5.41195333e-01 1.85116529e-01
-1.83585018e-01 -9.05446053e-01 9.46361065e-01 -4.90769655e-01
-8.12023640e-01 -6.16642786e-03 -9.12726462e-01 3.19967121e-01
7.19959080e-01 7.57357329e-02 -3.30214769e-01 -1.88997611e-02
-6.03386819e-01 6.80614769e-01 2.28595920e-02 5.14895976e-01
6.94294214e-01 -1.31764054e+00 -9.80448127e-01 3.20859462e-01
-2.01665252e-01 2.96888083e-01 -3.87398736e-03 5.72303653e-01
-1.03984749e+00 4.38851804e-01 -3.62655789e-01 -9.68736112e-02
-1.28310513e+00 9.56200957e-02 4.81252640e-01 -5.78870416e-01
-2.95208216e-01 1.60739946e+00 -2.31041357e-01 -1.05762517e+00
4.38512266e-01 -2.50071049e-01 -6.56927347e-01 -1.63645983e-01
8.99077237e-01 3.48483026e-01 3.27715605e-01 -4.66075182e-01
-1.88778147e-01 1.00743748e-01 -3.88134569e-01 2.39698030e-03
1.20544159e+00 -8.94127116e-02 -2.77929455e-01 6.13695920e-01
7.52823114e-01 9.84630436e-02 -8.89063954e-01 -3.08306277e-01
5.71565866e-01 -3.97603624e-02 -6.77306205e-02 -1.32771158e+00
-6.49475694e-01 1.29021394e+00 6.00169599e-01 -5.07943094e-01
8.86291385e-01 -4.97541904e-01 8.77773166e-01 2.29096755e-01
2.74715126e-01 -1.57741880e+00 -2.95366138e-01 1.66135299e+00
5.84931493e-01 -1.27080798e+00 -5.00197947e-01 -1.93270668e-01
-5.00421226e-01 1.19778955e+00 1.38280201e+00 -2.95729041e-01
6.36853695e-01 2.66568154e-01 3.96370202e-01 2.45790794e-01
-8.02164316e-01 -1.44520804e-01 2.52552956e-01 5.06671667e-01
1.27538764e+00 1.14806466e-01 -1.02011204e+00 1.01921606e+00
-3.06451946e-01 -2.01913714e-01 5.66381872e-01 1.04580045e+00
-7.95653522e-01 -1.63777721e+00 -7.54156569e-03 3.07142168e-01
-8.82440925e-01 -8.22613299e-01 -3.04523706e-01 5.24179876e-01
5.34621239e-01 1.00147152e+00 1.65494412e-01 -3.20868313e-01
4.00523871e-01 3.83357078e-01 4.92732882e-01 -1.41366279e+00
-1.52063024e+00 -2.60631353e-01 4.09906134e-02 -4.38598812e-01
-8.56554434e-02 -6.62966967e-01 -1.74053752e+00 -1.49752960e-01
-6.38874233e-01 4.10052121e-01 6.41111434e-01 1.19539785e+00
-1.87365487e-01 6.34409308e-01 1.14744596e-01 -2.41338402e-01
-5.83782494e-01 -1.62853837e+00 -4.06293958e-01 5.90529479e-03
2.64112115e-01 7.10154995e-02 1.01327300e-01 -2.70325571e-01]
|
[11.033177375793457, 10.800220489501953]
|
fdd664b1-2b51-4ab7-b88a-5efaec783286
|
interactive-submodular-bandit
| null | null |
http://papers.nips.cc/paper/6619-interactive-submodular-bandit
|
http://papers.nips.cc/paper/6619-interactive-submodular-bandit.pdf
|
Interactive Submodular Bandit
|
In many machine learning applications, submodular functions have been used as a model for evaluating the utility or payoff of a set such as news items to recommend, sensors to deploy in a terrain, nodes to influence in a social network, to name a few. At the heart of all these applications is the assumption that the underlying utility/payoff function is known a priori, hence maximizing it is in principle possible. In real life situations, however, the utility function is not fully known in advance and can only be estimated via interactions. For instance, whether a user likes a movie or not can be reliably evaluated only after it was shown to her. Or, the range of influence of a user in a social network can be estimated only after she is selected to advertise the product. We model such problems as an interactive submodular bandit optimization, where in each round we receive a context (e.g., previously selected movies) and have to choose an action (e.g., propose a new movie). We then receive a noisy feedback about the utility of the action (e.g., ratings) which we model as a submodular function over the context-action space. We develop SM-UCB that efficiently trades off exploration (collecting more data) and exploration (proposing a good action given gathered data) and achieves a $O(\sqrt{T})$ regret bound after $T$ rounds of interaction. Given a bounded-RKHS norm kernel over the context-action-payoff space that governs the smoothness of the utility function, SM-UCB keeps an upper-confidence bound on the payoff function that allows it to asymptotically achieve no-regret. Finally, we evaluate our results on four concrete applications, including movie recommendation (on the MovieLense data set), news recommendation (on Yahoo! Webscope dataset), interactive influence maximization (on a subset of the Facebook network), and personalized data summarization (on Reuters Corpus). In all these applications, we observe that SM-UCB consistently outperforms the prior art.
|
['Lin Chen', 'Andreas Krause', 'Amin Karbasi']
|
2017-12-01
| null | null | null |
neurips-2017-12
|
['movie-recommendation', 'data-summarization']
|
['miscellaneous', 'miscellaneous']
|
[ 2.04699650e-01 5.87122679e-01 -7.14789987e-01 -4.70718592e-01
-8.65401149e-01 -9.10247922e-01 2.06375182e-01 1.48319140e-01
-4.81137633e-01 9.58926737e-01 3.32297772e-01 -3.52880031e-01
-6.40898645e-01 -7.85953462e-01 -1.09956896e+00 -8.89185607e-01
-4.83359188e-01 7.01211870e-01 -3.30501884e-01 -6.38632625e-02
8.03933218e-02 4.42717634e-02 -1.11905932e+00 -3.69041301e-02
6.43970788e-01 1.44992185e+00 3.79278660e-01 4.87608522e-01
2.77577251e-01 4.15032923e-01 -3.92498493e-01 -3.96275252e-01
5.00099659e-01 -2.92953312e-01 -7.58011281e-01 3.86376292e-01
-2.18261912e-01 -5.20950258e-01 -6.69261217e-02 1.15488636e+00
1.87666908e-01 4.64476615e-01 4.88572806e-01 -1.05286419e+00
-2.37749368e-01 1.10796654e+00 -5.15173852e-01 5.60544478e-03
1.78952470e-01 -6.60184324e-02 1.59189034e+00 -2.77928233e-01
7.51449585e-01 9.94799972e-01 -6.53326511e-02 2.23803580e-01
-1.56258774e+00 -3.28861922e-01 6.75755322e-01 -2.84439951e-01
-9.00649667e-01 -3.26180398e-01 5.48816919e-01 -9.81649384e-02
2.88115770e-01 7.93023467e-01 7.53307462e-01 7.09225237e-01
-1.90865934e-01 1.11006689e+00 7.72684515e-01 8.35346803e-02
7.54187107e-01 3.42337698e-01 -1.09893993e-01 4.22091037e-01
1.99867785e-01 -9.09011811e-02 -7.10024118e-01 -4.45362270e-01
5.56498110e-01 1.91360191e-01 -4.77924764e-01 -4.31297898e-01
-9.16928172e-01 1.02420831e+00 3.97356987e-01 -1.90486535e-01
-7.06863999e-01 2.77774066e-01 -1.55983986e-02 7.12771535e-01
7.36696184e-01 7.27658987e-01 -5.73469400e-01 -2.90111840e-01
-7.99966395e-01 2.39470273e-01 1.02829611e+00 6.68793321e-01
6.52946830e-01 -3.66097242e-01 -2.09088117e-01 7.25137651e-01
1.51293874e-01 4.35830772e-01 -3.01841125e-02 -1.32694519e+00
6.51842713e-01 3.15853864e-01 7.75074482e-01 -8.02011251e-01
-2.48432592e-01 -4.10967946e-01 -5.97504914e-01 -3.76981497e-02
2.80846357e-01 -8.98944914e-01 -4.08522487e-01 1.85952878e+00
5.33269584e-01 -8.97936970e-02 -1.99123994e-01 1.25049770e+00
1.78523481e-01 8.76958966e-01 -5.91031253e-01 -8.74449849e-01
9.70197797e-01 -6.29010022e-01 -5.07764518e-01 -2.68520772e-01
5.54943264e-01 -1.77812994e-01 7.51851797e-01 9.31550622e-01
-1.31408966e+00 2.12487414e-01 -9.35454905e-01 2.99952954e-01
1.99404567e-01 -2.22226828e-01 9.40591097e-01 5.94281435e-01
-8.80644023e-01 7.43645072e-01 -6.75902247e-01 -9.40772891e-02
4.49882150e-01 6.72637880e-01 -1.70249492e-01 -1.75687730e-01
-9.82873738e-01 1.95545360e-01 6.96245059e-02 -2.89890952e-02
-1.17588449e+00 -5.75874627e-01 -2.77052730e-01 3.15931886e-01
1.25815201e+00 -5.99204719e-01 1.21677804e+00 -9.92608130e-01
-1.59681535e+00 3.29937756e-01 1.80606276e-01 -6.44486606e-01
6.38348639e-01 -2.45435573e-02 4.30889130e-02 -1.26236290e-01
-6.97966516e-02 3.92336696e-01 7.59034693e-01 -9.64855611e-01
-8.94686282e-01 -6.17092550e-01 8.19761872e-01 6.20612741e-01
-6.09900594e-01 -2.30511382e-01 -4.31808263e-01 -2.99460024e-01
6.06037863e-02 -1.10694170e+00 -6.23791277e-01 -4.24762480e-02
-4.66360301e-01 -2.06000075e-01 2.66895950e-01 -3.41891080e-01
1.21371925e+00 -2.01028109e+00 4.93432164e-01 4.68270004e-01
-5.02361953e-02 -4.45913017e-01 1.56815927e-02 4.23592538e-01
6.32927597e-01 4.04306173e-01 -6.44756407e-02 -5.12780368e-01
7.24460334e-02 2.93531805e-01 -1.29795969e-01 7.21912146e-01
-6.54940784e-01 5.56729853e-01 -7.68561542e-01 1.59933209e-01
-3.83593053e-01 -7.23013878e-02 -8.40654552e-01 -7.97578972e-03
-8.05024028e-01 2.66921580e-01 -8.31250250e-01 3.63323629e-01
3.61603260e-01 -5.87743878e-01 4.88898844e-01 3.80757630e-01
1.42022476e-01 3.53373170e-01 -1.60754001e+00 1.64598930e+00
-5.61537027e-01 2.57056177e-01 6.62235439e-01 -1.14315605e+00
5.28965771e-01 2.29336604e-01 7.39880443e-01 -1.53454825e-01
1.45218700e-01 -4.74369340e-02 -2.72039980e-01 -2.53356695e-01
4.35079247e-01 5.07933348e-02 -1.40190139e-01 9.23548818e-01
-3.27075720e-01 2.11486280e-01 1.74248800e-01 4.42730188e-01
1.12465978e+00 -4.98883605e-01 1.60215184e-01 -2.80708373e-01
-4.85645533e-02 -1.23655647e-01 5.11610985e-01 1.05151665e+00
4.10466999e-01 3.08735430e-01 1.04026020e+00 -7.00406134e-02
-6.49281025e-01 -6.80932760e-01 1.26527073e-02 1.38676810e+00
3.31824660e-01 -1.84291750e-01 -5.27470171e-01 -7.92662144e-01
3.43869209e-01 7.94243574e-01 -7.99765348e-01 1.85595259e-01
7.51771405e-02 -7.19831884e-01 -4.72723335e-01 6.27783984e-02
1.63703382e-01 -7.51441717e-01 -3.30718219e-01 3.32825631e-01
-1.40166730e-01 -7.50169754e-01 -9.69703913e-01 1.67922303e-01
-8.87973964e-01 -6.41723335e-01 -4.29149568e-01 -8.56835395e-02
7.82667756e-01 3.67942959e-01 8.22433949e-01 -2.55120724e-01
2.34319419e-01 6.04166210e-01 -2.17963561e-01 -2.92631209e-01
1.77172720e-01 -1.02082929e-02 1.44049749e-01 5.73461950e-01
-3.09730381e-01 -5.72073936e-01 -9.68061507e-01 5.07515311e-01
-8.20495486e-01 -1.66091129e-01 3.66157025e-01 6.35810554e-01
7.13541210e-01 1.22113109e-01 7.50114918e-01 -1.16119623e+00
8.38548899e-01 -1.04107130e+00 -1.00183225e+00 2.20641658e-01
-6.61605120e-01 4.39760052e-02 4.56060618e-01 -6.57104790e-01
-9.26899612e-01 3.79751027e-02 4.03431177e-01 -5.05649447e-02
4.26185489e-01 1.04626060e+00 -5.57458773e-02 2.44188473e-01
5.65081179e-01 -2.32812464e-02 3.28590162e-02 -5.71014345e-01
4.86568630e-01 7.36735761e-01 8.52938518e-02 -6.76316619e-01
3.01804334e-01 5.71897686e-01 7.19447359e-02 -6.95353448e-01
-9.40911889e-01 -4.14605260e-01 1.33081496e-01 -1.72178432e-01
3.64839792e-01 -8.93066823e-01 -1.21205521e+00 -2.19021052e-01
-7.74242163e-01 -3.41808856e-01 -4.96062636e-01 4.78383511e-01
-5.00634193e-01 -6.70649186e-02 -1.71372086e-01 -1.26326740e+00
-2.96373934e-01 -8.30560148e-01 6.48595393e-01 1.74276769e-01
-8.98591150e-03 -7.99053311e-01 -3.29867095e-01 7.18041062e-01
2.49953598e-01 2.04314560e-01 4.71456140e-01 -5.24706781e-01
-9.99000251e-01 -3.30001324e-01 2.10633084e-01 2.07902282e-01
-1.52008152e-02 -3.90425801e-01 -4.60287720e-01 -5.48981905e-01
-1.05142094e-01 -3.95530999e-01 7.34692574e-01 7.69657195e-01
1.38811600e+00 -1.12473941e+00 -3.05429548e-01 5.28706849e-01
9.97366667e-01 2.75293380e-01 4.00925651e-02 -4.10584435e-02
1.18296228e-01 5.15556037e-01 8.83696735e-01 1.06902182e+00
3.63889247e-01 7.96265781e-01 8.67263436e-01 4.06421989e-01
8.77058506e-01 -2.20904469e-01 6.60132945e-01 1.65731445e-01
-1.33765250e-01 -6.41983092e-01 1.16888294e-02 4.86174971e-01
-2.15928936e+00 -6.23782158e-01 5.23177564e-01 2.69728017e+00
9.33634937e-01 1.72602355e-01 4.22848791e-01 -2.98131615e-01
2.82613009e-01 1.52130619e-01 -1.31378639e+00 -4.02638078e-01
-3.53115015e-02 -3.47969919e-01 8.69812429e-01 6.94727540e-01
-8.23083758e-01 3.99899840e-01 4.58839846e+00 7.51612723e-01
-9.65274572e-01 2.28792980e-01 1.11631942e+00 -1.06418455e+00
-7.07156003e-01 -2.23791674e-02 -7.23182440e-01 5.25875032e-01
6.80601835e-01 -6.20590568e-01 1.31230521e+00 9.18536127e-01
4.09949422e-01 -3.59783828e-01 -1.30881906e+00 7.04951108e-01
-2.45211944e-01 -1.45750332e+00 -5.89728296e-01 6.51046157e-01
1.11928689e+00 2.14420423e-01 2.02182978e-01 2.54234057e-02
7.74940550e-01 -8.68755937e-01 5.95968843e-01 1.39788866e-01
6.87448084e-01 -8.81176114e-01 4.37542289e-01 8.82824898e-01
-5.23512840e-01 -5.03586888e-01 -4.25992876e-01 -1.15564816e-01
3.04858476e-01 8.35792482e-01 -8.86313736e-01 3.79001588e-01
5.48828900e-01 4.07229155e-01 3.62899810e-01 8.84603918e-01
-2.21943155e-01 7.06754327e-01 -8.16535175e-01 -5.67123830e-01
3.79424989e-01 -5.40036321e-01 7.21811056e-01 5.13705432e-01
4.62109923e-01 5.79089046e-01 3.79614085e-01 6.05462670e-01
-6.91146374e-01 3.20865333e-01 -2.21705183e-01 -1.85100734e-01
4.67502385e-01 1.29782879e+00 -4.41283047e-01 -5.74816077e-04
-9.41008478e-02 7.93966830e-01 1.67904049e-01 5.85254848e-01
-5.74016094e-01 2.10998997e-01 9.73163128e-01 3.05231661e-01
4.74102318e-01 1.83335200e-01 -3.48445565e-01 -9.94849086e-01
1.36284366e-01 -6.17705822e-01 6.62796438e-01 -4.16042715e-01
-9.51893330e-01 9.59279537e-02 -1.04041792e-01 -7.29278624e-01
-2.35671997e-01 -4.70210165e-02 -4.77727085e-01 5.70616007e-01
-1.00142264e+00 -3.59670758e-01 3.62409294e-01 3.88195723e-01
3.91512901e-01 4.74819057e-02 3.07442069e-01 -4.16829996e-02
-4.48353142e-01 4.41959083e-01 5.28526962e-01 -4.93681252e-01
2.89075971e-01 -1.24654841e+00 -7.18705654e-02 4.40020919e-01
2.42434666e-01 4.23570722e-01 8.13765168e-01 -3.84236276e-01
-1.85245967e+00 -8.44869733e-01 3.43802422e-01 -9.56628397e-02
6.50436044e-01 -5.01618922e-01 -3.38984311e-01 7.44895697e-01
-2.00666234e-01 7.07976297e-02 4.67137605e-01 5.04453838e-01
2.51773000e-01 -4.49503511e-01 -1.35055113e+00 6.25982523e-01
1.18370616e+00 2.19876450e-02 3.62958699e-01 6.85208261e-01
8.66832614e-01 -3.05816472e-01 -8.22254241e-01 1.34094387e-01
5.83562791e-01 -6.44487202e-01 6.18697524e-01 -9.49781895e-01
3.60399038e-01 -6.78225085e-02 -3.24259371e-01 -1.64447033e+00
-1.28130585e-01 -1.33454680e+00 -4.52453196e-01 8.92535031e-01
8.20009947e-01 -6.29423916e-01 1.14534009e+00 9.98014212e-01
3.92046124e-01 -1.24008036e+00 -1.16357613e+00 -5.40041029e-01
-3.43060195e-01 -3.20843637e-01 5.74890912e-01 5.96090436e-01
2.72839248e-01 3.56697857e-01 -8.59193027e-01 2.66578138e-01
6.47314131e-01 4.18645352e-01 7.48649299e-01 -1.11380887e+00
-9.05263901e-01 -1.84189916e-01 2.56277502e-01 -1.84910321e+00
-2.51460463e-01 -7.49047637e-01 -1.00212343e-01 -1.35836434e+00
3.49979788e-01 -6.24256313e-01 -3.25899720e-01 2.38930419e-01
4.68800478e-02 -3.43086809e-01 2.79821187e-01 -9.47097782e-03
-9.66485023e-01 3.76847893e-01 1.32794309e+00 -1.27261490e-01
-5.95230103e-01 8.00540149e-01 -1.26444936e+00 3.94636482e-01
5.12840033e-01 -5.23478627e-01 -6.87747121e-01 -2.53066450e-01
8.43989015e-01 9.68057215e-01 -1.97798073e-01 -9.30017382e-02
4.01393354e-01 -6.23028398e-01 -2.60550141e-01 -2.91078478e-01
5.53849578e-01 -8.84465814e-01 2.37713292e-01 1.39883667e-01
-7.38755107e-01 -5.91670454e-01 -5.44428885e-01 1.01061630e+00
2.02927575e-01 -2.56327271e-01 3.85595620e-01 -4.73726429e-02
7.11604580e-02 6.67652011e-01 -7.87600949e-02 6.80906847e-02
9.17625308e-01 1.69648930e-01 -2.35506117e-01 -1.24762893e+00
-7.62199342e-01 9.46262419e-01 1.99420378e-01 1.37258559e-01
3.89367729e-01 -1.08584940e+00 -7.30603397e-01 -2.84089208e-01
-1.96059853e-01 2.13526428e-01 2.64334351e-01 8.35602820e-01
3.55073899e-01 2.15935841e-01 4.19997662e-01 -3.09458107e-01
-1.24141228e+00 4.87013936e-01 5.61243556e-02 -3.52205038e-01
5.74836135e-02 1.07564032e+00 1.86728865e-01 -1.06737137e-01
6.35179877e-01 -3.04375917e-01 -1.15948759e-01 2.99535662e-01
4.32945132e-01 5.41906476e-01 -1.88766733e-01 1.57654464e-01
8.12910572e-02 -1.90260127e-01 -1.12432219e-01 -4.21301216e-01
1.65560424e+00 -3.94769907e-01 -9.53424945e-02 4.15131480e-01
1.13896632e+00 6.92690630e-03 -1.49950945e+00 -5.94011068e-01
-2.72711009e-01 -5.40238500e-01 2.88535029e-01 -9.70902622e-01
-1.28838623e+00 1.83033898e-01 8.94018486e-02 8.63409579e-01
1.06236351e+00 2.67686814e-01 5.99675417e-01 6.88274920e-01
5.87219536e-01 -1.31134093e+00 -1.67589635e-01 -3.58368978e-02
1.04265213e+00 -1.34213793e+00 2.43732557e-01 -1.30410805e-01
-8.95402789e-01 7.01230049e-01 6.43485188e-02 1.47076786e-01
9.65772748e-01 1.54864006e-02 -5.57898223e-01 -9.95547771e-02
-1.18600321e+00 2.08873644e-01 2.64302552e-01 -1.27091840e-01
-5.53795286e-02 6.36403739e-01 -3.73154283e-01 9.68874276e-01
-8.14288110e-02 -1.04958162e-01 6.30955935e-01 4.86676008e-01
-7.60629416e-01 -8.59466553e-01 -2.22313657e-01 1.01963878e+00
-7.15196133e-01 1.47131950e-01 -4.32524920e-01 2.23431773e-02
-2.97950357e-01 1.27833140e+00 -1.46350473e-01 -1.98464498e-01
1.71212986e-01 -5.24875939e-01 8.09826031e-02 -6.24683201e-01
-3.69023830e-01 2.83328086e-01 4.56144094e-01 -6.65230870e-01
-1.81803226e-01 -9.64649379e-01 -7.95372486e-01 -4.24415231e-01
-4.31506366e-01 3.94842893e-01 8.94730747e-01 9.08199608e-01
3.98393005e-01 5.73597178e-02 1.26767540e+00 -6.27725244e-01
-1.09181190e+00 -8.25534046e-01 -1.00083041e+00 1.58896506e-01
3.52694631e-01 -3.03047538e-01 -5.44471860e-01 -3.20107967e-01]
|
[4.7088117599487305, 3.454927921295166]
|
ea79cf11-793c-46a5-b626-fba0eed4dc85
|
open-set-representation-learning-through
|
2106.15278
| null |
https://arxiv.org/abs/2106.15278v3
|
https://arxiv.org/pdf/2106.15278v3.pdf
|
Open-Set Representation Learning through Combinatorial Embedding
|
Visual recognition tasks are often limited to dealing with a small subset of classes simply because the labels for the remaining classes are unavailable. We are interested in identifying novel concepts in a dataset through representation learning based on both labeled and unlabeled examples, and extending the horizon of recognition to both known and novel classes. To address this challenging task, we propose a combinatorial learning approach, which naturally clusters the examples in unseen classes using the compositional knowledge given by multiple supervised meta-classifiers on heterogeneous label spaces. The representations given by the combinatorial embedding are made more robust by unsupervised pairwise relation learning. The proposed algorithm discovers novel concepts via a joint optimization for enhancing the discrimitiveness of unseen classes as well as learning the representations of known classes generalizable to novel ones. Our extensive experiments demonstrate remarkable performance gains by the proposed approach on public datasets for image retrieval and image categorization with novel class discovery.
|
['Bohyung Han', 'Junoh Kang', 'Geeho Kim']
|
2021-06-29
|
open-set-representation-learning-through-1
|
http://openaccess.thecvf.com//content/CVPR2023/html/Kim_Open-Set_Representation_Learning_Through_Combinatorial_Embedding_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Kim_Open-Set_Representation_Learning_Through_Combinatorial_Embedding_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['novel-class-discovery', 'image-categorization', 'novel-class-discovery', 'novel-concepts']
|
['computer-vision', 'computer-vision', 'methodology', 'reasoning']
|
[ 5.43105364e-01 1.03317983e-02 -5.67984879e-01 -6.25131607e-01
-9.84630704e-01 -8.04366291e-01 7.68183649e-01 5.64574182e-01
-2.13271663e-01 7.32047617e-01 8.07966739e-02 1.56147435e-01
-5.15842736e-01 -6.25595033e-01 -5.63044250e-01 -1.02730405e+00
-1.40387073e-01 6.66709840e-01 -2.73933876e-02 2.87074417e-01
3.03349316e-01 5.79346776e-01 -1.88631189e+00 6.26437068e-01
5.64424634e-01 1.17604446e+00 -7.17730960e-03 2.77229428e-01
-4.23444510e-01 5.61100185e-01 -2.50659585e-01 3.87673005e-02
4.29580003e-01 -1.88859060e-01 -8.21848035e-01 4.32457387e-01
5.18954933e-01 2.27190435e-01 -8.94755423e-02 1.07122564e+00
1.66109294e-01 2.52845258e-01 9.67375517e-01 -1.44266367e+00
-9.62289572e-01 4.08631921e-01 -6.23642385e-01 1.20072275e-01
1.55039638e-01 -2.12689131e-01 1.41027820e+00 -1.18828905e+00
7.03253210e-01 1.20213985e+00 1.94672063e-01 5.71607947e-01
-1.57838774e+00 -8.79200578e-01 4.93086398e-01 5.39676070e-01
-1.70360196e+00 -3.48057538e-01 9.93163526e-01 -5.80829918e-01
5.23091853e-01 3.27478737e-01 3.55833285e-02 9.92768109e-01
-3.77804160e-01 7.63515651e-01 1.14071453e+00 -4.36563820e-01
3.14581156e-01 5.90682149e-01 6.03876173e-01 4.93132204e-01
2.39487320e-01 1.49662390e-01 -5.14813960e-01 -2.84216493e-01
-1.13696232e-01 5.21222889e-01 -1.36242405e-01 -9.58310425e-01
-1.26209784e+00 9.10244167e-01 6.19214416e-01 4.28757846e-01
-1.90234616e-01 -2.66359121e-01 2.85902292e-01 3.04608613e-01
5.67942798e-01 6.69109344e-01 -5.70173979e-01 6.18580997e-01
-6.61361694e-01 -1.63216278e-01 4.18845743e-01 1.21956730e+00
1.36191392e+00 -3.98840964e-01 -1.36392698e-01 9.52536464e-01
1.95767283e-01 2.72800893e-01 7.64768779e-01 -3.93101573e-01
1.63680777e-01 9.98232365e-01 -1.99973822e-01 -1.04099774e+00
-3.40331942e-01 -4.85593498e-01 -6.89194739e-01 -6.82386458e-02
8.36606324e-02 2.16798812e-01 -1.00370622e+00 1.70040846e+00
4.79244858e-01 1.85022429e-01 3.22716564e-01 5.64767420e-01
8.84310782e-01 7.42429316e-01 8.74240175e-02 -3.40249360e-01
1.04428720e+00 -9.38039720e-01 -3.61352921e-01 4.23335060e-02
5.57796597e-01 -6.14562452e-01 5.85278034e-01 1.34174287e-01
-3.35777938e-01 -7.42972672e-01 -1.03207552e+00 1.27183571e-01
-6.90035105e-01 1.62290215e-01 6.13432348e-01 3.21061015e-01
-5.45323372e-01 3.09737265e-01 -4.29379493e-01 -4.36322629e-01
7.44375467e-01 5.10881722e-01 -5.58558822e-01 -5.03238916e-01
-5.40562749e-01 5.05714118e-01 9.14506793e-01 -8.23780298e-02
-9.66081798e-01 -6.04930818e-01 -8.34865332e-01 7.09732696e-02
5.50839186e-01 -3.49194199e-01 5.27453423e-01 -1.21642637e+00
-8.51722836e-01 9.74084198e-01 -1.58901617e-01 -2.13782355e-01
1.51677965e-03 2.81717628e-01 -5.91781199e-01 3.18516761e-01
2.51257867e-01 7.82959163e-01 1.20705068e+00 -1.62361276e+00
-7.84986258e-01 -5.77070236e-01 -1.11231677e-01 1.24607787e-01
-7.29461312e-01 -3.12933385e-01 1.42003998e-01 -5.49928248e-01
6.14355028e-01 -8.77899706e-01 -1.07850060e-01 1.44170389e-01
-3.78715098e-01 -5.06301343e-01 1.20998383e+00 -1.47633284e-01
6.90620124e-01 -2.29001260e+00 2.63136923e-01 5.43812096e-01
4.95562375e-01 5.42771295e-02 -4.82470065e-01 3.52986485e-01
-4.35612321e-01 1.05864950e-01 -2.09144115e-01 2.86517311e-02
-5.63702248e-02 3.26888263e-01 -5.33954918e-01 3.54292989e-01
5.21973073e-01 7.65614927e-01 -1.03808987e+00 -4.65499789e-01
9.44486484e-02 5.80179878e-02 -1.73400119e-01 4.65451807e-01
-1.73960835e-01 4.75572526e-01 -5.58795154e-01 1.01459765e+00
6.01648092e-01 -3.18072289e-01 4.17255998e-01 -1.99554145e-01
2.43660361e-01 -1.36610821e-01 -1.23212397e+00 1.54222357e+00
-2.29255542e-01 3.41808230e-01 -4.97806013e-01 -1.48654938e+00
1.07132411e+00 8.29077438e-02 4.28457886e-01 -2.91069090e-01
-7.29010701e-02 2.55438417e-01 -6.60110870e-03 -6.64582849e-01
1.03020146e-01 -3.93736124e-01 1.39151037e-01 4.32619780e-01
4.81830448e-01 4.02887106e-01 7.32343346e-02 2.27570847e-01
8.18083167e-01 -2.41552666e-01 4.02753234e-01 -2.88453013e-01
5.32186627e-01 7.50084147e-02 5.93847752e-01 8.45728278e-01
-4.20888886e-02 4.25384432e-01 1.64254516e-01 -6.07639372e-01
-8.99386823e-01 -1.31688488e+00 -4.85702097e-01 1.07589173e+00
3.14792663e-01 -2.92469829e-01 5.07936329e-02 -1.12677598e+00
1.81263074e-01 2.46789321e-01 -8.70305121e-01 -3.67545575e-01
-4.63012159e-02 -7.64455497e-01 7.01670572e-02 4.83433098e-01
1.67453513e-01 -8.37453544e-01 -8.19189623e-02 -6.24692775e-02
1.61303744e-01 -9.52218354e-01 -1.83231369e-01 4.22520995e-01
-8.90060663e-01 -1.36299849e+00 -5.55077434e-01 -1.29944372e+00
1.06093156e+00 4.79061395e-01 5.88620722e-01 8.66546407e-02
-7.61556506e-01 5.56807399e-01 -6.79524064e-01 -7.74627104e-02
-2.19790339e-01 -1.69028357e-01 3.57341975e-01 6.93303227e-01
5.13801932e-01 -4.92860407e-01 -2.58066893e-01 3.31190616e-01
-9.49292481e-01 -2.90939838e-01 6.33515120e-01 1.26368582e+00
6.26150072e-01 2.63025582e-01 8.92246723e-01 -9.49172914e-01
5.08149192e-02 -8.46391737e-01 -3.99879336e-01 6.08364403e-01
-6.53144777e-01 2.57557392e-01 4.77538258e-01 -9.13557410e-01
-9.01850283e-01 2.43303329e-01 6.43047929e-01 -4.41565186e-01
-3.63475293e-01 5.32275736e-01 -2.60969251e-01 -2.64753819e-01
7.52792418e-01 2.99529076e-01 -7.78152421e-02 -4.48754519e-01
7.39352286e-01 6.90082550e-01 2.83660203e-01 -8.10575783e-01
1.05592430e+00 5.04517674e-01 8.56360570e-02 -9.54899013e-01
-8.67168725e-01 -1.01245189e+00 -1.03099287e+00 4.78335842e-02
6.48180187e-01 -8.72162879e-01 -4.43289250e-01 -1.01502798e-01
-8.02399099e-01 5.06517768e-01 -5.87997615e-01 5.63255250e-01
-2.99819767e-01 5.91039121e-01 4.74171713e-02 -6.98089063e-01
5.30740879e-02 -1.03754604e+00 1.01424515e+00 1.77102923e-01
-8.05776119e-02 -7.94899106e-01 1.16968490e-01 3.32264870e-01
-1.12131359e-02 2.21732020e-01 1.42530143e+00 -1.31094873e+00
-8.43495250e-01 -4.89081472e-01 -4.09959614e-01 4.00113285e-01
4.26953763e-01 -3.42197150e-01 -1.22730434e+00 -5.58724940e-01
-2.97494233e-01 -7.84062803e-01 1.14424253e+00 -2.04270352e-02
1.28387535e+00 -3.23004425e-01 -7.05297291e-01 3.89741838e-01
1.39673126e+00 2.36524507e-01 1.31472379e-01 9.64655504e-02
7.67594159e-01 8.87681365e-01 4.99193192e-01 1.94991976e-01
7.21553490e-02 4.58330840e-01 1.19543970e-01 1.04358450e-01
7.32921883e-02 -1.90078214e-01 -4.33323421e-02 6.33862078e-01
3.49897087e-01 -2.95077525e-02 -8.07208896e-01 6.64288521e-01
-1.76280236e+00 -8.69699061e-01 3.20018798e-01 2.24656677e+00
7.44772255e-01 -2.91490167e-01 -5.67271449e-02 1.28954068e-01
1.01086116e+00 -7.80926794e-02 -5.79596162e-01 4.72468697e-02
-3.05391490e-01 2.12551564e-01 3.73625457e-02 1.13303974e-01
-1.30937243e+00 6.82000458e-01 5.76568079e+00 8.62291336e-01
-8.31831932e-01 -4.72451858e-02 6.03240192e-01 1.24053903e-01
-2.59676695e-01 2.88215220e-01 -7.31310129e-01 -1.33682052e-02
4.66483027e-01 -2.52297491e-01 1.38150156e-01 8.20953727e-01
-4.65956092e-01 1.20537020e-01 -1.52582479e+00 1.20587564e+00
4.94569391e-01 -1.46428025e+00 6.65326297e-01 -1.28416792e-01
1.12576771e+00 -2.25704014e-01 2.69258022e-01 3.64137203e-01
8.13047290e-02 -8.43859136e-01 2.43191883e-01 6.08386397e-01
6.30039096e-01 -6.32029593e-01 5.57619572e-01 3.21947396e-01
-1.02842021e+00 -5.15934289e-01 -6.07694924e-01 1.65301755e-01
-6.30270660e-01 2.55632222e-01 -9.97295499e-01 4.73449022e-01
6.15723431e-01 1.22731078e+00 -9.82018590e-01 1.18955517e+00
2.12401338e-02 3.15668494e-01 -9.14968848e-02 -9.36371461e-03
2.08311439e-01 -2.06476510e-01 3.97903800e-01 7.93732285e-01
-1.37357898e-02 2.46677190e-01 6.27451241e-01 9.28871632e-01
-4.19693023e-01 3.19040030e-01 -8.20086181e-01 -3.64549339e-01
3.06892216e-01 1.41937518e+00 -8.40282142e-01 -4.09861714e-01
-2.70763546e-01 8.08579624e-01 5.23356140e-01 6.00174546e-01
-1.65345490e-01 -3.96865755e-01 4.03797865e-01 -3.78063649e-01
2.01816455e-01 -6.64287899e-03 -2.64866762e-02 -1.34337997e+00
2.05297992e-01 -5.95156193e-01 8.49273682e-01 -4.56447870e-01
-1.92241812e+00 5.71121991e-01 9.47063491e-02 -1.47966242e+00
1.37281362e-02 -7.58731127e-01 -3.24270844e-01 6.20687425e-01
-1.79858637e+00 -1.51377964e+00 -1.18034452e-01 6.08127534e-01
4.67960328e-01 -6.08388901e-01 1.02056527e+00 1.17720373e-01
-2.12701872e-01 6.78690910e-01 5.60068488e-01 4.11368208e-03
7.89635241e-01 -1.14164984e+00 -4.92059618e-01 4.66623127e-01
5.52165329e-01 6.29862309e-01 2.05944061e-01 -4.01753575e-01
-1.16803169e+00 -1.19839251e+00 9.40804422e-01 -4.70506132e-01
7.53535926e-01 -5.67107797e-01 -1.01825976e+00 5.51987290e-01
-1.26642957e-01 3.48261178e-01 1.24370611e+00 3.62163305e-01
-1.05191708e+00 -1.97924212e-01 -1.14867198e+00 2.25833893e-01
8.14386427e-01 -8.51409733e-01 -7.89577782e-01 6.96223795e-01
7.61075795e-01 2.76277751e-01 -6.94052875e-01 3.71247470e-01
4.84043956e-01 -1.99366987e-01 1.00845754e+00 -1.36777151e+00
1.97287515e-01 -5.02812505e-01 -7.02399909e-01 -1.15678084e+00
-3.02606970e-01 -2.47470941e-02 2.20790461e-01 1.34116995e+00
5.39666831e-01 -6.01363361e-01 6.45626247e-01 3.28592002e-01
8.15985426e-02 -4.76456642e-01 -8.82520199e-01 -9.88295972e-01
3.81887844e-03 -1.36069000e-01 4.08914179e-01 1.40902829e+00
2.65788436e-02 4.91943777e-01 -2.96414554e-01 4.09415454e-01
8.93088996e-01 8.14401686e-01 5.43188572e-01 -1.47942638e+00
-2.72576511e-01 -5.74495941e-02 -1.08190513e+00 -5.99801719e-01
5.92625618e-01 -1.62205410e+00 -1.75256785e-02 -9.10342872e-01
9.23700988e-01 -5.95736146e-01 -6.12523854e-01 6.50575519e-01
-1.14524908e-01 4.29216713e-01 1.58344984e-01 4.60826606e-01
-8.19573283e-01 7.66923547e-01 7.20125854e-01 -8.29998553e-01
-5.91646060e-02 -1.11205950e-01 -7.03032255e-01 6.09882951e-01
4.44762230e-01 -6.48388386e-01 -3.73839647e-01 -2.95920372e-01
-1.29810140e-01 -4.30608451e-01 5.23913085e-01 -8.04787517e-01
2.72160202e-01 -2.25655049e-01 5.97963333e-01 -4.36871946e-01
2.97328204e-01 -1.18881452e+00 -9.95231867e-02 2.10319668e-01
-8.97036791e-01 -4.45778191e-01 5.61577491e-02 1.11520028e+00
-2.75914758e-01 -2.62339324e-01 8.82806838e-01 1.07582673e-01
-1.09965527e+00 4.69089568e-01 7.19292695e-03 -4.96148355e-02
1.25961542e+00 -2.04586521e-01 -3.51259381e-01 -1.27290599e-02
-1.34215844e+00 2.52730727e-01 5.00602946e-02 7.38153994e-01
9.19151545e-01 -1.47118950e+00 -5.31587303e-01 3.84351403e-01
1.03518569e+00 -3.53450507e-01 1.60500124e-01 2.95910776e-01
1.02706507e-01 2.97248393e-01 -1.88429669e-01 -8.90641868e-01
-1.34701037e+00 1.13599050e+00 1.08302772e-01 1.40885711e-01
-4.30414826e-01 8.67598653e-01 4.74877387e-01 -7.40314186e-01
2.06410989e-01 7.42012635e-02 -2.91904688e-01 3.96501184e-01
4.17903572e-01 1.89431876e-01 -8.02793652e-02 -8.46417248e-01
-3.80234808e-01 7.15065718e-01 -4.07304049e-01 4.19347644e-01
1.32341027e+00 -1.91334575e-01 -2.83030123e-01 7.00518191e-01
1.82407153e+00 -4.16455537e-01 -6.96699440e-01 -8.74478281e-01
4.50687766e-01 -6.30676448e-01 -2.29652584e-01 -7.74649382e-01
-7.80772209e-01 8.70765150e-01 9.37458098e-01 -2.76456594e-01
1.16366482e+00 4.30818647e-01 2.40957588e-01 9.21571910e-01
3.33432525e-01 -8.02736759e-01 3.84047002e-01 2.10697025e-01
6.63613081e-01 -1.62937069e+00 -2.77465861e-02 -5.21737456e-01
-2.74010271e-01 1.37683570e+00 3.82755905e-01 3.23976651e-02
7.60382175e-01 -6.21935487e-01 -1.69452518e-01 -3.03220034e-01
-6.57190919e-01 -3.34574819e-01 7.14880109e-01 6.96081758e-01
-1.64019428e-02 1.94360092e-01 -1.29626974e-01 4.30652350e-01
4.52738434e-01 -5.72880626e-01 2.86130756e-01 9.37316477e-01
-4.74057525e-01 -1.18883729e+00 -2.06771612e-01 6.44410670e-01
9.23387986e-03 -5.13394885e-02 -5.51456273e-01 5.36384284e-01
3.80009115e-01 9.97319102e-01 -4.21405360e-02 -4.81676102e-01
1.72601473e-02 3.14241409e-01 3.17801028e-01 -1.03334224e+00
-9.07742530e-02 1.99562609e-02 -2.55654752e-01 -1.58475354e-01
-5.78153431e-01 -5.36209524e-01 -9.45372522e-01 4.93944198e-01
-5.00120103e-01 3.46728534e-01 4.03046072e-01 1.12063479e+00
2.03783140e-01 1.69497773e-01 1.20566249e+00 -5.87398708e-01
-8.10438752e-01 -5.40807843e-01 -9.04021263e-01 5.93634546e-01
4.53822345e-01 -9.01641607e-01 -5.26625872e-01 2.27992669e-01]
|
[9.653130531311035, 2.9196720123291016]
|
635b0cd3-2889-49e7-b160-198526a3a7c6
|
cat-nerf-constancy-aware-tx-2-former-for
|
2304.07915
| null |
https://arxiv.org/abs/2304.07915v1
|
https://arxiv.org/pdf/2304.07915v1.pdf
|
CAT-NeRF: Constancy-Aware Tx$^2$Former for Dynamic Body Modeling
|
This paper addresses the problem of human rendering in the video with temporal appearance constancy. Reconstructing dynamic body shapes with volumetric neural rendering methods, such as NeRF, requires finding the correspondence of the points in the canonical and observation space, which demands understanding human body shape and motion. Some methods use rigid transformation, such as SE(3), which cannot precisely model each frame's unique motion and muscle movements. Others generate the transformation for each frame with a trainable network, such as neural blend weight field or translation vector field, which does not consider the appearance constancy of general body shape. In this paper, we propose CAT-NeRF for self-awareness of appearance constancy with Tx$^2$Former, a novel way to combine two Transformer layers, to separate appearance constancy and uniqueness. Appearance constancy models the general shape across the video, and uniqueness models the unique patterns for each frame. We further introduce a novel Covariance Loss to limit the correlation between each pair of appearance uniquenesses to ensure the frame-unique pattern is maximally captured in appearance uniqueness. We assess our method on H36M and ZJU-MoCap and show state-of-the-art performance.
|
['Ram Nevatia', 'Wanrong Zheng', 'Zhaoheng Zheng', 'Haidong Zhu']
|
2023-04-16
| null | null | null | null |
['neural-rendering']
|
['computer-vision']
|
[-1.12126902e-01 -1.13466583e-01 -3.56119387e-02 -4.22588855e-01
1.48687571e-01 -2.65423328e-01 3.00501198e-01 -6.05759799e-01
-1.92610145e-01 4.29812580e-01 2.48957679e-01 3.56414735e-01
2.09430411e-01 -5.41059732e-01 -8.57577443e-01 -5.56454778e-01
1.90251525e-02 2.73339182e-01 2.88568318e-01 -4.70280468e-01
-4.18032974e-01 1.40265316e-01 -1.41479993e+00 1.73248902e-01
6.02565169e-01 1.20267570e+00 2.46757152e-03 6.94319129e-01
-6.57265410e-02 6.07235909e-01 -1.86530352e-01 -5.11237383e-01
6.04009449e-01 -7.28602767e-01 -7.43095517e-01 2.36656204e-01
1.21869838e+00 -3.96068871e-01 -5.47214389e-01 9.48254585e-01
4.39702183e-01 3.78891319e-01 6.70588374e-01 -1.26857531e+00
-1.00124848e+00 -4.23203930e-02 -7.27624536e-01 2.60676026e-01
3.05356741e-01 9.18798149e-02 7.07420468e-01 -5.40015101e-01
9.35701609e-01 1.49764228e+00 9.58418190e-01 1.22557926e+00
-1.17468357e+00 -7.21099198e-01 3.76298606e-01 3.37604620e-02
-1.07821918e+00 -4.79915410e-01 8.30852091e-01 -4.65502203e-01
4.13139313e-01 5.69807053e-01 1.03507650e+00 1.03973818e+00
5.95173061e-01 6.63600504e-01 9.25305367e-01 -4.64153364e-02
-2.08476603e-01 -4.90692347e-01 -2.02720106e-01 1.04681778e+00
-1.03913873e-01 6.82296082e-02 -7.37160027e-01 9.91012603e-02
1.55579066e+00 5.98872341e-02 -4.27231282e-01 -5.73129535e-01
-1.24217272e+00 4.05625701e-01 4.62673426e-01 1.22184874e-02
-1.23240963e-01 6.14078879e-01 2.51325577e-01 3.51347923e-01
4.44109619e-01 1.16851069e-01 -4.47616965e-01 1.22617006e-01
-7.15202928e-01 2.72487104e-01 2.56996512e-01 1.04664230e+00
4.83277559e-01 4.21919614e-01 -3.58285338e-01 7.42995024e-01
5.42299628e-01 6.22728348e-01 8.23399365e-01 -1.17730725e+00
1.69383347e-01 3.84636641e-01 -1.32965878e-01 -1.25058985e+00
-5.07284045e-01 -2.08106175e-01 -1.35258937e+00 3.53024513e-01
4.24671263e-01 -8.45642909e-02 -1.14586651e+00 2.29879713e+00
6.95357084e-01 5.74999928e-01 -2.47214943e-01 1.16738117e+00
1.21051013e+00 3.36573988e-01 9.65493992e-02 -1.56192794e-01
1.37169719e+00 -9.72380698e-01 -8.95583630e-01 -1.16508625e-01
-1.00588091e-01 -5.85020363e-01 9.42452431e-01 1.65798906e-02
-1.46827221e+00 -9.11229908e-01 -8.77281129e-01 -2.86153525e-01
8.98319632e-02 -8.68085399e-02 8.26843500e-01 4.27401692e-01
-1.09876060e+00 7.53661633e-01 -1.13233471e+00 -2.25340113e-01
1.83402225e-01 3.84102136e-01 -6.47074878e-01 3.44000846e-01
-1.03054333e+00 6.57523453e-01 -9.10921674e-03 3.49246234e-01
-5.63405633e-01 -7.22173393e-01 -1.08687568e+00 -3.31490964e-01
-1.87632088e-02 -1.38956320e+00 9.16283607e-01 -1.53376198e+00
-1.73353446e+00 9.71240819e-01 -2.40365490e-01 -1.40565157e-01
7.96594441e-01 -7.88859278e-02 -5.52522659e-01 -1.53746262e-01
-6.35619238e-02 8.33500385e-01 1.09083390e+00 -1.14495671e+00
-4.75871503e-01 -4.62306201e-01 -1.36383682e-01 4.14778978e-01
2.92803757e-02 -6.57563508e-02 -7.61984885e-01 -1.07016504e+00
7.88702965e-01 -1.19230747e+00 -6.86586648e-02 6.54493272e-01
-2.71369237e-02 3.13135870e-02 6.35242224e-01 -1.00631428e+00
1.04896903e+00 -2.01048899e+00 6.14346683e-01 1.41306922e-01
4.37412560e-01 -2.85330564e-01 -4.42478992e-02 -3.86875510e-01
-1.57663226e-01 -7.48312026e-02 -2.05531836e-01 -7.05387950e-01
-6.41883314e-02 5.53460002e-01 -7.36012459e-02 7.51815081e-01
4.82294150e-02 9.72202897e-01 -8.19693089e-01 -6.20380521e-01
1.43357605e-01 5.53430736e-01 -6.54610693e-01 3.52200031e-01
-1.18466891e-01 8.34964573e-01 -2.66838461e-01 3.59519392e-01
7.67974854e-01 -2.05389962e-01 1.81263953e-01 -5.55679202e-01
1.79651335e-01 -3.49524677e-01 -1.44848180e+00 2.08015752e+00
-2.82986492e-01 4.71781760e-01 1.22633725e-02 -4.68136251e-01
6.52558446e-01 3.28443289e-01 6.56863391e-01 -8.16969812e-01
1.72318295e-01 -1.32645875e-01 -5.26380353e-02 -4.04430121e-01
5.02180457e-01 -4.03156668e-01 2.17270851e-01 1.74616680e-01
1.60671905e-01 8.33024234e-02 -1.95936546e-01 -1.86112419e-01
5.67031920e-01 5.72826743e-01 -2.55629402e-02 -1.36360660e-01
3.60424072e-01 -6.68768466e-01 8.55872989e-01 2.99330980e-01
-2.25202605e-01 1.22818553e+00 -1.15511276e-01 -9.78058279e-01
-9.24884260e-01 -1.31542122e+00 -4.32354510e-02 1.04868698e+00
5.47509968e-01 -1.43685251e-01 -5.34748554e-01 -3.13422531e-01
-5.42834774e-03 1.49997994e-01 -1.04103827e+00 -2.28787109e-01
-9.13298309e-01 -6.41479194e-01 2.21886158e-01 6.21488750e-01
6.70388281e-01 -1.09091163e+00 -7.35128045e-01 1.72994271e-01
-5.52595854e-01 -9.48836148e-01 -9.98358965e-01 -3.47589880e-01
-8.52735877e-01 -9.15731370e-01 -8.88346612e-01 -6.66738868e-01
7.26888359e-01 -6.38980567e-02 1.18501770e+00 2.93235362e-01
-2.87519842e-01 4.52253819e-01 -2.66631739e-03 -6.40835389e-02
-3.05250883e-01 -4.25150931e-01 3.55184227e-01 2.54955083e-01
-3.48189622e-01 -7.72311389e-01 -8.37397754e-01 4.53907341e-01
-6.82559669e-01 3.27229500e-01 8.93970877e-02 7.26518631e-01
9.84634697e-01 -3.31993103e-01 -1.39136449e-01 -4.03589725e-01
2.59375244e-01 -1.49459407e-01 -3.63618210e-02 4.30973500e-01
-3.20664734e-01 1.40545502e-01 3.37983996e-01 -8.75930548e-01
-9.68282580e-01 4.72373031e-02 -1.83268607e-01 -9.05551493e-01
1.45178780e-01 -2.45753825e-02 -8.16665590e-02 -2.15645686e-01
4.73313332e-01 4.48550373e-01 2.62961507e-01 -4.28866923e-01
4.63589579e-01 -2.22312137e-01 1.04768932e+00 -7.52514362e-01
7.24404454e-01 7.39571869e-01 1.76556125e-01 -5.89461982e-01
-5.59126973e-01 -4.02456522e-01 -8.58242452e-01 -4.69066054e-01
1.18091142e+00 -9.02810037e-01 -8.30562532e-01 7.17345357e-01
-1.18985772e+00 -3.89519155e-01 -4.27256972e-01 4.44072992e-01
-8.34745705e-01 6.43364251e-01 -6.25778496e-01 -5.00508964e-01
-5.33308506e-01 -1.00559115e+00 1.14794767e+00 2.96271801e-01
-3.03606749e-01 -9.28057671e-01 3.91148686e-01 -5.01376279e-02
2.74341136e-01 7.25160360e-01 6.05150461e-01 2.68004656e-01
-3.36686492e-01 1.62002534e-01 -1.47571778e-02 1.50007010e-01
3.75619203e-01 -2.94433217e-02 -6.95902884e-01 -4.83227223e-01
2.85097342e-02 -1.20492373e-02 6.97020233e-01 7.55154669e-01
1.23768139e+00 -4.28127915e-01 -3.39121632e-02 1.35919893e+00
1.12634623e+00 2.17268001e-02 7.69554853e-01 1.81971207e-01
1.12043691e+00 4.49083000e-01 1.29625648e-01 4.11736369e-01
7.03616142e-01 9.64513421e-01 3.38788509e-01 -4.17917311e-01
-5.81858337e-01 -2.87842810e-01 2.18217209e-01 1.11917448e+00
-8.28557253e-01 3.03323623e-02 -3.06527317e-01 1.25679046e-01
-1.95291150e+00 -1.10298932e+00 4.18488272e-02 2.22572637e+00
8.97401154e-01 -2.24387810e-01 2.66131490e-01 -4.96002138e-01
5.57774663e-01 2.23623440e-01 -5.38583994e-01 -2.07656816e-01
-2.23106876e-01 1.10630922e-01 3.15131485e-01 4.54153329e-01
-1.09236169e+00 7.85966396e-01 6.45682144e+00 5.90680778e-01
-1.17085588e+00 2.27119327e-01 5.59090197e-01 -1.46869123e-01
-3.79234344e-01 -4.61098999e-01 -4.75161880e-01 5.37295818e-01
3.80039364e-01 8.90396163e-02 4.60323989e-01 5.64520299e-01
-4.65896502e-02 4.78489697e-01 -1.18422246e+00 1.37221956e+00
2.99897999e-01 -1.27461553e+00 2.10948691e-01 -2.09576041e-01
6.80426598e-01 -3.63739043e-01 2.65682966e-01 9.43994597e-02
1.73303619e-01 -1.07905626e+00 1.13581026e+00 1.02910173e+00
1.05404961e+00 -2.88040608e-01 2.21786216e-01 4.95070927e-02
-1.63915610e+00 3.14605504e-01 -4.83076870e-01 1.49671927e-01
1.71849012e-01 -9.11132619e-02 -2.64792666e-02 6.22030139e-01
9.30129290e-01 7.38987744e-01 -4.48578387e-01 7.54334390e-01
1.71640277e-01 7.30278715e-02 -3.82772565e-01 5.14515042e-01
-1.95360437e-01 -3.68662387e-01 5.49562812e-01 1.00104070e+00
3.28472108e-01 2.80676454e-01 2.37777591e-01 8.22734594e-01
5.54795749e-02 1.63414165e-01 -1.49573848e-01 5.47949135e-01
-1.81387484e-01 9.61884201e-01 -5.80172122e-01 -2.80367881e-01
-1.69455543e-01 1.42488575e+00 3.65460187e-01 2.68555194e-01
-9.93212640e-01 4.82380688e-01 8.54511023e-01 3.01542252e-01
2.59602871e-02 -3.68387431e-01 -3.65652591e-02 -1.37943017e+00
1.70893058e-01 -8.87278855e-01 4.19427127e-01 -7.93898463e-01
-1.30643117e+00 7.09076047e-01 -3.64736095e-02 -1.40774286e+00
-1.62696242e-01 -4.11731601e-01 -6.12567544e-01 7.33798504e-01
-1.04761529e+00 -1.49893332e+00 -4.94025081e-01 8.28456819e-01
6.70390248e-01 1.18539989e-01 6.25004172e-01 2.66729683e-01
-4.91117001e-01 9.35567260e-01 -1.93885818e-01 9.50876921e-02
8.32015872e-01 -1.26437569e+00 7.09449410e-01 7.18202591e-01
3.21971744e-01 6.07767105e-01 7.31365919e-01 -8.41553688e-01
-1.47524655e+00 -1.12133849e+00 4.79314178e-01 -6.93979263e-01
1.28457472e-01 -2.77475417e-01 -9.58043218e-01 8.23222220e-01
-1.12853006e-01 4.09909725e-01 4.70048100e-01 8.03569183e-02
-4.20962602e-01 2.91066654e-02 -1.04842961e+00 8.27767849e-01
1.48400545e+00 -2.62568384e-01 -5.34182012e-01 2.12318808e-01
8.18386436e-01 -1.09936774e+00 -8.94842446e-01 3.94402176e-01
1.25350988e+00 -8.75300229e-01 1.30473387e+00 -1.07811034e+00
4.76775318e-01 -4.70402539e-01 -2.75550842e-01 -1.06749523e+00
-7.74988830e-01 -5.83726645e-01 -3.06693375e-01 9.79929090e-01
-3.06201987e-02 -3.34522992e-01 7.29802370e-01 9.41521645e-01
-2.85999873e-03 -6.59514010e-01 -1.19512725e+00 -7.72574663e-01
9.85018015e-02 -1.88215986e-01 8.17339897e-01 1.07311785e+00
-5.05725205e-01 1.21979099e-02 -9.68061447e-01 1.69905365e-01
5.11139631e-01 7.14228228e-02 8.94748390e-01 -1.08856869e+00
-5.44430494e-01 -3.88572961e-01 -6.44100606e-01 -1.09821332e+00
2.20375225e-01 -7.87775397e-01 -3.61686349e-02 -1.27848768e+00
2.27991790e-01 -3.03651184e-01 -3.23428631e-01 4.14035708e-01
-2.72435308e-01 3.95296693e-01 3.49582434e-01 2.34329507e-01
-6.05556428e-01 8.23186040e-01 1.79779994e+00 -2.32939631e-01
-2.46601462e-01 -5.53082749e-02 -3.45332831e-01 9.53007340e-01
3.06369364e-01 -1.29416704e-01 -4.73455727e-01 -6.67640209e-01
2.70259306e-02 1.89950287e-01 5.16319931e-01 -1.04701161e+00
1.23955384e-01 -3.71972114e-01 9.38128769e-01 -2.26123944e-01
5.63598633e-01 -8.90066266e-01 8.15700531e-01 4.41969544e-01
-1.36404812e-01 5.17739832e-01 1.51048765e-01 6.55162394e-01
5.75885698e-02 3.57703179e-01 7.73436129e-01 -2.24582732e-01
-8.97335410e-01 1.06143856e+00 2.97260106e-01 2.06928253e-01
6.46065712e-01 -5.34438610e-01 -2.14557885e-03 -3.46069962e-01
-1.06174123e+00 1.35155007e-01 6.98467314e-01 7.12210000e-01
7.41347909e-01 -1.77158356e+00 -8.42741847e-01 3.56813192e-01
-1.40048757e-01 -5.63201569e-02 7.61682034e-01 6.75821066e-01
-4.96582896e-01 -3.47677112e-01 -6.16569757e-01 -6.95401669e-01
-1.45757532e+00 4.85897303e-01 7.76540220e-01 4.65939194e-02
-1.05299509e+00 1.02148926e+00 8.41322780e-01 -3.20786059e-01
-7.78255537e-02 -4.56898451e-01 -1.94849148e-01 -3.57312173e-01
4.08197910e-01 1.77373663e-01 -4.00874734e-01 -1.22744358e+00
-3.00719440e-01 1.19486213e+00 3.24227035e-01 4.01319861e-02
9.38226998e-01 -3.10190320e-01 -1.23795502e-01 5.07764757e-01
1.12270403e+00 -2.17845365e-01 -1.45686138e+00 -1.60892308e-01
-7.33181715e-01 -6.79261386e-01 -2.51208603e-01 -4.43885982e-01
-1.45713484e+00 5.68860173e-01 1.04370344e+00 -2.48507395e-01
1.02490473e+00 -1.52037010e-01 8.93912613e-01 -1.81321755e-01
3.48580599e-01 -9.54295456e-01 1.57320146e-02 2.66227543e-01
1.26677012e+00 -1.22721899e+00 5.21922335e-02 -4.67591405e-01
-6.60119116e-01 9.62245286e-01 9.48546290e-01 -1.71699524e-01
7.24434257e-01 -7.96645507e-02 6.92122951e-02 -2.36796096e-01
-4.37849283e-01 7.50393346e-02 9.55466211e-01 6.73411787e-01
4.69435513e-01 3.35352659e-01 -2.98221380e-01 5.99465847e-01
-4.40992534e-01 -2.14751855e-01 -2.53060702e-02 4.32025790e-01
5.54477237e-02 -8.91765714e-01 -2.07911298e-01 3.32359880e-01
-3.82503480e-01 1.33271605e-01 -5.35054468e-02 7.80328751e-01
5.47342300e-01 2.61619478e-01 4.23581570e-01 -3.50715399e-01
4.80913460e-01 -1.51797011e-01 8.64609063e-01 -2.66345561e-01
-4.47887540e-01 2.40366608e-01 -2.47560471e-01 -7.06095874e-01
-6.57212377e-01 -6.61856234e-01 -1.24549019e+00 -4.05195385e-01
6.65413728e-03 -5.03759742e-01 2.23073706e-01 6.75881863e-01
1.20695762e-01 6.14603817e-01 3.10776293e-01 -8.55825961e-01
-2.62691349e-01 -5.73764920e-01 -6.74622059e-01 9.79659557e-01
4.72404748e-01 -8.91899168e-01 -7.46022761e-02 5.62123418e-01]
|
[11.79651165008545, -0.8711130619049072]
|
c28a37d5-985c-42a1-b9cc-e28ba4cec03f
|
a-generic-self-supervised-learning-ssl
|
2306.15836
| null |
https://arxiv.org/abs/2306.15836v1
|
https://arxiv.org/pdf/2306.15836v1.pdf
|
A generic self-supervised learning (SSL) framework for representation learning from spectra-spatial feature of unlabeled remote sensing imagery
|
Remote sensing data has been widely used for various Earth Observation (EO) missions such as land use and cover classification, weather forecasting, agricultural management, and environmental monitoring. Most existing remote sensing data-based models are based on supervised learning that requires large and representative human-labelled data for model training, which is costly and time-consuming. Recently, self-supervised learning (SSL) enables the models to learn a representation from orders of magnitude more unlabelled data. This representation has been proven to boost the performance of downstream tasks and has potential for remote sensing applications. The success of SSL is heavily dependent on a pre-designed pretext task, which introduces an inductive bias into the model from a large amount of unlabelled data. Since remote sensing imagery has rich spectral information beyond the standard RGB colour space, the pretext tasks established in computer vision based on RGB images may not be straightforward to be extended to the multi/hyperspectral domain. To address this challenge, this work has designed a novel SSL framework that is capable of learning representation from both spectra-spatial information of unlabelled data. The framework contains two novel pretext tasks for object-based and pixel-based remote sensing data analysis methods, respectively. Through two typical downstream tasks evaluation (a multi-label land cover classification task on Sentienl-2 multispectral datasets and a ground soil parameter retrieval task on hyperspectral datasets), the results demonstrate that the representation obtained through the proposed SSL achieved a significant improvement in model performance.
|
['Liangxiu Han', 'Xin Zhang']
|
2023-06-27
| null | null | null | null |
['self-supervised-learning', 'weather-forecasting']
|
['computer-vision', 'miscellaneous']
|
[ 9.61066008e-01 -2.35836640e-01 -3.23449165e-01 -5.49114823e-01
-6.20786011e-01 -4.07808304e-01 5.96087992e-01 2.39296794e-01
-4.55949247e-01 8.87209117e-01 -3.78523052e-01 -7.03655899e-01
-3.40234995e-01 -1.32320213e+00 -3.02942902e-01 -1.03418374e+00
-1.28383517e-01 1.39636963e-04 -1.95654914e-01 -6.76548541e-01
-5.67519814e-02 6.23068929e-01 -1.96591580e+00 3.01814497e-01
1.14714658e+00 1.11989284e+00 6.30584240e-01 3.27317864e-01
-1.38111383e-01 4.16065603e-01 -7.81835169e-02 4.32532281e-01
4.79075402e-01 -2.88117915e-01 -7.42196262e-01 2.26320028e-01
-2.53461860e-02 -2.35827528e-02 3.03677410e-01 1.20732164e+00
4.65919822e-01 2.20917806e-01 7.29975760e-01 -1.13439596e+00
-2.71157026e-01 3.89954895e-01 -6.84258163e-01 -1.71731815e-01
-1.87343165e-01 -1.72272414e-01 9.45792913e-01 -5.89870572e-01
1.09450765e-01 1.06127083e+00 6.99969530e-01 1.03031583e-01
-1.13628745e+00 -7.05363572e-01 6.63777068e-02 3.27542867e-03
-1.58971786e+00 -1.30735800e-01 6.84223473e-01 -4.07922417e-01
7.28908896e-01 6.50154531e-01 7.63933122e-01 3.79051775e-01
-4.60641384e-02 5.83838761e-01 1.80519617e+00 -7.18022227e-01
2.44021446e-01 1.48315609e-01 6.86738193e-02 2.77589411e-01
2.86876172e-01 4.12405580e-01 -8.64311457e-02 2.73102112e-02
4.30379748e-01 4.12615478e-01 -2.69382328e-01 -2.39739954e-01
-1.00275993e+00 1.02049506e+00 1.17380750e+00 3.90207142e-01
-6.55681133e-01 -2.89699227e-01 5.46904579e-02 3.62757772e-01
8.75520885e-01 4.03763711e-01 -6.01569474e-01 4.54603851e-01
-1.00405443e+00 -3.30118626e-01 4.16362405e-01 5.68461657e-01
1.34915268e+00 -3.95898297e-02 4.41330999e-01 9.67762589e-01
4.86598313e-01 1.05172753e+00 5.61501980e-01 -3.94123852e-01
9.51502919e-02 1.01393676e+00 -2.97256410e-02 -9.71913934e-01
-5.14635265e-01 -6.29557192e-01 -1.11593795e+00 3.17671061e-01
-3.32845718e-01 1.01302343e-03 -1.13810885e+00 1.34083319e+00
2.46746302e-01 -2.00500444e-01 5.63492000e-01 9.57091868e-01
8.38158369e-01 9.56531882e-01 5.26039124e-01 -2.49520630e-01
1.04155862e+00 -5.82595348e-01 -4.46399599e-01 -4.79214817e-01
8.60941172e-01 -3.84639353e-01 9.80766654e-01 1.22311227e-01
6.33164570e-02 -5.54159939e-01 -1.12403858e+00 4.57249820e-01
-9.90119874e-01 4.11848873e-01 9.96542335e-01 8.18167448e-01
-8.08487773e-01 4.66364086e-01 -4.78425473e-01 -6.94282174e-01
6.23106003e-01 2.93255806e-01 -5.35770416e-01 -2.55622447e-01
-1.34211338e+00 8.63433659e-01 1.08622682e+00 5.65569401e-01
-5.17746449e-01 -1.51867718e-01 -8.50816071e-01 2.95530260e-03
1.64956704e-01 -7.69381551e-03 6.11714482e-01 -1.46718347e+00
-1.24023819e+00 1.20699942e+00 2.61604398e-01 -2.32859299e-01
6.53870180e-02 -2.60696914e-02 -6.78589582e-01 -2.27376409e-02
3.26317579e-01 4.84009385e-01 7.34484375e-01 -1.42058158e+00
-7.30242968e-01 -6.87670588e-01 -1.11821830e-01 4.02700245e-01
-5.37575006e-01 -2.42123902e-01 4.45848584e-01 -5.26464880e-01
7.03835666e-01 -1.02122557e+00 -3.68651986e-01 -1.96498334e-01
9.45590213e-02 1.78632692e-01 7.92593479e-01 -3.58557612e-01
8.43557179e-01 -2.26095057e+00 -4.59278226e-01 4.26809669e-01
-1.96458548e-01 6.99161291e-01 -2.85431623e-01 3.43031436e-01
-4.10956651e-01 2.79907018e-01 -5.56867063e-01 4.88558471e-01
-3.64506662e-01 4.60017025e-01 -2.83154726e-01 3.90240371e-01
2.58273304e-01 7.97479510e-01 -1.01694548e+00 -2.91139185e-01
5.06434739e-01 2.71662205e-01 1.45747602e-01 2.09418982e-01
-5.80831543e-02 3.50025862e-01 -5.57066679e-01 7.98395991e-01
8.84271383e-01 -2.07423091e-01 1.21506810e-01 -2.47255132e-01
-2.71578044e-01 -1.64524123e-01 -9.80328560e-01 1.36131823e+00
-5.20307779e-01 4.10344034e-01 -1.42230064e-01 -1.36171710e+00
1.16908181e+00 2.59268939e-01 4.11873370e-01 -6.20528877e-01
7.60954767e-02 5.08060575e-01 -1.82151765e-01 -7.12918341e-01
4.00447845e-01 -6.75162852e-01 2.32040375e-01 2.49180436e-01
-3.72131705e-01 -3.26257378e-01 -1.94449857e-01 -4.84648198e-01
4.65017200e-01 3.25740755e-01 5.53769886e-01 -3.64342868e-01
6.99614584e-01 6.49034142e-01 4.76138115e-01 5.19782782e-01
-1.20970860e-01 2.19301030e-01 -3.28020543e-01 -5.67051411e-01
-5.12132823e-01 -5.72573543e-01 -4.58225876e-01 1.48362231e+00
1.49848223e-01 1.44571841e-01 -1.99190274e-01 -4.09446269e-01
1.50770908e-02 6.83743894e-01 -5.01470506e-01 -1.18630275e-01
1.45407215e-01 -1.59441900e+00 4.16655421e-01 2.77419060e-01
1.10418606e+00 -1.11842775e+00 -6.56505942e-01 1.67444006e-01
-2.07152113e-01 -8.18402886e-01 6.63805902e-01 5.21053255e-01
-1.23336053e+00 -9.95325565e-01 -6.05398953e-01 -5.31079710e-01
7.03462124e-01 7.41395831e-01 7.30125189e-01 -3.46901827e-02
-2.17907697e-01 -4.68396954e-02 -7.76205719e-01 -7.99812019e-01
-3.88021201e-01 2.58753061e-01 -2.80035883e-01 2.64922678e-01
7.21254349e-01 -3.09855163e-01 -4.28493172e-01 3.79777640e-01
-1.28811884e+00 3.07801664e-01 1.07592344e+00 8.43111217e-01
5.93142867e-01 4.47122216e-01 6.90223038e-01 -9.05990124e-01
1.71151266e-01 -5.94976366e-01 -4.02173132e-01 4.88249123e-01
-7.75405884e-01 -1.66115507e-01 1.18171893e-01 -7.91992545e-02
-1.10026360e+00 3.52150500e-01 5.74552640e-02 2.46916115e-01
-5.02395809e-01 1.33788240e+00 -1.70136109e-01 -1.90133721e-01
1.05718589e+00 3.16808462e-01 1.55640736e-01 -3.17138970e-01
1.05235070e-01 1.33651078e+00 2.21716955e-01 -1.84576988e-01
7.05844820e-01 5.07604420e-01 2.80526578e-01 -1.22721720e+00
-1.08658957e+00 -9.49115038e-01 -7.78196752e-01 -3.19748312e-01
7.26317465e-01 -1.23465776e+00 -1.09993182e-01 6.63538575e-01
-5.08296072e-01 -3.47879589e-01 -1.87856376e-01 8.05776536e-01
-2.07704425e-01 1.92934200e-01 3.09000790e-01 -1.07966983e+00
-3.78306925e-01 -7.99049675e-01 9.04183686e-01 8.73381495e-02
3.48092258e-01 -9.16020453e-01 -8.19141865e-02 4.18277800e-01
6.71288431e-01 6.44470990e-01 7.83359766e-01 -4.28358167e-01
-1.86109558e-01 -4.79380101e-01 -6.07435167e-01 7.84914911e-01
7.04302371e-01 -2.54565656e-01 -1.27189696e+00 -3.28359872e-01
-5.39916120e-02 -6.65507674e-01 1.18517804e+00 2.43388817e-01
8.03832650e-01 8.91106576e-02 -2.28490651e-01 6.45929396e-01
1.81443501e+00 8.61953497e-02 5.75723350e-01 7.43603587e-01
6.12099648e-01 8.90326381e-01 1.08074892e+00 2.42857277e-01
2.04864874e-01 3.93440612e-02 8.28837097e-01 -6.48103714e-01
3.06475908e-01 1.42096400e-01 1.32626429e-01 3.51381749e-01
-5.30204535e-01 -4.27798219e-02 -1.17373765e+00 3.72855783e-01
-1.95185208e+00 -1.00966394e+00 -3.71348560e-01 2.16592622e+00
7.32714593e-01 -3.82854939e-01 -3.32823783e-01 6.03503108e-01
7.55710602e-01 2.90033281e-01 -5.52878141e-01 1.48525029e-01
-4.63224649e-01 3.07211399e-01 9.23980594e-01 1.38184547e-01
-1.67729831e+00 1.25665236e+00 5.57614088e+00 7.01742530e-01
-1.56634963e+00 -2.66695339e-02 4.36372250e-01 5.62072337e-01
-1.61640286e-01 -1.82040595e-02 -4.07788932e-01 -1.87844634e-01
8.57424796e-01 1.09240957e-01 2.72913635e-01 8.68120551e-01
4.16926056e-01 -3.99347603e-01 -4.87830579e-01 9.09652233e-01
-2.42103696e-01 -8.90588760e-01 2.03167170e-01 1.83366448e-01
8.54615748e-01 3.77553195e-01 -1.36147484e-01 3.21202427e-01
2.61513501e-01 -1.13396919e+00 1.98874742e-01 3.51825863e-01
9.27528083e-01 -4.55935389e-01 9.68402803e-01 5.44470727e-01
-1.22501183e+00 -2.56279588e-01 -6.22933924e-01 -3.31550330e-01
-4.42260295e-01 5.92645288e-01 -7.09307551e-01 9.29491401e-01
6.90017164e-01 1.09558094e+00 -6.19476259e-01 9.07427013e-01
-3.40074539e-01 7.15323269e-01 -4.39345956e-01 1.56621963e-01
4.70375240e-01 -3.94742191e-01 7.13020563e-02 9.49332893e-01
4.79513794e-01 4.48470116e-01 2.98728853e-01 4.16251510e-01
3.06608230e-01 4.77768302e-01 -7.78061748e-01 -2.72011578e-01
1.81343019e-01 1.36657178e+00 -7.55680025e-01 -1.63330331e-01
-2.24180713e-01 6.44792318e-01 -2.21711263e-01 4.74030912e-01
-3.47098559e-01 -2.79170007e-01 2.72843570e-01 -3.46022844e-02
4.10015397e-02 -1.33085117e-01 -2.75914729e-01 -1.09349513e+00
-3.19685251e-01 -6.52551651e-01 3.88281405e-01 -9.74641979e-01
-1.16819191e+00 6.29334688e-01 1.88083891e-02 -1.52722204e+00
-3.20794918e-02 -7.76984274e-01 -2.13052303e-01 1.16515028e+00
-2.53043771e+00 -1.63489640e+00 -9.20687139e-01 5.70953548e-01
-1.11363657e-01 -3.00117135e-01 1.27769721e+00 3.08024399e-02
-4.22746867e-01 -2.58834869e-01 4.19864953e-01 -6.00650385e-02
6.26210630e-01 -1.12053597e+00 -3.18394363e-01 7.34166026e-01
-2.21462592e-01 2.15241000e-01 2.79990315e-01 -4.74504739e-01
-1.06589341e+00 -1.64140558e+00 6.50532484e-01 4.46820855e-01
4.54955488e-01 4.15188491e-01 -7.89289296e-01 4.51192737e-01
-4.15711552e-01 2.47929066e-01 1.21336484e+00 -4.99139465e-02
-1.62929356e-01 -6.03808403e-01 -1.15623605e+00 1.96219366e-02
5.25705159e-01 -5.16342640e-01 -4.59306568e-01 5.35603464e-01
2.19164621e-02 2.37736061e-01 -8.30101788e-01 8.08466733e-01
6.00385904e-01 -7.65618205e-01 8.54078352e-01 -4.82836366e-01
4.04510707e-01 -5.08221567e-01 -6.58805728e-01 -1.26834309e+00
-3.84646833e-01 5.42623363e-02 7.61452258e-01 8.57135415e-01
3.79134387e-01 -8.63088667e-01 4.54582989e-01 4.05192822e-01
1.75490528e-02 -1.81594431e-01 -5.72395146e-01 -7.89820910e-01
-2.00350750e-02 -4.29447800e-01 7.15212882e-01 1.24148417e+00
-4.49339449e-01 2.76990443e-01 -4.63021472e-02 5.16447484e-01
5.43286383e-01 5.04195809e-01 7.16898024e-01 -1.94147611e+00
2.06972986e-01 -2.73082882e-01 -4.82585937e-01 -5.91037691e-01
8.88354704e-02 -1.12681556e+00 1.01782553e-01 -1.65474343e+00
5.82591072e-02 -1.01857245e+00 -5.71851075e-01 1.03688872e+00
-3.86161119e-01 5.15734017e-01 -1.30094767e-01 6.29891753e-01
-2.86166128e-02 7.41993010e-01 9.52517390e-01 -3.62133324e-01
-4.10653651e-01 2.49209985e-01 -6.00731254e-01 6.87564075e-01
1.00012946e+00 -4.82671440e-01 -4.42511082e-01 -1.86431736e-01
3.96432191e-01 -2.35609546e-01 4.35825408e-01 -8.76576841e-01
-3.39836717e-01 -6.79805815e-01 2.28808418e-01 -5.11699498e-01
5.40580638e-02 -1.22827399e+00 3.24092269e-01 5.15763521e-01
-8.77241716e-02 -6.99455738e-01 2.38363087e-01 3.46872538e-01
-3.56390804e-01 -2.23006651e-01 8.43381345e-01 -3.29323351e-01
-1.27557933e+00 2.18358487e-01 -3.27423424e-01 -5.86973488e-01
1.14659274e+00 -4.72293019e-01 -1.33333594e-01 2.05228534e-02
-6.90981507e-01 1.44164413e-01 3.11147809e-01 2.39878893e-01
4.19973761e-01 -1.24525881e+00 -9.01208282e-01 4.28886235e-01
7.45633841e-01 1.36097446e-01 8.47064704e-02 5.88197887e-01
-8.18708241e-01 3.23688328e-01 -4.16298181e-01 -7.44746327e-01
-1.26710248e+00 1.72166154e-01 3.91982943e-01 -1.48312882e-01
-1.76258460e-01 5.49911737e-01 -1.39898539e-01 -8.79192710e-01
-3.58877718e-01 -1.84317902e-01 -6.93157256e-01 3.48745227e-01
2.40298703e-01 1.63849890e-01 2.07278118e-01 -9.76711154e-01
-2.58087218e-01 5.46878397e-01 5.05686462e-01 8.11099783e-02
1.84401989e+00 -2.32278958e-01 -4.83013809e-01 6.68539822e-01
9.52382088e-01 -6.23382747e-01 -8.40860784e-01 -5.32071054e-01
1.83325987e-02 -4.19834614e-01 7.60995090e-01 -9.28325772e-01
-9.50618804e-01 1.00649405e+00 9.43018675e-01 4.30038691e-01
1.49186063e+00 -4.47137624e-01 2.50151724e-01 8.26838553e-01
4.87494230e-01 -1.14719641e+00 -4.66082186e-01 3.25333476e-01
8.51403654e-01 -1.96515119e+00 2.97984421e-01 -4.77665663e-01
-4.22010154e-01 1.19568467e+00 1.61584526e-01 3.60783100e-01
9.97963190e-01 -4.22700197e-01 4.74319309e-01 -3.22766215e-01
2.36500278e-02 -8.79997432e-01 3.53457749e-01 6.28675699e-01
5.00650764e-01 6.00233138e-01 -1.89019099e-01 1.91003740e-01
2.90840287e-02 2.22306728e-01 1.34400517e-01 1.29176712e+00
-8.66040111e-01 -1.06840181e+00 -6.57395363e-01 6.46120667e-01
-4.08282094e-02 -2.34406754e-01 -1.72899857e-01 5.69728494e-01
2.09374085e-01 1.23152971e+00 -2.78777093e-01 -3.23442578e-01
4.33153519e-03 2.09427252e-02 -1.04046101e-03 -8.27885568e-01
-2.68901974e-01 2.02600677e-02 -1.19662128e-01 -1.61951706e-01
-1.46875262e+00 -4.13988084e-01 -9.88162816e-01 -1.12633742e-01
-6.48920357e-01 1.34268507e-01 1.00617099e+00 1.10656440e+00
4.55464348e-02 3.29541713e-01 1.07925284e+00 -1.08921039e+00
-5.03194153e-01 -1.25908899e+00 -1.05532444e+00 2.84264594e-01
1.84346020e-01 -6.30729496e-01 -3.69228721e-01 7.59399980e-02]
|
[9.689863204956055, -1.4769763946533203]
|
ce2c10e8-c84f-433b-b5f9-b7587753440f
|
detr-with-additional-global-aggregation-for
|
2304.07082
| null |
https://arxiv.org/abs/2304.07082v1
|
https://arxiv.org/pdf/2304.07082v1.pdf
|
DETR with Additional Global Aggregation for Cross-domain Weakly Supervised Object Detection
|
This paper presents a DETR-based method for cross-domain weakly supervised object detection (CDWSOD), aiming at adapting the detector from source to target domain through weak supervision. We think DETR has strong potential for CDWSOD due to an insight: the encoder and the decoder in DETR are both based on the attention mechanism and are thus capable of aggregating semantics across the entire image. The aggregation results, i.e., image-level predictions, can naturally exploit the weak supervision for domain alignment. Such motivated, we propose DETR with additional Global Aggregation (DETR-GA), a CDWSOD detector that simultaneously makes "instance-level + image-level" predictions and utilizes "strong + weak" supervisions. The key point of DETR-GA is very simple: for the encoder / decoder, we respectively add multiple class queries / a foreground query to aggregate the semantics into image-level predictions. Our query-based aggregation has two advantages. First, in the encoder, the weakly-supervised class queries are capable of roughly locating the corresponding positions and excluding the distraction from non-relevant regions. Second, through our design, the object queries and the foreground query in the decoder share consensus on the class semantics, therefore making the strong and weak supervision mutually benefit each other for domain alignment. Extensive experiments on four popular cross-domain benchmarks show that DETR-GA significantly improves CSWSOD and advances the states of the art (e.g., 29.0% --> 79.4% mAP on PASCAL VOC --> Clipart_all dataset).
|
['Yi Yang', 'Si Liu', 'Yifan Sun', 'Zongheng Tang']
|
2023-04-14
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Tang_DETR_With_Additional_Global_Aggregation_for_Cross-Domain_Weakly_Supervised_Object_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Tang_DETR_With_Additional_Global_Aggregation_for_Cross-Domain_Weakly_Supervised_Object_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['weakly-supervised-object-detection']
|
['computer-vision']
|
[ 1.50971800e-01 7.05613717e-02 -5.29262722e-01 -5.09563565e-01
-1.13092113e+00 -5.32099605e-01 5.31703174e-01 -5.63266277e-02
-2.92406231e-01 4.85579163e-01 3.24921235e-02 3.21588106e-02
3.12358290e-01 -6.16003335e-01 -1.01143384e+00 -6.43665314e-01
3.33448946e-01 5.41141748e-01 1.04969347e+00 -2.00140566e-01
-1.09242633e-01 1.75231785e-01 -1.48910022e+00 7.62457550e-01
8.70862663e-01 1.25313723e+00 7.69295096e-01 2.20023155e-01
-2.65223652e-01 8.26762021e-01 -5.73920429e-01 -5.04688919e-01
3.08928877e-01 -3.34232271e-01 -5.75595438e-01 1.25010952e-01
7.21287072e-01 -3.27444166e-01 -1.61827922e-01 1.20400107e+00
4.70289141e-01 -1.88506961e-01 3.98806840e-01 -1.17073750e+00
-5.86388946e-01 5.01237988e-01 -1.00892699e+00 1.08067058e-01
1.33945316e-01 3.07442784e-01 1.12959695e+00 -1.17883527e+00
7.09168315e-01 1.18762589e+00 4.56009299e-01 6.46659791e-01
-1.30793655e+00 -8.09340477e-01 5.28947651e-01 1.33725226e-01
-1.46794772e+00 -3.08183223e-01 7.40343869e-01 -4.92987007e-01
9.21152294e-01 2.09443137e-01 2.15660617e-01 9.91027057e-01
-1.39726132e-01 1.24337864e+00 1.03484213e+00 -2.30273440e-01
8.54154378e-02 4.87366527e-01 2.43082792e-01 5.95575809e-01
1.30182534e-01 -6.95071667e-02 -8.11495960e-01 1.54549018e-01
4.90752697e-01 -1.88946351e-01 -1.20577857e-01 -6.29661918e-01
-1.33682024e+00 4.79865253e-01 6.17510915e-01 -3.51903178e-02
-1.20650947e-01 -6.02177009e-02 5.57184577e-01 1.67301491e-01
5.26221871e-01 4.00314219e-02 -5.64764738e-01 3.18766356e-01
-1.03577077e+00 1.67648882e-01 4.57176030e-01 1.48588300e+00
8.92313600e-01 -2.14176744e-01 -4.37623680e-01 8.86404395e-01
1.92157269e-01 6.65135801e-01 3.12652826e-01 -5.64189792e-01
8.06202888e-01 7.61118770e-01 1.04861129e-02 -6.65662467e-01
-1.41585413e-02 -6.84923530e-01 -5.86981893e-01 2.73671210e-01
2.84891754e-01 9.98856723e-02 -1.05813825e+00 1.86457026e+00
3.47105682e-01 1.00450695e-01 -3.59185180e-03 1.04445302e+00
8.03537488e-01 5.96936822e-01 3.64207387e-01 3.05671152e-02
1.52491593e+00 -1.24414933e+00 -3.97356004e-01 -7.10371315e-01
5.48208356e-01 -5.93231916e-01 1.20658123e+00 1.84180811e-01
-1.14244950e+00 -1.02182949e+00 -1.06877208e+00 -1.98594660e-01
-3.12576294e-01 5.48404932e-01 5.45778982e-02 1.06136166e-01
-7.86095560e-01 1.41932040e-01 -6.98678315e-01 -2.81311691e-01
5.11870325e-01 4.48496640e-01 -2.51358509e-01 -1.51312515e-01
-1.08962536e+00 6.38003886e-01 6.46707416e-01 -3.25995058e-01
-1.04002309e+00 -6.27805829e-01 -7.44902730e-01 1.95343476e-02
6.16197646e-01 -6.18341744e-01 1.23393941e+00 -1.39122355e+00
-1.12481809e+00 1.24022269e+00 -4.73105669e-01 -6.33596301e-01
6.36387467e-01 -4.07304287e-01 -4.73106235e-01 8.99277255e-02
7.51401246e-01 1.01039433e+00 9.06194687e-01 -1.35337853e+00
-1.21606505e+00 -5.20888567e-01 -1.26519501e-01 3.04206967e-01
-2.87361592e-01 -2.57707424e-02 -1.21595359e+00 -8.36103141e-01
1.07784055e-01 -7.92651534e-01 2.80233715e-02 1.85476571e-01
-4.71118063e-01 -3.16926479e-01 1.05994391e+00 -5.05805850e-01
1.24761772e+00 -2.59535074e+00 5.55348881e-02 -4.38639559e-02
3.03864211e-01 2.99512535e-01 -1.57444894e-01 -2.62396596e-02
-1.68318108e-01 -2.79311925e-01 -3.10123444e-01 -5.46490788e-01
8.33110977e-03 2.50486284e-01 -5.60365319e-01 2.86136717e-01
5.49844265e-01 9.70938563e-01 -9.81142163e-01 -7.32652009e-01
1.43388927e-01 3.99976186e-02 -5.23636222e-01 3.52664173e-01
-5.83155692e-01 7.96587393e-02 -3.13417673e-01 5.83161354e-01
1.00837708e+00 -4.76088524e-01 1.54928774e-01 -3.52500081e-01
-1.15096472e-01 3.04814339e-01 -1.20800221e+00 1.74799609e+00
-1.82839006e-01 5.10584354e-01 1.05003074e-01 -1.02453637e+00
8.98039937e-01 -5.22380583e-02 2.04495102e-01 -1.00117934e+00
-2.56043255e-01 4.38180625e-01 -4.08980370e-01 -1.16610840e-01
3.42692435e-01 8.28091204e-02 -2.32082039e-01 1.38326269e-02
3.75730991e-01 1.90180972e-01 1.50192142e-01 3.36847931e-01
7.91542888e-01 4.08540726e-01 2.24333286e-01 -3.98825377e-01
7.08038151e-01 3.88441741e-01 8.52315605e-01 6.98278189e-01
-1.68758750e-01 7.11584151e-01 6.14398837e-01 -1.05056942e-01
-9.16248143e-01 -1.47005153e+00 -2.68021580e-02 1.48032975e+00
7.64671326e-01 -1.95834711e-01 -7.65169859e-01 -1.07740355e+00
1.58786654e-01 5.85223198e-01 -4.70135450e-01 -1.82850868e-01
-5.99904239e-01 -4.41086292e-01 3.60301554e-01 9.00624275e-01
7.59087265e-01 -8.28461289e-01 -2.96349138e-01 2.57460177e-01
-2.21136376e-01 -1.52698505e+00 -6.89395845e-01 5.52493870e-01
-6.76441908e-01 -7.34012783e-01 -8.18639755e-01 -1.11030102e+00
6.78965271e-01 5.39700747e-01 1.22965086e+00 -2.71384001e-01
4.76893708e-02 -3.46796140e-02 -2.76353598e-01 -3.53067577e-01
-2.75659144e-01 1.18411258e-01 -6.88785240e-02 1.20379247e-01
6.61178708e-01 -1.17193893e-01 -5.85349202e-01 7.07266033e-01
-6.82608843e-01 2.49161437e-01 8.24839771e-01 9.52242494e-01
9.51161027e-01 -2.14402437e-01 4.97884989e-01 -9.84587550e-01
5.42056076e-02 -5.22659898e-01 -9.31049347e-01 2.33223438e-01
-3.65102708e-01 2.11553313e-02 6.44428253e-01 -3.31182361e-01
-1.31455040e+00 5.12493730e-01 -1.22805402e-01 -4.92481321e-01
-1.70228183e-01 -7.72685558e-02 -5.86290538e-01 1.87048316e-01
8.24733138e-01 5.21957219e-01 -2.06885695e-01 -5.69471061e-01
2.88250357e-01 7.16255486e-01 9.04469371e-01 -5.20520806e-01
7.63508677e-01 6.63012862e-01 -5.36818385e-01 -5.31156361e-01
-1.11750662e+00 -9.17980850e-01 -5.96749961e-01 7.38904849e-02
1.14909697e+00 -1.46146548e+00 -3.16343546e-01 4.79953945e-01
-1.22584963e+00 -3.31294298e-01 -4.90522474e-01 2.50398725e-01
-4.96943951e-01 2.38761663e-01 -4.26394641e-01 -4.31948245e-01
-7.01169968e-02 -1.31971800e+00 1.51102459e+00 9.29149389e-02
-1.24467157e-01 -6.04128242e-01 -3.93297315e-01 4.76867348e-01
5.71334064e-02 -9.21200961e-02 7.29096472e-01 -7.33601749e-01
-8.15196395e-01 1.30082726e-01 -7.52066731e-01 5.61415672e-01
-1.02553643e-01 -4.21984434e-01 -1.27871060e+00 -1.92291290e-01
-1.13864087e-01 -2.42029041e-01 1.06375468e+00 2.20358759e-01
1.19758964e+00 -1.18316058e-02 -6.84710860e-01 4.94902134e-01
1.50160646e+00 4.55138013e-02 4.72690016e-01 3.24954927e-01
6.93075836e-01 5.25644898e-01 9.91877854e-01 3.33921850e-01
3.22744250e-01 1.00823188e+00 5.02227902e-01 -4.98462826e-01
-5.85240543e-01 -3.53220493e-01 6.87677801e-01 4.25440490e-01
4.15041149e-01 -3.14225018e-01 -8.71606529e-01 8.76137257e-01
-2.02602768e+00 -7.65380681e-01 -1.74177796e-01 2.24966884e+00
9.69225287e-01 5.23145735e-01 3.24933052e-01 -3.20617110e-01
9.53497291e-01 1.30738690e-01 -7.59600520e-01 -4.43989262e-02
-3.35849822e-01 6.71885461e-02 6.25017345e-01 3.06394219e-01
-1.22548735e+00 1.08302522e+00 5.11648560e+00 1.21938503e+00
-1.05475438e+00 2.32472181e-01 6.02897525e-01 -1.61405236e-01
-1.26961708e-01 -1.81958705e-01 -1.38102806e+00 7.78459907e-01
4.14377868e-01 9.75411907e-02 -1.60279304e-01 1.28143287e+00
-1.18585110e-01 -1.70019343e-01 -1.26581550e+00 8.84674072e-01
-1.34384632e-01 -1.33844721e+00 2.62398005e-01 -1.02182008e-01
8.01178098e-01 3.82481724e-01 1.18906215e-01 3.61726910e-01
3.88607413e-01 -5.46218455e-01 1.13089144e+00 1.37581795e-01
1.05897582e+00 -5.64747095e-01 7.03250587e-01 5.26744425e-01
-1.35462594e+00 -6.79556355e-02 -4.16805893e-01 4.43907827e-01
9.31447223e-02 7.05620408e-01 -6.20869458e-01 5.55741787e-01
9.02337193e-01 7.01921105e-01 -4.13061529e-01 6.08638406e-01
-2.45668024e-01 5.12792289e-01 -2.88956344e-01 2.83980817e-01
3.64668429e-01 2.15945035e-01 5.51763356e-01 1.61260295e+00
2.29041819e-02 -1.39233559e-01 3.40797663e-01 9.74116027e-01
-8.29415694e-02 -2.05707595e-01 -3.12367290e-01 4.17936653e-01
3.55975330e-01 8.42604816e-01 -4.99883801e-01 -7.05492020e-01
-7.00185359e-01 1.26468563e+00 3.16490889e-01 3.30860525e-01
-1.23356795e+00 -3.17911208e-01 9.20613945e-01 3.30503702e-01
6.61250114e-01 3.68170701e-02 -5.31936526e-01 -1.14433980e+00
3.94335747e-01 -8.66912186e-01 4.94153500e-01 -7.24828720e-01
-1.20511854e+00 5.14863133e-01 -1.27866015e-01 -1.49452233e+00
1.89659446e-01 -6.75409853e-01 -3.37055981e-01 9.88189816e-01
-1.84016347e+00 -1.28829825e+00 -4.45930660e-01 7.54846692e-01
9.17208314e-01 -6.95369095e-02 3.35401297e-01 5.17515779e-01
-5.62620461e-01 7.27551818e-01 6.48865709e-03 2.86073864e-01
9.38842773e-01 -1.33009076e+00 4.48648512e-01 1.01338804e+00
1.60242110e-01 3.64970624e-01 4.07935053e-01 -6.73902988e-01
-1.01934564e+00 -1.42751217e+00 8.54801953e-01 -4.42065716e-01
5.69424510e-01 -6.64274573e-01 -1.04105365e+00 5.26914895e-01
3.63765545e-02 2.92391539e-01 2.93971241e-01 -2.44198710e-01
-5.26948810e-01 -5.23171067e-01 -9.89211082e-01 5.13366103e-01
1.28850782e+00 -6.50547683e-01 -7.18636513e-01 3.48807454e-01
9.58990753e-01 -4.72875923e-01 -4.13802534e-01 5.19503057e-01
3.00639480e-01 -1.06051815e+00 1.13501036e+00 -1.86870292e-01
4.76256549e-01 -7.31080532e-01 -2.78743207e-01 -9.25716043e-01
-4.38570827e-01 -2.95623481e-01 -1.69976875e-01 1.36797023e+00
5.06871581e-01 -4.71247375e-01 9.02878404e-01 2.30814457e-01
-2.76387066e-01 -5.93077779e-01 -8.23730290e-01 -1.07392287e+00
-2.51369655e-01 -4.93606478e-01 5.03805637e-01 7.22316146e-01
-3.38233262e-01 4.64788437e-01 -2.18809843e-01 4.88234431e-01
6.70267105e-01 3.51153195e-01 8.80964637e-01 -9.42230582e-01
-5.15769482e-01 -3.40729088e-01 -3.26812565e-01 -1.82637525e+00
3.86593561e-03 -7.81885743e-01 2.06842855e-01 -1.19731104e+00
3.23483467e-01 -3.95361036e-01 -4.20820147e-01 5.71696639e-01
-2.68234611e-01 3.07334185e-01 2.86655426e-01 3.73255283e-01
-9.44826961e-01 3.28641355e-01 1.04878032e+00 -3.12093914e-01
-8.77115056e-02 -1.07746743e-01 -7.39819884e-01 7.35232651e-01
4.64402527e-01 -5.68886876e-01 -3.14091146e-01 -6.98033929e-01
-1.78681999e-01 -2.54716128e-01 4.72253472e-01 -1.06054997e+00
4.02267843e-01 7.31866807e-02 3.85183841e-01 -7.04172552e-01
1.82293653e-01 -8.91516566e-01 -2.09381059e-01 4.06355053e-01
-2.39573762e-01 -3.00787002e-01 3.68594944e-01 7.33256221e-01
-5.94306350e-01 9.70535129e-02 1.14824176e+00 1.26362890e-02
-1.46174634e+00 2.18392134e-01 1.34649843e-01 2.46485308e-01
1.29797041e+00 -4.41059768e-01 -2.92174190e-01 1.19105931e-02
-7.42852330e-01 5.35367131e-01 4.51870620e-01 3.22774649e-01
4.52071846e-01 -1.24075854e+00 -6.31647944e-01 2.89260447e-01
6.15004241e-01 3.88668954e-01 1.01290897e-01 7.74699569e-01
-1.97336942e-01 3.45525354e-01 -4.54127640e-02 -1.03087115e+00
-1.32047606e+00 6.49323642e-01 2.09540039e-01 -3.64430517e-01
-3.73178482e-01 1.24419475e+00 1.06588781e+00 -2.14377642e-01
2.75643349e-01 -1.85137987e-01 1.84268594e-01 6.40918687e-02
4.40709502e-01 6.25563711e-02 4.45781462e-02 -6.03910089e-01
-6.40337825e-01 6.33882105e-01 -2.42316574e-01 9.90590304e-02
1.10514903e+00 -1.31130174e-01 1.20249689e-01 1.31860480e-01
1.22124350e+00 5.45950532e-02 -1.60452104e+00 -7.03114152e-01
1.83910340e-01 -4.35927004e-01 -1.71243459e-01 -8.38452935e-01
-1.05377340e+00 1.08547747e+00 6.21143103e-01 -5.92083111e-02
1.45975280e+00 4.01483625e-01 7.41571903e-01 -5.46245351e-02
3.06571662e-01 -1.22209036e+00 2.56230414e-01 4.47639555e-01
6.81795120e-01 -1.41909158e+00 -1.75690502e-01 -6.15260601e-01
-9.62338567e-01 8.46467972e-01 1.20435333e+00 -6.53519779e-02
2.11728886e-01 5.14198244e-01 -4.33183908e-02 -9.36181322e-02
-6.86397254e-01 -5.12071371e-01 4.14344966e-01 5.98181486e-01
2.43524164e-01 -5.65706119e-02 2.49097124e-02 6.29418015e-01
2.56306797e-01 -8.67862254e-02 -9.20214280e-02 5.69957078e-01
-6.46056294e-01 -9.90156531e-01 -2.47925311e-01 2.30845302e-01
-2.72062361e-01 -1.70038968e-01 -3.66283536e-01 7.46896744e-01
5.57671249e-01 5.68216443e-01 1.42424598e-01 -3.78793865e-01
5.88416040e-01 -6.16184697e-02 -2.68749949e-02 -8.80341172e-01
-4.69142914e-01 1.84505135e-01 2.74882577e-02 -8.05033207e-01
-1.69403151e-01 -5.23141861e-01 -1.36652434e+00 1.17221735e-01
-4.23338413e-01 6.19566441e-03 3.02311391e-01 7.48832047e-01
6.35257363e-01 4.57927406e-01 4.64432478e-01 -4.34561312e-01
-5.92624366e-01 -5.40359437e-01 -4.49437857e-01 5.26754022e-01
4.74275529e-01 -5.97398460e-01 -1.92880541e-01 2.23434642e-01]
|
[9.407964706420898, 1.3528774976730347]
|
c3004d17-7b6c-4f9d-930a-04a59ee63c05
|
graph-context-attention-networks-for-size
| null | null |
http://openaccess.thecvf.com//content/CVPR2022/html/Jiang_Graph-Context_Attention_Networks_for_Size-Varied_Deep_Graph_Matching_CVPR_2022_paper.html
|
http://openaccess.thecvf.com//content/CVPR2022/papers/Jiang_Graph-Context_Attention_Networks_for_Size-Varied_Deep_Graph_Matching_CVPR_2022_paper.pdf
|
Graph-Context Attention Networks for Size-Varied Deep Graph Matching
|
Deep learning for graph matching has received growing interest and developed rapidly in the past decade. Although recent deep graph matching methods have shown excellent performance on matching between graphs of equal size in the computer vision area, the size-varied graph matching problem, where the number of keypoints in the images of the same category may vary due to occlusion, is still an open and challenging problem. To tackle this, we firstly propose to formulate the combinatorial problem of graph matching as an Integer Linear Programming (ILP) problem, which is more flexible and efficient to facilitate comparing graphs of varied sizes. A novel Graph-context Attention Network (GCAN), which jointly capture intrinsic graph structure and cross-graph information for improving the discrimination of node features, is then proposed and trained to resolve this ILP problem with node correspondence supervision. We further show that the proposed GCAN model is efficient to resolve the graph-level matching problem and is able to automatically learn node-to-node similarity via graph-level matching. The proposed approach is evaluated on three public keypoint-matching datasets and one graph-matching dataset for blood vessel patterns, with experimental results showing its superior performance over existing state-of-the-art algorithms on the keypoint and graph-level matching tasks.
|
['Bryan M. Williams', 'Sue Black', 'Plamen Angelov', 'Hossein Rahmani', 'Zheheng Jiang']
|
2022-01-01
| null | null | null |
cvpr-2022-1
|
['graph-matching']
|
['graphs']
|
[ 9.25223008e-02 1.23686567e-01 -2.48530149e-01 -3.44990641e-01
-7.03370094e-01 -4.40939009e-01 4.49710965e-01 7.53037870e-01
-2.56966442e-01 8.26637521e-02 -1.82537958e-01 -3.77874583e-01
-4.99887884e-01 -9.07242179e-01 -6.49770498e-01 -5.25299191e-01
-2.68680245e-01 5.55008352e-01 2.28197336e-01 -1.85889360e-02
2.33139992e-01 8.10018659e-01 -1.35209918e+00 -1.27256051e-01
7.95270503e-01 1.18064213e+00 5.07798269e-02 5.30330122e-01
-2.71497816e-01 5.67542076e-01 -2.14879632e-01 -6.87854409e-01
6.70950532e-01 -3.25766355e-01 -7.24268377e-01 1.89153031e-01
1.21590686e+00 5.28313406e-02 -7.92066336e-01 1.33173192e+00
6.94883764e-01 1.13330886e-01 4.27472681e-01 -1.70008099e+00
-7.92302251e-01 4.33847785e-01 -6.68721199e-01 4.24512267e-01
2.40408167e-01 -3.02628987e-02 1.50117242e+00 -5.19676507e-01
5.69493353e-01 1.47050822e+00 5.63566267e-01 1.53612792e-01
-1.14339399e+00 -6.03032410e-01 3.11106980e-01 4.56982642e-01
-1.28179300e+00 8.56287330e-02 1.12559783e+00 -4.33267057e-01
7.09187627e-01 7.86877647e-02 8.89059126e-01 4.76902992e-01
1.55068234e-01 5.63299656e-01 6.77371979e-01 -3.59225512e-01
-1.80407375e-01 -3.37285399e-01 2.07231745e-01 9.37559724e-01
4.19859767e-01 1.53585806e-01 -1.48338348e-01 -1.25786409e-01
9.33657825e-01 3.34718317e-01 -3.52124184e-01 -8.84887338e-01
-1.21647799e+00 1.04759240e+00 1.22732556e+00 3.64043653e-01
-3.11399132e-01 2.78913856e-01 5.20112097e-01 4.94662851e-01
3.08026284e-01 4.48283970e-01 -8.08387920e-02 4.76634145e-01
-4.55336332e-01 3.58705878e-01 8.79433393e-01 1.04639041e+00
7.36866534e-01 -2.70185024e-01 -2.01398715e-01 7.67030239e-01
2.02318355e-01 3.59486461e-01 8.74871388e-02 -4.90652889e-01
6.46575689e-01 1.17385924e+00 -4.53986138e-01 -1.74656081e+00
-5.36185563e-01 -5.96416116e-01 -1.12385988e+00 1.09304331e-01
6.51823461e-01 3.72571439e-01 -8.64468038e-01 1.59237576e+00
4.46636945e-01 5.22342026e-01 -3.33072841e-01 9.82942045e-01
1.41548300e+00 1.93322375e-01 1.06301166e-01 3.22631478e-01
1.33823133e+00 -1.08070028e+00 -3.18643689e-01 -1.75301477e-01
7.06642389e-01 -4.89286214e-01 6.67387724e-01 -3.59958947e-01
-9.04674411e-01 -5.90045094e-01 -9.75250185e-01 -2.18000695e-01
-4.81765449e-01 -3.12208176e-01 8.20753396e-01 3.59815538e-01
-1.16073215e+00 6.05119884e-01 -3.63407731e-01 -5.10298908e-01
5.75185180e-01 4.00969982e-01 -7.18617976e-01 -2.70574033e-01
-1.25568604e+00 6.95035577e-01 1.62290514e-01 1.51511997e-01
-2.34681144e-01 -1.04803300e+00 -1.21255457e+00 3.65979493e-01
3.78623903e-01 -9.92923200e-01 5.09571731e-01 -7.62736738e-01
-9.81902719e-01 1.43587971e+00 2.84572423e-01 -4.15101230e-01
6.17412567e-01 4.05752659e-01 -3.41904819e-01 2.42490903e-01
1.56516209e-01 5.05774081e-01 7.40978301e-01 -8.01157951e-01
-4.65395540e-01 -5.72067559e-01 2.59923041e-01 7.18007088e-02
1.28449900e-02 -8.91607627e-02 -6.82855546e-01 -7.67087340e-01
1.77379340e-01 -1.00882137e+00 -4.01587576e-01 6.08982980e-01
-2.82957792e-01 -5.80385029e-01 3.66911411e-01 -5.00768542e-01
9.53955173e-01 -1.84404922e+00 2.53057092e-01 5.03078818e-01
7.46553957e-01 2.66732186e-01 -5.04345477e-01 3.69671673e-01
-3.85612905e-01 6.23703934e-04 -1.21487178e-01 -1.19071372e-01
4.63490561e-02 4.51695137e-02 1.17586479e-01 5.74678957e-01
8.97034556e-02 1.29944432e+00 -9.58112001e-01 -6.54005706e-01
3.76451224e-01 3.62956882e-01 -4.63284135e-01 4.84008014e-01
1.87611818e-01 2.61514008e-01 -3.99490237e-01 7.03706920e-01
8.92239630e-01 -6.96130514e-01 1.05683945e-01 -5.42431772e-01
3.20971012e-01 -3.23545575e-01 -1.25427389e+00 1.88594413e+00
-2.86479771e-01 5.72342157e-01 9.45116282e-02 -1.44664037e+00
1.07140338e+00 -1.81779847e-01 7.13560820e-01 -9.11623240e-01
1.86927438e-01 1.48436084e-01 2.51328349e-01 -3.23987186e-01
1.01021051e-01 2.75168836e-01 -7.05116689e-02 1.08325548e-01
7.72029832e-02 -3.09245531e-02 2.75604010e-01 3.24783534e-01
1.26093757e+00 -4.34104830e-01 4.51167762e-01 -3.87037098e-01
8.58989358e-01 -1.66581795e-01 4.95546132e-01 8.28732431e-01
-3.39248300e-01 4.62311387e-01 5.93591273e-01 -8.19958031e-01
-8.31232011e-01 -8.79333615e-01 -5.45712858e-02 8.27475190e-01
8.09244514e-01 -2.86296874e-01 -4.43071932e-01 -8.45913112e-01
5.84317446e-01 -1.79505900e-01 -6.65391266e-01 -1.03777796e-01
-8.76617908e-01 -3.62567127e-01 6.30539954e-02 4.15589392e-01
4.68747795e-01 -8.65064919e-01 -2.52155215e-01 2.52288669e-01
7.64259845e-02 -1.34290457e+00 -8.10745597e-01 -2.98726916e-01
-5.99333525e-01 -1.65637422e+00 -8.59900177e-01 -1.11066425e+00
8.13041031e-01 5.53945065e-01 1.56813562e+00 7.18528330e-01
-8.45719635e-01 4.90177244e-01 -3.96704376e-01 -1.18993692e-01
-2.81486422e-01 1.51667222e-01 -5.52385390e-01 2.33964160e-01
1.86779708e-01 -5.91196358e-01 -8.97309482e-01 4.84059274e-01
-7.62612760e-01 2.19740104e-02 6.48278236e-01 9.85686421e-01
7.53548980e-01 -1.87593088e-01 4.91794020e-01 -9.20289636e-01
4.61221546e-01 -3.52908641e-01 -8.35650921e-01 5.10093033e-01
-6.56271696e-01 2.20025241e-01 5.96641123e-01 -2.60502398e-01
-9.41232666e-02 2.03758641e-03 -3.67360078e-02 -6.94298685e-01
1.47254437e-01 4.84356880e-01 -2.74731964e-01 -9.13067579e-01
2.04894781e-01 -2.95315422e-02 1.84786946e-01 -3.34861726e-01
4.00571138e-01 9.37637091e-02 6.22351050e-01 -4.25455064e-01
9.62536097e-01 3.56059790e-01 6.37165189e-01 -4.39280093e-01
-5.57100773e-01 -7.52304018e-01 -7.12370157e-01 -2.43725121e-01
7.67700136e-01 -8.16791177e-01 -7.97042847e-01 5.15731752e-01
-9.73350644e-01 -9.80417132e-02 -2.14276128e-02 1.77866042e-01
-3.84296149e-01 7.94735372e-01 -3.28057647e-01 -9.80202481e-02
-6.82222128e-01 -1.23689497e+00 1.22880936e+00 3.20648193e-01
2.93981552e-01 -1.44052446e+00 2.07831729e-02 1.90741763e-01
4.19696003e-01 7.61471748e-01 1.20765436e+00 -7.52554893e-01
-1.00024760e+00 -5.38788617e-01 -8.79240870e-01 -2.44806260e-01
1.10625334e-01 -2.96562701e-01 -3.22654992e-01 -6.04348242e-01
-7.01961815e-01 1.54787572e-02 7.86725461e-01 3.21153969e-01
1.12697685e+00 -1.24129362e-01 -6.46332920e-01 1.01564550e+00
1.52135348e+00 -3.19655418e-01 3.12071115e-01 2.47890458e-01
1.05961704e+00 6.26917422e-01 3.62481087e-01 7.39592016e-02
5.97456753e-01 9.82159972e-01 7.22665370e-01 -6.06471241e-01
-4.24165547e-01 -2.92514086e-01 -5.96855640e-01 4.69736189e-01
2.48599738e-01 -2.01867655e-01 -7.17477143e-01 4.95234311e-01
-2.03501582e+00 -7.01261163e-01 -2.99229980e-01 2.06812859e+00
5.64935692e-02 -1.14628226e-01 1.42208979e-01 -1.10074572e-01
1.09714746e+00 3.87524903e-01 -6.95542514e-01 -1.53792813e-01
-9.65139642e-02 2.07229629e-01 5.46030223e-01 3.28757316e-01
-1.25448978e+00 7.11378694e-01 5.09710217e+00 6.61142588e-01
-1.04525316e+00 -1.89008728e-01 4.65914100e-01 5.81470311e-01
-3.42779636e-01 6.71343831e-03 -4.16033298e-01 5.35422444e-01
2.21968144e-01 -4.09493297e-01 3.66655439e-01 7.76659966e-01
-3.76456708e-01 3.75325918e-01 -1.35432684e+00 1.52315485e+00
1.85690835e-01 -1.49207842e+00 3.30985934e-01 3.10813007e-03
5.37055314e-01 -4.97366525e-02 -5.07565588e-02 2.17109919e-01
8.68341625e-02 -9.26213145e-01 1.76180243e-01 3.26565266e-01
7.04815030e-01 -5.54809451e-01 7.65381455e-01 -2.92374883e-02
-1.84859061e+00 1.37537140e-02 -5.27473390e-01 3.03222179e-01
4.29255888e-02 4.69002783e-01 -5.18153131e-01 9.18784976e-01
6.57948971e-01 1.00034046e+00 -7.71950960e-01 1.48487973e+00
-1.29415060e-03 -1.40244842e-01 -9.64129418e-02 4.59112227e-02
2.78187752e-01 -3.84082526e-01 6.87948048e-01 9.55999374e-01
1.73492640e-01 -1.91921234e-01 4.71680820e-01 9.44907606e-01
-3.46164972e-01 3.97424132e-01 -6.09628737e-01 1.90047622e-01
3.66717279e-01 1.50127816e+00 -7.61884630e-01 -9.04899836e-02
-5.56669891e-01 8.54964793e-01 7.92449892e-01 1.91361442e-01
-7.54087925e-01 -4.91414160e-01 7.71841049e-01 2.05541342e-01
2.45236635e-01 4.02348600e-02 2.98938185e-01 -1.15691268e+00
2.07018524e-01 -6.24985754e-01 9.56655324e-01 -4.61799830e-01
-1.80365968e+00 7.10747063e-01 -1.74953178e-01 -1.23842549e+00
-4.10216376e-02 -6.62894726e-01 -8.28033149e-01 8.83052826e-01
-1.83779681e+00 -1.56934500e+00 -9.62128639e-01 7.67226160e-01
1.46879122e-01 -1.32588774e-01 6.01397634e-01 5.91745555e-01
-3.07540655e-01 1.07589567e+00 -2.58266121e-01 5.18057704e-01
5.61005473e-01 -1.37010658e+00 7.10666716e-01 7.51640081e-01
1.35391101e-01 4.30611163e-01 3.32198918e-01 -3.84193331e-01
-1.83277321e+00 -1.31963170e+00 8.10140252e-01 -1.37140989e-01
7.69119024e-01 -4.33653831e-01 -1.03499961e+00 3.69754285e-01
-1.84165582e-01 8.80389810e-01 3.43326807e-01 -1.25898868e-01
-7.17927158e-01 -2.51593500e-01 -1.11597335e+00 2.39139333e-01
1.41623902e+00 -4.85420913e-01 -2.67605007e-01 4.67377603e-01
7.94577956e-01 -6.90321445e-01 -1.16342473e+00 5.79169750e-01
4.59178329e-01 -8.11127663e-01 1.26569772e+00 -7.40376532e-01
1.67518738e-03 -1.11156523e-01 1.09050855e-01 -1.26095593e+00
-6.29341722e-01 -6.74642861e-01 1.79228678e-01 9.80379224e-01
4.56544869e-02 -7.88766325e-01 9.47322071e-01 5.01745582e-01
2.40367861e-03 -7.40801454e-01 -1.03599358e+00 -6.76588595e-01
3.08377035e-02 1.58880487e-01 7.65753627e-01 1.06259644e+00
-2.24096715e-01 2.30021864e-01 -1.53789133e-01 2.73691893e-01
9.64165986e-01 8.13231528e-01 1.10905409e+00 -1.56320679e+00
-2.38693208e-01 -8.06233764e-01 -1.32146478e+00 -1.11353827e+00
5.83659112e-01 -1.54672766e+00 -4.17245984e-01 -1.69442880e+00
2.50633329e-01 -5.66142440e-01 -2.99802840e-01 1.81293786e-01
-4.86635178e-01 5.73322140e-02 3.77111673e-01 -5.37432730e-02
-5.59735000e-01 3.18297088e-01 1.41152406e+00 -6.09543860e-01
9.72100813e-03 6.05083518e-02 -5.69273114e-01 8.19562450e-02
3.35842162e-01 -2.42754132e-01 -1.96358457e-01 -2.19742000e-01
1.60225913e-01 2.59652048e-01 7.59922862e-01 -8.27423573e-01
5.44820905e-01 2.77992398e-01 -8.15342963e-02 -4.50097591e-01
-1.72766954e-01 -1.08926523e+00 1.52454793e-01 5.46402276e-01
-4.48340774e-01 2.74340421e-01 3.38870846e-02 9.13591087e-01
-4.11037803e-01 1.93270326e-01 8.61222327e-01 -6.85844719e-02
-9.33419287e-01 1.23073471e+00 3.88179958e-01 3.88043910e-01
1.05702555e+00 -2.60000050e-01 -2.94567317e-01 -2.73808390e-01
-5.59323251e-01 4.82820302e-01 3.21790785e-01 7.22277820e-01
6.48020029e-01 -1.53082585e+00 -9.97602701e-01 2.40345135e-01
7.54044771e-01 -7.77590498e-02 4.67056513e-01 7.46333182e-01
-5.38301229e-01 2.88374603e-01 -2.36550719e-01 -8.21326375e-01
-1.38934910e+00 7.83488274e-01 6.54856861e-01 -6.07128620e-01
-9.07744646e-01 8.83767366e-01 3.32276642e-01 -5.34018457e-01
2.61322320e-01 -3.37599784e-01 -2.49548316e-01 -1.82204351e-01
3.15874726e-01 2.39646956e-01 1.53801695e-01 -8.14476192e-01
-3.89575332e-01 1.15074432e+00 -1.38826529e-02 9.01210189e-01
1.08580148e+00 -6.86639473e-02 -2.08050549e-01 -2.89185762e-01
1.71443260e+00 -4.69483763e-01 -8.40154111e-01 -6.76272035e-01
-4.05133143e-02 -7.69610226e-01 -1.33374885e-01 -2.22478136e-01
-1.54500306e+00 8.90332639e-01 6.45806611e-01 2.90176213e-01
9.03802633e-01 1.95239112e-01 9.73900616e-01 1.00922719e-01
4.16840136e-01 -4.61483628e-01 -6.16020039e-02 1.11393556e-01
9.97906923e-01 -1.57365680e+00 3.50503586e-02 -7.59696186e-01
-6.69628680e-02 1.20706940e+00 6.90851927e-01 -4.22140241e-01
9.12931025e-01 -1.95306078e-01 -1.55924469e-01 -5.45696080e-01
-2.74039179e-01 -3.91960055e-01 8.49635780e-01 6.01987720e-01
1.12087682e-01 6.35349825e-02 -9.88790989e-02 1.27671778e-01
5.14627174e-02 -4.23856109e-01 -1.10939458e-01 4.74866837e-01
-1.37160376e-01 -1.15237951e+00 1.43036455e-01 6.15181208e-01
-2.69325107e-01 1.85369533e-02 -4.68659401e-01 1.01391399e+00
-1.84122682e-01 7.46574044e-01 1.61104441e-01 -2.06224144e-01
5.93106687e-01 -6.07780635e-01 5.11862516e-01 -3.18439841e-01
-7.68354833e-01 -3.39183629e-01 -1.39410615e-01 -9.79077458e-01
-4.17686343e-01 -2.50623971e-01 -9.80414689e-01 -2.38598034e-01
-3.76861155e-01 6.10426590e-02 2.62337595e-01 5.80281258e-01
5.34575939e-01 4.22046632e-01 6.75707042e-01 -7.82151580e-01
-3.46234620e-01 -4.52415556e-01 -5.91011643e-01 1.14077544e+00
3.53532255e-01 -7.84995675e-01 -4.83531475e-01 -5.51344812e-01]
|
[7.134178161621094, 6.3704633712768555]
|
47aa5316-c0eb-4f4a-8538-444216ce984e
|
shapes2toon-generating-cartoon-characters
|
2211.02141
| null |
https://arxiv.org/abs/2211.02141v1
|
https://arxiv.org/pdf/2211.02141v1.pdf
|
Shapes2Toon: Generating Cartoon Characters from Simple Geometric Shapes
|
Cartoons are an important part of our entertainment culture. Though drawing a cartoon is not for everyone, creating it using an arrangement of basic geometric primitives that approximates that character is a fairly frequent technique in art. The key motivation behind this technique is that human bodies - as well as cartoon figures - can be split down into various basic geometric primitives. Numerous tutorials are available that demonstrate how to draw figures using an appropriate arrangement of fundamental shapes, thus assisting us in creating cartoon characters. This technique is very beneficial for children in terms of teaching them how to draw cartoons. In this paper, we develop a tool - shape2toon - that aims to automate this approach by utilizing a generative adversarial network which combines geometric primitives (i.e. circles) and generate a cartoon figure (i.e. Mickey Mouse) depending on the given approximation. For this purpose, we created a dataset of geometrically represented cartoon characters. We apply an image-to-image translation technique on our dataset and report the results in this paper. The experimental results show that our system can generate cartoon characters from input layout of geometric shapes. In addition, we demonstrate a web-based tool as a practical implication of our work.
|
['Md. Faraz Kabir Khan', 'Mohtasim Hossain Shovon', 'Md. Hasib Al Zadid', 'Md. Shafiur Rahman', 'Mohammad Imrul Jubair', 'Simanta Deb Turja']
|
2022-11-03
| null | null | null | null |
['culture']
|
['speech']
|
[ 3.15529555e-01 1.51697934e-01 5.33002913e-01 -1.57186076e-01
9.61612910e-02 -7.93840766e-01 6.81186020e-01 -1.94873407e-01
9.16806702e-03 6.07072353e-01 -1.42836779e-01 -3.31293136e-01
4.10155147e-01 -1.34056902e+00 -9.51686084e-01 -2.32482955e-01
2.71403164e-01 5.87494314e-01 4.44308251e-01 -5.16907692e-01
4.57365602e-01 9.74112749e-01 -1.53665054e+00 4.27599758e-01
1.01225460e+00 2.77151078e-01 -5.03735058e-02 8.84404361e-01
-4.93441612e-01 7.36760437e-01 -8.63236904e-01 -1.10929763e+00
3.17750543e-01 -7.12980330e-01 -5.00877202e-01 7.46294274e-04
4.57265407e-01 -4.84201223e-01 1.39669850e-01 9.68085110e-01
3.03214341e-01 -1.19814187e-01 9.55779314e-01 -1.44675434e+00
-6.75643921e-01 7.58230269e-01 -6.52738035e-01 -6.44738555e-01
5.88947058e-01 -1.04537420e-01 3.93358916e-01 -6.54963493e-01
8.30291152e-01 1.28070164e+00 7.64056623e-01 7.01338410e-01
-1.23207593e+00 -8.97467732e-01 -2.31349230e-01 -2.05563590e-01
-1.25610423e+00 1.79284558e-01 1.06593382e+00 -4.03443366e-01
2.14812055e-01 6.97566569e-01 1.39464855e+00 8.57935667e-01
2.56430805e-01 8.91693473e-01 1.12783289e+00 -8.53274703e-01
2.79176682e-01 2.41552636e-01 -4.15402025e-01 6.28881872e-01
2.59549320e-01 -3.02639127e-01 7.38210157e-02 1.11130744e-01
1.48799193e+00 4.19150814e-02 1.01763226e-01 -6.37771368e-01
-8.96403372e-01 7.63595700e-01 4.83882695e-01 3.75485152e-01
-5.49278744e-02 5.38516521e-01 1.81718007e-01 2.13788807e-01
3.78464721e-03 4.38331306e-01 3.37630033e-01 -1.79831371e-01
-9.78466988e-01 5.42320549e-01 9.02691126e-01 1.25146699e+00
6.45218909e-01 1.69084087e-01 1.52055249e-01 8.10012221e-01
7.16655850e-02 6.04512393e-01 7.03716427e-02 -8.77169728e-01
1.71543434e-01 7.73113906e-01 -9.33410898e-02 -1.25570369e+00
-5.05273603e-02 1.40009850e-01 -9.00042653e-01 9.96202946e-01
3.57971132e-01 -3.70889753e-02 -8.15825462e-01 1.38937664e+00
3.84537041e-01 -4.70098145e-02 -2.90074557e-01 6.41425014e-01
8.20594370e-01 6.76103711e-01 1.53771718e-03 5.34557819e-01
1.30080914e+00 -9.85705137e-01 -4.97014314e-01 3.31229806e-01
1.99920863e-01 -1.04620469e+00 1.28333890e+00 6.06662512e-01
-1.30227208e+00 -5.17310798e-01 -9.48181868e-01 -2.00316384e-01
-4.97953057e-01 -3.16179469e-02 6.47652805e-01 1.09572077e+00
-1.08681870e+00 7.05308378e-01 -4.99281883e-01 -3.66441965e-01
2.46922612e-01 1.22328550e-01 -2.36071810e-01 2.09235847e-01
-7.83726513e-01 7.19112635e-01 2.96460152e-01 -3.44720721e-01
-2.30390936e-01 -6.51656628e-01 -6.24063194e-01 2.54138671e-02
-6.23649359e-02 -9.40652788e-01 1.28162837e+00 -1.31411386e+00
-1.72985148e+00 8.43297541e-01 4.00777996e-01 -3.62761796e-01
9.58714545e-01 -2.53984425e-02 -1.73890799e-01 3.78270805e-01
-3.40854004e-02 1.07208991e+00 1.05926704e+00 -1.58621001e+00
-5.20278335e-01 1.89489886e-01 4.04862523e-01 1.39488861e-01
6.29170472e-03 -3.58621925e-02 -5.46907187e-01 -1.18225622e+00
2.68656909e-02 -9.74285245e-01 -1.27418771e-01 3.28323960e-01
-6.96508527e-01 1.04655884e-01 9.85787809e-01 -6.86277747e-01
1.10811114e+00 -1.84018421e+00 1.59176793e-02 5.36355674e-01
1.08570019e-02 6.56775385e-02 6.67381287e-02 8.16982090e-01
-1.00325793e-01 3.33486497e-01 -2.76599526e-01 -2.18392015e-01
-5.93882799e-02 -5.54720312e-02 -3.11573714e-01 -7.76881054e-02
1.01781771e-01 9.36515927e-01 -8.36896956e-01 -5.56985795e-01
5.55880785e-01 8.29075575e-01 -7.19611704e-01 3.26221809e-02
-3.45465213e-01 3.82751346e-01 -3.60869437e-01 6.24975622e-01
7.60688007e-01 3.56586039e-01 7.04095289e-02 -1.74740076e-01
-8.30695853e-02 -2.95223027e-01 -1.18522966e+00 1.56745934e+00
-5.91591775e-01 6.58631682e-01 -3.39541823e-01 -5.48330724e-01
1.10692430e+00 7.43807182e-02 2.79166609e-01 -3.60462695e-01
1.65144458e-01 2.77400672e-01 2.40782406e-02 -2.02768490e-01
5.82763374e-01 -2.34580964e-01 -1.39554575e-01 6.68520927e-01
-5.40885270e-01 -1.04003453e+00 3.29678863e-01 4.20338690e-01
7.88293123e-01 6.99027836e-01 3.85390967e-01 1.98904369e-02
5.07384658e-01 7.48685375e-02 -2.98085332e-01 4.66113240e-01
6.17936492e-01 9.87872362e-01 6.34959757e-01 -7.17712522e-01
-1.57711053e+00 -1.01999891e+00 1.13447957e-01 5.42664468e-01
-1.72971949e-01 -1.98916197e-01 -1.20373631e+00 -2.83293515e-01
-2.16466084e-01 1.02387130e+00 -7.75457799e-01 2.92313606e-01
-8.04234445e-01 -7.43274093e-02 4.90099877e-01 5.28478742e-01
4.28158551e-01 -1.44332647e+00 -1.19484043e+00 2.66720086e-01
3.40689778e-01 -7.45253801e-01 -5.68129420e-01 -3.83826584e-01
-8.13493013e-01 -9.41077828e-01 -1.24411094e+00 -1.02432394e+00
1.19562972e+00 1.93854377e-01 1.18424177e+00 3.54567647e-01
-3.07924777e-01 6.25599563e-01 -5.25704801e-01 -6.70253158e-01
-9.64102328e-01 -2.30442330e-01 -4.53099698e-01 -5.19857593e-02
-1.08940214e-01 -1.09788549e+00 -6.46469831e-01 2.99782567e-02
-1.38943470e+00 9.47534978e-01 5.38495719e-01 3.80103827e-01
4.96639341e-01 -4.03917730e-02 1.27175316e-01 -1.19664395e+00
6.64840221e-01 3.96550894e-02 -5.09228885e-01 1.82790712e-01
1.87677607e-01 4.93109412e-02 1.01316357e+00 -6.52290285e-01
-6.95571125e-01 1.31522596e-01 -2.68281877e-01 -4.37065244e-01
-2.62460530e-01 -5.59958890e-02 -1.49574578e-01 -3.37767601e-01
3.60380501e-01 2.65158951e-01 -1.06157422e-01 -2.94322520e-01
8.40847433e-01 4.26915199e-01 5.98337710e-01 -9.08246696e-01
1.24388695e+00 4.74452585e-01 2.61711657e-01 -7.84264028e-01
4.00530212e-02 1.51810661e-01 -7.01668859e-01 -6.06817961e-01
7.72085607e-01 -3.58920127e-01 -7.11381614e-01 5.12025833e-01
-1.39410353e+00 -2.72420198e-01 -4.88661587e-01 -2.24925071e-01
-8.31246197e-01 2.22558007e-01 -3.64099324e-01 -6.90914690e-01
-3.34360719e-01 -1.07773185e+00 9.19539928e-01 3.63217145e-01
-3.96201074e-01 -8.72972488e-01 1.57240018e-01 -6.21936657e-02
2.31687412e-01 1.08447301e+00 1.01822579e+00 -2.48977050e-01
-7.08076298e-01 -2.59431392e-01 -7.32839182e-02 1.55882493e-01
9.71990824e-02 5.66994846e-01 -5.25951624e-01 7.49531537e-02
-5.03935456e-01 2.54560299e-02 8.39104876e-02 1.58665758e-02
1.18360341e+00 -4.06942576e-01 -2.00838789e-01 5.73668897e-01
1.61894405e+00 6.09093130e-01 1.05015886e+00 5.25315940e-01
9.23662901e-01 5.56494534e-01 2.47078300e-01 1.53151229e-01
1.91188559e-01 6.86178625e-01 2.24405706e-01 -1.01977281e-01
-3.52945954e-01 -8.07058156e-01 1.00394845e-01 6.55269325e-01
-4.74681407e-01 5.99253215e-02 -7.61760831e-01 2.43523896e-01
-1.35173595e+00 -1.14871955e+00 -4.17817175e-01 2.04014397e+00
7.96611011e-01 1.97820351e-01 3.78172100e-01 4.29678917e-01
7.56748915e-01 -3.17634821e-01 -1.75105333e-01 -1.08319867e+00
1.43151596e-01 6.66853011e-01 9.38356221e-02 2.68882543e-01
-7.68684566e-01 8.79483819e-01 5.56520748e+00 9.25144911e-01
-1.11827338e+00 -3.33798110e-01 5.73546648e-01 3.30371112e-01
-6.24132454e-01 -1.67617068e-01 -2.50642300e-01 6.06576622e-01
1.72621086e-01 -2.12509111e-01 5.39099932e-01 9.45916653e-01
6.34713992e-02 -1.13764644e-01 -1.04513907e+00 8.68670762e-01
1.06698990e-01 -1.37928522e+00 6.08094692e-01 -3.10725361e-01
9.33611989e-01 -1.05955136e+00 1.29616093e-02 9.26332101e-02
4.05844241e-01 -1.17779863e+00 1.22360134e+00 5.31388104e-01
1.06034875e+00 -1.02897167e+00 1.07266881e-01 3.33790064e-01
-1.17998362e+00 5.66125870e-01 -2.15247542e-01 1.27274334e-01
-4.82152291e-02 1.93941727e-01 -8.85614336e-01 5.13411403e-01
4.71463174e-01 8.71837959e-02 -5.81917346e-01 1.23797584e+00
-5.53661466e-01 1.90811574e-01 -3.86092335e-01 -4.74115878e-01
1.03090234e-01 -6.39310956e-01 3.33077192e-01 1.29880917e+00
6.28680289e-01 2.53408670e-01 -4.57037807e-01 1.11217701e+00
-1.15095481e-01 4.55599785e-01 -8.78045976e-01 1.32162973e-01
3.97313327e-01 1.20915377e+00 -1.26691389e+00 -2.72019267e-01
-3.20151687e-01 1.30382526e+00 1.02056954e-02 9.23375860e-02
-1.00682938e+00 -8.57200205e-01 9.17607844e-02 5.19981980e-01
1.62442461e-01 -1.93654612e-01 -5.96725583e-01 -7.52111375e-01
-2.03599587e-01 -1.11324203e+00 -3.35786790e-01 -1.33373010e+00
-7.85338640e-01 6.88504815e-01 6.64899945e-02 -1.49777937e+00
-2.26790532e-01 -4.45857376e-01 -1.11890996e+00 7.06625879e-01
-7.61496782e-01 -1.50398362e+00 -3.11384022e-01 3.55200589e-01
5.37919164e-01 5.34516647e-02 7.97121167e-01 8.19862112e-02
-2.51349397e-02 5.47419846e-01 6.37365207e-02 3.85063171e-01
3.02242666e-01 -1.36942434e+00 8.35452437e-01 7.89838731e-01
4.41977441e-01 5.17109215e-01 8.90833914e-01 -6.91645801e-01
-1.06094837e+00 -6.60362542e-01 4.86441612e-01 -3.04086179e-01
4.29981679e-01 -4.48689014e-01 -6.40054286e-01 5.76968431e-01
3.51315171e-01 -6.37123644e-01 3.94982457e-01 -6.21999681e-01
-1.96661174e-01 2.16748029e-01 -1.30176723e+00 1.23663032e+00
1.00424826e+00 -3.50074992e-02 -6.64795041e-01 1.71575733e-02
4.90026653e-01 -6.29943252e-01 -5.62731147e-01 5.73058464e-02
9.35382605e-01 -1.34699285e+00 1.12241793e+00 -3.01875919e-01
8.00957084e-01 -4.37358946e-01 2.76337177e-01 -1.39757919e+00
-8.24361667e-03 -7.94489801e-01 1.09057046e-01 1.15957201e+00
4.09848951e-02 -3.39591056e-01 8.85700107e-01 6.04624212e-01
9.25607458e-02 -6.65118635e-01 -4.80814904e-01 -6.41050339e-01
2.79358327e-01 -4.57148194e-01 9.50347543e-01 7.92464375e-01
-2.40882844e-01 -1.48591012e-01 -3.11558962e-01 -2.13882267e-01
4.93258029e-01 2.88418502e-01 1.26112604e+00 -1.03276145e+00
-1.27828881e-01 -5.57748556e-01 -5.07185221e-01 -6.89701259e-01
-2.96390086e-01 -7.26498485e-01 -3.78988385e-01 -1.64514899e+00
-1.32298902e-01 -6.44174755e-01 5.51736116e-01 2.04903707e-01
7.60324895e-02 6.27457678e-01 7.20603406e-01 -3.39208506e-02
9.34644863e-02 6.03941940e-02 1.71247780e+00 -6.15802873e-03
-1.95574909e-01 7.50270039e-02 -5.78717709e-01 8.82400155e-01
9.26565468e-01 -3.84659410e-01 -3.73931766e-01 -2.48627543e-01
3.23454827e-01 -9.73977149e-02 3.50604445e-01 -1.40742791e+00
-7.32010677e-02 9.97363776e-03 5.27226627e-01 -7.98241735e-01
3.43767047e-01 -1.08199251e+00 8.26187849e-01 7.88226485e-01
-7.67605528e-02 2.75559962e-01 2.69963115e-01 -2.32257117e-02
1.42264843e-01 -5.86604893e-01 7.90571094e-01 -2.72172511e-01
-3.65935326e-01 -8.08740482e-02 -2.70271569e-01 -1.71698436e-01
1.25992417e+00 -6.87805891e-01 -9.09452699e-03 -7.03821003e-01
-5.81286311e-01 -3.66869569e-01 9.89421427e-01 3.39603454e-01
6.42227411e-01 -1.68323016e+00 -6.09416664e-01 2.91788787e-01
-2.93701798e-01 -8.10240023e-03 1.74453892e-02 8.92978162e-02
-1.65557671e+00 1.22205563e-01 -7.78520465e-01 -2.13524207e-01
-1.26691103e+00 5.98871708e-01 1.41276225e-01 -6.21211044e-02
-7.69016445e-01 5.74124813e-01 2.77300864e-01 -2.64549315e-01
-5.43406641e-04 -5.69244146e-01 -2.29608387e-01 -1.17057391e-01
4.81043965e-01 3.23966444e-01 -4.02049035e-01 -4.01390791e-01
3.06941997e-02 9.70770299e-01 3.41495663e-01 -2.82857925e-01
1.16216385e+00 3.59604001e-01 -1.05230212e-01 2.28900656e-01
8.47492456e-01 6.43048584e-01 -1.01000357e+00 4.73464787e-01
-4.83511478e-01 -5.70930481e-01 -7.25222826e-01 -6.51955426e-01
-8.97307873e-01 9.77737010e-01 3.14239740e-01 5.56550384e-01
1.15791810e+00 -3.40681404e-01 7.07046986e-01 4.88930196e-02
7.49621272e-01 -5.92768908e-01 1.76141530e-01 6.99395165e-02
1.30551755e+00 -6.54869854e-01 3.98623832e-02 -7.40995646e-01
-5.94109058e-01 1.60167110e+00 3.88442814e-01 -5.39143741e-01
3.01910847e-01 4.80613917e-01 3.54884155e-02 -5.65578640e-02
-7.90253729e-02 1.29891843e-01 3.01373869e-01 7.49597490e-01
4.24376488e-01 2.12035894e-01 -5.43152452e-01 3.64856809e-01
-7.99592376e-01 3.58956270e-02 9.09801960e-01 9.12954450e-01
-3.35177869e-01 -1.49029982e+00 -8.82425606e-01 1.13970615e-01
-2.35763520e-01 -9.53748524e-02 -5.15439510e-01 1.17541242e+00
3.47382843e-01 3.87226969e-01 4.41957749e-02 -2.35703930e-01
4.45698887e-01 -1.11014977e-01 7.66964555e-01 -4.91609544e-01
-6.81480348e-01 -3.07404459e-01 -9.74514261e-02 -1.91631436e-01
-4.25199062e-01 -4.89275426e-01 -1.13626575e+00 -6.98639393e-01
2.36997634e-01 1.80978552e-02 7.69294083e-01 2.25111052e-01
-2.23488912e-01 5.16847372e-01 6.01145446e-01 -1.10152507e+00
1.27423480e-01 -5.53357720e-01 -3.09885383e-01 5.68909049e-01
-2.48140544e-01 -2.84931093e-01 1.41787246e-01 3.83074760e-01]
|
[11.720819473266602, -0.41440901160240173]
|
65204952-c7a4-4327-8c4e-5bd42c140954
|
deep-learning-for-monaural-speech-separation
| null | null |
https://ieeexplore.ieee.org/document/6853860
|
https://minjekim.com/papers/icassp2014_phuang.pdf
|
Deep learning for monaural speech separation
|
Monaural source separation is useful for many real-world applications though it is a challenging problem. In this paper, we study deep learning for monaural speech separation. We propose the joint optimization of the deep learning models (deep neural networks and recurrent neural networks) with an extra masking layer, which enforces a reconstruction constraint. Moreover, we explore a discriminative training criterion for the neural networks to further enhance the separation performance. We evaluate our approaches using the TIMIT speech corpus for a monaural speech separation task. Our proposed models achieve about 3.8~4.9 dB SIR gain compared to NMF models, while maintaining better SDRs and SARs.
|
['Mark Hasegawa-Johnson', 'Po-Sen Huang', 'Paris Smaragdis', 'Minje Kim']
|
2014-05-04
| null | null | null |
icassp-2014-5
|
['multi-speaker-source-separation']
|
['speech']
|
[ 7.50205070e-02 -2.45192811e-01 5.23507781e-02 -2.02661112e-01
-1.15871882e+00 -2.35763326e-01 2.73081541e-01 -2.49419659e-01
-1.90224722e-01 6.59582317e-01 5.30454636e-01 -3.02875578e-01
-1.09727062e-01 -1.17009126e-01 -5.38632691e-01 -9.58015025e-01
2.97494382e-01 -1.48598805e-01 2.83000502e-03 -1.22678950e-01
-2.04230815e-01 2.88213313e-01 -1.57338250e+00 6.70274571e-02
1.20019460e+00 1.24841738e+00 4.74957645e-01 7.48647094e-01
3.60645801e-01 7.25146890e-01 -9.17161047e-01 -7.60622621e-02
3.38174999e-01 -7.87868500e-01 -2.50933409e-01 -1.47901163e-01
4.86745805e-01 -3.78502421e-02 -6.33111298e-01 1.08386493e+00
1.13233352e+00 4.43052918e-01 6.34655952e-01 -7.04998910e-01
-3.55073512e-01 7.61583626e-01 -3.91149729e-01 4.92719889e-01
-2.45199502e-01 1.23706218e-02 1.10808039e+00 -5.64491451e-01
-5.13765872e-01 1.00998127e+00 3.52306187e-01 3.32493573e-01
-1.11577845e+00 -9.07886744e-01 -8.90395194e-02 4.45916504e-01
-1.32855177e+00 -1.14902127e+00 7.84654617e-01 -2.01577097e-01
9.36546147e-01 4.02462631e-01 3.04418921e-01 8.14400017e-01
-4.46738489e-02 9.68135417e-01 1.02339566e+00 -3.72319907e-01
6.12225905e-02 -1.17752939e-01 1.37423530e-01 3.86532880e-02
-2.36230671e-01 2.39474058e-01 -5.07163167e-01 1.30934730e-01
6.71137214e-01 -2.51953006e-01 -6.95379317e-01 3.24123800e-01
-7.04546809e-01 5.60759544e-01 4.88946676e-01 3.75338078e-01
-3.55779767e-01 1.18461236e-01 8.58432725e-02 3.11108500e-01
5.55781782e-01 5.08162141e-01 -3.31520289e-01 -1.17128417e-01
-1.08509588e+00 -6.20974861e-02 5.46707153e-01 5.99414110e-01
2.74072260e-01 9.19925094e-01 -2.83222407e-01 1.74173224e+00
3.70522648e-01 7.23154902e-01 6.84192836e-01 -8.28249097e-01
3.44835371e-01 -4.04209852e-01 -1.79850116e-01 -6.34207726e-01
-2.41865262e-01 -1.22497535e+00 -9.08576131e-01 -1.29544586e-01
-4.09235768e-02 -3.09612900e-01 -1.09297264e+00 1.73964024e+00
-8.00608322e-02 6.60179913e-01 3.25980484e-01 1.17034888e+00
8.93919945e-01 1.03263104e+00 -3.44957322e-01 -5.74422657e-01
1.09912014e+00 -1.37355971e+00 -1.11208725e+00 -4.40895587e-01
-4.85426858e-02 -1.13044024e+00 6.27491236e-01 5.92104197e-01
-1.42263675e+00 -7.36746907e-01 -1.04250026e+00 8.16496462e-02
2.25403383e-01 4.61783439e-01 1.91894978e-01 6.97254062e-01
-8.55517983e-01 2.73813903e-01 -6.71523988e-01 2.20751688e-01
-4.89588417e-02 3.91225547e-01 1.80477649e-01 2.33959064e-01
-1.26233351e+00 6.56835854e-01 1.28460124e-01 4.11082685e-01
-1.12003994e+00 -4.32565898e-01 -8.80386472e-01 3.88005555e-01
9.94367972e-02 -3.18534583e-01 1.58944476e+00 -7.42720306e-01
-1.90963793e+00 5.28094113e-01 -3.77894044e-01 -7.51247823e-01
-1.85223427e-02 -5.22675633e-01 -8.82828057e-01 -8.20916593e-02
-2.12203071e-01 2.02986032e-01 9.35629547e-01 -1.14378655e+00
-6.62587941e-01 -9.93550047e-02 -2.55703330e-01 4.20668036e-01
-3.42950642e-01 2.43195415e-01 -2.70733625e-01 -1.07407713e+00
3.59284490e-01 -6.77415788e-01 -2.68135428e-01 -8.72749746e-01
-4.77541625e-01 9.45757031e-02 4.49255705e-01 -1.13942432e+00
1.33457148e+00 -2.49685979e+00 2.42869824e-01 -1.52797857e-02
1.98429361e-01 5.93298376e-01 -1.18129365e-01 -1.21846400e-01
-1.52080394e-02 -3.97206992e-01 -2.05013782e-01 -6.12804174e-01
3.40732262e-02 -6.31301627e-02 -5.04034936e-01 3.64630729e-01
-1.09148294e-01 4.50771034e-01 -4.03894842e-01 -2.95940451e-02
1.21281587e-01 6.07431650e-01 -4.55849230e-01 7.46398151e-01
2.43853688e-01 3.66329461e-01 2.27447763e-01 3.92447531e-01
6.91656232e-01 1.90653592e-01 -3.62657756e-02 -1.19891256e-01
-8.54441673e-02 8.21218789e-01 -9.59617019e-01 1.46637714e+00
-8.29662681e-01 9.48758066e-01 4.22919214e-01 -1.10490739e+00
1.20849431e+00 6.69222534e-01 7.86215663e-02 -8.64278913e-01
4.47199732e-01 5.18000841e-01 4.75543618e-01 -1.52406707e-01
3.29732031e-01 -5.51519454e-01 2.39490315e-01 1.78081155e-01
3.23695779e-01 -1.72189385e-01 -2.95859665e-01 -3.89524907e-01
5.85567832e-01 -5.46968699e-01 -1.20837748e-01 -8.86932462e-02
5.43226898e-01 -8.13126922e-01 8.28892767e-01 5.55299640e-01
-2.18048587e-01 8.67172360e-01 6.35989085e-02 2.70896196e-01
-6.42044604e-01 -1.29044271e+00 -1.95689902e-01 9.86326575e-01
1.96221936e-02 -1.39445320e-01 -6.38087153e-01 -8.77458081e-02
-3.85962456e-01 8.21706653e-01 8.10577422e-02 -4.63918120e-01
-7.14732945e-01 -9.19457316e-01 7.73850501e-01 5.68659365e-01
4.05441791e-01 -6.33793533e-01 -4.97274324e-02 1.39179930e-01
-4.77061301e-01 -1.19430518e+00 -6.03317261e-01 5.20775259e-01
-5.06179988e-01 -3.20706993e-01 -1.01487386e+00 -1.09047639e+00
1.01091489e-01 3.98331523e-01 6.80001318e-01 -5.97437739e-01
1.76064506e-01 -1.89802125e-01 -7.54558668e-02 -4.07437861e-01
-3.29940856e-01 5.40371202e-02 3.82869661e-01 1.97810724e-01
8.99673402e-02 -1.02923203e+00 -5.95170617e-01 4.11289692e-01
-5.99468231e-01 -8.16358477e-02 6.30403519e-01 7.84958541e-01
4.71581459e-01 3.15274268e-01 8.16926479e-01 3.05755883e-02
7.31411695e-01 -2.78375983e-01 -6.57277226e-01 -1.33627504e-01
-2.08600372e-01 -1.10241033e-01 7.18827546e-01 -4.61413324e-01
-1.22632408e+00 -2.51678854e-01 -5.62485516e-01 -6.42154455e-01
-1.14673585e-01 4.25269157e-01 -4.07268107e-01 1.59845248e-01
4.39370334e-01 3.51604044e-01 -3.08425248e-01 -9.18935359e-01
5.70038781e-02 1.19492579e+00 7.42932200e-01 -2.94834644e-01
5.05798817e-01 -8.67980439e-03 -5.45451999e-01 -9.94635820e-01
-9.16328430e-01 -5.54483294e-01 -1.01691194e-01 2.43518706e-02
6.59516037e-01 -1.23288214e+00 -4.99912798e-01 5.92463017e-01
-1.13267756e+00 -2.37250224e-01 -8.47990625e-03 1.19287133e+00
-5.46784699e-01 3.62333134e-02 -7.62382567e-01 -1.09470379e+00
-4.23094302e-01 -1.30455446e+00 5.66293657e-01 3.36452633e-01
2.20533967e-01 -6.54143512e-01 1.53913841e-01 5.58358490e-01
6.54009879e-01 -3.00125480e-01 4.75530803e-01 -7.44477868e-01
-3.58623266e-01 1.70000091e-01 -6.79621249e-02 1.02920222e+00
3.91835749e-01 -5.07351935e-01 -1.48074746e+00 -3.32918644e-01
4.79232132e-01 -1.37756631e-01 1.09492123e+00 7.97120452e-01
1.09196949e+00 -4.04528707e-01 7.55857602e-02 9.07333016e-01
9.03861225e-01 7.72447586e-01 6.67224824e-01 -1.04345858e-01
6.85963094e-01 4.13802028e-01 2.30051547e-01 3.43617111e-01
-1.59758925e-02 7.32375443e-01 1.68536782e-01 -3.00889969e-01
-3.61153185e-01 9.33169350e-02 4.53801066e-01 1.70002806e+00
9.22918394e-02 -2.43920758e-01 -8.12347174e-01 8.23964000e-01
-1.70931566e+00 -6.75135136e-01 6.37474982e-03 2.34548306e+00
1.06611264e+00 6.60751984e-02 -2.39032339e-02 4.28873330e-01
7.47613847e-01 4.06116813e-01 -4.07708853e-01 -5.19880533e-01
-3.49064916e-01 5.60838759e-01 1.56561702e-01 6.38233364e-01
-1.07554460e+00 7.38444746e-01 6.53908634e+00 1.21334279e+00
-1.47058761e+00 2.34674498e-01 4.99387532e-01 -4.79364097e-01
-8.50556120e-02 -3.79484475e-01 -6.67931378e-01 3.58227849e-01
1.21584952e+00 -1.01297885e-01 5.73507488e-01 3.53012711e-01
2.96638429e-01 1.72243237e-01 -8.18964720e-01 1.25723159e+00
2.04595104e-01 -7.83724248e-01 -2.34629557e-01 -1.15706518e-01
6.33898854e-01 1.58280745e-01 4.26333904e-01 2.76643336e-01
1.98398232e-01 -1.24311578e+00 5.98126173e-01 2.37305716e-01
5.66785097e-01 -1.06594563e+00 5.62631309e-01 3.54634196e-01
-9.57018197e-01 -7.75937289e-02 -2.63587505e-01 -6.49308115e-02
3.24531227e-01 8.63981187e-01 -7.65183568e-01 5.59023440e-01
6.45626187e-01 5.18122256e-01 -1.00672364e-01 1.41637719e+00
-3.95164937e-01 1.06982028e+00 -2.48904929e-01 2.49379009e-01
1.46192774e-01 -7.16784224e-02 8.52859974e-01 1.34453845e+00
2.85483032e-01 -4.63083796e-02 -1.03047535e-01 7.17161536e-01
-1.29620776e-01 4.26776037e-02 -1.74556464e-01 -8.35811999e-03
3.70196164e-01 9.58713710e-01 -1.52065948e-01 2.42270250e-02
-1.79154888e-01 7.94338584e-01 3.46724652e-02 7.03299820e-01
-1.04084027e+00 -8.47610891e-01 1.04819214e+00 -3.19563031e-01
4.37689662e-01 -2.56474108e-01 -1.87075347e-01 -1.12520206e+00
-5.12886383e-02 -1.09331679e+00 -4.64604944e-02 -7.28925586e-01
-1.03820419e+00 9.10822153e-01 -4.12520289e-01 -1.21947789e+00
-2.66084552e-01 -3.88085246e-01 -6.63673282e-01 1.29291725e+00
-1.77727628e+00 -7.26308942e-01 1.00231119e-01 5.94925344e-01
4.96469527e-01 -3.60679716e-01 4.51429635e-01 7.86931098e-01
-8.45393240e-01 8.20683122e-01 5.00807822e-01 1.10391125e-01
7.55712688e-01 -1.04101801e+00 3.07768166e-01 1.08132052e+00
2.29205221e-01 5.57477653e-01 7.89031088e-01 -1.03935003e-01
-7.74870992e-01 -1.01816797e+00 6.61033690e-01 3.24162811e-01
5.14041185e-01 -5.04523158e-01 -9.13547635e-01 4.86688554e-01
5.65478504e-01 -3.06784183e-01 9.43939626e-01 2.14276105e-01
-2.99998641e-01 -4.15989518e-01 -5.38322747e-01 5.46666861e-01
6.84947908e-01 -7.07822382e-01 -8.18697512e-01 4.31761071e-02
1.16772366e+00 -4.72530514e-01 -5.54781199e-01 6.21896088e-01
4.15487736e-01 -8.27172756e-01 1.00971389e+00 -3.88517320e-01
2.55508035e-01 -3.94699335e-01 -5.42296946e-01 -1.80761981e+00
-2.88899273e-01 -9.44344163e-01 -3.01979512e-01 1.36138272e+00
4.98424679e-01 -7.33697832e-01 1.91642568e-01 -7.66992047e-02
-5.43221891e-01 -5.54175735e-01 -8.66108179e-01 -1.12662625e+00
-8.18169937e-02 -4.36781615e-01 3.29923749e-01 7.75405228e-01
-1.94286257e-01 5.53606451e-01 -7.00402141e-01 4.34077799e-01
4.92537349e-01 1.22983888e-01 4.81065184e-01 -7.82583654e-01
-7.89989114e-01 -7.04822302e-01 -1.38963267e-01 -1.54934096e+00
2.07231343e-01 -5.64626157e-01 3.71799558e-01 -1.43997657e+00
-3.03959072e-01 -2.02984393e-01 -9.18465614e-01 1.85799226e-01
-2.48258069e-01 1.93341464e-01 1.94933400e-01 5.29198069e-03
-2.55687118e-01 1.07963252e+00 9.18267548e-01 -1.79666758e-01
-4.02180344e-01 4.44821209e-01 -7.30746686e-01 7.20772207e-01
1.03402638e+00 -3.23335648e-01 -2.91572243e-01 -6.40492260e-01
-3.65290046e-01 2.30606675e-01 5.97450770e-02 -1.29937899e+00
2.91399956e-01 3.76224145e-02 -1.19934289e-03 -4.92375672e-01
8.18259001e-01 -5.27623892e-01 -2.01246329e-02 2.18368426e-01
-4.34952796e-01 -7.77967691e-01 5.27890086e-01 3.11716348e-01
-7.61382163e-01 -2.14720219e-02 9.99052823e-01 3.25474411e-01
-5.02459668e-02 2.77143456e-02 -4.12048131e-01 -8.16656798e-02
4.23861444e-01 2.38919467e-01 -1.79478690e-01 -6.05744362e-01
-7.70559907e-01 1.75533086e-01 -3.35036933e-01 4.36457545e-01
6.61763966e-01 -1.42852807e+00 -8.75915229e-01 2.56352216e-01
-2.76704758e-01 -1.04443178e-01 2.99690723e-01 1.03472054e+00
5.62061928e-02 4.28483635e-01 -1.28858700e-01 -5.74941337e-01
-1.30289423e+00 2.37089872e-01 6.86713338e-01 2.06174508e-01
-1.40026510e-01 1.07557905e+00 7.03319371e-01 -2.45157480e-01
5.82910240e-01 -2.17655480e-01 -3.87108624e-01 -1.16466254e-01
5.18610895e-01 5.55707395e-01 2.23442003e-01 -5.53517997e-01
-1.49585441e-01 2.78828532e-01 2.19407342e-02 -3.45457971e-01
1.30135942e+00 -2.76483744e-01 -4.04606424e-02 5.99183500e-01
1.33066010e+00 4.34461147e-01 -1.03864777e+00 -5.67239225e-01
-2.78880149e-01 -3.77801985e-01 5.00279188e-01 -8.14830124e-01
-1.23997867e+00 1.20396948e+00 5.26738644e-01 2.50008196e-01
1.63035882e+00 -5.74491806e-02 1.18098104e+00 1.19170889e-01
-2.48333931e-01 -1.03397250e+00 1.00617148e-02 7.28414357e-01
8.63134027e-01 -1.01421809e+00 -4.42678690e-01 -1.57778949e-01
-4.12162930e-01 6.97967947e-01 5.56784570e-01 -1.16152829e-03
6.31781816e-01 3.55784118e-01 3.32992196e-01 2.62209862e-01
-6.50589347e-01 -5.87893426e-01 5.61872482e-01 2.80315369e-01
5.77993929e-01 1.72999337e-01 -1.49216309e-01 9.80083466e-01
-5.02686858e-01 -7.16461539e-01 2.21227333e-01 3.40755433e-01
-6.89194679e-01 -7.65006065e-01 -6.24832928e-01 2.25132987e-01
-7.99919188e-01 -4.40470248e-01 -2.57690817e-01 -1.05377465e-01
-2.30602175e-01 1.33045506e+00 2.47147325e-02 -4.78174686e-01
4.36581671e-01 -4.21375632e-02 2.42256388e-01 -4.34060186e-01
-4.99929726e-01 1.07264769e+00 1.41773701e-01 -1.09448962e-01
-3.93007606e-01 -5.18891931e-01 -1.03846216e+00 2.30011363e-02
-7.19027698e-01 3.92037421e-01 5.51405132e-01 9.17705297e-01
1.95444003e-01 1.13930631e+00 1.09594488e+00 -5.58269620e-01
-3.67343366e-01 -1.04794312e+00 -8.01846981e-01 -1.47487536e-01
9.24470246e-01 -4.02640253e-01 -5.52648306e-01 3.83532420e-02]
|
[14.954517364501953, 5.860958099365234]
|
a54ce8b2-6ae2-498a-a210-165114ed6e6f
|
a-k-nearest-neighbor-approach-towards-multi
| null | null |
https://aclanthology.org/N19-2019
|
https://aclanthology.org/N19-2019.pdf
|
A k-Nearest Neighbor Approach towards Multi-level Sequence Labeling
|
In this paper we present a new method for intent recognition for complex dialog management in low resource situations. Complex dialog management is required because our target domain is real world mixed initiative food ordering between agents and their customers, where individual customer utterances may contain multiple intents and refer to food items with complex structure. For example, a customer might say {``}Can I get a deluxe burger with large fries and oh put extra mayo on the burger would you?{''} We approach this task as a multi-level sequence labeling problem, with the constraint of limited real training data. Both traditional methods like HMM, MEMM, or CRF and newer methods like DNN or BiLSTM use only homogeneous feature sets. Newer methods perform better but also require considerably more data. Previous research has done pseudo-data synthesis to obtain the required amounts of training data. We propose to use a k-NN learner with heterogeneous feature set. We used windowed word n-grams, POS tag n-grams and pre-trained word embeddings as features. For the experiments we perform a comparison between using pseudo-data and real world data. We also perform semi-supervised self-training to obtain additional labeled data, in order to better model real world scenarios. Instead of using massive pseudo-data, we show that with only less than 1{\%} of the data size, we can achieve better result than any of the methods above by annotating real world data. We achieve labeled bracketed F-scores of 75.46, 52.84 and 49.66 for the three levels of sequence labeling where each level has a longer word span than its previous level. Overall we achieve 60.71F. In comparison, two previous systems, MEMM and DNN-ELMO, achieved 52.32 and 45.25 respectively.
|
['Yue Chen', 'John Chen']
|
2019-06-01
| null | null | null |
naacl-2019-6
|
['intent-recognition']
|
['natural-language-processing']
|
[ 1.92831516e-01 4.25394833e-01 -4.38456088e-02 -7.88329244e-01
-4.69102293e-01 -6.60711706e-01 5.53321183e-01 3.11449111e-01
-8.55794966e-01 1.05735755e+00 4.03341353e-01 -3.37622732e-01
3.72317165e-01 -5.57437956e-01 -2.54704773e-01 -3.93351614e-01
3.23105454e-02 1.04314649e+00 1.84780002e-01 -6.33287191e-01
1.62951931e-01 -6.47565499e-02 -1.18009114e+00 3.63542616e-01
6.45963550e-01 6.19275928e-01 7.43400693e-01 8.88638258e-01
-5.28465807e-01 8.34315479e-01 -8.93486023e-01 -2.76970595e-01
9.36320648e-02 -3.08629185e-01 -1.11171591e+00 3.80461663e-01
-1.92460582e-01 -4.93987292e-01 8.39374214e-02 7.49209464e-01
3.57190609e-01 4.92702931e-01 6.26947999e-01 -1.16795957e+00
-5.53578854e-01 9.23183084e-01 -1.55382022e-01 7.09342305e-03
7.76397407e-01 1.77060395e-01 9.14936066e-01 -5.33320487e-01
5.27234256e-01 1.32561505e+00 4.91910964e-01 8.14490736e-01
-1.03426576e+00 -3.72279853e-01 1.93268523e-01 -6.85575679e-02
-9.31870222e-01 -3.12527657e-01 3.34028095e-01 -2.35676706e-01
1.45286369e+00 1.86785132e-01 2.04059899e-01 1.23246431e+00
5.80434017e-02 8.36912751e-01 1.14507914e+00 -7.10643589e-01
1.59842193e-01 4.72347379e-01 6.47276878e-01 6.74328208e-01
-5.17556146e-02 -1.60506368e-01 -2.33242378e-01 -2.94122905e-01
4.88656878e-01 1.59876361e-01 3.42124850e-02 2.26428896e-01
-1.23741496e+00 1.24327421e+00 -9.22520384e-02 6.25569046e-01
-3.27803165e-01 -3.43317181e-01 3.90692592e-01 3.67055476e-01
2.84406841e-01 5.06174505e-01 -9.93167281e-01 -3.51952106e-01
-4.10594672e-01 2.51279384e-01 1.43883431e+00 1.29276419e+00
6.97580755e-01 1.77571904e-02 -9.94941592e-02 9.81952429e-01
4.88942355e-01 2.58646876e-01 9.54693258e-01 -9.79969978e-01
5.39739668e-01 6.25650704e-01 5.54147780e-01 -3.83622497e-01
-7.34313905e-01 1.93720639e-01 -5.98403454e-01 -5.15728556e-02
8.14186096e-01 -5.86902142e-01 -1.05608690e+00 1.74278951e+00
1.71169177e-01 -4.02218789e-01 6.00440621e-01 6.95323110e-01
8.31063390e-01 8.96619022e-01 2.36989141e-01 -4.21214163e-01
1.83068788e+00 -1.28383601e+00 -1.07419193e+00 -4.74442065e-01
7.79884160e-01 -1.05979371e+00 1.30503929e+00 2.88737088e-01
-8.75508308e-01 -5.68229318e-01 -9.01781678e-01 2.22319551e-02
-6.11886203e-01 6.45304546e-02 7.07735837e-01 6.26139104e-01
-9.46646094e-01 3.30200881e-01 -7.34222174e-01 -7.37361848e-01
-5.04999161e-01 6.00706220e-01 -3.37305665e-01 -6.99926987e-02
-1.39208376e+00 1.25074065e+00 6.63364470e-01 -4.55982313e-02
-5.05649984e-01 -1.21607848e-01 -1.23343134e+00 -7.15605021e-02
4.78106171e-01 -3.00586611e-01 1.75526404e+00 -5.86525738e-01
-1.89446127e+00 6.11718953e-01 -1.90529704e-01 -5.58847845e-01
7.62599781e-02 -1.66640028e-01 -4.69486564e-01 -3.10184926e-01
8.16733297e-03 8.63135278e-01 1.97574019e-01 -1.04993713e+00
-9.12123144e-01 -1.91365972e-01 3.44131172e-01 3.12098920e-01
-1.32766604e-01 1.97570652e-01 3.34926806e-02 -2.84094572e-01
-1.02381177e-01 -1.02252603e+00 -5.27931213e-01 -7.39474118e-01
-3.83820385e-01 -5.29434085e-01 5.90346694e-01 -5.83316863e-01
9.66584802e-01 -1.92329144e+00 -1.76578730e-01 -2.99388111e-01
-3.35291438e-02 4.43604082e-01 -6.89443350e-02 7.96916068e-01
2.59360433e-01 -5.72520047e-02 -2.64438093e-01 -4.50146466e-01
3.34433407e-01 7.68008709e-01 1.11194059e-01 -4.49037887e-02
1.16093308e-01 6.25223458e-01 -7.96946347e-01 -5.24989843e-01
4.20697182e-01 1.30184412e-01 -3.67214322e-01 6.27842486e-01
-4.69748080e-01 2.76697040e-01 -2.76610345e-01 3.67916733e-01
4.61566627e-01 -1.17962606e-01 6.09045327e-01 -8.74024257e-02
-1.82581529e-01 4.19229776e-01 -1.13522935e+00 1.65547407e+00
-7.39489198e-01 2.17485979e-01 2.23277032e-01 -9.12226617e-01
1.01118708e+00 5.60763299e-01 1.60445631e-01 -2.71238774e-01
3.06936562e-01 1.54723838e-01 3.32849920e-01 -6.84932232e-01
7.75739193e-01 -2.99832433e-01 -5.47167957e-01 4.73549455e-01
2.84642160e-01 6.97445404e-03 4.58405674e-01 -2.08063293e-02
1.15354812e+00 -8.28244612e-02 6.24852061e-01 -2.65487880e-01
4.58188206e-01 2.57793516e-01 4.81304586e-01 6.32494450e-01
-4.19314563e-01 1.44794688e-01 2.00597972e-01 -4.39804286e-01
-1.01935792e+00 -6.78064764e-01 5.39134592e-02 1.51022172e+00
-1.40506318e-02 -1.70675665e-01 -6.54885411e-01 -6.89930797e-01
-3.33396554e-01 1.08351111e+00 -2.31354788e-01 3.29366952e-01
-7.96686172e-01 -8.89801264e-01 3.85770857e-01 3.64084840e-01
5.64138472e-01 -1.49490762e+00 -6.58859611e-01 6.61408722e-01
-2.99247384e-01 -1.31696296e+00 -6.23591065e-01 6.77282572e-01
-5.71250856e-01 -7.39845395e-01 -4.93936360e-01 -1.20806527e+00
3.89918298e-01 -5.52947409e-02 1.21429455e+00 -1.88455448e-01
-8.30205530e-02 1.39614325e-02 -7.27492809e-01 -2.84111440e-01
-7.83315063e-01 2.46527657e-01 2.33281419e-01 -4.37994897e-01
9.05583620e-01 -3.50069910e-01 -2.76469171e-01 3.83948684e-01
-8.66764247e-01 -4.09454741e-02 6.43593907e-01 1.13337326e+00
-8.65134299e-02 -1.64643034e-01 6.86958730e-01 -1.02985024e+00
8.60556781e-01 -4.14561063e-01 -2.29420349e-01 2.53120810e-01
-5.03504455e-01 3.14354360e-01 7.41287112e-01 -7.60432005e-01
-1.26127386e+00 2.21813664e-01 -5.36277890e-01 3.84627938e-01
-7.37853110e-01 4.12066430e-01 -2.29492635e-01 5.74161172e-01
4.74067181e-01 9.21563059e-02 1.25908554e-01 -5.21265209e-01
3.95388186e-01 1.15989804e+00 2.79590964e-01 -3.88364166e-01
2.14846864e-01 -1.63978532e-01 -6.49347901e-01 -9.11963761e-01
-7.35261261e-01 -5.82821429e-01 -5.21512985e-01 1.56449556e-01
1.04630518e+00 -6.25767052e-01 -1.09699965e+00 2.44605869e-01
-1.29718280e+00 -5.92613399e-01 -1.29023716e-01 6.70831323e-01
-5.11196017e-01 4.47539002e-01 -1.11272883e+00 -1.13914919e+00
-3.41459095e-01 -1.16010523e+00 7.37432897e-01 2.50706255e-01
-6.61223710e-01 -1.13538826e+00 4.99359518e-02 4.65264052e-01
3.84261638e-01 -3.87279913e-02 8.65181863e-01 -1.55784035e+00
5.53589463e-02 -1.96321420e-02 5.51607162e-02 3.55375528e-01
6.25648499e-01 -5.31438291e-01 -7.42749155e-01 -2.28069007e-01
3.49038422e-01 -5.57320714e-01 3.27962756e-01 9.63660255e-02
3.01904798e-01 -4.93637204e-01 -1.13651514e-01 -3.11743408e-01
1.27992332e+00 8.29793513e-01 2.83119023e-01 1.85117006e-01
3.40137094e-01 9.02600348e-01 9.22165275e-01 5.57027340e-01
7.03558683e-01 7.28595614e-01 5.16795516e-02 6.95116967e-02
2.13532254e-01 -1.36975780e-01 6.14323378e-01 1.11673379e+00
3.69548649e-01 -5.97427130e-01 -8.84866595e-01 5.22708654e-01
-1.99799228e+00 -6.45998240e-01 -1.18615016e-01 1.83982360e+00
1.13819659e+00 1.34962812e-01 2.20470443e-01 -7.94695616e-02
6.35562539e-01 -5.41535690e-02 -2.97586918e-01 -7.47521281e-01
1.28799260e-01 9.30701867e-02 5.28150678e-01 9.23476338e-01
-1.06524694e+00 1.11103165e+00 5.91369057e+00 6.54324293e-01
-5.59655666e-01 2.42491946e-01 5.24973333e-01 3.94673645e-01
9.60204601e-02 -1.20197982e-01 -1.15294039e+00 4.63507533e-01
1.38475430e+00 1.67678490e-01 3.79390925e-01 7.85499215e-01
6.18338734e-02 -4.43640709e-01 -1.19498682e+00 7.19050288e-01
6.86790720e-02 -8.19733322e-01 -3.81794542e-01 -6.73108399e-02
2.95953840e-01 -2.08943099e-01 -4.11728352e-01 7.67816544e-01
1.08221602e+00 -9.17676449e-01 2.19808161e-01 1.65139467e-01
1.80009812e-01 -4.68533754e-01 1.23625720e+00 7.59593546e-01
-1.06069088e+00 4.53413092e-03 -5.46157479e-01 -3.48277390e-01
5.13106108e-01 2.71668553e-01 -1.32580650e+00 5.30614257e-01
2.44623527e-01 1.44322659e-03 -1.05787476e-03 4.27526385e-01
7.75845125e-02 4.24116611e-01 -5.53393781e-01 -5.76113999e-01
5.00890613e-01 -2.12283626e-01 9.10352767e-02 1.34401274e+00
2.01227367e-01 2.99255341e-01 7.22392499e-01 3.91769052e-01
2.05631733e-01 1.93814889e-01 -5.77418566e-01 2.75246589e-03
3.03929806e-01 1.21345544e+00 -6.93471789e-01 -5.21310329e-01
-5.21916270e-01 1.13444340e+00 2.12744296e-01 8.58527347e-02
-6.62568033e-01 -3.20861280e-01 4.44143087e-01 -3.75477105e-01
1.46693900e-01 -3.94391984e-01 9.85076725e-02 -9.51226532e-01
-3.30580413e-01 -9.38206911e-01 3.43146265e-01 -3.65716428e-01
-1.52565312e+00 1.00896108e+00 1.47861034e-01 -8.22228134e-01
-8.67054284e-01 -7.63377428e-01 -3.90122890e-01 8.91012430e-01
-1.14028215e+00 -1.10901880e+00 3.88958342e-02 1.99313521e-01
1.27181435e+00 -2.71905005e-01 1.39001381e+00 2.22523659e-01
-4.29120839e-01 4.34039235e-01 -8.70635509e-02 1.65122867e-01
7.15288162e-01 -1.42484784e+00 2.65976846e-01 2.35284969e-01
3.52023393e-02 5.90499818e-01 1.09131837e+00 -6.91010416e-01
-1.22829497e+00 -8.12397003e-01 1.32764375e+00 -4.62662101e-01
6.65326536e-01 -6.74516141e-01 -8.24797511e-01 8.69092345e-01
7.67625690e-01 -5.42035043e-01 9.34113443e-01 1.28861412e-01
2.28032783e-01 3.22558969e-01 -1.44911718e+00 4.97517765e-01
7.58100748e-01 -9.00989994e-02 -1.01267612e+00 6.18039191e-01
1.06783617e+00 -2.73035854e-01 -9.63540077e-01 1.84574962e-01
3.30227196e-01 -7.30251014e-01 5.62750459e-01 -7.43381679e-01
6.35044649e-02 -7.90842250e-02 -5.12054682e-01 -1.30342960e+00
-1.58103891e-02 -5.71770906e-01 2.33180806e-01 1.47726476e+00
8.22748005e-01 -6.43116534e-01 8.30351233e-01 8.77409160e-01
-2.23045960e-01 -5.97728968e-01 -6.52167499e-01 -7.85348058e-01
-2.42530499e-02 -2.24959597e-01 6.56513512e-01 8.07471216e-01
5.40448010e-01 8.03017616e-01 -6.00845158e-01 2.02617655e-03
1.87458485e-01 3.65330651e-02 6.17679000e-01 -9.22113061e-01
-3.91382635e-01 1.18550166e-01 1.17911138e-02 -1.35636902e+00
2.95849532e-01 -4.25838590e-01 4.61182624e-01 -1.58872652e+00
-5.06217740e-02 -4.41415459e-01 1.95793938e-02 6.60298288e-01
-2.88717281e-02 6.39071409e-03 1.67564929e-01 -1.16262391e-01
-4.67366248e-01 3.38264108e-01 8.58508170e-01 -8.86998028e-02
-4.77829754e-01 1.51083574e-01 -3.31771523e-01 7.39467919e-01
1.10332549e+00 -3.96749049e-01 -4.85734135e-01 -1.01782098e-01
-3.46669286e-01 4.49575812e-01 -2.77958721e-01 -5.95794797e-01
1.60702422e-01 -2.49057621e-01 1.67481285e-02 -4.53234464e-01
7.16994882e-01 -8.94938707e-01 -2.32144296e-02 5.57153404e-01
-4.93318707e-01 3.94106448e-01 5.45059405e-02 2.88906157e-01
-1.79116279e-01 -8.69736731e-01 4.49932188e-01 -5.30299485e-01
-7.82236934e-01 -1.64641172e-01 -8.70601714e-01 -1.33709341e-01
9.50518489e-01 6.18744306e-02 -2.75234103e-01 -6.24807358e-01
-1.05598235e+00 4.37361807e-01 2.14646593e-01 4.08274144e-01
4.00060207e-01 -1.02141404e+00 -4.63227600e-01 8.03121030e-02
-7.92225972e-02 -1.54842690e-01 2.67598685e-02 4.27439332e-01
-4.24212158e-01 6.54301047e-01 -2.13103846e-01 -2.86984980e-01
-1.23013926e+00 5.84206820e-01 -1.56659335e-01 -8.00827742e-01
-3.37319702e-01 5.72179794e-01 -6.53322563e-02 -9.71177638e-01
2.97152489e-01 -5.95690250e-01 -5.59143305e-01 1.30218983e-01
3.36096734e-01 -1.95935965e-04 -1.48537144e-01 -5.27237952e-01
-3.07822287e-01 -1.09423980e-01 -2.90934056e-01 -3.92851472e-01
1.22886956e+00 -2.78857470e-01 8.98691118e-02 7.44817913e-01
1.09745955e+00 -2.50777870e-01 -9.48329985e-01 -2.54436821e-01
4.14800227e-01 -1.11824699e-01 -6.06231391e-01 -9.36390579e-01
-3.99320304e-01 6.05013013e-01 6.37207806e-01 6.72376394e-01
7.47034490e-01 9.63195786e-02 9.59930360e-01 6.02299690e-01
5.96515059e-01 -1.05267239e+00 8.70926306e-02 8.87258112e-01
6.07636154e-01 -1.46625423e+00 -3.54074568e-01 -3.85383040e-01
-1.06831014e+00 9.61105406e-01 8.51967633e-01 -2.22715270e-02
3.70684505e-01 4.01079059e-01 3.99505615e-01 -5.75866736e-02
-9.75930572e-01 -3.49370301e-01 -3.35441977e-01 6.35361016e-01
6.79687321e-01 1.74902782e-01 -4.62971449e-01 8.10272813e-01
-4.43522632e-01 -1.85031474e-01 6.90073609e-01 1.19766033e+00
-7.38718867e-01 -1.66854453e+00 -2.44243860e-01 4.96882409e-01
-4.93140876e-01 -1.19170755e-01 -7.60151818e-02 9.72184002e-01
-5.02787754e-02 1.47116721e+00 1.75649580e-02 -3.89492005e-01
4.82677728e-01 5.90926349e-01 1.76268160e-01 -9.90978181e-01
-8.20242763e-01 1.71869457e-01 7.32146740e-01 4.60399017e-02
-6.34893596e-01 -5.05982101e-01 -1.39938653e+00 -3.22594613e-01
-4.87323672e-01 4.90836054e-01 5.95749199e-01 1.08326936e+00
-6.83182105e-02 3.60797882e-01 4.48999554e-01 -7.93426812e-01
-8.41136694e-01 -1.69539845e+00 -5.38539350e-01 3.63031924e-01
-1.16731012e-02 -5.35197914e-01 -2.31168225e-01 1.22045524e-01]
|
[12.767327308654785, 7.813595294952393]
|
38292d46-3b2b-4a8e-a3b0-277e9d74ebb9
|
beyond-controlled-environments-3d-camera-re
|
2008.02004
| null |
https://arxiv.org/abs/2008.02004v1
|
https://arxiv.org/pdf/2008.02004v1.pdf
|
Beyond Controlled Environments: 3D Camera Re-Localization in Changing Indoor Scenes
|
Long-term camera re-localization is an important task with numerous computer vision and robotics applications. Whilst various outdoor benchmarks exist that target lighting, weather and seasonal changes, far less attention has been paid to appearance changes that occur indoors. This has led to a mismatch between popular indoor benchmarks, which focus on static scenes, and indoor environments that are of interest for many real-world applications. In this paper, we adapt 3RScan - a recently introduced indoor RGB-D dataset designed for object instance re-localization - to create RIO10, a new long-term camera re-localization benchmark focused on indoor scenes. We propose new metrics for evaluating camera re-localization and explore how state-of-the-art camera re-localizers perform according to these metrics. We also examine in detail how different types of scene change affect the performance of different methods, based on novel ways of detecting such changes in a given RGB-D frame. Our results clearly show that long-term indoor re-localization is an unsolved problem. Our benchmark and tools are publicly available at waldjohannau.github.io/RIO10
|
['Federico Tombari', 'Tommaso Cavallari', 'Torsten Sattler', 'Johanna Wald', 'Stuart Golodetz']
|
2020-08-05
| null |
https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/287_ECCV_2020_paper.php
|
https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123520460.pdf
|
eccv-2020-8
|
['camera-relocalization']
|
['computer-vision']
|
[-1.39023559e-02 -8.51831794e-01 -1.70944668e-02 -4.43007082e-01
-5.91214180e-01 -9.25628126e-01 6.55533910e-01 1.54334486e-01
-6.79009199e-01 5.34111619e-01 2.00227946e-01 3.42683047e-02
2.76270453e-02 -4.74742860e-01 -9.23054218e-01 -6.01005197e-01
4.26256768e-02 -9.51773077e-02 5.27826130e-01 -2.56133080e-01
2.09783837e-01 7.67409861e-01 -1.66124749e+00 -8.07128325e-02
3.03724527e-01 6.95549667e-01 4.19882268e-01 9.51114178e-01
3.34131092e-01 4.46743369e-01 -5.60671031e-01 -1.40380859e-01
4.43077981e-01 -2.94060826e-01 -6.02282166e-01 7.18451291e-02
9.22264576e-01 6.59528449e-02 -2.92052120e-01 1.15637171e+00
6.05742216e-01 3.50738049e-01 1.38091400e-01 -1.11458075e+00
-6.36966467e-01 2.69981474e-02 -3.14646184e-01 5.07420242e-01
1.08762956e+00 8.41699541e-02 5.52103400e-01 -7.87808955e-01
6.54756606e-01 8.99126709e-01 1.11884594e+00 2.55304456e-01
-1.11108148e+00 -4.77813840e-01 3.72088045e-01 3.92884195e-01
-1.54855072e+00 -6.39067471e-01 7.95976460e-01 -2.52059907e-01
8.80379319e-01 4.58470881e-01 5.35977423e-01 1.31698525e+00
3.62801194e-01 3.02697122e-01 1.29341483e+00 -4.81529593e-01
2.77925014e-01 5.24115115e-02 -1.06887832e-01 4.10064638e-01
2.71735191e-01 9.72269773e-02 -3.56756091e-01 2.91642338e-01
5.90505302e-01 4.67609435e-01 -5.01112819e-01 -8.53875995e-01
-1.69753706e+00 2.34690160e-01 8.46873403e-01 5.57975590e-01
-4.41744411e-03 4.91137177e-01 1.46398500e-01 2.58589923e-01
2.96181530e-01 5.40896237e-01 -4.41917747e-01 -3.66547525e-01
-6.99601948e-01 -3.24726515e-02 5.58528960e-01 1.15238988e+00
9.92325246e-01 -5.25639832e-01 2.86690056e-01 7.56272674e-01
2.27397069e-01 7.17125773e-01 6.75847471e-01 -7.12335229e-01
2.66693801e-01 2.96192944e-01 2.79665858e-01 -1.27394688e+00
-5.01082242e-01 -4.22969520e-01 -7.46531069e-01 5.06801642e-02
4.34138119e-01 2.90030748e-01 -6.64474070e-01 1.35030544e+00
2.87664324e-01 4.40093666e-01 1.03835411e-01 9.73857701e-01
7.34981358e-01 3.15064818e-01 -2.35804141e-01 1.41508654e-01
1.06905103e+00 -1.30447364e+00 -3.39169651e-01 -5.75638592e-01
4.19584125e-01 -1.02953231e+00 1.00334537e+00 1.92505673e-01
-4.08678859e-01 -7.95788586e-01 -1.03811479e+00 -1.83721166e-02
-8.57233882e-01 7.60456026e-02 3.89130443e-01 6.73960149e-01
-1.24342942e+00 3.92455071e-01 -7.61368155e-01 -1.24525964e+00
-1.83829859e-01 1.52264267e-01 -8.31790090e-01 -4.94179606e-01
-6.99353933e-01 1.00974512e+00 8.45378414e-02 1.49188206e-01
-7.83192694e-01 -4.12130624e-01 -8.26730609e-01 -5.06239891e-01
3.02445889e-01 -5.73380291e-01 1.14153314e+00 -7.28737116e-01
-1.40669048e+00 1.04717612e+00 -3.25747758e-01 -2.64171660e-01
5.61204851e-01 -7.73885176e-02 -6.18888855e-01 -3.71439397e-01
3.29568982e-01 4.20530885e-01 6.23602986e-01 -1.51932263e+00
-8.67809355e-01 -3.15484673e-01 3.10707331e-01 2.07949266e-01
5.30878082e-02 -1.28114283e-01 -8.70968521e-01 -4.45839107e-01
3.12510550e-01 -1.25099301e+00 -3.47666055e-01 -8.31723139e-02
-2.32539788e-01 1.81337938e-01 7.79388905e-01 -1.29669398e-01
8.55854213e-01 -2.22643518e+00 5.42799905e-02 -8.31534863e-02
-9.33429077e-02 -1.39367655e-01 -1.05258957e-01 3.39237750e-01
-2.18906149e-01 -1.92225855e-02 -1.54057257e-02 -7.80888379e-01
-3.39486003e-02 1.77646920e-01 -1.75144494e-01 1.00851369e+00
-5.66973686e-01 5.88250041e-01 -1.31960225e+00 -1.80602502e-02
9.71563697e-01 5.19427776e-01 -1.94417775e-01 9.57936719e-02
1.99719399e-01 7.07598507e-01 6.03276677e-02 8.52144182e-01
7.31964648e-01 -2.85490509e-02 -3.00160348e-01 -2.63564855e-01
-4.75795746e-01 -1.93046540e-01 -1.16074860e+00 2.23779202e+00
-7.76975632e-01 9.77501333e-01 -2.96373308e-01 -4.25408661e-01
7.16015100e-01 -3.14178467e-01 3.42734814e-01 -8.79429400e-01
1.12456838e-02 2.91017681e-01 -5.15822589e-01 -2.32682183e-01
8.61722350e-01 3.18613410e-01 -3.47945303e-01 -1.48886904e-01
-1.51436195e-01 -4.66177404e-01 2.75474727e-01 -8.01219493e-02
1.33469605e+00 2.09303886e-01 5.64437687e-01 -1.82712257e-01
5.88407099e-01 1.15821630e-01 3.44241589e-01 8.47591877e-01
-4.78915691e-01 1.00516212e+00 -3.28230947e-01 -5.32739997e-01
-9.02988851e-01 -9.74667609e-01 -9.68635082e-02 9.12458897e-01
8.29540372e-01 -4.68081027e-01 -3.61046255e-01 -4.53239441e-01
-4.93683368e-02 3.85275602e-01 -7.14215219e-01 4.33129855e-02
-2.89802223e-01 -5.83371043e-01 3.67956370e-01 2.73961604e-01
6.90178514e-01 -6.20978832e-01 -9.86216784e-01 6.57644421e-02
-6.61320537e-02 -1.53972960e+00 -3.59480679e-01 3.96131158e-01
-5.56516469e-01 -1.28093982e+00 -8.92003834e-01 -7.56544709e-01
6.74035728e-01 9.63863671e-01 1.20479405e+00 -2.85990387e-01
-3.95429343e-01 1.21710587e+00 -6.68189824e-01 -1.09435417e-01
-6.61189109e-02 4.16554101e-02 2.97385097e-01 -7.86369592e-02
2.96532720e-01 -4.40355211e-01 -7.64596939e-01 8.07025969e-01
-7.36226082e-01 -2.66600758e-01 3.48794281e-01 2.56851077e-01
8.18219244e-01 9.83820558e-02 -3.54224950e-01 -4.19852436e-01
3.12296450e-01 -4.67375368e-01 -9.28342879e-01 4.05695975e-01
-3.76362801e-01 -1.95470124e-01 5.90450108e-01 -2.98614293e-01
-6.53922558e-01 4.74109828e-01 1.69308372e-02 -4.54524577e-01
-5.81948459e-01 5.39564565e-02 -1.01522252e-01 -6.16306722e-01
7.34335601e-01 2.38533944e-01 -6.96331561e-01 -4.94330823e-01
4.18314517e-01 4.38120395e-01 8.42557847e-01 -3.18757236e-01
9.79516208e-01 7.38792896e-01 -5.91371097e-02 -9.38337266e-01
-7.97918439e-01 -1.13848603e+00 -8.26804638e-01 -2.97160178e-01
8.52010667e-01 -1.11143231e+00 -3.76418531e-01 6.41052008e-01
-1.15397632e+00 -4.58845466e-01 -3.21888179e-01 4.68072206e-01
-3.92752141e-01 3.62409174e-01 1.24498442e-01 -3.34289193e-01
2.76914328e-01 -1.41555583e+00 1.31284845e+00 4.15030837e-01
1.24086022e-01 -1.07515812e+00 7.40911424e-01 6.42737327e-03
5.22987485e-01 4.42323953e-01 -6.15846217e-02 -2.08535530e-02
-5.64217091e-01 -3.16405892e-01 -1.83169797e-01 2.20458597e-01
4.46826577e-01 -2.02663779e-01 -1.08286655e+00 -4.38972443e-01
-3.12470526e-01 1.75576016e-01 7.11028516e-01 1.20241702e-01
9.73542035e-01 2.13482141e-01 -5.26944816e-01 1.08381104e+00
1.79692042e+00 1.02436274e-01 6.00430071e-01 9.68877494e-01
8.36824894e-01 1.48529351e-01 6.71447992e-01 2.48168468e-01
6.01678848e-01 9.91571903e-01 6.56277359e-01 -1.81339160e-01
-3.11327606e-01 -1.78396016e-01 4.67049628e-01 6.09812617e-01
-1.37948900e-01 -3.27163815e-01 -1.02837765e+00 6.30499125e-01
-1.89739680e+00 -6.85647190e-01 -1.87806800e-01 2.36438942e+00
3.62376481e-01 -2.10559607e-01 -2.66500145e-01 -9.41305086e-02
6.53446555e-01 3.81647319e-01 -3.93841475e-01 1.23039410e-01
-4.76013452e-01 2.71652522e-03 1.20044231e+00 4.60010618e-01
-1.35135245e+00 9.75226879e-01 5.75845480e+00 4.72957104e-01
-1.25078106e+00 3.15324128e-01 2.39097789e-01 1.05066642e-01
1.90026090e-01 1.53880835e-01 -8.32268775e-01 4.72386241e-01
7.31771231e-01 9.56561714e-02 5.00321448e-01 1.17998922e+00
4.73149270e-02 -5.99967778e-01 -1.34982109e+00 1.69055176e+00
5.22673905e-01 -1.03939486e+00 -6.58780634e-01 -4.47940044e-02
1.15385449e+00 5.13498127e-01 2.84478515e-02 7.37614185e-02
2.57532299e-01 -6.89932108e-01 8.63684356e-01 6.49418414e-01
7.48411477e-01 -4.01188701e-01 8.00007880e-01 -1.22429192e-01
-1.54798448e+00 1.64498333e-02 -5.49585164e-01 1.84993297e-02
-9.55033116e-03 7.08920956e-01 -7.31738150e-01 4.99519944e-01
1.28214073e+00 1.10987067e+00 -1.38030851e+00 1.49008811e+00
-4.20319885e-01 -1.02901012e-01 -4.64945704e-01 5.31989448e-02
1.30050331e-01 1.47394929e-02 4.62404728e-01 1.43763351e+00
4.71328646e-01 -4.11406189e-01 1.69520244e-01 3.68884742e-01
-1.64598376e-02 -1.59303352e-01 -9.46488619e-01 5.26779711e-01
3.28424811e-01 1.12252641e+00 -8.93823624e-01 -3.53962323e-03
-3.63960624e-01 1.64831471e+00 2.94494573e-02 4.05452371e-01
-9.89445090e-01 -3.86330277e-01 8.14009845e-01 2.56255716e-02
1.84601426e-01 -7.40213037e-01 3.97052884e-01 -1.37748969e+00
5.25986366e-02 -5.50357580e-01 8.63923952e-02 -1.10743725e+00
-1.13380182e+00 6.16266787e-01 1.71653498e-02 -1.36614406e+00
2.51329020e-02 -7.11295128e-01 -2.20277771e-01 2.86030114e-01
-1.77295828e+00 -1.22622335e+00 -1.17684698e+00 9.41531897e-01
7.19658017e-01 2.70614415e-01 7.55182326e-01 4.62691545e-01
-5.66211164e-01 3.75661761e-01 5.91389954e-01 5.89738134e-03
1.09329498e+00 -1.31999958e+00 4.55747783e-01 1.33262539e+00
4.03408676e-01 5.20219386e-01 8.36643636e-01 -2.66380936e-01
-1.64697397e+00 -1.27735734e+00 5.44330120e-01 -9.52124715e-01
5.70203304e-01 -5.87531567e-01 -2.42795795e-01 8.54548693e-01
7.71268904e-02 5.84486842e-01 3.33276808e-01 -9.90194827e-02
-4.20515686e-01 -4.65998441e-01 -9.44394469e-01 4.45755213e-01
1.37028265e+00 -6.38978004e-01 -2.25385830e-01 5.18253863e-01
6.17887437e-01 -7.18999684e-01 -6.77651405e-01 2.69245982e-01
4.64337468e-01 -1.49938846e+00 1.27933538e+00 3.07175219e-01
-8.79192501e-02 -8.04140985e-01 -6.19597077e-01 -1.43763280e+00
-2.80625522e-01 -3.35710764e-01 3.13241929e-01 1.11724222e+00
-2.65462380e-02 -7.65556157e-01 4.12935138e-01 2.56017983e-01
-1.32370755e-01 -6.31260201e-02 -1.03877950e+00 -1.01524901e+00
-5.42254150e-01 -5.90972781e-01 6.52777195e-01 8.73549640e-01
-4.83385623e-01 -5.17714918e-02 -3.21753919e-01 4.18093234e-01
5.11742830e-01 2.76746079e-02 1.35616946e+00 -8.48060012e-01
-3.28780971e-02 -3.19510311e-01 -1.03892100e+00 -1.14689529e+00
-1.38160791e-02 -5.40062964e-01 3.40111703e-01 -1.59112477e+00
-6.77153235e-03 -5.60968041e-01 -3.01696509e-01 2.81607389e-01
2.24709347e-01 6.58922791e-01 1.68595836e-01 3.05983216e-01
-1.34445179e+00 4.37612504e-01 6.18862271e-01 -1.88467473e-01
-3.28776352e-02 -1.02002949e-01 -2.49226600e-01 7.01745689e-01
6.19391441e-01 -3.83553714e-01 5.97646984e-04 -7.64334679e-01
3.73134673e-01 -6.09383762e-01 8.09619009e-01 -1.88900554e+00
6.50370836e-01 -2.49412768e-02 4.24995959e-01 -5.23891568e-01
3.80403340e-01 -1.38474786e+00 4.63778496e-01 3.31416219e-01
2.39331454e-01 5.16727626e-01 2.61310250e-01 7.40440845e-01
-3.17493618e-01 4.81311139e-03 7.55733252e-01 -3.55592996e-01
-1.56475854e+00 9.12271515e-02 -8.53471458e-02 -2.75081873e-01
1.30505562e+00 -5.85597336e-01 -3.54977429e-01 -2.26020604e-01
-5.12091219e-01 -1.69757292e-01 1.16900349e+00 7.68592715e-01
5.28228641e-01 -1.26303196e+00 -2.22220227e-01 3.03991828e-02
7.19269216e-01 1.19908646e-01 9.57419574e-02 8.38052750e-01
-9.56232250e-01 5.34404516e-01 -1.11906774e-01 -9.42236543e-01
-1.37482190e+00 5.00987291e-01 5.04468441e-01 6.39593825e-02
-4.23691660e-01 9.43247795e-01 -7.65982941e-02 -6.08931720e-01
2.95778126e-01 -7.81709731e-01 2.76238650e-01 -1.20454274e-01
1.95297033e-01 4.38295901e-01 3.06695610e-01 -9.51819956e-01
-8.54480922e-01 1.08167040e+00 4.94752109e-01 2.33358696e-01
1.21146214e+00 -6.87437236e-01 -1.68137863e-01 6.67872608e-01
1.33603370e+00 3.25368553e-01 -1.11203790e+00 -1.78765193e-01
-1.92428159e-03 -9.22018290e-01 7.63536170e-02 -6.44935727e-01
-8.19384038e-01 4.67266589e-01 1.28750968e+00 -4.95948866e-02
1.10598636e+00 5.09167165e-02 4.16934073e-01 6.06556237e-01
1.07805681e+00 -8.18035305e-01 1.61744073e-01 6.99605286e-01
7.93889165e-01 -1.57243419e+00 2.69784540e-01 -4.43738997e-02
-1.91631183e-01 1.07170475e+00 5.09730935e-01 -1.63876310e-01
4.34049368e-01 5.50430901e-02 7.11687207e-02 1.29377782e-01
1.06001236e-01 -4.41185504e-01 5.84957041e-02 7.95827031e-01
4.97929044e-02 3.04009803e-02 4.52420056e-01 -1.51865594e-02
-2.42896497e-01 -2.35719383e-01 4.89159614e-01 1.05769241e+00
-1.36090949e-01 -1.05803311e+00 -8.38773429e-01 -2.59968966e-01
5.52400537e-02 1.03544220e-01 -5.28947175e-01 9.29274440e-01
5.42249501e-01 1.04722846e+00 -1.50137150e-03 -5.44308186e-01
5.30227065e-01 -5.44343770e-01 5.96338809e-01 -6.55983984e-01
-3.52364838e-01 -2.94773698e-01 -3.43443125e-01 -1.02824450e+00
-8.91422510e-01 -7.25687683e-01 -6.29578710e-01 -3.01968783e-01
-3.93393993e-01 -1.10473717e-02 1.26220703e+00 5.21779835e-01
4.31810707e-01 4.99509275e-01 5.78770041e-01 -1.38948655e+00
2.55778879e-01 -6.51798546e-01 -3.13276619e-01 3.80516917e-01
7.54730821e-01 -6.58657968e-01 -6.45841181e-01 1.00768663e-01]
|
[7.623378276824951, -2.034660816192627]
|
0e56b7e1-07f2-43d3-b294-7e4d386ce6e6
|
beyond-normal-on-the-evaluation-of-mutual
|
2306.11078
| null |
https://arxiv.org/abs/2306.11078v1
|
https://arxiv.org/pdf/2306.11078v1.pdf
|
Beyond Normal: On the Evaluation of Mutual Information Estimators
|
Mutual information is a general statistical dependency measure which has found applications in representation learning, causality, domain generalization and computational biology. However, mutual information estimators are typically evaluated on simple families of probability distributions, namely multivariate normal distribution and selected distributions with one-dimensional random variables. In this paper, we show how to construct a diverse family of distributions with known ground-truth mutual information and propose a language-independent benchmarking platform for mutual information estimators. We discuss the general applicability and limitations of classical and neural estimators in settings involving high dimensions, sparse interactions, long-tailed distributions, and high mutual information. Finally, we provide guidelines for practitioners on how to select appropriate estimator adapted to the difficulty of problem considered and issues one needs to consider when applying an estimator to a new data set.
|
['Alexander Marx', 'Niko Beerenwinkel', 'Julia E. Vogt', 'Frederic Grabowski', 'Paweł Czyż']
|
2023-06-19
| null | null | null | null |
['domain-generalization', 'mutual-information-estimation', 'benchmarking', 'benchmarking']
|
['methodology', 'methodology', 'miscellaneous', 'robots']
|
[ 1.37562945e-01 -1.09163336e-01 -3.36829454e-01 -6.58882976e-01
-5.34366369e-01 -3.93407851e-01 5.64311743e-01 2.84556776e-01
-4.04412478e-01 1.36860859e+00 1.88913316e-01 -1.38549671e-01
-8.29493642e-01 -7.75628388e-01 -3.80329281e-01 -7.52218723e-01
-5.01313031e-01 7.04101503e-01 -2.18656838e-01 2.18531668e-01
2.84938484e-01 4.42600697e-01 -1.60925138e+00 -2.94928521e-01
9.24360931e-01 5.90035141e-01 3.54429856e-02 6.89384043e-01
1.58983208e-02 6.21703565e-01 -6.85069799e-01 -3.59775186e-01
-2.44291633e-01 -6.41228437e-01 -8.37624609e-01 -5.00320017e-01
-7.20590446e-03 2.44434401e-01 -1.25292152e-01 1.02281535e+00
5.91873825e-01 1.83518022e-01 1.66110635e+00 -1.38707566e+00
-6.06758118e-01 8.73482466e-01 -4.43592966e-01 3.54209661e-01
4.01708484e-01 -1.52521044e-01 1.16588402e+00 -4.93119299e-01
5.83476961e-01 1.22487104e+00 7.83867717e-01 4.09437627e-01
-1.44706810e+00 -8.25450361e-01 -2.13918120e-01 1.74382508e-01
-1.45394897e+00 -1.12709992e-01 5.27603626e-01 -6.30011380e-01
8.47918928e-01 1.70803741e-01 3.01245034e-01 1.30228925e+00
2.77761936e-01 5.74324608e-01 1.03381073e+00 -2.65820891e-01
2.66034752e-01 6.50016740e-02 3.49162549e-01 2.02325284e-01
6.49479866e-01 1.62270829e-01 -6.20619714e-01 -5.42480409e-01
6.93983734e-01 -3.18000466e-01 -1.33292869e-01 -6.75107956e-01
-1.23949409e+00 9.55439389e-01 1.00062646e-01 6.48908854e-01
-2.45326042e-01 2.09492236e-01 3.76445383e-01 5.19666076e-01
5.63282549e-01 3.80124897e-01 -7.54828513e-01 -2.53889084e-01
-6.16595447e-01 3.71111929e-01 1.35811484e+00 8.90295327e-01
7.92597175e-01 -1.71015635e-01 2.55880747e-02 9.06403184e-01
3.68657589e-01 4.87554550e-01 3.78329247e-01 -6.64687335e-01
1.61175445e-01 1.90444797e-01 1.52723687e-02 -1.10601199e+00
-7.27428198e-01 -5.38297594e-01 -1.18873835e+00 -3.29167753e-01
7.24610865e-01 -3.78387302e-01 -3.43916148e-01 2.05350256e+00
1.83817342e-01 9.89820659e-02 1.84243858e-01 7.22096145e-01
7.52935052e-01 1.05187751e-01 1.75519269e-02 -3.34157407e-01
8.61186147e-01 -8.06267038e-02 -6.16526365e-01 -1.33383349e-01
8.14213872e-01 -5.55196941e-01 5.91152430e-01 1.72696292e-01
-7.25740433e-01 -5.03943302e-02 -7.74080873e-01 2.48616844e-01
-4.48192418e-01 -2.35391214e-01 7.48583436e-01 5.02261937e-01
-7.62086391e-01 8.75264645e-01 -5.53499579e-01 -6.43920600e-01
1.65758967e-01 5.29864550e-01 -5.64980984e-01 -1.43820569e-02
-1.48212552e+00 1.04750192e+00 5.45132041e-01 -3.98029685e-01
-4.95864958e-01 -6.16872191e-01 -6.90029562e-01 -1.01660810e-01
-1.71118736e-01 -8.53688538e-01 7.68007219e-01 -7.78409004e-01
-1.27139497e+00 9.07228529e-01 -6.91280738e-02 -4.73427653e-01
3.66976053e-01 -1.76924214e-01 -3.51506978e-01 -4.43421006e-01
1.20135225e-01 1.56885505e-01 3.25566918e-01 -8.82990181e-01
-1.80059865e-01 -4.39391315e-01 -3.38707507e-01 2.07939386e-01
-1.07911587e-01 -9.15199816e-02 2.33595986e-02 -5.03227651e-01
1.51689187e-01 -6.62765682e-01 -2.17177719e-01 -2.92806357e-01
-5.54247320e-01 -2.46389166e-01 3.05674821e-01 -2.64856577e-01
1.01803887e+00 -2.14108753e+00 3.77189189e-01 4.10107553e-01
1.27190098e-01 -3.93199712e-01 -2.83999592e-01 7.23545909e-01
-2.93866307e-01 2.22390592e-02 -5.90458512e-01 9.21891406e-02
-5.95005527e-02 2.31081188e-01 2.80122161e-01 8.05780649e-01
1.99548498e-01 4.62628901e-01 -1.02758455e+00 -4.29058880e-01
3.69697101e-02 5.96159875e-01 -4.38254684e-01 4.84926760e-01
1.27038464e-01 7.66680837e-01 -4.45736527e-01 2.74137884e-01
6.44089818e-01 -4.77620453e-01 3.98829043e-01 -4.72617149e-03
7.44789988e-02 2.86559850e-01 -1.27866602e+00 1.38690388e+00
-4.87914264e-01 8.47184122e-01 -3.29185039e-01 -1.24935484e+00
1.13885105e+00 -3.72349545e-02 5.08164704e-01 -2.88558245e-01
3.12820405e-01 1.09456800e-01 4.43570286e-01 -2.60292441e-01
-3.92079800e-02 -3.02052051e-01 -2.42040157e-02 6.30152702e-01
4.14380997e-01 -2.41612867e-01 2.39232346e-01 2.47665644e-02
1.25378978e+00 -1.74586654e-01 8.75440836e-01 -6.14397824e-01
3.73336583e-01 -7.02979922e-01 3.52358252e-01 8.98806453e-01
-3.42937298e-02 5.43443441e-01 9.42660391e-01 -7.55586103e-02
-8.33642185e-01 -1.20904326e+00 -6.27011120e-01 1.03002894e+00
-4.48704027e-02 -2.13138610e-01 -3.69971365e-01 -3.86883825e-01
2.81104535e-01 9.27559435e-01 -8.50026131e-01 -2.87988871e-01
1.09972740e-02 -1.09324205e+00 5.93255997e-01 1.70735419e-01
1.15131095e-01 -6.34846151e-01 -1.91943690e-01 1.28880516e-01
-2.24600011e-03 -6.48943424e-01 3.40193361e-02 7.53780365e-01
-8.26615036e-01 -1.37802124e+00 -6.80709243e-01 -4.10013199e-01
3.80325556e-01 -2.15039834e-01 1.33828330e+00 -5.23316681e-01
-2.06976295e-01 4.53118443e-01 -9.05446857e-02 -2.13858739e-01
-4.28698987e-01 2.58967906e-01 2.09523723e-01 -2.73889810e-01
5.42702615e-01 -8.64115119e-01 -3.37228715e-01 5.19552231e-01
-8.27787757e-01 -4.25026268e-01 6.70661986e-01 9.82562423e-01
2.46745840e-01 -1.11457437e-01 7.96134591e-01 -8.84502172e-01
9.53978539e-01 -1.04908895e+00 -2.51755178e-01 9.48381275e-02
-5.16467810e-01 5.46865284e-01 1.29692927e-01 -4.72820997e-01
-9.01915729e-01 -1.68764621e-01 1.72240451e-01 7.09816739e-02
-2.45465323e-01 7.59212732e-01 -2.93112993e-01 1.08759470e-01
8.60157371e-01 5.94060495e-02 2.21961632e-01 -3.40561152e-01
4.35803741e-01 6.12771809e-01 3.28136235e-01 -6.11307204e-01
2.06395388e-01 8.19794387e-02 2.57303864e-01 -7.96373427e-01
-8.22154999e-01 -6.16853416e-01 -9.28983092e-01 5.84333129e-02
4.60648656e-01 -6.87345386e-01 -8.01376760e-01 4.07676488e-01
-1.12744892e+00 -2.72281885e-01 -1.24275938e-01 7.63545454e-01
-7.85967410e-01 2.53957808e-01 -3.24326932e-01 -9.39716518e-01
-5.93137331e-02 -6.67897224e-01 8.08341444e-01 6.60712495e-02
-6.54208064e-01 -1.54558897e+00 6.00840092e-01 -1.84864387e-01
5.00657320e-01 2.02178016e-01 9.54883337e-01 -1.06059444e+00
1.66718155e-01 -4.67290357e-02 -3.36583853e-01 3.08376372e-01
4.48608637e-01 1.77535668e-01 -7.39066541e-01 -3.63930054e-02
-3.64214182e-01 -3.14888328e-01 9.09227133e-01 6.82838500e-01
9.91415918e-01 -2.86216944e-01 -5.93638122e-01 3.75880778e-01
1.06649280e+00 -1.99214637e-01 3.24178517e-01 -1.26544666e-02
3.91684592e-01 8.91832590e-01 3.53565723e-01 8.03373098e-01
3.98587078e-01 1.88282043e-01 2.21879363e-01 3.37238878e-01
3.96902859e-01 2.64401007e-02 -3.80125940e-02 9.82890904e-01
5.23255952e-02 -4.45740908e-01 -9.52500641e-01 5.54219544e-01
-1.93093741e+00 -9.81706560e-01 -2.78668880e-01 2.40322113e+00
1.21395612e+00 -4.42887433e-02 1.37998760e-01 3.03265005e-02
8.28306377e-01 -8.36517811e-02 -7.54052997e-01 -1.59821093e-01
-5.05924404e-01 -3.30155455e-02 4.79066581e-01 3.56645703e-01
-1.11579633e+00 6.26861095e-01 7.95047045e+00 7.03828514e-01
-8.28975797e-01 -6.79826811e-02 5.90393364e-01 1.04883932e-01
-2.72637039e-01 2.06030179e-02 -5.61635852e-01 4.47427213e-01
1.10738266e+00 -4.81784880e-01 8.01980793e-02 5.57008266e-01
-1.18941799e-01 -3.10785443e-01 -1.36840236e+00 1.03659821e+00
-1.38229057e-01 -7.52732396e-01 -4.10069615e-01 2.85077766e-02
8.89401078e-01 3.55835408e-01 -7.40950853e-02 3.00988276e-02
8.66125405e-01 -1.27443802e+00 -7.51330778e-02 1.01314127e+00
4.98773694e-01 -9.00628269e-01 8.67333472e-01 3.76922309e-01
-6.86696172e-01 2.52296954e-01 -4.50377345e-01 -4.06363934e-01
-1.82011947e-01 1.35078406e+00 -7.05421448e-01 3.46980989e-01
5.28872073e-01 1.00079083e+00 -2.03028947e-01 1.18183446e+00
-1.25038415e-01 5.84924221e-01 -5.52120507e-01 -4.07534629e-01
-2.44353160e-01 -2.41933599e-01 4.57100540e-01 1.42989159e+00
3.74978453e-01 -2.27268636e-01 -3.39075297e-01 8.12440336e-01
1.64855406e-01 2.14174986e-01 -9.57102656e-01 -1.81815028e-01
6.67926371e-01 1.03799617e+00 -3.68200362e-01 -1.25051200e-01
-3.81638259e-01 8.28626573e-01 5.90010941e-01 4.82946783e-01
-6.12680376e-01 -4.30101812e-01 1.06001377e+00 -1.60717964e-01
9.43076909e-02 -3.65045965e-01 -2.70885050e-01 -1.15164435e+00
-4.71759528e-01 -6.21570945e-01 5.80330312e-01 -3.49135756e-01
-2.02741265e+00 2.04257116e-01 3.39464575e-01 -1.05477405e+00
-6.99854255e-01 -6.05166197e-01 -4.48488951e-01 7.46506929e-01
-1.00833678e+00 -5.06154954e-01 -5.55229280e-03 3.56684148e-01
-3.12914073e-01 -2.55675286e-01 9.09711719e-01 1.38121143e-01
-5.03676414e-01 4.11626041e-01 7.55316198e-01 1.77256092e-02
9.12458062e-01 -1.34157670e+00 2.41349787e-02 1.32417813e-01
-2.77992897e-02 7.17433751e-01 1.18534374e+00 -5.21329284e-01
-1.19520223e+00 -7.85980880e-01 8.07236493e-01 -3.48913699e-01
1.07856059e+00 -2.22910419e-01 -8.81316066e-01 5.58231592e-01
9.52166691e-02 -6.34791106e-02 1.16844773e+00 6.09007061e-01
-4.49384749e-01 1.92599550e-01 -1.22362530e+00 3.13197494e-01
1.13646531e+00 -5.09796202e-01 -3.28717023e-01 3.68023276e-01
1.58349007e-01 -2.67923288e-02 -1.14810634e+00 4.43705171e-01
7.72285819e-01 -1.05097520e+00 7.00595140e-01 -8.06214571e-01
1.96605101e-01 4.57614176e-02 -3.39061499e-01 -1.58152390e+00
-3.66207242e-01 -4.08700168e-01 -9.20672622e-03 1.15874600e+00
3.77361715e-01 -8.01980555e-01 5.08567572e-01 5.42984962e-01
7.21196353e-01 -5.70349872e-01 -1.09525716e+00 -7.95745015e-01
4.79797125e-01 -7.09466755e-01 4.35932755e-01 1.20269585e+00
2.99327850e-01 5.06407738e-01 -4.32202518e-01 -1.05280273e-01
8.41306865e-01 -1.38145862e-02 9.09455419e-01 -1.70569313e+00
-3.29676002e-01 -5.46605706e-01 -8.15858662e-01 -9.38649774e-01
5.69896221e-01 -7.44282067e-01 1.23518370e-02 -1.14004910e+00
5.95420420e-01 -2.76294738e-01 -4.66152310e-01 4.90981117e-02
-1.06705517e-01 -9.26198214e-02 -4.96495545e-01 9.70175341e-02
-4.80339259e-01 6.64069176e-01 9.30307567e-01 -9.85148475e-02
-8.59819502e-02 1.90548569e-01 -6.83942854e-01 6.48597896e-01
8.67162108e-01 -5.90849042e-01 -3.39035749e-01 3.01735569e-02
4.19014513e-01 7.94771500e-03 2.58696049e-01 -7.48145759e-01
2.21847683e-01 -3.86991352e-01 3.38487148e-01 -4.83752161e-01
5.11240065e-02 -5.22836506e-01 3.34471792e-01 3.86926234e-01
-7.07809746e-01 -1.41498998e-01 -6.85847923e-02 6.33394420e-01
-2.58922100e-01 -3.34542304e-01 8.46816063e-01 7.40203857e-02
-2.52518952e-01 9.33788046e-02 -5.85699618e-01 3.04236978e-01
9.11337078e-01 9.25312862e-02 6.52304366e-02 -6.69636130e-01
-8.59859884e-01 7.46089369e-02 1.27691805e-01 3.24987531e-01
4.16648269e-01 -1.21467853e+00 -9.91457522e-01 -4.54953089e-02
2.25293398e-01 -3.84935498e-01 8.75723138e-02 9.14885998e-01
-9.51377302e-03 4.19760734e-01 -1.56728730e-01 -5.43434739e-01
-9.89893436e-01 2.08511591e-01 4.24920529e-01 -2.98736721e-01
1.06779806e-01 8.11727107e-01 4.36909586e-01 -8.55673909e-01
3.90511975e-02 -1.80153981e-01 -2.58099705e-01 3.32913458e-01
2.98960775e-01 4.78989959e-01 -2.50528783e-01 -6.43785596e-01
-6.22049332e-01 3.75622302e-01 9.88662168e-02 -4.65849452e-02
1.19987512e+00 -1.01579122e-01 -3.15141350e-01 1.00050032e+00
1.44093668e+00 -3.17427903e-01 -9.03319240e-01 -3.40185046e-01
3.58329862e-01 -1.81553185e-01 -1.19096160e-01 -6.22206748e-01
-8.84981990e-01 6.64474607e-01 5.07983446e-01 2.76478142e-01
6.52987123e-01 4.04742956e-01 -4.78846580e-02 4.83759850e-01
2.51490682e-01 -8.53215814e-01 -8.85569677e-02 7.85271287e-01
9.31524694e-01 -1.18689358e+00 1.68579265e-01 -2.53031433e-01
-5.02453923e-01 1.09522867e+00 4.00388509e-01 -1.45926923e-01
1.22893262e+00 4.38309103e-01 -2.74446368e-01 -8.57870430e-02
-8.91211808e-01 -4.56445158e-01 5.23122013e-01 1.06815088e+00
1.11058879e+00 2.45025247e-01 -6.15604818e-01 4.34097290e-01
-2.93671817e-01 -2.25981116e-01 2.13046178e-01 5.20724177e-01
-2.70468116e-01 -1.03474605e+00 1.00880442e-03 9.79038715e-01
-3.67056280e-01 -1.57603875e-01 -6.96439385e-01 8.18400502e-01
-1.67143703e-01 7.71142244e-01 1.96457982e-01 -1.71868220e-01
-1.17813788e-01 -7.27400184e-02 5.53253353e-01 -3.71895701e-01
-8.82059932e-02 -3.84358376e-01 2.53775418e-01 -3.04541945e-01
-6.16400301e-01 -1.08319247e+00 -1.08076572e+00 -2.94455290e-01
-5.89440286e-01 1.48907349e-01 5.96689224e-01 1.19820356e+00
4.18366343e-01 1.92684442e-01 3.61500293e-01 -5.29185355e-01
-4.09268707e-01 -1.33978534e+00 -9.30091619e-01 2.17122599e-01
2.04337299e-01 -9.92690921e-01 -6.37823462e-01 -3.27524185e-01]
|
[7.467804431915283, 4.202160358428955]
|
4b315624-98e1-4705-947e-6f94e46d5ac8
|
on-the-robustness-of-counterfactual
|
2201.09051
| null |
https://arxiv.org/abs/2201.09051v3
|
https://arxiv.org/pdf/2201.09051v3.pdf
|
On the Robustness of Sparse Counterfactual Explanations to Adverse Perturbations
|
Counterfactual explanations (CEs) are a powerful means for understanding how decisions made by algorithms can be changed. Researchers have proposed a number of desiderata that CEs should meet to be practically useful, such as requiring minimal effort to enact, or complying with causal models. We consider a further aspect to improve the usability of CEs: robustness to adverse perturbations, which may naturally happen due to unfortunate circumstances. Since CEs typically prescribe a sparse form of intervention (i.e., only a subset of the features should be changed), we study the effect of addressing robustness separately for the features that are recommended to be changed and those that are not. Our definitions are workable in that they can be incorporated as penalty terms in the loss functions that are used for discovering CEs. To experiment with robustness, we create and release code where five data sets (commonly used in the field of fair and explainable machine learning) have been enriched with feature-specific annotations that can be used to sample meaningful perturbations. Our experiments show that CEs are often not robust and, if adverse perturbations take place (even if not worst-case), the intervention they prescribe may require a much larger cost than anticipated, or even become impossible. However, accounting for robustness in the search process, which can be done rather easily, allows discovering robust CEs systematically. Robust CEs make additional intervention to contrast perturbations much less costly than non-robust CEs. We also find that robustness is easier to achieve for the features to change, posing an important point of consideration for the choice of what counterfactual explanation is best for the user. Our code is available at: https://github.com/marcovirgolin/robust-counterfactuals.
|
['Saverio Fracaros', 'Marco Virgolin']
|
2022-01-22
| null | null | null | null |
['counterfactual-explanation']
|
['miscellaneous']
|
[ 4.19272959e-01 3.79581869e-01 -3.75883013e-01 -3.09386849e-01
-6.04488969e-01 -7.27263689e-01 8.42938244e-01 1.34604067e-01
-2.98217237e-01 1.04043317e+00 5.45849323e-01 -6.92193747e-01
-4.20053095e-01 -6.36458695e-01 -9.91206288e-01 -5.55387497e-01
-8.91006738e-02 2.54783213e-01 -1.11182801e-01 -1.69722866e-02
4.83154297e-01 4.87702787e-01 -1.76063049e+00 2.91559100e-02
7.49440849e-01 2.59673953e-01 -1.00829363e-01 3.54677767e-01
3.62850726e-01 6.15496218e-01 -5.56585371e-01 -5.35764337e-01
4.44465131e-01 -5.33277631e-01 -9.01750684e-01 2.73902081e-02
1.85981899e-01 -2.79955387e-01 1.96150187e-02 8.49761367e-01
4.20566708e-01 3.24344397e-01 7.91593969e-01 -1.46516573e+00
-4.36934531e-01 8.38771939e-01 -4.04014856e-01 2.90295601e-01
4.73652512e-01 5.63680410e-01 1.18953347e+00 -2.72869498e-01
4.87515122e-01 1.31719196e+00 4.76110786e-01 5.21979213e-01
-1.50913382e+00 -7.59249866e-01 3.27252120e-01 -8.67030844e-02
-9.76841390e-01 -8.46306324e-01 5.47209561e-01 -4.05864984e-01
7.99392164e-01 8.52021158e-01 3.21865678e-01 1.19721234e+00
1.47270486e-01 4.67062920e-01 1.03406644e+00 -5.00905991e-01
2.87017316e-01 1.86519012e-01 -1.09811388e-01 3.74676585e-01
8.38617861e-01 5.94803095e-01 -2.21289530e-01 -6.73300207e-01
4.58848655e-01 -1.22547634e-01 -6.05789304e-01 -4.38464642e-01
-1.23179531e+00 8.79504681e-01 3.32565993e-01 3.80981229e-02
-4.01946455e-01 1.86059833e-01 4.10200804e-01 4.21173453e-01
2.45765463e-01 1.20564425e+00 -7.37030983e-01 -2.88922757e-01
-5.50909877e-01 6.48414433e-01 6.91757441e-01 5.80997705e-01
4.54997480e-01 -2.37763494e-01 -1.30085409e-01 5.07084668e-01
-4.54851314e-02 -8.83770362e-03 4.54434395e-01 -1.04037082e+00
2.67605156e-01 3.78454357e-01 6.00407422e-01 -7.90534019e-01
-3.72951001e-01 -2.05619559e-01 -4.46156174e-01 2.29890212e-01
7.32607067e-01 -3.29603463e-01 -6.57248378e-01 2.04088807e+00
2.84657389e-01 1.82690889e-01 9.98937711e-02 8.55833650e-01
1.71181455e-01 2.66482562e-01 1.96564734e-01 -5.39362490e-01
1.18941069e+00 -2.75859356e-01 -5.08038759e-01 -4.36619341e-01
9.27820563e-01 -7.07465172e-01 1.44670069e+00 2.25489110e-01
-9.43604529e-01 5.93138412e-02 -8.83319080e-01 4.14352149e-01
-1.36632875e-01 -4.84210491e-01 9.62696373e-01 6.24186337e-01
-4.65952277e-01 8.18452060e-01 -6.72263920e-01 -3.08605105e-01
2.54295319e-01 3.13191146e-01 -3.50095838e-01 8.80111307e-02
-1.29855621e+00 1.11489236e+00 2.49878064e-01 -7.06143826e-02
-6.61145806e-01 -8.96439254e-01 -7.54821539e-01 3.32098037e-01
8.61142099e-01 -6.93228424e-01 1.30822742e+00 -1.21621811e+00
-8.80973995e-01 4.30853814e-01 -1.98234335e-01 -4.55824465e-01
7.65404224e-01 -6.70519844e-02 -1.86030403e-01 -3.96050304e-01
2.88489074e-01 3.84561986e-01 5.55980921e-01 -1.13247693e+00
-4.57765847e-01 -1.76619217e-01 4.59129274e-01 2.64128178e-01
-2.33247671e-02 2.05765739e-01 -5.89370951e-02 -5.74841201e-01
-2.66006976e-01 -1.18866205e+00 -3.97908062e-01 -2.92119890e-01
-6.70729458e-01 -1.24590300e-01 4.34011757e-01 -2.72723585e-01
1.21833301e+00 -2.02128696e+00 -1.58427894e-01 2.12462083e-01
-1.12717129e-01 -7.00291842e-02 1.98740251e-02 4.39138144e-01
-5.66753745e-01 8.45442712e-01 -3.06214243e-01 1.52460739e-01
7.13473186e-02 -2.87656765e-02 -3.96769315e-01 5.48106790e-01
1.86539784e-01 5.53693831e-01 -7.33065069e-01 3.68819349e-02
1.21869810e-01 -8.87137055e-02 -7.43403018e-01 -7.54529312e-02
-2.35392585e-01 3.08664322e-01 -3.89107823e-01 2.04759419e-01
2.85903960e-01 -1.33048475e-01 2.26765230e-01 7.52697363e-02
-1.06831558e-01 7.87302852e-01 -1.51292896e+00 7.39493847e-01
-4.36599165e-01 4.19012100e-01 -3.06419134e-01 -9.59455729e-01
4.12038237e-01 2.85520524e-01 2.84587443e-01 -1.56139269e-01
-6.47454932e-02 6.12157434e-02 4.58871931e-01 -4.25616354e-01
1.87191427e-01 -3.35408032e-01 -7.41507411e-02 6.95833564e-01
-6.26349330e-01 -2.40147874e-01 1.95698291e-01 -8.70756134e-02
1.25812483e+00 -1.16726756e-01 7.79998541e-01 -4.06562954e-01
1.36775538e-01 1.27637416e-01 1.03817475e+00 9.03514385e-01
-2.51592640e-02 4.55695987e-01 8.20961416e-01 -3.06194454e-01
-1.05693161e+00 -5.96065640e-01 -2.05319062e-01 7.06584275e-01
-1.83649167e-01 -1.15345933e-01 -4.42710817e-01 -9.40128863e-01
2.83716232e-01 1.33310580e+00 -7.47932255e-01 -4.39375579e-01
-2.79045016e-01 -9.17261899e-01 3.62095207e-01 3.94645423e-01
1.07545801e-01 -1.06560159e+00 -8.73661816e-01 -6.35187477e-02
2.46360712e-02 -4.56118882e-01 -5.98369122e-01 2.52878666e-01
-6.78143144e-01 -1.42149615e+00 -7.47301131e-02 9.54815149e-02
8.38614047e-01 1.83197647e-01 9.96168256e-01 4.39816922e-01
-1.46367797e-03 2.94573069e-01 -2.79891044e-01 -6.30820811e-01
-3.68810624e-01 -2.72248209e-01 1.78663433e-01 -3.30884397e-01
2.86363989e-01 -5.51717758e-01 -3.93519104e-01 3.53016526e-01
-9.23384309e-01 -7.95378760e-02 4.17965978e-01 1.00526822e+00
1.93193242e-01 3.00532043e-01 7.40254402e-01 -1.29995203e+00
8.76067340e-01 -6.76361263e-01 -6.70472085e-01 1.91031396e-01
-8.84414136e-01 3.13269109e-01 6.80913448e-01 -5.30840099e-01
-1.05111170e+00 -4.44885492e-02 1.80298820e-01 -1.74898148e-01
-3.78280342e-01 6.23582780e-01 -2.54486203e-01 2.66233534e-01
1.02965355e+00 -2.90888727e-01 -9.36043188e-02 -3.26003343e-01
2.35701174e-01 5.62121272e-01 -4.76137502e-03 -7.52526999e-01
8.59780967e-01 2.86418289e-01 -1.13446988e-01 -4.28272635e-01
-6.18633866e-01 -3.25842835e-02 -6.45785034e-02 1.82960659e-01
3.12828511e-01 -6.12950742e-01 -6.91093624e-01 -2.10917830e-01
-7.33599782e-01 -4.75307167e-01 -3.67075801e-01 5.30095100e-01
-6.51864648e-01 1.72599107e-01 2.56625917e-02 -7.71637917e-01
1.08942986e-01 -1.30268967e+00 5.16032159e-01 -2.38671037e-03
-8.76253366e-01 -8.19675863e-01 -2.25180775e-01 1.50366932e-01
1.69408858e-01 3.82132292e-01 1.13434839e+00 -9.04645562e-01
-4.34177130e-01 -1.88329935e-01 2.31281489e-01 1.55265361e-01
4.55994129e-01 2.98022032e-01 -7.95239866e-01 -2.84960449e-01
-1.46554559e-01 -9.83828381e-02 6.45531952e-01 6.36371017e-01
1.21520758e+00 -8.60537946e-01 -4.24031883e-01 3.35541904e-01
1.15850341e+00 3.07604134e-01 5.70013702e-01 5.57043314e-01
2.02262983e-01 7.94266403e-01 7.57011354e-01 4.64146525e-01
2.41766766e-01 8.37495804e-01 3.49509865e-01 -3.37797310e-03
2.98519462e-01 -2.46401742e-01 3.61658186e-01 -1.95686474e-01
4.25153449e-02 -1.90832555e-01 -8.40126634e-01 7.18405247e-01
-1.82444632e+00 -1.22248435e+00 -5.47104105e-02 2.64690256e+00
9.11525846e-01 3.97189736e-01 2.56892145e-01 1.63293168e-01
6.09348238e-01 -1.67516954e-02 -6.53156221e-01 -8.04242611e-01
1.36973679e-01 -2.02831537e-01 6.83768213e-01 5.97951353e-01
-8.14079642e-01 5.27693212e-01 6.22587776e+00 3.09528053e-01
-1.07888246e+00 -1.51357993e-01 6.68807209e-01 -4.15862590e-01
-8.63926351e-01 4.14661050e-01 -4.14671481e-01 5.95213413e-01
9.40455019e-01 -6.71643794e-01 5.26267588e-01 7.13817239e-01
6.67198241e-01 -2.10478917e-01 -1.43744767e+00 2.25391254e-01
-5.10930538e-01 -1.30928445e+00 -6.54861480e-02 1.23120211e-01
6.57884657e-01 -2.41459429e-01 5.91571704e-02 1.36456728e-01
7.52693474e-01 -1.06600451e+00 8.03552389e-01 4.01511788e-01
5.82869411e-01 -9.18467820e-01 6.81484163e-01 5.69167554e-01
-4.79672343e-01 -3.82270545e-01 -3.30192149e-01 -4.61000949e-01
-1.79599240e-01 7.13461041e-01 -9.66987073e-01 3.65769506e-01
5.99010289e-01 3.65056962e-01 -3.21063608e-01 9.98690069e-01
-5.08385241e-01 8.80547345e-01 -3.06411982e-01 8.99425650e-04
2.83908676e-02 1.61165014e-01 7.47239590e-01 8.46864343e-01
2.87565619e-01 1.68767393e-01 -1.07372276e-01 8.86412203e-01
6.13879897e-02 5.59046445e-03 -1.04653561e+00 -3.08819283e-02
8.63574147e-01 7.93392003e-01 -5.30103564e-01 -9.15430933e-02
-4.10515815e-01 5.96925497e-01 1.48892999e-01 4.31169033e-01
-7.09012210e-01 -1.61069945e-01 9.79457796e-01 2.14924410e-01
3.76903564e-02 3.40951592e-01 -5.02327442e-01 -1.13869631e+00
4.29234207e-02 -1.32023776e+00 4.61663514e-01 -6.19253337e-01
-1.10995030e+00 4.51153470e-03 2.73391187e-01 -1.05325484e+00
-4.81489271e-01 -1.61494300e-01 -8.61554444e-01 9.72867429e-01
-1.14132333e+00 -6.11652970e-01 2.84015387e-01 1.83022827e-01
3.50206107e-01 3.02071124e-01 7.76373684e-01 -1.42344341e-01
-6.93546534e-01 6.34423614e-01 -2.16436774e-01 -3.46273988e-01
8.01507711e-01 -1.14480186e+00 3.24174374e-01 1.02948403e+00
4.57539335e-02 1.15892434e+00 1.25747669e+00 -7.91600227e-01
-1.02204835e+00 -9.36735392e-01 8.85866225e-01 -5.38625717e-01
4.82801259e-01 -2.30269507e-01 -8.70467424e-01 1.02847660e+00
-1.36421263e-01 -3.92222971e-01 6.25461578e-01 5.67073762e-01
-2.06552953e-01 1.21680565e-01 -1.25931144e+00 1.12999904e+00
1.07709992e+00 -1.79474145e-01 -7.14009404e-01 3.31205428e-01
7.12725103e-01 -1.72358423e-01 -6.03427589e-01 4.59550112e-01
3.32712352e-01 -9.48803246e-01 7.34135032e-01 -9.77710783e-01
3.93719822e-01 -3.73001069e-01 3.52613404e-02 -1.66139424e+00
-3.66708368e-01 -8.18529367e-01 3.11543614e-01 1.26142716e+00
8.29581797e-01 -8.77023697e-01 5.81761777e-01 1.29903495e+00
-2.88921017e-02 -7.38530517e-01 -8.37423027e-01 -8.25993419e-01
6.09594360e-02 -5.34139752e-01 1.05675387e+00 1.36550474e+00
2.29946300e-01 -7.61031406e-04 -3.75141859e-01 3.32737505e-01
2.15850607e-01 -5.44271655e-02 8.08328152e-01 -1.15568268e+00
-3.94125700e-01 -5.81202030e-01 -1.36962041e-01 -3.89076263e-01
2.48516098e-01 -6.69530749e-01 -1.15830228e-02 -1.22191012e+00
2.75404125e-01 -6.05294466e-01 -1.52918413e-01 9.06188250e-01
-6.13219321e-01 -4.45819080e-01 1.81875125e-01 1.40609637e-01
1.70985132e-01 2.37852275e-01 8.40900362e-01 1.65431619e-01
-3.67355794e-01 2.79921472e-01 -1.46285796e+00 8.86111021e-01
8.58061433e-01 -5.12522042e-01 -5.25805891e-01 -4.52990495e-02
3.62423897e-01 -9.94070526e-03 4.65881377e-01 -3.78963262e-01
-8.29958096e-02 -7.64125586e-01 1.04559660e-01 2.32110217e-01
-9.92188528e-02 -8.07427108e-01 5.99491179e-01 5.55879951e-01
-6.93699062e-01 7.59562999e-02 3.35180819e-01 3.78196210e-01
1.41264290e-01 -4.65488583e-01 5.80958009e-01 -1.56643644e-01
-4.29395288e-01 -1.34720162e-01 -3.31563979e-01 -8.38604048e-02
1.10011697e+00 -1.19849183e-01 -3.69054168e-01 -7.06042230e-01
-4.01669085e-01 2.32411370e-01 8.01525772e-01 3.66067916e-01
2.96349913e-01 -1.00975001e+00 -7.22552955e-01 -5.86562455e-02
1.07023641e-01 -2.60051906e-01 -4.86482829e-02 8.23433220e-01
1.57256991e-01 3.65999311e-01 1.14466809e-01 5.32500707e-02
-1.14882362e+00 7.57520139e-01 4.24968243e-01 -5.61578646e-02
-4.94810879e-01 7.15453088e-01 2.80030191e-01 -3.98837030e-01
7.62028247e-02 -2.75544524e-01 5.27046956e-02 -4.25863713e-02
3.53693664e-01 3.06319922e-01 5.37899993e-02 -1.73357666e-01
-3.85036170e-01 -4.06500287e-02 -1.92146882e-01 1.65769160e-02
1.43975079e+00 -5.62020345e-03 -1.73887275e-02 3.12240005e-01
6.26314998e-01 2.97535896e-01 -1.35552287e+00 1.80970788e-01
1.81034535e-01 -6.96414232e-01 2.84294263e-02 -1.08847547e+00
-7.03942120e-01 3.24642718e-01 1.57210201e-01 3.66065860e-01
1.10591400e+00 -1.64941400e-01 8.23930278e-03 2.64883101e-01
3.11395675e-01 -7.43370533e-01 -6.91766679e-01 1.28116067e-02
1.08073020e+00 -1.20492446e+00 2.56644070e-01 -2.85836160e-01
-6.36950552e-01 6.80499613e-01 5.20956337e-01 4.80460376e-02
2.39207923e-01 9.48278382e-02 -1.95792779e-01 -1.40011147e-01
-1.11040318e+00 -4.11246903e-02 2.15487972e-01 4.06624049e-01
5.88961065e-01 2.71274477e-01 -7.99540162e-01 5.23939788e-01
-4.35821980e-01 -2.45416954e-01 1.00598800e+00 7.53881693e-01
-6.53483495e-02 -1.06263483e+00 -5.75667799e-01 9.74855602e-01
-6.32620692e-01 -1.95018649e-01 -5.65439582e-01 9.95991886e-01
2.02578343e-02 1.15110743e+00 -1.00921914e-01 -1.72610655e-01
5.14578700e-01 6.88947886e-02 3.81814912e-02 -8.07310998e-01
-3.91765058e-01 -1.97704419e-01 6.01958334e-01 -6.28590882e-01
-2.36159787e-01 -1.11910093e+00 -1.13121021e+00 -4.62253660e-01
-5.67350864e-01 3.58418733e-01 3.01805675e-01 1.13146400e+00
3.68483335e-01 4.08753276e-01 5.79542518e-01 -4.56270039e-01
-8.90795767e-01 -7.97140479e-01 -3.30274552e-01 5.58373570e-01
4.29269731e-01 -9.28597927e-01 -8.34819436e-01 -1.04416259e-01]
|
[8.64247989654541, 5.599459171295166]
|
6911310b-d6d3-494a-a085-1235b9323aa2
|
benchmarking-foundation-models-with-language
|
2306.04181
| null |
https://arxiv.org/abs/2306.04181v1
|
https://arxiv.org/pdf/2306.04181v1.pdf
|
Benchmarking Foundation Models with Language-Model-as-an-Examiner
|
Numerous benchmarks have been established to assess the performance of foundation models on open-ended question answering, which serves as a comprehensive test of a model's ability to understand and generate language in a manner similar to humans. Most of these works focus on proposing new datasets, however, we see two main issues within previous benchmarking pipelines, namely testing leakage and evaluation automation. In this paper, we propose a novel benchmarking framework, Language-Model-as-an-Examiner, where the LM serves as a knowledgeable examiner that formulates questions based on its knowledge and evaluates responses in a reference-free manner. Our framework allows for effortless extensibility as various LMs can be adopted as the examiner, and the questions can be constantly updated given more diverse trigger topics. For a more comprehensive and equitable evaluation, we devise three strategies: (1) We instruct the LM examiner to generate questions across a multitude of domains to probe for a broad acquisition, and raise follow-up questions to engage in a more in-depth assessment. (2) Upon evaluation, the examiner combines both scoring and ranking measurements, providing a reliable result as it aligns closely with human annotations. (3) We additionally propose a decentralized Peer-examination method to address the biases in a single examiner. Our data and benchmarking results are available at: https://lmexam.com.
|
['Lei Hou', 'Juanzi Li', 'Jiayin Zhang', 'Haozhe Lyu', 'Yijia Xiao', 'Kaisheng Zeng', 'Jifan Yu', 'Xiaozhi Wang', 'Yuze He', 'Xin Lv', 'Yixin Cao', 'Jiahao Ying', 'Yushi Bai']
|
2023-06-07
| null | null | null | null |
['open-question']
|
['natural-language-processing']
|
[-1.31979227e-01 1.75583199e-01 9.38825011e-02 -6.39295220e-01
-1.55476630e+00 -1.16770554e+00 3.84253353e-01 2.16142386e-01
-4.09385532e-01 4.61443186e-01 2.21293926e-01 -5.23527563e-01
-1.91429138e-01 -7.66809702e-01 -6.98208869e-01 1.07940003e-01
5.84788263e-01 5.69806516e-01 5.39287031e-01 -4.31520671e-01
6.82438016e-02 -8.25354904e-02 -1.49060917e+00 5.64709127e-01
1.26403570e+00 7.25015461e-01 6.28901944e-02 8.10301304e-01
-1.34786040e-01 1.02663600e+00 -9.48294044e-01 -9.03365731e-01
8.80264267e-02 -2.95564473e-01 -1.41330898e+00 -2.46567965e-01
7.07827210e-01 -4.71043289e-01 1.12781107e-01 7.42959380e-01
5.98060787e-01 1.40812874e-01 1.64365336e-01 -9.70535338e-01
-9.35397625e-01 7.87379205e-01 1.49119347e-01 -1.05943628e-01
9.68937635e-01 5.77442408e-01 1.22541964e+00 -7.36145377e-01
6.07507944e-01 1.05021667e+00 4.04671282e-01 5.66309631e-01
-8.89095366e-01 -5.37312269e-01 1.62766859e-01 1.85361207e-01
-8.14714491e-01 -2.30490476e-01 4.36788142e-01 -2.90682614e-01
5.60867906e-01 6.53801203e-01 2.75015414e-01 1.23614049e+00
-3.16851348e-01 6.55605555e-01 1.41804385e+00 -4.93661016e-01
3.17431837e-01 5.10872006e-01 4.43738699e-01 4.40384895e-01
1.34228453e-01 -2.16057584e-01 -6.28685415e-01 -3.12063754e-01
3.38399000e-02 -2.79413313e-01 -3.39395791e-01 1.16760790e-01
-1.10184240e+00 5.99857211e-01 2.93167263e-01 2.85589874e-01
-4.20539588e-01 -1.89437166e-01 1.03995815e-01 6.77859247e-01
1.73103034e-01 1.07060611e+00 -5.33294797e-01 -4.43577528e-01
-8.80379915e-01 6.49773061e-01 1.22924793e+00 5.51449358e-01
7.48708963e-01 -6.33924663e-01 -7.22021759e-01 1.02415872e+00
4.43899482e-01 5.85905612e-01 4.41962183e-01 -1.26009846e+00
4.96263474e-01 1.11443985e+00 1.70211360e-01 -7.50227094e-01
4.52236384e-02 -5.69427192e-01 -7.26928040e-02 1.05830677e-01
4.91520524e-01 -2.11584255e-01 -4.00346845e-01 1.77643991e+00
5.41244924e-01 -2.13469163e-01 -5.65145575e-02 7.71754920e-01
1.11848676e+00 3.11076343e-01 5.52031118e-03 4.10236865e-01
1.76593816e+00 -1.10505188e+00 -6.35863423e-01 -1.02653518e-01
6.71071947e-01 -9.76334512e-01 1.76128948e+00 5.58035672e-01
-1.23309457e+00 -6.55964077e-01 -7.48210490e-01 -2.73231447e-01
-2.39073247e-01 -9.40977335e-02 8.48853737e-02 6.57835841e-01
-1.20017648e+00 2.30502933e-01 -4.95368123e-01 -3.49243462e-01
-4.64281142e-02 -1.13254815e-01 -1.97741941e-01 -2.19005883e-01
-1.39487541e+00 8.42962384e-01 -1.44974157e-01 -2.05805287e-01
-9.71965909e-01 -7.10013092e-01 -6.90574944e-01 4.63827215e-02
3.46466064e-01 -6.22600853e-01 1.98303175e+00 -3.80769342e-01
-1.55649412e+00 8.67285967e-01 -2.32492998e-01 -1.36452969e-02
5.25172770e-01 -3.74892175e-01 -3.32789510e-01 1.41507626e-01
3.24522376e-01 3.94131213e-01 1.25898927e-01 -1.15193093e+00
-5.01846135e-01 -1.15515798e-01 7.74569929e-01 8.92141834e-02
-5.04195869e-01 1.70604691e-01 -4.66667354e-01 -2.56492525e-01
-1.31980121e-01 -6.25212252e-01 -5.70069440e-02 -3.62469822e-01
-3.27675968e-01 -5.17893672e-01 4.17352736e-01 -7.47087538e-01
1.74841273e+00 -1.81087315e+00 -3.57905954e-01 6.81053549e-02
5.61350703e-01 4.36650723e-01 -4.03295517e-01 7.41248131e-01
1.76053703e-01 3.08879316e-01 -1.01872794e-01 -3.24430019e-01
4.74744856e-01 -1.82728976e-01 -3.98889661e-01 -3.19721907e-01
2.97348738e-01 1.08988571e+00 -1.04002738e+00 -4.38663840e-01
-3.05758685e-01 -4.75723296e-02 -8.26886177e-01 6.67666733e-01
-4.30937916e-01 2.92499870e-01 -4.98253912e-01 8.26494634e-01
4.31558430e-01 -6.73308492e-01 -3.90958413e-02 2.69601077e-01
1.61902532e-02 8.75744700e-01 -1.18031120e+00 1.50369287e+00
-6.12824559e-01 4.30883139e-01 8.39905515e-02 -4.23711777e-01
1.07677722e+00 3.99566591e-01 -1.52464494e-01 -8.80312741e-01
-2.01519072e-01 3.53903413e-01 -8.95124227e-02 -7.79988408e-01
5.42969704e-01 3.39669764e-01 -2.18256772e-01 1.00865817e+00
-3.93690914e-02 -5.20329550e-02 5.68945408e-01 4.50086683e-01
1.66421449e+00 -8.43613148e-02 -3.19691077e-02 -1.20889200e-02
7.43929744e-01 6.42238706e-02 2.90264368e-01 1.13069665e+00
-2.81198382e-01 4.13883597e-01 4.50858802e-01 -1.53965011e-01
-5.35619080e-01 -1.11346030e+00 -4.91339788e-02 1.43321478e+00
-1.83683008e-01 -5.43001056e-01 -9.77770507e-01 -8.72868419e-01
-8.78637433e-02 7.57982552e-01 -3.88227016e-01 8.61373991e-02
-4.11548913e-01 -1.56143224e-02 8.06510091e-01 2.58240998e-01
4.93792087e-01 -1.17350888e+00 -5.77569783e-01 2.74073839e-01
-7.97633410e-01 -9.38869834e-01 -2.76122421e-01 -3.03789735e-01
-4.00765151e-01 -1.22102380e+00 -4.21243310e-01 -6.01875842e-01
5.26214361e-01 1.70911893e-01 1.55860412e+00 4.24552292e-01
1.86315447e-01 8.80894899e-01 -5.81321239e-01 -3.72174740e-01
-7.92806327e-01 3.30751449e-01 -3.75454694e-01 -1.70205399e-01
5.50462127e-01 -2.55489737e-01 -8.65443587e-01 8.08049321e-01
-1.00562847e+00 -2.89424032e-01 5.21719098e-01 5.58252096e-01
4.15026814e-01 -5.94694972e-01 8.05167675e-01 -1.09576952e+00
1.35648656e+00 -6.96556389e-01 -5.60071886e-01 6.42407179e-01
-6.41244054e-01 -8.21522549e-02 5.69806397e-01 -3.95711124e-01
-1.00287938e+00 -5.65680027e-01 -4.83117849e-01 1.35824800e-01
-3.92214328e-01 6.55398846e-01 -5.36487401e-02 -9.76589993e-02
1.07218671e+00 6.02768622e-02 -2.15524197e-01 -5.71372449e-01
4.63162303e-01 9.88744736e-01 5.86928844e-01 -1.04105163e+00
1.06187952e+00 -8.55320506e-03 -8.69084895e-01 -1.42421260e-01
-1.01102746e+00 -5.42467833e-01 -1.28734708e-01 -4.53949690e-01
5.25395453e-01 -8.18582892e-01 -1.00913179e+00 2.40695745e-01
-1.14140511e+00 -5.32393694e-01 -3.68801117e-01 3.35557461e-01
-4.36725691e-02 2.05065474e-01 -7.33624220e-01 -7.00651348e-01
-4.27917153e-01 -1.26749051e+00 9.63999152e-01 4.39877689e-01
-7.76886106e-01 -9.27369237e-01 5.34679234e-01 1.07874036e+00
7.06582963e-01 1.85564801e-03 6.77756011e-01 -1.08573079e+00
-7.05999017e-01 -3.88538271e-01 9.21797305e-02 5.56858659e-01
-2.69670367e-01 5.10950685e-02 -1.12189376e+00 -1.01364851e-01
9.73733664e-02 -9.15574133e-01 2.79569060e-01 -4.70131099e-01
1.09535134e+00 -4.06983525e-01 2.10436597e-01 1.54663309e-01
9.25502181e-01 -1.43902540e-01 4.34266657e-01 6.46735549e-01
1.25438407e-01 8.45552623e-01 5.79712033e-01 5.14857136e-02
9.99233723e-01 3.59935582e-01 2.16485873e-01 2.43526213e-02
-1.13121882e-01 -5.22314608e-01 5.07628560e-01 1.20255709e+00
4.31552202e-01 -4.38301861e-01 -1.08922625e+00 6.45943046e-01
-1.53501642e+00 -7.12127447e-01 -3.28424156e-01 2.18835664e+00
1.18289173e+00 2.93818885e-03 -7.79955238e-02 -8.16547796e-02
3.82305950e-01 1.01281404e-01 -3.66548002e-01 -4.63578820e-01
8.87648389e-02 5.30770421e-01 -2.90845096e-01 5.61104238e-01
-3.70867789e-01 5.53744495e-01 6.55461311e+00 5.45750380e-01
-8.55560720e-01 3.69276434e-01 6.13764346e-01 -8.75829756e-02
-9.35777664e-01 3.50528181e-01 -8.46474767e-01 6.10130608e-01
1.09923053e+00 -2.29382217e-01 2.99123257e-01 7.14924574e-01
-1.26538165e-02 2.18176246e-02 -1.15083694e+00 3.91197979e-01
-5.09728380e-02 -1.10724330e+00 7.91994706e-02 -1.79839715e-01
6.88711643e-01 1.64280236e-01 2.55363155e-03 8.37897718e-01
5.65452099e-01 -8.33913803e-01 6.78544462e-01 7.00451851e-01
5.94321549e-01 -3.26242745e-01 7.29354501e-01 6.73869431e-01
-8.25458109e-01 -4.59508076e-02 3.55731919e-02 -1.65993303e-01
1.32086754e-01 5.28021753e-01 -8.27451229e-01 6.24168932e-01
7.44125068e-01 -2.02846497e-01 -9.87802863e-01 9.45659757e-01
-7.07496107e-01 1.03891039e+00 -1.74312517e-01 -4.04820889e-01
1.47722185e-01 9.11438689e-02 2.91819841e-01 1.06482971e+00
1.67318225e-01 6.15074113e-02 2.02077731e-01 1.05090010e+00
-4.98693675e-01 4.15576309e-01 -1.63185045e-01 -2.77134776e-02
8.69759142e-01 1.47530556e+00 -9.59031284e-02 -3.77917498e-01
-3.71652335e-01 4.47641313e-01 7.21102059e-01 3.48836094e-01
-6.23281181e-01 -4.85594183e-01 3.53240967e-01 3.52964163e-01
-1.37011990e-01 2.14761868e-01 -2.61643343e-02 -1.07690525e+00
5.66586733e-01 -1.66883314e+00 5.66346645e-01 -7.86926568e-01
-1.31278729e+00 6.99231267e-01 -2.60708630e-01 -9.07624483e-01
-4.42821264e-01 -3.93045217e-01 -8.63456190e-01 9.52615798e-01
-1.66131246e+00 -7.67266810e-01 -7.20036149e-01 4.65724677e-01
2.94964731e-01 7.89762735e-02 7.61246741e-01 4.64584291e-01
-4.75565821e-01 1.02190757e+00 -3.81625235e-01 1.20065868e-01
1.17388892e+00 -1.32659209e+00 4.49098766e-01 8.79160523e-01
1.91939667e-01 1.11172128e+00 3.99830341e-01 -5.62539577e-01
-1.22815657e+00 -9.05001581e-01 1.30174100e+00 -1.18076861e+00
7.45811224e-01 -3.67368877e-01 -1.37502050e+00 4.46967632e-01
2.95991927e-01 -2.45234236e-01 9.81596708e-01 3.81492585e-01
-3.94988000e-01 -2.20555350e-01 -9.27543938e-01 4.44528729e-01
7.52370775e-01 -8.60859811e-01 -9.10957694e-01 3.89218688e-01
9.13151920e-01 -5.76991320e-01 -9.52099800e-01 3.36671442e-01
4.32261825e-01 -1.09940648e+00 4.34096873e-01 -6.59474492e-01
6.90908611e-01 -3.93554211e-01 6.96092471e-02 -8.81551564e-01
-1.95301980e-01 -8.63094926e-01 -9.68104675e-02 1.54138076e+00
6.88142896e-01 -1.02244818e+00 3.06123942e-01 9.22302783e-01
2.26304457e-02 -1.06258667e+00 -6.21792555e-01 -5.46983361e-01
-6.50882348e-02 -5.96012592e-01 9.68848348e-01 7.88464248e-01
2.36729588e-02 2.11195394e-01 1.33284509e-01 2.82304972e-01
2.17080936e-01 8.08396190e-02 1.13916326e+00 -1.01519561e+00
-6.33712590e-01 -4.83452827e-01 2.23557681e-01 -1.20990503e+00
-1.15853339e-01 -9.33535516e-01 -1.59895804e-03 -1.59014976e+00
1.92542821e-01 -4.01023567e-01 -4.80528802e-01 3.76873851e-01
-7.16229916e-01 8.67679939e-02 8.77072364e-02 1.31756112e-01
-1.22255003e+00 2.26278573e-01 1.22908068e+00 2.14973673e-01
9.60444361e-02 7.63589442e-02 -1.37725866e+00 5.83502531e-01
5.94725788e-01 -5.50055742e-01 -3.11523408e-01 -7.31278062e-01
6.17786825e-01 -5.27375303e-02 4.49821234e-01 -1.00982404e+00
6.42191648e-01 1.07660696e-01 -3.72969471e-02 -3.57276022e-01
-2.14386389e-01 -3.00465733e-01 -1.76646456e-01 2.15220481e-01
-6.72111213e-01 4.57171649e-01 5.58986748e-03 1.61578119e-01
-4.96563047e-01 -5.52866340e-01 3.17992866e-01 -1.60943016e-01
-2.72215664e-01 1.57875627e-01 -7.90789276e-02 5.08563697e-01
6.53037727e-01 8.00886899e-02 -7.87287772e-01 -6.45076871e-01
-2.08670706e-01 7.36295581e-01 4.43669766e-01 4.68099475e-01
2.50396788e-01 -1.09329510e+00 -1.03402460e+00 -2.31332872e-02
5.01198947e-01 3.24314833e-02 3.10851723e-01 5.34095824e-01
-4.47899342e-01 4.27881420e-01 3.85878384e-01 -5.07569432e-01
-8.34817410e-01 -7.26300031e-02 2.77160466e-01 -8.40535522e-01
1.11782111e-01 9.26255524e-01 -2.04474702e-02 -1.18165660e+00
2.16039270e-01 -2.70538777e-01 -2.39439383e-01 -3.79071459e-02
7.83279896e-01 3.49336684e-01 3.30886722e-01 6.84425607e-02
-1.03265703e-01 1.24196000e-01 -9.58187580e-02 -2.46259362e-01
1.07620442e+00 -9.69347134e-02 -2.54259050e-01 4.47949588e-01
7.35162318e-01 5.44098735e-01 -8.60986233e-01 -5.39234161e-01
4.25220504e-02 -2.75735229e-01 -4.19646204e-01 -1.38803089e+00
-3.72919261e-01 6.31804228e-01 2.96704113e-01 5.12400508e-01
1.01803279e+00 9.79393572e-02 8.29794765e-01 5.31728685e-01
1.74188375e-01 -1.01277459e+00 3.44054788e-01 5.94670534e-01
9.91757572e-01 -1.27653646e+00 -4.33715075e-01 -4.17029411e-02
-4.06729102e-01 8.92726123e-01 1.09500575e+00 3.27020437e-01
-1.04402248e-02 -7.33414367e-02 6.30915582e-01 -3.12077761e-01
-1.17246401e+00 -4.35402803e-02 5.63619614e-01 2.35431716e-01
7.01109111e-01 2.66215652e-02 -3.69778961e-01 1.04169095e+00
-6.59542620e-01 -4.90146652e-02 3.83158207e-01 8.18591833e-01
-4.24144357e-01 -1.31993949e+00 -5.35016596e-01 3.76728565e-01
-4.52118009e-01 -2.47891285e-02 -5.60876846e-01 6.04010105e-01
4.49850895e-02 1.44008315e+00 -3.30686927e-01 -2.59311914e-01
7.37002611e-01 2.40546376e-01 -2.83799600e-02 -9.18868303e-01
-1.12583697e+00 -6.59758687e-01 2.70999879e-01 -8.16045523e-01
1.04224913e-01 -4.57670718e-01 -7.87302613e-01 -9.49404836e-02
-2.62221217e-01 5.41496277e-01 6.23893321e-01 9.18878913e-01
6.30434692e-01 4.68769521e-01 7.48515189e-01 2.38488019e-01
-1.09517026e+00 -1.09872711e+00 5.76106980e-02 7.05368400e-01
9.45082977e-02 -1.67683184e-01 -5.02072155e-01 -2.50762463e-01]
|
[11.350000381469727, 8.094339370727539]
|
42fcfc92-41ec-43bf-a6f7-7ba504f4be74
|
spatial-and-modal-optimal-transport-for-fast
|
2305.02774
| null |
https://arxiv.org/abs/2305.02774v1
|
https://arxiv.org/pdf/2305.02774v1.pdf
|
Spatial and Modal Optimal Transport for Fast Cross-Modal MRI Reconstruction
|
Multi-modal Magnetic Resonance Imaging (MRI) plays an important role in clinical medicine. However, the acquisitions of some modalities, such as the T2-weighted modality, need a long time and they are always accompanied by motion artifacts. On the other hand, the T1-weighted image (T1WI) shares the same underlying information with T2-weighted image (T2WI), which needs a shorter scanning time. Therefore, in this paper we accelerate the acquisition of the T2WI by introducing the auxiliary modality (T1WI). Concretely, we first reconstruct high-quality T2WIs with under-sampled T2WIs. Here, we realize fast T2WI reconstruction by reducing the sampling rate in the k-space. Second, we establish a cross-modal synthesis task to generate the synthetic T2WIs for guiding better T2WI reconstruction. Here, we obtain the synthetic T2WIs by decomposing the whole cross-modal generation mapping into two OT processes, the spatial alignment mapping on the T1 image manifold and the cross-modal synthesis mapping from aligned T1WIs to T2WIs. It overcomes the negative transfer caused by the spatial misalignment. Then, we prove the reconstruction and the synthesis tasks are well complementary. Finally, we compare it with state-of-the-art approaches on an open dataset FastMRI and an in-house dataset to testify the validity of the proposed method.
|
['Shihui Ying', 'Dinggang Shen', 'Qian Wang', 'Jun Shi', 'Zhijie Wen', 'Qi Wang']
|
2023-05-04
| null | null | null | null |
['mri-reconstruction']
|
['computer-vision']
|
[ 4.11292344e-01 5.81053570e-02 1.04970224e-01 -4.71190885e-02
-8.92824769e-01 -2.12008446e-01 4.95305151e-01 -4.15473849e-01
-3.40773255e-01 5.61552703e-01 3.64545614e-01 6.66309800e-03
-5.53314090e-01 -6.20117903e-01 -5.15431881e-01 -1.03593576e+00
2.89416108e-02 4.24063593e-01 4.65784997e-01 -1.05232045e-01
-9.86665860e-02 -7.85400718e-03 -1.09400129e+00 9.23226103e-02
1.03777874e+00 9.21147346e-01 5.41836619e-01 1.50847197e-01
2.62937993e-02 4.31320757e-01 -7.51399174e-02 4.36839648e-02
1.83900625e-01 -5.77460825e-01 -7.43352592e-01 1.38440296e-01
1.35258868e-01 -2.61432260e-01 -4.98096049e-01 1.23494852e+00
6.81111932e-01 2.20454812e-01 5.27265668e-01 -9.23171341e-01
-5.08996606e-01 6.18241429e-01 -9.96025205e-01 3.13596129e-01
-1.06998838e-01 1.51015639e-01 3.26009810e-01 -8.25264871e-01
1.00221992e+00 7.82100141e-01 4.40269291e-01 4.30659831e-01
-1.18736124e+00 -5.08928716e-01 -1.62380815e-01 3.26957405e-01
-1.07276928e+00 -1.54997751e-01 9.39782679e-01 -5.41509449e-01
1.26317635e-01 1.79384798e-01 4.79436308e-01 7.66336799e-01
3.31334144e-01 7.18121827e-01 1.51552641e+00 -2.01733172e-01
-2.71568060e-01 -3.58476698e-01 -4.04490791e-02 4.39845383e-01
-2.79958360e-02 1.27216682e-01 -2.94143170e-01 1.65103346e-01
9.04317200e-01 2.10832715e-01 -7.39369094e-01 -3.94793212e-01
-2.07154155e+00 3.21387142e-01 2.97573477e-01 9.09672558e-01
-5.87028563e-01 -1.34695351e-01 4.34130579e-01 1.16441511e-01
3.15093905e-01 7.12861195e-02 9.79604125e-02 1.69372693e-01
-9.44063902e-01 -4.22389992e-02 2.73154303e-03 7.20886230e-01
5.33704817e-01 -1.11545160e-01 -3.30968946e-01 7.57347286e-01
2.20397189e-01 7.06881762e-01 1.04612350e+00 -6.93549573e-01
2.95117795e-01 2.28417348e-02 -6.49726912e-02 -9.77665901e-01
-5.46503484e-01 -5.15985966e-01 -1.25989795e+00 -2.61918098e-01
4.90960449e-01 1.76874995e-02 -6.97774708e-01 1.87351179e+00
5.68371117e-01 5.17314374e-01 -1.91268951e-01 1.39928079e+00
6.73686028e-01 5.14698863e-01 -1.56013027e-01 -7.29740620e-01
1.39996982e+00 -1.07616568e+00 -1.15990496e+00 8.98386240e-02
3.94052297e-01 -8.66977990e-01 8.81290674e-01 2.81925291e-01
-1.27361035e+00 -5.34842253e-01 -8.97630811e-01 2.97815531e-01
3.25439781e-01 2.79676151e-02 3.20872217e-01 2.52820939e-01
-6.17083430e-01 5.14397502e-01 -1.05768180e+00 8.41382667e-02
3.23099531e-02 3.04488745e-02 -7.14171350e-01 -3.71917814e-01
-1.48815298e+00 1.11766684e+00 3.58020335e-01 3.82803649e-01
-7.97185898e-01 -1.18912935e+00 -5.30773163e-01 -5.95009685e-01
5.52013278e-01 -3.69022727e-01 8.51505816e-01 -5.04247904e-01
-1.23508453e+00 7.22446382e-01 -4.56641987e-02 9.98463631e-02
7.96636701e-01 1.86210781e-01 -7.31341839e-01 4.79913384e-01
2.54280567e-01 3.97328913e-01 9.95911360e-01 -1.10499668e+00
-2.39846706e-01 -6.48876607e-01 -3.95147651e-01 1.48500547e-01
-7.99624324e-02 -3.51977944e-01 -3.62567842e-01 -8.59330118e-01
7.02669203e-01 -1.02782142e+00 -5.67108132e-02 -2.44496167e-01
-5.22777021e-01 3.38452488e-01 5.92401624e-01 -1.02460825e+00
1.31508183e+00 -2.26878667e+00 5.45777738e-01 2.40820140e-01
5.98245680e-01 -7.79295191e-02 -1.71604633e-01 -1.07031666e-01
-5.18411875e-01 -2.31258333e-01 -5.83012521e-01 1.22211948e-01
-2.74153322e-01 -7.75441900e-02 -2.16780808e-02 8.28428864e-01
-2.65745521e-01 1.11195385e+00 -1.12511230e+00 -8.06172013e-01
1.77851543e-01 3.38314652e-01 -3.55792731e-01 1.38993800e-01
3.49891633e-01 1.19025540e+00 -5.25700927e-01 2.74159461e-01
9.33917999e-01 -1.06160812e-01 2.40581840e-01 -1.04549694e+00
-4.04742002e-01 -1.37269571e-01 -1.05022466e+00 2.27532482e+00
-4.05122578e-01 7.28135929e-02 2.43542984e-01 -1.31669164e+00
6.23239577e-01 5.83311141e-01 1.11519754e+00 -9.68183696e-01
1.26302123e-01 5.96951723e-01 2.50973076e-01 -8.41982126e-01
-1.81438938e-01 -8.62491429e-01 3.90358210e-01 6.87986851e-01
3.80623676e-02 -2.54870594e-01 1.50942773e-01 -1.53663084e-01
6.92719221e-01 8.94622952e-02 -1.45351559e-01 -5.28675914e-01
7.38642335e-01 -2.40902066e-01 3.78646433e-01 3.30556989e-01
-3.12495738e-01 8.17795336e-01 4.24053557e-02 -2.32883573e-01
-1.15939271e+00 -1.14625919e+00 -4.93116260e-01 2.45299622e-01
4.20361191e-01 2.00541779e-01 -7.13272274e-01 -5.68109930e-01
-4.01790351e-01 3.11230183e-01 -6.24702930e-01 -2.56443501e-01
-9.29119825e-01 -1.19110382e+00 5.01015306e-01 1.82039231e-01
6.68520272e-01 -7.68613517e-01 -5.21767795e-01 3.47162336e-01
-8.72181773e-01 -1.04144704e+00 -8.94531548e-01 -7.43734315e-02
-1.13882804e+00 -9.70085502e-01 -1.46715140e+00 -5.80359757e-01
5.84751189e-01 4.17726159e-01 8.19535315e-01 -3.86431552e-02
-8.98707658e-02 1.46967530e-01 -2.42429733e-01 1.52278736e-01
-1.89486891e-01 -3.95653173e-02 1.11776799e-01 2.67655104e-01
-2.88534790e-01 -8.04150760e-01 -8.59954834e-01 5.68884313e-01
-1.42079699e+00 5.45991004e-01 6.44843221e-01 1.02600110e+00
8.91510427e-01 2.16776669e-01 5.87895572e-01 -5.90336621e-01
2.65322536e-01 -4.27839816e-01 -1.25142783e-01 4.47857857e-01
-2.72535950e-01 2.52210945e-01 3.92708302e-01 -7.19629407e-01
-1.00217187e+00 -2.56269157e-01 -2.46959597e-01 -4.72839564e-01
3.52790579e-02 6.03797317e-01 -4.64321151e-02 -1.08446613e-01
4.09221023e-01 7.76415825e-01 4.48092669e-01 -4.62026298e-01
1.63327843e-01 2.48423755e-01 6.96814001e-01 -5.91252685e-01
8.87379110e-01 7.67920434e-01 2.38559395e-01 -8.08945000e-01
-8.03182364e-01 -9.78047103e-02 -6.70659900e-01 -4.58474368e-01
9.45950985e-01 -3.56640488e-01 -6.05923593e-01 6.49573743e-01
-8.59120846e-01 -1.40316576e-01 -1.81918949e-01 9.80385423e-01
-5.45185328e-01 7.52746284e-01 -6.32667422e-01 -1.41457483e-01
-3.39663833e-01 -1.64145684e+00 9.55634296e-01 -1.34577498e-01
1.79126337e-01 -9.04531121e-01 1.72457546e-01 3.86542737e-01
6.03768885e-01 3.84429932e-01 1.00772774e+00 -2.28830829e-01
-3.97696763e-01 2.43902266e-01 -1.31669447e-01 1.42960057e-01
1.30676687e-01 -6.94443345e-01 -4.86945093e-01 -2.08952084e-01
5.09817064e-01 -1.01340704e-01 5.16014099e-01 6.96809530e-01
1.14568782e+00 2.96312660e-01 6.90533146e-02 7.18745232e-01
1.16954064e+00 2.51618028e-01 7.13509321e-01 3.87593359e-01
6.91339612e-01 8.12388062e-01 7.33994544e-01 2.13855598e-02
3.44640076e-01 9.27100778e-01 2.03757823e-01 -2.75178343e-01
-2.02002585e-01 -1.85042009e-01 -9.43642706e-02 1.38241601e+00
-2.77921855e-01 4.24564511e-01 -1.03491211e+00 4.39607531e-01
-1.81911969e+00 -7.55078912e-01 -3.83136541e-01 2.18127084e+00
8.92030895e-01 -2.50878096e-01 -8.71415287e-02 1.15674481e-01
8.62566411e-01 2.20334262e-01 -6.63991749e-01 8.05371702e-01
-1.21858276e-01 1.47958547e-02 4.59274352e-01 6.39576912e-01
-8.03822815e-01 2.68371731e-01 5.78748035e+00 1.07391131e+00
-1.51203573e+00 6.15737379e-01 6.46181285e-01 -1.09690905e-01
-7.94987500e-01 -2.71564722e-01 3.49481367e-02 6.56964004e-01
5.80499768e-01 -3.32199037e-01 5.22268534e-01 1.95833623e-01
3.44467729e-01 -5.42099997e-02 -7.40937352e-01 1.18749559e+00
-4.51521249e-03 -1.16808105e+00 -5.59761189e-02 -2.97285542e-02
5.35935342e-01 -1.91172317e-01 1.17770799e-01 3.51714753e-02
-4.84312922e-01 -8.65243852e-01 7.07953870e-01 7.18325555e-01
1.28482878e+00 -5.24104953e-01 6.46169722e-01 3.33384305e-01
-1.17806923e+00 4.24620211e-01 -3.51223424e-02 6.40335083e-01
7.04497576e-01 9.72843111e-01 5.79223363e-03 1.11565232e+00
4.55539316e-01 6.88413739e-01 -2.27836564e-01 7.61616349e-01
3.81839438e-03 1.60813540e-01 -1.62231132e-01 7.82589018e-01
2.43878692e-01 -7.49276042e-01 5.76294899e-01 6.77178562e-01
6.59816623e-01 4.74610299e-01 1.08455218e-01 7.74385452e-01
3.26719642e-01 -1.83556490e-02 -4.08565730e-01 2.73957998e-01
1.56531617e-01 1.29342079e+00 -6.55564249e-01 -4.32037652e-01
-3.15484107e-01 9.24573481e-01 -3.15106273e-01 4.64857548e-01
-1.02351856e+00 -1.69456840e-01 -3.98503765e-02 3.07835400e-01
-2.60584295e-01 -1.87051117e-01 -8.62082690e-02 -1.54516447e+00
1.50038496e-01 -9.21333492e-01 8.95193741e-02 -7.54391432e-01
-1.15810502e+00 7.53653705e-01 2.32117355e-01 -1.37335324e+00
2.05270238e-02 -3.41088653e-01 -3.61077666e-01 9.73135471e-01
-1.62287390e+00 -9.08648908e-01 -5.07280350e-01 7.00023472e-01
1.35561690e-01 2.35996827e-01 3.05984229e-01 1.02232778e+00
-4.05896753e-01 2.02891439e-01 -2.66320817e-02 -1.39454395e-01
9.26682413e-01 -9.91748393e-01 -2.20245183e-01 5.89161336e-01
-4.38508421e-01 7.45715797e-01 5.29560566e-01 -7.06298411e-01
-1.59172904e+00 -8.89040768e-01 4.66633081e-01 3.94574367e-02
8.38458121e-01 1.39143109e-01 -1.05775189e+00 5.02980709e-01
-1.24766037e-01 3.30271870e-01 3.42039853e-01 -5.85283399e-01
1.36768132e-01 -6.24295846e-02 -1.21865249e+00 3.67892355e-01
9.74842906e-01 -5.25798261e-01 -6.57526314e-01 2.79235989e-01
6.84949636e-01 -6.67190254e-01 -1.31257057e+00 5.20101011e-01
7.50398934e-01 -9.07440066e-01 1.27255261e+00 -2.21261919e-01
7.04193771e-01 -6.00325286e-01 1.84716005e-02 -1.42966831e+00
-3.90029073e-01 -3.75157326e-01 1.25910029e-01 7.30394840e-01
2.19344441e-03 -7.36457825e-01 4.03089494e-01 2.69967467e-01
-3.45477909e-01 -6.68567121e-01 -1.12681329e+00 -7.17091143e-01
2.07538441e-01 -2.73793131e-01 5.06899178e-01 1.18280566e+00
-8.95014331e-02 2.16351002e-02 -6.33313119e-01 -2.57773616e-04
1.09434807e+00 2.59115875e-01 1.28333181e-01 -8.56327951e-01
-2.91978836e-01 -2.86014289e-01 4.94928956e-02 -1.11603260e+00
-2.66712233e-02 -1.02938724e+00 2.99770311e-02 -1.17713356e+00
4.22974318e-01 -6.59437895e-01 -4.43708092e-01 -3.60348411e-02
-2.98998207e-01 3.83369952e-01 1.30708352e-01 5.16564608e-01
-2.17255622e-01 7.00585008e-01 2.30222511e+00 -1.30446717e-01
1.71267182e-01 -4.72402215e-01 -2.87077278e-01 5.38243651e-01
3.13376606e-01 -1.51076615e-01 -4.88695174e-01 -4.28115129e-01
-1.61632337e-02 6.63600624e-01 1.42642081e-01 -7.97783434e-01
8.14119130e-02 -1.47701249e-01 2.34523192e-02 -5.83683133e-01
9.25437082e-04 -1.00034916e+00 5.34629405e-01 6.98158622e-01
-1.67841092e-01 3.92920561e-02 -1.41135022e-01 1.64714426e-01
-4.77276176e-01 -1.50519863e-01 1.12824845e+00 -2.20638752e-01
-2.59061307e-01 7.50836611e-01 -6.30965754e-02 2.05430448e-01
8.85178506e-01 3.37241031e-02 -1.73473030e-01 -1.25011817e-01
-9.40759838e-01 1.61611989e-01 2.02204630e-01 2.87660450e-01
6.99106276e-01 -1.62300611e+00 -7.44172513e-01 2.62050599e-01
-7.55840093e-02 -8.60604197e-02 1.11424184e+00 2.02245307e+00
-3.95398557e-01 2.19687998e-01 -4.43338186e-01 -8.10571074e-01
-6.95271075e-01 6.64665878e-01 6.38377070e-01 -5.46317220e-01
-1.01108527e+00 3.19036186e-01 5.70106745e-01 -6.62655652e-01
-2.63421327e-01 -9.24965069e-02 -1.90938622e-01 1.01962917e-01
7.52162635e-01 3.83790791e-01 2.26256773e-01 -9.10160184e-01
-4.26858813e-01 9.92533684e-01 2.80398905e-01 -3.59720320e-01
1.37479317e+00 -3.24853152e-01 -2.87579000e-01 5.39971769e-01
1.22170126e+00 -1.16993718e-01 -1.00668526e+00 -3.38315964e-01
-1.27553239e-01 -5.87025523e-01 2.71671027e-01 -4.30865914e-01
-1.45612073e+00 9.77981925e-01 7.21599042e-01 -4.01303284e-02
1.07146358e+00 -2.29259580e-01 1.17524481e+00 -2.35933930e-01
5.80559254e-01 -6.10311687e-01 6.51105940e-02 1.18068069e-01
8.85219336e-01 -1.14856565e+00 -1.75462186e-01 -4.21079993e-01
-7.21109867e-01 1.04071367e+00 2.94362545e-01 1.45352557e-01
6.32221520e-01 2.42296606e-01 9.88441184e-02 -3.56674492e-01
-2.89726555e-01 1.46648422e-01 4.47168231e-01 4.94529635e-01
5.25988400e-01 9.56887081e-02 -6.96558058e-01 4.43843842e-01
1.28854677e-01 1.48935527e-01 2.33455673e-01 5.81357479e-01
-1.71479266e-02 -1.01204026e+00 -6.37238145e-01 3.09113920e-01
-3.03604007e-01 -1.30996257e-02 4.89860862e-01 5.53509712e-01
1.54605567e-01 5.59605241e-01 -1.81585729e-01 -2.15172499e-01
4.75204319e-01 2.93867197e-02 7.29234636e-01 -1.73724443e-01
-3.29689056e-01 3.88260454e-01 -3.36832017e-01 -6.84980392e-01
-6.36749685e-01 -5.31668186e-01 -1.30692792e+00 -3.35971177e-01
-1.39510214e-01 1.81349933e-01 5.19163132e-01 1.04429984e+00
2.02522823e-03 7.95647681e-01 7.07475185e-01 -8.80188704e-01
-4.58064467e-01 -1.02435613e+00 -9.44038391e-01 8.26847970e-01
2.38313362e-01 -8.66376042e-01 -2.28798091e-01 -1.46877533e-02]
|
[13.560517311096191, -2.37461256980896]
|
926126e7-bcce-4a74-8f0d-cd6ab51d5863
|
dynamic-3d-gaze-from-afar-deep-gaze
| null | null |
http://openaccess.thecvf.com//content/CVPR2022/html/Nonaka_Dynamic_3D_Gaze_From_Afar_Deep_Gaze_Estimation_From_Temporal_CVPR_2022_paper.html
|
http://openaccess.thecvf.com//content/CVPR2022/papers/Nonaka_Dynamic_3D_Gaze_From_Afar_Deep_Gaze_Estimation_From_Temporal_CVPR_2022_paper.pdf
|
Dynamic 3D Gaze From Afar: Deep Gaze Estimation From Temporal Eye-Head-Body Coordination
|
We introduce a novel method and dataset for 3D gaze estimation of a freely moving person from a distance, typically in surveillance views. Eyes cannot be clearly seen in such cases due to occlusion and lacking resolution. Existing gaze estimation methods suffer or fall back to approximating gaze with head pose as they primarily rely on clear, close-up views of the eyes. Our key idea is to instead leverage the intrinsic gaze, head, and body coordination of people. Our method formulates gaze estimation as Bayesian prediction given temporal estimates of head and body orientations which can be reliably estimated from a far. We model the head and body orientation likelihoods and the conditional prior of gaze direction on those with separate neural networks which are then cascaded to output the 3D gaze direction. We introduce an extensive new dataset that consists of surveillance videos annotated with 3D gaze directions captured in 5 indoor and outdoor scenes. Experimental results on this and other datasets validate the accuracy of our method and demonstrate that gaze can be accurately estimated from a typical surveillance distance even when the person's face is not visible to the camera.
|
['Ko Nishino', 'Shohei Nobuhara', 'Soma Nonaka']
|
2022-01-01
| null | null | null |
cvpr-2022-1
|
['gaze-estimation']
|
['computer-vision']
|
[ 1.24451621e-02 2.07978189e-01 8.75747704e-04 -8.05122674e-01
-8.60632583e-02 -4.49188977e-01 4.12960708e-01 -4.88737255e-01
-3.28630120e-01 3.56568635e-01 3.04413885e-01 5.20368526e-03
2.55699366e-01 1.17393106e-01 -6.61815286e-01 -6.18096411e-01
-5.04827015e-02 -5.40301539e-02 9.45878625e-02 2.08380312e-01
2.95176834e-01 4.93321009e-02 -1.84056079e+00 -4.80700731e-01
4.52312291e-01 9.80994821e-01 -1.51522015e-03 1.18639588e+00
6.83585525e-01 6.80012286e-01 -5.63459456e-01 -2.25232959e-01
2.08904296e-01 -2.29165554e-01 -3.48083228e-01 5.09062469e-01
1.09595788e+00 -1.06187999e+00 -5.39642870e-02 8.45104218e-01
3.05549741e-01 3.09296343e-02 5.63900828e-01 -1.45195127e+00
-2.57642031e-01 -4.43859756e-01 -9.81826723e-01 2.78625816e-01
1.14030755e+00 3.62766594e-01 5.65376103e-01 -4.68431354e-01
4.42381501e-01 1.21338034e+00 6.74319923e-01 9.41035688e-01
-8.79807591e-01 -6.05408847e-01 6.45637631e-01 1.28913388e-01
-1.38464689e+00 -8.03681612e-01 6.49894416e-01 -6.37825966e-01
7.02144206e-01 9.03444514e-02 7.71212161e-01 1.33096683e+00
3.24363597e-02 7.39077389e-01 9.96232808e-01 -3.20076197e-01
-1.04479283e-01 9.39842090e-02 3.96336794e-01 6.52720988e-01
2.12782085e-01 1.80293903e-01 -9.38685000e-01 6.25035539e-02
4.57242966e-01 3.04244041e-01 -7.98629403e-01 -5.88708997e-01
-1.23304772e+00 3.27976376e-01 5.22796392e-01 -4.75228369e-01
-3.30433995e-01 -1.22336321e-01 -4.16155905e-01 -6.02790639e-02
6.95302486e-01 -1.85784131e-01 -2.54635036e-01 -2.38030642e-01
-9.38398540e-01 2.91536719e-01 7.48316765e-01 1.12758899e+00
5.93089044e-01 -5.07610023e-01 -1.68098971e-01 2.01243222e-01
9.51970935e-01 1.10790515e+00 -1.52788326e-01 -9.48671103e-01
2.48421595e-01 5.86032987e-01 6.15844667e-01 -8.37332845e-01
-5.93937695e-01 1.77538365e-01 -2.05510616e-01 3.70992869e-01
7.79466033e-01 -5.05246937e-01 -8.46864760e-01 1.95640230e+00
8.36399496e-01 1.54591620e-01 -3.32037300e-01 1.35470319e+00
7.60271609e-01 1.89843670e-01 -1.30626649e-01 -3.19902331e-01
1.55387664e+00 -7.40761518e-01 -9.29880381e-01 -6.85065210e-01
2.66597271e-01 -3.65305960e-01 5.91697156e-01 4.07220751e-01
-9.90064204e-01 -4.05949056e-01 -9.15310204e-01 -9.33837965e-02
-8.96691009e-02 1.11238346e-01 2.43071243e-01 8.61890674e-01
-1.32761717e+00 -1.50164008e-01 -1.17053795e+00 -8.06111395e-01
2.62360036e-01 4.69177812e-01 -3.64128560e-01 2.84544025e-02
-7.12525249e-01 9.43208635e-01 -2.41692886e-01 4.77640331e-01
-7.44727969e-01 -2.38614410e-01 -1.11103523e+00 -1.90703467e-01
4.01381105e-01 -8.34332466e-01 1.37753952e+00 -9.85880435e-01
-1.57829511e+00 1.04433632e+00 -9.96876776e-01 -1.32309452e-01
3.10270786e-01 -6.33054674e-01 -1.93706185e-01 2.34920725e-01
-1.62785649e-01 9.94408727e-01 1.20052958e+00 -9.89433348e-01
-6.61397755e-01 -9.09331501e-01 2.26151213e-01 5.98233402e-01
-2.02330664e-01 2.81221509e-01 -5.10979533e-01 9.84770879e-02
-6.23593144e-02 -1.20599592e+00 3.52964550e-01 4.85139638e-01
-5.63425899e-01 -1.48721680e-01 9.63909507e-01 -6.35014534e-01
1.04580474e+00 -1.91338396e+00 1.85317144e-01 -9.22236294e-02
6.47733748e-01 -6.75607845e-02 4.98648077e-01 -1.57593384e-01
5.81124201e-02 -3.25465232e-01 2.42731303e-01 -7.54308224e-01
-9.40536708e-03 -1.95869341e-01 -3.02786063e-02 9.03071940e-01
6.21584542e-02 7.86128998e-01 -8.31157625e-01 -4.76596206e-01
1.66695744e-01 7.96394527e-01 -4.90341544e-01 6.00790977e-01
7.99661875e-02 7.41111815e-01 -2.33615190e-01 6.33366466e-01
7.13286281e-01 -6.18627846e-01 7.29171261e-02 -2.31670514e-02
-7.44436532e-02 2.05730259e-01 -7.60161102e-01 1.52572811e+00
1.93547398e-01 1.04460549e+00 2.34146506e-01 -1.93728268e-01
5.39102077e-01 2.31952444e-01 -8.89625698e-02 -3.11993986e-01
3.02571177e-01 -4.47934687e-01 -1.49915218e-01 -8.14145982e-01
3.61169696e-01 2.19915718e-01 9.51014683e-02 6.70050144e-01
-6.49437383e-02 3.22162896e-01 6.06798707e-03 -1.08846828e-01
7.17929840e-01 5.37388027e-01 2.45013237e-01 -5.40035628e-02
4.27038938e-01 -5.47126234e-01 3.38833809e-01 4.06853467e-01
-6.59421146e-01 6.69257224e-01 5.34032762e-01 -4.63584185e-01
-5.63867331e-01 -9.86860454e-01 8.73287916e-02 1.34312391e+00
3.09681267e-01 -2.39751220e-01 -1.17940247e+00 -8.74306381e-01
-2.95488536e-01 3.62841338e-01 -9.95595336e-01 2.52103359e-01
-4.17577833e-01 -3.50026459e-01 -5.90486936e-02 2.76967734e-01
1.76952139e-01 -6.04175508e-01 -1.23813784e+00 -7.63167500e-01
-1.90543279e-01 -1.10856760e+00 -8.08798492e-01 -3.79808009e-01
-4.77774173e-01 -1.48185766e+00 -1.04998958e+00 -3.48824084e-01
1.13449717e+00 6.25140071e-01 8.63826394e-01 7.29310326e-03
3.20320725e-01 8.45098555e-01 9.28347651e-03 -7.17990041e-01
2.30968744e-01 -1.57809496e-01 5.91118753e-01 2.85104543e-01
9.43162978e-01 -1.44444332e-01 -8.97323847e-01 4.89966601e-01
-5.18877655e-02 6.23422265e-02 8.40068907e-02 3.58304977e-01
-5.33443205e-02 -7.23818600e-01 -1.20488569e-01 -4.28268492e-01
2.35720351e-01 -3.86283904e-01 -8.78134310e-01 3.09157073e-01
-2.87241459e-01 -2.88495868e-01 -2.86760688e-01 -2.36545265e-01
-1.20894277e+00 -4.52679843e-02 4.23876226e-01 -5.40688038e-01
-6.27199292e-01 -1.62366912e-01 7.79958814e-03 7.11299106e-02
5.44616222e-01 -2.27349311e-01 3.15400481e-01 -1.24973260e-01
-8.63766372e-02 8.33483338e-01 4.95422244e-01 -1.99859822e-03
5.17325342e-01 7.24358499e-01 -1.14059364e-02 -1.00474298e+00
-1.23244011e+00 -4.22280848e-01 -9.94767547e-01 -6.55092716e-01
1.08309877e+00 -1.12606907e+00 -1.56444371e+00 7.92669773e-01
-9.90818024e-01 -2.09958613e-01 5.29193640e-01 6.90515339e-01
-2.39685968e-01 3.96703303e-01 -7.29344338e-02 -1.26443613e+00
-3.10852826e-01 -1.03072560e+00 1.62991726e+00 7.52739429e-01
-4.97020394e-01 -1.08636272e+00 -1.32382112e-02 3.73890042e-01
1.81432635e-01 3.53973031e-01 -1.13603286e-01 -9.30183902e-02
-6.14988506e-01 -2.25991800e-01 -7.65966624e-02 -3.16047259e-02
1.12252831e-01 6.85418919e-02 -1.35200608e+00 -4.19308722e-01
1.45432716e-02 -2.95899659e-01 4.29904282e-01 1.01705885e+00
4.83553559e-01 -2.93609828e-01 -6.89052582e-01 8.26320648e-01
6.63152516e-01 -2.33897835e-01 2.21757721e-02 4.39584954e-03
7.84497917e-01 1.00317383e+00 5.57092130e-01 4.43637431e-01
8.10980082e-01 7.35028684e-01 5.54433644e-01 2.15344876e-01
1.09453432e-01 -4.94391806e-02 5.18079877e-01 2.48202980e-01
-2.34106287e-01 -5.15403748e-01 -9.45500970e-01 3.13957453e-01
-1.57605267e+00 -8.52788210e-01 -1.47120014e-01 2.40121722e+00
4.72802609e-01 7.79754743e-02 4.65612441e-01 -2.31442809e-01
8.55124474e-01 5.45533486e-02 -7.22179294e-01 1.04386367e-01
3.16600055e-01 -6.80273354e-01 2.92226940e-01 5.25948405e-01
-1.16403437e+00 4.62994695e-01 6.80240583e+00 -6.39659882e-01
-1.30814934e+00 -2.04434946e-01 3.90197247e-01 -8.20297897e-01
2.06952602e-01 -3.09808552e-01 -1.34407210e+00 6.55555427e-01
9.26419139e-01 3.64403099e-01 2.86781728e-01 4.41463530e-01
2.77886420e-01 -5.85434794e-01 -1.45853996e+00 1.23967695e+00
5.81406355e-01 -4.71847206e-01 -8.38023424e-01 3.43008608e-01
3.85871321e-01 1.54617131e-01 3.17030489e-01 -6.51261359e-02
3.03125773e-02 -9.26269054e-01 7.22898304e-01 1.01365793e+00
9.28909659e-01 -3.53475004e-01 4.29967701e-01 6.05124950e-01
-5.66845536e-01 -1.01010248e-01 1.26570866e-01 -4.93681699e-01
4.38113660e-01 -1.64463565e-01 -8.94593298e-01 -2.78472692e-01
1.05052936e+00 1.09108305e+00 -7.77610481e-01 8.38157952e-01
-5.37955642e-01 3.11698794e-01 -5.62857985e-01 -7.45707229e-02
-1.20905831e-01 4.37165983e-02 6.87672377e-01 7.70111918e-01
2.34773681e-01 2.10540146e-01 -2.51582533e-01 8.20066273e-01
1.65634483e-01 -6.97637022e-01 -6.70313418e-01 5.13501644e-01
2.93384433e-01 1.12750101e+00 -3.31156582e-01 2.68527661e-02
-5.20918071e-01 8.04248273e-01 2.13964060e-01 7.67192245e-01
-7.04155207e-01 -2.04781350e-02 8.86174083e-01 2.21190631e-01
5.78671217e-01 -1.49252191e-01 1.99959204e-01 -1.33519804e+00
2.92114228e-01 -4.43296462e-01 1.77456528e-01 -1.32130158e+00
-8.12678874e-01 5.60275733e-01 3.84792566e-01 -1.28584182e+00
-8.23203683e-01 -6.54125392e-01 -5.62169194e-01 1.09222901e+00
-1.58918190e+00 -1.27060819e+00 -9.04125869e-01 5.60527802e-01
2.74407208e-01 2.01324925e-01 7.49550700e-01 -2.42397785e-01
-8.04467916e-01 6.26162648e-01 -3.09665233e-01 6.65423200e-02
1.09827936e+00 -1.25606978e+00 3.76955509e-01 9.94457483e-01
-3.03207129e-01 8.61648858e-01 8.36355805e-01 -2.93237031e-01
-1.28582895e+00 -4.63578254e-01 9.99658763e-01 -1.38693285e+00
1.05409078e-01 -7.64990747e-01 -6.44185841e-01 1.01206708e+00
4.81254846e-01 3.72376919e-01 6.97002292e-01 4.00448352e-01
-1.70887932e-01 -4.42393087e-02 -9.77611184e-01 4.21441495e-01
9.02403474e-01 -3.77787918e-01 -7.38414824e-01 -1.77802257e-02
2.79852748e-01 -7.92140782e-01 -4.14615095e-01 6.70030937e-02
1.04755688e+00 -1.32054627e+00 8.15684021e-01 -4.15780336e-01
-7.63049796e-02 -4.58902448e-01 3.16740960e-01 -9.87103105e-01
6.38737381e-02 -8.00015986e-01 -6.26234710e-01 8.59965384e-01
-5.71859851e-02 -4.68400806e-01 7.08577752e-01 1.08284450e+00
4.54301625e-01 -3.18829447e-01 -6.91705167e-01 -6.54657558e-02
-6.85199499e-01 -3.29176098e-01 4.92394745e-01 4.87052053e-01
1.79625556e-01 5.05414784e-01 -6.90560341e-01 4.89510179e-01
9.70874608e-01 -4.11176905e-02 1.27284312e+00 -1.40161753e+00
7.58293644e-02 -8.73771132e-05 -4.29258704e-01 -1.66659498e+00
1.39315173e-01 1.41926274e-01 1.92322642e-01 -1.03292561e+00
2.23124981e-01 2.62097508e-01 2.84169931e-02 1.42130747e-01
-4.51086730e-01 2.49519676e-01 2.52266396e-02 2.85710782e-01
-9.46973264e-01 1.89165488e-01 1.04693043e+00 2.48012021e-01
-1.18275076e-01 3.80541861e-01 -6.40522003e-01 9.90531266e-01
2.72183269e-01 -2.00215295e-01 -4.76638526e-01 -7.45488584e-01
4.07646984e-01 1.72633931e-01 7.46075332e-01 -8.29643846e-01
6.35899305e-01 4.46756557e-02 7.97801495e-01 -8.51445675e-01
6.59756899e-01 -7.59534299e-01 -2.05993876e-01 -8.61937180e-02
-1.96632385e-01 1.00912852e-02 4.92965356e-02 8.35695863e-01
2.50669926e-01 1.42854616e-01 4.43742365e-01 1.37751549e-01
-4.43641126e-01 2.96317309e-01 -1.14765607e-01 -2.91980803e-02
1.06032598e+00 -6.40407979e-01 -3.16245496e-01 -7.78768539e-01
-7.28910506e-01 4.55601454e-01 9.67736304e-01 5.32447457e-01
5.00137091e-01 -8.79045606e-01 -4.70899314e-01 7.44260311e-01
1.43384874e-01 1.41213998e-01 2.20580801e-01 1.29058337e+00
-1.91113308e-01 7.28265703e-01 -1.01700656e-01 -1.40805972e+00
-1.69633555e+00 7.00443625e-01 5.47192037e-01 5.21664619e-01
-3.02417219e-01 1.05663967e+00 6.39705539e-01 -8.40836987e-02
4.86792356e-01 -4.84687388e-01 -5.30793548e-01 -8.02105069e-02
1.01129293e+00 2.20879868e-01 -4.62088108e-01 -1.24156260e+00
-5.36144376e-01 9.53894615e-01 -6.63495809e-02 2.73240004e-02
1.04671574e+00 -9.35336769e-01 2.01616347e-01 4.42962527e-01
9.29272711e-01 3.82733308e-02 -2.10354853e+00 -1.99798167e-01
-5.34872055e-01 -6.99263215e-01 -1.80158224e-02 -4.76124048e-01
-8.58313441e-01 1.09854901e+00 7.95677483e-01 1.91515818e-01
1.09856248e+00 7.83759356e-02 4.52141851e-01 3.51609886e-01
1.32189408e-01 -6.57464147e-01 -2.54751086e-01 2.69566000e-01
4.30943877e-01 -1.82994425e+00 8.50450993e-02 -1.29779384e-01
-5.59909999e-01 9.93079603e-01 8.09895992e-01 1.84109122e-01
8.29654634e-01 -1.20302290e-01 2.97864020e-01 -3.92791480e-01
-8.52292895e-01 -2.12899759e-01 7.71880984e-01 8.97678137e-01
3.58176798e-01 -2.43594244e-01 6.51075661e-01 5.64730503e-02
-1.59535855e-01 -4.16656584e-02 3.61832082e-01 9.20583546e-01
-1.88578799e-01 -3.41855168e-01 -6.64939284e-01 1.65332630e-01
-6.52622342e-01 3.14130425e-01 -3.48847002e-01 6.08626544e-01
-4.13915738e-02 1.16644669e+00 4.28065807e-01 -4.83595543e-02
7.84403905e-02 1.09878741e-01 6.83087885e-01 -4.63827729e-01
1.04206149e-02 9.37865749e-02 -1.95228994e-01 -8.08178008e-01
-1.06354439e+00 -9.24052000e-01 -5.01750767e-01 -1.19072594e-01
-3.80175889e-01 -3.29125196e-01 5.89131057e-01 9.23951685e-01
3.85219455e-01 1.12308592e-01 4.11549479e-01 -1.38066554e+00
-2.18766898e-01 -1.13973916e+00 -4.77989376e-01 3.22516590e-01
1.23686171e+00 -9.44413304e-01 -5.55101812e-01 5.69716275e-01]
|
[14.119274139404297, 0.08621423691511154]
|
1b086bff-8f99-44e1-b180-6b3d9c644790
|
decor-defy-knowledge-forgetting-by-predicting
|
2305.18441
| null |
https://arxiv.org/abs/2305.18441v1
|
https://arxiv.org/pdf/2305.18441v1.pdf
|
DeCoR: Defy Knowledge Forgetting by Predicting Earlier Audio Codes
|
Lifelong audio feature extraction involves learning new sound classes incrementally, which is essential for adapting to new data distributions over time. However, optimizing the model only on new data can lead to catastrophic forgetting of previously learned tasks, which undermines the model's ability to perform well over the long term. This paper introduces a new approach to continual audio representation learning called DeCoR. Unlike other methods that store previous data, features, or models, DeCoR indirectly distills knowledge from an earlier model to the latest by predicting quantization indices from a delayed codebook. We demonstrate that DeCoR improves acoustic scene classification accuracy and integrates well with continual self-supervised representation learning. Our approach introduces minimal storage and computation overhead, making it a lightweight and efficient solution for continual learning.
|
['Nima Mesgarani', 'Yinghao Aaron Li', 'Xilin Jiang']
|
2023-05-29
| null | null | null | null |
['acoustic-scene-classification', 'scene-classification']
|
['audio', 'computer-vision']
|
[ 1.99176744e-01 -3.33874524e-01 9.77008417e-03 -5.09672999e-01
-1.06240082e+00 -4.85021144e-01 2.97083735e-01 5.11861205e-01
-5.03075898e-01 9.96593833e-01 2.15759918e-01 2.76025273e-02
-1.18581541e-01 -6.65269673e-01 -7.39161789e-01 -5.79823792e-01
-3.10592502e-01 2.33675599e-01 4.97338176e-01 6.55860826e-02
1.93963915e-01 4.91935879e-01 -2.22471452e+00 5.01969457e-01
3.42205912e-01 1.02330089e+00 4.63110447e-01 1.03003502e+00
-3.90439361e-01 9.51346338e-01 -7.34067082e-01 5.69608025e-02
-7.66775832e-02 -3.22784573e-01 -8.25231075e-01 -1.92517355e-01
3.64068180e-01 -1.85022742e-01 -3.82844388e-01 5.97800612e-01
5.38160741e-01 3.62236381e-01 4.71492887e-01 -1.21571481e+00
-5.51681936e-01 6.62245393e-01 -1.50168344e-01 4.37468112e-01
2.97254205e-01 -3.86808068e-03 7.06334710e-01 -1.02703607e+00
2.93368578e-01 8.95761013e-01 1.09947002e+00 4.45407689e-01
-1.13040233e+00 -7.77938426e-01 3.14363986e-01 4.27592129e-01
-1.59650517e+00 -1.01922500e+00 7.66574800e-01 -3.06556106e-01
1.03371322e+00 1.70050859e-01 9.80287790e-01 6.03961825e-01
1.07140943e-01 7.21592665e-01 8.25879872e-01 -6.00451291e-01
5.71324050e-01 2.58254141e-01 8.75761807e-02 5.72383285e-01
7.65646622e-02 2.07002252e-01 -1.27231407e+00 -3.12944680e-01
1.99543580e-01 -5.87469600e-02 3.87824886e-02 -3.93148422e-01
-5.32093406e-01 3.51575017e-01 2.06517816e-01 3.31324339e-01
-3.53480369e-01 4.58067417e-01 3.99697125e-01 8.89961302e-01
4.13485020e-01 3.49888593e-01 -6.78983927e-01 -7.46711493e-01
-1.00680590e+00 -5.42368479e-02 5.24126291e-01 7.71325767e-01
1.19553578e+00 4.89450186e-01 2.68973172e-01 8.50654840e-01
-2.76767295e-02 4.91918355e-01 9.39526856e-01 -8.81093860e-01
-1.61982566e-01 -3.34676541e-02 -6.18209764e-02 -7.13629484e-01
-1.86299965e-01 -7.48415112e-01 -3.43949705e-01 4.11824398e-02
-6.11005649e-02 -9.94630530e-02 -7.91481912e-01 1.84536922e+00
2.86275387e-01 4.60871726e-01 6.76901639e-02 1.83847938e-02
2.85041660e-01 7.73053169e-01 -1.20353391e-02 -6.28254235e-01
4.79199022e-01 -7.46171534e-01 -5.72297752e-01 -2.23122776e-01
5.65402687e-01 -7.00599670e-01 1.22708547e+00 8.32414806e-01
-9.73284900e-01 -8.46507132e-01 -1.10982680e+00 1.82045296e-01
-2.26720974e-01 -5.41701376e-01 9.54094887e-01 7.45092332e-01
-1.28841567e+00 7.02312231e-01 -1.02626646e+00 -2.81308759e-02
4.95310932e-01 2.78705269e-01 -1.10007524e-02 -7.94993639e-02
-9.86176014e-01 4.63541865e-01 3.59525084e-01 -4.19083834e-01
-1.11226928e+00 -9.34405208e-01 -5.09985626e-01 -5.27407229e-02
1.20225079e-01 -3.37520450e-01 1.69612372e+00 -9.04205322e-01
-1.46727908e+00 1.79786514e-02 -5.47377229e-01 -6.95432305e-01
-1.28478697e-03 -4.65182394e-01 -5.60793698e-01 -1.02656566e-01
-7.66493753e-02 7.46831596e-01 1.30751908e+00 -1.18131769e+00
-9.22785997e-01 -1.35903537e-01 -3.02500427e-01 2.95035720e-01
-1.04306340e+00 -6.66749835e-01 -2.88534522e-01 -7.39269972e-01
1.70122176e-01 -6.33975625e-01 1.37874037e-02 1.06544688e-01
3.78188163e-01 -1.33090854e-01 9.03867245e-01 -3.80573779e-01
1.62511122e+00 -2.65707493e+00 -2.99203336e-01 1.21430330e-01
-2.12739468e-01 1.38320494e-02 -1.84232473e-01 3.73150676e-01
-4.15070727e-02 -1.45865887e-01 7.61798471e-02 -5.80159783e-01
-2.88324475e-01 4.74618345e-01 -7.04710424e-01 4.16027009e-02
-1.36269227e-01 6.68219507e-01 -1.06132889e+00 -3.78739476e-01
-1.00123353e-01 4.55831081e-01 -7.49575794e-01 2.70211566e-02
-1.91934004e-01 2.24674150e-01 4.84218188e-02 5.03470063e-01
2.09940419e-01 -1.27420664e-01 -1.38992444e-01 1.77290127e-01
-8.46954584e-02 4.84700322e-01 -1.24944460e+00 1.98931468e+00
-8.36970448e-01 6.48516417e-01 -3.60827893e-01 -9.09515977e-01
8.93058360e-01 2.75387585e-01 6.70677722e-01 -7.45161414e-01
-5.42495251e-01 1.97653547e-01 -4.81374085e-01 -1.90824375e-01
6.39945447e-01 -3.04171026e-01 -1.70746461e-01 7.17428386e-01
2.76023954e-01 -4.04192686e-01 -1.10099791e-02 1.12860277e-01
1.11018181e+00 -1.93318367e-01 2.46216759e-01 2.02114224e-01
6.35710284e-02 -9.10110548e-02 6.83037937e-01 9.97773111e-01
3.85680124e-02 2.04388440e-01 -1.71989575e-01 -5.25428057e-01
-8.36045980e-01 -1.25059128e+00 -8.58277157e-02 1.46717012e+00
-3.71010035e-01 -7.05205142e-01 -2.46441990e-01 -5.48315048e-01
1.80385157e-01 7.68729627e-01 -2.63321996e-01 -6.73009396e-01
-4.20525491e-01 -2.90750623e-01 4.63152230e-01 5.57169080e-01
2.33715668e-01 -8.91718805e-01 -6.84962392e-01 6.51361406e-01
1.46867381e-02 -4.42002863e-01 -3.63736629e-01 6.47730350e-01
-1.10257137e+00 -5.95547020e-01 -2.77823836e-01 -7.50714421e-01
2.41104424e-01 4.81524646e-01 9.29459691e-01 -2.09320597e-02
-3.86321634e-01 8.32529366e-01 -5.00711381e-01 -6.29235983e-01
-4.46930915e-01 1.65524587e-01 6.63885891e-01 -1.66561663e-01
1.60743684e-01 -1.04340148e+00 -2.06696570e-01 -1.66078702e-01
-7.91624904e-01 -1.79596245e-01 3.68965685e-01 9.00695384e-01
8.90230417e-01 5.91261625e-01 1.08910823e+00 -6.15640283e-01
5.01747608e-01 -3.82257193e-01 -3.03893209e-01 1.89937979e-01
-1.12975955e+00 1.69949710e-01 5.65443158e-01 -6.88618124e-01
-1.01261532e+00 3.90079975e-01 -1.32689133e-01 -3.09292763e-01
6.34960532e-02 5.27864695e-01 1.63111702e-01 -3.01208552e-02
9.01253760e-01 5.66403806e-01 -1.01347476e-01 -7.46809423e-01
4.80310172e-01 8.40649188e-01 7.43013322e-01 -5.37587881e-01
6.73391342e-01 3.11907530e-01 -3.91387939e-01 -9.24535334e-01
-1.01556122e+00 -2.82734990e-01 -7.42818415e-01 -2.18955114e-01
-4.61717360e-02 -9.66247380e-01 -2.29846135e-01 4.40703303e-01
-8.21594059e-01 -4.47552830e-01 -1.20746779e+00 4.22856718e-01
-6.01958036e-01 2.35420614e-01 -2.71005690e-01 -9.12613213e-01
-1.36966452e-01 -3.47794712e-01 6.73732638e-01 2.32359045e-03
-4.69342530e-01 -8.38386774e-01 3.06840748e-01 -3.02748412e-01
5.33484161e-01 -2.98061162e-01 8.24596286e-01 -2.11091265e-01
-3.91996413e-01 -2.43580431e-01 4.48879451e-01 5.93972504e-01
3.98690194e-01 -4.44145948e-01 -1.39694071e+00 -6.49242699e-01
-1.73800532e-02 -6.85939550e-01 1.08840346e+00 -7.11046383e-02
1.54401243e+00 -5.09913862e-01 -2.06844032e-01 4.30962831e-01
1.09912705e+00 5.58114111e-01 3.07292730e-01 6.94553256e-02
3.05358112e-01 5.63347004e-02 7.40863264e-01 9.16661024e-01
3.11108589e-01 3.37800980e-01 -3.96182463e-02 3.27592045e-01
-6.31197453e-01 -5.85316896e-01 6.07342541e-01 1.38781762e+00
4.75048184e-01 1.17566347e-01 -7.54227638e-01 7.52900183e-01
-1.42662394e+00 -8.82222235e-01 7.69268572e-01 2.40621495e+00
1.30882919e+00 2.89396048e-01 -1.30626366e-01 6.80748701e-01
2.31946394e-01 -1.15855433e-01 -8.76684666e-01 -3.51953030e-01
-3.86148505e-02 5.74677408e-01 2.91266833e-02 5.73816478e-01
-9.09985542e-01 9.79387641e-01 7.79387140e+00 9.36571479e-01
-1.23920131e+00 3.66550297e-01 2.94105113e-01 -4.23746467e-01
-4.10345614e-01 7.76653141e-02 -8.10612321e-01 2.84645379e-01
1.25708008e+00 -5.78943133e-01 5.76296628e-01 8.80975664e-01
-2.26540372e-01 -1.65417865e-01 -1.19709051e+00 1.23593342e+00
2.28474870e-01 -1.40717518e+00 2.95556039e-01 -3.45308125e-01
7.75343657e-01 1.02018258e-02 3.87088597e-01 4.90443558e-01
3.50226909e-01 -6.14440620e-01 8.37918162e-01 9.53345716e-01
1.00828934e+00 -9.62680757e-01 2.29676798e-01 4.37147856e-01
-1.40064740e+00 -5.51653683e-01 -4.50683385e-01 -2.40298912e-01
1.19985692e-01 7.53159702e-01 -1.06892717e+00 2.01941412e-02
9.61320460e-01 7.20530272e-01 -9.77541804e-01 1.14413655e+00
5.09125255e-02 8.87567163e-01 -3.91439527e-01 6.89703822e-02
-2.04795167e-01 7.20000505e-01 4.64696079e-01 1.11874843e+00
6.83473587e-01 -2.91884512e-01 2.54114717e-01 2.30840407e-03
8.07727203e-02 -4.38902469e-04 -6.84691608e-01 -4.84430231e-02
1.01824749e+00 5.71382105e-01 -3.01449746e-01 -5.74338675e-01
-6.31036386e-02 1.01866901e+00 3.75614971e-01 1.56430379e-01
-3.40614676e-01 -4.57141668e-01 5.16147256e-01 1.54798701e-01
4.23584491e-01 -4.49915349e-01 -1.24724023e-01 -7.70189226e-01
-7.38331378e-02 -6.21111155e-01 5.82692862e-01 -7.37005174e-01
-1.16921568e+00 4.74921823e-01 -1.93361312e-01 -1.31389511e+00
-6.22796893e-01 1.03223801e-01 -1.72088221e-01 2.45710537e-01
-1.48003781e+00 -7.83708692e-01 -7.34673440e-02 8.03011119e-01
7.29852378e-01 -4.95154709e-01 1.22044575e+00 3.67996275e-01
1.51541054e-01 8.56181681e-01 3.24879318e-01 -5.38139939e-01
9.15195584e-01 -1.07311058e+00 2.42643297e-01 4.83033806e-01
8.18196058e-01 4.62518692e-01 5.96763313e-01 -5.58072269e-01
-1.26307821e+00 -1.24266434e+00 9.79230642e-01 -2.87698805e-01
6.12476766e-01 -1.89056426e-01 -1.08115041e+00 3.87730867e-01
-2.34056264e-01 -2.02041231e-02 1.06801391e+00 3.42898041e-01
-4.94244933e-01 -8.22883368e-01 -8.29902768e-01 2.09854648e-01
1.10405040e+00 -8.96305680e-01 -4.79778558e-01 1.92235366e-01
9.60971177e-01 -1.84676684e-02 -7.41233706e-01 2.59394288e-01
6.76281154e-01 -6.14193320e-01 8.68082762e-01 -5.33609569e-01
-4.12487239e-01 -2.46090114e-01 -4.13042128e-01 -1.32394195e+00
-3.71319681e-01 -8.66001427e-01 -7.60352552e-01 1.28202116e+00
4.11269277e-01 -4.51280177e-01 7.08281457e-01 7.99955055e-02
-2.73675084e-01 -4.80488539e-01 -1.18047690e+00 -1.13559163e+00
-4.97447699e-02 -1.06030488e+00 9.05150175e-01 7.43507326e-01
-9.79583263e-02 2.51378894e-01 -5.49850523e-01 1.79259218e-02
5.57928920e-01 1.27054870e-01 7.35779047e-01 -1.41315913e+00
-5.58454156e-01 -3.76143567e-02 -5.31123638e-01 -1.27153599e+00
-6.80321157e-02 -9.95094180e-01 1.42168716e-01 -1.05141354e+00
-7.54088014e-02 -8.80468547e-01 -9.17539656e-01 8.91488731e-01
2.11420223e-01 4.40219820e-01 -7.08197616e-03 3.00995082e-01
-8.68353784e-01 7.32829928e-01 7.78960466e-01 -2.37932757e-01
-6.00393236e-01 1.64202332e-01 -8.15901697e-01 6.90720797e-01
8.37919772e-01 -7.76522934e-01 -8.21426213e-01 -4.37021464e-01
3.96978080e-01 -2.14984968e-01 7.97013417e-02 -1.63122702e+00
6.08508706e-01 1.17374295e-02 5.86067140e-01 -6.56917036e-01
6.78348005e-01 -6.21942461e-01 2.05479294e-01 3.94090801e-01
-4.48007315e-01 8.10465813e-02 3.47143024e-01 9.12407279e-01
-3.24598879e-01 -2.70581782e-01 6.70248330e-01 -1.55398045e-02
-1.00745595e+00 2.89498210e-01 -6.40576661e-01 -2.85925679e-02
6.71967506e-01 -2.67544806e-01 2.46979058e-01 -6.25217855e-01
-1.35297048e+00 -1.02555752e-01 2.70627141e-01 6.33796036e-01
9.08980548e-01 -1.40197444e+00 -2.38362372e-01 4.67824191e-01
1.07618354e-01 -2.11423025e-01 1.90803811e-01 1.12715274e-01
1.01203181e-01 1.75481930e-01 9.12542120e-02 -6.76936746e-01
-1.33339691e+00 5.89128077e-01 2.07431540e-02 8.56652409e-02
-5.52220702e-01 1.05750382e+00 -4.05578882e-01 -1.27698347e-01
6.11873627e-01 -1.65376663e-02 -1.25249876e-02 2.44528472e-01
7.57295609e-01 3.33305597e-01 1.42205000e-01 -1.66809335e-01
-7.89044276e-02 3.96559298e-01 -2.91457504e-01 -5.32771587e-01
1.51126182e+00 -5.52210093e-01 7.35653564e-02 1.48961759e+00
1.08708251e+00 -8.98553208e-02 -1.42402542e+00 -6.73983693e-01
6.69002756e-02 -5.94754219e-01 1.96072221e-01 -6.64987385e-01
-6.69779301e-01 9.44981754e-01 9.42448080e-01 -9.95804816e-02
1.15771568e+00 -3.59252952e-02 9.45766032e-01 1.00010550e+00
8.33921671e-01 -1.47697759e+00 7.80335367e-01 6.93615496e-01
8.01667452e-01 -6.28130615e-01 1.90223828e-02 2.64155328e-01
-4.13431227e-01 1.01374674e+00 4.42294300e-01 3.33305985e-01
1.23593867e+00 5.25287449e-01 -4.09924537e-02 2.32526422e-01
-1.17453706e+00 -3.18017267e-02 3.09874937e-02 8.79143119e-01
9.26661342e-02 -1.46353438e-01 3.95593941e-01 4.54239190e-01
-6.96689904e-01 1.90106869e-01 3.27583402e-01 1.39939141e+00
-9.73381400e-01 -1.28737867e+00 -9.40795913e-02 5.72788537e-01
1.96894426e-02 -1.11719362e-01 -3.37520428e-02 2.79203922e-01
4.82907414e-01 9.55066323e-01 2.75957793e-01 -5.32708168e-01
1.09531723e-01 5.84820390e-01 4.78259981e-01 -9.09255266e-01
-1.10277645e-01 1.32416606e-01 -1.76922366e-01 -2.44964898e-01
-1.63098082e-01 -9.55507815e-01 -1.29739153e+00 -3.70772779e-02
-3.44075352e-01 5.41185737e-01 6.42308652e-01 6.68961763e-01
5.37708521e-01 6.45621002e-01 1.11557877e+00 -4.66739625e-01
-5.59956133e-01 -6.38610661e-01 -5.14228284e-01 -9.86154899e-02
6.74637675e-01 -6.47250354e-01 -1.27810404e-01 4.30767000e-01]
|
[9.973753929138184, 3.5700104236602783]
|
f97b2f3e-b877-4687-bd12-6e67421bdaa9
|
rhetorical-structure-approach-for-online
| null | null |
https://aclanthology.org/2022.lrec-1.635
|
https://aclanthology.org/2022.lrec-1.635.pdf
|
Rhetorical Structure Approach for Online Deception Detection: A Survey
|
Most information is passed on in the form of language. Therefore, research on how people use language to inform and misinform, and how this knowledge may be automatically extracted from large amounts of text is surely relevant. This survey provides first-hand experiences and a comprehensive review of rhetorical-level structure analysis for online deception detection. We systematically analyze how discourse structure, aligned or not with other approaches, is applied to automatic fake news and fake reviews detection on the web and social media. Moreover, we categorize discourse-tagged corpora along with results, hence offering a summary and accessible introductions to new researchers.
|
['Thiago Pardo', 'Fabrício Benevenuto', 'Zohar Rabinovich', 'Jonas D‘Alessandro', 'Francielle Vargas']
| null | null | null | null |
lrec-2022-6
|
['deception-detection']
|
['miscellaneous']
|
[ 1.92758337e-01 5.48412025e-01 -9.10329759e-01 8.56088754e-03
-7.83587694e-01 -1.05966210e+00 1.09052753e+00 7.63513684e-01
-1.74846843e-01 1.04400301e+00 9.61934149e-01 -6.24767482e-01
3.88111711e-01 -4.24767852e-01 -3.29440594e-01 -2.37397119e-01
3.48857552e-01 8.98719952e-02 1.95433289e-01 -6.87481165e-01
1.19389510e+00 -6.49024397e-02 -9.98033822e-01 8.61234903e-01
1.18766069e+00 4.02729154e-01 -4.51044589e-01 3.67803663e-01
-5.65728724e-01 1.58134925e+00 -1.55917406e+00 -1.13187420e+00
-6.15107596e-01 -6.93858087e-01 -1.07316196e+00 1.67270880e-02
5.75913966e-01 -3.57027680e-01 -4.06879127e-01 1.32923174e+00
2.05718443e-01 -5.89325666e-01 6.57411635e-01 -7.76182055e-01
-9.53020930e-01 7.99785316e-01 -1.36867672e-01 6.84524179e-01
8.37268770e-01 -2.32066259e-01 7.50242889e-01 -6.61115885e-01
1.09590566e+00 1.33346283e+00 5.98140359e-01 5.50418198e-01
-4.44261521e-01 -1.12948254e-01 -1.36585146e-01 3.34196359e-01
-7.93414176e-01 -5.66640496e-01 1.08156836e+00 -8.41167390e-01
7.96286464e-01 4.69017804e-01 6.69843018e-01 1.96147144e+00
3.24917018e-01 8.36881101e-01 1.38943899e+00 -8.04176331e-01
-2.00370878e-01 6.50925994e-01 6.23682499e-01 7.73458302e-01
8.41046751e-01 -3.68495345e-01 -8.90206754e-01 -8.46878827e-01
-2.48283878e-01 -7.51376331e-01 -2.64178991e-01 4.05815303e-01
-7.74068058e-01 1.34916639e+00 1.36034563e-01 8.98089290e-01
-2.57022470e-01 -1.90871358e-01 1.06496894e+00 4.36764508e-01
1.15119374e+00 7.13854551e-01 1.70818299e-01 -4.78035569e-01
-7.37324357e-01 4.27550644e-01 1.30261862e+00 5.98820865e-01
-1.38821378e-01 2.50591356e-02 -1.11556157e-01 8.98595750e-01
2.58236825e-01 6.56343281e-01 6.33486688e-01 -7.40612864e-01
7.94956863e-01 4.59950954e-01 5.21784127e-01 -2.10099244e+00
-3.48218046e-02 -5.56231663e-02 -1.93147033e-01 -4.75143969e-01
4.69842106e-01 -2.68818915e-01 8.94244164e-02 8.33839536e-01
7.78764188e-02 -6.46567345e-01 -1.10426359e-01 8.20895016e-01
1.19725204e+00 5.77427864e-01 -1.50783777e-01 -6.99945867e-01
1.80169189e+00 -7.84056485e-01 -1.50182056e+00 -6.62725031e-01
1.05480349e+00 -1.23010349e+00 7.71957278e-01 2.10989282e-01
-9.77884948e-01 4.82788861e-01 -1.29257762e+00 -3.85599375e-01
-6.47955656e-01 1.35552753e-02 3.08775276e-01 1.12046063e+00
-2.67911166e-01 4.56273049e-01 -3.61441821e-01 -2.88761407e-01
6.85665786e-01 -7.05256224e-01 2.19474301e-01 9.42831561e-02
-1.46466982e+00 1.58937097e+00 1.85964666e-02 -2.11871222e-01
-2.04786062e-01 -9.09461733e-03 -7.59807587e-01 -7.02368796e-01
6.36006296e-01 -1.36921853e-01 1.27657890e+00 -1.27068758e+00
-1.38600063e+00 1.38010073e+00 -3.56673896e-01 -7.41512477e-01
5.86142421e-01 -2.99302667e-01 -6.64377749e-01 6.50854349e-01
2.42222637e-01 -4.81421381e-01 1.31759131e+00 -1.13061357e+00
-4.67941612e-02 -4.48189795e-01 6.66000769e-02 -1.15936277e-02
-6.16229475e-01 1.01945508e+00 6.38378978e-01 -7.97613502e-01
-1.74731389e-01 -4.16551858e-01 5.13467431e-01 -4.70926970e-01
-7.10842311e-01 -6.14912570e-01 9.74319220e-01 -1.48710418e+00
1.92566729e+00 -1.73670471e+00 -2.07194015e-01 -6.91896975e-02
7.67611921e-01 5.73463261e-01 6.17042303e-01 1.22289050e+00
4.25265461e-01 8.21387708e-01 -1.43492565e-01 -1.49702132e-01
2.07481816e-01 6.59318566e-02 -6.68599665e-01 1.12768686e+00
-3.26485217e-01 1.08815587e+00 -1.19619989e+00 -7.57274032e-01
4.05409969e-02 3.47881496e-01 1.63547531e-01 -3.63064706e-01
-6.31198958e-02 1.67602673e-01 -7.37181902e-01 7.24838197e-01
2.44533822e-01 -4.88470227e-01 4.14101243e-01 1.67099148e-01
-4.67781484e-01 1.36479175e+00 2.16655046e-01 5.96335888e-01
-1.75298065e-01 1.57855010e+00 2.42015064e-01 -8.58792365e-01
5.78287184e-01 5.44532776e-01 -1.97746247e-01 -5.44666708e-01
4.00791168e-01 4.95813102e-01 -2.52638668e-01 -7.63603389e-01
7.31571496e-01 5.34661599e-02 -4.31000352e-01 5.88048637e-01
-4.64638680e-01 -7.60620832e-02 3.19833532e-02 5.95852494e-01
8.76249433e-01 -5.51286995e-01 8.58461440e-01 -2.22686395e-01
8.62002790e-01 6.68376744e-01 -1.67492792e-01 8.11178923e-01
-5.04471600e-01 -3.09270799e-01 1.01109648e+00 -5.70899248e-01
-1.00466096e+00 -2.29912192e-01 -2.26443395e-01 7.86454380e-01
1.99485451e-01 -9.04195368e-01 -1.08882284e+00 -1.02009821e+00
-8.67677405e-02 1.14940786e+00 -6.11773193e-01 2.78309006e-02
-8.73570859e-01 -6.80322886e-01 7.97644556e-01 -5.27983665e-01
6.01297259e-01 -1.00366426e+00 -6.59699082e-01 6.31814077e-02
-9.03852761e-01 -1.13583493e+00 -2.14273095e-01 -7.62723505e-01
-3.92984748e-01 -1.38944376e+00 -2.29858235e-01 -5.96803427e-01
4.03213918e-01 5.65062523e-01 1.10070407e+00 6.85842395e-01
2.35794798e-01 1.53758720e-01 -7.84434021e-01 -3.69405121e-01
-1.09501255e+00 -1.29149050e-01 -3.05236965e-01 -2.31045723e-01
6.15997910e-01 -1.83772191e-01 -3.64343524e-02 -9.77226570e-02
-7.21151412e-01 -4.12315607e-01 -1.30609080e-01 8.63746643e-01
-6.37452364e-01 -2.36759454e-01 6.83803380e-01 -1.26428330e+00
1.50913262e+00 -7.29221642e-01 -2.36878440e-01 -5.49370609e-02
-5.37386596e-01 -4.52507019e-01 3.98113757e-01 -3.05653155e-01
-9.52101111e-01 -1.04625130e+00 -1.30419493e-01 4.86562461e-01
3.43153775e-01 6.68509185e-01 4.99022573e-01 -2.61543602e-01
1.28015637e+00 9.48378001e-04 1.13883838e-01 -3.00241470e-01
3.39419156e-01 1.29335189e+00 6.06427714e-02 -1.48294836e-01
6.99466050e-01 6.57206953e-01 -7.12007046e-01 -1.34984708e+00
-1.57908463e+00 -3.54964972e-01 -2.59825289e-01 -6.08901918e-01
5.05091488e-01 -4.10273284e-01 -5.17483950e-01 4.39013958e-01
-1.82118726e+00 9.71302912e-02 1.40347749e-01 4.77134399e-02
-9.22656283e-02 9.05818164e-01 -1.07869625e+00 -1.25310481e+00
-2.44340405e-01 -4.07343626e-01 6.53325915e-01 -4.01332021e-01
-7.87493408e-01 -1.41455412e+00 -8.95317346e-02 1.12146258e+00
1.67074323e-01 6.48713648e-01 7.24417269e-01 -1.08467054e+00
1.02983296e-01 -4.31041598e-01 -4.26277220e-02 3.30217212e-01
-1.03894994e-01 -1.85958728e-01 -5.91366768e-01 8.87248293e-02
5.93930304e-01 -7.60621428e-01 5.06088376e-01 -3.37592000e-03
5.42600930e-01 -1.58821404e+00 -3.99956793e-01 -6.67021811e-01
9.14776266e-01 -3.14907789e-01 6.36955500e-01 5.43337643e-01
3.46323878e-01 1.13929284e+00 4.47181135e-01 3.86027068e-01
3.43151212e-01 6.37270570e-01 1.92606553e-01 8.12302887e-01
-7.14843571e-02 -5.15594721e-01 6.37867987e-01 1.29726779e+00
1.28684267e-01 -6.64514422e-01 -7.98296690e-01 4.42130208e-01
-1.63373888e+00 -1.47093368e+00 -8.61800790e-01 1.48183405e+00
8.09519470e-01 2.64151722e-01 2.58574158e-01 3.06590885e-01
1.05111670e+00 9.06209767e-01 2.26412326e-01 -7.25735426e-01
-4.86434162e-01 -3.47900093e-01 5.78191996e-01 9.32621062e-01
-9.26484942e-01 1.08587694e+00 7.12782717e+00 1.06361175e+00
-7.07635641e-01 6.73258722e-01 2.92958349e-01 4.04048920e-01
-2.23871350e-01 -9.24549550e-02 -4.73952770e-01 7.52435744e-01
9.89255309e-01 -2.51831681e-01 2.59595782e-01 7.46560097e-01
3.60851049e-01 -2.15501234e-01 -2.30277658e-01 7.51639605e-01
7.72095263e-01 -1.75823927e+00 -1.68152377e-01 2.15804890e-01
6.99720323e-01 -2.14101449e-01 -5.64380810e-02 2.63596736e-02
1.45801514e-01 -7.40612447e-01 1.05291069e+00 1.75723940e-01
9.65280533e-02 -3.14250767e-01 1.10964608e+00 8.11055601e-01
1.23592027e-01 8.30723867e-02 -1.46389619e-01 -5.01878440e-01
5.34164965e-01 7.03771710e-01 -7.27794290e-01 2.76711613e-01
1.87408715e-01 9.39841390e-01 -4.73551542e-01 2.17663899e-01
-8.76389861e-01 1.06554568e+00 1.37636706e-01 -9.97433841e-01
2.19935074e-01 -9.94826928e-02 1.13394964e+00 1.33062768e+00
-2.68747896e-01 5.57027049e-02 -1.24352589e-01 6.39893055e-01
-8.90049413e-02 1.88969046e-01 -9.24743712e-01 -6.39767349e-01
7.20631003e-01 9.37141061e-01 -4.28074300e-01 -7.10391164e-01
-4.44935471e-01 9.52194571e-01 2.81255305e-01 -1.68702025e-02
-6.23381138e-01 -2.50260919e-01 1.27756611e-01 4.15182859e-01
-1.37277961e-01 -3.18549514e-01 -5.39937675e-01 -1.36253309e+00
1.91776991e-01 -1.23601043e+00 1.96891218e-01 -5.39628267e-01
-1.73665988e+00 4.76724714e-01 2.60771681e-02 -8.20857346e-01
-3.05777192e-01 -6.16317749e-01 -3.41210902e-01 2.76486039e-01
-1.15737116e+00 -7.18707144e-01 1.04860939e-01 -3.81193473e-03
4.18466121e-01 5.62178306e-02 6.18717909e-01 1.54601604e-01
-3.54782492e-01 1.46433473e-01 -7.08325347e-03 3.96613419e-01
6.70517862e-01 -7.79137850e-01 1.26564741e-01 4.76356804e-01
-1.11173064e-01 5.87336242e-01 1.19549680e+00 -1.03103840e+00
-1.00021982e+00 -3.99723679e-01 1.86865520e+00 -1.07171869e+00
1.35878623e+00 -3.40867221e-01 -9.59760964e-01 4.15325880e-01
7.21322775e-01 -6.56042755e-01 6.82781279e-01 -1.98738828e-01
-5.88754416e-01 8.88648152e-01 -1.41395378e+00 7.32475638e-01
1.01153278e+00 -8.39583635e-01 -1.37011063e+00 1.25276256e+00
5.67794144e-01 -5.08340597e-01 -5.23923397e-01 -3.11178476e-01
3.29974353e-01 -8.58787060e-01 5.99235296e-01 -7.51050293e-01
9.20733035e-01 2.26701647e-01 1.32722810e-01 -1.09072828e+00
1.34108663e-01 -9.59555447e-01 -4.46108609e-01 9.81176972e-01
2.91325033e-01 -9.98421431e-01 5.17153800e-01 2.32150882e-01
-2.91008651e-01 -3.04734260e-01 -1.06005764e+00 -5.99715233e-01
2.72708923e-01 -1.84053451e-01 -2.74988741e-01 1.58472013e+00
9.18901622e-01 6.41458929e-01 -5.42680502e-01 -3.77991766e-01
5.93818009e-01 -1.54919431e-01 4.24779832e-01 -9.89921153e-01
4.34725016e-01 -5.34551263e-01 -1.65669128e-01 -1.11912167e+00
4.22537237e-01 -6.19966269e-01 -4.36049223e-01 -1.18809700e+00
2.75951270e-02 4.62178409e-01 9.31150079e-01 -1.47518501e-01
1.98102519e-01 3.55944455e-01 -4.78278846e-03 6.53044105e-01
-5.81962705e-01 3.76042694e-01 1.37100363e+00 -8.53632316e-02
-2.79535968e-02 -2.77636498e-01 -9.48963821e-01 1.19770753e+00
8.57184052e-01 -7.17959821e-01 1.65857971e-01 -3.56206223e-02
7.20363438e-01 1.24760382e-01 5.71512282e-01 -1.76906452e-01
4.72847484e-02 -1.02050133e-01 -2.91843772e-01 -5.82993507e-01
3.40390027e-01 -2.91860253e-01 -5.80877960e-01 5.27727008e-01
-4.54539269e-01 1.25785425e-01 -2.45282918e-01 6.95934355e-01
-3.52689058e-01 -9.20204461e-01 7.21439540e-01 -3.61984134e-01
8.15925300e-02 -7.07134008e-01 -1.33754468e+00 5.57624817e-01
7.34231591e-01 -1.42825902e-01 -1.28400695e+00 -1.07194245e+00
-4.66932744e-01 -3.49837482e-01 4.20545697e-01 2.91915268e-01
3.10905904e-01 -1.24733245e+00 -8.97821724e-01 -7.58044660e-01
1.19529828e-01 -1.11735022e+00 -1.81622103e-01 1.00380754e+00
-6.40528738e-01 7.92087793e-01 3.09415400e-01 2.29570717e-02
-1.19491506e+00 5.98587692e-01 7.31303990e-02 -1.59332141e-01
-4.89034921e-01 2.74302661e-01 -6.38020635e-01 -1.53363384e-02
-3.45395565e-01 2.26005033e-01 -5.50290167e-01 2.31689885e-01
8.76484334e-01 6.80650949e-01 -5.23492359e-02 -1.22487664e+00
-3.35926414e-01 -2.66692638e-01 -3.81864682e-02 -2.56096870e-01
8.40495348e-01 -6.26785159e-01 -7.48725414e-01 6.86401188e-01
1.24772668e+00 8.43156993e-01 -1.40647918e-01 -1.95746616e-01
5.49667716e-01 -7.44382203e-01 -9.26062837e-02 -9.46885586e-01
-1.66397303e-01 3.90668929e-01 -5.75217962e-01 1.18517530e+00
1.91760287e-01 3.16131145e-01 7.75579691e-01 2.91198373e-01
1.60978600e-01 -1.47491038e+00 6.17366850e-01 6.20362520e-01
1.35119224e+00 -1.09243834e+00 3.15570712e-01 -1.05209231e+00
-6.22752905e-01 1.22863376e+00 1.69723719e-01 -3.16055566e-01
5.96255004e-01 -1.59784988e-01 8.46442729e-02 -8.73350263e-01
-4.00102019e-01 4.91419613e-01 2.42862001e-01 4.73414928e-01
6.90080702e-01 -2.55543999e-02 -1.41946745e+00 5.08148193e-01
-7.13455260e-01 -6.29505098e-01 1.16640341e+00 9.76977408e-01
-5.91720939e-01 -9.40076649e-01 -6.28044009e-01 4.26688254e-01
-1.10147345e+00 -1.75112769e-01 -1.55001736e+00 8.95520747e-01
-2.53574133e-01 1.53142905e+00 -3.28650028e-01 -1.42817333e-01
-1.57131881e-01 1.74103186e-01 4.04526383e-01 -4.94242519e-01
-9.29332256e-01 -3.30235630e-01 1.51472592e+00 -3.16115797e-01
-8.46190929e-01 -8.39491308e-01 -4.01534975e-01 -8.76840830e-01
-6.84793591e-01 4.46448296e-01 6.85307562e-01 1.27615523e+00
1.87072411e-01 -6.79774880e-02 2.43178293e-01 -1.58987775e-01
-5.24879634e-01 -1.33222425e+00 -1.76489025e-01 2.87121218e-02
5.40217936e-01 -5.83449602e-01 -9.73878145e-01 -1.69185534e-01]
|
[8.223597526550293, 10.251309394836426]
|
5d681c7d-66b1-4a97-9cd5-77050c01b637
|
salt-and-pepper-noise-removal-method-based-on
|
2110.09113
| null |
https://arxiv.org/abs/2110.09113v9
|
https://arxiv.org/pdf/2110.09113v9.pdf
|
Salt and pepper noise removal method based on stationary Framelet transform with non-convex sparsity regularization
|
Salt and pepper noise removal is a common inverse problem in image processing. Traditional denoising methods have two limitations. First, noise characteristics are often not described accurately. For example, the noise location information is often ignored and the sparsity of the salt and pepper noise is often described by L1 norm, which cannot illustrate the sparse variables clearly. Second, conventional methods separate the contaminated image into a recovered image and a noise part, thus resulting in recovering an image with unsatisfied smooth parts and detail parts. In this study, we introduce a noise detection strategy to determine the position of the noise, and a non-convex sparsity regularization depicted by Lp quasi-norm is employed to describe the sparsity of the noise, thereby addressing the first limitation. The morphological component analysis framework with stationary Framelet transform is adopted to decompose the processed image into cartoon, texture, and noise parts to resolve the second limitation. Then, the alternating direction method of multipliers (ADMM) is employed to solve the proposed model. Finally, experiments are conducted to verify the proposed method and compare it with some current state-of-the-art denoising methods. The experimental results show that the proposed method can remove salt and pepper noise while preserving the details of the processed image.
|
['Yanping Xu', 'Chaoqun Yu', 'Yuming Huang', 'Jianhua Song', 'Huiying Huang', 'Lingzhi Wang', 'Yingpin Chen']
|
2021-10-18
| null | null | null | null |
['salt-and-pepper-noise-removal']
|
['computer-vision']
|
[ 2.93892086e-01 -6.28643572e-01 1.74084470e-01 7.29720965e-02
-3.53949666e-01 1.73261575e-02 4.26100455e-02 -2.35074386e-01
-3.55615854e-01 6.08301938e-01 1.98847651e-01 2.17828408e-01
-1.67228088e-01 -6.87763691e-01 -1.35124773e-01 -1.30122185e+00
3.52926224e-01 -5.00156999e-01 9.62688401e-02 -2.10136995e-01
1.93446070e-01 2.15134114e-01 -1.26386583e+00 7.71357194e-02
1.12851787e+00 1.17748880e+00 4.79862869e-01 6.80233911e-02
-3.56257826e-01 7.15268135e-01 -5.03525794e-01 -1.12244911e-01
3.60705018e-01 -6.51266158e-01 1.03410609e-01 6.91134512e-01
-5.64132743e-02 -4.67709720e-01 -4.38431770e-01 1.81983614e+00
4.81051385e-01 2.04888061e-01 1.83419243e-01 -9.04355049e-01
-5.74749470e-01 1.11277536e-01 -1.16080546e+00 1.31575227e-01
-8.90046358e-02 -8.35169405e-02 4.01762515e-01 -1.05778623e+00
5.84004581e-01 1.38878417e+00 6.86245739e-01 -3.33914999e-03
-1.10070789e+00 -6.06556416e-01 2.02759802e-01 2.66626358e-01
-1.51006198e+00 -3.16771418e-01 1.17543101e+00 -2.37411141e-01
7.07914010e-02 3.43003958e-01 5.89285731e-01 6.07120693e-01
3.24362397e-01 6.69864714e-01 1.28398955e+00 -2.27616444e-01
9.02478322e-02 -8.68106559e-02 -1.49231786e-02 4.83133137e-01
6.34323716e-01 -8.93536285e-02 -1.55221432e-01 -1.88777316e-02
8.36410940e-01 3.66275191e-01 -7.95049548e-01 -1.75155669e-01
-1.02592373e+00 5.83956301e-01 2.79364288e-01 5.24176717e-01
-5.09614706e-01 -4.24267620e-01 4.62136179e-01 1.19382389e-01
4.75460410e-01 -2.61417538e-01 -4.18353565e-02 2.23009840e-01
-1.07746899e+00 1.20887980e-01 6.04164422e-01 7.00891495e-01
7.14070380e-01 4.55122501e-01 8.74082744e-02 1.00680208e+00
6.04958951e-01 6.56320095e-01 4.94208604e-01 -9.46870744e-01
4.85755235e-01 3.50714564e-01 1.70225710e-01 -1.74007642e+00
-6.99972212e-02 -5.88989615e-01 -1.48848116e+00 3.39969158e-01
2.46367157e-01 -1.19746752e-01 -7.68009245e-01 1.21929324e+00
2.24829987e-01 3.62299502e-01 2.30320636e-02 1.20642734e+00
8.26311588e-01 9.18734252e-01 -5.21136187e-02 -8.61459792e-01
1.36090183e+00 -7.17216909e-01 -1.49009609e+00 -2.12871179e-01
-3.05251721e-02 -1.20265472e+00 4.31243449e-01 6.60270154e-01
-1.09278488e+00 -7.22508430e-01 -1.03545034e+00 6.74242526e-02
9.13044736e-02 3.52312297e-01 4.75561358e-02 3.96480292e-01
-3.86353225e-01 4.29209679e-01 -7.48516619e-01 2.26905849e-02
2.73472995e-01 -2.11067781e-01 -2.78186262e-01 -5.60286820e-01
-1.02792072e+00 7.42001057e-01 3.60717028e-01 9.39077675e-01
-3.85596871e-01 -4.28839207e-01 -8.04104388e-01 1.06189795e-01
4.60432857e-01 -4.21309710e-01 5.28955102e-01 -1.05510962e+00
-1.17026126e+00 3.75478745e-01 -2.70418108e-01 3.84240411e-02
5.56201756e-01 -1.18868597e-01 -6.47405744e-01 2.41690248e-01
1.44511759e-01 -3.05309743e-01 1.27708590e+00 -1.57474101e+00
-5.28415084e-01 -4.05789286e-01 -5.35123408e-01 1.82189539e-01
-7.65468404e-02 -8.11738819e-02 -6.59265935e-01 -1.25689173e+00
9.15271759e-01 -4.32467848e-01 -3.40886980e-01 1.86400369e-01
-2.44120240e-01 5.72976112e-01 1.14136791e+00 -1.22703028e+00
1.44175446e+00 -2.71609187e+00 3.10975641e-01 3.47418576e-01
1.53265476e-01 3.10261816e-01 -1.02936953e-01 3.08089793e-01
-1.15263835e-01 -1.59356505e-01 -6.98930502e-01 -2.61252373e-01
-4.91879255e-01 4.17156696e-01 -9.06691700e-02 7.32263684e-01
1.00423638e-02 1.77860498e-01 -7.34603882e-01 -5.07211149e-01
3.08940679e-01 8.10821354e-01 -6.13377392e-02 1.27185389e-01
3.31704378e-01 6.61020160e-01 -5.68879902e-01 6.51140571e-01
1.45039928e+00 3.46100897e-01 6.05001785e-02 -8.32615614e-01
-3.76823604e-01 -5.31487226e-01 -1.96649349e+00 1.27453756e+00
-5.57819307e-02 1.51676819e-01 1.11690140e+00 -1.14125597e+00
1.12971509e+00 3.02526504e-01 5.44140339e-01 -6.10828280e-01
3.63012582e-01 4.28505421e-01 -1.57057136e-01 -9.66317773e-01
6.06791675e-02 -4.65635568e-01 6.22249067e-01 -7.22622797e-02
-3.60406637e-01 3.66544351e-03 2.37709492e-01 -1.77460074e-01
6.58514678e-01 7.12299496e-02 2.60487854e-01 -3.27352911e-01
9.89883959e-01 -1.50868967e-01 1.19078445e+00 3.71583343e-01
-2.46773660e-01 7.80349493e-01 3.09319794e-01 -2.78140664e-01
-8.08892608e-01 -6.84574425e-01 -1.58736885e-01 3.57875794e-01
6.05941832e-01 -4.36685160e-02 -6.40871108e-01 -2.11689219e-01
-2.09466338e-01 3.75672340e-01 -2.56393939e-01 6.47145906e-04
-7.63989747e-01 -1.00592577e+00 -6.17683399e-03 -2.55609420e-03
1.11920381e+00 -8.57352018e-01 -1.33755326e-01 3.84955853e-01
-4.99966323e-01 -1.06777918e+00 -4.60576952e-01 -2.28776321e-01
-1.02365100e+00 -1.19892323e+00 -1.00900590e+00 -1.05787838e+00
1.01967752e+00 7.51518726e-01 4.90192771e-01 4.63826120e-01
-2.25140125e-01 -3.11418548e-02 -3.51246566e-01 -3.78045514e-02
-1.64769039e-01 -7.47235715e-01 -1.33307189e-01 6.93122625e-01
7.12372689e-03 -5.71043730e-01 -6.85589492e-01 3.10777038e-01
-1.29699278e+00 -6.86886162e-02 7.98848629e-01 1.01312578e+00
7.73374259e-01 8.93561125e-01 1.39824405e-01 -7.45507658e-01
6.42807424e-01 -3.40559006e-01 -6.21435821e-01 3.02562527e-02
-3.60924184e-01 -3.98660541e-01 7.26180792e-01 -4.66747642e-01
-1.37334716e+00 7.63615072e-02 -1.12829797e-01 -4.55534041e-01
-3.15652937e-02 7.62241006e-01 -5.25904417e-01 -2.75735766e-01
1.67489946e-02 8.21764588e-01 4.77008104e-01 -8.11938286e-01
7.02922195e-02 4.68367249e-01 5.63778698e-01 -2.71637321e-01
1.10845077e+00 7.48489320e-01 1.70373723e-01 -1.13193810e+00
-5.78989148e-01 -5.18058479e-01 -3.27466369e-01 -1.75653115e-01
6.89905226e-01 -8.94526362e-01 -5.85949659e-01 8.01159918e-01
-1.05349517e+00 3.82391661e-01 -1.05310343e-01 6.09446108e-01
-2.23619416e-01 9.88465965e-01 -7.09350705e-01 -8.89495373e-01
-1.99562073e-01 -1.27216208e+00 5.73598921e-01 3.11642140e-01
4.43711430e-01 -7.23251641e-01 -3.65019292e-01 3.24190319e-01
6.90871835e-01 2.96845913e-01 7.72155046e-01 7.98537061e-02
-5.05152047e-01 -1.58597052e-01 -3.74971360e-01 7.41139591e-01
2.92187810e-01 -2.24202573e-01 -4.21721131e-01 -3.24543089e-01
9.42411244e-01 2.90919930e-01 7.86512911e-01 5.83091855e-01
8.94791901e-01 -4.13088858e-01 -7.06950650e-02 7.45651424e-01
1.79687548e+00 3.83547902e-01 1.02364206e+00 5.70817530e-01
5.29319227e-01 5.66944182e-01 7.27719128e-01 4.69994426e-01
-5.97455092e-02 2.65709490e-01 4.35377270e-01 -3.89214933e-01
-1.70146570e-01 3.14808846e-03 1.39620543e-01 1.14438665e+00
-7.43640438e-02 7.50376359e-02 -2.09793612e-01 4.24937010e-01
-1.84528780e+00 -8.48247588e-01 -4.66035634e-01 1.93525898e+00
5.82288384e-01 -5.41720502e-02 -4.39945489e-01 6.02232516e-01
9.60384786e-01 3.57278466e-01 -3.17936301e-01 2.23134950e-01
-4.53144819e-01 -5.49333692e-02 4.98077631e-01 5.02688289e-01
-1.04107428e+00 2.84257859e-01 5.00507259e+00 1.03613174e+00
-1.13972986e+00 -2.72843260e-02 3.94272506e-01 4.43669260e-01
-7.81414434e-02 8.85958523e-02 -3.19170654e-01 7.40874290e-01
-9.67392176e-02 -7.57985339e-02 4.98703718e-01 4.93673027e-01
8.02277684e-01 -2.21627563e-01 -1.69281512e-01 1.29684913e+00
2.67628059e-02 -7.34889388e-01 -4.79866704e-03 -1.35266244e-01
5.75793922e-01 -6.43368244e-01 -3.27274688e-02 -9.12279263e-02
-4.10676271e-01 -6.78484440e-01 6.15483999e-01 7.75458634e-01
3.02192062e-01 -6.41151547e-01 1.02194190e+00 5.28155327e-01
-1.19388807e+00 -1.02058627e-01 -5.72403789e-01 -7.11929873e-02
2.91336298e-01 1.10233319e+00 2.77021140e-01 6.91266060e-01
8.03361714e-01 9.67797458e-01 -8.11028257e-02 1.25352144e+00
-3.11582476e-01 5.71830332e-01 -2.64970332e-01 5.66466570e-01
1.01733021e-01 -1.05788052e+00 9.30479825e-01 9.94387925e-01
6.06378913e-01 6.50536180e-01 3.02783489e-01 7.94865608e-01
2.06883252e-01 3.73986006e-01 -1.91675782e-01 2.41254508e-01
2.07649961e-01 1.21019065e+00 -6.56207263e-01 -2.17900366e-01
-6.36033475e-01 8.95949662e-01 -5.32386959e-01 7.65846670e-01
-4.12646145e-01 -4.91440326e-01 4.62056041e-01 1.29924268e-01
5.71977317e-01 -3.13675225e-01 -4.93724674e-01 -1.41741729e+00
3.14336777e-01 -1.17690325e+00 1.42225564e-01 -6.02353871e-01
-1.29025567e+00 3.29637617e-01 -3.98850471e-01 -1.62358499e+00
6.97882533e-01 -3.53136897e-01 -6.74193203e-01 1.14135754e+00
-1.70334303e+00 -9.00427818e-01 -5.62515438e-01 4.98836398e-01
6.03521168e-01 -2.19432451e-02 3.83631974e-01 6.18380725e-01
-8.90867293e-01 -5.87271415e-02 4.85474080e-01 1.50952131e-01
4.79760289e-01 -6.11946166e-01 -4.09243226e-01 1.27404511e+00
-5.49210489e-01 7.04431057e-01 8.85948300e-01 -8.48397195e-01
-1.51716447e+00 -8.48581433e-01 5.71538508e-01 7.24396110e-01
5.02998173e-01 1.07844673e-01 -1.39976037e+00 2.99658716e-01
9.49815884e-02 2.77797550e-01 2.15730965e-01 -6.56850159e-01
-2.22588871e-02 -3.26461315e-01 -1.43056619e+00 5.10846257e-01
3.94326895e-01 5.68316393e-02 -5.31196237e-01 5.17818937e-03
2.27372706e-01 -3.68027061e-01 -8.29638243e-01 5.67017436e-01
2.80906171e-01 -1.01466906e+00 1.01295245e+00 8.69165882e-02
4.53340560e-01 -9.38085020e-01 -2.13024423e-01 -1.10653555e+00
-5.84729910e-01 -4.40060437e-01 -6.34189174e-02 1.43351364e+00
-9.34997424e-02 -6.09494507e-01 5.05078197e-01 2.15248078e-01
1.03204250e-01 -6.18260086e-01 -8.60216677e-01 -5.43913960e-01
-4.40984726e-01 -3.48588859e-04 1.79339916e-01 9.38750744e-01
-4.40600902e-01 2.16831528e-02 -6.80103660e-01 3.25986832e-01
1.11023533e+00 -3.59603018e-02 4.01799530e-01 -9.55940306e-01
6.45442009e-02 -3.35180759e-01 -2.77621150e-01 -1.04997706e+00
-2.88988113e-01 -3.32700998e-01 1.94633186e-01 -1.59179521e+00
8.71059000e-02 -2.39930108e-01 -2.75318384e-01 -4.12987471e-02
-4.96686459e-01 1.76044270e-01 1.97553501e-01 5.16541719e-01
-1.78135689e-02 7.25086808e-01 1.57225430e+00 -2.78065026e-01
-4.98721823e-02 -4.60772309e-03 -7.39066899e-01 8.82134616e-01
4.79569823e-01 -3.72515082e-01 -1.63265795e-01 -5.18194199e-01
-1.49748728e-01 2.29180053e-01 2.00240940e-01 -9.08940315e-01
2.24856451e-01 -2.37378091e-01 4.09310400e-01 -4.74955708e-01
3.22799951e-01 -1.35329115e+00 1.39141247e-01 4.77849871e-01
1.64739728e-01 -3.75577927e-01 2.89085098e-02 8.00654888e-01
-7.29657412e-01 -4.24832225e-01 1.06128728e+00 -3.54113311e-01
-7.09075689e-01 1.47624657e-01 -3.73040617e-01 -1.97159722e-01
8.77252221e-01 -3.46198320e-01 -1.69060841e-01 -3.41976285e-01
-7.67068744e-01 9.77756158e-02 2.87134975e-01 5.64168658e-05
1.00986779e+00 -1.18856573e+00 -9.81412113e-01 5.64629793e-01
-3.85695994e-01 -5.58249094e-02 7.12683320e-01 1.14319265e+00
-8.67299020e-01 -3.23816240e-01 -5.88118434e-02 -3.47424448e-01
-1.25353813e+00 6.83754444e-01 2.10926756e-01 -1.83517709e-01
-9.04527426e-01 4.20880467e-01 1.50909454e-01 7.63365552e-02
1.88492507e-01 -6.71987014e-04 -4.43694532e-01 7.91284069e-02
7.16642082e-01 6.22789621e-01 -1.85105622e-01 -1.02222383e+00
-9.03402045e-02 9.84834373e-01 2.31900975e-01 2.40403652e-01
1.39146912e+00 -4.47505921e-01 -7.62692690e-01 2.68209819e-02
1.25882792e+00 2.70974934e-01 -1.00162399e+00 -3.09760451e-01
-2.88946092e-01 -8.65634382e-01 3.53582203e-01 -3.26752901e-01
-1.50532103e+00 7.43646741e-01 6.52168036e-01 1.91999525e-01
1.65269685e+00 -8.70745242e-01 1.11239171e+00 -9.90654528e-02
1.14068836e-01 -1.00573146e+00 -2.47510582e-01 2.28382125e-01
1.00685477e+00 -1.06614172e+00 5.26700616e-01 -9.57797587e-01
-4.34421360e-01 1.17723787e+00 3.35549802e-01 -3.34352165e-01
7.83465326e-01 3.06715608e-01 3.22253585e-01 8.18969607e-02
3.13446224e-02 -1.66710809e-01 5.85281737e-02 3.45630556e-01
1.45816267e-01 -3.55755746e-01 -9.06765282e-01 7.45182753e-01
4.21756625e-01 7.15580285e-02 5.11079609e-01 1.01542771e+00
-6.50124490e-01 -9.06861424e-01 -1.08563888e+00 2.10123137e-01
-8.28738749e-01 -4.97259907e-02 2.50987142e-01 5.15196145e-01
4.50875849e-01 1.22347474e+00 -4.15385157e-01 -4.60978709e-02
5.84971666e-01 -2.72528052e-01 8.49179700e-02 -1.79597512e-01
-1.49360970e-01 7.56588697e-01 -4.15271878e-01 -2.90378213e-01
-5.71108639e-01 -3.90150726e-01 -9.89849925e-01 -1.32606745e-01
-3.34213793e-01 2.23825932e-01 6.33401453e-01 7.21310377e-01
-7.68077970e-02 4.14781481e-01 6.78622544e-01 -8.03610265e-01
-4.85266089e-01 -7.95964241e-01 -1.04162574e+00 6.05540872e-01
5.04272282e-01 -5.45037627e-01 -8.16873372e-01 3.47181767e-01]
|
[11.239115715026855, -2.470210313796997]
|
be6f6535-3460-4fa5-933b-e0ac9c1a7471
|
abstractive-text-summarization-by
|
1812.05407
| null |
http://arxiv.org/abs/1812.05407v1
|
http://arxiv.org/pdf/1812.05407v1.pdf
|
Abstractive Text Summarization by Incorporating Reader Comments
|
In neural abstractive summarization field, conventional sequence-to-sequence
based models often suffer from summarizing the wrong aspect of the document
with respect to the main aspect. To tackle this problem, we propose the task of
reader-aware abstractive summary generation, which utilizes the reader comments
to help the model produce better summary about the main aspect. Unlike
traditional abstractive summarization task, reader-aware summarization
confronts two main challenges: (1) Comments are informal and noisy; (2) jointly
modeling the news document and the reader comments is challenging. To tackle
the above challenges, we design an adversarial learning model named
reader-aware summary generator (RASG), which consists of four components: (1) a
sequence-to-sequence based summary generator; (2) a reader attention module
capturing the reader focused aspects; (3) a supervisor modeling the semantic
gap between the generated summary and reader focused aspects; (4) a goal
tracker producing the goal for each generation step. The supervisor and the
goal tacker are used to guide the training of our framework in an adversarial
manner. Extensive experiments are conducted on our large-scale real-world text
summarization dataset, and the results show that RASG achieves the
state-of-the-art performance in terms of both automatic metrics and human
evaluations. The experimental results also demonstrate the effectiveness of
each module in our framework. We release our large-scale dataset for further
research.
|
['Shen Gao', 'Piji Li', 'Zhaochun Ren', 'Xiuying Chen', 'Rui Yan', 'Dongyan Zhao', 'Lidong Bing']
|
2018-12-13
| null | null | null | null |
['reader-aware-summarization']
|
['natural-language-processing']
|
[ 4.78908956e-01 3.94719958e-01 -2.88845152e-02 -1.28243297e-01
-1.33834195e+00 -5.51130116e-01 8.02821517e-01 2.34328121e-01
-2.11462811e-01 9.28645611e-01 9.95185733e-01 -5.72374016e-02
4.39842999e-01 -4.96787906e-01 -7.40544081e-01 -3.58392179e-01
4.30761516e-01 4.79383916e-01 2.35670373e-01 -3.69940966e-01
6.36821508e-01 -2.57594287e-01 -1.15284026e+00 4.87713814e-01
1.31886244e+00 7.57291615e-01 4.86864477e-01 1.07515061e+00
-2.05995008e-01 1.03410995e+00 -1.35141969e+00 -4.58797932e-01
-9.42339823e-02 -1.10472918e+00 -7.52761185e-01 2.03291729e-01
3.38626921e-01 -4.23957348e-01 -3.03772509e-01 1.11298466e+00
8.35913479e-01 4.88104403e-01 7.53297985e-01 -1.04906476e+00
-8.77840877e-01 1.14333224e+00 -6.25595331e-01 2.46048912e-01
6.09044611e-01 1.58806413e-01 1.04481375e+00 -6.21616721e-01
5.39098442e-01 1.31308448e+00 3.79225224e-01 7.39256322e-01
-5.92297494e-01 -4.34931457e-01 5.66098452e-01 1.31772049e-02
-7.15739369e-01 -5.73247671e-01 1.02339983e+00 -3.70880477e-02
9.51179206e-01 5.05784988e-01 5.11078835e-01 1.23259401e+00
3.73976558e-01 1.20790613e+00 6.08172894e-01 -7.91712478e-02
2.91200876e-01 -4.09014933e-02 3.26165944e-01 3.22066277e-01
2.34449878e-01 -4.76162702e-01 -5.47775090e-01 -6.58174530e-02
1.94710478e-01 -2.77214646e-01 -3.61406207e-01 4.56191063e-01
-1.13200748e+00 8.76701355e-01 3.12421948e-01 9.20220837e-02
-7.16299534e-01 1.78869918e-01 7.28148282e-01 1.98830932e-01
6.70305669e-01 5.30617893e-01 -1.30178720e-01 -6.89819381e-02
-1.18933427e+00 7.07700133e-01 1.09930742e+00 1.23354876e+00
2.80306727e-01 2.68778741e-01 -8.20875943e-01 7.70913363e-01
1.74044639e-01 4.48666513e-01 7.87292123e-01 -6.67856574e-01
1.03830004e+00 5.70871592e-01 1.32202834e-01 -9.77075160e-01
-5.35847656e-02 -6.89169526e-01 -1.08047843e+00 -3.33328813e-01
-2.16334909e-01 -5.19528091e-01 -7.44612813e-01 1.55919850e+00
-3.29299830e-02 1.27778828e-01 4.03662443e-01 7.30386674e-01
1.50776029e+00 1.14635813e+00 -7.15519786e-02 -4.53280866e-01
1.37444389e+00 -1.66088283e+00 -1.05360854e+00 -6.97313368e-01
2.04338089e-01 -7.57788539e-01 9.27199066e-01 4.12827218e-03
-1.62922335e+00 -6.05843663e-01 -1.17969549e+00 -1.84832454e-01
4.20484804e-02 4.26928371e-01 3.72790173e-02 -1.00568064e-01
-9.95736122e-01 3.66555452e-01 -6.05041206e-01 -2.32632577e-01
3.70314896e-01 -9.68762487e-02 4.04756144e-02 2.70585686e-01
-1.11761284e+00 6.11556351e-01 6.99822485e-01 -5.03532998e-02
-7.58821368e-01 -5.92764914e-01 -1.05957925e+00 2.97401279e-01
5.70018530e-01 -1.29424965e+00 1.68096089e+00 -9.97707605e-01
-1.45240986e+00 4.41418946e-01 -3.43753219e-01 -5.64325154e-01
6.58335984e-01 -5.19137025e-01 -1.22169547e-01 2.05219418e-01
4.55957919e-01 4.17259961e-01 8.38744581e-01 -1.38413572e+00
-7.77347744e-01 -2.23221168e-01 -1.44762740e-01 5.51850379e-01
4.53832597e-02 -3.64189669e-02 -4.78623211e-01 -1.23017156e+00
-2.65035599e-01 -5.90128422e-01 -1.54060125e-01 -7.27391720e-01
-1.12312520e+00 -2.04459980e-01 7.74979293e-01 -1.17840636e+00
1.57474196e+00 -1.71485174e+00 2.70526946e-01 -5.19712448e-01
6.94008693e-02 6.02470696e-01 -2.90141374e-01 8.15743923e-01
1.05744815e-02 1.07669465e-01 -3.95197064e-01 -6.73106492e-01
-8.98230076e-02 -2.39570931e-01 -7.66556084e-01 -1.77926078e-01
1.97707966e-01 1.27581298e+00 -1.15027487e+00 -4.69362646e-01
-1.59213126e-01 -1.96284410e-02 -2.91994900e-01 6.71057582e-01
-6.15633428e-01 2.99192190e-01 -8.03673804e-01 1.80756122e-01
4.67106014e-01 -1.07491665e-01 -3.53093594e-01 -3.01427580e-02
1.01399034e-01 6.09510243e-01 -9.82484996e-01 1.69772267e+00
-1.87015370e-01 4.00405169e-01 -2.39306837e-02 -9.23169672e-01
9.31007683e-01 3.58603120e-01 -1.97565183e-01 -4.17350233e-01
2.58545667e-01 2.98058372e-02 -2.20909253e-01 -4.83267576e-01
1.11252558e+00 -6.38881847e-02 -3.28509033e-01 8.35297167e-01
7.12465569e-02 -4.00736630e-01 4.22965646e-01 7.84894526e-01
1.12057543e+00 -6.13347553e-02 6.12550497e-01 4.18763831e-02
7.17393637e-01 -5.24903508e-03 5.10037005e-01 9.71382856e-01
-1.59002710e-02 7.97435164e-01 7.27052927e-01 3.38307396e-02
-9.13290977e-01 -7.58645236e-01 8.63767624e-01 7.93013811e-01
2.21100897e-01 -6.01939499e-01 -9.76390243e-01 -8.60075414e-01
-3.93607706e-01 1.37438571e+00 -5.69145918e-01 -3.53183389e-01
-5.86807072e-01 -2.94047236e-01 5.61594903e-01 6.97387218e-01
8.48639548e-01 -1.37459195e+00 -4.36112672e-01 2.50490844e-01
-5.91756821e-01 -9.91960406e-01 -1.19589424e+00 -2.66116977e-01
-7.44282067e-01 -8.02280486e-01 -8.50305140e-01 -7.78483689e-01
5.02173841e-01 4.44153041e-01 1.02468038e+00 8.45458657e-02
4.48184609e-01 1.88791826e-01 -4.63579059e-01 -8.42155218e-01
-9.60687101e-01 3.04450035e-01 -4.46051776e-01 -1.56465217e-01
-2.05063775e-01 -4.24830437e-01 -5.07529795e-01 -1.62166014e-01
-1.15348232e+00 4.56821918e-01 7.98271298e-01 7.42834985e-01
4.09715205e-01 -1.63243432e-02 1.11757123e+00 -1.16283429e+00
1.34993696e+00 -4.69840169e-01 -1.29532320e-02 3.30174446e-01
-1.56465292e-01 4.31865938e-02 1.04087186e+00 -2.63945162e-01
-1.31483567e+00 -4.35461998e-01 -2.12652415e-01 -6.34750426e-02
3.15404348e-02 7.44399548e-01 -4.01807874e-01 9.17126715e-01
5.48478961e-01 8.39733005e-01 -3.01108463e-03 -2.30638742e-01
4.16019768e-01 8.19158554e-01 9.79860306e-01 -2.30348006e-01
8.38983953e-01 6.34265020e-02 -5.19141853e-01 -8.02925706e-01
-1.26521432e+00 -3.55700880e-01 -1.89103201e-01 -8.70852023e-02
7.48729527e-01 -9.34700727e-01 -1.75938517e-01 5.82296610e-01
-1.60248697e+00 -1.71429306e-01 -4.61769730e-01 1.79415140e-02
-7.55801976e-01 6.29526258e-01 -5.35832524e-01 -6.93526566e-01
-1.22963262e+00 -9.73853648e-01 1.23636448e+00 7.32590020e-01
-4.59497422e-01 -1.06594145e+00 8.22552517e-02 6.11919641e-01
3.04830164e-01 3.09980541e-01 7.02399135e-01 -1.20788503e+00
-2.81683922e-01 -3.62473786e-01 -1.82912964e-02 5.86033463e-01
1.24600567e-01 -7.29350671e-02 -6.06807053e-01 -9.11404863e-02
2.21389085e-01 -2.80324161e-01 1.08837140e+00 2.87540942e-01
9.02016401e-01 -9.64509189e-01 -1.44767791e-01 1.92636326e-01
9.43592668e-01 1.30018115e-01 6.47391021e-01 2.22756058e-01
7.01769352e-01 4.25663859e-01 7.53894210e-01 4.73127693e-01
6.08948648e-01 2.23188609e-01 2.55568206e-01 1.06279790e-01
-3.57403815e-01 -7.19846308e-01 5.23320317e-01 1.22696722e+00
2.53489614e-01 -9.66652215e-01 -3.28596652e-01 6.17657185e-01
-2.15145326e+00 -1.22869229e+00 -5.38845211e-02 1.73050797e+00
9.56954539e-01 2.24373549e-01 1.68276414e-01 -6.86582699e-02
7.77228653e-01 7.46771276e-01 -6.41221881e-01 -5.89999676e-01
9.02243238e-03 -2.92971134e-01 -1.19257234e-01 3.76849353e-01
-9.28034961e-01 9.88931239e-01 4.99855089e+00 9.45765436e-01
-8.47895205e-01 -2.05520079e-01 5.02410889e-01 6.26442432e-02
-5.74979067e-01 -9.97481793e-02 -7.38073111e-01 7.78584599e-01
6.07436121e-01 -6.83686376e-01 9.63504612e-02 7.58180380e-01
4.30156708e-01 -7.90103599e-02 -9.67240214e-01 5.79714835e-01
6.92706525e-01 -1.34418190e+00 6.95853114e-01 -4.00369734e-01
8.79247785e-01 -3.01427543e-01 -3.32395792e-01 4.21327084e-01
2.99376518e-01 -7.82545328e-01 9.72108483e-01 5.65360308e-01
5.14798939e-01 -9.31342125e-01 8.88514876e-01 8.79418790e-01
-9.56696391e-01 1.63063034e-01 -1.84347942e-01 4.23692763e-02
5.20256400e-01 4.36818421e-01 -7.32451737e-01 1.01866782e+00
4.55678552e-02 9.76628780e-01 -4.76343334e-01 9.04632807e-01
-7.85319388e-01 7.65523195e-01 2.07039997e-01 -2.37673461e-01
3.92269611e-01 -1.71078071e-01 1.18616223e+00 1.31080246e+00
2.38869920e-01 1.47190124e-01 2.22785845e-01 8.42064857e-01
-3.75395745e-01 -8.84706446e-04 -4.25815254e-01 -2.82893240e-01
5.17726302e-01 1.15201533e+00 -5.82162440e-01 -6.54786706e-01
-1.78282470e-01 1.26965463e+00 1.72349766e-01 3.61894518e-01
-7.68218756e-01 -8.97345543e-01 1.04177393e-01 -1.46917239e-01
3.49862307e-01 1.68654367e-01 -4.55476850e-01 -1.29913020e+00
2.32137024e-01 -1.08681893e+00 3.74738455e-01 -1.07161844e+00
-1.08306968e+00 7.66083181e-01 -1.45791456e-01 -1.04221463e+00
-4.16662723e-01 2.82508969e-01 -1.49941921e+00 9.55543816e-01
-1.43075979e+00 -1.17988801e+00 -4.65958685e-01 5.95981143e-02
1.39337885e+00 -3.32632393e-01 4.80615377e-01 -3.74335706e-01
-7.65826106e-01 4.93083596e-01 -1.37372166e-01 7.01009706e-02
6.78910553e-01 -1.36875284e+00 9.29330468e-01 1.24830770e+00
-1.88813478e-01 5.18661737e-01 1.02710521e+00 -1.07504308e+00
-1.20283151e+00 -1.47191596e+00 1.10622787e+00 -3.30086321e-01
5.14151454e-01 -6.20754696e-02 -9.01163995e-01 7.90507972e-01
7.56346345e-01 -8.95270169e-01 4.56230491e-01 -5.17205119e-01
6.42160699e-02 1.92727432e-01 -8.05418015e-01 8.65739584e-01
8.53368759e-01 -6.60291081e-03 -1.27356625e+00 2.47827590e-01
1.15689766e+00 -6.95816338e-01 -2.01690525e-01 1.12230644e-01
1.69194654e-01 -7.05873370e-01 5.63750565e-01 -6.44995928e-01
1.21294999e+00 -2.74503946e-01 2.07748473e-01 -1.85377192e+00
-1.28571570e-01 -9.63738203e-01 -3.24695170e-01 1.69524932e+00
2.18336195e-01 -4.14838344e-01 3.99814606e-01 2.36228526e-01
-6.84450269e-01 -8.73452783e-01 -4.34078157e-01 -5.06707489e-01
-6.74035847e-02 -7.07020462e-02 6.10498011e-01 3.92654568e-01
-5.47320955e-02 1.17302251e+00 -3.92336100e-01 -1.20755054e-01
3.79721820e-01 2.49068096e-01 1.01379788e+00 -6.57959998e-01
-3.10101420e-01 -5.69119096e-01 1.78007856e-01 -1.44692409e+00
3.41174394e-01 -7.81877398e-01 4.53861296e-01 -2.30534172e+00
5.76747656e-01 3.53513122e-01 3.49306971e-01 1.19316861e-01
-1.02010655e+00 -3.77304643e-01 3.05548847e-01 1.07354268e-01
-9.82820690e-01 9.70846593e-01 1.42690694e+00 -3.44846129e-01
-3.16050112e-01 3.37582409e-01 -1.52990091e+00 6.90326154e-01
7.43187368e-01 -3.45869035e-01 -6.86360061e-01 -4.33785826e-01
-2.49406090e-03 4.76406872e-01 1.25013947e-01 -8.69800508e-01
4.57771629e-01 -4.99002561e-02 1.39946744e-01 -1.12695026e+00
2.08064318e-02 -1.53424010e-01 -3.08439404e-01 4.28581834e-01
-8.54897261e-01 1.54644713e-01 9.41262171e-02 7.57540345e-01
-3.40908051e-01 -5.15546739e-01 5.18110454e-01 -3.01787585e-01
-3.07625383e-01 1.50731623e-01 -3.80207270e-01 6.84944689e-01
7.93508470e-01 -1.36200273e-02 -7.17538476e-01 -9.56263840e-01
-1.17800906e-01 6.65641785e-01 3.39122027e-01 5.03410697e-01
6.15450323e-01 -1.01351571e+00 -1.31352341e+00 -1.95977479e-01
-3.30131873e-02 3.98442060e-01 3.69131804e-01 4.36871320e-01
-4.69849050e-01 3.87392133e-01 1.86758175e-01 -1.23343870e-01
-1.23333848e+00 4.68895316e-01 -7.47669637e-02 -6.24897063e-01
-6.29183769e-01 7.72739291e-01 3.66424948e-01 -1.37558892e-01
2.69229144e-01 -2.97349513e-01 -4.23351288e-01 1.95122957e-01
9.94556248e-01 5.89667678e-01 -4.48156334e-02 -5.66165566e-01
2.19976739e-03 7.65469149e-02 -4.49935228e-01 -2.04917416e-01
1.29654276e+00 -2.37351462e-01 -2.72616595e-02 4.18994665e-01
9.64144647e-01 7.94856995e-02 -1.07460546e+00 -2.11761668e-01
-2.84987867e-01 1.85300391e-02 -2.23577783e-01 -1.00506651e+00
-6.92128420e-01 7.35462129e-01 -5.30916035e-01 2.67218560e-01
1.12274837e+00 4.78643104e-02 1.45268774e+00 2.51066953e-01
-2.82759875e-01 -1.05097723e+00 3.82412732e-01 7.81510055e-01
1.45275772e+00 -8.90249491e-01 1.07538447e-01 -2.33980805e-01
-1.23788476e+00 9.37940598e-01 6.65234029e-01 -3.68534476e-01
-8.15693289e-02 3.31194215e-02 2.10688170e-02 -2.61263363e-02
-1.12238121e+00 1.30680531e-01 3.99167210e-01 4.65338916e-01
2.44761735e-01 -1.34055421e-01 -4.84465450e-01 1.31300676e+00
-6.64933383e-01 -9.63547528e-02 8.98137033e-01 8.07490528e-01
-5.98644555e-01 -7.47997463e-01 -1.20807216e-01 6.01335049e-01
-5.92730403e-01 -9.99815166e-02 -7.89465785e-01 4.12004709e-01
-5.46774268e-01 1.22529936e+00 -1.05258018e-01 -2.22444296e-01
6.13838315e-01 -5.92095479e-02 -6.24867342e-03 -9.48424995e-01
-9.32932794e-01 5.59457429e-02 2.80586779e-01 -1.95173234e-01
-2.42371093e-02 -5.75485885e-01 -1.32328928e+00 -2.39350930e-01
-2.10812598e-01 4.34802115e-01 4.42814648e-01 1.03909636e+00
4.89848733e-01 1.07460678e+00 6.95276678e-01 -7.80662417e-01
-9.73148286e-01 -1.32843316e+00 -3.00076991e-01 4.38188374e-01
4.66010720e-01 1.03148110e-01 -3.21518213e-01 2.31115162e-01]
|
[12.512392044067383, 9.416106224060059]
|
8edaef41-e018-43e3-b150-e75e4f5cba11
|
a-study-of-n-gram-and-embedding
| null | null |
https://aclanthology.org/W17-5026
|
https://aclanthology.org/W17-5026.pdf
|
A study of N-gram and Embedding Representations for Native Language Identification
|
We report on our experiments with N-gram and embedding based feature representations for Native Language Identification (NLI) as a part of the NLI Shared Task 2017 (team name: NLI-ISU). Our best performing system on the test set for written essays had a macro F1 of 0.8264 and was based on word uni, bi and trigram features. We explored n-grams covering word, character, POS and word-POS mixed representations for this task. For embedding based feature representations, we employed both word and document embeddings. We had a relatively poor performance with all embedding representations compared to n-grams, which could be because of the fact that embeddings capture semantic similarities whereas L1 differences are more stylistic in nature.
|
['Sowmya Vajjala', 'Sagnik Banerjee']
|
2017-09-01
| null | null | null |
ws-2017-9
|
['native-language-identification']
|
['natural-language-processing']
|
[-2.70177215e-01 -4.38876227e-02 -4.66892332e-01 -1.90158896e-02
-7.17934370e-01 -7.80742586e-01 9.98672187e-01 4.07909364e-01
-7.50312746e-01 4.63575572e-01 9.19978559e-01 -7.46309042e-01
-1.92743838e-01 -5.13772190e-01 4.15457599e-02 -9.19545442e-02
7.55662844e-02 4.20041561e-01 -4.30753648e-01 1.09876283e-01
5.32588422e-01 3.77662897e-01 -1.09464335e+00 -8.05709511e-02
5.95840037e-01 4.17276859e-01 -2.23728135e-01 1.11387491e+00
-5.66281438e-01 4.76494670e-01 -6.81455016e-01 -4.91326362e-01
1.03480183e-02 -1.25951946e-01 -7.98002839e-01 -5.65038204e-01
6.47744834e-01 -7.16967434e-02 -7.59228885e-01 1.16136301e+00
7.69375265e-01 2.94634223e-01 1.31444979e+00 -7.36710250e-01
-1.37821245e+00 7.26400852e-01 -3.50809276e-01 5.60505569e-01
6.68629110e-01 1.39914984e-02 1.57130384e+00 -1.23789477e+00
5.99183619e-01 1.63648784e+00 8.27037692e-01 6.58457339e-01
-1.20485771e+00 -4.89546239e-01 -1.44545719e-01 2.01906830e-01
-1.14226735e+00 -3.17128390e-01 2.99634457e-01 -7.16076374e-01
1.49179471e+00 2.06845567e-01 2.54078001e-01 1.35948098e+00
4.81426895e-01 7.89152682e-01 1.27436471e+00 -8.69969368e-01
-3.00238043e-01 5.32643318e-01 1.02887177e+00 5.54133534e-01
5.07435858e-01 2.09256902e-01 -5.97042799e-01 -3.11580867e-01
4.29502219e-01 -1.19425490e-01 -6.35138899e-02 3.77161294e-01
-1.35090137e+00 1.27501595e+00 -1.25641257e-01 7.29656935e-01
-5.24574518e-02 1.14014573e-01 5.43375075e-01 4.87994313e-01
3.30087006e-01 8.27369094e-01 -5.09820580e-01 -6.15127385e-01
-7.04476297e-01 3.16024572e-02 9.49741483e-01 5.34011245e-01
5.09561479e-01 2.04955712e-01 -5.79055965e-01 1.07689583e+00
2.15863064e-01 2.03947753e-01 1.36984050e+00 -4.84942734e-01
3.44704479e-01 5.01261950e-01 -2.27548078e-01 -9.72543895e-01
-3.26538920e-01 -3.75464894e-02 -4.68804061e-01 -6.72353059e-02
4.75422114e-01 -1.26007766e-01 -8.03298116e-01 1.40926838e+00
-3.92484605e-01 -2.92093456e-01 1.92568645e-01 5.26687086e-01
1.29837441e+00 6.23538971e-01 2.82731652e-01 2.98840731e-01
1.82801378e+00 -1.13433611e+00 -7.99414456e-01 -2.36231863e-01
1.05939329e+00 -9.92561638e-01 9.21062708e-01 1.36825293e-02
-6.64903045e-01 -6.25638843e-01 -8.26534331e-01 -1.66403905e-01
-8.82556796e-01 3.85084957e-01 6.72559559e-01 1.00067127e+00
-1.12111294e+00 5.32223403e-01 -1.53418571e-01 -6.40742600e-01
1.24305598e-01 1.48616835e-01 -8.47740293e-01 -2.33449023e-02
-1.26618326e+00 1.13361824e+00 3.49064112e-01 -6.62084997e-01
-3.62582594e-01 -8.02799344e-01 -1.11747396e+00 1.20478962e-02
-3.90637010e-01 -3.43233272e-02 6.20802283e-01 -3.99935514e-01
-1.43667531e+00 1.19669259e+00 -2.99498320e-01 -2.73377568e-01
-1.28377169e-01 -3.74343954e-02 -5.82969785e-01 -5.82621060e-02
1.66049823e-01 6.62293792e-01 5.97346127e-01 -3.41667771e-01
-2.31750965e-01 -2.93525457e-01 -1.54751658e-01 -3.48842703e-02
-9.94064331e-01 6.47032201e-01 4.04207289e-01 -8.09289336e-01
-3.06060731e-01 -7.62644410e-01 1.00383811e-01 -4.04045641e-01
-1.93413526e-01 -8.81586075e-01 4.63815570e-01 -1.08842051e+00
1.45016861e+00 -2.18551588e+00 -1.24412313e-01 -7.76332617e-02
2.45054990e-01 4.65318471e-01 -4.53028291e-01 4.72370088e-01
-3.20870847e-01 6.91902816e-01 3.22190225e-01 -4.23727363e-01
3.10385555e-01 -5.08551672e-02 -1.26247808e-01 3.90586704e-01
1.82622761e-01 1.13500464e+00 -8.60892773e-01 -4.11227107e-01
-4.57993001e-02 3.32743585e-01 -3.37349445e-01 1.17301121e-01
3.69254857e-01 -1.14113808e-01 1.20773884e-02 6.72155976e-01
2.86523461e-01 2.30436951e-01 3.90546061e-02 1.64056584e-01
-3.06765854e-01 8.60243857e-01 -7.57479489e-01 1.37579131e+00
-6.51396573e-01 1.17798114e+00 -2.88126618e-01 -8.25728416e-01
1.18831944e+00 4.85329270e-01 1.69145949e-02 -2.53952980e-01
1.72556773e-01 2.96020478e-01 5.09434998e-01 -6.59451187e-02
6.79351747e-01 1.04971476e-01 -1.00751534e-01 8.75261903e-01
4.91418064e-01 2.22032771e-01 1.39864728e-01 2.48950407e-01
1.21340263e+00 -3.07497591e-01 6.76303029e-01 -8.22707057e-01
4.69815046e-01 -2.38254845e-01 3.11056554e-01 7.71280289e-01
-5.49271822e-01 5.65374792e-01 6.43289268e-01 -2.84164339e-01
-9.31798756e-01 -9.78120267e-01 -5.45283437e-01 1.11391342e+00
-5.86965978e-01 -7.69231260e-01 -2.29756862e-01 -7.16393292e-01
2.14382693e-01 8.48321915e-01 -7.35086262e-01 -1.30379930e-01
-2.17658252e-01 -5.73574007e-01 6.41534626e-01 7.54391313e-01
-1.85386524e-01 -1.16745496e+00 -5.88694736e-02 3.18039447e-01
4.82334584e-01 -9.25646305e-01 -8.58126581e-01 4.25924510e-01
-6.15662098e-01 -9.05837417e-01 -8.67144883e-01 -1.03884792e+00
4.88393009e-01 -7.26266429e-02 1.09276712e+00 -1.12563558e-01
-6.56969368e-01 6.46311998e-01 -5.79160631e-01 -4.74792868e-01
-3.82625371e-01 3.70402902e-01 4.49202061e-01 -5.75007260e-01
1.13977385e+00 -1.62415057e-01 -5.07103652e-02 -2.28544906e-01
-4.91710603e-01 -5.92008233e-01 3.80712569e-01 1.17246890e+00
1.70873981e-02 -3.86426806e-01 4.11202818e-01 -6.38083696e-01
1.11429739e+00 -3.96732986e-01 -2.53463187e-03 2.71894723e-01
-5.87030590e-01 -1.95123684e-02 4.56738174e-01 -9.81579661e-01
-4.03145641e-01 -5.44919968e-01 -1.10491641e-01 -6.57035261e-02
-2.88705796e-01 3.74220908e-01 1.28569663e-01 -2.24854857e-01
4.38607752e-01 1.30412981e-01 2.06377096e-02 -6.06680155e-01
1.30095705e-01 1.15006399e+00 9.66700912e-02 -6.97358489e-01
6.78465784e-01 -4.71642643e-01 -4.37067509e-01 -1.22666490e+00
-6.44607723e-01 -6.09838188e-01 -7.98673749e-01 3.24964046e-01
1.07736707e+00 -8.65004897e-01 -5.64352870e-01 1.12551048e-01
-1.32700479e+00 -5.01845069e-02 -2.54444182e-01 9.25147533e-01
3.52850668e-02 5.59442401e-01 -8.14904153e-01 -8.79638374e-01
-4.91430461e-01 -1.11925483e+00 5.81957519e-01 3.11951637e-01
-9.57483292e-01 -1.60108769e+00 4.66763794e-01 1.97451040e-01
5.70619345e-01 -1.22756600e-01 1.33820915e+00 -1.43225670e+00
3.73836666e-01 -3.75587374e-01 -4.93731856e-01 5.02247572e-01
1.88314706e-01 4.47262898e-02 -9.85076129e-01 -2.33313620e-01
-4.80334997e-01 -5.20128608e-01 1.05618155e+00 1.43731818e-01
9.48351979e-01 -3.51479471e-01 -5.02444059e-02 4.72462654e-01
1.30872381e+00 -1.47083312e-01 4.94788229e-01 5.88933289e-01
8.83432865e-01 5.62709093e-01 1.14266172e-01 3.94156158e-01
1.81140155e-01 4.04351592e-01 -4.59101886e-01 4.73338842e-01
-2.82521486e-01 -2.50459790e-01 7.32545555e-01 1.08418560e+00
2.90904641e-01 -2.25460995e-03 -1.24458528e+00 1.06749797e+00
-1.23016179e+00 -8.03937435e-01 -2.17041016e-01 1.98242223e+00
9.42215085e-01 5.83323687e-02 1.21510863e-01 1.35395154e-01
5.04011393e-01 3.33300412e-01 1.16289416e-02 -1.46830249e+00
-1.59476131e-01 5.81309617e-01 6.22864544e-01 7.28204012e-01
-9.25781131e-01 1.04020572e+00 7.08590746e+00 9.86180067e-01
-7.86801815e-01 2.46594459e-01 3.36492032e-01 2.58058280e-01
-6.32715285e-01 -2.02909112e-01 -1.37415147e+00 4.94537890e-01
1.35609269e+00 -4.51706976e-01 2.24926114e-01 7.22592533e-01
-5.75235069e-01 1.22905970e-01 -1.07961798e+00 9.20285583e-01
3.55474383e-01 -1.16110551e+00 2.06946999e-01 4.28098261e-01
7.80073762e-01 -1.47696994e-02 3.56567234e-01 6.34741127e-01
2.75490403e-01 -1.62186038e+00 3.39734524e-01 2.15593010e-01
1.18459666e+00 -7.50144660e-01 8.22757006e-01 6.92620203e-02
-9.28623140e-01 -6.98082522e-02 -6.88419759e-01 -3.56283218e-01
-3.70418400e-01 4.57739294e-01 -7.27367759e-01 2.30081335e-01
2.34892949e-01 8.65983069e-01 -7.59299636e-01 5.30206382e-01
-1.88425362e-01 9.29773211e-01 1.52868247e-02 -4.99330580e-01
4.60260808e-01 -1.52426332e-01 6.84469163e-01 1.75329328e+00
1.76968694e-01 -2.74922520e-01 -9.70034208e-03 6.99597538e-01
-7.98690617e-02 3.51121873e-01 -9.05288279e-01 -7.32955456e-01
5.63372850e-01 1.38687980e+00 -4.00296986e-01 -2.65298724e-01
-7.14164674e-01 1.05870569e+00 4.53384906e-01 8.43433142e-02
-3.41922075e-01 -7.79791892e-01 1.11564612e+00 -1.07797697e-01
1.02989435e-01 -4.74671841e-01 -4.60055858e-01 -1.30125391e+00
-4.22315419e-01 -9.45742905e-01 2.76552141e-01 3.87476161e-02
-1.68840921e+00 6.01698637e-01 -2.30182797e-01 -6.34219944e-01
-3.62958044e-01 -1.56428087e+00 -9.89259303e-01 1.32536578e+00
-1.39011478e+00 -8.68846536e-01 2.59070545e-01 -5.82720386e-03
4.93246704e-01 -1.11019897e+00 1.46668744e+00 6.43759295e-02
-6.44981146e-01 1.24223006e+00 4.23543870e-01 5.62806845e-01
1.06792045e+00 -1.55341506e+00 5.96502602e-01 4.41982180e-01
5.64722657e-01 9.65686142e-01 2.49542981e-01 -3.25486332e-01
-1.27592635e+00 -7.29709029e-01 1.77885354e+00 -7.64646232e-01
1.22797382e+00 -3.37466806e-01 -7.04749286e-01 6.58619940e-01
3.64610344e-01 -1.65431648e-01 1.27412450e+00 4.93853778e-01
-4.75600898e-01 6.36361957e-01 -9.89950120e-01 5.56609690e-01
8.20129812e-01 -9.55265999e-01 -9.50703382e-01 3.16753685e-01
6.28076792e-01 2.29560897e-01 -1.15014279e+00 -6.57006204e-02
5.84334970e-01 -4.56263363e-01 1.06290054e+00 -9.04173136e-01
5.85529566e-01 3.80937606e-01 -1.10055603e-01 -1.42387617e+00
-8.72510433e-01 -3.38269502e-01 -3.71918753e-02 1.58074820e+00
4.17560816e-01 -8.97058666e-01 3.42766553e-01 5.45956373e-01
2.33079538e-01 -8.58133674e-01 -8.32113266e-01 -1.15131032e+00
7.09740996e-01 -1.83529094e-01 2.99001187e-01 1.03953290e+00
2.76488513e-01 4.22664881e-01 -2.84271110e-02 -4.93759215e-01
3.78830135e-01 -2.79136121e-01 4.11693215e-01 -1.34975350e+00
-1.29577309e-01 -9.13585246e-01 -8.67876947e-01 -5.35893440e-01
1.08306682e+00 -1.35513568e+00 -4.82327014e-01 -1.32180929e+00
3.89214158e-01 -1.28665254e-01 -2.88382500e-01 4.94103879e-01
-5.59908926e-01 2.95872271e-01 1.72617301e-01 9.60159451e-02
-1.62536353e-01 3.83616894e-01 6.97771549e-01 -4.35220778e-01
1.19912319e-01 -4.46530223e-01 -8.51596415e-01 5.68833947e-01
9.42386746e-01 -4.21657205e-01 1.24555327e-01 -5.44970512e-01
-3.74997035e-02 -6.06054664e-01 -7.00805709e-02 -7.69431293e-01
-7.91206881e-02 1.48784593e-01 5.43959081e-01 -3.19035381e-01
1.96365133e-01 -2.88545638e-01 -7.13345289e-01 3.30059588e-01
-4.66401339e-01 2.86251813e-01 2.87962586e-01 1.52982816e-01
-3.05493653e-01 -9.22467887e-01 5.02453148e-01 -6.32507056e-02
-7.22421706e-01 2.26759359e-01 -7.58474946e-01 2.27919117e-01
6.41912460e-01 -3.70688826e-01 -9.19781327e-02 -1.05054833e-01
-2.84138083e-01 -1.03277765e-01 4.55868870e-01 6.34872079e-01
5.34212589e-01 -1.50941670e+00 -9.53421235e-01 3.14581603e-01
2.34750599e-01 -1.11592817e+00 -3.16510856e-01 5.87272167e-01
-5.14169097e-01 8.81032348e-01 -2.68578649e-01 1.57991201e-01
-1.38199258e+00 2.64731705e-01 -7.21848756e-02 -5.39782941e-01
-4.27175939e-01 9.99262512e-01 -4.14131843e-02 -8.30855668e-01
3.52829278e-01 5.72093986e-02 -6.74479663e-01 4.29086566e-01
8.45361173e-01 6.39226079e-01 -3.57374638e-01 -7.76379347e-01
-7.52378762e-01 6.01482034e-01 -1.82566330e-01 -2.53244162e-01
1.09828341e+00 3.21323603e-01 -2.72510916e-01 8.12737226e-01
1.71369720e+00 2.30394706e-01 -1.20980278e-01 -5.40980734e-02
2.16853991e-01 -4.76678967e-01 1.41683683e-01 -5.10564446e-01
-2.52349466e-01 1.11533546e+00 3.12229604e-01 -4.74611744e-02
3.20365168e-02 -2.24605560e-01 8.27162981e-01 2.43640035e-01
-1.54733688e-01 -1.28450513e+00 -3.21175195e-02 1.12543631e+00
6.58257127e-01 -1.40211165e+00 4.04219106e-02 2.90045232e-01
-5.90362072e-01 1.59843576e+00 4.65485662e-01 -2.17560604e-01
7.33500242e-01 1.07196249e-01 -6.52110651e-02 8.83237198e-02
-5.67417026e-01 -1.00180440e-01 7.40931153e-01 6.99506581e-01
1.23087740e+00 4.17828560e-01 -7.79622376e-01 7.98298538e-01
-3.94921094e-01 -7.79385746e-01 3.72346431e-01 3.79966736e-01
-4.44339156e-01 -1.31714368e+00 -4.08646852e-01 8.54657829e-01
-7.20422924e-01 -4.20532733e-01 -6.45808101e-01 6.82555079e-01
-9.44761187e-02 8.65520716e-01 2.87785888e-01 -4.38230723e-01
-2.06619039e-01 6.19890213e-01 4.90252912e-01 -1.01287937e+00
-8.84289742e-01 -7.94630647e-01 1.38975114e-01 -2.38863558e-01
2.29748026e-01 -7.74415970e-01 -8.37191522e-01 -4.71960276e-01
-1.79177254e-01 6.55117109e-02 7.90356517e-01 7.15425193e-01
2.77331397e-02 4.10002887e-01 3.58395845e-01 -6.45863175e-01
-9.58334386e-01 -1.48003614e+00 -7.40674973e-01 2.15521872e-01
4.60535139e-02 -5.10146797e-01 -6.96864128e-01 -4.64911908e-01]
|
[10.452022552490234, 10.282004356384277]
|
7b97bf20-d29b-4de7-882a-8d4851216614
|
learning-equational-theorem-proving
|
2102.05547
| null |
https://arxiv.org/abs/2102.05547v1
|
https://arxiv.org/pdf/2102.05547v1.pdf
|
Learning Equational Theorem Proving
|
We develop Stratified Shortest Solution Imitation Learning (3SIL) to learn equational theorem proving in a deep reinforcement learning (RL) setting. The self-trained models achieve state-of-the-art performance in proving problems generated by one of the top open conjectures in quasigroup theory, the Abelian Inner Mapping (AIM) Conjecture. To develop the methods, we first use two simpler arithmetic rewriting tasks that share tree-structured proof states and sparse rewards with the AIM problems. On these tasks, 3SIL is shown to significantly outperform several established RL and imitation learning methods. The final system is then evaluated in a standalone and cooperative mode on the AIM problems. The standalone 3SIL-trained system proves in 60 seconds more theorems (70.2%) than the complex, hand-engineered Waldmeister system (65.5%). In the cooperative mode, the final system is combined with the Prover9 system, proving in 2 seconds what standalone Prover9 proves in 60 seconds.
|
['Josef Urban', 'Mikoláš Janota', 'Tom Heskes', 'Jelle Piepenbrock']
|
2021-02-10
| null | null | null | null |
['automated-theorem-proving', 'automated-theorem-proving']
|
['miscellaneous', 'reasoning']
|
[-3.52308713e-02 8.26388538e-01 6.91070408e-02 1.39734417e-01
-1.06083214e+00 -7.15684474e-01 6.69463813e-01 -2.69513279e-01
-1.52566954e-01 1.19874072e+00 -4.56443787e-01 -1.14000821e+00
-1.82557508e-01 -7.53747821e-01 -1.34331608e+00 -2.94201791e-01
-5.00528216e-01 6.90158606e-01 1.31255090e-01 -5.59519470e-01
2.61577129e-01 -1.69705153e-01 -1.34753466e+00 3.13330084e-01
1.20302463e+00 6.57304585e-01 4.06360328e-02 1.41907883e+00
2.15073645e-01 1.72121787e+00 -5.58775127e-01 -5.19903600e-01
3.97218525e-01 -6.24851406e-01 -1.15227509e+00 -6.60574913e-01
5.68978250e-01 -6.78340673e-01 -2.43008524e-01 9.76934135e-01
6.82587698e-02 -3.56869608e-01 1.90896660e-01 -1.80672646e+00
-5.26280940e-01 1.44597018e+00 -1.60111517e-01 -3.98910828e-02
5.05903423e-01 3.00256491e-01 1.44326603e+00 -3.13342750e-01
9.01933968e-01 1.35586476e+00 6.21322751e-01 6.27409697e-01
-1.34728670e+00 -8.03470552e-01 -2.19088331e-01 6.11303866e-01
-1.00027478e+00 -3.91842902e-01 4.58393574e-01 -1.76889926e-01
1.18122470e+00 1.07202224e-01 3.20975810e-01 1.00669312e+00
7.08182216e-01 8.64984334e-01 1.20544946e+00 -3.62006605e-01
3.01307499e-01 -2.48939672e-04 1.75237842e-02 1.34865129e+00
1.89903408e-01 2.31968850e-01 -3.03241253e-01 -2.37637684e-01
4.18669254e-01 -5.77841520e-01 -4.80898209e-02 -5.56904793e-01
-1.49973428e+00 1.03382456e+00 3.28745693e-01 2.08691552e-01
-9.94253457e-02 8.07525635e-01 6.21856153e-01 9.50081706e-01
-3.57401744e-02 1.09471011e+00 -8.20748329e-01 -2.84103900e-01
-8.76773119e-01 7.87377715e-01 1.35426950e+00 8.42627287e-01
5.61868548e-01 8.44979882e-02 6.09089583e-02 -7.63705000e-02
8.65643770e-02 7.38563180e-01 -3.49835344e-02 -1.54144359e+00
5.01868904e-01 3.23888123e-01 2.17694014e-01 -4.79141265e-01
-4.41848069e-01 -6.52561843e-01 -4.85169351e-01 2.87685275e-01
6.24519289e-01 -5.02822697e-01 -1.76321089e-01 1.92000926e+00
2.52379179e-01 3.62774998e-01 5.15689194e-01 3.55383992e-01
5.48022985e-01 6.65129423e-01 -5.68601847e-01 -2.43913636e-01
9.84029889e-01 -1.12291968e+00 -3.90143663e-01 3.49898815e-01
1.12954354e+00 -4.16446209e-01 4.87886220e-01 7.18225479e-01
-1.49780571e+00 -3.24065030e-01 -1.35171270e+00 -1.16264522e-01
-1.78903103e-01 -1.23785906e-01 1.02455831e+00 8.54250863e-02
-1.28535426e+00 1.01835883e+00 -6.83687270e-01 2.10958347e-01
2.98624009e-01 4.61951613e-01 -1.45898521e-01 -7.35728964e-02
-1.56637251e+00 1.01513100e+00 2.81558901e-01 -2.26806879e-01
-1.43675172e+00 -1.22660565e+00 -1.12785912e+00 9.85071734e-02
8.00325155e-01 -7.30985880e-01 2.00279045e+00 -6.06440008e-01
-1.81930506e+00 6.12770021e-01 2.70010948e-01 -1.09931874e+00
7.17144191e-01 -6.66701421e-02 -2.86683086e-02 3.58502597e-01
3.35741788e-01 6.52673721e-01 6.26889527e-01 -8.43573213e-01
-6.50090575e-01 1.68358967e-01 8.48755658e-01 -2.81695038e-01
6.10912681e-01 -6.25892216e-03 5.15719712e-01 5.95130064e-02
-2.40652859e-01 -8.34414065e-01 1.59519821e-01 -2.08118960e-01
-2.00843036e-01 -8.34510267e-01 6.48554385e-01 -5.88213444e-01
6.30890489e-01 -1.55955017e+00 6.25300705e-01 -1.57922968e-01
6.07496738e-01 1.28593713e-01 -1.04584947e-01 4.89693820e-01
-7.57995471e-02 7.43002743e-02 2.29702428e-01 -1.00179106e-01
7.13511050e-01 1.65058017e-01 -4.37577337e-01 3.27075571e-01
3.83143902e-01 1.27724934e+00 -1.43248439e+00 -6.22695625e-01
-7.60030523e-02 -4.09314811e-01 -1.05390656e+00 1.44050717e-01
-1.04317200e+00 9.83483940e-02 -3.02059740e-01 4.74551976e-01
5.15125096e-01 -4.01636571e-01 3.99789453e-01 4.13622111e-01
1.51433386e-02 5.97058535e-01 -8.41238201e-01 1.86365032e+00
-5.11704981e-01 6.46066844e-01 2.53347307e-01 -1.04771245e+00
4.49366480e-01 3.61861795e-01 9.08991769e-02 -6.67179823e-01
1.62888825e-01 4.54949826e-01 6.08199239e-01 -4.01714742e-01
3.10915530e-01 -5.22831082e-02 -1.89985111e-01 7.65674412e-01
3.31522107e-01 -4.71681207e-01 5.95424771e-01 7.09328353e-01
1.75962782e+00 6.23066187e-01 8.05175900e-02 -3.26475173e-01
7.45927632e-01 3.14996205e-02 4.96765435e-01 1.19029129e+00
-1.22889638e-01 -3.14204156e-01 1.38966596e+00 -4.78696316e-01
-1.13702559e+00 -9.84445572e-01 3.59328091e-01 1.09598017e+00
-1.56849384e-01 -7.16824353e-01 -7.51485765e-01 -8.61734092e-01
2.05228537e-01 9.93546724e-01 -4.78602350e-01 -2.63874769e-01
-6.01030648e-01 5.06379306e-01 1.11428332e+00 2.37575024e-01
8.39321017e-01 -1.14297843e+00 -7.23397136e-01 5.74087873e-02
-1.30140036e-01 -9.28147137e-01 1.17687762e-01 1.67458981e-01
-2.38845810e-01 -1.48415494e+00 -2.39766091e-01 -5.97264051e-01
2.66415834e-01 -8.18985105e-02 1.01865005e+00 4.48726386e-01
4.50173803e-02 6.12044670e-02 -3.37249994e-01 -3.23268354e-01
-1.16045249e+00 1.95193723e-01 1.85977861e-01 -6.94465280e-01
-1.98050097e-01 -5.26942134e-01 6.70289025e-02 -6.17651939e-02
-3.73475373e-01 4.09905165e-01 5.00616908e-01 1.18663943e+00
-2.84202039e-01 -5.90794720e-02 4.99171197e-01 -6.59221232e-01
5.29853880e-01 -3.05128545e-01 -1.22098505e+00 3.96767110e-01
-5.11216402e-01 5.50130665e-01 1.19920170e+00 -2.24019602e-01
-5.51607251e-01 -5.71820378e-01 2.49794304e-01 -4.61689860e-01
3.41083765e-01 5.46902716e-01 3.44230354e-01 -9.00767520e-02
7.76611507e-01 3.30957919e-01 4.45565313e-01 2.19303966e-01
2.14177966e-01 2.42113441e-01 5.62541187e-01 -1.14297843e+00
1.28248668e+00 -2.86410123e-01 3.86771560e-01 -1.33855328e-01
-9.86436665e-01 3.41535598e-01 -1.62804440e-01 1.94393117e-02
4.52950984e-01 -7.08092391e-01 -1.56880605e+00 2.85911888e-01
-1.36961436e+00 -1.12620676e+00 -6.19066730e-02 2.74020702e-01
-1.04008293e+00 3.80711198e-01 -8.92693460e-01 -9.61242616e-01
-5.14078557e-01 -1.26213109e+00 1.14560461e+00 -5.16355895e-02
-2.49414429e-01 -7.73116469e-01 2.55568437e-02 4.70587850e-01
3.28620493e-01 3.50244433e-01 1.21221805e+00 -7.85152197e-01
-8.27348530e-01 1.25996266e-02 -2.62550235e-01 4.45517361e-01
-4.84956235e-01 -1.35236472e-01 -5.56248188e-01 -2.80334353e-01
-1.83282942e-01 -1.01380253e+00 5.09649396e-01 -6.66476637e-02
7.97315598e-01 -3.51084113e-01 1.82916075e-01 3.74366492e-01
8.70002508e-01 -1.88934967e-01 5.17156482e-01 3.54294211e-01
7.08028749e-02 -3.50766182e-02 6.11248791e-01 1.07359856e-01
6.58734441e-01 3.13280255e-01 5.79330027e-01 2.68868595e-01
1.92459598e-01 -5.13344109e-01 8.14550579e-01 3.40686470e-01
-9.75050181e-02 4.07758325e-01 -9.65781510e-01 8.74501280e-03
-2.02376699e+00 -1.38596213e+00 -4.79584672e-02 1.50908339e+00
1.28035951e+00 5.71270466e-01 8.76662731e-02 2.52025992e-01
2.31217474e-01 -1.94908932e-01 -5.43787599e-01 -7.74650514e-01
1.89457834e-01 7.28931844e-01 3.22169691e-01 8.85866821e-01
-8.77180338e-01 1.18891513e+00 6.08670759e+00 6.68017268e-01
-7.67776310e-01 -7.51125962e-02 4.46565822e-02 1.56217232e-01
-8.17670450e-02 3.87572140e-01 -3.49029392e-01 1.08228549e-01
1.21752119e+00 -5.00044703e-01 1.11763561e+00 1.00770199e+00
-1.80493183e-02 -2.16490418e-01 -1.76199353e+00 4.44715708e-01
-6.48790225e-02 -1.73107946e+00 -5.39796889e-01 1.84712186e-01
7.05789566e-01 -7.06470162e-02 -1.47623628e-01 1.23903894e+00
7.06221223e-01 -1.01779974e+00 8.54433477e-01 3.74912500e-01
7.52053618e-01 -8.60735774e-01 9.86072063e-01 7.32452452e-01
-4.27112848e-01 -3.02599907e-01 -1.07186750e-01 -7.74303615e-01
-4.23402160e-01 -2.29884852e-02 -1.23269737e+00 7.72705257e-01
2.26595715e-01 6.28673494e-01 -4.39335674e-01 3.29858452e-01
-8.23444843e-01 6.57374322e-01 -3.01513195e-01 -6.12340271e-01
5.08140206e-01 2.86209863e-02 5.66410005e-01 6.20503068e-01
-3.94026458e-01 -4.82836068e-02 1.42813608e-01 1.45696139e+00
-3.03702444e-01 -6.16553187e-01 -5.26999235e-01 -1.51125297e-01
6.32055849e-02 1.28869355e+00 -1.33262306e-01 -5.71992040e-01
1.95722133e-01 6.68770492e-01 5.41483223e-01 -1.76872741e-02
-1.41707039e+00 -6.06556833e-01 3.08591902e-01 -3.54237586e-01
6.45437837e-01 -3.15269232e-01 1.00770697e-01 -1.36186957e+00
3.16223949e-02 -1.47381258e+00 -1.36310309e-02 -1.08560896e+00
-7.75428236e-01 2.62590796e-02 7.38968924e-02 -4.58590657e-01
-7.69525647e-01 -8.81378651e-01 -5.96370101e-01 4.36766595e-01
-1.57844579e+00 -1.12232256e+00 1.43550113e-01 4.19050753e-01
3.86284471e-01 -3.71456623e-01 1.14312577e+00 -1.70407593e-01
-4.91245031e-01 7.99704373e-01 -4.86960262e-02 2.18833998e-01
2.95313120e-01 -1.69325423e+00 2.20708147e-01 7.58629680e-01
-1.40373573e-01 7.54518867e-01 9.09626126e-01 -4.33232158e-01
-2.21732259e+00 -7.53028512e-01 8.84109676e-01 -4.33163494e-01
1.55066586e+00 -7.05214560e-01 -3.51632386e-01 7.73741066e-01
3.99951130e-01 -1.33817405e-01 6.18671142e-02 3.04883182e-01
-7.34539211e-01 -1.30521163e-01 -1.18072724e+00 8.84273350e-01
9.93036687e-01 -6.18224502e-01 -1.02118969e+00 7.86297083e-01
1.06155324e+00 -8.13997924e-01 -9.42419350e-01 2.99098659e-02
5.38786769e-01 -8.13378513e-01 4.40436035e-01 -9.82106864e-01
1.32588601e+00 -2.56979436e-01 -7.13111013e-02 -1.10998940e+00
-1.44183218e-01 -1.45634079e+00 -6.96377277e-01 6.56756997e-01
4.90314096e-01 -8.86382580e-01 4.26595628e-01 2.27534637e-01
-2.80610882e-02 -8.97177458e-01 -1.09865248e+00 -7.59882271e-01
7.72290468e-01 -2.85957664e-01 3.41914177e-01 6.84860706e-01
7.81493843e-01 7.14900911e-01 -6.58280626e-02 8.21665972e-02
6.71984315e-01 3.97960991e-01 1.25081956e+00 -1.05192780e+00
-7.21854806e-01 -4.35153961e-01 -3.20398778e-01 -6.11868024e-01
9.71893907e-01 -1.39831662e+00 -1.29680678e-01 -1.06632864e+00
1.87568903e-01 -1.37262419e-01 4.32797745e-02 7.45971859e-01
1.48509905e-01 -3.02804917e-01 2.29322150e-01 -3.56318474e-01
-1.27899754e+00 3.83571863e-01 1.11179852e+00 -2.81534493e-01
3.17761898e-01 -9.85075161e-02 -7.03471363e-01 4.90731955e-01
8.16756427e-01 -1.59595683e-01 -2.02709541e-01 -6.61493745e-03
6.82192743e-01 3.82036597e-01 7.00112998e-01 -9.21795309e-01
3.37577641e-01 -2.75747366e-02 -6.91438392e-02 -3.49232733e-01
-2.59981602e-01 -2.98023313e-01 -3.32798451e-01 1.00089729e+00
-5.33935428e-01 1.60872146e-01 2.54851043e-01 1.21376462e-01
4.00876939e-01 -2.47599125e-01 4.63741690e-01 -1.41803026e-01
-2.00755492e-01 -2.22620711e-01 -3.30498606e-01 3.38404655e-01
1.05492675e+00 3.24883103e-01 -5.97441018e-01 -5.43923259e-01
-2.66164541e-01 8.44119966e-01 2.19262376e-01 9.65493396e-02
3.48447502e-01 -6.64600194e-01 -9.75250423e-01 -1.34343371e-01
-2.75214255e-01 1.20265327e-01 -2.14865625e-01 1.32711101e+00
-7.75096178e-01 6.64600313e-01 -2.44555771e-02 -4.33056056e-01
-1.12487435e+00 7.17164576e-01 7.17047036e-01 -9.14897680e-01
-7.21435368e-01 5.40759087e-01 -3.03986073e-01 -7.35102654e-01
1.47090420e-01 -6.89866304e-01 5.06814897e-01 -7.43960321e-01
4.06182468e-01 3.23841274e-01 -1.29562318e-01 3.04333419e-01
-1.58811077e-01 -1.42550066e-01 -3.23808968e-01 -1.32011324e-01
1.49359357e+00 8.73352289e-01 -4.70433146e-01 3.11488211e-01
8.15748036e-01 -2.27719009e-01 -8.17932189e-01 -9.01256502e-02
-2.32842080e-02 3.99368137e-01 -1.61135077e-01 -9.48392332e-01
-1.85544550e-01 4.83897001e-01 -3.23349327e-01 4.95515496e-01
5.13535440e-01 -6.60237297e-02 5.43759286e-01 1.20835495e+00
7.98253000e-01 -1.01164484e+00 -9.89673659e-03 9.84624505e-01
1.03238225e+00 -1.03546262e+00 2.24594176e-01 9.68530774e-02
-3.61480802e-01 1.31488335e+00 4.85708028e-01 -6.45009041e-01
-2.05298681e-02 4.24460471e-01 -5.29523253e-01 4.36531827e-02
-1.42560422e+00 1.86599612e-01 -3.19093019e-01 3.64125252e-01
7.23339915e-02 2.31581647e-02 -1.25024274e-01 4.83722508e-01
-7.96723843e-01 1.10526398e-01 7.95051396e-01 9.28243339e-01
-1.61275133e-01 -9.97222304e-01 -3.88724536e-01 1.46721542e-01
-3.39599878e-01 -1.34099469e-01 -4.85080749e-01 1.28915107e+00
8.86372402e-02 9.27367926e-01 -2.89911538e-01 -2.89723039e-01
-1.27947003e-01 2.86677837e-01 1.09426892e+00 -4.67933923e-01
-6.65775418e-01 -5.94408751e-01 4.14426297e-01 -6.39234662e-01
-6.86783865e-02 -7.65940428e-01 -1.49231529e+00 -8.74702394e-01
-4.70095336e-01 4.39600229e-01 4.48576570e-01 1.25991666e+00
2.26938531e-01 7.52887070e-01 7.98873127e-01 -7.18177497e-01
-1.49446654e+00 -1.03020823e+00 -3.25014174e-01 -1.82405695e-01
5.95570624e-01 -5.32550931e-01 -5.94791353e-01 -3.77243161e-01]
|
[8.92287826538086, 7.067637920379639]
|
465def8d-5bb9-4f69-bc87-24097437796f
|
unsupervised-heart-rate-estimation-in
|
1708.05356
| null |
http://arxiv.org/abs/1708.05356v1
|
http://arxiv.org/pdf/1708.05356v1.pdf
|
Unsupervised Heart-rate Estimation in Wearables With Liquid States and A Probabilistic Readout
|
Heart-rate estimation is a fundamental feature of modern wearable devices. In
this paper we propose a machine intelligent approach for heart-rate estimation
from electrocardiogram (ECG) data collected using wearable devices. The novelty
of our approach lies in (1) encoding spatio-temporal properties of ECG signals
directly into spike train and using this to excite recurrently connected
spiking neurons in a Liquid State Machine computation model; (2) a novel
learning algorithm; and (3) an intelligently designed unsupervised readout
based on Fuzzy c-Means clustering of spike responses from a subset of neurons
(Liquid states), selected using particle swarm optimization. Our approach
differs from existing works by learning directly from ECG signals (allowing
personalization), without requiring costly data annotations. Additionally, our
approach can be easily implemented on state-of-the-art spiking-based
neuromorphic systems, offering high accuracy, yet significantly low energy
footprint, leading to an extended battery life of wearable devices. We
validated our approach with CARLsim, a GPU accelerated spiking neural network
simulator modeling Izhikevich spiking neurons with Spike Timing Dependent
Plasticity (STDP) and homeostatic scaling. A range of subjects are considered
from in-house clinical trials and public ECG databases. Results show high
accuracy and low energy footprint in heart-rate estimation across subjects with
and without cardiac irregularities, signifying the strong potential of this
approach to be integrated in future wearable devices.
|
['Nikil Dutt', 'Siebren Schaafsma', 'Raj Thilak Rajan', 'Jeffrey L. Krichmar', 'Willemijn Groenendaal', 'Prathyusha Adiraju', 'Paruthi Pradhapan', 'Chris Van Hoof', 'Francky Catthoor', 'Anup Das']
|
2017-07-18
| null | null | null | null |
['heart-rate-estimation']
|
['medical']
|
[ 5.46216309e-01 -2.59545505e-01 2.96876937e-01 -1.27767503e-01
-1.91018730e-01 -4.99226421e-01 -3.41386050e-02 5.14071941e-01
-6.06045485e-01 9.73137558e-01 -3.43079150e-01 1.26329333e-01
-1.61300585e-01 -6.96287215e-01 -5.99404991e-01 -8.49849224e-01
-3.48818868e-01 3.68454486e-01 2.33366072e-01 3.64642553e-02
1.81267157e-01 4.78949696e-01 -1.70736778e+00 1.19830683e-01
8.90354216e-01 1.24440134e+00 1.88692694e-03 7.62956023e-01
5.76545179e-01 1.95080861e-01 -6.10139966e-01 1.70853034e-01
2.87563428e-02 -7.49581695e-01 2.79455762e-02 -5.34829557e-01
-3.36512536e-01 3.87807429e-01 8.88274834e-02 6.37559891e-01
1.09708917e+00 -3.71487528e-01 3.43231529e-01 -9.36483920e-01
1.98174361e-02 5.39447069e-01 5.95613047e-02 4.54291105e-01
2.76892066e-01 5.42606175e-01 1.17185540e-01 -3.73249859e-01
5.07415235e-01 2.62393594e-01 1.31723452e+00 6.61131620e-01
-1.62561524e+00 -7.22776949e-01 -6.72938287e-01 -1.84508875e-01
-1.51923168e+00 -3.50987762e-01 5.85018754e-01 -3.02928388e-01
1.31042504e+00 5.75847983e-01 1.52758574e+00 1.17793632e+00
7.49854803e-01 -2.78034853e-03 1.45789325e+00 -1.16468351e-02
9.93875206e-01 -4.95583788e-02 2.38259360e-01 3.69520545e-01
5.63731253e-01 1.21720083e-01 -8.40511203e-01 -8.06097865e-01
8.22390974e-01 -1.01481400e-01 -1.44084558e-01 6.69207349e-02
-1.35205746e+00 8.99203420e-02 4.42732545e-03 3.64999443e-01
-8.16863239e-01 7.43593693e-01 4.95691746e-01 6.51964471e-02
-2.10427828e-02 4.70695555e-01 -4.21305805e-01 -4.93685663e-01
-1.04760396e+00 3.42479110e-01 1.06196892e+00 4.22024429e-01
4.03853774e-01 2.63160437e-01 -2.89630145e-01 3.28330636e-01
2.48607069e-01 7.56090283e-01 7.66168952e-01 -9.05881882e-01
-3.27475458e-01 6.64530158e-01 -7.13248327e-02 -5.44844389e-01
-9.25572157e-01 -4.36319232e-01 -1.10188699e+00 6.89390078e-02
1.13764264e-01 -2.97907978e-01 -6.28985465e-01 1.62052751e+00
3.21788117e-02 6.30975962e-01 6.88917469e-03 7.59237468e-01
5.76451242e-01 3.65459472e-01 3.35988134e-01 -4.79430825e-01
1.43336189e+00 1.35621086e-01 -5.70976913e-01 2.26717100e-01
3.24155360e-01 5.19499518e-02 7.42764354e-01 7.02314496e-01
-1.23296273e+00 -2.37397939e-01 -1.21871877e+00 4.24054474e-01
-1.41469166e-01 -5.33363782e-02 6.59518123e-01 1.13031304e+00
-1.30250967e+00 1.06406629e+00 -1.35964060e+00 -6.03555799e-01
6.56929433e-01 8.54890823e-01 2.60133147e-01 8.75730574e-01
-9.44257915e-01 6.39571846e-01 2.56351948e-01 -1.47791654e-01
-4.75692749e-01 -9.55872893e-01 -2.09565505e-01 -1.11741237e-01
-5.49553216e-01 -1.16945338e+00 5.39977372e-01 -7.63138354e-01
-1.87278688e+00 9.38636482e-01 -1.31117210e-01 -9.06202912e-01
2.55572975e-01 3.62804383e-01 -3.29326302e-01 3.94646935e-02
-4.93469298e-01 4.02876049e-01 5.40459812e-01 -8.19352150e-01
1.71499059e-01 -5.07085085e-01 -9.07043159e-01 -1.42666688e-02
-4.60971445e-01 -7.46974573e-02 8.63515958e-02 -4.90267694e-01
4.35102433e-02 -1.08115923e+00 -3.78452450e-01 3.91252637e-02
-1.88065507e-02 1.33116096e-01 2.59213328e-01 -4.62580413e-01
1.28891015e+00 -1.88164687e+00 2.12005749e-01 6.18380189e-01
2.44287029e-01 2.54335403e-01 3.32952499e-01 3.31291348e-01
1.62465915e-01 -2.07379520e-01 -5.94269454e-01 -1.70816965e-02
-2.77452558e-01 8.13976079e-02 -1.17284879e-01 6.16922081e-01
9.41805094e-02 1.10970008e+00 -6.93288267e-01 -1.63854197e-01
1.31460512e-02 7.73474038e-01 -3.09196413e-01 -1.05657421e-01
5.19971400e-02 8.09983790e-01 -2.55908638e-01 6.20259345e-01
2.88351744e-01 -2.01349363e-01 2.17410788e-01 -1.46470442e-01
-2.90011525e-01 1.79792836e-01 -1.08122170e+00 1.97919965e+00
-3.95132117e-02 8.15623701e-02 -4.08832490e-01 -8.11123192e-01
1.46325612e+00 4.11077321e-01 8.58438849e-01 -8.61855924e-01
5.04597604e-01 6.42357528e-01 8.30215961e-02 -4.44254220e-01
-3.02658021e-01 -5.78882322e-02 -7.09750056e-02 6.19208574e-01
1.77871063e-01 -9.36727896e-02 -1.21817021e-02 -3.39170307e-01
1.58595407e+00 2.69695550e-01 1.09885655e-01 -8.19501638e-01
1.03797369e-01 -1.94027856e-01 5.61337650e-01 7.04143703e-01
-2.13591993e-01 6.49658978e-01 1.56374544e-01 -6.31229758e-01
-9.40700233e-01 -1.19731140e+00 -4.55964893e-01 3.00237209e-01
-3.15522403e-02 -2.32166857e-01 -1.02441585e+00 3.46854955e-01
-3.52205746e-02 3.78967941e-01 -6.39424980e-01 -3.53204042e-01
-5.82620561e-01 -1.21462941e+00 1.24296141e+00 4.10084367e-01
2.40730152e-01 -1.40865386e+00 -1.81563127e+00 6.93313301e-01
1.08371578e-01 -7.65230179e-01 1.44785017e-01 7.70473957e-01
-1.38168001e+00 -4.40536380e-01 -6.35499418e-01 -3.85931671e-01
3.28287542e-01 -8.35130155e-01 1.03920197e+00 1.49691299e-01
-9.58721101e-01 3.68449569e-01 1.05145268e-01 -8.33797574e-01
-1.71813279e-01 -7.53794014e-02 4.01773751e-01 -3.73476744e-03
3.68811011e-01 -1.21434820e+00 -9.84793246e-01 2.07806733e-02
-5.23309112e-01 -7.04985335e-02 3.01921517e-01 4.06973273e-01
1.01679921e+00 -4.35377270e-01 7.67745137e-01 -4.83463138e-01
5.57996929e-01 -4.07112300e-01 -4.86956745e-01 -8.07327405e-03
-8.21274936e-01 -2.96245124e-02 5.90955436e-01 -6.35948420e-01
-3.20078850e-01 5.06203651e-01 -7.56596848e-02 -9.16670784e-02
-4.63205352e-02 1.31083176e-01 4.11748499e-01 -1.74084708e-01
1.19115722e+00 6.28260851e-01 3.31725299e-01 3.83904912e-02
-4.93303329e-01 3.96407872e-01 6.16490066e-01 -5.88146269e-01
2.83445604e-02 6.16418242e-01 3.90178829e-01 -7.60734677e-01
1.47031039e-01 -1.20873354e-01 -3.54364485e-01 -2.53130823e-01
6.35610580e-01 -9.34498131e-01 -1.22698045e+00 7.77515352e-01
-9.27892327e-01 -7.17429340e-01 -5.81045747e-01 3.76245350e-01
-8.23206604e-01 1.01235062e-01 -6.95894361e-01 -1.13457334e+00
-1.16136205e+00 -7.50925064e-01 8.41536343e-01 4.80002075e-01
-7.29506314e-01 -7.95937777e-01 5.55671930e-01 -1.34462968e-01
8.06130111e-01 7.97487319e-01 3.85335952e-01 -4.12783116e-01
-1.61079168e-01 -1.98584199e-01 3.32450569e-01 -4.35789227e-02
-3.17625195e-01 -1.58707589e-01 -1.12824106e+00 -2.36640260e-01
3.25460821e-01 -1.29635468e-01 5.62061846e-01 6.71632767e-01
1.08378005e+00 1.28720194e-01 -5.60770810e-01 7.70803750e-01
1.57814026e+00 1.20564632e-01 1.04231012e+00 1.75354667e-02
2.70923972e-01 1.53742269e-01 -1.18613146e-01 8.39117706e-01
3.22100729e-01 5.71234584e-01 2.21448272e-01 2.65864320e-02
3.05991434e-02 2.08286643e-01 5.24684608e-01 8.01805615e-01
-5.90109944e-01 6.87759221e-02 -1.06725037e+00 3.41191798e-01
-1.82086313e+00 -8.73162687e-01 -5.06286621e-01 2.60016632e+00
9.97756124e-01 4.21425402e-02 4.50499237e-01 3.53925020e-01
6.04284227e-01 -7.81778276e-01 -9.12547171e-01 -5.91655076e-01
-3.78162354e-01 1.02769959e+00 5.00158548e-01 -3.06332231e-01
-5.19585252e-01 1.73185751e-01 5.85511446e+00 7.00564459e-02
-1.46319532e+00 3.01046371e-01 5.51564336e-01 -5.10848224e-01
2.98286527e-02 -1.27042472e-01 -4.96924341e-01 8.82893443e-01
1.71379840e+00 -5.18275127e-02 6.09961689e-01 2.37812385e-01
3.15728724e-01 -2.31724128e-01 -9.43866551e-01 1.32766914e+00
1.17463563e-02 -1.49458587e+00 -5.48831940e-01 -3.50092649e-02
4.01075929e-01 2.96668917e-01 -1.60036787e-01 -2.37626031e-01
-5.01603007e-01 -1.13294148e+00 7.04369128e-01 1.01756585e+00
9.94022846e-01 -6.08084559e-01 4.55659181e-01 3.01437795e-01
-9.86327291e-01 -1.88815251e-01 -1.09210148e-01 -1.67288944e-01
-2.47704302e-04 9.60767269e-01 -2.66044080e-01 4.19808142e-02
1.01330531e+00 5.82988441e-01 -4.56801891e-01 1.38046432e+00
3.84164959e-01 9.31094527e-01 -9.74760354e-01 -5.64214051e-01
-4.85139012e-01 -5.20541519e-02 5.13266087e-01 1.20751309e+00
7.79376686e-01 1.70505866e-01 -4.33070213e-01 1.31708872e+00
2.65949339e-01 6.42052526e-03 -3.77327830e-01 1.23333998e-01
5.74159563e-01 1.29302955e+00 -1.30370486e+00 -1.46176681e-01
3.19018304e-01 9.34108675e-01 -2.78136313e-01 -9.87214670e-02
-9.30089176e-01 -5.21113694e-01 2.84609348e-01 4.97863024e-01
-2.16091812e-01 -3.48056480e-02 -1.02764058e+00 -8.50175381e-01
2.13490412e-01 -4.82677311e-01 9.28328112e-02 -5.97809792e-01
-9.12560463e-01 6.15634978e-01 -4.95160609e-01 -1.17457616e+00
2.24401429e-02 -2.61378527e-01 -7.38191962e-01 8.67089331e-01
-9.93966520e-01 -5.89648306e-01 -4.61810052e-01 5.48426747e-01
-1.24574594e-01 1.13223732e-01 1.33605313e+00 2.43935093e-01
-3.85897905e-01 4.06680107e-01 -1.31705552e-01 -3.99706841e-01
3.18819404e-01 -1.00167394e+00 4.40046340e-01 4.22670811e-01
-1.22060075e-01 6.53128684e-01 6.33108795e-01 -6.77591801e-01
-1.74608386e+00 -9.96065140e-01 7.30279386e-01 -4.22854304e-01
3.62688690e-01 -6.61781728e-01 -1.01060081e+00 -1.16642319e-01
2.57092454e-02 2.34180957e-01 7.70500839e-01 -3.59208465e-01
1.39070377e-01 -2.39859506e-01 -1.29952919e+00 3.77854466e-01
9.79797363e-01 -1.23785533e-01 -2.56825894e-01 1.46490172e-01
-1.24196991e-01 -3.60543102e-01 -1.15905702e+00 5.27692795e-01
8.71838748e-01 -1.04087591e+00 7.84469604e-01 1.23325862e-01
-1.83249652e-01 -4.08354372e-01 3.63451809e-01 -9.64941740e-01
-7.65524730e-02 -1.08636606e+00 -5.17818093e-01 8.23569655e-01
2.36193240e-01 -9.64498401e-01 7.23944366e-01 5.07763505e-01
-1.02202399e-02 -9.90456462e-01 -1.23017061e+00 -7.90988564e-01
-1.70088902e-01 -2.86489248e-01 3.36004555e-01 4.55770344e-01
4.21896696e-01 -1.37718335e-01 5.20945117e-02 -5.58469817e-02
8.22596550e-01 -9.04907435e-02 1.31411478e-01 -1.59032798e+00
-5.64428627e-01 -4.25931722e-01 -8.07611704e-01 -5.19025438e-02
-2.63003349e-01 -1.05306411e+00 1.38094515e-01 -1.07302618e+00
2.87347943e-01 -7.12235749e-01 -6.97310925e-01 5.34120023e-01
1.50909528e-01 9.27283704e-01 -3.01954001e-01 1.36138573e-01
-4.13390636e-01 5.08606918e-02 3.73382837e-01 4.29741234e-01
-6.18400455e-01 -1.52794689e-01 -2.82244176e-01 4.36434597e-01
1.08211792e+00 -8.74673247e-01 -8.48418698e-02 9.19420868e-02
5.88304877e-01 1.91774264e-01 7.65678644e-01 -1.86750209e+00
4.58282143e-01 4.49612081e-01 6.60578430e-01 -1.04237318e-01
2.51421511e-01 -4.73076016e-01 8.46782506e-01 1.19638300e+00
-9.20260921e-02 1.00929558e-01 3.64813924e-01 4.59784657e-01
4.52762604e-01 8.49255398e-02 8.54411840e-01 1.18723854e-01
-3.73078063e-02 1.16279185e-01 -8.05229545e-01 -1.12716086e-01
1.02724814e+00 -7.18910456e-01 -1.13367453e-01 2.71613121e-01
-8.36055934e-01 -3.74231219e-01 7.12718189e-01 5.30833825e-02
4.49081898e-01 -1.07016718e+00 -6.42325699e-01 4.33622450e-01
1.48402959e-01 -5.33672035e-01 2.34959155e-01 1.34138703e+00
-5.78457892e-01 -2.70236954e-02 -7.40842283e-01 -1.07829869e+00
-9.49710488e-01 8.91639516e-02 3.97633940e-01 2.10533552e-02
-7.11742878e-01 5.02364814e-01 -8.19596112e-01 5.61144054e-02
2.99150646e-02 -4.48564500e-01 1.97969507e-02 -3.04637700e-02
4.03467089e-01 3.30429614e-01 4.40125138e-01 -1.67742699e-01
-6.74409270e-01 7.14914680e-01 8.69946718e-01 -7.78158084e-02
1.46483016e+00 1.26446649e-01 -2.75650144e-01 1.03114510e+00
6.55818939e-01 -4.53287750e-01 -1.05869317e+00 4.31711167e-01
-3.56731638e-02 4.48809385e-01 -2.34777316e-01 -1.01331675e+00
-8.43617022e-01 6.55941546e-01 1.37100887e+00 -8.64257477e-03
1.40663254e+00 -4.47724044e-01 9.20004606e-01 3.76704365e-01
9.22905266e-01 -1.14818275e+00 -1.50219381e-01 4.43463996e-02
3.52434605e-01 -5.94038010e-01 -3.87027226e-02 -2.08822396e-02
-4.02799696e-01 1.25662041e+00 8.26736540e-02 -5.84386826e-01
6.93581283e-01 1.03530478e+00 -1.27190933e-01 -1.77399293e-01
-8.55590284e-01 1.64791048e-01 -1.76240310e-01 7.58605301e-01
4.87101555e-01 -9.82696563e-03 -6.83748782e-01 1.04810023e+00
1.17185429e-01 7.69452155e-01 5.81391335e-01 1.11371374e+00
-2.93262690e-01 -8.86841893e-01 -1.23240590e-01 7.90459871e-01
-4.80611622e-01 -1.17912725e-01 -2.38222443e-02 -5.18596694e-02
2.07200840e-01 4.95757103e-01 1.27735749e-01 -2.17519745e-01
2.63158947e-01 4.41049069e-01 7.45279253e-01 -5.04754782e-01
-1.43940628e+00 -1.03750609e-01 -4.02060866e-01 -6.30317688e-01
-2.83704519e-01 -8.42160285e-01 -1.77790844e+00 2.39814863e-01
9.96989477e-03 -2.21511140e-01 1.18196189e+00 5.96355140e-01
1.00864851e+00 6.68027759e-01 1.97663963e-01 -1.03295636e+00
-2.43069470e-01 -5.96647561e-01 -7.68771827e-01 2.80064076e-01
-1.85436636e-01 -3.43778878e-01 -2.04591945e-01 3.18593681e-01]
|
[13.892457008361816, 3.164400100708008]
|
9aff249b-94f7-4dfb-afdf-7b0d2be1c50c
|
an-ensemble-quadratic-echo-state-network-for
|
1708.05094
| null |
http://arxiv.org/abs/1708.05094v1
|
http://arxiv.org/pdf/1708.05094v1.pdf
|
An Ensemble Quadratic Echo State Network for Nonlinear Spatio-Temporal Forecasting
|
Spatio-temporal data and processes are prevalent across a wide variety of
scientific disciplines. These processes are often characterized by nonlinear
time dynamics that include interactions across multiple scales of spatial and
temporal variability. The data sets associated with many of these processes are
increasing in size due to advances in automated data measurement, management,
and numerical simulator output. Non- linear spatio-temporal models have only
recently seen interest in statistics, but there are many classes of such models
in the engineering and geophysical sciences. Tradi- tionally, these models are
more heuristic than those that have been presented in the statistics
literature, but are often intuitive and quite efficient computationally. We
show here that with fairly simple, but important, enhancements, the echo state
net- work (ESN) machine learning approach can be used to generate long-lead
forecasts of nonlinear spatio-temporal processes, with reasonable uncertainty
quantification, and at only a fraction of the computational expense of a
traditional parametric nonlinear spatio-temporal models.
|
['Christopher K. Wikle', 'Patrick L. McDermott']
|
2017-08-16
| null | null | null | null |
['spatio-temporal-forecasting']
|
['time-series']
|
[-5.14306054e-02 -5.15048563e-01 2.23665163e-01 -1.98971346e-01
-3.69434685e-01 -7.65599370e-01 1.03068578e+00 4.47743565e-01
-2.97273666e-01 9.97491896e-01 1.55930951e-01 -7.79015124e-01
-8.11723113e-01 -8.21230412e-01 -3.08424026e-01 -6.60045087e-01
-7.72641838e-01 6.04994416e-01 2.79847980e-01 -2.73349434e-01
2.58676976e-01 8.17875743e-01 -1.29120970e+00 -2.41821289e-01
6.74701571e-01 9.58518982e-01 -1.82720855e-01 9.14976120e-01
-2.24732175e-01 6.71894968e-01 -2.76358426e-01 2.47318655e-01
3.90471071e-02 -2.58398145e-01 -4.89766479e-01 2.10261159e-03
-3.49335819e-01 4.62512001e-02 -3.92652988e-01 5.08164942e-01
2.27197424e-01 5.32703519e-01 7.83151984e-01 -1.01468039e+00
-1.65454999e-01 1.85326487e-01 -6.58844173e-01 5.61031699e-01
-8.09476674e-02 2.68998146e-01 5.39853692e-01 -7.01723099e-01
1.81622967e-01 1.37814760e+00 9.09009933e-01 -3.10236663e-01
-1.64111137e+00 -4.21820402e-01 1.61610655e-02 -2.38217369e-01
-1.27261162e+00 -2.98694104e-01 4.12346780e-01 -9.10422742e-01
8.97250593e-01 2.61098653e-01 7.93982148e-01 4.70897019e-01
6.74892306e-01 1.65510982e-01 1.34569120e+00 -2.27712259e-01
2.80541092e-01 -1.59784704e-01 -9.36001241e-02 1.18079469e-01
1.56842440e-01 7.04931259e-01 -4.17967886e-01 -6.25105977e-01
8.92874122e-01 7.22555071e-02 8.56888574e-03 -1.85039565e-01
-1.11831236e+00 8.18403959e-01 5.54643571e-03 4.72650081e-01
-6.51728630e-01 2.88026005e-01 3.38989347e-01 3.99305999e-01
9.05466795e-01 4.90986317e-01 -7.65902281e-01 -4.59604561e-01
-1.27644765e+00 6.34260118e-01 9.45237398e-01 4.59299237e-01
5.34147918e-01 3.50238264e-01 2.45225102e-01 4.88245159e-01
2.29326308e-01 1.03080046e+00 2.79473774e-02 -1.04828930e+00
1.54385343e-01 1.58262104e-02 6.73142135e-01 -1.18073463e+00
-5.87133884e-01 -3.27504098e-01 -1.16444802e+00 3.01364779e-01
5.49170375e-01 -5.04414260e-01 -7.48231709e-01 1.28596652e+00
2.27410555e-01 4.01406765e-01 1.61083221e-01 4.38313723e-01
-1.83258504e-01 1.24591422e+00 1.71127930e-01 -5.77143848e-01
1.01732898e+00 -1.57455429e-01 -8.96293104e-01 -1.32142052e-01
3.63859594e-01 -7.85728991e-01 4.29882824e-01 8.09743628e-02
-1.16117287e+00 -1.89457908e-01 -3.99649709e-01 5.14727652e-01
-4.79879826e-01 -5.04901230e-01 7.27805197e-01 2.07681179e-01
-1.12030852e+00 9.53315973e-01 -1.31518245e+00 -3.13371360e-01
-1.05307981e-01 -1.34734260e-02 2.99374238e-02 3.90755564e-01
-1.15846777e+00 1.05940986e+00 2.89222673e-02 5.47679901e-01
-6.78685904e-01 -9.32295978e-01 -6.80545211e-01 1.09640613e-01
2.93559313e-01 -2.71397889e-01 1.31840169e+00 -3.93574834e-01
-1.19979525e+00 7.99875036e-02 -5.11624694e-01 -2.40703151e-01
4.34266567e-01 1.09943673e-01 -7.09654212e-01 -1.79887518e-01
5.43187037e-02 -1.36690252e-02 6.03095055e-01 -1.29331553e+00
-6.90055549e-01 -9.96040925e-02 -4.39660847e-01 1.23947717e-01
3.40953946e-01 2.71955967e-01 1.70259118e-01 -6.26815379e-01
2.99181819e-01 -9.84241545e-01 -7.55312979e-01 6.23852834e-02
-5.02944225e-03 1.04136191e-01 7.65644491e-01 -6.46480680e-01
1.09628356e+00 -1.88228846e+00 -2.95155793e-02 4.27644312e-01
-1.16784088e-01 1.87209263e-01 1.68746233e-01 9.85297918e-01
5.91034070e-02 3.87270898e-01 -8.14181864e-01 -1.76665261e-01
-2.59657145e-01 3.31291586e-01 -7.62623131e-01 5.49888074e-01
3.11790973e-01 6.43731058e-01 -9.40209985e-01 -1.02596305e-01
4.32060868e-01 2.68115014e-01 -3.76528651e-02 -1.01579845e-01
-1.59105510e-02 6.50893211e-01 -4.23696339e-01 4.46995050e-01
4.37856138e-01 -2.62081712e-01 1.04264297e-01 5.34016728e-01
-7.97930539e-01 6.69636503e-02 -1.36365128e+00 6.54920220e-01
-5.81635475e-01 8.68359566e-01 3.38354200e-01 -1.08157456e+00
6.80673242e-01 5.60428023e-01 7.32967198e-01 -4.33984965e-01
-1.62825361e-01 2.74604887e-01 2.07625180e-01 -5.80272079e-01
6.16622388e-01 -7.66778529e-01 2.03993171e-02 6.54803813e-01
-3.86837393e-01 -4.88165706e-01 3.56376767e-02 -2.03994587e-01
8.96927059e-01 8.43091030e-03 4.78426665e-01 -6.32652581e-01
2.34874681e-01 3.49286497e-01 4.84826386e-01 7.30475068e-01
-4.56736274e-02 -4.37684264e-03 3.44666600e-01 -5.31967103e-01
-1.25061953e+00 -9.09605145e-01 -4.23955768e-01 6.76100731e-01
-5.70743345e-02 3.54488231e-02 1.35412499e-01 4.65144843e-01
4.08136189e-01 5.18272579e-01 -5.95421374e-01 1.41593665e-01
-6.57951176e-01 -9.61155593e-01 4.19381142e-01 6.16376579e-01
-5.90289123e-02 -1.04568863e+00 -5.00481665e-01 8.01716745e-01
3.03023815e-01 -8.88666749e-01 9.24572125e-02 3.64581674e-01
-1.39423871e+00 -6.25683904e-01 -4.73195881e-01 -9.59397703e-02
2.47008055e-01 5.91043457e-02 1.02246916e+00 -3.98537308e-01
-2.33914956e-01 4.54535812e-01 -2.44504865e-02 -7.02508092e-01
-5.93101263e-01 -2.45294794e-01 2.50742346e-01 -2.68566199e-02
2.69396175e-02 -8.00587893e-01 -3.15698534e-01 3.18514436e-01
-9.78799403e-01 -3.77183974e-01 4.93770480e-01 6.90351009e-01
4.35556829e-01 3.05246025e-01 7.16332138e-01 -4.89737928e-01
6.76429212e-01 -9.55740631e-01 -1.00828207e+00 -6.08646274e-02
-7.06861377e-01 1.46192983e-01 5.54382741e-01 -3.45244974e-01
-1.11973298e+00 -3.96922618e-01 4.07717049e-01 -3.08901876e-01
-4.37345684e-01 1.26690912e+00 5.90436459e-01 -4.16130088e-02
5.44827580e-01 3.51407588e-01 2.07687274e-01 -4.50775683e-01
1.85414702e-01 3.11166793e-01 1.61365986e-01 -7.14308143e-01
1.09578967e+00 8.18751097e-01 5.21374941e-01 -1.24133265e+00
-4.05302227e-01 -5.38843155e-01 -5.34228861e-01 -2.96117872e-01
5.66089630e-01 -7.44611144e-01 -6.30894780e-01 5.92627108e-01
-9.98160064e-01 -6.28786922e-01 -4.31687742e-01 7.98106790e-01
-4.78483140e-01 -1.03004448e-01 -6.22971892e-01 -1.50436687e+00
3.68457884e-01 -5.64086378e-01 8.40266407e-01 8.44989158e-03
-3.56605381e-01 -1.68815899e+00 3.54080170e-01 -4.61909890e-01
9.55168247e-01 5.14892638e-01 9.37333405e-01 -1.23897888e-01
-5.89696527e-01 -2.64301956e-01 -1.17112286e-01 -3.70094851e-02
3.33399713e-01 4.42057341e-01 -6.83599949e-01 -7.59619251e-02
1.22833118e-01 1.89308614e-01 6.36511981e-01 9.36325252e-01
7.81621397e-01 -2.84591734e-01 -3.86120051e-01 3.45151752e-01
1.47889602e+00 5.84682822e-01 2.13168591e-01 1.65800706e-01
3.65650386e-01 1.01328635e+00 4.42264587e-01 5.74556887e-01
2.06861734e-01 3.37880254e-01 1.14285931e-01 -3.05495322e-01
6.91367388e-01 1.94002047e-01 -4.27375128e-03 7.41082251e-01
-4.92429286e-01 -1.11536309e-01 -1.58232665e+00 8.77795219e-01
-1.82842839e+00 -1.47228050e+00 -5.33794999e-01 2.17702675e+00
6.07997596e-01 -7.71657601e-02 -9.05847922e-02 -1.99308991e-02
5.05657732e-01 5.14959574e-01 -4.10485506e-01 -5.43855906e-01
-1.84779182e-01 1.23741545e-01 9.22560334e-01 7.65858829e-01
-1.15504026e+00 5.49838722e-01 7.84410238e+00 2.32108697e-01
-1.18768191e+00 -9.59578454e-02 4.85784501e-01 2.97830887e-02
-2.61407256e-01 1.48478225e-01 -5.55593610e-01 3.04503381e-01
1.57046998e+00 -5.30349612e-01 4.91916180e-01 3.59648168e-01
1.14413905e+00 -4.43429142e-01 -7.19843447e-01 5.61960876e-01
-5.97325385e-01 -1.41502786e+00 -4.28733587e-01 4.59575564e-01
9.60495293e-01 -1.42469741e-02 2.22206302e-03 2.51038730e-01
5.87520242e-01 -1.19144070e+00 5.21652043e-01 1.12595546e+00
5.30733347e-01 -7.16317773e-01 5.85380495e-01 4.41851586e-01
-1.50553286e+00 -2.76692599e-01 -1.80987328e-01 -5.16553342e-01
8.61471891e-01 1.02888000e+00 -1.54847771e-01 4.79033113e-01
7.05724418e-01 5.93634188e-01 1.31119505e-01 1.22149742e+00
3.29027504e-01 1.03113592e+00 -6.20577872e-01 3.83160450e-02
5.51440656e-01 -5.41062951e-01 7.71523714e-01 1.03168511e+00
7.51544297e-01 3.69196951e-01 1.38175368e-01 6.62495494e-01
5.52593887e-01 -2.57617980e-01 -8.76974523e-01 -2.36744300e-01
5.37242115e-01 9.37194347e-01 -4.47981358e-01 -3.10841203e-01
-3.58667433e-01 1.98505461e-01 -2.73917884e-01 7.18207240e-01
-6.66122437e-01 7.73019902e-03 1.06517398e+00 5.62645614e-01
2.92087831e-02 -9.62812781e-01 -3.87832880e-01 -6.63324952e-01
-1.66930884e-01 -4.61815625e-01 1.44333586e-01 -6.71103120e-01
-1.53071797e+00 2.02051654e-01 5.60584962e-01 -1.31128550e+00
-5.24062753e-01 -5.19540370e-01 -6.80040836e-01 1.41538203e+00
-1.50187969e+00 -7.42648900e-01 1.52781159e-01 2.08952367e-01
1.79302260e-01 1.22278132e-01 8.12505484e-01 1.58251810e-03
-1.23720482e-01 -7.37381160e-01 8.70317698e-01 -2.52513856e-01
3.51480752e-01 -1.30747104e+00 7.20320404e-01 7.47420013e-01
-3.45266044e-01 4.77892131e-01 1.07391965e+00 -7.68742919e-01
-1.20650220e+00 -1.27736282e+00 1.12430596e+00 -3.34033310e-01
1.36444235e+00 -1.47016281e-02 -1.28919077e+00 6.41910017e-01
-7.34504312e-02 2.12169915e-01 4.54666972e-01 2.83874512e-01
2.60716200e-01 -4.39839512e-02 -7.11996198e-01 5.02889156e-01
6.01544261e-01 -6.46200836e-01 -4.76870865e-01 2.40882307e-01
1.90494955e-01 -1.19798914e-01 -9.26657140e-01 4.31975245e-01
4.36098933e-01 -4.74044353e-01 6.53224051e-01 -7.95126379e-01
3.01643908e-01 -4.38694507e-01 -4.33651879e-02 -1.65868795e+00
-4.21251506e-01 -7.42270887e-01 1.40346056e-02 9.90131557e-01
4.60361511e-01 -8.72667313e-01 2.76236296e-01 6.51189327e-01
2.37468388e-02 -5.42166591e-01 -1.04318047e+00 -1.27123010e+00
3.95409167e-01 -6.60747945e-01 3.53581697e-01 1.04885459e+00
-2.06069440e-01 -4.73963059e-02 -4.37270105e-01 5.05938411e-01
7.24087954e-01 1.67431638e-01 3.97612572e-01 -1.75593746e+00
-2.09792107e-01 -4.74993825e-01 -2.80189008e-01 -7.48057544e-01
-1.26266688e-01 -3.18651021e-01 4.00113970e-01 -1.49005783e+00
-3.16182286e-01 -9.08191860e-01 -1.58318430e-01 1.62002370e-01
-8.74629170e-02 -2.00294763e-01 -2.00507760e-01 5.42521596e-01
3.83853465e-01 7.14587748e-01 9.91512656e-01 2.25455433e-01
-4.64774370e-01 1.29200801e-01 -7.01166987e-02 5.94304085e-01
8.54647160e-01 -6.56448781e-01 -3.47992927e-01 -1.59643933e-01
3.04805964e-01 6.64699197e-01 5.70223689e-01 -9.71880078e-01
1.75181627e-01 -7.74429500e-01 1.68234974e-01 -6.16774023e-01
4.04234350e-01 -8.08668256e-01 5.39330423e-01 3.08109492e-01
-2.03305274e-01 4.02153611e-01 6.40734076e-01 9.17998970e-01
-5.08351505e-01 3.78287405e-01 8.35067987e-01 -1.30101442e-01
-4.93892699e-01 4.85964417e-01 -1.03297579e+00 -4.52396169e-04
1.05371249e+00 -8.63083899e-02 -1.64188758e-01 -7.90021896e-01
-8.76669347e-01 2.25483015e-01 9.20092389e-02 5.53923212e-02
2.65019219e-02 -9.32665169e-01 -7.08940387e-01 -1.09043792e-01
-1.51805609e-01 -8.59089717e-02 3.71766835e-01 1.14998317e+00
-5.24907351e-01 6.32475734e-01 4.47104685e-02 -7.12178648e-01
-5.98265290e-01 3.70955467e-01 6.57182097e-01 -3.51228297e-01
-5.72488785e-01 4.38904345e-01 1.01106703e-01 -5.26933074e-01
-4.84878629e-01 -4.86703336e-01 1.39809623e-01 2.36680225e-01
4.29439157e-01 6.26775503e-01 -2.15855733e-01 -7.08333910e-01
-2.52552301e-01 4.79056537e-01 7.81064272e-01 -6.93389595e-01
1.50373197e+00 -2.91996300e-01 -1.33093297e-01 1.41118979e+00
6.40513957e-01 -2.02669591e-01 -1.53202486e+00 -2.78889060e-01
2.70457149e-01 -2.70871162e-01 1.07867852e-01 -5.71726322e-01
-6.82799041e-01 1.09343052e+00 3.87483180e-01 8.78043830e-01
6.67449415e-01 -2.20240399e-01 3.24822664e-01 2.67521232e-01
2.98206061e-01 -9.14068818e-01 -4.97769147e-01 8.51403296e-01
1.03877711e+00 -1.08357930e+00 -2.16457210e-02 -2.07479119e-01
-3.76843333e-01 1.03661132e+00 -6.70047775e-02 -2.48542473e-01
1.34507453e+00 7.54881501e-01 1.08486116e-01 -1.89932466e-01
-8.58096659e-01 -3.27162780e-02 -3.62753905e-02 2.82975554e-01
3.97639036e-01 2.48491719e-01 -1.09103270e-01 -5.58792567e-03
1.28187433e-01 -6.93966970e-02 4.79822248e-01 1.12715650e+00
-5.08507013e-01 -8.42249751e-01 -6.72440827e-01 5.82818627e-01
-3.46001148e-01 -2.05933243e-01 2.62739986e-01 9.24552500e-01
-3.09167951e-01 1.13034868e+00 4.57115203e-01 2.25422621e-01
3.03552479e-01 2.41865531e-01 -1.86374784e-01 -3.42651844e-01
-3.63404930e-01 2.40427509e-01 1.49959117e-01 -4.98410344e-01
-4.75987583e-01 -1.13974428e+00 -1.06587219e+00 -7.52367795e-01
-5.77589013e-02 2.45289534e-01 7.99235225e-01 1.32254815e+00
2.63367832e-01 6.42315030e-01 6.64812863e-01 -1.22190762e+00
-5.04454136e-01 -1.09823084e+00 -9.80870962e-01 -1.70215368e-01
4.53602910e-01 -7.45119333e-01 -6.92710459e-01 9.98866409e-02]
|
[6.580275058746338, 3.3432230949401855]
|
17dc13d0-0640-4fe9-b306-a10ba47191d6
|
detailed-region-adaptive-normalization-for
|
2109.14525
| null |
https://arxiv.org/abs/2109.14525v4
|
https://arxiv.org/pdf/2109.14525v4.pdf
|
DRAN: Detailed Region-Adaptive Normalization for Conditional Image Synthesis
|
In recent years, conditional image synthesis has attracted growing attention due to its controllability in the image generation process. Although recent works have achieved realistic results, most of them have difficulty handling fine-grained styles with subtle details. To address this problem, a novel normalization module, named Detailed Region-Adaptive Normalization~(DRAN), is proposed. It adaptively learns both fine-grained and coarse-grained style representations. Specifically, we first introduce a multi-level structure, Spatiality-aware Pyramid Pooling, to guide the model to learn coarse-to-fine features. Then, to adaptively fuse different levels of styles, we propose Dynamic Gating, making it possible to adaptively fuse different levels of styles according to different spatial regions. Finally, we collect a new makeup dataset (Makeup-Complex dataset) that contains a wide range of complex makeup styles with diverse poses and expressions. To evaluate the effectiveness and show the general use of our method, we conduct a set of experiments on makeup transfer and semantic image synthesis. Quantitative and qualitative experiments show that equipped with DRAN, simple baseline models are able to achieve promising improvements in complex style transfer and detailed texture synthesis. Both the code and the proposed dataset will be available at https://github.com/Yueming6568/DRAN-makeup.git.
|
['Xu Wang', 'Bo Peng', 'Jing Dong', 'Jingna Sun', 'Peibin Chen', 'Yueming Lyu']
|
2021-09-29
| null | null | null | null |
['texture-synthesis', 'facial-makeup-transfer']
|
['computer-vision', 'computer-vision']
|
[ 1.17230058e-01 -5.57317436e-01 -2.31872261e-01 -6.38234317e-01
-5.16507566e-01 -6.28684461e-01 5.65277517e-01 -3.55664253e-01
-2.09819257e-01 7.38589942e-01 3.39615732e-01 1.73897877e-01
2.66953796e-01 -1.05148077e+00 -8.18426728e-01 -6.52738631e-01
5.78965545e-01 9.75097492e-02 2.91480720e-01 -4.60581064e-01
2.27998301e-01 6.48398459e-01 -1.35511696e+00 6.57111943e-01
9.00494516e-01 1.05993652e+00 5.42738497e-01 4.69912678e-01
-2.56639004e-01 3.59860510e-01 -5.62875986e-01 -4.66307253e-01
2.31949523e-01 -4.88101870e-01 -5.39018929e-01 1.76839411e-01
3.85532588e-01 -3.98872614e-01 -2.44554251e-01 1.15028536e+00
7.03311861e-01 1.47022635e-01 4.82657850e-01 -1.15328550e+00
-9.58019555e-01 5.41657507e-01 -7.20283329e-01 -1.54370936e-02
1.92116752e-01 3.23628157e-01 8.66761684e-01 -9.88142967e-01
5.95455647e-01 1.56675696e+00 3.07064474e-01 6.97217882e-01
-1.28083038e+00 -1.10503304e+00 4.67374891e-01 5.28557934e-02
-1.32236159e+00 -3.17296505e-01 9.19367254e-01 -2.31812477e-01
3.79949510e-01 2.13256016e-01 6.03531659e-01 1.39817870e+00
1.26567930e-01 9.18027878e-01 1.52689791e+00 -2.27801710e-01
8.36413130e-02 4.46532182e-02 -1.35200143e-01 6.88913226e-01
-3.16746011e-02 7.36505464e-02 -5.30356050e-01 3.16145897e-01
1.33099723e+00 7.94394165e-02 -3.30394179e-01 -2.45605022e-01
-1.41004038e+00 6.88597202e-01 7.40787923e-01 2.90344596e-01
-1.18362941e-01 2.51162887e-01 2.97530353e-01 1.44996196e-01
3.36168319e-01 2.95196891e-01 -3.59929174e-01 -5.47588542e-02
-6.70449615e-01 3.78760099e-01 4.50757235e-01 1.21452582e+00
9.20591354e-01 1.40452515e-02 -5.07643938e-01 1.19289017e+00
-4.89841029e-02 4.76914048e-01 4.57979023e-01 -8.02776992e-01
5.44866085e-01 4.84023452e-01 -5.65056838e-02 -1.01472235e+00
-1.85661718e-01 -3.80863369e-01 -1.22998285e+00 2.19203949e-01
2.58713394e-01 -1.82755306e-01 -1.12477589e+00 1.86293459e+00
1.55566767e-01 -3.27625424e-02 -3.50970179e-01 1.07238722e+00
8.35388422e-01 8.55893552e-01 2.17453882e-01 2.32306734e-01
1.38747156e+00 -1.16063893e+00 -7.39827991e-01 -1.37300119e-01
4.80010584e-02 -9.53941524e-01 1.48900521e+00 2.69502938e-01
-1.09741652e+00 -9.58642006e-01 -1.00767958e+00 -2.10872442e-01
-3.37800354e-01 2.92298973e-01 6.21678233e-01 3.53284627e-01
-1.07285666e+00 5.73403716e-01 -7.28410304e-01 -3.78161937e-01
5.98415911e-01 1.95298031e-01 -3.51960778e-01 -1.97395682e-01
-1.14481962e+00 4.87339079e-01 4.71034318e-01 -1.81327835e-02
-7.69851625e-01 -6.14780486e-01 -8.41715336e-01 3.95471742e-03
3.11582834e-01 -9.19075787e-01 1.09916198e+00 -1.15162241e+00
-1.77008629e+00 7.44761765e-01 -2.49617502e-01 1.41802624e-01
5.40060043e-01 -2.63427019e-01 -3.76464188e-01 6.25096932e-02
1.43365726e-01 1.08897638e+00 1.04598296e+00 -1.35679305e+00
-5.67896605e-01 -1.05805814e-01 2.74445593e-01 2.37991363e-01
-3.69685650e-01 -1.80167891e-02 -8.11201036e-01 -1.37905586e+00
-1.57688826e-01 -8.67687762e-01 -2.05781639e-01 1.50558338e-01
-4.62801278e-01 5.68119884e-02 6.02856517e-01 -3.97914827e-01
1.32474697e+00 -2.20120955e+00 4.94841248e-01 6.79263324e-02
7.90051650e-03 2.12265715e-01 -4.25402820e-01 2.80797839e-01
-1.67166721e-02 1.82401404e-01 -3.11565816e-01 -2.43256062e-01
7.97875002e-02 2.17730328e-01 -2.75539666e-01 -1.38172612e-01
4.89817441e-01 1.06318033e+00 -7.00980425e-01 -3.22112590e-01
3.17489892e-01 4.17095780e-01 -8.05567741e-01 3.73078227e-01
-2.78384328e-01 7.30016232e-01 -7.56172776e-01 7.95332551e-01
8.58327091e-01 -2.29322627e-01 -1.44011438e-01 -5.57919800e-01
-8.85971170e-03 -2.52613217e-01 -1.12498295e+00 1.90232468e+00
-7.07367480e-01 2.41417065e-01 1.02534436e-01 -6.45408869e-01
1.04670513e+00 -2.36108117e-02 -4.00173403e-02 -7.66715527e-01
2.43943766e-01 2.95503110e-01 -2.92986810e-01 -2.19241828e-01
4.92649883e-01 -2.58932114e-01 -2.23982483e-01 1.98007375e-02
1.68723315e-01 -3.80411834e-01 1.76293656e-01 8.37381557e-02
6.71563804e-01 3.51928532e-01 2.28905350e-01 -2.84149945e-01
6.36328459e-01 -4.28662390e-01 6.87417865e-01 4.65205193e-01
-1.39487550e-01 9.90762830e-01 4.28408653e-01 -3.79313886e-01
-9.80330825e-01 -1.14493203e+00 -1.22925863e-02 1.27583313e+00
4.17254597e-01 -4.17763323e-01 -8.79095256e-01 -5.54888070e-01
-1.07435994e-01 4.83164400e-01 -8.07763457e-01 -4.47006635e-02
-6.93850458e-01 -7.72110581e-01 2.15588480e-01 6.03604794e-01
1.03014660e+00 -1.34208763e+00 -1.11018986e-01 1.40830398e-01
-2.61963397e-01 -1.11054516e+00 -8.56694996e-01 -1.49114430e-01
-6.51488245e-01 -6.02709472e-01 -1.01751602e+00 -7.09111691e-01
6.22406065e-01 8.95457566e-02 1.13243282e+00 3.64797600e-02
-1.11130677e-01 8.96807108e-03 -4.79526699e-01 -1.96487710e-01
-2.79502183e-01 2.54103452e-01 -1.53859496e-01 3.19525301e-01
-9.10138637e-02 -7.42976487e-01 -8.28407228e-01 5.63931346e-01
-1.11465776e+00 4.76594836e-01 7.89378941e-01 9.58530724e-01
7.60653734e-01 -1.81814909e-01 5.57961166e-01 -9.47698176e-01
5.85603297e-01 -2.27610156e-01 -3.99532974e-01 1.99873909e-01
-1.23562634e-01 9.92830247e-02 8.19730282e-01 -4.02063638e-01
-1.27981162e+00 -2.01410741e-01 -4.91591811e-01 -5.21117151e-01
-3.82270962e-01 8.47461522e-02 -6.95055008e-01 -1.71151325e-01
4.17880505e-01 3.35758984e-01 -2.72950202e-01 -5.94982326e-01
6.56850934e-01 4.62789148e-01 4.52102363e-01 -1.03463840e+00
9.31803286e-01 3.73864740e-01 -2.45341271e-01 -5.75394630e-01
-9.55480635e-01 8.06467608e-02 -5.86545646e-01 -1.15168296e-01
9.86785531e-01 -1.05030501e+00 -2.64357746e-01 8.76744807e-01
-8.10360730e-01 -6.59130752e-01 -1.97363272e-01 2.19397899e-02
-5.23110271e-01 1.71720058e-01 -8.18248510e-01 -7.68134966e-02
-2.03464001e-01 -1.44882691e+00 1.20845544e+00 4.45508182e-01
-1.49472177e-01 -8.79769385e-01 -2.17345104e-01 1.61851138e-01
6.33020222e-01 3.57218802e-01 8.32213461e-01 -1.69052812e-03
-6.61208987e-01 2.43570447e-01 -5.83844244e-01 5.12399971e-01
4.25979346e-01 -6.01691846e-03 -8.74534309e-01 -3.87983322e-01
-3.79768819e-01 -4.09242272e-01 9.33256745e-01 2.59822309e-01
1.68262005e+00 -1.21597245e-01 -1.14107750e-01 1.11997294e+00
1.40714455e+00 6.54099435e-02 6.65489912e-01 3.10056806e-01
1.08467197e+00 4.00935739e-01 6.95827067e-01 4.83007491e-01
3.00149143e-01 8.25855076e-01 1.11464091e-01 -3.10976058e-01
-4.87067699e-01 -4.35601294e-01 2.04405576e-01 8.15829515e-01
-1.65662304e-01 -1.96417436e-01 -3.18137527e-01 3.13368589e-01
-1.60575867e+00 -6.91681921e-01 4.08869684e-01 1.76281154e+00
1.03228116e+00 2.29261145e-01 2.03537336e-03 -1.42018586e-01
7.76424170e-01 4.65894997e-01 -4.57761258e-01 -3.97699535e-01
-2.66525060e-01 4.25824702e-01 2.40532115e-01 3.98527652e-01
-1.08608925e+00 1.36112952e+00 5.06824589e+00 1.42782342e+00
-1.46217871e+00 2.36536027e-03 9.51347113e-01 5.49757853e-02
-4.63312685e-01 -2.62850940e-01 -8.58239353e-01 7.47083247e-01
3.08134764e-01 -5.20061329e-03 4.99999225e-01 6.10238492e-01
2.37903520e-01 1.59452736e-01 -8.57989609e-01 1.11774719e+00
-1.05589837e-01 -1.35323477e+00 5.89246631e-01 -3.85782033e-01
9.49723244e-01 -3.31153303e-01 2.31671229e-01 3.77328336e-01
3.04917485e-01 -9.25054371e-01 9.55138505e-01 4.82449174e-01
1.17554796e+00 -8.07802975e-01 3.80523324e-01 -2.07955623e-03
-1.39657772e+00 1.74286440e-02 -4.18448389e-01 4.51997109e-02
9.38700140e-02 5.32468021e-01 5.03530959e-03 6.26289606e-01
8.01941931e-01 8.86403978e-01 -6.72843933e-01 6.70122802e-01
-4.78667617e-01 3.07501674e-01 -6.02305308e-02 8.36813822e-02
2.39805683e-01 -2.12598741e-01 2.39933640e-01 1.27607179e+00
4.71473724e-01 2.80108601e-01 1.92730993e-01 8.95683706e-01
-3.73270303e-01 1.28003642e-01 -2.62272507e-01 1.17993802e-01
4.91765976e-01 1.43372488e+00 -7.25848615e-01 -4.81422573e-01
-3.46184522e-01 1.37264049e+00 3.87080133e-01 5.58486402e-01
-8.92261803e-01 -4.22598809e-01 9.06661808e-01 1.13628274e-02
3.55186701e-01 -1.23769224e-01 -4.25759166e-01 -1.47166264e+00
-8.50756690e-02 -1.02579141e+00 2.12574437e-01 -8.91223371e-01
-1.39913476e+00 8.16741765e-01 -8.80551804e-03 -1.24216890e+00
2.21854866e-01 -6.63169086e-01 -6.46894455e-01 1.04588425e+00
-1.45673788e+00 -1.41476762e+00 -6.28801048e-01 7.65008867e-01
8.32593083e-01 -6.64233193e-02 5.67907929e-01 4.11933810e-01
-7.13601232e-01 8.46961975e-01 -2.35117793e-01 6.99057281e-02
9.97812808e-01 -1.07438707e+00 5.06914198e-01 7.12315023e-01
-2.10838005e-01 6.31268203e-01 4.17257190e-01 -5.42088091e-01
-1.18283379e+00 -1.37355626e+00 3.38097066e-01 -2.77402282e-01
5.26535332e-01 -5.44258237e-01 -9.70196843e-01 5.61805964e-01
1.57636270e-01 1.76610947e-01 3.02038193e-01 -2.90156126e-01
-3.40400815e-01 -3.53087276e-01 -1.01396978e+00 8.08339655e-01
1.22918665e+00 -3.29420894e-01 -2.75536984e-01 -6.80393949e-02
8.21665823e-01 -5.04752338e-01 -8.29938769e-01 5.79430699e-01
4.54030305e-01 -1.08244967e+00 9.27258313e-01 -1.75735563e-01
7.27182746e-01 -3.65341127e-01 -2.24222526e-01 -1.48758674e+00
-6.58626378e-01 -4.58111465e-01 3.04194629e-01 1.47369254e+00
1.34208292e-01 -5.19410670e-01 6.26350343e-01 1.78641379e-01
-1.50731042e-01 -9.81168211e-01 -4.30254787e-01 -4.60896164e-01
2.01181576e-01 -1.82922423e-01 8.82162988e-01 9.42820489e-01
-5.73358476e-01 2.37628117e-01 -5.21314859e-01 -1.89758648e-04
4.36122894e-01 4.41838831e-01 8.15272093e-01 -6.62092090e-01
-3.85867536e-01 -6.76121354e-01 -2.43435279e-01 -1.23572206e+00
-2.23574489e-02 -7.73988843e-01 -1.34889230e-01 -1.44214880e+00
1.75710157e-01 -5.88419974e-01 -3.84918123e-01 4.79744256e-01
-5.34795046e-01 6.35567307e-01 3.38564277e-01 2.39237975e-02
-3.78049314e-01 8.24901462e-01 1.89340043e+00 -6.42735735e-02
-6.25877082e-02 -1.77935854e-01 -8.25459123e-01 6.69521511e-01
8.52434158e-01 4.24201824e-02 -3.86893839e-01 -5.94241917e-01
-8.88032690e-02 -1.01875171e-01 3.07192206e-01 -1.00532758e+00
-3.09643894e-01 -2.68954486e-01 7.26196051e-01 -3.17304581e-01
2.16067865e-01 -5.28433323e-01 1.86667621e-01 2.24982664e-01
-3.15094352e-01 4.20527197e-02 3.37259382e-01 2.71393031e-01
-4.97948766e-01 2.40257964e-01 1.15009546e+00 -1.59765825e-01
-1.08479249e+00 7.47075021e-01 3.94478329e-02 3.51529792e-02
8.78779233e-01 -3.05709764e-02 -1.97471473e-02 -2.39091352e-01
-7.81716108e-01 1.65341049e-01 7.33677626e-01 7.33434856e-01
6.27788663e-01 -1.66191733e+00 -7.00558722e-01 3.79811496e-01
2.34353974e-01 -2.66434141e-02 5.76333106e-01 4.20257628e-01
-6.27698302e-01 1.57345347e-02 -6.68335497e-01 -4.29494172e-01
-8.99946153e-01 5.60807407e-01 1.78842053e-01 -1.72184005e-01
-5.30419111e-01 1.09937119e+00 9.89273548e-01 -4.00020987e-01
-7.10444674e-02 -5.20969808e-01 -1.02878347e-01 -1.11410044e-01
5.07524908e-01 2.65173968e-02 -2.60257959e-01 -5.62500536e-01
-7.94243440e-02 9.30576921e-01 -9.89243388e-02 7.94567466e-02
1.09577942e+00 -2.59292543e-01 -1.19011343e-01 3.07697207e-01
1.13722837e+00 6.98338300e-02 -1.67129326e+00 -2.63321102e-01
-4.57047522e-01 -7.59562016e-01 -2.34617636e-01 -8.02404583e-01
-1.34713185e+00 1.03707564e+00 5.68243325e-01 -3.12293708e-01
1.57695889e+00 -1.17886685e-01 8.65810037e-01 -5.25467731e-02
5.37374139e-01 -7.39608467e-01 2.41250783e-01 5.05304396e-01
1.22095084e+00 -1.17963982e+00 -2.72363871e-01 -6.39359355e-01
-5.99604487e-01 9.94423628e-01 8.73067200e-01 -2.44544178e-01
4.80018198e-01 2.10908055e-01 2.12306798e-01 1.97192073e-01
-5.09793460e-01 -4.76558506e-03 3.35008204e-01 3.39945018e-01
4.62018758e-01 2.55138040e-01 -1.75504655e-01 6.68713391e-01
-3.36099595e-01 1.18661135e-01 2.61597514e-01 6.76932514e-01
-2.62320012e-01 -1.46410024e+00 -2.81350821e-01 3.62647057e-01
-3.04875344e-01 -1.40346944e-01 -7.55885392e-02 6.96527123e-01
3.56238782e-01 5.90385973e-01 8.95963330e-03 -4.29152578e-01
5.80496907e-01 -3.11759382e-01 6.08291030e-01 -5.67361593e-01
-3.86512518e-01 1.15883164e-01 -1.09890997e-01 -9.37808812e-01
-2.69911289e-01 -3.22568297e-01 -9.77397025e-01 -4.35501993e-01
1.86311245e-01 -1.78975299e-01 2.18666762e-01 6.22858346e-01
2.44423881e-01 8.06072414e-01 6.22534931e-01 -1.19154465e+00
-1.07279025e-01 -8.89356315e-01 -6.57418668e-01 7.08814561e-01
1.43485695e-01 -7.07351744e-01 3.19531858e-02 7.64111727e-02]
|
[11.475278854370117, -0.7550219893455505]
|
ad080fc6-3bb8-4daa-a7a7-0ba1903aa969
|
learning-entity-representations-for-few-shot
| null | null |
https://openreview.net/forum?id=BJgum4Qgu4
|
https://openreview.net/pdf?id=BJgum4Qgu4
|
Learning Entity Representations for Few-Shot Reconstruction of Wikipedia Categories
|
Language modeling tasks, in which words are predicted on the basis of a local context, have been very effective for learning word embeddings and context dependent representations of phrases. Motivated by the observation that efforts to code
world knowledge into machine readable knowledge bases tend to be entity-centric,
we investigate the use of a fill-in-the-blank task to learn context independent representations of entities from the contexts in which those entities were mentioned.
We show that large scale training of neural models allows us to learn extremely high fidelity entity typing information, which we demonstrate with few-shot reconstruction of Wikipedia categories. Our learning approach is powerful enough
to encode specialized topics such as Giro d’Italia cyclists.
|
['Tom Kwiatkowski', 'David Weiss', 'Livio Baldini Soares', 'Nicholas FitzGerald', 'Jeffrey Ling']
|
2019-03-20
| null | null | null |
iclr-workshop-lld-2019
|
['learning-word-embeddings']
|
['methodology']
|
[-3.56789827e-01 3.51740330e-01 -4.69363242e-01 -3.99001956e-01
-5.40171802e-01 -5.23788869e-01 7.88128436e-01 8.59836876e-01
-7.48754084e-01 7.28509486e-01 7.05332816e-01 -4.00583178e-01
1.01720780e-01 -1.21560931e+00 -8.69936466e-01 -5.53303547e-02
-2.59259820e-01 4.91821617e-01 1.12776995e-01 -3.84982795e-01
3.10299575e-01 1.11750439e-01 -1.29789412e+00 4.03577089e-01
6.27416670e-01 2.73683488e-01 3.22900474e-01 5.71792960e-01
-7.67329454e-01 8.31271648e-01 -3.16347897e-01 -6.42930925e-01
-3.24632943e-01 1.48530483e-01 -9.60624576e-01 -5.07552624e-01
3.00610960e-01 -4.08130847e-02 -6.80946410e-01 6.73875749e-01
1.36898877e-02 4.03030515e-01 9.20523643e-01 -7.23820627e-01
-1.28543687e+00 7.18060911e-01 -1.03349127e-01 3.68119478e-01
2.90108174e-01 -2.50806272e-01 1.37855911e+00 -1.17848468e+00
1.07289708e+00 8.75393093e-01 8.82747054e-01 5.68986118e-01
-1.10686195e+00 -1.87094077e-01 3.64784509e-01 3.24448645e-01
-1.39618874e+00 -1.31765336e-01 4.12304997e-01 -7.03598320e-01
1.52394986e+00 -1.51979774e-01 6.43070400e-01 1.00477517e+00
1.11242734e-01 5.93235910e-01 3.56058002e-01 -7.97645271e-01
6.17108084e-02 3.78285676e-01 7.18932033e-01 9.26181436e-01
8.92154515e-01 -9.95876938e-02 -3.28995287e-01 -2.56451696e-01
5.48159480e-01 3.97404693e-02 6.53110296e-02 -4.70911324e-01
-1.08273983e+00 9.69731629e-01 6.68831587e-01 5.56427121e-01
-2.35944241e-01 5.10759890e-01 5.28609157e-01 -5.73042147e-02
4.92921799e-01 8.15692067e-01 -5.99536121e-01 -1.38410196e-01
-6.94802940e-01 2.89627761e-01 9.33320820e-01 9.87089694e-01
1.02159441e+00 -1.37557328e-01 1.53711110e-01 8.17792416e-01
1.56278372e-01 1.37501135e-01 6.15359128e-01 -5.01940131e-01
4.31325614e-01 5.54924071e-01 3.20991009e-01 -9.53861594e-01
-3.61376911e-01 -2.73300469e-01 -1.91718787e-01 -4.02430624e-01
2.03096032e-01 -2.92999953e-01 -1.00068462e+00 1.76621461e+00
-2.12132605e-03 3.08242559e-01 2.59070039e-01 3.28216106e-01
7.98380554e-01 6.72425151e-01 7.52191603e-01 3.08752745e-01
1.35258901e+00 -6.92971647e-01 -4.81446803e-01 -5.00890195e-01
1.04435217e+00 -1.50110573e-01 8.89058471e-01 -2.23400593e-01
-6.93826139e-01 -4.78877753e-01 -8.11745405e-01 -4.24846530e-01
-1.18181920e+00 2.45702211e-02 8.96062732e-01 3.79775792e-01
-6.60102963e-01 5.66524327e-01 -8.75135779e-01 -7.44205952e-01
3.69558841e-01 -1.14777513e-01 -5.40707588e-01 -7.05211014e-02
-1.38066483e+00 1.46439600e+00 8.44425023e-01 -2.58447140e-01
-6.89592361e-01 -7.53480732e-01 -1.33036149e+00 2.44009882e-01
2.84829855e-01 -7.02123702e-01 1.07004070e+00 -6.19726121e-01
-6.45561457e-01 1.00980246e+00 -3.60977501e-01 -7.04688728e-01
-2.83799708e-01 -2.63196975e-01 -5.89517355e-01 -3.01634558e-02
1.66899204e-01 5.62074006e-01 2.92379767e-01 -9.86743569e-01
-6.09014452e-01 -1.31590530e-01 4.52997118e-01 8.31484422e-02
-7.42974162e-01 -6.25440404e-02 -3.37477833e-01 -6.80761218e-01
-2.39532709e-01 -6.93038523e-01 -3.64686489e-01 -2.74462793e-02
-6.78245351e-02 -4.44865942e-01 1.18675537e-01 -8.37910414e-01
1.47420704e+00 -1.89529121e+00 1.05879150e-01 4.65286486e-02
2.76125997e-01 1.49555475e-01 -1.04811907e-01 6.63705230e-01
4.82012928e-02 3.77600104e-01 7.68754557e-02 -6.83192387e-02
1.92464530e-01 5.73037863e-01 -5.30552208e-01 5.12740612e-02
2.02071488e-01 1.13022947e+00 -1.17410553e+00 -3.70038837e-01
7.03942077e-03 4.01392490e-01 -8.07585716e-01 2.43474077e-02
-3.00715208e-01 -3.66476387e-01 -5.06675482e-01 2.63251007e-01
-6.33260459e-02 -2.92786241e-01 3.05327773e-01 -1.40791550e-01
-7.78002515e-02 7.67699778e-01 -7.99973845e-01 1.75543404e+00
-7.84576952e-01 8.10725629e-01 -4.27319914e-01 -9.40777183e-01
7.00906456e-01 4.04976904e-01 1.27890229e-01 -2.97388256e-01
-1.87853426e-01 1.16612703e-01 -3.15351397e-01 -6.22137249e-01
1.08607471e+00 -4.02748168e-01 -3.98081541e-01 3.16863030e-01
5.99041820e-01 5.57642803e-02 3.96501243e-01 5.18595099e-01
9.37278867e-01 2.31506288e-01 5.21949768e-01 -1.83138430e-01
7.46547431e-02 2.97784388e-01 4.66065407e-01 7.20238030e-01
6.01535030e-02 3.72411221e-01 3.07444036e-01 -6.67447805e-01
-1.31699753e+00 -1.01976287e+00 -1.04869448e-01 1.36816490e+00
9.21352133e-02 -8.02053630e-01 -3.81476700e-01 -6.38482213e-01
2.50177681e-01 1.19721711e+00 -9.45795178e-01 -2.62482375e-01
-5.47452450e-01 -5.06403446e-01 4.86539483e-01 9.28344965e-01
-1.84543312e-01 -9.64791715e-01 -3.27540576e-01 4.46943074e-01
3.12475897e-02 -9.44289207e-01 -1.57669187e-02 4.03212965e-01
-7.14259267e-01 -1.13720524e+00 -5.94701767e-01 -9.59001362e-01
7.59204745e-01 2.52341125e-02 1.41932225e+00 1.67648531e-02
-3.12778860e-01 6.49615645e-01 -4.33110535e-01 -5.16308486e-01
-3.33838552e-01 2.65312165e-01 1.87177524e-01 -5.66697180e-01
1.00825572e+00 -5.99508643e-01 -1.09273195e-01 -3.58126521e-01
-7.77602673e-01 -1.28536329e-01 2.68307358e-01 8.63481998e-01
3.54298383e-01 -4.01673883e-01 4.75681901e-01 -1.35336351e+00
6.63265526e-01 -9.75004733e-01 -6.53592274e-02 5.32014430e-01
-4.44433063e-01 4.53084499e-01 4.06789690e-01 -3.54666322e-01
-9.22914326e-01 -2.32106194e-01 -3.04395616e-01 -9.25225765e-03
-8.58135894e-02 1.00893736e+00 3.52947086e-01 1.57869637e-01
1.08439553e+00 2.10640699e-01 -6.39788747e-01 -5.79512775e-01
9.59785581e-01 5.49002528e-01 5.76583624e-01 -9.96576428e-01
7.06223667e-01 4.30927314e-02 -4.94380802e-01 -8.80079806e-01
-9.82256591e-01 -4.76335764e-01 -8.22447896e-01 2.18934849e-01
9.78235066e-01 -1.30841362e+00 -1.46642610e-01 -2.09427401e-01
-1.12849808e+00 -1.03669219e-01 -5.66193342e-01 4.28221703e-01
-3.27269852e-01 2.11377800e-01 -6.69245601e-01 -4.42096859e-01
-6.06078021e-02 -2.90727913e-01 5.77531517e-01 3.88926923e-01
-4.87598151e-01 -1.42574644e+00 5.01044750e-01 -1.68835148e-01
4.52967107e-01 -3.53969894e-02 1.35586333e+00 -1.06341720e+00
-2.96869427e-01 -3.51663351e-01 -5.45164570e-02 1.49471074e-01
-2.30116304e-02 -7.51300082e-02 -6.99031293e-01 -1.89360622e-02
-6.77397728e-01 -2.28525624e-01 9.11652386e-01 -1.35889754e-01
1.03958809e+00 -3.46953869e-01 -7.07957745e-01 4.26117003e-01
1.70194554e+00 -2.94759303e-01 6.03730977e-01 4.75283355e-01
7.63258159e-01 1.98071226e-01 1.25834942e-01 1.85875580e-01
6.85665131e-01 3.61742288e-01 -4.26696874e-02 4.29641962e-01
-1.37926012e-01 -8.39277208e-01 -6.76695909e-03 8.66717160e-01
-7.92683363e-02 -1.14476137e-01 -1.14788258e+00 1.36598980e+00
-1.51853967e+00 -9.85540271e-01 1.71380132e-01 2.00609922e+00
9.99557257e-01 2.41630629e-01 -2.34219998e-01 -5.74987769e-01
5.15608728e-01 2.74758786e-01 -2.31710032e-01 -5.92627406e-01
1.28124744e-01 3.69695693e-01 5.58401167e-01 5.10894358e-01
-9.57988441e-01 1.01816618e+00 7.01020956e+00 5.29040456e-01
-6.60241783e-01 3.37498426e-01 2.05002446e-02 1.02389045e-01
-7.69273043e-01 2.74093300e-01 -9.13086772e-01 5.18277764e-01
1.31338155e+00 -4.83814389e-01 7.57295638e-02 1.06961894e+00
-3.04079980e-01 5.95188662e-02 -1.25700927e+00 6.76737309e-01
1.00079037e-01 -1.73251295e+00 2.72464871e-01 -1.43454552e-01
7.32266307e-01 2.36624807e-01 -2.94165373e-01 8.19282413e-01
8.06927562e-01 -9.56339896e-01 4.64609116e-01 6.45960510e-01
6.96451664e-01 -4.90859330e-01 3.69239688e-01 3.35705489e-01
-1.06935501e+00 -1.95865393e-01 -9.31541741e-01 -1.67308658e-01
1.62531286e-01 4.43745494e-01 -9.84005988e-01 3.83909911e-01
1.87341496e-01 8.23111296e-01 -4.84592646e-01 9.67626989e-01
-5.11263728e-01 6.32459104e-01 -1.95404798e-01 -2.61323780e-01
2.43914261e-01 2.38186359e-01 3.73898566e-01 1.70879960e+00
2.05360353e-01 2.05745935e-01 -6.22490756e-02 7.98662424e-01
-4.90939289e-01 1.18359633e-01 -9.37546611e-01 -4.41330761e-01
6.40110612e-01 9.66031492e-01 -2.09787712e-01 -4.96888161e-01
-7.27538824e-01 6.03271902e-01 8.89429569e-01 2.45866165e-01
-4.92292702e-01 -6.32054925e-01 9.51109648e-01 1.74338266e-01
6.45639718e-01 -6.80054903e-01 -2.04606101e-01 -1.55874693e+00
-1.52028576e-01 -1.79816231e-01 3.97714317e-01 -9.86817479e-01
-1.30829930e+00 4.28194940e-01 2.00958103e-02 -7.19881594e-01
-4.51692492e-01 -7.79473126e-01 -7.18966186e-01 8.31250072e-01
-1.47327244e+00 -1.09106040e+00 1.01614855e-01 1.38899803e-01
4.85634327e-01 -2.45994166e-01 1.22962201e+00 1.83919981e-01
-2.16568336e-01 3.71645212e-01 4.16699260e-01 6.43432498e-01
5.19632638e-01 -1.30158973e+00 6.84063017e-01 8.01218271e-01
8.84596944e-01 1.20173526e+00 7.12122321e-01 -6.69142663e-01
-1.06950724e+00 -9.57306743e-01 1.56045830e+00 -9.81333315e-01
8.79496634e-01 -4.11271900e-01 -1.11681783e+00 1.16376948e+00
7.04488158e-02 1.29781291e-01 9.37323332e-01 7.25149751e-01
-7.66044319e-01 2.35673502e-01 -7.65517473e-01 4.61473823e-01
8.22064042e-01 -1.12459505e+00 -1.57670534e+00 4.33847457e-01
6.84379816e-01 -2.32024908e-01 -8.77789915e-01 -1.65994540e-01
4.00117010e-01 -3.97524200e-02 1.03525722e+00 -1.38093626e+00
6.24971032e-01 -3.10552251e-02 -1.30048916e-01 -1.43294168e+00
-4.03726071e-01 -7.20446184e-02 -3.96417469e-01 1.09169793e+00
6.93679154e-01 -1.94096580e-01 7.18058705e-01 1.02561128e+00
-4.37122323e-02 -4.14244920e-01 -7.05315232e-01 -4.99302715e-01
4.99753803e-01 -2.57961690e-01 3.31286252e-01 1.16077149e+00
5.46935081e-01 4.88723427e-01 -1.29470631e-01 5.30493781e-02
3.08679581e-01 1.03451451e-02 3.19548547e-01 -1.42755878e+00
-1.43081561e-01 -3.32205556e-03 -4.95996028e-01 -1.04656053e+00
4.23530161e-01 -1.22434461e+00 3.43112499e-02 -1.82144761e+00
3.68379265e-01 -5.94200134e-01 -4.87694174e-01 6.43437266e-01
-4.69164670e-01 -3.43217775e-02 8.71634483e-02 3.90062518e-02
-8.28970075e-01 2.49509841e-01 3.55667204e-01 -1.70853138e-01
3.01991403e-01 -3.64577025e-01 -8.64328206e-01 7.52275467e-01
4.48248208e-01 -5.93169391e-01 -1.36149302e-01 -8.72740865e-01
7.65657425e-01 -2.97599882e-01 2.38838959e-02 -8.73614252e-01
4.24255461e-01 -7.00410828e-02 2.62765765e-01 -1.27654061e-01
3.08724076e-01 -6.81697547e-01 -1.83572456e-01 1.27541199e-01
-7.67218053e-01 -1.00376695e-01 2.73039937e-01 9.77620006e-01
-3.20828885e-01 -7.04951346e-01 2.58210987e-01 -5.56340575e-01
-1.53602946e+00 1.42960668e-01 -4.11188573e-01 3.83507371e-01
7.30057418e-01 -3.30821276e-02 -3.23721439e-01 -1.50440007e-01
-9.79176700e-01 1.00357883e-01 4.48932916e-01 5.01960874e-01
2.57494807e-01 -1.27168500e+00 -5.46327055e-01 -1.64670795e-01
5.18725216e-01 -3.56040299e-01 5.39261661e-02 2.83732209e-02
-5.50608993e-01 6.49268806e-01 -2.76479274e-01 8.40946361e-02
-8.09898794e-01 9.12938058e-01 2.26533055e-01 -4.20460016e-01
-5.30635178e-01 1.01215124e+00 1.77999344e-02 -5.28886139e-01
1.28624827e-01 -3.58864069e-01 -3.59506577e-01 1.43200457e-01
4.76903558e-01 -1.11125179e-01 5.34474477e-02 -3.76797318e-01
-3.68938327e-01 3.02362949e-01 -2.34566405e-01 1.33476749e-01
1.65395904e+00 -1.10308431e-01 1.07049614e-01 5.35517573e-01
1.23332751e+00 1.66220635e-01 -9.14366186e-01 -4.68280524e-01
3.66448939e-01 -3.35149735e-01 -1.45636901e-01 -7.04536557e-01
-3.80996495e-01 1.02966082e+00 9.58561823e-02 7.38762990e-02
3.80820423e-01 5.27796149e-02 7.47958839e-01 9.14895713e-01
6.69749320e-01 -1.06392741e+00 -3.72274756e-01 8.50754619e-01
3.52970302e-01 -1.06338739e+00 1.39792994e-01 -8.71765614e-02
-6.28661990e-01 1.20187211e+00 4.55131412e-01 -3.46170783e-01
7.99656808e-01 2.72474736e-02 -4.43167686e-01 -1.52771205e-01
-9.42037463e-01 -3.08252752e-01 2.19649553e-01 7.83266544e-01
5.73944926e-01 1.56135559e-01 -4.14097250e-01 8.66308451e-01
1.99791297e-01 1.00591831e-01 7.26254880e-01 1.00650859e+00
-8.20421398e-01 -1.16786170e+00 1.79941908e-01 5.62189102e-01
-4.80879992e-01 -4.95038629e-01 -8.15058872e-02 8.20230544e-01
8.43630806e-02 2.89166629e-01 2.64058501e-01 -2.80792326e-01
1.83221519e-01 7.57861018e-01 3.11149418e-01 -1.26773238e+00
-5.11534333e-01 -8.00427914e-01 5.15433848e-01 -2.00210705e-01
-2.01630473e-01 -4.12100136e-01 -1.20235884e+00 -2.06238359e-01
-6.85221702e-02 5.43995321e-01 6.92487180e-01 1.06444132e+00
5.40020168e-01 3.28352660e-01 5.46934903e-02 -4.73465919e-01
-3.33213061e-01 -8.52910578e-01 -5.44171512e-01 6.56026363e-01
1.54709607e-01 -7.04115570e-01 -3.58953811e-02 2.31735036e-01]
|
[9.950591087341309, 8.735747337341309]
|
7bc12d23-3455-4629-b743-6dd74e60e502
|
explore-contextual-information-for-3d-scene
|
2210.0624
| null |
https://arxiv.org/abs/2210.06240v2
|
https://arxiv.org/pdf/2210.06240v2.pdf
|
Explore Contextual Information for 3D Scene Graph Generation
|
3D scene graph generation (SGG) has been of high interest in computer vision. Although the accuracy of 3D SGG on coarse classification and single relation label has been gradually improved, the performance of existing works is still far from being perfect for fine-grained and multi-label situations. In this paper, we propose a framework fully exploring contextual information for the 3D SGG task, which attempts to satisfy the requirements of fine-grained entity class, multiple relation labels, and high accuracy simultaneously. Our proposed approach is composed of a Graph Feature Extraction module and a Graph Contextual Reasoning module, achieving appropriate information-redundancy feature extraction, structured organization, and hierarchical inferring. Our approach achieves superior or competitive performance over previous methods on the 3DSSG dataset, especially on the relationship prediction sub-task.
|
['Xin Yang', 'BaoCai Yin', 'Qiang Zhang', 'Bokai Liu', 'Zhaoxuan Zhang', 'Chengjiang Long', 'Yuanyuan Liu']
|
2022-10-12
| null | null | null | null |
['scene-graph-generation']
|
['computer-vision']
|
[ 3.44728827e-01 3.98339421e-01 -3.98325473e-01 -4.85369623e-01
-4.76732731e-01 -3.51143219e-02 6.79264605e-01 4.71512586e-01
2.91737199e-01 5.43052375e-01 4.23414297e-02 -2.49563262e-01
-3.10608327e-01 -1.11442959e+00 -1.69663936e-01 -4.10710782e-01
5.65343946e-02 7.44917810e-01 5.35240769e-01 1.72636602e-02
1.45058572e-01 7.03668058e-01 -1.71587050e+00 1.94601402e-01
9.50071454e-01 1.11111355e+00 -4.62227548e-03 2.85044968e-01
-3.98729652e-01 9.88680184e-01 -3.15981835e-01 -5.44777632e-01
2.03184903e-01 -2.85580635e-01 -1.03493559e+00 6.72961831e-01
5.98471761e-01 1.24876663e-01 -1.57630846e-01 1.06503737e+00
3.66702825e-01 -9.62533578e-02 7.35676169e-01 -1.44531310e+00
-5.54446340e-01 5.22955060e-01 -5.81086814e-01 -8.77215192e-02
5.72176874e-01 -4.09456342e-01 1.48651350e+00 -8.59834433e-01
9.46758568e-01 1.34841216e+00 5.52770019e-01 1.22195043e-01
-9.40030336e-01 -4.24490213e-01 4.10303622e-01 3.86281878e-01
-1.57826042e+00 -1.47619788e-02 9.48921382e-01 -5.54027915e-01
1.24885464e+00 1.13870874e-01 7.93346107e-01 4.89331514e-01
1.62882302e-02 5.03714442e-01 1.16616893e+00 -5.96182764e-01
6.86186105e-02 -1.32748270e-02 4.44329143e-01 1.02889061e+00
4.80621278e-01 -2.56879956e-01 -5.41658342e-01 -3.35903168e-02
6.50686383e-01 -7.88796842e-02 1.13725103e-01 -7.14730620e-01
-8.81224215e-01 5.88376462e-01 5.73883235e-01 3.01683635e-01
-3.22863102e-01 -1.56769961e-01 2.06367895e-01 1.78917393e-01
6.74064755e-01 3.72026771e-01 -4.04999048e-01 4.48130548e-01
-5.50450683e-01 3.05752128e-01 7.84823179e-01 1.57361364e+00
9.80489850e-01 -3.00766408e-01 -2.81097323e-01 7.31937706e-01
3.10175955e-01 1.36728466e-01 -1.83358505e-01 -6.28882647e-01
5.47772408e-01 1.49707985e+00 -2.51592457e-01 -1.51696062e+00
-8.25075924e-01 -7.19709635e-01 -8.94627929e-01 -1.78407371e-01
3.39686945e-02 4.58810538e-01 -8.34640801e-01 1.49935472e+00
8.30964506e-01 8.01635236e-02 -1.16759762e-01 6.93147182e-01
1.25540221e+00 2.14769915e-01 2.71883845e-01 -3.35347712e-01
1.30029154e+00 -9.39639390e-01 -6.68814480e-01 -1.32174894e-01
9.11170602e-01 -5.92304766e-01 9.18220639e-01 1.57530773e-02
-7.21778154e-01 -6.43279672e-01 -9.37669218e-01 -4.07909125e-01
-3.84008974e-01 2.83182655e-02 1.03478670e+00 4.76056814e-01
-9.19329762e-01 2.56568760e-01 -2.97545701e-01 -7.16965318e-01
5.46275675e-01 2.72661835e-01 -5.61989605e-01 -4.05497283e-01
-9.75975037e-01 9.33123350e-01 7.91817427e-01 -2.35501081e-01
-4.41179633e-01 -4.47667569e-01 -8.65750015e-01 -3.88052780e-03
8.04620981e-01 -9.34524357e-01 6.76754117e-01 -5.05092479e-02
-9.25073743e-01 1.24373162e+00 6.45285547e-02 -2.68157065e-01
1.42268687e-01 8.44148267e-03 -3.99700344e-01 3.37200686e-02
2.33720824e-01 4.53874737e-01 5.69222152e-01 -1.21992540e+00
-8.09214175e-01 -8.03615570e-01 3.20463032e-01 6.54913664e-01
-1.20193094e-01 -2.23915607e-01 -5.53316355e-01 -4.71248627e-01
6.75803185e-01 -8.77601266e-01 -3.41638505e-01 -5.40092528e-01
-6.69514000e-01 -7.51762271e-01 7.26779759e-01 -3.08231622e-01
1.32751000e+00 -1.89745390e+00 1.24790065e-01 3.32381606e-01
5.01134217e-01 -8.75358805e-02 1.06204264e-01 4.87070024e-01
-1.29694641e-01 3.80271338e-02 -8.11395198e-02 -4.63689230e-02
6.78694099e-02 3.30663323e-02 2.13123873e-01 2.02954218e-01
3.03863257e-01 9.68179107e-01 -8.80330503e-01 -1.03002965e+00
1.06286272e-01 -2.90638618e-02 -4.94765311e-01 2.04439446e-01
-2.76409566e-01 1.48416340e-01 -8.03037345e-01 1.03004408e+00
6.52693748e-01 -6.27435267e-01 5.71119606e-01 -4.86676753e-01
2.17165172e-01 2.36558095e-01 -1.33112144e+00 1.66536605e+00
-1.18107542e-01 4.18663844e-02 -2.69834518e-01 -9.98096943e-01
1.39402676e+00 3.77123021e-02 6.41401410e-01 -5.57407677e-01
3.53875086e-02 1.08532682e-01 -2.49867737e-01 -3.93113196e-01
4.23795611e-01 -3.00052166e-02 -2.92012155e-01 2.51164675e-01
-2.67177876e-02 -4.51095313e-01 2.14561284e-01 4.65779364e-01
1.33600891e+00 2.83612937e-01 8.46335828e-01 -2.53802687e-01
5.44817269e-01 3.32023531e-01 5.57653725e-01 5.09364188e-01
-3.52564789e-02 2.71823049e-01 6.17022514e-01 -4.10183460e-01
-7.94883013e-01 -6.20913506e-01 5.74866794e-02 7.85631478e-01
6.53282702e-01 -8.55556190e-01 -5.82111061e-01 -1.18622100e+00
1.42942846e-01 5.06792963e-01 -3.74493062e-01 -1.36513263e-01
-3.45627308e-01 -6.50780737e-01 2.07382664e-01 3.06575507e-01
6.30069613e-01 -8.68268728e-01 -1.41103610e-01 2.59622216e-01
-1.56445563e-01 -1.65315807e+00 -1.04158625e-01 9.88460407e-02
-8.63807321e-01 -1.37893701e+00 1.22739479e-01 -9.62562263e-01
8.82723808e-01 6.02855504e-01 1.35590887e+00 3.72926533e-01
-2.43937999e-01 2.01557159e-01 -5.99976003e-01 -1.93443745e-01
-1.97038144e-01 3.30728173e-01 -3.06879938e-01 -7.64926746e-02
4.21819359e-01 -5.96014142e-01 -1.77523315e-01 4.34986115e-01
-6.68377280e-01 5.49871683e-01 6.59550667e-01 6.40455723e-01
8.92246783e-01 3.85117918e-01 6.02153599e-01 -1.58793020e+00
4.97436792e-01 -2.37992585e-01 -4.78942305e-01 4.66006815e-01
-1.08005321e+00 -2.68356167e-02 2.37362742e-01 2.36201942e-01
-1.25636423e+00 1.79483965e-01 -1.58931598e-01 -3.04479897e-03
-4.32637364e-01 4.80199546e-01 -6.45325184e-01 -1.77197438e-02
5.37196994e-01 -1.59216017e-01 -5.57892203e-01 -4.84627843e-01
5.12336552e-01 5.59069455e-01 2.24288449e-01 -6.42024875e-01
9.21050072e-01 2.80536652e-01 6.15069151e-01 -5.56229770e-01
-1.27240050e+00 -6.47222936e-01 -1.09637320e+00 -1.98246419e-01
8.16863894e-01 -1.00049031e+00 -3.79911572e-01 4.37204927e-01
-9.29029703e-01 7.05643892e-02 -2.35583395e-01 1.79481372e-01
-4.26326573e-01 4.60946143e-01 -4.05766755e-01 -4.27814543e-01
-1.65881947e-01 -9.29429591e-01 1.26145959e+00 6.36560237e-03
1.27467439e-02 -8.65217626e-01 -1.82661027e-01 6.90373659e-01
-8.85958001e-02 5.74129224e-01 1.35953343e+00 -7.56215215e-01
-1.07861233e+00 -2.78878063e-01 -7.01573372e-01 -1.75349593e-01
2.78859556e-01 -3.55132699e-01 -6.52147174e-01 6.33619167e-03
-2.81544805e-01 -3.24615002e-01 5.79939187e-01 -4.60676011e-03
8.50838661e-01 1.61542464e-02 -6.31517470e-01 3.19437712e-01
1.38699269e+00 1.49194956e-01 2.68045336e-01 2.10583717e-01
1.15942228e+00 8.02219093e-01 1.18326211e+00 3.42225432e-01
9.14933741e-01 8.95025015e-01 5.35190165e-01 -1.26578461e-03
-5.54352939e-01 -3.08829993e-01 -3.92762810e-01 8.35162640e-01
-1.85567886e-01 -1.53310552e-01 -1.04540527e+00 4.43442643e-01
-2.11213779e+00 -6.51809216e-01 -5.29456377e-01 1.66081142e+00
4.34754819e-01 2.72206515e-01 -1.07593209e-01 2.37196848e-01
7.90856600e-01 1.51735678e-01 -4.60572839e-01 -5.15324324e-02
-2.02788934e-01 -1.81698188e-01 1.42189741e-01 3.26649785e-01
-1.16758752e+00 1.31781828e+00 5.52734661e+00 8.47706139e-01
-2.87847370e-01 -2.37538628e-02 4.70024705e-01 5.67950010e-01
-4.12188590e-01 3.89001608e-01 -1.15210867e+00 -1.62677363e-01
2.84123480e-01 4.68452275e-02 6.25306815e-02 9.91871715e-01
-2.85643727e-01 4.59505245e-02 -1.06017244e+00 1.07988715e+00
1.47627369e-01 -1.25907278e+00 3.51162016e-01 2.38945708e-01
8.68209720e-01 -4.37395006e-01 -4.79107082e-01 2.89794743e-01
2.80218154e-01 -8.28297317e-01 4.13162708e-01 4.09160256e-01
7.24241436e-01 -6.50871873e-01 8.49864781e-01 4.59956765e-01
-1.58783889e+00 7.00507835e-02 -2.20372170e-01 -6.87717199e-02
-3.18662054e-03 9.33837712e-01 -1.10243857e+00 1.17351496e+00
6.13048196e-01 9.20645654e-01 -8.90606463e-01 7.59045124e-01
-3.77610743e-01 1.81504905e-01 -1.59266382e-01 -7.61743784e-02
-3.09250206e-02 -2.57265121e-01 3.70725453e-01 1.08173668e+00
9.37669650e-02 3.88955653e-01 6.97587550e-01 1.79912105e-01
-1.70020893e-01 5.10832787e-01 -8.56312454e-01 2.43252844e-01
5.07119834e-01 1.35586441e+00 -9.66970861e-01 -2.94938177e-01
-5.95352590e-01 5.28309166e-01 8.06067467e-01 -1.63773447e-02
-6.62894666e-01 -1.82408288e-01 3.04016799e-01 1.12169780e-01
5.59819043e-02 -2.44231626e-01 -5.38023949e-01 -1.10033607e+00
-4.95116189e-02 -6.66644394e-01 8.58806133e-01 -6.92761183e-01
-1.28189719e+00 5.87949455e-01 1.41593054e-01 -1.09798467e+00
-3.23653594e-02 -4.84110028e-01 -1.57964521e-03 3.07185769e-01
-1.48509872e+00 -1.74338508e+00 -7.10951269e-01 7.06758559e-01
4.32772189e-01 -1.36351675e-01 7.50867128e-01 3.51571321e-01
-4.95066583e-01 3.04360121e-01 -6.21987343e-01 -1.88958943e-01
4.32238787e-01 -1.25631964e+00 3.95080745e-01 7.73050964e-01
3.06631684e-01 1.80318117e-01 3.51667345e-01 -9.96383667e-01
-1.21429801e+00 -1.39714491e+00 1.24625933e+00 -3.93986583e-01
3.95865321e-01 -2.98013031e-01 -7.51010120e-01 6.14395857e-01
-2.27716863e-01 6.74412102e-02 4.63744521e-01 5.00329018e-01
-4.48595554e-01 -2.02502921e-01 -1.10151100e+00 5.26710987e-01
2.00162625e+00 -3.74834955e-01 -5.02124667e-01 4.77898121e-01
8.51409376e-01 -4.60789293e-01 -1.27057683e+00 8.46284747e-01
1.59007579e-01 -9.52094734e-01 1.04855561e+00 -4.92891252e-01
1.13862947e-01 -5.68642378e-01 -4.26351219e-01 -9.15464640e-01
-5.82055330e-01 -1.79084435e-01 -1.56388953e-01 1.31234419e+00
3.06658536e-01 -3.50941271e-01 1.02480519e+00 3.85000288e-01
-3.34200382e-01 -8.08919489e-01 -7.04247713e-01 -5.68582296e-01
-7.23572254e-01 -5.20925701e-01 9.67073739e-01 1.10416770e+00
-1.84620678e-01 1.09030700e+00 -3.58550519e-01 3.63245487e-01
9.03516233e-01 8.31477821e-01 1.03194308e+00 -1.66803706e+00
1.09673059e-02 -2.11568132e-01 -7.50867248e-01 -8.42466772e-01
1.29580647e-01 -1.24321520e+00 -2.89702207e-01 -2.14318466e+00
3.81514430e-01 -6.72014773e-01 -1.82735249e-01 6.49008989e-01
-3.78405690e-01 2.22826868e-01 1.02682896e-01 2.00272217e-01
-9.79240596e-01 4.96639520e-01 1.54168642e+00 -1.85826629e-01
-1.85258240e-01 -1.22116528e-01 -7.40954161e-01 6.92279816e-01
6.36728168e-01 -3.88630807e-01 -7.43149340e-01 -9.45213884e-02
2.95204312e-01 1.47257224e-01 2.22037718e-01 -9.65136826e-01
1.75371692e-01 -3.36097330e-01 1.98032618e-01 -9.54162598e-01
2.56981641e-01 -9.79685128e-01 5.00567257e-01 1.11563824e-01
-1.60555914e-01 -2.54958689e-01 -2.93584615e-01 6.65662229e-01
-2.44758889e-01 3.80450208e-03 5.53679883e-01 -2.07882911e-01
-1.06139266e+00 6.18140340e-01 3.12603116e-01 1.02424391e-01
1.21580613e+00 -3.75078887e-01 -4.12752748e-01 7.55607411e-02
-7.55445123e-01 1.68944702e-01 5.17175555e-01 5.01098871e-01
6.80080712e-01 -1.49480665e+00 -6.53122842e-01 6.79176971e-02
6.25773370e-01 2.94993609e-01 1.63023174e-01 5.70405424e-01
-2.86240101e-01 3.81921083e-01 -2.31302287e-02 -6.57312095e-01
-1.38086331e+00 6.42159104e-01 -1.01102665e-01 -8.32961380e-01
-7.46800303e-01 6.60090923e-01 2.46074960e-01 -5.37731469e-01
2.07691342e-02 -9.99569446e-02 -5.26661038e-01 9.19179022e-02
-1.39065251e-01 2.89260775e-01 2.59076834e-01 -6.68382943e-01
-4.76365894e-01 9.41465974e-01 -6.72812611e-02 6.10693276e-01
1.13841188e+00 -3.61024588e-01 -2.18133509e-01 2.59550661e-01
7.99174070e-01 -4.78725322e-02 -8.09318006e-01 -4.95026737e-01
4.74023134e-01 -4.69915986e-01 -1.33511230e-01 -5.74816823e-01
-7.86606371e-01 5.75330019e-01 9.05285552e-02 6.37632549e-01
1.22112024e+00 5.16498566e-01 5.34540534e-01 3.50153118e-01
8.33303988e-01 -7.76311040e-01 1.15130283e-01 5.04889071e-01
6.78145170e-01 -1.25309658e+00 5.01105964e-01 -1.27921689e+00
-6.10584974e-01 8.15277457e-01 9.09571230e-01 2.45525632e-02
8.21782351e-01 -6.89923540e-02 -2.32167318e-01 -7.23536849e-01
-6.97397351e-01 -6.60070837e-01 6.07454181e-01 7.19331622e-01
1.39115304e-01 2.30878219e-01 -3.01338345e-01 3.86248291e-01
-1.14310175e-01 -2.37998083e-01 1.03716992e-01 7.80667305e-01
-5.91058850e-01 -1.23989093e+00 1.03671677e-01 5.62802494e-01
-1.36925042e-01 5.33322245e-02 -6.56419694e-01 1.05116165e+00
3.55644941e-01 9.42953169e-01 -3.78461778e-01 -5.51598907e-01
6.53767109e-01 -7.73440301e-02 4.64943409e-01 -9.92777646e-01
-2.43853077e-01 9.87447426e-03 7.16046095e-01 -5.60281217e-01
-8.62873256e-01 -6.21407151e-01 -1.27106166e+00 4.10071947e-02
-4.90056425e-01 1.56034097e-01 3.50542933e-01 9.94541347e-01
4.97465074e-01 5.10316908e-01 6.24452829e-01 -3.06546837e-01
4.53924723e-02 -8.32670093e-01 -8.28607976e-01 6.67197227e-01
-2.52867490e-01 -9.74379361e-01 5.85389957e-02 -6.20250069e-02]
|
[10.323127746582031, 1.6714763641357422]
|
eb7824ac-5654-442e-b8ef-29d5cb50cdda
|
improving-single-image-defocus-deblurring-how
|
2108.05251
| null |
https://arxiv.org/abs/2108.05251v2
|
https://arxiv.org/pdf/2108.05251v2.pdf
|
Improving Single-Image Defocus Deblurring: How Dual-Pixel Images Help Through Multi-Task Learning
|
Many camera sensors use a dual-pixel (DP) design that operates as a rudimentary light field providing two sub-aperture views of a scene in a single capture. The DP sensor was developed to improve how cameras perform autofocus. Since the DP sensor's introduction, researchers have found additional uses for the DP data, such as depth estimation, reflection removal, and defocus deblurring. We are interested in the latter task of defocus deblurring. In particular, we propose a single-image deblurring network that incorporates the two sub-aperture views into a multi-task framework. Specifically, we show that jointly learning to predict the two DP views from a single blurry input image improves the network's ability to learn to deblur the image. Our experiments show this multi-task strategy achieves +1dB PSNR improvement over state-of-the-art defocus deblurring methods. In addition, our multi-task framework allows accurate DP-view synthesis (e.g., ~39dB PSNR) from the single input image. These high-quality DP views can be used for other DP-based applications, such as reflection removal. As part of this effort, we have captured a new dataset of 7,059 high-quality images to support our training for the DP-view synthesis task. Our dataset, code, and trained models are publicly available at https://github.com/Abdullah-Abuolaim/multi-task-defocus-deblurring-dual-pixel-nimat.
|
['Michael S. Brown', 'Mahmoud Afifi', 'Abdullah Abuolaim']
|
2021-08-11
| null | null | null | null |
['reflection-removal']
|
['computer-vision']
|
[ 5.01480997e-01 -3.78155559e-01 4.38186899e-02 -3.36802691e-01
-8.86794865e-01 -6.55545354e-01 4.66642976e-01 -8.59943926e-01
-2.05324784e-01 5.49115121e-01 6.91668928e-01 -2.44362324e-01
1.57131970e-01 -3.81898075e-01 -1.01391423e+00 -9.81074750e-01
4.80105788e-01 -1.79307103e-01 8.48687589e-02 1.90575883e-01
3.99660915e-01 3.13473970e-01 -1.45854962e+00 4.36179280e-01
7.53065348e-01 7.22861052e-01 7.50471413e-01 1.15225923e+00
7.62391746e-01 9.29025054e-01 -3.90430212e-01 -2.78868638e-02
3.39994162e-01 -2.58337677e-01 -4.98709708e-01 -1.45307574e-02
1.12553477e+00 -1.35915279e+00 -7.20149755e-01 1.13443720e+00
7.01952279e-01 3.57783772e-02 4.37212974e-01 -7.92381942e-01
-6.63105011e-01 1.20438948e-01 -7.46824563e-01 3.86355788e-01
4.18940425e-01 4.65982020e-01 6.66953743e-01 -9.35337901e-01
3.76705348e-01 1.08661628e+00 4.69688833e-01 6.92473650e-01
-1.01748180e+00 -8.33209157e-01 -4.26725328e-01 3.32918525e-01
-9.54541087e-01 -9.38291609e-01 6.67507708e-01 -4.42372024e-01
1.10040402e+00 9.20410305e-02 2.42337942e-01 1.05979538e+00
5.80931902e-01 8.14172149e-01 1.14784813e+00 -2.82086492e-01
-6.54355437e-02 -4.13359582e-01 -6.97855698e-03 3.77036840e-01
2.75681555e-01 5.76628089e-01 -5.31223476e-01 2.00483561e-01
1.09299636e+00 7.14550689e-02 -9.90449309e-01 -1.45448759e-01
-1.29843473e+00 3.46774817e-01 3.65608484e-01 4.65606712e-02
-7.06494153e-02 4.07192141e-01 -1.78238451e-01 1.84459835e-01
5.50972402e-01 6.48988008e-01 -2.94636577e-01 -2.59899646e-01
-8.72742236e-01 2.04373032e-01 4.29500848e-01 8.09037864e-01
8.17457616e-01 3.51517424e-02 -3.14899422e-02 9.49060559e-01
1.47184938e-01 8.93416822e-01 2.96983898e-01 -1.56595755e+00
4.19987947e-01 -2.63338119e-01 4.43360448e-01 -7.04624951e-01
-2.37650618e-01 -3.21122110e-01 -8.42530608e-01 1.94058120e-01
1.60622373e-01 -4.21358854e-01 -9.97495890e-01 1.70853615e+00
-1.16229784e-02 4.69948471e-01 1.36274368e-01 1.43279207e+00
6.44972920e-01 7.78501093e-01 -8.13239932e-01 -1.96686566e-01
1.07183897e+00 -1.23738027e+00 -6.80727005e-01 -5.76519847e-01
3.52630764e-01 -1.00341558e+00 6.87244713e-01 7.41123080e-01
-1.22485757e+00 -5.64937294e-01 -1.10882926e+00 -2.86395848e-01
3.08663011e-01 2.45338485e-01 4.91847992e-01 3.47328931e-01
-1.39286733e+00 1.70655325e-01 -8.42817187e-01 -6.35000020e-02
3.62106860e-01 1.64227650e-01 -2.77407497e-01 -8.04284453e-01
-1.00771737e+00 8.23498130e-01 -2.17952803e-02 -1.68386012e-01
-1.38261771e+00 -8.81690323e-01 -7.21433699e-01 -8.24138075e-02
2.77700067e-01 -9.22540605e-01 1.46320367e+00 -6.78544283e-01
-1.27242982e+00 7.39538610e-01 -3.81110638e-01 -2.97837883e-01
1.90638050e-01 -6.08202696e-01 -2.48369873e-01 4.56637770e-01
5.99725060e-02 7.10509002e-01 1.28187191e+00 -1.39002478e+00
-6.77599072e-01 -2.23458916e-01 1.43415913e-01 4.29611862e-01
7.14082569e-02 -8.69354159e-02 -4.29536641e-01 -6.40672386e-01
4.99277227e-02 -8.02821994e-01 2.27253795e-01 -4.59656380e-02
-1.59409806e-01 4.59513813e-01 9.60703433e-01 -7.90650725e-01
9.77383614e-01 -2.34194207e+00 1.79043397e-01 -8.39327931e-01
5.00953078e-01 2.82008380e-01 -1.19827874e-01 1.63775578e-01
-2.65932769e-01 -2.89133906e-01 -1.94565982e-01 -5.10370493e-01
-5.00881076e-01 -9.94628593e-02 -3.30580562e-01 6.61310911e-01
-5.67197390e-02 6.93586767e-01 -9.22174335e-01 2.35872939e-01
3.63759100e-01 4.11096931e-01 -7.41098642e-01 5.44438243e-01
-1.79487541e-01 6.62009776e-01 9.57439318e-02 6.33253813e-01
1.12083864e+00 -3.51435542e-01 -2.52928197e-01 -7.79784799e-01
-2.39851817e-01 6.46660030e-02 -7.14607954e-01 1.91429532e+00
-4.97345477e-01 1.10903907e+00 4.67663974e-01 -5.94767988e-01
3.32891375e-01 2.92742431e-01 5.33319235e-01 -5.41578352e-01
1.12555780e-01 3.85331661e-01 -6.91185147e-02 -6.82825208e-01
7.09330320e-01 -2.25875527e-01 3.31605822e-01 6.63634121e-01
-4.41612191e-02 -4.78875965e-01 -3.51111516e-02 5.75365685e-02
1.25975358e+00 4.43639979e-03 6.27387129e-03 -4.56230752e-02
2.40697980e-01 -2.91537344e-01 2.78831989e-01 7.33163774e-01
-2.57923961e-01 1.23708367e+00 -1.07987076e-01 -2.84721494e-01
-1.15000558e+00 -1.07805228e+00 -8.67126659e-02 5.69350004e-01
4.66362596e-01 4.71391678e-02 -6.78934157e-01 -2.99146622e-01
-9.92403701e-02 8.86630177e-01 -1.39209598e-01 -1.81039438e-01
-6.16605043e-01 -6.35396838e-01 3.28346938e-01 1.03008017e-01
8.44249427e-01 -4.51893955e-01 -7.17857480e-01 -4.41920310e-02
-6.89977407e-01 -1.25892448e+00 -9.26667750e-01 8.52256939e-02
-6.63800240e-01 -1.06542993e+00 -1.01531935e+00 -4.70879257e-01
4.43721414e-01 1.15808439e+00 9.68592227e-01 -4.46905464e-01
7.43745302e-04 6.02904677e-01 -1.70129612e-01 9.33200419e-02
-3.22816700e-01 -3.90541255e-01 1.21431686e-01 -4.61610071e-02
2.32864469e-01 -7.50331700e-01 -9.73039567e-01 3.61510396e-01
-1.13596737e+00 4.09326136e-01 6.79551423e-01 8.58540773e-01
1.44653693e-01 -5.32427058e-02 1.42679393e-01 -3.44269156e-01
4.32650715e-01 -3.23127836e-01 -6.72628284e-01 -1.42376855e-01
-5.51637948e-01 -2.17154399e-02 3.09412897e-01 -4.22431856e-01
-1.38121200e+00 -2.61404335e-01 1.76523533e-02 -9.17184234e-01
-1.88448355e-01 2.15924770e-01 -6.78951070e-02 -2.41193727e-01
7.35209703e-01 3.58126521e-01 9.66805071e-02 -3.75933796e-01
1.75753728e-01 1.02855384e+00 9.05103445e-01 -2.85769314e-01
4.70067382e-01 5.85348666e-01 -6.73931688e-02 -6.93156362e-01
-1.01223242e+00 -5.44607043e-01 -1.08043544e-01 -3.64697605e-01
8.74208331e-01 -1.50354791e+00 -5.70973814e-01 1.23034811e+00
-1.25887895e+00 -5.53180575e-01 2.60698289e-01 7.52115309e-01
-6.57706201e-01 6.45607829e-01 -7.36877739e-01 -3.30906928e-01
-1.91716790e-01 -1.35525310e+00 1.19670331e+00 3.87498915e-01
3.31009001e-01 -7.72765577e-01 9.24335122e-02 8.43197405e-01
6.08032763e-01 -3.41668874e-01 4.74049211e-01 2.08706185e-01
-1.02809525e+00 2.57868737e-01 -4.70194429e-01 7.39355206e-01
4.85324621e-01 -4.33039606e-01 -1.33856785e+00 -7.50709414e-01
5.73156834e-01 -3.57046068e-01 1.20400333e+00 8.90589654e-01
9.51661348e-01 -2.36575156e-01 -2.35745147e-01 1.24545264e+00
1.48585880e+00 2.82029778e-01 8.87050331e-01 1.62250027e-01
8.78674924e-01 6.52035624e-02 6.04127467e-01 3.68770868e-01
3.56047362e-01 6.50504589e-01 7.91859508e-01 -2.60325707e-02
-6.49666369e-01 3.59799676e-02 5.06773710e-01 4.60449785e-01
1.58686548e-01 -6.73569798e-01 -7.29838490e-01 5.63987076e-01
-1.50422728e+00 -9.68508840e-01 3.00194994e-02 2.15632415e+00
1.03362465e+00 -2.54460067e-01 -4.93390769e-01 -3.65034729e-01
6.82723582e-01 6.53276563e-01 -8.11949551e-01 6.23561405e-02
-3.11360598e-01 -2.51827747e-01 7.45622516e-01 1.04865503e+00
-9.69013333e-01 8.13649416e-01 5.62727833e+00 5.38777769e-01
-1.39896560e+00 1.31809220e-01 6.33225262e-01 -3.97190273e-01
-3.33306193e-01 -9.59398597e-02 -9.52924490e-01 6.67132139e-01
7.25576818e-01 2.02255085e-01 9.31810141e-01 2.60761946e-01
4.24711287e-01 -6.60079360e-01 -1.17030787e+00 1.52048600e+00
4.85759318e-01 -1.29109919e+00 -1.59793064e-01 1.47384822e-01
1.05097556e+00 4.27985162e-01 2.76025712e-01 -2.94428557e-01
1.78586558e-01 -8.79354179e-01 4.30878431e-01 6.85717940e-01
1.21219754e+00 -3.84681672e-01 4.32462841e-01 3.59105259e-01
-4.39844728e-01 -1.55123264e-01 -2.71710515e-01 -4.91089486e-02
3.16294044e-01 9.21252370e-01 -6.53879941e-01 5.15110493e-01
8.96211386e-01 1.08785462e+00 -1.47350341e-01 9.82954144e-01
-9.52248201e-02 5.14586687e-01 -1.55893296e-01 6.29298806e-01
-1.42968461e-01 -1.14361055e-01 8.94566059e-01 8.31536293e-01
7.07158446e-01 3.61475110e-01 -4.66627628e-01 9.03242528e-01
-3.72293331e-02 -1.03051519e+00 -5.66380441e-01 1.42213151e-01
3.13868999e-01 1.01048768e+00 1.42653540e-01 -1.99988246e-01
-6.57613039e-01 1.17909014e+00 -1.23199999e-01 6.03371203e-01
-8.23316813e-01 -1.84735879e-01 1.12864494e+00 1.38438091e-01
4.14036065e-01 -3.88767421e-01 -1.47044927e-01 -1.71647418e+00
-1.26274928e-01 -1.03716600e+00 -2.72683352e-01 -1.72338939e+00
-1.06457722e+00 1.90707132e-01 1.10456450e-02 -1.25223589e+00
-3.03760797e-01 -6.54818535e-01 -2.11920604e-01 1.20882452e+00
-1.93118811e+00 -9.43439126e-01 -6.84108853e-01 4.79969084e-01
7.11302996e-01 8.23613331e-02 3.39077502e-01 4.56366420e-01
-2.60258883e-01 1.68755934e-01 3.16357553e-01 -1.84740022e-01
1.43742502e+00 -9.19642091e-01 3.23369622e-01 1.34067523e+00
-4.04358327e-01 3.41592789e-01 8.50964963e-01 -7.13723540e-01
-1.68470025e+00 -8.87348831e-01 3.18701118e-01 -4.21201050e-01
3.76135349e-01 -3.23175900e-02 -6.54700279e-01 7.75807083e-01
4.45200771e-01 9.28038824e-03 2.03229532e-01 -5.75234473e-01
-2.59183556e-01 -3.94474536e-01 -8.56454432e-01 4.72213089e-01
8.30277443e-01 -6.68367326e-01 -3.94295722e-01 2.31461152e-01
7.80747175e-01 -8.81673276e-01 -5.40058136e-01 5.38943231e-01
6.11115456e-01 -1.51322997e+00 1.05061567e+00 2.58793294e-01
1.00832808e+00 -4.17793840e-01 -4.22434032e-01 -1.65412343e+00
-5.10492384e-01 -7.41710365e-01 -3.58290076e-01 7.43423998e-01
-3.51367556e-02 -7.63078809e-01 5.63831806e-01 1.97455376e-01
-5.26300490e-01 -3.69111180e-01 -5.74458003e-01 -3.64152074e-01
-1.20504424e-01 -2.90645480e-01 2.26121277e-01 8.10822368e-01
-5.13362229e-01 4.40926999e-01 -8.00379336e-01 5.86545646e-01
9.56951439e-01 1.47466779e-01 7.52145231e-01 -5.50965071e-01
-7.21717656e-01 -1.12834794e-03 4.24195342e-02 -2.08365512e+00
-1.84462920e-01 -4.23667312e-01 1.73221916e-01 -1.39880800e+00
3.42618614e-01 -8.64170790e-02 1.57983646e-01 1.76788375e-01
-2.40893364e-01 1.24236979e-01 3.02911643e-02 5.32647014e-01
-1.60801545e-01 3.49009633e-01 1.61180151e+00 -1.76750377e-01
2.68322788e-02 1.10382186e-02 -8.56469154e-01 3.37140650e-01
5.40966749e-01 -1.88179791e-01 -6.11026049e-01 -1.28301573e+00
1.48610085e-01 5.30831695e-01 5.70517540e-01 -1.04692018e+00
5.77107430e-01 -1.26316726e-01 3.62957448e-01 -5.92182875e-01
7.33574748e-01 -7.55008399e-01 3.27288300e-01 1.97952300e-01
3.31200194e-03 -3.21991384e-01 1.49047405e-01 4.07617718e-01
-2.61212945e-01 -1.67979494e-01 1.03259480e+00 -2.34280288e-01
-5.81151426e-01 1.00943901e-01 -2.19097838e-01 5.78554831e-02
5.99347115e-01 -2.51686603e-01 -1.04522884e+00 -8.16129565e-01
-3.33946586e-01 -5.79797402e-02 8.95286381e-01 1.78333417e-01
8.06071579e-01 -8.96321952e-01 -7.08892822e-01 5.12261152e-01
-2.00416058e-01 2.18275934e-01 4.95932728e-01 9.61386144e-01
-6.17778480e-01 5.60729444e-01 -2.91185588e-01 -8.67545605e-01
-1.18820024e+00 3.67610008e-01 7.10495532e-01 1.87118083e-01
-3.72822016e-01 1.02616644e+00 6.96833491e-01 3.86053361e-02
-4.59261276e-02 -4.39922094e-01 2.69926399e-01 -3.99915546e-01
7.52335608e-01 3.77314091e-01 -5.55225275e-02 -3.65706801e-01
-4.63865139e-02 6.97616398e-01 -2.18398914e-01 -1.75805092e-01
1.41575944e+00 -5.68611145e-01 -1.52718499e-01 1.96445644e-01
1.33746064e+00 1.26418412e-01 -1.96788359e+00 -2.00989842e-01
-9.40571010e-01 -9.11293685e-01 5.50392628e-01 -9.23958123e-01
-1.16488016e+00 9.43048418e-01 6.61563218e-01 -7.30227977e-02
1.68061495e+00 -7.33042955e-02 7.82240212e-01 2.09690422e-01
2.71081269e-01 -4.21798944e-01 3.03226680e-01 6.63833976e-01
8.37426364e-01 -1.29760444e+00 6.70261234e-02 -2.61524647e-01
-3.31179351e-01 1.11860347e+00 6.00595176e-01 9.32892039e-02
4.92838711e-01 6.05671346e-01 6.56540319e-02 -1.84873603e-02
-9.39438224e-01 3.31075132e-01 1.09054998e-01 5.58418989e-01
2.33820796e-01 -2.81333119e-01 3.68295550e-01 8.96389037e-02
3.27235507e-03 1.63235098e-01 1.01058638e+00 7.59371161e-01
-6.14756763e-01 -6.28871322e-01 -4.95560914e-01 3.65680605e-01
-4.15442199e-01 -4.31911469e-01 4.74731214e-02 1.12715073e-01
-5.01559582e-03 1.07898784e+00 -1.13336127e-02 -4.47125226e-01
5.37922187e-03 -4.64599878e-01 7.51529336e-01 -4.27205235e-01
-6.53038919e-02 2.75486797e-01 -2.15559788e-02 -7.02655256e-01
-6.35913730e-01 -7.02243268e-01 -6.10075712e-01 -4.43285793e-01
-2.66086996e-01 -3.65074009e-01 4.66608375e-01 6.99363530e-01
5.32490015e-01 3.63185078e-01 7.46489286e-01 -1.17036593e+00
-4.52941805e-01 -1.05551779e+00 -4.69021022e-01 -3.23971882e-02
1.15353131e+00 -4.61810410e-01 -9.84190583e-01 3.47563267e-01]
|
[11.190256118774414, -2.7131805419921875]
|
76ecba2a-d753-4a15-923b-e7d7e594c23a
|
tacticzero-learning-to-prove-theorems-from
|
2102.09756
| null |
https://arxiv.org/abs/2102.09756v2
|
https://arxiv.org/pdf/2102.09756v2.pdf
|
TacticZero: Learning to Prove Theorems from Scratch with Deep Reinforcement Learning
|
We propose a novel approach to interactive theorem-proving (ITP) using deep reinforcement learning. The proposed framework is able to learn proof search strategies as well as tactic and arguments prediction in an end-to-end manner. We formulate the process of ITP as a Markov decision process (MDP) in which each state represents a set of potential derivation paths. This structure allows us to introduce a novel backtracking mechanism which enables the agent to efficiently discard (predicted) dead-end derivations and restart from promising alternatives. We implement the framework in the HOL4 theorem prover. Experimental results show that the framework outperforms existing automated theorem provers (i.e., hammers) available in HOL4 when evaluated on unseen problems. We further elaborate the role of key components of the framework using ablation studies.
|
['Amir Dezfouli', 'Christian Walder', 'Michael Norrish', 'Minchao Wu']
|
2021-02-19
| null |
http://proceedings.neurips.cc/paper/2021/hash/4dea382d82666332fb564f2e711cbc71-Abstract.html
|
http://proceedings.neurips.cc/paper/2021/file/4dea382d82666332fb564f2e711cbc71-Paper.pdf
|
neurips-2021-12
|
['automated-theorem-proving', 'automated-theorem-proving']
|
['miscellaneous', 'reasoning']
|
[ 1.58410579e-01 7.32675493e-01 -2.76664943e-01 2.95310169e-02
-9.02054846e-01 -1.07149208e+00 8.01207960e-01 9.36024934e-02
-1.65178627e-01 9.38075423e-01 -3.73019904e-01 -1.30901957e+00
-2.62565017e-01 -1.03102469e+00 -1.05709994e+00 -1.21568024e-01
-3.84515107e-01 7.26055324e-01 4.91477311e-01 5.63870966e-02
3.05421621e-01 2.76113570e-01 -1.40203798e+00 4.66270268e-01
1.04004812e+00 2.83930063e-01 1.20741148e-02 1.16252470e+00
5.70818372e-02 1.33274019e+00 -3.52895200e-01 -4.22405243e-01
3.78329754e-02 -5.32366097e-01 -1.48544729e+00 -5.76481283e-01
5.79167046e-02 -7.74426758e-01 1.75210238e-02 1.12110662e+00
-1.45677151e-02 -2.10740790e-01 3.46657842e-01 -1.68919194e+00
6.00875244e-02 1.35126138e+00 2.70365030e-02 -1.41286224e-01
6.90323293e-01 4.46039051e-01 1.30933058e+00 -3.66634667e-01
8.93489778e-01 1.47597051e+00 4.15772676e-01 8.03259611e-01
-1.48600900e+00 -6.66086376e-01 1.17221899e-01 7.86199927e-01
-8.37709248e-01 -4.04801875e-01 7.57081449e-01 -2.17151091e-01
1.12344062e+00 3.66504490e-01 7.22977698e-01 1.29252970e+00
4.30241793e-01 1.28616357e+00 1.15452111e+00 -7.65583932e-01
4.98886108e-01 5.64960465e-02 2.20394865e-01 1.13582706e+00
8.98778811e-02 6.02076948e-01 -3.19287300e-01 -3.77141416e-01
4.45555896e-01 -7.02775002e-01 1.60098255e-01 -7.67481267e-01
-1.27556252e+00 9.46244597e-01 -9.77190211e-02 -1.88865568e-02
-1.02295831e-01 5.58728993e-01 5.73632121e-01 5.51745594e-01
-5.32157779e-01 6.74101830e-01 -5.32465577e-01 -2.37522498e-01
-7.48423755e-01 8.69769573e-01 1.24604893e+00 8.03209245e-01
5.84415555e-01 -2.80667335e-01 -5.19700110e-01 -8.33670869e-02
2.88062096e-01 4.69663411e-01 -2.03709424e-01 -1.37510478e+00
3.84563535e-01 4.95575398e-01 5.44655144e-01 -4.51659501e-01
-4.02005017e-01 -1.94243211e-02 -1.20138504e-01 6.13076568e-01
3.99866223e-01 -4.89930362e-01 -5.16781330e-01 1.73487997e+00
5.29027581e-01 3.21895927e-01 4.25447166e-01 4.02791113e-01
5.31867564e-01 7.02144325e-01 -2.92249769e-02 -3.22670847e-01
1.02507353e+00 -1.03250372e+00 -5.94781697e-01 1.28862217e-01
7.79241621e-01 -1.61382183e-01 7.50943184e-01 7.07515657e-01
-1.48263609e+00 -1.50609106e-01 -1.06468987e+00 2.65139580e-01
-4.99187596e-02 -3.64893049e-01 9.54569042e-01 3.75963867e-01
-1.00570858e+00 8.40647697e-01 -1.06873322e+00 3.63680944e-02
4.17507678e-01 4.13099170e-01 3.09305135e-02 5.20172939e-02
-1.26280212e+00 8.31393719e-01 9.22273219e-01 8.63204077e-02
-1.73652577e+00 -5.99851072e-01 -1.02587557e+00 2.11193308e-01
9.79671240e-01 -8.71485054e-01 1.98277521e+00 -8.50187659e-01
-1.98857892e+00 4.19641048e-01 -7.08009973e-02 -1.01891947e+00
9.31511045e-01 -1.47724152e-01 -1.06721288e-02 3.56573761e-01
9.97940674e-02 7.01506555e-01 7.19638109e-01 -1.18256783e+00
-9.66533065e-01 1.47559598e-01 7.65220225e-01 -1.81759804e-01
4.38859403e-01 1.05928719e-01 1.37880355e-01 1.59827843e-01
-3.38938236e-01 -9.75735962e-01 -5.55165261e-02 -3.56654286e-01
-7.80670881e-01 -6.47490144e-01 6.00258231e-01 -3.48251700e-01
1.05390227e+00 -1.52893925e+00 4.20769930e-01 2.09216684e-01
1.64195597e-01 2.40549758e-01 -1.79788977e-01 6.66572154e-01
4.97314371e-02 2.41438553e-01 8.74894764e-03 2.47947603e-01
6.14776790e-01 1.33497924e-01 -2.95303524e-01 1.03602156e-01
2.61820912e-01 1.18967283e+00 -1.37771404e+00 -7.70223379e-01
3.43492210e-01 -3.71309757e-01 -9.04201090e-01 3.77649784e-01
-1.05311203e+00 3.29151511e-01 -7.13266551e-01 6.29509926e-01
6.06636226e-01 -6.19990267e-02 6.19010866e-01 5.14446020e-01
-3.02864671e-01 6.14416122e-01 -1.08611691e+00 1.66316926e+00
-4.48641807e-01 3.80598426e-01 -1.02578521e-01 -8.49105000e-01
2.11406887e-01 2.72326678e-01 -2.93237329e-01 -5.35634160e-01
1.30539104e-01 4.83912714e-02 3.97161245e-01 -5.03456771e-01
1.45648673e-01 -8.08797106e-02 -8.68953764e-02 6.38232052e-01
8.50103870e-02 5.67440828e-03 5.13863266e-01 2.73326427e-01
1.40372109e+00 7.42704451e-01 3.38822871e-01 -1.45480439e-01
7.80014753e-01 3.38396728e-01 6.27486408e-01 1.37555850e+00
-9.72221792e-02 -4.29640561e-01 1.28006172e+00 -5.14558971e-01
-8.18288803e-01 -1.00476241e+00 2.40676582e-01 8.42491627e-01
5.49603552e-02 -7.76228130e-01 -8.86475921e-01 -1.32473171e+00
-8.27156454e-02 1.42980373e+00 -5.24579227e-01 -2.45896906e-01
-6.40557051e-01 2.84410983e-01 7.34094679e-01 2.63363034e-01
4.39509749e-01 -1.76755023e+00 -1.18782103e+00 2.52610534e-01
-2.16779634e-01 -7.84548402e-01 4.57211316e-01 3.45976532e-01
-6.36990547e-01 -1.49405015e+00 3.14587593e-01 -7.88552880e-01
3.13128412e-01 -3.14247727e-01 1.04904270e+00 2.79015750e-01
2.08408132e-01 3.34958762e-01 -3.88429105e-01 -3.17757338e-01
-1.32770801e+00 1.40212864e-01 -2.85748452e-01 -6.81665838e-01
5.40882722e-03 -4.65226978e-01 -1.46927223e-01 -9.67243388e-02
-5.66775322e-01 5.82253873e-01 5.39389431e-01 1.07935798e+00
2.65664339e-01 4.23638761e-01 3.07047188e-01 -1.20354342e+00
8.32639039e-01 -1.39023602e-01 -1.44763160e+00 6.75043285e-01
-7.20163822e-01 5.56806922e-01 1.03396690e+00 -1.47791460e-01
-1.02281165e+00 -1.52121440e-01 9.03387517e-02 -2.57785559e-01
-3.70573908e-01 5.28132081e-01 1.86814144e-02 1.58050507e-01
5.00677526e-01 3.44553977e-01 -2.43889585e-01 -3.43615487e-02
4.18592840e-01 4.58266512e-02 4.57963884e-01 -1.24901927e+00
1.20459533e+00 -2.11203456e-01 3.01696539e-01 6.83203638e-02
-6.84037566e-01 2.26971865e-01 -3.91172856e-01 -5.35393581e-02
2.92582840e-01 -3.26140910e-01 -1.69996262e+00 -6.42805547e-02
-1.41898489e+00 -8.79246354e-01 -9.62177292e-02 5.98001927e-02
-9.76766765e-01 3.68729979e-01 -5.59606075e-01 -1.20984054e+00
-4.63027507e-01 -1.31880963e+00 7.65888035e-01 1.96099095e-02
-5.33286572e-01 -7.97188461e-01 2.24613622e-01 1.27785072e-01
-2.76340842e-01 3.88785630e-01 1.42658532e+00 -8.42631698e-01
-8.33948314e-01 -2.85672699e-03 -2.80624591e-02 1.75984904e-01
-4.04001117e-01 1.59322366e-01 -7.02317178e-01 -1.75692230e-01
-4.24900115e-01 -6.50038660e-01 5.06638467e-01 2.00038686e-01
1.04321742e+00 -7.99528658e-01 -4.33461100e-01 2.90324450e-01
1.17522764e+00 2.59099305e-01 6.16266370e-01 8.19862068e-01
8.32749903e-02 2.82173723e-01 9.52625155e-01 2.23977998e-01
2.89045125e-01 2.85132796e-01 6.93141341e-01 3.49214077e-01
2.55076170e-01 -6.90719545e-01 5.33827901e-01 -3.34262252e-01
7.99805894e-02 -1.15760803e-01 -1.19887018e+00 3.96171838e-01
-2.10501051e+00 -1.42033803e+00 1.05838396e-01 1.93379951e+00
1.10617244e+00 6.30937397e-01 1.95082739e-01 4.46787417e-01
3.74302655e-01 -2.66418159e-01 -5.03814340e-01 -9.87701237e-01
5.35472572e-01 6.07119262e-01 -4.70496248e-03 1.04159606e+00
-9.99180436e-01 1.35178471e+00 6.65273142e+00 4.04702514e-01
-5.50252855e-01 -3.89541268e-01 -1.04023345e-01 2.51361340e-01
-3.85660142e-01 5.65009177e-01 -5.65744579e-01 -1.67776307e-04
9.92673934e-01 -2.99943596e-01 8.33668888e-01 9.85442579e-01
5.86036313e-03 -1.41351312e-01 -1.78660166e+00 1.55763000e-01
-5.68605602e-01 -1.65872359e+00 1.37272611e-01 -3.60211611e-01
3.06383193e-01 -4.11938637e-01 -1.70023799e-01 8.31661344e-01
7.56622434e-01 -8.66783440e-01 1.03788269e+00 7.70220831e-02
5.54321170e-01 -1.21362805e+00 6.29885435e-01 8.01565170e-01
-4.97954577e-01 -5.58047175e-01 1.20786622e-01 -2.88782001e-01
-3.68730396e-01 -1.65968835e-01 -1.69768775e+00 7.55390227e-01
2.74302334e-01 5.15740335e-01 -3.31297070e-01 7.70686209e-01
-1.08574462e+00 7.87821412e-01 -1.13582246e-01 -4.32071030e-01
4.14754301e-01 -1.63384899e-02 7.99524724e-01 1.10286081e+00
-1.32263049e-01 -6.94826012e-03 1.17210090e-01 1.40259457e+00
1.16639599e-01 -5.29653549e-01 -4.33738530e-01 -9.01601836e-02
4.93130505e-01 1.21599984e+00 -5.93296707e-01 -3.72038782e-01
-1.44359013e-02 7.96430290e-01 4.86488670e-01 3.36857706e-01
-1.14540207e+00 -3.47342432e-01 1.17864296e-01 -1.73646510e-01
4.30231303e-01 7.54849315e-02 1.78694695e-01 -9.69542325e-01
9.86138657e-02 -1.34933817e+00 5.01299918e-01 -7.85891891e-01
-5.47964156e-01 2.22121388e-01 2.85941511e-01 -7.03678191e-01
-6.24251425e-01 -3.70907873e-01 -8.09612751e-01 5.44166625e-01
-1.92183232e+00 -1.19925106e+00 2.15083867e-01 3.93247217e-01
3.99146110e-01 -1.80144031e-02 1.14798713e+00 -4.70219582e-01
-4.89556909e-01 5.40190279e-01 -3.15028101e-01 9.30521712e-02
7.43221045e-02 -1.62623429e+00 3.22755516e-01 9.99908626e-01
1.72118135e-02 7.49129355e-01 9.70494628e-01 -4.69745576e-01
-1.94087088e+00 -8.36433709e-01 7.33468890e-01 -3.94010693e-01
9.62241769e-01 -3.84476840e-01 -5.05214393e-01 1.02079320e+00
3.56750727e-01 -3.96486312e-01 1.59734979e-01 5.71923018e-01
-2.82111466e-01 2.34753862e-02 -9.81055200e-01 9.68449593e-01
8.49005103e-01 -2.89755315e-01 -9.85442221e-01 4.14517492e-01
6.41608894e-01 -6.69791698e-01 -3.34541738e-01 3.18165869e-01
6.51671886e-01 -1.15408146e+00 6.86664462e-01 -8.94302070e-01
8.07188630e-01 -2.77624190e-01 3.89122248e-01 -9.76699829e-01
-2.90282547e-01 -1.41869378e+00 -7.89417744e-01 9.82933879e-01
5.04948795e-01 -4.15512532e-01 6.14984393e-01 5.09707689e-01
5.28488681e-03 -8.14421475e-01 -8.48022461e-01 -6.27815306e-01
2.68615603e-01 -5.56925416e-01 4.61715698e-01 5.35990238e-01
5.93760312e-01 5.94930112e-01 -7.04743862e-02 4.21009421e-01
6.96645141e-01 7.98446119e-01 8.59635592e-01 -1.17526913e+00
-5.50089777e-01 -5.08053124e-01 2.24899516e-01 -6.45553052e-01
1.00484753e+00 -1.25523114e+00 1.94526345e-01 -1.47036731e+00
2.57698923e-01 -3.23189914e-01 -3.39978427e-01 9.34713185e-01
8.66314843e-02 -6.58850670e-01 2.57944800e-02 -3.36577237e-01
-1.13501585e+00 2.42226422e-01 1.32598400e+00 -3.14366668e-01
-3.10072303e-01 2.80674040e-01 -3.50051552e-01 7.87156343e-01
7.60124326e-01 -5.29440999e-01 -3.76990795e-01 -1.17108516e-01
6.04833901e-01 8.03737044e-01 7.14494169e-01 -8.99691284e-01
1.97727323e-01 -5.77262402e-01 5.66174611e-02 -6.09636128e-01
-3.62703085e-01 -6.54974580e-01 -1.10260479e-01 8.82785857e-01
-7.22286403e-01 -2.03004256e-01 3.47033203e-01 3.89821798e-01
1.70145974e-01 -3.98911893e-01 4.12739903e-01 -5.93696199e-02
-6.26230061e-01 -1.41924813e-01 -6.33907199e-01 -1.51001319e-01
1.20169175e+00 1.12591401e-01 2.28541661e-02 -1.79729328e-01
-4.56104010e-01 8.21332097e-01 2.87814379e-01 1.40192121e-01
5.76345265e-01 -8.19359124e-01 -5.95869362e-01 -1.03394343e-02
-5.85281802e-03 1.49210975e-01 -2.37913504e-01 6.03529155e-01
-6.50142193e-01 5.66549659e-01 -2.78599352e-01 -3.37950319e-01
-1.38982046e+00 8.40782464e-01 5.10666847e-01 -8.01390946e-01
-8.27566206e-01 4.41492200e-01 -9.74070802e-02 -3.59920621e-01
3.67593646e-01 -5.47934949e-01 -7.31925517e-02 -5.83171785e-01
4.98444617e-01 1.77990600e-01 -1.64782822e-01 4.88506734e-01
-5.14276683e-01 -1.00777291e-01 -2.66910434e-01 -2.30843917e-01
1.37970269e+00 4.41245109e-01 -1.91654921e-01 2.24608749e-01
3.51292580e-01 -1.43079907e-01 -1.06530118e+00 1.26547143e-01
3.46020788e-01 2.15037912e-02 -6.82333261e-02 -1.30553317e+00
-1.36819869e-01 6.63087606e-01 -5.10118268e-02 3.64208281e-01
7.71918774e-01 -7.64978379e-02 5.64023316e-01 1.08857465e+00
5.54107666e-01 -9.58170354e-01 -1.67911813e-01 6.49630368e-01
6.00918949e-01 -9.46172237e-01 6.09377883e-02 -7.50015900e-02
-3.15904826e-01 1.38668883e+00 4.68144149e-01 -9.26295146e-02
-1.46120042e-01 3.75231236e-01 -3.76116872e-01 7.61684477e-02
-1.27029932e+00 9.95148346e-02 -2.98019618e-01 5.85935175e-01
1.76295117e-02 4.19730581e-02 -1.91245407e-01 5.99483192e-01
-2.02962488e-01 2.63247848e-01 6.26299381e-01 1.25952733e+00
-1.71792582e-01 -1.52494287e+00 -3.80098224e-01 -1.91791505e-01
-3.55149060e-01 -6.00178121e-03 -4.81601894e-01 1.15715098e+00
-4.12515886e-02 9.20156479e-01 -5.56659639e-01 -1.40763730e-01
5.36533371e-02 3.39558065e-01 1.14059758e+00 -4.80334729e-01
-6.51542306e-01 -2.57364124e-01 5.12017071e-01 -7.64386833e-01
-1.98034957e-01 -7.32997179e-01 -1.47719169e+00 -3.89847636e-01
-3.25861961e-01 5.78996062e-01 1.46061167e-01 9.10377681e-01
2.19442975e-02 7.04776049e-01 5.92963338e-01 -4.36577290e-01
-8.24398935e-01 -5.56897819e-01 -8.40537325e-02 -2.57427543e-02
6.02974236e-01 -5.47937334e-01 -1.80197895e-01 -6.32756278e-02]
|
[8.920601844787598, 7.082557201385498]
|
8760c806-129a-4da4-98de-3ddc900ac3a3
|
rrwavenet-a-compact-end-to-end-multi-scale
|
2208.08672
| null |
https://arxiv.org/abs/2208.08672v2
|
https://arxiv.org/pdf/2208.08672v2.pdf
|
RRWaveNet: A Compact End-to-End Multi-Scale Residual CNN for Robust PPG Respiratory Rate Estimation
|
Respiratory rate (RR) is an important biomarker as RR changes can reflect severe medical events such as heart disease, lung disease, and sleep disorders. Unfortunately, standard manual RR counting is prone to human error and cannot be performed continuously. This study proposes a method for continuously estimating RR, RRWaveNet. The method is a compact end-to-end deep learning model which does not require feature engineering and can use low-cost raw photoplethysmography (PPG) as input signal. RRWaveNet was tested subject-independently and compared to baseline in four datasets (BIDMC, CapnoBase, WESAD, and SensAI) and using three window sizes (16, 32, and 64 seconds). RRWaveNet outperformed current state-of-the-art methods with mean absolute errors at optimal window size of 1.66 \pm 1.01, 1.59 \pm 1.08, 1.92 \pm 0.96 and 1.23 \pm 0.61 breaths per minute for each dataset. In remote monitoring settings, such as in the WESAD and SensAI datasets, we apply transfer learning to improve the performance using two other ICU datasets as pretraining datasets, reducing the MAE by up to 21$\%$. This shows that this model allows accurate and practical estimation of RR on affordable and wearable devices. Our study also shows feasibility of remote RR monitoring in the context of telemedicine and at home.
|
['Theerawit Wilaiprasitporn', 'Emmanuel Mignot', 'Proadpran Punyabukkana', 'Tanut Choksatchawathi', 'Narin Kunaseth', 'Thee Mateepithaktham', 'Kawisara Ueafuea', 'Punnawish Thuwajit', 'Guntitat Sawadwuthikul', 'Pongpanut Osathitporn']
|
2022-08-18
| null | null | null | null |
['photoplethysmography-ppg', 'respiratory-rate-estimation']
|
['medical', 'medical']
|
[ 8.25948492e-02 -1.01565197e-01 3.87001820e-02 -2.75083870e-01
-7.50451148e-01 -1.81104258e-01 -3.36548418e-01 -7.25835413e-02
-6.42551899e-01 9.91135657e-01 -4.26707640e-02 -4.68193471e-01
-1.04255840e-01 -5.21426439e-01 -3.69158059e-01 -6.28504097e-01
-2.51235247e-01 1.83460675e-02 -2.10544795e-01 3.85454327e-01
-1.64469093e-01 4.33689594e-01 -7.17803836e-01 1.39228672e-01
5.94206929e-01 1.24232066e+00 -2.78353453e-01 1.04461575e+00
3.95021647e-01 6.25031412e-01 -9.24355984e-01 1.39961779e-01
3.19799900e-01 -8.01068664e-01 -2.36776665e-01 -6.36781394e-01
4.82141793e-01 -6.63663208e-01 -2.23842204e-01 5.60256205e-02
1.48505175e+00 1.00899480e-01 1.36466742e-01 -8.44255865e-01
-5.38495183e-02 2.69303918e-01 -3.74850422e-01 5.39365947e-01
1.22817114e-01 4.30786788e-01 3.17215443e-01 -4.62807417e-01
-1.47826016e-01 5.05486190e-01 1.37907588e+00 7.26274312e-01
-1.21800768e+00 -9.91985917e-01 -6.72627747e-01 5.61782010e-02
-1.37758064e+00 -3.80435944e-01 4.93546158e-01 -2.26191089e-01
1.13160169e+00 5.07252336e-01 6.57111287e-01 1.06681621e+00
6.87957525e-01 -7.62227401e-02 1.36863863e+00 -7.89130703e-02
1.06299065e-01 1.34432226e-01 -2.59847101e-03 4.15732861e-01
3.14654350e-01 2.40258593e-02 -6.09483480e-01 -3.02172631e-01
7.68957734e-01 4.78440315e-01 -6.37233019e-01 3.58799636e-01
-1.20769823e+00 4.94981110e-01 1.74538702e-01 2.27445289e-01
-5.80138564e-01 3.94242197e-01 3.76116067e-01 2.10542649e-01
3.09043497e-01 3.84328216e-01 -8.09293687e-01 -6.77452683e-01
-8.08895051e-01 -3.15500468e-01 9.35268104e-01 4.43832338e-01
2.22506493e-01 -1.04767501e-01 -4.59126353e-01 9.77633834e-01
1.91539198e-01 6.38995528e-01 6.15763545e-01 -1.08280349e+00
3.86485577e-01 2.49251440e-01 3.26013088e-01 -5.44680059e-01
-1.02803743e+00 -3.67797464e-01 -1.01132441e+00 -2.34990433e-01
1.89669654e-01 -4.66288477e-01 -7.26681471e-01 1.39407504e+00
1.14143141e-01 6.67426646e-01 -1.15247235e-01 8.94302547e-01
1.07666445e+00 2.40545094e-01 3.05366099e-01 -5.27694106e-01
1.32010436e+00 -4.26741481e-01 -7.39059746e-01 -8.59040953e-03
4.95183378e-01 -2.75001854e-01 1.18615913e+00 4.62028205e-01
-1.01216090e+00 -6.60124123e-01 -8.98853838e-01 1.84135035e-01
8.25910792e-02 7.18764290e-02 1.92608014e-01 9.93676484e-01
-1.01684153e+00 1.03438306e+00 -9.96708453e-01 -2.45436490e-01
5.03977060e-01 4.17038709e-01 1.39689520e-01 8.16281065e-02
-1.36156821e+00 8.80939066e-01 -2.15946913e-01 3.30971807e-01
-4.10488427e-01 -1.16311169e+00 -5.27611196e-01 1.15516484e-01
-3.70178431e-01 -1.00704002e+00 1.09163213e+00 -1.43653661e-01
-1.87888896e+00 6.16758704e-01 -6.90832287e-02 -5.07959187e-01
6.40825510e-01 -6.37280524e-01 -4.74816501e-01 4.18711364e-01
-4.88947481e-01 1.81070045e-01 4.45026189e-01 -4.75159615e-01
1.96180344e-02 -3.25961411e-01 -4.68757778e-01 7.52767473e-02
-2.21393377e-01 1.63293052e-02 2.56854951e-01 -1.83950260e-01
-1.61455721e-01 -8.61233532e-01 8.33550394e-02 7.82896727e-02
-4.77414876e-02 1.39718577e-01 5.45045435e-01 -6.53028727e-01
1.14860046e+00 -2.03078556e+00 -6.44105434e-01 -1.91159416e-02
3.72120559e-01 4.78138834e-01 2.77799755e-01 1.70897081e-01
-7.13554844e-02 2.82264560e-01 -2.59687513e-01 -3.11101079e-01
-3.09624016e-01 1.43639818e-01 8.32142681e-02 7.39966273e-01
-2.67293394e-01 9.89338219e-01 -7.73582041e-01 -2.75293350e-01
7.26353109e-01 9.13782716e-01 5.72764734e-03 5.36524951e-01
5.99598885e-01 7.69793212e-01 -3.95912901e-02 3.21730494e-01
4.17233467e-01 -3.63757700e-01 -6.23514652e-02 -3.30049992e-01
-9.79172066e-02 5.48246324e-01 -8.78067136e-01 1.53227246e+00
-7.20579624e-01 7.55266011e-01 -3.84045631e-01 -6.54996753e-01
1.19133949e+00 7.86537468e-01 1.01418960e+00 -8.42416823e-01
2.75610626e-01 6.76237717e-02 7.22753778e-02 -9.31938052e-01
-4.49280262e-01 -6.71601117e-01 2.47033358e-01 6.34582281e-01
-4.12449986e-01 -1.02569960e-01 -4.79922593e-01 -4.91312891e-01
1.29201341e+00 -2.85372257e-01 2.63989329e-01 -2.93633610e-01
2.63855577e-01 -6.97453499e-01 4.73685503e-01 8.50511372e-01
-4.75982755e-01 9.08772826e-01 -6.21857867e-03 -7.24323690e-01
-3.64483237e-01 -1.09110856e+00 -5.56288958e-01 5.87583959e-01
-1.42233387e-01 -2.41750926e-02 -4.67591256e-01 -2.94588715e-01
5.58663718e-02 6.53007269e-01 -4.04747874e-01 -3.29819679e-01
-6.85430944e-01 -1.18367028e+00 1.00857592e+00 8.70214283e-01
5.93281329e-01 -1.36474049e+00 -1.42019618e+00 4.35503066e-01
-3.75667870e-01 -1.03632414e+00 -3.92931342e-01 1.25885978e-01
-1.25272870e+00 -1.00064349e+00 -8.08557034e-01 -6.73854947e-02
-3.89486253e-02 -7.79678822e-02 1.15315282e+00 -7.35273585e-02
-8.32250357e-01 5.47978401e-01 2.65205037e-02 -8.72406960e-01
-2.62609534e-02 -1.35716051e-01 2.67847717e-01 -1.66793719e-01
5.35643637e-01 -7.82733679e-01 -1.37153733e+00 2.64804155e-01
-1.78213835e-01 -3.28577846e-01 4.24336880e-01 4.90242004e-01
6.65347278e-01 -6.36321723e-01 8.66712093e-01 -5.81833839e-01
8.85005116e-01 -4.83600348e-01 -1.85311601e-01 -1.91662967e-01
-1.17205083e+00 -5.03600359e-01 5.69955170e-01 -5.18116653e-01
-4.15679872e-01 -3.25619817e-01 2.42052856e-03 -5.83622694e-01
-3.17936987e-01 -5.99100254e-02 4.43567216e-01 1.66490003e-01
9.60234165e-01 1.17946364e-01 1.26370654e-01 -4.07678515e-01
-2.09659263e-01 7.93072224e-01 5.01225114e-01 -1.41296998e-01
3.58084649e-01 2.02827871e-01 1.68549359e-01 -7.98287570e-01
-4.54986989e-01 -5.77527046e-01 -4.67245609e-01 -2.09023386e-01
9.85130906e-01 -1.12450373e+00 -1.15551674e+00 4.21422660e-01
-7.76802480e-01 -8.07818651e-01 -5.91296315e-01 9.25595880e-01
-3.66653174e-01 2.41088867e-01 -7.45687246e-01 -8.85530531e-01
-1.28083181e+00 -4.36017990e-01 7.57278323e-01 2.77602404e-01
-6.74272299e-01 -9.57587779e-01 3.02732706e-01 2.78522968e-01
9.41040874e-01 8.14442813e-01 2.42285490e-01 -3.80014569e-01
-2.96333004e-02 -3.42594862e-01 -1.47252738e-01 6.30354106e-01
6.16452813e-01 -3.83384854e-01 -1.20655036e+00 -3.11681807e-01
3.69513035e-01 -2.86644399e-01 4.05340135e-01 5.55897236e-01
1.52973974e+00 -2.01888725e-01 -7.48519897e-02 7.04661310e-01
1.25679529e+00 5.27402103e-01 8.90654683e-01 -4.97587696e-02
6.01245582e-01 1.25125470e-02 9.24019665e-02 6.68193638e-01
3.54181856e-01 1.57682359e-01 9.66611803e-02 -5.06563008e-01
9.74384025e-02 3.29860330e-01 3.47182274e-01 6.48786604e-01
-2.85854638e-01 -9.64095369e-02 -9.24104869e-01 3.13172251e-01
-1.27828336e+00 -6.81265235e-01 -4.00906235e-01 2.37944436e+00
1.02562666e+00 -2.18477771e-01 2.26974919e-01 1.25828192e-01
5.83618939e-01 -1.00795418e-01 -7.94237316e-01 -7.28843033e-01
5.01759708e-01 8.99282932e-01 5.54057121e-01 4.20145877e-02
-6.75960839e-01 -4.28174157e-03 6.33479929e+00 -4.00491476e-01
-1.47295320e+00 2.69523948e-01 8.33500087e-01 -4.17016327e-01
4.24525619e-01 -6.41382158e-01 -4.44673687e-01 8.79029334e-01
1.79095519e+00 5.84876537e-02 3.31250995e-01 6.72357261e-01
6.68568075e-01 -1.70592174e-01 -1.26842880e+00 1.50118494e+00
6.09026663e-02 -8.83891940e-01 -9.50816572e-01 -1.26049384e-01
1.56305552e-01 5.30810773e-01 -2.45393440e-01 1.33252054e-01
-3.92076641e-01 -1.27791131e+00 -2.47409046e-01 6.87340319e-01
1.33465815e+00 -4.42976713e-01 1.05648434e+00 -2.51522660e-03
-7.74900556e-01 1.09168403e-01 -6.43417835e-02 -3.33682299e-02
-6.95871934e-02 8.47666979e-01 -9.16448832e-01 1.43813342e-01
9.81206715e-01 4.69661713e-01 -1.52141750e-01 1.11668742e+00
-1.52235463e-01 8.51705372e-01 -6.16680026e-01 9.61728022e-03
-4.22698677e-01 2.64362812e-01 2.30791971e-01 1.34353280e+00
4.10936654e-01 6.58136666e-01 -1.56240717e-01 9.66998041e-01
-2.34688327e-01 -2.37448722e-01 -4.27760422e-01 5.31501114e-01
3.54226589e-01 1.18112850e+00 -2.39504859e-01 -2.31947869e-01
-3.27649355e-01 6.32463753e-01 -4.17811215e-01 2.61912882e-01
-1.11963177e+00 -6.88924015e-01 5.73777199e-01 2.82942861e-01
-1.36905596e-01 6.75242692e-02 -7.93200433e-01 -7.70484149e-01
9.76517051e-02 -6.06954873e-01 4.83736187e-01 -7.27100194e-01
-1.04709399e+00 4.46516663e-01 -9.42436531e-02 -1.29844689e+00
-1.23301029e-01 -3.63769263e-01 -9.40879285e-01 1.27542400e+00
-1.77284300e+00 3.15193050e-02 -1.02642858e+00 6.90453708e-01
2.49667868e-01 5.08067787e-01 1.12124777e+00 4.80030417e-01
-6.43895090e-01 7.23200023e-01 -3.20835680e-01 -2.21990850e-02
9.00980353e-01 -1.15096855e+00 1.59544811e-01 3.07557136e-01
-5.90599179e-01 7.08545744e-01 2.85011709e-01 -2.27454513e-01
-1.35005832e+00 -1.17298579e+00 7.85932362e-01 -6.70094967e-01
2.72889525e-01 -1.32196561e-01 -9.03892934e-01 1.66762084e-01
3.93695235e-02 5.96954286e-01 1.17137301e+00 -1.88193962e-01
1.07842289e-01 -6.59990788e-01 -1.63121533e+00 2.37711128e-02
7.36848056e-01 -5.00542581e-01 -5.35357118e-01 2.33853698e-01
4.41460311e-01 -7.26984978e-01 -1.66037452e+00 6.75218642e-01
8.54297817e-01 -7.36940622e-01 8.70167315e-01 -2.02667311e-01
-1.79435499e-02 5.66424709e-03 3.35236222e-01 -1.08175218e+00
-1.57748759e-01 -9.79774594e-01 -4.86073881e-01 7.81895697e-01
1.26267388e-01 -1.16591465e+00 5.44130206e-01 1.10527289e+00
-1.87688962e-01 -8.56172740e-01 -8.96162927e-01 -6.99436843e-01
-1.06078751e-01 -3.67376089e-01 3.55495960e-01 8.63240063e-01
6.96857050e-02 2.51285255e-01 -3.16959113e-01 -8.98124948e-02
5.46155334e-01 -1.32737495e-02 4.89739090e-01 -1.25074220e+00
-2.36821502e-01 -2.94220317e-02 -1.27335091e-03 -3.81957293e-01
-4.95626032e-01 -5.10242641e-01 1.58839017e-01 -1.74848771e+00
-1.32672727e-01 -4.19889748e-01 -1.14063799e+00 7.94670641e-01
-2.82149732e-01 5.78823626e-01 -1.86301097e-01 5.92362061e-02
-9.90740061e-02 1.52505338e-01 9.68781769e-01 2.09338710e-01
-7.29798734e-01 2.66461074e-02 -4.43321437e-01 2.87599325e-01
1.44052863e+00 -6.71354055e-01 -4.36004430e-01 -1.10785745e-03
-2.02840626e-01 3.82489026e-01 4.66333181e-01 -1.13176250e+00
-8.73763487e-02 1.26952916e-01 8.70396197e-01 -2.13346869e-01
1.97790861e-01 -7.43631661e-01 3.59203070e-01 7.80094445e-01
-1.35723576e-01 2.41191164e-01 5.49382985e-01 1.14080288e-01
3.00471157e-01 3.39133471e-01 1.08396482e+00 -2.64824331e-01
2.88711302e-02 4.16118145e-01 -2.36528546e-01 4.49299484e-01
7.61969209e-01 -4.74353135e-01 -4.31485564e-01 -1.74966991e-01
-6.63835049e-01 8.67867619e-02 -1.53937265e-01 3.61498445e-01
7.84912705e-01 -9.04261827e-01 -6.96536779e-01 2.17915848e-01
1.82203706e-02 1.37091167e-02 3.27281475e-01 1.47121263e+00
-5.71174800e-01 2.41369784e-01 -9.64750815e-03 -7.90102780e-01
-1.10028076e+00 5.45704179e-02 8.52032244e-01 6.43578842e-02
-9.95320499e-01 4.72598463e-01 -2.53065109e-01 1.44286707e-01
4.33702320e-01 -6.52559519e-01 5.68925887e-02 -2.21759021e-01
6.66893601e-01 7.57557988e-01 3.53694171e-01 2.88325757e-01
-5.96424282e-01 7.32428253e-01 3.49793702e-01 2.52831280e-01
1.21971750e+00 -1.26108661e-01 -3.93823832e-02 6.48047388e-01
1.20787001e+00 -3.09994102e-01 -1.06100607e+00 2.36331224e-01
-2.33657420e-01 -2.34531477e-01 -2.73118410e-02 -1.20081222e+00
-1.03040826e+00 1.03362620e+00 1.48592067e+00 3.36536542e-02
1.41265082e+00 -3.31669778e-01 1.02110457e+00 3.90197754e-01
1.17819145e-01 -9.04592454e-01 3.73296365e-02 -9.75300893e-02
6.51065469e-01 -8.26055706e-01 -3.55687663e-02 2.44105011e-01
-4.89530265e-01 1.17128873e+00 4.99462962e-01 2.39544120e-02
6.50729597e-01 1.58626601e-01 4.62020963e-01 -5.46301976e-02
-5.65027893e-01 3.25883240e-01 9.19623747e-02 4.13447261e-01
7.46724010e-01 5.54268956e-02 -5.12807608e-01 3.33777100e-01
2.17600893e-02 5.85597694e-01 5.47997236e-01 6.89715326e-01
-1.74727604e-01 -5.96409678e-01 -1.52537338e-02 7.68552840e-01
-1.09933245e+00 -1.60925671e-01 -3.72226979e-03 6.43280029e-01
3.37734297e-02 1.15315282e+00 7.74594098e-02 -2.88857788e-01
6.41479850e-01 5.36751926e-01 4.26603019e-01 -6.03833973e-01
-9.38565850e-01 -1.09470300e-01 -5.83579354e-02 -7.30320275e-01
-4.80327189e-01 -3.63839656e-01 -1.42221808e+00 5.09510040e-02
-9.88123417e-02 3.90087184e-03 6.94851756e-01 4.87496674e-01
7.87616193e-01 8.19168091e-01 7.06446767e-01 -4.45202172e-01
-4.71008807e-01 -1.21339309e+00 -4.19395536e-01 1.72648117e-01
6.97768569e-01 -1.87129572e-01 -6.04421973e-01 -3.13756578e-02]
|
[13.927268981933594, 2.9805960655212402]
|
6c469607-b67f-4bfa-b8a3-850ae8c74520
|
iedit-localised-text-guided-image-editing
|
2305.05947
| null |
https://arxiv.org/abs/2305.05947v1
|
https://arxiv.org/pdf/2305.05947v1.pdf
|
iEdit: Localised Text-guided Image Editing with Weak Supervision
|
Diffusion models (DMs) can generate realistic images with text guidance using large-scale datasets. However, they demonstrate limited controllability in the output space of the generated images. We propose a novel learning method for text-guided image editing, namely \texttt{iEdit}, that generates images conditioned on a source image and a textual edit prompt. As a fully-annotated dataset with target images does not exist, previous approaches perform subject-specific fine-tuning at test time or adopt contrastive learning without a target image, leading to issues on preserving the fidelity of the source image. We propose to automatically construct a dataset derived from LAION-5B, containing pseudo-target images with their descriptive edit prompts given input image-caption pairs. This dataset gives us the flexibility of introducing a weakly-supervised loss function to generate the pseudo-target image from the latent noise of the source image conditioned on the edit prompt. To encourage localised editing and preserve or modify spatial structures in the image, we propose a loss function that uses segmentation masks to guide the editing during training and optionally at inference. Our model is trained on the constructed dataset with 200K samples and constrained GPU resources. It shows favourable results against its counterparts in terms of image fidelity, CLIP alignment score and qualitatively for editing both generated and real images.
|
['Loris Bazzani', 'Michael Donoser', 'Tae-Kyun Kim', 'Binod Bhattarai', 'Erhan Gundogdu', 'Rumeysa Bodur']
|
2023-05-10
| null | null | null | null |
['text-guided-image-editing']
|
['computer-vision']
|
[ 9.35960233e-01 3.29320699e-01 1.87898144e-01 -3.38912159e-01
-7.11437702e-01 -7.14726925e-01 9.58941221e-01 -2.93129355e-01
-4.08923507e-01 7.64843822e-01 -5.39144613e-02 -1.42714173e-01
-7.65718669e-02 -7.23141968e-01 -1.11917114e+00 -6.74615085e-01
2.76092798e-01 5.65207958e-01 8.95096287e-02 -6.43807128e-02
3.52491081e-01 4.63262528e-01 -1.28248763e+00 4.20410544e-01
1.31653917e+00 8.74077976e-01 7.05877066e-01 1.10091960e+00
3.86466198e-02 8.41752470e-01 -6.57700121e-01 -3.91641468e-01
5.88216484e-01 -7.67371774e-01 -6.97373986e-01 4.62711155e-01
6.67599380e-01 -3.31727892e-01 -2.23960906e-01 9.79556203e-01
6.48607612e-01 9.84523818e-02 9.30222988e-01 -1.22613430e+00
-9.75597799e-01 3.18936616e-01 -6.73182130e-01 -7.09527209e-02
1.40844554e-01 6.59728110e-01 4.99543995e-01 -7.90375888e-01
1.23263121e+00 1.01468074e+00 4.03780937e-01 6.50217772e-01
-1.72246444e+00 -3.35223645e-01 -2.03959256e-01 -1.69127628e-01
-1.22222710e+00 -4.90809888e-01 7.05999374e-01 -6.40691042e-01
5.14486969e-01 5.30048370e-01 4.11440730e-01 1.35352516e+00
1.87570557e-01 5.40559292e-01 1.48646319e+00 -6.13517404e-01
2.04481006e-01 4.70892787e-01 -7.31426954e-01 6.62154198e-01
-3.23844284e-01 4.51391637e-01 -2.30315849e-01 1.17846049e-01
1.01675737e+00 -4.16940540e-01 -3.89130145e-01 -4.45164412e-01
-1.44465041e+00 7.57995963e-01 3.64237249e-01 6.47830218e-02
-3.71130019e-01 1.27643496e-01 1.77523971e-01 4.42328066e-01
4.37383294e-01 4.89643276e-01 -1.47622824e-01 1.63815528e-01
-1.28636241e+00 2.91444361e-01 5.08277714e-01 1.20920730e+00
6.82331502e-01 1.48392543e-01 -8.28214645e-01 7.23438323e-01
-2.11675271e-01 6.33077025e-01 3.23684692e-01 -1.26574695e+00
5.47310531e-01 3.44430327e-01 3.24645609e-01 -9.19189990e-01
-1.11120017e-02 -1.79150417e-01 -1.11596274e+00 5.87909460e-01
4.62561250e-01 -2.94992715e-01 -1.20120406e+00 1.75661826e+00
3.34755629e-01 9.39592943e-02 -6.86058626e-02 8.91660869e-01
3.15790504e-01 8.35717618e-01 -2.14232296e-01 -3.43266338e-01
7.35463679e-01 -1.16869891e+00 -7.38714933e-01 -1.19019628e-01
3.54131162e-01 -8.47427607e-01 1.50983930e+00 2.63152987e-01
-1.30150187e+00 -6.51163042e-01 -7.94703782e-01 -1.67651623e-01
-2.59940475e-01 4.04631555e-01 1.13698959e-01 5.23025453e-01
-1.19627130e+00 5.81538081e-01 -4.85017240e-01 -1.86129048e-01
4.41067934e-01 2.27003589e-01 -1.95279077e-01 -4.17465195e-02
-9.21786964e-01 9.15399134e-01 2.80219793e-01 3.02878004e-02
-1.23606622e+00 -8.88312638e-01 -6.37284696e-01 -2.28018880e-01
3.32935005e-01 -6.94176137e-01 6.80195689e-01 -1.54699886e+00
-1.86741221e+00 9.69958961e-01 2.48032123e-01 -3.97206962e-01
1.28855348e+00 2.11294100e-01 -1.02221586e-01 1.35019511e-01
1.61481008e-01 1.35316002e+00 1.42397606e+00 -1.57108736e+00
-3.88308078e-01 2.04914555e-01 6.29046038e-02 1.61364287e-01
-1.27911428e-02 -1.34839922e-01 -6.92375422e-01 -1.12418604e+00
-6.37152076e-01 -1.18242264e+00 -4.66041893e-01 3.97132874e-01
-7.19302952e-01 5.97564042e-01 9.27711844e-01 -6.62461162e-01
8.86974037e-01 -2.05020452e+00 4.14866298e-01 3.31271172e-01
3.80872861e-02 3.09861422e-01 -6.30100548e-01 2.20250487e-01
4.78407815e-02 1.29432976e-01 -7.53787935e-01 -3.12778592e-01
-1.16785064e-01 1.59430370e-01 -4.55262601e-01 3.29052657e-01
3.91072810e-01 1.05805421e+00 -8.25272620e-01 -5.96161902e-01
3.71223778e-01 6.34381354e-01 -7.65393257e-01 4.50785667e-01
-5.46816409e-01 8.98637593e-01 -1.70756280e-01 7.76541308e-02
7.97551394e-01 -8.09988156e-02 -8.75882804e-02 -2.38763019e-01
-4.94706258e-02 -5.11155367e-01 -1.15304947e+00 1.70146310e+00
-6.10840023e-01 7.73969412e-01 2.87843108e-01 -7.62805760e-01
7.48981059e-01 3.37841697e-02 2.95182705e-01 -7.15996385e-01
1.80200785e-02 7.84503147e-02 -1.36629358e-01 -4.78996843e-01
4.30979431e-01 1.44039065e-01 1.41519085e-01 5.89395881e-01
-5.06246127e-02 -7.48992383e-01 3.60636413e-01 2.57694960e-01
7.71335304e-01 3.88896763e-01 -2.76627958e-01 -3.75841051e-01
4.68046874e-01 -1.21542020e-02 7.54725635e-02 9.60317791e-01
1.73465699e-01 1.19628489e+00 4.74378407e-01 -1.18010402e-01
-1.57760906e+00 -9.69141960e-01 -6.70254752e-02 7.42615283e-01
3.16743255e-02 -6.79806173e-02 -1.16304755e+00 -6.96098685e-01
-3.36924374e-01 8.21337223e-01 -7.59893715e-01 -1.08749464e-01
-5.58081806e-01 -5.39193690e-01 6.16686165e-01 1.46029457e-01
6.40386581e-01 -1.23338914e+00 -6.23386800e-01 1.43407613e-01
-1.22433759e-01 -1.07162905e+00 -1.11237144e+00 7.09668174e-02
-5.95159709e-01 -8.60519052e-01 -1.02201247e+00 -8.26828063e-01
1.17218149e+00 -2.69349605e-01 8.59572947e-01 -1.45916119e-01
-4.93767023e-01 3.33983749e-01 -1.83160510e-02 3.83207425e-02
-8.70792091e-01 -1.86192662e-01 -1.76304713e-01 3.05876315e-01
-8.38964164e-01 -4.00310457e-01 -6.55560374e-01 4.60248351e-01
-1.43110800e+00 5.99557161e-01 5.37760735e-01 1.02176583e+00
7.07631588e-01 -4.53401282e-02 2.10259601e-01 -1.02906144e+00
7.26070344e-01 -4.77123596e-02 -7.52599776e-01 3.66556495e-01
-5.14913321e-01 2.00678036e-01 8.95419896e-01 -7.61684179e-01
-1.33624697e+00 3.19161147e-01 2.28975460e-01 -6.26744986e-01
4.65293229e-02 9.22280475e-02 -1.01673409e-01 -2.73371935e-01
9.81908858e-01 3.63405585e-01 3.16952467e-02 -8.34044293e-02
8.01264286e-01 4.68782932e-01 7.05365181e-01 -7.16458976e-01
9.09092784e-01 5.48140228e-01 -1.13982983e-01 -7.66402185e-01
-4.55334663e-01 2.74500132e-01 -7.43590057e-01 -2.73193628e-01
9.83241618e-01 -5.88753700e-01 -2.98969388e-01 5.50306320e-01
-1.25995648e+00 -8.64412129e-01 -5.77346206e-01 1.18697181e-01
-9.79524732e-01 1.76065758e-01 -4.19300675e-01 -3.45952153e-01
-2.33641446e-01 -1.44779491e+00 1.19789124e+00 1.17758466e-02
-9.77762118e-02 -9.40585434e-01 -2.38364935e-01 2.91974664e-01
6.76827252e-01 5.76454222e-01 9.24833894e-01 4.87740198e-03
-8.96868944e-01 7.38701150e-02 -3.55307311e-01 6.08598590e-01
8.47824663e-02 1.89127877e-01 -7.25341976e-01 -3.12624782e-01
-8.83522034e-02 -3.73139083e-01 6.75593019e-01 5.00084102e-01
1.33598876e+00 -5.37719429e-01 -1.49459049e-01 7.91856945e-01
1.35141265e+00 1.45971164e-01 8.01828861e-01 9.01218429e-02
8.08311045e-01 7.28867888e-01 5.05414963e-01 2.78380752e-01
-1.15614608e-01 8.16235244e-01 8.16144198e-02 -3.01692456e-01
-4.67648715e-01 -3.53919774e-01 3.94953340e-01 3.61180276e-01
1.51475950e-03 -6.41475737e-01 -6.18633687e-01 4.70947295e-01
-1.67321897e+00 -8.16323459e-01 -1.18525334e-01 2.19976330e+00
1.18307495e+00 4.51182649e-02 -2.51760155e-01 -2.24133030e-01
9.33753848e-01 1.96135864e-01 -7.09649980e-01 -4.18679655e-01
-3.30609411e-01 1.08232073e-01 5.64219594e-01 8.95706594e-01
-8.16545427e-01 1.01816177e+00 6.15028763e+00 1.14981937e+00
-1.37159359e+00 1.88313946e-01 1.10129964e+00 -2.66270816e-01
-4.62825716e-01 -6.76003471e-02 -3.64554405e-01 5.67187667e-01
7.04770565e-01 -1.40676066e-01 6.54379070e-01 4.31375682e-01
6.72447801e-01 -1.67023242e-01 -9.96917069e-01 7.44331360e-01
5.11401296e-02 -1.64184344e+00 4.05382842e-01 2.40784418e-02
1.36135614e+00 -3.32194656e-01 3.95320266e-01 -1.17673472e-01
2.46056318e-01 -1.08345401e+00 9.06921208e-01 7.56855547e-01
1.37960780e+00 -5.03925145e-01 1.55766383e-01 3.45944226e-01
-6.39135540e-01 2.62434185e-01 -2.21344873e-01 3.58318776e-01
2.59693921e-01 5.57686985e-01 -7.12536335e-01 2.61019588e-01
3.76003653e-01 4.02799875e-01 -6.83796883e-01 6.30777895e-01
-1.62197426e-01 4.17188555e-01 -2.82877743e-01 3.95373046e-01
2.60544658e-01 -5.34385204e-01 6.71656907e-01 1.24341512e+00
4.22385782e-01 -8.12551677e-02 1.13255233e-01 1.35637331e+00
-9.36386585e-02 2.78530717e-02 -7.12998569e-01 -7.32215047e-02
2.24114001e-01 1.18127751e+00 -8.16283584e-01 -3.55662167e-01
1.62380442e-01 1.68577933e+00 8.61643180e-02 6.39235795e-01
-1.11689687e+00 -4.07384306e-01 7.10486695e-02 3.06263894e-01
4.09894913e-01 -5.76025955e-02 -3.73709053e-01 -9.21405077e-01
5.96820079e-02 -1.01243532e+00 -1.29777968e-01 -1.19757605e+00
-9.21299696e-01 7.43916690e-01 1.09554268e-02 -1.07969344e+00
-1.93550169e-01 -2.78686643e-01 -6.62885785e-01 9.52143371e-01
-1.15248656e+00 -1.33445239e+00 -3.99461120e-01 6.75935745e-01
6.02896988e-01 -1.11996002e-01 4.49032575e-01 2.98157722e-01
-3.58915597e-01 5.82963347e-01 3.24070156e-01 -1.78553879e-01
8.61227036e-01 -1.20805109e+00 3.82482320e-01 7.70753086e-01
-6.95606098e-02 2.63155282e-01 8.94233465e-01 -7.53491879e-01
-1.06529176e+00 -1.48711848e+00 5.20100832e-01 -5.55877030e-01
3.06083411e-01 -5.57161450e-01 -6.14673436e-01 6.31527305e-01
6.56980515e-01 1.77249521e-01 -1.05212077e-01 -8.75775874e-01
9.40511450e-02 -2.37370264e-02 -1.31151211e+00 8.55122566e-01
1.15421033e+00 -5.66273272e-01 6.62176218e-03 4.65284169e-01
8.06197464e-01 -6.94385290e-01 -6.55263066e-01 2.16774017e-01
1.19900703e-01 -8.12381029e-01 8.48499358e-01 -2.74613291e-01
7.35909879e-01 -4.46952611e-01 2.25066096e-01 -1.56556833e+00
-5.51358201e-02 -1.07589602e+00 3.24754983e-01 1.41401720e+00
5.57157159e-01 -1.25270203e-01 5.54155111e-01 5.99852860e-01
4.53106724e-02 -5.02619088e-01 -7.11761236e-01 -6.91499531e-01
2.20063962e-02 -2.21346363e-01 2.51379639e-01 8.82126451e-01
-5.84289789e-01 1.33671463e-01 -6.96400881e-01 -4.19283323e-02
5.96431077e-01 1.89407368e-03 7.81487465e-01 -4.99938846e-01
-2.26705104e-01 -4.00168747e-01 -1.48802670e-02 -9.36662912e-01
5.76842427e-02 -8.13129961e-01 1.36285976e-01 -1.24834609e+00
4.33275960e-02 -5.58560133e-01 4.47163999e-01 2.74075419e-01
-1.45510342e-02 5.11696339e-01 1.98678061e-01 1.89380705e-01
-2.81947166e-01 6.52885795e-01 1.95085800e+00 -3.78022850e-01
-2.78333485e-01 -2.27436081e-01 -2.78300852e-01 3.20316732e-01
5.40240884e-01 -4.51172829e-01 -7.16295481e-01 -4.62342054e-01
-7.33628543e-03 1.32447168e-01 4.86106932e-01 -8.35509896e-01
7.51675963e-02 -3.46968472e-01 3.53813529e-01 -1.96900830e-01
1.53872505e-01 -7.60689735e-01 6.08223081e-01 3.01169395e-01
-7.71317482e-01 -9.59091038e-02 1.00311249e-01 5.72539508e-01
-6.15497977e-02 -2.16563582e-01 1.14709854e+00 2.50701830e-02
-3.98065001e-01 3.27389449e-01 -2.96242088e-01 2.45741293e-01
1.27107608e+00 -3.11176211e-01 -2.25689828e-01 -4.87146169e-01
-7.23136663e-01 5.61944433e-02 8.13009262e-01 4.40390348e-01
4.79087144e-01 -1.34373927e+00 -7.76770949e-01 4.01642352e-01
-1.14367411e-01 1.66611914e-02 3.46100777e-01 7.55416751e-01
-7.39114463e-01 1.46993101e-01 -4.83258426e-01 -6.52377427e-01
-9.69488919e-01 6.58565819e-01 4.37705487e-01 -2.13783219e-01
-6.25511229e-01 6.78056419e-01 5.24144053e-01 -5.96020460e-01
4.84954864e-02 -3.48118991e-01 1.34303734e-01 -1.23503014e-01
3.14827830e-01 1.73177868e-01 -1.18087925e-01 -5.44770777e-01
1.91425055e-01 6.01688862e-01 -1.17386945e-01 -4.46138173e-01
1.10404336e+00 -2.87233144e-01 5.70908226e-02 3.42188179e-02
1.18622148e+00 9.20415819e-02 -2.01317596e+00 -1.83903351e-02
-2.79702365e-01 -6.90345287e-01 8.63659978e-02 -1.04387057e+00
-1.32700455e+00 6.84816718e-01 8.57725680e-01 -1.12096258e-01
1.16612363e+00 -3.43635052e-01 5.98232925e-01 8.04965794e-02
1.88280512e-02 -1.24133158e+00 4.11314964e-01 2.82192230e-01
1.42087853e+00 -1.17573023e+00 -3.50067139e-01 -3.55508298e-01
-1.02454543e+00 7.26522446e-01 6.04125500e-01 -1.34433240e-01
2.42931172e-01 4.07545894e-01 1.86696947e-01 -4.94345985e-02
-5.36519766e-01 1.43542886e-01 3.19064289e-01 9.37052727e-01
3.20839025e-02 -1.85253859e-01 -2.07934693e-01 1.95911527e-02
-1.43783346e-01 1.15398876e-01 6.48339570e-01 6.79526329e-01
2.76366640e-02 -1.04937720e+00 -3.89497668e-01 4.25695926e-01
-1.02158830e-01 -2.47975528e-01 -4.07303929e-01 6.12620354e-01
2.10168689e-01 7.05074370e-01 -1.22698247e-02 -9.29782465e-02
2.76020378e-01 -2.88461477e-01 6.57799363e-01 -3.80899101e-01
-5.81930757e-01 -3.57320122e-02 -4.00004610e-02 -5.10194361e-01
-3.69540781e-01 -4.51934159e-01 -9.82618392e-01 -2.81477809e-01
-2.08527267e-01 -3.07614524e-02 4.65794086e-01 6.86565757e-01
5.63542843e-01 4.99678940e-01 7.16148138e-01 -1.21099687e+00
-2.78368831e-01 -7.69799411e-01 -4.51523781e-01 6.38127327e-01
2.16306224e-01 -3.63218248e-01 -2.51187801e-01 6.54145539e-01]
|
[11.508194923400879, -0.4864347577095032]
|
2ce20c98-72cd-4522-bdfb-3e18afe19553
|
supercaustics-real-time-open-source
|
2107.11008
| null |
https://arxiv.org/abs/2107.11008v2
|
https://arxiv.org/pdf/2107.11008v2.pdf
|
SuperCaustics: Real-time, open-source simulation of transparent objects for deep learning applications
|
Transparent objects are a very challenging problem in computer vision. They are hard to segment or classify due to their lack of precise boundaries, and there is limited data available for training deep neural networks. As such, current solutions for this problem employ rigid synthetic datasets, which lack flexibility and lead to severe performance degradation when deployed on real-world scenarios. In particular, these synthetic datasets omit features such as refraction, dispersion and caustics due to limitations in the rendering pipeline. To address this issue, we present SuperCaustics, a real-time, open-source simulation of transparent objects designed for deep learning applications. SuperCaustics features extensive modules for stochastic environment creation; uses hardware ray-tracing to support caustics, dispersion, and refraction; and enables generating massive datasets with multi-modal, pixel-perfect ground truth annotations. To validate our proposed system, we trained a deep neural network from scratch to segment transparent objects in difficult lighting scenarios. Our neural network achieved performance comparable to the state-of-the-art on a real-world dataset using only 10% of the training data and in a fraction of the training time. Further experiments show that a model trained with SuperCaustics can segment different types of caustics, even in images with multiple overlapping transparent objects. To the best of our knowledge, this is the first such result for a model trained on synthetic data. Both our open-source code and experimental data are freely available online.
|
['Rolando Estrada', 'Mehdi Mousavi']
|
2021-07-23
| null | null | null | null |
['transparent-objects', 'transparent-object-depth-estimation', 'transparent-object-detection', 'physical-simulations']
|
['computer-vision', 'computer-vision', 'computer-vision', 'miscellaneous']
|
[ 2.92407662e-01 6.29128441e-02 5.55099785e-01 -1.62536219e-01
-4.92219806e-01 -4.79660362e-01 4.66880471e-01 -2.26097584e-01
-6.94127604e-02 5.56054533e-01 -4.10440713e-01 -4.72279191e-01
4.32732671e-01 -9.57679093e-01 -9.40723419e-01 -7.04385042e-01
-2.98592560e-02 6.59759223e-01 9.67293561e-01 -1.15544759e-01
-1.20412342e-01 6.50480151e-01 -1.86739755e+00 5.38802445e-01
1.21857560e+00 1.07189751e+00 3.50907534e-01 5.70980251e-01
-3.32680821e-01 5.83305895e-01 -6.44972086e-01 -3.27466756e-01
8.56812894e-01 -1.55278906e-01 -4.02026802e-01 -1.28934383e-01
1.00678039e+00 -6.15989864e-01 -1.22049280e-01 6.24096811e-01
6.63607717e-01 -1.47481099e-01 6.19371533e-01 -9.28641498e-01
-4.14198279e-01 1.65503442e-01 -3.53956550e-01 -2.03848109e-01
-5.56872645e-03 4.40880448e-01 4.28529888e-01 -7.01475382e-01
5.67559361e-01 8.70767951e-01 9.74480569e-01 6.55189395e-01
-1.15231907e+00 -5.03154039e-01 -1.88796654e-01 -3.16507012e-01
-9.67006266e-01 -3.71460944e-01 7.03251719e-01 -6.90680146e-01
8.57969165e-01 5.30818045e-01 1.00691736e+00 1.07065403e+00
1.08413003e-01 4.05296862e-01 1.27194309e+00 -3.55340153e-01
3.58231008e-01 2.88619310e-01 -1.79818273e-01 8.14040720e-01
2.40955383e-01 2.94889569e-01 -4.72543342e-03 -6.92796111e-02
8.83012652e-01 -2.37643123e-01 -4.47405785e-01 -4.29619730e-01
-1.15099871e+00 5.89538157e-01 6.80168390e-01 -3.03035766e-01
3.77098992e-02 2.74361014e-01 1.45174742e-01 6.75000018e-04
6.73329890e-01 3.23396683e-01 -3.15529168e-01 1.14883609e-01
-9.10132766e-01 1.52972594e-01 9.18146610e-01 8.78988743e-01
7.77046561e-01 3.16170305e-01 -5.58028072e-02 7.70898521e-01
2.41303757e-01 5.05808413e-01 1.57773122e-01 -1.10777605e+00
3.32144201e-01 5.66775382e-01 1.68173000e-01 -4.26086396e-01
-4.73596871e-01 -5.46675801e-01 -4.92717355e-01 1.02955747e+00
5.14277518e-01 -1.92680731e-01 -1.18653286e+00 1.16395915e+00
7.35045195e-01 2.70053416e-01 7.93042704e-02 9.92227077e-01
1.21505237e+00 7.97634900e-01 -4.82647061e-01 1.89978600e-01
1.00441301e+00 -1.23182631e+00 -1.27919137e-01 -1.66489363e-01
4.10418451e-01 -1.22039855e+00 1.42042994e+00 5.97476304e-01
-1.02522373e+00 -3.43923599e-01 -8.95112336e-01 -6.07683770e-02
-1.92292467e-01 1.41865999e-01 9.79749382e-01 8.05034459e-01
-1.02640307e+00 4.52898949e-01 -7.64955938e-01 -2.26382077e-01
6.40550256e-01 1.29807830e-01 4.85618226e-02 -5.80547843e-03
-7.56661475e-01 4.53834057e-01 3.86997350e-02 7.54472017e-02
-1.00959182e+00 -1.07884622e+00 -6.27339303e-01 -2.84032136e-01
1.58435568e-01 -6.99071169e-01 1.35945010e+00 -1.03565264e+00
-1.72013891e+00 8.82799149e-01 2.37469435e-01 -3.58563662e-01
1.04028594e+00 -2.74557531e-01 -1.00940399e-01 1.28704131e-01
-1.43368602e-01 4.94145989e-01 7.40754306e-01 -1.87736988e+00
-2.60279387e-01 7.79883340e-02 2.99876869e-01 -6.86968863e-02
-1.56361938e-01 7.14397132e-02 -4.22234625e-01 -6.47297442e-01
-1.30733386e-01 -1.03522253e+00 -2.71321237e-01 6.26590014e-01
-5.71263611e-01 2.56743163e-01 1.01566708e+00 -3.90630692e-01
4.90105987e-01 -1.83738804e+00 -5.42942524e-01 -5.59100509e-02
8.76633823e-02 6.63837612e-01 -4.50551994e-02 3.46296847e-01
1.50011808e-01 -3.82175222e-02 -6.00441992e-01 -5.08564591e-01
-2.20994115e-01 1.11156911e-01 -4.31136906e-01 4.25318182e-01
-1.60712123e-01 4.88947541e-01 -4.99184549e-01 -5.04142404e-01
4.21271950e-01 6.94792688e-01 -5.41579545e-01 1.72546014e-01
-6.93283319e-01 5.42646289e-01 -1.92957804e-01 5.49500346e-01
1.10393190e+00 -2.79483330e-02 -3.56458843e-01 -2.25243405e-01
-4.98886734e-01 2.05351472e-01 -1.06221521e+00 1.44933259e+00
-7.78911293e-01 9.78828073e-01 5.10185547e-02 -4.58791375e-01
8.16106021e-01 1.28757790e-01 4.72740948e-01 -6.95743322e-01
1.40810847e-01 3.93592358e-01 -6.63163811e-02 -6.38220608e-01
3.23168099e-01 1.40123323e-01 4.69014674e-01 5.21731079e-01
-5.35636425e-01 -7.95556843e-01 5.93945086e-02 -6.31998852e-02
1.09634078e+00 4.98330206e-01 -4.74198759e-01 -1.28703028e-01
-4.59075943e-02 3.35265219e-01 5.31225860e-01 6.00302517e-01
3.43760580e-01 1.37448990e+00 2.55066678e-02 -6.99712098e-01
-1.24969292e+00 -1.29851699e+00 -5.16415954e-01 6.63153887e-01
4.00253683e-01 -8.31471309e-02 -1.02180529e+00 -2.01034069e-01
-1.98084339e-01 5.86326659e-01 -4.04827833e-01 2.25210786e-01
-6.55255973e-01 -8.48146141e-01 4.03086156e-01 2.70146012e-01
8.13532889e-01 -1.20154226e+00 -1.17136848e+00 1.42630771e-01
1.97729617e-01 -1.38598502e+00 5.24668731e-02 -5.60609214e-02
-8.33578706e-01 -1.16609144e+00 -7.73993909e-01 -6.79921210e-01
6.33607268e-01 4.35646892e-01 1.34784162e+00 3.17258656e-01
-7.55814016e-01 -4.85068420e-03 -1.59850389e-01 -6.09849632e-01
-6.52964175e-01 -2.39800841e-01 -3.92224580e-01 -2.56061614e-01
-5.40607691e-01 -4.93465722e-01 -9.39577520e-01 5.84862769e-01
-9.88145053e-01 6.38156652e-01 4.01836157e-01 4.19087470e-01
5.35292804e-01 -2.71286350e-02 -3.58100608e-02 -9.93017673e-01
1.63458824e-01 -1.26272038e-01 -1.06797957e+00 1.34985477e-01
-1.40313953e-01 -1.98449433e-01 7.10807085e-01 -4.60075259e-01
-1.43647039e+00 9.90191326e-02 -2.09085360e-01 -2.72472382e-01
-2.90525645e-01 1.41386064e-02 5.79646342e-02 -3.82953554e-01
9.41022336e-01 -1.11485951e-01 -3.41652453e-01 -4.54443157e-01
8.81392211e-02 5.50714135e-01 3.04862648e-01 -6.32894397e-01
8.39044690e-01 1.07655263e+00 5.45332991e-02 -9.15526927e-01
-9.84648705e-01 -9.03964490e-02 -3.39703739e-01 -3.30405444e-01
6.71631038e-01 -7.55375445e-01 -5.41771114e-01 8.08854222e-01
-1.17030120e+00 -1.12973607e+00 -4.92434472e-01 4.70637679e-01
-5.42140543e-01 1.80083141e-01 -6.33700192e-01 -6.97445691e-01
-4.05969918e-01 -1.24669969e+00 1.10602212e+00 3.66929650e-01
2.45513707e-01 -7.17463970e-01 9.88540202e-02 5.02096832e-01
4.22643334e-01 6.52207673e-01 6.25150681e-01 1.91427082e-01
-1.08345699e+00 6.99054226e-02 -4.87152725e-01 3.33513498e-01
4.00765501e-02 6.69560790e-01 -1.41890383e+00 -8.74353573e-02
-6.35508597e-02 -5.49900532e-01 9.66958106e-01 3.02352488e-01
1.41863930e+00 -1.48598775e-01 -2.95668244e-01 1.01511228e+00
1.47248340e+00 -6.62863627e-02 7.37412751e-01 3.79512459e-01
1.00422668e+00 5.84962249e-01 5.20664692e-01 4.48957682e-01
2.19972223e-01 7.95138180e-01 9.50526178e-01 -6.41441822e-01
-6.85807526e-01 2.34964207e-01 1.28756166e-01 4.03569311e-01
-3.28759044e-01 -4.25163150e-01 -9.88460362e-01 3.42672616e-01
-1.45192182e+00 -8.00308585e-01 -8.84740770e-01 2.43071342e+00
8.04744840e-01 1.45349577e-01 -1.07146807e-01 2.88827624e-02
3.34907115e-01 -7.52807483e-02 -5.79675555e-01 -5.64234495e-01
-2.36267954e-01 2.76855707e-01 5.77665329e-01 3.89012814e-01
-9.26422119e-01 9.74117756e-01 5.71877813e+00 7.56093621e-01
-1.62442815e+00 1.02140263e-01 5.45106113e-01 -1.71690762e-01
-5.96351683e-01 -5.19641005e-02 -5.19346595e-01 4.23115432e-01
5.33610165e-01 5.33968866e-01 4.17911738e-01 6.86268568e-01
4.58445162e-01 -3.90249103e-01 -8.16601157e-01 7.36077547e-01
-2.16824841e-02 -1.49602056e+00 -1.61838248e-01 1.35391206e-01
1.20516992e+00 7.09347606e-01 6.83827102e-02 -3.33294392e-01
5.92453837e-01 -1.04945302e+00 1.08013701e+00 4.65744138e-01
8.65147650e-01 -3.78409863e-01 3.03331435e-01 3.61685336e-01
-9.38394904e-01 2.25856468e-01 -5.13557255e-01 1.94652840e-01
1.61553755e-01 1.09735441e+00 -1.01811397e+00 3.35988104e-01
9.06131089e-01 1.78321660e-01 -5.90000927e-01 1.58227134e+00
-1.99861690e-01 9.13324714e-01 -7.39033043e-01 1.13738924e-01
1.17706008e-01 -3.35497469e-01 3.86157125e-01 1.09438682e+00
4.74599808e-01 -3.59795272e-01 1.12118296e-01 1.10559750e+00
8.95756930e-02 -6.98905736e-02 -5.10625720e-01 4.03259814e-01
7.46815354e-02 1.28811944e+00 -1.01895833e+00 -1.73516721e-01
-3.28781903e-01 8.91883910e-01 3.47998530e-01 5.43272853e-01
-1.15983415e+00 -2.10071385e-01 5.46645582e-01 3.66846293e-01
4.57189381e-02 -2.49586955e-01 -5.73832750e-01 -1.03130269e+00
2.21918195e-01 -4.43029881e-01 -9.12197083e-02 -1.03219831e+00
-1.12977469e+00 9.48126197e-01 -1.40452906e-01 -1.47842157e+00
1.45050555e-01 -7.61675954e-01 -8.75506401e-01 7.76371300e-01
-1.67012513e+00 -1.36168301e+00 -8.67000818e-01 4.47820306e-01
5.20396054e-01 1.76207144e-02 8.16978097e-01 2.08042800e-01
-3.87063861e-01 1.77627757e-01 5.14079392e-01 -1.31455228e-01
5.21222055e-01 -1.07575965e+00 6.18306220e-01 7.65315115e-01
1.07586727e-01 2.41838574e-01 8.53537917e-01 -4.85151052e-01
-1.04475582e+00 -1.26190650e+00 -2.81441845e-02 -2.02413827e-01
3.19491565e-01 -6.51986897e-01 -9.46608841e-01 2.90808618e-01
5.05377389e-02 4.25301015e-01 4.26375568e-01 -3.71935606e-01
-2.56672084e-01 -1.66871369e-01 -1.32919252e+00 8.36814642e-01
1.26408458e+00 -3.07394452e-02 -5.47075085e-02 6.85683668e-01
8.25367808e-01 -8.36993515e-01 -6.06977463e-01 5.35704017e-01
6.91167414e-01 -1.75659752e+00 1.04869998e+00 1.64101273e-01
5.59307039e-01 -3.53947759e-01 1.33312881e-01 -1.33320940e+00
3.29818010e-01 -6.19812965e-01 -6.49418831e-02 1.17646205e+00
4.53413337e-01 -8.45890462e-01 9.79239404e-01 5.31196594e-01
-6.92323565e-01 -9.26721513e-01 -7.52533674e-01 -6.99492991e-01
-5.19851372e-02 -5.60602546e-01 6.62127614e-01 6.27410650e-01
-9.98061001e-01 -2.76400179e-01 -3.78808826e-02 2.44404286e-01
8.66456211e-01 7.09413648e-01 1.01209569e+00 -1.42643404e+00
-3.48794252e-01 -2.78336138e-01 -6.60292059e-02 -8.10573518e-01
-3.18739340e-02 -6.40552282e-01 2.02437162e-01 -1.99914134e+00
-1.12420693e-01 -1.37726760e+00 5.87409675e-01 3.09008151e-01
1.87015474e-01 8.11404288e-01 -5.73117286e-02 2.81246960e-01
1.87240653e-02 5.47166109e-01 1.59851408e+00 -2.32677072e-01
-2.26944342e-01 2.99433142e-01 -2.12799177e-01 9.48318243e-01
1.15990579e+00 -3.52647096e-01 -5.65337300e-01 -9.71940637e-01
2.11018324e-01 -5.13129354e-01 6.05051875e-01 -1.43629766e+00
-2.49166876e-01 -1.86452091e-01 3.29732269e-01 -5.33216178e-01
7.46697664e-01 -9.12986994e-01 3.14943820e-01 3.82497072e-01
1.68435991e-01 -5.92032731e-01 4.29090142e-01 6.61023930e-02
3.87217969e-01 -3.44763935e-01 9.78909016e-01 -2.07332432e-01
-3.78387451e-01 4.23657238e-01 -2.12243110e-01 1.60874538e-02
1.17123938e+00 -5.96861005e-01 -7.37763584e-01 -1.17454631e-02
-2.00999394e-01 -1.79366454e-01 1.10367346e+00 1.08209662e-01
6.39376581e-01 -8.33521605e-01 -5.97110152e-01 1.95983663e-01
1.25687167e-01 8.80902946e-01 1.96681008e-01 3.65034401e-01
-1.50084615e+00 -1.30154580e-01 -9.77618992e-02 -7.58638740e-01
-1.27672577e+00 2.86865532e-01 6.36401892e-01 1.82735547e-01
-1.02618110e+00 8.39640915e-01 5.67462564e-01 -6.98630393e-01
1.00059612e-02 -6.58021927e-01 2.84102499e-01 -3.68388385e-01
2.72322714e-01 2.80117810e-01 4.90021020e-01 -4.83459800e-01
5.12347408e-02 7.12850869e-01 3.42823297e-01 -1.17109127e-01
1.36982739e+00 1.93430945e-01 -2.86281481e-02 3.95184278e-01
8.99390280e-01 3.00750881e-01 -1.70146310e+00 6.73288703e-02
-7.67738223e-01 -5.73547363e-01 9.39013436e-02 -8.17159116e-01
-1.30938005e+00 1.07228160e+00 6.48830891e-01 2.03364819e-01
9.04433072e-01 -6.30532450e-04 9.90038216e-01 3.72663915e-01
5.54526150e-01 -8.15685749e-01 -7.69840628e-02 3.68879646e-01
9.57388639e-01 -1.18314636e+00 -4.89815287e-02 -9.59702253e-01
-2.83550709e-01 1.02622497e+00 8.83701742e-01 1.00433882e-02
5.00855446e-01 6.73494518e-01 7.14988172e-01 -1.25229105e-01
-4.25708145e-01 3.27999331e-02 3.65986973e-02 9.20960009e-01
3.32570523e-01 -5.55375330e-02 2.81923294e-01 -2.22347334e-01
-5.90481579e-01 -9.38227475e-02 6.69553041e-01 8.35671246e-01
-3.70469421e-01 -8.38419437e-01 -6.47214592e-01 4.70589370e-01
-2.19172224e-01 -1.19326167e-01 -1.47054762e-01 4.91599202e-01
2.74454623e-01 8.07695866e-01 2.05747768e-01 1.31742865e-01
7.00476915e-02 -7.61474013e-01 6.21758282e-01 -6.68668807e-01
-5.80579877e-01 -1.55063972e-01 1.83041230e-01 -5.74380994e-01
-4.00516212e-01 -3.48645389e-01 -1.42580390e+00 -9.04893428e-02
-2.09245399e-01 -1.21105224e-01 1.03743327e+00 4.79643315e-01
2.46630624e-01 6.67700291e-01 6.32468581e-01 -1.47657180e+00
-2.68473234e-02 -7.82466173e-01 -2.31398582e-01 2.38397166e-01
3.03159237e-01 -6.05016172e-01 -2.55141824e-01 1.08733021e-01]
|
[10.379950523376465, -1.2811797857284546]
|
f6dbb6be-16f8-4f89-af2f-e9c852363b34
|
a-new-twitter-verb-lexicon-for-natural
| null | null |
https://aclanthology.org/L12-1641
|
https://aclanthology.org/L12-1641.pdf
|
A New Twitter Verb Lexicon for Natural Language Processing
|
We describe in-progress work on the creation of a new lexical resource that contains a list of 486 verbs annotated with quantified temporal durations for the events that they describe. This resource is being compiled from more than 14 million tweets from the Twitter microblogging site. We are creating this lexicon of verbs and typical durations to address a gap in the available information that is represented in existing research. The data that is contained in this lexical resource is unlike any existing resources, which have been traditionally comprised from literature excerpts, news stories, and full-length weblogs. The kind of knowledge about how long an event lasts is crucial for natural language processing and is especially useful when the temporal duration of an event is implied. We are using data from Twitter because Twitter is a rich resource since people are publicly posting about real events and real durations of those events throughout the day.
|
['Graham Katz', 'Jennifer Williams']
|
2012-05-01
| null | null | null |
lrec-2012-5
|
['game-of-chess']
|
['playing-games']
|
[-1.43518165e-01 5.28695956e-02 -4.56795007e-01 -3.25926960e-01
-5.58195174e-01 -9.42537963e-01 9.19394553e-01 1.01399612e+00
-7.80758083e-01 8.93724680e-01 1.05519950e+00 -1.59972891e-01
1.89332232e-01 -8.38820636e-01 -2.27548912e-01 3.63817848e-02
-2.02419341e-01 2.34619841e-01 2.52241164e-01 -4.14622992e-01
3.96112531e-01 3.39887112e-01 -1.56817472e+00 3.50056350e-01
1.81214377e-01 8.33989501e-01 3.76347080e-02 2.52005816e-01
-3.84116203e-01 9.83386815e-01 -6.50098324e-01 -3.36188436e-01
-2.52753735e-01 -4.12878126e-01 -1.00870144e+00 -5.46891019e-02
-2.77844995e-01 2.63639241e-02 -2.62361377e-01 4.87957627e-01
1.69027507e-01 2.84804910e-01 4.65207577e-01 -1.15427434e+00
-2.10328028e-01 9.12087381e-01 -1.79774538e-01 9.96529281e-01
8.19399416e-01 -1.14602208e-01 1.08708370e+00 -5.07529676e-01
1.12723994e+00 9.67264175e-01 6.57458484e-01 3.47747020e-02
-6.82632387e-01 -5.62189341e-01 5.71741909e-02 -5.69759076e-03
-1.42175376e+00 -3.27198446e-01 6.24237001e-01 -6.78472400e-01
1.47788942e+00 1.27729729e-01 8.61794770e-01 1.45860231e+00
1.13814518e-01 2.48758033e-01 8.11755240e-01 -6.09274507e-01
-6.74746111e-02 2.39338115e-01 2.91268945e-01 1.16484679e-01
1.90191895e-01 -3.91378790e-01 -8.13577056e-01 -3.31724614e-01
4.46802229e-01 -1.06491357e-01 6.41113147e-02 5.11535585e-01
-1.25750577e+00 7.87284791e-01 -1.49621561e-01 8.74608457e-01
-5.23154259e-01 2.82089878e-02 9.35942948e-01 1.17077269e-01
1.00831664e+00 3.49877208e-01 -5.23344696e-01 -9.29878771e-01
-9.70835865e-01 5.66223443e-01 1.28464234e+00 9.65194702e-01
5.79082787e-01 -3.34955454e-01 8.56208354e-02 6.83784425e-01
2.37077504e-01 1.13505675e-02 5.36683679e-01 -7.56424189e-01
6.75998390e-01 8.18388939e-01 7.15266049e-01 -1.15442622e+00
-7.42328644e-01 7.87293538e-02 1.09037623e-01 -7.71482348e-01
4.98238593e-01 -3.96224201e-01 -3.55888695e-01 1.72268689e+00
3.73009205e-01 1.60231978e-01 -2.36328125e-01 2.97691464e-01
1.21689606e+00 6.57344222e-01 4.78681594e-01 -7.82106042e-01
1.70988524e+00 -9.49929655e-02 -1.28742564e+00 -3.21452469e-01
8.25473726e-01 -7.68717825e-01 1.08519840e+00 1.36481807e-01
-9.15870726e-01 1.62409365e-01 -7.86013126e-01 -2.71716326e-01
-1.02693021e+00 -5.49106359e-01 8.06352317e-01 5.73879838e-01
-5.17034888e-01 3.61937970e-01 -7.94634998e-01 -1.01794255e+00
7.91576505e-02 -3.38112593e-01 -2.76950896e-01 5.48872948e-01
-1.56923854e+00 1.17720640e+00 6.02926016e-01 -4.26791906e-01
-2.98731834e-01 -6.29097700e-01 -8.26279104e-01 -3.66041213e-01
6.28135622e-01 9.72371101e-02 1.71359372e+00 -6.29335701e-01
-9.27836597e-01 1.26417565e+00 -2.43394345e-01 -4.56871629e-01
2.36791730e-01 5.89570310e-03 -8.69457483e-01 1.00204572e-01
5.19374907e-01 5.30006401e-02 1.10413320e-01 -5.88205159e-01
-8.07635009e-01 -2.04816729e-01 1.96816593e-01 1.19828828e-01
-5.20782828e-01 8.73012781e-01 -4.68530625e-01 -9.25747812e-01
-3.23068112e-01 -6.92700565e-01 2.04083532e-01 -6.87562048e-01
-3.13652754e-01 -5.90274751e-01 6.68164492e-01 -6.96519196e-01
2.05149722e+00 -2.00978184e+00 -3.82044017e-01 -2.04864606e-01
-1.56061992e-01 -4.20839101e-01 4.94111031e-01 1.57580256e+00
4.62474078e-02 5.30580938e-01 1.53528720e-01 -1.51796803e-01
9.53561440e-02 2.23562062e-01 -6.59491837e-01 5.70573032e-01
-1.92635864e-01 6.10250413e-01 -1.16261339e+00 -7.46062636e-01
-1.02212764e-01 3.48452806e-01 -8.96447226e-02 -2.24904209e-01
-2.51012385e-01 2.23894447e-01 -6.11607790e-01 3.55223686e-01
-9.46864784e-02 -2.86321640e-01 8.83666873e-02 -5.21229114e-03
-7.48451710e-01 1.06550503e+00 -9.65770900e-01 1.35343695e+00
-4.36401635e-01 9.35840130e-01 -3.62057984e-01 -4.11120057e-01
4.99219030e-01 7.29703009e-01 7.10499406e-01 -4.40955490e-01
4.39594120e-01 -2.68396977e-02 -5.57445288e-01 -5.31874776e-01
8.90252292e-01 -1.34474918e-01 -6.20707095e-01 7.23618090e-01
-2.19081640e-01 -2.84027815e-01 9.09388959e-01 2.08847299e-01
1.12131083e+00 3.85274366e-02 9.09527123e-01 -2.49825418e-01
1.33486822e-01 4.65236217e-01 3.96703452e-01 4.34981436e-01
-3.19665253e-01 3.61577213e-01 7.46648431e-01 -4.90214527e-01
-1.04755270e+00 -5.97962618e-01 -5.02057254e-01 1.16645169e+00
-1.15533523e-01 -1.11475074e+00 -4.26635832e-01 -1.22588024e-01
-3.21969748e-01 9.14604664e-01 -7.56500483e-01 5.04899621e-01
-4.58721995e-01 -6.95680499e-01 5.50567865e-01 3.90094608e-01
1.53969422e-01 -1.35839045e+00 -8.49683464e-01 5.81226707e-01
-7.63652027e-01 -1.49471939e+00 -4.15226430e-01 4.68843766e-02
-3.21811438e-01 -1.02015162e+00 4.56555076e-02 -5.23756146e-01
7.71670975e-03 -4.96874005e-02 1.28475523e+00 -2.42076039e-01
3.05401674e-03 3.29746366e-01 -7.99761474e-01 -9.15736616e-01
-2.88875461e-01 3.95157933e-02 -1.27432585e-01 -4.45039809e-01
7.85936773e-01 -5.22152901e-01 -2.04965800e-01 -2.19108816e-03
-1.02723086e+00 -1.53799474e-01 -3.16216797e-01 1.43810824e-01
3.19181919e-01 2.94249713e-01 7.82066107e-01 -1.09028387e+00
8.59419167e-01 -1.31731951e+00 -2.92986959e-01 -3.23474824e-01
-2.66771674e-01 -5.54909348e-01 3.32698256e-01 -4.19043601e-01
-9.36031818e-01 -3.67306679e-01 -6.55901581e-02 5.32607794e-01
-3.85051697e-01 1.03603399e+00 4.55006093e-01 6.67320311e-01
6.51446283e-01 -2.06154034e-01 -3.91433001e-01 -3.96955878e-01
1.66454226e-01 8.11145544e-01 3.50551695e-01 -2.48923361e-01
3.27403337e-01 5.88975012e-01 -4.08392280e-01 -1.08817589e+00
-9.95662093e-01 -9.09330904e-01 -4.44222391e-01 -3.51820201e-01
6.60841763e-01 -9.89890993e-01 -8.28425765e-01 2.39318699e-01
-9.46001887e-01 -3.66927981e-01 -4.43214685e-01 4.66187388e-01
-4.63916928e-01 2.83261593e-02 -7.67072141e-01 -7.37137616e-01
-2.23048124e-02 -5.23860633e-01 5.67714751e-01 8.86180177e-02
-1.04857147e+00 -1.37246871e+00 1.60201430e-01 -1.22396350e-01
3.62673879e-01 8.19352984e-01 5.04515767e-01 -9.15116787e-01
2.00664520e-01 -5.02789319e-01 6.68561161e-02 -5.05227566e-01
3.83128673e-01 3.02896351e-01 -6.14713550e-01 5.92857487e-02
-3.07230595e-02 -3.91554326e-01 4.14065570e-01 2.68866152e-01
6.40444696e-01 -6.01034760e-01 -5.22036850e-01 -2.75636405e-01
1.36602080e+00 2.86311239e-01 3.28993946e-01 8.62466931e-01
1.61544666e-01 8.19286525e-01 6.89356685e-01 1.13094020e+00
8.98662984e-01 6.55996203e-01 2.04319790e-01 4.23065811e-01
3.12790364e-01 -3.05427641e-01 3.42453897e-01 6.60659075e-01
1.52888730e-01 -4.03449118e-01 -1.15087652e+00 1.04993701e+00
-1.68570018e+00 -1.35527861e+00 -3.02407175e-01 1.88567841e+00
1.20029807e+00 4.18256313e-01 4.42441106e-01 8.15064013e-02
5.45567930e-01 5.60553789e-01 -8.38158652e-02 -4.24497008e-01
8.43496770e-02 -4.55923053e-03 5.85255265e-01 5.13904870e-01
-1.19254494e+00 8.93453360e-01 7.44652081e+00 6.63646579e-01
-8.75644803e-01 1.71825022e-01 1.83462381e-01 -2.05942512e-01
-2.34135807e-01 1.88539639e-01 -8.84245634e-01 6.13855600e-01
1.45722663e+00 -6.45452678e-01 2.20312104e-01 4.91423279e-01
9.28823829e-01 -5.69304585e-01 -9.06655610e-01 6.60636008e-01
-3.77987325e-02 -1.31970477e+00 -3.62493813e-01 -5.23287170e-02
3.51507127e-01 1.69983298e-01 -3.48175615e-01 2.06150889e-01
2.89103121e-01 -7.12304175e-01 1.09234071e+00 3.55460107e-01
7.42358148e-01 -6.79277599e-01 5.09205997e-01 4.85471249e-01
-1.32332456e+00 2.00746462e-01 2.44940519e-01 -6.30712152e-01
5.85598290e-01 7.18848526e-01 -9.72610533e-01 1.45136103e-01
7.92037368e-01 1.05236578e+00 -3.80914778e-01 8.04056168e-01
-3.22388083e-01 9.25469100e-01 -5.72484851e-01 -4.06130821e-01
1.94417879e-01 -7.34620839e-02 8.00435483e-01 1.69508815e+00
3.10982000e-02 2.46684194e-01 6.94071800e-02 3.24809551e-01
-7.40822926e-02 3.26808572e-01 -9.12662983e-01 -6.89928353e-01
1.02772737e+00 1.28385675e+00 -1.43255687e+00 -2.23613247e-01
-6.54108167e-01 4.03020471e-01 2.28513628e-01 2.60272861e-01
-9.09375608e-01 -4.53958899e-01 4.63727862e-01 4.73989904e-01
1.11470811e-01 -5.34064353e-01 -3.78359109e-02 -8.89760792e-01
-3.08069829e-02 -5.87811172e-01 6.11051917e-01 -8.89205277e-01
-1.30840087e+00 5.20364702e-01 6.79887116e-01 -1.00656450e+00
-5.31864762e-01 -7.92960301e-02 -4.67952609e-01 5.73574662e-01
-8.68192732e-01 -8.88530731e-01 -1.26272619e-01 3.48812044e-01
5.28372765e-01 4.39183146e-01 7.65265584e-01 4.49599713e-01
-3.49986106e-01 8.72211754e-02 -3.60719383e-01 9.57205370e-02
8.56647670e-01 -1.00759637e+00 5.05374670e-01 5.22473812e-01
7.48602077e-02 7.96731293e-01 1.13023996e+00 -1.09243381e+00
-9.02220428e-01 -7.06467867e-01 1.69025254e+00 -8.60038519e-01
1.41171992e+00 -2.59862304e-01 -5.40628910e-01 1.27973235e+00
6.02457583e-01 -4.00534570e-01 1.00821316e+00 2.42162272e-01
-1.03378810e-01 3.55761170e-01 -9.72507238e-01 5.76300740e-01
8.46594870e-01 -7.27992713e-01 -1.01435399e+00 6.35674953e-01
7.04491973e-01 -6.85033560e-01 -9.20484006e-01 2.08339607e-03
4.11365539e-01 -4.17084903e-01 5.74215174e-01 -5.39830744e-01
3.35197955e-01 -4.35456745e-02 2.50617247e-02 -1.11776650e+00
2.04402372e-01 -7.87761331e-01 1.22617804e-01 1.70927012e+00
5.67041159e-01 -5.68957806e-01 2.02672645e-01 6.45850420e-01
-1.20783947e-01 -4.10976857e-01 -7.84709573e-01 -5.93082726e-01
-1.88471705e-01 -1.07740152e+00 4.33621109e-01 1.33225071e+00
6.15769625e-01 2.62891561e-01 -1.37054235e-01 -3.99016142e-01
-7.77676255e-02 -2.22097635e-01 4.30572510e-01 -1.26955831e+00
3.00165325e-01 -3.94944936e-01 -2.24347651e-01 -5.54328561e-01
6.41613975e-02 -6.92340791e-01 -2.75361210e-01 -1.70339501e+00
1.76870778e-01 -4.44918163e-02 3.21249843e-01 6.52661026e-01
2.42539689e-01 3.07572722e-01 -2.75407106e-01 2.32691973e-01
-5.05247653e-01 9.86959320e-03 7.57103443e-01 3.37191284e-01
-6.11505091e-01 -3.49972844e-01 -7.36669600e-01 9.66889262e-01
6.57032073e-01 -6.99906945e-01 -2.66484112e-01 2.21809939e-01
9.42274749e-01 -1.03689954e-01 3.62434722e-02 -5.79132915e-01
1.08821973e-01 -5.25679767e-01 -1.75554186e-01 -8.35870326e-01
3.31905901e-01 -6.66508615e-01 4.69079167e-01 -4.02384810e-02
-4.19197500e-01 3.28245789e-01 4.97227043e-01 4.83072817e-01
-3.54763746e-01 -1.57233730e-01 2.79299051e-01 -3.68378669e-01
-8.12349856e-01 1.06306009e-01 -9.62437749e-01 5.58143377e-01
1.27204633e+00 -3.14299375e-01 -3.12794358e-01 -7.21947670e-01
-8.99441242e-01 3.77559997e-02 4.05729830e-01 5.79519033e-01
4.70703002e-03 -1.23805296e+00 -4.44391042e-01 -4.77183461e-01
2.74166286e-01 -4.48703319e-01 -1.63518220e-01 7.07167566e-01
-5.46842158e-01 3.70377511e-01 -1.40718833e-01 1.39875218e-01
-8.51856053e-01 5.07122099e-01 -2.30772451e-01 -3.13989341e-01
-8.71178567e-01 5.16436934e-01 -4.91117686e-01 1.93711519e-01
1.04129873e-01 -4.79599088e-01 -7.89367497e-01 1.03186953e+00
7.01458693e-01 4.83784974e-01 1.26401614e-02 -1.08296263e+00
-5.59311271e-01 -8.18066113e-03 1.33068562e-01 -6.03337467e-01
1.60617268e+00 -5.94972908e-01 -4.00082558e-01 1.43918633e+00
1.15147662e+00 4.04639035e-01 -5.81356108e-01 -9.53128785e-02
5.20298600e-01 -1.05733790e-01 -1.57392263e-01 -4.14442271e-01
-2.72044808e-01 1.32919461e-01 -2.18630895e-01 1.07599247e+00
8.63553703e-01 2.18201429e-01 9.18627620e-01 1.68257747e-02
2.96635747e-01 -1.38537836e+00 -5.36792278e-02 9.91854429e-01
9.46214378e-01 -8.81654620e-01 2.40453586e-01 -4.40068007e-01
-5.86045146e-01 1.04371822e+00 1.63127437e-01 8.09667185e-02
1.10415006e+00 4.55196947e-01 3.64965685e-02 -6.39745772e-01
-7.02630162e-01 -4.38151479e-01 -5.89609370e-02 1.27299979e-01
8.89851511e-01 -7.23461360e-02 -1.29408610e+00 7.64006555e-01
-5.52067757e-01 -3.17216888e-02 8.81864250e-01 1.23217249e+00
-3.42224777e-01 -8.92273962e-01 -1.21826023e-01 3.51567954e-01
-1.17211473e+00 -1.72882363e-01 -5.45288861e-01 1.11736059e+00
-5.35755754e-02 1.29673600e+00 3.25445622e-01 -8.51121545e-02
1.96794003e-01 1.44523486e-01 9.70242247e-02 -1.00812387e+00
-7.74409473e-01 -1.14215106e-01 8.12168717e-01 -4.95847613e-01
-8.28994334e-01 -9.23160911e-01 -1.51113522e+00 -2.64482468e-01
-4.64496091e-02 2.33980834e-01 7.55224645e-01 1.23488939e+00
6.76171258e-02 3.59733462e-01 2.81558365e-01 -8.76092315e-01
2.15097755e-01 -1.00814891e+00 -5.80988586e-01 4.88374263e-01
1.32979974e-01 -8.85794759e-01 -4.40019488e-01 2.65104502e-01]
|
[9.098980903625488, 9.448710441589355]
|
c4d40367-7b07-44eb-b216-f2987f58dd31
|
robot-task-planning-and-situation-handling-in
|
2210.01287
| null |
https://arxiv.org/abs/2210.01287v1
|
https://arxiv.org/pdf/2210.01287v1.pdf
|
Robot Task Planning and Situation Handling in Open Worlds
|
Automated task planning algorithms have been developed to help robots complete complex tasks that require multiple actions. Most of those algorithms have been developed for "closed worlds" assuming complete world knowledge is provided. However, the real world is generally open, and the robots frequently encounter unforeseen situations that can potentially break the planner's completeness. This paper introduces a novel algorithm (COWP) for open-world task planning and situation handling that dynamically augments the robot's action knowledge with task-oriented common sense. In particular, common sense is extracted from Large Language Models based on the current task at hand and robot skills. For systematic evaluations, we collected a dataset that includes 561 execution-time situations in a dining domain, where each situation corresponds to a state instance of a robot being potentially unable to complete a task using a solution that normally works. Experimental results show that our approach significantly outperforms competitive baselines from the literature in the success rate of service tasks. Additionally, we have demonstrated COWP using a mobile manipulator. Supplementary materials are available at: https://cowplanning.github.io/
|
['Shiqi Zhang', 'Chad Esselink', 'Hao Yang', 'Nieqing Cao', 'Saeid Amiri', 'Xiaohan Zhang', 'Yan Ding']
|
2022-10-04
| null | null | null | null |
['common-sense-reasoning', 'robot-task-planning']
|
['reasoning', 'robots']
|
[ 4.20352906e-01 5.58513463e-01 -6.61173910e-02 -3.31878096e-01
-7.64249325e-01 -7.43340492e-01 4.99636799e-01 8.86847377e-02
-5.05838692e-01 1.01793754e+00 2.16305554e-01 -3.17153007e-01
-3.46852392e-01 -5.20345271e-01 -7.83079445e-01 -2.69374937e-01
-3.47400576e-01 1.18126225e+00 3.70863318e-01 -6.54504836e-01
3.87538254e-01 2.39304587e-01 -1.36883628e+00 1.25315756e-01
7.01512873e-01 4.39799964e-01 1.13606203e+00 4.53552037e-01
1.87266693e-01 8.57833564e-01 -4.43346411e-01 1.94613799e-01
6.27806664e-01 1.24391571e-01 -1.61732626e+00 1.29447058e-01
-3.12929600e-01 -2.68756121e-01 -3.30900520e-01 9.74602759e-01
1.03721857e-01 5.14874995e-01 5.22848547e-01 -1.79211617e+00
4.71935831e-02 7.30820298e-01 9.35001858e-03 -2.16457769e-01
9.09865975e-01 5.03565133e-01 6.75473988e-01 -2.53448457e-01
9.94774163e-01 1.58973634e+00 2.60065883e-01 4.93765354e-01
-9.08858180e-01 -3.10435951e-01 2.50552595e-01 3.34547698e-01
-1.27775955e+00 -3.64685804e-01 4.21351373e-01 -4.49319184e-01
1.73300850e+00 -1.34400398e-01 4.04864430e-01 1.17496741e+00
8.07179749e-01 7.90668845e-01 7.88827896e-01 -3.90916437e-01
2.37103567e-01 -3.67098778e-01 -4.88749258e-02 4.40186054e-01
2.31060028e-01 2.27329835e-01 -3.02951038e-01 -6.23176135e-02
7.04122305e-01 -1.01244994e-01 1.07472904e-01 -7.75248289e-01
-1.91176832e+00 4.19227391e-01 2.97643870e-01 2.49650910e-01
-6.21954501e-01 2.23110557e-01 5.47608733e-01 6.29876077e-01
-2.76586622e-01 1.08986127e+00 -7.48080432e-01 -5.50008595e-01
9.92847048e-03 8.27377021e-01 1.25264001e+00 1.70451450e+00
8.19723427e-01 -5.08423209e-01 5.72550856e-02 3.99910301e-01
6.76814318e-02 4.03845072e-01 2.18114495e-01 -1.74951708e+00
9.80081499e-01 6.38044894e-01 9.56434667e-01 -7.19907820e-01
-1.00260985e+00 3.04356337e-01 -1.14879183e-01 2.60483980e-01
4.93279785e-01 -1.38838142e-01 -9.00526881e-01 1.37996030e+00
3.51254463e-01 -2.47101605e-01 6.30817354e-01 8.88294995e-01
3.29498976e-01 6.44954324e-01 2.08177231e-02 -2.68725067e-01
1.36102009e+00 -1.24673057e+00 -1.03918135e+00 -8.37536991e-01
9.19486940e-01 -6.46059871e-01 1.05814540e+00 6.56902373e-01
-8.83159935e-01 -3.24427575e-01 -9.57636058e-01 -1.42066061e-01
-3.34135920e-01 -3.06545407e-01 7.71204054e-01 -1.61975622e-01
-9.41753447e-01 5.00468731e-01 -1.03093410e+00 -6.68851435e-01
-3.95456962e-02 3.94965589e-01 -6.59996390e-01 -4.90444124e-01
-1.09280431e+00 1.47411323e+00 1.09390581e+00 1.22580901e-01
-1.45944345e+00 6.07308000e-02 -1.32835758e+00 -1.79379866e-01
1.30893219e+00 -5.61934888e-01 2.09177685e+00 -1.70167565e-01
-1.53519249e+00 3.90147805e-01 4.30925004e-02 -4.64045703e-01
3.90799910e-01 -3.89887542e-01 -3.24655809e-02 -8.87045711e-02
6.92172825e-01 6.31939292e-01 2.72395670e-01 -1.30754101e+00
-9.50939775e-01 -3.95324171e-01 6.61909521e-01 6.53609157e-01
5.64146578e-01 -5.66429347e-02 -2.49532998e-01 1.18497340e-02
4.67417002e-01 -1.37127554e+00 -7.82682717e-01 -5.16726732e-01
-4.87138540e-01 -4.35175925e-01 2.96498090e-01 -1.79775506e-01
6.38232231e-01 -2.01745868e+00 4.21907812e-01 -9.28513892e-03
-9.27416682e-02 -3.31946045e-01 -3.29301745e-01 9.90086198e-01
2.83635348e-01 -2.69583434e-01 -2.09458783e-01 -4.93926816e-02
3.74495983e-01 7.25619614e-01 -3.10435236e-01 2.76043594e-01
-9.72111821e-02 7.37531185e-01 -1.39085209e+00 -5.71740210e-01
3.53448838e-01 -3.03439826e-01 -5.12631595e-01 1.44231886e-01
-7.13195264e-01 6.29937351e-01 -7.54453540e-01 5.80525637e-01
2.31755480e-01 1.44373566e-01 6.58923864e-01 5.60265422e-01
-2.59296477e-01 5.66999972e-01 -1.27451599e+00 2.43445158e+00
-6.19870126e-01 9.93504003e-02 1.48476258e-01 -4.61900741e-01
5.71404636e-01 3.78655910e-01 3.68089378e-01 -3.57015908e-01
9.73937511e-02 3.39725286e-01 1.38681471e-01 -7.34007299e-01
8.95291984e-01 -6.38851477e-03 -6.59897506e-01 3.31320703e-01
-4.87768278e-02 -9.42409098e-01 4.86170739e-01 5.33863083e-02
1.50923872e+00 5.16211092e-01 7.67434120e-01 -1.59485027e-01
9.51194093e-02 9.68331516e-01 5.92739463e-01 1.03987193e+00
-6.15149438e-01 3.78312543e-02 4.34470266e-01 -6.43012166e-01
-6.84722543e-01 -7.86113381e-01 3.33119094e-01 1.14216650e+00
7.68019140e-01 -5.22127509e-01 -4.60596770e-01 -4.36545461e-01
-3.69089425e-01 1.11300492e+00 -2.94910073e-01 3.32990214e-02
-7.53668904e-01 4.39737327e-02 3.62436980e-01 3.47894609e-01
3.99594456e-01 -1.70254195e+00 -1.37285554e+00 4.50170934e-01
-8.10697138e-01 -1.43743992e+00 -2.93422997e-01 5.71993113e-01
-7.23712444e-01 -1.36086500e+00 -1.46363020e-01 -1.11852705e+00
5.40312588e-01 4.64988023e-01 9.76211250e-01 -3.64326954e-01
1.09754885e-02 3.98699641e-01 -6.25795960e-01 -7.19799399e-01
-6.04718208e-01 1.99663252e-01 3.82339627e-01 -9.89055037e-01
1.94128305e-01 -2.66420037e-01 -1.67789496e-02 4.42713410e-01
-4.35045034e-01 1.43479705e-01 7.80743361e-01 6.45321369e-01
4.47907269e-01 5.70704758e-01 3.75250757e-01 -6.37723923e-01
8.59993160e-01 -5.26610494e-01 -4.62637097e-01 2.03714043e-01
-3.82340640e-01 9.42657068e-02 2.49782428e-01 -1.88471258e-01
-1.50571203e+00 5.36746383e-01 3.15954030e-01 -5.36928512e-03
-7.65066922e-01 8.17593932e-01 -8.31505060e-02 3.86025846e-01
7.41091728e-01 8.96558836e-02 -8.65920559e-02 -3.13426733e-01
2.03860238e-01 5.10033965e-01 6.09416068e-01 -9.65456307e-01
4.90990937e-01 2.24918082e-01 1.24450931e-02 -3.76677930e-01
-4.60407555e-01 -7.32811928e-01 -7.62112081e-01 -5.84162101e-02
6.76162183e-01 -8.02212715e-01 -8.47143412e-01 3.73652905e-01
-1.44069839e+00 -1.05792236e+00 -2.08957404e-01 4.71987575e-01
-1.43678379e+00 1.92802161e-01 -4.22432899e-01 -1.01798534e+00
2.91053355e-01 -1.41305840e+00 1.19428468e+00 -3.81938368e-02
-4.92150217e-01 -4.04872805e-01 -8.08073059e-02 1.90405518e-01
5.00575639e-04 1.88701555e-01 6.42115831e-01 -7.44495630e-01
-4.13397938e-01 -1.15881830e-01 2.01158360e-01 -3.94935519e-01
2.67966270e-01 -8.37715209e-01 -3.36949229e-01 -2.28085503e-01
5.93109876e-02 -6.13067210e-01 6.97799698e-02 1.81860656e-01
4.88894671e-01 -3.24010611e-01 -8.02476466e-01 -2.15636238e-01
1.00164402e+00 6.19355977e-01 6.94722831e-01 8.66510689e-01
1.69724047e-01 1.06164980e+00 1.66277170e+00 4.86075103e-01
7.19993949e-01 7.37939596e-01 8.34167778e-01 6.56916440e-01
3.17091316e-01 -1.96131349e-01 5.38076401e-01 2.43448511e-01
-1.56155601e-01 -3.20981801e-01 -1.44805944e+00 7.93556690e-01
-2.24495673e+00 -8.75792861e-01 -8.39146972e-02 1.77109683e+00
7.41253078e-01 4.08562720e-01 -1.08957805e-01 -4.93065752e-02
4.79460120e-01 -1.25120670e-01 -5.90726197e-01 -3.56285125e-01
5.42670846e-01 -3.95015776e-01 5.11877775e-01 6.18965626e-01
-1.08287668e+00 1.29236841e+00 5.76458263e+00 3.74098897e-01
-1.78329870e-01 1.40535742e-01 -4.17210132e-01 -7.11563602e-02
3.63406211e-01 1.04573742e-01 -4.97596711e-01 -1.04744256e-01
6.09438360e-01 -1.80098742e-01 5.17894328e-01 1.14457512e+00
1.63025558e-01 -6.78078532e-01 -1.53538442e+00 7.38494992e-01
-3.98534060e-01 -6.31964505e-01 -4.27835315e-01 -1.15336277e-01
3.81661028e-01 2.82435030e-01 -2.28768408e-01 7.73224533e-01
9.17172313e-01 -8.50768030e-01 1.06025696e+00 2.22621202e-01
4.74643886e-01 -5.36387563e-01 7.58154690e-01 1.13644195e+00
-1.00061107e+00 -5.06049454e-01 -3.41254145e-01 -7.31515944e-01
5.58481157e-01 -1.65887788e-01 -1.70499766e+00 9.08002377e-01
6.71918571e-01 5.87331474e-01 1.07422009e-01 8.02963734e-01
-4.43410426e-01 -2.23374054e-01 -3.46331894e-01 -5.24218008e-02
4.54962105e-01 -1.81056524e-03 9.33011055e-01 7.11346507e-01
1.39873460e-01 4.22269493e-01 8.77163649e-01 5.15187860e-01
5.97026169e-01 -4.22048986e-01 -1.18630552e+00 4.83021960e-02
5.66814542e-01 6.60010397e-01 -7.74818122e-01 -1.59656540e-01
-6.32998869e-02 9.17088628e-01 1.48008332e-01 3.59428763e-01
-6.82742000e-01 -4.43625957e-01 6.78104639e-01 -2.04369709e-01
-1.58618078e-01 -6.99870706e-01 -6.33051172e-02 -6.99207425e-01
2.39408225e-01 -9.32699025e-01 5.23083389e-01 -1.12576485e+00
-7.85898685e-01 5.85577071e-01 5.54780245e-01 -1.42837548e+00
-5.76956332e-01 -6.00450218e-01 1.91945257e-03 5.35281956e-01
-1.50936973e+00 -8.40440869e-01 -1.53410003e-01 4.93044019e-01
1.26916385e+00 1.14083923e-01 1.03569424e+00 -3.17281783e-01
-1.33112177e-01 -2.56835699e-01 -4.84122127e-01 -3.42765957e-01
7.34706998e-01 -9.80317116e-01 5.93873680e-01 5.67020178e-01
-4.34762716e-01 7.07330287e-01 1.15598595e+00 -1.22918618e+00
-1.75162232e+00 -8.29803348e-01 8.82047653e-01 -8.15718234e-01
6.76843166e-01 -1.49441168e-01 -4.28934753e-01 1.53695965e+00
6.64462149e-02 -4.04279053e-01 5.12077250e-02 1.41428739e-01
6.75797462e-02 5.46964109e-01 -1.20307744e+00 7.55663872e-01
1.69689536e+00 -1.04593359e-01 -1.34608495e+00 7.70941257e-01
1.04724443e+00 -1.08349848e+00 -5.32612979e-01 3.90449852e-01
3.66861403e-01 -4.06735986e-01 7.54414499e-01 -8.59387219e-01
3.46377999e-01 -5.07340491e-01 -3.39165479e-01 -1.51839542e+00
-3.48435909e-01 -7.90983498e-01 8.19028914e-02 4.66044098e-01
5.34800887e-01 -6.84258521e-01 1.94413245e-01 9.41516578e-01
-7.72920847e-01 -3.29843044e-01 -9.33898389e-01 -1.26956379e+00
-3.17678124e-01 -7.15826631e-01 6.64862692e-01 7.25399435e-01
7.57279277e-01 2.84856558e-01 -1.80362865e-01 5.24590671e-01
2.24406600e-01 3.65716517e-02 9.13859010e-01 -1.18288803e+00
2.32287109e-01 -3.68110999e-03 -2.24117249e-01 -1.08267713e+00
5.92536867e-01 -7.10756898e-01 1.09901547e+00 -2.03556514e+00
-5.56414656e-04 -7.16496944e-01 4.82449196e-02 9.26391780e-01
4.33136195e-01 -5.81063449e-01 4.10869688e-01 3.37245405e-01
-1.14781475e+00 3.71746927e-01 1.73767495e+00 -2.37765118e-01
-6.18460178e-01 3.27181518e-02 -4.61739361e-01 1.00442231e+00
1.04771459e+00 -3.53011340e-01 -5.95467746e-01 -7.57193923e-01
4.55476284e-01 4.64111686e-01 2.40437058e-03 -1.06624162e+00
5.96309721e-01 -9.16906059e-01 -4.58092153e-01 -2.76328206e-01
6.40404642e-01 -1.22729206e+00 3.05034876e-01 7.50383556e-01
-4.19795215e-01 5.72715662e-02 3.10400158e-01 8.33379447e-01
-2.77605467e-02 -4.13871020e-01 1.44642284e-02 -4.50985163e-01
-1.60770631e+00 -2.05322102e-01 -7.90338039e-01 -1.72093287e-01
1.56087255e+00 -2.03570515e-01 -3.10566902e-01 -2.60725260e-01
-9.39373076e-01 8.91335964e-01 4.97859627e-01 6.24558091e-01
5.85467041e-01 -8.83391380e-01 -4.99314040e-01 -1.69507086e-01
4.18702483e-01 6.22722387e-01 3.73674482e-02 7.01477468e-01
-5.37903607e-01 7.18719840e-01 -6.02136791e-01 -3.42362523e-01
-7.19213545e-01 8.09620142e-01 4.79392111e-02 -3.37030739e-01
-8.07297885e-01 5.14204204e-01 2.71856278e-01 -8.64665866e-01
3.87673020e-01 -7.68071949e-01 -3.00586462e-01 -3.60908091e-01
3.47885549e-01 4.37146485e-01 4.42878082e-02 -5.25883079e-01
-6.95508957e-01 3.98873761e-02 1.60514936e-02 -2.61587560e-01
1.13715410e+00 -4.16319638e-01 -6.06234148e-02 6.89764321e-01
2.38851085e-01 -6.59714937e-01 -1.20883751e+00 -2.70335436e-01
3.70707899e-01 -4.59673554e-01 -4.75986987e-01 -1.13104594e+00
-7.00242892e-02 2.36854285e-01 -6.77003413e-02 1.67950720e-01
6.88408911e-01 2.47586399e-01 6.64606929e-01 1.43258059e+00
1.65478015e+00 -1.29062080e+00 1.46789148e-01 1.36810946e+00
1.23731470e+00 -1.42746353e+00 -1.67777047e-01 -7.19081938e-01
-9.71501052e-01 8.92119467e-01 9.81040895e-01 2.10964605e-01
3.03434227e-02 3.48855674e-01 1.20244928e-01 -3.85176152e-01
-9.01080489e-01 -2.26783529e-01 -4.57626134e-01 9.45828259e-01
-2.50859439e-01 3.12597901e-01 -1.36648700e-01 6.50170982e-01
-4.60942507e-01 8.07607546e-02 9.39354837e-01 1.72737110e+00
-6.26760781e-01 -7.91743338e-01 -4.45387930e-01 2.57621706e-01
4.44931723e-02 3.39162767e-01 -4.66828614e-01 8.85664880e-01
-4.75644991e-02 1.41153145e+00 -5.32350726e-02 -2.09054619e-01
7.10341036e-01 2.21787155e-01 7.48796523e-01 -1.32969737e+00
-1.96176201e-01 -3.69663894e-01 6.60234272e-01 -1.15961087e+00
-4.67604518e-01 -8.31780016e-01 -1.94543087e+00 6.35142773e-02
-1.39611825e-01 -2.75588091e-02 5.23815215e-01 1.11705053e+00
1.85176507e-01 5.35917222e-01 5.23547176e-03 -1.40453994e+00
-7.90799975e-01 -1.02083933e+00 -4.65836138e-01 1.81796357e-01
2.96707422e-01 -1.16301513e+00 7.72209838e-02 1.30142868e-01]
|
[4.429544448852539, 0.910805881023407]
|
b343ae6d-53ed-4c6c-97eb-7904ece8a4a9
|
generalizing-fingerprint-spoof-detector
|
1901.03918
| null |
http://arxiv.org/abs/1901.03918v2
|
http://arxiv.org/pdf/1901.03918v2.pdf
|
Generalizing Fingerprint Spoof Detector: Learning a One-Class Classifier
|
Prevailing fingerprint recognition systems are vulnerable to spoof attacks.
To mitigate these attacks, automated spoof detectors are trained to distinguish
a set of live or bona fide fingerprints from a set of known spoof fingerprints.
Despite their success, spoof detectors remain vulnerable when exposed to
attacks from spoofs made with materials not seen during training of the
detector. To alleviate this shortcoming, we approach spoof detection as a
one-class classification problem. The goal is to train a spoof detector on only
the live fingerprints such that once the concept of "live" has been learned,
spoofs of any material can be rejected. We accomplish this through training
multiple generative adversarial networks (GANS) on live fingerprint images
acquired with the open source, dual-camera, 1900 ppi RaspiReader fingerprint
reader. Our experimental results, conducted on 5.5K spoof images (from 12
materials) and 11.8K live images show that the proposed approach improves the
cross-material spoof detection performance over state-of-the-art one-class and
binary class spoof detectors on 11 of 12 testing materials and 7 of 12 testing
materials, respectively.
|
['Joshua J. Engelsma', 'Anil K. Jain']
|
2019-01-13
| null | null | null | null |
['one-class-classifier']
|
['methodology']
|
[ 1.11741340e+00 -1.40768737e-01 -1.91099405e-01 -2.14355052e-01
-3.15319031e-01 -1.07268357e+00 6.18444383e-01 -2.08491132e-01
4.28373972e-03 5.25639713e-01 -5.46952963e-01 -4.32119876e-01
-3.87687907e-02 -1.13925028e+00 -1.24985111e+00 -7.11814404e-01
-7.54609704e-02 4.49396908e-01 2.79413432e-01 -1.97324641e-02
4.10369128e-01 7.86984205e-01 -1.31686437e+00 4.47309136e-01
6.43763363e-01 9.00734782e-01 -1.58802882e-01 1.04927027e+00
2.11859465e-01 4.10053015e-01 -8.49689364e-01 -7.84026563e-01
4.76217240e-01 -1.06649578e-01 -6.58882439e-01 -3.24165344e-01
9.38421667e-01 -5.76227665e-01 -4.16085452e-01 1.43354762e+00
2.10300863e-01 -6.29215539e-01 4.32204336e-01 -1.35358083e+00
-6.20900929e-01 5.06329715e-01 -4.41494524e-01 -3.46703641e-02
4.68832135e-01 -1.81367099e-02 4.52612072e-01 -4.72993523e-01
4.86579657e-01 1.21403146e+00 8.51642072e-01 8.18423569e-01
-1.30950773e+00 -1.41359472e+00 -7.18493104e-01 -2.95471281e-01
-9.43463802e-01 -8.00850213e-01 1.02052474e+00 -4.90050912e-01
4.66549069e-01 1.88236400e-01 1.95250884e-01 1.80167651e+00
5.99114299e-01 3.34414005e-01 1.37176311e+00 -3.89215857e-01
-3.12523335e-01 4.54509139e-01 2.09202737e-01 7.59148240e-01
9.66924310e-01 8.83889794e-01 -4.07366306e-01 -5.94481885e-01
8.53814662e-01 1.75081491e-01 -4.04839329e-02 -3.52292925e-01
-9.88210618e-01 6.89104736e-01 2.67071009e-01 7.01388195e-02
-3.31971109e-01 2.22632572e-01 2.10024819e-01 3.98879260e-01
-1.88018009e-01 4.18488443e-01 -3.37089822e-02 3.33087116e-01
-9.19999599e-01 2.18975693e-02 8.59922290e-01 5.61853886e-01
8.06498110e-01 -2.47204632e-01 3.25970232e-01 7.52201676e-01
5.72507441e-01 1.37402213e+00 1.42617494e-01 -4.20282423e-01
4.97454554e-01 4.37744819e-02 3.10375929e-01 -1.76368415e+00
2.58596241e-01 -2.52876073e-01 -5.24811566e-01 1.24704443e-01
1.91195592e-01 1.17441647e-01 -1.28524745e+00 1.20182943e+00
2.56209522e-02 1.62188455e-01 -1.30065791e-02 3.95337105e-01
3.15062582e-01 4.33801740e-01 -2.95347840e-01 1.95267454e-01
1.13331282e+00 -8.41612637e-01 -4.94004697e-01 -4.41329300e-01
-1.60691351e-01 -1.06549990e+00 4.45724636e-01 4.48421508e-01
-2.79083759e-01 -7.92947829e-01 -1.37856054e+00 7.05172181e-01
-5.11717021e-01 -3.30239624e-01 5.89948714e-01 1.81753361e+00
-5.48643887e-01 6.01030827e-01 -7.24378109e-01 -2.25297585e-01
6.38975024e-01 5.39763331e-01 -4.29341197e-01 -3.93713504e-01
-1.27456057e+00 3.31850380e-01 3.73469442e-01 -8.21729898e-02
-1.62404573e+00 -2.71798551e-01 -3.26581687e-01 -3.70698422e-01
-7.56751969e-02 -2.02786133e-01 6.70778573e-01 -9.95226562e-01
-1.32640707e+00 1.07429135e+00 1.54101908e-01 -4.92570311e-01
6.01765990e-01 -2.51268744e-01 -1.24932170e+00 2.56230295e-01
1.00393608e-01 -2.21678317e-02 1.57216227e+00 -1.72812724e+00
-4.35957015e-01 -3.00633907e-01 -2.64609069e-01 -1.00211287e+00
-2.36735463e-01 2.42842570e-01 1.80887394e-02 -4.52300966e-01
1.42988965e-01 -1.24146140e+00 4.67422754e-01 -2.86670715e-01
-9.17436898e-01 5.57536960e-01 1.28793120e+00 -8.55204344e-01
6.83521748e-01 -2.00010490e+00 -8.52295280e-01 8.24334025e-01
-1.22511253e-01 9.38510418e-01 -1.53372645e-01 2.17966199e-01
1.45386625e-02 2.77837813e-01 -1.31984487e-01 -2.14152858e-02
-3.17124128e-01 1.37340948e-01 -5.29559135e-01 8.36397469e-01
8.67788270e-02 5.66979051e-01 -1.12642384e+00 -4.12032634e-01
3.35452974e-01 7.18595862e-01 -3.63606334e-01 2.36731678e-01
2.76518553e-01 4.83064473e-01 -5.50518513e-01 1.10184252e+00
1.34534407e+00 -1.40100271e-01 -2.34551318e-02 -8.06778595e-02
4.05750722e-01 1.82187274e-01 -1.06181908e+00 1.05599272e+00
-2.12674886e-01 5.42786777e-01 1.29566088e-01 -7.19708383e-01
1.27866733e+00 2.07659110e-01 2.58360088e-01 -4.56644088e-01
-4.00219783e-02 5.70712984e-01 -1.34546131e-01 -5.93235195e-01
3.45987916e-01 7.78310597e-02 9.44318026e-02 4.64608490e-01
1.06811672e-01 3.46566349e-01 -3.09772253e-01 -1.14065729e-01
1.28311932e+00 -1.49854168e-01 -5.50212264e-01 -9.58067402e-02
5.68423748e-01 -3.88766438e-01 4.55699176e-01 1.52257979e+00
-4.09353435e-01 3.44959557e-01 -1.58567101e-01 -2.55056709e-01
-9.91523802e-01 -1.34864533e+00 -5.07156014e-01 6.70526266e-01
4.82087731e-01 1.41294688e-01 -4.75749284e-01 -9.02618766e-01
5.30360043e-01 9.27129611e-02 -6.96259975e-01 -6.59454837e-02
-7.31631577e-01 -5.85699618e-01 1.41933286e+00 1.31628197e-02
7.77541995e-01 -9.89752114e-01 -3.23127210e-01 4.02832687e-01
-1.18484706e-01 -1.05802608e+00 9.02683735e-02 -1.08792566e-01
-4.55447018e-01 -1.44336581e+00 -5.48078418e-01 -8.60811710e-01
7.88812339e-01 4.58194792e-01 6.61994219e-01 8.17706808e-02
-2.57808268e-01 9.39858928e-02 -2.04731405e-01 -3.75433475e-01
-1.18487167e+00 -3.11469704e-01 1.76655918e-01 4.97904927e-01
1.94069281e-01 -3.71330619e-01 -5.29831409e-01 8.11256230e-01
-8.03977013e-01 -1.68855846e-01 5.12523532e-01 8.88445437e-01
2.95508802e-01 3.42273712e-01 3.45681816e-01 -1.15429127e+00
2.22357586e-01 -5.41688025e-01 -5.33395886e-01 5.46875536e-01
-8.30323398e-01 3.30102779e-02 5.05706906e-01 -6.95314825e-01
-9.94419277e-01 -1.31499588e-01 2.84794793e-02 -2.92897046e-01
-1.28474846e-01 -2.01821834e-01 -2.96226650e-01 -1.00351572e+00
8.51415634e-01 3.39419067e-01 -1.86254263e-01 -3.96937758e-01
-5.21887280e-02 1.11318791e+00 1.06648350e+00 -6.56593800e-01
1.19480538e+00 6.46269679e-01 1.51470285e-02 -7.45553076e-01
-1.11627299e-02 -4.74266768e-01 -1.29075889e-02 -4.06989932e-01
3.50416094e-01 -4.34092999e-01 -8.43613029e-01 1.40198576e+00
-1.18784487e+00 2.82642215e-01 5.45977235e-01 4.41689864e-02
-7.52482712e-02 6.96611404e-01 -7.69514561e-01 -9.40184116e-01
-4.58149225e-01 -1.02653122e+00 8.17863166e-01 3.68465364e-01
2.02412933e-01 -7.69691229e-01 2.80099571e-01 6.40578032e-01
7.16715574e-01 7.72119701e-01 4.55575049e-01 -7.63788760e-01
-4.02247071e-01 -1.03282452e+00 -2.50051379e-01 7.32868075e-01
6.83851480e-01 2.73912966e-01 -1.25570536e+00 -5.80606580e-01
-7.73817347e-03 -1.98133454e-01 9.77389574e-01 -1.10116959e-01
8.94789338e-01 -3.37597847e-01 -9.17514563e-01 8.30072939e-01
1.61631477e+00 5.23474336e-01 7.49571323e-01 2.48651803e-01
9.38278198e-01 5.77218890e-01 3.13317180e-01 8.54564011e-02
-4.35544699e-01 5.04214466e-01 6.53922558e-01 1.71925366e-01
-4.96148095e-02 -7.31068313e-01 3.44800413e-01 -2.33274624e-02
1.28192782e-01 -5.82226276e-01 -1.08316481e+00 2.57092863e-01
-1.08128595e+00 -1.11631298e+00 -4.02240083e-02 2.43131614e+00
3.93897682e-01 3.89602125e-01 -6.35831133e-02 4.40071762e-01
1.49981618e+00 2.45095074e-01 -4.51992750e-01 -3.01469982e-01
-3.04908931e-01 2.96467245e-01 1.12333369e+00 6.02131486e-01
-1.45416844e+00 8.05663824e-01 6.06147861e+00 3.62911403e-01
-1.45219660e+00 -5.55445030e-02 2.91748762e-01 5.49140394e-01
-2.99469352e-01 -4.60248105e-02 -8.76698971e-01 9.22154903e-01
1.00876129e+00 4.83006895e-01 7.60924280e-01 6.76558018e-01
-4.57700461e-01 3.32380980e-01 -9.26663816e-01 8.33221197e-01
1.82752788e-01 -1.27835071e+00 -2.52351403e-01 3.80257815e-01
8.15431237e-01 -4.07681055e-02 2.26947203e-01 -1.25296146e-01
3.68386030e-01 -1.05886996e+00 5.05533516e-01 3.78605157e-01
8.11192214e-01 -5.28362632e-01 5.95589459e-01 1.09494887e-01
-8.79531622e-01 -1.08685501e-01 -2.67713487e-01 6.25630617e-01
4.51298170e-02 6.81095779e-01 -9.97652054e-01 4.71214950e-01
5.21170497e-01 1.94173843e-01 -3.58060598e-01 8.00178587e-01
-2.11883083e-01 8.13313067e-01 -3.92495871e-01 2.59975046e-01
-1.65397182e-01 6.25911534e-01 6.20780945e-01 1.43937337e+00
2.00606242e-01 -5.91012895e-01 -3.19111764e-01 8.79555225e-01
-2.21352354e-01 -5.91786206e-01 -7.71307111e-01 -5.66088259e-01
7.76516855e-01 7.34307110e-01 -6.65493071e-01 -1.31723106e-01
9.44425017e-02 1.21105623e+00 -3.19892943e-01 2.00295463e-01
-8.25978100e-01 -8.04582536e-01 5.76142788e-01 8.20323676e-02
3.20916206e-01 5.02131544e-02 -2.13264637e-02 -8.99023116e-01
4.29567546e-02 -1.04075480e+00 9.41659138e-02 -1.41685203e-01
-1.37475204e+00 3.81719649e-01 -6.03976905e-01 -7.95345902e-01
2.18380652e-02 -7.29302585e-01 -3.20227861e-01 9.28572416e-01
-1.43964183e+00 -1.32236516e+00 -2.85856336e-01 6.12020373e-01
-3.13011080e-01 -3.09125006e-01 1.00824499e+00 5.46643555e-01
-3.05979282e-01 9.59893405e-01 3.49638790e-01 6.30184114e-01
7.52321243e-01 -5.23223698e-01 9.66606617e-01 1.17160547e+00
4.97502461e-02 1.04361463e+00 6.96950316e-01 -1.18361914e+00
-1.62075257e+00 -8.53108108e-01 8.83184612e-01 -4.01021510e-01
3.81889343e-01 -2.70369202e-01 -8.13190758e-01 5.60489595e-01
-4.59369749e-01 5.27165651e-01 6.20381474e-01 -4.14574414e-01
-9.95745480e-01 -2.06922039e-01 -1.92180002e+00 -3.53769928e-01
8.55840743e-01 -1.05930650e+00 -1.79690063e-01 2.31553450e-01
3.61541629e-01 -2.91996747e-01 -7.40736246e-01 5.13406336e-01
1.53246737e+00 -1.03862524e+00 1.35901701e+00 -5.08779407e-01
3.28518003e-01 -2.82126844e-01 -2.22335547e-01 -6.56200707e-01
-1.46687418e-01 -8.65319967e-01 -3.22926760e-01 1.52658963e+00
1.12415301e-02 -1.00729811e+00 1.11155200e+00 -6.88106893e-03
3.92610252e-01 2.04114579e-02 -5.65315366e-01 -1.14525652e+00
-3.70620638e-01 -1.78277358e-01 8.37947428e-01 1.12697339e+00
-5.74255109e-01 -4.65400219e-01 -9.31099415e-01 7.85498977e-01
1.45838583e+00 -1.02483384e-01 8.42494667e-01 -1.27336907e+00
-6.62986457e-01 2.38598082e-02 -6.50074959e-01 -8.65878463e-01
-5.23355044e-03 -6.27092719e-01 9.11401361e-02 -6.89370275e-01
-1.35966213e-04 -8.77746940e-01 -8.09182823e-01 3.32730263e-01
7.70618692e-02 5.06405890e-01 -3.53531027e-03 4.78841513e-01
1.92898437e-01 -4.91452426e-01 9.46595788e-01 -5.69812953e-01
1.28009632e-01 3.75566453e-01 -2.42151290e-01 2.01794952e-01
5.82913518e-01 -1.04908848e+00 -1.46331206e-01 -2.36941561e-01
4.60164435e-02 2.39199787e-01 8.16184819e-01 -1.31924284e+00
-1.62954330e-02 -2.67154813e-01 4.34861958e-01 -2.73347735e-01
4.44659665e-02 -9.05071557e-01 6.47078037e-01 1.10575843e+00
-2.08365396e-01 -5.88715732e-01 -2.73214541e-02 6.21167839e-01
2.32352600e-01 -1.50883362e-01 8.98344159e-01 1.98534876e-01
-4.46784198e-01 2.27313861e-01 1.93357199e-01 -4.99253869e-01
6.65098429e-01 -5.73775649e-01 -9.41790521e-01 2.20285594e-01
-2.62252778e-01 -4.67960954e-01 8.47118974e-01 5.08360565e-01
6.37027144e-01 -1.11132014e+00 -5.53569674e-01 9.47395980e-01
2.05846637e-01 -5.74833870e-01 5.97146712e-02 -4.14790958e-02
-9.72778440e-01 5.78145146e-01 -7.82448828e-01 -5.64571798e-01
-1.28008080e+00 6.34979904e-01 3.81792426e-01 -2.27765515e-01
-7.47719631e-02 1.21445143e+00 -2.89492577e-01 -3.49007159e-01
1.47148296e-01 2.57452935e-01 3.40800405e-01 -6.81868374e-01
5.54000497e-01 5.56325376e-01 1.61775220e-02 -8.14774632e-01
-7.42459774e-01 4.41905171e-01 -2.37767547e-01 5.29496633e-02
8.72755647e-01 2.74865448e-01 -2.62198329e-04 -1.94139391e-01
1.35592163e+00 4.01920855e-01 -1.04268658e+00 -1.26636371e-01
-2.56271660e-01 -1.06138909e+00 -4.53377098e-01 -1.03674817e+00
-9.40213621e-01 6.83961749e-01 1.16990638e+00 5.49953938e-01
6.50905728e-01 -3.15179646e-01 1.07061756e+00 2.02612832e-01
8.73587966e-01 -5.24681807e-01 4.98181023e-02 1.42943501e-01
3.62338126e-01 -1.30516672e+00 -4.26448882e-01 -5.94875932e-01
2.53600270e-01 1.01463056e+00 1.40340716e-01 -3.98519844e-01
4.75661457e-01 2.00211957e-01 3.88972647e-02 -1.27416164e-01
3.90243709e-01 8.44280839e-01 2.35807337e-02 1.11573315e+00
-8.12159926e-02 5.20365298e-01 3.42147768e-01 -1.10239703e-02
-2.40306348e-01 1.35965738e-02 1.47327766e-01 1.26642847e+00
-3.95383537e-01 -1.41314590e+00 -9.39360976e-01 3.28806311e-01
-9.04099643e-01 3.65551710e-01 -3.69536132e-01 1.52973875e-01
4.99123096e-01 1.20563591e+00 -2.76183546e-01 -1.11086011e+00
-1.30452260e-01 -2.35829160e-01 7.15029061e-01 -7.89502710e-02
-5.21064699e-01 -5.88535845e-01 1.39112040e-01 -4.01093453e-01
-4.13444698e-01 -4.36345518e-01 -5.83498657e-01 -6.59547448e-01
-6.30499423e-01 -5.57443760e-02 1.24671984e+00 7.58782744e-01
8.94025937e-02 -1.60028394e-02 1.21704674e+00 -5.77243865e-01
-6.01609409e-01 -4.59424227e-01 -6.78123176e-01 6.54702306e-01
9.47824359e-01 -6.30453289e-01 -3.92243445e-01 5.77802807e-02]
|
[12.982288360595703, 1.122490406036377]
|
38ec22d7-d3bc-4a95-8e67-e343605ec9d6
|
biscuit-causal-representation-learning-from
|
2306.09643
| null |
https://arxiv.org/abs/2306.09643v1
|
https://arxiv.org/pdf/2306.09643v1.pdf
|
BISCUIT: Causal Representation Learning from Binary Interactions
|
Identifying the causal variables of an environment and how to intervene on them is of core value in applications such as robotics and embodied AI. While an agent can commonly interact with the environment and may implicitly perturb the behavior of some of these causal variables, often the targets it affects remain unknown. In this paper, we show that causal variables can still be identified for many common setups, e.g., additive Gaussian noise models, if the agent's interactions with a causal variable can be described by an unknown binary variable. This happens when each causal variable has two different mechanisms, e.g., an observational and an interventional one. Using this identifiability result, we propose BISCUIT, a method for simultaneously learning causal variables and their corresponding binary interaction variables. On three robotic-inspired datasets, BISCUIT accurately identifies causal variables and can even be scaled to complex, realistic environments for embodied AI.
|
['Efstratios Gavves', 'Taco Cohen', 'Yuki M. Asano', 'Sindy Löwe', 'Sara Magliacane', 'Phillip Lippe']
|
2023-06-16
| null | null | null | null |
['causal-discovery', 'causal-identification']
|
['knowledge-base', 'reasoning']
|
[ 4.27151442e-01 3.76126915e-01 -3.58784378e-01 -5.73012186e-03
-5.14822314e-03 -8.27558339e-01 7.75920033e-01 1.59608677e-01
-2.91163743e-01 1.20430136e+00 3.54085594e-01 -2.75525659e-01
-7.43314505e-01 -7.86628246e-01 -1.35097754e+00 -1.02890873e+00
-4.44706470e-01 6.77493036e-01 -1.50861636e-01 1.14989206e-01
-1.07757285e-01 2.53567666e-01 -1.43625808e+00 -4.47087258e-01
5.15126884e-01 2.03940108e-01 3.84513080e-01 7.52473176e-01
3.59779567e-01 9.42569733e-01 -4.81353015e-01 1.25429511e-01
-7.46303573e-02 -2.97381550e-01 -5.16660929e-01 -4.52509671e-01
-1.32246479e-01 -1.63845867e-01 -6.59872293e-01 1.10612965e+00
1.63759395e-01 2.02741977e-02 1.01684439e+00 -1.72089517e+00
-7.11386800e-01 1.08511770e+00 -4.35222059e-01 -1.84904963e-01
2.60367423e-01 2.80089855e-01 9.16171193e-01 1.02841802e-01
6.68106437e-01 1.83100665e+00 3.95779312e-01 4.58579242e-01
-1.85018456e+00 -8.54759037e-01 4.53424543e-01 2.01812685e-02
-8.95855606e-01 -3.55080724e-01 6.35837734e-01 -8.42919528e-01
1.65715292e-01 1.36201769e-01 4.13745552e-01 1.84540522e+00
4.56891924e-01 5.78384578e-01 1.08062828e+00 -1.78427160e-01
5.52270532e-01 -3.13561231e-01 -3.24309394e-02 5.08025110e-01
4.35748875e-01 9.07677948e-01 -5.43648362e-01 -4.32608813e-01
6.88725650e-01 -7.44428039e-02 -1.43240720e-01 -5.30924439e-01
-1.36807716e+00 8.50062072e-01 3.14903975e-01 -4.50699739e-02
-6.18316650e-01 1.05450618e+00 -3.13571654e-02 1.09196790e-01
1.76281277e-02 8.16416144e-01 -6.68185115e-01 -1.12423360e-01
6.77791834e-02 6.39568508e-01 6.97515547e-01 8.65158439e-01
6.12480164e-01 -3.10302198e-01 -1.31471947e-01 4.36706126e-01
4.31460351e-01 9.18206275e-01 -7.82768354e-02 -9.19822872e-01
-1.42477140e-01 2.87880808e-01 6.78778827e-01 -1.06398702e+00
-7.29107380e-01 1.48581311e-01 -7.81707942e-01 2.41520047e-01
5.32751858e-01 -7.94792831e-01 -9.89638269e-01 2.42202663e+00
4.36824709e-01 5.71297526e-01 -9.76030529e-02 8.15909326e-01
5.43814957e-01 3.84670675e-01 4.65104759e-01 -9.30665061e-02
1.00728452e+00 -1.00248791e-01 -7.90020645e-01 -5.05814791e-01
2.85496831e-01 -2.44504929e-01 5.83614945e-01 2.11119637e-01
-6.54683411e-01 4.36055921e-02 -8.16572428e-01 3.58509153e-01
-1.90309912e-01 -2.74809450e-01 1.03507328e+00 7.15236440e-02
-6.71625495e-01 4.27115649e-01 -1.07126689e+00 -3.42089742e-01
2.70260364e-01 5.40670991e-01 -3.98268729e-01 7.15946108e-02
-1.41375732e+00 8.44712794e-01 1.17972553e-01 -1.60115793e-01
-1.86095631e+00 -8.60624254e-01 -5.88215232e-01 -5.88742010e-02
8.44634950e-01 -8.51349473e-01 1.34153652e+00 -6.80841208e-01
-1.30450189e+00 2.89288402e-01 -1.40662845e-02 -3.17227334e-01
4.40904796e-01 -1.62963957e-01 4.59211506e-02 -2.89115816e-01
2.29047865e-01 4.67367947e-01 8.23806405e-01 -1.42320919e+00
-5.42837560e-01 -4.86556262e-01 4.66483027e-01 1.20758630e-01
1.10884145e-01 -1.12877399e-01 5.07498942e-02 -4.71743882e-01
-3.74657921e-02 -1.30429256e+00 -6.02619648e-01 -4.32341546e-02
-7.63970196e-01 7.33119203e-03 6.55760229e-01 -2.36853227e-01
6.00389063e-01 -1.92458653e+00 1.04761541e+00 7.81357512e-02
5.10569990e-01 -4.36394304e-01 -1.91809565e-01 3.85694534e-01
-1.67786375e-01 4.32294279e-01 -1.58045992e-01 -5.79691976e-02
1.91586718e-01 4.31250602e-01 -1.05070718e-01 8.80857944e-01
2.15511635e-01 8.88653457e-01 -1.25981688e+00 -5.08311056e-02
1.91634327e-01 3.48504066e-01 -4.98725712e-01 2.82398880e-01
-5.31760156e-01 8.14179897e-01 -8.86852086e-01 -2.65578050e-02
1.15347818e-01 1.81634441e-01 4.80497926e-01 3.20323080e-01
2.18473859e-02 8.85656402e-02 -1.21138966e+00 1.13300049e+00
-3.61700803e-01 6.72245741e-01 2.51572043e-01 -1.02773297e+00
4.68673885e-01 4.55373734e-01 6.14217162e-01 -2.21264884e-01
4.25786555e-01 -2.47697458e-01 4.61170763e-01 -7.34214127e-01
2.37455778e-02 -2.87325352e-01 -3.61102879e-01 3.34285915e-01
2.26631314e-02 -1.98684894e-02 -2.16545358e-01 1.11379914e-01
1.73637354e+00 -1.78678840e-01 4.13984597e-01 -3.11936796e-01
-3.02693099e-01 -3.62999417e-04 7.00791419e-01 1.18029630e+00
-1.76751181e-01 -3.34241763e-02 1.04939139e+00 5.61978817e-02
-9.05587673e-01 -1.08506382e+00 -7.15257600e-02 8.60019207e-01
3.33028883e-01 1.29314527e-01 -4.49530780e-01 -3.06831568e-01
4.19229329e-01 8.95875871e-01 -1.19767892e+00 -6.88890934e-01
-1.67430669e-01 -7.76954591e-01 5.14058948e-01 2.65716761e-01
-5.26389927e-02 -8.11114907e-01 -5.69193780e-01 2.10985228e-01
6.02322333e-02 -6.65182471e-01 1.70174822e-01 6.76258564e-01
-2.45887429e-01 -1.35290956e+00 1.53382979e-02 -1.27686828e-01
4.83663440e-01 3.57520330e-04 8.75969529e-01 -2.70972729e-01
-2.67527819e-01 4.90007967e-01 -2.88679004e-01 -8.28996599e-01
-4.96693581e-01 -2.76551962e-01 4.86042202e-01 -1.16386041e-01
1.26808388e-02 -6.94772899e-01 -2.06674010e-01 3.20324361e-01
-7.91320145e-01 -7.56388623e-03 4.29239482e-01 8.16464484e-01
2.52235174e-01 3.53857160e-01 5.26262224e-01 -8.48533988e-01
4.91569698e-01 -1.13133466e+00 -6.83720112e-01 1.31902307e-01
-8.08671266e-02 3.16788971e-01 3.23307008e-01 -1.06363559e+00
-8.82997930e-01 2.38138363e-01 7.05702245e-01 -3.34419936e-01
-5.17887771e-01 7.72364080e-01 -6.71583116e-01 3.78124416e-01
7.24833369e-01 -4.64238703e-01 -1.15790024e-01 -2.12077096e-01
5.42733967e-01 3.02293152e-01 4.33312863e-01 -9.91980851e-01
8.94000173e-01 3.64116907e-01 4.10028368e-01 -6.65727794e-01
-5.10832727e-01 8.29185024e-02 -4.34872836e-01 -2.76816666e-01
9.91447449e-01 -7.50776708e-01 -1.24380970e+00 4.01772380e-01
-1.13392711e+00 -1.01377153e+00 -6.34912625e-02 5.75760901e-01
-7.09366620e-01 -6.40713751e-01 -1.54061690e-01 -1.07362330e+00
8.13809574e-01 -1.19444096e+00 1.01259363e+00 1.20663568e-01
-6.50727451e-01 -1.09758854e+00 2.22832322e-01 -2.25900650e-01
-4.89833578e-03 5.00560343e-01 1.17586815e+00 -4.42899704e-01
-5.79409301e-01 -4.88411561e-02 1.37684733e-01 -4.78160858e-01
3.20651770e-01 2.78317362e-01 -6.75410330e-01 7.90935978e-02
-3.07037562e-01 -8.35768580e-02 5.02962291e-01 8.42014909e-01
8.33289862e-01 -4.83097613e-01 -9.16790485e-01 1.82336226e-01
7.95059800e-01 6.26030922e-01 2.95199394e-01 1.41054019e-01
7.24283814e-01 8.35075736e-01 4.56291944e-01 3.47473383e-01
3.90657932e-01 7.34739721e-01 9.17283833e-01 2.31511239e-02
2.89697081e-01 -4.88363832e-01 2.47210592e-01 -4.20580685e-01
2.87525952e-01 -4.92925644e-01 -8.87268484e-01 5.98815382e-01
-2.25485921e+00 -1.09310400e+00 -4.76083249e-01 2.00527501e+00
8.54783177e-01 -1.86542124e-01 3.96668017e-02 -3.15907210e-01
8.69348168e-01 -2.07829684e-01 -1.09702134e+00 2.05874652e-01
-2.92944908e-01 -2.74824053e-01 8.93511057e-01 6.42635942e-01
-1.15471816e+00 7.30373621e-01 6.61735201e+00 1.72287971e-01
-7.52571464e-01 1.41254023e-01 5.41161716e-01 -2.86667585e-01
-3.73757154e-01 5.64394109e-02 -3.82116973e-01 4.80230778e-01
8.70113015e-01 -4.57566679e-01 8.09430599e-01 5.46578407e-01
7.40193427e-01 -2.91075617e-01 -1.53318882e+00 5.57249665e-01
-6.29604757e-01 -1.04310977e+00 -4.89055306e-01 3.62412006e-01
7.45892406e-01 -1.21035025e-01 1.30708531e-01 1.66472252e-02
1.51083446e+00 -1.35368562e+00 9.16976750e-01 7.01106012e-01
5.41055501e-01 -5.57171464e-01 3.80548388e-01 4.26254928e-01
-4.00710523e-01 -2.74196655e-01 -6.71853647e-02 -4.57607508e-01
-2.22222367e-03 5.27611911e-01 -6.95201039e-01 8.77061114e-03
6.05993867e-01 5.91856778e-01 -4.65249084e-02 7.80756235e-01
-7.57615149e-01 8.95550609e-01 -3.78878623e-01 -3.12178582e-01
-4.14269157e-02 -1.87050655e-01 9.26871479e-01 5.30187666e-01
1.22917227e-01 4.84412819e-01 2.08753366e-02 1.21389222e+00
4.80040796e-02 -5.43941855e-01 -1.01226139e+00 -3.09900314e-01
7.43070841e-01 8.29107344e-01 -3.35892826e-01 -6.07063882e-02
-3.50120105e-02 5.05561769e-01 2.80787975e-01 7.20937490e-01
-1.03398216e+00 1.58648849e-01 1.43843210e+00 -2.11429387e-01
1.88011434e-02 -1.37687519e-01 -3.20493668e-01 -8.90890956e-01
-3.91754717e-01 -8.14043820e-01 -4.53604758e-02 -8.73583198e-01
-1.20823634e+00 -2.83706576e-01 3.58048648e-01 -4.43415672e-01
-1.25146270e-01 -5.14265597e-01 -4.35448647e-01 8.11793923e-01
-7.54607320e-01 -1.00217605e+00 8.10792074e-02 4.74133939e-01
-2.11335756e-02 8.79785419e-02 7.58228898e-01 -2.04680070e-01
-8.90718222e-01 -6.20550066e-02 1.57103464e-01 -1.33165553e-01
6.23699546e-01 -1.42170787e+00 9.47201159e-03 7.78561771e-01
-1.36485636e-01 7.54015863e-01 1.42014623e+00 -1.02785385e+00
-1.84620500e+00 -1.22931027e+00 1.22383162e-01 -5.42682052e-01
1.29910433e+00 -5.95245838e-01 -5.42170823e-01 1.10481036e+00
-2.17297509e-01 -2.95329094e-01 3.55793983e-01 4.96298611e-01
-2.73917466e-01 2.87247628e-01 -9.87136006e-01 1.20088756e+00
1.35128319e+00 -2.37751544e-01 -4.49181587e-01 3.34343642e-01
1.10480869e+00 -2.05271930e-01 -5.03254235e-01 2.93309808e-01
5.19084334e-01 -4.41451460e-01 8.88722301e-01 -1.10635149e+00
7.30806410e-01 -3.79785538e-01 -2.33988404e-01 -1.99457717e+00
-4.41154718e-01 -6.69874370e-01 -6.47329316e-02 1.27745903e+00
3.36044282e-01 -6.84011817e-01 3.44583005e-01 8.90277445e-01
2.58402735e-01 -3.63950361e-03 -9.22182322e-01 -5.01884043e-01
1.96694538e-01 -6.33779585e-01 7.53101945e-01 9.69928980e-01
-6.19098730e-02 5.24506986e-01 -6.07611835e-01 7.14009106e-01
8.40492010e-01 -2.95026451e-01 8.71448696e-01 -1.43194878e+00
-4.64170992e-01 -3.66005123e-01 -4.95167702e-01 -5.49768209e-01
6.12686872e-01 -4.79074389e-01 6.13058329e-01 -1.15254819e+00
4.22349542e-01 -8.77742529e-01 1.70720704e-02 5.61512053e-01
-5.05121708e-01 -6.05264843e-01 -1.69860691e-01 -2.12763786e-01
-6.06144108e-02 6.82195902e-01 8.60370815e-01 -3.08914632e-01
-1.42046452e-01 -1.85398147e-01 -7.74308264e-01 8.36936951e-01
6.25908077e-01 -7.97762334e-01 -5.15724242e-01 -2.69191563e-01
4.06539381e-01 4.08125460e-01 9.00911927e-01 -5.10312378e-01
1.83629259e-01 -9.57501948e-01 -7.95249082e-03 1.29547685e-01
1.62717357e-01 -1.05188596e+00 7.83290923e-01 5.15491009e-01
-7.08850801e-01 -3.05509806e-01 2.33552888e-01 1.08737552e+00
3.51677507e-01 1.38245849e-03 4.25228775e-01 1.83767781e-01
-4.19307828e-01 1.53875574e-01 -6.99699819e-01 -1.99125886e-01
1.24229479e+00 5.33667505e-01 -5.83116472e-01 -5.76141536e-01
-6.35008216e-01 5.16007602e-01 4.23823774e-01 5.71465075e-01
4.13580894e-01 -1.15507174e+00 -6.12876952e-01 -3.78905833e-01
1.00011285e-02 -5.32529876e-02 1.61757663e-01 8.29465628e-01
2.80294091e-01 1.31622121e-01 4.03231531e-02 -5.34138918e-01
-1.15529871e+00 8.47874045e-01 4.16941106e-01 1.94800496e-01
-1.32886231e-01 7.69335389e-01 9.37940121e-01 -5.14639378e-01
9.64629054e-02 -1.14970520e-01 -1.42884865e-01 -3.81068057e-05
3.53215009e-01 3.31894249e-01 -7.40255117e-01 -4.34255779e-01
-3.16563278e-01 3.63630019e-02 4.61165994e-01 -3.34947050e-01
1.54942536e+00 -1.81650072e-01 -3.41262162e-01 9.25947547e-01
7.63503671e-01 -1.34816647e-01 -1.46695936e+00 -2.66244879e-05
3.19588743e-02 -2.97830969e-01 1.31858647e-01 -1.01256168e+00
-7.78460145e-01 4.25364077e-01 3.26637715e-01 3.53428870e-01
7.49873459e-01 3.59727859e-01 -1.88526243e-01 2.33827263e-01
7.11399555e-01 -4.35625583e-01 -2.24425197e-01 2.88535148e-01
1.00924397e+00 -1.17427492e+00 -3.22624087e-01 -3.31266016e-01
-2.34570339e-01 5.70278406e-01 3.37744027e-01 -2.23283753e-01
6.70496881e-01 6.58079922e-01 -3.84348363e-01 -3.07471663e-01
-1.14089787e+00 -2.61271864e-01 -1.18632786e-01 8.79166543e-01
1.68912187e-01 7.77057230e-01 -6.54683337e-02 5.19723833e-01
-1.85889810e-01 -2.41657794e-01 8.71432245e-01 3.97582382e-01
-5.78336082e-02 -8.63837183e-01 -7.13498831e-01 5.09795785e-01
-1.47810012e-01 8.36585686e-02 -6.92195952e-01 9.37385082e-01
2.28459522e-01 1.29234564e+00 1.04147516e-01 -4.37265158e-01
4.24744308e-01 -2.71951169e-01 3.74690682e-01 -6.44872069e-01
-4.61962186e-02 -1.55735821e-01 1.33580014e-01 -7.44241118e-01
-5.11350036e-01 -1.31305981e+00 -1.09324276e+00 -3.71561825e-01
-7.72602260e-02 -5.08396104e-02 6.24572158e-01 1.09455800e+00
1.31859660e-01 9.33262706e-01 3.12602192e-01 -7.81445801e-01
-2.36889631e-01 -1.00548434e+00 -5.42510152e-01 2.87036449e-01
6.70431197e-01 -1.41642094e+00 -5.25962234e-01 1.69985384e-01]
|
[7.872957706451416, 5.270256042480469]
|
5a2db3ee-bdc8-4baf-b695-67ee01e5cfbb
|
learning-discriminative-motion-features
|
1812.04172
| null |
http://arxiv.org/abs/1812.04172v1
|
http://arxiv.org/pdf/1812.04172v1.pdf
|
Learning Discriminative Motion Features Through Detection
|
Despite huge success in the image domain, modern detection models such as
Faster R-CNN have not been used nearly as much for video analysis. This is
arguably due to the fact that detection models are designed to operate on
single frames and as a result do not have a mechanism for learning motion
representations directly from video. We propose a learning procedure that
allows detection models such as Faster R-CNN to learn motion features directly
from the RGB video data while being optimized with respect to a pose estimation
task. Given a pair of video frames---Frame A and Frame B---we force our model
to predict human pose in Frame A using the features from Frame B. We do so by
leveraging deformable convolutions across space and time. Our network learns to
spatially sample features from Frame B in order to maximize pose detection
accuracy in Frame A. This naturally encourages our network to learn motion
offsets encoding the spatial correspondences between the two frames. We refer
to these motion offsets as DiMoFs (Discriminative Motion Features).
In our experiments we show that our training scheme helps learn effective
motion cues, which can be used to estimate and localize salient human motion.
Furthermore, we demonstrate that as a byproduct, our model also learns features
that lead to improved pose detection in still-images, and better keypoint
tracking. Finally, we show how to leverage our learned model for the tasks of
spatiotemporal action localization and fine-grained action recognition.
|
['Gedas Bertasius', 'Du Tran', 'Lorenzo Torresani', 'Jianbo Shi', 'Christoph Feichtenhofer']
|
2018-12-11
| null | null | null | null |
['fine-grained-action-recognition']
|
['computer-vision']
|
[ 1.37522161e-01 -1.94032356e-01 -5.64177632e-01 -1.51782051e-01
-6.03080273e-01 -4.39741075e-01 2.86264598e-01 -2.62012243e-01
-6.06058300e-01 2.62390494e-01 4.96155947e-01 2.88416684e-01
2.13876113e-01 -5.11670530e-01 -8.45757365e-01 -4.89799857e-01
-2.92471856e-01 -5.63339964e-02 4.03880596e-01 -1.62410825e-01
2.56623358e-01 6.40558183e-01 -1.41389143e+00 3.83021414e-01
-6.87041357e-02 9.44379032e-01 2.01983809e-01 9.37343061e-01
6.09031498e-01 1.03973055e+00 -3.72542143e-01 2.55560905e-01
5.10679841e-01 -4.31156069e-01 -8.15646112e-01 3.53195757e-01
5.98404229e-01 -1.00786662e+00 -9.26111042e-01 5.56910217e-01
2.40456551e-01 5.09229958e-01 3.00543338e-01 -1.10961318e+00
-2.62292206e-01 -2.83222124e-02 -7.80058205e-01 4.49858129e-01
6.97695196e-01 3.88867646e-01 9.70491827e-01 -8.02066088e-01
8.60598326e-01 1.25904906e+00 6.65598512e-01 8.31858695e-01
-1.00335908e+00 -3.09339106e-01 2.05571800e-01 3.83775175e-01
-1.15238237e+00 -5.53470433e-01 8.17736328e-01 -4.62727368e-01
1.09144831e+00 8.76016840e-02 1.11748385e+00 1.18924654e+00
1.95170119e-01 1.15685403e+00 3.59747916e-01 -2.78738797e-01
2.77422480e-02 -5.35795331e-01 -1.78816140e-01 9.51799452e-01
-7.71981478e-02 2.95109242e-01 -9.60430205e-01 2.58127540e-01
1.43796480e+00 3.65796387e-01 -5.40622056e-01 -6.58309340e-01
-1.41728687e+00 7.63256550e-01 6.09669030e-01 1.03629805e-01
-3.39822531e-01 8.02113652e-01 3.22001487e-01 7.68458529e-04
1.54646724e-01 2.45071203e-01 -5.24938464e-01 -3.73744369e-01
-7.36422241e-01 3.44609946e-01 2.34636173e-01 6.59108877e-01
7.35683739e-01 4.91175912e-02 -7.17819929e-02 2.84220874e-01
2.27125838e-01 4.16990399e-01 4.88286853e-01 -1.52334964e+00
3.18897009e-01 3.74873638e-01 3.03532809e-01 -1.22254729e+00
-5.89403629e-01 6.79716915e-02 -3.41414362e-01 3.05137247e-01
7.41916299e-01 -3.75312567e-02 -6.67614877e-01 1.81666934e+00
3.36136431e-01 3.60647708e-01 -1.82049617e-01 1.31183958e+00
1.67334482e-01 4.31643039e-01 -4.81502302e-02 8.31459239e-02
1.12461460e+00 -7.82149255e-01 -3.26058477e-01 -4.57184166e-01
8.80275965e-01 -4.53322887e-01 9.74071443e-01 3.78756076e-01
-1.11921048e+00 -6.51139915e-01 -8.96508455e-01 -3.65883708e-01
-1.88285634e-02 2.83751637e-01 6.31486893e-01 2.64511973e-01
-8.58437538e-01 8.09100032e-01 -1.33877516e+00 -5.43092668e-01
3.96459371e-01 3.91193599e-01 -6.10757351e-01 2.12245032e-01
-7.19523132e-01 8.58768046e-01 2.64353245e-01 1.37971878e-01
-1.08020067e+00 -4.25669819e-01 -1.14444041e+00 -2.49189124e-01
4.63463753e-01 -7.81505764e-01 1.27175939e+00 -1.19268692e+00
-1.33504331e+00 8.13915968e-01 -3.21028829e-01 -6.12387180e-01
4.19359148e-01 -8.13243628e-01 1.03232153e-01 6.20386064e-01
3.34360003e-02 1.05619395e+00 9.90130782e-01 -8.80348265e-01
-8.73710155e-01 -4.12762612e-01 4.26520169e-01 2.51111001e-01
-2.79742807e-01 6.16685227e-02 -5.88419437e-01 -7.73149788e-01
9.54103544e-02 -1.00025237e+00 -2.43379608e-01 5.28854728e-01
-2.19883427e-01 -3.12016271e-02 1.06679738e+00 -7.41378725e-01
9.22985256e-01 -2.05239677e+00 4.41823602e-01 -3.60718742e-03
1.59953877e-01 1.44117415e-01 -1.90767840e-01 -2.79590748e-02
-1.44826502e-01 -2.88940072e-01 1.93907127e-01 -2.01918647e-01
-3.70249867e-01 2.71227449e-01 -3.89219373e-01 8.05219054e-01
4.88718629e-01 1.12009001e+00 -1.09282959e+00 -2.60356635e-01
7.10028231e-01 6.29809856e-01 -8.89618158e-01 2.31767595e-01
-3.93062443e-01 7.24133492e-01 -5.42412758e-01 6.04007065e-01
-4.38048281e-02 -1.97940603e-01 -9.60830133e-03 -3.77572805e-01
3.14135514e-02 1.66849978e-02 -1.02340984e+00 2.13530636e+00
-2.00706556e-01 8.87971997e-01 -1.99529633e-01 -1.01143682e+00
3.76806974e-01 2.51871496e-01 9.54864621e-01 -5.45454621e-01
2.45659903e-01 -1.41152903e-01 -1.31936818e-01 -6.24894798e-01
5.51745236e-01 1.94526181e-01 9.43285897e-02 2.76784360e-01
8.22187867e-03 2.59157360e-01 3.77543941e-02 4.01582643e-02
1.22402799e+00 7.48536527e-01 2.29574978e-01 2.28594899e-01
2.45041952e-01 2.21477188e-02 5.62541723e-01 5.86181879e-01
-5.00004590e-01 7.40431786e-01 2.49562860e-01 -6.62316859e-01
-1.14013898e+00 -8.76630127e-01 4.10872042e-01 1.19078755e+00
1.92876115e-01 -4.98464346e-01 -6.68213964e-01 -6.81377053e-01
-8.77706148e-03 5.36596105e-02 -7.09192634e-01 -2.41434112e-01
-1.14231515e+00 -1.05187997e-01 3.30364197e-01 1.08337462e+00
3.35678935e-01 -9.95520055e-01 -1.39504302e+00 1.15446597e-01
-1.45471275e-01 -1.21829808e+00 -8.53294969e-01 1.27911359e-01
-8.33377182e-01 -1.38741505e+00 -7.92285323e-01 -6.58564925e-01
4.65666145e-01 5.42787731e-01 8.09197664e-01 3.20128873e-02
-6.04162276e-01 8.86669636e-01 -3.55964184e-01 1.46121770e-01
3.30819078e-02 -3.59871149e-01 1.58501700e-01 8.58037546e-03
3.75287563e-01 -4.04726714e-01 -1.05218804e+00 3.04267019e-01
-7.86831200e-01 6.42972663e-02 4.68040705e-01 6.73901320e-01
5.15458167e-01 -4.82375085e-01 2.51358673e-02 -1.37508646e-01
-1.49720594e-01 -1.36666089e-01 -4.31843281e-01 8.92145112e-02
1.09227046e-01 8.19082409e-02 2.44147643e-01 -4.99996841e-01
-5.05015671e-01 5.76458335e-01 1.03095189e-01 -9.09732461e-01
-1.30025312e-01 9.94405076e-02 1.69271603e-01 -1.89521685e-01
6.29762769e-01 8.84033442e-02 5.27640283e-02 -1.56424239e-01
4.86321360e-01 2.09777340e-01 7.38052249e-01 -4.65410709e-01
6.17063344e-01 9.40038025e-01 1.01879135e-01 -7.87876129e-01
-9.28917468e-01 -7.25741506e-01 -9.63832319e-01 -4.53894764e-01
1.35404325e+00 -1.09361410e+00 -1.01384878e+00 3.49672198e-01
-1.18647122e+00 -6.41911566e-01 -3.09420586e-01 6.79078579e-01
-1.06892216e+00 4.67675984e-01 -6.45181239e-01 -7.27550268e-01
6.01749346e-02 -1.05815303e+00 1.48156548e+00 8.60442072e-02
-4.10625458e-01 -9.85035181e-01 -1.85364913e-02 1.62288994e-01
-6.18564785e-02 4.57765251e-01 2.59634048e-01 -1.67275465e-03
-8.14415216e-01 -1.58143833e-01 -2.43209153e-02 3.19480419e-01
6.81631640e-02 -5.16255759e-02 -8.28831553e-01 -3.71113658e-01
-1.67098284e-01 -4.88453954e-01 9.16541278e-01 7.26388276e-01
1.25893188e+00 -1.66215062e-01 -4.17725801e-01 7.56968975e-01
1.03925896e+00 -6.68451414e-02 5.30190945e-01 4.67215449e-01
9.62389171e-01 4.90016311e-01 7.67509818e-01 5.11574745e-01
1.42311037e-01 1.01745856e+00 4.06612992e-01 -3.42861633e-03
-2.42488980e-01 -3.63273233e-01 6.78826511e-01 1.98763818e-01
-3.14347655e-01 2.54692316e-01 -7.71762013e-01 4.49620903e-01
-2.16853356e+00 -1.13755989e+00 2.66698539e-01 2.22568297e+00
7.41512954e-01 7.15568382e-03 5.07968307e-01 -2.41855178e-02
3.97256553e-01 1.95642382e-01 -5.95792115e-01 2.15644255e-01
4.71588932e-02 1.12661220e-01 6.12524152e-01 5.15499115e-01
-1.30798042e+00 9.54182863e-01 6.26573038e+00 3.08314830e-01
-1.21819246e+00 -3.93256284e-02 4.81095582e-01 -5.33380091e-01
1.47853583e-01 2.69007962e-02 -6.04743958e-01 1.68658808e-01
6.03609741e-01 2.87859410e-01 3.50135446e-01 8.95511389e-01
5.52314222e-01 -1.70070782e-01 -1.56710005e+00 1.15777862e+00
1.00498490e-01 -1.37998235e+00 -3.52248810e-02 6.09513111e-02
5.20852506e-01 -7.15441853e-02 1.53805986e-02 3.87455113e-02
9.30121168e-02 -1.04220796e+00 1.01856351e+00 4.79092926e-01
4.82794225e-01 -6.42316878e-01 9.46220607e-02 2.72192031e-01
-1.34563625e+00 -3.06051672e-01 -3.41624349e-01 -2.76244998e-01
2.06750512e-01 5.68334721e-02 -5.11841238e-01 9.18408930e-02
7.45311141e-01 1.18635154e+00 -3.80196393e-01 8.89588058e-01
-2.08742470e-01 1.42957136e-01 -1.99940562e-01 2.67923951e-01
2.52811044e-01 2.06943471e-02 4.69544619e-01 1.03622258e+00
1.52052253e-01 8.30339715e-02 5.13056457e-01 5.36758959e-01
1.34511769e-01 -3.24731171e-01 -6.79139316e-01 1.11762986e-01
1.39733255e-01 1.03421617e+00 -6.54972136e-01 -1.51220545e-01
-5.53674161e-01 1.25834382e+00 3.62612396e-01 4.48427349e-01
-8.76523554e-01 6.43725172e-02 9.59669054e-01 3.58114511e-01
4.75906640e-01 -6.98682129e-01 1.77750438e-01 -1.31415427e+00
4.37114760e-02 -7.83936918e-01 2.64909327e-01 -9.22281563e-01
-7.93363392e-01 -1.98017759e-03 6.47641066e-03 -1.32545173e+00
-6.57162905e-01 -1.00091147e+00 -3.55308831e-01 3.31133872e-01
-1.28440547e+00 -1.10592628e+00 -2.37689793e-01 8.99760425e-01
7.02585042e-01 1.07706778e-01 6.24296010e-01 8.66653547e-02
-3.64361823e-01 3.49183083e-01 -2.80735850e-01 6.70294940e-01
6.28904164e-01 -1.18997812e+00 3.54263872e-01 9.39191997e-01
4.85574901e-01 6.54206872e-01 5.70958138e-01 -5.03258944e-01
-1.63714659e+00 -9.58112538e-01 2.46502563e-01 -8.65250349e-01
6.94340050e-01 -2.22021580e-01 -3.84821236e-01 9.71984208e-01
-2.70828962e-01 5.07253170e-01 3.78547460e-01 -3.08429562e-02
-4.60145742e-01 4.18854207e-02 -7.16552973e-01 6.10350370e-01
1.20607543e+00 -6.99159741e-01 -4.18762535e-01 3.99045199e-01
5.07205009e-01 -5.34983039e-01 -6.79666400e-01 2.93642521e-01
8.75212312e-01 -9.06869233e-01 1.32718241e+00 -8.00653279e-01
4.56412464e-01 -3.74726713e-01 -3.09242308e-01 -8.22838724e-01
-2.76443481e-01 -6.21405244e-01 -5.15078485e-01 4.45245862e-01
-2.61084557e-01 -6.40695617e-02 1.06317103e+00 4.52518493e-01
7.76461810e-02 -8.44659328e-01 -9.77929533e-01 -7.16482520e-01
-2.10850164e-01 -5.37162781e-01 -4.32520621e-02 5.98642528e-01
1.50226504e-02 9.61910188e-03 -6.14692986e-01 9.56663340e-02
5.27719676e-01 5.60934469e-02 9.64642882e-01 -6.97016120e-01
-6.26500845e-01 -3.86886954e-01 -8.27881277e-01 -1.60416067e+00
7.39158839e-02 -3.20097029e-01 2.80661762e-01 -1.03922462e+00
3.22464883e-01 -9.73091647e-03 -1.71034545e-01 5.99100530e-01
-1.95398599e-01 5.97360849e-01 3.30919713e-01 4.36684102e-01
-6.95966065e-01 3.07695419e-01 1.21504092e+00 1.65143847e-01
-3.94666716e-02 -2.46649221e-01 -2.56956369e-01 9.41188931e-01
6.14830017e-01 -1.87749788e-01 -3.66384476e-01 -5.29066861e-01
1.73161086e-02 2.01933697e-01 1.02873230e+00 -1.15753472e+00
6.77090734e-02 -3.13753217e-01 8.49627495e-01 -4.29931998e-01
6.04318678e-01 -6.54925704e-01 -1.86955824e-01 5.25596440e-01
-4.22586232e-01 5.17781489e-02 -4.16992866e-02 7.11081088e-01
1.00988917e-01 1.15242086e-01 6.01520061e-01 -2.26599544e-01
-1.18666637e+00 5.10344684e-01 -1.32329792e-01 3.19768302e-02
1.16960609e+00 -4.83821034e-01 -8.94901436e-03 -4.70846146e-01
-8.56910646e-01 5.02922293e-03 7.83702254e-01 6.24235690e-01
7.61731267e-01 -1.34200323e+00 -3.46147329e-01 2.21896484e-01
8.22823346e-02 5.05027696e-02 7.16862381e-02 1.02948308e+00
-5.69252312e-01 4.23236638e-01 -3.14229727e-01 -9.01339412e-01
-1.21438479e+00 5.33790112e-01 4.97322172e-01 4.31926036e-03
-9.40563142e-01 8.99366319e-01 3.88047844e-01 1.44240275e-01
4.39372510e-01 -5.84409416e-01 9.22636539e-02 -3.04449439e-01
6.62970543e-01 2.07317412e-01 -2.21842095e-01 -8.84911954e-01
-3.97662431e-01 7.02025950e-01 6.33412004e-02 -2.65094727e-01
1.22796142e+00 -1.43650204e-01 3.23869795e-01 3.14834386e-01
1.44650090e+00 -1.69790760e-01 -2.06767678e+00 -1.09390114e-02
-9.27596539e-02 -8.65712285e-01 -3.48081738e-02 -2.51888424e-01
-1.13350499e+00 8.57451856e-01 6.61060929e-01 -2.60974646e-01
9.45785165e-01 1.61111295e-01 8.93274426e-01 5.39203465e-01
4.35077757e-01 -1.24466968e+00 8.59936237e-01 4.10385251e-01
6.34328246e-01 -1.21356606e+00 -1.32210433e-01 -5.15973307e-02
-4.61889565e-01 1.32285881e+00 5.53791106e-01 -4.70834285e-01
2.77788758e-01 1.86455876e-01 -7.32404813e-02 -1.72810808e-01
-4.75695074e-01 -4.12557840e-01 3.46566975e-01 8.68953168e-01
4.28386748e-01 -2.52863258e-01 3.01774234e-01 6.62698671e-02
8.99042785e-02 1.84968814e-01 3.42980564e-01 1.16317558e+00
-7.06737816e-01 -9.50651705e-01 -3.66650194e-01 7.22710714e-02
-3.12498838e-01 3.33009601e-01 -2.29411855e-01 7.56069541e-01
1.37118652e-01 7.76617587e-01 1.29905835e-01 -4.79900002e-01
1.96506545e-01 -1.85049191e-01 9.69157934e-01 -7.09895670e-01
-1.68846682e-01 8.20041299e-02 -2.33769402e-01 -1.37421167e+00
-6.69114470e-01 -8.10004890e-01 -1.41067767e+00 -1.76350117e-01
-1.62427455e-01 -3.79773140e-01 3.18532020e-01 9.80156600e-01
9.29526612e-02 4.11655039e-01 3.90399545e-01 -1.40023506e+00
-5.02127409e-01 -5.17035306e-01 -4.76246655e-01 6.01312459e-01
7.18866467e-01 -7.94405162e-01 4.13268656e-02 3.65304768e-01]
|
[8.099140167236328, 0.24702244997024536]
|
99e5e5ba-9bf6-4adf-8880-5a8ac49acfb7
|
do-not-sleep-on-linear-models-simple-and
|
2207.07753
| null |
https://arxiv.org/abs/2207.07753v3
|
https://arxiv.org/pdf/2207.07753v3.pdf
|
Do Not Sleep on Traditional Machine Learning: Simple and Interpretable Techniques Are Competitive to Deep Learning for Sleep Scoring
|
Over the last few years, research in automatic sleep scoring has mainly focused on developing increasingly complex deep learning architectures. However, recently these approaches achieved only marginal improvements, often at the expense of requiring more data and more expensive training procedures. Despite all these efforts and their satisfactory performance, automatic sleep staging solutions are not widely adopted in a clinical context yet. We argue that most deep learning solutions for sleep scoring are limited in their real-world applicability as they are hard to train, deploy, and reproduce. Moreover, these solutions lack interpretability and transparency, which are often key to increase adoption rates. In this work, we revisit the problem of sleep stage classification using classical machine learning. Results show that competitive performance can be achieved with a conventional machine learning pipeline consisting of preprocessing, feature extraction, and a simple machine learning model. In particular, we analyze the performance of a linear model and a non-linear (gradient boosting) model. Our approach surpasses state-of-the-art (that uses the same data) on two public datasets: Sleep-EDF SC-20 (MF1 0.810) and Sleep-EDF ST (MF1 0.795), while achieving competitive results on Sleep-EDF SC-78 (MF1 0.775) and MASS SS3 (MF1 0.817). We show that, for the sleep stage scoring task, the expressiveness of an engineered feature vector is on par with the internally learned representations of deep learning models. This observation opens the door to clinical adoption, as a representative feature vector allows to leverage both the interpretability and successful track record of traditional machine learning models.
|
['Nicolas Vandenbussche', 'Sofie Van Hoecke', 'Gilles Vandewiele', 'Michael Rademaker', 'Emiel Deprost', 'Jonas Van Der Donckt', 'Jeroen Van Der Donckt']
|
2022-07-15
| null | null | null | null |
['sleep-stage-detection', 'multimodal-sleep-stage-detection', 'sleep-staging', 'automatic-sleep-stage-classification']
|
['medical', 'medical', 'medical', 'medical']
|
[ 2.16970250e-01 2.47550830e-01 -4.12229806e-01 -5.32002270e-01
-7.81855643e-01 -3.57442290e-01 3.92728627e-01 5.12942791e-01
-6.72079563e-01 7.40948856e-01 2.66441941e-01 -5.51398695e-01
-2.46794954e-01 -3.34158808e-01 -2.14219466e-01 -5.83806872e-01
2.12358451e-03 4.68391478e-01 7.01752082e-02 -1.19653039e-01
-1.52704874e-02 2.01232001e-01 -1.45525956e+00 2.96073496e-01
7.66252518e-01 1.15367961e+00 4.27958965e-02 6.95544541e-01
1.05071709e-01 6.90363944e-01 -5.61483800e-01 -4.24669117e-01
-9.77936685e-02 -3.19430828e-01 -9.09307718e-01 -3.34745318e-01
2.52210438e-01 -2.32789919e-01 3.04895919e-02 5.61878264e-01
6.15597785e-01 -2.47210532e-01 4.07194376e-01 -1.01611650e+00
-2.71451294e-01 2.42023081e-01 -1.93020999e-01 3.85735273e-01
2.06996739e-01 1.55949771e-01 1.18563175e+00 -3.20674390e-01
3.69226217e-01 4.58442420e-01 1.04203725e+00 8.41543138e-01
-1.28486025e+00 -4.59054679e-01 -1.96000025e-01 2.29541987e-01
-1.06946301e+00 -6.13496423e-01 2.62388915e-01 -3.27592283e-01
1.15442801e+00 4.27903473e-01 9.87992525e-01 1.14623225e+00
4.95373607e-01 7.21466541e-01 1.21721578e+00 -3.76939625e-01
3.56054038e-01 3.11474353e-01 3.27438116e-01 8.99472237e-01
3.03470314e-01 -3.68882529e-02 -6.00389361e-01 -9.21359658e-02
1.19775645e-01 3.49641323e-01 -9.34724733e-02 -1.50307879e-01
-9.54267502e-01 8.83697152e-01 4.67102051e-01 5.62972069e-01
-2.41256714e-01 6.50915205e-02 4.42075461e-01 2.86900669e-01
4.88325179e-01 5.51157236e-01 -5.95294297e-01 -5.12409270e-01
-1.57720923e+00 -7.12129893e-03 6.33283436e-01 2.84273565e-01
4.99329060e-01 -2.63211936e-01 -1.81726485e-01 6.19133294e-01
2.58079290e-01 3.44940186e-01 8.84087026e-01 -7.78780043e-01
5.56512699e-02 8.69190991e-01 -1.11760311e-01 -5.21706343e-01
-1.16700757e+00 -8.35666001e-01 -8.55157316e-01 1.47790015e-01
2.80995250e-01 1.40883356e-01 -6.94974303e-01 1.66249657e+00
-4.56137210e-02 -1.21344276e-01 3.89402509e-02 4.52286094e-01
9.24327433e-01 7.95822442e-02 5.15740253e-02 -1.07357554e-01
1.44677675e+00 -9.51226592e-01 -5.64212084e-01 -5.88740945e-01
9.23598588e-01 -5.49177706e-01 1.24348211e+00 7.00045347e-01
-1.20808578e+00 -2.78509051e-01 -1.16298449e+00 -3.61582279e-01
-3.63378316e-01 3.35613817e-01 8.42881322e-01 8.27192664e-01
-1.36128998e+00 7.59838223e-01 -1.14205015e+00 -5.29955864e-01
7.75274873e-01 8.56904685e-01 -4.83347863e-01 1.13497876e-01
-7.83368289e-01 1.17516482e+00 -2.96227261e-02 -4.92873341e-02
-7.14600682e-01 -8.28904808e-01 -5.43590128e-01 1.68819830e-01
-7.71524990e-03 -1.08086956e+00 1.43939865e+00 -8.53480577e-01
-1.30736554e+00 1.18311012e+00 -2.56740272e-01 -7.28568733e-01
4.50726926e-01 -3.81632596e-01 -3.95013332e-01 1.19389910e-02
7.36515149e-02 4.16334331e-01 6.97821498e-01 -6.57770991e-01
-7.04896986e-01 -3.89135182e-01 1.77489132e-01 -1.44852877e-01
-5.19167125e-01 -1.90309975e-02 -1.23956636e-01 -1.69315621e-01
-3.77964646e-01 -1.02458036e+00 -3.05792928e-01 3.18659283e-02
-2.74890840e-01 -1.34696454e-01 2.04548448e-01 -3.92095923e-01
1.60764289e+00 -2.24075985e+00 -6.73330948e-02 -8.89204443e-02
7.01572418e-01 4.92292494e-01 2.07611248e-01 2.68474817e-01
-9.89583507e-02 7.93806463e-02 -1.79660633e-01 -8.07245672e-01
-7.40636187e-03 2.50989377e-01 1.63849756e-01 5.80139339e-01
6.92838430e-02 1.01882064e+00 -8.70920062e-01 -4.42082882e-01
2.01400980e-01 3.44764620e-01 -8.31249714e-01 4.63090576e-02
2.67544359e-01 1.43859506e-01 -2.62439489e-01 4.26703960e-01
2.01907322e-01 -6.39473259e-01 2.18120784e-01 -1.16562545e-01
-1.52236640e-01 6.03747010e-01 -5.71920455e-01 1.86803734e+00
-7.28082895e-01 5.59190571e-01 -2.08991557e-01 -9.44225073e-01
5.88589907e-01 1.01461448e-01 6.17345572e-01 -6.71969235e-01
2.29508355e-01 3.92176419e-01 2.61111259e-01 -4.78013635e-01
2.93430567e-01 -4.79239970e-01 -1.50560319e-01 5.21152675e-01
2.79405445e-01 -3.38879274e-03 1.54483706e-01 1.08942360e-01
1.47580612e+00 -3.03446651e-01 6.52460515e-01 -3.25818121e-01
3.80754739e-01 -1.07000284e-01 4.15555865e-01 6.12907648e-01
-4.15253639e-01 6.19829357e-01 6.06211245e-01 -6.06619954e-01
-6.88294351e-01 -9.74206328e-01 -2.49312922e-01 1.05849195e+00
-3.45706642e-01 -9.60221529e-01 -7.54837036e-01 -8.41493130e-01
-1.93203136e-01 6.16914451e-01 -1.03686130e+00 -5.51744640e-01
-3.50174606e-01 -9.76208270e-01 6.63900614e-01 6.62843108e-01
5.87166548e-02 -9.67612982e-01 -9.88240957e-01 1.53875694e-01
-2.29251813e-02 -8.88253868e-01 -7.60137439e-02 5.32139599e-01
-1.01656044e+00 -1.11944520e+00 -4.97961044e-01 -2.96752393e-01
3.57168019e-01 3.34300026e-02 1.36259687e+00 4.75315779e-01
-3.73152882e-01 1.17922202e-01 -2.18061775e-01 -5.92664719e-01
-4.47681129e-01 5.88598788e-01 1.43836230e-01 -1.39501661e-01
5.11772990e-01 -6.53700054e-01 -1.04961383e+00 1.16388639e-02
-7.29774237e-01 1.15654193e-01 9.23952103e-01 9.48185027e-01
3.20845127e-01 -4.82051432e-01 5.28678417e-01 -9.13733661e-01
4.81922954e-01 -4.52934891e-01 -6.31009787e-02 8.37775543e-02
-1.16357517e+00 1.52382655e-02 6.43071532e-01 -1.91626325e-02
-5.41536093e-01 -8.36610198e-02 -5.77573717e-01 1.94916744e-02
-1.59541324e-01 4.45202082e-01 3.93704712e-01 1.12507902e-01
7.57654130e-01 2.33457640e-01 1.57016411e-01 -5.72319627e-01
1.31808579e-01 7.31735766e-01 3.94072354e-01 9.34948027e-03
5.07339358e-01 5.06904721e-01 7.73321018e-02 -6.43524289e-01
-1.25208473e+00 -5.84438145e-01 -5.59897423e-01 1.66833952e-01
9.56486285e-01 -8.37951481e-01 -6.52378261e-01 1.70566797e-01
-5.38826227e-01 -4.41263944e-01 -4.32880908e-01 3.52530926e-01
-5.49154818e-01 1.79773510e-01 -4.17367518e-01 -4.82770026e-01
-7.56069005e-01 -1.09189153e+00 1.32093120e+00 1.43711910e-01
-7.83931494e-01 -1.11935949e+00 3.74571532e-01 6.32107437e-01
7.06075430e-01 1.03744857e-01 1.08491039e+00 -8.87681484e-01
-1.80564374e-02 -3.02384913e-01 -7.24410862e-02 3.03042620e-01
2.12698936e-01 -9.10525769e-02 -1.26705313e+00 -2.51540452e-01
-4.34592664e-02 -4.18464512e-01 8.07362556e-01 4.09695357e-01
1.02652717e+00 -6.59358352e-02 -2.38946810e-01 7.11714089e-01
1.16359496e+00 -1.05892383e-01 4.44962949e-01 5.79531193e-01
2.63256222e-01 1.94854200e-01 2.06452832e-01 3.46179187e-01
4.85839128e-01 6.79637611e-01 4.17261958e-01 -3.75085682e-01
-1.25322580e-01 1.24743387e-01 3.20787460e-01 6.50030255e-01
-2.62134790e-01 1.48457110e-01 -9.52258587e-01 3.77104670e-01
-1.65845156e+00 -7.49467373e-01 -8.60549435e-02 2.10435891e+00
7.26412416e-01 4.72395778e-01 3.69093835e-01 4.65129733e-01
-1.49705797e-01 4.02675644e-02 -4.83145416e-01 -7.40314603e-01
1.28455237e-01 7.25873113e-01 6.96712211e-02 9.99726877e-02
-9.19436276e-01 5.29200554e-01 6.73723412e+00 5.50335109e-01
-1.30286002e+00 5.74274123e-01 5.49394667e-01 -6.28341317e-01
1.05565963e-02 -3.90598446e-01 -6.59159958e-01 6.52049184e-01
1.55227065e+00 4.27281521e-02 3.21279526e-01 8.44760358e-01
3.21540564e-01 -8.80476832e-02 -1.22587740e+00 9.52926755e-01
2.60579623e-02 -1.43688643e+00 -4.43849832e-01 2.21743837e-01
2.67773867e-01 3.64036620e-01 1.52199790e-01 6.09620214e-01
6.17237156e-03 -1.24083722e+00 6.07795417e-01 3.74110878e-01
8.64715040e-01 -4.28610504e-01 1.04331064e+00 1.86409786e-01
-7.39205778e-01 -3.57739508e-01 -1.28394678e-01 -2.15671360e-01
-1.34037063e-01 6.66457891e-01 -9.01963830e-01 5.02408743e-01
9.60067630e-01 8.37143302e-01 -1.02607882e+00 9.76043046e-01
-1.55986115e-01 7.52995789e-01 -3.45906585e-01 -5.21723963e-02
2.94869959e-01 2.10016966e-01 -9.66396928e-02 1.23100269e+00
2.15279862e-01 -3.51970702e-01 -2.18629673e-01 5.43848157e-01
-8.79187733e-02 8.12040269e-02 -3.37732911e-01 -6.10885918e-02
-1.69890389e-01 1.54493487e+00 -6.03527606e-01 -2.20848873e-01
-5.75747669e-01 8.20707798e-01 4.62052017e-01 -2.23131210e-01
-8.88367236e-01 9.14735943e-02 7.38073766e-01 2.81065136e-01
1.57512844e-01 1.57043427e-01 -4.67061609e-01 -1.19891429e+00
-9.87474918e-02 -8.77930224e-01 5.01759708e-01 -4.59429443e-01
-1.14553523e+00 7.67768979e-01 -2.75509119e-01 -1.07905006e+00
-2.56452024e-01 -6.06678009e-01 -6.92514002e-01 4.92647976e-01
-1.48180723e+00 -1.22081125e+00 -3.18174958e-01 4.78981346e-01
3.66098106e-01 -7.91731626e-02 1.09889317e+00 4.82623577e-01
-7.54832923e-01 7.51872063e-01 1.97930381e-01 -2.18927398e-01
7.27252722e-01 -1.46128750e+00 1.94499716e-01 4.21325237e-01
1.49201795e-01 8.07195783e-01 4.26234782e-01 -6.11418253e-03
-1.15972960e+00 -9.19206262e-01 1.22256291e+00 -7.18243182e-01
7.19626606e-01 -3.98933947e-01 -5.85570037e-01 5.79983234e-01
1.43077716e-01 -2.37484090e-02 1.32034469e+00 6.00539923e-01
-1.87780894e-02 -3.95167410e-01 -1.00382650e+00 3.58289778e-01
8.54935050e-01 -4.78384167e-01 -6.58772111e-01 2.64266819e-01
1.36447623e-01 -1.88209444e-01 -8.61278534e-01 3.58634233e-01
7.69014001e-01 -1.40916491e+00 8.56116593e-01 -7.57850409e-01
5.10997057e-01 3.30095142e-02 5.72293922e-02 -1.03404117e+00
-3.59051347e-01 -6.21226311e-01 -2.61587262e-01 9.39703941e-01
5.22763610e-01 -5.50881386e-01 9.06537592e-01 6.32324100e-01
-3.72829854e-01 -1.37900269e+00 -1.03033948e+00 -4.99667019e-01
1.18657775e-01 -3.55655491e-01 4.40326035e-01 6.13639176e-01
1.08434655e-01 6.21374309e-01 -2.30759338e-01 -3.50033790e-01
5.96826077e-02 2.88271517e-01 6.65524244e-01 -1.37820292e+00
-5.01978517e-01 -6.42083466e-01 -5.57217598e-01 -4.31662887e-01
-7.24661425e-02 -1.00132179e+00 -2.29155138e-01 -1.68046939e+00
5.65890074e-01 -4.71060842e-01 -7.87198961e-01 8.31579447e-01
-2.68320769e-01 5.79038978e-01 8.57441723e-02 1.97083935e-01
-7.69561768e-01 2.46543974e-01 6.83803201e-01 2.49035340e-02
-1.64193630e-01 3.21660399e-01 -1.05242622e+00 9.56501544e-01
1.00084054e+00 -5.87382078e-01 -4.19434398e-01 -2.08888665e-01
4.12620962e-01 -4.18051541e-01 2.58613795e-01 -1.18197727e+00
5.82159963e-03 3.33075792e-01 3.41278285e-01 -2.42784902e-01
4.89440918e-01 -5.99423826e-01 1.20968543e-01 8.62919748e-01
-1.73095644e-01 6.34149089e-02 2.04076424e-01 2.96158999e-01
2.78391130e-02 -2.76711375e-01 7.55289674e-01 -5.77754974e-02
-3.83452833e-01 1.68041766e-01 -5.03617406e-01 5.80850765e-02
7.05327868e-01 -2.06987396e-01 -3.55289370e-01 -1.09465763e-01
-8.35596204e-01 -1.35178775e-01 4.82667178e-01 2.47388676e-01
2.29847118e-01 -7.78017223e-01 -4.22380030e-01 3.43007773e-01
1.86174765e-01 -2.40751714e-01 1.89900234e-01 1.41822731e+00
-4.72511441e-01 4.28086877e-01 -2.19941586e-01 -7.09228814e-01
-1.34169126e+00 4.11630839e-01 4.18826580e-01 -8.56391847e-01
-5.72206557e-01 6.71997249e-01 1.52832597e-01 -5.92610501e-02
1.52596667e-01 -6.73838615e-01 -5.95120303e-02 1.35356843e-01
5.69213212e-01 2.93092996e-01 6.38590634e-01 -2.32264206e-01
-7.10692644e-01 3.75581771e-01 -1.71110451e-01 3.80995989e-01
1.65254653e+00 6.48002252e-02 9.08646807e-02 4.16237831e-01
1.07309616e+00 1.95294525e-02 -7.20512986e-01 1.90482438e-01
9.68073756e-02 2.15711398e-03 2.38507330e-01 -1.06775498e+00
-9.58740711e-01 1.07234251e+00 8.95637929e-01 1.70911580e-01
1.31387711e+00 1.08637802e-01 9.13210511e-01 4.07471925e-01
2.03817159e-01 -7.71589220e-01 -6.74705207e-02 1.68978930e-01
3.74188811e-01 -1.24067342e+00 2.09759101e-01 6.70719519e-02
-4.59027082e-01 1.07665622e+00 1.12466261e-01 1.11888155e-01
5.53581059e-01 2.23969549e-01 1.65571183e-01 -3.49449843e-01
-8.28533292e-01 -2.00655878e-01 3.24887723e-01 3.71339053e-01
5.80957592e-01 3.92729193e-02 -2.62498200e-01 1.08516586e+00
-5.68741024e-01 4.27498490e-01 4.07757014e-01 8.14824760e-01
-3.45898241e-01 -1.26243889e+00 2.51896262e-01 7.95967162e-01
-9.75626051e-01 -3.60775679e-01 -3.16930473e-01 7.76575923e-01
3.53343666e-01 1.03364730e+00 -1.91511884e-01 -4.51176167e-01
2.86032528e-01 1.67521015e-01 3.81370455e-01 -7.45522320e-01
-9.96126235e-01 -1.94271430e-01 1.98243633e-01 -8.14909279e-01
-5.06616473e-01 -5.85182786e-01 -9.88690972e-01 -2.90916651e-01
-2.29945481e-01 1.28841534e-01 5.40325105e-01 1.06432569e+00
5.02804935e-01 6.21385872e-01 3.17071527e-01 -6.46065772e-01
-5.10125518e-01 -8.26679289e-01 -5.22410452e-01 3.52178454e-01
5.37673235e-01 -6.99467480e-01 -3.03134590e-01 -3.63811441e-02]
|
[13.53534984588623, 3.5360617637634277]
|
d3aecd41-23a8-44a1-a8bc-335c77200cd6
|
an-empirical-study-of-multitask-learning-to
|
2304.08115
| null |
https://arxiv.org/abs/2304.08115v1
|
https://arxiv.org/pdf/2304.08115v1.pdf
|
An Empirical Study of Multitask Learning to Improve Open Domain Dialogue Systems
|
Autoregressive models used to generate responses in open-domain dialogue systems often struggle to take long-term context into account and to maintain consistency over a dialogue. Previous research in open-domain dialogue generation has shown that the use of \emph{auxiliary tasks} can introduce inductive biases that encourage the model to improve these qualities. However, most previous research has focused on encoder-only or encoder/decoder models, while the use of auxiliary tasks in \emph{decoder-only} autoregressive models is under-explored. This paper describes an investigation where four different auxiliary tasks are added to small and medium-sized GPT-2 models fine-tuned on the PersonaChat and DailyDialog datasets. The results show that the introduction of the new auxiliary tasks leads to small but consistent improvement in evaluations of the investigated models.
|
['Richard Johansson', 'Mehrdad Farahani']
|
2023-04-17
| null | null | null | null |
['dialogue-generation', 'dialogue-generation']
|
['natural-language-processing', 'speech']
|
[ 5.27093112e-02 9.23033357e-01 1.98617682e-01 -6.41995192e-01
-8.35238338e-01 -5.68911016e-01 1.25250828e+00 -9.47581083e-02
-4.78514612e-01 1.28130758e+00 8.50665808e-01 -1.84969127e-01
5.98809905e-02 -4.82421517e-01 -2.11524710e-01 -3.27747703e-01
1.29909292e-01 1.15624011e+00 -1.04247265e-01 -8.98174047e-01
1.48568094e-01 -2.14468852e-01 -9.74795043e-01 5.06837189e-01
9.07597363e-01 3.26337874e-01 2.67467767e-01 1.02462256e+00
-1.84437856e-01 1.16701674e+00 -1.05233872e+00 -6.36387467e-01
-6.43863007e-02 -7.99373925e-01 -1.38660824e+00 -8.07740614e-02
8.48360639e-03 -4.71452087e-01 -1.09160341e-01 4.49778080e-01
9.78974700e-01 4.87134576e-01 6.80938840e-01 -8.24162304e-01
-1.01912701e+00 1.07053876e+00 2.13317141e-01 9.76023823e-02
9.24807549e-01 3.54899764e-01 8.73141527e-01 -4.84379172e-01
7.88011849e-01 1.52210450e+00 7.09408939e-01 9.98313606e-01
-1.46453965e+00 -2.73788184e-01 4.79302034e-02 -2.26657361e-01
-7.27228940e-01 -6.01182520e-01 6.95015073e-01 -2.40181565e-01
1.42783880e+00 2.19977438e-01 2.90515959e-01 1.79957199e+00
2.40371414e-02 4.44856673e-01 1.56814682e+00 -5.16213000e-01
-6.33102134e-02 8.10134530e-01 4.43298280e-01 1.38316125e-01
-1.68779805e-01 2.19057515e-01 -4.38334823e-01 -2.34177575e-01
6.20937824e-01 -8.89894128e-01 -7.57740662e-02 3.54793996e-01
-1.01029348e+00 1.25675619e+00 9.06858444e-02 5.36339641e-01
-6.36389732e-01 -2.86556304e-01 6.44176185e-01 6.72256649e-01
8.78989995e-01 9.60948706e-01 -4.91043836e-01 -8.46098363e-01
-5.89821458e-01 6.77798986e-01 1.44246078e+00 8.93312454e-01
4.49896157e-01 2.96892107e-01 -7.92392850e-01 1.41665483e+00
2.69036114e-01 2.96300530e-01 8.20482433e-01 -9.17589486e-01
6.13664448e-01 3.27580243e-01 2.36261576e-01 -7.67013371e-01
-5.12497127e-01 -9.56265181e-02 -5.59780180e-01 -2.99874395e-01
5.73146701e-01 -9.66844857e-01 -5.48901081e-01 1.88859451e+00
1.25821590e-01 -5.46288550e-01 4.93779927e-01 7.56648600e-01
1.29329860e+00 8.27835083e-01 2.95909137e-01 -2.07658932e-01
1.26258910e+00 -9.32586551e-01 -1.05569756e+00 -4.92987126e-01
8.07548344e-01 -6.72662437e-01 1.22983897e+00 2.13672638e-01
-1.42777514e+00 -7.02591777e-01 -6.44776702e-01 -3.74868244e-01
-3.36800337e-01 8.31050798e-03 5.37869930e-01 7.66243577e-01
-1.15345490e+00 4.73441988e-01 -1.51231915e-01 -4.19051677e-01
-3.62245768e-01 2.06740156e-01 -2.71243691e-01 4.25770849e-01
-1.94183421e+00 1.53791881e+00 3.41065735e-01 2.72763148e-02
-6.51670039e-01 -3.41509610e-01 -9.99977827e-01 -1.64025888e-01
-2.59460136e-02 -7.45607436e-01 1.80576265e+00 -1.02866423e+00
-2.13950157e+00 8.48917603e-01 -1.56659335e-02 -7.99754143e-01
5.32483160e-01 -3.68581474e-01 -1.47181138e-01 -1.65776059e-01
-1.06790654e-01 8.05292726e-01 4.93069112e-01 -9.79805470e-01
-2.65974611e-01 -1.38526142e-01 4.41079021e-01 5.69894016e-01
-2.54165322e-01 1.93663672e-01 1.69672206e-01 -6.13400817e-01
-5.99627376e-01 -1.00614870e+00 -2.69838572e-01 -1.07740724e+00
-2.33612537e-01 -5.85185111e-01 3.58477145e-01 -8.99437606e-01
1.38804042e+00 -1.50212646e+00 2.47246981e-01 -3.10804486e-01
-2.40046531e-01 3.28679681e-01 -6.12592921e-02 7.86640644e-01
-4.19287346e-02 7.07713217e-02 -1.08386087e-03 -4.07034546e-01
1.76907137e-01 4.54822123e-01 -1.51424453e-01 -1.55386567e-01
2.07295984e-01 8.84409249e-01 -8.16449642e-01 -3.56953710e-01
1.79608166e-01 4.01152551e-01 -6.79895401e-01 5.32754779e-01
-4.54903096e-01 5.88456511e-01 -3.91486675e-01 -3.16421151e-01
2.20722482e-01 -1.12782968e-02 1.16688497e-01 4.90110248e-01
-9.35587883e-02 1.11615837e+00 -7.88060486e-01 1.60471690e+00
-7.69846737e-01 6.49083436e-01 -6.52630329e-02 -5.18535733e-01
1.15985155e+00 8.59082699e-01 -5.51493987e-02 -7.48380363e-01
2.35298499e-01 1.10776052e-01 3.62472713e-01 -6.49052024e-01
1.06324732e+00 -4.21528280e-01 -2.88996279e-01 7.20386028e-01
3.24577957e-01 -4.19783592e-01 2.62645304e-01 2.41363660e-01
6.93366289e-01 2.49031499e-01 3.10895890e-01 -2.30354950e-01
7.54343271e-01 6.41829595e-02 1.94717690e-01 9.00350094e-01
-1.73049327e-02 5.36123455e-01 5.00790715e-01 -7.72681832e-02
-1.08315980e+00 -4.14484560e-01 -5.12352623e-02 1.54302979e+00
-5.91561079e-01 -3.94729435e-01 -1.07636321e+00 -6.11823916e-01
-3.86593521e-01 1.38140082e+00 -6.25485778e-01 -9.76405516e-02
-5.60389400e-01 -7.98453987e-01 8.79209042e-01 2.54636735e-01
4.01832998e-01 -1.44491589e+00 -3.88689429e-01 5.76769650e-01
-6.39696062e-01 -9.21489656e-01 -1.26982257e-01 1.41959846e-01
-8.57424796e-01 -4.38400179e-01 -8.61944020e-01 -6.01978421e-01
1.92406431e-01 -3.53685677e-01 1.31010664e+00 -2.47114435e-01
4.85410750e-01 3.31425458e-01 -5.81836641e-01 -6.26259685e-01
-1.13126791e+00 5.53791344e-01 -2.92400777e-01 -3.61845732e-01
4.95444506e-01 -3.33273411e-01 4.99399900e-02 1.28550336e-01
-5.53937137e-01 1.38150305e-01 3.02463859e-01 1.26650965e+00
-4.20245022e-01 -7.66990483e-01 9.74949658e-01 -1.41466558e+00
1.76625884e+00 -5.38995385e-01 3.32863778e-02 -6.05017431e-02
-6.93263292e-01 3.17928970e-01 5.23657501e-01 -4.77034599e-01
-1.84143543e+00 -5.13662040e-01 -6.53156161e-01 3.18250448e-01
-4.72269624e-01 5.58827400e-01 1.47471517e-01 3.56505007e-01
1.06816566e+00 5.09190001e-02 2.67584413e-01 -2.90353060e-01
4.75012034e-01 8.79411280e-01 1.64523855e-01 -6.65184438e-01
3.03143591e-01 -3.30974013e-01 -6.37976706e-01 -9.50528324e-01
-5.80673397e-01 -1.45077676e-01 -4.77625608e-01 -2.15617463e-01
7.58989930e-01 -8.51588905e-01 -3.13468933e-01 3.07758033e-01
-1.38146591e+00 -7.72409618e-01 -3.72195631e-01 3.30943525e-01
-8.48089933e-01 2.81950146e-01 -8.72495532e-01 -1.05767274e+00
-5.10191143e-01 -9.95489120e-01 6.98211372e-01 2.29185224e-01
-1.13093257e+00 -1.50304151e+00 5.61663628e-01 5.67976892e-01
6.75205410e-01 -1.70516342e-01 7.24154472e-01 -1.35443151e+00
1.70327932e-01 -1.06698200e-01 3.60265166e-01 4.76974219e-01
-3.58450273e-03 -2.92521924e-01 -1.28182673e+00 1.43955007e-01
2.64858991e-01 -7.52741337e-01 3.29986244e-01 1.86873838e-01
2.00992405e-01 -6.57679796e-01 9.41999182e-02 -6.84376433e-02
5.91293812e-01 1.65409565e-01 8.47800553e-01 2.83725619e-01
1.56015530e-01 1.06022060e+00 7.51280069e-01 5.62019467e-01
6.06388569e-01 6.88501477e-01 -1.67349711e-01 1.78516190e-02
1.12935491e-01 -3.58025551e-01 6.10028684e-01 7.38647163e-01
-2.31354058e-01 -3.37738812e-01 -6.89784944e-01 5.54034710e-01
-1.70245385e+00 -1.01479924e+00 -4.83411729e-01 1.95160532e+00
1.34655464e+00 1.12944521e-01 1.81747824e-01 -1.56633794e-01
5.53226531e-01 3.87117863e-01 8.03289786e-02 -1.22509599e+00
-1.40474483e-01 2.19235152e-01 -7.42902607e-02 9.41839039e-01
-7.17945635e-01 1.05016983e+00 6.95151138e+00 4.25748378e-01
-8.10107112e-01 2.68624634e-01 6.19373977e-01 9.85387340e-02
-3.70008498e-01 4.41282205e-02 -9.29672599e-01 4.16953802e-01
1.64611030e+00 -2.81963527e-01 1.37967438e-01 8.05750012e-01
5.51069081e-01 -1.54742733e-01 -1.23369646e+00 3.90482575e-01
1.23426035e-01 -8.79977226e-01 5.92275374e-02 7.32106790e-02
7.29851842e-01 -3.31905037e-01 -9.53760222e-02 9.33897793e-01
8.24723959e-01 -1.09581542e+00 3.61825317e-01 5.64070344e-01
2.64162511e-01 -4.97977465e-01 9.21379805e-01 5.85872769e-01
-1.76805899e-01 4.77316938e-02 -2.76505768e-01 -5.59700191e-01
5.48690140e-01 -2.98189539e-02 -1.57615364e+00 3.77173811e-01
1.34735838e-01 1.89349115e-01 -3.83778572e-01 3.41529071e-01
-4.06287372e-01 8.04051757e-01 1.41974930e-02 -4.93137687e-01
5.22052526e-01 -3.00620079e-01 6.99005425e-01 1.49152946e+00
-1.05539255e-01 1.27883554e-01 -1.16326377e-01 7.67932951e-01
2.41812438e-01 2.36354098e-01 -8.18610489e-01 -8.91861320e-02
1.98224172e-01 1.05793059e+00 5.10097183e-02 -5.07354975e-01
-2.45239392e-01 1.01297843e+00 3.60116303e-01 1.54547155e-01
-7.46410608e-01 -7.01669380e-02 4.58826631e-01 1.62628487e-01
-4.34745029e-02 -7.39828050e-02 -3.19082797e-01 -6.86648548e-01
-2.40634590e-01 -1.29678082e+00 4.24610883e-01 -8.55714202e-01
-1.27868271e+00 9.21680152e-01 1.84971899e-01 -5.51828504e-01
-1.34367633e+00 -3.21822673e-01 -6.06627285e-01 1.32898211e+00
-1.06908596e+00 -1.01923525e+00 -5.28658442e-02 4.73145455e-01
1.02117860e+00 -2.26794422e-01 1.34445953e+00 -1.15856272e-03
-3.00608128e-01 5.52350342e-01 -1.00073248e-01 1.84443966e-01
1.05630267e+00 -1.42857611e+00 4.69230562e-01 3.61388892e-01
-2.39736900e-01 8.63052666e-01 1.17997074e+00 -6.87499464e-01
-7.55201340e-01 -6.51817560e-01 1.49881721e+00 -7.48766780e-01
4.84984696e-01 -2.92227447e-01 -1.07091916e+00 8.45156193e-01
1.09158909e+00 -8.74628484e-01 6.82441235e-01 4.79410380e-01
1.04019880e-01 3.74707699e-01 -1.05173981e+00 5.10411799e-01
6.16967082e-01 -5.77236354e-01 -1.24024284e+00 3.88398230e-01
7.20059514e-01 -5.99319518e-01 -1.02747321e+00 4.71237637e-02
3.22641939e-01 -8.68716896e-01 7.53055513e-01 -9.68224764e-01
6.90664828e-01 5.37882686e-01 3.41351241e-01 -1.88587546e+00
-3.20903718e-01 -9.75170910e-01 -6.14195550e-03 1.57205343e+00
8.12162578e-01 -7.44866729e-01 4.31787431e-01 1.01166475e+00
-2.61004061e-01 -2.08469138e-01 -6.86390817e-01 -3.02608073e-01
5.00292897e-01 -4.92621288e-02 2.79146850e-01 8.71091723e-01
4.63496983e-01 1.12649071e+00 -6.18314266e-01 -4.99124557e-01
-1.14490025e-01 -5.31267703e-01 1.18715489e+00 -1.21797276e+00
-3.19925785e-01 -2.60977924e-01 1.92889661e-01 -1.17058480e+00
2.99378872e-01 -5.35945117e-01 2.79393256e-01 -1.61453283e+00
-2.37692505e-01 -4.47355717e-01 4.39512312e-01 1.35031030e-01
-5.04025459e-01 -2.10618451e-01 7.39123300e-02 -2.37742692e-01
-2.17972770e-01 8.02666187e-01 1.02316010e+00 2.41286263e-01
-4.14214939e-01 2.11137846e-01 -9.31783080e-01 4.62825090e-01
8.17575097e-01 -3.21250826e-01 -6.10479891e-01 -3.42525333e-01
1.54766425e-01 3.71135533e-01 3.53953019e-02 -6.89939380e-01
5.52947819e-02 9.91786569e-02 1.41108319e-01 -1.36995643e-01
6.89936161e-01 -2.01882139e-01 7.64254853e-02 1.64336190e-01
-8.21953297e-01 3.48841012e-01 3.56363088e-01 2.47768179e-01
-2.95843869e-01 -6.34724557e-01 5.30964375e-01 -5.56990743e-01
-4.55418229e-01 -3.91545594e-01 -9.28459048e-01 4.29123282e-01
6.67090714e-01 -2.21057385e-01 -2.37495080e-01 -1.04086494e+00
-7.13139832e-01 2.97704339e-01 2.15398404e-03 6.75898612e-01
2.31483325e-01 -9.78900492e-01 -1.18151891e+00 -1.08627722e-01
-2.72838529e-02 -2.15832442e-01 3.02239448e-01 7.25022256e-01
-3.06108624e-01 8.91493082e-01 -2.16625243e-01 -3.69296610e-01
-1.34885359e+00 1.96423039e-01 4.11893189e-01 -8.29618812e-01
-3.86308044e-01 9.08613801e-01 -9.98245627e-02 -8.79868269e-01
1.50796503e-01 2.19220333e-02 -7.54560709e-01 3.52360278e-01
2.67314672e-01 3.14069778e-01 -4.06620204e-02 -7.07576156e-01
3.28786492e-01 -3.54703516e-01 -3.62030923e-01 -7.86609232e-01
1.16725302e+00 -3.82603437e-01 1.48143291e-01 7.17861414e-01
6.39876723e-01 -1.95819631e-01 -9.65814412e-01 -1.59475267e-01
3.72573882e-02 -1.23555653e-01 -2.24224135e-01 -1.03422320e+00
-2.08094358e-01 8.81582260e-01 3.94074304e-04 4.46564585e-01
4.93858486e-01 -3.41262698e-01 3.74257147e-01 4.89200175e-01
1.78109601e-01 -1.41328466e+00 3.04769576e-02 1.21817291e+00
1.30320954e+00 -1.07616770e+00 -3.42418820e-01 -1.16728865e-01
-1.46531928e+00 8.78418267e-01 8.60168993e-01 1.46671562e-02
1.47904977e-01 1.76284544e-03 2.56570190e-01 -1.44496903e-01
-1.32214141e+00 2.16091778e-02 1.06978659e-02 8.08144271e-01
1.16989183e+00 6.65950179e-02 -7.54017591e-01 7.83888042e-01
-7.91207552e-01 -1.25931054e-01 8.27890396e-01 6.72004282e-01
-1.39097840e-01 -1.24270570e+00 -3.44905615e-01 2.94453084e-01
-5.07001221e-01 -2.89873779e-01 -1.04048657e+00 9.46662903e-01
-3.23111624e-01 1.30162668e+00 6.72654286e-02 -9.96990874e-02
3.87218744e-01 7.93402970e-01 2.95801193e-01 -8.44892561e-01
-1.46608436e+00 -2.07564473e-01 1.12354898e+00 -8.32414627e-02
-3.78839016e-01 -7.59995461e-01 -8.96406591e-01 -2.93398798e-01
-4.49698895e-01 6.56474173e-01 4.50322360e-01 8.83895159e-01
2.09468916e-01 6.17614269e-01 5.75828552e-01 -5.44870257e-01
-9.90525603e-01 -2.07558250e+00 -2.06867054e-01 4.55042779e-01
5.23997657e-02 -4.64413553e-01 -1.61978632e-01 9.91740264e-03]
|
[12.711736679077148, 8.131341934204102]
|
dc738d86-f11b-4e14-8068-7f7ac87053e8
|
icdar-2019-robust-reading-challenge-on
|
1912.09641
| null |
https://arxiv.org/abs/1912.09641v1
|
https://arxiv.org/pdf/1912.09641v1.pdf
|
ICDAR 2019 Robust Reading Challenge on Reading Chinese Text on Signboard
|
Chinese scene text reading is one of the most challenging problems in computer vision and has attracted great interest. Different from English text, Chinese has more than 6000 commonly used characters and Chinesecharacters can be arranged in various layouts with numerous fonts. The Chinese signboards in street view are a good choice for Chinese scene text images since they have different backgrounds, fonts and layouts. We organized a competition called ICDAR2019-ReCTS, which mainly focuses on reading Chinese text on signboard. This report presents the final results of the competition. A large-scale dataset of 25,000 annotated signboard images, in which all the text lines and characters are annotated with locations and transcriptions, were released. Four tasks, namely character recognition, text line recognition, text line detection and end-to-end recognition were set up. Besides, considering the Chinese text ambiguity issue, we proposed a multi ground truth (multi-GT) evaluation method to make evaluation fairer. The competition started on March 1, 2019 and ended on April 30, 2019. 262 submissions from 46 teams are received. Most of the participants come from universities, research institutes, and tech companies in China. There are also some participants from the United States, Australia, Singapore, and Korea. 21 teams submit results for Task 1, 23 teams submit results for Task 2, 24 teams submit results for Task 3, and 13 teams submit results for Task 4. The official website for the competition is http://rrc.cvc.uab.es/?ch=12.
|
['Baoguang Shi', 'Qi Song', 'Mingkun Yang', 'Kai Zhou', 'Yongsheng Zhou', 'C. V. Jawahar', 'Shijian Lu', 'Rui Zhang', 'Nan Li', 'Minghui Liao', 'Dong Wang', 'Xi Liu', 'Xiang Bai', 'Qianyi Jiang', 'Lei Wang', 'Dimosthenis Karatzas']
|
2019-12-20
| null | null | null | null |
['line-detection']
|
['computer-vision']
|
[ 2.33974665e-01 -6.58736467e-01 -4.27378193e-02 -2.84075439e-01
-7.24975407e-01 -6.92558527e-01 4.87329036e-01 -3.02640051e-01
-5.68022370e-01 5.49499393e-01 2.85938978e-01 -2.46071860e-01
3.31977129e-01 -2.37498686e-01 -5.31651199e-01 -6.25029683e-01
4.83215898e-01 2.67877489e-01 3.83944154e-01 3.44699733e-02
6.54561341e-01 1.96672156e-01 -1.16238010e+00 7.21093357e-01
1.00509107e+00 8.63412678e-01 4.90461558e-01 8.72976542e-01
-2.13789091e-01 5.98844349e-01 -8.79021287e-01 -6.01071656e-01
1.43371895e-01 -7.00842619e-01 -4.97863770e-01 4.18908536e-01
6.64489388e-01 -1.41605377e-01 -2.00482130e-01 1.14487910e+00
7.12932110e-01 1.73904784e-02 6.04640782e-01 -1.07992041e+00
-6.74722016e-01 5.71718514e-01 -8.75699222e-01 -1.78260624e-01
4.06067550e-01 -1.55582055e-01 9.06159401e-01 -1.18088734e+00
5.81141472e-01 8.89810860e-01 4.86671776e-01 5.34520805e-01
-2.33183518e-01 -5.15664816e-01 3.30415666e-01 2.60086864e-01
-1.12495899e+00 -2.99940437e-01 3.73991251e-01 -4.06072438e-01
5.12417078e-01 5.37700474e-01 4.09924984e-01 1.12254953e+00
-9.38891992e-02 1.46231413e+00 1.22380424e+00 -8.76272440e-01
-3.35679166e-02 -2.48481799e-02 3.04557860e-01 5.72889745e-01
2.38055944e-01 -4.64667082e-01 -4.66249645e-01 3.33739907e-01
6.59320354e-01 -2.37953037e-01 -5.10586381e-01 4.61034663e-02
-1.60665309e+00 6.01411879e-01 -3.27759497e-02 4.21423674e-01
2.07025796e-01 -2.91319519e-01 5.96127808e-01 1.81326821e-01
-3.44022363e-02 1.87657312e-01 -4.47164893e-01 -3.51933122e-01
-7.33883142e-01 2.28127092e-01 6.48265481e-01 1.31851172e+00
1.67630725e-02 -6.39021471e-02 -3.69648755e-01 1.25102127e+00
1.96853742e-01 9.38554585e-01 7.26196170e-01 -3.94439697e-01
1.14399052e+00 3.25361550e-01 2.24324912e-01 -9.03152823e-01
-4.49877262e-01 -9.67190322e-03 -1.05138052e+00 -7.74128735e-02
5.57611585e-01 -5.29684722e-01 -9.83939767e-01 8.25219214e-01
-1.53391898e-01 -1.92422405e-01 -5.48006557e-02 1.13295007e+00
9.23439920e-01 7.19383955e-01 -5.83499849e-01 5.58922738e-02
1.33990347e+00 -1.59945130e+00 -8.32532227e-01 -9.83306319e-02
5.03526032e-01 -1.48133123e+00 1.18948734e+00 7.55053401e-01
-9.50343609e-01 -5.52033126e-01 -1.11250925e+00 -4.25831266e-02
-3.25797409e-01 1.08490407e+00 2.21779153e-01 7.50347316e-01
-5.59803009e-01 -5.36887124e-02 -5.35447001e-01 -4.65524524e-01
2.00848073e-01 -4.69533429e-02 -9.03934017e-02 -2.84678757e-01
-8.41484785e-01 7.05523729e-01 2.03348339e-01 5.34796417e-01
-4.57772613e-02 8.64247233e-02 -4.88968760e-01 -4.06781793e-01
4.24386978e-01 -1.73906803e-01 1.33574867e+00 -1.07066357e+00
-1.64782643e+00 8.70253026e-01 -7.38011152e-02 1.11774486e-02
1.12776303e+00 -3.86621982e-01 -9.45800006e-01 2.65840739e-02
1.50012732e-01 4.45013732e-01 8.07341516e-01 -8.28492820e-01
-9.56428885e-01 -1.72587186e-01 -4.88020629e-01 3.67106140e-01
-1.05359130e-01 4.96138334e-01 -1.19773602e+00 -9.79443491e-01
1.51977688e-01 -9.89439905e-01 1.45676196e-01 -1.95895433e-01
-8.48890424e-01 -3.81341763e-02 9.26332474e-01 -9.50780571e-01
1.09786153e+00 -1.99392092e+00 -2.16988042e-01 8.24648142e-02
-2.06300363e-01 4.97072399e-01 -1.16428718e-01 3.39193344e-01
2.01944724e-01 1.10848054e-01 -1.32192239e-01 -3.22619319e-01
1.80707961e-01 -2.79397517e-01 -3.44415724e-01 3.67970437e-01
-1.74877882e-01 8.95889103e-01 -4.87145811e-01 -4.72170681e-01
2.78626114e-01 1.22375991e-02 1.36378720e-01 -1.55020833e-01
-1.32073998e-01 2.15513095e-01 -5.76053381e-01 8.71001720e-01
8.99670899e-01 -2.55134434e-01 3.87823507e-02 -7.81488419e-02
-4.15458858e-01 -1.80495322e-01 -1.44369519e+00 1.29876602e+00
-4.66815196e-02 1.22666442e+00 -1.56041160e-01 -5.39578736e-01
1.00195634e+00 1.14485212e-01 1.74554452e-01 -9.07579303e-01
1.32177696e-01 3.43220174e-01 -8.75393003e-02 -4.21631157e-01
9.54498649e-01 5.48503876e-01 -2.05217510e-01 2.21346885e-01
-6.56746805e-01 -2.23188788e-01 7.25647092e-01 7.91560039e-02
5.26785553e-01 3.26236427e-01 4.35515220e-04 5.41125871e-02
6.83539093e-01 1.10395350e-01 4.79089767e-01 7.68563569e-01
-2.10096970e-01 1.23885417e+00 4.27652687e-01 -3.97213608e-01
-1.15625596e+00 -8.09432209e-01 -1.79199815e-01 8.68349910e-01
3.10715437e-01 -4.71438199e-01 -7.40849376e-01 -5.08717299e-01
-4.46222544e-01 2.48682439e-01 -3.39663982e-01 4.37766880e-01
-8.25842798e-01 -5.93404472e-01 8.42603505e-01 6.61937833e-01
1.27638960e+00 -1.19269919e+00 -4.48498845e-01 -1.66819245e-01
-6.44281209e-01 -1.45960104e+00 -1.12792897e+00 -1.80308580e-01
-5.66629827e-01 -1.18866289e+00 -1.53753686e+00 -1.23081076e+00
8.06927025e-01 3.85173082e-01 5.55443048e-01 -4.86320369e-02
-4.10911798e-01 1.59817755e-01 -8.03016245e-01 -7.33912230e-01
-1.83777377e-01 1.14424594e-01 -1.93497702e-01 4.49350886e-02
3.13160092e-01 8.08342516e-01 -1.97980553e-01 7.75250614e-01
-6.48816288e-01 3.51093411e-01 6.29196823e-01 8.66471589e-01
4.60958362e-01 -1.61851063e-01 1.13800958e-01 -8.21379602e-01
5.39881229e-01 4.57508147e-01 -8.72550130e-01 6.48208261e-01
-1.21993944e-01 -4.33998048e-01 6.51251853e-01 -2.28004292e-01
-1.12671244e+00 4.77510737e-03 7.03069940e-02 1.73377678e-01
-2.84811139e-01 3.17808956e-01 -4.19170499e-01 1.82801485e-01
5.34675956e-01 3.84204745e-01 -3.93820167e-01 -4.52229619e-01
2.12828740e-02 1.16712284e+00 4.69045609e-01 -3.49512070e-01
6.08157814e-01 1.86689883e-01 -3.87905627e-01 -1.31614482e+00
-6.92507803e-01 -4.90011752e-01 -6.95252657e-01 -3.46331686e-01
8.81124437e-01 -8.31195116e-01 -5.16412079e-01 1.60866725e+00
-1.07162595e+00 -4.21937704e-01 3.22164863e-01 7.14035571e-01
-2.25530967e-01 8.21945488e-01 -6.61384821e-01 -5.16055107e-01
-1.75716639e-01 -1.01219940e+00 1.07267177e+00 6.26368642e-01
1.51761174e-01 -7.77444422e-01 -3.08596492e-01 6.12370670e-01
2.13185713e-01 2.08186987e-03 4.20810103e-01 -3.77412498e-01
-5.88077247e-01 -4.93631154e-01 -5.17247498e-01 4.72843528e-01
8.01499039e-02 2.92978615e-01 -7.77739823e-01 -2.13540822e-01
-5.94221652e-01 -3.34721595e-01 9.80528653e-01 5.13925374e-01
1.22525156e+00 6.46953508e-02 -2.08376393e-01 2.96291858e-01
1.17539918e+00 3.66859108e-01 8.06753218e-01 5.02800703e-01
9.11783755e-01 4.60665107e-01 8.39309216e-01 5.54384053e-01
3.15671921e-01 8.44481170e-01 -1.61818132e-01 -1.22966364e-01
-2.07832307e-01 -1.76839530e-01 5.73963463e-01 9.00531292e-01
-2.22504631e-01 -6.08381331e-01 -9.82006848e-01 1.35198385e-01
-1.75532639e+00 -8.51372182e-01 -9.29800093e-01 2.33954310e+00
5.56220651e-01 8.89723897e-02 2.03724876e-01 1.52609497e-01
1.03098202e+00 7.65335485e-02 -4.04439121e-01 -1.86597243e-01
-8.40743244e-01 -3.03819865e-01 7.38643765e-01 2.65065879e-01
-1.41456544e+00 1.08171940e+00 5.30113983e+00 8.99876952e-01
-1.22325981e+00 -4.73727584e-01 7.24590719e-01 2.00068235e-01
1.69641003e-01 -1.14539862e-01 -1.17057955e+00 6.21940732e-01
3.25211436e-01 9.32620466e-03 1.63004383e-01 4.46003705e-01
1.90441966e-01 -3.09194148e-01 -5.33568025e-01 1.26205420e+00
5.32524705e-01 -1.34698677e+00 -1.34527341e-01 -4.91060577e-02
1.20123255e+00 1.96099356e-01 3.98447439e-02 1.13020614e-02
9.28521082e-02 -9.75399435e-01 9.67537463e-01 6.28652334e-01
8.03809524e-01 -4.52788353e-01 8.55303824e-01 2.51480162e-01
-1.14685035e+00 9.41416398e-02 -4.18610334e-01 1.35219023e-01
2.32500449e-01 3.79058272e-01 -3.42949569e-01 6.82396233e-01
6.63666010e-01 8.90084505e-01 -8.02484572e-01 1.47958708e+00
-5.81642210e-01 6.15500033e-01 -1.10331684e-01 -6.80550337e-01
2.24358127e-01 -4.65692133e-01 1.91897705e-01 1.43308568e+00
2.46177748e-01 -2.83276349e-01 3.04674655e-01 2.36119002e-01
2.66757868e-02 3.92971694e-01 -1.84125632e-01 -7.58600384e-02
1.29906997e-01 1.08386803e+00 -1.10562706e+00 -3.29270542e-01
-5.35042465e-01 1.36721599e+00 -4.32239324e-01 6.91713035e-01
-8.12424362e-01 -1.03407490e+00 -8.45993683e-02 -5.14496207e-01
4.82972562e-01 -3.21480751e-01 -5.13126612e-01 -1.37256682e+00
4.67027754e-01 -1.10441279e+00 3.31975013e-01 -9.27031696e-01
-1.18920863e+00 5.99684000e-01 -4.02185082e-01 -1.52353048e+00
9.97785255e-02 -1.04508507e+00 -5.31558275e-01 9.52470303e-01
-1.17178130e+00 -1.21139336e+00 -5.48513949e-01 4.30207342e-01
9.33472991e-01 -4.89318132e-01 5.32943368e-01 2.27371663e-01
-8.24509323e-01 8.64581943e-01 7.06105709e-01 7.36387432e-01
9.90723968e-01 -1.09677982e+00 5.28848708e-01 9.36661065e-01
3.31701100e-01 2.28397578e-01 3.74927014e-01 -6.00765347e-01
-1.11187029e+00 -9.61045921e-01 1.19611883e+00 -3.36344123e-01
4.33249623e-01 -5.10428667e-01 -3.79075795e-01 4.67325568e-01
2.55891353e-01 -3.64560753e-01 4.30411011e-01 -3.18985969e-01
-1.18613914e-01 9.84652713e-02 -5.46798825e-01 7.98282146e-01
6.12402380e-01 -1.72370479e-01 -3.64151001e-01 5.32410145e-01
1.13707431e-01 -9.49395239e-01 -2.54523635e-01 -6.19486570e-02
6.45015538e-01 -7.41770625e-01 3.76784295e-01 -1.85424522e-01
4.55891818e-01 -4.89789963e-01 -2.02372134e-01 -8.78987670e-01
7.69519359e-02 -4.50626165e-01 5.19074440e-01 1.09571278e+00
6.47179484e-01 -5.46695888e-01 8.46550107e-01 3.11808556e-01
-4.80993718e-01 -3.17835599e-01 -5.36809623e-01 -8.79239261e-01
1.37086660e-01 -3.78041476e-01 2.03033566e-01 6.94170177e-01
-1.14854150e-01 3.33038419e-01 -6.16246402e-01 -1.85357347e-01
4.29466814e-01 3.60981584e-01 9.25005913e-01 -8.56827557e-01
-1.59689829e-01 -7.82051206e-01 -1.85087651e-01 -1.49602091e+00
-1.62209347e-01 -5.79818368e-01 1.36623144e-01 -1.82832289e+00
-3.55206728e-02 -5.18637523e-02 3.31314802e-01 4.13196623e-01
-1.28975645e-01 2.69972771e-01 6.34225905e-01 2.10441500e-01
-7.42116392e-01 3.22441995e-01 1.56170774e+00 -3.36735725e-01
-1.38316778e-02 3.84553909e-01 -3.11492294e-01 6.96825504e-01
1.00054431e+00 2.52211094e-01 7.60146677e-02 -7.66544998e-01
1.79861188e-01 -3.85565519e-01 1.98117197e-01 -7.98496246e-01
5.11945426e-01 -7.41078481e-02 7.45449722e-01 -1.19534230e+00
1.61387190e-01 -4.64679986e-01 -4.57233131e-01 3.33055615e-01
-3.31744462e-01 9.18501765e-02 1.13420166e-01 1.77726999e-01
-3.87293875e-01 -4.07801747e-01 6.68489039e-01 -4.05879021e-02
-9.42704260e-01 5.72303087e-02 -5.84656775e-01 2.17907444e-01
9.94856954e-01 -6.10171318e-01 -5.56854784e-01 -2.51383394e-01
-4.15353537e-01 5.04239082e-01 2.97197044e-01 8.23425591e-01
7.45839000e-01 -1.19073582e+00 -9.63204324e-01 2.40566246e-02
2.78903037e-01 -1.72022954e-01 3.99072021e-01 8.79853547e-01
-1.06338263e+00 7.24022329e-01 -1.41357675e-01 -4.56801146e-01
-1.45890188e+00 -4.27192114e-02 1.89499497e-01 -3.06218062e-02
-6.59383953e-01 8.14813435e-01 -8.75580609e-02 -3.79872739e-01
4.76712346e-01 -5.77196181e-01 -3.03274840e-01 4.91909496e-02
7.73699164e-01 5.28315306e-01 1.24143988e-01 -6.14675403e-01
-2.43478149e-01 9.94026601e-01 -2.69539058e-01 -2.46360019e-01
8.63651812e-01 -1.01967640e-01 1.16099864e-01 4.28099602e-01
9.10195112e-01 2.79514492e-01 -1.05164063e+00 -7.24857207e-03
1.92085832e-01 -5.62918782e-01 -4.55711186e-01 -1.19099975e+00
-9.21251357e-01 9.73605156e-01 3.49221021e-01 -2.91430265e-01
1.07269454e+00 -4.40826595e-01 7.46371925e-01 5.50552487e-01
3.22094262e-01 -1.76384151e+00 1.83386818e-01 1.04382050e+00
9.82969105e-01 -1.23241889e+00 -1.37628078e-01 -3.82065475e-01
-1.15374470e+00 1.44523227e+00 6.67508006e-01 1.77092060e-01
2.05299020e-01 1.00822926e-01 4.47840631e-01 3.11926007e-01
-2.39059880e-01 -9.27602500e-02 4.10233766e-01 6.47117019e-01
5.84954977e-01 2.19749019e-01 -5.33096790e-01 6.14648402e-01
-2.27272511e-01 -1.21904291e-01 9.09776092e-01 9.24359441e-01
-2.81812459e-01 -1.23211133e+00 -5.90151727e-01 4.83856529e-01
-2.02586293e-01 -2.23586887e-01 -7.78009892e-01 7.16602385e-01
-1.09905422e-01 1.02084982e+00 -1.22444533e-01 -2.12178677e-01
4.53373313e-01 -3.27624023e-01 3.96192342e-01 -2.33693421e-01
-4.55082208e-01 2.93560386e-01 1.68460041e-01 -3.24241146e-02
-1.55569166e-01 -9.29608822e-01 -1.37979102e+00 -1.39316738e-01
-5.26655614e-01 9.17088017e-02 9.28078055e-01 7.21521020e-01
6.91235811e-03 3.33606988e-01 6.24573350e-01 -3.61418486e-01
-4.69978034e-01 -9.70177412e-01 -6.76079035e-01 2.43869536e-02
-1.99021734e-02 -1.39958546e-01 9.22229700e-03 3.64475518e-01]
|
[11.951560020446777, 2.23593807220459]
|
7b4236f0-7ae7-408b-9f7b-18ec605dff25
|
efficient-video-scene-text-spotting-unifying
|
1903.03299
| null |
https://arxiv.org/abs/1903.03299v3
|
https://arxiv.org/pdf/1903.03299v3.pdf
|
You Only Recognize Once: Towards Fast Video Text Spotting
|
Video text spotting is still an important research topic due to its various real-applications. Previous approaches usually fall into the four-staged pipeline: text detection in individual images, framewisely recognizing localized text regions, tracking text streams and generating final results with complicated post-processing skills, which might suffer from the huge computational cost as well as the interferences of low-quality text. In this paper, we propose a fast and robust video text spotting framework by only recognizing the localized text one-time instead of frame-wisely recognition. Specifically, we first obtain text regions in videos with a well-designed spatial-temporal detector. Then we concentrate on developing a novel text recommender for selecting the highest-quality text from text streams and only recognizing the selected ones. Here, the recommender assembles text tracking, quality scoring and recognition into an end-to-end trainable module, which not only avoids the interferences from low-quality text but also dramatically speeds up the video text spotting process. In addition, we collect a larger scale video text dataset (LSVTD) for promoting the video text spotting community, which contains 100 text videos from 22 different real-life scenarios. Extensive experiments on two public benchmarks show that our method greatly speeds up the recognition process averagely by 71 times compared with the frame-wise manner, and also achieves the remarkable state-of-the-art.
|
['ShiLiang Pu', 'Zhanzhan Cheng', 'Jing Lu', 'Shuigeng Zhou', 'Yi Niu', 'Fei Wu']
|
2019-03-08
| null | null | null | null |
['text-spotting']
|
['computer-vision']
|
[ 4.43354607e-01 -9.29177821e-01 -1.35245427e-01 -1.69094950e-01
-9.02366042e-01 -2.71018147e-01 4.71831143e-01 -2.25532934e-01
-3.96067321e-01 4.74806875e-02 2.12774947e-01 -1.24607556e-01
1.33765310e-01 -6.05643928e-01 -5.72774291e-01 -7.40920126e-01
5.03002226e-01 4.50491428e-01 5.36790371e-01 2.61306018e-01
4.46639270e-01 2.86951140e-02 -1.68765414e+00 3.70008945e-01
1.02725530e+00 1.15713978e+00 3.89382690e-01 8.34451020e-01
-4.24428046e-01 6.94257975e-01 -4.87397522e-01 -3.71741116e-01
1.52417943e-01 -2.42055550e-01 -1.34457633e-01 5.84819257e-01
6.38481796e-01 -6.49047077e-01 -6.14973903e-01 9.25796986e-01
4.97059822e-01 1.47975832e-01 3.79122555e-01 -1.09135437e+00
-3.10982049e-01 5.14164090e-01 -9.44558442e-01 1.31089374e-01
5.38314819e-01 2.37825900e-01 1.03703940e+00 -1.27448583e+00
3.84379804e-01 1.08454597e+00 4.60731596e-01 3.92029643e-01
-6.48075104e-01 -8.38925660e-01 2.87415445e-01 2.15131953e-01
-1.46262193e+00 -6.43495381e-01 5.77685118e-01 -4.78818119e-01
6.61122859e-01 4.11140919e-01 6.57223642e-01 9.70843196e-01
-1.87072344e-02 1.37150061e+00 4.42307532e-01 -1.71131656e-01
-5.52604720e-02 -1.97193608e-01 -1.20280482e-01 7.66644061e-01
3.37416283e-03 -4.60550636e-01 -9.08717453e-01 5.02340607e-02
6.36290669e-01 5.21247327e-01 -4.31039065e-01 -6.53954521e-02
-1.68856406e+00 3.25491607e-01 -2.44624034e-01 3.85868490e-01
-2.83818334e-01 -7.71158189e-03 5.19788802e-01 3.42794806e-01
6.10773146e-01 -3.20825756e-01 -2.01551378e-01 -4.43847597e-01
-1.56935930e+00 -1.27111927e-01 3.69814575e-01 9.11607265e-01
5.36524713e-01 -1.79551423e-01 -5.67440212e-01 9.05331373e-01
3.03656161e-01 9.15035009e-01 5.84028065e-01 -2.23195568e-01
1.11768198e+00 7.66912758e-01 9.51098800e-02 -1.09386098e+00
-4.13926840e-02 -4.06323816e-04 -8.47660065e-01 -3.52891058e-01
3.52811307e-01 -5.00122197e-02 -8.16766143e-01 9.02267277e-01
3.24627429e-01 5.00965297e-01 -4.10557568e-01 1.02174819e+00
6.67262197e-01 1.07393003e+00 -1.56224996e-01 -4.82569158e-01
1.34798908e+00 -1.13742435e+00 -7.21543908e-01 -7.10789636e-02
6.57928824e-01 -1.10366452e+00 1.29458773e+00 5.72119415e-01
-7.38554418e-01 -3.75428200e-01 -8.37764919e-01 -1.19390367e-02
-2.66652759e-02 7.35379815e-01 1.72176689e-01 6.39848053e-01
-8.71384144e-01 3.08394164e-01 -8.04577112e-01 -3.82118285e-01
4.40591276e-01 2.48538792e-01 -5.25977127e-02 -2.65337229e-01
-6.92743540e-01 4.49526198e-02 6.76227212e-02 1.21579006e-01
-6.88267410e-01 -3.33573252e-01 -5.27571678e-01 1.21472277e-01
9.39975679e-01 -4.73277956e-01 1.00432563e+00 -1.13495898e+00
-1.65835643e+00 4.90319014e-01 -5.00228941e-01 -4.34243605e-02
8.51262450e-01 -4.14225847e-01 -5.57793558e-01 3.41819048e-01
1.92867741e-01 3.82711262e-01 1.32725620e+00 -4.88682568e-01
-1.06489575e+00 -3.13015312e-01 -4.75104928e-01 5.58850288e-01
-9.35408831e-01 2.94421136e-01 -1.32033396e+00 -9.90720928e-01
1.64153472e-01 -6.09895825e-01 3.03384513e-01 2.92181134e-01
-3.89395386e-01 -4.84404713e-01 1.14001358e+00 -7.59438872e-01
1.56281412e+00 -2.22741985e+00 6.31594360e-02 -2.25391816e-02
3.60972971e-01 2.38024265e-01 -5.04337363e-02 2.02992320e-01
2.78209150e-01 -5.22217788e-02 1.90467000e-01 -5.49223006e-01
6.00521639e-02 -3.85429889e-01 -3.76366913e-01 4.54351485e-01
-7.36326096e-04 7.03058302e-01 -7.78260529e-01 -8.66630793e-01
5.98093688e-01 3.42404068e-01 -2.59611577e-01 1.53763935e-01
-3.63428414e-01 7.78472871e-02 -7.13862658e-01 1.01256526e+00
6.19987845e-01 -3.82861823e-01 -1.61223665e-01 -2.86051810e-01
-1.46585718e-01 8.22442695e-02 -1.26742911e+00 1.63659215e+00
-1.93749443e-01 8.10273647e-01 -6.19359175e-03 -7.30168521e-01
6.95818007e-01 2.73517936e-01 6.97357178e-01 -7.42747068e-01
2.79862642e-01 1.91513389e-01 -6.20476663e-01 -7.94736981e-01
7.29169130e-01 2.95251995e-01 9.63880420e-02 6.92660868e-01
-2.66569763e-01 3.92222315e-01 3.55567694e-01 3.04618806e-01
1.08428478e+00 1.70349613e-01 -2.34012961e-01 2.72300631e-01
6.71132982e-01 -1.74192622e-01 2.89433509e-01 5.10865569e-01
-1.75282508e-01 7.38756061e-01 3.05792987e-01 -2.48486489e-01
-8.74781609e-01 -4.66228008e-01 1.50148332e-01 1.41224360e+00
4.35770363e-01 -7.31520057e-01 -6.86341405e-01 -9.82243717e-01
-1.95018858e-01 2.09915176e-01 -4.29325730e-01 1.31251663e-01
-5.53080022e-01 -6.44261658e-01 5.23264885e-01 3.56204718e-01
5.41034877e-01 -8.02846909e-01 -4.12858278e-01 -5.16558765e-03
-5.33841729e-01 -1.34576440e+00 -1.23289561e+00 -2.83851922e-01
-6.78783000e-01 -8.93739283e-01 -1.18960941e+00 -7.53342688e-01
5.80496848e-01 9.86176372e-01 5.79376101e-01 2.74353504e-01
-6.26612827e-02 2.11800158e-01 -6.85118794e-01 2.45078191e-01
6.17693029e-02 -5.87236956e-02 -1.31006509e-01 6.33122265e-01
4.36511666e-01 8.71547759e-02 -5.98958969e-01 7.35853255e-01
-1.03158236e+00 4.00747150e-01 6.66819632e-01 7.79239357e-01
6.33594990e-01 4.44231898e-01 1.35954201e-01 -4.77075517e-01
3.31151426e-01 -2.12517083e-01 -5.72140455e-01 4.19214875e-01
-4.07373697e-01 -3.59715044e-01 7.67307401e-01 -6.29950583e-01
-9.22689438e-01 2.88632303e-01 9.86671299e-02 -7.96877921e-01
-4.19505872e-02 2.35462278e-01 -2.62871951e-01 -4.33010384e-02
1.82559416e-01 8.44758332e-01 -3.54845554e-01 -4.46579516e-01
-3.11388392e-02 9.58164811e-01 2.14799255e-01 -3.44860137e-01
1.03379118e+00 4.14234936e-01 -4.28290397e-01 -9.33671057e-01
-4.32893932e-01 -8.90557885e-01 -3.20921957e-01 -4.47935224e-01
6.64955437e-01 -1.11611676e+00 -6.08392894e-01 8.26699376e-01
-8.55093956e-01 -1.59815028e-01 3.55953187e-01 3.80087972e-01
-2.90568292e-01 9.33084667e-01 -5.78621984e-01 -7.61957943e-01
-6.47678673e-01 -1.27570570e+00 1.67118382e+00 1.74973592e-01
4.05438542e-01 -4.27350789e-01 -3.27289790e-01 5.26610553e-01
2.35277176e-01 -4.58718121e-01 3.69017631e-01 -4.00306821e-01
-8.59305084e-01 -3.55008155e-01 -5.59693754e-01 -1.24362290e-01
-3.30334567e-02 1.93094626e-01 -7.57381320e-01 -3.00738901e-01
-2.70481557e-01 -1.16020977e-01 9.10673022e-01 2.25081697e-01
1.13295829e+00 -6.08678311e-02 -3.35899770e-01 6.30732656e-01
1.23792803e+00 1.64624318e-01 4.44685489e-01 3.12022328e-01
1.00102103e+00 3.01318347e-01 1.11263084e+00 7.21160114e-01
2.74982065e-01 8.04052114e-01 1.03260994e-01 1.67407822e-02
6.14470951e-02 -2.80297756e-01 7.43414223e-01 8.41852069e-01
2.84196436e-01 -7.08647728e-01 -7.38058448e-01 3.42602700e-01
-2.13614321e+00 -9.04213369e-01 -1.63096562e-01 2.19420218e+00
5.99989712e-01 1.53582171e-01 4.02939767e-01 4.14537668e-01
1.11168265e+00 2.66797692e-01 -6.34253681e-01 5.88735819e-01
-6.59489706e-02 -2.01753125e-01 4.32965040e-01 5.72818033e-02
-1.30322719e+00 1.11680675e+00 5.01359844e+00 1.42355657e+00
-1.30684769e+00 -4.93401662e-02 7.28354871e-01 -5.25442064e-01
1.71820685e-01 -3.16295803e-01 -9.11685348e-01 8.52786958e-01
6.08985364e-01 -1.88712865e-01 4.28651005e-01 6.00656271e-01
5.80548525e-01 -2.10780874e-01 -8.37117016e-01 1.53154027e+00
3.94882083e-01 -1.22951591e+00 2.65842766e-01 -1.83030188e-01
4.51025277e-01 -4.95273173e-02 -1.55799836e-01 1.04884379e-01
-2.89774895e-01 -6.89233720e-01 9.69090104e-01 4.04448032e-01
1.01862359e+00 -5.78951955e-01 5.04309833e-01 3.16448331e-01
-1.71747375e+00 1.50190452e-02 -3.43898624e-01 3.57769400e-01
4.21318337e-02 8.94273102e-01 -5.57683229e-01 4.51802790e-01
8.60353231e-01 1.26703775e+00 -6.42185032e-01 1.31559932e+00
2.13376924e-01 6.51069701e-01 -5.12110472e-01 -4.13199991e-01
1.43984139e-01 -1.56887114e-01 4.18581754e-01 1.24587858e+00
6.39255285e-01 4.44263825e-03 4.05913800e-01 4.31560397e-01
-1.64731964e-01 5.72323263e-01 -1.97385803e-01 -3.00523371e-01
3.27212960e-01 1.36372817e+00 -1.16414094e+00 -5.55693686e-01
-5.75556040e-01 1.25480020e+00 -2.01620668e-01 1.86657384e-01
-9.78248775e-01 -6.95585668e-01 1.16288781e-01 8.41469243e-02
5.93592167e-01 -1.33985475e-01 -1.94950968e-01 -1.74548137e+00
5.70415378e-01 -1.05485785e+00 1.38503924e-01 -8.83290112e-01
-9.45015907e-01 3.66452932e-01 -6.40779078e-01 -1.65703535e+00
3.24531585e-01 -4.47404414e-01 -5.46286166e-01 5.22242188e-01
-1.30833459e+00 -1.04273295e+00 -7.37419724e-01 8.86104643e-01
1.25553167e+00 -5.11347800e-02 6.01299629e-02 8.00627649e-01
-1.14116216e+00 8.51635337e-01 2.97562063e-01 2.80664146e-01
1.19582701e+00 -7.22996294e-01 4.39685613e-01 1.06946182e+00
2.04020157e-01 1.57830432e-01 3.50152761e-01 -8.11405361e-01
-1.87759089e+00 -1.18782854e+00 8.09319854e-01 -3.21686804e-01
8.16499174e-01 -4.97900188e-01 -8.58340323e-01 2.05372050e-01
-1.00604817e-01 -2.56653696e-01 1.94551870e-01 -3.51128042e-01
-7.05086738e-02 -2.20757380e-01 -6.91131294e-01 6.49977922e-01
1.04057300e+00 -4.44886863e-01 -1.95529521e-01 5.31428695e-01
5.53766727e-01 -3.92894924e-01 -4.06865567e-01 4.95438725e-02
5.89640617e-01 -9.64862406e-01 6.71455026e-01 2.85406709e-01
3.66210312e-01 -5.57884693e-01 1.36693688e-02 -6.93403840e-01
3.55215892e-02 -7.32567370e-01 -6.34485260e-02 1.51988637e+00
1.45262286e-01 -2.73905873e-01 1.07997942e+00 1.77437156e-01
1.44340433e-02 -6.05076849e-01 -8.77532661e-01 -3.66996855e-01
-6.70595646e-01 -6.03083968e-01 5.20776153e-01 7.83206344e-01
1.35807306e-01 3.34363878e-01 -7.72088587e-01 -3.93757410e-02
4.17500854e-01 2.52774924e-01 8.17084014e-01 -1.09781253e+00
-2.44462833e-01 -6.20298624e-01 -3.50942880e-01 -1.86540747e+00
-1.17086500e-01 -4.54366088e-01 4.40180540e-01 -1.24304903e+00
4.24029738e-01 -3.04742813e-01 -1.70600954e-02 3.61316979e-01
-4.87688452e-01 1.15834258e-01 1.35082111e-01 5.40036619e-01
-1.30452514e+00 5.96706331e-01 1.10718679e+00 -3.12838018e-01
-4.12380807e-02 7.12421536e-02 -3.21050018e-01 5.60307384e-01
3.58066857e-01 -4.01059151e-01 -3.24205220e-01 -6.78273976e-01
2.88399041e-01 3.35168093e-01 6.99508786e-02 -1.15750122e+00
5.50226688e-01 -2.05619916e-01 5.53865671e-01 -1.12437630e+00
1.33887947e-01 -1.01155293e+00 -1.86797157e-01 1.48331374e-01
-1.51727453e-01 -4.92322929e-02 -1.71340629e-02 7.37214744e-01
-5.95750436e-02 -1.60242394e-01 5.85568368e-01 3.61605793e-01
-6.85685098e-01 5.73356748e-01 -4.18699563e-01 -2.19899908e-01
1.03797221e+00 -2.80316651e-01 -3.45635235e-01 -2.78749943e-01
-2.09653705e-01 4.32217836e-01 6.85375631e-01 6.02095425e-01
7.17553973e-01 -1.09439635e+00 -5.65319300e-01 3.46857876e-01
1.91341579e-01 -1.40516922e-01 3.90540063e-01 1.09775782e+00
-4.25477356e-01 3.52928877e-01 2.68152922e-01 -9.46184993e-01
-1.49860454e+00 5.07022798e-01 9.89908651e-02 -2.00536251e-01
-8.50575447e-01 6.75518513e-01 2.17281476e-01 1.84472531e-01
5.88370204e-01 -2.80189395e-01 -1.03797697e-01 9.26313549e-02
7.85924435e-01 3.85655552e-01 1.52043641e-01 -7.62713909e-01
-2.57372499e-01 1.07465434e+00 -2.11139500e-01 1.61102131e-01
9.40353036e-01 -5.13080060e-01 3.22252274e-01 2.84368098e-01
9.30844247e-01 5.51221296e-02 -1.35911608e+00 -4.49466467e-01
-8.06502923e-02 -8.71597469e-01 2.83647478e-01 -5.26208639e-01
-1.36206436e+00 9.23502386e-01 5.98189056e-01 2.82968730e-02
1.37543941e+00 -3.99406821e-01 1.26418865e+00 4.67399508e-01
2.18990356e-01 -1.23318553e+00 3.39561343e-01 3.65033269e-01
3.43668669e-01 -1.28846359e+00 2.24098912e-03 -3.73366952e-01
-6.06323004e-01 1.24285245e+00 5.41707575e-01 9.64692086e-02
3.00409108e-01 2.06176162e-01 -1.84159633e-02 9.78034288e-02
-9.12134647e-01 -1.31512016e-01 3.21509957e-01 8.64602774e-02
3.91872942e-01 -1.53098822e-01 -2.03348435e-02 4.25315827e-01
2.63100594e-01 1.46280736e-01 3.15103740e-01 6.82544053e-01
-8.02294254e-01 -8.75487745e-01 -5.56570947e-01 8.70856345e-01
-4.88008350e-01 -3.04972321e-01 -3.02707613e-01 2.76120156e-01
-1.00385636e-01 1.10929978e+00 2.18768953e-03 -7.62856126e-01
1.87340692e-01 -3.69983464e-02 -1.23364385e-02 -3.07308942e-01
-4.49025571e-01 8.41137767e-01 -1.52414843e-01 -5.33176482e-01
-2.95497894e-01 -7.92221725e-01 -1.15536833e+00 -5.27212381e-01
-7.13348031e-01 -8.60191649e-04 5.28242230e-01 9.33504045e-01
4.51907188e-01 3.65340382e-01 8.96893799e-01 -1.00399828e+00
-1.62301823e-01 -9.16687250e-01 -3.64584714e-01 3.40231746e-01
2.34814137e-01 -5.33616900e-01 -4.27362204e-01 2.70038754e-01]
|
[11.986751556396484, 2.148066759109497]
|
fadb4694-c94e-453a-bf47-0219cd1bd463
|
graph-encoder-embedding
|
2109.13098
| null |
https://arxiv.org/abs/2109.13098v3
|
https://arxiv.org/pdf/2109.13098v3.pdf
|
One-Hot Graph Encoder Embedding
|
In this paper we propose a lightning fast graph embedding method called one-hot graph encoder embedding. It has a linear computational complexity and the capacity to process billions of edges within minutes on standard PC -- making it an ideal candidate for huge graph processing. It is applicable to either adjacency matrix or graph Laplacian, and can be viewed as a transformation of the spectral embedding. Under random graph models, the graph encoder embedding is approximately normally distributed per vertex, and asymptotically converges to its mean. We showcase three applications: vertex classification, vertex clustering, and graph bootstrap. In every case, the graph encoder embedding exhibits unrivalled computational advantages.
|
['Carey E. Priebe', 'Qizhe Wang', 'Cencheng Shen']
|
2021-09-27
| null | null | null | null |
['stochastic-block-model']
|
['graphs']
|
[ 3.81720960e-02 6.24403298e-01 -5.39051771e-01 7.02886134e-02
-3.42710644e-01 -7.36949086e-01 4.65423346e-01 5.77742755e-01
-2.18002528e-01 4.03215557e-01 -1.44972175e-01 -8.13081741e-01
1.58175513e-01 -1.09693062e+00 -8.47487271e-01 -6.04457498e-01
-6.93624914e-01 8.35030615e-01 9.65673029e-02 -7.31634498e-02
-1.29514650e-01 3.58724177e-01 -9.02847648e-01 -2.71697968e-01
5.92595398e-01 5.05374908e-01 -7.02496469e-02 1.21721578e+00
-1.92305401e-01 6.72816098e-01 -2.41591796e-01 -8.48106861e-01
3.12842041e-01 -1.36122465e-01 -7.02646971e-01 7.61618791e-03
2.60789484e-01 -1.00921385e-01 -1.02194965e+00 1.32014000e+00
4.79303509e-01 -2.38706067e-01 5.53987443e-01 -1.66763115e+00
-9.45108533e-01 7.98939705e-01 -5.46887875e-01 1.65330604e-01
6.50577784e-01 -2.73129582e-01 1.53747654e+00 -6.44963622e-01
8.05554152e-01 1.08841240e+00 8.08619976e-01 1.45182550e-01
-1.58787334e+00 -2.40038812e-01 -1.50857061e-01 2.12041721e-01
-1.47114813e+00 -1.23679385e-01 7.63475835e-01 -4.64941174e-01
8.49068046e-01 2.80087203e-01 9.04390931e-01 7.26605654e-01
3.51439327e-01 7.36905217e-01 7.05562770e-01 -4.57688987e-01
2.48879060e-01 7.98219442e-02 9.44142118e-02 1.10806322e+00
6.49388969e-01 -4.41388339e-02 -1.62984371e-01 -5.27489245e-01
6.86708748e-01 -3.16976868e-02 -2.63817936e-01 -6.57527328e-01
-1.21924138e+00 1.06208277e+00 4.73694265e-01 1.19519562e-01
-5.57076409e-02 7.70427585e-01 6.10687912e-01 8.53961229e-01
6.07536733e-01 7.53755793e-02 5.98261431e-02 -1.49698511e-01
-4.75473464e-01 -7.73161128e-02 1.24134755e+00 1.41969848e+00
1.06335521e+00 1.57033205e-01 2.79446959e-01 2.86086440e-01
1.13453627e-01 5.96390486e-01 1.79816857e-01 -4.11692739e-01
4.88453805e-01 7.01364756e-01 -4.35486645e-01 -1.36303329e+00
-2.32413411e-01 -4.58060168e-02 -1.12781298e+00 -2.76267409e-01
-2.89970869e-03 9.98109207e-02 -6.74568355e-01 1.40464437e+00
1.96850389e-01 4.54315603e-01 -2.38072842e-01 2.59565622e-01
6.78372979e-01 6.46034420e-01 -2.93666154e-01 -1.76418170e-01
1.15955913e+00 -7.02262402e-01 -7.21863151e-01 -3.47252011e-01
9.14724767e-01 -3.95498276e-01 9.36550975e-01 -2.42289249e-02
-9.73485231e-01 -1.69446170e-01 -1.05327284e+00 -5.35911834e-03
-5.35626471e-01 -2.74132699e-01 9.93896127e-01 8.88876438e-01
-1.55012524e+00 5.58356464e-01 -7.60189056e-01 -3.82302254e-01
-3.34319472e-02 4.35536742e-01 -9.01871443e-01 -3.48792732e-01
-9.59304571e-01 4.67011452e-01 4.83821243e-01 -2.51934499e-01
-4.92677540e-01 -3.83919954e-01 -1.20287871e+00 1.99017301e-01
2.30538368e-01 -6.53445661e-01 6.77493334e-01 -5.67279518e-01
-1.21562445e+00 8.88093293e-01 -2.30761245e-01 -6.53485835e-01
2.87653685e-01 4.05294508e-01 -4.48418945e-01 3.09268683e-01
1.51194753e-02 2.14373991e-01 1.05830646e+00 -6.94253206e-01
-2.23443955e-02 -3.81118208e-01 -9.08568967e-03 -1.43707857e-01
-4.76936132e-01 -6.80494383e-02 -5.46585679e-01 -5.87553442e-01
2.59672165e-01 -1.23493576e+00 -4.12742674e-01 -2.22288415e-01
-4.02713865e-01 -5.77524863e-02 6.08768582e-01 -5.65608263e-01
1.45119357e+00 -2.32960749e+00 2.30666071e-01 5.73019803e-01
7.67375946e-01 -1.26160860e-01 -1.17580213e-01 9.77928102e-01
-5.04964769e-01 3.54713500e-01 -1.80391535e-01 1.56366751e-02
2.48608887e-01 3.32532346e-01 -2.08449662e-01 1.03124881e+00
-2.17738580e-02 1.29339421e+00 -1.06966174e+00 -4.37162757e-01
2.23616153e-01 3.49639326e-01 -6.55530155e-01 8.72649625e-02
2.12286130e-01 -3.90303016e-01 -3.35795060e-02 4.21423644e-01
6.36485338e-01 -7.43398190e-01 5.39594948e-01 -6.85970783e-02
6.63286746e-01 -6.85675517e-02 -1.29784179e+00 1.20233393e+00
-3.12457770e-01 9.71491814e-01 1.22855835e-01 -9.98093247e-01
8.80196869e-01 2.83339530e-01 5.38968861e-01 5.32834083e-02
1.21704310e-01 7.00065792e-02 -2.17385739e-02 -2.30253879e-02
3.67485315e-01 1.40610918e-01 -1.43514529e-01 8.04810822e-01
3.55544329e-01 1.38667664e-02 3.44306201e-01 8.45901310e-01
1.56517851e+00 -8.25025976e-01 5.94931245e-01 -2.34651119e-01
1.03944108e-01 -4.09576952e-01 1.05665140e-01 5.37736535e-01
-1.91211551e-02 2.98217326e-01 8.97565424e-01 -5.14353812e-01
-1.24838984e+00 -1.17901242e+00 3.24887961e-01 7.85585821e-01
-8.04207381e-03 -1.11360800e+00 -6.86360657e-01 -5.49871802e-01
2.91670263e-01 2.02702612e-01 -6.09946370e-01 -5.89912474e-01
-2.19323143e-01 -6.50498271e-01 4.21736270e-01 4.69621330e-01
-5.47237322e-02 -5.27113795e-01 2.60979533e-01 2.71069944e-01
1.50464177e-01 -1.15155792e+00 -7.61098266e-01 1.79994375e-01
-8.33336651e-01 -1.34325516e+00 -4.28047985e-01 -1.08259559e+00
8.68969858e-01 3.72042805e-01 1.29260576e+00 2.58879304e-01
-4.02606457e-01 6.82234764e-01 -2.93846995e-01 8.43084753e-02
-6.23873889e-01 3.01663995e-01 1.45832673e-01 1.35530248e-01
4.01888132e-01 -7.92776585e-01 -2.26369113e-01 -1.39707178e-01
-8.46780181e-01 -1.89722717e-01 2.81309992e-01 7.53720403e-01
5.63400626e-01 2.74459064e-01 1.33983642e-01 -1.24470818e+00
1.06808281e+00 -6.05078697e-01 -7.33899653e-01 3.09810728e-01
-1.02206326e+00 2.34454095e-01 7.39719808e-01 -2.62085080e-01
1.29215911e-01 1.48904383e-01 9.73726660e-02 -4.77607489e-01
3.61340404e-01 6.61548495e-01 1.47449402e-02 -4.43408012e-01
3.86061221e-01 2.85725862e-01 2.13445276e-01 -1.79554194e-01
7.43752003e-01 4.77864921e-01 3.63018513e-01 -1.82161614e-01
1.12110209e+00 3.91865611e-01 1.39143884e-01 -1.02649975e+00
6.57819619e-04 -5.38553357e-01 -5.67565143e-01 -9.13045779e-02
5.94932139e-01 -8.48001778e-01 -8.74828339e-01 -2.57586148e-02
-1.07166815e+00 -2.75090396e-01 -3.11374813e-01 3.45471203e-01
-5.84052384e-01 7.69140720e-01 -9.65587020e-01 -5.30925810e-01
-4.03610915e-01 -8.30293953e-01 8.44358623e-01 -3.09901714e-01
5.60223088e-02 -1.52122593e+00 3.29439074e-01 -5.20129979e-01
1.28832057e-01 3.22577894e-01 1.06896043e+00 -6.46707296e-01
-4.71797556e-01 -9.66242313e-01 -3.06521118e-01 1.86260834e-01
7.11026192e-02 2.00214699e-01 -4.39342529e-01 -7.76735485e-01
-6.36519551e-01 1.00901254e-01 7.22664654e-01 2.35342965e-01
1.11310720e+00 -3.50985706e-01 -4.87493455e-01 9.16939795e-01
1.54883862e+00 -4.79024917e-01 5.08276284e-01 -1.28054664e-01
9.24287081e-01 3.62162362e-03 -8.04978013e-02 4.05016541e-01
3.61747056e-01 3.51634890e-01 3.47877085e-01 -9.43645835e-03
9.89023820e-02 -5.19623280e-01 3.60457391e-01 1.54358935e+00
-4.91563268e-02 -3.12158227e-01 -9.59915161e-01 5.49167752e-01
-1.88895929e+00 -9.17299271e-01 -4.33266103e-01 2.14832020e+00
3.47742081e-01 8.69919136e-02 2.42128760e-01 2.26294234e-01
1.07734489e+00 4.53223944e-01 -2.20272556e-01 -6.87625587e-01
-1.71100363e-01 2.49176815e-01 1.18323493e+00 7.42752075e-01
-8.00587296e-01 9.39835310e-01 8.21836567e+00 5.13272822e-01
-5.42136014e-01 1.38740525e-01 3.04452479e-01 4.41912413e-01
-5.90799212e-01 2.72029161e-01 -2.79011488e-01 5.01800776e-01
1.23596370e+00 -9.75543618e-01 8.94527614e-01 1.11855435e+00
-5.30767560e-01 4.71178412e-01 -1.25196862e+00 1.41052794e+00
3.36853936e-02 -1.32522869e+00 1.34910151e-01 4.25844729e-01
6.15792453e-01 1.32640556e-01 9.77709983e-03 1.88488513e-01
6.61511719e-01 -1.15702415e+00 1.24152359e-02 -1.00600824e-01
1.21511197e+00 -9.58379567e-01 7.26995647e-01 8.09236467e-02
-1.76212764e+00 1.07716136e-01 -7.52808690e-01 8.80861655e-02
2.40209013e-01 7.66924560e-01 -1.13737130e+00 5.15146255e-01
3.56379896e-01 7.73519874e-01 -6.54840231e-01 9.21914577e-01
-2.32224315e-01 7.23553836e-01 -4.36706305e-01 2.90619750e-02
7.17835948e-02 -5.88640034e-01 5.50685883e-01 1.11184490e+00
3.54386359e-01 -1.48862556e-01 3.32369894e-01 3.13642204e-01
-3.59382629e-01 3.00541759e-01 -1.26236200e+00 -6.68383896e-01
5.76653779e-01 1.14568508e+00 -9.14241195e-01 -5.15274823e-01
-3.89150620e-01 1.36115885e+00 5.73986948e-01 3.42420340e-01
-8.27837527e-01 -7.63582945e-01 5.11369646e-01 9.11098123e-02
3.52521390e-01 -4.94493723e-01 1.57303467e-01 -1.34110045e+00
-3.11264992e-02 -5.90443134e-01 4.72441941e-01 -4.17865038e-01
-1.20867109e+00 6.93777204e-01 -2.96616137e-01 -9.59726572e-01
-2.20017239e-01 -9.41790342e-01 -6.83612049e-01 5.62078714e-01
-1.00071204e+00 -7.85413206e-01 -1.32725954e-01 9.30355787e-01
-3.83817017e-01 -1.75291285e-01 9.93769765e-01 3.48963112e-01
-4.69464362e-01 8.07548761e-01 3.28907043e-01 1.38068169e-01
4.03527796e-01 -1.73840261e+00 1.06623554e+00 7.94087648e-01
6.51751578e-01 4.58184093e-01 7.93706656e-01 -6.82171881e-01
-2.12861586e+00 -1.14741015e+00 9.84032989e-01 -1.53504878e-01
1.10095465e+00 -7.22986460e-01 -8.03850710e-01 1.18170500e+00
-1.50186615e-02 6.00806653e-01 8.19182098e-01 3.19709271e-01
-4.86142755e-01 2.69568283e-02 -8.99141073e-01 7.44701684e-01
1.04169977e+00 -1.05278957e+00 -5.96691743e-02 9.22786713e-01
9.20285940e-01 -2.67573655e-01 -1.20344961e+00 -7.47812316e-02
2.30867922e-01 -5.18984735e-01 9.17203605e-01 -6.65406883e-01
-1.49381220e-01 7.43291108e-03 -1.41037917e-02 -1.35910869e+00
-7.14452028e-01 -1.15873957e+00 -6.21103048e-01 6.85957670e-01
3.62762868e-01 -1.03286469e+00 9.89279151e-01 2.71477103e-01
3.09530675e-01 -4.68883485e-01 -1.04282570e+00 -9.81605947e-01
-4.10824835e-01 -3.98553699e-01 7.37217605e-01 1.02351630e+00
3.80298108e-01 4.43856329e-01 -2.97549278e-01 2.35789284e-01
8.06778610e-01 1.06061690e-01 9.96481597e-01 -1.22834563e+00
-4.36607659e-01 -3.45552921e-01 -1.17412937e+00 -9.66456175e-01
4.07282621e-01 -1.52318370e+00 -4.07661200e-01 -1.53904569e+00
1.21996738e-01 -1.37178063e-01 -1.26074165e-01 1.09671570e-01
-2.65909266e-02 2.25124314e-01 -1.05655625e-01 -2.43597522e-01
-5.34581065e-01 4.36835319e-01 7.24068344e-01 -1.50837466e-01
5.61895147e-02 -4.24594805e-02 -3.97537172e-01 3.51860106e-01
7.34650373e-01 -4.17599648e-01 -5.29072225e-01 -6.44491166e-02
7.88877189e-01 6.17196485e-02 2.08233684e-01 -6.52529776e-01
1.97800428e-01 1.91465557e-01 -3.29859823e-01 -4.14499700e-01
1.76219232e-02 -8.34236443e-01 3.75210434e-01 4.71954763e-01
-5.12425527e-02 6.88260436e-01 -1.12847023e-01 1.07468164e+00
-3.03467900e-01 -3.09010465e-02 4.44995016e-01 -1.01890434e-02
-4.96796936e-01 9.00341272e-01 -2.41926253e-01 2.19994813e-01
1.10254514e+00 -2.93513834e-01 -1.53457105e-01 -7.39515722e-01
-7.78728664e-01 -9.72236390e-04 7.79431522e-01 1.61779955e-01
7.27754354e-01 -1.81298029e+00 -6.58031166e-01 3.12508970e-01
2.44478479e-01 -4.34528172e-01 -9.19169784e-02 8.30367208e-01
-8.63527834e-01 1.33133173e-01 2.26119995e-01 -5.22813320e-01
-1.23587966e+00 1.10565877e+00 -5.88721558e-02 -2.61198252e-01
-8.87984633e-01 6.89408600e-01 -7.19723701e-02 -4.60009038e-01
-7.46794418e-02 -1.73037350e-01 4.08744186e-01 -2.63446808e-01
6.71706378e-01 4.48945731e-01 8.82915035e-03 -6.02859259e-01
-2.98457474e-01 2.13802397e-01 5.11934273e-02 1.29711583e-01
1.13594508e+00 9.52960849e-02 -6.62406743e-01 5.39301217e-01
1.76836765e+00 7.82288089e-02 -6.52466714e-01 -7.88215026e-02
6.56987801e-02 -5.15705287e-01 -9.92415324e-02 2.19642609e-01
-1.01973474e+00 7.71535039e-01 1.42235473e-01 9.44491625e-01
6.79262102e-01 2.87777215e-01 6.98751867e-01 5.60778737e-01
5.52497029e-01 -9.16502416e-01 -2.73961931e-01 3.95979017e-01
5.54742157e-01 -1.03784394e+00 7.54058510e-02 -7.36039460e-01
-3.48942667e-01 1.09582841e+00 2.84978934e-02 -6.79516852e-01
1.02634215e+00 3.42098236e-01 -5.83148301e-01 -2.43465990e-01
-9.21079338e-01 -3.89023609e-02 2.96850353e-01 6.69389009e-01
2.33768582e-01 5.46472430e-01 -7.03682564e-03 2.71341670e-02
-3.90679777e-01 -4.72209394e-01 6.83931470e-01 5.91947734e-01
-4.06287789e-01 -1.13506639e+00 9.49520245e-02 8.17938328e-01
-1.16257407e-01 -3.35099131e-01 -5.39546967e-01 6.82369769e-01
-6.28360212e-01 6.08353913e-01 -1.83954462e-02 -7.08390772e-01
-6.78698197e-02 3.12041733e-02 4.79113221e-01 -7.08677709e-01
-1.05851397e-01 -3.43330979e-01 1.29372343e-01 -7.01224864e-01
1.15743041e-01 -3.12321663e-01 -9.33149278e-01 -1.03239286e+00
-5.82314730e-01 3.01733106e-01 5.85832059e-01 2.70552367e-01
5.12916505e-01 2.34722778e-01 6.81515753e-01 -5.76772153e-01
-4.15065914e-01 -6.60652101e-01 -1.03076875e+00 3.94856244e-01
1.65383369e-01 -3.29233468e-01 -6.78174555e-01 -2.16869891e-01]
|
[7.1660027503967285, 5.852138519287109]
|
bfac278c-c999-4553-83d7-cb3f250040ba
|
evaluating-generalization-in-classical-and
|
2201.0877
| null |
https://arxiv.org/abs/2201.08770v3
|
https://arxiv.org/pdf/2201.08770v3.pdf
|
Generalization Metrics for Practical Quantum Advantage in Generative Models
|
As the quantum computing community gravitates towards understanding the practical benefits of quantum computers, having a clear definition and evaluation scheme for assessing practical quantum advantage in the context of specific applications is paramount. Generative modeling, for example, is a widely accepted natural use case for quantum computers, and yet has lacked a concrete approach for quantifying success of quantum models over classical ones. In this work, we construct a simple and unambiguous approach to probe practical quantum advantage for generative modeling by measuring the algorithm's generalization performance. Using the sample-based approach proposed here, any generative model, from state-of-the-art classical generative models such as GANs to quantum models such as Quantum Circuit Born Machines, can be evaluated on the same ground on a concrete well-defined framework. In contrast to other sample-based metrics for probing practical generalization, we leverage constrained optimization problems (e.g., cardinality-constrained problems) and use these discrete datasets to define specific metrics capable of unambiguously measuring the quality of the samples and the model's generalization capabilities for generating data beyond the training set but still within the valid solution space. Additionally, our metrics can diagnose trainability issues such as mode collapse and overfitting, as we illustrate when comparing GANs to quantum-inspired models built out of tensor networks. Our simulation results show that our quantum-inspired models have up to a $68 \times$ enhancement in generating unseen unique and valid samples compared to GANs, and a ratio of 61:2 for generating samples with better quality than those observed in the training set. We foresee these metrics as valuable tools for rigorously defining practical quantum advantage in the domain of generative modeling.
|
['Alejandro Perdomo-Ortiz', 'Marta Mauri', 'Kaitlin Gili']
|
2022-01-21
| null | null | null | null |
['tensor-networks']
|
['methodology']
|
[ 3.64037573e-01 3.50825995e-01 6.94393143e-02 -1.43752530e-01
-1.11046481e+00 -5.18127143e-01 8.55108559e-01 -2.01009855e-01
-2.04262674e-01 9.80800688e-01 -1.39212400e-01 -4.66237158e-01
-3.77870321e-01 -1.28226709e+00 -5.70244849e-01 -1.23180223e+00
1.35403976e-01 6.67207897e-01 -3.96858782e-01 -4.20045316e-01
2.57989317e-01 2.46472523e-01 -1.43804348e+00 -1.38114333e-01
1.18123043e+00 7.74542987e-01 -2.51554638e-01 6.66269779e-01
2.62204200e-01 2.35633299e-01 -5.17815471e-01 -5.99199831e-01
2.71249682e-01 -1.06053495e+00 -8.52524400e-01 -2.52791762e-01
6.84996620e-02 -2.42745727e-01 -5.06156325e-01 1.30844498e+00
4.52055961e-01 2.86828857e-02 7.24177063e-01 -1.02215731e+00
-7.52492368e-01 9.28521156e-01 3.16882014e-01 1.36534080e-01
1.03133902e-01 5.74789166e-01 1.28171563e+00 -1.99194893e-01
5.82135618e-01 8.15270662e-01 2.32386723e-01 9.94632006e-01
-1.69760466e+00 -4.58565503e-01 -8.36641788e-01 1.13389619e-01
-1.24960542e+00 -4.08730358e-01 5.83403051e-01 -3.90511125e-01
7.60670602e-01 3.48323017e-01 4.45238173e-01 1.17549205e+00
1.25842288e-01 9.41867083e-02 1.47017348e+00 -7.15508282e-01
5.86272717e-01 2.02548862e-01 1.00537069e-01 4.88065749e-01
2.79646873e-01 7.35701680e-01 -6.83636248e-01 -1.53015479e-01
7.18327224e-01 -2.97128052e-01 -2.70971328e-01 -2.06033170e-01
-1.21221685e+00 9.76450503e-01 4.70723659e-01 4.11785007e-01
-1.68551430e-01 6.62105918e-01 4.72392477e-02 1.81966007e-01
1.27552301e-01 8.86056781e-01 -8.95102769e-02 -4.59745198e-01
-1.00966203e+00 3.79726917e-01 5.80106974e-01 8.91779065e-01
1.08573890e+00 7.26845860e-02 -2.42187917e-01 1.46701634e-01
2.00583581e-02 5.28409243e-01 2.23292634e-01 -9.89928663e-01
5.28628267e-02 2.57951647e-01 6.49042130e-02 -1.05750784e-01
-1.65301695e-01 -7.59694993e-01 -7.34213352e-01 1.34505451e-01
4.73186314e-01 -2.23179236e-02 -6.76437795e-01 2.03333473e+00
8.08494985e-02 -2.68844254e-02 3.65773976e-01 6.70978844e-01
4.30866003e-01 4.67052341e-01 -1.59673229e-01 -2.03036398e-01
1.08718967e+00 -3.34553093e-01 -2.39114806e-01 1.74661532e-01
1.06978524e+00 -4.43927974e-01 1.15612686e+00 3.92954737e-01
-1.02054679e+00 -2.94969231e-01 -1.19230998e+00 8.58245343e-02
-1.29824221e-01 -2.73846954e-01 1.14270127e+00 1.43174982e+00
-1.09705091e+00 1.25156903e+00 -8.57578754e-01 -3.90300840e-01
3.76696467e-01 3.51749212e-01 -2.80069262e-02 2.87058577e-02
-1.07073951e+00 9.10084009e-01 5.59914351e-01 -5.20908497e-02
-1.07385707e+00 -4.54598546e-01 -2.48222724e-01 1.90794870e-01
1.38365090e-01 -1.20545125e+00 1.05716729e+00 -3.55088860e-01
-1.77464461e+00 6.25518262e-01 4.40225080e-02 -6.16854370e-01
1.39694884e-01 3.96622211e-01 -3.96873951e-01 2.19134279e-02
-1.40534863e-01 4.05594975e-01 5.71385086e-01 -8.98976028e-01
-3.80227901e-02 -3.45387995e-01 4.95028466e-01 -1.43198833e-01
-1.73831493e-01 -2.57063597e-01 3.83924127e-01 2.58323532e-02
2.73562908e-01 -1.17912495e+00 -2.85158396e-01 -7.88581371e-01
-5.04993081e-01 -7.32540116e-02 2.96570987e-01 -9.73608121e-02
1.05158639e+00 -1.80438268e+00 3.57269526e-01 2.72301286e-01
3.05353373e-01 -7.97815770e-02 1.24522440e-01 7.93805659e-01
1.67506501e-01 4.64992404e-01 -3.95973265e-01 -1.39701128e-01
4.28610444e-01 2.93563336e-01 -4.23556298e-01 3.87808681e-01
2.61464357e-01 1.20191169e+00 -8.98149550e-01 1.02130324e-02
1.79760754e-01 3.74733806e-01 -9.19912815e-01 6.99227154e-02
-2.90920228e-01 9.72063422e-01 -5.34331501e-01 3.47843170e-01
4.58623469e-01 -3.94509315e-01 1.14173703e-01 6.68752119e-02
1.49446011e-01 6.48281634e-01 -7.43685901e-01 1.67033684e+00
-5.36865115e-01 4.32967365e-01 -4.38998103e-01 -1.02504587e+00
5.54531336e-01 6.63326606e-02 2.90141180e-02 -8.01375091e-01
1.71209931e-01 5.27211845e-01 5.28178334e-01 -1.31995335e-01
6.26725495e-01 -8.30710113e-01 -2.56456077e-01 6.36126935e-01
5.70273757e-01 -5.41470528e-01 1.98921680e-01 3.38444531e-01
1.06999874e+00 1.55919880e-01 5.27695082e-02 -4.05066043e-01
2.15122938e-01 -2.18135521e-01 6.07210323e-02 1.08875561e+00
-4.27008867e-02 5.36227047e-01 5.84341586e-01 -8.33137110e-02
-1.44762444e+00 -1.22581244e+00 -6.24963045e-01 5.55909455e-01
1.06254071e-01 -6.84232116e-01 -1.02014136e+00 -2.42728472e-01
-4.09651935e-01 1.13120282e+00 -5.61330140e-01 -5.24509490e-01
-1.53270036e-01 -1.37612355e+00 7.40393400e-01 6.38349429e-02
4.26987469e-01 -6.63485348e-01 -3.30508441e-01 3.31889726e-02
6.29617646e-02 -1.02675521e+00 2.23151445e-01 3.48118573e-01
-1.03043568e+00 -8.26396763e-01 -1.90456972e-01 1.46769404e-01
4.96937335e-01 -3.16978216e-01 1.19766903e+00 7.46754631e-02
-2.71782279e-01 3.08191925e-01 -3.81470680e-01 -2.93978620e-02
-1.09853971e+00 8.35924894e-02 2.43267879e-01 -2.54977226e-01
2.28410274e-01 -8.20661306e-01 -7.27846801e-01 -9.14794765e-03
-9.95970964e-01 -2.48945691e-02 5.89227915e-01 1.09671295e+00
2.63847113e-01 3.66577432e-02 4.57824618e-01 -7.76098847e-01
4.97069657e-01 -3.08883011e-01 -5.35362005e-01 1.48032144e-01
-1.00689507e+00 5.53222835e-01 7.84640133e-01 1.11021176e-01
-6.84517682e-01 -6.19962513e-01 -2.99702615e-01 1.34557500e-01
-2.30207339e-01 4.36513990e-01 -8.97668824e-02 -3.58442992e-01
1.11913371e+00 5.23204386e-01 -1.50774196e-01 -9.22378823e-02
7.16647565e-01 4.13741827e-01 2.83103466e-01 -1.06855345e+00
1.04645932e+00 3.93404663e-01 6.48667514e-01 -8.46834302e-01
-7.04945683e-01 -1.81379374e-02 -4.83252168e-01 6.87106550e-02
9.82305586e-01 -4.92517889e-01 -8.47265780e-01 2.07529709e-01
-8.00342977e-01 -1.19323410e-01 -7.28031397e-01 5.93959153e-01
-9.60688889e-01 4.05452967e-01 -5.54220498e-01 -1.20439458e+00
-2.43502527e-01 -1.33525836e+00 1.09783459e+00 3.13562512e-01
2.20645204e-01 -1.12673116e+00 3.77157293e-02 4.41410094e-01
6.27525985e-01 2.84793109e-01 1.11812556e+00 -5.43021321e-01
-1.09462905e+00 -1.28142387e-01 1.05897486e-01 4.71128523e-01
-3.49907845e-01 -5.14348075e-02 -1.29198766e+00 -3.92002225e-01
2.69073159e-01 -4.71330673e-01 8.48182261e-01 1.23528764e-01
9.19199169e-01 -1.48796305e-01 7.12492110e-05 7.62690723e-01
1.57209074e+00 4.94174771e-02 1.07129049e+00 -9.93738472e-02
4.56633240e-01 1.22032598e-01 -1.10793814e-01 1.85307086e-01
-7.92391449e-02 7.27010727e-01 5.12509167e-01 4.55364525e-01
1.55715644e-01 -3.18820119e-01 3.09658885e-01 9.69640076e-01
-4.42557842e-01 -4.91843745e-02 -6.12804651e-01 -4.35600579e-02
-1.33965409e+00 -1.23568201e+00 -1.84012666e-01 2.74878931e+00
6.98176682e-01 3.14904660e-01 9.46526304e-02 1.81944564e-01
4.61980492e-01 -1.85333341e-01 -3.44868660e-01 -4.86546874e-01
-2.07377255e-01 8.98460984e-01 4.12432879e-01 2.80563325e-01
-6.01651788e-01 7.07502961e-01 6.55652046e+00 1.25552535e+00
-1.07361698e+00 4.31999952e-01 7.33392954e-01 -2.89697037e-03
-8.43325615e-01 7.46388555e-01 -7.24940121e-01 4.62394655e-01
1.48741937e+00 -2.36966789e-01 1.08312213e+00 5.73875844e-01
-2.59317756e-01 -1.09512754e-01 -1.50494778e+00 1.01511836e+00
-3.40695113e-01 -1.64883745e+00 -1.50731131e-02 6.70959353e-01
8.55661571e-01 8.56873766e-02 2.78376341e-01 4.76604789e-01
9.87121556e-03 -1.39128590e+00 7.52784789e-01 5.21510065e-01
1.04476988e+00 -5.98912776e-01 5.42905033e-01 5.45854151e-01
-3.78302962e-01 1.30365819e-01 -5.17096341e-01 -2.66933650e-01
9.78816971e-02 7.77649641e-01 -7.97731161e-01 8.19751441e-01
1.81309685e-01 1.69695839e-01 -4.82582152e-01 5.93395770e-01
-1.25034824e-01 9.60187316e-01 -4.84725028e-01 -2.25899935e-01
2.83987820e-01 -8.42554688e-01 5.87647438e-01 6.86067522e-01
6.48349702e-01 1.46831032e-02 -4.80612040e-01 1.80900538e+00
-1.93251241e-02 -1.70109153e-01 -5.92528880e-01 -6.77137792e-01
2.79845417e-01 1.30606127e+00 -6.32930994e-01 -2.05369160e-01
5.84977418e-02 7.54463613e-01 9.39159021e-02 2.64511555e-01
-1.02449727e+00 -1.77774414e-01 3.18660259e-01 -6.59659356e-02
8.82239938e-02 -4.18089807e-01 -3.97288293e-01 -1.45800340e+00
-1.25817999e-01 -7.69186318e-01 -1.35531142e-01 -4.49731410e-01
-1.23210001e+00 4.47627664e-01 -1.63096294e-01 -1.02390599e+00
-3.74805182e-01 -7.43203461e-01 -7.20772147e-01 1.11092615e+00
-1.07635915e+00 -9.83070552e-01 -9.02953297e-02 1.17802866e-01
-3.54699612e-01 -5.67748733e-02 1.15308917e+00 5.51096862e-03
-5.05044341e-01 6.49121821e-01 5.65549910e-01 -2.79332668e-01
-1.09131690e-02 -1.41595566e+00 3.86304796e-01 9.61173296e-01
6.03947520e-01 1.09494758e+00 9.28103149e-01 -1.57051414e-01
-1.78302848e+00 -5.72403669e-01 4.48588371e-01 -8.57359350e-01
7.28990555e-01 -4.94461000e-01 -4.00720447e-01 4.66020197e-01
2.24151984e-02 -2.47946382e-01 8.53443980e-01 4.62627798e-01
-3.70965600e-01 1.57204315e-01 -1.15420198e+00 4.98385459e-01
1.23780251e+00 -9.45320725e-01 -2.19546109e-01 6.18588984e-01
5.00213146e-01 -2.38133788e-01 -1.14361286e+00 2.52032757e-01
3.38581860e-01 -1.57059550e+00 8.00396621e-01 -7.27947950e-01
6.35949194e-01 -4.38206270e-02 -5.10424733e-01 -1.24861383e+00
-1.25703216e-01 -1.08829761e+00 6.51072860e-02 9.68067527e-01
4.31283534e-01 -8.77712131e-01 7.98523486e-01 6.00079060e-01
-2.69873738e-01 -8.11197937e-01 -1.33989513e+00 -9.66790259e-01
6.83806300e-01 -6.60044074e-01 6.91117942e-01 6.87810123e-01
8.36145654e-02 3.20731282e-01 -1.10622257e-01 2.66927835e-02
9.10584569e-01 1.00608692e-01 7.74675071e-01 -8.99149776e-01
-8.05206120e-01 -7.65382409e-01 -7.39720404e-01 -8.44126463e-01
-1.56091973e-01 -1.40817523e+00 -3.53346139e-01 -1.09144449e+00
3.99388671e-01 -7.53502131e-01 -1.96867794e-01 -3.26062530e-01
6.89556152e-02 2.73511082e-01 2.09362447e-01 2.51205623e-01
-2.78691322e-01 7.02222288e-01 1.17380393e+00 1.82529926e-01
2.25775272e-01 -1.16077229e-01 -7.55983710e-01 2.05075651e-01
5.59878826e-01 -5.05278587e-01 -2.52532631e-01 6.90382868e-02
7.32201397e-01 1.71170503e-01 7.76347756e-01 -1.33123589e+00
-6.62065074e-02 -1.77364752e-01 -3.12245339e-02 4.83407602e-02
3.99781734e-01 -8.24137628e-02 4.43654001e-01 4.37527955e-01
-1.72179520e-01 -5.53133845e-01 -3.44057590e-01 4.69528437e-01
1.41039327e-01 -6.24338150e-01 6.98515356e-01 -2.80302584e-01
-9.93576720e-02 2.44825110e-01 1.32842034e-01 2.15708688e-01
7.90541649e-01 -1.50054663e-01 -6.56735122e-01 -3.26280862e-01
-6.68918371e-01 -3.58295977e-01 8.59925389e-01 -2.94538796e-01
1.26006514e-01 -1.18240535e+00 -5.57746232e-01 1.46581933e-01
1.30680054e-01 -3.03370178e-01 3.96608561e-01 9.60633039e-01
-4.62745130e-01 6.91833556e-01 -6.11930937e-02 -7.87568152e-01
-2.66622454e-01 4.95606214e-01 5.87990344e-01 -3.08890134e-01
-2.70288020e-01 7.55011857e-01 2.27560967e-01 -4.34271097e-01
-6.59715831e-01 -4.48495358e-01 6.36249423e-01 -4.54457551e-01
2.87666805e-02 3.73878866e-01 2.82533824e-01 -4.10528004e-01
-1.70730293e-01 2.22662434e-01 3.93106997e-01 -2.91799337e-01
1.09560394e+00 1.97397828e-01 -2.61857629e-01 5.01704514e-01
1.08667016e+00 6.13809526e-02 -8.98235142e-01 4.33447771e-02
-2.93824673e-01 -1.96778715e-01 -8.59453976e-02 -6.90635264e-01
-4.15634423e-01 1.20204985e+00 4.50547874e-01 8.05611074e-01
7.44785547e-01 3.66616279e-01 7.01897264e-01 4.47734803e-01
9.32380676e-01 -9.37411666e-01 -7.62415156e-02 2.69164681e-01
4.44181710e-01 -1.09393823e+00 -1.32166520e-01 -1.21195227e-01
-1.74699739e-01 1.11545324e+00 5.29959053e-02 -1.86968431e-01
1.71364471e-01 -1.98182389e-01 -5.14341533e-01 -3.04805338e-01
-5.96907318e-01 -3.08787942e-01 2.22181126e-01 4.76044446e-01
3.84643495e-01 4.55098659e-01 -2.97137380e-01 5.02575874e-01
-8.33239377e-01 -2.61735976e-01 9.48808551e-01 5.13056695e-01
-3.88613820e-01 -1.47590637e+00 -1.23136997e-01 5.35962284e-01
-3.82105321e-01 -4.00957495e-01 -1.41357159e-04 6.21297836e-01
1.03762262e-01 9.07774031e-01 -1.95649266e-01 -6.63821578e-01
-1.65812522e-01 4.72061068e-01 1.16049159e+00 -7.44713962e-01
-3.45934093e-01 -3.56301337e-01 1.23728022e-01 -3.81649643e-01
-4.70991760e-01 -5.73056996e-01 -1.10641921e+00 -6.16582632e-01
-7.13556588e-01 3.73217493e-01 7.91736305e-01 1.18716609e+00
3.26016366e-01 3.98405850e-01 5.92566907e-01 -8.28220129e-01
-1.30760074e+00 -8.83428693e-01 -7.31771827e-01 4.27966237e-01
1.23312753e-02 -6.34450197e-01 -6.48048997e-01 -4.05378103e-01]
|
[5.605358600616455, 4.977115631103516]
|
98c27969-0e2e-4706-a393-f19562b14537
|
end-to-end-lane-detection-with-one-to-several
|
2305.00675
| null |
https://arxiv.org/abs/2305.00675v4
|
https://arxiv.org/pdf/2305.00675v4.pdf
|
End-to-End Lane detection with One-to-Several Transformer
|
Although lane detection methods have shown impressive performance in real-world scenarios, most of methods require post-processing which is not robust enough. Therefore, end-to-end detectors like DEtection TRansformer(DETR) have been introduced in lane detection.However, one-to-one label assignment in DETR can degrade the training efficiency due to label semantic conflicts. Besides, positional query in DETR is unable to provide explicit positional prior, making it difficult to be optimized. In this paper, we present the One-to-Several Transformer(O2SFormer). We first propose the one-to-several label assignment, which combines one-to-many and one-to-one label assignment to solve label semantic conflicts while keeping end-to-end detection. To overcome the difficulty in optimizing one-to-one assignment. We further propose the layer-wise soft label which dynamically adjusts the positive weight of positive lane anchors in different decoder layers. Finally, we design the dynamic anchor-based positional query to explore positional prior by incorporating lane anchors into positional query. Experimental results show that O2SFormer with ResNet50 backbone achieves 77.83% F1 score on CULane dataset, outperforming existing Transformer-based and CNN-based detectors. Futhermore, O2SFormer converges 12.5x faster than DETR for the ResNet18 backbone.
|
['Rui Zhou', 'Kunyang Zhou']
|
2023-05-01
| null | null | null | null |
['lane-detection']
|
['computer-vision']
|
[-1.88585743e-02 -1.16564564e-01 -4.13544089e-01 -6.10304832e-01
-8.96666169e-01 -4.21247095e-01 1.48380399e-01 -1.61692545e-01
-4.90728259e-01 4.67892468e-01 -2.20527370e-02 -4.22807097e-01
1.69011652e-01 -5.72393656e-01 -7.21568823e-01 -5.90291083e-01
1.80692002e-01 -2.69607361e-02 1.05993783e+00 -1.85310006e-01
1.91889837e-01 2.82006264e-01 -1.23502564e+00 1.42379895e-01
9.06296372e-01 1.08835649e+00 6.09798014e-01 5.01317620e-01
-1.34042442e-01 5.35109103e-01 -3.48408610e-01 -2.12874532e-01
2.94461846e-01 -1.00594431e-01 -2.09175214e-01 -9.25254300e-02
6.94719732e-01 -4.17777717e-01 -4.14393097e-01 1.12229168e+00
8.34491789e-01 -1.14770472e-01 4.02385205e-01 -1.14608335e+00
-2.31441483e-01 3.89643997e-01 -1.10581720e+00 2.56827831e-01
-1.08396389e-01 2.70092785e-01 9.16107476e-01 -1.21273077e+00
3.41097206e-01 1.26275527e+00 8.54609489e-01 1.69458807e-01
-8.17084372e-01 -8.79510283e-01 3.39194655e-01 3.75455499e-01
-1.67142475e+00 -2.75961757e-01 8.46068382e-01 -2.95024008e-01
6.77776277e-01 -3.96129452e-02 2.69283265e-01 6.87630475e-01
1.08341172e-01 1.07918215e+00 8.72396648e-01 -2.07267314e-01
-1.75574869e-01 -5.69996424e-02 1.53533727e-01 1.08313835e+00
9.61705446e-02 -5.11389561e-02 -5.02091348e-01 3.69850069e-01
7.05768764e-01 -1.19558230e-01 -9.62845981e-02 -9.61474851e-02
-1.18805909e+00 6.10054195e-01 8.05364430e-01 -5.50159886e-02
-9.99713391e-02 2.75025785e-01 4.90892738e-01 -4.63356311e-03
9.49634239e-02 -6.73537049e-03 -3.78062278e-01 1.11580156e-01
-1.05727971e+00 3.96564119e-02 1.39175892e-01 1.29453766e+00
7.93538034e-01 1.07073121e-01 -4.22868997e-01 9.21757042e-01
5.04362404e-01 5.50421357e-01 2.05959603e-01 -6.78194821e-01
8.13466489e-01 5.03812969e-01 -2.31078267e-01 -9.83756185e-01
-7.48084247e-01 -9.03733969e-01 -7.56253421e-01 5.46608157e-02
2.72711545e-01 -2.50864834e-01 -1.07623959e+00 1.60307574e+00
1.03289366e-01 4.41397786e-01 -1.38763204e-01 1.19173157e+00
7.24358082e-01 6.99577928e-01 1.64878085e-01 -5.85131720e-03
1.55731261e+00 -1.40548503e+00 -5.09162784e-01 -6.08738363e-01
1.03030670e+00 -9.36355412e-01 1.01517189e+00 2.61482060e-01
-5.93701482e-01 -6.19483590e-01 -1.28131568e+00 -1.28249839e-01
-1.57128468e-01 6.04859948e-01 1.62963033e-01 5.08053839e-01
-8.87034118e-01 -5.44290729e-02 -4.83653337e-01 -3.18319321e-01
1.92782640e-01 3.72465014e-01 -2.12566450e-01 -1.20732047e-01
-1.27846897e+00 7.07515061e-01 5.25345266e-01 3.71316075e-01
-6.69856608e-01 -4.01299447e-01 -7.58504331e-01 -2.16090716e-02
7.28305042e-01 -2.83884794e-01 1.31209779e+00 -4.33800846e-01
-1.20394146e+00 5.49153268e-01 -4.49344575e-01 -5.17277837e-01
6.21349812e-01 1.11686662e-02 -6.05103970e-01 -3.65492105e-02
3.37191373e-01 1.08894765e+00 5.49228907e-01 -1.23118448e+00
-1.22940254e+00 -4.45222221e-02 -4.20768037e-02 3.63882095e-01
-3.37207407e-01 -1.47993848e-01 -9.94353831e-01 -6.86350226e-01
4.19169515e-01 -8.48980427e-01 -2.38158852e-01 2.30878696e-01
-8.11613798e-01 -2.16758698e-01 1.11552477e+00 -4.01935399e-01
1.66633832e+00 -2.25660205e+00 -5.43636322e-01 5.57239763e-02
3.48425746e-01 4.72183019e-01 -1.75511792e-01 6.68183118e-02
1.60786882e-01 -3.80331464e-02 1.13854885e-01 -4.95815188e-01
4.72103134e-02 -3.17666754e-02 -6.41009510e-02 4.62671995e-01
1.14432715e-01 7.64399946e-01 -8.89361024e-01 -7.95275688e-01
4.42035139e-01 3.19887459e-01 -4.50441033e-01 -1.41225308e-01
4.27423120e-02 3.27674486e-02 -5.22906780e-01 8.33732545e-01
8.85084033e-01 -1.28307015e-01 -1.69368446e-01 -6.45387888e-01
-4.16331977e-01 2.04575181e-01 -9.77436304e-01 1.36316943e+00
-4.34299290e-01 1.01984441e+00 -9.83467475e-02 -5.66853821e-01
1.05674660e+00 -1.13678118e-02 2.14618042e-01 -1.08167565e+00
2.68686920e-01 2.84047604e-01 -3.54542546e-02 -3.34215581e-01
6.66126430e-01 1.85615979e-02 -4.35867310e-01 -3.76306295e-01
-4.62892354e-01 6.14269137e-01 1.39985889e-01 1.69717789e-01
1.11374974e+00 -1.59734413e-01 -1.07973091e-01 -1.42826915e-01
7.39724457e-01 -2.97676384e-01 1.04240263e+00 7.00119376e-01
-5.12425482e-01 7.77013838e-01 3.90812188e-01 -1.88265339e-01
-8.79626811e-01 -9.37878788e-01 -4.69195284e-02 1.10085750e+00
5.06897092e-01 -3.61856341e-01 -6.39929831e-01 -7.82041907e-01
-1.35497212e-01 7.41416752e-01 -2.42148265e-01 -2.30621174e-01
-7.25722969e-01 -5.27193248e-01 9.17539954e-01 7.37974465e-01
1.03844392e+00 -6.16743088e-01 -5.85915685e-01 5.15247405e-01
-1.61061376e-01 -1.38910460e+00 -9.42335486e-01 1.63643345e-01
-5.39405107e-01 -8.18917572e-01 -6.31808162e-01 -9.10634518e-01
5.68685591e-01 6.63766444e-01 5.72233498e-01 -5.48485331e-02
1.79417357e-02 -2.17126027e-01 -3.94529104e-01 -2.93586731e-01
1.77243233e-01 4.34542090e-01 -1.52663201e-01 -7.68902004e-02
3.72925878e-01 -1.59831315e-01 -8.13049972e-01 7.48379648e-01
-4.45636690e-01 3.11921954e-01 1.01189363e+00 5.73443234e-01
6.31273031e-01 -1.15854308e-01 6.59564316e-01 -6.78522825e-01
3.56429607e-01 -2.53350079e-01 -6.63478851e-01 1.20338440e-01
-7.44043529e-01 8.34792629e-02 8.11750174e-01 -3.56680930e-01
-8.33773375e-01 4.04216677e-01 -2.98099130e-01 -6.68564796e-01
5.48999496e-02 4.41112429e-01 -3.27026814e-01 -5.27685359e-02
4.30207431e-01 2.22013131e-01 -3.08520883e-01 -2.96749830e-01
2.22530812e-01 7.27021635e-01 7.91648448e-01 -6.09641038e-02
8.28203380e-01 2.67527878e-01 1.17370598e-02 -6.51170075e-01
-8.54846656e-01 -6.70590520e-01 -3.04131448e-01 -4.19211447e-01
9.55988228e-01 -1.14428151e+00 -7.15865195e-01 5.60775101e-01
-1.05107450e+00 -1.92310855e-01 3.24823588e-01 3.87097776e-01
-1.90028146e-01 2.56009221e-01 -4.32770580e-01 -5.23429453e-01
-3.18501264e-01 -1.38515604e+00 1.05085301e+00 3.95143151e-01
3.42305869e-01 -6.37217760e-01 -5.37742019e-01 2.34348878e-01
3.27810705e-01 4.61269729e-02 4.51769859e-01 -3.46605062e-01
-8.84240568e-01 -3.88081014e-01 -8.49031985e-01 1.19260199e-01
-1.92436293e-01 -2.20977724e-01 -8.97408247e-01 -2.24645317e-01
-5.81152320e-01 1.21327385e-01 1.14433408e+00 2.45084584e-01
1.06397867e+00 5.30351922e-02 -5.60249865e-01 6.01430893e-01
1.36440468e+00 1.51124686e-01 7.97233045e-01 5.86260438e-01
1.12925112e+00 2.31229469e-01 9.87045586e-01 3.04762155e-01
8.53137493e-01 9.04709876e-01 5.10933995e-01 -3.85596722e-01
-5.80687046e-01 -3.83827895e-01 5.20170391e-01 5.71357608e-01
5.50550282e-01 -5.65964937e-01 -7.67677367e-01 5.11109293e-01
-1.77252066e+00 -6.90914690e-01 -6.42481685e-01 2.03033972e+00
4.79496866e-01 8.86039853e-01 1.12934887e-01 7.04480708e-02
9.26765442e-01 3.44430774e-01 -6.35062575e-01 -6.44068271e-02
-1.39255702e-01 -6.87264681e-01 1.27594507e+00 5.81814051e-01
-1.19352710e+00 1.29826546e+00 4.81732750e+00 1.28885555e+00
-1.38176215e+00 1.67708546e-01 4.19818908e-01 1.73394039e-01
-3.68785299e-02 8.81640464e-02 -1.56253660e+00 6.79488659e-01
5.72213411e-01 2.70325422e-01 -8.83182511e-02 8.29610467e-01
5.98698080e-01 -1.67947203e-01 -7.95061111e-01 9.71802652e-01
6.39515221e-02 -1.09222841e+00 -1.76745161e-01 8.94534867e-03
4.36847955e-01 2.43278965e-01 5.10540940e-02 4.41383660e-01
-1.39791787e-01 -5.37472129e-01 1.17030823e+00 2.28925005e-01
1.00477242e+00 -7.64458239e-01 6.78674996e-01 5.05300581e-01
-1.74844718e+00 -1.76352158e-01 -3.19501549e-01 2.83847243e-01
5.06275475e-01 7.03011811e-01 -1.12423277e+00 3.06204259e-01
4.79233503e-01 7.02326775e-01 -6.94526076e-01 1.36292803e+00
-3.59940559e-01 6.80066824e-01 -4.45804745e-01 -7.60763064e-02
6.73936188e-01 2.48412088e-01 4.88212168e-01 1.53861010e+00
3.56111646e-01 -8.86193290e-02 5.65149665e-01 3.33733112e-01
-2.39314854e-01 -1.58752039e-01 -2.24189401e-01 6.88866317e-01
9.01455998e-01 1.21492672e+00 -9.19479728e-01 -1.64553210e-01
-4.55511451e-01 8.43621850e-01 1.50952399e-01 2.33839154e-01
-1.25220311e+00 -7.24090636e-01 3.39830071e-01 4.97261018e-01
4.42499131e-01 -2.53314734e-01 -2.11086243e-01 -6.53531790e-01
9.71594006e-02 -4.34375256e-01 2.20586479e-01 -7.65801609e-01
-9.85342026e-01 4.46866900e-01 -3.60653214e-02 -1.56957817e+00
3.24868917e-01 -5.25506914e-01 -5.46101749e-01 5.89858234e-01
-1.89868689e+00 -1.24930286e+00 -4.67849016e-01 2.79594332e-01
8.63665581e-01 1.22865535e-01 3.00267581e-02 7.98587382e-01
-9.86470819e-01 1.13102496e+00 3.39924432e-02 3.08710277e-01
1.03639925e+00 -9.16613579e-01 3.66805375e-01 1.03055882e+00
-4.07492965e-01 1.59923017e-01 6.73778951e-01 -5.34734547e-01
-1.11629152e+00 -1.43989670e+00 8.11555147e-01 5.49698845e-02
4.76171851e-01 -3.45593721e-01 -8.28783751e-01 5.88575423e-01
-1.05478249e-01 2.27158368e-01 1.11694150e-01 -2.79955298e-01
-3.10548842e-01 -4.62009311e-01 -9.64977086e-01 6.82710528e-01
1.26544929e+00 -3.53419334e-01 -1.33093223e-01 2.01394573e-01
6.63924158e-01 -5.88558555e-01 -3.95662785e-01 5.90426028e-01
5.18320203e-01 -8.91713738e-01 9.59294319e-01 3.46863061e-01
1.00351758e-01 -9.27844703e-01 -1.30513221e-01 -8.37894082e-01
-4.77915913e-01 -3.53734076e-01 1.21857643e-01 1.25736809e+00
6.46460533e-01 -7.19825804e-01 9.74268794e-01 1.79407761e-01
-7.54070878e-01 -1.00222838e+00 -8.98796678e-01 -9.63600159e-01
-2.07619905e-01 -4.81756449e-01 3.54977131e-01 6.64642572e-01
-4.89988595e-01 5.67330837e-01 -5.08026958e-01 5.53931057e-01
5.59488058e-01 -3.18233967e-01 6.40130103e-01 -1.09525836e+00
-5.55116162e-02 -5.39367855e-01 -4.96358961e-01 -1.87724936e+00
9.30316281e-03 -7.75609672e-01 4.42384183e-01 -1.74046254e+00
-7.19500519e-03 -7.73462057e-01 -3.50299120e-01 7.14985609e-01
-9.23840702e-02 4.54516798e-01 2.33860463e-01 1.54083520e-01
-9.58587527e-01 4.70558316e-01 1.21700847e+00 -1.71406373e-01
-1.77128047e-01 -3.31755936e-01 -4.75365102e-01 8.30896437e-01
8.76500070e-01 -4.44018424e-01 -5.24655223e-01 -6.44861281e-01
1.98633000e-01 -8.15063342e-03 3.61593723e-01 -1.27789950e+00
6.78260028e-01 1.54350281e-01 3.13694537e-01 -1.30664611e+00
2.81180501e-01 -7.49157727e-01 -2.58321136e-01 3.36230665e-01
-1.60274029e-01 2.09235996e-01 2.53527373e-01 6.98454499e-01
-1.05304234e-01 -2.23420396e-01 8.13649595e-01 3.27327102e-01
-1.41089535e+00 2.51085043e-01 -4.20184940e-01 1.56730525e-02
1.17732692e+00 -6.10629499e-01 -6.28646493e-01 -1.69345319e-01
-2.93820679e-01 8.94816518e-01 2.01222375e-01 4.43494916e-01
6.62275612e-01 -1.33257389e+00 -5.71260273e-01 1.92682017e-02
3.40971738e-01 1.59586981e-01 3.08336645e-01 1.08401918e+00
-6.75529480e-01 5.38186550e-01 1.05693474e-01 -6.81473017e-01
-1.31372678e+00 3.55183780e-01 4.68881071e-01 -1.70155421e-01
-6.23881400e-01 1.04252326e+00 3.13019633e-01 -2.02771395e-01
3.82770836e-01 -2.37211615e-01 -9.61982310e-02 -7.75591880e-02
2.99068153e-01 3.82431030e-01 -5.47825024e-02 -8.35258842e-01
-5.41065574e-01 7.71895051e-01 -4.11554635e-01 5.59984297e-02
6.21571362e-01 -3.65324676e-01 3.32888126e-01 2.31040254e-01
1.21555901e+00 -3.60570811e-02 -1.51045609e+00 -1.40441284e-01
-9.08255279e-02 -4.35658723e-01 4.27344859e-01 -6.73142374e-01
-1.17694569e+00 9.63996530e-01 8.65233839e-01 -2.35895306e-01
1.17537034e+00 -3.60867083e-01 1.21998107e+00 3.54486167e-01
4.53897864e-01 -1.07878363e+00 7.15286881e-02 6.42853677e-01
4.18545872e-01 -1.24013817e+00 -1.66926786e-01 -8.18487823e-01
-4.45983320e-01 9.43664014e-01 9.84777927e-01 1.10672846e-01
5.10328412e-01 3.04965794e-01 1.18824124e-01 1.86855029e-02
-6.02127016e-01 -3.10091704e-01 2.90207714e-01 1.75948232e-01
2.36931473e-01 -1.23614296e-02 -3.83530855e-01 2.76022971e-01
1.12732813e-01 -1.89719185e-01 3.82953763e-01 8.53090227e-01
-8.94979596e-01 -1.06973469e+00 -2.60398239e-01 4.74464595e-01
-8.35924745e-02 -8.63184556e-02 8.58843792e-03 6.89742267e-01
5.06979823e-01 9.23678339e-01 -1.08991541e-01 -7.81456113e-01
5.83209395e-01 -2.17398539e-01 -7.21593127e-02 -2.90995717e-01
-2.27832288e-01 3.39617223e-01 2.39111304e-01 -4.87936854e-01
6.84224218e-02 -5.03340304e-01 -1.69735706e+00 -3.12670887e-01
-7.14113176e-01 -2.99814254e-01 5.56375206e-01 8.40861797e-01
4.97000575e-01 7.76103258e-01 6.23663485e-01 -6.64146006e-01
-3.04300249e-01 -8.11594248e-01 -2.76721090e-01 -2.26359904e-01
3.22737783e-01 -8.32838476e-01 -1.40492633e-01 -2.77482718e-01]
|
[8.114404678344727, -1.3732094764709473]
|
ea7543ad-f71e-405e-b0cc-f9d01a1b6ca1
|
how-does-generative-retrieval-scale-to
|
2305.11841
| null |
https://arxiv.org/abs/2305.11841v1
|
https://arxiv.org/pdf/2305.11841v1.pdf
|
How Does Generative Retrieval Scale to Millions of Passages?
|
Popularized by the Differentiable Search Index, the emerging paradigm of generative retrieval re-frames the classic information retrieval problem into a sequence-to-sequence modeling task, forgoing external indices and encoding an entire document corpus within a single Transformer. Although many different approaches have been proposed to improve the effectiveness of generative retrieval, they have only been evaluated on document corpora on the order of 100k in size. We conduct the first empirical study of generative retrieval techniques across various corpus scales, ultimately scaling up to the entire MS MARCO passage ranking task with a corpus of 8.8M passages and evaluating model sizes up to 11B parameters. We uncover several findings about scaling generative retrieval to millions of passages; notably, the central importance of using synthetic queries as document representations during indexing, the ineffectiveness of existing proposed architecture modifications when accounting for compute cost, and the limits of naively scaling model parameters with respect to retrieval performance. While we find that generative retrieval is competitive with state-of-the-art dual encoders on small corpora, scaling to millions of passages remains an important and unsolved challenge. We believe these findings will be valuable for the community to clarify the current state of generative retrieval, highlight the unique challenges, and inspire new research directions.
|
['Vinh Q. Tran', 'Donald Metzler', 'Jimmy Lin', 'Honglei Zhuang', 'Adam D. Lelkes', 'Jai Gupta', 'Kai Hui', 'Ronak Pradeep']
|
2023-05-19
| null | null | null | null |
['passage-ranking']
|
['natural-language-processing']
|
[ 3.64041984e-01 -2.22471088e-01 -4.47989069e-02 -1.60998851e-02
-1.68277323e+00 -7.82955587e-01 1.09623957e+00 7.19748586e-02
-4.16588902e-01 7.40062177e-01 5.03221214e-01 -3.94533783e-01
-4.64434415e-01 -5.78219056e-01 -7.02692866e-01 -5.40605307e-01
-1.13081131e-02 8.46663296e-01 1.59780398e-01 -4.96722609e-01
6.49476171e-01 2.11785093e-01 -1.52494347e+00 2.92927057e-01
5.10900199e-01 6.80345058e-01 1.64691299e-01 9.71005023e-01
2.36134361e-02 6.06785417e-01 -9.80956316e-01 -5.17823279e-01
2.61747360e-01 -3.43926698e-01 -1.04488969e+00 -1.77464351e-01
5.31806767e-01 -7.07430243e-01 -7.54730701e-01 7.84900784e-01
8.63622606e-01 7.52448142e-02 6.34313822e-01 -1.01136589e+00
-1.33562899e+00 8.06083024e-01 -2.00830147e-01 5.82967281e-01
2.87557542e-01 -8.84449184e-02 1.29347539e+00 -8.79259825e-01
6.91774547e-01 1.22920561e+00 2.21762359e-01 3.72485787e-01
-9.66588080e-01 -4.16326433e-01 -1.26624465e-01 2.56955475e-01
-1.58801675e+00 -7.67342567e-01 2.17027083e-01 -8.68871361e-02
1.40598702e+00 5.72519183e-01 3.07877690e-01 1.18388438e+00
7.84496404e-03 8.87943447e-01 5.78211248e-01 -5.47461987e-01
-4.00767475e-02 -1.47517487e-01 2.38264322e-01 3.39324325e-01
4.42434520e-01 -5.21317795e-02 -7.09302366e-01 -4.20722663e-01
5.34736454e-01 -6.85056150e-02 -3.30931216e-01 1.43821493e-01
-1.09532964e+00 1.01861155e+00 4.50867563e-02 2.74883568e-01
-2.03242093e-01 4.59126830e-01 5.17981291e-01 5.19846439e-01
3.89338136e-01 6.78143680e-01 -3.93430777e-02 -4.86931026e-01
-1.25468123e+00 5.85577190e-01 8.59675467e-01 1.23271322e+00
4.45651114e-01 -1.11532189e-01 -3.59016299e-01 7.88129687e-01
1.70044973e-01 5.62220752e-01 7.89221406e-01 -8.82704377e-01
4.85418409e-01 2.35716209e-01 -7.99184944e-03 -7.27753997e-01
7.54841566e-02 -6.30052447e-01 -3.99710268e-01 -6.36456251e-01
2.05728784e-02 7.97464550e-02 -7.73462713e-01 1.42768466e+00
-1.97525054e-01 -2.22226709e-01 1.35643214e-01 8.35991383e-01
5.16571939e-01 8.84495556e-01 -2.45963469e-01 -6.08080178e-02
1.25049841e+00 -1.09029639e+00 -4.96153504e-01 -2.00876027e-01
6.35983884e-01 -1.22341967e+00 1.03282213e+00 1.12606339e-01
-1.49617660e+00 -3.09004575e-01 -9.94675994e-01 -4.12513494e-01
-2.77182013e-01 -3.47554162e-02 5.97373724e-01 5.08736968e-01
-1.45988417e+00 3.10905457e-01 -7.74241328e-01 -3.86146396e-01
-5.66847734e-02 1.45656824e-01 1.27683505e-01 -3.10418248e-01
-1.32260466e+00 7.29833126e-01 3.22934359e-01 -6.97716177e-02
-8.52678955e-01 -7.21805215e-01 -3.09991747e-01 3.43030125e-01
2.46570721e-01 -8.88953567e-01 1.30227053e+00 -2.06759810e-01
-1.05822098e+00 4.98881906e-01 -2.19055265e-01 -6.01140440e-01
2.75943071e-01 -4.78195250e-01 -4.15893823e-01 3.86421263e-01
2.79879887e-02 4.87897545e-01 5.51317573e-01 -1.07307363e+00
-3.40881407e-01 -1.82813838e-01 1.92498043e-01 4.11353171e-01
-5.67806363e-01 8.87407362e-02 -9.41922247e-01 -7.48070121e-01
2.20367946e-02 -1.09730577e+00 3.22751775e-02 -6.70190573e-01
-2.77765900e-01 -1.41527236e-01 6.63529098e-01 -5.63727796e-01
1.60518348e+00 -2.17786765e+00 7.58690909e-02 4.11474705e-02
1.68412104e-02 2.86008924e-01 -4.69828665e-01 1.07097554e+00
3.66189808e-01 4.09537226e-01 1.72184139e-01 -3.28626364e-01
3.36346030e-01 -4.21384014e-02 -8.15060556e-01 1.01541430e-01
-8.85737911e-02 1.39393568e+00 -7.90630400e-01 -4.93010551e-01
-1.80254504e-01 7.12343514e-01 -5.46587944e-01 1.36527807e-01
-2.19471127e-01 -2.12119386e-01 -6.01305187e-01 5.76571763e-01
1.97331473e-01 -6.97257757e-01 6.88558891e-02 8.45848843e-02
1.97789297e-01 8.20960402e-01 -7.81838894e-01 1.75982940e+00
-3.94861788e-01 7.43633330e-01 -3.34508091e-01 -5.53481579e-01
6.97848260e-01 3.35309327e-01 2.95979083e-01 -9.29886699e-01
-2.34074593e-01 4.20956135e-01 -2.32582629e-01 -2.90705979e-01
1.36550260e+00 4.52377349e-02 -1.55013248e-01 8.97639155e-01
9.38675553e-03 -1.57172345e-02 6.43336356e-01 6.13319933e-01
1.33656871e+00 -1.05032027e-01 -2.29449645e-01 -3.47898118e-02
-7.08165299e-03 2.26127300e-02 -3.03637862e-01 1.19189501e+00
3.73105288e-01 7.12630153e-01 1.31690979e-01 -7.26438239e-02
-1.21670890e+00 -9.87425447e-01 -1.35900557e-01 1.38519585e+00
8.62335786e-02 -7.50032961e-01 -6.58226252e-01 -1.19357839e-01
-5.04281037e-02 5.87816358e-01 -3.53075355e-01 -5.20965576e-01
-7.57048488e-01 -1.12630832e+00 9.98301387e-01 5.38263977e-01
2.20886484e-01 -8.78353059e-01 -5.68508923e-01 3.48814428e-01
-3.64803314e-01 -9.76928771e-01 -7.41688490e-01 3.90491597e-02
-8.30589056e-01 -6.90652311e-01 -9.65442717e-01 -6.32811964e-01
2.78997511e-01 5.04634082e-01 1.49873340e+00 3.82667065e-01
-3.87521535e-01 6.77993178e-01 -6.40962124e-01 -4.09559049e-02
-6.25185668e-01 6.30848169e-01 -2.70799160e-01 -6.07146144e-01
3.43910277e-01 -5.05183339e-01 -8.70545566e-01 1.81058068e-02
-1.46329200e+00 -3.08248907e-01 8.31951797e-01 7.90986359e-01
3.89100105e-01 -2.83886433e-01 6.47867382e-01 -8.05859506e-01
1.22696126e+00 -5.44869304e-01 -2.97505081e-01 6.04549348e-01
-9.93709028e-01 3.73481482e-01 2.83306807e-01 -3.75368327e-01
-7.75031924e-01 -7.42271006e-01 -7.10397437e-02 -3.92877251e-01
3.35283577e-01 6.23531997e-01 5.89995027e-01 2.15220600e-01
6.31325364e-01 6.51480973e-01 -1.74664631e-01 -4.65253383e-01
5.80741167e-01 6.90025687e-01 4.27085787e-01 -6.63657963e-01
6.58104837e-01 1.24154352e-01 -1.42484292e-01 -5.90499520e-01
-6.31685555e-01 -6.26865625e-01 -1.71591863e-01 2.17531919e-01
2.63753623e-01 -9.58885610e-01 -3.67234915e-01 4.47787493e-02
-9.99610662e-01 -4.74980585e-02 -2.10046977e-01 1.24128312e-01
-4.61930633e-01 6.50045097e-01 -1.04064953e+00 -5.35831988e-01
-8.76168013e-01 -1.15196788e+00 1.56397259e+00 -1.91241607e-01
-3.10662895e-01 -8.69768739e-01 3.18021208e-01 4.90847230e-01
8.75295877e-01 -4.65511173e-01 9.75859761e-01 -6.48803830e-01
-8.82851779e-01 -3.93138677e-01 -1.60951585e-01 7.35128820e-02
-1.37729630e-01 -1.00239389e-01 -7.57288098e-01 -5.92288435e-01
-3.36651087e-01 -6.16927743e-01 9.85647261e-01 -5.11019714e-02
8.64704013e-01 -3.13948333e-01 -2.79069215e-01 3.80185574e-01
1.45066476e+00 3.75282496e-01 7.29129136e-01 5.54090559e-01
2.54460156e-01 -2.18838006e-02 3.04379433e-01 4.96700823e-01
2.78282940e-01 7.58260429e-01 1.30045295e-01 3.28228921e-01
-3.98765028e-01 -3.97883534e-01 2.31063202e-01 1.24063683e+00
7.33905882e-02 -8.32413852e-01 -7.63304055e-01 6.19201064e-01
-1.56147552e+00 -9.76167262e-01 3.93658340e-01 2.17816877e+00
9.76875186e-01 1.55542791e-01 -1.55400619e-01 1.08472416e-02
4.28003281e-01 2.48693034e-01 -3.62252057e-01 -2.64001161e-01
-2.08770201e-01 4.96914148e-01 4.72828835e-01 5.15953302e-01
-7.00710654e-01 8.66835415e-01 7.55143929e+00 9.36288714e-01
-1.26279926e+00 -1.01932213e-02 4.81309950e-01 -6.10008657e-01
-6.39767826e-01 1.23532720e-01 -1.18265522e+00 4.58464086e-01
1.41047180e+00 -4.78613377e-01 7.10348725e-01 6.06895149e-01
-3.67563844e-01 1.92200348e-01 -1.22155905e+00 7.94984102e-01
4.20123279e-01 -1.42201626e+00 5.10839939e-01 2.11263701e-01
5.83701968e-01 2.80198663e-01 2.83855915e-01 5.51591158e-01
1.29239514e-01 -1.14822745e+00 7.65612304e-01 5.49505591e-01
7.28439093e-01 -4.39623207e-01 5.18615544e-01 1.92912474e-01
-8.60514343e-01 1.96022630e-01 -4.94026184e-01 1.23781912e-01
3.24745685e-01 3.12150329e-01 -9.10693347e-01 4.68531311e-01
4.75613266e-01 2.32543334e-01 -8.45697224e-01 9.00695562e-01
3.41372788e-01 6.33000791e-01 -4.37975556e-01 -2.81837404e-01
4.31874514e-01 6.73806742e-02 3.35188597e-01 1.38208711e+00
5.75117946e-01 -2.76921362e-01 -1.69155255e-01 5.92922986e-01
-2.82832414e-01 -8.02354515e-02 -4.47888464e-01 -4.27373499e-01
7.39085853e-01 8.93709600e-01 -4.78141814e-01 -5.11590540e-01
-1.95760980e-01 9.72723603e-01 3.36914659e-01 5.42809248e-01
-7.15475738e-01 -2.86496967e-01 2.81538844e-01 9.24184024e-02
3.61089975e-01 -3.00046086e-01 8.82712454e-02 -1.02437937e+00
3.77523988e-01 -9.70529139e-01 2.83837080e-01 -8.15594256e-01
-1.04061556e+00 8.17841291e-01 2.02540144e-01 -9.66722548e-01
-8.92343640e-01 -1.01828083e-01 9.72484425e-03 8.72118652e-01
-1.54726624e+00 -7.93011129e-01 1.12998830e-02 2.47757807e-01
6.66845918e-01 -7.85538182e-02 1.09732521e+00 6.23677433e-01
-2.27696270e-01 7.97272682e-01 6.49912596e-01 -1.64032832e-01
6.99054480e-01 -9.15718019e-01 6.39674425e-01 6.61511600e-01
5.24472415e-01 1.15024137e+00 5.44543207e-01 -2.84199119e-01
-1.82356930e+00 -7.44051874e-01 1.06398177e+00 -6.48070574e-01
6.69428170e-01 -2.77743578e-01 -8.89254391e-01 6.91453874e-01
4.15579349e-01 -2.87760556e-01 6.15813494e-01 -3.10573243e-02
-4.78135884e-01 1.05959907e-01 -5.17910242e-01 7.41410196e-01
1.09129047e+00 -7.34720111e-01 -4.90367502e-01 4.53345656e-01
9.71837819e-01 -5.34964323e-01 -1.07908750e+00 1.90875024e-01
6.98714018e-01 -5.91047406e-01 1.18187773e+00 -5.64709008e-01
5.33526540e-01 1.63732469e-01 -3.81201178e-01 -9.77867186e-01
-3.27489406e-01 -7.18921840e-01 -3.78779858e-01 1.15040159e+00
4.93570507e-01 -3.50433230e-01 6.26176417e-01 5.53749263e-01
-2.26267442e-01 -9.25030649e-01 -8.45438063e-01 -6.99667752e-01
2.89223760e-01 -2.35476896e-01 7.16770649e-01 4.61344033e-01
-2.18476281e-01 6.20032907e-01 -3.11399072e-01 -1.98308095e-01
2.99294919e-01 3.16104203e-01 6.58573091e-01 -7.20463872e-01
-5.89735866e-01 -5.65650225e-01 -7.44834989e-02 -1.50450885e+00
-3.14345881e-02 -9.44950402e-01 -1.71509027e-01 -1.38830853e+00
5.46072364e-01 -3.90467554e-01 -2.76611120e-01 1.73306450e-01
-2.11601049e-01 4.39197868e-01 2.42248774e-01 5.65837204e-01
-8.90470207e-01 5.38681388e-01 1.13234937e+00 -2.84965158e-01
2.81819850e-01 -3.59872311e-01 -1.03766525e+00 -9.68061313e-02
5.81459403e-01 -2.70572543e-01 -7.91209519e-01 -9.02611315e-01
5.61527610e-01 1.80684745e-01 2.67296076e-01 -7.74058759e-01
3.77411395e-01 1.65570363e-01 1.77508332e-02 -5.64123631e-01
5.52673757e-01 -3.82164091e-01 1.81415588e-01 1.60879597e-01
-6.88471138e-01 5.85309803e-01 2.01981232e-01 4.47773397e-01
-4.45118368e-01 -4.45428133e-01 8.05211887e-02 -1.26866430e-01
-5.02097130e-01 5.72485253e-02 -1.59692630e-01 4.73969311e-01
4.31790978e-01 -1.56067265e-02 -7.59405851e-01 -5.79156518e-01
-2.52703488e-01 -8.71842802e-02 4.06940937e-01 6.66834533e-01
5.72289705e-01 -1.14374518e+00 -7.92731285e-01 8.22036341e-02
1.12220980e-01 -3.18595886e-01 1.34577127e-02 3.31414551e-01
-4.95305091e-01 1.20531356e+00 3.25944006e-01 -3.93114775e-01
-1.13993156e+00 4.02155071e-01 -8.99085030e-02 -6.20849967e-01
-6.09011829e-01 7.27070808e-01 -6.00726437e-03 2.02375412e-01
2.92646021e-01 6.89647719e-02 2.11307600e-01 -1.07005008e-01
5.60858011e-01 4.61713225e-01 5.51660597e-01 -2.94644207e-01
-8.34444612e-02 3.34084272e-01 -5.63091338e-01 -3.43995541e-01
1.23750293e+00 -2.18811616e-01 -7.26140067e-02 1.32309258e-01
1.51125956e+00 -5.55059053e-02 -6.23808086e-01 -2.44139373e-01
-1.11992441e-01 -2.62029976e-01 5.87383285e-02 -8.44604850e-01
-8.01318705e-01 6.25424802e-01 3.42502713e-01 3.08621556e-01
1.12610531e+00 1.21201962e-01 1.08127820e+00 7.62109518e-01
5.57463825e-01 -7.80775785e-01 2.51844287e-01 6.87587559e-01
8.43370855e-01 -8.15209687e-01 1.76350668e-01 6.55069351e-02
-2.73936957e-01 8.85182858e-01 2.54455376e-02 4.05331096e-03
3.13731223e-01 2.87686735e-01 -7.30737597e-02 -4.17471826e-01
-1.15593588e+00 1.18535258e-01 3.57740968e-01 -2.65721772e-02
8.13289940e-01 -1.49712011e-01 -4.59590435e-01 1.11916892e-01
-5.31174481e-01 -3.70790847e-02 2.83751905e-01 1.11321223e+00
-3.00014555e-01 -1.29625428e+00 -2.49601737e-01 6.88215196e-01
-7.14320302e-01 -7.67926693e-01 -1.43626913e-01 7.52680302e-01
-6.94592535e-01 7.95861065e-01 1.25920922e-01 -2.40643695e-01
1.96510423e-02 3.72642308e-01 5.25789320e-01 -4.67525810e-01
-5.96274137e-01 2.26337284e-01 1.04529560e-01 -3.17146093e-01
-3.29458237e-01 -6.70471787e-01 -7.49845982e-01 -3.99927706e-01
-3.31645846e-01 4.96627241e-01 6.89262331e-01 7.33534396e-01
7.98943400e-01 4.67704475e-01 3.15555155e-01 -4.97078031e-01
-1.09436858e+00 -1.28450406e+00 -3.35761607e-01 3.74659210e-01
2.62907237e-01 -3.67871314e-01 -3.95185828e-01 7.29832277e-02]
|
[11.461040496826172, 7.6132378578186035]
|
8ea15346-24ed-48d5-aabe-a13cbaca3b01
|
neural-attribution-for-semantic-bug
| null | null |
http://papers.nips.cc/paper/9358-neural-attribution-for-semantic-bug-localization-in-student-programs
|
http://papers.nips.cc/paper/9358-neural-attribution-for-semantic-bug-localization-in-student-programs.pdf
|
Neural Attribution for Semantic Bug-Localization in Student Programs
|
Providing feedback is an integral part of teaching. Most open online courses on programming make use of automated grading systems to support programming assignments and give real-time feedback. These systems usually rely on test results to quantify the programs' functional correctness. They return failing tests to the students as feedback. However, students may find it difficult to debug their programs if they receive no hints about where the bug is and how to fix it. In this work, we present NeuralBugLocator, a deep learning based technique, that can localize the bugs in a faulty program with respect to a failing test, without even running the program. At the heart of our technique is a novel tree convolutional neural network which is trained to predict whether a program passes or fails a given test. To localize the bugs, we analyze the trained network using a state-of-the-art neural prediction attribution technique and see which lines of the programs make it predict the test outcomes. Our experiments show that NeuralBugLocator is generally more accurate than two state-of-the-art program-spectrum based and one syntactic difference based bug-localization baselines.
|
['Aditya Kanade', 'Rahul Gupta', 'Shirish Shevade']
|
2019-12-01
| null | null | null |
neurips-2019-12
|
['fault-localization']
|
['computer-code']
|
[-2.67079651e-01 -2.51216181e-02 -4.13294494e-01 -5.82117379e-01
-7.30533957e-01 -7.09360540e-01 -3.52700561e-01 7.41224170e-01
2.07384571e-01 2.48579860e-01 -4.00430501e-01 -1.05161238e+00
8.64948556e-02 -1.15401888e+00 -1.18164361e+00 -8.21924303e-03
1.21954679e-02 1.43679976e-01 5.85011244e-01 -3.97606909e-01
5.40867805e-01 4.30028811e-02 -1.62933218e+00 7.32156038e-01
1.08387959e+00 6.07709825e-01 -6.61138967e-02 1.05519092e+00
-1.51305765e-01 1.36734700e+00 -1.11950040e+00 -2.45657131e-01
-1.87319070e-01 -3.03772062e-01 -1.06089926e+00 -5.37202477e-01
9.26853001e-01 -5.58975577e-01 2.27322392e-02 1.36060619e+00
-2.74276417e-02 -1.76566169e-01 -5.20485900e-02 -1.46013010e+00
-9.17417824e-01 8.83867145e-01 -4.28776115e-01 4.07670200e-01
6.88624084e-01 8.51026997e-02 1.16130912e+00 -3.27826649e-01
3.32558006e-01 7.08608389e-01 9.46101248e-01 4.50286359e-01
-9.98191297e-01 -4.93101835e-01 -7.11163282e-02 3.07618469e-01
-8.23008299e-01 1.65791690e-01 8.02531004e-01 -8.20359528e-01
1.02246487e+00 5.60936779e-02 6.11904263e-01 7.03559101e-01
7.03262091e-01 5.85642219e-01 8.14823747e-01 -7.33941197e-01
1.24358229e-01 6.73588961e-02 8.23109150e-01 1.60159540e+00
1.98440179e-01 -9.24985260e-02 -3.98706257e-01 -2.49035016e-01
3.38949770e-01 1.87994063e-01 -3.98863465e-01 -3.11126471e-01
-8.01682711e-01 7.00951695e-01 6.42167568e-01 2.92758197e-01
2.29068324e-01 5.33514559e-01 3.94614249e-01 8.65927696e-01
1.61754057e-01 7.41547167e-01 -8.96124065e-01 -4.73929554e-01
-9.93763983e-01 3.87972176e-01 9.30290878e-01 4.86426681e-01
7.34420836e-01 2.83615254e-02 -1.28465533e-01 4.41769928e-01
5.18783689e-01 -3.69049013e-01 7.45661497e-01 -7.71380067e-01
4.41875994e-01 1.35405326e+00 -3.57331604e-01 -7.82479405e-01
4.71488535e-02 -3.93555731e-01 1.74441248e-01 9.26598608e-01
4.80901599e-01 -7.41336541e-03 -8.03557932e-01 1.55334949e+00
-6.37580641e-03 3.45903903e-01 -3.69833887e-01 5.41923761e-01
9.40963566e-01 3.75283211e-01 -3.61623526e-01 6.01683438e-01
1.06500256e+00 -1.21648431e+00 -1.49560377e-01 -5.24327196e-02
1.43660009e+00 -7.39178896e-01 1.38524592e+00 7.35854983e-01
-9.26930726e-01 -6.62101388e-01 -1.42175615e+00 -4.85192984e-02
-4.64305341e-01 1.53124154e-01 4.02003109e-01 9.36695337e-01
-1.34204662e+00 1.25054693e+00 -1.04263949e+00 9.78559032e-02
9.25782844e-02 2.11043611e-01 -3.30605686e-01 -1.03288770e-01
-6.48584127e-01 8.44950199e-01 1.10214517e-01 -4.80201989e-01
-1.04579818e+00 -1.15132463e+00 -1.11888337e+00 6.21392906e-01
9.01656672e-02 -1.92256629e-01 1.99720478e+00 -8.02054286e-01
-1.23810363e+00 8.38376820e-01 8.56762230e-02 -2.42697205e-02
1.93304390e-01 -2.86644101e-01 2.87994999e-03 -4.27018493e-01
1.78282514e-01 2.78170615e-01 3.91956389e-01 -6.60986841e-01
-5.47682345e-01 -2.13094562e-01 4.02191877e-01 -5.40608883e-01
-2.83002079e-01 5.19445390e-02 2.48889858e-03 -1.32113084e-01
1.27680957e-01 -4.61234331e-01 5.09612076e-02 -1.29261985e-01
-2.82865167e-01 -6.60847723e-01 9.63549137e-01 -6.35717630e-01
1.53880441e+00 -1.90519023e+00 -2.51921833e-01 1.33702502e-01
5.45456946e-01 2.16429710e-01 -1.96868896e-01 1.60610706e-01
-6.07959688e-01 4.26744252e-01 1.64225638e-01 7.35292137e-02
1.52725920e-01 -2.14097664e-01 -4.49219495e-01 3.13066572e-01
3.29060078e-01 4.65936035e-01 -9.54464436e-01 -4.65169027e-02
-2.84766816e-02 -4.85462472e-02 -8.98761213e-01 4.40655410e-01
-5.24323940e-01 -6.03735335e-02 -2.35890612e-01 1.01480019e+00
3.21862012e-01 -1.43982202e-01 -2.52774060e-01 5.86106777e-01
-2.00059339e-01 9.54770744e-01 -7.38141596e-01 1.38617599e+00
-4.64960307e-01 1.01665950e+00 -2.36241192e-01 -9.79844213e-01
1.02534246e+00 3.50309998e-01 -1.82976946e-01 -4.57002938e-01
-1.05873793e-01 4.28266555e-01 3.47408682e-01 -6.31015718e-01
4.94939178e-01 5.32856703e-01 -1.83279812e-01 8.59295070e-01
3.19221765e-01 -1.10349603e-01 3.78963858e-01 1.76788181e-01
1.99430156e+00 5.57918370e-01 7.79046938e-02 -3.71046513e-02
2.83053815e-01 -1.25214932e-02 5.94239235e-01 8.40615392e-01
-4.10578340e-01 4.62050855e-01 1.20978940e+00 -7.28542149e-01
-7.93396771e-01 -6.18691623e-01 1.91068277e-01 1.48624182e+00
-5.58001518e-01 -7.34043181e-01 -9.09864843e-01 -1.03666830e+00
1.32922307e-01 6.23899877e-01 -7.34149516e-01 -4.69817787e-01
-4.23402607e-01 6.34890571e-02 5.43132901e-01 6.15698159e-01
2.12805811e-02 -9.90318298e-01 -8.47431540e-01 1.10126697e-01
1.70942575e-01 -2.50332773e-01 -3.94041628e-01 5.52566051e-01
-6.78453147e-01 -1.37153864e+00 4.54658009e-02 -9.43872690e-01
8.56444299e-01 1.53453931e-01 1.55207610e+00 1.10438526e+00
-2.99961746e-01 1.41524568e-01 -3.53392690e-01 -3.09569776e-01
-6.40453875e-01 5.45434796e-06 -3.68545651e-01 -8.79281938e-01
6.94345832e-01 -5.23021102e-01 -3.75611819e-02 3.92294526e-02
-5.36967993e-01 -2.98228979e-01 2.19402537e-01 6.82367980e-01
1.94939703e-01 2.63854951e-01 5.71370199e-02 -1.11358500e+00
8.76531661e-01 -5.35394788e-01 -8.89847040e-01 5.01559556e-01
-6.55694902e-01 3.34817380e-01 7.23521292e-01 -1.90558791e-01
-2.23716825e-01 -2.20477685e-01 -2.93089688e-01 -4.08221662e-01
-1.89088643e-01 1.02529621e+00 1.99349597e-01 -5.53302526e-01
1.00162327e+00 -1.08881518e-01 -3.73338073e-01 -3.41781348e-01
-3.24729860e-01 3.44184965e-01 4.95343983e-01 -9.10622776e-01
5.42442143e-01 -6.36785686e-01 -3.43590677e-01 -3.52942497e-02
-9.15349483e-01 -3.44137639e-01 -4.15923208e-01 -3.92407477e-01
3.76050383e-01 -6.06243074e-01 -1.02259290e+00 3.55768532e-01
-1.37270141e+00 -6.89774632e-01 1.58529952e-01 8.60780571e-03
-2.11694956e-01 7.41263926e-02 -8.45184684e-01 -5.54093421e-01
1.60087105e-02 -1.64294863e+00 6.13500953e-01 3.70677501e-01
-4.59096193e-01 -8.99697006e-01 5.49750268e-01 4.77173507e-01
4.11959529e-01 1.57624409e-01 1.31555474e+00 -8.17710757e-01
-7.11304843e-01 -5.02032399e-01 1.86859995e-01 3.75605017e-01
-6.32008314e-02 7.21794486e-01 -8.84784758e-01 -1.28814317e-02
-8.39671418e-02 -5.64517379e-01 7.47521520e-01 1.01619691e-01
1.36974430e+00 -2.07655102e-01 -1.45085514e-01 5.61567724e-01
1.41445386e+00 8.67895856e-02 4.44261253e-01 6.06652856e-01
6.23888791e-01 5.07026792e-01 2.49731213e-01 -4.17593829e-02
3.38408619e-01 3.00124139e-01 7.28824496e-01 4.06328440e-01
1.93877205e-01 -4.43463564e-01 8.64852130e-01 8.02499413e-01
3.71308923e-01 -1.36356317e-02 -1.42032731e+00 6.16261125e-01
-1.66450810e+00 -7.12059379e-01 -5.83806634e-01 2.07957196e+00
1.08349681e+00 4.20391411e-01 -1.73827082e-01 5.31491339e-01
4.68084276e-01 -2.73276150e-01 -1.84418574e-01 -1.04163837e+00
8.13808620e-01 6.46032035e-01 -1.63332198e-03 5.43324113e-01
-9.76547182e-01 6.49699450e-01 5.82212305e+00 2.31078148e-01
-1.40099239e+00 1.28085807e-01 5.36580145e-01 2.66032457e-01
-3.23286831e-01 2.54553333e-02 -8.00221443e-01 2.18287557e-01
1.34309626e+00 1.12117715e-01 2.07725197e-01 1.51709855e+00
-2.16187328e-01 -2.01485828e-01 -1.63140607e+00 2.07329512e-01
1.77051634e-01 -1.37688696e+00 -6.36934757e-01 -2.50032842e-01
7.16426611e-01 -4.88177827e-03 -2.45036297e-02 9.33556676e-01
7.45850027e-01 -1.15041649e+00 8.65398049e-01 2.57332176e-01
1.39043987e-01 -7.30167568e-01 9.54137862e-01 3.94411296e-01
-7.11045265e-01 -9.32274684e-02 -3.30321729e-01 -6.27406180e-01
-8.68472993e-01 5.16884744e-01 -1.16535926e+00 -7.70293474e-02
1.06490016e+00 3.85896921e-01 -1.11243427e+00 1.33284557e+00
-9.08580542e-01 9.74153280e-01 1.99704170e-01 -3.03086191e-01
5.39575890e-02 2.34568909e-01 4.69501168e-02 8.64227533e-01
6.60275280e-01 -3.99914443e-01 5.75679660e-01 1.42305744e+00
-2.30554774e-01 -2.27925748e-01 -6.38838053e-01 -2.71093011e-01
2.92043388e-01 1.13382411e+00 -5.82687497e-01 -4.19238321e-02
-3.60799223e-01 5.40974796e-01 8.09874415e-01 -6.39101267e-02
-5.95183015e-01 -8.38154972e-01 5.77036500e-01 1.25797227e-01
2.63307482e-01 -4.45490354e-04 -7.22153127e-01 -9.62078035e-01
3.45399410e-01 -9.91250396e-01 1.38020694e-01 -1.09127688e+00
-7.31213868e-01 2.96962798e-01 -6.96673810e-01 -8.81918848e-01
-1.61070406e-01 -8.68273973e-01 -1.54660177e+00 9.24628496e-01
-1.30166650e+00 -5.62427163e-01 -4.08963978e-01 6.82293847e-02
1.57088906e-01 -2.31199697e-01 8.86847436e-01 1.32546619e-01
-7.64783442e-01 9.24739420e-01 -3.40591550e-01 4.14873838e-01
7.63490736e-01 -1.81621110e+00 3.55327010e-01 1.15367341e+00
4.00583353e-03 9.52923179e-01 6.62727773e-01 -6.73417091e-01
-1.36809337e+00 -1.04141045e+00 1.14571130e+00 -7.35570788e-01
9.75274086e-01 -5.25839217e-02 -1.35121131e+00 9.27480936e-01
1.90385163e-01 2.28493840e-01 8.38235557e-01 5.40411055e-01
-8.15768301e-01 -3.78557220e-02 -9.15053248e-01 2.11556271e-01
2.63251007e-01 -6.98359191e-01 -7.50488698e-01 3.15124482e-01
8.76787543e-01 -7.82686591e-01 -7.23631680e-01 1.22015812e-01
4.04161692e-01 -1.26019895e+00 2.70567328e-01 -7.37748325e-01
1.35881388e+00 -3.47099513e-01 2.26088166e-01 -1.48495615e+00
-1.54338017e-01 -2.44787619e-01 -5.15568219e-02 1.10490334e+00
5.48730254e-01 -2.33178645e-01 1.13388383e+00 5.81354618e-01
-7.60844827e-01 -8.72832298e-01 -4.39310580e-01 -3.14612329e-01
1.50970995e-01 -5.49918771e-01 5.84883213e-01 1.25773144e+00
6.21954858e-01 4.59905826e-02 4.45500255e-01 3.05581808e-01
1.34863541e-01 1.14417046e-01 7.38614321e-01 -1.44692624e+00
-6.24682307e-01 -8.38154376e-01 -8.15246463e-01 -5.88910818e-01
3.91951203e-01 -9.26211953e-01 3.60427946e-01 -1.08323336e+00
-2.34311353e-03 -2.90084779e-01 -4.18208003e-01 1.18791950e+00
-3.47827911e-01 -4.29533869e-02 -2.77897894e-01 -4.18698639e-01
-6.35326385e-01 -1.84894368e-01 7.06604958e-01 -2.32927456e-01
2.03536339e-02 1.53217912e-01 -6.53730571e-01 7.68120408e-01
6.20364010e-01 -6.76456392e-01 -1.27182126e-01 -6.08092308e-01
8.72711062e-01 3.23961735e-01 4.25986856e-01 -1.24508822e+00
4.91864771e-01 -7.63176382e-02 2.61397004e-01 -4.43789631e-01
-6.10803127e-01 -4.75510716e-01 -4.87444550e-01 7.93343365e-01
-6.15859449e-01 4.80171949e-01 3.74732852e-01 4.10882533e-02
-4.05550748e-01 -1.04961729e+00 5.38886070e-01 -1.82752013e-01
-6.46404445e-01 -8.07348415e-02 -4.27831352e-01 -3.00145894e-01
9.16619539e-01 -4.63772379e-02 -7.88986444e-01 -5.95644824e-02
-2.62539655e-01 1.96876526e-01 5.08935452e-01 3.00567091e-01
6.53848052e-01 -1.16450465e+00 -4.57132339e-01 4.39562052e-01
2.22025156e-01 -1.78424299e-01 -2.91151609e-02 6.18597984e-01
-1.06186366e+00 4.41581607e-01 -2.94759303e-01 -6.62062466e-01
-1.47159755e+00 2.78484970e-01 3.91742319e-01 -4.70578521e-01
-2.38841608e-01 1.37121320e+00 -4.32031631e-01 -1.18131137e+00
4.27856833e-01 -9.94835675e-01 -6.86667040e-02 -4.46987182e-01
7.57415354e-01 -1.31275933e-02 7.49961674e-01 3.74290466e-01
-1.50484055e-01 -5.94686344e-02 -1.37422979e-01 4.20148134e-01
1.45249677e+00 9.43681359e-01 -5.92109442e-01 6.09626114e-01
9.66278493e-01 1.18948027e-01 -7.98479021e-01 -4.56731208e-02
2.58937687e-01 -4.44879442e-01 1.79599747e-01 -1.06190753e+00
-9.89534736e-01 1.12519860e+00 3.88244122e-01 4.49988723e-01
7.31020093e-01 -3.21541429e-01 4.30373937e-01 5.44327021e-01
4.66568381e-01 -6.01462603e-01 3.29632878e-01 9.61607337e-01
4.31906372e-01 -1.38020635e+00 -2.40758732e-01 -1.76326782e-01
3.19489181e-01 1.66919410e+00 1.36838055e+00 -3.79851818e-01
2.75503367e-01 4.21429306e-01 -2.59509742e-01 -2.24277109e-01
-1.22126639e+00 4.68629599e-01 2.40220055e-01 3.54606807e-01
1.22327542e+00 5.59807457e-02 1.59891531e-01 8.18708777e-01
-6.48859680e-01 5.22625931e-02 1.22456539e+00 1.43411732e+00
-9.23127711e-01 -1.25735593e+00 -6.80499554e-01 4.86102611e-01
-5.80776334e-01 -1.83011740e-01 -6.19292140e-01 4.34582353e-01
1.96517751e-01 8.22545886e-01 -2.29594652e-02 -7.72086620e-01
2.05517232e-01 4.74011123e-01 6.70535147e-01 -1.37040472e+00
-1.17425001e+00 -7.24237621e-01 -2.98981279e-01 -6.05329394e-01
3.18789691e-01 -4.13457304e-01 -1.18687558e+00 -6.33877814e-01
-4.53123808e-01 2.45097786e-01 7.04575360e-01 1.03270650e+00
1.28880918e-01 1.16078556e+00 3.26171815e-01 -2.42235765e-01
-7.54359186e-01 -8.84009540e-01 4.47769724e-02 -1.20632418e-01
6.09330654e-01 -4.20442015e-01 -2.14243799e-01 -6.59365132e-02]
|
[7.772510528564453, 7.721486568450928]
|
f73c4ccc-547e-45de-a097-1738b741cf92
|
membership-inference-on-word-embedding-and
|
2106.11384
| null |
https://arxiv.org/abs/2106.11384v1
|
https://arxiv.org/pdf/2106.11384v1.pdf
|
Membership Inference on Word Embedding and Beyond
|
In the text processing context, most ML models are built on word embeddings. These embeddings are themselves trained on some datasets, potentially containing sensitive data. In some cases this training is done independently, in other cases, it occurs as part of training a larger, task-specific model. In either case, it is of interest to consider membership inference attacks based on the embedding layer as a way of understanding sensitive information leakage. But, somewhat surprisingly, membership inference attacks on word embeddings and their effect in other natural language processing (NLP) tasks that use these embeddings, have remained relatively unexplored. In this work, we show that word embeddings are vulnerable to black-box membership inference attacks under realistic assumptions. Furthermore, we show that this leakage persists through two other major NLP applications: classification and text-generation, even when the embedding layer is not exposed to the attacker. We show that our MI attack achieves high attack accuracy against a classifier model and an LSTM-based language model. Indeed, our attack is a cheaper membership inference attack on text-generative models, which does not require the knowledge of the target model or any expensive training of text-generative models as shadow models.
|
['Marcello Hasegawa', 'Esha Ghosh', 'Melissa Chase', 'Huseyin A. Inan', 'Saeed Mahloujifar']
|
2021-06-21
| null | null | null | null |
['membership-inference-attack']
|
['computer-vision']
|
[ 3.64582956e-01 4.29048032e-01 -9.92515683e-03 -1.14850923e-01
-9.02315438e-01 -1.09099233e+00 9.53474879e-01 6.64573848e-01
-5.44273496e-01 3.52274656e-01 2.01502562e-01 -9.40267086e-01
3.34393919e-01 -1.15859091e+00 -8.92540812e-01 -7.66283751e-01
-1.49249196e-01 2.92007983e-01 3.40989262e-01 -7.42247924e-02
2.56576240e-01 5.93905389e-01 -8.57479453e-01 2.84712434e-01
2.75757730e-01 7.00849414e-01 -6.08828008e-01 1.11768889e+00
-4.44728769e-02 5.77211559e-01 -9.22480881e-01 -8.49028885e-01
2.91767627e-01 1.71621308e-01 -7.41622686e-01 -4.18010235e-01
3.16371679e-01 -2.67744392e-01 -3.91900271e-01 1.28954887e+00
3.96103084e-01 -2.82820791e-01 5.94044685e-01 -1.29577863e+00
-6.01723313e-01 1.08942354e+00 -1.52850881e-01 9.28782448e-02
1.61746591e-01 3.39391619e-01 9.78018343e-01 -6.39584541e-01
3.72465938e-01 1.04525936e+00 4.60455239e-01 6.35154068e-01
-1.44383609e+00 -7.58868217e-01 -1.15721270e-01 -2.87192851e-01
-1.10064852e+00 -3.34584117e-01 5.36182523e-01 -4.67029721e-01
9.96112287e-01 3.91476154e-01 -1.34133711e-01 1.80391538e+00
5.82192063e-01 6.21947765e-01 9.53153610e-01 -6.42539799e-01
4.63660032e-01 7.10543633e-01 5.50785422e-01 4.31773722e-01
7.11965263e-01 2.76031494e-01 -4.15670693e-01 -9.82819736e-01
7.67582143e-03 -1.62698984e-01 -5.13558626e-01 -1.88901007e-01
-8.09649825e-01 1.29049611e+00 4.96217012e-02 4.14396077e-01
1.52781472e-01 3.75610769e-01 7.14573085e-01 3.15471888e-01
4.73684043e-01 5.93212426e-01 -6.09502733e-01 -1.93423517e-02
-6.60073221e-01 3.02594770e-02 1.36018181e+00 5.47475219e-01
4.64977235e-01 6.10055774e-02 1.09587677e-01 -2.09744312e-02
4.13405538e-01 6.15402043e-01 6.35515630e-01 1.25797331e-01
6.90049231e-01 1.30451992e-01 -1.42966002e-01 -1.09399402e+00
-2.51448631e-01 -1.30441234e-01 -4.80465025e-01 1.65559888e-01
5.53315341e-01 -5.64411819e-01 -6.52145624e-01 1.80599260e+00
6.91361874e-02 8.78113210e-02 2.93529689e-01 3.98179829e-01
1.02178298e-01 9.39755082e-01 2.54887164e-01 1.30450204e-01
1.64794660e+00 -2.99892575e-01 -5.70700109e-01 -3.76101583e-01
1.01838326e+00 -6.99781954e-01 9.86537397e-01 4.41124737e-01
-9.07979906e-01 -3.46840941e-04 -1.44496906e+00 -7.90320784e-02
-1.07755721e+00 -3.35307658e-01 5.41342258e-01 1.49345219e+00
-6.43477261e-01 3.96385938e-01 -8.62077832e-01 -2.67022580e-01
2.96948642e-01 4.52695251e-01 -2.93065637e-01 2.15842232e-01
-1.99393237e+00 9.16579247e-01 3.53097409e-01 -2.01660931e-01
-8.41410041e-01 -6.84336305e-01 -9.95143473e-01 2.53068954e-01
1.24500796e-01 -2.79847533e-01 9.95562553e-01 -4.33079749e-01
-1.49383259e+00 6.83790267e-01 2.11525053e-01 -9.40042377e-01
5.48831701e-01 -2.33789876e-01 -4.81276900e-01 7.91601762e-02
-4.53876883e-01 -4.30423766e-02 1.31591284e+00 -1.06168163e+00
-1.91406056e-01 -4.40334290e-01 3.27830791e-01 -5.06389320e-01
-8.77884805e-01 1.70494020e-01 4.35821831e-01 -5.31262159e-01
-6.32417798e-01 -7.55955935e-01 -1.98809430e-01 2.23144214e-03
-7.21600592e-01 5.62527142e-02 1.22193396e+00 -3.57377321e-01
1.15463758e+00 -2.29242134e+00 -4.51068163e-01 5.17245889e-01
1.13176659e-01 8.32094610e-01 6.39407858e-02 8.01207066e-01
-7.12160394e-02 7.71047890e-01 -2.48471692e-01 -5.36803603e-01
4.66629028e-01 1.04691610e-01 -1.50421965e+00 7.58838415e-01
3.82016987e-01 8.14785600e-01 -5.76518297e-01 -1.56165570e-01
1.95844054e-01 5.62539756e-01 -4.72336948e-01 3.16768661e-02
-3.68578345e-01 -3.10147077e-01 -5.45589805e-01 9.35429037e-02
6.28093302e-01 7.04976097e-02 1.60980508e-01 -5.27782068e-02
1.98300898e-01 6.04214191e-01 -1.00073862e+00 1.28107715e+00
-7.92740762e-01 8.79949093e-01 -1.15580313e-01 -7.82558024e-01
4.96428519e-01 4.47142720e-01 -3.38136941e-01 1.41097441e-01
4.88670081e-01 5.06568812e-02 7.34180659e-02 -2.75721908e-01
6.10491812e-01 -1.68329597e-01 -4.62331057e-01 1.16925728e+00
7.46728480e-02 -4.07134555e-02 -3.30803573e-01 4.74793077e-01
1.30110312e+00 -4.87512916e-01 2.33748913e-01 -1.78876668e-01
5.74334145e-01 -2.28685558e-01 -6.97805583e-02 8.79408956e-01
-1.33726299e-02 2.54606694e-01 7.10919321e-01 -7.77443498e-02
-9.52619672e-01 -1.26085520e+00 -6.21369839e-01 9.53067720e-01
-2.57245719e-01 -6.94773436e-01 -8.60492349e-01 -9.04380858e-01
-1.10539779e-01 1.04414976e+00 -7.61894405e-01 -7.87557662e-01
-4.06534433e-01 -8.67650926e-01 1.39575934e+00 5.22139907e-01
1.78058550e-01 -6.28121436e-01 -6.43864095e-01 1.42540902e-01
3.82221639e-01 -1.19605124e+00 -5.33599317e-01 4.89307672e-01
-5.40457904e-01 -8.78363967e-01 1.72025692e-02 -3.70157003e-01
5.95685542e-01 -3.06612104e-01 6.25334203e-01 -1.43448770e-01
-3.16655725e-01 4.23261166e-01 -2.04557627e-01 -6.45201087e-01
-8.93340826e-01 3.54837060e-01 7.87223503e-02 2.11293906e-01
8.27796161e-01 -5.77874899e-01 4.59551774e-02 -1.16712995e-01
-1.47891533e+00 -6.83282077e-01 2.58190423e-01 7.00808346e-01
-1.96358278e-01 1.85930490e-01 2.41547853e-01 -1.35192227e+00
9.44611192e-01 -5.37115097e-01 -6.57880604e-01 2.62413472e-01
-2.12546185e-01 3.29726517e-01 1.07873046e+00 -9.12433803e-01
-5.60958803e-01 -2.12931529e-01 -2.71093518e-01 -2.17334196e-01
-2.04942092e-01 3.48157346e-01 -5.23493588e-01 -3.55236940e-02
1.04461467e+00 3.20956796e-01 -2.37552151e-01 -1.28957182e-01
4.91976321e-01 8.86194706e-01 1.22838795e-01 -6.40226722e-01
1.53347516e+00 5.47072709e-01 -2.07565576e-02 -1.01432347e+00
-5.07742763e-01 2.52048105e-01 -2.25108311e-01 4.54479814e-01
8.59151125e-01 -5.39268970e-01 -6.30946040e-01 3.05725604e-01
-1.50847828e+00 -2.18659446e-01 -2.66214043e-01 3.64298463e-01
-2.47595206e-01 5.31644762e-01 -6.28266096e-01 -9.10328448e-01
-3.70906055e-01 -1.21045220e+00 8.23415458e-01 -2.24166453e-01
-4.70307291e-01 -1.40576804e+00 -4.68895063e-02 -5.21001890e-02
4.99423385e-01 1.75473019e-01 1.31660855e+00 -1.62775779e+00
-3.61607522e-01 -9.31349874e-01 1.73903391e-01 5.01027942e-01
-8.58101994e-02 3.40228498e-01 -1.47466779e+00 -3.13956320e-01
5.34421623e-01 -4.06010032e-01 7.38045931e-01 -1.96232140e-01
9.81849790e-01 -7.00179696e-01 -2.06332043e-01 4.19998944e-01
1.33858812e+00 -1.46562949e-01 6.74562216e-01 -6.77871555e-02
3.91860157e-01 5.12274623e-01 -8.01075399e-02 4.02329147e-01
-2.96307415e-01 3.93628091e-01 3.97102475e-01 1.55441701e-01
5.88592172e-01 -5.61917603e-01 8.18241656e-01 2.73834467e-01
7.68624485e-01 -4.67488021e-01 -7.49573112e-01 1.80350453e-01
-1.34596992e+00 -9.17746544e-01 -9.70348716e-02 2.43346548e+00
9.80432570e-01 6.19059443e-01 -2.31731921e-01 4.32670474e-01
6.24639809e-01 4.40560758e-01 -2.34394789e-01 -1.09733999e+00
6.82006776e-02 6.56621814e-01 8.85229588e-01 7.98246503e-01
-1.24882007e+00 9.00367618e-01 5.81717539e+00 7.79308558e-01
-1.26477849e+00 4.75412339e-01 3.62406224e-01 7.82513246e-03
-4.39105839e-01 6.16148636e-02 -9.11068082e-01 6.28315866e-01
1.33665740e+00 -4.01546091e-01 3.99675295e-02 8.92730355e-01
-3.27413410e-01 1.72183827e-01 -1.36312664e+00 3.52224648e-01
1.55310586e-01 -1.02838945e+00 1.70976758e-01 4.74363297e-01
1.05860196e-02 -3.87404829e-01 5.01135111e-01 3.17005247e-01
7.76746213e-01 -1.06127059e+00 4.91271138e-01 -2.01434389e-01
4.90249753e-01 -1.11018062e+00 6.35700524e-01 4.71077800e-01
-6.10390306e-01 2.54178811e-02 -6.43582284e-01 1.79903716e-01
1.04121409e-01 8.36857438e-01 -7.63009608e-01 1.15765527e-01
-1.82226626e-03 -1.09346487e-01 -5.32827795e-01 1.68485686e-01
-7.09906936e-01 1.04952765e+00 -4.91957843e-01 -1.65775806e-01
4.36870843e-01 4.86856103e-01 6.38001382e-01 1.54059303e+00
-1.50744945e-01 -2.03584120e-01 -5.28822420e-04 1.06847060e+00
-2.39310145e-01 -2.86592841e-01 -9.96630728e-01 -4.11321700e-01
4.43091124e-01 1.26025915e+00 -3.26385736e-01 -1.29183486e-01
-2.30367362e-01 7.51146913e-01 1.16210282e-01 3.10774207e-01
-9.14195299e-01 -6.72974348e-01 9.09878254e-01 5.44307008e-03
1.24074712e-01 -4.62360531e-01 -2.44216934e-01 -1.26451683e+00
-6.82471469e-02 -7.36372292e-01 3.84664893e-01 -1.88918278e-01
-1.38119769e+00 3.81686836e-01 -1.87354714e-01 -7.77093709e-01
-4.67189074e-01 -8.62970412e-01 -8.28515947e-01 9.64523554e-01
-1.28715885e+00 -9.62489903e-01 7.21422374e-01 6.07469261e-01
-6.93483725e-02 -7.91670531e-02 1.09023893e+00 -6.91296160e-02
-5.91693819e-01 1.11258423e+00 5.91287725e-02 6.63832128e-01
5.04550338e-01 -1.12978017e+00 6.20216548e-01 1.41444397e+00
6.10657096e-01 1.02543437e+00 9.16849911e-01 -3.42664838e-01
-1.68355775e+00 -1.18298352e+00 1.09371150e+00 -6.89665258e-01
1.19154680e+00 -1.25163794e+00 -1.03743601e+00 1.05958962e+00
2.03407541e-01 -1.16386957e-01 1.10365689e+00 1.17137916e-02
-9.03734982e-01 2.96928942e-01 -1.24195051e+00 7.92827547e-01
4.26787227e-01 -1.09930158e+00 -8.42167079e-01 4.86226290e-01
1.04268873e+00 -1.08387135e-01 -6.92287922e-01 -2.26073995e-01
3.01029563e-01 -4.37261224e-01 9.14099813e-01 -9.86554444e-01
3.79551858e-01 -1.03654020e-01 -3.92291635e-01 -1.15336728e+00
2.65059143e-01 -7.67913461e-01 -3.29485059e-01 1.45724726e+00
6.16248310e-01 -1.28637874e+00 5.07068515e-01 1.04054415e+00
4.44660217e-01 -2.08348587e-01 -1.03391457e+00 -7.94386148e-01
5.85606694e-01 -8.01019311e-01 4.90970314e-01 1.06552410e+00
3.78987402e-01 2.41550714e-01 -1.59542367e-01 8.48068416e-01
4.83809143e-01 -1.76367626e-01 8.03026736e-01 -8.57563376e-01
-5.13858736e-01 -1.93139538e-01 -5.50375402e-01 -6.27546489e-01
5.29387772e-01 -1.07854533e+00 -1.16397165e-01 -6.63744867e-01
-4.19135392e-01 -9.68102664e-02 -1.10243773e-02 4.44078475e-01
1.81446318e-02 -2.62721851e-02 2.60043502e-01 -4.95734029e-02
1.31446674e-01 2.81913579e-01 2.52570987e-01 -2.87209690e-01
2.14117393e-01 8.26288909e-02 -6.92753375e-01 7.18488276e-01
9.36441779e-01 -9.37534392e-01 -4.72661644e-01 -1.77817672e-01
4.22760069e-01 -3.98307472e-01 4.49781030e-01 -9.32419360e-01
4.38570976e-01 9.22302604e-02 -2.92008966e-02 -1.28972426e-01
1.78114355e-01 -9.73117352e-01 -3.71629387e-01 7.01378942e-01
-6.28313363e-01 -2.66263247e-01 2.00958923e-01 5.70860922e-01
1.31635502e-01 -6.57220900e-01 7.76923656e-01 1.78862795e-01
1.99239142e-03 2.53733039e-01 -8.13218057e-01 1.51852071e-01
9.81914163e-01 4.72742617e-02 -5.81403553e-01 -2.43136466e-01
-5.88259041e-01 -9.14261490e-02 3.12737226e-01 2.62384862e-01
5.37050605e-01 -8.67740273e-01 -5.04682302e-01 4.38194394e-01
5.47989830e-02 -1.96859658e-01 -2.30809286e-01 3.45218271e-01
-1.76782683e-01 4.02197570e-01 3.76966685e-01 -1.93796977e-01
-1.11003947e+00 1.00648570e+00 2.72223920e-01 -5.32892168e-01
-4.41161543e-01 6.91490054e-01 4.30190593e-01 -2.66245931e-01
2.08602205e-01 -5.18963933e-01 2.80600041e-01 4.99227159e-02
9.86500263e-01 -9.50397104e-02 1.49036825e-01 -1.96730450e-01
-4.51731205e-01 1.36728808e-01 -3.65137279e-01 -3.65344882e-01
7.89104581e-01 2.62836337e-01 2.85075679e-02 4.80728209e-01
1.58895755e+00 4.43771541e-01 -6.27750576e-01 -3.38779032e-01
4.16106544e-02 -4.24656838e-01 -1.34071708e-01 -6.03812099e-01
-7.85269082e-01 1.60173392e+00 2.47923970e-01 6.00351632e-01
5.96525490e-01 -1.81240216e-01 9.27229404e-01 5.60221612e-01
3.69195640e-01 -9.19724166e-01 -1.65048555e-01 4.13260430e-01
5.38948357e-01 -7.14105964e-01 -1.48895621e-01 -2.64036834e-01
-2.58578837e-01 1.29358125e+00 1.44551560e-01 -2.06975564e-01
9.38110173e-01 8.10621202e-01 -1.03420988e-02 1.10549092e-01
-9.19086993e-01 3.46113682e-01 -2.81040221e-01 6.74932003e-01
1.12696610e-01 -6.25574961e-02 -1.01223446e-01 7.46016383e-01
-2.66263336e-01 -3.02440107e-01 9.44265306e-01 1.07487833e+00
-4.65180039e-01 -1.29127467e+00 -4.65138257e-01 2.39561900e-01
-9.26155269e-01 -4.84613180e-01 -4.89445567e-01 7.18995810e-01
-2.00976938e-01 9.56445634e-01 -2.78312974e-02 -3.89750302e-01
5.16842268e-02 4.44329590e-01 -1.67165585e-02 -7.17423916e-01
-9.86003220e-01 -6.33031487e-01 5.81169985e-02 -2.55875498e-01
2.96510726e-01 -4.44095522e-01 -1.14891601e+00 -6.75450325e-01
-4.18047190e-01 1.64890155e-01 7.50622511e-01 8.50671470e-01
2.66466141e-01 -2.28864178e-02 7.72812128e-01 -4.28329438e-01
-1.19470596e+00 -7.67784059e-01 -7.61938572e-01 1.58443332e-01
2.68983990e-01 -1.43398091e-01 -9.54692543e-01 -1.54687852e-01]
|
[5.978590965270996, 7.713857650756836]
|
cf1a3cd4-4cd4-42d6-bb6a-77ff455fe0aa
|
attribute-based-classification-for-zero-shot
| null | null |
https://ieeexplore.ieee.org/document/6571196
|
https://hannes.nickisch.org/papers/articles/lampert13attributes.pdf
|
Attribute-Based Classification for Zero-Shot Visual Object Categorization
|
We study the problem of object recognition for categories for which we have no training examples, a task also called
zero-data or zero-shot learning. This situation has hardly been studied in computer vision research, even though it occurs frequently;
the world contains tens of thousands of different object classes, and image collections have been formed and suitably annotated for
only a few of them. To tackle the problem, we introduce attribute-based classification: Objects are identified based on a high-level
description that is phrased in terms of semantic attributes, such as the object’s color or shape. Because the identification of each such
property transcends the specific learning task at hand, the attribute classifiers can be prelearned independently, for example, from
existing image data sets unrelated to the current task. Afterward, new classes can be detected based on their attribute representation,
without the need for a new training phase. In this paper, we also introduce a new data set, Animals with Attributes, of over 30,000
images of 50 animal classes, annotated with 85 semantic attributes. Extensive experiments on this and two more data sets show that
attribute-based classification indeed is able to categorize images without access to any training images of the target classes.
|
['Stefan Harmeling', 'Hannes Nickisch', 'Christoph H. Lampert']
|
2013-07-30
| null | null | null |
ieee-transactions-on-pattern-analysis-and-17
|
['object-categorization']
|
['computer-vision']
|
[ 6.51560426e-01 6.67325500e-03 -2.11999178e-01 -7.77121723e-01
-3.14088523e-01 -5.58771849e-01 7.61801422e-01 7.45250463e-01
-5.48624992e-01 6.59899533e-01 -3.38572323e-01 9.68247205e-02
-2.06918105e-01 -9.79294896e-01 -5.76099455e-01 -8.72221649e-01
-7.43252337e-02 9.16677356e-01 3.11959207e-01 -6.50445595e-02
2.94633687e-01 2.71926254e-01 -2.28075194e+00 3.43932629e-01
5.56037188e-01 1.21460462e+00 3.42180580e-01 4.29113090e-01
-3.95198256e-01 5.42246699e-01 -5.02714157e-01 -3.68430495e-01
8.41224641e-02 -3.42833430e-01 -9.62945521e-01 4.57816929e-01
5.77488482e-01 2.67899409e-02 2.19117850e-01 1.19854999e+00
5.73022151e-03 1.71006843e-01 9.98260856e-01 -1.49470925e+00
-5.66954195e-01 1.28662676e-01 -1.48225695e-01 5.43669276e-02
1.61698833e-01 -2.20657840e-01 1.12668729e+00 -7.76353002e-01
5.70498049e-01 1.06842017e+00 2.54941374e-01 6.01642311e-01
-1.55802953e+00 -4.19084489e-01 4.12085764e-02 5.04382908e-01
-1.39310396e+00 -2.05192104e-01 6.83411062e-01 -6.12709701e-01
4.14287537e-01 4.99561056e-02 5.91462195e-01 8.57127190e-01
-2.14189380e-01 5.26575863e-01 1.01693654e+00 -6.15014672e-01
4.96156871e-01 4.61655229e-01 4.04000729e-01 3.79882693e-01
3.31379354e-01 -1.76703334e-02 -2.13567540e-01 -1.07791740e-02
3.18880081e-01 7.35276490e-02 -6.21357262e-02 -9.68763709e-01
-1.42803168e+00 8.64163399e-01 4.81958210e-01 4.66420561e-01
-2.88738728e-01 -2.73311913e-01 5.19989967e-01 3.77890706e-01
3.91166568e-01 5.82598984e-01 -5.35786390e-01 2.22450748e-01
-3.91103357e-01 -2.80193659e-03 8.49338293e-01 1.22538865e+00
1.32880652e+00 -1.84792846e-01 1.98416457e-01 9.32226896e-01
-1.77989334e-01 3.28330666e-01 5.81435859e-01 -7.23597348e-01
-1.74091458e-02 8.18179250e-01 2.44431067e-02 -8.62348437e-01
-2.64869034e-01 -1.15871653e-01 -8.70315731e-01 1.35419905e-01
4.79413062e-01 4.05758888e-01 -1.06683719e+00 1.67236447e+00
3.76536131e-01 1.90821841e-01 2.51920879e-01 7.76051462e-01
8.04950178e-01 5.93208432e-01 3.91722560e-01 -7.31954426e-02
1.57092273e+00 -4.77739930e-01 -5.71380734e-01 -3.74335587e-01
5.26490569e-01 -4.57543582e-01 1.03852737e+00 3.86520118e-01
-2.11774483e-01 -5.97151101e-01 -1.09891117e+00 1.71430260e-01
-9.69864428e-01 -3.24235894e-02 7.53227949e-01 4.77651238e-01
-5.56488752e-01 4.45782423e-01 -2.06113592e-01 -6.43503487e-01
4.17563677e-01 3.56521934e-01 -6.97587371e-01 -8.82982761e-02
-9.87850845e-01 8.73372674e-01 9.08201933e-01 -3.33990753e-01
-8.85178387e-01 -3.23343188e-01 -9.49238479e-01 -9.51233879e-02
6.10697031e-01 -4.31794763e-01 1.01296759e+00 -1.48291171e+00
-8.21436584e-01 1.46132374e+00 7.80281425e-03 -2.98276961e-01
-3.83570120e-02 1.13437377e-01 -6.00416183e-01 1.42458171e-01
4.71018702e-01 5.98566592e-01 1.03346348e+00 -1.46095383e+00
-8.79775941e-01 -6.28669620e-01 1.88353047e-01 -1.30979801e-02
-5.39350867e-01 -1.09162830e-01 -1.93362147e-01 -5.76209605e-01
2.18845800e-01 -8.98157775e-01 -1.31068587e-01 1.43619969e-01
-2.55361021e-01 -3.78210366e-01 7.64533401e-01 7.55687654e-02
5.80298126e-01 -2.42925358e+00 4.06946689e-02 1.93764478e-01
1.97841808e-01 1.87261090e-01 -1.24860615e-01 8.11606497e-02
-3.22275877e-01 -2.31809653e-02 -5.86988211e-01 8.16714913e-02
-2.29681686e-01 6.29224360e-01 -1.56406268e-01 3.35377306e-01
2.63715059e-01 4.04952645e-01 -1.01982021e+00 -6.73846424e-01
2.58527339e-01 1.24626681e-01 -2.37357691e-01 2.95048863e-01
-2.03566521e-01 3.35049450e-01 -5.74303389e-01 5.70303500e-01
3.00199449e-01 -1.51748642e-01 1.37730181e-01 -2.27532446e-01
2.11964503e-01 -2.13028073e-01 -1.15264416e+00 1.35801125e+00
-4.77197260e-01 4.06451017e-01 -3.11562270e-01 -1.61555243e+00
1.15178406e+00 3.06154758e-01 5.23002803e-01 -3.93045068e-01
8.44241530e-02 2.71186143e-01 1.71027794e-01 -5.46290159e-01
2.46547416e-01 -3.88565540e-01 -3.24790299e-01 3.42406571e-01
4.03695792e-01 -1.35831222e-01 4.13436830e-01 -8.37624595e-02
6.35856748e-01 -5.04146628e-02 7.32917666e-01 -2.63535172e-01
6.65297508e-01 3.35736424e-01 5.67321897e-01 6.70838535e-01
-2.01869130e-01 5.16808271e-01 2.16982976e-01 -8.80145490e-01
-1.28794706e+00 -9.95260656e-01 -5.50831914e-01 1.46273458e+00
4.20954406e-01 -2.10390598e-01 -6.02520525e-01 -5.95127463e-01
2.37921681e-02 5.60974479e-01 -8.95708799e-01 -2.38506585e-01
-2.19839677e-01 -6.64249659e-01 8.56960099e-03 3.84487212e-01
3.82345796e-01 -1.37642443e+00 -7.88358510e-01 2.65949965e-01
-6.65653422e-02 -1.13135147e+00 4.01493423e-02 5.06290078e-01
-7.60090947e-01 -1.28502464e+00 -4.47343647e-01 -9.51085627e-01
1.04441667e+00 2.29075983e-01 1.31369543e+00 2.58557111e-01
-5.14226496e-01 4.62720871e-01 -6.68401778e-01 -5.98680258e-01
-5.15542269e-01 -1.58174187e-01 1.96736962e-01 5.66693246e-01
7.51634300e-01 -4.55599725e-01 -1.85698226e-01 4.30474818e-01
-8.34993958e-01 -1.12261586e-01 8.64310145e-01 9.06336129e-01
6.91759467e-01 -1.51033867e-02 7.67995834e-01 -1.14599574e+00
-3.70380171e-02 -4.66114134e-01 -3.94127995e-01 3.47813547e-01
-3.06264848e-01 2.02695712e-01 5.79711199e-01 -6.32540703e-01
-7.88068175e-01 5.07780254e-01 1.71798229e-01 -2.07893386e-01
-8.23951125e-01 2.83474028e-01 -5.37553370e-01 1.90278217e-02
7.05825210e-01 2.48453930e-01 2.27870364e-02 -3.92835617e-01
4.39122409e-01 8.29306304e-01 8.29653800e-01 -6.06400549e-01
6.74479544e-01 2.80960411e-01 1.54507354e-01 -9.37081814e-01
-1.20156205e+00 -8.24001670e-01 -1.30613554e+00 -2.50573874e-01
1.08912957e+00 -5.71083724e-01 -5.20577133e-01 3.65346372e-01
-9.64836240e-01 1.43229678e-01 -4.63079780e-01 5.64822495e-01
-9.14582193e-01 7.93496817e-02 -8.02294761e-02 -5.42839527e-01
6.69739097e-02 -9.52255666e-01 8.58619750e-01 1.15637407e-02
-1.44869789e-01 -7.04696000e-01 -2.43037343e-01 7.77724907e-02
1.64678007e-01 2.27349758e-01 1.44096911e+00 -1.34848320e+00
-3.61803383e-01 -4.68350083e-01 -1.44330963e-01 5.20567954e-01
5.48348248e-01 -3.49945575e-01 -1.05051446e+00 -1.93890721e-01
8.44817236e-02 -5.11078596e-01 7.57174313e-01 8.84583686e-03
1.49518180e+00 -2.23004639e-01 -4.52753603e-01 2.79159278e-01
1.44703901e+00 4.03800398e-01 3.11232299e-01 3.71140569e-01
6.64899468e-01 9.15528357e-01 8.61393332e-01 3.35241079e-01
6.27007484e-02 7.30208158e-01 6.09600842e-01 5.71606271e-02
1.06931120e-01 1.24124520e-01 -3.10032398e-01 3.09447318e-01
-2.10330456e-01 5.57617052e-03 -8.28811646e-01 5.70361912e-01
-1.60722423e+00 -9.26463187e-01 -5.05989492e-02 2.67227650e+00
8.67314279e-01 1.16975918e-01 1.28117368e-01 3.60059083e-01
1.08670402e+00 -1.71544060e-01 -7.31237054e-01 -1.04969941e-01
-1.86605584e-02 -5.00506721e-02 2.62500226e-01 1.11099303e-01
-1.50217509e+00 7.06167281e-01 6.12416458e+00 7.15339959e-01
-8.62302959e-01 -1.43203452e-01 6.60699248e-01 5.70604920e-01
1.35844639e-02 1.00330546e-01 -6.64532542e-01 3.67403477e-01
6.82905793e-01 -4.24451768e-01 1.50537938e-01 1.12343752e+00
-5.21873057e-01 -1.89435646e-01 -1.57634091e+00 9.43714857e-01
4.00873780e-01 -9.38335061e-01 4.06724244e-01 -5.92358559e-02
2.16103584e-01 -5.97575426e-01 -2.38689482e-01 3.49450052e-01
1.17688969e-01 -9.11575854e-01 5.72596967e-01 3.64373833e-01
8.31489444e-01 -7.45150566e-01 7.53685415e-01 5.49036801e-01
-1.13721168e+00 -1.82324305e-01 -6.70014799e-01 1.04919076e-01
-3.72247785e-01 2.56037831e-01 -7.54665136e-01 4.40513253e-01
7.40923166e-01 8.56591105e-01 -7.69258976e-01 1.18363893e+00
-1.80476010e-01 3.90217632e-01 -1.20209128e-01 -4.51970771e-02
1.11388326e-01 -2.51454771e-01 3.98939669e-01 9.00384188e-01
3.35030258e-01 3.84933650e-01 3.92007500e-01 6.33018076e-01
4.81683202e-02 1.20671280e-01 -8.78157914e-01 2.03967914e-01
4.79543895e-01 1.31516755e+00 -9.08888102e-01 -6.74953699e-01
-4.67559695e-01 8.39877248e-01 1.35360837e-01 8.38093907e-02
-4.68062848e-01 -5.75424671e-01 5.34263790e-01 3.17258537e-02
2.30263248e-01 7.13879988e-02 7.01219812e-02 -9.63952839e-01
-2.25630417e-01 -6.50993586e-01 5.92362881e-01 -6.30311430e-01
-1.59683442e+00 7.34538198e-01 2.03054264e-01 -1.60137129e+00
-4.34913099e-01 -8.49511802e-01 -1.32249609e-01 6.22988462e-01
-1.27208865e+00 -1.17215097e+00 -4.20246333e-01 7.79171348e-01
6.14020705e-01 -3.18423867e-01 1.22445142e+00 2.84562677e-01
-9.37787890e-02 1.27439216e-01 9.76465195e-02 2.75465310e-01
9.25991714e-01 -1.49168670e+00 -1.04476944e-01 3.36133391e-01
4.21633273e-01 3.24778795e-01 8.75826895e-01 -2.62387723e-01
-1.00743425e+00 -1.12260044e+00 9.38819230e-01 -4.99071687e-01
6.56141281e-01 -4.99938756e-01 -1.31684434e+00 5.46007991e-01
-1.16138525e-01 4.66903239e-01 8.10107052e-01 2.03428343e-01
-5.20526648e-01 -3.32512051e-01 -1.09122169e+00 1.26368797e-03
7.99436986e-01 -4.62569028e-01 -9.24476266e-01 4.78669375e-01
5.26897848e-01 1.55564100e-01 -8.05381596e-01 3.89129013e-01
2.20330760e-01 -6.54029012e-01 9.87199664e-01 -8.96262705e-01
2.10186139e-01 -3.96341294e-01 -3.88828158e-01 -1.40722239e+00
-3.50102901e-01 2.88895965e-01 4.41668004e-01 1.30448365e+00
3.02930266e-01 -5.48717737e-01 5.54927886e-01 4.81656462e-01
-1.39495581e-01 -2.50008166e-01 -9.17927146e-01 -1.15925205e+00
-1.15029633e-01 -2.10506350e-01 6.20371222e-01 1.17836261e+00
-3.40035796e-01 7.41578877e-01 -2.90957004e-01 -1.95687860e-02
8.06846976e-01 6.44628942e-01 6.68022871e-01 -1.89109242e+00
6.01372309e-02 -2.51932979e-01 -1.03845489e+00 -4.19148982e-01
3.10078233e-01 -9.48975623e-01 3.20928156e-01 -1.26275110e+00
4.96482611e-01 -6.49818838e-01 -1.96534574e-01 7.84158826e-01
-1.87852338e-01 4.39331800e-01 2.08335998e-03 4.11558628e-01
-5.62004924e-01 4.30143535e-01 9.30070221e-01 -4.68394935e-01
2.03102991e-01 1.59439117e-01 -4.37833726e-01 8.40230227e-01
6.22643948e-01 -6.34352505e-01 -3.79101664e-01 2.07571760e-02
-1.67476982e-01 -1.17228568e-01 5.92871308e-01 -1.13464391e+00
1.40464589e-01 -2.98215598e-01 3.71924490e-01 -3.01897377e-01
4.31803524e-01 -1.16906893e+00 7.81824067e-02 2.22757131e-01
-5.01759708e-01 -1.77738607e-01 -2.15941235e-01 7.92071640e-01
-4.24182504e-01 -6.64264143e-01 1.04779792e+00 -2.58927315e-01
-1.53313291e+00 3.82124484e-01 -3.17473799e-01 3.04930611e-04
1.47591591e+00 -4.18036878e-01 -1.37535095e-01 -9.56961960e-02
-1.13454175e+00 -4.41783257e-02 4.74363714e-01 4.38683689e-01
5.64657748e-01 -1.40832710e+00 -6.59877539e-01 2.18471169e-01
1.00430822e+00 -9.93254334e-02 -9.25084352e-02 2.78776407e-01
-3.20914574e-02 4.44690622e-02 -7.25061476e-01 -8.84486496e-01
-1.26464570e+00 1.25024807e+00 1.43336162e-01 3.29086870e-01
-6.97389901e-01 4.29157764e-01 5.51393688e-01 -4.26780730e-01
6.85279816e-02 7.46733099e-02 -6.33841693e-01 3.86057913e-01
4.93958682e-01 -1.49368361e-01 1.24985196e-01 -9.75567043e-01
-3.68982553e-01 6.68528795e-01 -2.17638612e-02 2.87678272e-01
1.28682590e+00 -7.22210854e-02 -2.34555334e-01 8.98323476e-01
1.14556026e+00 -6.57231867e-01 -8.94672573e-01 -5.18556058e-01
4.18349028e-01 -7.09863603e-01 -4.01318878e-01 -6.47230983e-01
-9.41662014e-01 9.43663538e-01 6.81395590e-01 4.75105494e-01
1.25178993e+00 3.41540009e-01 1.09577611e-01 7.23826885e-01
7.52004147e-01 -1.04024589e+00 1.93077326e-01 4.33852971e-01
7.19326735e-01 -1.53908265e+00 -1.52999267e-01 -5.40514708e-01
-6.59482896e-01 1.11182737e+00 7.35720515e-01 1.67525224e-02
7.09064960e-01 -8.19883049e-02 -7.84376636e-02 -2.26477072e-01
-5.78251123e-01 -4.76093799e-01 2.70886332e-01 8.74246597e-01
2.57363886e-01 2.45061502e-01 -1.22198328e-01 3.68373424e-01
-9.36038047e-02 -3.37326556e-01 4.74033177e-01 8.77108335e-01
-8.82987797e-01 -1.15458274e+00 -2.69398540e-01 7.28750110e-01
-1.53981686e-01 2.21082754e-02 -3.32337171e-01 7.67256618e-01
3.64552677e-01 7.18380392e-01 4.84508455e-01 -1.68655112e-01
3.33215743e-01 1.14227198e-01 3.48849207e-01 -9.40740764e-01
7.44424947e-03 -3.82074386e-01 7.76125342e-02 -2.35025689e-01
-7.29695559e-01 -6.18496358e-01 -1.10227191e+00 7.85597935e-02
-1.34963512e-01 3.92606974e-01 5.44841707e-01 1.11246288e+00
-1.11722127e-01 1.70491517e-01 6.70267820e-01 -8.06771219e-01
-3.07891786e-01 -7.22168267e-01 -7.70861983e-01 8.94002557e-01
3.41885000e-01 -9.94559884e-01 -3.66574913e-01 5.78299522e-01]
|
[9.704559326171875, 2.5364162921905518]
|
3c1703a3-c0b3-4874-934c-8ec57f14f808
|
rigid-soft-interactive-learning-for-robust
|
2003.01584
| null |
https://arxiv.org/abs/2003.01584v1
|
https://arxiv.org/pdf/2003.01584v1.pdf
|
Rigid-Soft Interactive Learning for Robust Grasping
|
Inspired by widely used soft fingers on grasping, we propose a method of rigid-soft interactive learning, aiming at reducing the time of data collection. In this paper, we classify the interaction categories into Rigid-Rigid, Rigid-Soft, Soft-Rigid according to the interaction surface between grippers and target objects. We find experimental evidence that the interaction types between grippers and target objects play an essential role in the learning methods. We use soft, stuffed toys for training, instead of everyday objects, to reduce the integration complexity and computational burden and exploit such rigid-soft interaction by changing the gripper fingers to the soft ones when dealing with rigid, daily-life items such as the Yale-CMU-Berkeley (YCB) objects. With a small data collection of 5K picking attempts in total, our results suggest that such Rigid-Soft and Soft-Rigid interactions are transferable. Moreover, the combination of different grasp types shows better performance on the grasping test. We achieve the best grasping performance at 97.5\% for easy YCB objects and 81.3\% for difficult YCB objects while using a precise grasp with a two-soft-finger gripper to collect training data and power grasp with a four-soft-finger gripper to test.
|
['Xiaobo Liu', 'Fang Wan', 'Linhan Yang', 'Yujia Liu', 'Jia Pan', 'Haokun Wang', 'Chaoyang Song']
|
2020-02-29
| null | null | null | null |
['small-data']
|
['computer-vision']
|
[-2.59180605e-01 -1.72452509e-01 -1.16170704e-01 -2.59060621e-01
-1.47117972e-01 -8.55080366e-01 1.31676480e-01 -6.66799366e-01
-4.52767134e-01 4.39083159e-01 -4.95177984e-01 3.07273138e-02
-7.61733413e-01 -6.00737810e-01 -9.15537298e-01 -1.03197730e+00
-3.30262631e-01 8.24090123e-01 6.00727260e-01 -2.93740749e-01
2.58037060e-01 6.49160206e-01 -1.70309639e+00 6.13730729e-01
8.09419334e-01 1.08512175e+00 9.95454967e-01 4.97207999e-01
1.16926305e-01 -1.16715468e-01 -3.11791986e-01 -4.91925985e-01
7.33669460e-01 7.82898128e-01 -6.75185323e-01 -3.74620229e-01
4.14794125e-02 -7.65954137e-01 -7.82386065e-02 8.05459321e-01
4.20720041e-01 7.38160238e-02 9.78258669e-01 -1.08007157e+00
-7.50909150e-01 9.63414311e-01 -4.83650565e-01 -3.85545522e-01
4.18606728e-01 2.70084679e-01 5.48745036e-01 -8.50937784e-01
4.73852694e-01 1.59342790e+00 5.54441273e-01 9.02516186e-01
-8.76436472e-01 -5.70645213e-01 3.81524153e-02 2.41583772e-02
-9.06078577e-01 -9.18682218e-02 7.78316498e-01 -5.44705093e-01
6.36053503e-01 4.91816759e-01 4.45666790e-01 1.52249229e+00
4.12563503e-01 7.41385579e-01 1.14822018e+00 -5.01507223e-01
1.42815217e-01 6.10034242e-02 3.88390750e-01 4.95264858e-01
4.52570587e-01 -4.56417969e-04 6.09019063e-02 -6.14971742e-02
1.08941293e+00 3.38788927e-01 -2.02722371e-01 -2.48075098e-01
-1.16448426e+00 3.15918863e-01 5.72594404e-01 3.97334486e-01
-1.63182616e-01 9.32903364e-02 2.41210327e-01 3.12844723e-01
4.95940037e-02 4.01099682e-01 -8.01811814e-01 -2.19371766e-02
-5.32841757e-02 4.76383388e-01 1.24568355e+00 1.50212431e+00
1.09966561e-01 -3.67336631e-01 -2.13653043e-01 1.18121672e+00
4.85143870e-01 9.91848767e-01 1.99382201e-01 -8.27993572e-01
8.16797316e-01 2.99294233e-01 5.27121067e-01 -8.92851532e-01
-3.39206159e-01 7.39544451e-01 -6.81719363e-01 5.45824647e-01
8.10910821e-01 1.63684621e-01 -1.02620196e+00 1.15658832e+00
2.29420185e-01 -7.23790228e-01 -1.82025433e-01 1.09403980e+00
9.66288090e-01 2.35373780e-01 -1.25109674e-02 9.75190029e-02
1.34219122e+00 -9.05282795e-01 -4.71545517e-01 2.73142129e-01
6.88595995e-02 -9.13580775e-01 1.62491381e+00 8.27261150e-01
-1.12782288e+00 -9.37267303e-01 -6.69007182e-01 1.01819880e-01
-5.22701442e-01 3.61792415e-01 9.26765501e-01 4.20011550e-01
-4.39732611e-01 1.06360257e+00 -9.13920224e-01 -1.63648695e-01
1.52251855e-01 5.53092062e-01 -2.32440025e-01 -1.78750247e-01
-8.37313592e-01 8.96943212e-01 5.69746256e-01 2.08945662e-01
-5.58654666e-01 -2.84418762e-01 -9.70129445e-02 -7.74468109e-02
4.92868036e-01 -1.35259330e-01 8.52261603e-01 -2.82307029e-01
-1.92409933e+00 8.24531376e-01 6.34950578e-01 3.83586049e-01
8.30815375e-01 -7.47046351e-01 1.83733806e-01 9.91580561e-02
-3.20148647e-01 3.60658318e-01 1.24335885e+00 -1.86938918e+00
-7.66195282e-02 -3.78883421e-01 3.39558870e-01 -1.36590555e-01
-3.40265185e-01 -4.50839847e-02 -4.30646151e-01 -8.63024175e-01
4.13098961e-01 -1.31643867e+00 2.77441263e-01 1.71367675e-01
-6.20381296e-01 -8.21427345e-01 1.04228890e+00 -5.09076834e-01
5.47390521e-01 -2.04545069e+00 6.72702849e-01 4.05366898e-01
9.45047243e-04 4.14185226e-01 -2.69361943e-01 3.46630007e-01
2.38385007e-01 1.14002414e-02 2.54606873e-01 -1.34398475e-01
3.93311709e-01 4.61384147e-01 -4.34629709e-01 -6.20840527e-02
-3.20773363e-01 8.09402227e-01 -8.38779688e-01 -5.92293739e-01
2.04407901e-01 -3.95608507e-03 -6.08763933e-01 6.80680513e-01
-2.49820501e-01 1.94679961e-01 -8.85999501e-01 1.54889989e+00
9.26224113e-01 1.28749028e-01 5.39227873e-02 -6.25089169e-01
4.86643352e-02 -4.24696058e-01 -1.13927770e+00 1.40300810e+00
-2.95282573e-01 -1.49420097e-01 5.84342420e-01 -7.18144715e-01
1.15186644e+00 -5.77232540e-02 4.20321822e-01 -3.03541958e-01
1.52370930e-01 5.88461876e-01 1.40780106e-01 -1.25581479e+00
2.55214453e-01 1.06711403e-01 1.83898315e-01 1.96547925e-01
-7.26146102e-02 -6.52759254e-01 9.47361067e-02 -3.96357358e-01
6.93617702e-01 5.57028770e-01 -3.45153987e-01 -6.54865086e-01
-1.81788564e-01 -5.42380393e-01 2.71598790e-02 8.24942529e-01
1.30694315e-01 5.31955063e-01 -1.44318361e-02 -4.52418685e-01
-9.52894568e-01 -1.40226209e+00 -4.19593662e-01 1.56522298e+00
5.28471529e-01 1.83849946e-01 -7.05505729e-01 -5.07005036e-01
7.13004827e-01 1.50221422e-01 -4.36934173e-01 1.70618296e-03
-8.14437509e-01 -5.49045444e-01 2.10782159e-02 6.78746581e-01
4.84112531e-01 -1.53093112e+00 -6.75694764e-01 -3.82695422e-02
7.42205000e-03 -8.48027170e-01 -9.44959652e-03 2.68577427e-01
-8.36841404e-01 -1.17619014e+00 -9.57901478e-01 -7.96991110e-01
6.94660664e-01 1.94365785e-01 8.38317692e-01 2.98003167e-01
-4.24508274e-01 5.26725769e-01 -1.13517785e+00 -5.28239429e-01
-2.06262037e-01 -2.84660161e-02 5.76240897e-01 -5.44736981e-01
-1.32465586e-01 -5.05486906e-01 -4.76156682e-01 7.79214203e-01
-7.16542423e-01 -7.06528276e-02 7.69323170e-01 7.28209555e-01
1.84533685e-01 -1.32058278e-01 2.74349093e-01 -2.83813417e-01
4.07593727e-01 -2.71974534e-01 -2.65264541e-01 6.28674090e-01
9.99874845e-02 -9.19977799e-02 5.23633778e-01 -1.25581408e+00
-8.98150742e-01 -3.13048661e-01 8.33466724e-02 -7.97571778e-01
-3.33691873e-02 -5.48673654e-03 -1.17480323e-01 -4.36616033e-01
3.93069893e-01 -1.85176298e-01 -2.28259981e-01 -9.14716780e-01
3.41202207e-02 9.79003072e-01 1.64678887e-01 -1.44148207e+00
8.53615403e-01 1.23698249e-01 -2.61831731e-01 -7.07924664e-01
-2.21896246e-01 1.61414623e-01 -1.03638566e+00 -4.71232206e-01
8.97786677e-01 4.43216525e-02 -1.69199812e+00 9.94479835e-01
-1.21670520e+00 -8.47619414e-01 -2.47757994e-02 5.73537827e-01
-7.99504697e-01 3.02983373e-01 -9.13191557e-01 -8.22937310e-01
-5.25540888e-01 -1.40208995e+00 1.21753240e+00 1.10425629e-01
1.15896791e-01 -3.77974689e-01 -5.92318058e-01 3.83601815e-01
3.73339117e-01 1.48337260e-01 1.05317521e+00 -3.14533710e-01
-6.07986569e-01 -2.37342343e-01 -2.65445292e-01 5.18154442e-01
4.69600916e-01 2.95700580e-01 -6.17451131e-01 -7.45547354e-01
-9.92734823e-03 -7.66168892e-01 7.07538068e-01 2.73181170e-01
1.69013977e+00 -3.07063639e-01 -5.97461283e-01 3.24614882e-01
1.23716104e+00 5.59730768e-01 6.47355497e-01 -1.62582658e-02
8.38641822e-01 5.37923932e-01 9.76249039e-01 4.27336246e-01
-1.17708556e-01 7.73801029e-01 4.49705571e-01 1.66607782e-01
8.95664543e-02 2.90145606e-01 2.27143735e-01 9.38306093e-01
-1.21886921e+00 -2.29218453e-01 -9.23977971e-01 -5.28112650e-02
-1.63862455e+00 -7.96850562e-01 1.55291960e-01 2.14332175e+00
1.22117448e+00 4.46662754e-01 5.98909659e-03 1.37325138e-01
3.89703095e-01 -5.03131151e-01 -5.61398983e-01 -1.12532854e-01
3.53185594e-01 1.93449989e-01 2.73697317e-01 2.48479053e-01
-9.17166471e-01 7.55306542e-01 6.19349432e+00 9.50226307e-01
-1.16542566e+00 -3.63512665e-01 5.63655011e-02 -7.00778216e-02
1.27944157e-01 -7.28736401e-01 -8.41364264e-01 1.11626470e+00
-2.44188607e-01 6.40040338e-01 8.51630330e-01 1.16638017e+00
-3.13196272e-01 -2.70878047e-01 -1.59071445e+00 9.43432510e-01
-3.47439855e-01 -5.94381452e-01 7.76541457e-02 -3.93872380e-01
1.21403269e-01 -3.73029500e-01 -7.00806677e-02 4.19516414e-01
1.84444577e-01 -9.23561811e-01 8.84379864e-01 1.12187779e+00
8.31949711e-01 -1.49562478e-01 5.82128882e-01 4.23238844e-01
-1.14622891e+00 -4.35845524e-01 -7.35800862e-01 8.38192627e-02
-3.20656508e-01 4.86458480e-01 -4.07735735e-01 5.40573716e-01
1.55898881e+00 2.11181387e-01 -2.58528478e-02 6.07745230e-01
2.20264852e-01 6.03294745e-02 -4.47371453e-01 -4.16582704e-01
-3.55889201e-01 -2.97332674e-01 5.25226891e-01 1.19053483e+00
1.55310258e-01 5.96070528e-01 4.70880240e-01 1.09821677e+00
2.27212936e-01 -3.25011462e-01 -5.00525594e-01 8.46804976e-02
5.68710923e-01 1.27443528e+00 -7.80874848e-01 -3.03981066e-01
7.04363063e-02 7.56854832e-01 3.19755554e-01 3.34161490e-01
-7.19798267e-01 -6.27552509e-01 2.01092646e-01 -1.27694666e-01
3.37563843e-01 -6.40549302e-01 -2.25979581e-01 -1.04819119e+00
5.32004774e-01 -8.17038536e-01 -3.11504185e-01 -1.05791688e+00
-1.82009757e+00 4.15888458e-01 5.79758465e-01 -1.21633959e+00
4.11481291e-01 -1.40889442e+00 -4.21940327e-01 8.22658718e-01
-7.39066124e-01 -1.15260541e+00 -6.71893060e-01 5.36160648e-01
6.05521619e-01 -8.07914734e-02 6.97336197e-01 -1.33466255e-02
-6.90981969e-02 6.21969819e-01 1.90833151e-01 9.74435806e-02
7.06159592e-01 -1.04619849e+00 -2.75330812e-01 -2.78298378e-01
-6.28266692e-01 9.79238093e-01 5.27393520e-01 -6.84535027e-01
-2.07906175e+00 -3.25917214e-01 -1.10781729e-01 -5.96740305e-01
3.42859477e-01 -7.44971633e-01 -1.02017248e+00 3.85023266e-01
-3.20184916e-01 -4.40411456e-02 8.55560601e-02 1.16202861e-01
-2.58958548e-01 -1.31315932e-01 -1.70490527e+00 5.60462236e-01
1.43340778e+00 -1.73039168e-01 -8.79639447e-01 5.92302442e-01
6.96338713e-01 -6.88509047e-01 -1.30616176e+00 1.02549481e+00
1.30298603e+00 -6.00211024e-01 1.35220337e+00 -6.43982470e-01
6.36698663e-01 2.20175877e-01 -4.13465381e-01 -9.85904634e-01
-6.28879130e-01 -1.58509940e-01 -2.47297898e-01 9.73625064e-01
-2.44738773e-01 -5.44497550e-01 6.18895590e-01 7.91009486e-01
-1.81951314e-01 -1.10353923e+00 -7.40585446e-01 -1.22788477e+00
1.93422645e-01 1.44037545e-01 2.87198991e-01 7.32578516e-01
2.35711887e-01 -4.54468548e-01 -1.28630221e-01 -6.58869818e-02
7.01052129e-01 4.73707289e-01 7.17546463e-01 -1.60303974e+00
-5.95091224e-01 -1.76303953e-01 1.67975083e-01 -9.75171864e-01
5.34699969e-02 -6.89410925e-01 3.30300599e-01 -1.14574170e+00
4.12866384e-01 -1.40665221e+00 -6.33727247e-03 9.19030190e-01
-1.70772731e-01 -8.95749629e-02 5.17209470e-01 5.13008118e-01
-1.74436182e-01 1.77793279e-01 1.93143773e+00 -2.89383531e-01
-2.85146058e-01 3.15120131e-01 2.41587877e-01 8.16550851e-01
5.38059056e-01 -2.54189894e-02 -9.68920439e-02 -5.84291458e-01
-1.59869030e-01 1.29832029e-01 3.99024725e-01 -8.02155614e-01
-2.61466295e-01 -7.17388511e-01 3.88278782e-01 -3.78449649e-01
3.95450890e-01 -1.18557274e+00 -1.33554935e-01 7.87395597e-01
-1.53179809e-01 -5.11041045e-01 2.82006949e-01 2.30749562e-01
6.07750952e-01 -6.43524289e-01 6.94984436e-01 -3.28245550e-01
-2.82484204e-01 1.75028458e-01 1.09859541e-01 -4.88298625e-01
1.20168567e+00 -3.83864999e-01 -3.11166853e-01 3.49266887e-01
-1.06231773e+00 6.18435792e-04 3.20192307e-01 6.79412007e-01
4.29378331e-01 -1.29029262e+00 -3.74385685e-01 2.83419341e-01
-3.73130888e-01 2.52014548e-01 1.22977085e-01 3.84471774e-01
-3.65707219e-01 1.01468652e-01 -6.69321060e-01 -5.90736747e-01
-1.50403047e+00 7.22082615e-01 5.71593903e-02 8.77465755e-02
-5.77770174e-01 9.85144734e-01 -8.64454284e-02 -5.26367188e-01
7.32642591e-01 -1.04263401e+00 2.52929404e-02 -2.54539251e-01
-6.29667332e-03 7.70447612e-01 -1.35962442e-01 3.17872465e-01
-2.26167992e-01 9.95121896e-01 -4.77906577e-02 6.35309696e-01
1.46952581e+00 6.13417208e-01 -2.22218171e-01 4.09156322e-01
7.82363772e-01 -1.30312219e-01 -1.34103990e+00 8.85358825e-02
-2.94078678e-01 -6.63821936e-01 -8.72618318e-01 -1.06580901e+00
-8.96885276e-01 6.04595959e-01 5.71391642e-01 6.08481169e-01
7.27772415e-01 4.81273770e-01 6.33905828e-01 1.17106497e+00
1.12720668e+00 -1.25932336e+00 6.82556093e-01 5.57802200e-01
1.86671257e+00 -1.19490194e+00 -1.89916398e-02 -8.42524052e-01
-2.16924056e-01 1.33759677e+00 9.72006977e-01 -3.18725824e-01
7.69725621e-01 7.18214214e-01 -4.23492551e-01 -6.22027554e-02
-2.35468119e-01 4.27218169e-01 4.23361748e-01 6.17340863e-01
3.34850810e-02 4.69715804e-01 -1.51347846e-01 7.99833000e-01
-1.57359958e-01 2.59275764e-01 -1.24098584e-01 1.29835737e+00
-6.00002825e-01 -1.02818489e+00 -4.74811554e-01 7.51154363e-01
8.21061283e-02 3.36061716e-01 -8.82338211e-02 8.34996104e-01
2.35038087e-01 5.38738668e-01 1.55891599e-02 -7.00998187e-01
6.57346547e-01 -7.70982578e-02 1.16874015e+00 -3.44161570e-01
-3.80967706e-01 -2.50677317e-01 -4.65398878e-01 -6.30793154e-01
-3.04621875e-01 -3.26440364e-01 -1.29099679e+00 -4.31640208e-01
-5.06248891e-01 -2.74858624e-01 7.01734722e-01 7.99855530e-01
1.71446726e-02 1.70656294e-01 6.75687671e-01 -2.09058905e+00
-1.20267618e+00 -1.36498630e+00 -7.71963716e-01 6.31492078e-01
-2.55714823e-02 -1.41400743e+00 -3.07560802e-01 4.00056541e-02]
|
[5.8287763595581055, -0.8778322339057922]
|
8e0f81b1-7bf4-48b8-a25a-10eef30718d3
|
knowledge-graph-informed-fake-news
|
2110.10457
| null |
https://arxiv.org/abs/2110.10457v2
|
https://arxiv.org/pdf/2110.10457v2.pdf
|
Knowledge Graph informed Fake News Classification via Heterogeneous Representation Ensembles
|
Increasing amounts of freely available data both in textual and relational form offers exploration of richer document representations, potentially improving the model performance and robustness. An emerging problem in the modern era is fake news detection -- many easily available pieces of information are not necessarily factually correct, and can lead to wrong conclusions or are used for manipulation. In this work we explore how different document representations, ranging from simple symbolic bag-of-words, to contextual, neural language model-based ones can be used for efficient fake news identification. One of the key contributions is a set of novel document representation learning methods based solely on knowledge graphs, i.e. extensive collections of (grounded) subject-predicate-object triplets. We demonstrate that knowledge graph-based representations already achieve competitive performance to conventionally accepted representation learners. Furthermore, when combined with existing, contextual representations, knowledge graph-based document representations can achieve state-of-the-art performance. To our knowledge this is the first larger-scale evaluation of how knowledge graph-based representations can be systematically incorporated into the process of fake news classification.
|
['Blaž Škrlj', 'Senja Pollak', 'Marko Robnik-Šikonja', 'Timen Stepišnik-Perdih', 'Boshko Koloski']
|
2021-10-20
| null | null | null | null |
['news-classification']
|
['natural-language-processing']
|
[ 1.42329514e-01 2.06949085e-01 -5.84433973e-01 -2.28947788e-01
-8.86795104e-01 -6.40503645e-01 9.46113288e-01 8.55478823e-01
1.41426297e-02 6.37578726e-01 4.73739624e-01 -5.07494688e-01
-2.94066817e-01 -1.03245032e+00 -7.82339573e-01 -2.22488776e-01
1.27946436e-02 7.33308375e-01 1.95303142e-01 -7.23033965e-01
7.08298206e-01 5.30293763e-01 -1.54651308e+00 7.73428202e-01
5.30205727e-01 8.24108839e-01 -5.45494735e-01 1.82538569e-01
-2.57421792e-01 1.41305017e+00 -9.94810581e-01 -5.79460204e-01
-2.87427574e-01 -3.31634104e-01 -8.99100125e-01 -2.67706037e-01
6.13360286e-01 -4.16513950e-01 -6.13017082e-01 1.02322185e+00
1.55261040e-01 1.07487485e-01 9.59471524e-01 -1.05508852e+00
-1.63923943e+00 1.00845671e+00 -2.31978968e-01 4.61386472e-01
6.04459405e-01 -2.86601335e-01 1.24464965e+00 -8.14599752e-01
8.90737116e-01 1.43772995e+00 6.67612076e-01 3.07277352e-01
-1.17986655e+00 -6.02350593e-01 1.40558198e-01 5.00592887e-01
-1.07340419e+00 -2.88552076e-01 9.89663959e-01 -5.44069707e-01
1.20520926e+00 2.47790962e-01 7.67047286e-01 1.66710591e+00
8.91954750e-02 8.41733992e-01 1.25454390e+00 -6.63122714e-01
-1.06746830e-01 4.04622912e-01 6.35097742e-01 9.01482999e-01
9.28453445e-01 1.61605537e-01 -8.61392200e-01 -5.09173870e-01
3.69788915e-01 -1.60478786e-01 -3.21697384e-01 -3.96608412e-01
-1.21100748e+00 1.33430052e+00 8.14837337e-01 6.77487314e-01
-8.07243288e-02 6.00023925e-01 5.13087451e-01 5.40494323e-01
7.71953762e-01 1.07248306e+00 -1.42663717e-01 -2.49044504e-02
-8.18251729e-01 6.16538167e-01 7.67305553e-01 8.07204902e-01
4.08434242e-01 1.07089810e-01 -3.73626709e-01 6.59899950e-01
2.02132627e-01 3.27287883e-01 7.80282140e-01 -2.38057375e-01
5.94072759e-01 7.37659693e-01 1.20295808e-01 -1.81045723e+00
-3.35307598e-01 -5.61242819e-01 -3.26352090e-01 -1.19054452e-01
3.67601961e-01 5.07437706e-01 -6.91149354e-01 1.11536074e+00
-1.67359468e-02 -1.94964901e-04 7.24885315e-02 8.65396917e-01
1.14081800e+00 5.50458729e-01 -1.02617085e-01 1.62684135e-02
1.47429287e+00 -7.56924450e-01 -7.83410907e-01 -2.98905522e-01
9.27735627e-01 -5.65058708e-01 7.43154407e-01 4.23185587e-01
-6.27990603e-01 -2.11698823e-02 -1.23102593e+00 -3.17268133e-01
-1.18209314e+00 -6.74645007e-02 9.40463543e-01 7.70038188e-01
-6.58968151e-01 8.45138192e-01 -3.50129277e-01 -2.03611657e-01
7.27535307e-01 -1.28945615e-02 -5.28643012e-01 -3.17016840e-01
-1.48396051e+00 1.38722062e+00 6.15584075e-01 -2.49258250e-01
-7.45734334e-01 -4.66053277e-01 -8.82989883e-01 -4.58091386e-02
5.98340333e-01 -4.67517376e-01 1.02357733e+00 -8.13685536e-01
-1.03154147e+00 1.03427660e+00 2.59315431e-01 -6.05006695e-01
3.64012331e-01 -2.16221616e-01 -6.39033556e-01 2.18942881e-01
1.39844283e-01 6.08986542e-02 1.21383965e+00 -1.53689301e+00
1.50317281e-01 -4.66100335e-01 -4.14346792e-02 -1.52978122e-01
-5.01536369e-01 3.05282325e-01 2.77109921e-01 -1.07220876e+00
3.15640479e-01 -7.55876899e-01 3.03505898e-01 -1.22093201e-01
-5.85745931e-01 -3.04260850e-01 8.95804167e-01 -7.91240692e-01
1.21548927e+00 -1.77608347e+00 2.26268262e-01 2.12699816e-01
3.92969191e-01 4.23667818e-01 1.44147396e-01 7.22338796e-01
-7.69653991e-02 4.55890328e-01 -3.58072445e-02 -1.64682969e-01
7.81574845e-02 1.12470448e-01 -8.64421368e-01 5.95514715e-01
2.14019090e-01 1.23590636e+00 -1.26145220e+00 -2.77817190e-01
-1.62345171e-02 4.13475692e-01 -2.66354203e-01 -2.90366411e-01
-3.37992996e-01 1.36985540e-01 -5.70755303e-01 8.41928840e-01
1.91366449e-01 -2.60517746e-01 1.40786722e-01 -1.00924775e-01
2.50550032e-01 8.36234331e-01 -5.63021958e-01 1.13613975e+00
-1.76071212e-01 1.06877887e+00 -7.07625091e-01 -1.49285352e+00
9.63976502e-01 1.20412387e-01 -9.43494588e-02 -5.90989351e-01
2.06866160e-01 3.37685376e-01 -1.98299959e-01 -2.83738613e-01
1.05759871e+00 -2.65939891e-01 -2.70190179e-01 7.11013019e-01
2.40121469e-01 -4.15857732e-01 -7.03541562e-02 5.37280798e-01
9.55995142e-01 9.91748795e-02 5.71750820e-01 -4.71015833e-02
9.27984118e-02 2.61541814e-01 -1.59284875e-01 1.08649790e+00
1.43078342e-01 3.99769604e-01 8.43970537e-01 -4.12517369e-01
-7.63634384e-01 -6.81571603e-01 -1.47851303e-01 1.13095784e+00
-4.89704981e-02 -6.52167499e-01 -1.04781836e-01 -8.73116195e-01
5.92883468e-01 1.17722809e+00 -9.46129620e-01 -5.78830004e-01
-4.26707953e-01 -6.25887215e-01 8.74668598e-01 4.93538350e-01
-9.79166664e-03 -1.04693711e+00 -3.13569278e-01 3.60543460e-01
-6.23660274e-02 -8.56108606e-01 1.54331326e-01 5.86114563e-02
-9.27010953e-01 -1.06968594e+00 -2.84623206e-01 -3.66020918e-01
4.85765338e-01 5.39590359e-01 1.25868475e+00 4.79990274e-01
-1.35465398e-01 4.65269893e-01 -7.94512093e-01 -4.59435225e-01
-8.19600940e-01 -3.77747208e-01 -6.86047003e-02 -1.59803584e-01
2.69900650e-01 -2.16408819e-01 -4.49173711e-02 -9.39139128e-02
-9.02036369e-01 -2.46077269e-01 2.63676703e-01 1.09769261e+00
2.81519622e-01 6.61898628e-02 6.14869416e-01 -1.26279557e+00
1.04021871e+00 -7.31183708e-01 -2.86120534e-01 3.08747083e-01
-6.48376644e-01 1.35572970e-01 5.46699047e-01 -4.35507059e-01
-6.41696692e-01 -7.56199360e-01 4.11878914e-01 -3.64037454e-01
2.99931943e-01 9.14744437e-01 4.62602884e-01 -3.50619674e-01
1.37032902e+00 2.73997694e-01 -1.21601805e-01 -3.12275261e-01
8.85122359e-01 7.01305330e-01 2.71128863e-01 -6.87615573e-01
6.78425550e-01 4.82287973e-01 -1.01817966e-01 -8.02429378e-01
-1.11228693e+00 -4.97061819e-01 -4.97633725e-01 -1.88395046e-02
2.75819808e-01 -6.98981404e-01 -1.30924374e-01 2.75095701e-01
-1.17159355e+00 1.17242791e-01 -1.86011255e-01 1.39647901e-01
-3.79221171e-01 5.37122369e-01 -6.23097956e-01 -7.90602863e-01
-2.48763878e-02 -7.66808271e-01 1.08399642e+00 -4.01944816e-01
-2.91499764e-01 -9.83953476e-01 -2.13549718e-01 7.29158163e-01
3.17588419e-01 5.07540107e-01 1.08768976e+00 -1.13473606e+00
-4.19467330e-01 -7.51065671e-01 -3.52347821e-01 2.91543156e-01
-1.00171313e-01 -6.39240071e-02 -8.94130766e-01 -2.09491923e-01
-2.09890127e-01 -7.26680577e-01 1.32577193e+00 -1.77539125e-01
1.08024383e+00 -7.01332390e-01 -4.77033913e-01 3.85878026e-03
1.13116133e+00 -1.55647159e-01 6.03365064e-01 4.78908271e-01
9.05331552e-01 5.58091164e-01 4.00591969e-01 3.23755264e-01
3.09101075e-01 8.18840623e-01 3.78583133e-01 6.23363972e-01
-2.65400887e-01 -5.43651104e-01 3.05298567e-01 6.64552391e-01
-1.15224011e-01 -5.25104702e-01 -1.11441064e+00 5.22865176e-01
-1.92626047e+00 -1.36447144e+00 -2.14722514e-01 2.03749347e+00
6.80506766e-01 1.55347034e-01 -9.88857597e-02 3.69712323e-01
5.72064281e-01 3.73753130e-01 -7.20330551e-02 -4.78054196e-01
-2.41801575e-01 1.36522144e-01 4.68671083e-01 1.79180473e-01
-1.27656066e+00 1.22165978e+00 6.54772568e+00 6.98098421e-01
-9.18225586e-01 2.49225646e-01 9.17770565e-02 2.25585043e-01
-6.16445661e-01 -9.05181766e-02 -4.12095070e-01 2.59878248e-01
9.94361043e-01 -1.66773781e-01 4.06825960e-01 1.05428994e+00
-3.80853146e-01 1.50826126e-01 -1.15088415e+00 1.02062786e+00
6.40592694e-01 -1.91760767e+00 4.96991813e-01 1.84433255e-02
7.12654829e-01 -1.17126338e-01 1.50052384e-01 5.12838662e-01
3.68835449e-01 -1.29757786e+00 1.01882052e+00 5.78885615e-01
3.70791972e-01 -5.57630599e-01 8.54587078e-01 2.36958072e-01
-5.12475133e-01 -2.35710114e-01 -4.49102521e-01 -3.92381698e-02
-1.05706736e-01 4.93044466e-01 -8.90397549e-01 8.16842496e-01
3.60281765e-01 1.13946033e+00 -8.60550523e-01 7.35590339e-01
-5.13897240e-01 6.40114486e-01 3.95223796e-02 -5.32559931e-01
3.23318958e-01 3.16767693e-01 6.42064273e-01 1.37547493e+00
1.83387384e-01 5.34637012e-02 5.14159054e-02 9.90999937e-01
-2.03184620e-01 1.27818719e-01 -1.19036722e+00 -7.70248473e-01
4.03474778e-01 6.88609660e-01 -6.94212556e-01 -6.25572383e-01
-2.24768251e-01 8.09626520e-01 8.52892339e-01 3.79710495e-02
-5.58460712e-01 -1.06618054e-01 2.86005765e-01 9.89223942e-02
2.64159679e-01 -2.29341522e-01 -1.69792861e-01 -1.37170959e+00
-1.21294141e-01 -1.06135213e+00 5.36555290e-01 -9.57672834e-01
-1.75068963e+00 3.60670537e-01 1.94230780e-01 -1.09593785e+00
-2.32014999e-01 -9.71764445e-01 -1.72571793e-01 2.70487756e-01
-1.36058021e+00 -1.39675009e+00 -1.20540764e-02 4.93171662e-01
2.98783667e-02 -4.25992697e-01 1.18552363e+00 -1.22062571e-01
-7.66517147e-02 4.24509704e-01 2.92420715e-01 3.34513992e-01
5.48675716e-01 -1.18020117e+00 2.62401968e-01 4.71548438e-01
7.78651476e-01 8.67272675e-01 6.43127322e-01 -7.85990953e-01
-1.47728336e+00 -8.05654645e-01 9.18772161e-01 -1.09186685e+00
1.14338076e+00 -3.50548148e-01 -1.15206313e+00 8.40762913e-01
-3.04612458e-01 2.55735785e-01 8.24657679e-01 3.38294417e-01
-1.35071659e+00 3.84072274e-01 -1.18528235e+00 3.83266985e-01
9.80502725e-01 -9.76637900e-01 -1.19937813e+00 9.53070581e-01
6.92327857e-01 -5.22931397e-01 -5.91859341e-01 -2.03129835e-03
5.19252002e-01 -9.12752986e-01 1.17405868e+00 -1.32089663e+00
6.78150415e-01 -7.50064403e-02 -2.48061597e-01 -1.45939302e+00
-3.84072691e-01 -3.21274906e-01 -5.48629165e-01 8.60740483e-01
3.87433320e-01 -9.16443706e-01 2.73887157e-01 4.03146185e-02
-1.34960383e-01 -5.80179214e-01 -1.03702867e+00 -1.00189245e+00
2.87736207e-01 -6.28904581e-01 4.38754261e-01 1.48759997e+00
4.60703313e-01 2.77680546e-01 -5.31950712e-01 1.56800970e-02
5.04070938e-01 2.37722680e-01 4.17146146e-01 -1.39026678e+00
-2.73015290e-01 -6.66472137e-01 -1.03729653e+00 -6.03503585e-01
4.10147280e-01 -1.38055885e+00 -2.91618645e-01 -1.71623921e+00
-1.78535040e-02 -2.58134961e-01 -3.05250973e-01 6.39960647e-01
4.90259193e-02 2.55359262e-01 2.18463898e-01 2.92375743e-01
-3.73045146e-01 3.82392794e-01 1.09211087e+00 -5.52615464e-01
2.49189869e-01 -2.74872839e-01 -9.81180370e-01 5.74010491e-01
7.45124578e-01 -8.06523919e-01 -3.05049151e-01 -2.16867656e-01
5.06595790e-01 -5.42299487e-02 7.07326889e-01 -5.68311095e-01
3.93603649e-03 4.63115796e-02 2.69409239e-01 -1.78449601e-01
4.63270575e-01 -4.95181948e-01 -2.42844626e-01 2.69055337e-01
-5.16225398e-01 -1.61942184e-01 2.09190786e-01 1.04383707e+00
-2.65158594e-01 -3.92832816e-01 4.03880596e-01 -4.02866691e-01
-5.64240694e-01 -2.06794217e-01 -3.36505383e-01 1.13283925e-01
7.83913195e-01 -1.31605104e-01 -1.12701881e+00 -5.25395691e-01
-4.45300907e-01 -5.11197865e-01 1.59784377e-01 7.12371051e-01
8.62653077e-01 -1.43361413e+00 -9.08410311e-01 -1.53777137e-01
6.71697974e-01 -6.84578300e-01 -1.31850407e-01 5.06889880e-01
-5.54680228e-01 6.40029848e-01 -3.22847604e-03 -1.61181346e-01
-1.02881289e+00 8.12788546e-01 8.43225271e-02 -2.29988679e-01
-8.76284957e-01 8.69750261e-01 -4.38652873e-01 -3.93058687e-01
-1.19169779e-01 -4.38955039e-01 -2.70082772e-01 4.42685515e-01
5.59847951e-01 2.44948924e-01 2.47169256e-01 -9.17442679e-01
-5.20767510e-01 2.08857551e-01 -2.62258440e-01 9.98158823e-04
1.25928247e+00 5.35778999e-01 -4.11717951e-01 6.23925626e-01
1.07029057e+00 2.66093258e-02 -2.47440204e-01 -2.84746349e-01
2.05579594e-01 -8.57065856e-01 3.16361547e-01 -1.13441849e+00
-6.79485798e-01 6.50317013e-01 -1.25804991e-02 6.32048070e-01
3.32452595e-01 3.12432885e-01 2.83560991e-01 8.24655533e-01
6.11128926e-01 -8.59838843e-01 5.13532698e-01 4.44666922e-01
1.37387061e+00 -1.33560085e+00 4.54252571e-01 -4.18813586e-01
-6.35365486e-01 1.36863995e+00 1.16758697e-01 -3.24462295e-01
4.25789177e-01 -4.31538165e-01 -2.66880304e-01 -8.04021239e-01
-4.66118872e-01 1.33267539e-02 7.26923585e-01 7.58313835e-01
5.49698412e-01 2.36786693e-01 -1.23973742e-01 5.75167120e-01
-4.13481623e-01 -3.40665579e-01 8.40495408e-01 1.03924179e+00
-4.26871032e-01 -8.42920482e-01 -4.59139407e-01 7.25871205e-01
-3.61872852e-01 -2.74875402e-01 -9.03768480e-01 9.52324748e-01
-1.55456960e-01 1.03626275e+00 -2.96071470e-01 -2.21634120e-01
2.64247298e-01 2.52838284e-01 6.93610013e-01 -8.55889976e-01
-7.41164804e-01 -7.14693725e-01 6.04994297e-01 -5.06253064e-01
-3.79056394e-01 -5.31114399e-01 -9.42496181e-01 -3.78128976e-01
-6.57484531e-01 6.98216185e-02 7.16045022e-01 1.03071904e+00
2.76437163e-01 4.27504390e-01 -6.38878718e-03 -7.68465698e-01
-7.35787511e-01 -9.28294778e-01 -5.13585985e-01 5.20700872e-01
5.08438408e-01 -1.07823408e+00 -4.70176101e-01 -4.08024192e-01]
|
[8.183954238891602, 10.237748146057129]
|
57362925-bfa4-4e3b-a007-158715d4250f
|
sonnet-generation-by-training-on-non-poetic
| null | null |
https://openreview.net/forum?id=6wKqI-x0Vb
|
https://openreview.net/pdf?id=6wKqI-x0Vb
|
Sonnet Generation by Training on Non-poetic Texts with Discourse-level Coherence and Poetic Features
|
Poetry generation, and creative language generation in general, usually suffers from the lack of large training data. In this paper, we present a novel framework to generate sonnets that does not require training on poems. We design a hierarchical framework which plans the poem sketch before decoding. Specifically, a content planning module is trained on non-poetic texts to obtain discourse-level coherence; then a rhyme module generates rhyme words and a polishing module introduces imagery and similes for aesthetics purposes. Finally, we design a constrained decoding algorithm to impose the meter-and-rhyme constraint of the generated sonnets. Automatic and human evaluation show that our multi-stage approach without training on poem corpora generates more coherent, poetic, and creative sonnets than several strong baselines.
|
['Anonymous']
|
2022-01-16
| null | null | null |
acl-arr-january-2022-1
|
['sonnet-generation']
|
['natural-language-processing']
|
[ 2.90807098e-01 4.78962898e-01 9.29595530e-02 3.36795785e-02
-6.73825085e-01 -5.51842391e-01 1.09297431e+00 -4.25551981e-01
-1.19069135e-02 6.80870593e-01 8.20588946e-01 -3.02518271e-02
3.28853160e-01 -1.04500246e+00 -4.06679213e-01 -4.11970645e-01
3.74374419e-01 7.47226894e-01 4.65953834e-02 -6.01754725e-01
3.27259392e-01 3.87067646e-02 -1.02049088e+00 5.02533913e-01
8.77121627e-01 6.07971191e-01 4.70169395e-01 6.99030101e-01
-2.63366282e-01 1.09481871e+00 -9.26214337e-01 -6.65517628e-01
1.18542142e-01 -1.34484661e+00 -7.96207070e-01 1.29968673e-01
-2.15482675e-02 -8.47774968e-02 -9.83994901e-02 6.04001641e-01
6.27410114e-01 2.78251588e-01 7.34564364e-01 -7.30045557e-01
-5.96673369e-01 1.47037363e+00 -2.20621705e-01 -2.61373073e-01
3.75576079e-01 8.40092674e-02 1.54672313e+00 -8.33466530e-01
9.63067710e-01 1.00273597e+00 6.71145618e-01 7.19554007e-01
-1.39186478e+00 -3.64776522e-01 -4.76633132e-01 4.38440070e-02
-1.27946794e+00 -7.00322390e-01 1.31289482e+00 -2.40527406e-01
9.36463714e-01 4.15293664e-01 1.12558603e+00 1.17226779e+00
-2.03794688e-01 1.01416016e+00 8.13418090e-01 -6.62016153e-01
2.17386052e-01 -3.40446800e-01 -6.88935518e-01 4.67886895e-01
-5.53292394e-01 -6.58189431e-02 -8.49331141e-01 3.43819380e-01
8.48470509e-01 -7.55533218e-01 -6.69740960e-02 3.66124332e-01
-1.28972030e+00 8.08316648e-01 1.43088371e-01 7.28122354e-01
-5.21979749e-01 2.73260087e-01 4.80910778e-01 1.04510769e-01
1.90729558e-01 1.16342163e+00 2.95202643e-01 -4.42400426e-01
-1.64263558e+00 4.24631745e-01 9.30010319e-01 1.01201248e+00
4.07575816e-01 3.08970571e-01 -3.71279478e-01 1.07694530e+00
3.43453661e-02 1.24411605e-01 5.76910496e-01 -8.82115006e-01
5.28234482e-01 1.36867091e-01 -3.00216854e-01 -8.47077847e-01
-2.46364176e-01 -3.55103165e-01 -6.55999422e-01 -2.03834161e-01
2.86831204e-02 -1.15798712e-01 -5.93177319e-01 1.52884781e+00
-1.23134091e-01 8.94875824e-02 5.28900288e-02 8.65714490e-01
1.02573156e+00 1.04069316e+00 1.04016609e-01 -3.14866155e-01
1.32759154e+00 -1.36533952e+00 -8.40253830e-01 -2.71123409e-01
3.03204775e-01 -1.01915538e+00 1.54121220e+00 3.28868032e-01
-1.81353271e+00 -4.76693600e-01 -1.28146160e+00 -4.30578589e-01
2.78986841e-01 2.14961961e-01 5.11092365e-01 1.93213329e-01
-6.84755087e-01 9.13152635e-01 -6.45810425e-01 -1.60617650e-01
3.85609269e-01 -2.90908724e-01 -1.67724434e-02 3.68039638e-01
-9.60277200e-01 9.43154633e-01 7.13320255e-01 -1.15542002e-01
-5.50849915e-01 -5.22032797e-01 -9.17126954e-01 1.19869530e-01
-1.40766865e-02 -9.69909668e-01 1.46778870e+00 -9.85350072e-01
-2.11684752e+00 1.10223055e+00 1.81864843e-01 -3.84813130e-01
6.23068988e-01 8.67255554e-02 -8.98148715e-02 2.40471616e-01
-8.16289559e-02 7.66548574e-01 5.44596732e-01 -1.13100553e+00
-4.61304605e-01 2.95926780e-01 -1.17419116e-01 5.01363277e-01
-5.17411530e-02 1.84939742e-01 -5.56592822e-01 -1.01973748e+00
1.66295364e-01 -6.17567062e-01 -5.94307035e-02 -6.70462549e-01
-6.60429358e-01 -1.91339970e-01 1.49469778e-01 -8.50702763e-01
1.53099132e+00 -1.90935075e+00 2.79993385e-01 8.85373950e-02
-6.71267211e-02 -7.80632123e-02 -3.06063086e-01 7.49196172e-01
3.06474596e-01 2.51785852e-02 -2.85561323e-01 -6.32533908e-01
5.30660808e-01 2.38371864e-01 -4.12810385e-01 -5.35990298e-02
3.61522168e-01 1.35925746e+00 -8.86813760e-01 -7.10361004e-01
4.75315861e-02 3.59587908e-01 -8.61627102e-01 2.98257738e-01
-7.47129679e-01 5.70129275e-01 -3.44411992e-02 3.79496872e-01
1.48091286e-01 -2.31922790e-02 4.32628721e-01 5.88725992e-02
-3.34569722e-01 1.19540381e+00 -7.45202124e-01 2.12008595e+00
-7.89907575e-01 7.56176770e-01 -4.45979923e-01 -6.67620420e-01
1.01155198e+00 5.05058765e-01 1.17138356e-01 -8.72615039e-01
4.11105603e-01 2.21082762e-01 1.97526604e-01 -3.09062123e-01
9.82234836e-01 -7.88927317e-01 -4.29702669e-01 6.80524349e-01
1.74744129e-01 -9.04686451e-01 5.47525883e-01 -7.72639140e-02
1.09877622e+00 5.49677253e-01 5.54751337e-01 -2.29929574e-02
3.14821422e-01 2.48128206e-01 5.24486065e-01 3.20104897e-01
5.77218831e-01 9.77529645e-01 6.23769820e-01 -2.83027738e-01
-1.30857289e+00 -1.22105825e+00 2.66562313e-01 1.09220123e+00
-1.31345898e-01 -1.04629755e+00 -1.05140841e+00 -8.85603875e-02
-7.60122359e-01 1.23093247e+00 -3.05012256e-01 3.00409019e-01
-1.14101529e+00 -4.64557499e-01 8.21503997e-01 3.95365864e-01
4.49998826e-01 -1.76927090e+00 -7.23848224e-01 5.63182294e-01
-5.20758510e-01 -1.10196984e+00 -5.41010618e-01 -2.87290752e-01
-6.13519728e-01 -7.58011699e-01 -4.32000011e-01 -1.15574431e+00
3.63177508e-01 -3.99665177e-01 1.44788432e+00 1.34845659e-01
7.29676932e-02 -4.16976273e-01 -4.43001509e-01 -2.55534500e-01
-8.52859855e-01 3.36944431e-01 -6.32047951e-01 -4.78697032e-01
-4.75880876e-02 -1.04656053e+00 -2.74372369e-01 -1.37415990e-01
-6.70885086e-01 9.95764077e-01 4.79564428e-01 8.57488751e-01
5.64523220e-01 -1.54616505e-01 4.31598336e-01 -9.86306012e-01
7.36524880e-01 -5.48101179e-02 -3.08129311e-01 -1.99571997e-02
-1.45539641e-01 -1.00019999e-01 7.94715226e-01 -3.93438876e-01
-1.08591795e+00 -7.26744011e-02 -5.93591392e-01 1.94530249e-01
2.60469526e-01 5.61860621e-01 -4.72977340e-01 6.36403501e-01
9.12685931e-01 2.09011033e-01 -4.60076451e-01 -4.43913788e-01
7.57181168e-01 5.10938644e-01 9.86246169e-01 -7.54017234e-01
1.04710984e+00 9.30808410e-02 -1.44360557e-01 -8.71257842e-01
-7.38865733e-01 1.37761369e-01 -5.12233973e-01 -2.91316304e-02
8.19152653e-01 -7.86007047e-01 -3.43246579e-01 2.98222266e-02
-1.50755620e+00 -7.73752749e-01 -9.31830704e-01 2.62285352e-01
-1.17390466e+00 4.22430858e-02 -1.00077844e+00 -4.83756751e-01
-7.29793727e-01 -6.16227627e-01 7.26842642e-01 2.37588018e-01
-9.98444736e-01 -9.10927474e-01 3.21137041e-01 3.28220546e-01
2.24454433e-01 4.74518806e-01 9.22443509e-01 -2.77992427e-01
-5.95566452e-01 2.19252437e-01 4.94544161e-03 6.00148886e-02
7.65510052e-02 -1.35267571e-01 -8.75665188e-01 3.91577065e-01
-1.55791134e-01 -2.39654168e-01 6.45012021e-01 1.14557268e-02
5.15666366e-01 -8.36848676e-01 1.54547960e-01 1.01217914e+00
1.06401932e+00 -9.87733826e-02 1.03542888e+00 3.81092638e-01
6.38394177e-01 4.90171641e-01 3.24233860e-01 6.95296049e-01
6.41439676e-01 5.70342898e-01 -2.39597276e-01 4.77732532e-03
-6.90152645e-01 -9.76865053e-01 5.42093396e-01 1.38960409e+00
-4.35088307e-01 -4.18321878e-01 -7.46956766e-01 8.64851773e-01
-1.69652903e+00 -1.40249550e+00 -5.04095517e-02 1.51844597e+00
1.35084820e+00 7.49920532e-02 2.79827982e-01 2.90855795e-01
4.40524012e-01 4.89851266e-01 2.03004971e-01 -4.38755304e-01
-3.69361937e-01 9.08993661e-01 -2.03827068e-01 6.59748554e-01
-7.54721940e-01 1.72612548e+00 6.08459187e+00 9.74922836e-01
-9.86701012e-01 2.74278671e-01 1.87575072e-01 -1.08419292e-01
-7.34191298e-01 2.38807306e-01 -3.51351410e-01 3.15384507e-01
5.73133469e-01 -3.29899639e-01 5.93153119e-01 5.95730245e-01
3.46770734e-01 1.72158599e-01 -1.02931046e+00 9.86291409e-01
2.52570361e-01 -1.88423002e+00 -1.38153583e-01 -2.04455242e-01
8.84206116e-01 -2.74436623e-01 -3.13166946e-01 1.50335640e-01
3.55306178e-01 -1.07660437e+00 1.26415265e+00 3.47477794e-01
9.92728174e-01 -7.17581868e-01 1.36717573e-01 3.38859618e-01
-1.17576098e+00 4.61630255e-01 -1.57719091e-01 -3.99689555e-01
7.12924182e-01 5.41157015e-02 -1.07006001e+00 2.04278633e-01
-2.07906187e-01 6.57649875e-01 -3.16896915e-01 8.99143934e-01
-1.34174097e+00 8.99129212e-01 -2.33589768e-01 -1.45818636e-01
2.27516785e-01 -2.74229765e-01 6.52282059e-01 1.65317428e+00
4.50238645e-01 1.96316198e-01 5.71512058e-02 1.36705267e+00
-1.28944322e-01 6.13781095e-01 -3.36967409e-01 -3.58612984e-01
4.41423476e-01 1.42733872e+00 -9.44204986e-01 -1.76669836e-01
-3.51172723e-02 1.52131176e+00 2.38454789e-01 1.75910786e-01
-6.98983848e-01 -3.54263872e-01 1.32699832e-01 4.16548371e-01
3.03767622e-01 -4.21694636e-01 -7.61817276e-01 -1.07113266e+00
-1.01382926e-01 -6.62038624e-01 2.04196393e-01 -8.92366648e-01
-1.04990232e+00 7.33231843e-01 -4.49251801e-01 -1.05020893e+00
-8.06303322e-01 -6.36092946e-02 -1.30365598e+00 7.04125464e-01
-1.34152186e+00 -1.66450047e+00 -3.33016962e-02 4.52480540e-02
7.49460995e-01 -2.16700330e-01 9.00008500e-01 2.46769469e-02
-1.66665718e-01 4.42883193e-01 -5.57835519e-01 4.49714005e-01
4.17715281e-01 -1.32024467e+00 8.38807464e-01 9.49278116e-01
6.25050664e-01 3.81208301e-01 9.13612247e-01 -5.25586247e-01
-8.10642004e-01 -7.49662399e-01 1.62936437e+00 -1.44181401e-01
7.09767759e-01 -4.71754104e-01 -4.58883405e-01 6.26854718e-01
6.14120305e-01 -5.64941525e-01 7.70936012e-01 1.02592260e-01
-1.20251596e-01 3.53356361e-01 -6.92519486e-01 9.78823185e-01
1.29955506e+00 -4.14745092e-01 -1.08416235e+00 2.52605826e-01
5.63515306e-01 -6.11136854e-01 -7.87831783e-01 -7.05381632e-02
4.73322630e-01 -8.57769132e-01 6.36769950e-01 -2.74455398e-01
1.13254952e+00 -4.20814782e-01 8.11854675e-02 -1.47481918e+00
-3.57076913e-01 -1.39467776e+00 9.43370908e-02 1.47374749e+00
5.25601923e-01 2.29939893e-01 7.17784107e-01 6.51398376e-02
-6.07187510e-01 -2.62710303e-01 -6.14414275e-01 -5.96596479e-01
1.13295745e-02 -4.79393244e-01 6.22223377e-01 9.35898006e-01
5.33755004e-01 1.12246227e+00 -3.31438065e-01 -4.36839163e-01
2.59596497e-01 2.33389258e-01 8.93360853e-01 -8.80481124e-01
-8.04358423e-01 -9.14411843e-01 -3.20736803e-02 -9.65747237e-01
2.77872771e-01 -1.30955350e+00 3.40153813e-01 -1.86195993e+00
-1.25345305e-01 -2.89052039e-01 2.65910298e-01 4.75751013e-01
3.83889191e-02 6.93365216e-01 7.11914420e-01 3.15199137e-01
-4.59874749e-01 7.52095759e-01 1.49088562e+00 6.52899221e-02
-7.77099252e-01 -3.31684917e-01 -7.36398637e-01 8.23384583e-01
8.84174883e-01 -4.02431428e-01 -2.17414781e-01 -3.96666586e-01
5.16235530e-01 9.10038203e-02 1.18172199e-01 -1.06435835e+00
1.41477451e-01 -2.47969493e-01 2.89876580e-01 -5.28766155e-01
3.48000079e-01 3.29144448e-02 4.21856493e-01 2.10107833e-01
-3.11636686e-01 -6.20406941e-02 -8.10805336e-02 -4.27083135e-01
-2.02342287e-01 -4.33271855e-01 8.28245163e-01 -4.30465698e-01
-3.68506074e-01 -1.50512859e-01 -3.80108595e-01 3.73722434e-01
4.76154834e-01 -2.29992479e-01 -9.17422473e-02 -3.70881766e-01
-6.38309598e-01 -6.81141838e-02 5.80256343e-01 2.76789069e-01
5.80363691e-01 -1.69248629e+00 -1.03009164e+00 2.32554842e-02
-1.87920071e-02 1.55922532e-01 -6.17588125e-02 5.26620090e-01
-1.01263988e+00 -1.71817362e-01 -2.75242746e-01 -1.31741911e-02
-8.36026788e-01 7.39379004e-02 1.20910831e-01 -5.11037827e-01
-9.09743786e-01 7.39796579e-01 1.03846513e-01 -2.25395963e-01
-1.11000508e-01 4.84387539e-02 -1.66241333e-01 2.82440782e-01
5.95260799e-01 3.02559614e-01 -3.71289045e-01 -7.22330153e-01
1.50054500e-01 3.27343941e-01 4.57189769e-01 -9.21643496e-01
1.51237285e+00 7.91883022e-02 -2.70100534e-01 4.28916931e-01
5.91309607e-01 5.19703805e-01 -1.17488706e+00 5.27875088e-02
3.82078558e-01 -4.84844409e-02 -2.14880317e-01 -8.06717277e-01
-5.04916251e-01 8.60197484e-01 -6.15719199e-01 -6.16107583e-02
1.07546842e+00 6.48909733e-02 1.25612438e+00 1.40378669e-01
-5.56479208e-02 -1.54179990e+00 2.22979367e-01 9.25740361e-01
1.21993911e+00 -4.15294021e-01 -2.66306698e-01 -2.85580218e-01
-1.04526079e+00 1.19685876e+00 3.87436807e-01 -3.61818820e-01
2.81976466e-03 3.86652052e-01 -7.99996592e-03 -1.64555743e-01
-7.29666829e-01 -4.75654989e-01 3.61806005e-01 4.87436175e-01
8.71128201e-01 1.94377929e-01 -8.98395479e-01 1.17441034e+00
-1.52273166e+00 -3.88016403e-02 5.69347620e-01 4.97658402e-01
-5.30033827e-01 -1.33338606e+00 -1.00210600e-01 -3.03631663e-01
-2.67761737e-01 -5.53958356e-01 -5.12794673e-01 6.20731950e-01
4.74583387e-01 6.96562111e-01 3.63358617e-01 -4.04052079e-01
3.52613062e-01 1.51564732e-01 9.97740388e-01 -9.66955125e-01
-9.29946244e-01 6.19737208e-01 6.65265501e-01 -1.98323235e-01
-2.18452245e-01 -5.94887793e-01 -1.68727088e+00 -3.99512827e-01
7.73363337e-02 1.40932724e-01 4.52473015e-01 9.70310807e-01
-6.56111836e-02 6.71756387e-01 4.46008176e-01 -1.02118528e+00
4.69999462e-02 -1.14946115e+00 -5.59619784e-01 4.36698526e-01
-2.54657120e-01 7.45598003e-02 1.09027721e-01 3.57098550e-01]
|
[11.591136932373047, 9.377100944519043]
|
6e6f544f-c61a-467a-84aa-4747a5687a66
|
sparse-representation-based-classification
|
1607.04942
| null |
http://arxiv.org/abs/1607.04942v1
|
http://arxiv.org/pdf/1607.04942v1.pdf
|
Sparse Representation-Based Classification: Orthogonal Least Squares or Orthogonal Matching Pursuit?
|
Spare representation of signals has received significant attention in recent
years. Based on these developments, a sparse representation-based
classification (SRC) has been proposed for a variety of classification and
related tasks, including face recognition. Recently, a class dependent variant
of SRC was proposed to overcome the limitations of SRC for remote sensing image
classification. Traditionally, greedy pursuit based method such as orthogonal
matching pursuit (OMP) are used for sparse coefficient recovery due to their
simplicity as well as low time-complexity. However, orthogonal least square
(OLS) has not yet been widely used in classifiers that exploit the sparse
representation properties of data. Since OLS produces lower signal
reconstruction error than OMP under similar conditions, we hypothesize that
more accurate signal estimation will further improve the classification
performance of classifiers that exploiting the sparsity of data. In this paper,
we present a classification method based on OLS, which implements OLS in a
classwise manner to perform the classification. We also develop and present its
kernelized variant to handle nonlinearly separable data. Based on two
real-world benchmarking hyperspectral datasets, we demonstrate that class
dependent OLS based methods outperform several baseline methods including
traditional SRC and the support vector machine classifier.
|
['Saurabh Prasad', 'Minshan Cui']
|
2016-07-18
| null | null | null | null |
['sparse-representation-based-classification', 'remote-sensing-image-classification']
|
['computer-vision', 'miscellaneous']
|
[ 6.18205369e-01 -3.95265937e-01 -1.52484626e-01 -4.95904744e-01
-5.73977888e-01 -1.44028604e-01 2.80705124e-01 -5.68465330e-02
-1.16949817e-02 7.39761233e-01 -2.17289627e-02 -4.24102247e-02
-3.88159961e-01 -5.75688064e-01 -2.94272363e-01 -1.00056612e+00
-1.12945147e-01 -1.95513010e-01 -7.56463408e-02 -1.46093350e-02
2.60540634e-01 8.48573744e-01 -1.87408185e+00 2.76702613e-01
9.99288440e-01 1.25122714e+00 2.09129021e-01 -3.26741561e-02
3.51674892e-02 5.09572268e-01 -1.66260257e-01 7.92018473e-02
4.03842658e-01 -4.15367544e-01 -6.40177503e-02 2.12023094e-01
4.53390718e-01 2.56001472e-01 -4.09759581e-01 9.94759440e-01
5.97404778e-01 3.52366030e-01 7.59928286e-01 -1.08029270e+00
-5.51272571e-01 3.23287696e-01 -8.91084194e-01 2.38182172e-01
2.27950558e-01 -2.51184523e-01 7.08710194e-01 -1.29654443e+00
2.13461325e-01 8.93998265e-01 8.67204905e-01 1.76305130e-01
-1.21333599e+00 -8.11193287e-01 -1.19133696e-01 4.75798815e-01
-1.61396098e+00 -7.41521239e-01 1.19452596e+00 -3.65839362e-01
7.32956767e-01 4.55451280e-01 4.26905781e-01 8.43541563e-01
-2.78603584e-02 7.54310668e-01 1.46450770e+00 -6.27772272e-01
4.38298821e-01 3.17805141e-01 1.66892365e-01 6.29414320e-01
3.58776897e-01 4.37087640e-02 -7.29222894e-01 -3.60703766e-01
4.04879451e-01 2.79258728e-01 -7.03418434e-01 -4.64615881e-01
-9.12050724e-01 1.06403756e+00 2.88803220e-01 3.50271463e-01
-6.98497474e-01 -2.65885532e-01 1.18369997e-01 2.21513376e-01
6.94038093e-01 2.72724777e-01 -1.82811171e-01 3.44302684e-01
-1.35065401e+00 -8.44831839e-02 7.90117741e-01 2.97677457e-01
7.86622822e-01 5.57209671e-01 -1.45754337e-01 1.23665035e+00
3.68554115e-01 6.63642704e-01 7.22544670e-01 -5.68681180e-01
1.66390225e-01 4.97681707e-01 -3.26476514e-01 -1.41403496e+00
-3.03737551e-01 -9.88225043e-01 -1.18652892e+00 -2.61320192e-02
-4.53349128e-02 1.70029208e-01 -7.33337402e-01 1.27495313e+00
3.09231281e-01 8.46963882e-01 3.94313663e-01 9.88143504e-01
8.11045110e-01 6.60843372e-01 -1.51690632e-01 -4.03573364e-01
1.10195875e+00 -5.94786942e-01 -5.34742296e-01 -8.60898271e-02
3.95690531e-01 -7.62426019e-01 5.57460666e-01 6.31967425e-01
-4.81266081e-01 -3.82087380e-01 -9.63897467e-01 4.96647000e-01
-8.66179615e-02 5.30974388e-01 9.18764949e-01 9.17552650e-01
-6.73722863e-01 5.56679666e-01 -7.11005867e-01 -4.17310536e-01
5.98459721e-01 2.56471813e-01 -7.29705930e-01 -3.66442561e-01
-8.73392642e-01 5.12367666e-01 1.14134967e-01 3.79719347e-01
-4.38880950e-01 -7.26666749e-01 -1.10245049e+00 8.79122168e-02
1.84229702e-01 -1.67124167e-01 5.01879513e-01 -9.24153209e-01
-1.68800378e+00 6.70136273e-01 -3.34905893e-01 -3.59408170e-01
-1.89085409e-01 -1.77359935e-02 -5.22634029e-01 3.46546054e-01
6.55459613e-02 1.50142476e-01 1.40532064e+00 -9.01775062e-01
-2.77710021e-01 -5.68318963e-01 -5.32595932e-01 -7.66053945e-02
-6.17034912e-01 1.63898263e-02 1.41114682e-01 -8.68591726e-01
7.11242199e-01 -1.03572440e+00 -1.06773458e-01 9.41206980e-03
-2.44065344e-01 -1.09199464e-01 1.06170845e+00 -5.96353114e-01
7.87305355e-01 -2.57476759e+00 1.35881558e-01 3.23454171e-01
-1.08183518e-01 3.64331245e-01 -2.15424478e-01 3.51800025e-01
-5.17462909e-01 -5.36981046e-01 -7.39770830e-01 -3.04362953e-01
-4.24675941e-01 6.98656403e-03 -5.26996851e-01 1.07729065e+00
3.62729251e-01 4.46673870e-01 -4.96287435e-01 -1.14487499e-01
1.84064806e-01 6.85206711e-01 -4.82090175e-01 6.64741397e-02
2.23560929e-01 5.10506094e-01 -3.44146729e-01 9.95440543e-01
9.23614502e-01 -2.18024433e-01 -6.33901358e-02 -4.42918599e-01
-1.70818180e-01 -3.45060229e-01 -1.35500407e+00 1.59721410e+00
-3.70274037e-01 6.73288047e-01 2.16931164e-01 -1.80208504e+00
1.21087694e+00 3.15812916e-01 6.19524121e-01 -3.04970920e-01
-1.86273009e-01 5.08995235e-01 -1.64464965e-01 -5.84609747e-01
1.15125075e-01 -3.89605403e-01 6.17335498e-01 3.15328799e-02
-1.70866251e-01 2.47341990e-02 -2.16904044e-01 -1.51717186e-01
8.71824145e-01 1.08426243e-01 5.96279383e-01 -2.69736052e-01
8.31580818e-01 4.17816415e-02 6.75854743e-01 3.16311359e-01
-6.74032122e-02 6.35095835e-01 -9.82073545e-02 -2.47116745e-01
-3.41743022e-01 -4.81113821e-01 -6.14326119e-01 7.13571727e-01
-9.00100693e-02 -1.70706888e-03 -1.14768334e-01 -4.46963340e-01
3.02709550e-01 5.78733265e-01 -3.61089617e-01 -8.29736292e-02
-3.98362726e-01 -1.04315281e+00 3.63089293e-01 2.77309269e-01
5.61032474e-01 -8.50068033e-01 -3.94624799e-01 1.23619355e-01
2.47138470e-01 -1.16671193e+00 7.14057162e-02 2.85795599e-01
-1.05096459e+00 -1.00010121e+00 -9.95638609e-01 -5.38176656e-01
7.26794660e-01 9.83411729e-01 2.90592492e-01 -2.38250390e-01
-4.48782742e-01 3.88314635e-01 -6.42273486e-01 -2.35576198e-01
1.07459843e-01 -1.86608836e-01 2.51110107e-01 8.11049283e-01
3.25455487e-01 -7.34552860e-01 -4.29791301e-01 1.85867831e-01
-9.50075448e-01 -2.58444428e-01 8.76211345e-01 9.88446057e-01
5.10143101e-01 2.65649766e-01 6.01881146e-01 -7.70941436e-01
2.52103835e-01 -8.62478375e-01 -5.11486053e-01 3.22168134e-02
-4.99695927e-01 3.16007920e-02 6.92577660e-01 -3.53626162e-01
-9.10912395e-01 3.10054481e-01 1.26911700e-01 -7.55862117e-01
-1.50142103e-01 1.07917070e+00 8.93868208e-02 -6.37584448e-01
7.24211872e-01 7.63291001e-01 3.21110219e-01 -7.30382740e-01
-1.42287672e-01 9.57184255e-01 1.63679719e-01 -3.81675363e-01
1.02924764e+00 6.81475461e-01 4.39105570e-01 -1.55492043e+00
-9.01997030e-01 -9.44276035e-01 -2.34013468e-01 1.36875600e-01
3.40914220e-01 -1.21733701e+00 -2.62219876e-01 3.21260840e-01
-7.66599178e-01 1.37160912e-01 -2.70848960e-01 9.29880798e-01
-2.31413022e-01 7.16792941e-01 -1.66998412e-02 -1.02479196e+00
-3.25221896e-01 -9.34796274e-01 1.02419770e+00 2.60768920e-01
2.55472511e-01 -7.40225077e-01 -3.37423161e-02 4.00461227e-01
6.82512283e-01 2.15866551e-01 5.43967426e-01 -7.10127056e-01
-2.23654777e-01 -2.62443274e-01 -4.43290114e-01 5.50497532e-01
2.80375391e-01 -3.57626975e-01 -1.13388562e+00 -5.84005117e-01
3.28336865e-01 -1.86772153e-01 1.09431708e+00 4.96464998e-01
1.08326781e+00 -1.02837704e-01 -4.30157661e-01 1.05571938e+00
1.65853095e+00 -8.38763192e-02 6.61878884e-01 3.18135440e-01
5.60563743e-01 7.82875597e-01 4.59563017e-01 5.72592199e-01
-8.21456239e-02 7.86410511e-01 2.96631306e-01 -6.76018298e-02
9.44062881e-03 1.23235106e-01 4.66445804e-01 5.49166262e-01
-1.43622860e-01 3.68929058e-01 -6.00486219e-01 3.00759256e-01
-1.74517143e+00 -1.11552489e+00 -2.91767091e-01 2.22408295e+00
4.64607984e-01 -5.36665618e-01 -1.61191359e-01 4.79953408e-01
5.13457417e-01 2.30554283e-01 -6.13190591e-01 1.46911770e-01
-4.45889711e-01 5.41748643e-01 3.96582872e-01 1.72234494e-02
-1.12265813e+00 7.22183228e-01 5.72734213e+00 8.23421776e-01
-1.55747485e+00 2.62752175e-02 3.29496890e-01 3.10436755e-01
-1.75351242e-03 4.39435840e-02 -7.84072161e-01 3.43248308e-01
6.66312218e-01 1.31678179e-01 3.56021047e-01 1.01292527e+00
3.20374250e-01 2.72339676e-02 -7.13568687e-01 1.44075513e+00
4.24642593e-01 -1.14704394e+00 -2.15349980e-02 3.63670960e-02
8.01760495e-01 -1.35111198e-01 2.07990304e-01 3.11979502e-01
-4.95732516e-01 -1.09171772e+00 3.71563435e-01 4.41030622e-01
7.72153556e-01 -4.18388724e-01 7.15916455e-01 4.30743754e-01
-1.21440947e+00 -2.84782469e-01 -7.53722072e-01 -1.50055096e-01
-2.48562902e-01 1.22105777e+00 -5.22259474e-01 7.84951270e-01
4.75301385e-01 1.29450798e+00 -4.17749107e-01 1.24136245e+00
-3.33162136e-02 8.57650280e-01 -3.42320234e-01 3.77326876e-01
-2.11134087e-02 -4.97580290e-01 8.30432892e-01 1.05700290e+00
7.66789198e-01 3.95683199e-01 2.43283540e-01 7.62441397e-01
3.22354794e-01 4.59241867e-01 -7.82231569e-01 -1.20036341e-01
3.40277344e-01 1.33681440e+00 -5.76462924e-01 1.47892684e-02
-5.93589425e-01 8.79743338e-01 3.32787521e-02 4.77124780e-01
-5.29793322e-01 -2.57771313e-01 6.29626334e-01 4.39279675e-02
6.48703277e-01 -4.08219755e-01 -4.57524844e-02 -1.42160165e+00
-9.72046237e-03 -1.04971552e+00 3.56141031e-01 -4.44594115e-01
-1.28535903e+00 4.90306884e-01 -1.15431793e-01 -1.46535039e+00
1.73166215e-01 -5.85027099e-01 -5.30908704e-01 9.36695755e-01
-2.22488594e+00 -1.03610909e+00 -5.51217556e-01 6.84721470e-01
5.53534567e-01 -6.87601626e-01 8.37297440e-01 2.15503573e-01
-8.90707850e-01 3.24489713e-01 3.57202619e-01 -1.97560176e-01
6.35888994e-01 -7.59520948e-01 -5.34249127e-01 8.60808074e-01
2.89875090e-01 6.99548006e-01 6.25845730e-01 -3.66093636e-01
-1.84542990e+00 -1.10513616e+00 4.64478076e-01 3.28088075e-01
4.19509113e-01 1.43378109e-01 -9.75428045e-01 3.50439131e-01
-2.84252077e-01 5.62744379e-01 1.18122935e+00 -1.24627829e-01
-4.89643723e-01 -4.76289093e-01 -1.18196821e+00 -1.15549788e-02
6.50052845e-01 -5.08306682e-01 -4.77186263e-01 4.66120839e-01
1.02049150e-01 -4.20086645e-03 -7.07541525e-01 6.37545884e-01
4.31688637e-01 -1.04678357e+00 1.05561805e+00 -3.55552644e-01
2.28935510e-01 -4.84882593e-01 -6.51432693e-01 -1.26297867e+00
-4.31016713e-01 -3.69968355e-01 -2.12812245e-01 8.01796854e-01
1.47602156e-01 -1.09025347e+00 9.26270247e-01 8.68312269e-02
-1.55149490e-01 -6.44135654e-01 -1.04817009e+00 -9.68628347e-01
-5.60476124e-01 -1.81247637e-01 2.57084846e-01 1.25461209e+00
-4.95979227e-02 1.70909539e-01 -3.53329837e-01 5.83151996e-01
9.97496247e-01 5.59177995e-01 5.52859187e-01 -1.60070801e+00
-5.00600219e-01 -3.78797017e-02 -8.30604196e-01 -7.10412920e-01
5.15497744e-01 -1.24138021e+00 -2.13698715e-01 -1.12491560e+00
1.74269810e-01 -6.34965956e-01 -4.89475071e-01 6.73483431e-01
-1.23173177e-01 5.88920891e-01 8.16767514e-02 7.24779487e-01
-8.90520029e-03 7.86745667e-01 6.59468472e-01 -3.26062649e-01
-3.28359544e-01 6.78341016e-02 -7.39970088e-01 4.81938213e-01
6.56210303e-01 -5.42177916e-01 -2.98055321e-01 4.73708734e-02
-8.06997716e-02 3.83793823e-02 3.28271747e-01 -1.34784257e+00
2.33625278e-01 -1.74793229e-01 4.01692659e-01 -1.71534091e-01
5.96881151e-01 -9.51751590e-01 4.06940579e-01 5.72521627e-01
-1.03199165e-02 -5.80906212e-01 1.68740198e-01 8.21848094e-01
-6.23921037e-01 -3.16913486e-01 9.03361261e-01 1.36926994e-01
-7.80922055e-01 3.38613391e-01 -7.74928406e-02 -6.92041993e-01
1.18505621e+00 -4.97441858e-01 4.99419086e-02 -1.38682261e-01
-5.57347655e-01 -3.89417142e-01 6.01295382e-02 -9.21622813e-02
9.62327361e-01 -1.05896544e+00 -1.11861563e+00 6.44441903e-01
2.06598863e-01 -4.84617025e-01 2.72360384e-01 1.33605957e+00
-3.34085286e-01 5.90850234e-01 -1.28512159e-01 -7.78506458e-01
-1.29081035e+00 3.07842314e-01 1.10085361e-01 1.44182056e-01
-8.13187897e-01 8.38854373e-01 2.53107212e-02 -1.68736815e-01
-9.18238759e-02 7.80812353e-02 -5.59340596e-01 9.57930908e-02
6.40628397e-01 4.10101742e-01 2.16904074e-01 -7.65126526e-01
-5.40695131e-01 7.67098188e-01 1.99699506e-01 2.33559147e-01
1.87163007e+00 2.67496943e-01 -3.74472111e-01 2.05991879e-01
1.20435309e+00 1.90869406e-01 -9.72628891e-01 -3.47916603e-01
-5.79922926e-03 -7.63432503e-01 5.76590478e-01 -2.82660842e-01
-1.25617087e+00 6.97821438e-01 6.07495785e-01 1.06099069e-01
1.38110483e+00 -4.83269483e-01 5.24760365e-01 3.88351679e-01
6.06170237e-01 -4.15253669e-01 -2.17885196e-01 1.99981496e-01
9.57592726e-01 -1.45510566e+00 3.30473244e-01 -8.03509712e-01
-3.95934463e-01 1.31259477e+00 7.05981329e-02 -1.22727245e-01
8.35412323e-01 -8.94150957e-02 -2.25499168e-01 -2.23363221e-01
-3.73980433e-01 -2.78798908e-01 4.56549048e-01 6.49821520e-01
4.53531027e-01 -3.08506526e-02 -3.97201538e-01 3.58485311e-01
2.17554510e-01 1.26653403e-01 2.36073062e-01 9.13049936e-01
-5.88138759e-01 -1.03568685e+00 -8.11462522e-01 5.64618051e-01
-2.96469897e-01 -1.14508800e-01 -3.28833796e-02 1.86034620e-01
-2.58983403e-01 9.96089995e-01 -4.63429719e-01 -2.04744548e-01
1.33708134e-01 1.47166431e-01 3.24641883e-01 -8.85228753e-01
-1.28677174e-01 -1.90891534e-01 -2.15967491e-01 -5.57077348e-01
-7.10080802e-01 -6.67565584e-01 -8.98721039e-01 5.26886731e-02
-5.86118400e-01 3.29527408e-01 1.05761611e+00 9.25462663e-01
5.15699923e-01 3.99758816e-02 1.06396747e+00 -1.01328456e+00
-9.44912255e-01 -8.57559025e-01 -1.12492788e+00 1.48854181e-01
2.34766588e-01 -8.54019761e-01 -7.32792079e-01 -3.11529525e-02]
|
[12.44079303741455, 0.40752479434013367]
|
8b4f24d2-6279-442a-9d2b-563a2f7f1616
|
multimodal-relation-extraction-with-cross
|
2305.16166
| null |
https://arxiv.org/abs/2305.16166v1
|
https://arxiv.org/pdf/2305.16166v1.pdf
|
Multimodal Relation Extraction with Cross-Modal Retrieval and Synthesis
|
Multimodal relation extraction (MRE) is the task of identifying the semantic relationships between two entities based on the context of the sentence image pair. Existing retrieval-augmented approaches mainly focused on modeling the retrieved textual knowledge, but this may not be able to accurately identify complex relations. To improve the prediction, this research proposes to retrieve textual and visual evidence based on the object, sentence, and whole image. We further develop a novel approach to synthesize the object-level, image-level, and sentence-level information for better reasoning between the same and different modalities. Extensive experiments and analyses show that the proposed method is able to effectively select and compare evidence across modalities and significantly outperforms state-of-the-art models.
|
['Philip S. Yu', 'Irwin King', 'Zhiyang Teng', 'Zhijiang Guo', 'Xuming Hu']
|
2023-05-25
| null | null | null | null |
['cross-modal-retrieval', 'relation-extraction']
|
['miscellaneous', 'natural-language-processing']
|
[ 2.30544254e-01 -1.91579103e-01 -3.76109689e-01 -3.19146305e-01
-1.04385459e+00 -3.43715847e-01 8.78360808e-01 8.23541045e-01
-2.95523196e-01 6.19283140e-01 3.16210270e-01 -6.82582408e-02
-2.79038936e-01 -7.03591645e-01 -4.96398300e-01 -2.24943697e-01
2.59489745e-01 1.90851659e-01 4.83464360e-01 1.41548580e-02
5.13199806e-01 3.55570436e-01 -1.69774163e+00 9.97209966e-01
7.72214651e-01 1.14436007e+00 4.33397740e-01 4.36959058e-01
-4.62472647e-01 1.17987740e+00 -4.53875870e-01 -6.38164878e-01
-3.38746130e-01 -5.26513100e-01 -1.02055383e+00 2.69469142e-01
6.39073670e-01 -2.92063773e-01 -2.40146711e-01 1.05633438e+00
3.66511226e-01 -2.01384246e-01 6.41979814e-01 -1.15123177e+00
-7.16126978e-01 4.15039957e-01 -4.72373128e-01 2.26346865e-01
8.71451914e-01 -4.45476741e-01 8.67667735e-01 -1.04501164e+00
1.02591348e+00 1.25577140e+00 -2.45554075e-02 1.08016878e-01
-6.57299161e-01 -3.69336993e-01 7.22460747e-02 7.23073304e-01
-1.40258741e+00 -2.30684385e-01 9.26315308e-01 -4.05834496e-01
7.60417938e-01 2.65056700e-01 5.40542543e-01 7.97868609e-01
9.01889801e-02 1.01859331e+00 1.36433029e+00 -7.71029770e-01
-2.67883509e-01 4.41155016e-01 2.69529790e-01 9.74730253e-01
1.11152381e-01 -4.17894810e-01 -1.03716207e+00 -5.19753844e-02
5.50383270e-01 -2.96025872e-01 -2.71321923e-01 -1.72403276e-01
-1.39818549e+00 4.16132450e-01 2.91447252e-01 5.46372175e-01
-4.05629754e-01 -2.26193026e-01 4.53977168e-01 8.00617784e-02
1.87101021e-01 3.33505154e-01 -1.71438769e-01 1.95556134e-01
-7.82346725e-01 -3.51037383e-02 6.91995621e-01 9.09099996e-01
5.85915267e-01 -6.68651342e-01 -4.24422801e-01 8.00117910e-01
5.35516500e-01 5.20417750e-01 2.00918674e-01 -8.14778626e-01
7.45389163e-01 9.95725095e-01 2.99481750e-02 -1.42047966e+00
-8.92515555e-02 -8.28308985e-02 -4.25752223e-01 -2.30643958e-01
-4.80046868e-02 3.53465438e-01 -7.18550861e-01 1.14952099e+00
3.23280692e-01 -1.03474893e-01 3.03205550e-01 1.01869023e+00
1.68897879e+00 5.01538992e-01 4.93147105e-01 -3.63768548e-01
1.70245790e+00 -1.03643739e+00 -1.12588930e+00 -2.98580706e-01
4.69117373e-01 -1.10157406e+00 7.24061668e-01 8.28764439e-02
-1.15642452e+00 -5.95349014e-01 -1.08973885e+00 -1.39954835e-01
-8.37164640e-01 7.61777759e-01 4.06082571e-01 8.12049061e-02
-6.33958161e-01 1.09091826e-01 -8.41173902e-02 -4.87520367e-01
1.11154146e-01 -2.18533874e-02 -6.87780619e-01 -4.95274007e-01
-1.20745337e+00 1.47984266e+00 7.07184076e-01 2.39618301e-01
-5.57843089e-01 2.76909973e-02 -1.05055094e+00 -8.70423019e-03
5.71070254e-01 -1.03299546e+00 6.93823099e-01 -7.27819443e-01
-8.11167598e-01 1.09190333e+00 -3.07701468e-01 -9.55518633e-02
6.60497472e-02 -6.22953586e-02 -6.70864403e-01 8.61534238e-01
2.35305473e-01 7.16616392e-01 4.99416828e-01 -1.84171271e+00
-7.87632346e-01 -4.75832611e-01 2.33167857e-01 6.61206961e-01
-3.59479278e-01 3.15031797e-01 -9.38277781e-01 -4.14265215e-01
4.21048909e-01 -6.36334002e-01 3.77297819e-01 -1.71783686e-01
-2.20952213e-01 -2.63850152e-01 8.41825724e-01 -7.75522053e-01
1.15982032e+00 -1.87018585e+00 1.32958114e-01 1.86741993e-01
-5.93874790e-02 1.12881698e-01 -2.15747058e-01 4.54086304e-01
1.13388024e-01 1.07280582e-01 2.08635971e-01 -2.72498667e-01
-1.97297245e-01 2.58073986e-01 -3.32478359e-02 -6.76544243e-03
1.59228131e-01 8.79410982e-01 -9.15475011e-01 -1.38021958e+00
2.45617315e-01 2.98198432e-01 3.71915013e-01 1.63883820e-01
-8.41940343e-02 3.11458230e-01 -4.73056257e-01 1.02745819e+00
3.41240644e-01 -3.46197963e-01 3.16619188e-01 -7.49756992e-01
2.44655475e-01 9.23909321e-02 -1.05720615e+00 1.67128801e+00
-3.57687384e-01 7.06274807e-01 -3.19537789e-01 -7.53592074e-01
7.90376544e-01 4.39628214e-01 3.32959533e-01 -9.54550326e-01
1.91153020e-01 1.79445088e-01 -2.90554941e-01 -1.26243401e+00
5.24400473e-01 -1.19636327e-01 3.61088179e-02 -1.22707181e-01
1.35190010e-01 -1.15846053e-01 4.57998961e-01 3.83306921e-01
6.67609692e-01 3.51716131e-01 3.40066880e-01 3.63004535e-01
8.23539972e-01 6.80796579e-02 1.23072185e-01 6.01049125e-01
5.03607206e-02 2.51916289e-01 3.99358362e-01 3.47205671e-03
-5.72764695e-01 -9.56730127e-01 2.06381548e-02 8.33047986e-01
9.44635510e-01 -6.41260982e-01 -4.97812867e-01 -7.94396818e-01
-3.07867497e-01 7.37844944e-01 -6.47041857e-01 1.14062868e-01
-1.56375200e-01 -4.16766971e-01 1.98477864e-01 4.10639912e-01
8.06604564e-01 -9.46035683e-01 -3.58931184e-01 -2.41735041e-01
-8.21022630e-01 -1.57875383e+00 -7.23876357e-02 -4.81213450e-01
-8.31724882e-01 -1.28928900e+00 -3.16754580e-01 -8.90746832e-01
7.75651515e-01 3.73317510e-01 1.16616583e+00 4.84207362e-01
-1.25650674e-01 7.28937089e-01 -6.06265664e-01 -3.12856793e-01
-3.63441944e-01 -2.97478408e-01 -4.78682846e-01 1.34860098e-01
1.33366615e-01 7.62178004e-02 -4.77476180e-01 2.97227353e-01
-9.86352682e-01 3.83994162e-01 8.90539408e-01 5.32774866e-01
7.16160119e-01 2.36001909e-01 2.98576534e-01 -4.11177754e-01
7.62235522e-01 -4.12063479e-01 3.44997495e-02 1.19407523e+00
-5.88613272e-01 1.21549934e-01 -2.10066840e-01 -3.88834476e-01
-1.33765125e+00 -8.50569680e-02 3.22760373e-01 -2.57040292e-01
-2.71830678e-01 9.15846109e-01 -1.82331830e-01 -6.63500726e-02
8.18758309e-02 1.96293414e-01 -2.94795364e-01 -9.12428498e-02
4.43691015e-01 5.22149086e-01 3.73645067e-01 -7.69950330e-01
1.41069248e-01 4.39417154e-01 3.96002501e-01 -5.82043707e-01
-1.05054760e+00 -8.07359159e-01 -7.78650045e-01 -8.98631394e-01
9.54654276e-01 -8.54970813e-01 -5.41925550e-01 -3.78912315e-02
-1.43821251e+00 4.02248055e-01 1.66867808e-01 6.92990184e-01
-3.14406991e-01 5.49427986e-01 -5.64720809e-01 -8.43430281e-01
-1.61127269e-01 -1.17194152e+00 1.36974061e+00 2.38470614e-01
4.76027355e-02 -8.52616966e-01 -3.38081717e-01 1.00107574e+00
-1.02034941e-01 1.46531403e-01 1.00952482e+00 -4.21465755e-01
-8.04597974e-01 -4.06237572e-01 -6.17148817e-01 4.05890495e-02
-1.15553364e-01 1.59714937e-01 -4.78403032e-01 4.09035355e-01
-4.23295125e-02 -4.25898552e-01 7.84372389e-01 -8.73023570e-02
8.27553630e-01 -2.34534275e-02 -5.80722392e-01 -3.39931130e-01
1.47955227e+00 1.99795306e-01 7.21091092e-01 3.95779014e-01
5.86740553e-01 8.94243240e-01 1.26949990e+00 -1.61919668e-02
6.81526124e-01 7.18675733e-01 3.68994981e-01 7.41784796e-02
-4.36172336e-01 -2.35942364e-01 7.39005134e-02 9.06944036e-01
-2.54432142e-01 -3.42596471e-01 -8.92366409e-01 5.95675349e-01
-2.25180149e+00 -1.18742776e+00 -3.60803366e-01 1.60175431e+00
7.71106839e-01 2.84150634e-02 -3.45266700e-01 -1.95034638e-01
7.07427263e-01 -1.47432074e-01 2.05557570e-02 -8.89989808e-02
-4.70269322e-01 -3.77948493e-01 2.64067370e-02 3.47050279e-01
-8.94693911e-01 7.85568774e-01 6.44124413e+00 1.01947808e+00
-6.46877587e-01 1.29503384e-01 4.37974334e-01 2.23281384e-01
-3.87927979e-01 1.57406360e-01 -7.30418205e-01 1.65513888e-01
4.99407709e-01 2.00099632e-01 -4.05843407e-02 3.75161648e-01
-6.72334284e-02 -8.65189672e-01 -1.01451850e+00 1.08555365e+00
7.47719944e-01 -1.30922759e+00 4.68077034e-01 -1.94627628e-01
5.02952456e-01 -6.33958519e-01 -9.81814489e-02 3.81596982e-02
-3.31754476e-01 -8.89445841e-01 7.90357828e-01 1.18034852e+00
3.63673151e-01 -4.06079054e-01 1.14250708e+00 4.15663093e-01
-1.30747974e+00 1.80093125e-01 1.21259362e-01 2.78839916e-01
1.80450350e-01 1.41633675e-01 -8.58608961e-01 1.15766585e+00
7.48106301e-01 5.11982620e-01 -1.01611269e+00 8.03159595e-01
-5.12518048e-01 1.68035626e-01 -1.36454431e-02 -2.09212378e-01
-1.54076025e-01 3.38985845e-02 3.04793298e-01 1.17649341e+00
2.27882117e-01 2.92139947e-01 1.00426145e-01 6.56274498e-01
-2.34927274e-02 3.59430790e-01 -6.12613678e-01 7.52420425e-02
2.91014910e-01 1.24333656e+00 -9.66980636e-01 -6.35340095e-01
-6.39553607e-01 9.00531888e-01 3.32909226e-01 2.32845366e-01
-7.25115657e-01 -6.71150163e-02 -3.27433735e-01 -1.60493135e-01
2.01910809e-02 -3.00183833e-01 -3.53683263e-01 -1.07891142e+00
3.22050989e-01 -5.54228306e-01 7.19147682e-01 -1.60840416e+00
-1.33469474e+00 5.59540510e-01 4.91522998e-01 -1.29518938e+00
6.43873066e-02 -5.03626049e-01 -1.06176168e-01 6.93304121e-01
-1.47338617e+00 -1.60190070e+00 -3.47102493e-01 5.36277831e-01
5.55049837e-01 -1.04767084e-01 7.94052184e-01 2.78549314e-01
-2.15839103e-01 3.14231664e-02 -3.72767031e-01 2.10492879e-01
7.70432174e-01 -9.16464150e-01 -8.60336185e-01 7.92190373e-01
4.21822369e-01 6.94180667e-01 5.89592099e-01 -9.34032440e-01
-1.21411204e+00 -4.29744929e-01 1.10285354e+00 -5.20193756e-01
5.37708819e-01 2.92021900e-01 -6.54734910e-01 2.19962791e-01
6.36137784e-01 -1.97235256e-01 7.74505734e-01 6.97597116e-02
-4.32906568e-01 -1.72994554e-01 -9.65644419e-01 7.51534462e-01
6.32371724e-01 -8.75098526e-01 -1.05302286e+00 2.78513759e-01
4.63018477e-01 -3.96975636e-01 -1.11425853e+00 8.41005385e-01
5.70650101e-01 -7.78668582e-01 1.12036812e+00 -3.53337020e-01
8.27362895e-01 -4.99394000e-01 -4.26513195e-01 -6.66626930e-01
1.67434439e-01 4.21197623e-01 -3.27120781e-01 1.43337822e+00
5.56993067e-01 9.47575048e-02 1.90845400e-01 6.14818871e-01
2.62695044e-01 -8.20088804e-01 -7.62435615e-01 -4.62644637e-01
-7.17089295e-01 -3.19329113e-01 2.70257205e-01 1.09690261e+00
2.87964225e-01 6.91375315e-01 -6.50488958e-02 4.66118068e-01
3.96320492e-01 5.12272239e-01 3.20801735e-01 -8.43769968e-01
1.57055423e-01 -5.19532561e-01 -5.80077410e-01 -5.86583972e-01
2.91863650e-01 -6.62377715e-01 3.30286026e-02 -2.24220324e+00
7.80385852e-01 -5.42815402e-02 -4.23803896e-01 3.59956592e-01
-3.98897290e-01 3.31897467e-01 3.30286711e-01 3.93350571e-01
-1.14561248e+00 3.86419326e-01 1.48153937e+00 -2.72498369e-01
3.46752465e-01 -4.10534710e-01 -3.12024534e-01 6.78433955e-01
4.62303042e-01 -3.28051895e-01 -3.94319475e-01 -2.12499514e-01
4.50121254e-01 4.17134047e-01 7.02506363e-01 -7.15641916e-01
5.50407588e-01 -2.78721780e-01 4.55604136e-01 -9.26849902e-01
5.41299284e-01 -1.05535591e+00 -7.26721762e-03 1.02163181e-01
-5.55228472e-01 7.14070275e-02 1.72308698e-01 6.80131078e-01
-7.52679288e-01 -6.26652122e-01 1.81134447e-01 -2.01991796e-01
-9.93988693e-01 -7.11633638e-02 -1.07813217e-01 -4.16577607e-01
1.03589952e+00 -1.70375422e-01 -6.34030104e-01 -3.21724117e-01
-7.77913094e-01 2.71409243e-01 1.48460627e-01 5.50888598e-01
1.16595566e+00 -1.46893513e+00 -5.14941633e-01 -5.03836751e-01
5.03929794e-01 -3.50248486e-01 4.67662007e-01 9.17588115e-01
-3.78191829e-01 3.85838985e-01 -1.51982918e-01 -4.39845413e-01
-1.95758629e+00 6.25382423e-01 1.92927092e-01 -3.17958027e-01
-2.78142020e-02 4.67593938e-01 -2.38345489e-01 7.80963600e-02
2.06261396e-01 1.22534312e-01 -7.09227324e-01 4.03664351e-01
2.44433090e-01 8.47784504e-02 -7.55783021e-02 -9.91796374e-01
-3.93540651e-01 7.06107438e-01 1.62512466e-01 -4.17227745e-01
7.68015623e-01 -4.66126770e-01 -5.81693351e-01 6.37312353e-01
9.88826990e-01 -5.33541255e-02 -3.92937303e-01 -3.91518682e-01
4.58373763e-02 -5.20645440e-01 2.42418200e-02 -1.01130807e+00
-6.24903560e-01 7.36100197e-01 5.15856922e-01 2.47712612e-01
1.28017843e+00 5.36945343e-01 4.63754177e-01 2.62664437e-01
3.68616223e-01 -1.18856215e+00 5.17607808e-01 1.82643473e-01
9.37223196e-01 -1.55267251e+00 5.84025621e-01 -9.44790840e-01
-8.29239428e-01 1.28793800e+00 7.91499853e-01 5.84092259e-01
4.47778106e-01 -8.18599984e-02 -2.50670284e-01 -5.89520216e-01
-6.33794904e-01 -5.54868460e-01 1.02637088e+00 1.79527104e-01
5.14012516e-01 -1.22280039e-01 -5.27566552e-01 4.44359243e-01
6.20063066e-01 -7.02510849e-02 -2.62160394e-02 1.10785794e+00
-3.90521437e-01 -1.09233618e+00 -5.45451880e-01 2.92383730e-01
-2.50782609e-01 -7.29578435e-02 -6.29407048e-01 7.24960804e-01
4.32889313e-01 1.25149095e+00 -1.98983461e-01 -3.11363131e-01
3.34053636e-01 4.67696115e-02 8.98141146e-01 -3.04621011e-01
-2.22169206e-01 7.22649693e-03 4.96124119e-01 -2.83187479e-01
-1.38090849e+00 -5.29759109e-01 -1.22429919e+00 1.55852854e-01
-5.28709650e-01 -3.49209458e-02 8.20613742e-01 1.41914773e+00
2.14950129e-01 5.09138882e-01 4.97979261e-02 -3.69113117e-01
2.70276755e-01 -8.21763277e-01 -1.98621452e-01 6.85744882e-01
-8.88923928e-02 -7.92431593e-01 3.98701876e-02 3.81961793e-01]
|
[10.709062576293945, 1.4351511001586914]
|
6e4a8cac-48cc-4c83-b846-36426e63cb49
|
boundary-detection-and-categorization-of
| null | null |
https://aclanthology.org/2022.argmining-1.12
|
https://aclanthology.org/2022.argmining-1.12.pdf
|
Boundary Detection and Categorization of Argument Aspects via Supervised Learning
|
Aspect-based argument mining (ABAM) is the task of automatic _detection_ and _categorization_ of argument aspects, i.e. the parts of an argumentative text that contain the issue-specific key rationale for its conclusion. From empirical data, overlapping but not congruent sets of aspect categories can be derived for different topics. So far, two supervised approaches to detect aspect boundaries, and a smaller number of unsupervised clustering approaches to categorize groups of similar aspects have been proposed. With this paper, we introduce the Argument Aspect Corpus (AAC) that contains token-level annotations of aspects in 3,547 argumentative sentences from three highly debated topics. This dataset enables both the supervised learning of boundaries and categorization of argument aspects. During the design of our annotation process, we noticed that it is not clear from the outset at which contextual unit aspects should be coded. We, thus, experiment with classification at the token, chunk, and sentence level granularity. Our finding is that the chunk level provides the most useful information for applications. At the same time, it produces the best performing results in our tested supervised learning setups.
|
['Gregor Wiedemann', 'Mattes Ruckdeschel']
| null | null | null | null |
argmining-acl-2022-10
|
['boundary-detection', 'argument-mining']
|
['computer-vision', 'natural-language-processing']
|
[ 1.77580297e-01 8.44201982e-01 -5.17542481e-01 -4.85000253e-01
-8.88000071e-01 -8.15126359e-01 1.02942300e+00 1.12912500e+00
-3.63711238e-01 7.12648630e-01 8.35985422e-01 -8.56085300e-01
-2.59598196e-01 -6.36438847e-01 -3.83018464e-01 -4.10230577e-01
2.36666694e-01 7.06917524e-01 2.41659686e-01 -6.81506246e-02
8.24370027e-01 -1.61816880e-01 -1.57221746e+00 1.02166820e+00
8.96411955e-01 6.89938486e-01 -1.49108917e-01 2.85756201e-01
-7.94091880e-01 8.90329421e-01 -1.02260149e+00 -5.39014637e-01
-4.07765716e-01 -6.08701348e-01 -1.33609569e+00 4.49593328e-02
8.40330198e-02 4.74262834e-01 7.04126000e-01 6.52588606e-01
2.81468123e-01 -2.25620061e-01 9.07749712e-01 -9.87269461e-01
-1.80746719e-01 1.34003055e+00 -3.12990218e-01 1.80874869e-01
4.61900622e-01 -4.80932951e-01 1.49413276e+00 -5.88018954e-01
1.01533091e+00 1.33589756e+00 5.81622720e-01 1.38441831e-01
-9.98239517e-01 -3.63614224e-02 2.62086660e-01 8.52567926e-02
-5.00738144e-01 -2.15736181e-01 8.78309965e-01 -6.53293192e-01
8.87612522e-01 3.76120448e-01 8.17268491e-01 8.23449850e-01
6.08232580e-02 8.85453105e-01 1.57441747e+00 -7.80720651e-01
6.39858007e-01 5.02148032e-01 7.43526936e-01 -4.10581268e-02
6.59513950e-01 -6.99170649e-01 -4.73003030e-01 -7.85715759e-01
-2.13183120e-01 -6.89273536e-01 -1.97628140e-02 -7.57973865e-02
-9.79479730e-01 1.18949795e+00 -1.88628256e-01 8.57179642e-01
-5.39549828e-01 -2.95941085e-01 8.41086328e-01 3.63550693e-01
7.48971701e-01 4.93741184e-01 -7.98079014e-01 -4.10634458e-01
-7.22882569e-01 3.45475107e-01 1.27110922e+00 6.00831270e-01
3.96692544e-01 -4.92099285e-01 -7.90682435e-02 6.65968060e-01
3.06685477e-01 4.76669669e-02 3.44682723e-01 -7.01097786e-01
7.28658378e-01 1.27906048e+00 4.00543064e-02 -6.91837370e-01
-3.64188462e-01 -2.24644795e-01 -2.49372393e-01 6.41281204e-03
5.03891051e-01 -2.66221970e-01 -4.55129266e-01 1.38500476e+00
6.19567156e-01 -7.85369098e-01 4.17321384e-01 5.13224185e-01
1.08405852e+00 2.56532520e-01 3.74923468e-01 -3.36133540e-01
1.89778924e+00 -3.71158242e-01 -9.28824663e-01 -4.83195484e-02
9.48677838e-01 -1.20978260e+00 1.01660967e+00 1.88719571e-01
-9.23287630e-01 -1.20016217e-01 -8.74997199e-01 1.95026219e-01
-5.14268577e-01 -8.01660642e-02 9.33477938e-01 6.84449673e-01
-4.77911681e-01 3.17900807e-01 -4.00328398e-01 -3.26135844e-01
3.61540556e-01 -1.28306642e-01 -2.06380069e-01 4.58331555e-01
-1.00736761e+00 7.45519042e-01 5.30161321e-01 -4.01588351e-01
4.26960327e-02 -4.15727049e-01 -7.30468631e-01 -2.29947898e-03
6.44170880e-01 -3.38995039e-01 1.09170711e+00 -9.94407058e-01
-7.71970212e-01 1.18208742e+00 -2.79878825e-01 -6.43517375e-01
1.88977078e-01 -3.39284807e-01 -1.94165304e-01 2.16103733e-01
5.88917732e-01 2.76614249e-01 7.57306993e-01 -1.24174428e+00
-7.87324309e-01 -4.39029127e-01 2.06603408e-01 1.56677544e-01
-1.49701253e-01 5.01077354e-01 2.03562528e-01 -6.69488668e-01
2.10840240e-01 -8.19926262e-01 -4.09717821e-02 -7.12198257e-01
-5.35530210e-01 -8.07095706e-01 9.07665431e-01 -5.40098965e-01
1.24503732e+00 -1.92906809e+00 -6.17735833e-02 1.25453934e-01
1.84720010e-01 8.40208530e-02 5.22610366e-01 7.22878575e-01
-9.08503905e-02 4.54958439e-01 -2.57144898e-01 1.33318320e-01
5.69351278e-02 2.79814690e-01 -5.23604929e-01 2.10438758e-01
-5.71877472e-02 7.93631375e-01 -6.48758471e-01 -9.08016026e-01
-2.79614359e-01 4.36071232e-02 -6.40166581e-01 -1.07389353e-01
-5.25657773e-01 2.96829611e-01 -7.74521589e-01 5.46347737e-01
3.80137444e-01 -1.54814245e-02 5.37884295e-01 -1.84696078e-01
-2.78920174e-01 1.19637144e+00 -1.10980141e+00 1.28589082e+00
-3.07378441e-01 8.61719310e-01 7.39721656e-02 -1.28306615e+00
8.74210894e-01 6.25161648e-01 5.37375331e-01 -3.76751184e-01
2.19300658e-01 3.63333493e-01 3.43025804e-01 -4.02023107e-01
6.68890536e-01 -6.10963348e-03 -6.33240581e-01 9.98223543e-01
-1.44022450e-01 -3.73747982e-02 4.59731877e-01 1.48530334e-01
8.79701316e-01 -1.26227066e-01 8.88417125e-01 -7.49565959e-01
5.39512992e-01 6.29904687e-01 6.40110552e-01 4.62689251e-01
9.62625444e-02 2.89087564e-01 1.16879559e+00 -6.14335954e-01
-9.49839711e-01 -6.77589417e-01 -4.06244040e-01 8.67538035e-01
-2.24900335e-01 -9.10384595e-01 -7.77526259e-01 -8.73550773e-01
-2.03891158e-01 8.68900716e-01 -8.54978740e-01 5.24299741e-01
-7.77900040e-01 -7.30485916e-01 2.33475000e-01 9.15818289e-02
4.68813092e-01 -1.31104422e+00 -1.13211310e+00 3.11350018e-01
-4.85118330e-01 -9.40201521e-01 2.54978567e-01 5.47900200e-01
-7.58094311e-01 -1.77075696e+00 -6.19958416e-02 -5.59342384e-01
4.64428514e-01 -4.27797809e-02 1.41569376e+00 3.49381417e-01
4.52333726e-02 1.32181466e-01 -8.55801761e-01 -8.28611314e-01
-8.49892437e-01 2.20485687e-01 -4.56145018e-01 -3.69505852e-01
5.78876734e-01 -5.25502324e-01 -2.78417647e-01 8.81391242e-02
-7.55355597e-01 -2.50860780e-01 4.50124204e-01 6.48213387e-01
4.40583229e-01 -6.71176985e-02 4.96238977e-01 -1.44218206e+00
1.10912263e+00 -3.74002874e-01 -4.86357249e-02 2.16393217e-01
-6.39277160e-01 1.62655964e-01 3.08263421e-01 -2.94374406e-01
-1.15014434e+00 -6.88435912e-01 -3.01776052e-01 8.86267066e-01
-5.11072040e-01 6.24345005e-01 -1.82046145e-01 6.78108215e-01
7.56027222e-01 -2.79940039e-01 -9.17720795e-02 -3.28936458e-01
4.18415308e-01 9.13650632e-01 -1.60436749e-01 -8.74827027e-01
3.56449813e-01 4.68936294e-01 -1.66111887e-01 -9.69370306e-01
-1.26405871e+00 -6.06101811e-01 -5.73388934e-01 -1.70045599e-01
7.88694322e-01 -5.81007838e-01 -4.38096553e-01 -1.84813261e-01
-1.25128734e+00 -1.14087895e-01 -5.94997287e-01 3.85509580e-01
-6.00446522e-01 3.84443671e-01 -3.98817569e-01 -7.26339340e-01
-5.56206763e-01 -8.82294297e-01 8.08366537e-01 -6.97993189e-02
-1.16320539e+00 -9.61588740e-01 8.20028111e-02 6.33882880e-01
1.65443987e-01 4.64162260e-01 1.42444384e+00 -1.32594097e+00
1.64375260e-01 1.24117859e-01 1.81792051e-01 -4.73433658e-02
2.78717428e-01 -1.39965475e-01 -8.48651111e-01 2.68422782e-01
4.34280276e-01 5.39206574e-03 7.02404082e-01 3.75669003e-01
4.50223684e-01 -6.07628047e-01 -3.79146069e-01 -2.02885091e-01
1.04205453e+00 2.97619283e-01 5.28727174e-01 1.00301778e+00
1.34510592e-01 1.09231758e+00 1.17487311e+00 3.15078676e-01
2.16378048e-01 5.77446163e-01 1.98120415e-01 2.13977784e-01
-4.82413545e-03 1.14381112e-01 1.88677683e-01 5.00398815e-01
-9.01196431e-03 -9.73051786e-02 -9.01815951e-01 7.79505610e-01
-1.88907778e+00 -1.18897355e+00 -6.17803514e-01 1.81499279e+00
7.89309978e-01 6.62524283e-01 2.66270965e-01 6.39245868e-01
6.58578753e-01 3.37794691e-01 9.38402265e-02 -9.46443081e-01
-1.57252103e-01 9.00310427e-02 -1.69631273e-01 4.08852994e-01
-1.17678225e+00 6.84733510e-01 5.09824228e+00 7.18298316e-01
-4.04698461e-01 2.53851622e-01 7.06518173e-01 2.19203368e-01
-7.51230478e-01 5.77794611e-01 -7.53986180e-01 3.30712527e-01
7.54332900e-01 -1.72435809e-02 -4.56040323e-01 1.06093204e+00
2.46083304e-01 -5.57328522e-01 -7.31033325e-01 1.67818934e-01
-3.15796621e-02 -1.49124861e+00 2.49304473e-02 1.27034009e-01
5.49485862e-01 -2.68428147e-01 -4.14025068e-01 1.15143247e-01
9.08457413e-02 -4.77514952e-01 9.24223542e-01 -1.61583722e-01
5.72408848e-02 -6.81907475e-01 1.01643276e+00 3.93247724e-01
-1.04431987e+00 6.87579857e-03 -2.24461749e-01 -1.93234533e-01
2.09603980e-01 9.64859605e-01 -7.31855392e-01 6.15513802e-01
6.11883402e-01 4.18632895e-01 -4.15540904e-01 6.53789103e-01
-6.36595607e-01 1.00767887e+00 -5.96714616e-02 -2.72602707e-01
3.77464920e-01 -2.46074647e-01 7.45486438e-01 1.23695374e+00
-2.07257029e-02 -7.01312721e-02 -1.57899340e-03 5.47791064e-01
2.87130922e-01 5.83092749e-01 -5.67463756e-01 1.17068604e-01
4.31605905e-01 1.31203234e+00 -1.27867591e+00 -4.42097604e-01
-2.38166347e-01 1.55560419e-01 2.22611859e-01 -8.84310827e-02
-5.71777463e-01 -7.53428936e-02 5.56801558e-01 3.72024626e-01
3.99342507e-01 -6.86209742e-03 -7.57842839e-01 -8.91401947e-01
2.84543723e-01 -8.50891888e-01 8.49260926e-01 -1.55774355e-01
-8.66948664e-01 7.03874350e-01 2.20739409e-01 -1.10739028e+00
-5.43784261e-01 -3.70551944e-01 -1.03448105e+00 3.59894067e-01
-1.23087871e+00 -9.05509770e-01 1.10661231e-01 9.35799256e-02
5.65177560e-01 -9.01582986e-02 7.99105525e-01 -1.97813615e-01
1.35624707e-01 -1.33258179e-02 -2.76288271e-01 1.54319465e-01
5.46840549e-01 -1.35553265e+00 9.49922949e-02 4.79379624e-01
3.04991007e-01 7.09353268e-01 1.08071887e+00 -7.85220921e-01
-7.46181190e-01 -6.86522007e-01 1.64639604e+00 -5.20422101e-01
7.44260013e-01 -1.52875140e-01 -8.54369104e-01 3.49567741e-01
6.86404288e-01 -7.72297084e-01 1.10547090e+00 6.91022456e-01
-3.17719609e-01 2.80346692e-01 -1.04380560e+00 4.21901017e-01
6.53152108e-01 -3.41013759e-01 -1.54743302e+00 2.96470702e-01
5.42092323e-01 -4.68856618e-02 -9.05569613e-01 1.08965635e-01
2.41253525e-01 -1.19417465e+00 6.45960331e-01 -5.33453345e-01
5.54364026e-01 -1.74439609e-01 -4.21581343e-02 -1.01946187e+00
2.69227415e-01 -3.95528316e-01 1.41113877e-01 1.87091303e+00
7.81338513e-01 -6.44239604e-01 7.63476014e-01 3.25260252e-01
-1.83364481e-01 -8.40411901e-01 -9.95822668e-01 -3.87028426e-01
2.30886742e-01 -5.70855975e-01 5.73223770e-01 8.90761554e-01
5.63517272e-01 7.90759027e-01 4.11976933e-01 -4.18759614e-01
5.56579709e-01 9.78258371e-01 6.47906721e-01 -1.51998532e+00
-1.20494097e-01 -6.50348425e-01 -9.11193341e-02 -2.46723846e-01
2.25018218e-01 -7.63821781e-01 -3.35514158e-01 -1.74662399e+00
2.19730109e-01 -5.39102554e-01 3.06414127e-01 2.66856194e-01
-2.83253759e-01 -3.08877826e-01 -3.71102500e-03 4.47405398e-01
-5.05477130e-01 1.57782286e-01 7.66565204e-01 -2.47277364e-01
-4.23549652e-01 2.81772733e-01 -1.02050471e+00 1.14846134e+00
1.18557656e+00 -7.14638829e-01 -2.12151542e-01 7.18973577e-02
5.13229489e-01 -3.66454601e-01 -2.12787129e-02 -5.88753462e-01
-1.44341066e-01 -1.18668646e-01 -6.97297752e-02 -9.53772247e-01
-8.24991837e-02 -6.30010188e-01 -1.00652546e-01 6.56640887e-01
-5.71503401e-01 1.24257304e-01 -5.95167726e-02 1.99218214e-01
-4.62812930e-01 -6.50735378e-01 2.81050831e-01 -2.05268234e-01
-4.34073150e-01 -5.21180809e-01 -7.39044309e-01 6.44028306e-01
8.67425263e-01 -1.35251254e-01 -4.55229521e-01 -2.28804052e-01
-5.08120179e-01 -1.63823694e-01 2.10987210e-01 1.92202404e-01
3.00883085e-01 -9.62438345e-01 -7.53443182e-01 -4.67446297e-01
1.62091225e-01 -1.48847951e-02 -2.40571737e-01 9.38921452e-01
-1.82822853e-01 5.98839045e-01 1.54769510e-01 -4.55306500e-01
-1.64088404e+00 5.14235258e-01 -4.02031153e-01 -6.42945647e-01
-7.17654228e-01 1.33049965e-01 -2.36386940e-01 -3.24814945e-01
-4.43051085e-02 -3.45027626e-01 -8.46803904e-01 8.15756619e-01
3.34274769e-01 1.88179210e-01 -9.59999580e-03 -8.16703975e-01
-3.28343868e-01 3.97114336e-01 3.93940844e-02 -2.64015615e-01
1.50560486e+00 -5.97255901e-02 -4.73398745e-01 6.14332378e-01
8.06043565e-01 3.86056572e-01 -3.54234159e-01 4.18446353e-03
7.91935802e-01 -1.53418584e-02 -4.74257737e-01 -5.77794611e-01
-3.62975210e-01 5.08050859e-01 -2.14308038e-01 1.03650331e+00
6.83457792e-01 7.33057797e-01 3.39109004e-01 2.64560163e-01
7.62000978e-02 -1.42034066e+00 -3.39785635e-01 5.82408011e-01
7.90475309e-01 -9.87459064e-01 3.35239053e-01 -7.03139186e-01
-7.33936071e-01 1.12338853e+00 1.87729150e-01 1.54111475e-01
6.07407570e-01 3.10475349e-01 1.69906855e-01 -7.63090730e-01
-8.84975970e-01 -2.29546621e-01 2.32448682e-01 4.22922790e-01
9.08702970e-01 1.63314790e-01 -1.45005703e+00 7.10532665e-01
-5.24967134e-01 -7.18400002e-01 5.48777103e-01 1.17360628e+00
-7.75160551e-01 -1.32233882e+00 -5.93734086e-01 5.94401896e-01
-9.95229840e-01 -1.36410594e-01 -7.85017252e-01 1.26640427e+00
4.56665382e-02 1.30009365e+00 2.25703418e-02 9.96972769e-02
2.04670012e-01 2.49684468e-01 9.79649052e-02 -6.32719040e-01
-1.08164775e+00 2.87285950e-02 9.58555818e-01 -2.46091649e-01
-1.02953720e+00 -9.65715230e-01 -1.31088257e+00 1.03733860e-01
-2.82357752e-01 1.17620802e+00 7.40755975e-01 1.38078034e+00
1.11483656e-01 2.48220757e-01 3.35874379e-01 -2.11824432e-01
-2.67896563e-01 -1.11873317e+00 -2.62314439e-01 6.25551462e-01
-2.86551476e-01 -5.06927609e-01 -4.46644068e-01 -4.31899615e-02]
|
[9.432713508605957, 9.629117012023926]
|
f7f193f7-4864-4f22-b094-216aba03bf74
|
learning-where-to-learn-gradient-sparsity-in
|
2110.14402
| null |
https://arxiv.org/abs/2110.14402v1
|
https://arxiv.org/pdf/2110.14402v1.pdf
|
Learning where to learn: Gradient sparsity in meta and continual learning
|
Finding neural network weights that generalize well from small datasets is difficult. A promising approach is to learn a weight initialization such that a small number of weight changes results in low generalization error. We show that this form of meta-learning can be improved by letting the learning algorithm decide which weights to change, i.e., by learning where to learn. We find that patterned sparsity emerges from this process, with the pattern of sparsity varying on a problem-by-problem basis. This selective sparsity results in better generalization and less interference in a range of few-shot and continual learning problems. Moreover, we find that sparse learning also emerges in a more expressive model where learning rates are meta-learned. Our results shed light on an ongoing debate on whether meta-learning can discover adaptable features and suggest that learning by sparse gradient descent is a powerful inductive bias for meta-learning systems.
|
['João Sacramento', 'Nicolas Zucchet', 'Massimo Caccia', 'Simon Schug', 'Seijin Kobayashi', 'Dominic Zhao', 'Johannes von Oswald']
|
2021-10-27
| null |
http://proceedings.neurips.cc/paper/2021/hash/2a10665525774fa2501c2c8c4985ce61-Abstract.html
|
http://proceedings.neurips.cc/paper/2021/file/2a10665525774fa2501c2c8c4985ce61-Paper.pdf
|
neurips-2021-12
|
['sparse-learning']
|
['methodology']
|
[ 3.43312293e-01 1.05463207e-01 -5.20002067e-01 -4.96111661e-01
-3.50779861e-01 -4.52665031e-01 6.71781063e-01 1.51893228e-01
-5.95526576e-01 7.65256166e-01 3.65176022e-01 -1.20354502e-03
-4.54310924e-01 -9.24275577e-01 -7.14421809e-01 -8.14025104e-01
-1.70868948e-01 2.24410877e-01 1.70924410e-03 -6.09239221e-01
5.16739190e-01 3.41821045e-01 -1.70915687e+00 4.58931983e-01
7.50556231e-01 5.81667185e-01 2.87218690e-01 5.62955558e-01
-2.99443334e-01 8.29310060e-01 -4.36957419e-01 -1.44122526e-01
4.55461383e-01 -6.89018548e-01 -8.37342322e-01 8.25663209e-02
5.52780807e-01 2.63102829e-01 -9.98743549e-02 9.97789681e-01
3.17463994e-01 4.94811743e-01 5.23282170e-01 -9.83609319e-01
-7.08098769e-01 9.07906115e-01 -5.35416543e-01 5.22809505e-01
5.36498986e-02 1.95650429e-01 1.18956971e+00 -8.05038869e-01
7.42347598e-01 1.00742841e+00 9.73037302e-01 8.12199116e-01
-1.47897899e+00 -5.78658223e-01 4.91438299e-01 7.95901716e-02
-9.43942249e-01 -5.81677973e-01 8.50186586e-01 -3.31137210e-01
9.55501556e-01 2.81445771e-01 9.14779902e-01 1.12180030e+00
2.04274565e-01 6.45457447e-01 1.04290962e+00 -7.49894142e-01
6.31746233e-01 2.13431224e-01 1.08715594e-01 9.07935977e-01
4.68204677e-01 2.56041974e-01 -6.98646009e-01 -1.99072167e-01
5.71563482e-01 3.89459282e-01 -1.16545662e-01 -4.34320360e-01
-9.03572142e-01 9.78602231e-01 3.74498069e-01 6.12914324e-01
-3.03122103e-01 2.59072363e-01 3.63258153e-01 9.03299153e-01
4.79268491e-01 1.17838573e+00 -7.33585954e-01 3.45691927e-02
-1.05072403e+00 2.30253503e-01 8.66638601e-01 5.09721696e-01
1.26070023e+00 4.05819982e-01 6.04558513e-02 1.06083727e+00
-7.78810233e-02 1.42471626e-01 8.98133337e-01 -1.14454114e+00
2.32795030e-01 4.44023013e-01 -1.73274279e-01 -7.72255838e-01
-4.54842508e-01 -7.38172293e-01 -5.86118460e-01 4.90490228e-01
3.42904180e-01 -5.49506187e-01 -8.99802804e-01 2.09449029e+00
1.13694943e-01 1.63563654e-01 -4.26769145e-02 5.91986179e-01
2.54591674e-01 6.47790909e-01 1.69832557e-01 -4.78211135e-01
7.39692569e-01 -9.05434906e-01 -3.25591058e-01 -4.95233536e-01
8.71630609e-01 -2.60295451e-01 9.91646945e-01 4.60645199e-01
-1.13497603e+00 -4.11822200e-01 -9.35708642e-01 4.03849721e-01
-4.76421088e-01 -5.03887177e-01 1.00162983e+00 6.55041933e-01
-9.98069048e-01 9.63977218e-01 -6.72250509e-01 -6.08594477e-01
5.28104961e-01 3.89043897e-01 -1.49745688e-01 6.25516567e-03
-8.87401760e-01 1.00729597e+00 5.49401164e-01 -3.22549522e-01
-8.11054945e-01 -1.03223920e+00 -6.68128431e-01 1.35360733e-01
3.75921279e-01 -7.59234548e-01 1.07999325e+00 -1.69555080e+00
-1.37114537e+00 7.49340236e-01 -1.60089016e-01 -5.22270381e-01
1.28741831e-01 1.45067200e-01 -3.94492686e-01 -1.55970916e-01
-1.43613204e-01 6.83576465e-01 1.23737049e+00 -1.33072758e+00
-8.58681619e-01 -2.51048595e-01 3.16350833e-02 1.01938836e-01
-8.28912497e-01 -3.28838974e-01 1.40847296e-01 -7.21408427e-01
-1.01412244e-01 -8.66639435e-01 -5.93638897e-01 -2.08466336e-01
1.82243690e-01 -2.97236770e-01 5.07347941e-01 1.26474544e-01
1.43111086e+00 -2.03776908e+00 2.44084746e-01 4.47988957e-01
4.33263421e-01 1.25566751e-01 -5.57880580e-01 5.68103969e-01
-2.34273896e-01 8.41900781e-02 -1.91883191e-01 1.43695533e-01
-2.57352829e-01 3.48402411e-01 -3.23403507e-01 3.08613688e-01
1.77385911e-01 9.35273349e-01 -1.14241171e+00 -1.10122532e-01
-1.47651404e-01 2.44601592e-02 -9.27385867e-01 1.88791975e-02
-2.60346740e-01 1.86065584e-02 -4.61147279e-01 6.62669778e-01
1.06663242e-01 -5.84646940e-01 2.11133540e-01 1.16238207e-01
-3.39081973e-01 6.01799823e-02 -1.13683486e+00 1.75260711e+00
-5.96319616e-01 6.94336772e-01 -1.07180707e-01 -1.43132329e+00
7.75572777e-01 5.55441827e-02 6.40364408e-01 -5.54610550e-01
4.09558751e-02 1.42790794e-01 2.85575658e-01 -5.58127463e-01
3.02507401e-01 -4.79075432e-01 2.04232931e-01 7.50084639e-01
5.10658681e-01 -7.03978539e-02 4.06696439e-01 8.92611369e-02
1.23761690e+00 -2.77899086e-01 3.02076250e-01 -5.31021476e-01
2.11504642e-02 2.60221690e-01 6.56244874e-01 1.21179438e+00
1.44074574e-01 3.70679587e-01 8.39922503e-02 -8.17361176e-01
-1.11742675e+00 -7.79095173e-01 -8.01886842e-02 1.82685995e+00
-2.82151878e-01 -4.06263739e-01 -4.65314031e-01 -5.28283834e-01
1.35983542e-01 6.39773011e-01 -1.12102056e+00 -5.38881123e-01
-5.56039810e-01 -1.01284599e+00 2.41816435e-02 3.63878250e-01
1.29978612e-01 -1.19703579e+00 -5.88512897e-01 4.40708429e-01
5.80427349e-01 -3.30139786e-01 -3.21588606e-01 8.86556864e-01
-1.38737500e+00 -8.63481224e-01 -6.91479862e-01 -8.87854218e-01
9.06906128e-01 4.39562827e-01 1.21154082e+00 2.92014033e-01
-3.31181318e-01 6.62027180e-01 -3.63089800e-01 -4.75776702e-01
-2.13791132e-01 5.52209675e-01 2.37184882e-01 -1.61770433e-01
3.21096718e-01 -1.06735551e+00 -4.63395745e-01 -6.50550425e-02
-8.30353141e-01 -1.91119790e-01 5.91059983e-01 1.09070694e+00
2.69497484e-01 -1.20233856e-01 9.54988420e-01 -1.45877433e+00
1.06688845e+00 -8.58327746e-01 -3.13801646e-01 4.21053022e-01
-1.16629159e+00 4.39225346e-01 6.94672465e-01 -8.52845609e-01
-9.97920871e-01 3.13587114e-02 9.63620991e-02 -2.48428181e-01
-2.09152475e-02 5.73089421e-01 4.79511321e-01 -4.09050107e-01
1.30349827e+00 1.98275819e-01 -8.36114213e-02 -2.98420012e-01
6.03111923e-01 1.20474301e-01 7.59884436e-03 -8.36586714e-01
7.95004368e-01 4.10962105e-01 -1.97780594e-01 -8.02132368e-01
-1.22721612e+00 -3.18277150e-01 -3.58971268e-01 -1.13432199e-01
2.99275577e-01 -8.01358163e-01 -3.58338416e-01 -3.95331457e-02
-6.36693120e-01 -6.96932375e-01 -1.01625776e+00 2.95460761e-01
-5.44551075e-01 -2.39366323e-01 -3.94566000e-01 -5.82084537e-01
-1.75655320e-01 -5.73972106e-01 3.07829887e-01 3.17911565e-01
-4.98206794e-01 -1.35141444e+00 4.82564241e-01 -2.88813293e-01
7.52096176e-01 -6.79507479e-02 8.57459188e-01 -6.72922254e-01
-2.66856223e-01 8.61828774e-02 2.75404066e-01 8.80816877e-02
4.00196612e-01 -1.25259206e-01 -8.91414046e-01 -4.37063813e-01
1.20861782e-02 -7.48456717e-01 1.21386313e+00 3.49326670e-01
1.22840869e+00 -5.21337867e-01 -2.96756029e-01 8.52588236e-01
1.50590682e+00 6.77832142e-02 2.72661328e-01 4.64691073e-01
5.87925851e-01 4.00714219e-01 -4.61947210e-02 4.66668576e-01
-6.86183497e-02 1.36783216e-02 -2.97114300e-03 3.15151848e-02
-6.96326196e-02 -1.77383661e-01 1.26792222e-01 9.08862770e-01
-3.05466145e-01 1.56521499e-01 -8.41120720e-01 5.54919422e-01
-1.95081902e+00 -1.52091444e+00 5.21961808e-01 1.90861917e+00
1.29269278e+00 2.74018794e-01 2.67172158e-01 -2.18035765e-02
5.06012738e-01 3.24973226e-01 -7.67768562e-01 -6.02513731e-01
-2.28413180e-01 6.12879932e-01 3.72995228e-01 4.47496563e-01
-6.57302380e-01 1.05099225e+00 8.09015751e+00 6.57418251e-01
-1.26443160e+00 1.54594883e-01 5.18121123e-01 -5.09974062e-01
-6.70917034e-01 -4.94842753e-02 -7.47709930e-01 2.50851721e-01
9.86336648e-01 -3.35228890e-01 7.62877822e-01 9.02940810e-01
-2.93613151e-02 8.60702544e-02 -1.14060867e+00 6.77810371e-01
3.86763141e-02 -1.85763657e+00 1.67589083e-01 -7.75917768e-02
1.36481297e+00 2.34768271e-01 1.36458948e-01 4.38837707e-01
7.58121967e-01 -9.03626919e-01 4.43767697e-01 5.70599377e-01
4.93095070e-01 -7.23016679e-01 3.51763293e-02 5.41746438e-01
-8.85986149e-01 -7.09805310e-01 -6.79308772e-01 -3.29115361e-01
-2.72924125e-01 5.94098091e-01 -4.35265660e-01 -2.14460254e-01
4.10329640e-01 8.67574275e-01 -5.72956860e-01 1.24814415e+00
2.10705251e-02 8.76461208e-01 -6.74992055e-02 -2.85617888e-01
3.90427023e-01 -8.98792893e-02 4.85605121e-01 1.38866997e+00
3.30879867e-01 2.33972758e-01 2.51685649e-01 7.57611811e-01
-2.35164806e-01 1.51220756e-02 -8.03469002e-01 -2.61887964e-02
4.29590762e-01 1.15697968e+00 -6.20911181e-01 -2.75100976e-01
-3.64389181e-01 7.49269605e-01 7.20842838e-01 5.43085873e-01
-1.14889525e-01 -3.77219379e-01 5.88438392e-01 4.56126332e-02
4.87307757e-01 -1.00966096e-01 -3.46424818e-01 -1.26881790e+00
-4.00701910e-01 -9.26743686e-01 5.44848502e-01 -4.12189931e-01
-1.46162760e+00 2.13601053e-01 -1.34153038e-01 -8.91202629e-01
-2.78345764e-01 -3.97082120e-01 -9.25594568e-01 4.36818093e-01
-1.36831081e+00 -5.56799233e-01 -8.52999464e-03 5.82034767e-01
7.50860691e-01 -4.92379218e-01 7.87573338e-01 -1.41125560e-01
-4.54721719e-01 6.41409993e-01 2.72298217e-01 -1.68739825e-01
4.71063197e-01 -1.11414099e+00 2.55864393e-02 4.22090828e-01
7.30557561e-01 9.37107384e-01 7.72521973e-01 -3.67770135e-01
-1.49074519e+00 -8.97845745e-01 6.69840813e-01 -2.83355564e-01
8.59664917e-01 -1.58761397e-01 -1.07757258e+00 5.90586662e-01
2.85647631e-01 2.11304903e-01 9.04199243e-01 7.14999735e-01
-5.36026001e-01 -3.08741212e-01 -1.08967006e+00 5.47225893e-01
1.33032179e+00 -3.56854945e-01 -8.38600874e-01 3.40271205e-01
5.55539846e-01 9.36718658e-03 -5.64000010e-01 -3.06144804e-02
4.73301679e-01 -8.32431197e-01 8.45849037e-01 -1.22171605e+00
5.23360431e-01 1.01744249e-01 -1.11648940e-01 -1.82882941e+00
-8.60493779e-01 -6.65772378e-01 -3.47061723e-01 9.02690351e-01
7.36778975e-01 -7.39609003e-01 8.72282803e-01 6.48802102e-01
-6.58433959e-02 -8.04136336e-01 -5.80900192e-01 -8.37291181e-01
3.17816645e-01 -1.79169685e-01 2.86862910e-01 1.35305631e+00
3.50429416e-01 1.99091524e-01 -1.72442555e-01 -3.16222817e-01
6.11560225e-01 1.38623610e-01 2.90646404e-01 -1.44189417e+00
-2.92116165e-01 -5.83960235e-01 -1.66470259e-01 -7.20077336e-01
3.66304010e-01 -1.14972472e+00 -1.95476487e-01 -1.00849807e+00
2.83778250e-01 -5.24056911e-01 -8.27343166e-01 5.90246081e-01
-1.11335903e-01 5.35382703e-02 4.48366776e-02 3.16525698e-01
-7.38250375e-01 3.48946244e-01 1.06859136e+00 -1.90057769e-01
-5.16236365e-01 -1.17966980e-01 -1.20591664e+00 7.20602393e-01
1.08119607e+00 -8.13636124e-01 -5.14835775e-01 -4.84513432e-01
6.40151739e-01 -3.52029175e-01 -4.80317287e-02 -1.01363206e+00
5.18413544e-01 -7.47650981e-01 5.68393230e-01 2.33751535e-01
5.14191128e-02 -5.87685108e-01 -2.35280931e-01 5.03407657e-01
-8.44945967e-01 -1.36263808e-03 7.95167238e-02 6.06825471e-01
1.10198550e-01 -7.27314711e-01 8.27893794e-01 -6.56585217e-01
-8.31880867e-01 3.33317041e-01 -5.04297554e-01 5.56530178e-01
7.03069568e-01 -3.39616299e-01 -5.41920103e-02 -1.02384597e-01
-8.33212256e-01 1.23746566e-01 2.97552615e-01 2.40030080e-01
4.44200128e-01 -1.44447434e+00 -5.89632154e-01 1.40879467e-01
2.34795004e-01 -5.56074262e-01 -8.70245174e-02 4.35854048e-01
1.25084370e-01 -8.61863941e-02 -3.04776460e-01 -3.17847639e-01
-7.75163710e-01 3.75031680e-01 4.19974089e-01 5.19484319e-02
-6.76099658e-01 1.15044558e+00 -2.92605400e-01 -2.88864523e-01
3.33497494e-01 -6.23515472e-02 6.26653014e-03 3.42177063e-01
6.58913970e-01 2.78727263e-01 -1.88180313e-01 1.31177008e-01
-2.58439314e-02 4.98121113e-01 -4.33475614e-01 -1.21398987e-02
1.92904699e+00 1.29424438e-01 4.44107838e-02 7.38699138e-01
1.09746909e+00 -8.68958235e-02 -1.45322096e+00 -4.16053742e-01
1.77301720e-01 -4.12043661e-01 9.51582640e-02 -7.19761848e-01
-1.18748116e+00 5.38037717e-01 4.53551471e-01 4.31093782e-01
9.70221341e-01 -1.61907841e-02 4.13426906e-01 1.13294423e+00
4.12029028e-01 -1.60023034e+00 3.59706521e-01 5.91544688e-01
6.77637696e-01 -1.43215346e+00 2.06892937e-01 4.79738295e-01
-4.20085460e-01 1.35731137e+00 6.32804930e-01 -4.16944593e-01
8.31290483e-01 2.11060941e-01 -2.17720032e-01 -3.18918914e-01
-1.16916561e+00 -2.15935439e-01 1.41967118e-01 6.97682202e-01
4.96562243e-01 -1.27940401e-01 -6.14251867e-02 3.48320723e-01
-1.90520301e-01 1.53972566e-01 4.34111863e-01 1.06680524e+00
-1.07674420e+00 -1.22497928e+00 -6.63986951e-02 7.68976152e-01
-1.08380303e-01 -1.69553995e-01 -2.99469143e-01 5.34464598e-01
2.19365120e-01 6.86432660e-01 4.43682261e-02 -3.33573878e-01
1.46404520e-01 3.56044382e-01 7.53583670e-01 -1.06995308e+00
-9.70594943e-01 -4.48476017e-01 -1.26755893e-01 -4.35362577e-01
-5.63010633e-01 -6.10116363e-01 -1.03112245e+00 -2.81924218e-01
-2.71135747e-01 4.00677353e-01 3.11628491e-01 1.00745416e+00
2.25807190e-01 3.94722909e-01 8.62509847e-01 -8.24576080e-01
-7.43399322e-01 -6.41618252e-01 -6.51599884e-01 4.16224390e-01
5.21803081e-01 -3.91665399e-01 -5.20354569e-01 1.35296151e-01]
|
[8.887541770935059, 3.2533509731292725]
|
5ecefef2-fa4b-4b18-838b-36191d1e93c1
|
reference-based-oct-angiogram-super
|
2305.05835
| null |
https://arxiv.org/abs/2305.05835v1
|
https://arxiv.org/pdf/2305.05835v1.pdf
|
Reference-based OCT Angiogram Super-resolution with Learnable Texture Generation
|
Optical coherence tomography angiography (OCTA) is a new imaging modality to visualize retinal microvasculature and has been readily adopted in clinics. High-resolution OCT angiograms are important to qualitatively and quantitatively identify potential biomarkers for different retinal diseases accurately. However, one significant problem of OCTA is the inevitable decrease in resolution when increasing the field-of-view given a fixed acquisition time. To address this issue, we propose a novel reference-based super-resolution (RefSR) framework to preserve the resolution of the OCT angiograms while increasing the scanning area. Specifically, textures from the normal RefSR pipeline are used to train a learnable texture generator (LTG), which is designed to generate textures according to the input. The key difference between the proposed method and traditional RefSR models is that the textures used during inference are generated by the LTG instead of being searched from a single reference image. Since the LTG is optimized throughout the whole training process, the available texture space is significantly enlarged and no longer limited to a single reference image, but extends to all textures contained in the training samples. Moreover, our proposed LTGNet does not require a reference image at the inference phase, thereby becoming invulnerable to the selection of the reference image. Both experimental and visual results show that LTGNet has superior performance and robustness over state-of-the-art methods, indicating good reliability and promise in real-life deployment. The source code will be made available upon acceptance.
|
['Hao Chen', 'Carol Y. Cheung', 'An Ran Ran', 'Ziqi Tang', 'Dawei Yang', 'Yuyan Ruan']
|
2023-05-10
| null | null | null | null |
['texture-synthesis', 'reference-based-super-resolution']
|
['computer-vision', 'computer-vision']
|
[ 4.06073272e-01 -1.27482623e-01 1.21635027e-01 -1.80533215e-01
-6.32690132e-01 -2.09467590e-01 2.17929170e-01 -1.35492593e-01
-2.49989659e-01 8.10353339e-01 -1.67117402e-01 -3.02315563e-01
-1.22341178e-01 -9.63051796e-01 -4.17761296e-01 -8.45628262e-01
3.00415128e-01 1.87469080e-01 5.48394442e-01 1.42818570e-01
2.68557012e-01 6.92874908e-01 -1.73712707e+00 3.03007811e-01
1.26266837e+00 1.18651605e+00 5.82719326e-01 3.29013854e-01
-6.19264208e-02 5.13037026e-01 -4.22539473e-01 -1.08058661e-01
2.57672846e-01 -5.66945374e-01 -4.66896474e-01 2.61751950e-01
6.79209948e-01 -4.52577740e-01 1.99778397e-02 1.07681894e+00
7.43388057e-01 5.93526810e-02 3.23149294e-01 -3.22496057e-01
-3.03082734e-01 7.97895044e-02 -6.11447334e-01 3.87748986e-01
-9.15934667e-02 3.15052956e-01 6.10814214e-01 -7.15965033e-01
6.28630280e-01 9.05202866e-01 1.02462836e-01 5.72045982e-01
-1.43884456e+00 -4.85616088e-01 3.73245627e-02 1.40554354e-01
-1.34143806e+00 -4.96760219e-01 3.62759382e-01 -6.43299401e-01
3.98986131e-01 3.88286650e-01 9.87500072e-01 5.44345081e-01
3.24234992e-01 1.99143931e-01 1.56320369e+00 -3.96637797e-01
2.26921439e-01 8.57593343e-02 -2.50247985e-01 7.66314030e-01
3.66743416e-01 2.59899169e-01 -2.10747570e-01 3.92813571e-02
1.56560338e+00 -3.33677173e-01 -4.56907272e-01 -3.44171226e-01
-8.69991004e-01 5.07982731e-01 3.30196053e-01 9.58568007e-02
-5.01282990e-01 -1.50735393e-01 3.80681306e-02 -1.09110199e-01
3.62433583e-01 6.04831338e-01 1.18263856e-01 -8.50049555e-02
-8.63273799e-01 -2.26706527e-02 1.78513184e-01 4.08308685e-01
6.31194770e-01 -4.11415398e-02 -3.98823947e-01 1.05229938e+00
1.73639115e-02 4.02375191e-01 3.97776902e-01 -9.32789862e-01
7.87545964e-02 7.36062169e-01 2.05647260e-01 -7.66668856e-01
-1.09590493e-01 -8.31123173e-01 -8.64621043e-01 5.34757674e-01
5.65158904e-01 5.72912693e-02 -9.42310631e-01 1.12997842e+00
4.43063825e-01 3.37449163e-01 -3.81819129e-01 1.11886716e+00
7.33446360e-01 1.80910707e-01 -9.51535553e-02 -1.83870897e-01
1.38885486e+00 -6.58529878e-01 -3.11253965e-01 -2.30009586e-01
1.06381625e-01 -8.70465279e-01 1.18467820e+00 4.43428606e-01
-1.05276799e+00 -5.49709260e-01 -8.76596332e-01 7.41286799e-02
2.28172913e-01 6.40932381e-01 5.30367732e-01 5.20851672e-01
-9.38816845e-01 3.48815501e-01 -9.34998989e-01 -2.01412797e-01
6.23476326e-01 1.96908295e-01 -2.10561275e-01 -2.73612350e-01
-5.65330267e-01 7.57680357e-01 2.45025847e-02 3.31963301e-01
-4.60899264e-01 -6.34138763e-01 -5.01561642e-01 -2.81799398e-03
2.07135558e-01 -8.54259074e-01 8.72968137e-01 -5.25229514e-01
-1.85238135e+00 8.17785919e-01 -6.52821064e-01 -3.73874485e-01
5.97759247e-01 -2.63053905e-02 -2.54053116e-01 5.55769980e-01
1.02354482e-01 3.27343255e-01 1.03618705e+00 -1.05878603e+00
-5.90676963e-01 -2.42398202e-01 -1.13805659e-01 -1.48631647e-01
1.36190102e-01 -1.55369580e-01 -4.68792230e-01 -4.40595597e-01
3.12912911e-01 -7.91882098e-01 -4.14860427e-01 8.94593373e-02
-3.23080063e-01 2.32816815e-01 3.10778022e-01 -4.11613166e-01
1.26019490e+00 -1.99131906e+00 -2.71834046e-01 3.73580664e-01
3.84399354e-01 5.38069606e-01 -2.14357957e-01 -1.11912288e-01
-6.45155385e-02 -1.26847038e-02 5.16887307e-02 1.13637619e-01
-7.13972509e-01 -5.64182512e-02 -1.57293871e-01 3.67485374e-01
3.62743556e-01 7.70227849e-01 -8.55350554e-01 -5.13528943e-01
6.39924526e-01 5.67130864e-01 -5.43585479e-01 1.63926676e-01
-1.59232482e-01 1.21667790e+00 -7.54756570e-01 4.89617169e-01
8.75904381e-01 -6.39468014e-01 2.39472210e-01 -3.64020526e-01
-4.49098825e-01 2.55705595e-01 -1.12102652e+00 1.13080955e+00
-5.18509626e-01 4.30253208e-01 -3.64599764e-01 -5.95159888e-01
1.14787376e+00 7.60888532e-02 4.98991430e-01 -1.06118810e+00
1.75925359e-01 3.89413446e-01 2.46561497e-01 -5.48132896e-01
6.80341646e-02 -1.17278896e-01 7.25580394e-01 2.76593447e-01
-4.71972466e-01 1.03642726e-02 2.92499602e-01 -4.17654306e-01
7.15259731e-01 7.96524882e-02 3.36473167e-01 1.00504588e-02
7.87079871e-01 -7.98989385e-02 5.46058834e-01 7.09652781e-01
1.05899557e-01 7.31526852e-01 6.14676952e-01 -5.88704228e-01
-1.17577875e+00 -1.00576603e+00 -6.20020807e-01 1.07825801e-01
2.34566763e-01 -1.40955791e-01 -5.62475085e-01 -2.70887852e-01
-2.64024049e-01 3.06974560e-01 -5.64158916e-01 3.50607097e-01
-5.02298832e-01 -8.11253071e-01 5.93020804e-02 1.89885587e-01
8.03197265e-01 -6.89786971e-01 -1.07647526e+00 2.87472367e-01
-2.33941346e-01 -1.14948082e+00 -2.60240346e-01 -6.62507951e-01
-1.04443192e+00 -1.16502225e+00 -6.54760063e-01 -4.35985327e-01
1.07842195e+00 2.67956793e-01 7.74931967e-01 1.38538908e-02
-6.68986082e-01 -7.52894804e-02 -2.09921837e-01 -1.37005299e-01
-2.36166686e-01 -1.95946172e-01 -4.24653649e-01 5.36809742e-01
2.64747173e-01 -5.36353648e-01 -1.23184752e+00 4.64884520e-01
-4.80869412e-01 3.07056248e-01 8.35935891e-01 1.04530585e+00
1.19133067e+00 1.09464221e-01 3.84493828e-01 -9.30744052e-01
2.33201861e-01 6.85633570e-02 -1.12390733e+00 1.13812901e-01
-5.83773732e-01 -1.61282539e-01 5.14751375e-01 -3.91704738e-01
-1.12952793e+00 -1.02635130e-01 6.95980415e-02 -3.88110012e-01
-1.53763279e-01 4.58384275e-01 1.21166907e-01 -3.97446603e-01
6.82247758e-01 3.26036811e-01 4.37865555e-01 -5.09243011e-01
2.46152524e-02 5.14449239e-01 3.83404404e-01 -4.75676984e-01
4.37065840e-01 7.90144980e-01 3.26498240e-01 -8.78051579e-01
-7.62687147e-01 -3.60012740e-01 -4.92235780e-01 -3.29923213e-01
6.12274051e-01 -7.35578597e-01 -8.47670615e-01 5.10838151e-01
-7.95455039e-01 -2.62070179e-01 -1.89301595e-01 6.75104260e-01
-3.52410406e-01 3.06852549e-01 -3.76460910e-01 -6.94819808e-01
-3.01825613e-01 -1.43289495e+00 1.03039384e+00 4.91917402e-01
-1.78495757e-02 -7.56114185e-01 -2.59341508e-01 4.63606358e-01
5.47895312e-01 3.53016704e-01 1.31995976e+00 1.35260910e-01
-1.11566794e+00 -1.78857580e-01 -6.07035160e-01 4.64167863e-01
3.71283144e-01 5.19653410e-02 -8.35634172e-01 -2.97654629e-01
-1.18010277e-02 1.15234502e-01 7.00279772e-01 8.38628888e-01
1.25799453e+00 -1.14870854e-01 -2.40984365e-01 7.30748534e-01
1.71316504e+00 3.33243310e-01 1.11803198e+00 4.82164383e-01
4.26202953e-01 6.12262905e-01 6.69546843e-01 4.01945025e-01
-5.49761318e-02 7.97419608e-01 1.87879026e-01 -4.78248328e-01
-4.46250618e-01 -1.68008924e-01 -2.48986363e-01 2.33717024e-01
-4.18981820e-01 1.37706429e-01 -6.72188044e-01 3.90619636e-01
-1.57047307e+00 -8.16370308e-01 -1.39022991e-01 2.79000449e+00
8.50492299e-01 -4.02712114e-02 -1.09407872e-01 -2.89857954e-01
7.88818955e-01 -1.34164587e-01 -7.19924510e-01 -2.66370308e-02
-4.35255170e-02 7.81839252e-01 3.38454425e-01 3.69361520e-01
-7.97790825e-01 8.24810684e-01 5.99735022e+00 7.04299808e-01
-1.75912499e+00 -1.21225677e-01 7.16573656e-01 -2.48752549e-01
-2.80708149e-02 1.02214515e-01 -6.59045875e-01 4.42620695e-01
4.60255355e-01 -1.08803920e-01 3.81814569e-01 4.67075020e-01
8.33963871e-01 -5.94266474e-01 -7.64958918e-01 1.04260218e+00
-3.53600919e-01 -1.51600003e+00 2.55748928e-01 2.17781529e-01
5.90224266e-01 -1.52061164e-01 1.64932013e-01 -4.46328789e-01
-2.55644500e-01 -1.20994210e+00 1.16640307e-01 7.62028933e-01
1.36097145e+00 -4.75104630e-01 7.71305978e-01 -6.13494404e-02
-9.43969190e-01 8.41860548e-02 -5.67897558e-01 2.00230420e-01
2.08968855e-02 9.00856256e-01 -1.02394867e+00 4.36073869e-01
5.17682374e-01 5.99401295e-01 -7.07714975e-01 1.47522807e+00
-2.14327335e-01 5.97143173e-01 -4.77470225e-03 2.85599113e-01
-1.25145391e-01 -5.67600965e-01 8.20981979e-01 5.59635282e-01
4.50749546e-01 -7.12053031e-02 5.99604994e-02 1.21993399e+00
4.91333097e-01 2.73380667e-01 -1.54902637e-01 3.57871354e-02
6.50190234e-01 1.22301936e+00 -6.52432501e-01 -1.04957514e-01
-4.23873186e-01 3.86368066e-01 7.84619004e-02 5.22416890e-01
-5.05064487e-01 -3.34165782e-01 4.30456340e-01 8.44186962e-01
3.46495748e-01 7.32487962e-02 -3.72948796e-01 -1.04429579e+00
2.43229017e-01 -8.12642634e-01 1.56256407e-02 -7.58714020e-01
-1.20372880e+00 9.25102711e-01 -2.37354666e-01 -1.66516590e+00
-4.94517609e-02 -4.38785106e-01 -4.09629822e-01 1.49602580e+00
-1.78886425e+00 -1.10235918e+00 -6.80385470e-01 3.56066346e-01
1.30394444e-01 -1.46873698e-01 7.25334346e-01 2.29702219e-01
-6.11079454e-01 4.18433130e-01 -4.30195034e-02 -1.58102531e-02
6.43675268e-01 -9.42118585e-01 6.19991347e-02 9.91687953e-01
-4.35394019e-01 9.02702510e-01 4.00135368e-01 -6.41731083e-01
-9.42136288e-01 -1.17453551e+00 5.07319510e-01 -1.58668607e-01
4.48285252e-01 6.05457798e-02 -8.87628973e-01 2.28741378e-01
-3.47242385e-01 2.61344820e-01 4.80570644e-01 -2.43583426e-01
1.72583722e-02 -4.58637267e-01 -1.06429851e+00 6.66996181e-01
6.34947181e-01 -2.50590086e-01 -1.11407220e-01 -3.99855338e-02
8.71767253e-02 -8.39767456e-01 -8.87780488e-01 5.02638519e-01
7.25890219e-01 -1.16138971e+00 6.49686217e-01 -1.62469491e-01
5.14146507e-01 -5.57739437e-01 2.77672231e-01 -9.59712923e-01
-2.41680115e-01 -6.69360876e-01 1.95207521e-01 8.56440604e-01
2.24875018e-01 -1.29323506e+00 8.44924748e-01 4.52235699e-01
5.14158346e-02 -9.87260878e-01 -9.62925017e-01 -6.37817562e-01
-2.98472315e-01 1.42463028e-01 6.67467892e-01 4.63052392e-01
-4.40806717e-01 1.26034036e-01 -3.95415202e-02 3.88146937e-01
7.78041780e-01 4.85757202e-01 7.61860192e-01 -1.34107375e+00
-4.18217033e-01 -3.34336698e-01 -4.44438756e-01 -1.14432502e+00
-4.45394337e-01 -6.51844740e-01 -1.20907396e-01 -1.55406821e+00
1.10591583e-01 -9.05418217e-01 -1.85428932e-01 1.78641081e-01
-2.45127022e-01 3.33490014e-01 -1.61056876e-01 3.28442633e-01
1.14607997e-01 2.13027373e-01 2.18376875e+00 2.14998305e-01
-4.97201383e-01 3.47829640e-01 -7.27713645e-01 4.19461668e-01
6.14500821e-01 -1.44488946e-01 -6.43413544e-01 -2.31807783e-01
1.15810204e-02 1.89108089e-01 7.25879669e-01 -8.19543540e-01
7.42779598e-02 -1.65269941e-01 4.46119308e-01 -3.22037160e-01
1.44567356e-01 -3.74428421e-01 2.94891417e-01 2.21490100e-01
-1.03144888e-02 -4.77250934e-01 1.62126318e-01 3.28600824e-01
-3.16187859e-01 -1.14391290e-01 1.23297513e+00 -1.74131230e-01
-4.14876640e-01 4.59632814e-01 -2.25210771e-01 -1.06095806e-01
7.01043367e-01 -6.51399851e-01 -5.97256422e-01 -3.35876942e-02
-7.38832295e-01 -7.40434900e-02 6.06835425e-01 -4.75964174e-02
7.93187499e-01 -8.02335799e-01 -7.90277898e-01 4.98899370e-01
8.49718899e-02 2.58178622e-01 6.99525714e-01 1.22433710e+00
-6.91321313e-01 4.62586164e-01 -3.53366524e-01 -8.06916595e-01
-1.09256899e+00 1.21315755e-01 7.14643002e-01 -1.43460497e-01
-1.03091419e+00 5.61736166e-01 4.02713448e-01 2.87228793e-01
-1.77572057e-01 -3.45194370e-01 -3.69337857e-01 -3.60797793e-01
8.13624382e-01 2.74688840e-01 1.43370569e-01 -2.98857272e-01
6.14519231e-02 9.15898681e-01 -2.48172328e-01 1.83372200e-01
1.17287230e+00 -2.36902654e-01 -4.39157933e-01 5.57111762e-02
5.21006346e-01 2.23099723e-01 -1.26743364e+00 -2.30241716e-01
-4.28318828e-01 -9.72280204e-01 2.59107739e-01 -8.46224964e-01
-1.09037232e+00 8.52564037e-01 8.46474171e-01 -1.02925718e-01
1.38573027e+00 -2.99995512e-01 7.54040301e-01 -3.14053118e-01
6.08763218e-01 -6.58988714e-01 -7.42399991e-02 2.67236382e-02
7.62439609e-01 -8.60839427e-01 5.22487238e-02 -9.86545324e-01
-5.72229266e-01 1.16057444e+00 7.25796878e-01 -7.34508634e-02
3.13933253e-01 5.01250057e-03 1.77487284e-01 -1.20909259e-01
-5.39433479e-01 -3.25463355e-01 3.68301660e-01 7.00193107e-01
3.76201361e-01 -8.10334384e-02 -4.64169234e-01 2.16529131e-01
-1.17244646e-01 4.08940762e-01 6.42111361e-01 3.83440018e-01
-3.47819895e-01 -1.34274912e+00 -6.60322681e-02 7.69288123e-01
-3.70823503e-01 -1.94690451e-01 2.39313424e-01 5.08708417e-01
1.92439601e-01 7.91476011e-01 1.52945369e-01 1.24054011e-02
3.32228959e-01 -3.48782271e-01 7.03329921e-01 -9.07376468e-01
-6.49206787e-02 5.04257917e-01 -1.12859113e-02 -6.49254024e-01
-3.87162805e-01 -5.50351322e-01 -9.03809607e-01 1.32316742e-02
-2.86728919e-01 2.46102475e-02 2.85605282e-01 6.54985189e-01
6.34805441e-01 5.58956146e-01 6.54778361e-01 -4.96945053e-01
-1.39760241e-01 -8.07991445e-01 -6.88045204e-01 3.58379595e-02
4.09299135e-01 -7.80215442e-01 -1.28859207e-01 9.33023989e-02]
|
[15.73158073425293, -3.9257242679595947]
|
e4600439-17c7-4e1e-b644-fefe9f865196
|
accuracy-privacy-trade-off-in-deep-ensemble
|
2105.05381
| null |
https://arxiv.org/abs/2105.05381v4
|
https://arxiv.org/pdf/2105.05381v4.pdf
|
Accuracy-Privacy Trade-off in Deep Ensemble: A Membership Inference Perspective
|
Deep ensemble learning has been shown to improve accuracy by training multiple neural networks and averaging their outputs. Ensemble learning has also been suggested to defend against membership inference attacks that undermine privacy. In this paper, we empirically demonstrate a trade-off between these two goals, namely accuracy and privacy (in terms of membership inference attacks), in deep ensembles. Using a wide range of datasets and model architectures, we show that the effectiveness of membership inference attacks increases when ensembling improves accuracy. We analyze the impact of various factors in deep ensembles and demonstrate the root cause of the trade-off. Then, we evaluate common defenses against membership inference attacks based on regularization and differential privacy. We show that while these defenses can mitigate the effectiveness of membership inference attacks, they simultaneously degrade ensemble accuracy. We illustrate similar trade-off in more advanced and state-of-the-art ensembling techniques, such as snapshot ensembles and diversified ensemble networks. Finally, we propose a simple yet effective defense for deep ensembles to break the trade-off and, consequently, improve the accuracy and privacy, simultaneously.
|
['Xin Liu', 'Zubair Shafiq', 'Shahbaz Rezaei']
|
2021-05-12
|
accuracy-privacy-trade-off-in-deep-ensemble-a
|
https://openreview.net/forum?id=wxVpa5z4DU1
|
https://openreview.net/pdf?id=wxVpa5z4DU1
| null |
['membership-inference-attack']
|
['computer-vision']
|
[ 6.88057840e-02 -2.29736254e-01 3.31227839e-01 -5.02558470e-01
-6.77655995e-01 -1.12181413e+00 4.71492738e-01 7.17864197e-04
-4.95691329e-01 8.77103508e-01 -6.38142228e-02 -6.92738175e-01
-2.70713747e-01 -8.10103953e-01 -7.88547277e-01 -8.65901887e-01
-3.35580975e-01 -1.11459307e-01 -4.35623646e-01 -2.47859687e-01
2.13089645e-01 5.71408093e-01 -1.46536994e+00 2.57425427e-01
1.04356551e+00 9.92850244e-01 -1.09499991e+00 7.00905561e-01
4.42888439e-01 5.89481235e-01 -1.07183933e+00 -8.65811288e-01
6.88637316e-01 -3.98615658e-01 -3.55709195e-01 -5.46679020e-01
7.50030518e-01 -6.92944765e-01 -4.01445001e-01 1.18511152e+00
6.26757741e-01 1.15979770e-02 7.58310139e-01 -1.61432362e+00
-4.47504997e-01 6.99226677e-01 -4.98524636e-01 1.41670212e-01
-2.76341081e-01 7.11420402e-02 9.29850519e-01 -2.43241087e-01
-1.39642909e-01 9.78444159e-01 1.01673079e+00 6.97134614e-01
-1.44065750e+00 -1.33724451e+00 5.70738241e-02 -3.57797652e-01
-1.36238730e+00 -7.17382967e-01 4.49309051e-01 -2.65781522e-01
4.14953560e-01 7.06572711e-01 2.85811931e-01 1.31543720e+00
2.10249349e-01 4.53963816e-01 1.24094665e+00 -1.15529805e-01
4.47412521e-01 2.11579621e-01 4.03177619e-01 4.72403288e-01
7.37060070e-01 5.23558497e-01 -3.12920779e-01 -8.88665438e-01
5.01830101e-01 -1.64437331e-02 -3.34091425e-01 -3.14427823e-01
-3.41774017e-01 7.89016485e-01 2.22307324e-01 -6.72509670e-02
-1.13137081e-01 3.42954099e-01 5.85538685e-01 5.86271703e-01
4.98487353e-01 6.73424959e-01 -5.58366060e-01 1.65287599e-01
-9.18910265e-01 2.86566883e-01 1.19546938e+00 4.43187118e-01
5.44818759e-01 -1.23302275e-02 -1.84483573e-01 3.66020888e-01
1.09941565e-01 3.79606068e-01 -3.39198075e-02 -1.09866202e+00
3.21702033e-01 1.78015843e-01 2.33064532e-01 -1.07522345e+00
-2.25951746e-02 -6.78645849e-01 -1.07843542e+00 5.97863734e-01
6.96385086e-01 -7.12544322e-01 -5.59301257e-01 2.43083119e+00
2.08889276e-01 4.24996853e-01 1.10418543e-01 5.18050969e-01
5.02626523e-02 9.09701139e-02 1.36941999e-01 -7.07977191e-02
1.09172213e+00 -3.31427574e-01 -4.57292795e-01 2.60243624e-01
5.71857929e-01 -8.14541876e-02 6.33388102e-01 3.11094463e-01
-8.43677104e-01 -4.56499383e-02 -1.41852725e+00 3.13277215e-01
-4.39699352e-01 -2.41397128e-01 5.57681262e-01 1.50984395e+00
-9.71698642e-01 9.89878833e-01 -9.60238814e-01 1.99564043e-02
6.77488565e-01 6.46398127e-01 -2.01351225e-01 4.85651225e-01
-1.45470548e+00 6.88505113e-01 1.57207455e-02 -1.53294532e-02
-8.54751825e-01 -7.61970282e-01 -4.68222111e-01 3.61780107e-01
1.38533428e-01 -9.80060518e-01 8.88604581e-01 -1.05377877e+00
-1.26511669e+00 5.54402769e-01 3.05768132e-01 -1.05107391e+00
9.44693446e-01 -5.76307833e-01 -5.06719887e-01 -4.05690938e-01
-4.10366654e-01 1.44211829e-01 8.74676585e-01 -1.31845856e+00
-3.19340616e-01 -8.64779830e-01 2.03777507e-01 4.22369726e-02
-7.62788117e-01 -2.17484117e-01 4.47547674e-01 -5.66161811e-01
-5.36717236e-01 -9.38252509e-01 -2.97930121e-01 -8.93978626e-02
-6.08101249e-01 2.24215433e-01 1.01025486e+00 -3.73429298e-01
1.40343428e+00 -2.21930528e+00 -2.30348498e-01 5.40173471e-01
4.96409059e-01 3.15603286e-01 4.59841564e-02 1.98166162e-01
1.18512429e-01 6.01922512e-01 -4.37115103e-01 -5.29256523e-01
9.93437394e-02 2.05140486e-01 -6.23259962e-01 6.75983727e-01
-6.67244792e-02 5.84258974e-01 -4.57117438e-01 8.76015499e-02
-4.12788354e-02 4.50625896e-01 -7.49031007e-01 -3.35871875e-02
-9.73385498e-02 5.45973539e-01 -2.79610991e-01 3.13262939e-01
9.17890131e-01 -8.34436119e-02 4.50510740e-01 -1.18217833e-01
7.17584118e-02 2.34419450e-01 -8.66441011e-01 9.63496149e-01
-3.86437416e-01 4.84165639e-01 3.84177685e-01 -9.76288974e-01
6.72933757e-01 1.45077571e-01 1.47306114e-01 1.37657337e-02
3.93655956e-01 3.01821440e-01 2.96898484e-01 7.47543722e-02
4.12028909e-01 3.22676189e-02 -2.39240989e-01 8.47119033e-01
-3.04874152e-01 4.41077530e-01 -3.37796926e-01 1.77452058e-01
1.14465344e+00 -2.32926846e-01 1.51229545e-01 -6.09351754e-01
4.90717918e-01 -5.88369012e-01 3.57363552e-01 1.43229795e+00
-4.35332775e-01 2.14788675e-01 5.37488043e-01 -3.62662107e-01
-9.80110109e-01 -1.04441154e+00 -1.50980830e-01 1.32529175e+00
1.04182944e-01 -3.04648191e-01 -1.01711202e+00 -1.02408051e+00
4.86636311e-01 7.88682580e-01 -9.45412815e-01 -4.12116349e-01
-3.37684184e-01 -1.07017136e+00 1.39004731e+00 4.81494784e-01
7.92972922e-01 -3.12163472e-01 -7.03374922e-01 -3.39119822e-01
4.64984067e-02 -6.67076707e-01 -3.45276684e-01 6.44406602e-02
-8.23365927e-01 -8.81525934e-01 -1.92478150e-01 1.49537742e-01
5.60137987e-01 -2.88054608e-02 1.19893754e+00 3.39742631e-01
2.06647385e-02 3.78970116e-01 6.48678541e-02 -4.82280791e-01
-6.70334637e-01 1.64082378e-01 5.57129323e-01 1.08638085e-01
2.02085719e-01 -1.19641089e+00 -6.87794447e-01 2.94903040e-01
-9.80720282e-01 -4.85580236e-01 3.21752876e-01 9.15087819e-01
1.71839707e-02 4.41176184e-02 7.35062540e-01 -1.02937138e+00
9.76900399e-01 -5.22189498e-01 -6.83550060e-01 5.53115129e-01
-6.61498129e-01 3.41971815e-01 6.84551477e-01 -4.42969054e-01
-7.81671107e-01 -2.84702778e-01 -3.37114751e-01 -5.99416792e-01
-9.46459733e-03 1.82021543e-01 -2.83735335e-01 -3.48240614e-01
9.18463349e-01 1.18962467e-01 2.58213490e-01 -3.09730709e-01
5.39433241e-01 7.51658559e-01 5.67070603e-01 -9.99884903e-01
5.37912190e-01 6.58929408e-01 2.87847165e-02 -4.01944011e-01
-8.92121494e-01 2.36020684e-01 -2.39924043e-01 4.66637649e-02
4.64208454e-01 -8.47853720e-01 -1.12660205e+00 6.25658035e-01
-8.82134199e-01 -2.71841846e-02 -5.81283234e-02 6.51878305e-03
-3.32346052e-01 5.20234823e-01 -6.68932796e-01 -1.17281783e+00
-7.83331633e-01 -9.54617321e-01 5.46616912e-01 1.57496557e-01
-1.72878131e-01 -8.74288857e-01 -4.51179929e-02 1.49440661e-03
6.35379791e-01 6.03877962e-01 7.71821797e-01 -1.26083648e+00
-3.30805540e-01 -1.83935508e-01 -1.34741440e-01 5.49829483e-01
1.47668850e-02 1.14210159e-01 -1.13967657e+00 -6.08266592e-01
1.31657645e-01 -2.93946624e-01 1.03047419e+00 1.60046205e-01
1.65234983e+00 -7.75421083e-01 -7.91690275e-02 9.37936723e-01
1.19101191e+00 8.27179626e-02 7.03852475e-01 1.87067211e-01
4.29615676e-01 4.50815290e-01 3.21583077e-02 6.00784361e-01
-3.58686480e-03 3.85325670e-01 5.33219039e-01 1.40968695e-01
6.25219047e-01 -1.51844665e-01 3.66810113e-01 2.19331369e-01
-4.93661463e-02 -2.08832309e-01 -6.28584087e-01 2.17928644e-02
-1.84666431e+00 -1.07112086e+00 2.27128372e-01 2.53358865e+00
8.67043495e-01 5.88948987e-02 4.35766488e-01 3.52174751e-02
6.72902405e-01 2.53138393e-02 -8.59350920e-01 -4.79629874e-01
-1.73199520e-01 1.49611980e-01 8.53495121e-01 3.72004539e-01
-1.33905637e+00 5.57936490e-01 6.66165400e+00 7.26348102e-01
-1.02118051e+00 1.57395214e-01 1.19954193e+00 -2.52751380e-01
-2.67412186e-01 -1.77877799e-01 -5.95845938e-01 5.60594559e-01
1.01786160e+00 -2.59976923e-01 5.33605576e-01 7.98393071e-01
-4.72607195e-01 4.16516304e-01 -1.25079095e+00 7.26398349e-01
-8.67271051e-02 -1.29444921e+00 -2.28109941e-01 4.27186728e-01
6.70025945e-01 -2.98797768e-02 5.41554511e-01 4.63058054e-01
9.00415838e-01 -1.12592518e+00 3.69816124e-01 1.26243979e-01
8.69701743e-01 -1.22089565e+00 5.54399133e-01 3.75887871e-01
-5.76576114e-01 -3.91277909e-01 2.53794505e-03 -1.72638267e-01
-1.52027115e-01 5.72092891e-01 -1.45467982e-01 5.96216500e-01
6.80322051e-01 1.05744220e-01 -3.75053644e-01 5.64346671e-01
4.22578910e-03 5.20573795e-01 -6.87700927e-01 1.01382114e-01
1.93919577e-02 -2.53638893e-01 6.96790695e-01 8.87314737e-01
2.96175838e-01 1.37977466e-01 -1.61637992e-01 1.08930182e+00
-3.34517539e-01 -4.85221893e-01 -6.97685003e-01 -5.55106997e-02
1.03617179e+00 9.83824909e-01 -5.44329770e-02 -2.54046202e-01
1.75031394e-01 7.86066473e-01 4.75860834e-01 3.37670982e-01
-9.67873752e-01 -2.27859080e-01 1.44086826e+00 -1.75591022e-01
6.03177510e-02 -1.15100712e-01 -7.71654665e-01 -1.25855422e+00
-3.33548814e-01 -1.27549148e+00 6.08582497e-01 4.72413376e-02
-1.65371525e+00 3.82813960e-01 -1.27184704e-01 -8.53773475e-01
-4.58560064e-02 -4.22004133e-01 -6.49354100e-01 8.83411288e-01
-7.44377613e-01 -8.28040063e-01 8.05648938e-02 5.16528845e-01
-2.98342735e-01 -1.54017359e-01 9.25489843e-01 2.36862987e-01
-8.85948122e-01 1.49605644e+00 5.12761354e-01 2.45565027e-01
5.56846738e-01 -1.02134943e+00 5.78198016e-01 1.00764143e+00
2.29928598e-01 1.20218992e+00 6.90364480e-01 -4.39779341e-01
-1.14272523e+00 -1.04616475e+00 1.21716713e-03 -8.74154985e-01
5.00094533e-01 -4.98993248e-01 -8.94213140e-01 6.50398910e-01
8.06373507e-02 -1.59743398e-01 9.95201826e-01 4.99289930e-01
-9.86815512e-01 -1.13817185e-01 -1.61652470e+00 7.34616637e-01
9.25977826e-01 -5.84856451e-01 -1.54711574e-01 -1.90626279e-01
6.64863646e-01 -3.01758587e-01 -8.27651322e-01 7.86456585e-01
1.15976155e+00 -1.46236300e+00 1.11679769e+00 -1.00229180e+00
3.06457043e-01 2.09380835e-02 -3.81221026e-01 -1.24799705e+00
4.45270725e-02 -7.37427056e-01 -6.17597342e-01 1.18169761e+00
4.22591358e-01 -1.13772321e+00 8.28969717e-01 8.32060218e-01
4.54641819e-01 -7.79183328e-01 -9.64925289e-01 -7.56927788e-01
4.93889928e-01 -1.09716102e-01 1.01147449e+00 1.19018495e+00
-3.09259593e-01 9.27561074e-02 -7.04606175e-01 4.48628455e-01
9.58335280e-01 -8.31974521e-02 9.37685609e-01 -1.10357392e+00
-4.39953119e-01 -5.29904306e-01 -2.18002453e-01 -7.38442481e-01
2.35033080e-01 -4.18161541e-01 -2.39141077e-01 -3.69801790e-01
6.28372952e-02 -6.91314161e-01 -7.45243132e-01 4.26415026e-01
-3.62345636e-01 1.39453068e-01 2.80855060e-01 3.13983172e-01
-4.19546604e-01 2.64646858e-01 4.53082085e-01 1.27102062e-01
-2.33181536e-01 1.64927199e-01 -1.22948551e+00 5.28198004e-01
1.02683401e+00 -4.25715595e-01 -3.60547096e-01 -2.19469532e-01
1.99019581e-01 -1.32133067e-01 5.19981205e-01 -1.04736936e+00
1.31319448e-01 1.48565784e-01 4.14618611e-01 -3.65479797e-01
2.28412926e-01 -8.01233411e-01 3.39219570e-01 5.99298418e-01
-6.12798154e-01 -2.34575104e-02 3.05905759e-01 8.01232338e-01
3.08050156e-01 2.91590452e-01 9.50780332e-01 1.91762537e-01
1.06436916e-01 3.24918389e-01 -8.09422880e-02 1.13326401e-01
1.01850712e+00 2.91996039e-02 -6.18529975e-01 -6.19705081e-01
-6.58847034e-01 3.78099769e-01 5.78605354e-01 1.47232279e-01
2.13130862e-01 -1.10272121e+00 -8.33017826e-01 3.33102852e-01
-2.99860924e-01 -4.06371295e-01 3.44737619e-01 8.69602859e-01
-2.37530395e-01 -2.85055190e-01 -1.96542412e-01 -3.48299950e-01
-1.52361166e+00 3.32695961e-01 8.15133095e-01 -3.69056106e-01
-1.63908139e-01 1.03765070e+00 3.31242383e-01 -5.39404750e-01
3.62094611e-01 1.94094092e-01 2.61678010e-01 -2.66952962e-01
6.49412096e-01 4.55495417e-01 -1.93199106e-02 -1.72156870e-01
-4.93453771e-01 -1.68495566e-01 -2.66035914e-01 -1.46481484e-01
9.14017379e-01 -5.13765328e-02 -2.31086001e-01 -7.53347874e-02
1.02923536e+00 1.69134721e-01 -1.25745523e+00 1.42524585e-01
-2.14949459e-01 -5.58649778e-01 -2.17462450e-01 -1.00902426e+00
-9.49277222e-01 8.70369196e-01 6.18033350e-01 5.96479416e-01
1.19294548e+00 -5.39038956e-01 7.25933492e-01 3.66336018e-01
2.83685446e-01 -5.79101086e-01 -3.98332000e-01 4.31023300e-01
4.37529057e-01 -9.48400855e-01 8.55646506e-02 -2.25311577e-01
-3.69329423e-01 6.42246604e-01 6.24401510e-01 -3.09718341e-01
7.38815367e-01 6.10363543e-01 -6.62682131e-02 1.46977946e-01
-7.14144230e-01 3.68095994e-01 -3.22388634e-02 5.20000994e-01
1.65913746e-01 2.84920752e-01 -2.07546011e-01 8.53736103e-01
-3.50428402e-01 -3.62865210e-01 2.59708315e-01 6.54037952e-01
-5.24181649e-02 -1.10453367e+00 -5.30034065e-01 6.73307717e-01
-1.02137613e+00 -1.65923327e-01 -7.50298917e-01 6.51382208e-01
6.45461902e-02 8.58623505e-01 2.03251675e-01 -7.25312471e-01
-3.21249478e-02 3.10810525e-02 4.51419502e-01 7.43725598e-02
-1.04064846e+00 -5.08297086e-01 3.00258458e-01 -2.80900091e-01
-9.60681587e-02 -4.20881897e-01 -4.89113182e-01 -9.69256043e-01
-4.66770738e-01 2.24631980e-01 4.66388762e-01 8.26545537e-01
8.09931576e-01 2.83087581e-01 8.36321592e-01 -4.65709329e-01
-1.61286378e+00 -6.61592662e-01 -6.43317461e-01 3.64942998e-01
4.77850914e-01 -4.73805398e-01 -8.37341249e-01 -6.20146334e-01]
|
[5.99124813079834, 7.094431400299072]
|
9ec450d1-02f1-41af-97cb-046a7a72418e
|
sherlock-scalable-fact-learning-in-images
|
1511.04891
| null |
http://arxiv.org/abs/1511.04891v4
|
http://arxiv.org/pdf/1511.04891v4.pdf
|
Sherlock: Scalable Fact Learning in Images
|
We study scalable and uniform understanding of facts in images. Existing
visual recognition systems are typically modeled differently for each fact type
such as objects, actions, and interactions. We propose a setting where all
these facts can be modeled simultaneously with a capacity to understand
unbounded number of facts in a structured way. The training data comes as
structured facts in images, including (1) objects (e.g., $<$boy$>$), (2)
attributes (e.g., $<$boy, tall$>$), (3) actions (e.g., $<$boy, playing$>$), and
(4) interactions (e.g., $<$boy, riding, a horse $>$). Each fact has a semantic
language view (e.g., $<$ boy, playing$>$) and a visual view (an image with this
fact). We show that learning visual facts in a structured way enables not only
a uniform but also generalizable visual understanding. We propose and
investigate recent and strong approaches from the multiview learning literature
and also introduce two learning representation models as potential baselines.
We applied the investigated methods on several datasets that we augmented with
structured facts and a large scale dataset of more than 202,000 facts and
814,000 images. Our experiments show the advantage of relating facts by the
structure by the proposed models compared to the designed baselines on
bidirectional fact retrieval.
|
['Walter Chang', 'Mohamed Elhoseiny', 'Brian Price', 'Ahmed Elgammal', 'Scott Cohen']
|
2015-11-16
| null | null | null | null |
['multiview-learning']
|
['computer-vision']
|
[-1.67600840e-01 2.51877010e-01 -2.83619225e-01 -6.93852544e-01
-6.56580687e-01 -7.56450891e-01 7.94215918e-01 3.37465584e-01
-1.30948182e-02 8.41477931e-01 4.06640291e-01 -2.45298773e-01
-1.35874733e-01 -9.13996994e-01 -1.60118842e+00 -6.47976875e-01
-3.62813622e-01 6.33388162e-01 1.22430986e-02 -1.14320092e-01
1.62913322e-01 2.70703416e-02 -2.07894611e+00 8.07263613e-01
3.10011655e-01 1.30591345e+00 3.79787356e-01 4.13922161e-01
-4.06747133e-01 1.40370929e+00 -4.65489149e-01 -1.00655520e+00
2.45986760e-01 -1.45222902e-01 -1.02352595e+00 3.18161428e-01
1.20707130e+00 -5.42339087e-01 -3.81844521e-01 1.03160059e+00
-3.12699303e-02 1.05153546e-01 6.71022773e-01 -1.33435237e+00
-1.36578345e+00 7.99885333e-01 -7.99230456e-01 4.18701380e-01
6.74484193e-01 1.33295462e-01 1.45148718e+00 -1.03428066e+00
1.00435436e+00 1.46630096e+00 -1.44672617e-01 2.38827154e-01
-8.61521721e-01 -7.02518582e-01 7.13214338e-01 5.68711162e-01
-1.27810669e+00 -8.45796317e-02 5.65628886e-01 -4.60215867e-01
8.86338115e-01 2.61547506e-01 8.23226392e-01 9.99081194e-01
-4.18755189e-02 9.73258972e-01 1.00317824e+00 -1.21250585e-01
7.72949308e-02 2.90727466e-01 2.39041850e-01 9.59342837e-01
3.71599376e-01 -1.64670661e-01 -9.22952592e-01 1.91969648e-01
5.67639768e-01 2.20264256e-01 -1.30943760e-01 -7.18838930e-01
-1.24304962e+00 7.22601235e-01 7.00667679e-01 1.66076615e-01
-3.31854820e-01 2.56274492e-01 3.11199397e-01 3.09677541e-01
3.01470190e-01 -1.45817772e-01 -5.42633951e-01 2.68668622e-01
-3.57858241e-01 1.73896536e-01 5.12117624e-01 1.45948601e+00
1.06872582e+00 1.25037000e-01 1.12562634e-01 6.85461104e-01
-7.11698383e-02 8.37235868e-01 6.61543831e-02 -1.00105155e+00
8.82730901e-01 8.09324026e-01 2.26949394e-01 -1.32313943e+00
-1.69542968e-01 -1.55152127e-01 -8.76855850e-01 7.74936937e-03
1.15659788e-01 3.27919811e-01 -1.05546534e+00 1.86575353e+00
2.97761083e-01 6.40637353e-02 2.30675325e-01 7.14430034e-01
1.42157793e+00 7.30132937e-01 2.35126063e-01 -1.07185185e-01
1.60297787e+00 -9.27751064e-01 -4.46020812e-01 -1.92073643e-01
3.98164690e-01 -5.81200123e-01 1.27030432e+00 3.01006615e-01
-1.32407188e+00 -5.97398400e-01 -5.66474617e-01 -4.79216099e-01
-9.97103274e-01 3.10195774e-01 8.82989109e-01 3.28461140e-01
-1.17584109e+00 -4.39041592e-02 -2.23499477e-01 -5.33343613e-01
6.33798599e-01 2.01769873e-01 -8.22271347e-01 -6.92829311e-01
-8.32780123e-01 7.29516864e-01 5.57299316e-01 -2.78908193e-01
-1.22048771e+00 -6.26738429e-01 -1.16474140e+00 2.00055510e-01
9.73610640e-01 -1.09713340e+00 7.32508600e-01 -1.04415858e+00
-6.45920575e-01 1.49814582e+00 -3.32827896e-01 -4.09678847e-01
1.75349116e-01 -3.92651916e-01 -5.13597608e-01 5.09313285e-01
4.04289842e-01 7.52518833e-01 7.66738832e-01 -2.02381277e+00
-9.21931565e-01 -5.80639243e-01 1.08831251e+00 3.19100857e-01
-1.36084333e-01 -3.49333473e-02 -4.09148365e-01 -3.70166183e-01
4.73418273e-02 -5.40679395e-01 5.13458788e-01 9.80771855e-02
-5.90678573e-01 -1.21579701e-02 9.35432553e-01 -6.95510566e-01
8.42194498e-01 -2.01101756e+00 1.15355857e-01 4.67203222e-02
5.13936043e-01 9.69127864e-02 4.03460339e-02 5.42114437e-01
-3.32201302e-01 1.43292457e-01 -7.86795840e-02 -4.01086628e-01
-6.87488541e-02 7.08749592e-01 -4.68458295e-01 8.99415463e-02
-2.02727243e-01 1.03123069e+00 -1.00295830e+00 -6.44541085e-01
3.40037912e-01 3.30966562e-01 -3.25527698e-01 1.43875077e-01
-3.48453641e-01 8.77230987e-02 -3.91286939e-01 1.02569211e+00
6.67795122e-01 -6.37773752e-01 1.28424868e-01 -4.00676131e-01
1.67322829e-01 -8.12332630e-02 -1.27901769e+00 1.56218600e+00
-4.65872556e-01 3.42633873e-01 3.48058692e-03 -1.19388330e+00
6.83084548e-01 2.77953625e-01 2.08173707e-01 -5.46275616e-01
-1.03530861e-01 -1.92893758e-01 -3.61370057e-01 -7.19937444e-01
4.15363997e-01 -3.15783292e-01 8.67266059e-02 3.45070899e-01
2.88936883e-01 -3.43545973e-02 2.99055547e-01 9.70211565e-01
7.30916560e-01 -1.68755762e-02 4.11867738e-01 -6.66876957e-02
4.64731246e-01 -3.26190554e-02 2.47541144e-01 7.40395665e-01
8.82381797e-02 2.88719356e-01 4.59688514e-01 -7.95578122e-01
-7.26147294e-01 -1.44283605e+00 4.07506227e-01 1.16025198e+00
6.35368943e-01 -5.09867847e-01 -1.10197626e-03 -7.19311297e-01
3.07925582e-01 6.14947617e-01 -9.43722665e-01 9.85936597e-02
-3.59543175e-01 -9.96089913e-03 8.49976856e-03 8.91401470e-01
8.61948252e-01 -9.88170385e-01 -6.53613985e-01 -3.59190434e-01
-4.51022446e-01 -1.28386903e+00 -3.36695969e-01 -8.98470208e-02
-5.48659980e-01 -1.31105351e+00 -2.62224644e-01 -8.44697177e-01
7.30375051e-01 7.02966332e-01 1.81275833e+00 5.79696894e-01
-3.10240358e-01 1.19598627e+00 -3.71851414e-01 -5.05657196e-01
3.67316306e-01 -7.65520573e-01 1.15477340e-02 3.57040614e-01
1.98161043e-02 -4.86819774e-01 -6.87920809e-01 1.62625179e-01
-1.03787458e+00 8.52470323e-02 6.82036757e-01 7.95169294e-01
7.82497466e-01 1.64232209e-01 2.02048570e-01 -1.15408361e+00
-8.73842984e-02 -7.68521607e-01 -2.39768416e-01 8.38438153e-01
-7.44577646e-02 -2.06708685e-01 5.84121406e-01 -3.72935951e-01
-1.41935515e+00 -9.45325717e-02 3.70739788e-01 -7.18707561e-01
-3.32719117e-01 6.08983040e-01 -3.77729774e-01 2.63816357e-01
4.50309843e-01 3.62136096e-01 -5.79817057e-01 -6.91004246e-02
9.66337979e-01 1.84767321e-03 7.80191362e-01 -8.90704930e-01
8.12350988e-01 9.52728450e-01 9.17540565e-02 -6.66334331e-01
-9.66999292e-01 -4.62202698e-01 -5.98703563e-01 -3.07336539e-01
8.02580595e-01 -1.30169272e+00 -1.10269868e+00 1.13621779e-01
-9.26650107e-01 5.53249791e-02 -2.30762690e-01 2.05856100e-01
-6.92759097e-01 2.05211669e-01 -4.42706108e-01 -7.98714578e-01
1.72607414e-02 -6.64283574e-01 1.17969882e+00 1.37589216e-01
3.21450949e-01 -9.67587173e-01 -5.34817219e-01 8.03636789e-01
9.67048779e-02 3.32691282e-01 1.29873562e+00 -3.23340744e-01
-1.17485356e+00 1.94372967e-01 -5.77766299e-01 1.54282808e-01
1.58883229e-01 8.61718208e-02 -7.83691227e-01 -7.52325207e-02
-3.00395340e-01 -8.40416074e-01 9.21236157e-01 3.09857816e-01
1.14730275e+00 -3.53128850e-01 -4.88444775e-01 2.36558333e-01
1.56262243e+00 4.29621488e-01 6.74426258e-01 -1.33886442e-01
8.25513124e-01 7.18462110e-01 5.89610279e-01 4.32987154e-01
1.00611556e+00 3.69235754e-01 7.21456110e-01 -8.13784674e-02
-2.23691225e-01 -2.99612820e-01 7.50384480e-02 3.45425487e-01
-5.72887778e-01 -2.33589604e-01 -8.70808423e-01 8.48934829e-01
-1.67921805e+00 -1.38026023e+00 9.82599426e-03 1.90003681e+00
6.17846251e-01 -1.27668172e-01 1.14121705e-01 -2.65372843e-01
6.16025388e-01 3.49658370e-01 -5.11229277e-01 -9.68612954e-02
-4.38083947e-01 -8.86759162e-03 -4.35450859e-02 3.83531839e-01
-9.34753835e-01 1.06686890e+00 5.31798172e+00 6.96815431e-01
-6.76332474e-01 -5.76423854e-02 6.97221160e-01 -1.64990723e-01
-6.96348846e-01 2.15490386e-01 -6.42478585e-01 1.67416885e-01
1.82158370e-02 -1.81012139e-01 3.96912009e-01 7.92492568e-01
-5.43685198e-01 -5.37543535e-01 -1.19381511e+00 1.45466268e+00
6.03501558e-01 -1.58194768e+00 7.95628130e-01 -2.36242265e-01
8.69843841e-01 -5.81409812e-01 2.91582614e-01 4.30249929e-01
7.12279201e-01 -1.01587558e+00 1.02773440e+00 5.10487616e-01
8.91955137e-01 -5.33743620e-01 4.34066683e-01 3.46664250e-01
-1.58259046e+00 -2.30009839e-01 -2.80757904e-01 -1.88767299e-01
1.37013227e-01 5.36337018e-01 -1.37371868e-01 1.19985771e+00
1.14761078e+00 1.02076054e+00 -5.53249419e-01 3.42323303e-01
-3.18193287e-01 9.88025665e-02 -9.24731791e-02 2.49598801e-01
7.41290376e-02 -8.37313831e-02 1.10506594e-01 8.31114471e-01
1.37475595e-01 6.89429224e-01 1.63101554e-01 7.76096523e-01
-3.07880551e-01 4.86577749e-02 -1.11217940e+00 1.35663584e-01
2.78438836e-01 9.27414775e-01 -7.03334272e-01 -9.48931277e-01
-8.15483868e-01 7.28420138e-01 6.78620160e-01 6.61544681e-01
-7.21070826e-01 1.40791103e-01 6.28500462e-01 4.07773964e-02
4.29278523e-01 -3.65661569e-02 -2.67237220e-02 -1.47853887e+00
3.85286629e-01 -7.64456391e-01 9.80894148e-01 -1.36247313e+00
-1.35271060e+00 4.25295115e-01 3.11331689e-01 -9.44479704e-01
8.58092234e-02 -6.59476340e-01 -3.19972932e-01 5.22609890e-01
-1.56197894e+00 -1.59169710e+00 -3.99864316e-01 1.36941218e+00
5.18186688e-01 1.38081359e-02 4.89530116e-01 1.17917545e-01
-4.65129875e-02 1.06434114e-01 -4.53612477e-01 -1.61272343e-02
5.25229216e-01 -1.34835505e+00 -7.38832504e-02 5.71061432e-01
7.16930509e-01 7.73557007e-01 5.81611753e-01 -7.46699095e-01
-1.45490015e+00 -7.87842989e-01 8.45651746e-01 -7.96923041e-01
6.01326108e-01 -4.75710541e-01 -8.35338414e-01 1.50649786e+00
4.71541852e-01 2.58253902e-01 5.75234354e-01 3.53043646e-01
-9.11134422e-01 -1.99372262e-01 -1.14368379e+00 4.77805674e-01
1.44679976e+00 -5.89835584e-01 -7.60923743e-01 4.02496636e-01
5.33004105e-01 -6.14357889e-01 -4.54304039e-01 1.72068030e-01
6.05537236e-01 -1.29749870e+00 1.33649075e+00 -9.29981709e-01
8.16560686e-01 -6.08565137e-02 -5.40459692e-01 -1.15140653e+00
-1.12424038e-01 1.57080412e-01 -2.23239332e-01 1.21780217e+00
2.34797612e-01 -5.32173991e-01 4.62492377e-01 6.86357141e-01
5.15086576e-03 -7.15266109e-01 -8.83326590e-01 -7.85669982e-01
-6.61715865e-02 -1.94690481e-01 6.57714427e-01 9.70136821e-01
-3.46656479e-02 1.48913443e-01 -8.03157687e-01 3.78944725e-01
4.76918936e-01 7.45949686e-01 8.67599487e-01 -9.98320103e-01
-2.57360488e-01 -6.94055930e-02 -6.15134776e-01 -1.20271719e+00
1.39292419e-01 -7.13213623e-01 -3.97101134e-01 -1.84988606e+00
8.00298572e-01 -1.19569361e-01 -1.62902683e-01 6.29043877e-01
-2.95483589e-01 -1.44147992e-01 3.56029421e-01 -3.76740657e-02
-8.85563791e-01 3.44842196e-01 1.36717355e+00 -4.67709154e-01
5.27605176e-01 -5.45760572e-01 -6.68506801e-01 8.04080427e-01
4.69883889e-01 -1.36386722e-01 -7.71387637e-01 -7.21175313e-01
5.63880920e-01 3.21634501e-01 8.97001982e-01 -5.71953714e-01
1.72662050e-01 -2.78032392e-01 3.46405655e-01 -5.08199990e-01
7.57174551e-01 -1.04975617e+00 5.24777472e-02 1.93587430e-02
-2.26144463e-01 2.30686948e-01 7.37138018e-02 9.51259434e-01
-5.80493748e-01 3.27886850e-01 2.83991069e-01 -6.68552279e-01
-1.09624612e+00 3.45654607e-01 2.29011744e-01 7.17198849e-02
1.25325668e+00 -2.40645543e-01 -1.02638137e+00 -8.64181399e-01
-1.00450110e+00 4.62467134e-01 1.20284714e-01 3.24595481e-01
8.35633457e-01 -1.29605722e+00 -3.94461632e-01 -4.68369238e-02
5.74505210e-01 -1.67398483e-01 7.80628681e-01 4.60966706e-01
6.81606755e-02 3.49062920e-01 -3.56182277e-01 -4.37595963e-01
-1.43344307e+00 1.31065798e+00 -3.84802371e-02 -1.52525455e-01
-5.95981359e-01 1.07535160e+00 8.80053937e-01 -1.17527157e-01
3.05313587e-01 -4.54955399e-01 -2.33360440e-01 1.03143737e-01
5.11174798e-01 1.40484214e-01 -3.35453391e-01 -9.95453477e-01
-3.75001431e-01 9.38918650e-01 -4.52532060e-03 2.52622962e-01
1.01241410e+00 -4.03308451e-01 -2.45324541e-02 6.74332500e-01
9.94866014e-01 -1.83677897e-01 -7.77284265e-01 -5.25236011e-01
-4.75769758e-01 -9.95563865e-01 -5.50415337e-01 -9.81097817e-01
-1.45626116e+00 8.65169704e-01 2.70981759e-01 2.47460321e-01
1.16573894e+00 7.73581266e-01 4.93041277e-01 3.58833224e-01
1.03130651e+00 -8.28190207e-01 4.24960494e-01 9.25640464e-02
1.09139562e+00 -1.63992476e+00 3.43874171e-02 -7.24688947e-01
-1.01466572e+00 7.65304446e-01 6.30353212e-01 2.14207694e-01
5.92093825e-01 -1.15621760e-01 1.21031851e-01 -9.80680048e-01
-1.03619552e+00 -6.37749851e-01 1.82340324e-01 4.59408164e-01
2.24021506e-02 3.76881599e-01 7.26076439e-02 4.22107190e-01
-9.73523334e-02 -1.87478751e-01 2.64876604e-01 9.78725791e-01
-3.25458467e-01 -6.79417074e-01 -3.34484667e-01 5.45978725e-01
-1.43503606e-01 -1.66475698e-01 -3.68264884e-01 1.10983849e+00
5.91433287e-01 1.14544761e+00 -9.56708416e-02 -1.97363883e-01
4.95487809e-01 1.53545812e-01 5.25602460e-01 -6.61996484e-01
-3.76778930e-01 -5.12857080e-01 -8.76105651e-02 -7.07743406e-01
-9.08521533e-01 -5.80016017e-01 -1.32366824e+00 -2.36252442e-01
1.39537677e-01 -2.26949766e-01 2.81054497e-01 1.07599187e+00
2.29448035e-01 7.24490583e-02 1.59530848e-01 -2.55480021e-01
2.24929482e-01 -4.46036786e-01 -1.03481877e+00 1.14505267e+00
4.76529121e-01 -1.18627405e+00 -4.09822196e-01 7.51241684e-01]
|
[10.453036308288574, 1.692186713218689]
|
57627e2a-59e5-4336-8457-e67d89964e76
|
deepcontour-a-deep-convolutional-feature
| null | null |
http://openaccess.thecvf.com/content_cvpr_2015/html/Shen_DeepContour_A_Deep_2015_CVPR_paper.html
|
http://openaccess.thecvf.com/content_cvpr_2015/papers/Shen_DeepContour_A_Deep_2015_CVPR_paper.pdf
|
DeepContour: A Deep Convolutional Feature Learned by Positive-Sharing Loss for Contour Detection
|
Contour detection serves as the basis of a variety of computer vision tasks such as image segmentation and object recognition. The mainstream works to address this problem focus on designing engineered gradient features. In this work, we show that contour detection accuracy can be improved by instead making the use of the deep features learned from convolutional neural networks (CNNs). While rather than using the networks as a blackbox feature extractor, we customize the training strategy by partitioning contour (positive) data into subclasses and fitting each subclass by different model parameters. A new loss function, named positive-sharing loss, in which each subclass shares the loss for the whole positive class, is proposed to learn the parameters. Compared to the sofmax loss function, the proposed one, introduces an extra regularizer to emphasizes the losses for the positive and negative classes, which facilitates to explore more discriminative features. Our experimental results demonstrate that learned deep features can achieve top performance on Berkeley Segmentation Dataset and Benchmark (BSDS500) and obtain competitive cross dataset generalization result on the NYUD dataset.
|
['Zhijiang Zhang', 'Wei Shen', 'Xinggang Wang', 'Yan Wang', 'Xiang Bai']
|
2015-06-01
| null | null | null |
cvpr-2015-6
|
['contour-detection']
|
['computer-vision']
|
[ 2.28442580e-01 8.19493234e-02 -4.84221101e-01 -7.74631023e-01
-5.00513971e-01 -3.70338321e-01 2.69258887e-01 9.55766067e-02
-7.23058641e-01 4.60159421e-01 -2.98441827e-01 -1.72211424e-01
2.68470824e-01 -1.10785210e+00 -7.16024399e-01 -8.16262782e-01
8.38002041e-02 6.06679767e-02 6.36457682e-01 7.56677845e-03
2.49362126e-01 5.71930408e-01 -1.12517881e+00 9.26540792e-02
1.04361653e+00 1.66172636e+00 1.35225147e-01 1.84910089e-01
-4.35563773e-01 4.61784303e-01 -4.82957751e-01 -5.06030560e-01
3.92066330e-01 -2.48618498e-01 -6.05571985e-01 1.03869326e-01
4.06744003e-01 -1.84498340e-01 -1.11881875e-01 1.18414903e+00
4.75580812e-01 -8.28582272e-02 6.21912003e-01 -1.15860295e+00
-6.79659128e-01 3.47990125e-01 -9.27255154e-01 5.63643947e-02
-2.76307702e-01 -1.70578435e-01 1.15948474e+00 -7.34861255e-01
3.37112993e-01 1.01121116e+00 7.87060797e-01 4.80677783e-01
-9.67022240e-01 -8.16154361e-01 2.66555309e-01 1.62486002e-01
-1.32740939e+00 1.89469546e-01 9.88561451e-01 -3.84567618e-01
2.49304488e-01 7.65069649e-02 7.80861139e-01 5.71765780e-01
-4.63221297e-02 1.28743422e+00 7.96604276e-01 -2.92792886e-01
1.00915357e-01 2.54710793e-01 5.39473355e-01 9.06184912e-01
2.60294020e-01 -6.34680316e-02 -1.11629732e-01 1.58795953e-01
9.01252031e-01 3.55040282e-02 -5.32048821e-01 -6.19771719e-01
-7.38893211e-01 1.14716113e+00 7.74312377e-01 -3.76340076e-02
-9.51792523e-02 6.09887913e-02 4.52246547e-01 4.18650508e-02
4.45103437e-01 6.57441542e-02 -6.62636638e-01 2.64275908e-01
-7.42755592e-01 7.66650215e-02 5.92378497e-01 9.36466157e-01
1.13991654e+00 -6.96523562e-02 -3.79403174e-01 1.00734591e+00
2.78270841e-01 2.71031529e-01 6.23187602e-01 -5.99871933e-01
3.77321422e-01 1.00627935e+00 -2.59160280e-01 -8.74347031e-01
-5.31180084e-01 -7.48910069e-01 -8.50273848e-01 2.44098350e-01
4.66576487e-01 -3.44596207e-01 -1.24449062e+00 1.66034341e+00
3.46753061e-01 2.02717800e-02 -2.13888347e-01 9.83335078e-01
7.98912883e-01 5.74072957e-01 -4.23462391e-02 1.04998993e-02
1.12964725e+00 -1.12785554e+00 -4.59329903e-01 -1.93165451e-01
4.30190235e-01 -6.14086449e-01 1.16237509e+00 3.69804531e-01
-8.88738334e-01 -6.06513619e-01 -1.34532905e+00 -1.72580153e-01
-4.17089373e-01 3.17761779e-01 9.74759936e-01 6.65406942e-01
-7.67064810e-01 6.42404199e-01 -8.54167044e-01 1.06209651e-01
8.09229076e-01 3.91316861e-01 -4.41084988e-02 1.24241523e-01
-1.07873011e+00 4.09383535e-01 5.19129336e-01 3.32315683e-01
-4.85804260e-01 -6.59674168e-01 -9.49607909e-01 2.15499550e-01
3.79904002e-01 -2.68768460e-01 8.22600424e-01 -1.24120593e+00
-1.61652851e+00 9.38876331e-01 1.39833689e-01 -4.71382022e-01
7.65142918e-01 -3.00873369e-01 -5.53966910e-02 1.52845919e-01
2.49631740e-02 8.97232831e-01 8.40426981e-01 -1.09734249e+00
-7.51773000e-01 -3.05066258e-01 -8.14088061e-02 -1.00373968e-01
-6.13971710e-01 -3.15300882e-01 -8.15826714e-01 -8.85019600e-01
3.79654467e-01 -7.39456654e-01 -1.55891389e-01 4.89651650e-01
-6.31231964e-01 -4.53987151e-01 9.99728858e-01 -3.57913762e-01
1.12139964e+00 -2.22783446e+00 -4.60550375e-02 4.52616423e-01
1.35999829e-01 4.51086104e-01 1.52196735e-02 -1.15845487e-01
7.65960142e-02 1.01856440e-01 -5.67682624e-01 -2.84838080e-01
-1.10625491e-01 1.83000833e-01 -2.45946959e-01 4.13922340e-01
6.02309942e-01 8.28273416e-01 -5.18339813e-01 -5.66681325e-01
1.20073243e-03 4.20153588e-01 -5.59878588e-01 8.14638510e-02
-2.13440642e-01 7.23061413e-02 -7.02380061e-01 4.14618909e-01
1.02905059e+00 -4.67040688e-01 -1.01039536e-01 -3.04237336e-01
-5.88956028e-02 -1.19829088e-01 -1.10420823e+00 1.57379782e+00
-2.17986807e-01 4.15448010e-01 4.00575474e-02 -1.22616863e+00
1.22097790e+00 -8.44047293e-02 2.46940643e-01 -7.08419144e-01
1.96584642e-01 1.11710586e-01 2.23722551e-02 -2.80492306e-01
1.11691825e-01 -4.28943150e-02 7.06542209e-02 -1.04932925e-02
1.18767560e-01 -1.94492787e-02 9.08413753e-02 -2.41972115e-02
5.04089057e-01 1.26750365e-01 -1.17706217e-01 -3.88838261e-01
6.85742974e-01 -8.31418782e-02 9.81376171e-01 4.73789603e-01
-2.69967973e-01 7.32750356e-01 6.33887053e-01 -4.21695739e-01
-6.98474824e-01 -9.54404831e-01 -4.96268868e-01 1.09919024e+00
5.28834820e-01 -2.90569831e-02 -7.76063025e-01 -8.12525094e-01
1.56177938e-01 1.88165307e-01 -8.36146772e-01 -2.02783033e-01
-7.36648083e-01 -1.09974527e+00 4.83476907e-01 8.95966828e-01
1.00170970e+00 -9.62939680e-01 -5.45258701e-01 3.29757892e-02
1.78171605e-01 -8.14613819e-01 -7.11660087e-01 4.61019039e-01
-8.53244424e-01 -9.78181541e-01 -8.51848304e-01 -1.16093028e+00
5.64478517e-01 9.64337811e-02 6.77121580e-01 1.04834534e-01
-4.70435560e-01 -5.32636605e-02 -3.79872322e-01 -3.79403889e-01
3.09279948e-01 2.90715814e-01 -6.23488247e-01 3.16025913e-01
2.43136287e-01 -3.61618042e-01 -9.24193442e-01 2.09744737e-01
-9.47170138e-01 -3.17477621e-02 6.04759157e-01 1.06505084e+00
7.42610395e-01 -2.02347413e-01 7.97793269e-01 -9.85280752e-01
2.82507777e-01 -2.77764678e-01 -6.76569045e-01 1.83119610e-01
-6.91224933e-01 3.47494334e-02 6.47299528e-01 -3.03367734e-01
-9.35548306e-01 1.60344675e-01 -2.86173493e-01 -2.29084715e-01
7.04745203e-02 5.19150376e-01 -3.25639486e-01 -2.40439281e-01
2.13139459e-01 2.77894199e-01 5.70444688e-02 -7.03400970e-01
1.87471434e-01 4.44723696e-01 4.19466943e-01 -5.72508633e-01
5.49483955e-01 5.57599187e-01 2.95853689e-02 -6.26687646e-01
-9.58367646e-01 -4.58406538e-01 -4.27421451e-01 4.07766216e-02
1.11505628e+00 -6.98365152e-01 -6.77392602e-01 7.03551352e-01
-9.92555976e-01 -3.24537098e-01 -1.36317328e-01 2.96564847e-01
-2.88706273e-01 3.80124867e-01 -6.87131166e-01 -6.18042409e-01
-3.54688704e-01 -1.16485369e+00 9.91059959e-01 6.94000244e-01
4.37584311e-01 -8.69131804e-01 -1.18499324e-01 7.43338615e-02
2.99686968e-01 2.94719219e-01 1.06707335e+00 -6.97757006e-01
-5.04910946e-01 -2.94929206e-01 -6.54247642e-01 7.79308140e-01
1.74486741e-01 7.97559023e-02 -9.50210810e-01 -3.26448441e-01
-1.34250715e-01 -5.91585875e-01 1.37137234e+00 4.13255244e-01
1.58870435e+00 8.28011781e-02 -3.07049841e-01 1.06199396e+00
1.49290359e+00 3.02114069e-01 6.00252211e-01 3.20978254e-01
7.17305422e-01 5.13112605e-01 3.62502396e-01 2.74776399e-01
2.74454355e-01 4.53788936e-01 4.77370650e-01 -5.19887745e-01
5.66851646e-02 -3.06271832e-04 -1.05332091e-01 2.76287109e-01
8.91709775e-02 -3.65712941e-02 -6.88923001e-01 3.65178555e-01
-1.79160130e+00 -5.38350105e-01 -5.47748711e-03 2.01923084e+00
9.06035960e-01 4.29399222e-01 8.36400837e-02 7.85657540e-02
7.60829568e-01 1.48537219e-01 -7.34864175e-01 -9.79166105e-02
-2.03215316e-01 4.15983081e-01 4.40906882e-01 3.27763706e-01
-1.54370606e+00 9.23279524e-01 5.41534042e+00 1.08838999e+00
-1.56789553e+00 -2.28279009e-01 9.03889596e-01 2.35070720e-01
-2.60384530e-02 -1.97670639e-01 -9.15191770e-01 4.50325727e-01
2.12299570e-01 5.67743741e-02 -1.08315982e-02 9.01631594e-01
-1.00022413e-01 1.14213616e-01 -8.45304906e-01 6.38273716e-01
-1.96985900e-01 -1.23532379e+00 2.98466831e-02 -2.01637700e-01
4.94930238e-01 -3.95664088e-02 2.19871297e-01 4.57188964e-01
5.97142940e-03 -9.00260329e-01 8.94957900e-01 1.90828398e-01
5.66352129e-01 -8.11666906e-01 6.80889547e-01 2.34973893e-01
-1.31501317e+00 -2.70279292e-02 -5.66403329e-01 7.84641132e-02
-9.10426825e-02 5.04997253e-01 -4.10794646e-01 3.98140132e-01
7.35261738e-01 7.63863385e-01 -5.49343109e-01 1.34742999e+00
-1.33444697e-01 6.94748402e-01 -2.86300898e-01 -7.80643895e-02
4.25801784e-01 -4.63957489e-01 4.78367321e-02 1.22860670e+00
-1.14568882e-01 -1.33801252e-01 4.87075150e-01 9.49230552e-01
-2.93159187e-01 3.34409446e-01 -1.18753731e-01 2.57628560e-01
1.48168325e-01 1.40008330e+00 -8.69726956e-01 -2.36395627e-01
-5.41136742e-01 8.88698936e-01 4.98538077e-01 3.54720503e-01
-8.40525270e-01 -8.49882066e-01 3.86171907e-01 -7.85036162e-02
6.50761545e-01 -4.44827834e-03 -5.21046519e-01 -9.89186227e-01
1.57371253e-01 -2.33266130e-01 3.89042497e-01 -2.30740339e-01
-1.31044126e+00 4.73962098e-01 -2.66738743e-01 -1.16701674e+00
3.62121016e-01 -9.08451080e-01 -9.55057025e-01 8.43584955e-01
-1.88624287e+00 -1.30087125e+00 -3.93913656e-01 4.09277231e-01
3.26082975e-01 1.51994601e-01 3.62432390e-01 3.91519606e-01
-8.24348629e-01 8.14415932e-01 7.62650296e-02 6.55956030e-01
5.95592320e-01 -1.45031083e+00 6.87506497e-02 4.72136497e-01
-1.99968919e-01 4.94570613e-01 2.01914221e-01 -3.86800677e-01
-8.13921988e-01 -1.25819659e+00 3.33267927e-01 2.96223223e-01
5.39139211e-01 -4.90248829e-01 -1.08196938e+00 5.18603981e-01
4.27802578e-02 5.37718177e-01 6.47439659e-01 -9.68945399e-02
-3.42591792e-01 -3.64896089e-01 -1.11225438e+00 3.07277083e-01
9.68979537e-01 -1.07581988e-02 -2.56443828e-01 1.14506505e-01
8.28202188e-01 -2.44831681e-01 -7.61871815e-01 7.24121332e-01
5.84107876e-01 -9.64766204e-01 8.88735056e-01 -7.61620522e-01
4.33368921e-01 -2.13952154e-01 4.86261807e-02 -1.04392731e+00
-1.25574604e-01 -1.49155736e-01 2.55477756e-01 1.20362031e+00
6.13299489e-01 -7.69791126e-01 1.21489573e+00 3.65146577e-01
-4.35103923e-01 -1.26441169e+00 -8.53447080e-01 -5.87680101e-01
5.18613756e-01 -1.76632494e-01 5.15659869e-01 8.26007366e-01
-3.98932070e-01 2.31095836e-01 -2.63232943e-02 4.48175669e-02
4.82625842e-01 5.39369464e-01 4.97735709e-01 -1.44369960e+00
-2.46766835e-01 -6.97528005e-01 -4.93805885e-01 -1.33541048e+00
1.23570465e-01 -8.72432888e-01 8.47433209e-02 -1.21476936e+00
2.47438878e-01 -7.99308479e-01 -5.26936471e-01 4.79955345e-01
-3.74989212e-01 3.11709136e-01 2.10581705e-01 -7.39523992e-02
-4.98254210e-01 8.43627036e-01 1.62531531e+00 -3.06905895e-01
-1.33655354e-01 3.08845878e-01 -6.08030498e-01 9.49417055e-01
7.32981920e-01 -3.07177871e-01 -3.47302347e-01 -4.32825714e-01
2.54166052e-02 -2.24813357e-01 4.57733780e-01 -8.80353034e-01
2.49619901e-01 -1.34968325e-01 6.05204523e-01 -6.48004770e-01
2.76442945e-01 -7.60281622e-01 -6.04709625e-01 4.61105794e-01
-3.29490989e-01 -3.75812739e-01 2.90013969e-01 6.83111548e-01
-4.06706125e-01 -3.03908467e-01 1.07972217e+00 4.43817079e-02
-6.28853619e-01 5.36520660e-01 1.70760229e-01 2.35178620e-01
1.04473960e+00 -2.15168655e-01 -4.12830889e-01 2.60832272e-02
-6.90549791e-01 4.85970944e-01 2.85840631e-01 2.16676161e-01
7.04981625e-01 -1.31217873e+00 -4.64488566e-01 3.59376818e-01
3.11872968e-03 4.17725772e-01 7.37369210e-02 6.53326333e-01
-5.89853525e-01 1.48566976e-01 -1.80214003e-01 -7.50207126e-01
-7.66094685e-01 2.86306232e-01 5.12992263e-01 -2.75272012e-01
-6.27111495e-01 1.16308236e+00 6.29750907e-01 -4.51160222e-01
4.93118525e-01 -3.58991325e-01 -3.73518199e-01 1.10901631e-02
3.44801188e-01 2.51608163e-01 7.95610100e-02 -2.63157755e-01
-2.47691870e-01 7.65358746e-01 -3.85135502e-01 2.91948229e-01
1.35121810e+00 1.27018884e-01 2.21650070e-03 2.00064778e-01
1.57975984e+00 -2.18942627e-01 -1.56210279e+00 -2.42492303e-01
1.55473650e-01 -2.34281018e-01 2.06271961e-01 -7.21383274e-01
-1.69303846e+00 9.99962568e-01 8.49639833e-01 1.43327907e-01
1.20407975e+00 -1.30422533e-01 1.02127409e+00 3.09147090e-01
2.12007627e-01 -1.13310802e+00 1.67972252e-01 4.57185239e-01
6.29959702e-01 -1.44264865e+00 -3.98381442e-01 -7.03065932e-01
-5.38806081e-01 1.25569999e+00 8.11421454e-01 -5.87382019e-01
9.36240375e-01 2.33430251e-01 8.80120695e-02 -1.00625090e-01
-2.44382486e-01 -2.61910200e-01 3.14747751e-01 3.31663460e-01
3.87864321e-01 2.30963081e-02 -4.83666301e-01 9.17863429e-01
1.60999581e-01 -5.75186983e-02 1.64592370e-01 9.90744352e-01
-7.08633840e-01 -1.10707784e+00 3.77755389e-02 7.42451310e-01
-5.10774076e-01 -3.53764780e-02 -2.34464526e-01 9.22888935e-01
3.06195289e-01 4.95545954e-01 2.85606176e-01 -1.94028810e-01
2.85622329e-01 -8.84406492e-02 2.94070452e-01 -6.08253777e-01
-4.92358714e-01 1.74387798e-01 -3.04790705e-01 -3.60542923e-01
-2.17043504e-01 -3.60720724e-01 -1.51816595e+00 1.37231261e-01
-6.02682233e-01 1.52853638e-01 4.53685135e-01 8.09867680e-01
3.58361192e-02 5.89844942e-01 6.97880030e-01 -7.65837133e-01
-7.79102623e-01 -7.82373786e-01 -7.19401836e-01 5.00802636e-01
4.24931109e-01 -8.27929437e-01 -1.82944670e-01 -4.55491208e-02]
|
[9.562115669250488, 0.42912471294403076]
|
8c7738d1-2221-4193-8386-92f29db7bd7b
|
technology-report-smartphone-based-pedestrian
|
2301.03471
| null |
https://arxiv.org/abs/2301.03471v1
|
https://arxiv.org/pdf/2301.03471v1.pdf
|
Technology Report : Smartphone-Based Pedestrian Dead Reckoning Integrated with Data-Fusion-Adopted Visible Light Positioning
|
Pedestrian dead-reckoning (PDR) is a potential indoor localization technology that obtains location estimation with the inertial measurement unit (IMU). However, one of its most significant drawbacks is the accumulation of its measurement error. This paper proposes a visible light positioning (VLP)-integrated PDR system, which could achieve real-time and accurate indoor positioning using IMU and the camera sensor of our smartphone. A multi-frame fusion method is proposed in the encoding and decoding process of the system, reaching 98.5% decoding accuracy with a 20-bit-long ID at the height of 2.1 m, which allows the variation in the shutter speeds of cameras and heights of the LED. Meanwhile, absolute locations and step length could be calibrated with the help of a single light-emitting diode (LED), promising average accuracy within 0.5 meters in a 108-meter walk.
|
['Xuecong Fang', 'Yingcong Chen', 'Danlan Yuan', 'Ziyang Ge', 'ShangSheng Wen']
|
2023-01-06
| null | null | null | null |
['indoor-localization']
|
['computer-vision']
|
[ 3.78650911e-02 -6.49263918e-01 9.90898609e-02 -1.64720237e-01
-6.10828221e-01 -4.83410656e-01 1.99920744e-01 -2.11063512e-02
-6.73233926e-01 1.01251137e+00 5.92794095e-04 -4.13856298e-01
3.36937994e-01 -9.63541508e-01 -5.64161479e-01 -6.40566230e-01
4.59227026e-01 -2.82733947e-01 1.76980004e-01 -6.00758642e-02
1.13267362e-01 3.82618278e-01 -1.48446310e+00 -4.04291540e-01
1.01327324e+00 1.06903863e+00 4.77164805e-01 9.34946835e-01
5.62992275e-01 3.67227942e-01 -8.43850017e-01 -2.78463960e-01
-6.64498061e-02 8.69488623e-03 2.82825172e-01 -3.15653950e-01
1.56003878e-01 -6.47692323e-01 -6.12582743e-01 6.15407348e-01
1.01811099e+00 -7.25727007e-02 2.36905381e-01 -1.12259221e+00
-4.87856120e-01 -1.53136358e-01 -5.02692103e-01 -5.63701950e-02
1.11903536e+00 9.65253562e-02 1.49133643e-02 -4.95836258e-01
-7.59180682e-03 6.11774623e-01 1.01949215e+00 1.29502535e-01
-7.00578451e-01 -4.93771344e-01 -6.98212087e-01 4.67759848e-01
-1.93607295e+00 -3.84128392e-01 5.39534509e-01 -2.10194468e-01
7.97835410e-01 4.85250294e-01 9.37172413e-01 9.19456303e-01
6.79797769e-01 -1.29736096e-01 1.10766566e+00 -3.15903693e-01
3.48819822e-01 4.65858914e-02 -3.44023742e-02 4.67815965e-01
9.59786773e-01 1.11006625e-01 -6.28681123e-01 1.21710300e-01
7.08166063e-01 3.71815383e-01 -5.41499197e-01 1.10383593e-01
-1.30807912e+00 1.72093198e-01 5.18495321e-01 1.43794626e-01
-8.73313472e-02 3.65053356e-01 -1.93504825e-01 -2.65058696e-01
1.35354146e-01 -2.26170123e-01 -9.08547267e-02 -8.77262533e-01
-6.64167762e-01 -6.28826499e-01 3.66048306e-01 1.25990117e+00
6.52300775e-01 -1.35414451e-01 6.35195002e-02 3.47933531e-01
6.93652272e-01 1.38751948e+00 4.05851543e-01 -8.94249320e-01
6.32724166e-01 4.27848279e-01 6.78712726e-01 -1.05756748e+00
-4.87795442e-01 -1.42567262e-01 -8.04221749e-01 1.38317883e-01
5.41879237e-01 -7.25154877e-01 -2.74256200e-01 1.00669920e+00
2.53901213e-01 3.76146495e-01 -6.39191121e-02 8.01137269e-01
6.31039441e-01 4.84350860e-01 -5.68710387e-01 -2.87352800e-01
1.42221463e+00 -6.05355799e-01 -7.88075507e-01 -3.20947260e-01
5.13607681e-01 -7.85203516e-01 9.17729914e-01 1.25020161e-01
-5.24561167e-01 -8.34221125e-01 -1.37177265e+00 -2.71648645e-01
-1.34386599e-01 7.52920091e-01 -2.05625650e-02 1.28968501e+00
-9.87637043e-01 -1.67378411e-01 -8.75780046e-01 -4.31071728e-01
-2.68776387e-01 2.70039052e-01 -1.43128306e-01 -2.92473227e-01
-9.47283387e-01 8.31158519e-01 -2.66093582e-01 2.66252339e-01
1.63322598e-01 -3.62388700e-01 -5.92311621e-01 -3.81475538e-02
-1.02774672e-01 -9.02062595e-01 8.15992594e-01 -2.12259918e-01
-1.99359500e+00 3.49188834e-01 -7.68894374e-01 -5.40357530e-01
6.16424382e-01 -4.91734505e-01 -6.29820108e-01 -3.29005718e-03
8.71046036e-02 -1.18980408e-02 4.17726725e-01 -9.58985507e-01
-5.86244106e-01 -5.93720973e-01 -1.77184060e-01 3.00534606e-01
-4.55074832e-02 -4.64817703e-01 -2.27589831e-01 8.29296634e-02
2.80832708e-01 -1.08765912e+00 -7.00553507e-02 -1.36839643e-01
-1.80880800e-01 4.52218503e-01 5.88213146e-01 -6.91143870e-01
1.38484240e+00 -2.16031289e+00 -8.06153476e-01 -8.54977965e-02
-1.27147183e-01 1.80528432e-01 7.48292327e-01 2.85633266e-01
6.46537125e-01 -3.67052078e-01 2.15451509e-01 -4.92665082e-01
-2.92184949e-01 1.32467762e-01 -1.37032688e-01 8.14875722e-01
-1.05729795e+00 8.05608094e-01 -1.04218936e+00 -9.94113535e-02
8.87127578e-01 9.92446125e-01 -5.60138375e-02 -1.76929384e-01
8.02829921e-01 7.01035976e-01 1.28810462e-02 6.76511109e-01
9.10592437e-01 3.11318785e-02 6.33173510e-02 -1.10299900e-01
-7.72601902e-01 3.78251284e-01 -1.39493299e+00 1.50401807e+00
-5.92901826e-01 1.03203535e+00 -3.54227036e-01 -3.73018114e-03
1.16325796e+00 2.45950013e-01 3.06616485e-01 -8.11107218e-01
1.80076957e-01 3.75756353e-01 -6.60666764e-01 -2.71842390e-01
9.13594604e-01 5.12981832e-01 2.85075791e-02 7.58210719e-02
-7.65249670e-01 1.69900164e-01 -2.16792569e-01 -2.61713356e-01
1.13895798e+00 1.16648234e-03 4.90076065e-01 8.35392401e-02
7.14846730e-01 -3.72300267e-01 3.51674646e-01 7.21806884e-01
-2.99106270e-01 6.86035573e-01 -5.33694386e-01 -1.72324225e-01
-7.64183819e-01 -9.10683155e-01 -1.29543826e-01 1.44022092e-01
7.97757328e-01 -5.54132819e-01 -5.70316792e-01 1.10827349e-01
5.63537441e-02 7.33968556e-01 4.25773822e-02 5.76306432e-02
-5.45985401e-01 -5.45026958e-01 4.65209782e-01 4.24826533e-01
1.18278849e+00 6.87512085e-02 -1.02701783e+00 2.18418956e-01
-6.08876169e-01 -1.23921537e+00 -3.61384153e-01 -3.81598264e-01
-4.85549539e-01 -9.39492583e-01 -7.03801334e-01 -3.06985021e-01
7.38547444e-01 1.16017413e+00 4.54246968e-01 -1.62307680e-01
1.53388053e-01 8.08304846e-01 -2.08370402e-01 -8.31625462e-02
3.37959588e-01 -2.46718034e-01 3.06391895e-01 2.17965506e-02
5.76714694e-01 -3.29750448e-01 -9.11977828e-01 6.33808911e-01
2.53237724e-01 8.56193453e-02 8.17973614e-02 2.44577661e-01
4.76780772e-01 1.43735306e-04 -7.83901811e-02 1.87345654e-01
1.77509591e-01 -2.93764025e-01 -9.76488888e-01 4.46851552e-02
-7.25990295e-01 -6.31965637e-01 4.95609194e-01 -7.81388804e-02
-9.50883508e-01 1.12473562e-01 -2.63131320e-01 4.43309963e-01
-2.78069139e-01 -1.47877157e-01 -4.30416852e-01 -3.83291066e-01
5.64244151e-01 5.26019812e-01 -2.08526567e-01 -1.40087068e-01
3.07985544e-01 1.22862446e+00 8.31588268e-01 -8.32781382e-03
4.91601735e-01 6.95784688e-01 2.25675896e-01 -1.21347308e+00
3.34239379e-02 -5.37958145e-01 -6.09254718e-01 -5.58079720e-01
6.81904733e-01 -1.47507381e+00 -1.30404127e+00 8.74564946e-01
-1.13599038e+00 -7.46090412e-02 6.78249672e-02 8.45314860e-01
-2.53332853e-01 6.58092678e-01 -4.86463249e-01 -8.53629410e-01
-1.27880931e-01 -1.06322873e+00 9.02013838e-01 5.27917624e-01
2.30370183e-02 -7.21181929e-01 4.65849899e-02 6.93744719e-01
3.78587395e-01 1.33024201e-01 -1.65914446e-01 6.07131600e-01
-7.67298520e-01 -6.07814014e-01 -1.25817999e-01 -7.05632716e-02
3.91545832e-01 -1.72156051e-01 -8.88138175e-01 -2.00430021e-01
1.24256529e-01 7.16618657e-01 2.51772106e-01 7.86923945e-01
5.68468511e-01 -2.78689086e-01 -5.90787053e-01 8.49905193e-01
1.59608877e+00 4.52377945e-01 1.10801733e+00 5.87982178e-01
8.12224865e-01 -4.65166748e-01 5.94913960e-01 5.91622889e-01
9.35362637e-01 1.11201370e+00 2.49196112e-01 2.89908111e-01
-2.35782668e-01 -4.12105739e-01 6.04744136e-01 6.54440761e-01
-5.30285001e-01 -4.58618075e-01 -7.77494609e-01 -1.30022869e-01
-1.61490786e+00 -9.94855523e-01 -9.76427019e-01 2.86438131e+00
3.84387732e-01 -1.05221041e-01 -5.80214150e-02 4.69229460e-01
9.14035201e-01 1.11750606e-02 -2.13586852e-01 -1.10332631e-01
-4.73286435e-02 -4.13550943e-01 1.35173404e+00 1.02365363e+00
-5.72271168e-01 2.53554136e-01 5.78165197e+00 3.86089087e-01
-1.13455439e+00 7.29159042e-02 1.27922460e-01 1.87110469e-01
-6.47091046e-02 -2.20828474e-01 -1.21185231e+00 1.40928447e+00
1.15869439e+00 3.39331059e-03 4.52027947e-01 8.88906896e-01
5.57829499e-01 -8.78374696e-01 -5.62159002e-01 1.54257190e+00
1.01039074e-01 -9.55918014e-01 -8.04652154e-01 5.12382150e-01
5.88536382e-01 -1.10082090e-01 4.32687961e-02 -3.55260104e-01
-1.17461309e-01 -2.83319563e-01 6.97330892e-01 6.11102998e-01
1.08159423e+00 -8.24684501e-01 7.31655121e-01 4.45468694e-01
-1.65114462e+00 -1.21388100e-01 -4.95006382e-01 -6.96926117e-01
7.01713741e-01 9.70958531e-01 -7.22136140e-01 5.11890233e-01
5.66188335e-01 6.80764854e-01 -5.86385131e-01 1.16642964e+00
-6.93618774e-01 5.50704122e-01 -7.51532674e-01 -3.20552498e-01
-4.58441406e-01 -6.32221699e-01 4.51092273e-01 7.97811031e-01
1.24878156e+00 1.08565301e-01 -3.32152277e-01 1.40683219e-01
2.11428180e-01 -3.37509483e-01 -7.45106101e-01 9.77268696e-01
1.07688606e+00 9.30229127e-01 -3.35496157e-01 -3.08633417e-01
-4.87178981e-01 1.01746500e+00 -4.66378391e-01 3.18024725e-01
-1.11717558e+00 -5.88856101e-01 5.91336787e-01 4.25287247e-01
6.78665712e-02 -1.08751619e+00 -7.91047573e-01 -1.19845736e+00
1.94932237e-01 4.50497754e-02 -3.88590932e-01 -1.00299060e+00
-4.51060116e-01 4.63514030e-02 -5.62458932e-01 -1.71176946e+00
1.14032729e-02 -3.25607747e-01 -2.34915420e-01 7.38572478e-01
-1.37284982e+00 -8.28252137e-01 -9.35074687e-01 7.36870229e-01
1.86742768e-02 2.17383668e-01 6.78642333e-01 8.04375827e-01
-5.98248780e-01 5.26594162e-01 7.95606554e-01 -5.09570055e-02
6.61059201e-01 -9.59331155e-01 3.61079663e-01 1.18439531e+00
-1.79128587e-01 5.29706895e-01 5.55245876e-01 -6.84714079e-01
-1.62850916e+00 -7.73917019e-01 1.25823665e+00 -5.40629864e-01
2.05692247e-01 -1.79721668e-01 -3.23573239e-02 5.87209284e-01
-6.90239817e-02 6.09677518e-03 7.75148630e-01 -3.54689717e-01
2.29851350e-01 -5.93337297e-01 -1.35858631e+00 5.44699967e-01
1.04693317e+00 -5.58437705e-01 -1.20499246e-01 2.97208838e-02
3.78007978e-01 -7.46484399e-01 -4.61987168e-01 1.16103264e-02
9.11534607e-01 -1.05955350e+00 1.27792358e+00 1.06855428e+00
-4.97994214e-01 -8.16164613e-01 -3.91581625e-01 -8.43303204e-01
-3.49592835e-01 -5.82391918e-01 -2.66787916e-01 1.33937287e+00
-2.40027174e-01 -1.33128941e+00 4.82370377e-01 7.51719952e-01
4.13791202e-02 -7.81068206e-02 -1.46181619e+00 -7.02674925e-01
-1.00349164e+00 -5.57721615e-01 7.34728515e-01 1.93576902e-01
1.55808672e-01 4.49816883e-01 -7.76778042e-01 5.82891345e-01
7.38926589e-01 -4.81959224e-01 1.06315911e+00 -1.04056752e+00
1.17373012e-01 3.12992334e-01 -7.18216419e-01 -1.85404110e+00
-5.74570239e-01 -1.65973991e-01 -4.66785952e-02 -1.80888844e+00
-5.34310520e-01 -2.39234135e-01 2.09799483e-01 -1.01787709e-01
-7.77921900e-02 2.98201978e-01 1.81499608e-02 1.17367059e-01
-4.34718519e-01 4.14247751e-01 5.92501700e-01 4.77254950e-02
-3.36234808e-01 6.43206000e-01 -2.46942163e-01 6.70289814e-01
8.17264378e-01 -1.46462172e-01 -2.61584818e-01 -7.02660918e-01
3.97557437e-01 9.47089419e-02 6.85458660e-01 -1.78011310e+00
6.00375891e-01 1.80977225e-01 6.73139095e-01 -8.57767165e-01
7.36546993e-01 -1.27603197e+00 6.51442349e-01 7.91134894e-01
5.75106502e-01 2.24446550e-01 -7.26237372e-02 4.58288074e-01
3.67180169e-01 1.04658544e-01 5.48704386e-01 2.21389890e-01
-6.87277317e-01 -1.48206502e-01 -6.00305021e-01 -6.23417974e-01
1.19123936e+00 -7.34043598e-01 -8.48840475e-01 -3.86175603e-01
1.21444300e-01 -3.05256397e-01 9.68972981e-01 -8.51504877e-02
6.64630711e-01 -1.68643069e+00 4.55444232e-02 4.01988804e-01
-2.18987256e-01 -1.72480702e-01 3.59677136e-01 8.90439749e-01
-9.60395336e-01 7.82296896e-01 1.87359378e-02 -8.76472831e-01
-1.41948223e+00 2.51847357e-01 2.04197466e-01 4.02280599e-01
-5.60930073e-01 3.57100666e-01 -8.02985489e-01 8.85882825e-02
-7.46542439e-02 -3.69815320e-01 -8.86124745e-02 -2.75818348e-01
8.17479312e-01 1.20756984e+00 -1.12169877e-01 -1.05039191e+00
-7.04021454e-01 1.23195684e+00 9.72804844e-01 -2.23255411e-01
6.66005015e-01 -1.17960918e+00 2.56270498e-01 3.06557208e-01
9.29742992e-01 5.90969443e-01 -1.30928814e+00 1.46091655e-01
-4.98640507e-01 -9.41169083e-01 -1.55084878e-01 -4.50728029e-01
-3.85055125e-01 6.86090469e-01 1.17251897e+00 -4.15963158e-02
9.19597030e-01 -5.55694222e-01 1.06634319e+00 4.08388138e-01
1.20809853e+00 -1.00682938e+00 -5.98623157e-01 2.98152626e-01
7.77530074e-02 -1.06566441e+00 2.21553311e-01 -2.31694058e-01
-1.57651469e-01 1.15326512e+00 2.55239159e-01 2.65084356e-01
4.32441592e-01 2.92916387e-01 2.81606801e-02 6.17667317e-01
1.40204713e-01 -1.65216327e-01 -1.32282645e-01 1.05955350e+00
2.04654589e-01 3.75822395e-01 -5.45322657e-01 3.38886827e-01
-2.65334427e-01 2.17767671e-01 6.13011539e-01 9.85773146e-01
-7.71818161e-01 -8.21851492e-01 -8.60588253e-01 -1.64507926e-01
-1.23346888e-01 2.18824178e-01 4.19588000e-01 2.89890021e-01
4.24228430e-01 1.48934841e+00 -1.88294575e-01 -7.75408030e-01
3.76977175e-01 -3.60331237e-01 4.21254814e-01 9.36854184e-02
6.74250945e-02 -2.17726544e-01 -1.02534458e-01 -6.57277465e-01
-3.35802794e-01 -6.37200058e-01 -1.09590411e+00 -8.39799285e-01
-2.03034982e-01 -2.24952176e-02 1.01412153e+00 7.47081816e-01
7.57737696e-01 2.33339220e-01 9.22286153e-01 -8.57433975e-01
1.53985247e-01 -5.59441686e-01 -6.65060461e-01 -2.89196044e-01
4.64118868e-01 -3.89335632e-01 -3.20859253e-01 3.61706913e-02]
|
[6.432774066925049, 0.8467540144920349]
|
f58ed07e-fdf1-41fc-829a-b90a395411f5
|
traffic-sign-detection-with-event-cameras-and
|
2207.13345
| null |
https://arxiv.org/abs/2207.13345v1
|
https://arxiv.org/pdf/2207.13345v1.pdf
|
Traffic Sign Detection With Event Cameras and DCNN
|
In recent years, event cameras (DVS - Dynamic Vision Sensors) have been used in vision systems as an alternative or supplement to traditional cameras. They are characterised by high dynamic range, high temporal resolution, low latency, and reliable performance in limited lighting conditions -- parameters that are particularly important in the context of advanced driver assistance systems (ADAS) and self-driving cars. In this work, we test whether these rather novel sensors can be applied to the popular task of traffic sign detection. To this end, we analyse different representations of the event data: event frame, event frequency, and the exponentially decaying time surface, and apply video frame reconstruction using a deep neural network called FireNet. We use the deep convolutional neural network YOLOv4 as a detector. For particular representations, we obtain a detection accuracy in the range of 86.9-88.9% mAP@0.5. The use of a fusion of the considered representations allows us to obtain a detector with higher accuracy of 89.9% mAP@0.5. In comparison, the detector for the frames reconstructed with FireNet is characterised by an accuracy of 72.67% mAP@0.5. The results obtained illustrate the potential of event cameras in automotive applications, either as standalone sensors or in close cooperation with typical frame-based cameras.
|
['Tomasz Kryjak', 'Piotr Wzorek']
|
2022-07-27
| null | null | null | null |
['traffic-sign-detection']
|
['computer-vision']
|
[ 9.48983356e-02 -8.13501701e-02 4.30571824e-01 -2.03584418e-01
-2.91632742e-01 -1.90786988e-01 9.50897157e-01 2.65542150e-01
-1.06573355e+00 6.75481915e-01 -5.36860466e-01 -8.62753019e-02
-7.70578831e-02 -7.74283350e-01 -5.33645511e-01 -8.51320744e-01
1.93456784e-01 2.39540607e-01 9.15753961e-01 -3.10540468e-01
-1.05471291e-01 1.02627289e+00 -2.26051426e+00 1.97591677e-01
5.01560509e-01 1.15329802e+00 3.03605437e-01 5.51848829e-01
1.96029786e-02 6.66400552e-01 -7.96586215e-01 -3.13297063e-01
5.27957790e-02 -9.43206921e-02 2.52603497e-02 -1.42228752e-02
2.63167679e-01 -4.40684408e-01 -3.32091331e-01 9.57024992e-01
1.51706010e-01 -1.45830330e-03 6.85092390e-01 -1.08013046e+00
3.59741032e-01 -6.58043772e-02 -2.62125552e-01 5.35706401e-01
4.27622050e-01 4.51573968e-01 3.49770159e-01 -5.76313674e-01
7.37757981e-01 7.92703807e-01 3.62827063e-01 4.35847878e-01
-1.15849233e+00 -3.64214838e-01 -2.80158132e-01 9.24901962e-01
-1.30122423e+00 -6.00453973e-01 5.68978310e-01 -6.94038630e-01
9.19029474e-01 -8.42892677e-02 6.74343526e-01 1.33449757e+00
4.30918217e-01 3.64082992e-01 1.17617548e+00 -1.38942257e-01
2.73714125e-01 3.65154386e-01 3.44159216e-01 3.12863708e-01
4.34081614e-01 3.92462730e-01 -2.07581073e-01 3.76908064e-01
3.77959132e-01 5.79272509e-02 -1.85005367e-01 -7.93658569e-02
-1.02166939e+00 7.21552730e-01 3.12005103e-01 6.74546421e-01
-6.90305769e-01 3.19205672e-01 5.82620978e-01 2.48791829e-01
1.20022207e-01 -1.66367278e-01 -1.52183287e-02 -2.77237803e-01
-7.95794725e-01 3.30161273e-01 6.16599381e-01 3.37746918e-01
6.02340579e-01 4.46041763e-01 7.21171731e-03 2.67739832e-01
1.59792170e-01 5.97345769e-01 4.77301568e-01 -4.18996453e-01
7.41413012e-02 4.30903435e-01 1.06387503e-01 -8.25433075e-01
-6.64598465e-01 -3.48879188e-01 -6.93361819e-01 9.35675025e-01
6.60172164e-01 -5.83660416e-02 -4.11263913e-01 1.26273096e+00
2.42202297e-01 2.84751981e-01 4.05553252e-01 8.87465060e-01
5.62959313e-01 6.72640800e-01 6.89208554e-03 -2.76577443e-01
1.52683771e+00 -8.49135518e-02 -7.77496576e-01 9.87727195e-04
2.00672835e-01 -7.63506889e-01 4.94353563e-01 7.39569128e-01
-7.99749911e-01 -8.65399480e-01 -1.22517443e+00 3.57719421e-01
-6.28469825e-01 4.95757610e-01 -1.75746337e-01 5.45118928e-01
-8.89988005e-01 5.14416575e-01 -6.93369627e-01 -3.88564944e-01
-6.33750781e-02 1.36130810e-01 -3.86017144e-01 2.34238356e-01
-1.07676172e+00 1.27756941e+00 5.58934510e-01 2.34393284e-01
-6.59620106e-01 -2.16898263e-01 -6.33562088e-01 1.12395853e-01
3.18187028e-01 -2.51177400e-01 1.12356317e+00 -9.50930715e-01
-1.52947986e+00 9.59967792e-01 5.51735237e-03 -1.12736762e+00
9.64346111e-01 -1.11225143e-01 -7.68900335e-01 5.02660155e-01
-3.50132227e-01 3.57788652e-01 1.05415714e+00 -9.05529380e-01
-7.81986296e-01 -1.29724875e-01 6.79609776e-02 -3.17625821e-01
-3.56470160e-02 -4.63927053e-02 1.78386562e-03 -7.52571747e-02
-5.04171669e-01 -7.59952426e-01 1.13031626e-01 -2.07277052e-02
1.65061444e-01 -1.52602881e-01 9.96696591e-01 -5.22031486e-01
6.94671988e-01 -2.13847613e+00 -7.25032538e-02 8.88174400e-02
1.36169866e-01 7.85039902e-01 1.99161619e-01 8.68676454e-02
-2.27069538e-02 -8.01732540e-01 1.10983746e-02 -9.65278819e-02
-2.28987873e-01 1.29591644e-01 -2.70917658e-02 5.76641381e-01
4.08438623e-01 4.74664360e-01 -7.28861570e-01 -4.55171764e-01
1.19427097e+00 7.93707907e-01 3.78882550e-02 -1.10625871e-01
-2.23760307e-01 3.36516827e-01 -2.28017017e-01 1.20885648e-01
5.51807642e-01 4.32753980e-01 -1.27250940e-01 -2.61095971e-01
-6.05123103e-01 -5.23447156e-01 -1.34733510e+00 8.35850537e-01
-5.07499635e-01 1.23135746e+00 3.56646255e-03 -1.05891097e+00
1.45701361e+00 4.15848643e-01 5.61806321e-01 -1.05091846e+00
4.43602979e-01 4.21198100e-01 -1.51899606e-02 -6.35319769e-01
5.06326616e-01 -4.51263972e-02 3.72065812e-01 -1.40444711e-01
4.68507297e-02 1.44066334e-01 4.90165949e-01 -3.89222056e-01
1.01722944e+00 -1.00419670e-01 3.49649429e-01 2.90936474e-02
1.19006801e+00 -3.77431542e-01 -8.48535076e-02 3.60326409e-01
-9.64535922e-02 3.97862554e-01 5.56024611e-01 -4.42047447e-01
-9.83261764e-01 -9.42383826e-01 -3.82448912e-01 1.87870249e-01
1.02725543e-01 -5.68894558e-02 -7.05337524e-01 -1.93939865e-01
-1.09237939e-01 7.86542237e-01 -2.99276173e-01 -3.32903385e-01
-5.85154831e-01 -4.20332819e-01 4.19322461e-01 3.09577644e-01
5.35669267e-01 -1.01277041e+00 -1.35917795e+00 3.96417528e-01
8.35662931e-02 -1.47974348e+00 5.64857423e-01 2.22549997e-02
-5.93278229e-01 -1.24888241e+00 -7.99375534e-01 -2.03472883e-01
1.96947102e-02 1.96510941e-01 9.08799946e-01 -3.57610404e-01
-6.01138771e-01 5.79169989e-01 -3.51326227e-01 -3.96084338e-01
-8.69551599e-01 -2.85877794e-01 -8.26968551e-02 5.92609227e-01
4.95780706e-01 -3.58236223e-01 -6.34686708e-01 3.47149372e-01
-1.02931702e+00 -2.95600563e-01 4.42225099e-01 3.31629515e-01
2.81211197e-01 -3.63174140e-01 3.71776640e-01 -3.74349892e-01
4.09603000e-01 -2.71510750e-01 -1.28642046e+00 -1.13987923e-03
-3.42756599e-01 -1.36330649e-01 8.92937362e-01 -3.68821144e-01
-1.00440729e+00 3.82531583e-01 -5.71684718e-01 -5.25914013e-01
-6.92950547e-01 -9.96977910e-02 1.57777533e-01 -1.11028075e-01
1.00014246e+00 1.47881344e-01 1.74353331e-01 -3.17584485e-01
2.42120307e-02 5.97492158e-01 6.71839833e-01 -3.75247784e-02
5.18152118e-01 5.88663161e-01 3.25376928e-01 -1.29103112e+00
1.77135356e-02 -5.42213082e-01 -3.43946815e-01 -8.82535636e-01
1.01183188e+00 -8.20548594e-01 -1.04159641e+00 6.75196409e-01
-1.41689038e+00 4.24832199e-03 -5.96636415e-01 7.23251939e-01
-5.89076042e-01 3.94250691e-01 -2.53335595e-01 -1.14990997e+00
-5.14333919e-02 -1.21158361e+00 9.54580486e-01 3.92739207e-01
8.62776190e-02 -7.18138397e-01 -1.28029987e-01 5.62784821e-02
4.03404742e-01 6.72492146e-01 9.69999358e-02 -3.91895413e-01
-5.59228778e-01 -6.52868807e-01 -2.27627963e-01 6.31362259e-01
-3.13384503e-01 2.10778654e-01 -1.26594865e+00 1.20107137e-01
2.35677939e-02 3.00136387e-01 1.11630881e+00 4.39618886e-01
8.00894916e-01 4.40476120e-01 -1.44607946e-01 2.37758219e-01
1.65272844e+00 5.63094437e-01 7.47949183e-01 3.69011104e-01
-2.99008396e-02 6.19738638e-01 7.72983372e-01 6.95973992e-01
-1.11124054e-01 1.22475743e+00 6.75114512e-01 1.72082454e-01
-2.06450775e-01 4.68919784e-01 5.41205466e-01 -6.17249757e-02
-3.12954277e-01 -2.32904285e-01 -8.66212547e-01 4.50078607e-01
-1.92363727e+00 -1.28086913e+00 -7.28486776e-01 2.39648795e+00
7.26916194e-02 8.36844862e-01 4.20207530e-01 5.62425554e-01
9.15921271e-01 1.31022453e-01 -1.53708294e-01 -6.21408165e-01
-1.81967512e-01 2.76242852e-01 5.35611868e-01 3.81788552e-01
-9.35295463e-01 2.90727168e-01 4.59834957e+00 7.03529775e-01
-1.30136633e+00 1.08111069e-01 7.27912039e-02 -2.38496765e-01
4.30302590e-01 -3.58819932e-01 -9.21611071e-01 8.97270441e-01
1.46775365e+00 -1.25730649e-01 4.41595400e-03 8.10511708e-01
4.89678562e-01 -4.47463453e-01 -9.27087128e-01 1.26425600e+00
1.25503361e-01 -1.14212751e+00 -2.94134796e-01 2.00466067e-03
1.96021199e-01 -1.58647355e-02 -1.00624152e-01 -5.34141064e-02
-4.66938406e-01 -6.06349647e-01 8.96845341e-01 9.27204311e-01
7.12838471e-01 -7.69778013e-01 1.05859923e+00 2.41980657e-01
-1.27528882e+00 -1.95500761e-01 -2.31670290e-01 1.71650890e-02
7.47908294e-01 7.33382344e-01 -6.66390359e-01 5.56101263e-01
4.84062493e-01 5.56642473e-01 -3.55522603e-01 1.14177847e+00
-3.14403363e-02 2.98500359e-01 -5.87944806e-01 -2.58129507e-01
3.08789611e-01 -1.79162011e-01 8.25151265e-01 1.49267375e+00
4.30960059e-01 -4.45266634e-01 -3.72811407e-01 7.84614563e-01
4.53379154e-01 -2.59638637e-01 -8.42606187e-01 3.68632406e-01
8.69372219e-04 1.20998156e+00 -6.81832492e-01 -4.02343988e-01
-4.19926077e-01 8.01515877e-01 -3.29493731e-01 1.91801384e-01
-1.02249718e+00 -5.34193456e-01 7.09371209e-01 2.60648608e-01
5.40760398e-01 -3.38473529e-01 2.44904026e-01 -8.21018279e-01
1.04303651e-01 -1.40821859e-01 -2.28707213e-03 -8.68761539e-01
-7.25112438e-01 7.60459900e-01 3.11007828e-01 -1.47392416e+00
-6.74394786e-01 -1.10297775e+00 -4.97372597e-01 6.76818311e-01
-1.50420046e+00 -7.16709077e-01 -5.95289648e-01 7.46663213e-01
5.01622677e-01 -3.00646544e-01 3.32957476e-01 5.16140938e-01
-4.55028534e-01 9.95325521e-02 2.58184820e-01 -2.88308203e-01
2.56729603e-01 -1.07949531e+00 1.43374324e-01 8.77010882e-01
-8.49481001e-02 -2.71358043e-01 9.67700541e-01 -2.47423917e-01
-9.28343773e-01 -8.24758828e-01 7.98204362e-01 8.55995566e-02
5.20742714e-01 -2.04927605e-02 -9.10214782e-01 1.17399015e-01
9.06804204e-02 2.65314758e-01 -9.86650810e-02 -6.49835646e-01
1.04697160e-02 -5.90320528e-01 -1.10485613e+00 2.66875774e-01
4.42416579e-01 -5.47835350e-01 -5.56666911e-01 8.96900520e-02
1.75378501e-01 -2.73222655e-01 -6.65582299e-01 2.12102041e-01
5.45863807e-01 -1.41298115e+00 1.02733266e+00 7.61111081e-02
3.62323225e-02 -4.02699023e-01 2.88879216e-01 -8.10007691e-01
9.37857255e-02 -2.97582477e-01 -1.74036309e-01 9.66192901e-01
-1.21526435e-01 -9.43293869e-01 4.13543254e-01 6.95794821e-02
-2.02620849e-02 -2.04848304e-01 -1.41186905e+00 -8.50445330e-01
-5.24814844e-01 -6.06860936e-01 1.15339749e-01 1.88455001e-01
-5.45104325e-01 1.23384818e-01 -1.31469607e-01 -5.45976348e-02
5.28162241e-01 -2.95518339e-01 5.69678128e-01 -1.47568214e+00
-2.13739961e-01 -6.45687938e-01 -1.26005185e+00 -3.72359157e-01
1.09462395e-01 -3.67862314e-01 -2.16012374e-01 -1.18136728e+00
-5.47668576e-01 1.63207799e-01 -2.64954418e-01 -1.83709785e-01
3.72211128e-01 3.57675076e-01 3.32599014e-01 -1.12639211e-01
-2.81053752e-01 1.85198084e-01 4.12666321e-01 1.69401482e-01
2.64722239e-02 2.07928017e-01 3.91056001e-01 7.76760519e-01
8.19757700e-01 -2.08742674e-02 -6.95964471e-02 1.68426037e-01
1.92999482e-01 4.72086743e-02 9.14315581e-01 -1.59724832e+00
3.93161386e-01 3.80154669e-01 2.86340415e-01 -6.46302283e-01
7.20921934e-01 -1.26211703e+00 4.70236063e-01 8.68848205e-01
5.09128422e-02 -1.04895629e-01 4.11033094e-01 6.11531019e-01
-4.17407751e-01 -5.60553849e-01 1.06308734e+00 -1.94478920e-03
-1.26302624e+00 -2.76803464e-01 -1.01151693e+00 -5.05140185e-01
1.41316891e+00 -7.69669592e-01 -2.88495660e-01 -1.73237264e-01
-7.52831459e-01 -3.23729217e-01 2.15435520e-01 2.71846175e-01
7.17678249e-01 -1.04085553e+00 -6.91531837e-01 1.73802808e-01
1.99494496e-01 -4.39307779e-01 6.06178761e-01 9.02307510e-01
-7.55609512e-01 5.42484581e-01 -5.94871879e-01 -8.65220368e-01
-1.48306668e+00 6.39095306e-01 5.79851747e-01 1.75179631e-01
-7.11283088e-01 6.81621358e-02 -1.46757096e-01 5.06082833e-01
1.12875015e-01 -4.33999062e-01 -5.79419374e-01 4.55428302e-01
5.22386134e-01 7.57651508e-01 4.02438521e-01 -7.67767966e-01
-4.82914150e-01 7.94866621e-01 4.13084686e-01 -9.71138477e-02
1.14011276e+00 7.84845054e-02 4.16534275e-01 5.71116447e-01
1.11756158e+00 -2.70102084e-01 -1.39109945e+00 8.77114087e-02
-7.03724548e-02 -1.73122779e-01 1.75470203e-01 -3.76355976e-01
-8.39183152e-01 1.00253093e+00 1.11087132e+00 5.27227402e-01
1.18047035e+00 -1.30220294e-01 4.58633840e-01 2.53734708e-01
1.31981686e-01 -9.82893288e-01 -2.17281401e-01 2.89579868e-01
7.88373709e-01 -9.78002727e-01 -2.37441480e-01 -1.52620003e-01
-4.34212804e-01 1.84828377e+00 3.69065017e-01 -2.32936770e-01
4.15057570e-01 3.45051378e-01 -2.05434337e-01 3.49853113e-02
-6.78872824e-01 -9.87083316e-01 1.58399671e-01 6.44613087e-01
1.45528361e-01 7.70215392e-02 -6.52307093e-01 2.02888660e-02
1.84022591e-01 4.19752657e-01 6.00609362e-01 5.44948876e-01
-4.48431164e-01 -7.50795364e-01 -7.23615944e-01 2.28461981e-01
-8.94257426e-02 4.08879668e-01 -9.13791880e-02 9.92317080e-01
5.58987498e-01 1.03475654e+00 3.11727732e-01 -4.21855211e-01
8.38551342e-01 2.58447707e-01 4.17507470e-01 1.19887255e-01
-6.60776258e-01 -2.16623709e-01 6.32357076e-02 -6.55547619e-01
-5.84019125e-01 -9.24722314e-01 -8.77201319e-01 -3.81944060e-01
-1.77719399e-01 -2.62423128e-01 1.13243866e+00 8.84299159e-01
4.02364023e-02 6.69032991e-01 4.10083443e-01 -9.54637289e-01
-2.41236389e-01 -5.74881375e-01 -4.35183138e-01 2.93002367e-01
5.94574511e-01 -7.81829894e-01 -3.58723938e-01 1.09668814e-01]
|
[8.434290885925293, -1.0616436004638672]
|
36242baf-e461-4e82-8df9-95ced55ba066
|
enhancing-data-security-against-cyberattacks
|
2305.11652
| null |
https://arxiv.org/abs/2305.11652v1
|
https://arxiv.org/pdf/2305.11652v1.pdf
|
Enhancing data security against cyberattacks in artificial intelligence based smartgrid systems with crypto agility
|
A new paradigm of electricity generation at the distribution level, with renewable and alternative sources, is possible with microgrids. The main idea is to have microgrids deployed on low- or medium-voltage active distribution networks. They can be advantageous in many different ways, such as improving the energy efficiency and reliability of the system and reducing transmission losses and network congestion. There are challenges in implementing MGs with DER units, those are related to power quality and stability issues voltage and fault level changes, energy management, low inertia, further complex protection schemes, load and generation forecasting, cyber-attacks, and cyber security. This paper shows the deep utilization of advanced, accurate, and fast methodologies such as artificial intelligence-based techniques. They guarantee efficient, optimal, safe, and reliable operation of smart grids safe against cyberattacks. AI refers to the computer-based system's ability to perform tasks with intelligence typically associated with human decision-making, they can learn from past experiences and solve problems.
|
['Wilson Lopes', 'Emmanuel Anti', 'Sayawu Diaba', 'Mazaher Karimi', 'Mike Mekkanen', 'Tero Vartiainen', 'Mohammed Elmusrati', 'Marcelo Simoes']
|
2023-05-19
| null | null | null | null |
['energy-management']
|
['time-series']
|
[-3.86148125e-01 -2.91671753e-01 1.23686576e-02 1.91203147e-01
3.46678764e-01 -1.01329112e+00 3.36637437e-01 2.35485241e-01
4.36855406e-01 1.32062399e+00 -6.23700731e-02 -1.49846911e-01
-5.15833020e-01 -9.74361062e-01 9.56328064e-02 -1.19140780e+00
-5.81413209e-01 5.09839058e-01 2.57984921e-02 -4.55312371e-01
4.57411587e-01 8.40015709e-01 -1.52766836e+00 -4.04693604e-01
1.26482749e+00 9.67609942e-01 -1.04654968e-01 3.27127934e-01
5.73022246e-01 4.48804170e-01 -1.09021270e+00 3.01476240e-01
1.87228680e-01 -3.28129023e-01 -4.43491757e-01 -1.69514805e-01
-1.28712714e+00 -2.83048153e-01 -3.62773091e-02 1.33475745e+00
4.73199278e-01 2.58694977e-01 7.21583486e-01 -2.00810814e+00
-5.60450554e-01 4.96361762e-01 -5.59302628e-01 4.57183450e-01
5.17407596e-01 3.89072329e-01 3.39454532e-01 -2.96715647e-01
1.00021228e-01 9.98311222e-01 2.74039268e-01 -1.20680127e-02
-1.03908157e+00 -5.23132384e-01 -8.92990560e-04 1.01509714e+00
-1.20369577e+00 2.33257830e-01 6.50091648e-01 -1.45911276e-01
1.40861988e+00 4.26116705e-01 1.19101167e+00 1.41306251e-01
7.91317642e-01 5.63463986e-01 1.04401910e+00 -3.89468253e-01
7.65453577e-01 -5.67525029e-02 -2.69796699e-02 -3.40857744e-01
7.49296069e-01 7.78110772e-02 5.58899902e-02 3.70100662e-02
-9.47450548e-02 -2.33918130e-02 -8.73312891e-01 -1.44784003e-01
-8.05658877e-01 6.52077317e-01 3.08522910e-01 9.35784638e-01
-5.32024145e-01 -4.28398579e-01 3.39025378e-01 4.86055583e-01
1.02096103e-01 5.98810196e-01 -7.22851932e-01 -1.09401114e-01
-5.86248040e-01 5.48398420e-02 9.56397891e-01 7.64553249e-01
2.76936501e-01 1.02456951e+00 5.01939833e-01 -1.87132075e-01
2.74697691e-01 6.08401060e-01 1.06754243e+00 -5.03894329e-01
-3.49170357e-01 7.26290703e-01 2.25584075e-01 -7.18320251e-01
-7.14549601e-01 -4.52409208e-01 -1.06889546e+00 9.05630827e-01
-2.41918683e-01 -6.07309043e-01 -4.32008475e-01 1.18338084e+00
2.21462786e-01 -1.02374375e-01 2.00621575e-01 4.27860230e-01
-9.30358171e-02 1.15387189e+00 -3.71884823e-01 -9.85659778e-01
1.14859438e+00 -4.99690354e-01 -1.37981725e+00 2.00334102e-01
3.41097742e-01 -6.71575367e-01 -4.13779467e-02 8.56362581e-01
-1.30471122e+00 -2.92215794e-01 -1.56708908e+00 6.64313495e-01
-1.07642913e+00 -4.28244531e-01 2.83139169e-01 6.56061709e-01
-1.08027923e+00 8.95332456e-01 -5.98590136e-01 -1.37300789e-01
1.20115094e-02 4.62147057e-01 2.46804766e-02 3.74964714e-01
-1.41030586e+00 1.81341827e+00 1.05871928e+00 2.85922080e-01
-4.39366519e-01 -7.42414951e-01 -7.94495106e-01 2.74206638e-01
2.22998470e-01 -4.27112758e-01 8.78840208e-01 -6.37226164e-01
-1.38637710e+00 -2.79989779e-01 2.73551911e-01 -9.77951527e-01
1.52470976e-01 -1.66260600e-02 -9.10167336e-01 1.71783596e-01
-1.84670463e-01 -4.41339910e-01 3.36647987e-01 -8.67508411e-01
-7.58236647e-01 -4.97398674e-01 -6.68293417e-01 9.76144299e-02
-6.20463677e-02 -3.66539121e-01 1.19405270e+00 -5.11458158e-01
-1.12230219e-02 -5.48502386e-01 -2.96977460e-01 -6.96378589e-01
-1.25434309e-01 -5.83518267e-01 1.61630535e+00 -9.08595741e-01
1.01243281e+00 -1.69510448e+00 2.66740590e-01 6.65453076e-01
-4.17372376e-01 5.51591039e-01 6.34543300e-01 1.09394515e+00
-2.74507761e-01 5.77418879e-02 1.47602990e-01 6.50057077e-01
3.36247563e-01 6.62862897e-01 -1.39196560e-01 4.64761198e-01
-1.69991180e-01 6.33541644e-01 -8.57548177e-01 5.61475277e-01
7.93036163e-01 2.24807397e-01 2.31066808e-01 1.72003388e-01
-5.43123810e-03 2.63745993e-01 -2.28489384e-01 5.17656267e-01
7.67325342e-01 2.04152558e-02 3.59249175e-01 -3.93821627e-01
-2.61481315e-01 -2.54947841e-01 -1.58895922e+00 1.13469422e+00
-1.64851353e-01 4.47105020e-01 2.48220563e-01 -1.56863034e+00
6.30376101e-01 7.07265079e-01 5.40430844e-01 -7.56081522e-01
7.29628801e-02 3.39583129e-01 2.06709266e-01 -4.19078529e-01
6.33040816e-02 3.19016337e-01 3.91779125e-01 6.47012234e-01
2.29054064e-01 -5.46353221e-01 7.76388466e-01 2.35977266e-02
4.64813352e-01 -1.41230404e-01 8.25320065e-01 -9.57222342e-01
8.09962332e-01 -4.50031728e-01 8.13732982e-01 -1.84969202e-01
7.00323582e-02 -4.41520512e-01 1.22539826e-01 -3.32687944e-01
-7.34274328e-01 -1.01728106e+00 -2.21464276e-01 9.64111760e-02
1.91454679e-01 1.08929195e-01 -4.40298587e-01 -1.82378948e-01
2.50601649e-01 1.43806696e+00 -8.35521594e-02 -5.72921395e-01
-4.14932132e-01 -1.21316504e+00 -4.06821787e-01 3.93261582e-01
5.76185942e-01 -5.74658275e-01 -6.66389585e-01 7.17135549e-01
4.70300019e-01 -6.40465438e-01 6.94514960e-02 3.77100319e-01
-6.53370440e-01 -1.14443910e+00 -3.81989121e-01 -5.87133706e-01
7.53867030e-01 -2.35234350e-02 1.14490044e+00 3.50093395e-01
-4.69136924e-01 -5.85574061e-02 -3.26703310e-01 -6.32279754e-01
-2.12728351e-01 -3.72182935e-01 6.50093973e-01 -5.39450943e-01
1.01870373e-01 -1.20987952e+00 -7.55186498e-01 2.10635796e-01
-7.12291479e-01 -1.73247337e-01 5.40179133e-01 8.39077175e-01
-4.64172773e-02 1.31479919e+00 1.53708756e+00 -1.84291676e-01
6.94356978e-01 -7.20842957e-01 -9.53570664e-01 3.97303641e-01
-1.39311337e+00 -1.93486527e-01 1.06440079e+00 5.09385392e-03
-6.84144795e-01 -5.03776193e-01 -7.87984207e-02 2.07889482e-01
-2.03013137e-01 2.40067855e-01 -6.04372501e-01 -3.53545129e-01
-8.38165879e-02 4.95007187e-01 9.02485251e-02 -4.28205073e-01
1.76689446e-01 5.16439438e-01 3.19367200e-01 4.31611687e-02
1.31413519e+00 3.49413082e-02 4.63780105e-01 -4.81919050e-01
-1.46927565e-01 2.51706969e-02 -5.85177422e-01 -3.27690244e-01
4.67612475e-01 -5.13339818e-01 -1.21172273e+00 8.32777917e-01
-6.96330786e-01 2.38461703e-01 -2.74015844e-01 4.32716697e-01
3.64411399e-02 3.43018919e-01 -5.15345871e-01 -9.56943035e-01
-6.99957728e-01 -1.07143342e+00 -1.25436768e-01 9.53228831e-01
4.92072888e-02 -1.41646659e+00 -2.68140852e-01 -1.13346525e-01
7.46916533e-01 8.01865637e-01 1.05916047e+00 -7.00242817e-01
-5.29727340e-01 -2.85733584e-02 4.00538802e-01 8.11619043e-01
6.57242656e-01 1.68270856e-01 -3.75990659e-01 -9.56222713e-01
2.35607207e-01 8.16951767e-02 -4.46930349e-01 1.71288997e-01
7.13114858e-01 -7.23658919e-01 -3.84591937e-01 1.01097375e-01
1.88908398e+00 1.25798321e+00 4.71900642e-01 3.95532250e-01
-7.38828778e-02 2.90507585e-01 4.76660967e-01 5.74328244e-01
3.14152420e-01 1.66458532e-01 5.64119339e-01 5.56256175e-02
6.41773880e-01 5.11867464e-01 9.49480459e-02 9.56228316e-01
-1.00569546e-01 -3.41604739e-01 -4.84200120e-01 5.22282720e-01
-1.91222572e+00 -9.27621126e-01 2.98820231e-02 1.87299359e+00
4.65586394e-01 4.47911352e-01 -1.08152904e-01 1.09495759e+00
4.89051253e-01 -3.94021571e-01 -6.35817587e-01 -6.10078275e-01
-3.67009789e-01 6.55790120e-02 3.38074028e-01 2.68982023e-01
-4.21189278e-01 -1.78355962e-01 5.68273449e+00 7.72114754e-01
-9.17248189e-01 -1.23215623e-01 5.73483825e-01 2.13412017e-01
-7.67337829e-02 -1.54064283e-01 -2.80779272e-01 1.11300170e+00
1.16104460e+00 -1.34369969e+00 6.52509868e-01 6.86811626e-01
5.65976143e-01 -6.37521565e-01 -9.25568521e-01 3.85410666e-01
3.67032713e-03 -1.13492441e+00 -4.91688661e-02 -9.67825800e-02
1.15813375e+00 -2.68435895e-01 -6.82656348e-01 -9.20896381e-02
4.18934315e-01 -6.83521569e-01 5.24751365e-01 4.56110835e-01
-4.73778576e-01 -1.53390825e+00 1.26123106e+00 3.41456711e-01
-1.20249140e+00 -4.06501681e-01 1.69194862e-02 -4.66460913e-01
7.53672719e-01 6.95801914e-01 -3.22448969e-01 1.48568463e+00
9.86185431e-01 5.74642003e-01 -3.91968995e-01 1.12120295e+00
-5.06001174e-01 2.43171185e-01 -3.94395977e-01 -1.86119542e-01
-1.89930163e-02 -6.79636538e-01 3.18916619e-01 2.96704620e-01
3.00848365e-01 3.03775251e-01 4.09602582e-01 5.28155327e-01
5.39301634e-01 -3.69745523e-01 -4.60761696e-01 4.37347174e-01
7.46929944e-01 1.34000707e+00 -8.08425844e-01 -5.60804069e-01
-3.35126854e-02 2.57710338e-01 -6.87580943e-01 4.46496844e-01
-4.24551755e-01 -8.43202531e-01 6.19425595e-01 -3.16131949e-01
-4.17782634e-01 -2.30990320e-01 -5.46237826e-01 -8.98091793e-01
-1.28703699e-01 -7.71034360e-01 5.27083695e-01 -9.76326644e-01
-1.62567568e+00 3.35156858e-01 2.31368378e-01 -1.06315064e+00
-1.05291688e+00 -5.11625707e-01 -1.28857791e+00 1.07241881e+00
-1.67477739e+00 -6.01323605e-01 1.12151854e-01 7.66837478e-01
5.23652971e-01 -2.68920422e-01 9.12267447e-01 2.42668584e-01
-6.89580739e-01 -1.60644874e-01 6.51058495e-01 -2.50309408e-01
-3.83162081e-01 -1.54683673e+00 -2.98735291e-01 1.39598465e+00
-5.78224063e-01 -1.65669069e-01 1.01553798e+00 -2.87928164e-01
-1.86389565e+00 -3.69375885e-01 2.93062121e-01 6.04908764e-01
9.38652456e-01 -2.27936283e-02 -9.83988464e-01 5.71559131e-01
1.34468627e+00 -5.87880194e-01 5.55105686e-01 -6.19940519e-01
7.01934636e-01 -4.24117774e-01 -1.78012824e+00 5.01184583e-01
7.64559060e-02 4.44757827e-02 -9.58670557e-01 5.58767915e-01
1.05160557e-01 9.17007588e-03 -1.11369407e+00 5.30608118e-01
3.11013125e-02 -5.11874318e-01 8.63559961e-01 -1.06852248e-01
-7.71606445e-01 -6.99525535e-01 3.52571368e-01 -2.14207125e+00
-3.47485781e-01 -1.07558000e+00 -4.92870241e-01 1.27544332e+00
1.84891462e-01 -1.58239508e+00 1.99203238e-01 4.90610331e-01
-1.95814669e-01 -5.89184761e-01 -1.23839760e+00 -9.32678521e-01
1.78092197e-02 4.42831784e-01 1.18141913e+00 1.27102506e+00
7.84692347e-01 1.23413950e-02 1.65668920e-01 6.92301869e-01
1.04403913e+00 3.03353518e-01 2.26944592e-03 -1.32233298e+00
2.67442375e-01 -4.57980156e-01 -5.33245802e-01 1.75183371e-01
-3.07848640e-02 -1.01732172e-01 -5.27209938e-01 -2.17712331e+00
-8.29309404e-01 3.00647438e-01 -3.46278399e-01 4.97921437e-01
1.80835709e-01 -3.59581858e-02 2.30637729e-01 -2.69203298e-02
8.71683732e-02 6.49451196e-01 8.22331548e-01 -5.68983145e-02
2.25104928e-01 1.31002799e-01 1.00627735e-01 9.07022059e-01
1.40749669e+00 2.23954469e-01 -7.43079722e-01 -3.07593704e-03
-3.65329348e-02 2.42401212e-01 -3.40105832e-01 -1.21907830e+00
4.53219891e-01 -2.35020563e-01 7.05098629e-01 -9.31533575e-01
-2.21152782e-01 -1.32831693e+00 8.46239090e-01 1.50189781e+00
7.13418305e-01 8.26626956e-01 8.68014619e-02 1.61114439e-01
-3.26072603e-01 -4.23104912e-01 8.66219044e-01 -1.92442104e-01
-8.50006580e-01 -2.22772434e-01 -9.77138102e-01 -5.27767897e-01
1.92073929e+00 -1.61831945e-01 -7.48033047e-01 -2.54971087e-01
-9.31125998e-01 1.04594052e+00 2.16922402e-01 6.88490629e-01
2.13387594e-01 -1.05579579e+00 -4.74656075e-01 3.43685091e-01
-7.25055158e-01 -2.40964249e-01 1.18433766e-01 4.38413471e-01
-6.49861217e-01 3.65272105e-01 -7.55173862e-01 -3.75078946e-01
-8.92679393e-01 5.69188714e-01 4.85450596e-01 -2.33649492e-01
-3.53664786e-01 -1.16350666e-01 -6.93267345e-01 3.51159006e-01
-8.33909735e-02 -1.75580993e-01 -4.60653186e-01 4.32247072e-01
7.95391500e-01 9.59292710e-01 5.74792564e-01 -1.23826467e-01
-3.79969090e-01 3.39054555e-01 1.93399534e-01 3.25978756e-01
1.41210723e+00 -3.60628605e-01 -5.91573894e-01 3.15480351e-01
5.96680462e-01 -6.03668272e-01 -6.75068855e-01 4.84271944e-01
2.13666767e-01 -3.98362786e-01 1.96605816e-01 -1.33497608e+00
-1.30338538e+00 2.68655777e-01 6.71046615e-01 1.38217998e+00
1.45076668e+00 -1.00642645e+00 4.51377541e-01 6.33272529e-02
1.01741636e+00 -1.46621740e+00 -3.96419913e-01 1.87045082e-01
6.34074271e-01 -6.84890151e-01 2.33500674e-01 3.80160846e-03
7.81634375e-02 1.32791245e+00 4.47295338e-01 -2.69885927e-01
9.95063424e-01 6.28854632e-01 -3.76735657e-01 2.39876047e-01
-9.60451424e-01 2.44237378e-01 -1.23072781e-01 8.31533074e-01
-1.69831906e-02 1.54749691e-01 -7.67237484e-01 2.52547204e-01
-2.33696356e-01 1.38934664e-02 9.49177623e-01 1.30131721e+00
-3.54483455e-01 -1.12554145e+00 -5.52538037e-01 4.22655463e-01
-7.28500605e-01 3.86669636e-01 6.17804050e-01 8.49474072e-01
3.74906629e-01 1.24398339e+00 -2.76956186e-02 2.80612558e-01
4.45429265e-01 -9.39672068e-02 -1.70670636e-03 -5.52804060e-02
-6.18111432e-01 -1.35244310e-01 -3.02793622e-01 1.71331540e-02
-1.74843401e-01 -7.80170798e-01 -1.44258046e+00 -6.37545884e-01
-4.43921208e-01 1.06147420e+00 9.80492115e-01 1.06682789e+00
1.63462088e-01 6.67533159e-01 1.22954583e+00 -1.01472461e+00
-7.44434118e-01 -9.68034506e-01 -1.01280141e+00 9.29959640e-02
-6.73162937e-02 -5.48984826e-01 -9.09144163e-01 -4.08538938e-01]
|
[5.776973724365234, 2.568315267562866]
|
81d6e3fd-1214-4427-8b10-af9ed767a04a
|
learning-visual-n-grams-from-web-data
|
1612.09161
| null |
http://arxiv.org/abs/1612.09161v2
|
http://arxiv.org/pdf/1612.09161v2.pdf
|
Learning Visual N-Grams from Web Data
|
Real-world image recognition systems need to recognize tens of thousands of
classes that constitute a plethora of visual concepts. The traditional approach
of annotating thousands of images per class for training is infeasible in such
a scenario, prompting the use of webly supervised data. This paper explores the
training of image-recognition systems on large numbers of images and associated
user comments. In particular, we develop visual n-gram models that can predict
arbitrary phrases that are relevant to the content of an image. Our visual
n-gram models are feed-forward convolutional networks trained using new loss
functions that are inspired by n-gram models commonly used in language
modeling. We demonstrate the merits of our models in phrase prediction,
phrase-based image retrieval, relating images and captions, and zero-shot
transfer.
|
['Laurens van der Maaten', 'Allan Jabri', 'Ang Li', 'Armand Joulin']
|
2016-12-29
|
learning-visual-n-grams-from-web-data-1
|
http://openaccess.thecvf.com/content_iccv_2017/html/Li_Learning_Visual_N-Grams_ICCV_2017_paper.html
|
http://openaccess.thecvf.com/content_ICCV_2017/papers/Li_Learning_Visual_N-Grams_ICCV_2017_paper.pdf
|
iccv-2017-10
|
['zero-shot-transfer-image-classification']
|
['computer-vision']
|
[ 3.75951082e-01 -1.17149040e-01 -2.48217732e-01 -4.07234371e-01
-8.93222988e-01 -6.47447050e-01 6.91690266e-01 3.72083724e-01
-4.68663126e-01 4.63908672e-01 2.65961707e-01 -6.56019866e-01
4.26664054e-01 -6.99596584e-01 -9.68560874e-01 -3.41049731e-01
5.46207242e-02 3.78428996e-01 2.99142743e-03 1.19934157e-01
4.16381985e-01 4.64546055e-01 -1.53821886e+00 7.96122670e-01
8.81823078e-02 1.10479546e+00 1.77208737e-01 1.06677032e+00
-4.74490076e-01 6.66333318e-01 -3.47023457e-01 -7.01839089e-01
1.18091762e-01 -3.20913553e-01 -7.73929715e-01 4.75989521e-01
1.03749633e+00 -4.35177654e-01 -2.76312351e-01 1.04381430e+00
3.18146527e-01 6.61738291e-02 1.00012910e+00 -1.21763492e+00
-1.33389866e+00 1.74037933e-01 -5.05176544e-01 1.88503057e-01
2.73195565e-01 1.44196093e-01 1.24583173e+00 -1.26734233e+00
5.67338109e-01 1.30320716e+00 4.70236689e-01 5.48390567e-01
-1.16990733e+00 -4.99384880e-01 1.29268067e-02 3.22095543e-01
-1.16779292e+00 -1.97562844e-01 5.00245929e-01 -8.03440750e-01
1.09091949e+00 1.84583008e-01 4.89243597e-01 1.23909152e+00
1.82363138e-01 8.09250832e-01 7.98206568e-01 -6.68948293e-01
-1.13738375e-02 4.89365518e-01 2.20824361e-01 8.56964290e-01
-8.40043426e-02 -9.43772346e-02 -7.97225118e-01 -1.50207013e-01
7.21404850e-01 6.88220859e-02 7.93267936e-02 -4.73786265e-01
-1.00310230e+00 1.03381157e+00 6.71335697e-01 -1.65618695e-02
-1.51238769e-01 3.32965016e-01 6.14908516e-01 1.72534510e-01
5.75839877e-01 2.17193455e-01 -1.16529010e-01 2.11029097e-01
-7.29542077e-01 8.19494054e-02 7.09718764e-01 1.11032367e+00
9.27297771e-01 -2.06008047e-01 -3.00871462e-01 7.68655479e-01
2.64069498e-01 5.17665386e-01 5.56877732e-01 -5.38573921e-01
3.14792305e-01 3.80422294e-01 1.09287068e-01 -1.09264123e+00
-3.60033140e-02 -1.83965400e-01 -8.75991702e-01 1.51113123e-01
1.14818178e-01 3.19314480e-01 -1.23067939e+00 1.12601149e+00
-1.29700691e-01 -1.17714377e-02 -6.94671720e-02 8.64988148e-01
8.39605153e-01 9.69961166e-01 5.96259236e-01 5.24967313e-02
1.51885676e+00 -1.16556931e+00 -2.86245078e-01 -3.85235101e-01
3.39768291e-01 -8.98328424e-01 1.31963885e+00 4.57725115e-02
-8.88505816e-01 -7.05931664e-01 -6.33613586e-01 -3.01549971e-01
-9.35534835e-01 -2.76408549e-02 4.90308672e-01 3.86283308e-01
-1.19189227e+00 2.41459280e-01 -2.56780297e-01 -5.94448030e-01
6.41351819e-01 2.55070180e-01 -3.25728834e-01 -2.28256077e-01
-8.53554547e-01 7.61218131e-01 3.18664968e-01 -1.97085902e-01
-7.48040378e-01 -5.28522134e-01 -8.92600358e-01 3.82763147e-01
-1.68865114e-01 -6.79902911e-01 1.15909076e+00 -1.62911892e+00
-7.50742555e-01 1.40788805e+00 -3.91391546e-01 -5.03302753e-01
1.85486868e-01 -6.87986836e-02 -2.81543404e-01 5.58656693e-01
1.10215947e-01 1.34026790e+00 1.25742769e+00 -1.33478451e+00
-5.61795831e-01 -3.67402652e-04 1.26708001e-01 1.64020851e-01
-1.00458014e+00 3.34800243e-01 -4.94841099e-01 -6.39530003e-01
-4.46553558e-01 -8.20172668e-01 -3.11747849e-01 4.83503431e-01
-1.94200397e-01 -2.79488444e-01 9.10066128e-01 -5.68165481e-01
8.08234394e-01 -2.24735546e+00 -1.08267687e-01 1.99635580e-01
1.74809843e-01 3.56002420e-01 -6.34717941e-01 3.90945047e-01
-1.02880761e-01 4.00112927e-01 1.38063669e-01 -2.41735026e-01
1.14132715e-02 2.31412932e-01 -7.11709023e-01 6.91557303e-04
3.96527559e-01 1.28426290e+00 -8.47747624e-01 -7.10015476e-01
3.62756968e-01 4.74157959e-01 -3.35852772e-01 3.24342519e-01
-4.73477274e-01 5.64318672e-02 -4.02330548e-01 4.76004213e-01
2.68188149e-01 -8.92963231e-01 -1.91761211e-01 -2.82963187e-01
1.39740229e-01 -5.78147992e-02 -3.38136137e-01 1.27895641e+00
-6.20897770e-01 1.06977713e+00 -4.49837208e-01 -8.15290272e-01
7.60450661e-01 3.92287672e-01 2.97412157e-01 -7.19231904e-01
2.58014705e-02 4.86281700e-02 -4.91937339e-01 -7.00304747e-01
5.88139236e-01 -2.17163995e-01 -2.25918135e-03 5.51185966e-01
2.36118048e-01 -5.10523543e-02 1.83917165e-01 3.66413683e-01
7.27486432e-01 -2.58641094e-01 2.62026548e-01 6.25685006e-02
3.99085969e-01 1.16354622e-01 -2.85174370e-01 8.46905112e-01
1.99057050e-02 7.52956629e-01 3.04360211e-01 -8.47444654e-01
-1.63730085e+00 -9.87982750e-01 1.09356999e-01 1.46212614e+00
8.72410834e-02 -3.17824781e-01 -3.85363847e-01 -7.26957798e-01
2.96339635e-02 4.56672281e-01 -6.47353530e-01 3.48314308e-02
-1.34363845e-01 -2.93171138e-01 1.77550524e-01 6.50857806e-01
9.01048481e-02 -1.42609859e+00 -3.39310378e-01 -2.51933448e-02
-2.54920665e-02 -1.26631057e+00 -5.57925582e-01 1.09372929e-01
-4.40813363e-01 -8.70284855e-01 -9.86968517e-01 -1.36413646e+00
1.03269255e+00 5.34156203e-01 1.40727174e+00 2.48089984e-01
-6.22855842e-01 8.41053605e-01 -5.34350753e-01 -4.94744003e-01
-4.27075088e-01 -1.17794815e-02 -2.79040635e-01 1.22036554e-01
8.51165175e-01 -3.75615835e-01 -5.81454515e-01 -4.13750578e-03
-8.97388637e-01 2.04766631e-01 8.57012570e-01 9.02559876e-01
5.63719511e-01 -5.08789539e-01 3.91271442e-01 -7.94222832e-01
6.91518128e-01 -5.23948789e-01 -4.06975091e-01 6.71285331e-01
-3.59982491e-01 -1.91057175e-01 6.63640141e-01 -7.26914048e-01
-5.83193421e-01 3.97682160e-01 -9.07851569e-03 -7.14022517e-01
-3.54623139e-01 5.28647542e-01 4.49798882e-01 -3.47625196e-01
7.35757291e-01 3.31270039e-01 -2.64089584e-01 -7.44823068e-02
6.87670410e-01 8.69937301e-01 4.64417130e-01 -2.49100119e-01
7.86939979e-01 2.99981534e-01 -1.40782461e-01 -1.06377292e+00
-1.12544465e+00 -1.05420840e+00 -6.98826015e-01 -3.87809813e-01
1.19391596e+00 -1.02004683e+00 -5.36340415e-01 2.65889205e-02
-1.47335792e+00 -6.32538367e-03 -2.23359540e-01 1.31208390e-01
-5.69579422e-01 4.68042195e-01 -6.59457564e-01 -7.06635177e-01
-5.99188149e-01 -8.56920123e-01 1.26839101e+00 1.67776078e-01
-2.34905720e-01 -1.00090396e+00 -1.60050660e-01 3.60849798e-01
3.03512573e-01 -2.91877359e-01 1.15568042e+00 -5.92289567e-01
-7.29845047e-01 -4.65466201e-01 -7.50280142e-01 4.79162037e-01
-1.66757852e-01 -1.21295586e-01 -1.04581928e+00 -1.32944062e-01
-5.69733322e-01 -1.06447530e+00 8.64552498e-01 1.93139911e-01
1.56870317e+00 -5.06339371e-01 -2.08449557e-01 2.73956358e-01
1.63747144e+00 -1.91842243e-01 7.15722919e-01 2.53615201e-01
8.56263340e-01 5.58909953e-01 3.00654173e-01 2.52978742e-01
1.77892923e-01 4.09609854e-01 4.43966717e-01 -3.94484997e-01
2.88971653e-03 -3.82547557e-01 1.27310798e-01 6.91891313e-01
2.76978552e-01 -4.67419922e-01 -1.03307056e+00 8.30512643e-01
-1.82598174e+00 -9.53565896e-01 -6.41811118e-02 1.77181244e+00
7.06786990e-01 1.48211122e-01 4.72374037e-02 -6.37098968e-01
6.79100394e-01 5.22012003e-02 -4.51915354e-01 -6.14171922e-01
2.06763409e-02 3.06149423e-01 5.85036278e-01 2.03528017e-01
-1.27741635e+00 9.40330327e-01 7.22551441e+00 7.06705749e-01
-1.23424149e+00 -6.41904101e-02 1.04651904e+00 1.92938015e-01
-2.51984894e-01 -1.90095872e-01 -8.10639262e-01 2.38865018e-01
1.14566946e+00 -1.42222077e-01 8.49293247e-02 1.10524774e+00
1.02685235e-01 1.37302443e-01 -1.23170686e+00 1.26304436e+00
5.21810174e-01 -1.54843843e+00 7.79501081e-01 -5.40104806e-02
8.15204799e-01 7.60048628e-02 4.35045660e-01 9.26579610e-02
1.90713957e-01 -1.25274825e+00 5.81134915e-01 3.09066355e-01
7.99934685e-01 -3.92181575e-01 4.65644538e-01 7.75067210e-02
-8.77853215e-01 -1.62162036e-01 -7.07069755e-01 1.01365089e-01
1.46288261e-01 2.58014113e-01 -9.32528853e-01 -7.89508671e-02
8.25091541e-01 7.96961188e-01 -8.56293857e-01 1.07257009e+00
-6.63303584e-02 5.20917952e-01 4.23851795e-02 -3.46298158e-01
5.01039624e-01 3.23322276e-03 5.80940358e-02 1.57346439e+00
1.71750218e-01 -2.05871798e-02 3.60637605e-01 6.75000787e-01
-3.87502700e-01 3.06825221e-01 -7.73313046e-01 -3.34905565e-01
3.27073932e-02 1.48050880e+00 -6.28955066e-01 -6.21696174e-01
-8.29771817e-01 1.00611663e+00 4.16500747e-01 4.79895532e-01
-4.92632002e-01 -4.60464619e-02 2.92413771e-01 3.06722254e-01
4.69429195e-01 -2.42236733e-01 -1.35891721e-01 -1.05149877e+00
-1.61141250e-02 -8.08893800e-01 2.50024378e-01 -1.27658129e+00
-1.82281816e+00 5.60477138e-01 -2.87294388e-01 -1.40354371e+00
-2.63516963e-01 -1.17939758e+00 -5.24930358e-01 8.32863688e-01
-1.64364874e+00 -1.63742065e+00 -3.29920828e-01 5.39379239e-01
9.02599156e-01 -1.49806201e-01 1.24285245e+00 2.60539651e-01
5.89251034e-02 4.19017971e-01 2.42660090e-01 4.08784091e-01
7.56457388e-01 -1.08638000e+00 6.44926667e-01 4.77276802e-01
4.91637617e-01 4.04440463e-01 6.30558014e-01 -1.51183814e-01
-9.59375739e-01 -1.25159228e+00 1.13697696e+00 -3.11765105e-01
1.06329572e+00 -4.83811229e-01 -8.50592256e-01 7.14277923e-01
3.79923671e-01 2.13446379e-01 9.75213826e-01 -4.20335829e-02
-7.71780670e-01 1.40680209e-01 -6.29204810e-01 6.22176766e-01
5.44746339e-01 -1.08615434e+00 -3.58483940e-01 8.30717921e-01
5.90905488e-01 2.72033513e-01 -5.47250211e-01 -2.06598237e-01
6.35459960e-01 -4.00666058e-01 1.33236170e+00 -1.07885790e+00
8.88559937e-01 1.44615084e-01 -7.07126483e-02 -1.03275454e+00
-4.06479537e-01 -1.42079189e-01 4.28474322e-02 9.42022026e-01
5.34548461e-01 1.26857683e-02 9.42768931e-01 7.31932700e-01
2.16150045e-01 -3.49249363e-01 -6.13613963e-01 -5.16106963e-01
1.03522636e-01 -6.47397265e-02 -9.34764594e-02 9.36888695e-01
1.59177691e-01 6.62292004e-01 -5.07065833e-01 -5.81630580e-02
5.69539726e-01 1.21863462e-01 6.57646775e-01 -1.00510442e+00
-3.30196053e-01 -4.09259975e-01 -6.74104333e-01 -1.08793616e+00
4.13035810e-01 -8.07449520e-01 2.55294383e-01 -1.61925471e+00
6.17650926e-01 -3.06145370e-01 -2.89037168e-01 6.03591025e-01
-1.02755450e-01 7.50815988e-01 2.66215891e-01 2.55058378e-01
-8.79119277e-01 1.97763428e-01 9.77513313e-01 -6.29811406e-01
3.70746940e-01 -1.34231046e-01 -3.19549799e-01 6.94053531e-01
6.98131680e-01 -3.03708345e-01 -4.04998928e-01 -6.04158044e-01
4.95143950e-01 -1.60466254e-01 8.45746517e-01 -6.73402429e-01
2.66662747e-01 -2.19515443e-01 6.05735779e-01 -6.52970791e-01
4.28181112e-01 -1.04831409e+00 -3.77880484e-01 1.98485583e-01
-7.93393850e-01 6.24930970e-02 2.68983275e-01 7.10963547e-01
-3.98837864e-01 -5.02857327e-01 6.41446471e-01 -5.88977695e-01
-1.15425944e+00 3.65799576e-01 -5.15223563e-01 -2.76316971e-01
1.01878297e+00 -8.86705890e-02 -4.25360769e-01 -6.73807681e-01
-8.43728065e-01 6.79145455e-02 2.93816239e-01 6.11382008e-01
1.00092649e+00 -1.22140813e+00 -6.09477937e-01 -4.87193093e-02
6.87121511e-01 -5.01096368e-01 1.16289838e-03 2.06604674e-01
-5.85129142e-01 4.75935757e-01 -3.94714743e-01 -6.27350867e-01
-1.55941379e+00 9.93386269e-01 7.04532638e-02 -1.23698331e-01
-4.49889660e-01 8.78186405e-01 5.24348617e-01 -7.05597550e-02
3.11806053e-01 3.97979841e-02 -2.24235266e-01 -8.86986703e-02
7.80362010e-01 -3.76254380e-01 -1.82113200e-01 -7.68268168e-01
-7.11190235e-03 4.71957296e-01 -3.25086057e-01 -2.75051706e-02
1.29242480e+00 -1.91021830e-01 -1.44274816e-01 4.29841340e-01
1.66196764e+00 -6.88868403e-01 -1.01327693e+00 -2.80558676e-01
-3.88461165e-02 -3.58982623e-01 -1.95753664e-01 -6.27362370e-01
-7.18505919e-01 1.39913690e+00 7.42743850e-01 2.27658838e-01
9.27123666e-01 1.81837127e-01 6.85643792e-01 6.25976920e-01
9.93200243e-02 -9.25818622e-01 5.93666017e-01 5.18262982e-01
7.40867972e-01 -1.57695150e+00 -1.11569978e-01 -4.08927202e-01
-5.57597339e-01 1.41335964e+00 5.34355521e-01 -1.39859110e-01
6.53225303e-01 -6.38286471e-02 1.23741351e-01 -2.52640396e-01
-7.81316161e-01 -2.53873914e-01 6.82508647e-01 7.27362990e-01
4.04183924e-01 -7.26513043e-02 1.58524029e-02 -7.49656633e-02
2.47355491e-01 -8.00395161e-02 4.63479906e-01 8.76896024e-01
-6.71720147e-01 -1.04795766e+00 -2.71102607e-01 7.00879514e-01
-4.42691803e-01 -6.20342731e-01 -4.52766746e-01 1.97442874e-01
-3.24009150e-01 5.84033132e-01 3.40052545e-01 -2.28665546e-01
-1.09181479e-01 2.45630354e-01 2.29467869e-01 -9.55241382e-01
-4.73512501e-01 5.23018697e-03 -5.91900237e-02 -5.52621484e-02
-5.66295445e-01 -2.47299820e-01 -8.71745586e-01 -2.22409561e-01
-8.74198303e-02 -8.53231698e-02 8.11591327e-01 7.59784162e-01
2.72655666e-01 1.55300662e-01 4.34037775e-01 -9.81518626e-01
-3.85753721e-01 -7.25125909e-01 -3.84991735e-01 8.73693645e-01
3.43870014e-01 -8.73241425e-02 -1.61327913e-01 9.33093905e-01]
|
[10.380906105041504, 1.8809481859207153]
|
5c461c84-e394-49c9-af78-75ef9dd4049f
|
multi-task-learning-framework-for-extracting
|
2211.03742
| null |
https://arxiv.org/abs/2211.03742v1
|
https://arxiv.org/pdf/2211.03742v1.pdf
|
Multi-Task Learning Framework for Extracting Emotion Cause Span and Entailment in Conversations
|
Predicting emotions expressed in text is a well-studied problem in the NLP community. Recently there has been active research in extracting the cause of an emotion expressed in text. Most of the previous work has done causal emotion entailment in documents. In this work, we propose neural models to extract emotion cause span and entailment in conversations. For learning such models, we use RECCON dataset, which is annotated with cause spans at the utterance level. In particular, we propose MuTEC, an end-to-end Multi-Task learning framework for extracting emotions, emotion cause, and entailment in conversations. This is in contrast to existing baseline models that use ground truth emotions to extract the cause. MuTEC performs better than the baselines for most of the data folds provided in the dataset.
|
['Ashutosh Modi', 'Ashwani Bhat']
|
2022-11-07
| null | null | null | null |
['causal-emotion-entailment']
|
['natural-language-processing']
|
[ 2.46192813e-01 3.72020870e-01 -2.47838661e-01 -9.67064857e-01
-8.33804131e-01 -5.25151014e-01 5.63927531e-01 9.82859284e-02
-1.41255185e-01 9.92785454e-01 1.08674848e+00 7.69505790e-03
2.13243634e-01 -5.72011173e-01 -7.77214348e-01 -3.09414148e-01
2.02192262e-01 1.76443189e-01 -7.16580868e-01 -2.00733945e-01
5.95204234e-02 -1.25073344e-01 -1.47569895e+00 1.13236952e+00
5.65579116e-01 1.20695686e+00 -4.58264202e-01 7.69295573e-01
-3.62425566e-01 1.64073873e+00 -6.39739037e-01 -8.05902123e-01
-4.57085073e-01 -7.94852853e-01 -1.40585792e+00 3.42762843e-02
1.78617790e-01 -1.00482844e-01 1.00528225e-01 4.88189608e-01
5.35120487e-01 7.48899281e-02 9.40834045e-01 -1.54495299e+00
-4.46944326e-01 1.25509036e+00 -3.29347432e-01 -6.65386766e-02
6.55366540e-01 -4.34769392e-01 1.57486629e+00 -7.87656903e-01
4.35833395e-01 1.67486501e+00 5.06318986e-01 6.89959168e-01
-9.61678326e-01 -6.58740103e-01 3.59433323e-01 4.34203804e-01
-6.98705673e-01 -3.04304659e-01 9.72995162e-01 -2.46720597e-01
1.58397651e+00 2.45647714e-01 2.56025672e-01 1.67793274e+00
2.84701735e-01 1.32999432e+00 1.05586648e+00 -4.57163513e-01
6.91176131e-02 -5.68631403e-02 3.62174809e-01 3.35084319e-01
-6.49935365e-01 -2.18326882e-01 -7.71315217e-01 -1.57846883e-01
8.41196328e-02 -3.40389073e-01 -1.94264978e-01 5.88288963e-01
-1.01794410e+00 9.50037956e-01 1.46267325e-01 2.41593748e-01
-7.06214666e-01 4.09485728e-01 7.35374749e-01 3.66835177e-01
1.09226525e+00 4.89602178e-01 -1.09397149e+00 -5.10631859e-01
-4.95743990e-01 4.09360111e-01 1.21158779e+00 6.65162683e-01
4.56670105e-01 -2.25913048e-01 -1.97961912e-01 9.77689147e-01
6.98486716e-02 1.88312717e-02 4.02998447e-01 -7.90436506e-01
4.08512354e-01 8.35299015e-01 -9.78996605e-02 -9.28498745e-01
-7.05780208e-01 -6.50524572e-02 -7.15286970e-01 -5.75747848e-01
4.35656123e-02 -1.00420964e+00 -5.09190679e-01 1.91596758e+00
2.84709394e-01 5.07835865e-01 5.17681241e-01 6.93116009e-01
1.19761753e+00 9.79476571e-01 1.39660388e-01 -3.29688162e-01
1.41940844e+00 -1.00959754e+00 -1.24112976e+00 -2.43932202e-01
7.34256327e-01 -7.44977415e-01 1.10762000e+00 6.33420169e-01
-6.70698524e-01 -2.01612115e-01 -6.62343144e-01 -2.47255892e-01
-5.22115827e-01 1.86854601e-01 8.94165516e-01 2.74419576e-01
-5.09122968e-01 3.43879998e-01 -2.44207770e-01 -1.52050987e-01
-2.62293275e-02 3.17725651e-02 -7.73304701e-02 1.54586673e-01
-1.80267262e+00 9.50951934e-01 3.32104772e-01 3.37330392e-03
-7.49791801e-01 -7.54166067e-01 -1.04925442e+00 -6.71807975e-02
4.06891912e-01 -4.73503113e-01 1.67374039e+00 -1.36373210e+00
-1.53912330e+00 8.17294240e-01 -5.21427810e-01 -4.27880585e-01
1.35967601e-02 -8.09283853e-01 -5.04763305e-01 -2.70267606e-01
-2.42613226e-01 6.18616581e-01 5.04126847e-01 -1.06319141e+00
-6.38289094e-01 -2.67051756e-01 1.41521841e-01 1.42940134e-01
-3.93154949e-01 4.82716948e-01 -6.16672821e-02 -4.20789897e-01
-6.05953693e-01 -8.13003182e-01 -2.16961019e-02 -8.80810380e-01
-8.31994832e-01 -1.09759271e+00 6.69578791e-01 -5.14882505e-01
1.23761630e+00 -1.94666171e+00 3.26974094e-01 -2.66966045e-01
-2.07048282e-02 -4.50993717e-01 -1.96416691e-01 6.66753888e-01
-3.83913994e-01 1.85407907e-01 -1.07783603e-03 -5.21348000e-01
3.65749180e-01 3.78488004e-01 -8.22291851e-01 -2.13358272e-02
5.58844328e-01 7.93957591e-01 -6.88726246e-01 -3.99402201e-01
-1.52723074e-01 4.79597718e-01 -5.95516860e-01 6.99337065e-01
-6.07506275e-01 2.65980870e-01 -3.44804466e-01 3.43729764e-01
2.28424430e-01 -1.44521119e-02 1.03997394e-01 -1.86709642e-01
1.20504498e-01 7.37143338e-01 -6.68453693e-01 1.41330862e+00
-9.21720028e-01 5.91084599e-01 -1.40900046e-01 -8.36442769e-01
8.93150508e-01 9.74496663e-01 4.59409803e-01 -3.01197290e-01
5.00806093e-01 -2.58242749e-02 -1.37141764e-01 -9.33297813e-01
4.86384094e-01 -6.69946790e-01 -6.72549844e-01 6.38218760e-01
1.56790763e-01 -3.16453874e-01 1.44385725e-01 1.11148335e-01
1.13733983e+00 1.46966174e-01 4.02946889e-01 3.01327884e-01
4.10934836e-01 -1.45579427e-01 6.78900599e-01 1.87101156e-01
1.40910968e-03 3.03295553e-01 9.50021744e-01 -5.06755710e-01
-5.42637944e-01 -4.70730841e-01 3.35866399e-02 1.39032340e+00
-3.11762869e-01 -6.60834134e-01 -6.94527268e-01 -8.86637330e-01
-1.48116559e-01 1.20297813e+00 -7.79013157e-01 -8.91929679e-03
-4.79766697e-01 -7.78220475e-01 8.66351724e-01 5.55595577e-01
3.35536420e-01 -1.61954904e+00 -4.02308524e-01 3.79836082e-01
-1.05008340e+00 -1.51504982e+00 -1.21328786e-01 5.30784786e-01
-4.02270228e-01 -1.02821863e+00 -1.31230429e-01 -5.57787418e-01
7.52590820e-02 -5.12736499e-01 1.59207809e+00 -3.55572850e-01
-1.49298310e-01 3.07183564e-01 -6.77870452e-01 -1.03558040e+00
-3.19041669e-01 4.67342958e-02 -1.82052106e-01 2.18884498e-01
9.65038121e-01 -4.29761797e-01 -8.64333659e-02 6.39576167e-02
-7.28199244e-01 1.52442485e-01 3.43490243e-01 5.89161992e-01
3.24212730e-01 -3.48536372e-02 1.13048732e+00 -9.30600345e-01
1.07451463e+00 -8.25749397e-01 2.15755850e-01 -1.59151442e-02
-2.80470133e-01 -5.19641824e-02 7.19141006e-01 -2.48781472e-01
-1.36929154e+00 -1.01325866e-02 -4.10591781e-01 -2.27243658e-02
-7.19555020e-01 8.51156533e-01 -4.27231610e-01 9.82857108e-01
3.18176448e-01 -2.87701190e-01 -4.75215614e-01 -2.30753511e-01
5.87886810e-01 9.69117641e-01 2.26974010e-01 -7.66326785e-01
-1.77379310e-01 9.00747031e-02 -3.48378360e-01 -6.71370506e-01
-1.61295378e+00 -4.09417331e-01 -3.64750057e-01 -3.63863409e-01
9.52404797e-01 -1.04733968e+00 -8.18478942e-01 3.40723813e-01
-1.67211282e+00 -3.89015198e-01 8.43917280e-02 5.21515727e-01
-6.71979725e-01 -2.23620653e-01 -8.97791743e-01 -1.07898259e+00
-6.04456782e-01 -6.96765959e-01 1.23911536e+00 -3.78435627e-02
-1.00767386e+00 -1.01819730e+00 1.31791234e-01 4.52412039e-01
2.07395889e-02 6.60823345e-01 1.04715002e+00 -7.87200332e-01
2.26774260e-01 -1.73767969e-01 -4.13687304e-02 2.89169341e-01
1.43638402e-01 4.85281721e-02 -1.23798144e+00 4.45242316e-01
9.29462630e-03 -1.25179410e+00 9.72018838e-01 3.68515551e-01
1.25140452e+00 -6.95150435e-01 -5.97213916e-02 4.10476327e-02
8.97530854e-01 1.68963999e-01 5.43435872e-01 4.82533574e-02
4.54739004e-01 1.22775400e+00 5.99832833e-01 7.28166401e-01
7.39781439e-01 3.08898836e-01 3.91815454e-01 -3.20368677e-01
3.45779806e-01 -8.54387134e-02 5.81619382e-01 6.53076351e-01
1.10381976e-01 -5.34255624e-01 -6.44828618e-01 7.09423542e-01
-2.06017256e+00 -1.08080947e+00 -4.11384642e-01 9.66964602e-01
1.38589609e+00 -1.93191499e-01 6.33109957e-02 9.44901332e-02
4.32801902e-01 3.20136338e-01 -3.20727885e-01 -1.09825099e+00
-1.69195458e-01 2.88950056e-02 -2.58007586e-01 4.94819313e-01
-1.19103706e+00 9.81569111e-01 5.74837255e+00 5.29095173e-01
-9.56209779e-01 -1.77339800e-02 1.01374674e+00 -4.02677685e-01
-3.81281197e-01 -1.88184693e-01 -7.83585370e-01 4.20948640e-02
1.09184861e+00 7.27918297e-02 2.27734536e-01 8.86965156e-01
4.65228468e-01 6.12302423e-02 -1.48726869e+00 8.16063404e-01
3.25758129e-01 -8.51322293e-01 -1.14107169e-01 -2.71607280e-01
7.03435063e-01 -1.91007823e-01 -3.03644598e-01 7.50604749e-01
4.94695514e-01 -1.21291602e+00 5.17678082e-01 2.95067847e-01
4.55842197e-01 -1.19681013e+00 1.10368061e+00 3.77887100e-01
-6.98657393e-01 5.09319417e-02 -1.12781666e-01 -6.57481372e-01
4.37306792e-01 1.07458115e+00 -9.96373534e-01 3.15950453e-01
5.84219813e-01 8.45148683e-01 -7.07994848e-02 1.49063915e-01
-7.37405956e-01 1.18404543e+00 -4.44325022e-02 -4.82310385e-01
4.36222553e-01 2.39544749e-01 2.51900822e-01 1.77520466e+00
-6.89023808e-02 2.41008043e-01 2.54073799e-01 8.51419747e-01
-5.56770027e-01 3.22161287e-01 -6.36007428e-01 -2.28964522e-01
1.00935418e-02 1.49605989e+00 -2.47371748e-01 -3.57908577e-01
-3.44703913e-01 1.13892961e+00 4.43525016e-01 2.86878824e-01
-9.59303498e-01 -3.73207062e-01 6.94628477e-01 -7.34464288e-01
2.06309501e-02 2.63540685e-01 -2.90619463e-01 -9.22075450e-01
-1.02164306e-01 -9.40702796e-01 5.16261578e-01 -8.24619949e-01
-1.64925802e+00 7.11539268e-01 -7.44830295e-02 -4.99425024e-01
-6.91665769e-01 -5.79228580e-01 -9.67672527e-01 7.04139590e-01
-1.57826424e+00 -1.19522429e+00 -6.46554381e-02 5.90935469e-01
8.83586109e-01 2.61403203e-01 1.20897400e+00 -4.00522500e-02
-6.20131195e-01 3.58919889e-01 -6.90589428e-01 3.58550012e-01
1.01426876e+00 -1.42222202e+00 3.68287824e-02 5.43444276e-01
1.28697485e-01 4.50901598e-01 9.30914760e-01 -7.05103517e-01
-1.16754651e+00 -1.08996737e+00 1.53547537e+00 -5.34407616e-01
6.85805202e-01 -4.27656263e-01 -6.31536961e-01 8.53611946e-01
1.04013515e+00 -4.21362311e-01 1.27920651e+00 9.32819009e-01
-4.94098276e-01 1.40976205e-01 -9.36920345e-01 3.88983339e-01
6.19805634e-01 -3.94686282e-01 -7.74364173e-01 2.90976942e-01
1.11332476e+00 -2.19190642e-01 -6.76602960e-01 4.57004368e-01
5.74431300e-01 -7.61144996e-01 4.09763396e-01 -1.11127818e+00
1.51632404e+00 2.44991690e-01 -3.09647530e-01 -1.79582465e+00
4.36076075e-02 -5.84219933e-01 -1.33305281e-01 1.74917734e+00
8.68679881e-01 2.27869488e-02 3.96279722e-01 8.10271263e-01
-1.65615022e-01 -1.05280757e+00 -6.08634830e-01 -1.91336423e-01
2.80680180e-01 -1.03361213e+00 5.80234885e-01 1.15022850e+00
5.49275875e-01 1.15526605e+00 -6.80268347e-01 -2.21376419e-01
1.88389570e-01 4.64703381e-01 7.28080392e-01 -9.74808395e-01
-2.36616224e-01 -3.12867224e-01 3.14948022e-01 -6.02965355e-01
9.52251732e-01 -8.58058572e-01 3.76544058e-01 -1.77758861e+00
2.39081621e-01 2.35227197e-01 -1.08986199e-01 8.33506584e-01
-4.22277302e-01 -1.62011817e-01 -4.83698957e-02 -5.45896947e-01
-6.42514884e-01 7.53590345e-01 9.47425485e-01 9.76488926e-03
-8.28430951e-02 -2.01548487e-01 -8.36205661e-01 1.07453275e+00
9.81178403e-01 -4.11018521e-01 -2.03474805e-01 -1.88934281e-01
7.95304418e-01 4.72397655e-02 1.68798521e-01 -2.60388315e-01
-1.10240243e-01 -3.42979610e-01 3.32034528e-01 -8.08100998e-01
3.87371361e-01 -6.75687969e-01 -5.05915999e-01 -1.88762426e-01
-9.27614570e-01 5.96384890e-02 2.46874169e-01 2.58456349e-01
-5.10105908e-01 -8.16993490e-02 3.09149355e-01 -1.04021288e-01
-5.08753896e-01 -1.22670248e-01 -7.09892929e-01 2.37584978e-01
6.54395878e-01 5.27511060e-01 -1.74969882e-01 -8.69211972e-01
-3.55420083e-01 4.07912552e-01 -5.27985692e-01 8.47189903e-01
7.31043696e-01 -1.16799581e+00 -1.10581553e+00 -5.11782348e-01
1.46181390e-01 -1.40743375e-01 8.56449530e-02 7.86025167e-01
4.97437626e-01 3.20864618e-01 3.48050296e-01 -3.96027155e-02
-1.38856757e+00 4.46285158e-01 2.01318145e-01 -4.46302265e-01
-1.98836237e-01 9.79933619e-01 9.99884978e-02 -7.62972534e-01
3.37242782e-01 -5.20802140e-01 -4.86324787e-01 5.35019517e-01
7.47803509e-01 6.25967234e-02 -1.24689028e-01 -3.88634890e-01
-3.37130845e-01 1.42155543e-01 -4.45993431e-02 -2.80921400e-01
1.63117635e+00 -9.27578062e-02 -5.18823743e-01 9.79878604e-01
1.42225003e+00 -5.07215178e-03 -7.61853099e-01 -1.56436518e-01
2.61848539e-01 2.92749405e-01 1.11446142e-01 -1.32917047e+00
-6.97828174e-01 8.90950322e-01 -1.89240113e-01 2.31329918e-01
1.08130217e+00 2.72888333e-01 1.05639553e+00 4.94884640e-01
-9.67992917e-02 -1.27458966e+00 2.68924087e-01 1.12307000e+00
1.50570357e+00 -1.27479601e+00 -4.22170699e-01 -4.25299317e-01
-1.07208991e+00 1.13731039e+00 8.77759159e-01 -4.75517586e-02
3.74623001e-01 7.40586162e-01 3.29959750e-01 -4.26333517e-01
-1.39304316e+00 -2.51716435e-01 2.42871970e-01 7.23445266e-02
1.25284779e+00 1.84566021e-01 -3.91860902e-01 1.31277490e+00
-4.77887213e-01 -7.33553916e-02 4.23094839e-01 6.08408689e-01
-1.62869826e-01 -8.94173801e-01 -3.31200153e-01 3.52826059e-01
-9.85044777e-01 -3.35567713e-01 -1.48422837e+00 3.16466749e-01
1.74900845e-01 1.60039496e+00 -1.39244990e-02 -4.75609452e-01
3.96247923e-01 6.34398818e-01 2.20926240e-01 -4.54933882e-01
-9.83109176e-01 1.74532294e-01 9.36999619e-01 -5.26789427e-01
-7.13193476e-01 -4.25927311e-01 -1.80992055e+00 8.49064291e-02
-1.05788827e-01 2.75948286e-01 6.43445373e-01 1.33463061e+00
3.16461384e-01 9.31743503e-01 7.71520019e-01 -6.14916861e-01
-4.46025804e-02 -1.45275331e+00 -3.10149670e-01 7.00847268e-01
2.19526932e-01 -3.45753431e-01 -5.58730721e-01 2.07765013e-01]
|
[12.753632545471191, 6.288743019104004]
|
7e2b7532-d259-4ea9-9b09-0134ac63f90d
|
energy-efficient-deployment-of-multiple-uavs
|
2003.05668
| null |
http://arxiv.org/abs/2003.05668v1
|
http://arxiv.org/pdf/2003.05668v1.pdf
|
Energy-efficient Deployment of Multiple UAVs Using Ellipse Clustering to Establish Base Stations
|
The demand for future wireless communication systems is being satisfied for
various circumstances through unmanned aerial vehicles (UAVs), which act as
flying base stations (BSs). In this letter, we propose an ellipse clustering
algorithm that maximizes the user coverage probability of UAV- BSs and avoids
inter-cell interference with minimal transmit power. We obtain the coverage of
each UAV by adjusting its antenna half-power beamwidth, orientation, and 3D
location by minimizing the path loss of the cell-edge user. Simulation results
confirm that the proposed algorithm achieves high system throughput and
coverage probability with lower transmit power compared to conventional
algorithms.
|
[]
|
2020-03-12
| null | null | null | null |
['user-simulation']
|
['natural-language-processing']
|
[-2.30622470e-01 1.78056791e-01 -2.60290027e-01 4.62248176e-01
4.51110393e-01 -1.00928009e+00 9.20052733e-03 -1.93885162e-01
-1.68768704e-01 1.05510855e+00 -4.49197888e-01 -6.94045782e-01
-3.98377627e-01 -8.89831781e-01 -1.57977507e-01 -1.06699359e+00
-4.23959166e-01 -2.04034612e-01 2.33517408e-01 4.41566743e-02
-2.27046728e-01 7.62947679e-01 -7.80172467e-01 -9.04892862e-01
1.21509433e+00 1.07438731e+00 2.14839384e-01 5.18098772e-01
8.78287971e-01 -9.41136777e-02 -7.83549607e-01 -1.23724647e-01
2.92231083e-01 -1.31746054e-01 -2.53236949e-01 6.33866727e-01
-6.11058176e-01 -2.73534805e-01 -3.96107286e-01 9.35915530e-01
2.09331572e-01 -1.38313532e-01 7.74282813e-01 -1.56697690e+00
-3.88050109e-01 2.24377096e-01 -1.06380391e+00 -6.01522587e-02
1.70345560e-01 -2.82419115e-01 1.41109928e-01 -2.74396300e-01
2.62804210e-01 5.09694576e-01 5.13006806e-01 -5.78497313e-02
-3.63603204e-01 -5.17535627e-01 -3.96312624e-02 -1.55368015e-01
-1.95960891e+00 -2.75175244e-01 2.76242167e-01 -7.11657777e-02
3.94806057e-01 5.25568902e-01 1.02431905e+00 -9.48846806e-03
7.42411017e-01 2.91398227e-01 1.70821965e-01 -4.43637639e-01
1.78648278e-01 -6.66185170e-02 -5.11656284e-01 7.72317529e-01
1.40498877e+00 4.66018431e-02 3.76303047e-01 -4.67219092e-02
9.23322022e-01 1.46371573e-02 -8.32285941e-01 -5.48465908e-01
-1.34243703e+00 5.90503514e-01 7.01428950e-01 3.31827402e-01
-7.68658102e-01 -7.41083398e-02 -5.05722702e-01 1.10281095e-01
7.17601797e-05 4.39974010e-01 -3.52501661e-01 5.76551497e-01
-6.77650273e-01 -6.87051043e-02 7.91522801e-01 1.83752835e+00
1.96847066e-01 2.75892079e-01 -3.49345692e-02 2.51269460e-01
6.29120231e-01 1.07511222e+00 -2.04909563e-01 -9.10266697e-01
-2.58436780e-02 6.02999151e-01 5.77022970e-01 -1.08037710e+00
-4.63001668e-01 -1.25337327e+00 -1.09679365e+00 -2.34944493e-01
-2.65739471e-01 -1.24323320e+00 -5.10366917e-01 1.20813453e+00
5.29558361e-01 -7.01943189e-02 4.64406610e-01 7.67072082e-01
2.59418815e-01 7.87975073e-01 -3.51407200e-01 -9.63867545e-01
9.23423111e-01 -7.69185901e-01 -6.44640148e-01 -2.24620178e-01
4.13006246e-01 -6.95639431e-01 -1.68297797e-01 1.69394150e-01
-9.86294329e-01 -2.34784745e-02 -1.26065147e+00 1.09051108e+00
-6.54476359e-02 7.53157914e-01 3.53587091e-01 9.43223894e-01
-9.02961075e-01 -3.85821760e-01 -5.66737592e-01 -5.55864394e-01
2.85369039e-01 5.31204104e-01 9.39441249e-02 -1.07580289e-01
-6.89697802e-01 3.23627472e-01 3.05010498e-01 -1.07064128e-01
-5.32369792e-01 -1.49818406e-01 -6.60509706e-01 -9.26815048e-02
5.40925264e-01 -1.01393902e+00 7.38647759e-01 -1.40756369e-01
-1.13932347e+00 3.60315330e-02 -4.42148671e-02 -5.43910384e-01
-3.09913754e-01 3.43598992e-01 -3.18570733e-01 2.77639657e-01
2.47287583e-02 4.94434565e-01 2.65698791e-01 -1.42477787e+00
-9.76963937e-01 -4.29306448e-01 4.84255366e-02 4.52567935e-01
-2.93595880e-01 -4.13149893e-01 -2.75919378e-01 -4.00931656e-01
3.81417692e-01 -1.23084581e+00 -5.69763303e-01 -3.10069233e-01
-6.89574063e-01 1.32664368e-01 1.31771040e+00 1.16574034e-01
1.29710412e+00 -2.01595545e+00 3.05379957e-01 2.10278094e-01
1.77230109e-02 1.23053588e-01 1.69853255e-01 3.21781248e-01
6.56384170e-01 -2.08352543e-02 -2.95994859e-02 3.91099364e-01
-5.44106066e-01 -1.51827306e-01 -2.18269061e-02 8.17113936e-01
-5.50097525e-01 3.84744436e-01 -7.39066780e-01 -2.49690711e-01
2.18091100e-01 3.51975143e-01 -3.81646842e-01 6.73938915e-02
4.16050315e-01 3.03218186e-01 -8.78919423e-01 1.10489726e+00
9.56795216e-01 6.07316801e-03 1.38434261e-01 -7.17331320e-02
-4.64373589e-01 -9.05989826e-01 -8.59958589e-01 9.38950062e-01
-4.04589698e-02 4.77016509e-01 7.96622872e-01 -7.20405340e-01
7.31140375e-01 4.03775811e-01 8.39362323e-01 4.22382593e-01
8.27668607e-01 -6.05843887e-02 -1.28653541e-01 -2.08290100e-01
4.08749640e-01 2.82195479e-01 -7.62651637e-02 -7.95732886e-02
-6.08481131e-02 -2.74088204e-01 7.57952929e-02 2.42401525e-01
8.06915641e-01 -7.05528796e-01 6.81660354e-01 -5.14421582e-01
4.65238452e-01 4.71078865e-02 4.08905715e-01 2.41182104e-01
-1.89658508e-01 -1.55057803e-01 -1.84895694e-01 1.55193165e-01
-4.82504457e-01 -9.30369854e-01 -1.75022781e-01 3.76474708e-02
1.17038059e+00 -1.07436672e-01 -9.99313235e-01 -1.87912911e-01
-1.16587698e-01 8.13638687e-01 -9.99563634e-02 -2.68918514e-01
4.41293269e-01 -9.30420339e-01 3.28296572e-01 -1.74208209e-01
6.95396721e-01 1.13540307e-01 -1.04239309e+00 4.71900105e-02
-1.40184835e-01 -1.50056124e+00 -4.03233081e-01 -2.37999082e-01
-6.59793317e-01 -1.28142679e+00 -6.57218635e-01 -7.61694312e-01
9.91807044e-01 1.21821284e+00 2.35544428e-01 1.78470492e-01
-1.78132597e-02 8.46735060e-01 -5.21625698e-01 -8.15823793e-01
3.50494206e-01 1.64028779e-01 4.44053948e-01 -1.22976795e-01
-5.12440167e-02 -3.15146923e-01 -8.30485225e-01 7.77689457e-01
-4.25822228e-01 -2.52096534e-01 5.63073337e-01 1.46365911e-01
5.87233782e-01 1.01814747e+00 5.85416436e-01 9.29695889e-02
5.41952491e-01 -9.23804581e-01 -9.25341904e-01 2.89115429e-01
-2.95884997e-01 -9.23737049e-01 2.76190847e-01 1.84305057e-01
-5.36634624e-01 2.42298976e-01 5.49971521e-01 -5.54065891e-02
-2.13480648e-02 4.02460247e-01 -3.67319345e-01 -6.72506392e-01
9.86568406e-02 3.63107882e-02 -2.24461868e-01 4.18689340e-01
4.88559417e-02 8.92553270e-01 2.21393362e-01 2.20123962e-01
9.06498909e-01 3.64231884e-01 5.51932812e-01 -1.45794189e+00
-3.78989398e-01 -1.32539496e-01 -4.18763250e-01 -7.41585016e-01
7.68498063e-01 -1.32526898e+00 -7.27723897e-01 1.79317176e-01
-1.02403986e+00 8.37357342e-02 3.23022544e-01 1.01436901e+00
-1.01271167e-01 8.21077153e-02 1.35310784e-01 -1.28036904e+00
-2.75676161e-01 -1.09332871e+00 6.92149103e-01 6.08700752e-01
2.25417182e-01 -7.89340913e-01 -5.64743042e-01 -2.64364779e-01
4.13138360e-01 7.80547738e-01 3.08162391e-01 2.88127244e-01
-9.57863092e-01 -5.29415667e-01 -4.80840616e-02 -3.91601562e-01
4.92234439e-01 -1.97019354e-02 -9.57287326e-02 -7.59333730e-01
-2.16929018e-01 4.31127459e-01 3.26773912e-01 1.17047179e+00
7.87023187e-01 -5.02212346e-01 -1.20533955e+00 8.47240984e-01
1.48526418e+00 7.10326254e-01 2.27480039e-01 -1.52000129e-01
2.03607097e-01 1.54416785e-01 1.03769195e+00 7.38067746e-01
1.85319155e-01 2.41530359e-01 1.03732979e+00 2.85606813e-02
4.95125353e-01 1.71425074e-01 1.09361477e-01 4.99408752e-01
-2.16180414e-01 -1.40688145e+00 -3.70207042e-01 5.37473381e-01
-1.55722356e+00 -5.26485503e-01 -3.28754395e-01 2.15527177e+00
-9.42011699e-02 -3.31538647e-01 8.15321729e-02 1.24781750e-01
7.32428849e-01 -9.51688811e-02 -4.01135862e-01 3.62719953e-01
-1.11637980e-01 -8.05333495e-01 1.28834355e+00 5.93696177e-01
-1.06816363e+00 5.84433377e-01 6.49381256e+00 5.50499737e-01
-7.95576394e-01 -3.20094347e-01 2.56738782e-01 -2.03807786e-01
-1.29224002e-01 -3.39036554e-01 -6.69581532e-01 2.38803208e-01
3.90324980e-01 -5.29642224e-01 3.94950509e-01 5.35943210e-01
2.02615693e-01 -4.49894279e-01 -7.75398090e-02 8.53142083e-01
-2.27086991e-02 -1.00318336e+00 1.06185954e-02 5.08946896e-01
8.92084122e-01 -2.30673745e-01 -2.93831807e-02 -5.06147742e-01
3.30723643e-01 -2.86705434e-01 3.93723458e-01 5.49985647e-01
9.36569750e-01 -1.12204218e+00 7.43041754e-01 5.96037984e-01
-1.42495763e+00 -5.68008244e-01 -6.92225277e-01 3.58895846e-02
4.98354465e-01 9.29039180e-01 -7.20796347e-01 8.33460212e-01
4.80830699e-01 5.20887077e-01 9.80021060e-02 1.36139429e+00
-1.01684541e-01 3.72902393e-01 -5.01398444e-01 -3.57499570e-01
3.04607689e-01 -6.20003045e-01 1.36046755e+00 4.30245787e-01
1.32509243e+00 8.90069962e-01 2.93040574e-01 3.70481133e-01
-4.79929857e-02 -1.19586423e-01 -1.17349613e+00 6.20051026e-02
1.26633012e+00 1.20042109e+00 -9.08803701e-01 2.42804736e-01
-2.54520327e-01 7.40908563e-01 -4.84699041e-01 6.76310718e-01
-7.59355903e-01 -8.78338695e-01 6.21495962e-01 1.78100064e-01
3.01789790e-01 -8.45695436e-01 -1.33947939e-01 -5.51049590e-01
-5.32803953e-01 -1.44124478e-01 -1.47668167e-03 -6.29569709e-01
-2.14031592e-01 6.23724461e-01 -1.45435348e-01 -1.42178535e+00
3.68033439e-01 -3.57203186e-01 -5.66642284e-01 1.88338131e-01
-1.07554507e+00 -9.29885268e-01 -6.00934327e-01 5.54806411e-01
1.61218792e-01 -6.35462284e-01 5.26557148e-01 -1.40291557e-01
-7.96756089e-01 1.18184678e-01 2.09433615e-01 -2.96161085e-01
1.97424442e-01 -5.11383593e-01 -5.58969021e-01 1.06210446e+00
-3.46201479e-01 4.13129210e-01 7.08944917e-01 -7.15621650e-01
-1.69937158e+00 -1.33948672e+00 2.69754946e-01 1.10522062e-01
1.17202334e-01 2.44160861e-01 4.35397536e-01 4.65744436e-01
8.22671950e-01 -5.05959153e-01 9.92344499e-01 -5.14285088e-01
1.04186177e+00 -4.97266240e-02 -1.39641964e+00 7.82672822e-01
9.17418599e-01 6.62620246e-01 2.44499967e-01 4.83717114e-01
8.72708440e-01 -3.47221166e-01 -8.59788895e-01 8.13832283e-01
5.56566536e-01 -5.74012637e-01 8.28045785e-01 2.08150610e-01
-5.80079615e-01 -7.75155127e-01 -4.79864597e-01 -1.76890790e+00
-7.72513449e-01 -7.09590316e-01 1.45755559e-01 7.85775244e-01
4.21766549e-01 -4.94703978e-01 3.83942127e-01 -8.68076906e-02
-1.83034062e-01 -8.35597277e-01 -7.21275926e-01 -6.33102715e-01
-7.15361595e-01 2.01120511e-01 7.57063448e-01 6.18394852e-01
2.16359451e-01 7.11287260e-01 -3.62757683e-01 1.09991264e+00
8.31287861e-01 -8.76590759e-02 7.73337364e-01 -1.43378830e+00
4.39113826e-01 7.05821021e-03 -1.72364905e-01 -1.27161741e+00
-9.27674621e-02 -3.90617907e-01 -2.21128359e-01 -1.78355014e+00
-5.59898734e-01 -2.83457935e-01 2.18105033e-01 8.60931948e-02
2.73592621e-01 2.00669914e-01 -1.93917483e-01 -1.42470226e-01
-5.91430783e-01 6.13311052e-01 1.48297000e+00 5.94156347e-02
-2.78065741e-01 6.06141925e-01 -7.43739724e-01 9.43436980e-01
1.02635014e+00 -4.11324427e-02 -9.88930404e-01 -5.13966203e-01
-4.32788841e-02 5.56398690e-01 -6.40768409e-02 -1.53446972e+00
4.17291313e-01 -4.23649102e-01 5.95861793e-01 -8.13551009e-01
2.01939151e-01 -1.57439256e+00 2.88354754e-01 8.15965354e-01
6.01156116e-01 -1.35918083e-02 1.96992218e-01 1.12538648e+00
1.35594666e-01 3.78667153e-02 8.11018825e-01 3.73585463e-01
1.10830860e-02 5.33675909e-01 -1.23796296e+00 -7.02404678e-01
2.16810155e+00 -4.40872073e-01 -2.12828889e-01 -7.04378605e-01
-3.60768914e-01 6.81856573e-01 5.56047738e-01 -1.13630481e-01
6.52614415e-01 -1.19862521e+00 -5.03138781e-01 4.58789110e-01
-6.54095337e-02 -1.74620703e-01 1.85466498e-01 1.00938642e+00
-5.06867528e-01 9.61090088e-01 -1.41620040e-01 -5.34233987e-01
-1.33206522e+00 2.63970405e-01 4.04050350e-01 5.43106437e-01
2.70993829e-01 9.21998858e-01 -1.98873967e-01 1.04666829e-01
7.84036070e-02 5.95752336e-02 -4.46980774e-01 -2.95964420e-01
3.23434293e-01 6.06144011e-01 -4.60405439e-01 -7.70612001e-01
-2.99020261e-01 7.33536839e-01 6.25464678e-01 2.17708036e-01
9.66751635e-01 -8.85035932e-01 -3.23829770e-01 -4.03925180e-01
3.34848017e-01 4.83084410e-01 -8.84606063e-01 3.84651303e-01
-6.64602876e-01 -5.97779512e-01 6.85001493e-01 -4.57247764e-01
-1.26210380e+00 1.02187969e-01 2.38032758e-01 6.60963655e-01
1.30650699e+00 -2.05743447e-01 5.46329200e-01 4.46637154e-01
1.24302649e+00 -7.33308017e-01 -4.32970941e-01 1.67631403e-01
6.63175225e-01 -5.49525380e-01 1.91377491e-01 -9.35106575e-01
-3.08223456e-01 9.76491213e-01 6.18624985e-01 -4.28529680e-02
1.05173504e+00 4.57919568e-01 -3.59037280e-01 -1.30222291e-01
-5.28859198e-01 -3.94018620e-01 -8.95926058e-02 9.58279312e-01
8.83696601e-02 4.12037641e-01 -5.37470102e-01 7.16374755e-01
-2.33093008e-01 -2.60686904e-01 7.07108319e-01 7.75601685e-01
-1.31409001e+00 -8.17828000e-01 -4.94579166e-01 5.62613487e-01
-2.75999129e-01 3.09673131e-01 -4.32591796e-01 5.95607400e-01
2.11800024e-01 1.65989339e+00 -3.23056504e-02 -6.22255325e-01
2.80253619e-01 -7.95509160e-01 4.05251265e-01 -3.30642283e-01
7.59081125e-01 3.25821400e-01 7.10331928e-03 1.54368132e-01
-5.64124703e-01 -2.88058728e-01 -1.04004729e+00 -2.56883681e-01
-8.67948234e-01 8.06825697e-01 3.42023462e-01 7.17768431e-01
6.51919305e-01 6.08876109e-01 1.17490017e+00 -4.08761799e-01
2.89694756e-01 -6.34410739e-01 -1.14783740e+00 -1.16390109e+00
2.51417428e-01 -9.45617437e-01 -2.45520756e-01 -1.31832823e-01]
|
[5.973454475402832, 1.4558496475219727]
|
4874766c-b8fd-4e33-a95b-00d5cf4609c8
|
leveraging-denoised-abstract-meaning
|
2307.02127
| null |
https://arxiv.org/abs/2307.02127v1
|
https://arxiv.org/pdf/2307.02127v1.pdf
|
Leveraging Denoised Abstract Meaning Representation for Grammatical Error Correction
|
Grammatical Error Correction (GEC) is the task of correcting errorful sentences into grammatically correct, semantically consistent, and coherent sentences. Popular GEC models either use large-scale synthetic corpora or use a large number of human-designed rules. The former is costly to train, while the latter requires quite a lot of human expertise. In recent years, AMR, a semantic representation framework, has been widely used by many natural language tasks due to its completeness and flexibility. A non-negligible concern is that AMRs of grammatically incorrect sentences may not be exactly reliable. In this paper, we propose the AMR-GEC, a seq-to-seq model that incorporates denoised AMR as additional knowledge. Specifically, We design a semantic aggregated GEC model and explore denoising methods to get AMRs more reliable. Experiments on the BEA-2019 shared task and the CoNLL-2014 shared task have shown that AMR-GEC performs comparably to a set of strong baselines with a large number of synthetic data. Compared with the T5 model with synthetic data, AMR-GEC can reduce the training time by 32\% while inference time is comparable. To the best of our knowledge, we are the first to incorporate AMR for grammatical error correction.
|
['Dongyan Zhao', 'Hejing Cao']
|
2023-07-05
| null | null | null | null |
['grammatical-error-correction']
|
['natural-language-processing']
|
[ 2.61655331e-01 3.03140253e-01 5.21765351e-01 -7.97297478e-01
-1.28904057e+00 -3.35074514e-01 3.87171298e-01 2.00961843e-01
-4.66557890e-01 8.37698579e-01 3.81817073e-01 -1.44405156e-01
4.07075793e-01 -7.36300707e-01 -9.83994782e-01 -2.35998169e-01
5.49599946e-01 4.15461868e-01 3.89075801e-02 -4.22994941e-01
2.53589123e-01 -3.88388515e-01 -1.50931501e+00 4.62159753e-01
1.44788384e+00 5.73007941e-01 4.02270079e-01 5.14242947e-01
-2.32854038e-01 6.97647333e-01 -8.22190940e-01 -8.88994634e-01
-2.18057573e-01 -6.54395521e-01 -9.64880407e-01 -4.17060167e-01
4.80693191e-01 1.86911523e-01 3.15346807e-01 1.41885626e+00
5.36735892e-01 1.00594364e-01 4.73407894e-01 -8.11579645e-01
-7.55817533e-01 8.93066764e-01 -1.96622774e-01 -4.45909798e-02
4.51117873e-01 -3.20988595e-02 9.89348650e-01 -1.14894867e+00
8.19932520e-01 1.69800997e+00 9.80277717e-01 1.05147517e+00
-1.05523980e+00 -5.66351116e-01 1.17726073e-01 1.30940735e-01
-1.27607441e+00 -4.72896725e-01 3.89685243e-01 3.19479816e-02
1.25555444e+00 3.60386014e-01 1.63809076e-01 1.31747591e+00
9.15566385e-02 6.19167864e-01 1.09631526e+00 -6.96417809e-01
3.45906913e-01 -2.06956849e-01 1.15101419e-01 6.01022124e-01
3.19887727e-01 -1.84473693e-01 -3.70219380e-01 4.24419567e-02
-9.60920677e-02 -3.79903734e-01 -3.70368212e-01 6.16421103e-01
-9.90129530e-01 8.18415046e-01 2.44083539e-01 3.14497024e-01
-1.68300763e-01 5.96288741e-01 6.19643748e-01 4.62631315e-01
9.36575532e-01 5.85672796e-01 -5.00346720e-01 -4.71500725e-01
-9.52016115e-01 5.35929680e-01 6.65411353e-01 1.08169317e+00
4.17314410e-01 2.72201836e-01 -1.95602655e-01 1.11383212e+00
4.85511899e-01 7.57837176e-01 4.14377272e-01 -1.25072885e+00
8.03578377e-01 3.25291008e-01 -1.73806977e-02 -8.90826404e-01
-8.11054409e-02 -3.88692051e-01 -1.00705719e+00 -9.60842595e-02
3.27273607e-01 1.17937237e-01 -9.24850106e-01 1.98336494e+00
-6.00560904e-02 1.80095565e-02 1.89913511e-01 7.23812163e-01
1.03366268e+00 6.92553818e-01 3.55384171e-01 -2.12238785e-02
1.12222993e+00 -8.99896324e-01 -1.03251398e+00 -5.03277957e-01
8.97619843e-01 -9.02423084e-01 1.29925334e+00 2.96700180e-01
-1.18882859e+00 -4.77283895e-01 -8.58135760e-01 -3.22438926e-01
-2.19549790e-01 1.25865594e-01 3.33110899e-01 6.44930899e-01
-1.07753968e+00 7.51125753e-01 -6.75863385e-01 -4.20904785e-01
2.23492742e-01 -3.11884463e-01 -2.75919288e-01 -3.81614327e-01
-1.52218330e+00 1.10210657e+00 4.25528139e-01 1.23687424e-01
-6.44160509e-01 -6.15553558e-01 -1.15583384e+00 -7.46919066e-02
1.50060117e-01 -8.74114990e-01 1.49116004e+00 -9.48452353e-01
-1.18075347e+00 9.48039234e-01 -5.74987888e-01 -6.35886192e-01
4.09029722e-01 -3.58707219e-01 -4.46530879e-01 -1.99424312e-01
4.35238421e-01 7.44483769e-01 5.35618901e-01 -1.19521439e+00
-3.99027914e-01 -2.38721639e-01 -2.07259163e-01 9.42321047e-02
2.50191659e-01 4.10132855e-01 -2.26392344e-01 -1.15909934e+00
2.83581018e-01 -8.24265778e-01 -1.40853822e-01 -3.57948422e-01
-2.64871180e-01 -4.74826783e-01 2.96136439e-01 -1.07776392e+00
1.38304973e+00 -1.94719124e+00 1.84746161e-01 -7.94770643e-02
-7.78570771e-02 3.91060442e-01 -2.92956710e-01 2.41352946e-01
-5.15174791e-02 6.02775872e-01 -6.98733926e-01 -9.85062122e-01
3.26167271e-02 3.64650816e-01 -3.57095897e-01 2.49977205e-02
3.48152459e-01 9.99085426e-01 -1.12110996e+00 -3.10567230e-01
-6.90741614e-02 3.39076370e-01 -6.52747512e-01 -1.31112840e-02
-2.81242490e-01 3.49422783e-01 -3.61455791e-02 4.97109443e-01
6.86660409e-01 -4.03593071e-02 4.11284268e-02 1.68228835e-01
2.23558649e-01 7.47385204e-01 -9.53856230e-01 1.96951783e+00
-6.85689211e-01 3.24207813e-01 -1.62557185e-01 -8.03686917e-01
9.91723239e-01 2.99611956e-01 -3.66028875e-01 -7.51509786e-01
4.13580332e-03 7.32707739e-01 -1.60321862e-01 -3.17235649e-01
7.73440063e-01 -3.21839958e-01 -3.96173656e-01 3.36205870e-01
1.24060564e-01 -5.55713654e-01 4.22624409e-01 4.51379746e-01
1.20453537e+00 1.41909704e-01 1.44499660e-01 -2.03597784e-01
4.56515044e-01 -7.13417381e-02 1.01468599e+00 7.65279830e-01
-1.05309412e-01 1.14583874e+00 1.78778812e-01 -1.61695316e-01
-1.00163424e+00 -9.30280149e-01 2.65812539e-02 6.01786137e-01
-9.81876552e-02 -7.84298956e-01 -1.16105187e+00 -8.05166483e-01
-2.22994819e-01 1.46406376e+00 -4.63011533e-01 -2.38536537e-01
-7.37167120e-01 -6.81421578e-01 8.06358755e-01 4.89477605e-01
6.46106660e-01 -1.24258995e+00 5.29776104e-02 3.88789713e-01
-8.88309956e-01 -1.15323389e+00 -5.82381010e-01 -1.98065802e-01
-8.53236973e-01 -9.45156693e-01 -3.51007849e-01 -7.23792374e-01
7.92717278e-01 1.55395372e-02 1.53243613e+00 4.28081036e-01
9.79767665e-02 5.39126582e-02 -9.43254948e-01 -5.43324947e-01
-9.93435919e-01 -2.39065006e-01 -7.41874948e-02 -4.64896947e-01
4.86481071e-01 -2.76692927e-01 -1.74181551e-01 -4.54455353e-02
-6.79379404e-01 2.19231725e-01 1.82203487e-01 8.42069685e-01
7.80017555e-01 -1.95176661e-01 8.69628429e-01 -1.35683918e+00
7.64686286e-01 -2.99435377e-01 3.34464875e-03 3.59888107e-01
-7.48303056e-01 2.82822307e-02 7.55668819e-01 1.28408343e-01
-1.25894880e+00 -4.10756737e-01 -7.36043453e-01 -1.73539612e-02
4.38985005e-02 5.82787573e-01 -3.05550575e-01 3.49004179e-01
7.52348721e-01 1.66660234e-01 -1.32718712e-01 -5.50139785e-01
3.97719711e-01 7.38151312e-01 6.45582259e-01 -8.04535866e-01
5.03868043e-01 5.00870422e-02 -3.43091249e-01 -3.12099785e-01
-1.31058180e+00 -1.09034561e-01 -1.04072683e-01 -9.11904648e-02
6.77336276e-01 -1.12383568e+00 8.21976662e-02 7.08725214e-01
-1.80381548e+00 -1.73968196e-01 -3.60850334e-01 2.67786890e-01
-4.73302126e-01 6.42323136e-01 -7.32435644e-01 -4.77781296e-01
-8.38107586e-01 -9.04735446e-01 1.22058451e+00 -1.24410264e-01
-5.61957657e-01 -8.00478518e-01 -1.09360978e-01 5.24328232e-01
5.92006564e-01 1.92566857e-01 7.43153453e-01 -3.36684376e-01
-2.57483035e-01 -1.47679955e-01 -6.38934597e-02 8.57245624e-01
-2.46508699e-02 -1.05795063e-01 -8.09577942e-01 -1.88669667e-01
9.76240709e-02 -2.42791817e-01 1.16902268e+00 2.76577294e-01
1.39151812e+00 -3.64783764e-01 8.70373920e-02 3.65263492e-01
1.26810598e+00 -9.10315812e-02 1.07077789e+00 1.40468389e-01
5.31294644e-01 4.25863475e-01 8.36754739e-01 8.84836763e-02
6.83019757e-01 5.55513680e-01 2.87799567e-01 2.16391355e-01
-5.28622210e-01 -4.51151431e-01 5.13345718e-01 1.53830802e+00
-1.27821043e-01 -5.46888053e-01 -8.26512098e-01 6.94439173e-01
-1.84213161e+00 -9.73889947e-01 -5.87115586e-01 2.01399636e+00
1.21849585e+00 1.30511016e-01 -6.23917997e-01 2.65458256e-01
8.41200590e-01 -6.27939552e-02 5.72513789e-02 -7.31040776e-01
-4.98661250e-01 4.64672476e-01 5.15858456e-02 6.75542593e-01
-9.22975600e-01 1.07269621e+00 5.83233929e+00 1.13348222e+00
-4.94369388e-01 6.07466340e-01 5.87286651e-01 1.04265638e-01
-6.98075235e-01 2.30385944e-01 -8.86277854e-01 8.96239519e-01
1.14416575e+00 -5.47464658e-03 3.43111515e-01 6.31317556e-01
4.29255247e-01 -1.05614915e-01 -9.09268796e-01 9.11376715e-01
3.36609244e-01 -1.19262779e+00 5.44331148e-02 -5.97274303e-01
9.13496494e-01 -4.86474000e-02 -3.61098468e-01 5.55697381e-01
4.72000659e-01 -1.18300724e+00 1.09584427e+00 6.71766222e-01
8.45593393e-01 -6.00316405e-01 1.07853293e+00 3.97582740e-01
-8.90653908e-01 2.58742750e-01 -5.79861701e-01 -2.03463495e-01
4.78076279e-01 1.00954950e+00 -1.26798645e-01 5.78295290e-01
9.71611440e-01 8.97526085e-01 -8.29522491e-01 6.71476007e-01
-9.27961409e-01 1.12372184e+00 -1.52685001e-01 -4.59788367e-02
8.93446803e-03 -1.01279788e-01 5.82768917e-01 1.31615233e+00
6.25721991e-01 -6.15322683e-03 -2.94766307e-01 1.10088742e+00
-3.28435063e-01 8.16978589e-02 -3.85987490e-01 1.73635587e-01
8.10168982e-01 8.87513161e-01 -1.70293063e-01 -4.15041327e-01
-3.86558086e-01 1.22912550e+00 5.25425255e-01 1.78698555e-01
-8.12193096e-01 -5.36263943e-01 5.04754663e-01 -2.27650348e-02
1.90261416e-02 4.97578904e-02 -6.06385171e-01 -1.28156674e+00
3.07396233e-01 -1.11365473e+00 1.65720150e-01 -1.00867438e+00
-1.56120431e+00 6.30995691e-01 -3.95570010e-01 -1.07616603e+00
-2.67104298e-01 -2.04683796e-01 -6.49470150e-01 1.08037567e+00
-1.61339796e+00 -1.06222379e+00 -2.17869282e-01 9.38699543e-02
8.08787942e-01 -9.28562209e-02 1.08225906e+00 3.97462606e-01
-4.99190420e-01 7.25518942e-01 9.89067256e-02 3.93161587e-02
8.78815532e-01 -1.49702728e+00 9.80465472e-01 1.28081512e+00
-1.49100482e-01 8.16676378e-01 9.45870697e-01 -9.31769669e-01
-7.17666626e-01 -1.68351388e+00 1.74123383e+00 -6.38524294e-01
1.76540062e-01 -3.26372981e-01 -1.22099233e+00 6.29818976e-01
1.61123142e-01 -2.10470669e-02 2.66624272e-01 7.55950585e-02
-4.45622444e-01 7.54296780e-03 -1.19973230e+00 4.36329901e-01
1.16641986e+00 -4.06630129e-01 -8.94090235e-01 3.23530346e-01
1.04856229e+00 -5.62206328e-01 -6.41244888e-01 5.40340006e-01
-7.80845806e-02 -7.17570543e-01 6.37960076e-01 -3.49708945e-01
8.18270266e-01 -5.47635019e-01 -3.02547306e-01 -1.75046706e+00
-1.00207843e-01 -3.62203062e-01 7.62066320e-02 1.65126729e+00
5.27102888e-01 -5.48534989e-01 2.47758448e-01 4.19852108e-01
-6.60617232e-01 -4.31868672e-01 -1.07275760e+00 -9.64715064e-01
4.59018558e-01 -7.43380964e-01 6.99230850e-01 7.51829326e-01
-2.69796997e-01 1.62346706e-01 -2.55949974e-01 -5.58864437e-02
6.05977774e-01 -2.18293667e-01 4.53315556e-01 -1.05543113e+00
-1.29307061e-01 -1.59616128e-01 -8.80651325e-02 -7.10602105e-01
3.91924202e-01 -1.31592560e+00 4.48365480e-01 -1.75176537e+00
2.41826445e-01 -3.94427240e-01 1.71237156e-01 5.21322608e-01
-6.24611437e-01 1.16480716e-01 8.02668631e-02 -3.74542847e-02
-7.18144119e-01 8.46003234e-01 1.00278413e+00 -1.48058251e-01
2.77306885e-01 -3.32498521e-01 -7.94400454e-01 7.45829463e-01
9.38806951e-01 -8.79872203e-01 -6.46844804e-02 -8.61622334e-01
6.97638750e-01 -4.11238045e-01 3.46138924e-01 -7.45168805e-01
-2.41604671e-02 2.07840234e-01 -1.40803531e-01 -2.84259170e-01
-1.00353584e-01 -2.86230028e-01 3.15262944e-01 2.86929756e-01
-2.98585802e-01 9.87652168e-02 -1.95245980e-03 6.20024502e-01
-5.98751247e-01 -6.28268719e-01 8.42291355e-01 -3.56075615e-01
-6.41856313e-01 -2.18283936e-01 -4.03273612e-01 5.75647056e-01
5.79839945e-01 4.58278395e-02 -5.09003937e-01 -3.23247612e-01
-3.89201283e-01 2.30467230e-01 5.12959480e-01 5.39872825e-01
7.46043324e-01 -1.36253893e+00 -1.16448832e+00 -1.51095092e-01
2.36681595e-01 2.70296007e-01 2.96772689e-01 5.81997454e-01
-5.70334136e-01 1.60301000e-01 4.64743674e-01 -3.30994636e-01
-1.22374487e+00 -1.19453659e-02 3.51352662e-01 -3.09942782e-01
-6.47004545e-01 9.02886629e-01 -2.97900438e-01 -1.05194747e+00
9.53337178e-03 -3.08392853e-01 6.62963688e-02 -3.09602559e-01
5.20740986e-01 3.66159290e-01 3.76866221e-01 -6.10674202e-01
-3.53254974e-01 4.34922993e-01 -2.59789471e-02 6.90922365e-02
1.31700480e+00 -3.74307930e-01 -4.40998495e-01 1.85706183e-01
8.45649898e-01 -7.39069209e-02 -6.96450591e-01 -2.50314683e-01
1.89353734e-01 -3.95902365e-01 -1.07824475e-01 -1.17029929e+00
-9.09220397e-01 9.48343277e-01 2.61286795e-01 -2.89826035e-01
1.11462986e+00 -2.63074227e-02 9.48009491e-01 2.38257542e-01
5.40957987e-01 -1.35051095e+00 -1.08169101e-01 9.31109846e-01
1.31420088e+00 -1.35399556e+00 -3.50517958e-01 -7.10258424e-01
-7.42352188e-01 8.88943136e-01 6.39396787e-01 -5.82098588e-02
3.41117561e-01 7.81313479e-02 8.54036659e-02 -5.61970398e-02
-7.99039245e-01 -3.40682790e-02 6.30404353e-02 3.46898675e-01
6.46931350e-01 2.04888150e-01 -9.24720585e-01 1.00430405e+00
-3.98327053e-01 -1.22315213e-01 6.92331076e-01 6.25507593e-01
-4.10699069e-01 -1.34271920e+00 -5.97517118e-02 3.53806168e-01
-5.89230180e-01 -4.73800302e-01 -1.49756864e-01 4.23508972e-01
2.53690302e-01 1.35571563e+00 -1.53932616e-01 -1.33701280e-01
4.58834887e-01 2.48300329e-01 4.09279108e-01 -9.35558379e-01
-7.32361615e-01 -4.36901569e-01 5.22306502e-01 -6.21305406e-01
-4.51063603e-01 -7.19159782e-01 -1.46906269e+00 -4.62867141e-01
-3.37753683e-01 8.04421827e-02 7.27882028e-01 9.94190693e-01
5.15712738e-01 6.57165706e-01 2.88394928e-01 -2.08162501e-01
-7.01504290e-01 -1.44088995e+00 -4.09896493e-01 8.24259877e-01
-2.49661133e-01 -3.11704457e-01 -6.17181122e-01 1.91003457e-02]
|
[11.081777572631836, 10.705124855041504]
|
52d49634-0909-40e2-b72e-db475d5794f1
|
xtarnet-learning-to-extract-task-adaptive
|
2003.08561
| null |
https://arxiv.org/abs/2003.08561v2
|
https://arxiv.org/pdf/2003.08561v2.pdf
|
XtarNet: Learning to Extract Task-Adaptive Representation for Incremental Few-Shot Learning
|
Learning novel concepts while preserving prior knowledge is a long-standing challenge in machine learning. The challenge gets greater when a novel task is given with only a few labeled examples, a problem known as incremental few-shot learning. We propose XtarNet, which learns to extract task-adaptive representation (TAR) for facilitating incremental few-shot learning. The method utilizes a backbone network pretrained on a set of base categories while also employing additional modules that are meta-trained across episodes. Given a new task, the novel feature extracted from the meta-trained modules is mixed with the base feature obtained from the pretrained model. The process of combining two different features provides TAR and is also controlled by meta-trained modules. The TAR contains effective information for classifying both novel and base categories. The base and novel classifiers quickly adapt to a given task by utilizing the TAR. Experiments on standard image datasets indicate that XtarNet achieves state-of-the-art incremental few-shot learning performance. The concept of TAR can also be used in conjunction with existing incremental few-shot learning methods; extensive simulation results in fact show that applying TAR enhances the known methods significantly.
|
['Do-Yeon Kim', 'Jun Seo', 'Sung Whan Yoon', 'Jaekyun Moon']
|
2020-03-19
| null |
https://proceedings.icml.cc/static/paper_files/icml/2020/6928-Paper.pdf
|
https://proceedings.icml.cc/static/paper_files/icml/2020/6928-Paper.pdf
|
icml-2020-1
|
['novel-concepts']
|
['reasoning']
|
[ 4.73701119e-01 1.07385777e-01 -3.18092316e-01 -3.40927422e-01
-8.80928636e-01 -6.47243187e-02 5.82361698e-01 1.33244425e-01
-5.95143557e-01 6.58174217e-01 9.02861208e-02 5.21072984e-01
-7.78708160e-02 -7.27894127e-01 -5.92880487e-01 -7.41781056e-01
-2.13577986e-01 3.12719703e-01 6.61076486e-01 -2.14016765e-01
3.29441518e-01 8.31327811e-02 -2.03646898e+00 5.63371658e-01
5.89962125e-01 1.09985507e+00 5.37979066e-01 6.23677015e-01
-3.31353128e-01 9.30312335e-01 -3.53264421e-01 -2.42947713e-01
4.46917385e-01 -5.19954443e-01 -7.92108357e-01 1.66198209e-01
3.11454833e-01 -2.26813972e-01 -2.18259528e-01 7.28716135e-01
5.55400670e-01 8.39506805e-01 5.34352660e-01 -1.13998842e+00
-5.63481569e-01 6.15790904e-01 -4.62243736e-01 5.37619054e-01
9.73737761e-02 2.06480175e-01 7.73816228e-01 -1.37182438e+00
8.81160736e-01 9.03531849e-01 7.30364203e-01 8.98181438e-01
-9.20801401e-01 -4.51762110e-01 3.10000360e-01 7.12601483e-01
-1.18177497e+00 -5.34286261e-01 6.51754200e-01 -2.66550094e-01
1.07253623e+00 -2.15117887e-01 5.98027945e-01 7.79887736e-01
2.41579980e-01 9.24181223e-01 8.54406655e-01 -6.87207699e-01
7.22530007e-01 3.69599044e-01 5.24903893e-01 6.64362788e-01
1.01123929e-01 -1.28339410e-01 -8.76159966e-01 -4.11508754e-02
1.54690519e-01 7.63160646e-01 -9.16087627e-02 -8.19592416e-01
-9.13091838e-01 8.90041173e-01 5.31990886e-01 5.88227391e-01
-4.51660603e-01 8.08300301e-02 7.55609155e-01 6.60343528e-01
5.30725002e-01 4.55942899e-01 -6.00232482e-01 -8.31001028e-02
-8.14542294e-01 -4.84850407e-02 4.69855666e-01 1.06791747e+00
1.31770742e+00 1.44918799e-01 -3.67134631e-01 1.11325347e+00
-2.49650717e-01 3.60581651e-02 1.11752319e+00 -7.04696357e-01
2.67577469e-01 6.54894114e-01 -1.78832740e-01 -3.84190530e-01
-1.38484284e-01 -5.92737317e-01 -4.70328361e-01 5.71188070e-02
-2.13369310e-01 -1.43076420e-01 -1.25224864e+00 1.70102370e+00
4.53418314e-01 6.71410024e-01 2.26299688e-01 2.63366103e-01
6.42621994e-01 7.39805043e-01 2.35282302e-01 -5.90595424e-01
1.00686038e+00 -1.22229636e+00 -5.16828537e-01 -4.90268677e-01
6.13837600e-01 -2.41162613e-01 7.95568347e-01 4.00728546e-02
-9.38976526e-01 -8.86085570e-01 -1.16276824e+00 2.05960914e-01
-7.52851248e-01 -4.58146483e-01 4.00557488e-01 3.00130963e-01
-8.63282204e-01 7.21775413e-01 -4.93894756e-01 -6.46304429e-01
7.97116220e-01 1.60792768e-01 -2.69704133e-01 -5.91545403e-01
-1.01852047e+00 1.05294168e+00 6.88715160e-01 -4.07869458e-01
-1.24149072e+00 -9.09731686e-01 -1.05735421e+00 1.67819589e-01
6.38066649e-01 -7.38665998e-01 1.56068623e+00 -1.04207003e+00
-1.35518503e+00 5.29537320e-01 -2.20420569e-01 -5.91295123e-01
2.05442414e-01 -1.20025605e-01 -3.89603466e-01 3.41513842e-01
3.69595677e-01 6.39051318e-01 1.37427270e+00 -1.06517851e+00
-1.04960656e+00 -3.93387616e-01 -8.14609602e-02 4.43946153e-01
-8.45817864e-01 -3.22315156e-01 -2.89652765e-01 -4.30018485e-01
-2.33422801e-01 -6.28422141e-01 -3.75475347e-01 -6.62543401e-02
3.53253663e-01 -3.24682921e-01 1.01922512e+00 -2.02165648e-01
1.04230428e+00 -2.18737912e+00 1.46592915e-01 -1.71244189e-01
2.04123169e-01 5.57297230e-01 -3.64820689e-01 4.76522356e-01
-1.58590525e-01 -5.07961750e-01 -2.81759620e-01 -2.87210345e-01
-3.31317365e-01 9.21698287e-02 -3.05106819e-01 4.16834727e-02
1.22109689e-01 1.00673902e+00 -1.37555802e+00 -3.61774623e-01
3.08059305e-01 9.49579328e-02 -2.17743382e-01 3.17089230e-01
-5.90677373e-02 -9.56311673e-02 -2.77833223e-01 5.73408902e-01
5.00107110e-01 -2.29303077e-01 -1.47569880e-01 9.77547616e-02
1.04885563e-01 -3.00138324e-01 -1.08050811e+00 2.05261850e+00
-5.29595971e-01 4.59902793e-01 -5.53547204e-01 -1.33797240e+00
8.74471128e-01 3.07246447e-01 3.59631002e-01 -5.06274641e-01
9.17452127e-02 -4.53439690e-02 -1.42502919e-01 -5.89471102e-01
4.84398454e-01 -4.55902368e-01 9.23379418e-03 6.40136063e-01
9.75226820e-01 7.61456788e-02 4.07926828e-01 4.21491206e-01
1.35567033e+00 -1.44833727e-02 8.39584827e-01 1.23356335e-01
2.03157619e-01 1.46673426e-01 7.02028096e-01 1.04547930e+00
-4.60376173e-01 3.09622675e-01 -1.92840144e-01 -6.02319956e-01
-9.18898404e-01 -1.18155682e+00 2.32505966e-02 1.65104532e+00
-4.23546806e-02 -4.65768367e-01 -4.14987355e-01 -1.06578743e+00
4.37814817e-02 9.12542641e-01 -1.06031680e+00 -7.99493551e-01
-1.68460533e-01 -5.71036398e-01 -2.24126995e-01 7.37121582e-01
5.98447502e-01 -1.12546289e+00 -1.07976949e+00 5.95733047e-01
1.58865988e-01 -7.60524809e-01 -3.24513525e-01 5.68710089e-01
-1.10255241e+00 -1.11397898e+00 -8.89613867e-01 -8.99047077e-01
7.50741243e-01 9.85318899e-01 8.00315380e-01 -2.65544623e-01
-6.17621541e-01 8.03296268e-01 -7.89116204e-01 -6.16526186e-01
-1.70197085e-01 -3.11016068e-02 1.55711725e-01 2.68271714e-01
5.72967529e-01 -5.82659781e-01 -3.48165542e-01 3.70913334e-02
-8.70732605e-01 -1.08654857e-01 7.76572227e-01 1.14573956e+00
4.32224393e-01 5.40505461e-02 1.16834605e+00 -1.17906117e+00
2.10666344e-01 -8.56193721e-01 5.05270846e-02 3.74380618e-01
-6.31757438e-01 -3.02152354e-02 6.91927254e-01 -8.83988798e-01
-1.45425534e+00 1.27988666e-01 4.25129414e-01 -7.02839494e-01
5.81003763e-02 5.55108190e-01 1.88404396e-01 -5.65056689e-02
8.62300217e-01 5.28337479e-01 -1.13550223e-01 -4.29657370e-01
7.49858677e-01 6.53013945e-01 3.76408637e-01 -1.47616953e-01
9.38696325e-01 5.61587632e-01 -3.40505570e-01 -8.32206249e-01
-1.26265395e+00 -9.53824937e-01 -9.85658467e-01 -3.13224465e-01
3.51162821e-01 -9.66414034e-01 1.59792468e-01 3.74308169e-01
-6.22973084e-01 -2.62109041e-01 -1.06480217e+00 3.14277291e-01
-6.37278676e-01 6.05172329e-02 -2.38529578e-01 -6.80003047e-01
-4.81443614e-01 -5.27482927e-01 6.84352100e-01 4.83248174e-01
-4.13771793e-02 -9.96371984e-01 3.72801542e-01 5.13353944e-02
6.04720592e-01 -2.79409811e-02 7.97788441e-01 -1.00515604e+00
-2.71506876e-01 -4.51903760e-01 1.39389753e-01 3.03511858e-01
3.62472266e-01 -5.44996917e-01 -1.31221974e+00 -6.62202477e-01
2.24213004e-01 -7.80822039e-01 1.39651930e+00 1.65093869e-01
9.03878272e-01 -1.11048169e-01 -4.58531857e-01 2.93133169e-01
1.63348246e+00 3.07055891e-01 2.86638439e-01 2.48322815e-01
4.24185753e-01 3.00766945e-01 8.27461183e-01 5.47979891e-01
1.89422607e-01 3.27063322e-01 4.42694388e-02 3.76744151e-01
-2.96591431e-01 -7.22861588e-02 3.68870676e-01 8.65584552e-01
2.40604207e-02 5.25356591e-01 -7.29566216e-01 9.25438106e-01
-2.09290123e+00 -1.47980404e+00 8.81837249e-01 2.06206942e+00
8.97180974e-01 1.55711353e-01 -4.73508425e-02 8.09073374e-02
8.36955428e-01 6.89115822e-02 -1.03428209e+00 -3.09185594e-01
3.30907196e-01 4.99567419e-01 -1.56089412e-02 -5.36030065e-03
-1.00861907e+00 9.01476860e-01 5.94146585e+00 1.00685823e+00
-8.79048109e-01 5.45741320e-01 3.55146766e-01 -4.29657191e-01
1.24260329e-01 2.24325731e-02 -8.94332290e-01 1.25869140e-01
1.04765368e+00 -9.37957525e-01 1.70887947e-01 1.35281122e+00
-4.04293239e-01 -2.05167621e-01 -1.31365943e+00 8.99128616e-01
7.14402139e-01 -1.53022921e+00 2.39707783e-01 -4.29110616e-01
1.23551297e+00 1.92193061e-01 1.39434993e-01 1.08754206e+00
3.39327633e-01 -3.27975541e-01 3.22019219e-01 6.53473496e-01
8.17617476e-01 -7.31906474e-01 7.12501228e-01 4.62481737e-01
-1.50406277e+00 -7.22268701e-01 -7.40128160e-01 -4.75548618e-02
-1.59121409e-01 3.60768855e-01 -1.05715942e+00 5.31959653e-01
6.61806405e-01 1.11063385e+00 -7.65528321e-01 1.37070239e+00
-1.98633373e-02 4.11630929e-01 1.51935622e-01 3.07748914e-02
2.42910519e-01 4.30612653e-01 5.85503995e-01 1.07049263e+00
2.29871213e-01 1.25328720e-01 2.27431431e-01 4.23752457e-01
-2.93148428e-01 5.30538373e-02 -9.43919182e-01 7.84185976e-02
6.58220291e-01 1.41630006e+00 -6.46671832e-01 -9.73895967e-01
-6.27238274e-01 1.15761030e+00 7.33148217e-01 2.90492386e-01
-2.81087995e-01 -6.23971820e-01 3.00147653e-01 -1.63048625e-01
6.79849923e-01 2.22968325e-01 9.72307324e-02 -1.25655651e+00
-2.28379622e-01 -5.12393832e-01 8.10223222e-01 -7.61469781e-01
-1.49275458e+00 4.16093290e-01 2.67157942e-01 -1.57198596e+00
-3.77087325e-01 -2.84632891e-01 -9.17965710e-01 4.07938391e-01
-1.65367985e+00 -1.09099448e+00 -5.22664011e-01 7.56684780e-01
1.41624725e+00 -6.83962345e-01 1.03692651e+00 -1.05286874e-01
-4.95099008e-01 5.15007973e-01 3.17196906e-01 -1.82134315e-01
8.18756878e-01 -1.20414531e+00 2.45422333e-01 8.63949478e-01
1.64090484e-01 5.75069070e-01 4.33101475e-01 -5.98374605e-01
-1.28565359e+00 -1.59125459e+00 5.74931026e-01 -3.41481358e-01
5.15552819e-01 -1.89643234e-01 -1.08574772e+00 8.82620871e-01
2.15576351e-01 2.93721348e-01 9.26090240e-01 -1.09644104e-02
-4.77391720e-01 -3.37995887e-01 -1.20242059e+00 4.49230373e-01
1.02116501e+00 -5.68716049e-01 -1.25639117e+00 2.05653965e-01
9.16353703e-01 9.71405134e-02 -4.50863451e-01 2.14903161e-01
2.96586603e-01 -6.19801879e-01 6.81544065e-01 -8.73215616e-01
1.81414500e-01 2.80114692e-02 -2.05330372e-01 -1.82458889e+00
-6.24176323e-01 -3.67026299e-01 -6.26311064e-01 1.00308669e+00
1.21056117e-01 -5.31212151e-01 6.72603905e-01 4.54064637e-01
-2.40183622e-01 -6.32838011e-01 -9.12314773e-01 -1.19414937e+00
-1.84505612e-01 -2.67516941e-01 2.59481609e-01 1.03321946e+00
4.09463048e-01 6.70599163e-01 -3.98935139e-01 -4.30456072e-01
8.94750237e-01 1.09744124e-01 7.73255467e-01 -1.32099175e+00
-1.19431131e-01 2.89036464e-02 -6.12893999e-01 -4.87505734e-01
9.59640667e-02 -1.18235981e+00 9.19363648e-02 -1.45614529e+00
7.90456533e-01 -8.17083567e-02 -8.47055912e-01 8.52247059e-01
-4.64628875e-01 1.25454754e-01 3.92584801e-01 1.84277639e-01
-1.18066990e+00 8.62379670e-01 8.98020089e-01 -3.53618711e-01
-4.18413699e-01 2.34071258e-02 -7.47540712e-01 7.47237206e-01
7.50984788e-01 -6.43430114e-01 -7.47966886e-01 1.44213550e-02
-3.14610094e-01 -3.72922450e-01 -1.94229428e-02 -1.35374856e+00
6.24367774e-01 -6.97292015e-02 6.49746239e-01 -4.10529017e-01
4.86259162e-01 -7.60927081e-01 -3.60115618e-01 6.95739925e-01
-3.83836776e-01 -3.42492461e-01 1.03026025e-01 9.89773989e-01
-1.07245371e-01 -5.26020467e-01 1.07571113e+00 -5.83667338e-01
-1.59740889e+00 5.64375103e-01 -2.38029718e-01 1.13620199e-01
1.52370858e+00 -4.38143611e-01 -2.47976035e-01 -8.09823722e-02
-1.01119649e+00 3.14570487e-01 1.38386086e-01 4.92152482e-01
1.11064208e+00 -1.40197206e+00 -5.74856579e-01 2.44175106e-01
7.90632844e-01 -2.83949763e-01 5.31751037e-01 4.36985821e-01
3.80515069e-01 -1.33518782e-02 -5.30798256e-01 -4.48660672e-01
-1.09006500e+00 1.03222537e+00 6.35747537e-02 -1.71983317e-01
-7.86500096e-01 9.05785978e-01 -2.61634681e-02 -1.72764793e-01
2.17063144e-01 2.29622647e-01 -3.85408729e-01 5.66251397e-01
1.07677019e+00 5.80108881e-01 -3.69535908e-02 -3.02469969e-01
-1.33268401e-01 4.42879021e-01 -6.99304223e-01 -1.88412536e-02
1.73330235e+00 -2.24890739e-01 3.04148138e-01 1.24329400e+00
1.26609182e+00 -7.08822727e-01 -1.30190921e+00 -1.04635060e+00
6.86099380e-02 -5.91493487e-01 -8.59210193e-02 -7.79845595e-01
-7.50057280e-01 7.95593381e-01 8.59381676e-01 -2.23081470e-01
1.07116055e+00 -4.78620492e-02 6.97624326e-01 9.39078271e-01
6.98867917e-01 -1.60568774e+00 7.68796146e-01 5.56764424e-01
6.76226795e-01 -1.39146829e+00 -7.62042403e-02 5.69322444e-02
-7.34979272e-01 1.10300434e+00 8.70332658e-01 -1.99504822e-01
8.00921917e-01 -1.67345300e-01 -2.10071191e-01 -7.45670646e-02
-1.26029003e+00 -4.96231765e-01 1.66052595e-01 7.59847999e-01
-1.85893372e-01 -2.35269994e-01 9.26941112e-02 5.70035219e-01
3.52636963e-01 2.91803062e-01 5.36380172e-01 1.44097924e+00
-1.14452386e+00 -7.97670424e-01 1.89259537e-02 8.96170676e-01
2.06055064e-02 -4.98766117e-02 -4.24794853e-02 4.48953718e-01
3.33197176e-01 7.77499378e-01 3.78536135e-02 -4.75186169e-01
2.63855428e-01 6.50944114e-01 4.16340709e-01 -1.42913687e+00
-4.43408728e-01 -3.75091970e-01 -2.74736017e-01 -5.17444193e-01
-3.64861667e-01 -6.55494153e-01 -1.05265057e+00 1.60995334e-01
-4.04429018e-01 2.74266869e-01 3.50165784e-01 1.15966058e+00
3.53713930e-01 6.02160752e-01 9.54637647e-01 -9.20723855e-01
-7.30077505e-01 -1.12607384e+00 -6.82531536e-01 4.89371061e-01
4.43544209e-01 -8.86289060e-01 -3.99343729e-01 3.01306158e-01]
|
[9.98175048828125, 3.1111693382263184]
|
21f3cedc-d4f1-4bf1-a9b9-4e540ba1f5bb
|
multi-granularity-chinese-word-embedding
| null | null |
https://aclanthology.org/D16-1100
|
https://aclanthology.org/D16-1100.pdf
|
Multi-Granularity Chinese Word Embedding
| null |
['Rongchao Yin', 'Peng Li', 'Rui Li', 'Bin Wang', 'Quan Wang']
|
2016-11-01
| null | null | null |
emnlp-2016-11
|
['learning-word-embeddings']
|
['methodology']
|
[-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]
|
[-7.3876447677612305, 3.8253791332244873]
|
f721465e-3d04-4981-8fbe-2d6898cc5422
|
sense-a-shared-encoder-network-for-scene-flow-1
|
1910.12361
| null |
https://arxiv.org/abs/1910.12361v1
|
https://arxiv.org/pdf/1910.12361v1.pdf
|
SENSE: a Shared Encoder Network for Scene-flow Estimation
|
We introduce a compact network for holistic scene flow estimation, called SENSE, which shares common encoder features among four closely-related tasks: optical flow estimation, disparity estimation from stereo, occlusion estimation, and semantic segmentation. Our key insight is that sharing features makes the network more compact, induces better feature representations, and can better exploit interactions among these tasks to handle partially labeled data. With a shared encoder, we can flexibly add decoders for different tasks during training. This modular design leads to a compact and efficient model at inference time. Exploiting the interactions among these tasks allows us to introduce distillation and self-supervised losses in addition to supervised losses, which can better handle partially labeled real-world data. SENSE achieves state-of-the-art results on several optical flow benchmarks and runs as fast as networks specifically designed for optical flow. It also compares favorably against the state of the art on stereo and scene flow, while consuming much less memory.
|
['Erik Learned-Miller', 'Deqing Sun', 'Zhaoyang Lv', 'Huaizu Jiang', 'Jan Kautz', 'Varun Jampani']
|
2019-10-27
|
sense-a-shared-encoder-network-for-scene-flow
|
http://openaccess.thecvf.com/content_ICCV_2019/html/Jiang_SENSE_A_Shared_Encoder_Network_for_Scene-Flow_Estimation_ICCV_2019_paper.html
|
http://openaccess.thecvf.com/content_ICCV_2019/papers/Jiang_SENSE_A_Shared_Encoder_Network_for_Scene-Flow_Estimation_ICCV_2019_paper.pdf
|
iccv-2019-10
|
['occlusion-estimation', 'scene-flow-estimation']
|
['computer-vision', 'computer-vision']
|
[-7.08222166e-02 -8.38075131e-02 -6.49966598e-01 -5.49063981e-01
-4.91850615e-01 -5.19061446e-01 3.84719610e-01 -2.60888368e-01
-4.67689306e-01 8.65995049e-01 6.48522615e-01 -3.52698378e-02
1.61447570e-01 -7.78371334e-01 -7.47323632e-01 -3.91696066e-01
-1.04465999e-01 3.53263825e-01 2.87501872e-01 1.28497601e-01
1.90127552e-01 5.29831648e-01 -1.66593421e+00 3.46475124e-01
7.89646447e-01 8.96006823e-01 2.46695787e-01 8.71626318e-01
-9.71734077e-02 1.56019557e+00 -4.82026525e-02 -5.04230082e-01
4.01920825e-01 -1.26537442e-01 -1.19699323e+00 8.50302279e-02
1.18804407e+00 -9.18275833e-01 -8.29808235e-01 7.50611305e-01
4.15036410e-01 4.68410522e-01 3.67194921e-01 -1.17053747e+00
-2.91987240e-01 3.40957969e-01 -5.13098180e-01 3.77271831e-01
2.65547007e-01 4.19393897e-01 1.23730457e+00 -5.59407294e-01
8.36312473e-01 1.72321630e+00 6.37387514e-01 6.25665843e-01
-1.41509604e+00 -4.82433408e-01 2.46965304e-01 2.05630764e-01
-9.22122300e-01 -7.61459112e-01 4.66209888e-01 -5.26864946e-01
1.13048363e+00 -6.54479340e-02 5.98342657e-01 8.14202368e-01
8.43744799e-02 1.21315861e+00 6.11092389e-01 8.01663920e-02
-1.41944483e-01 -6.68743029e-02 3.61023732e-02 1.06132746e+00
1.21634886e-01 1.84133321e-01 -5.99195361e-01 2.44093150e-01
9.69506681e-01 -2.42451839e-02 -5.62933207e-01 -4.59950536e-01
-1.04133904e+00 7.90164948e-01 6.97437048e-01 -1.89681381e-01
6.76176473e-02 6.40736163e-01 6.79580688e-01 2.59241730e-01
5.65286279e-01 4.41793442e-01 -5.07016182e-01 -3.07044268e-01
-9.73737657e-01 2.95686960e-01 8.65775168e-01 8.92966151e-01
1.38964593e+00 1.87301919e-01 -2.93509364e-01 7.55778909e-01
3.59713346e-01 4.58573520e-01 1.33395851e-01 -1.76995516e+00
6.52842522e-01 2.53937870e-01 -6.27252609e-02 -7.42274940e-01
-4.18563992e-01 -3.96645904e-01 -6.88826799e-01 3.94362450e-01
5.77334166e-01 -2.73604482e-01 -9.24125075e-01 2.02689338e+00
-6.81504188e-03 5.82520843e-01 -1.20776929e-01 9.20638502e-01
7.31895566e-01 6.19013190e-01 -1.38184026e-01 2.24332541e-01
9.22136307e-01 -1.44902790e+00 -3.30716491e-01 -6.09790862e-01
7.49445796e-01 -7.02289402e-01 9.30141628e-01 1.05269954e-01
-1.27525711e+00 -6.53582513e-01 -9.08184290e-01 -8.87549341e-01
4.52393554e-02 -2.15489835e-01 1.20235693e+00 3.04116726e-01
-1.43792772e+00 8.19557965e-01 -1.01251853e+00 -1.92737579e-01
9.15345669e-01 3.94821644e-01 -2.88784921e-01 -4.64723527e-01
-7.72128880e-01 6.83155656e-01 1.63164929e-01 -2.43673190e-01
-8.62981617e-01 -1.15879571e+00 -1.48787296e+00 8.17521811e-02
1.64898545e-01 -1.32347775e+00 1.21970952e+00 -8.77788007e-01
-1.62862885e+00 1.01535583e+00 -6.55923426e-01 -5.64760029e-01
5.87069035e-01 -4.00297880e-01 3.79663594e-02 3.25070679e-01
3.30869555e-01 1.37793469e+00 7.14987099e-01 -7.30729163e-01
-8.13300431e-01 -1.21881388e-01 3.56018633e-01 1.30490765e-01
-2.61301380e-02 -3.42690855e-01 -6.10439777e-01 -4.55201656e-01
-3.14890683e-01 -8.02315176e-01 -2.63988376e-01 6.74783707e-01
-3.67247552e-01 1.41586205e-02 7.62094319e-01 -2.95525819e-01
8.97675693e-01 -2.15861416e+00 1.20184757e-01 -2.57530928e-01
5.18041968e-01 2.51058996e-01 -2.71031201e-01 -7.95638561e-02
1.44650131e-01 -1.34341300e-01 -2.54587829e-01 -7.54060745e-01
-1.36005387e-01 5.44861257e-01 -3.48385751e-01 4.15961295e-01
3.83029312e-01 1.09532595e+00 -1.07522047e+00 -5.64824343e-01
6.30509377e-01 4.13770199e-01 -1.21859014e+00 1.98693424e-01
-2.40229279e-01 5.89636147e-01 -2.79091626e-01 4.04738069e-01
7.34305441e-01 -6.54886842e-01 -2.07888871e-01 -1.64178878e-01
8.49700253e-03 6.99835718e-01 -1.18102884e+00 2.09864378e+00
-7.19734788e-01 1.21025562e+00 9.89154875e-02 -8.64254713e-01
5.18695712e-01 2.29631960e-02 5.03425717e-01 -5.77474773e-01
7.93011859e-03 1.34956673e-01 -2.47821972e-01 -4.37342167e-01
5.47137141e-01 9.69682187e-02 4.81359243e-01 5.35584569e-01
5.24865270e-01 -2.19478905e-01 5.23568332e-01 2.11410284e-01
1.04871523e+00 3.30958605e-01 1.35592306e-02 -3.30900550e-01
5.11603773e-01 -2.33843446e-01 7.34401941e-01 6.63262248e-01
-4.31092113e-01 6.61007524e-01 4.39937770e-01 -7.93013990e-01
-7.67861485e-01 -1.21809125e+00 -9.91833210e-02 1.07572138e+00
4.12212640e-01 -3.85416925e-01 -3.59968245e-01 -5.55533946e-01
3.37772518e-01 1.83480456e-01 -4.92807746e-01 4.94289631e-03
-7.86209583e-01 -4.55138087e-01 3.92403454e-01 7.37058640e-01
7.32745707e-01 -9.73500252e-01 -6.65683389e-01 2.99669504e-01
-3.37125629e-01 -1.47458768e+00 -6.15515590e-01 3.37978303e-02
-1.07288039e+00 -1.24994683e+00 -6.95637822e-01 -7.68153071e-01
2.60053366e-01 3.38161170e-01 1.61412120e+00 1.13082463e-02
-4.36361998e-01 2.22905681e-01 2.04870567e-01 4.24739607e-02
-1.45602554e-01 2.88912952e-01 -4.65269566e-01 -3.96849364e-02
1.81068763e-01 -5.97764075e-01 -9.21075940e-01 6.54240325e-02
-7.63113439e-01 1.07399754e-01 9.19505209e-02 8.34600270e-01
2.97311544e-01 -5.03271699e-01 1.24388285e-01 -9.62181747e-01
5.70138805e-02 -1.41548067e-01 -6.54927790e-01 -7.74157792e-02
-2.22062930e-01 5.87940395e-01 6.79096818e-01 7.93041438e-02
-1.28567898e+00 -1.44626172e-02 -1.85390368e-01 -5.68914413e-01
-1.11237429e-01 -1.40188590e-01 1.74934432e-01 -3.05440366e-01
5.22439659e-01 -2.01578289e-01 2.07617447e-01 -3.75018924e-01
5.50964117e-01 2.53092498e-01 6.64223135e-01 -5.68577886e-01
4.57311839e-01 9.97422874e-01 3.02477896e-01 -6.37437820e-01
-1.18940854e+00 -5.27790904e-01 -4.58923966e-01 8.54866728e-02
9.12093937e-01 -1.25856817e+00 -9.17352855e-01 7.48387456e-01
-1.27037835e+00 -6.61565185e-01 -5.42051017e-01 5.40828645e-01
-8.65666866e-01 2.46066883e-01 -1.11681187e+00 -2.74855137e-01
-1.32485211e-01 -1.41232693e+00 1.16568851e+00 2.52660275e-01
-1.61759228e-01 -1.47212529e+00 5.00638336e-02 3.58595580e-01
3.71768564e-01 1.32921383e-01 6.27838671e-01 6.40234947e-02
-1.12381089e+00 3.99793714e-01 -6.09848082e-01 6.10688925e-01
1.71697527e-01 -2.63458360e-02 -1.30577469e+00 -3.36143047e-01
-3.87734473e-01 -7.37428010e-01 1.62743032e+00 6.96529388e-01
1.29491186e+00 -3.32390331e-03 -3.01531255e-01 1.44371915e+00
1.43930006e+00 -3.07870030e-01 7.41325438e-01 2.18326952e-02
1.01933551e+00 5.83367527e-01 1.81966513e-01 4.62427884e-01
6.93198383e-01 4.16164070e-01 4.78567094e-01 -2.98638135e-01
-5.16863048e-01 -1.65075898e-01 2.58248717e-01 4.82792437e-01
9.31737274e-02 -3.06705654e-01 -6.10615253e-01 5.67530334e-01
-1.92075992e+00 -1.11102593e+00 4.81356978e-02 1.89258194e+00
7.59705544e-01 8.18555802e-02 -1.83550902e-02 -1.96492761e-01
5.01262069e-01 8.02441061e-01 -6.37759984e-01 -5.00718534e-01
-1.54842332e-01 5.37743270e-01 8.03742707e-01 1.01523888e+00
-1.31669390e+00 1.23563504e+00 7.07576704e+00 5.17916501e-01
-1.22557688e+00 -6.03093766e-03 9.87742543e-01 -2.75529921e-01
-3.63585055e-01 -4.33298349e-02 -8.05548012e-01 4.22621995e-01
4.79082167e-01 1.74355563e-02 5.75858951e-01 5.83783090e-01
1.69045046e-01 -1.49726555e-01 -1.47407830e+00 1.16764843e+00
-8.20935071e-02 -1.93765211e+00 1.72044650e-01 -8.97618607e-02
9.09033656e-01 6.66634738e-01 -8.13399330e-02 -2.43504513e-02
6.16742074e-01 -9.42101836e-01 3.83259267e-01 4.95098501e-01
8.15592527e-01 -5.51789343e-01 5.64196110e-01 -8.53149518e-02
-1.27677143e+00 -4.08731885e-02 -4.10561979e-01 -4.47177798e-01
4.66591239e-01 6.80166423e-01 -1.98712125e-01 2.71374226e-01
5.85827351e-01 1.65899694e+00 -3.02427053e-01 1.23787868e+00
-1.75558537e-01 2.50768602e-01 -2.51973987e-01 4.76628900e-01
4.73331422e-01 -1.03750475e-01 3.81299615e-01 1.32452989e+00
-4.79118526e-02 -4.61446255e-01 3.15936327e-01 9.66829538e-01
-3.89824182e-01 -2.80221403e-01 -6.10462010e-01 3.87000680e-01
3.33826184e-01 8.53411198e-01 -3.58403295e-01 -5.50951838e-01
-6.28407836e-01 9.93985355e-01 6.17286146e-01 5.22122502e-01
-6.39909744e-01 -3.60104740e-01 1.45505214e+00 4.68294509e-02
3.49456638e-01 -8.21716636e-02 -2.73460746e-01 -1.54356229e+00
-7.91232958e-02 -3.94869715e-01 3.33100587e-01 -7.53450871e-01
-1.14856565e+00 2.80158848e-01 -2.22827852e-01 -1.06085384e+00
-4.44404721e-01 -8.54198575e-01 -6.12168610e-01 8.66226912e-01
-2.24189782e+00 -6.92739308e-01 -4.19917136e-01 7.85994947e-01
5.32691777e-01 1.60550699e-01 3.99846911e-01 6.11480534e-01
-5.98858595e-01 4.71763343e-01 -1.36057481e-01 3.06360453e-01
9.70661402e-01 -1.28221226e+00 6.23565078e-01 1.00874078e+00
1.92426950e-01 1.68427587e-01 2.20258996e-01 -1.80414841e-02
-1.08677173e+00 -1.22106791e+00 8.13241899e-01 -3.01994890e-01
6.03028715e-01 -1.52118072e-01 -6.39059484e-01 9.39206302e-01
1.31112546e-01 6.89652264e-01 3.22119534e-01 4.67908802e-04
-6.23659074e-01 -3.04202139e-01 -1.00031042e+00 4.66534585e-01
1.46853590e+00 -7.76518822e-01 -2.01289088e-01 2.22028539e-01
8.17049026e-01 -6.89468563e-01 -4.96379346e-01 2.34884948e-01
4.89661664e-01 -1.42360437e+00 1.17152834e+00 -5.32054067e-01
8.18841636e-01 -7.29007050e-02 2.32452806e-02 -1.23541987e+00
-4.04881209e-01 -8.99062097e-01 -2.52325088e-01 8.50885272e-01
2.19903201e-01 -8.21601212e-01 1.03319204e+00 5.35641074e-01
-3.21073294e-01 -6.50075734e-01 -6.64273918e-01 -6.66827261e-01
2.99953818e-02 -4.86501396e-01 5.08018970e-01 8.58372808e-01
-3.76792789e-01 2.21928805e-01 -3.95909518e-01 -1.93357900e-01
7.60821700e-01 2.93034375e-01 8.38368773e-01 -1.19744790e+00
-4.21998918e-01 -6.21571362e-01 -7.26517141e-01 -1.97779083e+00
5.75813234e-01 -9.84639704e-01 -6.74524531e-02 -1.52906322e+00
6.29594997e-02 -4.51420575e-01 -1.99136645e-01 5.12997270e-01
-2.42531136e-01 4.68535900e-01 4.03175145e-01 3.60561535e-02
-7.06579983e-01 4.64803815e-01 1.78055811e+00 -9.68026072e-02
-2.06747144e-01 -1.15115762e-01 -6.67833388e-01 8.10562551e-01
5.38332701e-01 -1.58572927e-01 -6.27944529e-01 -1.07763398e+00
-4.36446480e-02 1.60012245e-02 4.44618464e-01 -1.07577705e+00
1.44516766e-01 -8.54179338e-02 2.15244904e-01 -3.00183028e-01
3.98828328e-01 -4.35855299e-01 -4.44761693e-01 3.31911564e-01
-2.91743815e-01 -1.66673496e-01 3.36313605e-01 3.24108750e-01
-5.27584791e-01 7.83821270e-02 1.07203329e+00 -2.50270277e-01
-1.15398192e+00 7.16181755e-01 -3.26000340e-02 8.74902129e-01
6.04127765e-01 -2.93440282e-01 -5.77467740e-01 -5.41795969e-01
-5.41196764e-01 4.74118143e-01 4.67760116e-01 4.18416560e-01
3.74060214e-01 -1.06226492e+00 -6.48170292e-01 4.04006243e-01
-5.27486056e-02 3.61254096e-01 2.56209493e-01 6.02694273e-01
-1.02153754e+00 5.12506008e-01 -4.15397435e-01 -8.73424292e-01
-8.43299568e-01 1.61450505e-01 5.99191487e-01 -3.09960276e-01
-7.17473745e-01 1.13745940e+00 5.28181553e-01 -2.35431746e-01
2.99244791e-01 -6.20051384e-01 1.47329226e-01 -1.29344210e-01
5.61271548e-01 6.21827662e-01 -2.12824017e-01 -4.62822825e-01
-3.02687049e-01 7.66252577e-01 -1.41815504e-03 1.96088478e-01
1.14112997e+00 -2.82994002e-01 -1.23685107e-01 1.96527824e-01
1.74357724e+00 -2.77438939e-01 -2.05610728e+00 -4.44695532e-01
-3.79899532e-01 -8.31415355e-01 2.58935630e-01 -4.06856626e-01
-1.65726352e+00 1.24851394e+00 2.25749001e-01 -2.82762498e-01
1.01933551e+00 -8.38914812e-02 1.11485398e+00 3.64942908e-01
2.68725395e-01 -7.84226179e-01 7.54244328e-02 8.41144323e-01
2.75629073e-01 -1.54219306e+00 -1.57290101e-01 -6.96456194e-01
-3.86240631e-01 1.06755459e+00 7.09284246e-01 -3.98040205e-01
7.49079227e-01 4.88810599e-01 1.78843997e-02 9.26869810e-02
-8.47700059e-01 -4.10812825e-01 2.58885443e-01 5.36051154e-01
5.07783234e-01 -1.32877842e-01 4.79677498e-01 -3.98001939e-01
-1.33137375e-01 2.60384917e-01 4.79301244e-01 7.03765392e-01
-3.75417829e-01 -1.09549224e+00 2.58232355e-01 5.94523370e-01
-4.26827222e-01 -2.45753810e-01 1.67879447e-01 6.74757659e-01
5.09805046e-02 8.16378415e-01 6.75051451e-01 2.73641106e-02
-2.04275884e-02 -2.22135097e-01 7.53465950e-01 -6.46552384e-01
-3.10625136e-01 -3.57321858e-01 1.82664409e-01 -1.33387911e+00
-6.88684344e-01 -3.55281532e-01 -1.14264870e+00 -8.06693494e-01
8.20260197e-02 -3.15825790e-01 4.36816402e-02 9.95032609e-01
6.61849678e-01 5.37450016e-01 6.04751408e-01 -1.08231771e+00
-6.82370141e-02 -4.09870088e-01 -4.22044516e-01 6.16047680e-01
9.54794943e-01 -6.00046158e-01 -4.75311428e-01 1.49891645e-01]
|
[8.726972579956055, -1.9060944318771362]
|
d3f4b7d1-8e9d-46e8-9e65-43da364576f2
|
cross-domain-human-parsing-via-adversarial
|
1801.0126
| null |
http://arxiv.org/abs/1801.01260v2
|
http://arxiv.org/pdf/1801.01260v2.pdf
|
Cross-domain Human Parsing via Adversarial Feature and Label Adaptation
|
Human parsing has been extensively studied recently due to its wide
applications in many important scenarios. Mainstream fashion parsing models
focus on parsing the high-resolution and clean images. However, directly
applying the parsers trained on benchmarks to a particular application scenario
in the wild, e.g., a canteen, airport or workplace, often gives
non-satisfactory performance due to domain shift. In this paper, we explore a
new and challenging cross-domain human parsing problem: taking the benchmark
dataset with extensive pixel-wise labeling as the source domain, how to obtain
a satisfactory parser on a new target domain without requiring any additional
manual labeling? To this end, we propose a novel and efficient cross-domain
human parsing model to bridge the cross-domain differences in terms of visual
appearance and environment conditions and fully exploit commonalities across
domains. Our proposed model explicitly learns a feature compensation network,
which is specialized for mitigating the cross-domain differences. A
discriminative feature adversarial network is introduced to supervise the
feature compensation to effectively reduce the discrepancy between feature
distributions of two domains. Besides, our model also introduces a structured
label adversarial network to guide the parsing results of the target domain to
follow the high-order relationships of the structured labels shared across
domains. The proposed framework is end-to-end trainable, practical and scalable
in real applications. Extensive experiments are conducted where LIP dataset is
the source domain and 4 different datasets including surveillance videos,
movies and runway shows are evaluated as target domains. The results
consistently confirm data efficiency and performance advantages of the proposed
method for the cross-domain human parsing problem.
|
['Defa Zhu', 'Yu Chen', 'Si Liu', 'Jiashi Feng', 'Guanghui Ren', 'Yao Sun', 'Jizhong Han']
|
2018-01-04
| null | null | null | null |
['human-parsing']
|
['computer-vision']
|
[ 0.4325786 0.04377211 0.20091961 -0.69627786 -0.9310189 -0.73504484
0.22315875 -0.27156672 -0.30263332 0.6445311 0.01289222 -0.02663219
0.09622125 -0.6717509 -0.7946985 -0.63776046 0.390803 0.25382444
0.41609102 -0.13574961 -0.15880157 0.21283558 -1.1690885 0.19749686
0.9863909 0.79947555 0.43714872 0.59977204 -0.02442191 0.417596
-0.7125741 -0.5828928 0.6537379 -0.53522915 -0.69130516 0.51488996
0.609122 -0.3155015 -0.1987945 1.392882 0.60204065 0.03707041
0.35363272 -1.0206199 -0.6593626 0.05940315 -0.8741031 -0.03663307
0.38842687 0.24161442 0.7116389 -0.4183593 0.65130347 1.5001093
0.637075 0.7587863 -1.1282239 -0.8866705 0.5152915 -0.17992665
-0.96283436 -0.16932246 1.1409745 -0.25132054 0.13660103 -0.08205877
0.16453062 1.3517308 0.24789453 0.63994014 1.354445 -0.3091707
0.13150653 0.2380485 0.07906365 0.71461755 0.18365963 0.11082199
-0.27039555 0.11778908 0.706833 -0.01197106 0.02189462 -0.5590805
-0.9235665 0.6282708 0.4226699 -0.02757511 -0.24321878 -0.39804652
0.53228277 0.06773848 0.21776593 0.0530324 -0.6921128 0.29607245
-0.53711337 0.31606057 0.42072332 1.1667532 0.6452836 -0.11224317
-0.1599821 0.98787034 0.17993543 0.52843404 0.3676566 -0.9726709
0.88750577 0.679578 0.16925253 -1.138505 -0.41276437 -0.29650193
-0.85894805 0.21354875 0.75219244 -0.42705262 -0.9857761 2.0973027
0.8649091 0.15957482 0.28430623 1.0046065 0.82879514 0.42153162
0.59363383 0.03158432 1.7326318 -1.0526152 -0.55179584 -0.7289451
0.31809387 -0.8208091 1.1550751 0.33991167 -0.7859686 -1.0985752
-1.0018299 -0.11951622 -0.2505136 0.4069786 0.33152238 0.6339677
-0.5416014 0.28965726 -0.67947555 -0.5984386 0.39431083 0.31224442
-0.5876708 -0.43459237 -1.1830653 0.27106434 0.4433223 0.2167732
-0.7204605 -0.40750292 -0.96760553 -0.17286648 0.597191 -0.5758028
1.1472977 -1.4067937 -1.346543 1.125265 0.11076089 -0.10765417
0.55817735 -0.37125558 -0.5792129 0.02456445 0.47414368 0.6699648
0.8029017 -1.3174878 -0.85159147 -0.6706425 0.21182866 0.30058575
-0.11692213 0.08159614 -0.7035489 -0.70796245 0.03709783 -1.0463724
-0.19837873 -0.12121281 -0.6119276 0.09869276 0.80270064 -1.0637578
0.7659647 -2.282284 0.10618361 -0.24241598 -0.05593942 0.2915849
-0.24810135 -0.10072632 -0.07453705 -0.24703236 -0.43977258 -0.20420547
-0.16565302 0.29881573 -0.06336138 0.33784956 0.51833177 0.5475475
-0.8698085 -0.7657812 0.08769853 0.316484 -0.43057021 0.7239742
-0.0374641 0.9521219 -0.81850064 0.6732287 1.1374714 0.05557828
0.37789494 -0.37870878 0.15986884 -0.3326131 -1.3230875 1.9150413
-0.27556092 0.0515551 0.49989128 -1.1513292 1.1403087 -0.06180478
0.2003652 -0.94856125 0.13800222 -0.02536382 -0.06128589 -0.57346094
0.23590359 -0.19348748 -0.6616538 0.0555493 0.08203228 0.17100945
0.05369766 -0.13071147 1.086706 0.42776296 0.3624782 -0.18593158
0.7451138 0.12601356 1.150796 0.41251218 -0.63121116 0.79042715
0.50202566 -0.49670276 -0.84284854 -1.1654629 0.11806508 1.2793914
0.5949013 0.12084328 -1.119935 -1.3174679 -0.25707576 0.34987783
-0.6802782 -0.10501129 -0.76531035 -0.55761254 0.4444732 0.6073442
1.1367375 -1.1429833 -0.6736836 0.0664387 -0.19999383 -1.6339524
-0.55830586 0.22507931 -0.5969184 -1.2289898 -0.7440853 -1.1801438
0.7548457 0.16843006 1.126354 -0.3049225 -0.298799 0.27081615
-0.39911366 -0.1884732 -0.3565996 0.02081315 -0.23747037 0.11069231
0.4928607 -0.21661772 -0.71114445 0.6352487 -0.9577845 0.073466
0.74992764 1.0287703 0.7458875 0.25071222 0.48980975 -1.3528548
0.4715532 -0.49269432 -0.5831738 0.48235187 -0.07894652 0.02398753
0.9662563 -0.4654538 -1.5779481 0.5428075 -0.04686716 -0.13063578
-0.69599646 -0.11814602 -0.889079 0.2662382 0.42269987 -0.0287077
-0.26623988 -0.47082055 0.19989638 0.5313449 0.9136674 -0.96830344
0.9811718 0.4102894 -0.14311598 -0.3565308 -0.9900009 -0.27351552
-0.8832875 0.03842931 1.4070649 -1.0154706 -0.11236527 0.5455154
-1.2370095 -0.17363337 -0.02316968 0.06233881 -0.40693295 0.52641565
-0.27385968 -0.5653981 -0.2735277 -1.2864658 1.332873 0.6285021
0.15002201 -0.9225472 0.0329803 0.55797386 -0.18085156 0.680332
0.90848804 -0.71424866 -0.27492544 -0.04401063 -0.48244146 0.5793532
0.3249772 -0.4262054 -0.9613865 -0.21341917 0.03001366 -0.3433325
0.362831 0.1972582 1.0241654 0.0863765 -0.30061528 0.56271887
1.3144773 0.2932757 0.4104889 0.40970144 0.7317311 1.0123612
1.148738 0.20688713 0.44212118 0.6605859 0.30943722 -0.4358411
-0.22833912 -0.4807593 0.33035004 0.33890256 0.37474436 -0.17175576
-0.7103457 0.481783 -1.7549423 -0.5803777 0.16515563 2.0922492
0.6635993 0.22293627 0.16169119 -0.27298984 1.1210006 -0.04075116
-0.80612946 -0.40400654 -0.00528699 0.18667266 0.47981086 0.04264679
-1.5792348 0.9770567 5.0145783 0.632588 -0.95704305 0.20007916
0.8050235 0.33198285 0.23687464 -0.201369 -0.8111114 0.5404907
0.5695198 0.39739972 0.25411054 1.0397637 0.09596299 0.01858141
-1.0563002 0.9153657 -0.10363915 -0.52650535 -0.29116583 -0.14123707
0.54960257 -0.57248247 0.19022228 0.6058546 0.35815206 -0.82233995
0.58690053 0.14704335 0.81048954 -0.76301855 0.79547215 0.43965828
-1.3577588 -0.20590714 -0.37718678 0.18218563 0.21813375 0.1994851
-0.5572821 0.6812393 0.9590983 0.4914546 -0.7606155 0.40600052
-0.19477706 0.32800096 -0.01050953 0.56392634 0.24121249 -0.30240276
0.30637363 1.1880692 0.19687584 0.14702214 0.52027833 0.6748613
0.04002987 0.04883849 -0.6291466 0.25905153 0.35016567 1.3260934
-0.7811891 -0.13979863 -0.69194955 1.2280375 0.35960844 0.30548415
-1.1186721 -0.31876627 0.70466137 0.15573199 0.49024275 0.09106031
-0.25331232 -0.98784804 0.16213307 -1.1850908 0.57369614 -0.49069604
-1.5139548 0.77034855 0.04906253 -1.2504225 -0.09485268 -0.7145685
-0.536632 0.8216503 -1.6774473 -1.4698929 -0.6245372 0.82034063
0.9361342 -0.22333862 0.62645 0.5097163 -0.9254981 0.78730994
-0.19708484 0.40375373 1.0548425 -1.2285206 0.38037243 1.228327
-0.28489637 0.5182654 0.69444174 -0.72803557 -1.1070068 -1.4329052
0.30098084 -0.33057207 0.22609994 -0.46581942 -0.92627674 0.64812565
0.23836657 0.37380424 0.44608575 -0.11942459 -0.35409376 -0.36237168
-1.5679114 0.29089284 1.1398449 -0.08372693 -0.48114228 0.19184905
0.70761925 -0.6815115 -0.63805974 0.50473106 0.39567524 -1.1109453
0.98019725 -0.6698817 0.5010754 -0.3639995 -0.31503528 -1.089034
-0.19230726 -0.37219316 0.32282263 1.7389511 -0.04606405 -0.40366253
0.85949975 0.7232992 0.03995468 -0.4620347 -0.84481925 -0.52968997
0.18469799 -0.12663752 0.63865155 0.675981 -0.73308575 0.41090715
-0.5435663 0.64663506 0.8482447 0.11271646 1.1720046 -1.1901108
-0.4341998 0.20715575 -0.336508 -1.132902 0.38492316 -0.4664989
0.37126625 -1.1897646 0.3339342 -0.58638775 -0.32780862 0.23607099
-0.44692913 0.03777973 0.27380425 -0.23029962 -0.69447297 0.2703857
1.3548982 -0.12578511 -0.08552483 0.08068989 -0.9938292 0.84756535
0.6754477 -0.50286514 -0.60490453 -0.73659104 -0.37654835 0.12813228
0.39685094 -1.0359035 0.0739266 -0.14882073 0.5431739 -0.26944768
0.12977433 -1.1865505 0.02400963 0.1349084 -0.18629652 0.1314945
0.27031058 0.8619409 -0.2638318 -0.0925658 1.0980082 -0.21348843
-1.0985701 0.19180307 0.34775412 0.47739816 1.366186 -0.36241877
-0.11633699 -0.01782896 -0.65512896 0.2733707 0.3980371 0.5916731
0.3902059 -1.2232348 -0.59014946 0.3842305 0.18849042 0.3152712
0.71827734 0.38211966 -0.30920106 0.06629673 -0.55588233 -0.6367866
-1.4059768 0.6293792 0.27889407 -0.6640403 -0.5297614 0.78136486
0.9376914 -0.81415194 0.05530022 -0.23711531 -0.13445501 -0.29952398
0.32836366 0.10459355 -0.21378008 -0.8108797 -0.3987664 0.9096031
-0.04278911 0.3154123 1.1367185 -0.39610147 0.36544195 -0.13706988
1.136174 0.01315913 -1.794936 -0.34721988 -0.16683301 -0.4546897
-0.39093295 -0.9176304 -1.3055854 0.84702444 0.9716394 -0.1684485
1.567584 0.03402445 1.1028315 -0.08653087 0.362472 -1.160786
0.10010866 0.15862198 0.5391252 -1.6028101 -0.13189328 -0.65829736
-1.1144191 1.0045793 1.052823 -0.29058534 0.3243384 0.13993207
0.2941926 -0.01758758 -0.1517211 -0.03063083 0.12806183 1.0909405
0.18911561 0.11954547 -0.17496619 1.1322665 0.01473194 -0.20961402
0.14215173 0.8297153 -0.19541764 -1.3101528 -0.58730406 -0.10682593
-0.446689 0.47956017 -0.24710967 1.1181695 0.69148695 0.97698253
-0.16160248 -0.37932673 0.7135842 -0.09498665 0.24701503 -0.6118878
-0.56933665 0.08033576 -0.03291311 -0.57102215 -0.45669258 -0.76152885
-1.2481186 0.10357919 0.12539841 -0.18857822 0.37225676 0.77277404
0.36541605 0.66286445 0.56727237 -0.5695636 -0.62153006 -0.6960258
-0.61873895 0.9119985 0.13254143 -0.9120771 0.05452003 0.33926302]
|
[9.25375747680664, 0.7842438817024231]
|
5694a2f0-2089-4d9d-973b-e2c57c897fbb
|
clusterq-semantic-feature-distribution
|
2205.00179
| null |
https://arxiv.org/abs/2205.00179v2
|
https://arxiv.org/pdf/2205.00179v2.pdf
|
Towards Feature Distribution Alignment and Diversity Enhancement for Data-Free Quantization
|
To obtain lower inference latency and less memory footprint of deep neural networks, model quantization has been widely employed in deep model deployment, by converting the floating points to low-precision integers. However, previous methods (such as quantization aware training and post training quantization) require original data for the fine-tuning or calibration of quantized model, which makes them inapplicable to the cases that original data are not accessed due to privacy or security. This gives birth to the data-free quantization method with synthetic data generation. While current data-free quantization methods still suffer from severe performance degradation when quantizing a model into lower bit, caused by the low inter-class separability of semantic features. To this end, we propose a new and effective data-free quantization method termed ClusterQ, which utilizes the feature distribution alignment for synthetic data generation. To obtain high inter-class separability of semantic features, we cluster and align the feature distribution statistics to imitate the distribution of real data, so that the performance degradation is alleviated. Moreover, we incorporate the diversity enhancement to solve class-wise mode collapse. We also employ the exponential moving average to update the centroid of each cluster for further feature distribution improvement. Extensive experiments based on different deep models (e.g., ResNet-18 and MobileNet-V2) over the ImageNet dataset demonstrate that our proposed ClusterQ model obtains state-of-the-art performance.
|
['Shuicheng Yan', 'Jicong Fan', 'Haijun Zhang', 'Richang Hong', 'Zhao Zhang', 'Yangcheng Gao']
|
2022-04-30
| null | null | null | null |
['data-free-quantization', 'data-free-quantization']
|
['computer-vision', 'methodology']
|
[ 1.76441576e-02 -3.94644320e-01 -1.82456806e-01 -6.14388108e-01
-6.08586609e-01 -3.02670896e-01 3.50326419e-01 1.30555689e-01
-7.10871339e-01 7.73450315e-01 -2.32091114e-01 -2.42142305e-01
-1.26570016e-01 -1.08939826e+00 -8.19244504e-01 -8.61644387e-01
1.87687173e-01 2.29781091e-01 2.87057072e-01 -5.89318424e-02
2.56374419e-01 3.12945247e-01 -1.74051750e+00 5.26064560e-02
1.17271543e+00 1.35459280e+00 4.09563690e-01 1.79682642e-01
-3.39956492e-01 3.83181065e-01 -8.34150791e-01 -3.34028810e-01
3.56738001e-01 -4.30640653e-02 -3.43052775e-01 -1.42259330e-01
3.50887984e-01 -7.21099079e-01 -4.70001847e-01 1.59228241e+00
5.36281288e-01 -9.00638998e-02 3.15206826e-01 -1.56560874e+00
-4.79368359e-01 4.10734117e-01 -4.05807257e-01 1.87594682e-01
-3.16451371e-01 9.95698720e-02 6.35553122e-01 -8.33049536e-01
2.26940498e-01 1.33727860e+00 4.81160164e-01 5.18280447e-01
-1.01984811e+00 -1.27817822e+00 2.63191145e-02 4.12766874e-01
-1.85282350e+00 -5.35480678e-01 5.85499585e-01 -1.18165649e-01
7.41553605e-01 1.33259833e-01 6.58993661e-01 6.48304999e-01
8.69365260e-02 6.05826855e-01 7.31107056e-01 -1.34497136e-01
5.82017839e-01 2.87292004e-02 -3.56460363e-02 6.52033567e-01
3.82555038e-01 -2.25924462e-01 -4.18800265e-01 -1.44297332e-01
9.72006142e-01 2.82968700e-01 -3.82327080e-01 -1.88315421e-01
-1.09619701e+00 7.23520994e-01 7.75192201e-01 -4.22047377e-02
-4.86736506e-01 4.65741098e-01 6.18073821e-01 8.20897669e-02
1.00399889e-01 -1.95609033e-01 -4.22350705e-01 -2.19110280e-01
-1.11485100e+00 4.73633230e-01 2.45871961e-01 1.08572602e+00
1.06143165e+00 2.65646309e-01 -1.12465940e-01 7.93890834e-01
5.48113585e-01 5.54906011e-01 1.06325150e+00 -8.58520865e-01
8.21183860e-01 5.89519382e-01 2.90700793e-02 -1.27073836e+00
-1.66587412e-01 -5.47041178e-01 -1.29295886e+00 -4.67235930e-02
-2.42400132e-02 -5.30656129e-02 -1.07678854e+00 1.72190499e+00
2.56468475e-01 3.93076271e-01 2.04444021e-01 7.44095981e-01
6.85499966e-01 7.89480448e-01 3.28375660e-02 -1.42051220e-01
1.31230867e+00 -8.33188951e-01 -7.87329674e-01 1.45285679e-02
6.47428274e-01 -3.85311484e-01 1.08898401e+00 4.11952347e-01
-7.77160943e-01 -7.77864456e-01 -1.39227271e+00 5.17389812e-02
-3.53919983e-01 5.85319214e-02 4.26439494e-01 8.48361909e-01
-9.76859748e-01 5.61200380e-01 -1.03330207e+00 2.38447726e-01
8.00189018e-01 6.14828527e-01 -1.65843055e-01 -4.55819629e-02
-1.52165973e+00 3.66124600e-01 9.55531240e-01 8.44742730e-02
-5.92767596e-01 -6.95478559e-01 -7.05309272e-01 2.18072519e-01
7.26389373e-03 -3.69901806e-01 1.16545343e+00 -7.79382467e-01
-1.37472785e+00 1.98812559e-01 -1.42456114e-01 -6.52061224e-01
3.84994656e-01 6.08701520e-02 -4.53330845e-01 8.24621618e-02
-1.16430469e-01 1.04469573e+00 8.89864206e-01 -1.05294967e+00
-7.94966519e-01 -2.93406695e-01 -1.02875724e-01 2.07862720e-01
-1.00722504e+00 -4.57669675e-01 -6.25896394e-01 -7.29617834e-01
3.68083596e-01 -6.31024778e-01 -2.29033202e-01 9.14938822e-02
-2.94950634e-01 -9.98653546e-02 8.16120386e-01 -4.02100831e-01
1.61364925e+00 -2.22558451e+00 -4.53159213e-01 3.39088291e-01
3.60254735e-01 6.05247378e-01 2.26452984e-02 -1.34714574e-01
1.17894769e-01 3.42432439e-01 -1.25777215e-01 -4.36415046e-01
5.69263287e-02 5.21086454e-01 -1.92582071e-01 2.42257282e-01
1.92744970e-01 6.98611081e-01 -8.27301264e-01 -8.12380552e-01
2.30963334e-01 5.14000356e-01 -8.13109100e-01 -1.26772031e-01
-9.75050926e-02 -5.86933573e-04 -5.75446367e-01 4.05239463e-01
1.18891633e+00 -2.40122631e-01 -1.46237478e-01 -5.38405001e-01
1.57265261e-01 2.23157465e-01 -1.36991107e+00 1.55619478e+00
-3.96461129e-01 2.15840608e-01 -9.40493494e-02 -9.13646281e-01
1.06289792e+00 1.97546318e-01 3.09952289e-01 -8.55397224e-01
1.66976169e-01 2.81603336e-01 -1.29352972e-01 -1.27528593e-01
7.83259511e-01 1.03347242e-01 1.26739725e-01 4.39290218e-02
-9.21277776e-02 9.29609910e-02 -2.40064263e-02 1.64312627e-02
6.95987046e-01 -2.97810555e-01 -1.70686096e-02 -1.50667727e-01
3.82588089e-01 -2.29801342e-01 8.76512825e-01 4.14881468e-01
-2.93123245e-01 5.86757898e-01 2.50692785e-01 -3.32269490e-01
-9.53553855e-01 -9.09805834e-01 -4.24394369e-01 7.12018430e-01
3.37965220e-01 -5.55244148e-01 -1.04668736e+00 -5.14317393e-01
-1.15056276e-01 5.39881170e-01 -2.45289594e-01 -4.97463912e-01
-4.78330076e-01 -9.42711532e-01 6.82219386e-01 4.67776775e-01
1.13666070e+00 -7.76433706e-01 -6.76432848e-01 4.03895050e-01
-2.40732402e-01 -9.97486949e-01 -3.02605629e-01 6.27456680e-02
-9.79704142e-01 -5.84140241e-01 -6.22753620e-01 -8.45284522e-01
6.15948558e-01 2.34184131e-01 6.87840879e-01 2.05724448e-01
1.80124909e-01 -4.16631758e-01 -3.24737370e-01 -2.73098737e-01
-2.28529721e-01 2.90347010e-01 2.44867086e-01 2.60249544e-02
3.82159948e-01 -6.70273066e-01 -9.26489592e-01 1.58976033e-01
-1.22532034e+00 1.23152867e-01 4.51532423e-01 9.69806135e-01
8.69470894e-01 5.19210279e-01 8.05217087e-01 -4.25269812e-01
6.30903840e-01 -4.72033083e-01 -6.20702207e-01 1.57667413e-01
-6.87481463e-01 1.29182994e-01 1.07226157e+00 -4.33539808e-01
-6.47677183e-01 -6.55776188e-02 -3.17907780e-01 -8.10778081e-01
1.60357788e-01 4.77106839e-01 -4.57791328e-01 -8.22246969e-02
5.01126766e-01 5.08831978e-01 5.37042646e-03 -2.73753583e-01
2.64992356e-01 1.04272830e+00 6.45006120e-01 -3.47535193e-01
6.99195027e-01 1.59820169e-01 -8.34724084e-02 -5.57994187e-01
-2.24070787e-01 -2.62295129e-03 -3.04021895e-01 1.96553499e-01
6.95124328e-01 -1.15888786e+00 -6.40107989e-01 7.38778412e-01
-1.13163662e+00 -7.86616132e-02 -6.73104823e-02 4.22201365e-01
-2.74642587e-01 3.19089830e-01 -3.62533927e-01 -5.09531021e-01
-6.11013174e-01 -1.53857946e+00 1.01608324e+00 4.70559061e-01
3.41818124e-01 -7.34478354e-01 -5.18507659e-01 -1.18680328e-01
4.37372565e-01 8.75681192e-02 9.21737254e-01 -4.64436322e-01
-6.81028605e-01 -1.95236862e-01 -5.87630033e-01 5.35114408e-01
2.54841179e-01 -1.10860877e-01 -9.01183426e-01 -3.88250589e-01
-8.90685022e-02 -4.06790107e-01 7.25899279e-01 6.54688179e-02
1.77410555e+00 -5.34613311e-01 -2.34875411e-01 9.26538885e-01
1.36600459e+00 1.73173636e-01 7.12638676e-01 3.57438833e-01
8.60989392e-01 -2.29223855e-02 7.69117773e-01 8.23564768e-01
5.95253766e-01 6.02961242e-01 7.19370544e-01 1.24952264e-01
9.53864157e-02 -3.13078195e-01 -1.13931543e-03 9.54767346e-01
4.28096652e-01 -2.70388961e-01 -7.76853204e-01 4.18984383e-01
-1.63377655e+00 -5.84512889e-01 1.96727023e-01 2.29499340e+00
1.14039493e+00 4.56407875e-01 -1.96910143e-01 4.52382982e-01
7.49227524e-01 1.35699511e-01 -7.37950981e-01 -1.43581897e-01
1.79702967e-01 1.87451139e-01 8.03965688e-01 9.53682959e-02
-1.02709842e+00 9.01611745e-01 5.11659765e+00 1.59242272e+00
-1.45799148e+00 9.06643942e-02 7.71555781e-01 -8.59294739e-03
-3.01886827e-01 -4.29677308e-01 -1.12854326e+00 9.42280054e-01
9.77732122e-01 -2.51329213e-01 4.84222829e-01 9.40678895e-01
1.83389634e-01 2.06113845e-01 -7.71492422e-01 1.42016006e+00
-2.71611989e-01 -1.37462747e+00 4.19991553e-01 -1.94738116e-02
7.12927163e-01 1.15075693e-01 1.77215174e-01 3.89199317e-01
-7.54520446e-02 -9.18356657e-01 9.08529282e-01 2.69506186e-01
1.15903938e+00 -9.86847222e-01 7.66961336e-01 4.40518141e-01
-1.32676375e+00 -1.31142855e-01 -8.27029467e-01 1.92289457e-01
-8.54562446e-02 6.45392239e-01 -7.79271126e-01 4.97270703e-01
7.47093022e-01 6.72214806e-01 -6.78586841e-01 9.89533961e-01
3.38238657e-01 5.13548434e-01 -5.48838913e-01 -5.10785170e-02
3.13573360e-01 4.92584780e-02 2.84494786e-03 7.91722119e-01
5.28014183e-01 -5.84734343e-02 1.12093501e-01 6.98255301e-01
-3.24423403e-01 4.50313278e-02 -1.08462028e-01 1.61233008e-01
1.28466272e+00 1.02359200e+00 -6.65817559e-01 -5.73137164e-01
-1.10365577e-01 8.10829997e-01 2.09456280e-01 2.09236562e-01
-9.85847235e-01 -7.69193411e-01 8.25705707e-01 -1.43553928e-01
3.80805939e-01 -1.42910793e-01 -3.77739698e-01 -1.02838600e+00
2.20091403e-01 -8.84607196e-01 9.29436740e-03 -4.78528857e-01
-8.22131515e-01 6.32089317e-01 7.21824691e-02 -1.45410097e+00
-2.47455657e-01 -3.46347153e-01 -3.89207065e-01 9.26547766e-01
-1.63880563e+00 -7.15364158e-01 -5.46116948e-01 7.16419101e-01
3.99677575e-01 -1.53578356e-01 7.10762739e-01 6.79326296e-01
-5.95022202e-01 1.28655386e+00 4.68961179e-01 7.79570043e-02
5.09453058e-01 -8.59891295e-01 6.21983469e-01 5.86946011e-01
-2.13146597e-01 6.97954535e-01 3.94672662e-01 -3.73411924e-01
-1.27737951e+00 -1.60548580e+00 4.52705503e-01 2.10590452e-01
3.21765006e-01 -3.86768013e-01 -1.08816075e+00 3.97555023e-01
-3.96908760e-01 3.77401769e-01 3.50949168e-01 -6.34024024e-01
-1.27342865e-01 -7.01133907e-01 -1.47632456e+00 5.94834924e-01
8.41157377e-01 -3.52710307e-01 -1.95710093e-01 2.15941757e-01
1.02868485e+00 -4.01316851e-01 -1.06916559e+00 4.74902779e-01
4.58263397e-01 -8.98022950e-01 9.88550425e-01 -7.37217218e-02
1.87270507e-01 -5.19695520e-01 -4.09291893e-01 -1.17961645e+00
-1.51228845e-01 -3.55530918e-01 -1.47300869e-01 1.37997043e+00
2.69535873e-02 -8.81093681e-01 9.69985247e-01 4.82608259e-01
-1.02978095e-01 -9.27035987e-01 -1.15088129e+00 -6.97519183e-01
7.73187820e-03 -2.85668015e-01 1.24037147e+00 8.60487819e-01
-4.30975139e-01 -1.13470726e-01 -7.92941153e-02 2.14227274e-01
6.36012316e-01 -2.89436877e-01 6.84899092e-01 -1.01098847e+00
-2.28104871e-02 -4.32168126e-01 -7.86062956e-01 -1.32795048e+00
4.35379939e-03 -7.38641798e-01 8.06128010e-02 -1.22656584e+00
-1.50841177e-01 -1.07008612e+00 -4.83396590e-01 5.26448786e-01
-2.51560152e-01 5.12998819e-01 2.18927354e-01 3.71580929e-01
-6.56633139e-01 9.32046175e-01 1.31451821e+00 -4.18197699e-02
-1.68199807e-01 -2.31595784e-01 -6.01910055e-01 5.27280152e-01
9.94277298e-01 -5.59880197e-01 -7.67835498e-01 -6.94600761e-01
5.98778315e-02 -9.97634307e-02 3.44068080e-01 -1.48097169e+00
3.10099781e-01 -1.39287263e-02 4.35713649e-01 -7.10603476e-01
3.55508119e-01 -8.25585008e-01 5.78112751e-02 5.16871810e-01
-9.47984755e-02 2.65598059e-01 2.23004401e-01 5.51466942e-01
-4.39114869e-01 -3.29872310e-01 7.49670565e-01 1.04279950e-01
-7.94333220e-01 5.99127471e-01 -3.20185162e-02 -4.61263210e-02
7.91997194e-01 -3.42752576e-01 -3.86590838e-01 -1.95838451e-01
-1.72956571e-01 4.52248693e-01 4.15721864e-01 2.70749837e-01
7.52786398e-01 -1.69300902e+00 -4.17820603e-01 5.93455970e-01
1.87556669e-02 5.64977944e-01 3.60706925e-01 5.06919742e-01
-7.83195198e-01 2.61028767e-01 -2.54013479e-01 -7.43996084e-01
-9.18494105e-01 4.16385889e-01 2.94814229e-01 -1.18087903e-01
-3.12464505e-01 7.88360238e-01 1.07112847e-01 -3.77125472e-01
3.58647883e-01 -7.00628161e-01 -1.01181410e-01 -1.61669657e-01
7.55991459e-01 3.03541660e-01 3.20803583e-01 -5.38978100e-01
-3.30411106e-01 4.05100554e-01 -2.74240017e-01 1.17256112e-01
9.43464696e-01 -1.93862215e-01 -5.43289399e-03 1.32375777e-01
1.53125548e+00 -5.27258873e-01 -1.36826992e+00 -2.14187786e-01
-4.40403521e-01 -4.45635259e-01 2.99702555e-01 -3.10431093e-01
-1.39258027e+00 1.08630037e+00 9.41636801e-01 1.17737874e-01
1.33358479e+00 -6.74621344e-01 1.21250069e+00 5.46787262e-01
5.92743039e-01 -9.33835864e-01 -1.95135131e-01 4.42362458e-01
3.60061109e-01 -1.06428325e+00 -3.65760475e-02 -1.50928944e-01
-3.42852771e-01 9.24341202e-01 8.22614789e-01 -2.35281941e-02
8.34492981e-01 1.06580824e-01 7.21519999e-03 2.71316946e-01
-5.76098204e-01 3.67108881e-01 -1.55692548e-01 5.54409683e-01
-8.99770558e-02 8.25116858e-02 -1.74791262e-01 7.49344289e-01
-6.20978713e-01 2.29084432e-01 2.50593960e-01 9.27641153e-01
-5.47503293e-01 -9.92163420e-01 -3.10881048e-01 6.94427729e-01
-3.35824877e-01 -2.19806105e-01 3.79580975e-01 4.76247042e-01
4.43914682e-01 7.90371239e-01 3.83911252e-01 -7.46231616e-01
1.02927513e-01 -1.45293877e-01 1.34508714e-01 -2.47676104e-01
-2.96381325e-01 -9.00790840e-02 -4.08849448e-01 -3.80648553e-01
-5.00255376e-02 -2.13224471e-01 -1.74051499e+00 -4.37975615e-01
-4.22669888e-01 1.65282190e-01 7.93597877e-01 7.69500732e-01
5.64608097e-01 7.01690257e-01 6.91505253e-01 -9.19900060e-01
-9.65354502e-01 -9.07668650e-01 -4.91426885e-01 2.19412565e-01
4.02950317e-01 -6.59044027e-01 -6.76192418e-02 -1.44665025e-03]
|
[8.70576000213623, 3.0287702083587646]
|
1d52ff78-2480-4f03-b0b2-4877eb81c12a
|
foveation-based-deep-video-compression
|
2203.1649
| null |
https://arxiv.org/abs/2203.16490v1
|
https://arxiv.org/pdf/2203.16490v1.pdf
|
Foveation-based Deep Video Compression without Motion Search
|
The requirements of much larger file sizes, different storage formats, and immersive viewing conditions of VR pose significant challenges to the goals of acquiring, transmitting, compressing, and displaying high-quality VR content. At the same time, the great potential of deep learning to advance progress on the video compression problem has driven a significant research effort. Because of the high bandwidth requirements of VR, there has also been significant interest in the use of space-variant, foveated compression protocols. We have integrated these techniques to create an end-to-end deep learning video compression framework. A feature of our new compression model is that it dispenses with the need for expensive search-based motion prediction computations. This is accomplished by exploiting statistical regularities inherent in video motion expressed by displaced frame differences. Foveation protocols are desirable since only a small portion of a video viewed in VR may be visible as a user gazes in any given direction. Moreover, even within a current field of view (FOV), the resolution of retinal neurons rapidly decreases with distance (eccentricity) from the projected point of gaze. In our learning based approach, we implement foveation by introducing a Foveation Generator Unit (FGU) that generates foveation masks which direct the allocation of bits, significantly increasing compression efficiency while making it possible to retain an impression of little to no additional visual loss given an appropriate viewing geometry. Our experiment results reveal that our new compression model, which we call the Foveated MOtionless VIdeo Codec (Foveated MOVI-Codec), is able to efficiently compress videos without computing motion, while outperforming foveated version of both H.264 and H.265 on the widely used UVG dataset and on the HEVC Standard Class B Test Sequences.
|
['Alan C. Bovik', 'Richard Webb', 'Meixu Chen']
|
2022-03-30
| null | null | null | null |
['foveation']
|
['computer-vision']
|
[ 2.79934347e-01 -1.98407874e-01 3.00722364e-02 -7.55642578e-02
-3.72014970e-01 -3.47734183e-01 3.49375039e-01 1.82041118e-03
-5.96371889e-01 3.84630293e-01 1.52542248e-01 -3.31470162e-01
-7.13290945e-02 -6.77166879e-01 -8.14272106e-01 -5.79966009e-01
-2.58984596e-01 -3.74536514e-01 3.15357298e-01 1.98648330e-02
4.95877087e-01 4.59662795e-01 -1.93032110e+00 3.36387813e-01
6.24017298e-01 1.07686090e+00 8.79461050e-01 1.04998350e+00
2.94005841e-01 9.13803101e-01 -4.57103461e-01 -2.22282827e-01
4.01393890e-01 -3.39489549e-01 -5.06821573e-01 1.12151414e-01
7.28054762e-01 -8.74445975e-01 -8.40086877e-01 7.68117547e-01
6.13710821e-01 4.01874810e-01 2.15200901e-01 -8.07098329e-01
-3.95203084e-01 1.36560440e-01 -5.35560250e-01 7.18406737e-01
4.38181967e-01 1.77706778e-01 8.29617262e-01 -5.65572381e-01
7.63411283e-01 7.66930580e-01 2.20530748e-01 4.90271926e-01
-1.02740562e+00 -1.45969167e-01 -1.34415209e-01 5.03145158e-01
-1.52521825e+00 -6.82583570e-01 3.94441009e-01 -2.79343188e-01
1.23041034e+00 3.80135268e-01 8.05674493e-01 5.97021341e-01
4.10932183e-01 4.34730530e-01 5.59630036e-01 -3.59769851e-01
4.22484994e-01 -1.02315255e-01 -4.42068547e-01 5.71018100e-01
3.35789099e-02 1.72133297e-01 -6.55431509e-01 3.14307421e-01
9.62878764e-01 -2.93517411e-02 -9.42043304e-01 -4.87206936e-01
-9.94212210e-01 4.97323424e-01 2.77496338e-01 1.71622664e-01
-3.24639142e-01 3.43838006e-01 2.11791337e-01 3.07094783e-01
1.34993479e-01 1.49598882e-01 -2.20460415e-01 -5.83791792e-01
-1.25958204e+00 3.24701220e-01 4.19459164e-01 8.75550687e-01
3.62210780e-01 2.52424389e-01 2.32112892e-02 6.39623404e-01
2.01757535e-01 4.30652708e-01 6.00930035e-01 -1.60333967e+00
4.25701916e-01 -1.66455153e-02 -2.61479672e-02 -1.08297551e+00
-1.01510935e-01 -4.37104523e-01 -6.15023673e-01 4.76407081e-01
1.64999381e-01 1.17734455e-01 -5.51088095e-01 1.71765757e+00
1.06294960e-01 2.17028707e-01 6.67319000e-02 1.03574109e+00
4.17416602e-01 7.88707674e-01 -3.00663203e-01 -4.26296979e-01
1.10306442e+00 -5.04640937e-01 -6.27217472e-01 -5.72446920e-03
6.14405870e-01 -5.99969268e-01 1.12175417e+00 5.74747443e-01
-1.46820867e+00 -6.58562899e-01 -1.33637416e+00 -2.89336771e-01
1.13249093e-01 -1.74203530e-01 3.67632687e-01 7.30388105e-01
-1.30379879e+00 7.04820156e-01 -1.00017524e+00 1.82047859e-02
4.20042396e-01 4.73597080e-01 -3.68566871e-01 -3.80522460e-01
-7.31160223e-01 6.23646617e-01 1.76352695e-01 -2.22677529e-01
-5.24560750e-01 -6.77288949e-01 -7.30027795e-01 5.64174175e-01
2.55691886e-01 -6.13000154e-01 1.08773685e+00 -8.96588326e-01
-1.36238945e+00 4.94421214e-01 -3.01342130e-01 -6.67377710e-01
4.21315938e-01 -2.17164382e-01 -1.50196075e-01 6.68663025e-01
-4.69486356e-01 9.42510366e-01 9.58868563e-01 -9.18987036e-01
-7.95414686e-01 -3.51045996e-01 1.73530638e-01 5.39346814e-01
-3.54351431e-01 -1.69894591e-01 -6.29171669e-01 -4.10832912e-01
-4.20424715e-02 -9.10246134e-01 1.20479487e-01 1.98621750e-01
2.21991882e-01 2.45586053e-01 1.02011025e+00 -7.97317863e-01
1.33696258e+00 -2.29227304e+00 2.78850317e-01 -5.47442995e-02
5.05178452e-01 4.24166113e-01 -9.92310271e-02 5.74715883e-02
8.59138090e-03 -6.49647862e-02 5.66492416e-02 -1.85349241e-01
-5.19625902e-01 8.73426124e-02 -2.40842566e-01 4.93733913e-01
-2.67319709e-01 6.18889868e-01 -7.32884347e-01 -3.83574665e-01
4.36014086e-01 7.99488664e-01 -1.10809028e+00 2.16541484e-01
1.68586001e-02 1.67921275e-01 1.15438052e-01 3.46184134e-01
5.84702194e-01 -2.96082377e-01 1.58630952e-01 -2.21478850e-01
-2.29748204e-01 2.03638360e-01 -9.93763983e-01 1.78538716e+00
-4.30532247e-01 1.25714910e+00 -9.36539471e-02 -7.74455667e-01
3.89684856e-01 3.19373220e-01 4.92187589e-01 -1.05634701e+00
1.77733183e-01 -1.86196645e-03 1.76857054e-01 -5.35063565e-01
8.30203831e-01 3.77588272e-01 7.04032660e-01 3.57534200e-01
-6.79559410e-02 1.58886448e-01 3.65520835e-01 2.33786494e-01
1.15887761e+00 1.14726871e-01 1.79677233e-01 2.70852804e-01
1.93080321e-01 -4.62789446e-01 3.24684620e-01 4.45581079e-01
-2.33277276e-01 9.67626035e-01 2.96217054e-01 -1.51642442e-01
-1.29468501e+00 -1.01255119e+00 -1.19872473e-01 7.63033450e-01
1.80909052e-01 -7.07355738e-01 -7.10218608e-01 2.08205245e-02
-4.55116659e-01 6.32315993e-01 -1.90134346e-01 -1.04813531e-01
-6.60487652e-01 -1.75962865e-01 1.30133197e-01 3.12017798e-01
6.07804954e-01 -7.84957528e-01 -1.62779200e+00 2.35148389e-02
-4.47302818e-01 -1.22546506e+00 -4.80758071e-01 -1.69755965e-01
-1.04393470e+00 -8.54056001e-01 -8.56760800e-01 -4.53160554e-01
3.33242506e-01 7.96764910e-01 8.76025736e-01 -5.42840511e-02
-2.96129704e-01 5.63210487e-01 -4.27525252e-01 6.31019399e-02
-1.99784368e-01 -2.03148931e-01 -1.93272457e-01 -3.39469224e-01
6.58416227e-02 -6.60172939e-01 -9.53889132e-01 1.82097144e-02
-1.08335197e+00 2.42184028e-01 4.04586136e-01 5.08542895e-01
6.25836313e-01 2.61441201e-01 3.99316736e-02 -3.11891168e-01
2.68541217e-01 -2.60769844e-01 -7.03606248e-01 -5.40194735e-02
-5.95129550e-01 -1.64809614e-01 4.47666943e-01 -4.50413436e-01
-7.20541656e-01 -2.34826744e-01 -6.24205433e-02 -8.24170172e-01
1.13421485e-01 3.68484765e-01 2.53145322e-02 -2.00147539e-01
5.89237511e-01 3.62124175e-01 1.85790017e-01 -1.93447210e-02
2.44096801e-01 7.02588141e-01 7.04330385e-01 1.27623752e-01
3.92283916e-01 4.99136150e-01 7.79669583e-02 -1.25698125e+00
5.67116104e-02 -2.70174801e-01 -3.03105861e-01 -3.47040683e-01
8.91149879e-01 -9.43513274e-01 -8.40114057e-01 2.31981382e-01
-9.60553586e-01 -3.78114671e-01 -2.53021598e-01 6.99115992e-01
-8.51550043e-01 6.74300194e-01 -5.74103713e-01 -5.98570108e-01
-1.56784043e-01 -1.26831591e+00 6.41869128e-01 1.44777879e-01
-5.90905398e-02 -6.51576281e-01 -1.86956882e-01 4.02128696e-01
5.77702403e-01 -6.21222220e-02 9.21438575e-01 1.35488570e-01
-1.09430361e+00 -1.01158492e-01 -1.85232729e-01 4.76206005e-01
-1.10963896e-01 -1.12111114e-01 -8.97423029e-01 -4.43861604e-01
1.55080676e-01 -1.25229895e-01 5.86635113e-01 7.42519319e-01
1.26325822e+00 -1.96004167e-01 6.69799596e-02 1.02028883e+00
1.49427342e+00 4.71226215e-01 1.10834646e+00 2.43805155e-01
4.95372504e-01 4.56385791e-01 4.59611684e-01 5.82087636e-01
2.87304699e-01 9.53871012e-01 5.21029294e-01 1.58337727e-01
-3.79339188e-01 1.91028528e-02 1.47941262e-01 5.93263686e-01
-4.30512786e-01 -5.22082269e-01 -5.68727732e-01 2.21439704e-01
-1.37842822e+00 -1.28386641e+00 2.32088625e-01 2.60370851e+00
6.36981726e-01 -3.87493894e-02 -1.91848531e-01 3.16028714e-01
3.03226024e-01 1.65990293e-01 -4.92522985e-01 -3.85664761e-01
-1.30229011e-01 1.77430049e-01 5.75399160e-01 4.54185933e-01
-7.07563937e-01 5.67164660e-01 5.66196442e+00 5.04488528e-01
-1.52601802e+00 -2.15955436e-01 5.47600091e-01 -7.00266600e-01
-1.81749821e-01 -1.37140468e-01 -4.86262321e-01 5.35431266e-01
1.30328214e+00 -1.85103238e-01 8.66073966e-01 7.31242239e-01
5.61072409e-01 -4.45069760e-01 -1.09038424e+00 1.51149487e+00
2.79660165e-01 -1.52325237e+00 -4.16995734e-02 4.66467947e-01
3.59828681e-01 1.17282085e-01 4.34215933e-01 -1.03787899e-01
-4.64560598e-01 -1.07064199e+00 7.46294022e-01 2.89024413e-01
1.15422690e+00 -7.41973639e-01 4.16609079e-01 1.89844623e-01
-9.13432360e-01 -2.66226768e-01 -6.70676947e-01 7.69943446e-02
1.93590775e-01 2.56066233e-01 -5.81382215e-01 2.46156558e-01
8.14353049e-01 4.88980770e-01 -3.85811508e-01 1.16003466e+00
2.94413030e-01 3.97663951e-01 -2.06920922e-01 4.77015615e-01
-1.63477853e-01 7.22321719e-02 4.62305844e-01 8.23464155e-01
7.03192294e-01 2.73040652e-01 -5.15934229e-01 5.07323682e-01
-2.94273607e-02 2.99671013e-03 -6.09653473e-01 -1.45133529e-02
4.16298866e-01 7.81896591e-01 -5.27403355e-01 -2.38913912e-02
-7.13459551e-01 1.10604405e+00 6.44563809e-02 5.00414789e-01
-6.77336752e-01 -3.31352770e-01 4.83399212e-01 4.72169012e-01
7.30801642e-01 -4.88037944e-01 -6.52929470e-02 -1.21068156e+00
-2.76939641e-03 -1.03352392e+00 -9.26944017e-02 -1.13451850e+00
-9.66049731e-02 6.81267798e-01 2.81070992e-02 -1.31987751e+00
-5.24503648e-01 -3.45173955e-01 -1.37164339e-01 7.54399061e-01
-1.50879717e+00 -4.65144068e-01 -5.39688170e-01 7.64571905e-01
6.47256613e-01 -1.07886933e-01 5.81537604e-01 5.26651323e-01
-9.77531597e-02 6.36662900e-01 8.94040149e-03 -3.37686330e-01
5.20294070e-01 -7.62638450e-01 -2.94221584e-02 9.75134432e-01
2.56119430e-01 5.39438009e-01 8.40050757e-01 -3.92615318e-01
-1.67062306e+00 -7.60296881e-01 7.15781987e-01 -4.40278053e-02
-2.84617972e-02 -1.16428979e-01 -1.00473511e+00 4.58196372e-01
1.58614263e-01 5.00620343e-02 7.15962768e-01 -3.01280707e-01
-2.04375133e-01 -1.53772369e-01 -9.98833060e-01 6.43555522e-01
8.02897274e-01 -4.40642446e-01 -1.75820887e-01 -1.10416291e-02
6.90114439e-01 -4.48165774e-01 -5.62191904e-01 2.05166250e-01
7.24372387e-01 -1.45706320e+00 9.97363150e-01 3.76562588e-02
7.43248880e-01 -1.80478558e-01 -5.83271384e-01 -8.91981363e-01
-1.88264027e-01 -5.09160101e-01 -5.45376897e-01 6.79103017e-01
-6.29457161e-02 -1.95831388e-01 8.42107773e-01 6.75486803e-01
-1.23262472e-01 -6.85794592e-01 -9.41038489e-01 -5.31395912e-01
-3.39072108e-01 -4.99916285e-01 6.35638013e-02 5.27165771e-01
3.56718339e-02 -3.01175527e-02 -3.65689307e-01 1.07165642e-01
4.23925221e-01 -9.17324126e-02 7.07746446e-01 -8.12088728e-01
-8.12995970e-01 -2.12441102e-01 -7.97775209e-01 -1.50672185e+00
-4.16286200e-01 -5.98639727e-01 -2.77147561e-01 -1.20453918e+00
3.90627868e-02 -1.48520947e-01 -7.68450573e-02 -5.30735143e-02
2.27638528e-01 2.70944476e-01 4.33816642e-01 3.78741711e-01
-3.36218089e-01 4.32179332e-01 1.21419692e+00 3.08657199e-01
-3.03111762e-01 -8.25515315e-02 -3.78636867e-01 4.36692595e-01
5.84520936e-01 -1.97674502e-02 -9.90608394e-01 -7.52852380e-01
3.17089200e-01 5.64238131e-01 2.58146316e-01 -1.26345873e+00
2.37014055e-01 7.53794014e-02 3.85972708e-01 -5.70717752e-01
5.34442186e-01 -7.60168612e-01 1.39389172e-01 4.36894149e-01
-4.42290127e-01 2.63692260e-01 8.74361843e-02 5.63162088e-01
-1.31232560e-01 -1.56109899e-01 9.15532649e-01 5.40608428e-02
-7.45213151e-01 1.64664671e-01 -5.32778382e-01 -8.86844844e-02
9.51688051e-01 -6.45093560e-01 -1.41377881e-01 -6.02373481e-01
-4.57747638e-01 -2.81184316e-01 7.29270339e-01 2.45622501e-01
1.03797889e+00 -7.89478302e-01 -4.35992330e-01 4.40737277e-01
-2.91212559e-01 -1.01244375e-01 6.22138917e-01 6.30444169e-01
-8.49926412e-01 5.77530503e-01 -3.89593571e-01 -6.44297779e-01
-1.60077465e+00 6.73153102e-01 2.18669176e-01 1.47158101e-01
-9.25944865e-01 7.55519331e-01 2.33689874e-01 7.30827093e-01
4.33271319e-01 -2.76022077e-01 -2.76636690e-01 -3.67158294e-01
9.43451881e-01 5.04037142e-01 6.45378307e-02 -5.03859103e-01
1.81200564e-01 3.98493946e-01 -1.63026869e-01 -2.98486888e-01
1.12885296e+00 -4.76913959e-01 3.58698398e-01 1.63785398e-01
1.13959384e+00 2.29685679e-01 -1.59437490e+00 7.24293813e-02
-5.43531179e-01 -9.07531559e-01 4.90185946e-01 -4.32186902e-01
-1.30899751e+00 1.03738952e+00 1.02460861e+00 -3.44888009e-02
1.45927668e+00 -3.31876099e-01 9.81547296e-01 2.23045439e-01
4.17180330e-01 -8.81047249e-01 5.92206381e-02 1.63641870e-01
5.34494221e-01 -8.58388305e-01 1.67261232e-02 -3.04441750e-01
-4.99042511e-01 1.07901001e+00 3.38846862e-01 2.71772463e-02
2.72492588e-01 2.35189378e-01 -2.27602705e-01 8.82111341e-02
-9.62413609e-01 1.15170263e-01 1.48797417e-02 8.37209105e-01
4.31398034e-01 -3.97719502e-01 -2.95835674e-01 -1.33246288e-01
-3.70524764e-01 2.32907996e-01 9.17185187e-01 9.49732244e-01
-6.46880627e-01 -6.41906977e-01 8.69875774e-03 3.54744583e-01
-5.85564077e-01 -2.57767677e-01 5.06764412e-01 5.62278807e-01
-3.52132991e-02 9.57136333e-01 4.60451484e-01 -3.92778397e-01
-1.02168232e-01 -3.04324508e-01 8.24161708e-01 -3.21855932e-01
5.27395830e-02 2.44225636e-01 -3.09587002e-01 -8.99729967e-01
-2.37811521e-01 -5.91808915e-01 -1.19127965e+00 -5.41983843e-01
-4.33689430e-02 -2.27179959e-01 8.73683035e-01 6.56162977e-01
5.11271536e-01 5.12368262e-01 4.70181644e-01 -8.94226670e-01
-4.01996106e-01 -5.04358232e-01 -5.76768517e-01 1.77969262e-01
4.82316762e-01 -3.65009099e-01 -2.79700965e-01 3.47386360e-01]
|
[11.366376876831055, -1.6833715438842773]
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.