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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
ba3e6497-a832-4a94-a573-c98f7639697d | end-to-end-emotion-cause-pair-extraction | null | null | https://aclanthology.org/2020.emnlp-main.290 | https://aclanthology.org/2020.emnlp-main.290.pdf | End-to-End Emotion-Cause Pair Extraction based on Sliding Window Multi-Label Learning | Emotion-cause pair extraction (ECPE) is a new task that aims to extract the potential pairs of emotions and their corresponding causes in a document. The existing methods first perform emotion extraction and cause extraction independently, and then perform emotion-cause pairing and filtering. However, the above methods ignore the fact that the cause and the emotion it triggers are inseparable, and the extraction of the cause without specifying the emotion is pathological, which greatly limits the performance of the above methods in the first step. To tackle these shortcomings, we propose two joint frameworks for ECPE: 1) multi-label learning for the extraction of the cause clauses corresponding to the specified emotion clause (CMLL) and 2) multi-label learning for the extraction of the emotion clauses corresponding to the specified cause clause (EMLL). The window of multi-label learning is centered on the specified emotion clause or cause clause and slides as their positions move. Finally, CMLL and EMLL are integrated to obtain the final result. We evaluate our model on a benchmark emotion cause corpus, the results show that our approach achieves the best performance among all compared systems on the ECPE task. | ['Jianfei Yu', 'Rui Xia', 'Zixiang Ding'] | null | null | null | null | emnlp-2020-11 | ['emotion-cause-pair-extraction'] | ['natural-language-processing'] | [ 1.11145951e-01 -1.95852946e-02 -1.98217720e-01 -5.33718169e-01
-1.03905821e+00 -6.92144394e-01 4.19463605e-01 4.01126772e-01
-1.90883949e-01 5.99195659e-01 2.74246544e-01 1.47316888e-01
-8.67415071e-02 -3.44151765e-01 -2.14944690e-01 -8.71689796e-01
-3.27672921e-02 2.90676266e-01 -9.88738611e-02 1.70393407e-01
-7.79298833e-03 1.32517740e-01 -1.69431412e+00 7.43678033e-01
8.19295943e-01 1.11902976e+00 -1.55446678e-01 4.67511833e-01
-4.94671077e-01 1.03090441e+00 -7.41656482e-01 -3.63584459e-01
-2.57574439e-01 -6.26731455e-01 -9.37528551e-01 -1.73847660e-01
-1.74805209e-01 9.65496749e-02 5.96976638e-01 1.07635796e+00
4.89137828e-01 -1.85314938e-01 8.26245010e-01 -1.72674966e+00
-1.13204472e-01 6.54017985e-01 -6.66700363e-01 -2.30158806e-01
6.08095765e-01 -6.53831244e-01 1.36204803e+00 -8.68294239e-01
6.37247801e-01 1.31706655e+00 3.95632625e-01 5.07195413e-01
-5.68954289e-01 -8.49385381e-01 4.13028330e-01 3.06740582e-01
-1.13654900e+00 -5.68542369e-02 9.45419908e-01 -1.85160726e-01
9.87095475e-01 2.99622685e-01 5.50780773e-01 8.56773853e-01
2.89659917e-01 1.10783482e+00 1.26194823e+00 -3.93793583e-01
2.68667459e-01 1.67419270e-01 4.85598117e-01 5.14043212e-01
-4.09893095e-01 -2.92368859e-01 -4.74410951e-01 -4.97350156e-01
3.36852893e-02 -2.94619888e-01 -2.35218629e-01 1.70583114e-01
-8.82026851e-01 7.17979848e-01 -1.54268956e-02 4.19125497e-01
-4.20062125e-01 -7.35551715e-02 7.14689791e-01 2.71164238e-01
5.15956461e-01 2.13800088e-01 -9.22251701e-01 -7.30271861e-02
-8.07789683e-01 4.23768133e-01 1.00362813e+00 7.30667710e-01
6.82167351e-01 -6.53743148e-01 -1.09913819e-01 8.12418759e-01
2.70489186e-01 2.57757276e-01 2.70179003e-01 -5.15184522e-01
2.17870504e-01 8.75052750e-01 1.93771333e-01 -1.00423670e+00
-6.56430781e-01 -3.17250222e-01 -5.72651684e-01 -1.80171162e-01
7.05767348e-02 -6.24342263e-01 -6.45671189e-01 1.77188540e+00
6.40168786e-01 3.59794199e-01 4.53374565e-01 8.97222817e-01
1.36342013e+00 9.71577227e-01 3.44861418e-01 -8.02770913e-01
1.68341744e+00 -9.52562809e-01 -1.24274743e+00 -2.89936662e-01
8.55659366e-01 -8.80541384e-01 7.26335347e-01 6.33015096e-01
-7.53349662e-01 -8.01600292e-02 -8.15870643e-01 1.96600263e-03
-3.75304729e-01 4.26213056e-01 7.17834353e-01 8.13098773e-02
-4.38752443e-01 1.40827686e-01 -2.59927869e-01 -4.87403534e-02
-1.63618565e-01 3.83627713e-01 -2.86989212e-01 1.39179230e-01
-1.71490932e+00 7.35428691e-01 4.26384211e-01 1.35772020e-01
-4.08761859e-01 -4.56008464e-01 -7.41837263e-01 9.74793062e-02
3.89147580e-01 -1.79363593e-01 1.08320582e+00 -1.22481775e+00
-9.45619702e-01 9.27633643e-01 -4.89058852e-01 1.13599740e-01
-4.51839603e-02 -4.88113552e-01 -7.33636081e-01 -3.78666967e-02
1.82128161e-01 5.16488254e-01 6.93381727e-01 -1.45529366e+00
-1.06894147e+00 -1.97216690e-01 -1.72449991e-01 4.59403247e-01
-9.11599472e-02 6.16008162e-01 -7.49549627e-01 -2.00362667e-01
5.10795265e-02 -6.74733162e-01 6.85584024e-02 -7.23506093e-01
-4.83589441e-01 -9.66588974e-01 1.14954960e+00 -5.01741409e-01
1.58324909e+00 -2.44565487e+00 1.48429468e-01 1.75076723e-01
2.05334216e-01 -4.56644408e-02 -9.03745294e-02 2.84600854e-01
-4.83407229e-01 6.15684912e-02 3.84073332e-02 -2.16458961e-01
1.75137177e-01 2.05212593e-01 -4.84428495e-01 7.68171623e-02
3.49629700e-01 6.05829477e-01 -9.50152457e-01 -9.44269896e-01
-1.18750386e-01 2.62553751e-01 -2.80039698e-01 5.15609324e-01
-2.66741753e-01 3.78546342e-02 -5.82055688e-01 5.50497651e-01
6.03520572e-01 7.29939155e-03 2.21576408e-01 -5.38253963e-01
-4.97422703e-02 3.72559726e-01 -1.51249766e+00 8.98566186e-01
-2.70430744e-01 7.09852204e-02 2.17552409e-01 -5.15146077e-01
1.13356924e+00 7.09268749e-01 7.18696296e-01 -3.76177967e-01
2.19183996e-01 2.90525198e-01 -1.35366753e-01 -8.63545954e-01
1.64469644e-01 -4.09364820e-01 -6.89960778e-01 5.96901774e-01
2.03039926e-02 -1.24369457e-01 3.03796589e-01 1.32191107e-01
1.08556485e+00 6.22712970e-02 5.33970296e-01 5.29668331e-02
6.92306221e-01 -1.40022442e-01 1.14574289e+00 7.81398937e-02
-4.23696637e-01 1.72719419e-01 1.16963625e+00 -3.71760994e-01
-3.78129482e-01 -7.62750089e-01 5.09322174e-02 1.07142329e+00
2.26527601e-01 -6.49399459e-01 -5.32112420e-01 -1.16425169e+00
-2.89635897e-01 7.65125096e-01 -5.26958048e-01 -5.65055124e-02
-5.05125523e-01 -8.85415733e-01 4.44975197e-01 1.94479764e-01
4.06452447e-01 -1.33809507e+00 -4.91814792e-01 2.32245669e-01
-7.51451671e-01 -1.14138234e+00 -8.57462287e-02 6.61021173e-01
-7.80778751e-02 -1.21362579e+00 -9.94835421e-02 -1.10378325e+00
5.63881993e-01 -4.16802227e-01 1.03966224e+00 1.02761555e-02
1.24374963e-03 7.85465073e-03 -5.18343449e-01 -6.06634915e-01
-2.07555249e-01 -1.09241188e-01 -2.04441324e-01 1.68573588e-01
7.01473236e-01 -3.31212014e-01 -1.12086080e-01 9.83141586e-02
-5.63487172e-01 1.84424847e-01 4.65175778e-01 5.63405335e-01
7.45467603e-01 4.61985022e-01 7.97839999e-01 -9.57323909e-01
9.81475711e-01 -5.61434507e-01 -1.35282710e-01 4.86916602e-01
-4.52325642e-01 -1.43988609e-01 5.85936308e-01 -3.76148164e-01
-1.21316445e+00 3.64888668e-01 -1.81874663e-01 -1.57639056e-01
-4.69459176e-01 6.56659961e-01 -5.61200738e-01 6.45406365e-01
1.31658306e-02 -8.31269473e-02 -6.51434958e-01 -1.35934457e-01
3.55725884e-01 7.63681948e-01 5.16864181e-01 -4.37845945e-01
2.84369081e-01 1.23343933e-02 -1.88093320e-01 -1.33535430e-01
-1.16814351e+00 -5.96648514e-01 -4.72732842e-01 -4.63677704e-01
1.10599339e+00 -8.61527741e-01 -7.04534769e-01 3.57365102e-01
-1.49220848e+00 1.41943872e-01 1.77507743e-01 5.07275224e-01
-2.88955361e-01 5.26569150e-02 -8.36178422e-01 -9.53284979e-01
-4.69672561e-01 -9.26877141e-01 1.24637091e+00 3.05336058e-01
-7.19803393e-01 -8.20531011e-01 5.74849434e-02 1.11119673e-02
-2.73347497e-01 3.88137370e-01 1.43336844e+00 -9.67313826e-01
3.48369516e-02 -2.50243217e-01 -3.03768069e-01 -1.40882581e-02
2.63906658e-01 2.18585804e-01 -8.96067142e-01 2.26666525e-01
1.93306774e-01 -6.86709344e-01 5.23327470e-01 1.40349567e-01
7.77349830e-01 -2.29130521e-01 -4.71306562e-01 1.58910185e-01
1.33890510e+00 4.70177323e-01 3.27230871e-01 -5.59805799e-03
5.10311186e-01 1.00079608e+00 1.05797124e+00 4.33352679e-01
5.68229020e-01 4.13204640e-01 3.33687782e-01 -4.71165568e-01
2.89519638e-01 7.58027285e-02 4.09723848e-01 9.25846934e-01
2.26265252e-01 -3.45769256e-01 -7.89415598e-01 5.15271366e-01
-1.92907894e+00 -7.60305166e-01 -5.49613297e-01 1.51023388e+00
1.10329616e+00 1.25763295e-02 -1.41027242e-01 2.51034826e-01
7.46446609e-01 1.68455020e-01 -3.78384531e-01 -7.14317799e-01
-1.79766551e-01 -9.53068808e-02 -3.24513555e-01 4.19523120e-01
-1.35924792e+00 1.03816092e+00 6.10482979e+00 9.25129354e-01
-9.59530294e-01 7.52758831e-02 6.29816771e-01 -1.45585090e-01
-3.17022324e-01 2.16214228e-02 -9.18135107e-01 2.45754853e-01
5.27131736e-01 1.94129646e-02 6.64891154e-02 8.21512401e-01
1.77125186e-01 -2.18195647e-01 -1.21474087e+00 1.04700625e+00
1.53455690e-01 -5.44661462e-01 -2.72098660e-01 -3.95512849e-01
5.96537352e-01 -3.77103567e-01 -3.95851940e-01 4.93423909e-01
1.36822775e-01 -7.71005690e-01 6.11747444e-01 3.54475945e-01
4.38324630e-01 -1.18413937e+00 1.08831549e+00 2.78216511e-01
-1.41377580e+00 5.19898534e-02 -2.15085633e-02 -9.76835843e-03
3.35066199e-01 1.07584941e+00 -3.95964533e-01 5.92469513e-01
7.33702719e-01 6.24887824e-01 -2.99284846e-01 6.23077512e-01
-8.21370184e-01 6.42964125e-01 -1.71003044e-01 -2.48709336e-01
9.62305889e-02 6.48569316e-02 4.79151547e-01 1.45749819e+00
1.14494815e-01 3.00828278e-01 3.38957965e-01 7.77481496e-01
-4.58607562e-02 5.17367542e-01 -1.14334509e-01 1.16288669e-01
4.68122452e-01 1.79250956e+00 -8.86534631e-01 -4.37918961e-01
-3.21496874e-01 1.00395942e+00 4.82556164e-01 2.68828958e-01
-1.07389688e+00 -5.97917795e-01 2.14002490e-01 -7.14491487e-01
1.16779394e-01 3.54550540e-01 -1.80265337e-01 -9.22009826e-01
1.77637860e-01 -9.81860757e-01 6.03758872e-01 -7.51705170e-01
-1.41121280e+00 7.39158034e-01 -1.09456979e-01 -9.36859250e-01
-4.93567824e-01 -3.47507983e-01 -9.12239492e-01 6.97318196e-01
-1.43720186e+00 -1.08760679e+00 -9.17064101e-02 5.60478568e-01
2.21958444e-01 3.11986387e-01 1.05956972e+00 3.98122221e-01
-7.21745312e-01 3.75313789e-01 -6.11991465e-01 1.91515997e-01
1.07969630e+00 -1.27672696e+00 -4.43902850e-01 7.05562115e-01
-1.67499438e-01 3.14354181e-01 5.90114713e-01 -9.02573228e-01
-8.66143227e-01 -9.57423508e-01 1.76091135e+00 -1.72411352e-01
4.23578918e-01 -5.66506982e-01 -8.68121624e-01 3.77679259e-01
3.50613266e-01 -1.85335532e-01 9.48570490e-01 4.96494949e-01
-4.43577588e-01 -8.49266052e-02 -1.00977457e+00 4.60901618e-01
4.42801476e-01 -3.73206496e-01 -7.33468115e-01 3.10454875e-01
7.49797821e-01 -2.84170985e-01 -6.89503372e-01 5.69885135e-01
5.33846676e-01 -8.84127498e-01 4.10124272e-01 -3.99190724e-01
8.60307693e-01 -4.39279556e-01 4.29181680e-02 -1.18453312e+00
-1.73021942e-01 -4.37706858e-01 -6.31734682e-03 1.98458755e+00
7.23225594e-01 -1.95483327e-01 1.83453828e-01 7.60514975e-01
-4.04594131e-02 -1.14671171e+00 -8.08592439e-01 -2.42966995e-01
-2.71136045e-01 -5.90681732e-01 7.70647466e-01 1.23501527e+00
4.59909320e-01 7.23739445e-01 -4.37128931e-01 2.01216176e-01
2.68516660e-01 6.47778332e-01 2.41990566e-01 -1.06431353e+00
-1.65481389e-01 -4.45963949e-01 2.10415393e-01 -3.68223608e-01
5.07024944e-01 -7.08653331e-01 3.92600864e-01 -1.64015102e+00
4.03571457e-01 -4.21923578e-01 -2.97533453e-01 9.38289046e-01
-6.81080997e-01 -1.84896305e-01 2.73445621e-02 6.36489764e-02
-8.33853126e-01 4.11975771e-01 8.91141176e-01 1.99804291e-01
-3.61702889e-01 -1.84832111e-01 -5.93205869e-01 1.13198829e+00
6.12950861e-01 -6.83277667e-01 -2.77336717e-01 1.02639310e-01
5.98304451e-01 1.06433786e-01 6.25132322e-02 -5.30067205e-01
1.90385401e-01 -2.42562637e-01 2.42171466e-01 -8.21927726e-01
-7.85204321e-02 -8.47726703e-01 -1.90759271e-01 1.86886922e-01
-3.35244209e-01 -1.93412136e-02 1.63525030e-01 6.53594807e-02
-4.67131972e-01 -3.29258233e-01 4.63785231e-01 1.62050620e-01
-7.13073075e-01 -6.23369552e-02 -4.32943374e-01 1.12562058e-02
1.09581876e+00 3.53375435e-01 -1.59994990e-01 -3.25346291e-01
-7.18892455e-01 5.41748464e-01 -2.03452900e-01 5.12578130e-01
4.65871215e-01 -1.36840451e+00 -6.02417767e-01 -1.40663132e-01
6.98261410e-02 -8.54391381e-02 3.87746282e-02 8.72602999e-01
4.35056120e-01 1.42690748e-01 2.75945306e-01 -1.71043471e-01
-1.75598383e+00 4.99540359e-01 2.58068413e-01 -8.12587142e-01
-5.95389679e-02 8.97978425e-01 3.05847228e-01 -5.98713875e-01
2.98944414e-01 -1.25230625e-01 -7.40814626e-01 5.79207063e-01
3.94658208e-01 2.33024321e-02 -1.09953038e-01 -6.29237831e-01
-7.17730820e-01 5.60357094e-01 -7.39907622e-02 -8.89203846e-02
1.17663515e+00 -1.01368763e-01 -9.07159626e-01 6.30597770e-01
1.25145805e+00 1.78924143e-01 -6.17666543e-01 3.92449759e-02
1.71727806e-01 2.24335238e-01 2.73255482e-02 -1.11418200e+00
-8.60530555e-01 5.40422618e-01 3.19210738e-01 -3.28431390e-02
1.64999533e+00 3.53832066e-01 8.24186027e-01 -1.86700895e-02
8.82709175e-02 -1.26963592e+00 -1.26927048e-01 6.92803204e-01
8.89443636e-01 -9.25505579e-01 -1.04142845e-01 -9.11330462e-01
-7.23902822e-01 1.07726860e+00 8.38320792e-01 1.71881661e-01
5.57801723e-01 7.94149518e-01 4.36147988e-01 -4.97072846e-01
-1.15164089e+00 -2.48255894e-01 4.64355856e-01 5.11781797e-02
6.79368377e-01 9.85108688e-02 -8.31122160e-01 1.36271977e+00
-1.43312693e-01 -1.55231073e-01 1.93123911e-02 8.14281344e-01
-2.51625299e-01 -1.24640465e+00 -4.21945035e-01 3.28092724e-01
-6.84357643e-01 -1.29329652e-01 -9.94328678e-01 6.39690101e-01
8.60688984e-01 1.24744856e+00 -2.24965643e-02 -5.66246688e-01
2.23172799e-01 4.78594065e-01 1.50302509e-02 -4.83796924e-01
-9.76054668e-01 7.00946629e-01 4.17159319e-01 -5.49488485e-01
-7.20267713e-01 -7.02798545e-01 -2.03216577e+00 3.58105510e-01
-4.84913617e-01 5.31728864e-01 3.36313963e-01 1.47972870e+00
2.03900546e-01 5.91307044e-01 7.85483241e-01 -1.92544326e-01
1.25966787e-01 -6.72580779e-01 -6.10661685e-01 5.80362678e-01
2.03207672e-01 -4.28272724e-01 -5.40859342e-01 1.23388603e-01] | [12.641186714172363, 6.215455055236816] |
84b4cc44-7b7e-4ebf-aad6-b65bdb207bca | synthesizing-artistic-cinemagraphs-from-text | 2307.03190 | null | https://arxiv.org/abs/2307.03190v1 | https://arxiv.org/pdf/2307.03190v1.pdf | Synthesizing Artistic Cinemagraphs from Text | We introduce Artistic Cinemagraph, a fully automated method for creating cinemagraphs from text descriptions - an especially challenging task when prompts feature imaginary elements and artistic styles, given the complexity of interpreting the semantics and motions of these images. Existing single-image animation methods fall short on artistic inputs, and recent text-based video methods frequently introduce temporal inconsistencies, struggling to keep certain regions static. To address these challenges, we propose an idea of synthesizing image twins from a single text prompt - a pair of an artistic image and its pixel-aligned corresponding natural-looking twin. While the artistic image depicts the style and appearance detailed in our text prompt, the realistic counterpart greatly simplifies layout and motion analysis. Leveraging existing natural image and video datasets, we can accurately segment the realistic image and predict plausible motion given the semantic information. The predicted motion can then be transferred to the artistic image to create the final cinemagraph. Our method outperforms existing approaches in creating cinemagraphs for natural landscapes as well as artistic and other-worldly scenes, as validated by automated metrics and user studies. Finally, we demonstrate two extensions: animating existing paintings and controlling motion directions using text. | ['Jun-Yan Zhu', 'Sergey Tulyakov', 'Hsin-Ying Lee', 'Aliaksandr Siarohin', 'Aniruddha Mahapatra'] | 2023-07-06 | null | null | null | null | ['image-animation'] | ['computer-vision'] | [ 7.21119702e-01 -1.85503252e-02 1.71457231e-01 -2.91799545e-01
-3.03217173e-01 -1.07872450e+00 9.20337379e-01 -2.96827346e-01
2.84471631e-01 4.60997641e-01 3.65978330e-01 -6.25380427e-02
2.57002503e-01 -6.60486162e-01 -6.09212577e-01 -2.70752817e-01
8.72901604e-02 4.21452790e-01 4.32814747e-01 -5.22234499e-01
4.43262428e-01 3.74864399e-01 -1.60639584e+00 5.13308764e-01
6.05624020e-01 6.92186058e-01 3.96822810e-01 1.31136596e+00
-4.76992518e-01 1.00804794e+00 -7.30335593e-01 -5.47526121e-01
2.31719971e-01 -7.84965158e-01 -7.39897072e-01 4.66859013e-01
9.45452869e-01 -5.87910593e-01 -2.75666326e-01 9.15650845e-01
9.40293521e-02 1.84203889e-02 5.43758810e-01 -1.48316634e+00
-6.85354769e-01 3.59455854e-01 -6.81585848e-01 -1.19339064e-01
8.65045130e-01 4.55254495e-01 8.46563041e-01 -4.47960138e-01
1.29257381e+00 1.52651834e+00 4.94216740e-01 5.08431673e-01
-1.49048734e+00 -5.19242644e-01 3.44430774e-01 -9.28689614e-02
-9.89403427e-01 -5.22349954e-01 1.22142231e+00 -6.00570858e-01
3.91391814e-01 6.10394776e-01 1.26799917e+00 1.35938811e+00
-3.10500208e-02 9.10332561e-01 1.10740340e+00 -3.11072022e-01
2.75660098e-01 -2.88114101e-01 -5.99147320e-01 5.51708639e-01
-4.22687680e-01 -2.96829883e-02 -5.41273952e-01 3.78150516e-03
1.39486277e+00 -2.29202703e-01 -2.75160402e-01 -4.88658905e-01
-1.54323041e+00 3.54770601e-01 1.53821195e-02 1.12569675e-01
-3.44755799e-01 6.84615612e-01 2.20374137e-01 6.06917515e-02
3.37028772e-01 5.98916113e-01 8.50350782e-02 -5.73543251e-01
-1.46076214e+00 6.24436438e-01 6.41044140e-01 1.00571156e+00
4.54634517e-01 2.82426715e-01 -5.58220148e-02 4.64572251e-01
-8.31422806e-02 6.13628864e-01 4.16713394e-02 -1.55934584e+00
2.89725095e-01 4.44205374e-01 5.27442038e-01 -1.43737817e+00
-5.40728979e-02 2.97567874e-01 -6.57684326e-01 4.39941645e-01
6.43467844e-01 1.09297186e-01 -7.56937325e-01 1.39914227e+00
1.29755884e-01 2.06325382e-01 -3.54944587e-01 9.73990858e-01
4.97964501e-01 9.10217881e-01 7.91272894e-02 7.21740630e-03
1.36235559e+00 -9.92493927e-01 -9.39631283e-01 -4.36598659e-01
1.27403364e-01 -1.03025925e+00 1.45071292e+00 3.93332988e-01
-1.48945975e+00 -4.94102985e-01 -9.22978699e-01 -2.00170860e-01
-5.28472289e-02 -1.51180318e-02 5.19251406e-01 4.64864135e-01
-1.01746643e+00 7.03352690e-01 -5.58862627e-01 -1.56731427e-01
1.87210947e-01 -2.93630064e-01 -1.90670162e-01 9.35072601e-02
-8.80898833e-01 5.53982139e-01 1.21730357e-01 -8.27082619e-02
-7.10493863e-01 -9.18248951e-01 -7.42809772e-01 -1.68647438e-01
1.39323294e-01 -7.01948941e-01 1.44606411e+00 -1.68609166e+00
-1.73188114e+00 9.20863092e-01 -1.19897082e-01 -8.55810270e-02
1.03586626e+00 -8.82029384e-02 -5.73274970e-01 4.74462062e-01
1.01771630e-01 1.20459771e+00 1.01561880e+00 -1.61596680e+00
-6.76124096e-01 6.24818951e-02 3.25610042e-01 2.00328737e-01
-1.76270604e-01 -1.68756709e-01 -7.53792822e-01 -1.09304047e+00
-1.44000858e-01 -7.25961089e-01 -1.86990336e-01 5.70962489e-01
-5.75593829e-01 5.05679667e-01 1.34184861e+00 -7.74936080e-01
1.35548174e+00 -1.94108510e+00 1.93718880e-01 3.14389355e-02
1.22236036e-01 3.01084649e-02 -2.17014343e-01 4.91183907e-01
2.01000553e-02 1.59563184e-01 -1.63190410e-01 -4.55402941e-01
-7.18404949e-02 2.91431788e-02 -5.70809901e-01 8.20659846e-02
3.04728150e-02 1.08206451e+00 -1.30417132e+00 -6.64145052e-01
6.52805269e-01 4.56472993e-01 -5.15786707e-01 1.81410909e-01
-7.29229033e-01 6.99307144e-01 -1.62432462e-01 5.32605231e-01
6.57739103e-01 -1.74197346e-01 3.78507107e-01 -5.22022367e-01
-5.71286827e-02 -1.67763799e-01 -1.25195479e+00 1.97144115e+00
-4.94287938e-01 1.34661627e+00 -6.32757246e-02 -2.80693263e-01
8.71931970e-01 5.37868813e-02 4.64083850e-01 -9.55899656e-01
-6.33012429e-02 -1.33341067e-02 -5.92353582e-01 -8.13254178e-01
1.07906878e+00 1.00365318e-01 -2.11966798e-01 6.53475106e-01
-7.08592892e-01 -9.22938645e-01 3.23628962e-01 4.10863519e-01
7.39019632e-01 7.43364811e-01 1.13385334e-03 -2.28674952e-02
4.47272360e-02 4.44534332e-01 9.76928249e-02 5.33479631e-01
6.77839890e-02 1.16509593e+00 6.63945258e-01 -9.07351971e-01
-1.70373392e+00 -1.08302605e+00 4.33838457e-01 7.66997278e-01
6.11567080e-01 -6.47485912e-01 -9.80224490e-01 -3.43516409e-01
-2.23044887e-01 9.14072275e-01 -8.46880615e-01 3.47883582e-01
-8.43475044e-01 -1.78271547e-01 4.55095321e-01 4.67197686e-01
4.69022095e-01 -9.92028892e-01 -1.18057764e+00 1.92036614e-01
-5.07542074e-01 -1.21360838e+00 -9.32004750e-01 -5.75995982e-01
-5.64443409e-01 -1.03664279e+00 -8.57180059e-01 -7.52700448e-01
6.69696212e-01 3.95622939e-01 1.40707374e+00 -7.36345574e-02
-4.59071875e-01 4.51124936e-01 -2.25594327e-01 2.11075574e-01
-8.12067688e-01 -5.30015767e-01 -4.20592904e-01 9.19674262e-02
-4.74571913e-01 -8.10598731e-01 -1.03531516e+00 3.65057379e-01
-1.20446134e+00 1.22197235e+00 1.10495582e-01 5.04814804e-01
4.80792731e-01 -2.40416214e-01 -2.99590211e-02 -9.61137891e-01
4.68232840e-01 -3.81213464e-02 -4.27318364e-01 2.74730921e-01
-1.03537701e-01 5.12193237e-03 8.23304236e-01 -7.31021166e-01
-1.27060258e+00 3.21770906e-01 3.53255540e-01 -6.12650335e-01
-9.36485603e-02 -1.58408076e-01 1.98616404e-02 2.24326327e-01
6.62944615e-01 1.08909070e-01 -1.69379637e-01 -2.10399166e-01
9.71952260e-01 3.71574461e-01 1.16114604e+00 -6.09722853e-01
8.61450911e-01 8.11789036e-01 -2.31074125e-01 -7.64739990e-01
-4.31889385e-01 8.71505365e-02 -8.85436535e-01 -7.29931891e-01
9.37312961e-01 -6.20701075e-01 -5.46894550e-01 3.31135601e-01
-1.50482118e+00 -6.86953664e-01 -4.09638554e-01 -2.08923310e-01
-9.08178866e-01 3.70887667e-01 -3.56471360e-01 -7.73567438e-01
-2.53158584e-02 -1.22360814e+00 1.63339949e+00 9.23458710e-02
-8.73065829e-01 -1.12819552e+00 6.05995674e-03 2.47650206e-01
2.57181466e-01 9.73679602e-01 9.85616624e-01 4.94415581e-01
-7.04056799e-01 -1.49644434e-01 -2.15726510e-01 -2.43622005e-01
1.77713692e-01 6.63727820e-01 -8.46867085e-01 2.40351140e-01
-6.84198439e-01 -1.57769248e-01 2.65571833e-01 4.02707040e-01
1.19332016e+00 -4.58385706e-01 -1.68822899e-01 6.34167731e-01
1.33383620e+00 3.02491277e-01 7.83864081e-01 3.45711559e-01
9.94689822e-01 6.97176039e-01 6.02608621e-01 4.78879184e-01
2.99045771e-01 1.00914633e+00 3.75496417e-01 -3.06185782e-01
-4.75004256e-01 -8.77610028e-01 2.94386834e-01 4.18641269e-01
-9.79207605e-02 -3.51157248e-01 -7.53307521e-01 5.33253551e-01
-1.84866703e+00 -1.39263546e+00 -4.10099775e-01 1.91022933e+00
7.93443680e-01 -5.79349063e-02 2.83071578e-01 4.44227532e-02
5.55078328e-01 5.47832966e-01 -4.43486750e-01 -4.76442754e-01
-4.61566508e-01 -1.90349922e-01 3.42248261e-01 6.27354383e-01
-9.23552334e-01 1.16093802e+00 6.78635788e+00 7.43724525e-01
-1.05428433e+00 -1.16277352e-01 9.10409510e-01 -2.00065687e-01
-8.90300572e-01 2.74596661e-01 -9.00808796e-02 5.42414129e-01
4.73451376e-01 -1.49020806e-01 6.37322247e-01 5.76937556e-01
8.48830998e-01 -3.61591995e-01 -1.14382899e+00 1.26838398e+00
2.72280961e-01 -1.85953832e+00 3.27794224e-01 -3.34491402e-01
1.06441748e+00 -9.44916368e-01 1.32923409e-01 -4.18582529e-01
5.23999512e-01 -1.11435676e+00 1.44830668e+00 6.59939647e-01
1.14580858e+00 -4.22839195e-01 -1.76394463e-01 -1.55131416e-02
-1.35578156e+00 2.17077196e-01 2.71499634e-01 -9.87896621e-02
8.44261408e-01 5.91442436e-02 -5.72193325e-01 1.94276989e-01
6.05423987e-01 8.28701615e-01 -7.75041759e-01 6.38036788e-01
-9.39117521e-02 9.34343711e-02 2.19889749e-02 2.20725015e-02
2.68417120e-01 -3.67820114e-01 5.04352331e-01 1.39393938e+00
2.70852387e-01 -9.66076404e-02 -3.18569839e-02 1.24506640e+00
1.76731169e-01 -9.66303125e-02 -6.61657631e-01 -3.11805427e-01
5.76370895e-01 1.20355988e+00 -1.12610030e+00 -4.45269138e-01
1.41056702e-01 1.77016473e+00 -1.69037730e-01 5.77487946e-01
-9.48194921e-01 -2.58415729e-01 4.73359108e-01 3.39243829e-01
-3.84314395e-02 -3.84734839e-01 -4.64248717e-01 -1.00992262e+00
-1.02380030e-01 -9.08160686e-01 -2.34830260e-01 -1.61245084e+00
-6.84380829e-01 7.15328753e-01 1.51027292e-01 -1.47017741e+00
-4.93775219e-01 -1.56734169e-01 -8.63962829e-01 4.37139571e-01
-8.43464553e-01 -1.46884930e+00 -7.75735378e-01 2.43649229e-01
9.69487429e-01 4.12522644e-01 6.63335860e-01 1.32886767e-01
-2.70073056e-01 1.49202600e-01 1.40159413e-01 -8.82523805e-02
6.73931360e-01 -1.25972378e+00 1.05746043e+00 7.74315000e-01
2.06694528e-01 1.75737757e-02 1.15867484e+00 -6.44431829e-01
-1.39260960e+00 -9.83187020e-01 4.49605078e-01 -6.28207445e-01
5.78472733e-01 -6.93740010e-01 -5.24798214e-01 4.75947559e-01
3.40566844e-01 -2.34191224e-01 -1.07231047e-02 -7.10026920e-01
-2.59761691e-01 2.51670241e-01 -7.79219210e-01 1.28876936e+00
1.35625434e+00 -5.01995623e-01 -2.63433784e-01 3.13199371e-01
5.92768013e-01 -6.14523590e-01 -7.12975264e-01 -2.22299412e-01
1.11455262e+00 -1.14135730e+00 1.22237706e+00 -3.70701551e-01
9.99657094e-01 -4.65042561e-01 9.84030291e-02 -1.29250503e+00
9.60203111e-02 -1.20234704e+00 5.89851178e-02 1.13396454e+00
1.27657596e-02 3.73614699e-01 8.76245141e-01 7.21807599e-01
2.83436865e-01 -2.33641326e-01 -3.87620181e-01 -5.87093830e-01
-3.12461972e-01 -5.28697729e-01 6.64864957e-01 1.10701036e+00
-1.53863847e-01 5.81004731e-02 -8.42775524e-01 -3.24056119e-01
5.82744837e-01 4.10993278e-01 1.25488973e+00 -9.34523284e-01
-2.22202167e-01 -6.29722536e-01 -3.19211602e-01 -1.15760291e+00
-1.08447038e-01 -3.59888822e-01 -8.33380595e-02 -1.46891963e+00
-6.29540831e-02 -3.62843543e-01 9.08463359e-01 2.20993776e-02
-8.55641067e-02 6.38836920e-01 5.10462821e-01 3.27427030e-01
-5.49829900e-01 2.99082160e-01 1.69361711e+00 -1.91478297e-01
-4.92205054e-01 -5.66619635e-01 -2.73497373e-01 9.39468503e-01
5.07211447e-01 1.13392388e-02 -6.33219957e-01 -5.65258503e-01
2.86510706e-01 4.20586765e-01 6.87503219e-01 -8.88849437e-01
9.15180370e-02 -5.74031532e-01 6.78144753e-01 -6.17954373e-01
6.33911490e-01 -5.81429064e-01 8.61601233e-01 3.54326546e-01
-5.75934649e-01 5.35474181e-01 1.23772532e-01 6.71835005e-01
4.56781760e-02 1.55314863e-01 7.03798294e-01 -1.87039748e-01
-9.85446990e-01 1.17182575e-01 -3.64783227e-01 1.63433909e-01
9.67441201e-01 -7.51027703e-01 -5.58394611e-01 -9.67321515e-01
-6.26110137e-01 5.71888126e-02 1.06796885e+00 7.56088972e-01
8.11892867e-01 -1.47559416e+00 -3.69035780e-01 1.71033323e-01
3.15982997e-02 -5.45608103e-02 5.61014593e-01 4.96605277e-01
-1.27643144e+00 -5.00262082e-02 -5.44110775e-01 -7.43157268e-01
-1.26877987e+00 3.51451457e-01 3.02838981e-01 2.64580548e-01
-1.01500666e+00 2.98483640e-01 6.25697970e-01 1.67641342e-01
9.23493654e-02 -4.48079973e-01 1.60790950e-01 -6.82120547e-02
6.42270744e-01 2.75090069e-01 -4.56377387e-01 -7.74303317e-01
8.28024521e-02 7.74278820e-01 2.76708305e-01 -7.32795835e-01
1.12564290e+00 -4.48148966e-01 3.16214025e-01 4.33975846e-01
1.00066030e+00 2.00284541e-01 -1.92963970e+00 2.15090811e-01
-2.33101964e-01 -8.37898254e-01 -1.56965926e-01 -7.83408999e-01
-9.52380121e-01 9.22674954e-01 5.01152158e-01 1.28450319e-01
1.02039075e+00 -2.84186691e-01 9.11369920e-01 -2.22676411e-01
2.33955994e-01 -1.13228142e+00 5.91433108e-01 -1.07078282e-02
1.04846656e+00 -9.05667067e-01 -1.67911813e-01 -4.43099111e-01
-9.96686757e-01 1.38914502e+00 5.93306959e-01 -8.42468068e-02
1.09281614e-01 4.60068256e-01 1.85102016e-01 -2.15158105e-01
-5.33991396e-01 2.41579026e-01 2.84132510e-01 7.40233660e-01
2.05499649e-01 3.29095796e-02 2.35289320e-01 2.12914005e-01
-5.24100244e-01 -8.20793360e-02 8.79301965e-01 8.13166499e-01
-7.32904077e-02 -9.42634463e-01 -5.88076055e-01 -3.07267942e-02
5.12735993e-02 -2.30797548e-02 -7.91344702e-01 8.07931066e-01
-3.28362286e-02 6.96913540e-01 3.94769102e-01 -4.29708272e-01
2.60788709e-01 -2.98152417e-01 5.90798557e-01 -1.21505037e-01
-3.86859328e-01 1.54342100e-01 2.44002759e-01 -8.50791454e-01
-5.92685342e-01 -5.23422599e-01 -9.56478834e-01 -6.31519258e-01
3.11087608e-01 -2.76798010e-01 7.31571555e-01 5.51975965e-01
2.87205964e-01 5.49941182e-01 4.21197027e-01 -1.53165901e+00
3.24171633e-01 -4.76192415e-01 -3.01635176e-01 1.05125785e+00
3.30490410e-01 -3.24689984e-01 -1.46984398e-01 8.63881171e-01] | [11.170757293701172, 0.04060830920934677] |
04d84b37-eb8f-4824-8a9c-23ff50de9b05 | h4d-human-4d-modeling-by-learning-neural | 2203.01247 | null | https://arxiv.org/abs/2203.01247v2 | https://arxiv.org/pdf/2203.01247v2.pdf | H4D: Human 4D Modeling by Learning Neural Compositional Representation | Despite the impressive results achieved by deep learning based 3D reconstruction, the techniques of directly learning to model 4D human captures with detailed geometry have been less studied. This work presents a novel framework that can effectively learn a compact and compositional representation for dynamic human by exploiting the human body prior from the widely used SMPL parametric model. Particularly, our representation, named H4D, represents a dynamic 3D human over a temporal span with the SMPL parameters of shape and initial pose, and latent codes encoding motion and auxiliary information. A simple yet effective linear motion model is proposed to provide a rough and regularized motion estimation, followed by per-frame compensation for pose and geometry details with the residual encoded in the auxiliary code. Technically, we introduce novel GRU-based architectures to facilitate learning and improve the representation capability. Extensive experiments demonstrate our method is not only efficacy in recovering dynamic human with accurate motion and detailed geometry, but also amenable to various 4D human related tasks, including motion retargeting, motion completion and future prediction. Please check out the project page for video and code: https://boyanjiang.github.io/H4D/. | ['Yanwei Fu', 'xiangyang xue', 'Xingkui Wei', 'yinda zhang', 'Boyan Jiang'] | 2022-03-02 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Jiang_H4D_Human_4D_Modeling_by_Learning_Neural_Compositional_Representation_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Jiang_H4D_Human_4D_Modeling_by_Learning_Neural_Compositional_Representation_CVPR_2022_paper.pdf | cvpr-2022-1 | ['motion-retargeting'] | ['computer-vision'] | [-1.14768051e-01 7.64443725e-02 -2.87834585e-01 -5.04423864e-02
-5.40374935e-01 -2.26855546e-01 3.36226583e-01 -5.08286595e-01
-1.90376982e-01 3.29728842e-01 6.75620139e-01 3.52966815e-01
2.07359090e-01 -3.54050249e-01 -7.05397904e-01 -5.32491565e-01
-1.98879704e-01 6.56347096e-01 4.33058143e-02 -2.67990291e-01
-2.74203777e-01 5.01217544e-01 -1.33465719e+00 3.25681716e-02
4.65597957e-01 6.04920208e-01 2.98562050e-01 8.17363083e-01
5.33733964e-01 8.50918055e-01 -2.75542825e-01 -1.62377939e-01
5.23197591e-01 -5.02129316e-01 -8.00493658e-01 4.23310190e-01
4.47641581e-01 -8.27837288e-01 -1.05810475e+00 4.30580586e-01
8.33703637e-01 4.73256469e-01 5.72969735e-01 -9.47872460e-01
-4.82974559e-01 7.86006227e-02 -4.61351424e-01 -8.07508081e-02
8.60369265e-01 4.62751120e-01 6.50285840e-01 -9.30959582e-01
8.58890712e-01 1.47783947e+00 8.03044081e-01 9.04239595e-01
-1.05387855e+00 -3.85031760e-01 -6.78463653e-03 2.64025718e-01
-1.43605316e+00 -3.58033627e-01 9.61906135e-01 -6.23685896e-01
8.54468584e-01 1.87373996e-01 1.14716601e+00 1.28912961e+00
2.19066024e-01 1.19390666e+00 2.82504618e-01 -2.48276025e-01
-2.37333879e-01 -5.33638656e-01 -3.10536325e-01 1.12596869e+00
1.25410603e-02 2.98273027e-01 -5.93579471e-01 -2.09758710e-03
1.44219327e+00 2.45056346e-01 -5.57439864e-01 -8.69391918e-01
-1.62039459e+00 6.27994001e-01 3.92063588e-01 -6.67410716e-02
-4.57515478e-01 5.38057625e-01 2.47951090e-01 -7.62448832e-02
2.78104901e-01 6.73349351e-02 -3.48668456e-01 -9.83128324e-02
-8.21208060e-01 6.96413577e-01 4.57347840e-01 1.05049551e+00
5.70085168e-01 3.05530071e-01 -2.26032868e-01 6.76180422e-01
2.92906255e-01 7.59013355e-01 4.38374430e-01 -1.55891180e+00
4.56491411e-01 2.36904591e-01 2.74731159e-01 -1.18682969e+00
-6.94338262e-01 -3.62667888e-01 -1.13080418e+00 -2.72870902e-02
3.03678334e-01 1.13706430e-03 -9.50771093e-01 1.72606039e+00
4.79149282e-01 1.78938180e-01 -2.90226012e-01 1.25563443e+00
7.89850712e-01 6.37458742e-01 3.05502638e-02 1.72911093e-01
1.13556147e+00 -1.22137833e+00 -5.19403577e-01 -3.33001196e-01
7.33821988e-01 -5.28042614e-01 8.42775345e-01 4.60900031e-02
-1.29112029e+00 -8.53279829e-01 -8.43299031e-01 -5.55902183e-01
3.67630273e-01 3.04913372e-01 5.96033990e-01 1.28732890e-01
-1.01941073e+00 6.61722481e-01 -1.16277909e+00 -3.44679892e-01
1.15583159e-01 2.77612448e-01 -6.26845658e-01 -2.25229502e-01
-1.19156837e+00 8.28448355e-01 2.70112187e-01 2.83732235e-01
-1.10531604e+00 -6.42703295e-01 -1.28954554e+00 -2.71811664e-01
2.56999195e-01 -1.51688862e+00 1.17497802e+00 -4.16859537e-01
-1.61667192e+00 9.42490339e-01 -3.98978293e-02 -3.29442322e-01
8.16350996e-01 -6.77439690e-01 6.85260817e-02 5.31939685e-01
7.21652806e-02 9.67051268e-01 9.94287372e-01 -1.15026927e+00
-1.80061147e-01 -2.50691354e-01 -5.12102358e-02 5.24190068e-01
1.04627639e-01 -3.71083677e-01 -8.90481353e-01 -1.17232740e+00
2.26173565e-01 -1.22742307e+00 -4.33239251e-01 4.21255410e-01
-2.14779630e-01 1.09918371e-01 4.38027710e-01 -1.28006780e+00
1.17111576e+00 -1.91147768e+00 8.26977253e-01 -3.23453844e-02
2.92068124e-01 8.46345052e-02 -5.10671996e-02 1.64406627e-01
-4.77702171e-02 -2.50554681e-01 -4.33700502e-01 -5.92993975e-01
-3.02958339e-02 2.27118567e-01 -6.34462237e-02 6.59238875e-01
5.83829507e-02 1.27625883e+00 -9.28699374e-01 -3.82368267e-01
6.24582231e-01 8.85265946e-01 -8.23547244e-01 4.35748070e-01
-1.57763576e-03 1.12360394e+00 -4.95402992e-01 6.27878368e-01
3.70192707e-01 -3.79012138e-01 6.83190376e-02 -4.05003995e-01
2.79668331e-01 -8.70003849e-02 -1.08409083e+00 2.25545120e+00
-1.54689029e-01 3.03011328e-01 7.53873810e-02 -8.10711384e-01
6.63159132e-01 3.90323281e-01 9.47088182e-01 -3.65647703e-01
4.30829749e-02 6.84598684e-02 -5.57016134e-01 -6.90946460e-01
5.56200802e-01 -9.87635404e-02 -1.87016547e-01 1.47011623e-01
1.41890692e-02 -1.53174639e-01 -3.08031470e-01 9.39692557e-02
8.96056890e-01 8.29281092e-01 3.96773398e-01 6.98945075e-02
5.45356810e-01 -2.48120110e-02 5.53903043e-01 2.57225364e-01
-3.89141858e-01 1.03273344e+00 -7.81479105e-02 -6.10038579e-01
-1.37314641e+00 -1.12061203e+00 3.17937255e-01 7.57950127e-01
2.31048867e-01 -3.87699276e-01 -5.27925670e-01 -2.56635338e-01
8.91293138e-02 3.03388476e-01 -5.12337267e-01 -3.16416562e-01
-1.24328196e+00 -3.03941399e-01 4.01366144e-01 7.42988527e-01
6.17334187e-01 -8.44143271e-01 -7.55049527e-01 1.62032858e-01
-7.97636390e-01 -1.10886610e+00 -9.00240600e-01 -4.11558062e-01
-1.07460916e+00 -9.70621467e-01 -1.42450070e+00 -8.96154642e-01
5.45700729e-01 3.86775613e-01 9.97601449e-01 1.96788535e-01
-4.58410978e-01 7.04734743e-01 -2.90054172e-01 4.28760171e-01
-2.29389518e-01 -2.36192435e-01 2.51590550e-01 -2.17988625e-01
-2.55733192e-01 -6.27233207e-01 -1.00728321e+00 2.72211999e-01
-6.77616775e-01 5.25640666e-01 5.43487787e-01 8.03737879e-01
6.57027483e-01 -3.63867402e-01 -3.33607048e-02 -3.87706161e-01
-2.38695025e-01 -3.61208826e-01 8.09596330e-02 -1.04015708e-01
-8.48450586e-02 3.79443020e-02 2.85989702e-01 -3.46188635e-01
-1.01532638e+00 3.97009820e-01 -4.43186790e-01 -8.94617856e-01
-3.91876064e-02 8.65865424e-02 -2.80499697e-01 1.22686341e-01
5.30710101e-01 3.64733696e-01 1.48430705e-01 -6.88816309e-01
4.88065630e-01 -1.73521452e-02 1.00115514e+00 -5.56680739e-01
9.69891608e-01 7.28023887e-01 8.36112797e-02 -6.61739051e-01
-7.86154091e-01 -5.21045446e-01 -1.20512402e+00 -3.32015008e-01
1.09265101e+00 -1.46781754e+00 -4.49435890e-01 5.80454290e-01
-1.11419284e+00 -7.45918572e-01 -3.29941094e-01 7.15963721e-01
-1.15800548e+00 8.36843014e-01 -9.88725543e-01 -4.70824480e-01
-3.10197860e-01 -8.76298368e-01 1.39917505e+00 -3.00772429e-01
-5.31684816e-01 -9.34261680e-01 1.07818782e-01 5.37751734e-01
-7.94089511e-02 6.61107838e-01 6.54944241e-01 1.86615691e-01
-6.93615556e-01 -2.56602645e-01 3.03627133e-01 1.81615561e-01
-1.36204921e-02 -5.00560343e-01 -4.53614175e-01 -6.45563185e-01
-1.42962158e-01 -1.84440523e-01 8.65025342e-01 6.45061255e-01
1.00315475e+00 -3.24231118e-01 -3.97473812e-01 1.13661611e+00
1.01791728e+00 -3.24149162e-01 6.14374697e-01 3.34450185e-01
1.34995854e+00 6.76246703e-01 6.38591528e-01 7.38267004e-01
6.25665843e-01 9.69722986e-01 1.84862480e-01 4.10648547e-02
-6.54982388e-01 -7.09541321e-01 5.29228866e-01 9.41612244e-01
-5.89030981e-01 -3.14529426e-02 -8.06579351e-01 4.49615031e-01
-1.92779696e+00 -1.03668165e+00 1.12356074e-01 2.08401036e+00
5.46562135e-01 -2.44809449e-01 3.33920836e-01 -6.19094111e-02
6.21516168e-01 2.94076979e-01 -6.21274829e-01 4.28252578e-01
-7.91348219e-02 -8.99734646e-02 3.37979585e-01 6.85847640e-01
-1.04237485e+00 9.58202779e-01 6.27342463e+00 5.29965997e-01
-6.98858976e-01 9.06540528e-02 2.62773633e-01 -2.56676018e-01
-3.02350670e-01 -1.32997677e-01 -6.41717851e-01 2.39093050e-01
5.81106722e-01 7.95362401e-04 1.81637228e-01 7.25887358e-01
4.43333030e-01 3.00589234e-01 -1.08241582e+00 1.40693462e+00
2.02052742e-01 -1.17928958e+00 4.37650800e-01 7.45033920e-02
7.38389015e-01 -3.76214623e-01 4.16471856e-03 2.34971926e-01
-4.48156334e-02 -8.11665535e-01 1.17441905e+00 8.86805296e-01
1.03397489e+00 -6.85887456e-01 3.38123351e-01 4.52874601e-01
-1.46219671e+00 -7.36158490e-02 -2.78270364e-01 -8.61519873e-02
5.80878675e-01 1.02627233e-01 -2.48701781e-01 6.41222537e-01
7.36674726e-01 1.22764850e+00 -3.58848155e-01 8.48566413e-01
-3.00237328e-01 1.37257025e-01 -1.93591848e-01 6.20006323e-01
-1.26047567e-01 6.66844426e-03 9.21405077e-01 1.07096696e+00
5.31555712e-01 3.40846300e-01 4.36554432e-01 5.35802960e-01
1.37227356e-01 -5.33685759e-02 -5.51957965e-01 3.09453428e-01
8.15276280e-02 8.02897871e-01 -3.44301105e-01 -4.09944683e-01
-1.15142040e-01 1.46178794e+00 2.03639925e-01 4.65200901e-01
-8.41206789e-01 2.49967754e-01 5.81769288e-01 4.39716101e-01
3.85265261e-01 -8.43322873e-01 5.12724370e-02 -1.51842690e+00
1.00933880e-01 -6.70858920e-01 4.36022460e-01 -8.07728291e-01
-8.99949729e-01 3.03647280e-01 3.38476002e-01 -1.55887103e+00
-6.71007156e-01 -4.24994916e-01 -1.69891328e-01 6.52919114e-01
-1.03859401e+00 -1.42052758e+00 -4.97329712e-01 8.99574220e-01
7.18855858e-01 -8.19675028e-02 6.41903937e-01 3.79827619e-01
-3.55442017e-01 5.44721067e-01 -2.37863615e-01 2.15977222e-01
7.33349562e-01 -9.52063859e-01 6.02196693e-01 7.32593298e-01
-1.98181912e-01 4.26266015e-01 7.30910242e-01 -9.11920428e-01
-1.51926708e+00 -1.16105199e+00 6.03822231e-01 -7.26629198e-01
1.62066832e-01 -1.66039824e-01 -7.14444339e-01 9.42530453e-01
-3.60426635e-01 3.22508700e-02 4.71549988e-01 -4.38976645e-01
-1.67655692e-01 2.61298418e-01 -8.94563556e-01 7.17476606e-01
1.64436150e+00 -4.07904148e-01 -6.49144113e-01 3.02904278e-01
7.77803481e-01 -7.74059772e-01 -9.54889476e-01 6.61924958e-01
7.37822592e-01 -7.68893182e-01 1.45060229e+00 -4.04776722e-01
5.16377747e-01 -4.48352218e-01 -2.40825787e-01 -9.52793181e-01
-7.91247129e-01 -7.80544817e-01 -8.66671026e-01 5.83904624e-01
-3.21683019e-01 -6.64642900e-02 1.11434555e+00 5.94973028e-01
-3.08702618e-01 -7.53494084e-01 -7.93133438e-01 -6.52123690e-01
-5.76197682e-03 -3.28869909e-01 2.63816625e-01 6.82755470e-01
-3.35519105e-01 6.60057738e-02 -1.19896209e+00 1.35856956e-01
8.72828901e-01 2.22536419e-02 1.11275160e+00 -8.10317874e-01
-5.45153677e-01 -2.14962095e-01 -6.57896519e-01 -1.88908231e+00
1.62496269e-01 -8.36013436e-01 1.02909133e-01 -1.53094304e+00
1.85316667e-01 -9.14442446e-03 2.78151274e-01 3.18061233e-01
-2.33773008e-01 3.83145630e-01 4.25592363e-01 7.49443114e-01
-5.07456958e-01 1.00202179e+00 1.69455171e+00 -1.10351862e-02
-2.17355385e-01 7.44379461e-02 -2.04215556e-01 8.38500023e-01
4.75804925e-01 -2.28440799e-02 -3.76211196e-01 -5.92527568e-01
-2.05185637e-01 4.43218529e-01 9.82919574e-01 -1.20452213e+00
1.21252671e-01 -1.20421816e-02 7.67736673e-01 -8.64625335e-01
7.09097922e-01 -5.72005570e-01 5.68647981e-01 7.45108604e-01
-2.67610013e-01 9.69611555e-02 1.71298981e-02 8.66236567e-01
4.82672639e-02 2.89853930e-01 6.97585821e-01 -4.74009484e-01
-1.14936435e+00 8.79731119e-01 -1.54268876e-01 -6.98696449e-02
7.84013271e-01 -3.90625894e-01 3.59048456e-01 -6.82038009e-01
-1.28826594e+00 2.90806770e-01 7.47689009e-01 4.89257038e-01
9.91268456e-01 -1.68602717e+00 -8.01822960e-01 1.66327178e-01
4.95965406e-02 3.30920935e-01 8.13886344e-01 7.32030451e-01
-9.19657469e-01 4.22797978e-01 -4.31576461e-01 -7.96490252e-01
-9.88399744e-01 5.70123315e-01 4.43103641e-01 -2.41277054e-01
-1.40511084e+00 7.39887118e-01 5.08146346e-01 -4.68505681e-01
2.62089729e-01 -5.44454940e-02 5.12515232e-02 -4.86511618e-01
4.16821718e-01 5.74868321e-01 -5.04328966e-01 -1.15456104e+00
-2.85516679e-01 9.35519755e-01 4.16174293e-01 -7.15459585e-02
1.21524930e+00 -6.65500760e-01 2.24738240e-01 2.69038737e-01
1.32957578e+00 -2.29443178e-01 -1.93519008e+00 -2.76146978e-01
-3.29855561e-01 -6.37993515e-01 -4.68444914e-01 -2.37744823e-01
-1.01701915e+00 8.14854145e-01 4.91527498e-01 -6.81431472e-01
9.44559216e-01 1.01360686e-01 1.24108219e+00 2.78423399e-01
6.39551997e-01 -8.29368293e-01 4.09468263e-01 5.58608234e-01
1.20458841e+00 -1.04326916e+00 2.41013408e-01 -5.25190830e-01
-6.37717724e-01 1.01882386e+00 5.82111776e-01 -3.86664957e-01
6.32147372e-01 -2.56329924e-01 -1.32465973e-01 -1.15992792e-01
-4.69539493e-01 -1.59241334e-01 5.97062945e-01 8.56660962e-01
3.36793959e-01 7.32362494e-02 9.11453217e-02 4.17202473e-01
-3.24590236e-01 -2.86328271e-02 2.55063206e-01 7.97806978e-01
-3.05637926e-01 -8.48290205e-01 -4.22839522e-01 -8.05098116e-02
-9.67660025e-02 3.04766953e-01 7.58938491e-02 8.15368652e-01
6.20921031e-02 3.68458271e-01 -1.49303481e-01 -5.63725054e-01
4.99092221e-01 -7.12211877e-02 8.19977939e-01 -6.46114469e-01
7.99664855e-03 2.19955325e-01 -2.18934547e-02 -9.23129141e-01
-6.18828416e-01 -6.61146700e-01 -1.34865153e+00 -4.31156963e-01
3.59826684e-01 -2.98057973e-01 2.23939568e-02 8.00893188e-01
2.85107344e-01 3.77991438e-01 3.17711204e-01 -1.69915879e+00
-4.37474340e-01 -7.25468576e-01 -5.59652150e-01 7.69250154e-01
5.69783270e-01 -1.01531255e+00 -1.49234608e-01 3.86969298e-01] | [7.189009666442871, -0.653018057346344] |
0a6a81d0-6035-4dd7-9c7e-329926243d1d | understanding-interlocking-dynamics-of | 2110.13880 | null | https://arxiv.org/abs/2110.13880v1 | https://arxiv.org/pdf/2110.13880v1.pdf | Understanding Interlocking Dynamics of Cooperative Rationalization | Selective rationalization explains the prediction of complex neural networks by finding a small subset of the input that is sufficient to predict the neural model output. The selection mechanism is commonly integrated into the model itself by specifying a two-component cascaded system consisting of a rationale generator, which makes a binary selection of the input features (which is the rationale), and a predictor, which predicts the output based only on the selected features. The components are trained jointly to optimize prediction performance. In this paper, we reveal a major problem with such cooperative rationalization paradigm -- model interlocking. Interlocking arises when the predictor overfits to the features selected by the generator thus reinforcing the generator's selection even if the selected rationales are sub-optimal. The fundamental cause of the interlocking problem is that the rationalization objective to be minimized is concave with respect to the generator's selection policy. We propose a new rationalization framework, called A2R, which introduces a third component into the architecture, a predictor driven by soft attention as opposed to selection. The generator now realizes both soft and hard attention over the features and these are fed into the two different predictors. While the generator still seeks to support the original predictor performance, it also minimizes a gap between the two predictors. As we will show theoretically, since the attention-based predictor exhibits a better convexity property, A2R can overcome the concavity barrier. Our experiments on two synthetic benchmarks and two real datasets demonstrate that A2R can significantly alleviate the interlock problem and find explanations that better align with human judgments. We release our code at https://github.com/Gorov/Understanding_Interlocking. | ['Tommi S. Jaakkola', 'Shiyu Chang', 'Yang Zhang', 'Mo Yu'] | 2021-10-26 | null | http://proceedings.neurips.cc/paper/2021/hash/6a711a119a8a7a9f877b5f379bfe9ea2-Abstract.html | http://proceedings.neurips.cc/paper/2021/file/6a711a119a8a7a9f877b5f379bfe9ea2-Paper.pdf | neurips-2021-12 | ['hard-attention'] | ['methodology'] | [ 2.02499747e-01 6.96715057e-01 -4.41572905e-01 -4.99926060e-01
-5.79100430e-01 -3.78911257e-01 2.92529792e-01 -1.43444389e-01
-1.18452922e-01 5.94813585e-01 2.34504566e-01 -1.77738011e-01
2.74571814e-02 -6.12763941e-01 -9.08147037e-01 -7.22388327e-01
3.22090030e-01 5.19037127e-01 -1.99989579e-03 -2.46913150e-01
3.01669896e-01 2.52036215e-03 -1.60312319e+00 5.01674533e-01
9.43675697e-01 1.19790971e+00 3.99049193e-01 5.39435446e-01
2.73730874e-01 8.97651732e-01 -2.12020203e-01 -4.00547206e-01
3.74101371e-01 -4.85688508e-01 -7.10017860e-01 -2.27612615e-01
1.49232149e-01 3.34146693e-02 9.18125510e-02 9.03526485e-01
1.00926243e-01 -2.49522831e-02 5.96507847e-01 -1.43244767e+00
-8.28253806e-01 8.61873567e-01 -3.28325480e-01 -1.66668788e-01
-7.27566481e-02 2.50677586e-01 1.66300952e+00 -9.34595764e-01
2.32112557e-01 1.22049975e+00 5.77310741e-01 8.25852096e-01
-1.43743813e+00 -6.34323537e-01 4.98019665e-01 1.06175251e-01
-1.23287737e+00 -4.70423341e-01 8.51348937e-01 -5.42707086e-01
9.61331844e-01 3.77499253e-01 6.39667034e-01 1.01753342e+00
2.38249376e-01 8.23942780e-01 6.30901754e-01 -2.70134181e-01
2.77400702e-01 4.65016603e-01 3.31609219e-01 7.15932310e-01
1.39774606e-01 3.34826320e-01 -6.29527509e-01 -2.58482456e-01
7.46144414e-01 5.61044216e-02 -3.86031717e-01 -2.66822815e-01
-7.84238577e-01 9.82504249e-01 6.14092052e-01 -5.53785190e-02
-6.80083573e-01 1.98034465e-01 -1.38513610e-01 4.19097096e-01
1.91811070e-01 1.00202942e+00 -8.58795702e-01 1.56568348e-01
-6.25008225e-01 3.76898021e-01 6.68186545e-01 6.70613527e-01
9.47462976e-01 -2.13463623e-02 -2.10282236e-01 7.73906827e-01
5.55882394e-01 9.17367488e-02 6.72567785e-01 -7.88038433e-01
3.60737681e-01 1.01196170e+00 2.00714558e-01 -9.74865735e-01
-3.41991663e-01 -7.90622890e-01 -7.73914099e-01 1.47538960e-01
4.29192573e-01 -1.63847923e-01 -9.16372895e-01 2.32666111e+00
-1.01503640e-01 -7.43318945e-02 2.97403466e-02 1.00386524e+00
5.80999911e-01 5.52203655e-01 1.04773015e-01 -1.91535324e-01
1.02449811e+00 -1.18930709e+00 -1.54329985e-01 -6.44258618e-01
5.45216858e-01 -4.04486716e-01 1.34552467e+00 2.71135479e-01
-1.24508429e+00 -4.45036501e-01 -1.10865247e+00 7.89001398e-03
5.01860343e-02 2.77320981e-01 4.01930302e-01 1.30661041e-01
-1.09767044e+00 8.20438743e-01 -7.09990740e-01 -1.48101142e-02
3.06903034e-01 6.15105152e-01 -1.67521670e-01 4.40094709e-01
-1.27130079e+00 1.06873047e+00 1.94231927e-01 6.00083657e-02
-9.14166331e-01 -7.43059397e-01 -5.72208762e-01 5.93283653e-01
4.13994908e-01 -9.14641201e-01 1.43514311e+00 -1.52077031e+00
-1.47514033e+00 6.63055480e-01 -3.91297489e-01 -5.89031816e-01
5.44296920e-01 -3.19476277e-01 7.27649182e-02 -5.11311054e-01
1.09555751e-01 7.42147148e-01 1.05341184e+00 -1.20315278e+00
-6.27855539e-01 -2.17433736e-01 -1.73328161e-01 2.70247608e-01
-1.43621072e-01 -2.94255853e-01 -4.43896145e-01 -4.79457349e-01
9.52060223e-02 -1.01223910e+00 -4.28040534e-01 -3.77291560e-01
-7.34493673e-01 -4.05663788e-01 3.48062664e-01 -2.55774200e-01
1.44441664e+00 -1.97123706e+00 4.06808376e-01 3.14805359e-01
5.55870891e-01 -4.10375446e-02 -1.97375551e-01 1.78282991e-01
-5.11741817e-01 3.44681710e-01 -1.96485385e-01 -4.65824783e-01
-3.58695574e-02 1.07919820e-01 -6.76733553e-01 2.20220193e-01
4.70578939e-01 1.06135535e+00 -5.04449069e-01 9.05133504e-03
-3.52558434e-01 2.77683616e-01 -9.69073415e-01 5.27039647e-01
-3.40319633e-01 6.02925271e-02 -4.59122628e-01 3.43914747e-01
2.47694552e-01 -7.58433461e-01 1.57410249e-01 2.46371087e-02
-1.72565714e-01 5.64406991e-01 -1.05502987e+00 8.56974840e-01
-1.17707439e-01 3.54602396e-01 -1.01740085e-01 -9.86832976e-01
1.01047325e+00 2.12567136e-01 2.20997483e-01 -3.69453728e-01
1.99979439e-01 3.07331115e-01 3.09716046e-01 -3.11643451e-01
3.55585635e-01 -3.23451489e-01 1.52771950e-01 6.30460143e-01
-4.02045362e-02 2.81515539e-01 -1.42329812e-01 1.22328289e-01
9.24248993e-01 7.35401362e-03 5.53947985e-01 -2.77806789e-01
3.26993316e-01 4.20306101e-02 1.10992134e+00 9.23394620e-01
2.46890616e-02 7.56791532e-01 9.62523997e-01 -6.29080057e-01
-1.04327989e+00 -6.16301894e-01 3.23110074e-01 1.31267083e+00
9.30577144e-02 -5.14156282e-01 -6.01430774e-01 -8.26605678e-01
1.53721660e-01 8.76461148e-01 -1.20563281e+00 -4.46807504e-01
-5.12221038e-01 -7.29463279e-01 1.11552872e-01 6.46380901e-01
2.41803065e-01 -1.00164807e+00 -8.89247894e-01 9.61399451e-02
-6.59703836e-02 -4.32986856e-01 -5.31260431e-01 7.92291343e-01
-6.84900165e-01 -1.05692708e+00 -2.01378420e-01 -4.46659952e-01
7.54237592e-01 6.79636002e-02 1.27589965e+00 4.71371323e-01
3.20874363e-01 -1.92778960e-01 -2.24563107e-02 -5.38858771e-01
-4.15311545e-01 3.27694058e-01 3.74168530e-03 1.23503640e-01
1.50222465e-01 -4.87610132e-01 -5.64280033e-01 5.11668622e-01
-4.84030336e-01 6.51622355e-01 7.33872473e-01 1.18149948e+00
6.05336189e-01 -2.50419378e-01 6.58529818e-01 -1.00245774e+00
5.96712172e-01 -6.40687168e-01 -5.31493545e-01 3.13952774e-01
-1.08080339e+00 4.80784416e-01 9.13038433e-01 -5.11178970e-01
-7.96214044e-01 2.55724818e-01 4.00210172e-02 -5.19421458e-01
2.17831321e-02 5.93188345e-01 -2.74965495e-01 3.32311779e-01
6.69411719e-01 2.42316537e-02 2.38318034e-02 -4.98376340e-01
1.58075884e-01 2.45935813e-01 2.95860440e-01 -3.30738634e-01
6.25084043e-01 9.18401182e-02 -3.70203018e-01 -3.09023023e-01
-9.94784594e-01 -1.95429232e-02 -2.92506009e-01 -4.96339314e-02
6.69134140e-01 -7.32367814e-01 -7.73597956e-01 1.86678171e-01
-1.23243093e+00 -4.94686723e-01 -5.36457658e-01 2.36386389e-01
-6.27557278e-01 -4.78826553e-01 -2.68666983e-01 -7.63673544e-01
-3.03810894e-01 -1.34650314e+00 6.50488794e-01 3.35818350e-01
-5.42991638e-01 -7.75473952e-01 2.51364391e-02 6.51431456e-02
3.94041806e-01 -1.04178168e-01 1.29989982e+00 -1.10952222e+00
-4.75897491e-01 -1.76249981e-01 -8.49615932e-02 3.21821451e-01
-2.04970226e-01 6.00594170e-02 -1.08081865e+00 2.47679800e-02
2.34821439e-01 -2.97097594e-01 1.16469395e+00 5.55418253e-01
1.04869246e+00 -5.56424081e-01 -2.47255877e-01 6.18747413e-01
1.16083872e+00 2.73519903e-01 4.26299244e-01 3.49967301e-01
6.59625351e-01 5.85780621e-01 3.17006439e-01 3.20208937e-01
4.84133422e-01 6.41781092e-01 7.95919061e-01 -2.32038990e-01
1.78295255e-01 -5.99092424e-01 4.61549848e-01 4.08293098e-01
-7.73430541e-02 -1.86475322e-01 -9.81045485e-01 1.95528641e-01
-2.29178834e+00 -9.30030704e-01 5.66330962e-02 2.22629499e+00
9.35541749e-01 2.44137526e-01 8.76727253e-02 1.24022141e-02
6.38648450e-01 -7.55464286e-02 -1.03026950e+00 -4.25913930e-01
-1.21810071e-01 -2.53842086e-01 2.97732711e-01 7.26634443e-01
-8.42457712e-01 9.91996229e-01 6.22629023e+00 5.85581481e-01
-1.30719209e+00 -3.05554383e-02 1.06428635e+00 -1.90841988e-01
-6.21055067e-01 2.21080482e-01 -9.28259254e-01 2.95430869e-01
7.10876226e-01 -3.19363534e-01 5.05803585e-01 1.14010632e+00
3.79753709e-01 1.61256105e-01 -1.52323496e+00 6.34229481e-01
-2.04604924e-01 -1.53328526e+00 9.12547410e-02 5.35851121e-02
5.35173774e-01 1.21732652e-01 3.39359641e-01 3.37761045e-01
3.68809611e-01 -1.28139603e+00 1.17880738e+00 4.32354927e-01
4.74687487e-01 -5.49097955e-01 4.79738772e-01 6.09060824e-01
-7.86011457e-01 -5.22383273e-01 -3.34328145e-01 -3.36582601e-01
-1.66634455e-01 3.88706118e-01 -9.86564755e-01 -4.97944392e-02
4.55607951e-01 7.61672378e-01 -4.46574241e-01 6.28390253e-01
-5.16768634e-01 7.57737637e-01 -1.09533824e-01 6.23374730e-02
1.36762902e-01 -6.58628196e-02 6.06979132e-01 8.01809371e-01
-3.99236614e-03 3.87687124e-02 -8.35652184e-03 1.32909751e+00
-2.07960084e-01 3.95216234e-02 -3.58854383e-01 1.80444203e-03
3.77548963e-01 9.65923786e-01 -4.09222841e-01 -3.78877312e-01
-1.91844046e-01 5.26339471e-01 8.33825350e-01 5.39361417e-01
-9.29590046e-01 4.17363085e-02 7.28374958e-01 2.47644126e-01
1.01839699e-01 3.52261543e-01 -9.94955301e-01 -1.24295986e+00
-1.05052307e-01 -9.60838139e-01 3.21311146e-01 -6.61188304e-01
-1.10649860e+00 8.30542445e-01 -2.99510062e-01 -8.77644300e-01
-3.09979022e-01 -4.17457819e-01 -8.02952409e-01 1.07156491e+00
-1.34611130e+00 -8.91407311e-01 -1.23091146e-01 3.41612071e-01
6.43207073e-01 -2.02211037e-01 7.53431618e-01 -1.44424751e-01
-9.50377345e-01 7.36430228e-01 -3.48098934e-01 -3.55120152e-01
4.65408325e-01 -1.31979680e+00 2.86380440e-01 6.59130394e-01
-8.74535143e-02 8.86674702e-01 8.26409280e-01 -6.28704667e-01
-1.00714099e+00 -9.48716819e-01 1.14588571e+00 -4.34633553e-01
5.00377893e-01 -1.29319459e-01 -1.14281559e+00 7.98316300e-01
-2.67285500e-02 -4.35102999e-01 7.77854681e-01 3.18511218e-01
-5.13990343e-01 -1.19579257e-02 -8.03039730e-01 6.67948842e-01
7.52000332e-01 -3.67905870e-02 -5.10671675e-01 5.35575077e-02
6.70369804e-01 -1.10079229e-01 -3.03398073e-01 4.84421879e-01
8.03072393e-01 -1.04523158e+00 5.58661044e-01 -9.79256272e-01
8.77996385e-01 -8.12117383e-02 -1.84389636e-01 -1.48965287e+00
-6.69418097e-01 -7.35125184e-01 -1.31346613e-01 8.22740972e-01
1.06274486e+00 -8.65860045e-01 8.69308710e-01 1.08302271e+00
-1.43140242e-01 -1.36315525e+00 -6.92147911e-01 -4.47549760e-01
7.29611740e-02 -3.51521492e-01 7.46988595e-01 8.18529010e-01
5.26438504e-02 8.14602792e-01 -5.18698871e-01 -5.51564358e-02
3.62749457e-01 3.91105950e-01 5.68349123e-01 -1.25779819e+00
-6.28090262e-01 -8.47095788e-01 1.55939057e-01 -1.24266613e+00
7.01119825e-02 -9.40110028e-01 3.04846019e-01 -1.10681975e+00
4.57792103e-01 -7.10076272e-01 -4.80741203e-01 1.06939518e+00
-4.70136255e-01 -1.66060522e-01 3.58358353e-01 6.36004984e-01
-3.62840295e-01 6.45015478e-01 1.07327282e+00 1.51468873e-01
-6.84521019e-01 1.81598470e-01 -1.47899830e+00 1.08251977e+00
8.94692004e-01 -6.76130235e-01 -4.06619668e-01 -5.57253480e-01
5.10836244e-01 -4.76042926e-02 2.84179211e-01 -4.91195291e-01
2.41106495e-01 -4.48806643e-01 3.52912813e-01 -1.48940027e-01
2.08233625e-01 -5.59856653e-01 2.17144340e-01 4.76840168e-01
-9.92189348e-01 3.36722940e-01 -2.34913632e-01 4.14576739e-01
-2.46253330e-02 6.22034632e-03 1.01525688e+00 4.50363703e-04
-1.89732298e-01 2.24985048e-01 -1.22114100e-01 -5.51560000e-02
8.77149343e-01 -5.08740693e-02 -4.00976390e-01 -4.35411990e-01
-7.33641863e-01 4.99588937e-01 5.05187750e-01 3.36965531e-01
7.11425841e-01 -1.16104567e+00 -6.87395096e-01 5.28768420e-01
-1.08879820e-01 -5.63533716e-02 -1.87117442e-01 9.13651824e-01
1.66127324e-01 5.14322460e-01 -6.73072936e-04 -3.55822802e-01
-9.54302371e-01 2.45389894e-01 5.63979149e-01 -4.66222614e-01
-2.97882736e-01 1.19418287e+00 7.03255236e-01 -1.50076166e-01
1.63840011e-01 -3.51147383e-01 -2.45416924e-01 3.38263847e-02
3.47288132e-01 1.78413376e-01 -1.55337676e-01 -5.13435185e-01
-3.68403286e-01 2.58933216e-01 -2.45459601e-01 -6.48806468e-02
1.53623331e+00 3.04067582e-02 3.53300781e-03 4.87491608e-01
8.10709655e-01 -2.05798313e-01 -1.50903976e+00 -1.30436271e-01
8.05160552e-02 -2.68223166e-01 -6.75756782e-02 -9.46768224e-01
-1.06151867e+00 7.18992531e-01 2.38672234e-02 5.05529106e-01
1.10304117e+00 2.41199154e-02 4.72021222e-01 2.18181491e-01
-7.51100704e-02 -9.81420636e-01 8.26423988e-02 7.00857282e-01
1.33265734e+00 -1.19671273e+00 -4.18634862e-01 -1.98534146e-01
-1.10334682e+00 9.36838508e-01 9.29209650e-01 -3.28427166e-01
5.62318683e-01 1.63646936e-01 -9.16201994e-03 -1.80700570e-01
-1.68419814e+00 1.87836420e-02 3.20202202e-01 1.01904184e-01
4.13702756e-01 2.03555390e-01 -2.94014812e-01 1.37470341e+00
-4.36700612e-01 -1.57460064e-01 2.58084685e-01 3.78075600e-01
-5.63334167e-01 -9.64673638e-01 -1.95963979e-01 6.35987759e-01
-2.42411628e-01 -1.95562586e-01 -6.65272951e-01 4.56992418e-01
1.95987955e-01 8.34898114e-01 -8.06192867e-03 -9.23844397e-01
3.46254826e-01 1.90235302e-01 -1.52610376e-01 -8.11448514e-01
-8.44428003e-01 1.17669366e-01 -9.37329605e-02 -7.17191994e-01
2.00430915e-01 -5.39065301e-01 -1.24745119e+00 -6.64188862e-02
-4.30363476e-01 3.82825136e-01 4.07783836e-01 8.43758821e-01
5.66369414e-01 5.14740705e-01 7.94564068e-01 -6.31554425e-01
-9.44028258e-01 -7.91384578e-01 -2.89884031e-01 2.55048513e-01
5.32902718e-01 -6.77301168e-01 -5.07714868e-01 2.26814430e-02] | [9.001008987426758, 5.772383689880371] |
4e630bec-f707-4d47-9dc5-6ef4e9a58c7e | graph-and-graphon-neural-network-stability | 2010.12529 | null | https://arxiv.org/abs/2010.12529v4 | https://arxiv.org/pdf/2010.12529v4.pdf | Graph and graphon neural network stability | Graph neural networks (GNNs) are learning architectures that rely on knowledge of the graph structure to generate meaningful representations of large-scale network data. GNN stability is thus important as in real-world scenarios there are typically uncertainties associated with the graph. We analyze GNN stability using kernel objects called graphons. Graphons are both limits of convergent graph sequences and generating models for deterministic and stochastic graphs. Building upon the theory of graphon signal processing, we define graphon neural networks and analyze their stability to graphon perturbations. We then extend this analysis by interpreting the graphon neural network as a generating model for GNNs on deterministic and stochastic graphs instantiated from the original and perturbed graphons. We observe that GNNs are stable to graphon perturbations with a stability bound that decreases asymptotically with the size of the graph. This asymptotic behavior is further demonstrated in an experiment of movie recommendation. | ['Alejandro Ribeiro', 'Zhiyang Wang', 'Luana Ruiz'] | 2020-10-23 | null | null | null | null | ['movie-recommendation'] | ['miscellaneous'] | [ 1.32422596e-01 6.42631650e-01 2.85707563e-01 6.76591229e-03
8.14279318e-02 -7.97677517e-01 3.82403970e-01 1.82892784e-01
1.44627899e-01 3.78593624e-01 7.65883625e-02 -5.22000253e-01
-4.77307111e-01 -1.05293965e+00 -1.21908140e+00 -5.47222316e-01
-8.06444883e-01 3.45051140e-01 1.46338508e-01 -4.01344270e-01
-2.90320307e-01 5.89220583e-01 -1.00047994e+00 -1.29658818e-01
5.01798749e-01 6.87179685e-01 -1.76561221e-01 1.36305881e+00
2.49591038e-01 9.10788596e-01 -5.86558223e-01 -3.70072335e-01
7.13871717e-01 -6.45310879e-01 -4.38814551e-01 2.73054782e-02
3.54692221e-01 1.51970267e-01 -9.03678000e-01 1.76883328e+00
4.39413875e-01 3.19816023e-01 7.13391423e-01 -1.55248511e+00
-1.08525991e+00 1.39556611e+00 -1.14992246e-01 2.76602954e-01
1.29490480e-01 1.28665671e-01 1.09793675e+00 -2.42221743e-01
6.08247638e-01 1.34270895e+00 1.23348904e+00 4.93241280e-01
-1.71888804e+00 -4.96373475e-01 8.78176168e-02 -1.11663587e-01
-1.13894105e+00 -5.21394685e-02 8.95594180e-01 -5.49332798e-01
7.40964234e-01 1.18172713e-01 7.90320694e-01 1.39499593e+00
6.16921246e-01 3.57922077e-01 3.04964930e-01 -4.13610786e-01
5.11057556e-01 -4.44000751e-01 3.92418623e-01 7.69956529e-01
7.30821490e-01 3.90439391e-01 -7.91097507e-02 -3.48301306e-02
1.04578638e+00 -2.26934344e-01 -4.73589778e-01 -6.32042468e-01
-8.28795552e-01 9.45118725e-01 7.31131077e-01 1.95973083e-01
-5.33252180e-01 9.53009427e-01 3.71813565e-01 7.63890386e-01
1.97495788e-01 5.57478070e-01 1.28756776e-01 2.54937589e-01
-3.60629886e-01 -7.87075423e-03 1.13980651e+00 1.11243141e+00
5.24625301e-01 6.59931540e-01 2.38434091e-01 2.61122882e-01
3.90586331e-02 5.67229569e-01 4.74510163e-01 -7.28561759e-01
2.31111795e-01 6.33002579e-01 -4.18234617e-01 -1.42713141e+00
-8.36929142e-01 -6.65646017e-01 -1.23207700e+00 -9.39027071e-02
7.97065854e-01 -5.25434017e-01 -1.05225325e+00 2.06450844e+00
-2.47199878e-01 5.17539084e-01 2.08978895e-02 4.39778298e-01
6.98484242e-01 5.70591152e-01 -2.97816664e-01 1.84558015e-02
5.86109877e-01 -2.34685972e-01 -4.40777570e-01 -2.32111495e-02
6.08575344e-01 9.84265432e-02 9.00828719e-01 2.42176756e-01
-8.92514527e-01 -2.74337292e-01 -1.11460066e+00 4.26191360e-01
-2.80361891e-01 -5.59906721e-01 4.63725299e-01 7.65855670e-01
-1.60805452e+00 1.07230937e+00 -7.82084882e-01 -6.08132362e-01
-3.37543748e-02 2.97082901e-01 -3.01575422e-01 2.12616220e-01
-1.34409094e+00 4.12114501e-01 6.36992335e-01 4.46481884e-01
-1.10454142e+00 -5.31777263e-01 -9.18696225e-01 2.96977937e-01
1.94524974e-01 -4.97826189e-01 1.11423218e+00 -1.21876800e+00
-1.12446928e+00 2.83590406e-01 5.96879423e-01 -1.05950689e+00
4.33171868e-01 3.27653080e-01 -6.62966311e-01 3.07875484e-01
-3.10433209e-01 3.14412341e-02 1.10389304e+00 -9.59705472e-01
1.41899824e-01 -8.99341851e-02 1.41974330e-01 -2.46993706e-01
-7.38908574e-02 -4.17479366e-01 2.98438311e-01 -6.39296353e-01
2.62665927e-01 -1.17633450e+00 -5.21948338e-01 -5.12187004e-01
-7.60088444e-01 4.44117375e-02 4.38539565e-01 -3.63765866e-01
1.18363965e+00 -1.82459259e+00 1.49381250e-01 8.92588675e-01
8.75154436e-01 -6.90820143e-02 -4.49981660e-01 6.83235705e-01
-6.36701107e-01 2.89426565e-01 1.96924627e-01 3.90665501e-01
2.16375321e-01 1.37905434e-01 -5.26899219e-01 7.32797265e-01
-1.03237160e-01 1.28238618e+00 -1.11765671e+00 3.89621675e-01
-8.20742026e-02 1.54153913e-01 -5.04939437e-01 -4.34239358e-02
-5.16877651e-01 -8.51036087e-02 -1.59788966e-01 3.20431963e-02
2.55214423e-01 -7.18686759e-01 4.06781465e-01 -1.05381139e-01
7.71290004e-01 -3.70751888e-01 -1.13641465e+00 9.30736005e-01
3.92342098e-02 8.20138276e-01 -5.07330615e-03 -1.21259415e+00
8.77736032e-01 1.81586564e-01 3.05636287e-01 -8.47839117e-02
4.61758494e-01 -5.43298014e-03 4.37658995e-01 -4.01321501e-02
2.42614537e-01 -1.97478160e-01 -1.91635519e-01 7.88134038e-01
3.79215688e-01 -4.28775176e-02 3.28790367e-01 8.18525434e-01
1.70078623e+00 -6.70003057e-01 1.68214127e-01 -5.62528968e-01
7.62522370e-02 -5.09173155e-01 8.24258178e-02 1.26180673e+00
-3.84566896e-02 4.94216502e-01 1.18261611e+00 -4.57098514e-01
-1.18201566e+00 -1.57694149e+00 4.53541577e-01 7.16423273e-01
1.78464781e-02 -5.50399125e-01 -1.00843251e+00 -4.62009877e-01
1.59747139e-01 6.54335022e-01 -9.25000489e-01 -9.00711834e-01
-2.12456390e-01 -6.99316978e-01 7.55658209e-01 5.14717460e-01
-5.00325449e-02 -1.00672567e+00 7.74168298e-02 3.15904319e-01
3.35795492e-01 -1.11538982e+00 -6.48342788e-01 1.47608310e-01
-7.61735618e-01 -1.37906170e+00 -3.59087169e-01 -7.19416797e-01
8.54377985e-01 -7.11459517e-02 1.24891198e+00 1.01363845e-01
-6.12676702e-02 1.02675390e+00 -8.15105140e-02 -3.27513754e-01
-1.16431534e+00 -3.13615799e-02 4.77788568e-01 3.04939687e-01
-9.84309092e-02 -1.17645335e+00 -1.89346716e-01 4.50026542e-02
-1.12294090e+00 -2.72466213e-01 1.14480257e-01 6.30400181e-01
1.47808328e-01 4.53019977e-01 5.99184752e-01 -9.62283671e-01
1.42700577e+00 -6.93955898e-01 -9.85893726e-01 2.76677519e-01
-6.43034160e-01 4.13619906e-01 1.18764496e+00 -6.93714261e-01
-1.38223812e-01 -9.74319503e-02 4.86864954e-01 -7.61168599e-01
3.88550401e-01 8.15945327e-01 9.35382769e-02 -1.73408270e-01
1.34223068e+00 6.15227707e-02 1.68804705e-01 5.04381359e-02
7.77228594e-01 -5.56043908e-02 6.02107286e-01 -4.57987130e-01
9.97806311e-01 2.25497067e-01 4.89216417e-01 -1.10098374e+00
-3.27973276e-01 1.51117191e-01 -4.02120531e-01 -5.64704359e-01
4.84621406e-01 -4.52732801e-01 -9.95445728e-01 5.16943157e-01
-9.90286529e-01 -5.18505275e-01 -6.89224601e-01 3.68296742e-01
-6.22065425e-01 4.24170077e-01 -9.71981466e-01 -7.66259730e-01
-2.07766145e-01 -5.09922445e-01 4.53368217e-01 1.23949677e-01
-9.89249125e-02 -1.37523377e+00 2.51200318e-01 -8.04976165e-01
3.48907620e-01 5.74910104e-01 1.04710865e+00 -9.27792370e-01
-6.29801631e-01 -6.72364831e-01 -8.90659019e-02 3.92528594e-01
-2.28824660e-01 5.26477359e-02 -4.50843930e-01 -4.91429418e-01
-8.86192769e-02 -5.37016355e-02 6.31641746e-01 8.30852568e-01
6.25957489e-01 -4.87774163e-01 -6.59256503e-02 6.07474267e-01
1.42866230e+00 -4.30684984e-02 3.97971034e-01 -3.06969047e-01
9.51345861e-01 3.82617682e-01 -7.09144652e-01 1.21786311e-01
-1.46209821e-01 -1.02881961e-01 4.65642869e-01 3.31866980e-01
-1.39688198e-02 -7.84996986e-01 5.36308825e-01 9.59520400e-01
-2.20371913e-02 -6.64748132e-01 -1.03791308e+00 3.30798090e-01
-1.90961647e+00 -1.08192039e+00 -1.76007047e-01 2.06947756e+00
1.14345744e-01 4.01107401e-01 2.44866654e-01 4.13755588e-02
1.32587373e+00 1.97459713e-01 -8.25306118e-01 -2.64470756e-01
-4.13917661e-01 1.28385872e-01 6.93977296e-01 6.93211377e-01
-7.17262983e-01 6.44759119e-01 7.25288963e+00 3.45558077e-01
-1.02850747e+00 -3.51040959e-01 4.52402204e-01 2.04077400e-02
-4.04050440e-01 -1.19350217e-01 -9.04975757e-02 3.12405348e-01
1.40769911e+00 -1.03791428e+00 8.60652328e-01 8.02697361e-01
1.09946504e-02 4.86542344e-01 -1.41144514e+00 8.92500877e-01
-1.75305545e-01 -1.35507786e+00 2.09507868e-01 2.10961536e-01
7.91930318e-01 2.96479970e-01 1.06457293e-01 3.48623604e-01
1.35355902e+00 -1.29665899e+00 5.68595767e-01 4.82253313e-01
6.40324235e-01 -6.43699944e-01 4.41083670e-01 2.30389044e-01
-1.26562583e+00 -1.32320493e-01 -5.61580002e-01 -2.45867353e-02
2.08216071e-01 7.90250123e-01 -9.36617911e-01 3.23054135e-01
2.13798910e-01 7.62475550e-01 -6.67418301e-01 7.78636396e-01
-2.58439839e-01 9.89190638e-01 -4.80788499e-01 -2.14218453e-01
1.29196331e-01 -5.34802556e-01 9.45701540e-01 7.62070477e-01
2.68309504e-01 -3.43250521e-02 -2.68733967e-03 1.12333190e+00
-3.71273696e-01 -2.08867222e-01 -1.03663135e+00 -7.66021132e-01
2.03045428e-01 1.00726402e+00 -1.08127081e+00 -3.84853818e-02
5.01317903e-03 7.82017291e-01 4.73004758e-01 8.30295384e-01
-6.83996975e-01 -6.35827661e-01 2.51922160e-01 1.52910993e-01
1.81723058e-01 -2.45551512e-01 1.35911599e-01 -1.19335866e+00
-5.96720204e-02 -7.37479925e-01 3.53987545e-01 -8.13197792e-01
-1.48139131e+00 9.22490358e-01 -3.43116820e-02 -9.91539836e-01
-6.11475408e-01 -8.00992846e-01 -7.44629323e-01 4.53014702e-01
-2.69951701e-01 -8.71226311e-01 -1.03559591e-01 6.43541753e-01
-4.09663320e-01 -2.87196159e-01 5.17603099e-01 -2.92761207e-01
-4.46815580e-01 4.71245527e-01 3.08543026e-01 4.53604639e-01
-2.22761244e-01 -1.57894337e+00 1.10857987e+00 1.34555638e+00
5.17515659e-01 7.38514662e-01 9.48538303e-01 -9.85788882e-01
-1.62222695e+00 -1.16644180e+00 1.39383465e-01 -5.72162807e-01
1.36408985e+00 -6.87606633e-01 -7.24024415e-01 1.17829192e+00
-2.54485875e-01 4.06594396e-01 1.89736024e-01 1.55351341e-01
-4.51855779e-01 7.73684010e-02 -9.38355744e-01 9.53734696e-01
1.47176898e+00 -7.25801885e-01 -1.55939683e-01 4.09862399e-01
1.00432062e+00 -2.44272321e-01 -6.46911681e-01 1.96717888e-01
4.00670409e-01 -9.99174476e-01 7.45916247e-01 -1.18754292e+00
8.72608200e-02 -1.46249115e-01 -1.15072131e-01 -1.68062401e+00
-5.88119447e-01 -1.31763673e+00 -4.37521130e-01 4.80201691e-01
4.37336892e-01 -7.86976576e-01 1.00061083e+00 5.84338546e-01
1.30238041e-01 -1.57937571e-01 -7.35983729e-01 -1.20583522e+00
5.00576794e-02 -7.21459866e-01 4.66687262e-01 6.67008519e-01
1.22991316e-01 5.18194258e-01 -1.62552059e-01 3.49338591e-01
8.66096854e-01 -1.96350947e-01 6.54448688e-01 -1.48104584e+00
-4.76751238e-01 -7.53588974e-01 -1.09865475e+00 -6.10182762e-01
3.07306677e-01 -1.33914661e+00 -4.18593027e-02 -1.19129324e+00
-2.22956583e-01 1.74397111e-01 -2.81815767e-01 -2.31708810e-01
1.68912888e-01 -8.88630971e-02 2.76239365e-01 -2.30415434e-01
-5.13303280e-01 3.26093882e-01 7.36756623e-01 1.53485462e-02
-2.42121667e-01 1.96633279e-01 -7.50555277e-01 6.59267485e-01
6.74166143e-01 -2.14639679e-01 -1.00410211e+00 5.64455539e-02
7.51816571e-01 -6.77822111e-03 4.05140042e-01 -1.13078320e+00
3.24040622e-01 2.25051299e-01 1.66024640e-01 -1.27606317e-01
-2.62502998e-01 -6.71977401e-01 6.20612919e-01 5.77126086e-01
-5.12354612e-01 3.69859576e-01 3.18208709e-02 1.22804010e+00
2.46214867e-01 -4.86901350e-04 7.92662740e-01 -9.43847224e-02
-1.79724261e-01 6.50949657e-01 -4.89285856e-01 4.04862612e-01
8.53576124e-01 -5.44674210e-02 -2.87808031e-01 -1.29084611e+00
-1.12906253e+00 4.75454591e-02 3.04397941e-01 1.91578373e-01
5.29340625e-01 -1.37412775e+00 -4.42772210e-01 4.62746531e-01
-1.81326389e-01 -2.14169949e-01 1.27124846e-01 4.35921907e-01
-6.72142684e-01 1.07789382e-01 -5.41265272e-02 -5.87471306e-01
-7.88984060e-01 9.02189910e-01 8.64848197e-01 -2.26421341e-01
-7.23295093e-01 8.75157893e-01 3.55201244e-01 -4.83365357e-01
3.04983079e-01 -8.69743288e-01 1.81342050e-01 -3.02055210e-01
3.15284550e-01 4.18976933e-01 -6.32135496e-02 -3.40439349e-01
7.51712918e-02 1.78708494e-01 1.31638825e-01 -5.55974208e-02
1.14448214e+00 7.27301240e-02 -2.09168464e-01 5.86636961e-01
1.08498788e+00 -9.33986232e-02 -1.00733447e+00 -1.64406180e-01
2.52807915e-01 2.09117442e-01 -3.39948416e-01 -1.74697429e-01
-1.07030964e+00 5.49238086e-01 3.36089402e-01 1.16178179e+00
8.07381392e-01 7.93606713e-02 4.48138654e-01 6.45048380e-01
2.30001479e-01 -8.29986513e-01 5.57497442e-02 6.69075370e-01
1.13712490e+00 -5.68624079e-01 -3.06402236e-01 -9.06353295e-02
-9.39532220e-02 1.25790858e+00 3.03051531e-01 -8.86602044e-01
1.23992515e+00 1.25596613e-01 -3.43461543e-01 -5.30848503e-01
-7.11347103e-01 1.34102538e-01 3.13766956e-01 8.51782620e-01
-1.86136931e-01 2.69717395e-01 2.87976235e-01 7.82031775e-01
-5.44673026e-01 -2.15400100e-01 1.18701506e+00 2.26773560e-01
-3.75834286e-01 -4.10074830e-01 -1.56684682e-01 7.69077897e-01
-6.39828444e-02 -2.66151339e-01 -7.11874425e-01 6.93629146e-01
-7.76795447e-01 6.61934435e-01 -4.98140864e-02 -9.33839440e-01
2.96394825e-01 -4.84728515e-02 5.60112536e-01 -6.33489788e-01
-1.86573133e-01 -4.20320302e-01 5.21243215e-02 -6.22217417e-01
1.56407043e-01 -4.16643739e-01 -1.02313602e+00 -6.07066929e-01
-3.34027469e-01 1.21350326e-01 3.00412774e-01 5.78198671e-01
4.66531038e-01 4.32801664e-01 4.06096995e-01 -7.84057796e-01
-9.22191322e-01 -9.47140396e-01 -1.24350798e+00 4.32170928e-01
4.89052832e-01 -2.17886686e-01 -7.79754460e-01 -1.81818590e-01] | [6.827754020690918, 6.060400485992432] |
c6ab8906-26b4-4eb4-b33d-dd38116b3488 | optimal-timing-for-power-plant-maintenance-in | 2302.00185 | null | https://arxiv.org/abs/2302.00185v3 | https://arxiv.org/pdf/2302.00185v3.pdf | Optimal timing for power plant maintenance in the Electricity Reliability Council of Texas in a changing climate | We analyzed data for the Electricity Reliability Council of Texas (ERCOT) to assess shoulder seasons -- that is, the 45 days of lowest total energy use and peak demand in the spring and fall -- and whether their occurrence has changed over time. Over the period 1996--2022, the shoulder seasons never started earlier than late March nor later than mid-October, corresponding well with the minimum of total degree days. In the temperature record 1959--2022, the minimum in degree days in the spring moved earlier, from early March to early February, and in the fall moved later, from early to mid-November. Warming temperatures might cause these minima in degree days to merge into a single annual minimum in December or January by the mid-2040s, a time when there is a non-trivial risk of 1-day record energy use and peak demand from winter storms. | ['Aidan Pyrcz', 'Michael E. Webber', 'Joshua D. Rhodes', 'Hugh Daigle'] | 2023-02-01 | null | null | null | null | ['total-energy'] | ['miscellaneous'] | [-3.34428340e-01 -8.59766603e-02 1.38341472e-01 1.00111449e-02
-2.81537592e-01 -8.44268143e-01 7.18351245e-01 2.74745554e-01
-2.72100180e-01 1.17754209e+00 1.88748036e-02 -7.81547487e-01
-3.42190713e-01 -8.71170640e-01 5.70029989e-02 -5.39770544e-01
-5.09905636e-01 1.98914617e-01 -2.22953081e-01 -5.68795085e-01
-3.81293334e-02 5.76828241e-01 -9.84338582e-01 -1.82283357e-01
9.00875390e-01 4.86384749e-01 -1.13958038e-01 1.99627534e-01
2.37906516e-01 5.23486435e-02 -5.79916954e-01 3.45884293e-01
3.32750261e-01 -5.21069050e-01 -3.07067633e-01 -4.18885708e-01
-5.21952868e-01 -6.10918701e-01 -2.72664458e-01 5.09301126e-01
9.56705883e-02 9.31848437e-02 4.01681691e-01 -1.44761086e+00
-1.63446497e-02 6.43720269e-01 -9.65115309e-01 3.71009231e-01
5.10334253e-01 2.76834637e-01 6.84844911e-01 -5.71330130e-01
2.05236763e-01 4.58732188e-01 6.83999002e-01 9.92459711e-03
-1.16794145e+00 -9.73686039e-01 -9.23590958e-02 -1.54481962e-01
-1.52471387e+00 -2.62250572e-01 6.88790143e-01 -3.65438700e-01
1.24028051e+00 4.94355828e-01 1.03904104e+00 2.65473276e-01
4.97208059e-01 4.03727740e-02 1.11962962e+00 -3.45370978e-01
4.95202899e-01 -1.21824361e-01 -1.23954125e-01 -1.56437248e-01
4.97601122e-01 1.70108646e-01 -2.75619924e-01 -4.81715560e-01
-6.75611198e-02 -1.97077051e-01 -4.29897040e-01 5.61899066e-01
-9.48294222e-01 5.15272975e-01 1.40010938e-01 6.12656951e-01
-4.82379258e-01 -5.73817194e-02 2.14300543e-01 3.18015218e-01
5.51299453e-01 2.45524794e-01 -8.47524524e-01 -4.83530521e-01
-1.51185668e+00 3.23429257e-01 7.81530499e-01 7.45756984e-01
6.24543548e-01 3.06696624e-01 5.13412714e-01 1.97012663e-01
2.33083531e-01 7.75075018e-01 -5.08744121e-02 -7.36457705e-01
5.56252182e-01 1.18178584e-01 6.82464302e-01 -2.42058098e-01
-8.65564704e-01 -5.18609643e-01 -6.96532607e-01 -9.37601924e-03
6.57211304e-01 -6.44648552e-01 -6.25157297e-01 1.79598653e+00
-1.89051479e-01 -1.35019794e-01 7.96932057e-02 5.26315928e-01
-1.11318059e-01 1.09807098e+00 1.08195551e-01 -1.21251392e+00
1.19170940e+00 1.02794856e-01 -5.59612393e-01 -3.49560291e-01
4.28497642e-01 -7.43090212e-01 4.46851790e-01 7.62602314e-02
-9.88886178e-01 -1.41608883e-02 -1.13980079e+00 7.88690448e-01
-3.38716686e-01 -4.89342600e-01 2.11286157e-01 6.51235700e-01
-8.16531479e-01 7.17194259e-01 -9.51472700e-01 -6.99041009e-01
-2.61904031e-01 -9.02259946e-02 1.01944640e-01 6.71021938e-01
-1.19994974e+00 1.02279234e+00 2.36430645e-01 4.74547327e-01
-2.79678375e-01 -9.39965487e-01 -5.61330259e-01 2.94039756e-01
-2.06995644e-02 4.64042909e-02 1.26934659e+00 -3.97481084e-01
-9.70991969e-01 2.21234605e-01 2.87237950e-02 -3.28417033e-01
9.14338827e-02 1.33033857e-01 -1.01708686e+00 -2.88826853e-01
3.58516648e-02 -8.52967948e-02 -2.39189550e-01 -1.04427528e+00
-6.33705735e-01 -4.91917849e-01 -1.05690777e-01 3.07257712e-01
4.36725244e-02 2.11217329e-01 2.40808129e-01 -3.14244330e-01
2.88114280e-01 -1.01901400e+00 -2.59001881e-01 -8.46712232e-01
-2.74338514e-01 -3.51058036e-01 8.14907730e-01 -6.52644157e-01
1.35712600e+00 -2.06328893e+00 -7.85950601e-01 5.13146222e-01
-2.15464145e-01 -3.64376575e-01 5.38442373e-01 1.04618263e+00
-5.14309227e-01 2.35583782e-01 -4.43843663e-01 4.73408192e-01
9.76379886e-02 3.85557741e-01 -4.77945298e-01 6.38922513e-01
3.28788757e-02 2.87909389e-01 -9.88255262e-01 4.32304680e-01
2.34929487e-01 -2.39803210e-01 2.35383913e-01 -2.22055897e-01
1.96055800e-01 1.18062526e-01 -1.37797669e-01 5.35692215e-01
1.11328328e+00 3.09272945e-01 3.41308832e-01 2.87705839e-01
-9.55728590e-01 5.61392248e-01 -9.01713908e-01 1.23058403e+00
-4.32264954e-01 7.71188378e-01 -2.14628913e-02 -7.35089302e-01
6.38730645e-01 4.33199763e-01 8.73270094e-01 -1.27344096e+00
-2.22975478e-01 5.13909042e-01 2.90483266e-01 -1.83612600e-01
6.65872931e-01 -1.35729641e-01 -3.05208713e-01 6.58903241e-01
-7.07032502e-01 -6.75550699e-01 6.32945299e-01 1.54988859e-02
9.38988328e-01 -2.74694264e-02 3.88944987e-03 -9.74964440e-01
-1.58925995e-01 3.30573320e-02 1.19527984e+00 -9.44993868e-02
-6.17052130e-02 2.92724222e-01 8.24435949e-01 -2.90380269e-01
-1.00597882e+00 -1.19986045e+00 -4.62949395e-01 5.02346277e-01
1.11915044e-01 -4.41233724e-01 -4.01954770e-01 -1.35905802e-01
8.90672556e-04 1.54974961e+00 -3.88191193e-01 3.13740559e-02
-5.02394795e-01 -1.23451722e+00 2.56527185e-01 1.85419649e-01
5.24992168e-01 -7.69270599e-01 -9.35998440e-01 4.01677191e-01
-3.20058405e-01 -7.02466607e-01 -3.96541685e-01 8.25599790e-01
-7.87326336e-01 -8.09889972e-01 -7.57554352e-01 7.36409193e-03
7.95726001e-01 -2.18291491e-01 1.34237921e+00 -2.03864962e-01
-6.01475872e-02 -2.09084541e-01 6.99069863e-03 -6.37155056e-01
-1.79100614e-02 -1.25964388e-01 2.65826821e-01 -8.18780422e-01
2.89786845e-01 -8.40796292e-01 -8.28338444e-01 2.82777518e-01
-5.57608902e-01 -1.99101344e-01 -2.77633866e-04 7.07108453e-02
-1.18424460e-01 2.50156760e-01 1.13838887e+00 -2.47677624e-01
2.30342343e-01 -8.56116116e-01 -8.41732025e-01 2.07533523e-01
-1.15181303e+00 -2.51591742e-01 5.05207658e-01 4.12972927e-01
-6.86145067e-01 -1.95671916e-01 7.20176473e-02 4.93294269e-01
1.07906610e-01 6.32522762e-01 1.67370558e-01 6.06218398e-01
3.20254594e-01 4.44382206e-02 -5.91139257e-01 -4.47787762e-01
-1.61095813e-01 8.10141623e-01 4.68807697e-01 -4.55915272e-01
1.11141634e+00 1.51854292e-01 -1.98231637e-03 -8.96715939e-01
-1.68251038e-01 -1.62514046e-01 -3.21155161e-01 -3.49483252e-01
7.25807369e-01 -1.06754243e+00 -5.63426375e-01 6.10990286e-01
-5.99009514e-01 -4.26315933e-01 -4.03470516e-01 5.21900952e-01
3.65814939e-02 2.29737282e-01 -4.28171962e-01 -1.10501266e+00
-2.57399440e-01 -5.98988771e-01 2.04586253e-01 7.44598150e-01
-4.77666348e-01 -7.18957007e-01 2.89831638e-01 -4.81930792e-01
3.67268264e-01 6.53461993e-01 1.01677847e+00 -2.16519386e-01
-2.42177825e-02 2.72356153e-01 1.15683796e-02 1.04197852e-01
5.34485102e-01 6.40471220e-01 -5.49744546e-01 -5.95661104e-01
-2.68819690e-01 3.10843349e-01 1.52877003e-01 3.70381683e-01
3.24911177e-01 2.21649799e-02 -4.32732016e-01 2.27679938e-01
1.79094219e+00 1.11182404e+00 6.25344217e-01 5.03443837e-01
-4.69631046e-01 2.47249176e-04 4.28275526e-01 8.04758728e-01
4.55097884e-01 4.35563058e-01 2.67637163e-01 -2.49573588e-01
5.50888896e-01 2.28273407e-01 1.87992528e-01 6.62855029e-01
-8.60918686e-02 -1.87357202e-01 -1.10645521e+00 1.25716376e+00
-1.37829661e+00 -7.08102047e-01 -4.32196617e-01 2.85683417e+00
7.67484128e-01 4.48586017e-01 -5.58370203e-02 -2.27235164e-02
2.86807925e-01 2.85250783e-01 -2.99372554e-01 -8.32421958e-01
-6.36139438e-02 1.75061777e-01 8.73690307e-01 4.80141759e-01
-4.11725402e-01 -3.64094526e-02 7.04421997e+00 -5.36506176e-02
-8.63585413e-01 -2.35916436e-01 5.72706699e-01 -8.12298432e-03
-7.92288303e-01 5.89773536e-01 -3.57863635e-01 8.33204567e-01
1.55878437e+00 -9.10258651e-01 4.09623951e-01 3.15227449e-01
9.53472018e-01 -9.78005230e-01 -7.78031290e-01 2.76620597e-01
-6.28960252e-01 -6.52121961e-01 -7.93042123e-01 4.01650071e-01
9.96466637e-01 1.41653985e-01 -5.87284029e-01 1.20108835e-01
2.46615410e-01 -5.86988509e-01 7.68838167e-01 5.27020752e-01
9.20360386e-01 -1.31459558e+00 4.68413234e-01 5.06894886e-01
-1.40975201e+00 -2.68111862e-02 -1.29324093e-01 -2.64491737e-01
4.02555704e-01 1.28178859e+00 -5.36115408e-01 1.11332309e+00
8.43971372e-01 6.69446960e-02 2.93316007e-01 6.99186444e-01
-1.20869443e-01 9.02749062e-01 -1.16802132e+00 2.81290948e-01
2.74403006e-01 -4.30523217e-01 4.07195628e-01 6.79320097e-01
5.69801986e-01 2.69394070e-01 -9.41792652e-02 5.81824660e-01
1.14986427e-01 -7.08859205e-01 -5.72479725e-01 1.02820285e-01
8.80419791e-01 1.05827451e+00 -6.92938030e-01 -1.12207055e-01
-6.29743457e-01 3.93058568e-01 -7.60223448e-01 6.26918018e-01
-9.81755316e-01 -9.99072134e-01 6.56664610e-01 3.79342824e-01
-3.44279379e-01 -4.15147781e-01 -5.98142207e-01 -7.70542502e-01
1.26720428e-01 -1.20458283e-01 1.26274610e-02 -7.82042682e-01
-7.39523947e-01 1.80667236e-01 3.61952603e-01 -1.15652120e+00
-7.91192710e-01 2.97837317e-01 -1.20940399e+00 1.20210791e+00
-1.19782400e+00 -3.70345473e-01 1.39318913e-01 1.72642514e-01
8.34299624e-02 1.98197916e-01 8.83583546e-01 2.98251510e-01
-4.33505327e-01 2.95970052e-01 1.01025355e+00 9.15000122e-03
2.45338112e-01 -1.25868678e+00 7.02797532e-01 9.11701977e-01
-4.58458453e-01 2.90203691e-01 1.00363839e+00 -7.54033566e-01
-9.95584190e-01 -4.91256326e-01 1.36026955e+00 2.18593657e-01
7.32633352e-01 -2.52565652e-01 -5.05202174e-01 5.95899403e-01
9.25594151e-01 -5.91607451e-01 4.75923926e-01 6.41899854e-02
3.62411261e-01 -4.69704539e-01 -1.10264397e+00 4.31769103e-01
1.94028497e-01 -6.82893693e-01 -4.35984701e-01 2.83838481e-01
-1.31917357e-01 -1.76845774e-01 -1.02214289e+00 5.03319800e-01
6.99819922e-01 -6.51156127e-01 1.50896206e-01 2.19958723e-01
-1.44370243e-01 -5.12918413e-01 7.40596130e-02 -1.83968151e+00
-1.88556567e-01 -1.17448521e+00 4.80451077e-01 1.47207224e+00
8.15359712e-01 -9.24154162e-01 4.39375490e-01 7.46448040e-01
-3.29746813e-01 -7.21087098e-01 -1.51765370e+00 -6.82690680e-01
2.69036174e-01 -1.28296018e-01 9.17606235e-01 1.01761150e+00
4.43308711e-01 -1.84560016e-01 8.34273547e-02 3.81606311e-01
6.69002652e-01 2.41544113e-01 6.93087652e-02 -1.03202474e+00
3.14902395e-01 -3.37847739e-01 1.48409531e-01 -4.83083934e-01
-5.95714986e-01 -4.47754830e-01 1.73239708e-01 -1.83176148e+00
2.08969578e-01 -3.22860181e-01 -3.42058331e-01 9.02969420e-01
2.05882311e-01 -6.61573485e-02 1.12679698e-01 8.77953172e-02
6.05808735e-01 4.83940333e-01 5.29238939e-01 4.92782034e-02
-2.18790278e-01 -1.83045045e-01 -3.38492215e-01 4.58612740e-01
8.88912201e-01 -4.71887648e-01 -5.83971262e-01 -2.08455682e-01
5.14786661e-01 7.38652050e-01 -1.36375830e-01 -1.09056067e+00
-2.82575011e-01 -5.46053708e-01 4.86855775e-01 -1.00707614e+00
-9.70175043e-02 -1.18193376e+00 7.83496618e-01 6.98528528e-01
5.22661448e-01 5.12836754e-01 5.87194085e-01 1.21996729e-02
2.37251282e-01 -2.95275934e-02 5.35582006e-01 2.11463079e-01
-2.17002869e-01 -1.33728638e-01 -1.10793328e+00 -3.22824679e-02
1.45534527e+00 -4.04939801e-01 -5.86090863e-01 -2.61949271e-01
-8.52426469e-01 8.68068159e-01 6.24721050e-01 4.92728889e-01
-1.26086399e-01 -1.20134461e+00 -8.28556180e-01 1.24351971e-01
-3.79521698e-01 -5.04344821e-01 2.96518058e-01 8.96330774e-01
-5.20616710e-01 1.59432769e-01 -2.33141810e-01 -3.16105455e-01
-4.93907660e-01 -2.47430146e-01 7.93235421e-01 -3.50813180e-01
-5.76033115e-01 3.20421278e-01 -1.44469649e-01 2.84309834e-01
-2.36789852e-01 -6.31491125e-01 1.73166126e-01 3.19983989e-01
2.03643844e-01 2.73362845e-01 3.10698748e-01 -1.73924282e-01
-8.19463193e-01 4.71470892e-01 2.17978373e-01 -1.41296297e-01
1.29537714e+00 -2.28796124e-01 -2.26271585e-01 8.02000761e-01
9.76648629e-01 -2.60001794e-02 -1.13328242e+00 6.33222878e-01
2.88066834e-01 -1.00727245e-01 1.02552891e-01 -1.06059957e+00
-1.21772254e+00 8.49024132e-02 7.62958467e-01 8.50816488e-01
1.39789474e+00 -2.46430323e-01 9.41887856e-01 1.60946511e-02
3.66485089e-01 -1.57931817e+00 -1.09209442e+00 5.82268178e-01
6.93230033e-01 -2.64860660e-01 2.75166273e-01 2.61580884e-01
-9.08974931e-02 1.19993567e+00 1.09790720e-01 3.62472236e-01
1.01325727e+00 6.37008607e-01 -2.15573251e-01 -1.31693287e-02
-8.81327212e-01 6.97012842e-02 -3.22040915e-01 -6.61971271e-02
4.72596645e-01 5.07843912e-01 -8.35170388e-01 5.36433578e-01
-3.63931209e-01 8.17204267e-03 9.92654204e-01 1.24877346e+00
-3.07676166e-01 -7.40980983e-01 -3.14563245e-01 6.13280833e-01
-4.43489164e-01 -7.81882256e-02 -1.98339783e-02 9.47110474e-01
3.24970961e-01 1.31432736e+00 8.23597074e-01 -1.70129642e-01
4.25308257e-01 4.20317173e-01 9.87660587e-02 -1.42262518e-01
-6.38308346e-01 3.38263392e-01 4.80201423e-01 -2.81075984e-01
-1.04966708e-01 -1.15057290e+00 -1.78354096e+00 -9.85664904e-01
-6.84640050e-01 6.85322523e-01 6.48590267e-01 1.17064166e+00
-2.87018269e-01 3.21716607e-01 1.30487609e+00 -5.02026975e-01
-4.09644634e-01 -1.17475355e+00 -1.31334269e+00 -4.23809528e-01
3.62604618e-01 -1.01565979e-01 -9.32900012e-01 -2.02267468e-01] | [6.001705646514893, 2.718507766723633] |
7377659c-4283-4aa5-99a7-cb3e32136fa9 | auxiliary-functions-as-koopman-observables | 2303.01483 | null | https://arxiv.org/abs/2303.01483v3 | https://arxiv.org/pdf/2303.01483v3.pdf | Auxiliary Functions as Koopman Observables: Data-Driven Analysis of Dynamical Systems via Polynomial Optimization | We present a flexible data-driven method for dynamical system analysis that does not require explicit model discovery. The method is rooted in well-established techniques for approximating the Koopman operator from data and is implemented as a semidefinite program that can be solved numerically. Furthermore, the method is agnostic of whether data is generated through a deterministic or stochastic process, so its implementation requires no prior adjustments by the user to accommodate these different scenarios. Rigorous convergence results justify the applicability of the method, while also extending and uniting similar results from across the literature. Examples on discovering Lyapunov functions, performing ergodic optimization, and bounding extrema over attractors for both deterministic and stochastic dynamics exemplify these convergence results and demonstrate the performance of the method. | ['Giovanni Fantuzzi', 'Jason J. Bramburger'] | 2023-03-02 | null | null | null | null | ['model-discovery'] | ['miscellaneous'] | [-1.67330988e-02 -1.24954529e-01 -3.90349627e-01 -2.50688884e-02
-5.64762712e-01 -9.06478584e-01 7.32756317e-01 -1.68046832e-01
-2.76056975e-01 9.30955529e-01 1.56071171e-01 -3.47301751e-01
-7.95496821e-01 -3.70330900e-01 -4.15564358e-01 -8.40806901e-01
-5.61485946e-01 6.20113015e-01 -2.01537594e-01 -4.46021676e-01
2.59053200e-01 5.68533063e-01 -1.24056697e+00 -5.92200816e-01
7.23277152e-01 7.67843187e-01 -1.67365953e-01 8.27084422e-01
3.74905646e-01 6.17766023e-01 -3.22100073e-01 -9.97575969e-02
5.30564070e-01 -3.19821775e-01 -6.18037999e-01 3.38991165e-01
-2.22640216e-01 -1.57570139e-01 -3.70759726e-01 9.66938376e-01
4.12595451e-01 1.98110670e-01 7.58541465e-01 -1.32372153e+00
-6.42209888e-01 1.19048685e-01 -3.42293121e-02 4.14933413e-01
2.54520744e-01 3.83862942e-01 1.00856173e+00 -9.38402534e-01
5.86162329e-01 1.03301227e+00 7.81662285e-01 5.20444632e-01
-1.73907924e+00 -2.03747988e-01 1.26529485e-01 -4.23717380e-01
-1.36533868e+00 -6.97581947e-01 5.97538412e-01 -8.37445259e-01
6.04483187e-01 5.52210689e-01 9.79626954e-01 9.48669851e-01
2.52833843e-01 6.93841279e-01 1.01280761e+00 -1.46886170e-01
4.88424480e-01 5.57770319e-02 1.91114828e-01 7.33939767e-01
4.62418616e-01 4.74409163e-01 -2.47199699e-01 -5.63905835e-01
8.48805904e-01 -1.11411765e-01 -2.15660051e-01 -9.85916734e-01
-1.13697898e+00 7.64458239e-01 -3.55653048e-01 -7.44337365e-02
-3.21291327e-01 3.19103390e-01 3.74380022e-01 3.98308188e-01
4.46577430e-01 5.66462874e-01 -1.47614270e-01 -3.72928828e-01
-1.05534387e+00 4.89430994e-01 1.13738930e+00 9.38701630e-01
2.69675493e-01 4.10707980e-01 1.55052409e-01 1.34690106e-01
4.13370937e-01 7.76805639e-01 2.74572700e-01 -1.17409301e+00
1.37038529e-01 4.71642762e-01 5.21245122e-01 -9.78012204e-01
-5.34942985e-01 -3.32484156e-01 -7.54567564e-01 2.79274490e-02
4.89817262e-01 -3.47452193e-01 -5.92129052e-01 1.93959236e+00
2.99209386e-01 -1.24227740e-01 1.00498088e-01 8.71481895e-01
-3.64666991e-02 7.13457167e-01 -5.24308324e-01 -7.11050034e-01
6.44135416e-01 -2.85763413e-01 -8.61531496e-01 3.04340482e-01
4.31385934e-01 -3.58558089e-01 9.12871480e-01 3.17362905e-01
-1.24540591e+00 3.01806647e-02 -8.28799129e-01 2.15751752e-01
-1.37835786e-01 -1.12924807e-01 6.00498676e-01 6.97544277e-01
-1.25686133e+00 6.29300952e-01 -1.11446297e+00 -3.21366400e-01
-1.30301127e-02 6.68450058e-01 -8.40290263e-03 7.71495700e-01
-1.06492519e+00 8.85754764e-01 2.84793258e-01 3.03336382e-01
-1.08709681e+00 -7.72310793e-01 -6.13211453e-01 -6.54089684e-03
2.20415354e-01 -7.30087996e-01 1.20368862e+00 -7.68947124e-01
-1.77804792e+00 5.98711491e-01 -3.20331126e-01 -7.57649183e-01
7.02284336e-01 2.01062560e-01 -2.68294662e-01 4.89804745e-02
-2.49158032e-02 6.05791062e-02 8.17348123e-01 -9.91863787e-01
-1.26337439e-01 1.46649346e-01 -1.24982744e-01 1.35384008e-01
-4.55579311e-01 -4.71918620e-02 -6.38827011e-02 -6.00642562e-01
-1.73318878e-01 -1.11534679e+00 -5.86905360e-01 -3.72515507e-02
-6.15548730e-01 1.06962688e-01 7.73223042e-01 -2.75565118e-01
1.58237326e+00 -1.90006661e+00 3.34354371e-01 5.82344413e-01
1.30761564e-01 -1.66124284e-01 1.22076571e-01 1.13422966e+00
-1.67426895e-02 2.15394840e-01 -5.25234282e-01 -2.24876300e-01
2.57821083e-01 1.37531057e-01 -8.59672070e-01 8.62930834e-01
1.41416147e-01 7.56725252e-01 -9.43502545e-01 -5.61677702e-02
2.39921138e-01 -4.89388928e-02 -5.82038343e-01 -2.17182338e-01
1.04884906e-02 5.57985246e-01 -3.99229109e-01 3.62940282e-01
1.91484302e-01 -4.02375758e-01 3.13777655e-01 2.02438653e-01
-4.38758194e-01 1.82728231e-01 -1.48385179e+00 1.21943045e+00
-2.07207590e-01 7.82589436e-01 1.57155216e-01 -1.08814895e+00
7.49786377e-01 2.40923196e-01 9.41606164e-01 -9.17661563e-02
1.57310352e-01 2.69352138e-01 1.09716184e-01 -3.72222632e-01
4.14425135e-01 -2.97967523e-01 -2.12517038e-01 7.53700316e-01
-1.37240499e-01 -2.93429196e-01 5.93971312e-01 1.64857998e-01
9.69091713e-01 -7.21511394e-02 3.07530910e-01 -1.01724064e+00
4.36962813e-01 3.79037797e-01 5.04186630e-01 6.43902302e-01
-2.32195213e-01 1.83572978e-01 5.82466602e-01 -2.76609659e-01
-1.26263380e+00 -1.26238763e+00 -4.99825239e-01 6.62322640e-01
2.30616733e-01 -4.49847430e-01 -6.04260921e-01 -1.23484358e-01
1.80676639e-01 4.02743846e-01 -7.60180891e-01 -2.89697468e-01
-3.56792271e-01 -1.03611505e+00 4.33394730e-01 3.26034456e-01
3.28952298e-02 -5.82164168e-01 -4.15483475e-01 3.17919701e-01
-1.22615419e-01 -7.75078893e-01 -7.33839869e-01 5.46075143e-02
-1.04120803e+00 -1.14675033e+00 -5.96151650e-01 -4.68476921e-01
8.33171725e-01 3.26209925e-02 8.91445994e-01 -4.09804195e-01
-3.81223589e-01 9.92081821e-01 4.16355819e-01 -2.74755061e-01
-5.83325386e-01 -1.06679760e-01 8.01842332e-01 1.24021851e-01
-3.64558212e-02 -4.75993842e-01 -4.43667889e-01 3.35638702e-01
-7.03523695e-01 -2.90147394e-01 4.54387516e-02 8.88285160e-01
6.58611000e-01 7.79965222e-02 6.45737767e-01 -4.94869560e-01
1.16800475e+00 -6.37172401e-01 -1.06089449e+00 -2.39223372e-02
-8.07747066e-01 2.74123281e-01 8.19149673e-01 -5.31812966e-01
-5.04857838e-01 3.56002808e-01 5.57594717e-01 -6.28699362e-01
5.12386799e-01 5.94222426e-01 3.62189472e-01 -1.15340598e-01
5.26821911e-01 5.54355383e-01 5.99943042e-01 -1.51008174e-01
3.07210028e-01 3.74190032e-01 2.89277345e-01 -5.59565306e-01
1.02504265e+00 8.74854386e-01 2.96263278e-01 -1.07149565e+00
-4.06802714e-01 -3.16999704e-01 -4.98030961e-01 -2.46226579e-01
5.00352919e-01 -7.12167323e-01 -1.27008808e+00 1.75302416e-01
-6.71689749e-01 -3.94188046e-01 -7.25859821e-01 4.76801634e-01
-1.11006057e+00 1.77901834e-01 -4.19850767e-01 -1.31015980e+00
-1.85540646e-01 -1.01058710e+00 6.71973228e-01 -5.89597300e-02
-3.47263306e-01 -1.47329652e+00 6.02390230e-01 -3.63441557e-01
4.71765667e-01 5.12775898e-01 5.93477726e-01 -5.62377751e-01
-4.88631487e-01 -3.38986665e-01 2.97822028e-01 2.14589387e-01
2.29247123e-01 6.25158846e-01 -4.49116588e-01 -6.20403409e-01
1.16896458e-01 -1.38037145e-01 3.61773342e-01 6.83180869e-01
6.25172913e-01 -7.39805698e-01 -1.54130474e-01 4.31419224e-01
1.35910523e+00 -4.13766839e-02 7.56725892e-02 2.52153635e-01
2.42184639e-01 6.77067280e-01 2.95018792e-01 7.00477421e-01
3.80026549e-01 4.42378610e-01 3.77885640e-01 5.83448075e-02
6.01265490e-01 -1.05349913e-01 6.42741382e-01 8.87083411e-01
-4.15590638e-03 -4.45897691e-02 -1.00870287e+00 6.40607536e-01
-2.05558038e+00 -1.29292846e+00 -2.80642062e-01 2.32070279e+00
9.42828715e-01 -6.69431910e-02 6.53077662e-01 -2.30793208e-02
6.22681499e-01 -1.16849706e-01 -8.82998705e-01 -4.67428625e-01
-1.48505449e-01 -6.42699888e-03 7.00053155e-01 5.78449845e-01
-1.03487134e+00 7.35967338e-01 8.53929710e+00 5.24833143e-01
-9.05468941e-01 -3.40467423e-01 4.48988676e-01 -4.11184639e-01
-2.04859644e-01 1.40191332e-01 -6.51527762e-01 6.32990062e-01
1.08028114e+00 -9.84064102e-01 7.23747194e-01 6.87204480e-01
9.51710522e-01 6.08416684e-02 -1.01958692e+00 7.85116851e-01
-5.11656165e-01 -1.55667520e+00 -8.85118246e-02 4.81321096e-01
8.94553781e-01 -8.05260465e-02 4.58901823e-01 -6.47195429e-02
6.40667915e-01 -9.52406943e-01 7.76513398e-01 8.84324551e-01
6.29756749e-01 -7.47803152e-01 1.89290702e-01 3.80584300e-01
-1.09795666e+00 -2.84005582e-01 -1.92723453e-01 -2.98645288e-01
3.47513348e-01 5.89598179e-01 -4.88782704e-01 1.19795360e-01
2.63904482e-01 7.72830427e-01 -7.24904612e-02 1.02186275e+00
2.59531051e-01 7.52940774e-01 -6.86697662e-01 -2.13386148e-01
2.65050769e-01 -6.52505994e-01 1.11497366e+00 1.12904465e+00
1.84391707e-01 -9.16050747e-02 1.79306433e-01 9.43377733e-01
3.43000054e-01 3.35379653e-02 -9.83561873e-01 -3.78272951e-01
5.14255524e-01 9.45686877e-01 -5.45682132e-01 -5.35782687e-02
-7.29115531e-02 4.54459161e-01 -7.89395941e-04 4.73169506e-01
-7.32052803e-01 -2.46322170e-01 9.31824744e-01 1.28533378e-01
1.46830171e-01 -7.38692999e-01 -3.87956411e-01 -1.25915122e+00
7.26548862e-03 -8.05141449e-01 2.82649040e-01 -8.54039267e-02
-1.31027949e+00 1.58279166e-01 7.75316730e-02 -1.60715520e+00
-6.51670516e-01 -5.89505434e-01 -6.01861656e-01 6.38502002e-01
-8.00002396e-01 -4.11458433e-01 3.29523534e-01 6.85611010e-01
1.74855679e-01 -1.36748627e-01 6.44756258e-01 -3.89533117e-03
-9.17899311e-01 2.62034684e-01 8.27221274e-01 -1.79635778e-01
2.22962961e-01 -1.60469460e+00 2.92630792e-01 1.00327170e+00
-7.64346793e-02 9.76519585e-01 1.08614540e+00 -5.42175114e-01
-1.84858239e+00 -9.23417270e-01 4.10893679e-01 -6.63359344e-01
1.46272838e+00 -4.35484648e-01 -4.79807705e-01 5.42027593e-01
-4.24228832e-02 -2.16441393e-01 6.37511194e-01 9.15789157e-02
2.33629316e-01 9.50944200e-02 -9.96240377e-01 8.86713445e-01
8.77723753e-01 -3.50502521e-01 -2.54255295e-01 5.50327420e-01
2.08818406e-01 -3.38295966e-01 -9.05628979e-01 1.37639835e-01
3.64459544e-01 -6.12599015e-01 1.00584126e+00 -9.93375003e-01
1.86722055e-02 -3.60847741e-01 -7.61282369e-02 -1.13046920e+00
-2.44184017e-01 -1.57758880e+00 -7.49155223e-01 8.76387417e-01
4.85912859e-01 -8.13162625e-01 5.29216290e-01 8.33963275e-01
1.39252111e-01 -9.47055459e-01 -1.00483096e+00 -1.45221174e+00
1.91709980e-01 -2.20370382e-01 2.55516976e-01 7.93722689e-01
2.89932549e-01 6.10664561e-02 -5.41327536e-01 1.42794430e-01
8.11266780e-01 2.64735341e-01 8.17189991e-01 -1.14804673e+00
-1.57560855e-01 -6.16556764e-01 -3.97564769e-01 -1.15791190e+00
8.71684104e-02 -9.88516569e-01 3.41903716e-02 -1.07512391e+00
1.29901662e-01 -5.68367958e-01 -9.83857065e-02 1.11457519e-01
1.01680815e-01 1.69789970e-01 7.07055023e-03 7.13650584e-01
-3.67540181e-01 8.76238227e-01 1.03660452e+00 2.64547676e-01
-5.61339140e-01 3.37753862e-01 -6.55372262e-01 7.34183788e-01
8.55966270e-01 -3.84538203e-01 -6.43162608e-01 2.02008188e-01
3.96171123e-01 2.39637494e-02 5.48299730e-01 -7.32961118e-01
2.92386621e-01 -4.67400610e-01 -1.37412459e-01 -3.34549457e-01
1.67825073e-01 -6.18187308e-01 1.96102604e-01 6.55926406e-01
-3.81456971e-01 3.68016690e-01 2.13658601e-01 1.03236103e+00
1.98090281e-02 -1.30485207e-01 8.78825068e-01 2.65129745e-01
-2.85900742e-01 4.95251328e-01 -9.43456650e-01 5.07809818e-01
1.20114172e+00 -2.78406024e-01 -7.53773078e-02 -8.06393385e-01
-1.08797681e+00 4.35026586e-01 4.47965384e-01 8.93788785e-02
2.92722076e-01 -1.39116883e+00 -4.54597920e-01 -9.31641385e-02
-2.59890318e-01 -4.18734550e-01 -2.40485117e-01 1.18691540e+00
-3.25913817e-01 6.12524867e-01 1.85141459e-01 -7.02314675e-01
-7.89701343e-01 7.01186836e-01 6.31501853e-01 -2.38591388e-01
-4.37394828e-01 2.32755318e-01 -8.14557672e-02 -5.03387570e-01
-4.71217744e-02 -3.67255002e-01 3.74962598e-01 8.82496014e-02
1.75601363e-01 6.53487146e-01 -3.10304433e-01 -3.11480641e-01
-3.70469332e-01 2.98097491e-01 2.83511549e-01 -5.47450423e-01
1.21600139e+00 -4.87324595e-01 -5.43060564e-02 8.51537585e-01
9.73183215e-01 2.06911638e-02 -1.50535226e+00 -1.29596342e-03
1.31569877e-01 2.06186045e-02 -1.49088591e-01 -3.26583058e-01
-5.80260515e-01 4.90513444e-01 1.98122531e-01 7.37753093e-01
8.78788471e-01 -4.20379996e-01 4.77089554e-01 6.88213468e-01
3.63983780e-01 -1.21854830e+00 7.05635687e-03 7.56138623e-01
8.08303118e-01 -9.51964498e-01 -4.51845601e-02 -1.64905086e-01
-5.16314507e-01 1.30892062e+00 5.41746542e-02 -5.75756073e-01
8.49662006e-01 4.21814948e-01 -2.90361762e-01 -1.84273914e-01
-9.52083290e-01 -1.87497139e-01 2.16256976e-01 3.46570402e-01
5.35232797e-02 -1.84205085e-01 -4.49955672e-01 1.40176013e-01
-2.36369953e-01 -7.89162144e-02 7.72776067e-01 1.03598368e+00
-2.58475453e-01 -7.92640865e-01 -1.27850935e-01 5.08253396e-01
-1.81663468e-01 -1.21424153e-01 -4.78092164e-01 9.81448412e-01
-7.10587144e-01 8.59298587e-01 -1.37596847e-02 -1.67281747e-01
1.76177472e-01 -2.98152287e-02 2.85601854e-01 -4.32433367e-01
-3.91887188e-01 -4.58961539e-02 -8.98187533e-02 -5.54936588e-01
-2.65962690e-01 -1.05098009e+00 -1.07021761e+00 -5.91632605e-01
-4.95255709e-01 3.27677518e-01 3.56052428e-01 9.13034141e-01
6.12455666e-01 1.05097532e-01 1.05732536e+00 -7.24704087e-01
-9.44507241e-01 -3.88799399e-01 -5.94699442e-01 -8.52414407e-04
5.90566754e-01 -6.62137806e-01 -6.74679339e-01 1.56677395e-01] | [6.618040561676025, 3.542778730392456] |
28c89c2a-1558-4bbe-981f-475eeda062b3 | weatherbench-probability-a-benchmark-dataset | 2205.00865 | null | https://arxiv.org/abs/2205.00865v1 | https://arxiv.org/pdf/2205.00865v1.pdf | WeatherBench Probability: A benchmark dataset for probabilistic medium-range weather forecasting along with deep learning baseline models | WeatherBench is a benchmark dataset for medium-range weather forecasting of geopotential, temperature and precipitation, consisting of preprocessed data, predefined evaluation metrics and a number of baseline models. WeatherBench Probability extends this to probabilistic forecasting by adding a set of established probabilistic verification metrics (continuous ranked probability score, spread-skill ratio and rank histograms) and a state-of-the-art operational baseline using the ECWMF IFS ensemble forecast. In addition, we test three different probabilistic machine learning methods -- Monte Carlo dropout, parametric prediction and categorical prediction, in which the probability distribution is discretized. We find that plain Monte Carlo dropout severely underestimates uncertainty. The parametric and categorical models both produce fairly reliable forecasts of similar quality. The parametric models have fewer degrees of freedom while the categorical model is more flexible when it comes to predicting non-Gaussian distributions. None of the models are able to match the skill of the operational IFS model. We hope that this benchmark will enable other researchers to evaluate their probabilistic approaches. | ['Nils Thuerey', 'Stephan Rasp', 'Sagar Garg'] | 2022-05-02 | null | null | null | null | ['weather-forecasting'] | ['miscellaneous'] | [-3.43855470e-01 -1.39111429e-01 -8.25857446e-02 -7.88451552e-01
-1.14933729e+00 -9.32901800e-01 1.12607110e+00 3.04406643e-01
-8.17654282e-02 1.08857405e+00 5.26191533e-01 -8.24607491e-01
-3.67529273e-01 -8.44318926e-01 -2.64600456e-01 -8.67696643e-01
-4.48835641e-01 7.89133012e-01 2.57308573e-01 -2.14340732e-01
3.76820058e-01 4.48626429e-01 -1.64448297e+00 9.99048650e-02
1.12198067e+00 9.32606578e-01 -2.00514600e-01 6.71004236e-01
6.25717640e-02 4.88182694e-01 -4.18917328e-01 -7.13564157e-02
1.34412378e-01 1.04965210e-01 -3.76921058e-01 -1.07684803e+00
3.56581986e-01 -1.60881519e-01 3.07584792e-01 7.17936695e-01
5.70868611e-01 2.74344802e-01 1.25795817e+00 -1.42268622e+00
-1.19982399e-01 6.49445474e-01 -3.74508917e-01 3.30654621e-01
3.52983803e-01 2.59902298e-01 9.07229245e-01 -5.96129417e-01
9.48715284e-02 1.13957524e+00 1.23221111e+00 -1.05129495e-01
-1.55625010e+00 -7.62871683e-01 -1.72873605e-02 -2.09647804e-01
-1.40939450e+00 -3.76603693e-01 1.08941682e-01 -1.09075940e+00
1.10053086e+00 4.14714336e-01 1.38789773e-01 8.40141237e-01
4.25677955e-01 -5.88588379e-02 1.62979960e+00 -1.06260695e-01
5.94544828e-01 2.32526168e-01 2.90327251e-01 9.85994656e-03
2.77509898e-01 1.04648089e+00 -2.16690272e-01 -8.17279756e-01
-2.82468814e-02 -2.25950181e-01 -4.34086204e-01 6.13742210e-02
-1.03632104e+00 8.55099440e-01 1.56508433e-03 2.62806296e-01
-4.06588227e-01 1.01002278e-02 7.95966387e-02 5.58510572e-02
9.39967275e-01 5.65453589e-01 -1.06609905e+00 -3.65328521e-01
-1.80033493e+00 9.04249310e-01 1.16844678e+00 4.99500841e-01
6.93983734e-01 2.45682761e-01 -5.48178375e-01 3.08526605e-01
4.58108723e-01 1.13897431e+00 1.35925755e-01 -8.18267703e-01
2.66764492e-01 -7.52586648e-02 8.55760574e-01 -7.10348189e-01
-7.03391969e-01 -2.57493645e-01 -7.53932238e-01 5.27797282e-01
5.92034996e-01 -5.18222928e-01 -9.71768796e-01 1.50398147e+00
-6.71183839e-02 4.09162134e-01 1.61859602e-01 5.74593544e-01
7.91328728e-01 1.16408420e+00 2.75094867e-01 -9.15749818e-02
9.10898983e-01 -3.59453470e-01 -3.66561532e-01 2.21312732e-01
5.51409423e-01 -8.51398349e-01 8.83605719e-01 1.33036211e-01
-6.88361287e-01 -4.28840309e-01 -9.33049738e-01 6.30468547e-01
-7.64531791e-01 -1.88873187e-01 4.25412178e-01 9.60120261e-01
-1.26861250e+00 1.10455525e+00 -9.29795563e-01 6.09902143e-02
-2.54436940e-01 -3.96979637e-02 -1.57536700e-01 2.62211323e-01
-1.55541730e+00 1.21246088e+00 3.60439330e-01 -1.11688659e-01
-8.33766758e-01 -1.27683735e+00 -8.55893910e-01 3.11409295e-01
-5.64404666e-01 -4.12149072e-01 1.22766769e+00 -1.14129625e-01
-1.57120287e+00 3.55997294e-01 -3.10008734e-01 -4.09440875e-01
4.38386708e-01 2.03309464e-03 -6.35341585e-01 -5.55384815e-01
-3.74548770e-02 3.80772263e-01 6.42955005e-01 -1.23458576e+00
-7.54652500e-01 -2.75935531e-02 -5.04500389e-01 -4.44663987e-02
4.48411077e-01 -1.66464783e-03 3.89765799e-01 -5.67228913e-01
-6.54159114e-02 -7.33567178e-01 -3.90067995e-01 -6.76864684e-01
-3.53864402e-01 -1.59597561e-01 3.38540316e-01 -7.88164139e-01
1.24769044e+00 -1.82879293e+00 -2.98687786e-01 5.28981745e-01
-2.04350099e-01 -6.10537492e-02 3.86714861e-02 6.73903704e-01
-2.13554993e-01 4.91758555e-01 -5.05078971e-01 -3.05907130e-01
2.62138158e-01 7.64972940e-02 -8.77395213e-01 4.61142153e-01
7.18939453e-02 4.84718680e-01 -6.74793899e-01 -6.97874874e-02
3.39960277e-01 3.99505824e-01 -2.37053320e-01 2.02332228e-01
-2.46263281e-01 4.79025424e-01 -7.58514255e-02 3.86950374e-01
1.25481224e+00 9.14249271e-02 -2.31879368e-01 9.29993913e-02
-5.39809823e-01 4.05820459e-01 -1.25960326e+00 7.43883431e-01
-3.33785117e-01 3.63701433e-01 -3.60533118e-01 -3.24160457e-01
1.03629577e+00 3.74405622e-01 -4.06101085e-02 -1.55349404e-01
-3.42070580e-01 2.34470442e-01 -1.80914864e-01 -1.47070080e-01
7.42433429e-01 -2.63688803e-01 -1.05230339e-01 3.13712120e-01
-8.83994401e-02 -7.84441233e-01 -2.47150771e-02 -5.51906750e-02
7.39190280e-01 3.24100971e-01 8.43601152e-02 -7.28061557e-01
-1.36442743e-02 1.28000230e-01 5.27157545e-01 1.17669654e+00
-4.62299697e-02 8.70267093e-01 5.62126577e-01 -5.08421898e-01
-9.59116280e-01 -1.39139760e+00 -6.42418444e-01 1.06874561e+00
-3.36024821e-01 -2.50251561e-01 -1.32713780e-01 -2.90042907e-01
5.15155554e-01 1.51269913e+00 -5.50538838e-01 3.65693182e-01
1.94578543e-01 -1.45265996e+00 5.90661407e-01 3.18744481e-01
-2.06609935e-01 -7.23487437e-01 -3.08872789e-01 9.76878703e-02
-1.67404637e-02 -7.41588712e-01 1.53645933e-01 2.71101356e-01
-7.83818066e-01 -6.36830449e-01 -6.84352815e-01 1.12133250e-02
1.32997036e-01 -5.42998731e-01 1.49858558e+00 -4.63577420e-01
4.05272543e-01 2.26318426e-02 -4.10116553e-01 -5.42316914e-01
-4.76491034e-01 5.46466038e-02 1.18070923e-01 -4.09110755e-01
2.05657959e-01 -7.59983599e-01 -5.32546282e-01 3.85327965e-01
-4.31629360e-01 -2.18224287e-01 3.27270962e-02 6.49899840e-01
2.53043711e-01 3.70048210e-02 3.75423104e-01 -5.23884654e-01
4.80291605e-01 -7.74685860e-01 -1.09961295e+00 1.55186668e-01
-8.58346879e-01 2.09471583e-01 3.50487858e-01 3.22919227e-02
-1.07121515e+00 -1.07926942e-01 -2.68820673e-01 9.84292701e-02
-3.93212259e-01 9.28947687e-01 2.91245371e-01 6.20799027e-02
7.85649359e-01 1.95690662e-01 -4.48254436e-01 -6.61511123e-01
2.24984601e-01 4.98429567e-01 6.28345251e-01 -7.06262827e-01
9.05501604e-01 2.58206993e-01 3.59923877e-02 -5.71070552e-01
-5.83446145e-01 -4.09882307e-01 -3.65959108e-01 -6.58163205e-02
6.02950871e-01 -9.82678413e-01 -4.54748243e-01 6.22724533e-01
-1.03325760e+00 -5.40512502e-01 -1.38993144e-01 7.78624475e-01
-1.43381909e-01 7.85263926e-02 -3.38729262e-01 -1.38057756e+00
-2.88720638e-01 -9.84750569e-01 1.21839333e+00 3.92556041e-01
-2.82129735e-01 -1.16849899e+00 7.19223619e-01 -3.85058939e-01
9.78088140e-01 6.45413160e-01 9.07611310e-01 -8.06835115e-01
2.74108034e-02 -3.56799781e-01 -2.03019261e-01 5.41735031e-02
-1.42201856e-01 5.41874290e-01 -1.29846394e+00 -2.45266348e-01
-3.20409298e-01 -1.45878136e-01 1.12704754e+00 8.07100892e-01
8.28938901e-01 -1.42987981e-01 -4.13538098e-01 6.98154092e-01
1.35528886e+00 1.05383106e-01 7.08718479e-01 1.90393060e-01
2.44706020e-01 6.38895273e-01 3.84616673e-01 5.56095481e-01
5.46667397e-01 2.85033464e-01 2.99835175e-01 -9.22838449e-02
6.90039277e-01 -4.03695628e-02 2.64420450e-01 3.72501016e-01
-3.46172512e-01 -3.59478205e-01 -1.69234550e+00 4.31645066e-01
-1.64033198e+00 -1.24804187e+00 -3.16195786e-01 2.62181544e+00
9.27050412e-01 2.99440414e-01 -2.86429018e-01 -4.68198117e-03
5.14701128e-01 5.20343721e-01 -9.14893858e-03 -4.99030054e-01
-1.42592251e-01 2.73466229e-01 8.25176656e-01 8.96457553e-01
-1.31215107e+00 6.71310961e-01 7.18251896e+00 8.81505311e-01
-1.06303227e+00 -9.49239731e-03 9.48966742e-01 3.20400268e-01
-5.67683935e-01 3.62344176e-01 -1.14478993e+00 7.16801047e-01
1.63228154e+00 -1.85468748e-01 -3.55753279e-03 6.89971507e-01
5.02919674e-01 -7.17458606e-01 -8.42088878e-01 4.48453546e-01
-4.86287534e-01 -1.45896900e+00 -2.59691507e-01 -8.83854702e-02
1.07229364e+00 4.28378910e-01 -1.54495472e-02 5.99255204e-01
8.28770638e-01 -1.44071233e+00 6.44836128e-01 1.34220088e+00
8.70637894e-01 -6.51483119e-01 1.27118897e+00 4.55073446e-01
-1.03063226e+00 3.77932787e-01 -1.97330922e-01 -3.46954346e-01
3.25194389e-01 1.12274539e+00 -4.61955249e-01 7.79547513e-01
1.03447533e+00 3.48963618e-01 -3.93410623e-01 1.33584976e+00
-5.44490013e-03 1.22247601e+00 -8.99035573e-01 1.44112736e-01
2.45538786e-01 -2.81172916e-02 5.02940953e-01 1.26456988e+00
5.52329183e-01 7.66265839e-02 1.63683608e-01 6.54657185e-01
6.29384339e-01 -1.83212101e-01 -4.78135347e-01 3.60837907e-01
7.03864634e-01 1.08354914e+00 -3.84939909e-01 -3.71869951e-01
-8.82477090e-02 1.92998797e-01 -2.19156861e-01 5.66295385e-01
-7.18760490e-01 -3.38466495e-01 5.12280464e-01 -4.36407588e-02
2.40674868e-01 -1.82230607e-01 -5.49860716e-01 -8.74790251e-01
-4.39924270e-01 -5.30377865e-01 1.74327254e-01 -9.15464520e-01
-1.55489159e+00 7.21322536e-01 5.90999961e-01 -1.04492843e+00
-5.87643921e-01 -4.94673312e-01 -1.17379987e+00 1.71097302e+00
-1.65990543e+00 -8.90691221e-01 -3.32525074e-01 -8.72049257e-02
-2.44368032e-01 -1.00123785e-01 1.23615956e+00 9.94154159e-03
-1.84144661e-01 2.87259102e-01 8.35279465e-01 -3.14254016e-01
8.63224804e-01 -1.56461692e+00 8.36704493e-01 4.43888187e-01
-3.22454661e-01 2.50762582e-01 1.24255562e+00 -7.68542528e-01
-6.06914103e-01 -1.27299905e+00 1.25002670e+00 -7.16954052e-01
6.65512204e-01 -1.29026666e-01 -1.03830814e+00 3.54141057e-01
4.61902954e-02 8.06157216e-02 4.87217218e-01 3.84238034e-01
-2.76053101e-01 1.79912597e-02 -1.29115701e+00 9.25120041e-02
7.09539428e-02 -2.67159224e-01 -8.31348002e-01 3.90627950e-01
3.59391749e-01 -5.04002810e-01 -1.15780997e+00 9.33146775e-01
8.56656194e-01 -1.30841303e+00 6.12944841e-01 -4.75664616e-01
3.33560586e-01 -5.06168067e-01 -3.96294445e-01 -1.79797733e+00
-2.62203842e-01 -5.50809920e-01 3.39040875e-01 1.49662280e+00
1.04150164e+00 -1.04686666e+00 4.08433110e-01 6.97722673e-01
-2.85956729e-02 -5.99669695e-01 -1.09833097e+00 -9.00745809e-01
5.90690732e-01 -7.93085694e-01 1.07271063e+00 1.11368024e+00
-2.60575920e-01 -3.50360096e-01 -4.12315995e-01 6.70447648e-01
6.08687401e-01 3.84690821e-01 5.64071119e-01 -1.55242777e+00
-2.10745394e-01 -5.48972547e-01 -1.44677788e-01 -4.72927064e-01
3.12157329e-02 -5.17309308e-01 2.44407579e-01 -1.37841427e+00
-1.43033147e-01 -6.58108890e-01 -2.54462689e-01 4.87114608e-01
-2.27707371e-01 3.36210690e-02 -2.63532072e-01 1.99971259e-01
-1.24089941e-02 7.42667437e-01 3.13657224e-01 1.53662041e-01
-2.94454724e-01 4.69685882e-01 -7.65084289e-03 7.76439011e-01
9.29104507e-01 -7.51394689e-01 -2.33755112e-02 6.36127517e-02
2.23936245e-01 1.60087287e-01 4.20461327e-01 -1.09871805e+00
1.70803234e-01 -3.41402888e-01 6.31011605e-01 -1.09505665e+00
-7.96889770e-04 -2.29850218e-01 3.50733042e-01 2.11001381e-01
-5.89607470e-02 4.06487465e-01 5.42278707e-01 2.64905959e-01
-2.60579646e-01 -8.72457176e-02 7.80062497e-01 1.15623057e-01
-4.68634307e-01 6.13522567e-02 -6.36618316e-01 2.48476863e-01
5.93552351e-01 -2.45051514e-02 -5.42084038e-01 -6.05678499e-01
-7.91942418e-01 5.49650788e-01 2.98682719e-01 9.85209420e-02
1.07818842e-02 -8.43962908e-01 -1.17815685e+00 6.74328804e-02
3.49722989e-02 -4.60681349e-01 2.05473006e-01 7.74382949e-01
-6.04359627e-01 4.51022536e-01 1.31023183e-01 -7.14336038e-01
-5.68469584e-01 -7.71989301e-02 6.84146464e-01 -4.64560211e-01
-1.51030868e-01 7.65620410e-01 -8.70939493e-02 -1.02312171e+00
7.64746368e-02 -8.99384737e-01 -3.24566305e-01 2.44646221e-01
5.14694512e-01 4.86678630e-01 1.15001298e-01 -4.80248392e-01
-4.48094338e-01 5.79875946e-01 4.54123944e-01 -5.43505192e-01
1.18734872e+00 1.13055564e-01 5.71506061e-02 8.44544172e-01
5.18886149e-01 3.82687226e-02 -1.22050166e+00 2.37420171e-01
2.34224930e-01 -2.08535179e-01 3.17576885e-01 -1.27896774e+00
-5.07460654e-01 8.16829801e-01 8.10320497e-01 4.12646383e-01
6.95267081e-01 -4.07013118e-01 1.02553077e-01 5.86790703e-02
2.62546450e-01 -7.01886535e-01 -1.16157246e+00 9.65034902e-01
8.92936647e-01 -1.36137342e+00 2.58443087e-01 1.18100397e-01
-7.09801674e-01 8.79029751e-01 2.24338859e-01 -5.42825498e-02
1.49435186e+00 6.18085086e-01 1.31678060e-01 8.19848105e-02
-8.35649967e-01 -1.76399469e-01 6.73484862e-01 6.05596840e-01
4.32263136e-01 3.37370306e-01 -2.04001263e-01 8.60670805e-01
-6.10918462e-01 -1.15191840e-01 2.06532851e-01 5.10284364e-01
-3.33813876e-01 -6.41197264e-01 -6.16562426e-01 8.61881316e-01
-3.42466712e-01 -7.26540864e-01 2.37005591e-01 6.84572697e-01
-3.09721697e-02 1.04996705e+00 3.87742192e-01 -3.20351005e-01
3.22965890e-01 4.63337392e-01 -2.45589241e-01 -2.86047518e-01
-6.16077662e-01 -2.23784357e-01 3.31501365e-01 -4.38762844e-01
-9.46080834e-02 -8.82416368e-01 -6.97979152e-01 -6.12993300e-01
-2.22165167e-01 6.74112558e-01 7.53112376e-01 9.68759775e-01
3.29438061e-01 7.25707933e-02 5.81710696e-01 -1.51786149e+00
-8.89783084e-01 -1.39683402e+00 -8.31697464e-01 -1.57092422e-01
4.72616851e-01 -7.51271665e-01 -9.61973190e-01 -3.28023016e-01] | [6.599457263946533, 3.1427485942840576] |
bd4bc3b9-7bde-436a-8eed-fabe73a470e0 | a-survey-on-knowledge-graphs-for-healthcare | 2306.04802 | null | https://arxiv.org/abs/2306.04802v1 | https://arxiv.org/pdf/2306.04802v1.pdf | A Survey on Knowledge Graphs for Healthcare: Resources, Applications, and Promises | Healthcare knowledge graphs (HKGs) have emerged as a promising tool for organizing medical knowledge in a structured and interpretable way, which provides a comprehensive view of medical concepts and their relationships. However, challenges such as data heterogeneity and limited coverage remain, emphasizing the need for further research in the field of HKGs. This survey paper serves as the first comprehensive overview of HKGs. We summarize the pipeline and key techniques for HKG construction (i.e., from scratch and through integration), as well as the common utilization approaches (i.e., model-free and model-based). To provide researchers with valuable resources, we organize existing HKGs (The resource is available at https://github.com/lujiaying/Awesome-HealthCare-KnowledgeBase) based on the data types they capture and application domains, supplemented with pertinent statistical information. In the application section, we delve into the transformative impact of HKGs across various healthcare domains, spanning from fine-grained basic science research to high-level clinical decision support. Lastly, we shed light on the opportunities for creating comprehensive and accurate HKGs in the era of large language models, presenting the potential to revolutionize healthcare delivery and enhance the interpretability and reliability of clinical prediction. | ['Carl Yang', 'Fei Wang', 'Joyce Ho', 'Chen Ling', 'Xuan Kan', 'Yue Yu', 'Shaojun Yu', 'Wenjing Ma', 'ran Xu', 'Shiyu Wang', 'Jiaying Lu', 'Hejie Cui'] | 2023-06-07 | null | null | null | null | ['knowledge-graphs'] | ['knowledge-base'] | [-1.06357664e-01 4.37215388e-01 -7.73747802e-01 -3.25256884e-01
-7.42434561e-01 -4.12599355e-01 1.33638298e-02 8.00351083e-01
-1.04405209e-02 6.76612079e-01 6.29829347e-01 -5.88607907e-01
-5.02278268e-01 -8.72276187e-01 -3.29297751e-01 -4.81354952e-01
-1.73635229e-01 6.33054733e-01 -2.05314398e-01 -1.19147055e-01
-8.43288079e-02 2.33468562e-01 -1.06642747e+00 4.19986248e-01
1.02954006e+00 5.66849530e-01 2.31027100e-02 4.15928155e-01
-1.14099026e-01 1.08949327e+00 -3.85066986e-01 -5.82180262e-01
-3.76005113e-01 -5.18053114e-01 -1.24967384e+00 -1.79354340e-01
-2.20339775e-01 8.66465867e-02 -3.51760477e-01 6.37974024e-01
4.35684264e-01 -3.71929795e-01 3.06866974e-01 -1.15779209e+00
-9.15826440e-01 6.85210943e-01 3.92798185e-02 9.99222100e-02
5.32248080e-01 -3.27178389e-02 8.38366807e-01 -6.44922197e-01
1.07335055e+00 8.09910655e-01 7.81902313e-01 7.68542528e-01
-8.03098619e-01 -3.65822136e-01 1.87807620e-01 5.51952720e-01
-1.54451096e+00 -3.75107676e-01 2.77960330e-01 -6.26250565e-01
1.14772093e+00 6.25960290e-01 1.03331709e+00 7.86303639e-01
3.07785869e-01 7.15288460e-01 8.86928260e-01 -5.28635859e-01
3.64067852e-02 3.88196856e-02 4.92160767e-01 9.67679977e-01
6.04894340e-01 -1.64538041e-01 -6.06506705e-01 -6.64730787e-01
5.27946293e-01 3.64213645e-01 -3.18577707e-01 -1.55485481e-01
-1.25355196e+00 6.68418109e-01 3.36290121e-01 3.91852081e-01
-4.16749090e-01 -2.70307034e-01 4.03929323e-01 5.57829216e-02
2.32905060e-01 4.31566238e-01 -6.52668118e-01 -1.03304321e-02
-7.17225194e-01 1.12976119e-01 9.91273582e-01 1.11844754e+00
3.57425958e-01 -3.98006171e-01 -4.17186320e-02 7.27155387e-01
2.00907379e-01 2.35197008e-01 4.14743334e-01 -6.76324189e-01
4.88774702e-02 8.82429004e-01 -1.73354223e-01 -1.02090311e+00
-8.20163727e-01 -4.43112791e-01 -9.71529365e-01 -7.15871572e-01
6.75578937e-02 -2.12546140e-02 -9.40454006e-01 1.51529574e+00
4.29938227e-01 -3.56624797e-02 2.47463599e-01 5.63600123e-01
1.50577414e+00 4.04144190e-02 4.30562139e-01 -2.46426582e-01
1.89250910e+00 -8.58045280e-01 -1.06563640e+00 1.03873769e-02
1.11432648e+00 -4.32583064e-01 7.14601517e-01 2.79613584e-01
-1.05281198e+00 4.98225633e-03 -5.51266611e-01 -2.47134089e-01
-7.40981936e-01 -1.73184335e-01 8.78661871e-01 3.91722947e-01
-1.07223260e+00 2.18492180e-01 -1.39341724e+00 -7.31331408e-01
6.04392052e-01 3.67127769e-02 -5.33817291e-01 -3.83440673e-01
-1.46964538e+00 1.19432831e+00 5.73723853e-01 -6.13503940e-02
-4.84676391e-01 -9.90916252e-01 -9.25455809e-01 -2.85199225e-01
5.67230225e-01 -1.33486915e+00 9.84904289e-01 -1.30802363e-01
-8.10811996e-01 9.99659181e-01 -3.57357144e-01 -3.45441103e-01
1.10199019e-01 -4.84775379e-02 -7.56951690e-01 1.68508604e-01
-1.42999347e-02 1.95646867e-01 -3.29761058e-01 -9.44041908e-01
-6.38973117e-01 -5.20821273e-01 -1.56788126e-01 1.75629035e-01
-2.85223812e-01 8.24538711e-03 -8.70412230e-01 -5.19918442e-01
-1.24403536e-01 -7.68202364e-01 -5.04007936e-01 -2.42106780e-01
-5.83032608e-01 -1.66757017e-01 1.44409359e-01 -1.08988702e+00
1.99586260e+00 -1.83010387e+00 1.31559730e-01 3.38773161e-01
7.77118802e-01 7.27248043e-02 4.44421381e-01 1.02086544e+00
-1.51325405e-01 2.99910575e-01 -2.18404695e-01 -2.04053354e-02
-3.34391057e-01 4.76244271e-01 3.09702724e-01 2.99221665e-01
6.75940420e-03 1.50603855e+00 -1.15245283e+00 -6.38086855e-01
1.88700706e-01 5.15540183e-01 -3.64298761e-01 -1.42385634e-02
1.16186403e-01 4.54236418e-01 -5.64236939e-01 1.03686690e+00
1.30083710e-01 -9.86438751e-01 7.34630883e-01 -1.75867707e-01
1.82722183e-03 3.14803213e-01 -6.17748320e-01 1.46809912e+00
-2.07269132e-01 7.75564834e-02 -8.19722563e-02 -8.47971916e-01
5.31318307e-01 4.65431899e-01 7.11295187e-01 -2.83708692e-01
-2.65537292e-01 1.66012630e-01 -1.07168421e-01 -6.26722515e-01
1.26756951e-01 -1.83213964e-01 -1.06614269e-01 8.99393931e-02
-2.67740846e-01 3.93640518e-01 1.56215206e-01 3.85737747e-01
1.29593706e+00 -3.48248959e-01 1.21427679e+00 -2.88255304e-01
3.57098132e-01 4.30939466e-01 6.54047489e-01 5.80366492e-01
5.10690510e-02 2.28276253e-01 3.28112721e-01 -6.24022424e-01
-6.13524437e-01 -8.95775020e-01 -5.23194075e-01 5.73552847e-01
-1.33394197e-01 -1.00038481e+00 -5.07246673e-01 -5.58785021e-01
1.95855469e-01 3.98699522e-01 -8.64808202e-01 -1.75828218e-01
-3.50173175e-01 -1.11374784e+00 6.51755393e-01 7.17245758e-01
1.45159855e-01 -8.34522963e-01 -3.86146247e-01 9.92717445e-02
-6.28765106e-01 -9.59816396e-01 -8.35244581e-02 -1.37611091e-01
-8.85287583e-01 -1.54618597e+00 -3.35739553e-01 -8.22350204e-01
9.21725392e-01 -1.02627598e-01 1.37783587e+00 4.16846097e-01
-6.06955945e-01 5.95300794e-01 -4.63636070e-01 -5.42919219e-01
-4.19094026e-01 3.36339697e-02 -1.37280405e-01 -7.22608984e-01
5.63823104e-01 -1.12589039e-01 -8.83263826e-01 7.21925572e-02
-9.62162852e-01 3.62020254e-01 6.85197294e-01 9.06715035e-01
9.55577731e-01 3.15540172e-02 6.92112207e-01 -1.49846339e+00
7.36643076e-01 -8.48589480e-01 -5.79990670e-02 7.08079100e-01
-1.19749641e+00 -1.85138911e-01 3.62230539e-01 3.27431977e-01
-6.30776703e-01 -1.80210039e-01 -3.15604538e-01 8.78727734e-02
-1.78594321e-01 1.21336174e+00 1.28921077e-01 1.38814181e-01
5.90130687e-01 1.63004428e-01 1.28848046e-01 -6.56465650e-01
4.75144953e-01 7.28112698e-01 2.51272559e-01 -2.50925869e-01
2.13016972e-01 5.54092407e-01 -4.65474688e-02 -5.31314969e-01
-8.39313865e-01 -7.50356793e-01 -4.63234484e-01 1.02649264e-01
8.90428782e-01 -8.62794161e-01 -5.91479719e-01 2.35795155e-01
-6.31354034e-01 -2.02777922e-01 -2.70780265e-01 3.81602407e-01
-2.62767762e-01 4.70163763e-01 -9.72242832e-01 -2.96433717e-01
-5.54400444e-01 -9.98496830e-01 6.98770165e-01 -9.96856391e-02
-6.57595098e-01 -1.49288666e+00 6.47415370e-02 7.10354924e-01
1.70745462e-01 5.18614113e-01 1.39604950e+00 -7.47301042e-01
-2.14950010e-01 -2.84501106e-01 1.10282756e-01 -9.65108275e-02
5.61748028e-01 -1.23223834e-01 -5.20926476e-01 -8.20001960e-02
-2.20552564e-01 -4.13815603e-02 5.63833535e-01 4.85961378e-01
1.39679956e+00 -5.26725829e-01 -9.05468047e-01 5.46407342e-01
1.26747823e+00 2.81119019e-01 4.96221185e-01 2.41830170e-01
8.13470900e-01 5.10857522e-01 4.84653622e-01 3.48855168e-01
1.07513750e+00 5.13002455e-01 -9.36683714e-02 -4.61855859e-01
-1.36797428e-01 -2.03471020e-01 -3.47246766e-01 1.30479956e+00
-4.12294924e-01 -9.70772952e-02 -1.45337284e+00 5.54197252e-01
-2.03013635e+00 -5.90601683e-01 -1.14562519e-01 2.07845712e+00
1.15273798e+00 -1.96157590e-01 1.10798784e-01 -7.39980191e-02
4.25579518e-01 -2.55134463e-01 -4.68955159e-01 -2.89239734e-01
-8.04836974e-02 1.19922429e-01 4.20821667e-01 5.10069788e-01
-8.13867986e-01 8.43804717e-01 6.94594479e+00 6.10446632e-01
-6.86905146e-01 1.41811550e-01 5.87086439e-01 8.86274427e-02
-4.38971341e-01 -1.52037963e-01 -6.30040586e-01 3.06608468e-01
1.11754560e+00 -4.19708222e-01 2.27893904e-01 7.14497924e-01
3.19809586e-01 4.41573784e-02 -8.67140830e-01 8.23375881e-01
-6.10381849e-02 -1.81735849e+00 6.33255243e-02 1.58771947e-01
5.12531400e-01 5.70314415e-02 -2.99612492e-01 -1.38569474e-02
3.97585392e-01 -1.13609767e+00 1.05774090e-01 7.96058118e-01
1.00596273e+00 -3.75430405e-01 9.63723540e-01 1.88890789e-02
-1.26603878e+00 1.86150357e-01 3.96345481e-02 8.81138071e-02
1.67367324e-01 7.45108128e-01 -1.00863481e+00 1.27288568e+00
8.34804177e-01 1.04514301e+00 -4.10263240e-01 8.26671898e-01
-6.33894280e-02 5.54363906e-01 1.61769018e-01 2.32245207e-01
-2.56518185e-01 3.37742381e-02 7.26086423e-02 1.49476111e+00
6.69118762e-02 5.93028665e-01 2.68355101e-01 3.23695868e-01
1.16618127e-01 3.79198730e-01 -5.26706398e-01 -3.41515988e-01
6.84709013e-01 1.06263232e+00 -7.56748557e-01 -5.93842685e-01
-7.58856475e-01 4.80756402e-01 3.35624695e-01 2.73404896e-01
-4.92690623e-01 -2.16596827e-01 6.30996346e-01 2.72367477e-01
-2.83598870e-01 2.37513501e-02 -3.55285406e-01 -1.18163061e+00
-1.48347020e-01 -1.07966137e+00 1.15487301e+00 -4.17100817e-01
-1.32086468e+00 5.94751716e-01 1.35306358e-01 -9.30156350e-01
-3.16521198e-01 -5.34880280e-01 1.68354496e-01 6.85504138e-01
-1.31211138e+00 -1.09438205e+00 -4.30835932e-01 6.25304937e-01
9.04021934e-02 2.26697803e-01 1.33633804e+00 3.82316887e-01
-6.99092865e-01 5.40086925e-01 3.11460122e-02 2.51202345e-01
5.58564067e-01 -1.07939303e+00 3.37894052e-01 2.73750335e-01
-2.56610304e-01 1.16227674e+00 2.64724046e-01 -9.99872088e-01
-1.42997777e+00 -1.11660099e+00 1.24855828e+00 -9.49210346e-01
5.99981427e-01 -1.42032042e-01 -1.02387714e+00 8.65659535e-01
-1.09966308e-01 -1.44492000e-01 1.52950883e+00 4.36547726e-01
-3.31169851e-02 1.32948264e-01 -1.07016313e+00 5.47821105e-01
1.35745859e+00 -4.23657328e-01 -5.45939147e-01 4.91319567e-01
5.93262672e-01 -5.70347667e-01 -1.58854949e+00 5.94881713e-01
6.13600075e-01 -6.09589696e-01 9.35162961e-01 -1.08734810e+00
2.84642756e-01 -2.32572481e-01 2.67584413e-01 -9.98994291e-01
-7.04851925e-01 -3.49187255e-01 -3.66497934e-01 5.70440948e-01
6.52717710e-01 -8.32958400e-01 5.89347243e-01 7.69993782e-01
-3.82634521e-01 -1.54024541e+00 -5.81907153e-01 -4.09783781e-01
-1.51633725e-01 -3.66355747e-01 7.01292694e-01 1.34493339e+00
7.66740680e-01 4.45877910e-02 -2.17744544e-01 1.49872094e-01
3.17830533e-01 1.41625851e-01 3.28296930e-01 -1.12345338e+00
-1.81026116e-01 -2.61129826e-01 -5.68110466e-01 -4.93969470e-01
-3.54658663e-01 -1.28608966e+00 -4.83626962e-01 -2.38755798e+00
6.53873146e-01 -3.63553613e-01 -4.87066269e-01 9.53013301e-01
-3.29591811e-01 7.67572746e-02 -2.94634759e-01 4.62798387e-01
-6.55455768e-01 -6.82397261e-02 1.24700046e+00 2.13756636e-01
-1.75094873e-01 -1.57549307e-01 -1.24133170e+00 6.21814966e-01
8.75392020e-01 -2.98203856e-01 -5.19110918e-01 -2.50890315e-01
3.87831539e-01 1.65648162e-01 2.46878326e-01 -4.04826045e-01
3.63722801e-01 -4.06623721e-01 2.66112298e-01 -3.40409905e-01
1.38794053e-02 -5.66651464e-01 6.98565900e-01 8.91117334e-01
-2.47857660e-01 3.22485179e-01 3.09936285e-01 4.73590463e-01
-3.02237421e-01 3.02524596e-01 3.16045284e-01 -4.15902227e-01
-8.22892845e-01 3.29237968e-01 -2.51136452e-01 1.93679541e-01
1.00058389e+00 -3.89247648e-02 -6.60314441e-01 -1.99202076e-01
-1.15763783e+00 5.53394437e-01 3.48452717e-01 3.87451291e-01
6.88089132e-01 -1.04593873e+00 -7.01627553e-01 5.10887615e-03
6.25115633e-01 4.82962700e-03 5.90635300e-01 1.18698883e+00
-5.79576552e-01 9.73966062e-01 9.67688560e-02 -1.60507664e-01
-1.32109666e+00 7.22259879e-01 2.34077007e-01 -4.63351578e-01
-8.69888186e-01 5.00588775e-01 1.21153034e-01 -3.25776190e-01
2.05399901e-01 -4.27004933e-01 -3.06482613e-01 -2.06933856e-01
5.72589874e-01 3.74039888e-01 3.36040765e-01 -3.19249213e-01
-9.19144154e-01 2.75414556e-01 -1.43653721e-01 4.67577100e-01
1.29967332e+00 -1.85617566e-01 -5.39942086e-01 4.78489488e-01
8.03215444e-01 7.83761684e-03 -1.68550879e-01 -1.63668111e-01
2.17870161e-01 -1.50342500e-02 -2.14442536e-01 -1.21204245e+00
-8.84138823e-01 3.58791113e-01 1.89717755e-01 1.28810033e-01
1.21179104e+00 5.92570722e-01 5.52080691e-01 1.54311985e-01
3.67259592e-01 -7.91157484e-01 -4.96295094e-01 1.12391479e-01
7.09093153e-01 -1.12829447e+00 2.55799204e-01 -8.45025539e-01
-7.25817978e-01 8.18766892e-01 2.41855294e-01 5.76804578e-01
1.06997406e+00 3.11541736e-01 3.39999527e-01 -7.65498757e-01
-9.33951259e-01 -1.02368385e-01 4.84082580e-01 6.32743239e-01
7.90927708e-01 5.06463110e-01 -6.21666670e-01 9.52535927e-01
-3.41819018e-01 4.13282603e-01 1.53686062e-01 1.01289165e+00
-3.96244926e-03 -1.38299143e+00 -8.30523893e-02 1.03898060e+00
-5.27903557e-01 -4.62038875e-01 -4.82733607e-01 9.01587009e-01
1.78392589e-01 9.68952656e-01 -3.84373903e-01 -4.31001723e-01
4.13727611e-01 9.35341641e-02 2.46845618e-01 -8.06340635e-01
-4.17139113e-01 4.08145692e-03 4.30137455e-01 -5.86241722e-01
-4.80656683e-01 -6.00370109e-01 -1.45442617e+00 -3.21675986e-01
-8.15116093e-02 4.02627319e-01 2.46327609e-01 8.28413188e-01
8.61502469e-01 7.00811028e-01 -1.78895235e-01 3.33783805e-01
7.58937448e-02 -5.55664182e-01 -3.89862508e-01 1.31119728e-01
1.26480803e-01 -4.44795966e-01 4.17024083e-02 1.87772855e-01] | [8.449039459228516, 8.590336799621582] |
e867c811-4f31-4778-b1e9-1e865cc997f4 | dual-task-mutual-learning-for-semi-supervised | 2103.04708 | null | https://arxiv.org/abs/2103.04708v2 | https://arxiv.org/pdf/2103.04708v2.pdf | Dual-Task Mutual Learning for Semi-Supervised Medical Image Segmentation | The success of deep learning methods in medical image segmentation tasks usually requires a large amount of labeled data. However, obtaining reliable annotations is expensive and time-consuming. Semi-supervised learning has attracted much attention in medical image segmentation by taking the advantage of unlabeled data which is much easier to acquire. In this paper, we propose a novel dual-task mutual learning framework for semi-supervised medical image segmentation. Our framework can be formulated as an integration of two individual segmentation networks based on two tasks: learning region-based shape constraint and learning boundary-based surface mismatch. Different from the one-way transfer between teacher and student networks, an ensemble of dual-task students can learn collaboratively and implicitly explore useful knowledge from each other during the training process. By jointly learning the segmentation probability maps and signed distance maps of targets, our framework can enforce the geometric shape constraint and learn more reliable information. Experimental results demonstrate that our method achieves performance gains by leveraging unlabeled data and outperforms the state-of-the-art semi-supervised segmentation methods. | ['Jicong Zhang', 'Yichi Zhang'] | 2021-03-08 | null | null | null | null | ['semi-supervised-medical-image-segmentation'] | ['computer-vision'] | [ 3.75025809e-01 5.92838109e-01 -3.07609946e-01 -7.55746722e-01
-1.22837627e+00 -3.59117568e-01 3.06739844e-02 3.02841008e-01
-7.08481610e-01 4.88570005e-01 -1.83666557e-01 -2.30628252e-01
7.72806257e-02 -6.96878731e-01 -7.61218011e-01 -8.67446303e-01
3.10922891e-01 6.88328207e-01 4.14154261e-01 1.32990554e-01
1.76377427e-02 2.97679573e-01 -1.02093065e+00 -1.02571510e-02
1.37467527e+00 8.92861485e-01 4.34114814e-01 2.58979052e-01
-3.34292114e-01 6.74581707e-01 -1.93411902e-01 -3.32030579e-02
8.98751542e-02 -4.11382169e-01 -1.08146906e+00 5.25906622e-01
9.97207686e-02 -2.38946795e-01 5.66753969e-02 1.16119576e+00
4.14786756e-01 1.64339185e-01 7.27992237e-01 -8.71308267e-01
-5.50416112e-01 5.57701230e-01 -8.74832213e-01 -9.27552059e-02
-1.26058623e-01 -9.22345817e-02 8.95512044e-01 -6.88898504e-01
3.11335653e-01 7.21195519e-01 6.06095493e-01 4.36515480e-01
-1.23566532e+00 -7.20279694e-01 2.38968596e-01 -2.21037060e-01
-1.32290983e+00 6.50242642e-02 1.01179338e+00 -5.73118269e-01
2.13871866e-01 4.94821556e-02 4.96914089e-01 5.19026637e-01
-3.25161278e-01 1.36390936e+00 1.06223750e+00 -3.85366261e-01
1.38760567e-01 3.03853840e-01 1.68999642e-01 9.95035648e-01
2.22857241e-02 -1.76862970e-01 -8.78785849e-02 1.10962525e-01
1.11331618e+00 1.95413172e-01 -2.39565879e-01 -7.75892019e-01
-1.08434057e+00 8.41831744e-01 7.19086349e-01 3.34433824e-01
-4.68143076e-02 -5.63837364e-02 9.37170386e-02 -1.14075445e-01
7.12476015e-01 2.96862781e-01 -5.84380388e-01 1.83705136e-01
-1.20709836e+00 -3.32368881e-01 6.33017719e-01 8.79564524e-01
1.10888302e+00 -2.43614256e-01 -1.17283789e-02 7.07503200e-01
6.01315916e-01 3.03146899e-01 4.42321986e-01 -6.60688162e-01
2.80175745e-01 9.51471448e-01 -7.33770579e-02 -6.06596053e-01
-3.07265371e-01 -4.05221134e-01 -8.49208832e-01 7.79207796e-02
6.66878223e-01 -2.55386174e-01 -1.34782565e+00 1.45235896e+00
6.24803782e-01 4.67053831e-01 -1.50924817e-01 9.31786954e-01
9.19378161e-01 2.89629847e-01 1.00903526e-01 -6.34923577e-02
1.10564005e+00 -1.32673478e+00 -5.67516208e-01 -1.87313020e-01
9.20496583e-01 -6.18786514e-01 8.12614083e-01 2.23886758e-01
-1.01955628e+00 -4.89527941e-01 -9.25471961e-01 -1.38492987e-01
-3.56381945e-02 1.59677252e-01 6.65889859e-01 4.52162236e-01
-8.32186222e-01 5.86955845e-01 -1.19713569e+00 1.04647703e-01
1.04162669e+00 5.86694837e-01 -2.00495318e-01 -2.54282337e-02
-7.78365970e-01 4.64585692e-01 3.03042859e-01 1.18289433e-01
-7.23593950e-01 -8.37298751e-01 -1.09682596e+00 -1.13080896e-01
5.83227813e-01 -4.40566212e-01 1.29702759e+00 -1.11896145e+00
-1.53374970e+00 1.25646710e+00 1.09458074e-01 -1.44454151e-01
8.02000523e-01 -2.43033081e-01 2.55688697e-01 3.20478290e-01
2.31142104e-01 8.38497996e-01 6.68443739e-01 -1.40991461e+00
-4.80928570e-01 -5.59319735e-01 -2.68918246e-01 3.21406215e-01
-1.78891703e-01 -3.51137847e-01 -6.45140171e-01 -6.45332754e-01
4.04611677e-01 -8.26555371e-01 -7.19441235e-01 2.71146834e-01
-5.40728629e-01 -3.81813258e-01 7.47520924e-01 -5.74110091e-01
9.26943123e-01 -1.96930635e+00 1.53538987e-01 4.03688282e-01
3.67531776e-01 3.33641201e-01 1.22068860e-01 -2.58446664e-01
1.40349135e-01 1.34353489e-01 -7.74529040e-01 -4.35061067e-01
-3.81990850e-01 3.39209437e-01 1.06256947e-01 5.93324482e-01
1.85657337e-01 1.12339687e+00 -1.22434938e+00 -1.04372466e+00
1.79211155e-01 5.06372273e-01 -3.81616414e-01 3.70303929e-01
-2.02716231e-01 1.06914568e+00 -8.58498693e-01 5.09818494e-01
4.89487708e-01 -7.24052072e-01 2.18965277e-01 -1.53060397e-03
7.29108080e-02 -2.86449268e-02 -9.20887172e-01 2.20195484e+00
-4.39603627e-01 3.43514770e-01 2.36025155e-01 -1.46634829e+00
8.70262384e-01 3.56207788e-01 7.52245545e-01 -4.43648130e-01
2.75646955e-01 2.83008546e-01 -2.32081696e-01 -5.37258327e-01
-1.32600650e-01 -2.73170948e-01 8.47105905e-02 7.25956440e-01
1.94475025e-01 -4.01044279e-01 -8.43143165e-02 6.69438019e-02
5.18377423e-01 4.14601237e-01 1.06521495e-01 -3.91059935e-01
5.22627890e-01 -1.47321850e-01 9.05578315e-01 4.06489313e-01
-2.20863327e-01 6.52328074e-01 1.99024498e-01 -3.46145689e-01
-6.43807173e-01 -9.72923696e-01 -1.94497392e-01 9.77920592e-01
4.29006636e-01 1.10540621e-01 -9.12892997e-01 -1.27770567e+00
-9.83956829e-02 2.49390721e-01 -8.02485883e-01 1.49786755e-01
-6.87303364e-01 -6.34718299e-01 2.52445847e-01 8.54981840e-01
4.40740496e-01 -1.00441134e+00 -5.20757079e-01 7.60221779e-02
-1.17371976e-01 -1.06662977e+00 -7.87313938e-01 1.60013586e-01
-1.16490352e+00 -1.28424573e+00 -1.22360718e+00 -1.28599763e+00
1.41989207e+00 3.36985648e-01 9.93529856e-01 4.49356884e-01
-4.24006641e-01 2.89319247e-01 -2.08527446e-01 -4.53020304e-01
-2.00222269e-01 1.29402488e-01 -4.20585155e-01 1.56513214e-01
1.19181454e-01 -6.03211880e-01 -7.58991003e-01 5.07375538e-01
-9.87241924e-01 4.39663768e-01 8.28058541e-01 8.39403629e-01
1.08342206e+00 -9.83886570e-02 5.68537056e-01 -1.43972242e+00
-8.03118944e-03 -2.60518551e-01 -5.53952932e-01 4.56406891e-01
-6.23068988e-01 2.45757952e-01 1.38324231e-01 -3.70752156e-01
-1.29165828e+00 5.22400618e-01 -1.25882894e-01 -3.52702349e-01
-3.13053638e-01 5.47279775e-01 -1.86173514e-01 -2.91876644e-01
2.19625250e-01 1.63029209e-01 1.95045158e-01 -5.50992846e-01
3.59139472e-01 5.03809631e-01 4.44167763e-01 -6.17830396e-01
7.39604115e-01 7.27174044e-01 -2.19342276e-01 -6.53948188e-01
-1.44312680e+00 -7.89812565e-01 -1.14887261e+00 -1.47670671e-01
1.19555688e+00 -8.80771756e-01 -4.20452505e-01 7.08107233e-01
-8.49988997e-01 -7.82729566e-01 -3.36787611e-01 4.63695109e-01
-3.89182568e-01 5.72677612e-01 -6.27705276e-01 -4.79446560e-01
-3.02144289e-01 -1.52713597e+00 1.22249544e+00 4.76746231e-01
1.86626047e-01 -1.38938272e+00 -2.00580768e-02 8.42923045e-01
2.71371715e-02 2.48900205e-01 5.70125818e-01 -9.14213359e-01
-6.49832129e-01 -1.51163921e-01 -4.00984287e-01 3.53042006e-01
4.16391760e-01 -3.68647635e-01 -7.61192679e-01 -2.58793950e-01
1.15111226e-03 -6.84526682e-01 1.08247137e+00 5.53522348e-01
1.43231940e+00 -1.90400202e-02 -5.95857084e-01 7.22357333e-01
1.16455388e+00 -1.50786504e-01 2.11552098e-01 -7.75512382e-02
1.25321329e+00 8.75738680e-01 6.85480952e-01 1.39872387e-01
5.93606055e-01 3.10422480e-01 2.99915761e-01 -7.37393618e-01
1.10421009e-01 -1.98725611e-01 -3.18149447e-01 9.63494539e-01
1.28350019e-01 2.54763156e-01 -1.07414258e+00 6.63979828e-01
-1.92250633e+00 -2.73997843e-01 2.42734634e-04 2.10462189e+00
1.37319934e+00 1.02047384e-01 2.39648409e-02 -5.96517883e-02
6.54734075e-01 -1.03939325e-01 -7.79730082e-01 3.43719929e-01
3.56391102e-01 2.02367157e-01 5.17156720e-01 5.52091360e-01
-1.37023199e+00 9.52899039e-01 5.25981188e+00 8.59369457e-01
-1.22144639e+00 1.12782016e-01 1.06064105e+00 3.05865705e-01
-3.49963844e-01 -2.28601262e-01 -5.83327174e-01 3.34800035e-01
2.88099051e-01 3.50894511e-01 -8.76529664e-02 7.32281685e-01
-7.30596781e-02 -2.44191214e-01 -1.07954574e+00 8.11467946e-01
2.33740844e-02 -1.31972575e+00 -2.17172027e-01 1.44935653e-01
1.13403201e+00 -3.38205062e-02 -5.00606373e-02 9.96794999e-02
5.86634099e-01 -1.06576419e+00 3.46024275e-01 3.84497255e-01
7.04336166e-01 -6.75401151e-01 6.58931434e-01 5.29936135e-01
-1.17039466e+00 3.34810525e-01 4.51312922e-02 2.93321162e-01
1.77490219e-01 7.96532273e-01 -6.80124700e-01 5.07836759e-01
5.22965789e-01 8.95223200e-01 -3.50456119e-01 1.06587589e+00
-6.04062676e-01 7.44599462e-01 -2.14588568e-01 1.92177802e-01
3.60937446e-01 -5.23267031e-01 2.11607426e-01 9.35805917e-01
-9.88009349e-02 1.85196027e-01 8.23415875e-01 1.01568794e+00
-1.84712023e-01 1.21196866e-01 -8.21195692e-02 1.62268028e-01
1.90442905e-01 1.39204371e+00 -1.30013585e+00 -2.43025124e-01
-4.07648265e-01 9.67374086e-01 4.35183465e-01 2.41544753e-01
-6.11596406e-01 -1.45729855e-01 -2.48582605e-02 -7.09007457e-02
2.99216300e-01 -7.25337937e-02 -5.49703121e-01 -1.00333238e+00
-1.48183271e-01 -4.17899549e-01 4.21256304e-01 -2.77912676e-01
-1.23802757e+00 3.46950918e-01 -1.68412372e-01 -1.06436503e+00
1.17943324e-01 -4.56995428e-01 -7.91333079e-01 7.38514781e-01
-1.79993951e+00 -1.53997886e+00 -2.94518113e-01 4.36379790e-01
4.49098408e-01 1.82521075e-01 6.66232765e-01 2.11577460e-01
-6.75572157e-01 4.96491313e-01 -2.62366943e-02 6.70283198e-01
7.02872097e-01 -1.68817163e+00 7.04019144e-02 5.19351840e-01
2.49588773e-01 4.01344985e-01 2.04837963e-01 -6.48323834e-01
-8.40001225e-01 -1.18563235e+00 4.42454785e-01 -1.46818548e-01
4.66220170e-01 -1.62532970e-01 -1.12162757e+00 6.96949184e-01
-6.69540744e-03 5.09953380e-01 1.04263318e+00 9.57307220e-03
-9.06756297e-02 1.69838041e-01 -1.05726695e+00 2.63934046e-01
7.68486142e-01 -3.71684819e-01 -4.27977800e-01 5.53027630e-01
6.96691513e-01 -7.21196532e-01 -8.40035677e-01 5.25992274e-01
2.13061497e-01 -6.45677507e-01 8.75182986e-01 -4.83945578e-01
2.52049655e-01 -1.68085039e-01 4.52503800e-01 -1.17649853e+00
2.53769606e-01 -5.41506767e-01 2.04381973e-01 1.14222479e+00
5.35355687e-01 -4.17725146e-01 1.21229184e+00 7.85015345e-01
-2.61812210e-01 -1.27350974e+00 -5.26559353e-01 -4.36023593e-01
3.63019884e-01 -3.07774335e-01 1.72625273e-01 1.15982711e+00
-9.68641639e-02 2.79886574e-01 5.31664640e-02 1.38055861e-01
8.79375815e-01 4.61794376e-01 4.98079538e-01 -1.44934464e+00
-3.36037815e-01 -3.66096675e-01 -1.22647788e-02 -1.30955040e+00
3.82324547e-01 -1.05371559e+00 3.59389305e-01 -1.68769157e+00
3.37848127e-01 -9.06642497e-01 -2.18105987e-01 6.36854231e-01
-5.27564585e-01 3.10915858e-01 -2.43610293e-01 2.82495826e-01
-8.95212054e-01 5.19194603e-01 1.75354111e+00 -1.87938824e-01
-3.33003461e-01 2.67136723e-01 -6.72632217e-01 1.03282714e+00
7.63952911e-01 -3.71453762e-01 -5.13646483e-01 -5.43717384e-01
-1.29642859e-02 1.65220156e-01 1.70216873e-01 -6.18574798e-01
4.75953758e-01 -1.73551932e-01 2.67631859e-01 -6.33318841e-01
-2.96675116e-02 -8.25100541e-01 -5.34136295e-01 3.29982042e-01
-4.28973049e-01 -7.15503693e-01 -5.48381656e-02 7.31863081e-01
-2.63621032e-01 -2.36431867e-01 8.48357379e-01 -3.90986770e-01
-4.66097295e-01 6.02480054e-01 -1.58739500e-02 3.40148419e-01
1.13640428e+00 -1.59786984e-01 2.37654984e-01 -1.30306110e-01
-1.04181480e+00 7.56543338e-01 2.16813624e-01 8.01374242e-02
6.90252125e-01 -9.80342150e-01 -5.20703435e-01 7.59025291e-02
-8.57390389e-02 1.06203723e+00 3.55317682e-01 1.10160995e+00
-3.69990021e-01 1.62330434e-01 5.52438013e-02 -9.13245261e-01
-1.14707851e+00 3.84084970e-01 3.73361140e-01 -7.07898855e-01
-5.13964713e-01 1.29191446e+00 5.84269226e-01 -8.51579428e-01
4.03909892e-01 -2.87216157e-01 -2.58093566e-01 -3.76656726e-02
3.15721929e-01 1.69898584e-01 -1.63367257e-01 -5.77849030e-01
-1.43872291e-01 6.02761686e-01 -2.72765130e-01 1.08297832e-01
1.36880469e+00 -9.62364003e-02 -1.05999991e-01 2.94855684e-01
1.32357943e+00 -2.92954385e-01 -1.63130307e+00 -6.55564010e-01
1.97814718e-01 -1.10856272e-01 3.12215656e-01 -7.58086920e-01
-1.49069285e+00 1.03794360e+00 4.11498249e-01 -2.02712715e-01
9.36419785e-01 1.20300077e-01 1.11371636e+00 1.52043015e-01
2.55207717e-01 -1.13626635e+00 4.04245317e-01 1.16473056e-01
3.67027134e-01 -1.80754304e+00 2.89364103e-02 -9.08009052e-01
-5.84553778e-01 8.66911232e-01 7.75110006e-01 -1.84767231e-01
9.60967898e-01 2.45280161e-01 3.72207850e-01 -2.78709650e-01
-2.52907295e-02 -2.77116269e-01 8.64609957e-01 4.53348041e-01
6.03838563e-01 1.51813969e-01 9.67044197e-03 7.68528283e-01
2.29713261e-01 -8.79047364e-02 2.32191524e-03 1.09437585e+00
-4.45755482e-01 -1.26965153e+00 -1.36520445e-01 5.36086261e-01
-5.70407569e-01 1.32379718e-02 -2.58719891e-01 6.46610677e-01
1.62172243e-02 6.02163732e-01 3.48143056e-02 5.88328987e-02
-1.32463992e-01 -1.80508286e-01 5.59808731e-01 -1.02657747e+00
-4.41146642e-01 2.73920208e-01 -4.95595753e-01 -4.10166591e-01
-7.17375100e-01 -5.97647846e-01 -1.79351783e+00 4.48484659e-01
-6.75470352e-01 2.79149860e-01 4.75955665e-01 1.35301769e+00
-2.48828046e-02 6.56693280e-01 6.33982182e-01 -9.40987885e-01
-3.77482444e-01 -7.54056931e-01 -5.10048330e-01 3.77705961e-01
3.07978719e-01 -6.12978637e-01 -3.45223360e-02 3.64909261e-01] | [14.620734214782715, -2.11541485786438] |
ba6dd1f3-ad56-4d0a-aa4a-d70b5145a152 | deep-gaussian-process-for-crop-yield | null | null | https://www.aaai.org/ocs/index.php/AAAI/AAAI17/paper/view/14435 | https://www.aaai.org/ocs/index.php/AAAI/AAAI17/paper/viewFile/14435/14067 | Deep Gaussian Process for Crop Yield Prediction Based on Remote Sensing Data | Agricultural monitoring, especially in developing countries,
can help prevent famine and support humanitarian efforts.
A central challenge is yield estimation, i.e., predicting crop
yields before harvest.
We introduce a scalable, accurate, and inexpensive method
to predict crop yields using publicly available remote sensing data. Our approach improves existing techniques in three
ways. First, we forego hand-crafted features traditionally
used in the remote sensing community and propose an approach based on modern representation learning ideas. We
also introduce a novel dimensionality reduction technique
that allows us to train a Convolutional Neural Network or
Long-short Term Memory network and automatically learn
useful features even when labeled training data are scarce.
Finally, we incorporate a Gaussian Process component to
explicitly model the spatio-temporal structure of the data
and further improve accuracy. We evaluate our approach on
county-level soybean yield prediction in the U.S. and show
that it outperforms competing techniques. | ['Xiaocheng Li', 'Melvin Low', 'Jiaxuan You', 'Stefano Ermon', 'David Lobell'] | 2017-02-12 | null | null | null | aaai-2017-2017-2 | ['crop-yield-prediction', 'crop-yield-prediction'] | ['computer-vision', 'miscellaneous'] | [ 7.97939748e-02 -1.90333202e-01 -3.19482893e-01 -3.19601029e-01
-5.44678748e-01 -6.90877974e-01 6.23217523e-01 3.67632568e-01
-2.13988423e-01 7.95168161e-01 3.66292834e-01 -6.98157609e-01
-1.01577625e-01 -1.44203365e+00 -6.66761041e-01 -7.75781691e-01
-2.34714597e-01 4.59770858e-02 -4.39855963e-01 -4.23178434e-01
1.04929656e-02 8.74232590e-01 -1.42042148e+00 -4.07648742e-01
9.81078625e-01 7.97510505e-01 5.82404077e-01 7.63331711e-01
1.30847096e-01 6.17434561e-01 -3.20494398e-02 -3.30498591e-02
1.62812397e-01 9.10104439e-02 -6.42417550e-01 -1.54353734e-02
-1.23459570e-01 -7.14004040e-01 -1.58349365e-01 9.04289067e-01
4.09446210e-01 1.62460618e-02 8.96350384e-01 -8.64914298e-01
-8.77728283e-01 5.40824056e-01 -9.42221701e-01 -1.06838733e-01
-2.43161738e-01 5.24369534e-03 6.50518656e-01 -7.35095620e-01
2.85654306e-01 1.20543849e+00 8.46271873e-01 5.25190979e-02
-1.39445913e+00 -6.54607713e-01 8.85606632e-02 -2.19701067e-01
-1.18446934e+00 -4.75617200e-01 5.70061088e-01 -5.45635521e-01
9.22672272e-01 2.01674588e-02 5.55255473e-01 8.29284072e-01
3.47153157e-01 7.94372082e-01 8.36204350e-01 -3.86652380e-01
8.21988806e-02 -4.06459808e-01 7.98005536e-02 8.08290541e-01
3.36358637e-01 3.93720716e-01 -5.89914620e-02 -3.72551471e-01
9.08368170e-01 4.39809382e-01 9.91875008e-02 -3.75302345e-01
-1.02709961e+00 1.15132022e+00 5.82398236e-01 2.51246721e-01
-9.58583176e-01 2.48263523e-01 1.91564068e-01 5.43205515e-02
9.64925826e-01 3.54699105e-01 -9.53710496e-01 3.57977189e-02
-1.39793468e+00 1.97050214e-01 6.97264493e-01 5.54911435e-01
9.42384124e-01 1.94664791e-01 1.24425493e-01 6.53033257e-01
2.07196116e-01 1.15741169e+00 1.14088409e-01 -9.55557108e-01
2.09387034e-01 3.58759493e-01 5.44722080e-01 -1.41936707e+00
-7.78696001e-01 -2.33652234e-01 -1.19442284e+00 1.28666028e-01
3.40763003e-01 -5.63359439e-01 -8.60736191e-01 1.65654469e+00
5.86463080e-04 -4.69744857e-03 2.34489933e-01 5.67517221e-01
3.47906888e-01 9.19072509e-01 3.29387814e-01 -1.66033730e-01
1.08626425e+00 -6.27209961e-01 -6.22324705e-01 -1.45863175e-01
8.09390962e-01 -4.90250170e-01 6.54285729e-01 2.08534077e-01
-6.19152963e-01 -2.19165042e-01 -6.84013367e-01 1.46649033e-01
-8.25286031e-01 7.77240276e-01 1.24225736e+00 4.27758992e-01
-1.09342587e+00 1.00683856e+00 -1.29400289e+00 -5.64921558e-01
5.73263586e-01 -1.25033762e-02 -4.32065398e-01 -4.03076075e-02
-9.52197373e-01 9.54219162e-01 4.71583575e-01 4.21612293e-01
-7.92864621e-01 -7.02130914e-01 -1.07204366e+00 4.66740161e-01
1.24200925e-01 -2.98967183e-01 9.32683885e-01 -6.93928063e-01
-1.27833021e+00 6.08568132e-01 -3.09261173e-01 -5.38680255e-01
-1.57710649e-02 -5.36612391e-01 -3.32862109e-01 -1.46247353e-02
1.81888327e-01 6.26978815e-01 7.48453200e-01 -1.00828969e+00
-5.70662379e-01 -5.81518352e-01 -3.66246223e-01 -2.18445107e-01
-6.06011569e-01 1.26255304e-01 4.29345459e-01 -8.38379920e-01
2.58616716e-01 -9.56934631e-01 -6.88423812e-01 1.01240128e-01
-1.26769871e-01 1.59981042e-01 6.35697663e-01 -1.20511460e+00
1.07715702e+00 -2.04565454e+00 -8.34836513e-02 2.74501294e-01
1.13374628e-01 4.12939399e-01 -3.96931440e-01 5.26280761e-01
-1.14912288e-02 7.51416162e-02 -3.87284577e-01 1.74532726e-01
-9.62390900e-02 3.99051279e-01 -5.62058806e-01 3.44299436e-01
8.70772958e-01 1.04318237e+00 -1.20376599e+00 5.74855171e-02
3.15332711e-01 5.27079105e-01 -2.02341348e-01 3.63927819e-02
-2.94428673e-02 2.33584046e-01 -6.09851480e-01 9.31963027e-01
1.00142634e+00 -1.88154757e-01 2.07381129e-01 9.75153521e-02
-4.27457571e-01 -1.18305743e-01 -7.58685172e-01 1.58127379e+00
-5.79288423e-01 4.74869430e-01 -4.85430397e-02 -1.32261527e+00
1.25314343e+00 1.74567536e-01 5.95560133e-01 -2.01340422e-01
-1.10948130e-01 2.48051256e-01 -4.55239296e-01 -3.01149815e-01
6.72822356e-01 1.98570937e-01 -8.81535560e-02 4.43296641e-01
8.46308395e-02 5.57706356e-02 -1.51078656e-01 -1.77506119e-01
9.72746491e-01 5.97285569e-01 6.19673073e-01 -6.44266784e-01
-1.41353700e-02 1.52242541e-01 6.44549906e-01 6.19203210e-01
-1.96269616e-01 3.31389189e-01 5.95725417e-01 -8.79053950e-01
-8.23496282e-01 -7.88399100e-01 -8.91035348e-02 1.50300336e+00
-5.51097631e-01 -9.56737772e-02 -2.91583925e-01 -6.59644306e-01
3.81929159e-01 6.59587443e-01 -7.73603380e-01 -5.77516779e-02
-2.46524110e-01 -1.33992803e+00 5.14418125e-01 9.46382701e-01
3.83745760e-01 -1.00802457e+00 -7.82112479e-01 5.56690454e-01
-1.97532907e-01 -6.63680553e-01 2.85453290e-01 6.23309672e-01
-1.09322667e+00 -6.48605168e-01 -1.00212038e+00 -3.86596769e-01
3.94474566e-01 4.10197556e-01 1.24798036e+00 -4.40945297e-01
-1.93318561e-01 -1.42542869e-01 -3.50412935e-01 -7.06405997e-01
-1.68917805e-01 5.63728809e-01 4.25423235e-02 -4.68198270e-01
8.47330749e-01 -6.55512512e-01 -3.69616002e-01 -2.92687029e-01
-6.97847188e-01 -5.76433726e-02 8.20316076e-01 9.34548497e-01
2.47163817e-01 4.82487455e-02 7.70026684e-01 -5.62067986e-01
2.66945809e-01 -8.43410730e-01 -8.46078932e-01 5.12248516e-01
-5.33397496e-01 4.20266271e-01 1.94423154e-01 -3.68299574e-01
-8.48343551e-01 6.93632960e-01 4.46655042e-02 -1.77648708e-01
-3.95678699e-01 1.07166874e+00 -1.43165160e-02 1.77120164e-01
4.89331365e-01 1.44279137e-01 1.32824536e-02 -6.66185975e-01
5.81219494e-01 6.57190621e-01 4.73633260e-01 -4.28815722e-01
8.53343546e-01 3.97502661e-01 1.68937638e-01 -1.00206232e+00
-6.47229791e-01 -3.58611822e-01 -9.67276096e-01 1.91878751e-01
6.01121545e-01 -1.21127295e+00 -3.80833656e-01 5.69087267e-01
-1.13612437e+00 -5.02739131e-01 -1.12989880e-01 6.84884667e-01
-5.96049666e-01 -1.03277201e-02 -4.39367473e-01 -9.35147226e-01
-5.41366518e-01 -5.54402173e-01 1.25146985e+00 1.72033221e-01
1.80284545e-01 -1.06112027e+00 1.38605908e-01 -4.63943094e-01
8.10646892e-01 7.34323442e-01 8.36506128e-01 -2.41305709e-01
2.59961840e-02 -4.09837067e-02 -7.76733458e-01 1.46940589e-01
6.88517749e-01 1.92910105e-01 -1.09026623e+00 -2.09142104e-01
-4.89899248e-01 -3.22126478e-01 1.31137121e+00 9.96642232e-01
1.02637446e+00 -3.02215815e-01 -4.50599045e-01 6.48721457e-01
1.41026270e+00 -2.05962017e-01 4.04208839e-01 2.04758927e-01
5.27533054e-01 8.49778771e-01 7.23618567e-01 7.17594087e-01
4.82751667e-01 1.14454422e-02 3.82711321e-01 -5.67485988e-01
5.02300441e-01 -3.07165772e-01 2.31705993e-01 4.18433219e-01
-3.51240486e-01 5.25942072e-02 -1.19356406e+00 8.90945911e-01
-2.14583302e+00 -1.11204731e+00 1.25897869e-01 1.91305041e+00
7.60959566e-01 -4.74841237e-01 5.81324622e-02 8.68828967e-02
2.72598028e-01 2.34501556e-01 -5.14799535e-01 -4.03742343e-01
-3.09546590e-01 5.44874370e-01 1.18459582e+00 3.01393688e-01
-1.69706595e+00 1.43285465e+00 6.76674747e+00 4.53644931e-01
-1.19233549e+00 -2.30723172e-01 6.70300007e-01 1.60657734e-01
-6.07692972e-02 -1.51919618e-01 -6.72728717e-01 -1.05378531e-01
9.82706845e-01 -1.64373945e-02 2.51537204e-01 8.49420369e-01
6.19908750e-01 -1.36622518e-01 -6.22188628e-01 4.59311277e-01
-3.13004106e-01 -1.22062492e+00 -2.98366547e-01 2.50233412e-01
6.78949177e-01 2.34246001e-01 7.52787367e-02 2.72691488e-01
1.03517580e+00 -1.22404313e+00 6.53425008e-02 8.83069217e-01
7.60615945e-01 -1.00745654e+00 6.81175053e-01 4.07145292e-01
-1.49075246e+00 -1.77908033e-01 -8.52267742e-01 -4.09708589e-01
-2.42126584e-01 9.02857602e-01 -3.53759587e-01 6.12520158e-01
6.87574387e-01 1.22270536e+00 -5.82579374e-01 5.62324882e-01
-1.84717879e-01 8.45875144e-01 -5.22400796e-01 2.96756655e-01
5.85219026e-01 -3.04266155e-01 3.93492803e-02 1.13214540e+00
7.23388195e-01 1.19009204e-01 2.67327607e-01 8.23359609e-01
1.37124762e-01 8.00301656e-02 -1.14936578e+00 -3.05714577e-01
2.61278570e-01 1.27481067e+00 -4.34679210e-01 9.48492214e-02
-3.31868768e-01 7.83440113e-01 2.97155201e-01 4.76949722e-01
-3.86416405e-01 -5.06826699e-01 7.51942694e-01 -2.67593801e-01
5.31259775e-01 -5.55941105e-01 -1.37109339e-01 -1.27143216e+00
-1.84242576e-01 -5.83729446e-01 -2.97880173e-02 -6.16348863e-01
-1.15766740e+00 1.53453708e-01 -2.19708443e-01 -8.82893264e-01
-6.04936540e-01 -6.56315625e-01 -4.94922519e-01 1.21486771e+00
-2.20862246e+00 -1.65557301e+00 -2.21302077e-01 1.77360162e-01
1.53497666e-01 -2.45183662e-01 1.42619038e+00 -1.09663822e-01
-5.70469916e-01 8.27279985e-02 7.04157293e-01 2.69244432e-01
4.67549890e-01 -1.14062846e+00 8.11603785e-01 1.01562107e+00
-1.16463855e-01 3.78748238e-01 3.95928681e-01 -6.54304743e-01
-1.29254413e+00 -1.65278912e+00 1.26735234e+00 -2.83357743e-02
9.40543473e-01 -8.09860155e-02 -8.54144394e-01 7.63950169e-01
-2.01860890e-01 2.37162709e-01 7.17779636e-01 5.31095266e-01
-3.66785973e-01 -1.13709375e-01 -1.15886784e+00 2.16774479e-01
6.39146328e-01 -3.03296089e-01 -2.43603662e-01 2.53844559e-01
4.92072821e-01 3.48778009e-01 -9.35505986e-01 4.29773033e-01
8.53274405e-01 -2.71095306e-01 8.80553842e-01 -8.28085124e-01
6.65900588e-01 1.46948010e-01 -4.71153080e-01 -1.64426172e+00
-6.95892513e-01 -5.08540630e-01 -2.50623465e-01 1.19393134e+00
3.39760393e-01 -4.66663659e-01 6.54487669e-01 4.64866817e-01
5.09202778e-01 -3.45400989e-01 -2.32710928e-01 -8.52790356e-01
7.58256257e-01 -1.85176805e-01 1.02717161e+00 1.08923328e+00
-1.15346968e-01 -1.88851476e-01 -5.95914185e-01 5.21350980e-01
4.92834121e-01 3.79437834e-01 6.23689055e-01 -1.65020764e+00
1.60649493e-01 -2.76704222e-01 -3.57785344e-01 -7.22608984e-01
4.13372874e-01 -4.89109814e-01 3.36534232e-02 -1.36753500e+00
4.66845669e-02 -6.61719263e-01 -3.62457693e-01 1.09981334e+00
-5.12653708e-01 -2.44110972e-02 -8.65516737e-02 -6.01869449e-02
2.31314570e-01 5.94639182e-01 6.81653619e-01 -7.97797069e-02
-3.76774371e-01 1.14815503e-01 -8.10270488e-01 7.42884755e-01
1.35700893e+00 -4.70914632e-01 -6.76218048e-02 -6.97325647e-01
1.74286097e-01 6.12391457e-02 4.81187075e-01 -8.30736578e-01
-4.06925291e-01 -7.13896692e-01 7.49988079e-01 -8.26789796e-01
-1.56603113e-01 -7.65539587e-01 -2.84918785e-01 5.58563650e-01
-1.36129111e-01 -2.22365116e-03 4.54711348e-01 4.26288694e-01
1.62345737e-01 4.66347784e-02 4.04972583e-01 1.40246479e-02
-6.52808964e-01 3.26477349e-01 -6.07392192e-01 -6.34008408e-01
6.95548177e-01 4.44065332e-01 -1.14513732e-01 -1.83828264e-01
-6.50355577e-01 4.15346891e-01 2.94140995e-01 5.44017792e-01
3.96362752e-01 -1.23702753e+00 -1.30430090e+00 4.63556468e-01
2.90849507e-01 -3.62135470e-01 -2.77777880e-01 5.10358751e-01
-6.46672964e-01 4.71161395e-01 -3.85431468e-01 -3.32477272e-01
-7.58788049e-01 5.57371199e-01 1.16470847e-02 -5.64415038e-01
-5.62824667e-01 5.13239384e-01 -2.08713278e-01 -6.82610035e-01
-6.20579831e-02 -6.35312378e-01 -5.49823165e-01 3.09032977e-01
7.57365167e-01 1.74418390e-01 -8.99156928e-02 -6.25788987e-01
-2.27626711e-01 2.31807664e-01 2.25190431e-01 -2.67951172e-02
2.01084614e+00 -1.08465195e-01 1.32425334e-02 4.89743263e-01
9.15435612e-01 -5.25402725e-01 -1.35487235e+00 -4.02648062e-01
3.62950236e-01 -3.78699601e-01 6.27930701e-01 -7.17200458e-01
-1.20535612e+00 1.18163168e+00 8.34338963e-01 5.60832955e-02
1.32613909e+00 -5.03810644e-01 6.24346673e-01 9.54624414e-01
2.43207768e-01 -9.45030391e-01 -5.85772216e-01 5.86236894e-01
6.64887369e-01 -1.63885248e+00 -1.83651801e-02 4.76686321e-02
-3.82742256e-01 1.28049541e+00 -7.88158625e-02 -3.08732688e-01
1.05110681e+00 4.86351103e-01 3.28038968e-02 7.27992505e-02
-5.67057669e-01 -6.92061365e-01 -6.88990280e-02 1.10565984e+00
6.60778105e-01 6.11974120e-01 3.88967968e-03 5.53228021e-01
-6.48121089e-02 4.00578052e-01 2.75723994e-01 1.13064945e+00
-7.41208971e-01 -9.48723733e-01 -3.88718992e-01 7.57952869e-01
-3.31709236e-01 -5.13574600e-01 -1.16280787e-01 3.76289368e-01
-2.58295238e-01 9.67800915e-01 3.59893590e-02 -1.10046498e-01
-1.49703503e-01 1.82000741e-01 2.19064176e-01 -4.61357236e-01
-2.75778860e-01 8.83431137e-02 -2.94285238e-01 -5.58118224e-01
-5.91897428e-01 -7.73902535e-01 -6.30346298e-01 -7.33237207e-01
-5.04523039e-01 -1.15920275e-01 8.68435264e-01 9.15846288e-01
4.70483243e-01 2.03330666e-01 9.66577470e-01 -1.21112871e+00
-6.91954613e-01 -1.13506925e+00 -9.17025745e-01 -2.48358816e-01
6.00150883e-01 -6.08251154e-01 1.55839652e-01 1.44554541e-01] | [9.365687370300293, -1.589890480041504] |
c5e1a419-b2df-4849-8861-1ffd0ebe63dd | reducing-information-loss-for-spiking-neural | 2307.04356 | null | https://arxiv.org/abs/2307.04356v1 | https://arxiv.org/pdf/2307.04356v1.pdf | Reducing Information Loss for Spiking Neural Networks | The Spiking Neural Network (SNN) has attracted more and more attention recently. It adopts binary spike signals to transmit information. Benefitting from the information passing paradigm of SNNs, the multiplications of activations and weights can be replaced by additions, which are more energy-efficient. However, its ``Hard Reset" mechanism for the firing activity would ignore the difference among membrane potentials when the membrane potential is above the firing threshold, causing information loss. Meanwhile, quantifying the membrane potential to 0/1 spikes at the firing instants will inevitably introduce the quantization error thus bringing about information loss too. To address these problems, we propose to use the ``Soft Reset" mechanism for the supervised training-based SNNs, which will drive the membrane potential to a dynamic reset potential according to its magnitude, and Membrane Potential Rectifier (MPR) to reduce the quantization error via redistributing the membrane potential to a range close to the spikes. Results show that the SNNs with the ``Soft Reset" mechanism and MPR outperform their vanilla counterparts on both static and dynamic datasets. | ['Zhe Ma', 'Xuhui Huang', 'Yuanyuan Ou', 'Xinyi Tong', 'Xiaode Liu', 'Liwen Zhang', 'Yuanpei Chen', 'Yufei Guo'] | 2023-07-10 | null | null | null | null | ['quantization'] | ['methodology'] | [ 5.75985253e-01 -2.83677518e-01 6.17215037e-02 -1.80629492e-01
2.75032246e-04 -3.04094374e-01 3.07491839e-01 3.69877517e-01
-7.02802718e-01 8.98142457e-01 -1.27188653e-01 4.40606549e-02
2.21983328e-01 -1.07874346e+00 -7.65052736e-01 -1.28356063e+00
1.57361969e-01 -2.74542212e-01 6.79883897e-01 -1.50628313e-01
5.56977808e-01 6.01011291e-02 -1.43534911e+00 3.62867028e-01
8.11249495e-01 1.21708298e+00 2.73478806e-01 3.47225443e-02
-3.26365799e-01 4.05673951e-01 -6.49673760e-01 -2.30105426e-02
2.42260724e-01 -4.95579839e-01 -1.17089279e-01 -7.47855306e-01
-3.49691510e-01 -7.09453374e-02 -5.97372770e-01 1.35688007e+00
6.80526316e-01 -3.53973210e-02 4.19544160e-01 -1.18547809e+00
-5.60701907e-01 8.63816977e-01 -5.34867644e-01 4.97712016e-01
-7.88868442e-02 3.73127580e-01 4.40468192e-01 -4.88807321e-01
5.06924331e-01 7.47693837e-01 5.45766413e-01 5.45724452e-01
-1.47122490e+00 -9.89139259e-01 1.66399866e-01 1.19256325e-01
-1.64781010e+00 -3.66884410e-01 8.32206309e-01 -3.75623368e-02
9.70917165e-01 -2.11215932e-02 9.38128710e-01 7.72427738e-01
5.54428518e-01 5.17319679e-01 1.03408909e+00 -6.05316870e-02
7.43097544e-01 -1.26475558e-01 -9.72371772e-02 1.62391141e-02
5.63187122e-01 -1.73033431e-01 -5.51601529e-01 1.13606885e-01
9.24849391e-01 1.51438996e-01 -5.61868131e-01 1.27197206e-01
-9.64004636e-01 3.07009190e-01 7.58078039e-01 4.71312851e-01
-4.67268080e-01 5.08741498e-01 3.52856487e-01 2.28582606e-01
-2.77631342e-01 5.73072195e-01 -2.98949718e-01 -1.06441185e-01
-8.72844875e-01 1.59015521e-01 4.40932214e-01 3.36931467e-01
7.85455525e-01 2.10114196e-01 -2.79660851e-01 5.64684331e-01
2.46922940e-01 3.89014244e-01 8.56322110e-01 -8.29674125e-01
2.34747857e-01 1.04877663e+00 5.90791777e-02 -7.99532950e-01
-2.87536353e-01 -3.36187631e-01 -1.21671259e+00 2.79703885e-01
5.60904980e-01 -1.57396853e-01 -9.34276760e-01 2.06621289e+00
-1.76082537e-01 2.94642955e-01 1.33737504e-01 8.84115934e-01
5.97966194e-01 7.87583292e-01 1.51579797e-01 -4.44009960e-01
1.18436241e+00 -5.83899878e-02 -9.09161210e-01 -4.04392987e-01
2.95965284e-01 -2.28128240e-01 7.94183493e-01 -2.85970811e-02
-1.23056424e+00 -2.31468663e-01 -1.29775703e+00 1.44661114e-01
-3.39476496e-01 -4.30964261e-01 4.38973278e-01 2.98718482e-01
-9.07743990e-01 7.79863298e-01 -9.12719369e-01 -6.25952482e-02
6.12977266e-01 4.37983066e-01 5.44097275e-02 4.31697965e-01
-1.44768023e+00 7.92594612e-01 4.22113687e-01 2.46645808e-01
-4.55021322e-01 -6.17842674e-01 -5.25575280e-01 3.10347438e-01
-1.75478593e-01 -3.81656259e-01 7.71261632e-01 -7.91655660e-01
-1.34232140e+00 6.31274045e-01 -2.09857062e-01 -7.65347838e-01
-5.51351048e-02 4.87313449e-01 -1.32042542e-01 -1.52143717e-01
-2.96829581e-01 8.22820127e-01 3.90426457e-01 -8.88817489e-01
-9.68751460e-02 -3.96276355e-01 -4.16615129e-01 1.50985196e-01
-3.37502569e-01 -4.15683359e-01 -7.97805405e-05 -5.94050407e-01
4.48073149e-01 -5.87059736e-01 -1.76867962e-01 1.23624213e-01
-1.50518119e-01 -8.94840434e-02 5.97787797e-01 -6.22821711e-02
1.32038629e+00 -2.35804534e+00 -3.81549029e-03 1.87172249e-01
-2.29087397e-02 3.26245010e-01 2.40702420e-01 2.68387854e-01
3.92768495e-02 1.43936247e-01 -5.88017762e-01 3.45556349e-01
-3.72493178e-01 2.65337914e-01 -2.67776996e-01 3.22738111e-01
3.36468071e-01 8.48331630e-01 -8.81112814e-01 -1.88390657e-01
7.74664339e-03 6.92187905e-01 -4.31165725e-01 -2.03842446e-01
-7.60975899e-03 3.15737098e-01 -2.33251899e-01 3.61850351e-01
8.50472450e-01 2.31575370e-02 8.25297013e-02 -3.33586156e-01
-6.77506387e-01 5.99890590e-01 -1.18436086e+00 1.38658237e+00
9.47565734e-02 5.64989805e-01 9.60502103e-02 -1.05271327e+00
1.35204685e+00 2.65671939e-01 4.26595598e-01 -1.21649647e+00
5.68705142e-01 3.26016396e-01 4.07074392e-01 3.79461311e-02
-1.03806160e-01 -1.60756499e-01 -2.45396164e-04 1.71444461e-01
-3.46778452e-01 1.47013431e-02 7.21152648e-02 -5.72146215e-02
1.14280021e+00 -1.20503582e-01 -9.45890248e-02 -4.13026899e-01
3.98611158e-01 -2.76974320e-01 1.10114753e+00 3.59100759e-01
-3.61876607e-01 2.81094223e-01 6.84698105e-01 -2.93733239e-01
-9.19570923e-01 -1.10697865e+00 -4.04963255e-01 4.95496273e-01
5.71628869e-01 2.41174712e-03 -8.36344719e-01 3.38516444e-01
-9.57766175e-02 6.56525970e-01 -4.95745569e-01 -6.45676851e-01
-5.46785533e-01 -9.74132001e-01 7.10387230e-01 5.69302082e-01
9.34084177e-01 -1.29841840e+00 -1.02244937e+00 5.51047444e-01
-2.11264580e-01 -7.24000216e-01 -2.33512282e-01 8.58751714e-01
-9.49195385e-01 -5.66087425e-01 -5.74306846e-01 -7.65897274e-01
8.91316295e-01 -2.77674288e-01 3.90478194e-01 1.50558695e-01
-3.08540940e-01 -5.86418271e-01 -5.64569347e-02 -5.03175974e-01
2.95285173e-02 -1.10463895e-01 -9.14221257e-02 -7.45978132e-02
4.89749908e-01 -9.62509811e-01 -1.04602575e+00 2.57100314e-02
-1.07555807e+00 1.81359857e-01 4.68441963e-01 5.91696858e-01
9.33506668e-01 1.21987998e-01 7.32084334e-01 -2.86031574e-01
4.26111966e-01 -2.41111651e-01 -4.62393731e-01 -1.45481169e-01
-5.89843214e-01 4.42800641e-01 7.77629316e-01 -6.02489531e-01
-8.38047147e-01 6.92721903e-02 -1.91732273e-01 -3.00377663e-02
3.07059824e-01 -5.20777926e-02 -2.40565211e-01 -1.87810957e-01
5.72510123e-01 6.25823259e-01 -1.80880770e-01 -1.01291858e-01
-2.00638890e-01 3.10019523e-01 6.40063345e-01 -2.39944637e-01
4.68768209e-01 5.48312783e-01 5.63734695e-02 -8.44840854e-02
-1.42907187e-01 8.77072141e-02 -2.36377478e-01 -8.53042230e-02
5.67538857e-01 -6.12917185e-01 -1.10804486e+00 9.41890180e-01
-1.14959347e+00 -1.92671403e-01 -3.61657262e-01 3.68179679e-01
-2.87708163e-01 5.83721735e-02 -8.41723800e-01 -7.84558058e-01
-7.37067938e-01 -9.35096085e-01 2.60745555e-01 8.16919744e-01
4.96039391e-02 -4.93765324e-01 1.53281882e-01 -5.98498821e-01
8.78540158e-01 2.26364940e-01 9.91568863e-01 -5.04726768e-01
-4.06037927e-01 5.74794300e-02 -3.72735202e-01 1.20078482e-01
-1.05334476e-01 -5.11932001e-03 -9.63766813e-01 -4.15858552e-02
1.75126106e-01 4.12524454e-02 9.04574335e-01 5.14055610e-01
1.21874809e+00 -2.20011771e-01 -5.58334231e-01 6.56749070e-01
1.67004156e+00 6.93189025e-01 1.06305790e+00 3.11139286e-01
1.98429480e-01 1.41453324e-02 -2.27895364e-01 4.47444469e-01
3.34540129e-01 2.84064472e-01 5.87675750e-01 2.05502217e-03
7.54908919e-02 -2.98500329e-01 2.81796664e-01 7.97886133e-01
-5.85298799e-02 -7.81775936e-02 -6.32498860e-01 5.34627080e-01
-1.47801864e+00 -8.94357085e-01 -2.30814055e-01 2.71413994e+00
1.35189927e+00 4.51983571e-01 -5.44002771e-01 4.75653678e-01
9.11825836e-01 3.07083335e-02 -9.85074520e-01 -6.58035278e-01
-3.52270216e-01 3.56960148e-01 8.17533791e-01 -4.04089913e-02
-2.21297875e-01 4.54276919e-01 6.16582537e+00 3.83116752e-01
-1.52392685e+00 -2.45242640e-01 4.58264470e-01 -9.28413197e-02
-5.89163244e-01 6.74270540e-02 -7.26660371e-01 1.30822194e+00
8.86030257e-01 -1.93569809e-01 7.22458661e-01 2.51823783e-01
1.79826602e-01 -3.55358481e-01 -7.28745639e-01 1.11176848e+00
-5.60890973e-01 -1.42830086e+00 8.75573084e-02 -2.42578134e-01
4.31151301e-01 7.70337656e-02 -1.74511299e-01 2.77639538e-01
-7.64843747e-02 -8.04726005e-01 5.57474196e-01 5.85343480e-01
5.68788171e-01 -6.36171818e-01 7.76812792e-01 4.61636603e-01
-1.36132479e+00 -1.19513065e-01 -6.25872254e-01 -3.88580590e-01
1.34966031e-01 8.01887751e-01 -3.16771775e-01 -7.03671426e-02
8.40451300e-01 2.49152645e-01 -2.56243020e-01 1.21107829e+00
-5.39575145e-03 3.56534988e-01 -7.26844430e-01 -1.70381576e-01
-1.43395394e-01 -3.41085017e-01 1.54663220e-01 7.92207003e-01
3.58935088e-01 2.67656237e-01 -4.17863071e-01 1.16628015e+00
-4.51035529e-01 -2.30145901e-01 -3.39061737e-01 -2.15709750e-02
1.06907248e+00 9.78754520e-01 -9.75162864e-01 -3.01784545e-01
9.52359010e-03 8.50175619e-01 -1.51556116e-02 3.87502611e-01
-6.79891229e-01 -9.24972832e-01 4.63748276e-01 2.75408506e-01
4.31686360e-03 -4.35205828e-03 -8.65564167e-01 -6.73401654e-01
1.79738909e-01 5.96810691e-02 8.15917552e-02 -6.93361640e-01
-6.77736104e-01 4.63943273e-01 -4.40198064e-01 -9.52959836e-01
2.37034947e-01 -2.36800879e-01 -1.02307999e+00 8.74283910e-01
-1.76095974e+00 -2.47808155e-02 -3.35151017e-01 5.06269157e-01
7.85757154e-02 6.25346065e-01 6.11876488e-01 3.93119395e-01
-6.45328045e-01 4.48545396e-01 -6.08032662e-03 2.09523335e-01
5.13669789e-01 -7.97729492e-01 3.55551004e-01 5.83665192e-01
-5.78043520e-01 6.48905754e-01 6.39990687e-01 -3.85705709e-01
-1.44397843e+00 -7.70789027e-01 9.57082510e-01 1.99295536e-01
3.57293606e-01 -4.86627460e-01 -1.46616411e+00 2.00019807e-01
4.67796549e-02 -4.69295569e-02 3.99702132e-01 -8.27436268e-01
-2.22312003e-01 -4.82181937e-01 -1.45039380e+00 6.50040925e-01
7.76690543e-01 -1.48559287e-01 -5.34515262e-01 -2.16423050e-01
4.78656650e-01 -1.86275482e-01 -7.08358884e-01 6.65464222e-01
5.25774539e-01 -8.20533097e-01 5.93014121e-01 1.12634726e-01
5.48573136e-02 -7.27551341e-01 7.35312626e-02 -1.12744129e+00
-2.09360451e-01 -4.14603949e-01 2.59744078e-01 1.19252002e+00
4.41345423e-01 -8.56628537e-01 4.69367564e-01 6.38184309e-01
6.95940778e-02 -8.42504680e-01 -1.23759830e+00 -3.97280842e-01
5.10541983e-02 -9.28096399e-02 4.11271751e-01 4.50541079e-01
3.37735802e-01 -1.47356465e-02 1.70074806e-01 -5.44953309e-02
4.27262276e-01 -3.17450315e-01 -1.35166958e-01 -1.16348207e+00
2.05056608e-01 -7.23687947e-01 -6.19484007e-01 -1.03012347e+00
-3.89417559e-01 -9.25081909e-01 4.90323246e-01 -1.47684979e+00
1.54598445e-01 -3.91681641e-01 -7.97343373e-01 6.29386008e-01
-7.97070935e-02 3.79224628e-01 5.07430956e-02 1.95517585e-01
-2.33456671e-01 4.09948856e-01 1.05095804e+00 1.45710446e-03
-4.00285006e-01 -2.80717194e-01 -4.94356483e-01 5.64422309e-01
8.88036907e-01 -6.26720607e-01 -1.56735033e-01 -3.16779673e-01
5.30465007e-01 -1.99811742e-01 3.13900650e-01 -1.05156982e+00
5.84684610e-01 -2.93090213e-02 6.06392145e-01 -3.50278020e-01
1.68275818e-01 -5.80971479e-01 2.94164896e-01 8.96439135e-01
-2.29328543e-01 -2.89719086e-02 3.12211335e-01 3.53850186e-01
1.69394966e-02 -1.59889713e-01 1.06031477e+00 -5.77224232e-02
-3.81673276e-01 1.78173766e-01 -8.01181734e-01 6.62863553e-02
1.10081553e+00 -6.62666440e-01 -4.61724997e-01 1.60778925e-01
-3.34648609e-01 2.65906334e-01 5.61297476e-01 8.90585333e-02
6.01114213e-01 -1.34713244e+00 -4.26757962e-01 7.69904017e-01
-1.99542508e-01 7.72062913e-02 2.97384709e-01 9.10323322e-01
-2.49699473e-01 2.24003375e-01 -5.81583619e-01 -4.80571687e-01
-5.77146351e-01 3.44086885e-01 4.98314112e-01 1.11198165e-01
-4.35028940e-01 8.53780329e-01 -1.53203771e-01 2.06426561e-01
2.29003489e-01 -4.36985463e-01 -1.64407939e-01 -1.28475651e-01
5.54335058e-01 2.59211212e-01 7.27657378e-02 1.39889424e-03
-6.44825816e-01 5.55630684e-01 1.60750926e-01 1.44582298e-02
1.08110309e+00 -8.73791948e-02 -2.06331134e-01 5.81728220e-01
8.88445258e-01 -4.96686190e-01 -1.46762061e+00 -5.41274324e-02
-1.12027578e-01 -1.71073139e-01 1.13155998e-01 -6.82963789e-01
-1.19883823e+00 9.02330756e-01 5.71262300e-01 1.73928499e-01
1.36282897e+00 -2.63039231e-01 1.13288021e+00 1.87185377e-01
3.67993951e-01 -1.21595132e+00 -2.70334065e-01 2.86349386e-01
3.29560161e-01 -6.34677827e-01 -4.93923992e-01 4.70217783e-03
-4.25143123e-01 1.07809508e+00 6.01178288e-01 -3.01095724e-01
5.06167293e-01 8.72938037e-01 -2.44454622e-01 5.79352416e-02
-8.48163843e-01 6.31534085e-02 -4.69847679e-01 5.05316257e-01
3.27478379e-01 -1.80924028e-01 -7.58284271e-01 6.85616255e-01
2.00173438e-01 3.73927146e-01 4.51036960e-01 8.63587320e-01
-8.75931621e-01 -8.12454224e-01 -1.45755783e-01 5.15740097e-01
-2.95716792e-01 -2.38144696e-01 -5.91388568e-02 1.59444124e-01
3.83436263e-01 5.58031559e-01 6.64369762e-01 -1.72566295e-01
3.75095338e-01 2.83815771e-01 3.08864266e-01 -2.60305047e-01
-5.90108931e-01 -4.46967222e-02 -7.44523823e-01 -2.79091179e-01
-1.25911817e-01 -3.65637064e-01 -2.36149001e+00 -3.63731861e-01
-3.47019941e-01 -2.32860763e-02 8.95694792e-01 8.15302730e-01
3.19875091e-01 7.25657403e-01 6.21751070e-01 -4.11371648e-01
-5.00253320e-01 -6.97938204e-01 -5.69736838e-01 3.57514203e-01
1.03541739e-01 -4.92469430e-01 -6.52529538e-01 -1.22427270e-01] | [8.2382230758667, 2.5096311569213867] |
b923ecdb-7cf7-4cac-8b65-5801ae79f542 | automatic-hemisphere-segmentation-in-rodent | 2108.01941 | null | https://arxiv.org/abs/2108.01941v3 | https://arxiv.org/pdf/2108.01941v3.pdf | Automatic cerebral hemisphere segmentation in rat MRI with lesions via attention-based convolutional neural networks | We present MedicDeepLabv3+, a convolutional neural network that is the first completely automatic method to segment cerebral hemispheres in magnetic resonance (MR) volumes of rats with lesions. MedicDeepLabv3+ improves the state-of-the-art DeepLabv3+ with an advanced decoder, incorporating spatial attention layers and additional skip connections that, as we show in our experiments, lead to more precise segmentations. MedicDeepLabv3+ requires no MR image preprocessing, such as bias-field correction or registration to a template, produces segmentations in less than a second, and its GPU memory requirements can be adjusted based on the available resources. We optimized MedicDeepLabv3+ and six other state-of-the-art convolutional neural networks (DeepLabv3+, UNet, HighRes3DNet, V-Net, VoxResNet, Demon) on a heterogeneous training set comprised by MR volumes from 11 cohorts acquired at different lesion stages. Then, we evaluated the trained models and two approaches specifically designed for rodent MRI skull stripping (RATS and RBET) on a large dataset of 655 MR rat brain volumes. In our experiments, MedicDeepLabv3+ outperformed the other methods, yielding an average Dice coefficient of 0.952 and 0.944 in the brain and contralateral hemisphere regions. Additionally, we show that despite limiting the GPU memory and the training data, our MedicDeepLabv3+ also provided satisfactory segmentations. In conclusion, our method, publicly available at https://github.com/jmlipman/MedicDeepLabv3Plus, yielded excellent results in multiple scenarios, demonstrating its capability to reduce human workload in rat neuroimaging studies. | ['Jussi Tohka', 'Riccardo de Feo', 'Artem Shatillo', 'Juan Miguel Valverde'] | 2021-08-04 | null | null | null | null | ['skull-stripping'] | ['medical'] | [-3.74905735e-01 -6.28601685e-02 3.04806143e-01 -3.32868576e-01
-6.52185798e-01 -3.50547612e-01 2.78861433e-01 2.31698573e-01
-9.62569773e-01 7.74305582e-01 4.04836610e-03 -4.67591226e-01
9.60254893e-02 -5.91633618e-01 -8.42773497e-01 -4.33946103e-01
-6.09836817e-01 6.67424262e-01 5.47538042e-01 -1.91017613e-01
1.43012926e-01 8.98024082e-01 -8.31784546e-01 2.07529232e-01
8.63816261e-01 7.61694193e-01 4.60831821e-01 5.22779882e-01
1.26149490e-01 5.85607231e-01 -7.10041374e-02 -3.56629491e-01
3.34769219e-01 -3.59020196e-02 -1.06067979e+00 -6.35387659e-01
4.45445836e-01 -5.95909774e-01 -5.29397190e-01 8.61022294e-01
1.00533986e+00 -3.72121371e-02 5.25068820e-01 -7.44803727e-01
-5.49889863e-01 7.82922983e-01 -7.11114943e-01 9.77453947e-01
-3.46165925e-01 5.22166967e-01 2.05937773e-01 -1.03049588e+00
9.59680915e-01 8.07834029e-01 1.08111227e+00 5.80695868e-01
-1.42233384e+00 -9.37302232e-01 -3.04617912e-01 1.97970703e-01
-1.32862651e+00 -2.97776639e-01 2.48570800e-01 -9.16602552e-01
1.05385840e+00 2.99946759e-02 7.80284286e-01 9.43878591e-01
9.61004615e-01 3.91541123e-01 1.38428223e+00 2.76486725e-01
2.12238699e-01 -4.66411591e-01 5.16134799e-01 8.33487809e-01
2.48074189e-01 6.77697957e-02 2.39785612e-02 -2.18709856e-01
1.13170540e+00 -2.04800338e-01 -4.96934474e-01 -3.75992835e-01
-1.61603475e+00 8.76347840e-01 9.38709915e-01 5.27535439e-01
-7.43194997e-01 3.38957012e-01 8.40872765e-01 -1.73284262e-01
5.72851062e-01 4.63673562e-01 -3.53533506e-01 2.85740077e-01
-1.21652818e+00 1.34672269e-01 2.91195810e-01 8.20720613e-01
2.27652237e-01 1.74708933e-01 -4.01544094e-01 7.92012930e-01
9.31550097e-03 2.68711478e-01 8.91863704e-01 -9.63323236e-01
3.51141959e-01 1.14677958e-01 -3.98259133e-01 -6.81091964e-01
-1.05595398e+00 -6.63883388e-01 -1.08574796e+00 4.19012368e-01
2.77799219e-01 -4.66181487e-01 -1.33698750e+00 1.51491714e+00
1.41116172e-01 -3.52007121e-01 -5.53101122e-01 1.09718370e+00
1.23600197e+00 2.09324047e-01 2.70192027e-01 1.24593332e-01
1.29608083e+00 -1.11247039e+00 -4.33611125e-01 -1.95087746e-01
8.83587837e-01 -4.20799464e-01 7.70777881e-01 2.06998125e-01
-1.30521560e+00 -1.38194859e-01 -1.05850708e+00 -1.02064982e-01
-4.20938998e-01 4.30542380e-02 7.41645753e-01 4.63019133e-01
-1.60759890e+00 9.37567592e-01 -1.34551752e+00 -2.09938601e-01
9.49880540e-01 6.10539854e-01 -7.50809908e-01 1.64597377e-01
-9.35169637e-01 1.34858656e+00 3.67200524e-01 1.87102109e-01
-1.09768796e+00 -1.43715179e+00 -6.50796831e-01 -1.15990095e-01
-1.21301226e-01 -7.93769479e-01 1.04295182e+00 -2.59110123e-01
-1.16585231e+00 1.12996185e+00 3.33227307e-01 -9.58930969e-01
1.03937781e+00 1.48743644e-01 -9.41164419e-02 4.76948887e-01
2.15473667e-01 1.13010538e+00 1.30888626e-01 -9.03675914e-01
1.87388286e-01 -5.37772715e-01 -5.41433990e-01 -2.09579706e-01
5.13841867e-01 3.36909801e-01 -1.60889924e-01 -6.42552853e-01
9.18592885e-02 -7.02308953e-01 -5.17954588e-01 1.15449108e-01
-1.40949205e-01 2.91738182e-01 2.93592691e-01 -1.22639871e+00
6.00620210e-01 -1.68372893e+00 3.48808356e-02 2.53560990e-01
8.97329032e-01 3.77703249e-01 -8.44022930e-02 -2.06821829e-01
-6.43981278e-01 2.34050065e-01 -7.43178666e-01 1.84743721e-02
-3.42447370e-01 -1.85896337e-01 2.17139661e-01 1.04085243e+00
-4.19639677e-01 1.18434250e+00 -7.80066252e-01 -5.93446553e-01
3.64547253e-01 7.21237242e-01 -7.35121429e-01 -1.09489206e-02
5.87525666e-01 8.25589597e-01 -3.55973765e-02 2.54735649e-01
8.35851431e-01 -6.48031756e-02 -1.57510594e-01 -1.96672142e-01
-1.80116400e-01 -2.61269093e-01 -5.14190495e-01 1.88189983e+00
-4.75747854e-01 5.14596581e-01 4.88208354e-01 -9.72149849e-01
6.44051015e-01 3.74362320e-01 9.75056946e-01 -9.30845976e-01
6.82606399e-01 5.91186583e-01 3.78632724e-01 -3.96481454e-01
-9.40412953e-02 -7.16812313e-02 4.36409533e-01 3.71700704e-01
6.89570725e-01 -3.25341642e-01 3.33938181e-01 1.44183323e-01
1.12449634e+00 -9.81860086e-02 6.35915250e-02 -9.64326799e-01
3.86380851e-01 -2.83111315e-02 3.09557855e-01 5.93284011e-01
-7.16589212e-01 9.43054140e-01 4.75466847e-01 -5.54334700e-01
-1.27148807e+00 -9.64542627e-01 -6.97939157e-01 7.09984601e-01
-3.02580953e-01 -6.69303015e-02 -1.25713611e+00 -5.55888057e-01
-1.86509445e-01 6.40534222e-01 -9.68240678e-01 2.08793506e-01
-9.01407182e-01 -1.09239709e+00 6.62418604e-01 5.71355164e-01
5.78012109e-01 -1.16529572e+00 -6.95216596e-01 5.02716422e-01
-7.45063052e-02 -1.00635827e+00 -5.03451169e-01 3.01726013e-01
-9.94744718e-01 -1.08624899e+00 -1.34725106e+00 -7.06883729e-01
6.93683743e-01 -1.46519765e-01 1.14126074e+00 2.17454344e-01
-6.09828293e-01 -2.47772671e-02 -3.17211449e-02 -2.24478573e-01
-1.33939549e-01 1.26324207e-01 -6.23190925e-02 -6.11265659e-01
-4.82007042e-02 -9.43921983e-01 -1.14890718e+00 1.87623262e-01
-6.80682898e-01 1.97711095e-01 3.87323707e-01 7.58819222e-01
6.73635364e-01 -8.30852926e-01 5.25209785e-01 -8.15907300e-01
5.40686131e-01 -6.60397589e-01 -6.16385818e-01 -8.86822939e-02
-4.06071395e-01 -1.98534355e-01 5.79502821e-01 -2.02071384e-01
-5.51187277e-01 -8.88025090e-02 -6.68626726e-01 -4.09542590e-01
-3.10531855e-02 3.77479643e-01 5.24249673e-01 -5.33901870e-01
9.46224630e-01 7.01847821e-02 2.11251244e-01 -3.43624532e-01
2.04578504e-01 1.53783560e-01 7.13518620e-01 -3.31771523e-01
1.42428309e-01 4.80143219e-01 -1.09221630e-01 -4.39459056e-01
-1.35114998e-01 -2.24547327e-01 -1.00824606e+00 -2.91513771e-01
1.29978836e+00 -6.50955617e-01 -5.31538248e-01 3.97789925e-01
-1.12808096e+00 -8.40214193e-01 -4.42936085e-02 7.52480209e-01
-6.08195603e-01 2.38930970e-01 -9.78300989e-01 5.81648499e-02
-1.29064286e+00 -1.88803315e+00 5.67400932e-01 -1.82735696e-01
-3.04296523e-01 -8.98423254e-01 4.52111624e-02 3.02865297e-01
9.21583056e-01 5.49502313e-01 9.81442451e-01 -8.87660384e-01
-9.17914063e-02 -3.24419397e-03 -4.93904531e-01 1.21571936e-01
-3.75057995e-01 -4.22028810e-01 -5.93889594e-01 -4.40849900e-01
-1.43088251e-01 -2.60885268e-01 9.63955939e-01 9.02837992e-01
1.63425994e+00 2.67537031e-02 -2.01696053e-01 1.04393208e+00
1.29098511e+00 2.07176417e-01 6.29683375e-01 5.05492449e-01
6.05516374e-01 1.05410583e-01 -2.30080739e-01 7.83364012e-05
2.33293235e-01 4.49293137e-01 4.77038503e-01 -4.87240762e-01
-5.66102684e-01 3.77076626e-01 -1.37465537e-01 9.53304231e-01
-3.65428418e-01 2.97650009e-01 -1.33102667e+00 7.23182082e-01
-1.32865584e+00 -6.93522036e-01 -5.01676500e-01 2.02399111e+00
8.70361567e-01 -2.48838812e-01 4.81187403e-02 -4.63712394e-01
7.66118705e-01 6.54189512e-02 -5.94363153e-01 -4.78122592e-01
2.58776881e-02 7.29216874e-01 9.27305102e-01 4.88804966e-01
-1.00363958e+00 7.93335974e-01 6.32931089e+00 6.72666669e-01
-1.41326392e+00 8.23867798e-01 8.87790084e-01 -1.83874220e-01
2.08452314e-01 -5.14249504e-01 -2.91464090e-01 4.97513026e-01
8.46763670e-01 -1.52158827e-01 5.47047734e-01 6.84005320e-01
1.54407144e-01 4.56680864e-04 -9.30119634e-01 8.36973548e-01
-9.92880091e-02 -1.52493691e+00 -2.91310281e-01 -1.79044940e-02
5.49787760e-01 1.16177535e+00 -5.27472645e-02 2.88979530e-01
3.39660823e-01 -1.30767643e+00 7.95963526e-01 1.93348691e-01
1.07204819e+00 -5.73202670e-01 9.98754978e-01 -3.21710855e-02
-6.70523584e-01 3.90185207e-01 -4.11151350e-01 6.60935819e-01
2.65111953e-01 6.01276457e-01 -8.05993497e-01 3.13245475e-01
8.91075134e-01 2.86804736e-01 -7.21796334e-01 1.35729969e+00
-4.93108444e-02 4.57137555e-01 -1.65574208e-01 4.45625126e-01
4.42029744e-01 -1.37424767e-01 4.43924457e-01 1.56119859e+00
3.50076526e-01 4.35952336e-01 -1.25965998e-01 1.17588222e+00
-2.38302946e-01 4.44613993e-01 -2.12452143e-01 5.60611069e-01
-5.57584222e-04 1.56504607e+00 -1.09194148e+00 -3.79470348e-01
-7.54147843e-02 6.19977593e-01 5.05984604e-01 1.87762529e-01
-1.09161389e+00 -4.15078223e-01 2.31669933e-01 4.50269312e-01
1.67251304e-01 -2.96682626e-01 -6.23167515e-01 -1.09947515e+00
-3.13287854e-01 -6.11276984e-01 1.98838383e-01 -1.03079104e+00
-9.97047842e-01 1.06729019e+00 -1.47647196e-02 -6.82777405e-01
2.98047930e-01 -5.71897268e-01 -5.94652116e-01 1.05944216e+00
-1.35090446e+00 -8.18766952e-01 -2.35942423e-01 6.01311505e-01
3.25707763e-01 -5.18010035e-02 5.91411173e-01 4.90734607e-01
-5.24531603e-01 3.45744967e-01 1.26454204e-01 4.08292681e-01
4.94440049e-01 -1.10246825e+00 5.69707632e-01 8.41299951e-01
-6.29185796e-01 7.46521652e-01 4.46912497e-01 -7.76794255e-01
-9.33463633e-01 -1.32638824e+00 4.36912149e-01 -6.07252903e-02
8.30109000e-01 -2.11201355e-01 -1.06850791e+00 1.00917029e+00
4.74679649e-01 6.45966530e-01 4.12572652e-01 -5.19188762e-01
-4.36749309e-03 2.86284417e-01 -1.57555819e+00 5.30908525e-01
1.09478688e+00 -2.35179849e-02 -7.06984043e-01 5.87927699e-01
5.07040262e-01 -9.48108494e-01 -1.33808529e+00 5.15099168e-01
4.62863088e-01 -8.89563322e-01 1.15454376e+00 -4.25196230e-01
6.66105926e-01 3.50185670e-02 2.73613095e-01 -1.67023790e+00
-5.57586908e-01 -1.83260858e-01 2.41092414e-01 5.27433157e-01
5.29078782e-01 -8.96970451e-01 3.77037466e-01 7.31939197e-01
-6.11026525e-01 -9.27156091e-01 -1.20383573e+00 -5.56169748e-01
9.36529040e-01 -2.68241078e-01 5.60992479e-01 8.45881641e-01
-2.15820730e-01 -2.83950001e-01 1.88798770e-01 -1.36001632e-01
7.18770087e-01 -2.67747223e-01 2.40785182e-01 -9.66681063e-01
3.22833002e-01 -8.25051129e-01 -4.17218834e-01 -3.82059515e-01
2.50905484e-01 -1.60107493e+00 -5.12701124e-02 -1.78647125e+00
3.93134236e-01 -3.89558792e-01 -3.14185113e-01 7.03756750e-01
1.49319634e-01 5.76924145e-01 1.34635448e-01 9.60191339e-03
-3.32364365e-02 3.46550018e-01 1.65259981e+00 -1.86770752e-01
-1.00681156e-01 -6.17340386e-01 -3.90057206e-01 7.31842399e-01
9.89967942e-01 -4.43075567e-01 7.67562091e-02 -6.38347805e-01
-4.37106699e-01 1.21170647e-01 6.19593322e-01 -1.20933199e+00
1.08882502e-01 3.94887388e-01 8.53510618e-01 -5.03962696e-01
-1.11368477e-01 -4.66022879e-01 1.58609405e-01 7.43218660e-01
-1.75250828e-01 3.18758637e-01 4.82675284e-01 -2.95866907e-01
2.42494583e-01 -2.59545110e-02 1.36677909e+00 -4.09439206e-01
-3.47398043e-01 8.85983527e-01 -7.16285706e-01 1.28991157e-01
8.64191711e-01 -1.57805532e-02 -3.78681719e-01 1.78136557e-01
-1.04197061e+00 2.43464485e-01 1.76151514e-01 5.85152358e-02
5.50427258e-01 -1.24411380e+00 -9.62384045e-01 9.21361595e-02
-2.76124418e-01 -1.08592793e-01 7.16727555e-01 1.73040259e+00
-1.35793996e+00 6.24830782e-01 -8.59542310e-01 -7.17296422e-01
-9.87591743e-01 5.69062948e-01 8.76060486e-01 -3.18523765e-01
-1.12669957e+00 7.28605330e-01 3.03607583e-01 -8.89341056e-01
-1.66202962e-01 -6.02775216e-01 -3.08789819e-01 -2.07841620e-01
4.71787184e-01 4.39979881e-01 6.82573259e-01 -8.85064840e-01
-5.03410041e-01 2.98160970e-01 -1.90894529e-01 -8.40068702e-03
1.81257439e+00 1.07651278e-01 -3.75696003e-01 -2.61605918e-01
1.36118305e+00 -3.42945039e-01 -9.57455754e-01 6.31286651e-02
-2.60080665e-01 -1.31160654e-02 6.42704666e-01 -9.58085299e-01
-1.81738234e+00 8.88861775e-01 9.57203269e-01 -5.27947843e-01
7.12249339e-01 -2.05075353e-01 9.83179927e-01 -1.87647879e-01
6.80078924e-01 -7.21755445e-01 -5.59832692e-01 6.13922775e-01
1.32249236e+00 -1.06398273e+00 6.98423684e-02 -8.57484639e-02
-6.00056827e-01 9.80306685e-01 7.13490307e-01 -3.24286014e-01
7.73532629e-01 3.02217364e-01 1.08934455e-02 -6.28186703e-01
-1.16347291e-01 1.66187197e-01 1.77994594e-01 5.40874600e-01
7.17743456e-01 1.89492032e-01 -6.85971856e-01 7.65484929e-01
-5.81827283e-01 1.14884250e-01 6.04102015e-01 7.41453409e-01
-1.66845128e-01 -4.65673059e-01 -2.64193207e-01 7.08557725e-01
-8.36011171e-01 -3.75059247e-01 4.85582687e-02 8.28164816e-01
7.92067796e-02 3.18098962e-01 -1.21544920e-01 1.05247743e-01
3.68579060e-01 -2.44048730e-01 5.88567972e-01 -3.45046163e-01
-1.06867433e+00 -1.18185237e-01 -1.66923374e-01 -8.42992663e-01
7.21587315e-02 -4.57444012e-01 -1.50370181e+00 -5.83083034e-01
7.22522736e-02 -7.80298263e-02 7.48463690e-01 8.40316236e-01
4.10248578e-01 8.26394796e-01 1.09646045e-01 -1.24629259e+00
-2.22712141e-02 -1.10517251e+00 -5.88757396e-01 7.41803870e-02
7.45868236e-02 -7.40226090e-01 -4.31323275e-02 -2.26608321e-01] | [14.150733947753906, -2.388071298599243] |
511ba1d8-d552-4714-b6cb-ef5bd56fcb26 | bongard-logo-a-new-benchmark-for-human-level | 2010.00763 | null | https://arxiv.org/abs/2010.00763v4 | https://arxiv.org/pdf/2010.00763v4.pdf | Bongard-LOGO: A New Benchmark for Human-Level Concept Learning and Reasoning | Humans have an inherent ability to learn novel concepts from only a few samples and generalize these concepts to different situations. Even though today's machine learning models excel with a plethora of training data on standard recognition tasks, a considerable gap exists between machine-level pattern recognition and human-level concept learning. To narrow this gap, the Bongard problems (BPs) were introduced as an inspirational challenge for visual cognition in intelligent systems. Despite new advances in representation learning and learning to learn, BPs remain a daunting challenge for modern AI. Inspired by the original one hundred BPs, we propose a new benchmark Bongard-LOGO for human-level concept learning and reasoning. We develop a program-guided generation technique to produce a large set of human-interpretable visual cognition problems in action-oriented LOGO language. Our benchmark captures three core properties of human cognition: 1) context-dependent perception, in which the same object may have disparate interpretations given different contexts; 2) analogy-making perception, in which some meaningful concepts are traded off for other meaningful concepts; and 3) perception with a few samples but infinite vocabulary. In experiments, we show that the state-of-the-art deep learning methods perform substantially worse than human subjects, implying that they fail to capture core human cognition properties. Finally, we discuss research directions towards a general architecture for visual reasoning to tackle this benchmark. | ['Animashree Anandkumar', 'Yuke Zhu', 'Ankit B. Patel', 'Lei Mao', 'Zhiding Yu', 'Weili Nie'] | 2020-10-02 | null | http://proceedings.neurips.cc/paper/2020/hash/bf15e9bbff22c7719020f9df4badc20a-Abstract.html | http://proceedings.neurips.cc/paper/2020/file/bf15e9bbff22c7719020f9df4badc20a-Paper.pdf | neurips-2020-12 | ['novel-concepts'] | ['reasoning'] | [ 4.25785750e-01 1.52697548e-01 9.51601043e-02 -4.21115279e-01
-2.49064267e-01 -6.39317513e-01 8.83786321e-01 3.63317817e-01
-1.69020548e-01 3.77490312e-01 3.17536108e-02 -6.38504088e-01
-1.55344307e-01 -9.04527903e-01 -7.17926502e-01 -2.69597441e-01
-9.99735370e-02 4.97705638e-01 2.23282248e-01 -5.17149508e-01
3.60473424e-01 3.10086370e-01 -2.04080939e+00 8.79628360e-01
9.22811925e-01 1.15874815e+00 2.49552876e-01 6.26150191e-01
-3.09365243e-01 1.11649394e+00 -8.63480747e-01 -2.84973592e-01
2.17364579e-01 -5.61197221e-01 -1.03946722e+00 1.45867497e-01
8.13376427e-01 2.29257327e-02 4.57515977e-02 9.89457667e-01
9.60030332e-02 3.82140130e-01 8.65149796e-01 -1.46153152e+00
-1.57030225e+00 3.32646430e-01 -2.26626113e-01 1.91861719e-01
6.93265438e-01 4.27321792e-01 1.10684383e+00 -8.12045157e-01
3.32418621e-01 1.77815282e+00 4.07212168e-01 8.76467466e-01
-1.29988539e+00 -3.28031927e-01 4.07793760e-01 6.69609308e-01
-1.10197043e+00 -2.39764713e-02 5.11822045e-01 -6.23200893e-01
1.28272510e+00 3.44304532e-01 9.72351491e-01 1.01348114e+00
2.52095550e-01 8.63003433e-01 1.46474957e+00 -6.48278236e-01
6.35053992e-01 6.24527708e-02 2.48284549e-01 6.67793274e-01
2.53012508e-01 3.33296090e-01 -5.37968576e-01 9.35878009e-02
7.09917068e-01 2.15624526e-01 -3.64562348e-02 -4.86306518e-01
-1.25397384e+00 7.50871539e-01 8.96036148e-01 2.14474723e-01
-1.89846024e-01 1.70215681e-01 3.23988885e-01 5.60697794e-01
-1.30948380e-01 7.93877959e-01 -3.10058206e-01 2.38915309e-01
-3.57221842e-01 4.77351725e-01 7.62097239e-01 9.82213557e-01
8.01189899e-01 1.62149861e-01 -1.95613414e-01 6.98579192e-01
5.06166033e-02 4.17367071e-01 7.84344435e-01 -9.76654828e-01
3.13411728e-02 8.90348732e-01 -1.52138844e-01 -9.88652050e-01
-2.85926431e-01 -3.68307024e-01 -8.69495392e-01 5.84892809e-01
3.80634755e-01 3.47060800e-01 -1.08108723e+00 1.74515021e+00
-5.63503504e-02 -1.86034501e-01 3.43097627e-01 8.87842953e-01
8.12318444e-01 6.56767249e-01 3.30311626e-01 2.93894857e-02
1.51247251e+00 -1.14687622e+00 -1.73365712e-01 -7.47290134e-01
3.00431013e-01 -2.25619748e-01 1.67928028e+00 5.45568526e-01
-8.54418635e-01 -9.43832874e-01 -1.22353184e+00 -3.18570882e-01
-8.83078575e-01 -3.76438349e-01 1.19542420e+00 6.31049275e-01
-1.10349357e+00 1.99260950e-01 -2.39563644e-01 -5.60150981e-01
7.09338009e-01 2.37870812e-02 -8.41815323e-02 -2.97696412e-01
-9.37897503e-01 1.08585930e+00 6.80360496e-01 -2.38389850e-01
-1.05293894e+00 -7.07382560e-01 -8.34065557e-01 3.16509366e-01
6.64330781e-01 -1.02907968e+00 1.36741209e+00 -1.46061850e+00
-9.90063190e-01 1.19549584e+00 3.42241041e-02 -5.99045753e-01
2.08350793e-01 -1.68889970e-01 -5.29998839e-01 8.77787769e-02
1.37745842e-01 9.02755141e-01 8.97546053e-01 -1.45344365e+00
-7.08546937e-01 -2.11979210e-01 7.67147601e-01 1.33542612e-01
-5.18692248e-02 -4.19731110e-01 9.63592753e-02 -6.22567773e-01
-3.38994572e-03 -7.61605024e-01 -1.51947409e-01 4.49682713e-01
7.94091355e-03 -5.89286506e-01 5.58992028e-01 -2.55430508e-02
6.75474346e-01 -1.92428946e+00 1.39371991e-01 -3.28761935e-02
4.16670501e-01 2.74192423e-01 -3.79806727e-01 2.75381446e-01
-1.38132706e-01 1.09693296e-02 -2.13451162e-01 1.90443829e-01
3.04882497e-01 4.80966151e-01 -8.17445636e-01 -2.36336544e-01
1.73610449e-01 1.22647381e+00 -1.38835001e+00 -2.92875141e-01
1.96529195e-01 4.42736633e-02 -6.88398242e-01 2.25374430e-01
-6.35881662e-01 1.84960306e-01 -2.42580429e-01 7.32508838e-01
3.25287133e-01 -5.40795922e-01 2.37742245e-01 1.59275308e-01
3.00022990e-01 5.98207451e-02 -8.27982485e-01 1.80616128e+00
-5.55135429e-01 7.39582837e-01 -5.72704136e-01 -1.31972754e+00
9.31043983e-01 -4.30869311e-02 -2.77509212e-01 -8.96878660e-01
-1.15350537e-01 -8.03182721e-02 4.12713468e-01 -5.67514241e-01
3.56617749e-01 -4.78194922e-01 -1.31981820e-01 4.02304053e-01
-1.92780308e-02 -6.08653426e-01 3.07154655e-01 1.40916318e-01
9.50441003e-01 1.05431341e-02 8.62905502e-01 -4.08610165e-01
5.55850506e-01 2.56178081e-01 3.84917676e-01 1.15148067e+00
-3.01830173e-01 3.93059850e-01 4.82621908e-01 -1.03162706e+00
-6.56536937e-01 -1.47316003e+00 2.81576574e-01 1.54528582e+00
1.88406512e-01 -4.26955909e-01 -6.72540188e-01 -4.06799763e-01
-1.47185475e-02 9.10788000e-01 -7.75313735e-01 -4.27938759e-01
-2.67624676e-01 -3.93976450e-01 3.95522416e-01 9.26134408e-01
6.75710082e-01 -1.50891817e+00 -1.19035065e+00 -8.55859146e-02
4.50808108e-02 -8.62556636e-01 6.20445833e-02 9.21562687e-03
-7.60540426e-01 -1.29055917e+00 -3.87339532e-01 -9.35675919e-01
6.81244016e-01 5.14440358e-01 1.70370448e+00 4.82772291e-01
-5.88174105e-01 7.97317207e-01 -4.05366510e-01 -8.29497814e-01
-4.94374126e-01 -5.48048735e-01 4.50931676e-02 -3.70188743e-01
7.02908218e-01 -5.70448339e-01 -5.49476743e-01 1.30878881e-01
-1.05724704e+00 2.54131347e-01 6.80053234e-01 1.00220931e+00
4.34560597e-01 -4.47370932e-02 6.67560279e-01 -4.85802203e-01
1.04106534e+00 -3.62451643e-01 -3.54748100e-01 6.83893025e-01
-4.93713707e-01 1.52609959e-01 9.36747789e-01 -7.29103923e-01
-1.05867994e+00 -2.42978051e-01 3.78955215e-01 -1.88872561e-01
-3.83836359e-01 5.99998653e-01 -1.45135731e-01 2.18805745e-01
1.09935760e+00 4.84385103e-01 -2.51231790e-01 -1.55167188e-02
9.15097058e-01 3.94947231e-01 7.25182176e-01 -1.12678146e+00
5.42338908e-01 4.75725323e-01 2.97349337e-02 -9.87943292e-01
-1.24331248e+00 -5.65027334e-02 -6.31756306e-01 -5.56938052e-02
9.03577149e-01 -6.81125700e-01 -9.46294188e-01 1.13819927e-01
-1.19521570e+00 -3.51890504e-01 -6.25356257e-01 -1.58176824e-01
-7.91107476e-01 2.45034426e-01 -1.77246660e-01 -7.38456547e-01
-4.19201814e-02 -8.72297764e-01 8.07266533e-01 3.62424701e-01
-4.93521303e-01 -8.87249231e-01 -1.42175987e-01 4.29165781e-01
3.32403392e-01 1.68810070e-01 1.54425502e+00 -5.95823944e-01
-7.29452312e-01 2.30172426e-01 -4.36723769e-01 4.97855186e-01
3.65106799e-02 -4.79964733e-01 -9.47579026e-01 -2.64257044e-01
-1.73346251e-02 -8.88406038e-01 8.88222277e-01 3.68097275e-02
1.59904838e+00 -1.32279739e-01 -1.52496964e-01 3.17994893e-01
1.23204267e+00 5.74344575e-01 8.05234551e-01 2.28390411e-01
3.98024917e-01 5.56443810e-01 3.91818583e-01 1.05426848e-01
4.20793027e-01 3.44955117e-01 3.53799790e-01 2.26324692e-01
-1.97838157e-01 -3.68073136e-01 6.89446032e-02 3.82885516e-01
-4.38482463e-01 -1.30935654e-01 -1.16645217e+00 5.85458755e-01
-1.89832580e+00 -1.18310487e+00 3.22376609e-01 2.00158358e+00
8.05375457e-01 2.84702867e-01 -3.93806621e-02 1.39816105e-01
3.56310576e-01 1.33252470e-02 -7.22166657e-01 -6.83252096e-01
-1.70655742e-01 5.14235854e-01 -3.97526294e-01 2.43820027e-01
-7.47100294e-01 1.06105769e+00 6.76736116e+00 6.92054152e-01
-9.54214513e-01 -6.71627671e-02 4.38011616e-01 1.92510173e-01
-2.81759173e-01 1.98628396e-01 -2.47444615e-01 1.97643824e-02
5.19914627e-01 -3.27428699e-01 5.33948123e-01 1.10318327e+00
-3.97625566e-01 -7.96631649e-02 -1.68704557e+00 1.28003669e+00
3.57261360e-01 -1.49224567e+00 6.96045458e-01 -1.98039889e-01
6.47004068e-01 -4.53812629e-01 2.08567962e-01 8.67745817e-01
3.25320005e-01 -1.57499814e+00 8.15144539e-01 6.07453167e-01
5.10228753e-01 -4.57027346e-01 3.13117094e-02 3.84965628e-01
-1.13627803e+00 -4.74815190e-01 -8.36972117e-01 -7.29127526e-01
-3.34174663e-01 2.33418316e-01 -5.54715633e-01 1.41320094e-01
6.69433475e-01 5.94029963e-01 -8.31212819e-01 7.41239727e-01
-6.16940439e-01 1.93634197e-01 2.34855026e-01 -3.81714791e-01
2.33847946e-01 9.10982341e-02 3.02879155e-01 1.03319550e+00
1.26016438e-01 3.40745270e-01 2.18401819e-01 1.44645572e+00
1.39629483e-01 -4.05570537e-01 -5.68527281e-01 -1.17948875e-01
3.59961182e-01 8.57400537e-01 -6.70006037e-01 -6.82521164e-01
-6.02486610e-01 1.04669714e+00 5.70321739e-01 5.38077712e-01
-5.28371632e-01 -2.42452383e-01 7.63558149e-01 -4.21118066e-02
1.04445353e-01 -2.71258533e-01 -3.09874982e-01 -1.24062514e+00
-4.12645721e-04 -1.24928129e+00 5.68004370e-01 -1.21533215e+00
-1.62708521e+00 5.38082123e-01 1.70823589e-01 -1.10638964e+00
-3.54747295e-01 -1.34185612e+00 -7.59637415e-01 5.84168434e-01
-1.16531992e+00 -1.30494142e+00 -5.01582980e-01 6.48117959e-01
8.65458429e-01 -4.83944654e-01 1.11366785e+00 -4.39607799e-01
1.73609421e-01 2.88715661e-01 -4.00301397e-01 -6.59067603e-03
2.81317204e-01 -1.50045073e+00 6.40517235e-01 5.97867489e-01
3.83338839e-01 1.09327328e+00 5.83461463e-01 -3.16709667e-01
-1.25535691e+00 -7.55463481e-01 6.69615567e-01 -6.74574912e-01
5.19888759e-01 -5.27152836e-01 -1.01082742e+00 8.16343904e-01
2.08394080e-01 -1.56781133e-02 7.53029048e-01 3.65934759e-01
-1.03014481e+00 -4.94848341e-02 -9.61693287e-01 1.06732023e+00
1.32689989e+00 -7.64357150e-01 -1.56416512e+00 4.10322249e-01
7.82945216e-01 -7.51610473e-02 -3.21589053e-01 6.09527193e-02
6.55120492e-01 -1.16361523e+00 1.34398854e+00 -1.24052787e+00
7.40768373e-01 -3.64799649e-01 -3.41777861e-01 -1.51332629e+00
-4.88941550e-01 -3.83738220e-01 -2.42173687e-01 5.61331868e-01
1.40529022e-01 -6.20786428e-01 4.64844018e-01 6.75163448e-01
-1.11052856e-01 -6.16351485e-01 -7.06176460e-01 -1.06861508e+00
4.73434061e-01 -6.02757335e-01 7.09468067e-01 8.93553436e-01
3.02993327e-01 6.75528288e-01 3.19679417e-02 -1.39382571e-01
6.09373569e-01 5.69180131e-01 6.95178509e-01 -1.42104447e+00
-5.28028607e-01 -6.25944853e-01 -6.27730191e-01 -1.28812075e+00
3.52310330e-01 -9.68049526e-01 -1.94500126e-02 -1.57913089e+00
3.31574678e-01 -3.58498469e-02 -3.52344126e-01 6.40008390e-01
-1.20508514e-01 -1.38599994e-02 4.30145860e-01 3.10083497e-02
-6.95654869e-01 3.38606656e-01 1.45796418e+00 -5.69437265e-01
7.00834487e-03 -3.07599962e-01 -1.09791684e+00 9.53852236e-01
5.72932243e-01 -5.14969453e-02 -8.43566775e-01 -5.69918513e-01
6.26381695e-01 -2.86961496e-01 1.08260167e+00 -1.27960360e+00
2.25649953e-01 -5.89839995e-01 5.81645787e-01 -3.03954512e-01
3.77562582e-01 -7.43749559e-01 -3.78253758e-01 6.54877245e-01
-6.13618374e-01 3.51194739e-02 2.74178267e-01 6.50032520e-01
-1.12974204e-01 -8.03600475e-02 7.42415011e-01 -6.31497204e-01
-1.61455643e+00 -1.29377812e-01 -5.52216887e-01 3.34850281e-01
8.95591557e-01 -4.71549898e-01 -7.17730165e-01 -4.48642284e-01
-7.72953629e-01 1.23546071e-01 3.15521300e-01 7.78663158e-01
9.07777190e-01 -1.25803530e+00 -5.21665037e-01 7.52254128e-02
6.16229415e-01 -1.85829520e-01 1.93664819e-01 4.53896858e-02
-3.49087179e-01 4.83395606e-01 -6.81198418e-01 -6.20682120e-01
-9.11565721e-01 9.80357766e-01 3.80211145e-01 1.86131150e-01
-6.27189636e-01 8.37480009e-01 7.86158741e-01 -4.69177186e-01
1.01654947e-01 -7.68321633e-01 -4.35570627e-03 -3.77947420e-01
9.14864659e-01 1.84045792e-01 -2.69050837e-01 -2.61183113e-01
-3.19806725e-01 3.52087647e-01 3.77928168e-02 1.34346902e-01
1.01464176e+00 1.51191175e-01 -1.23429947e-01 5.50470650e-01
7.15720117e-01 -6.56307697e-01 -1.03594708e+00 -1.32337019e-01
6.08770289e-02 -4.21922892e-01 -4.27197278e-01 -1.16928411e+00
-4.47312146e-01 1.17545807e+00 5.65675557e-01 2.52600133e-01
1.17794621e+00 1.52331486e-01 3.45635831e-01 1.06826639e+00
4.98026878e-01 -1.02089620e+00 6.50121152e-01 8.17340970e-01
1.39022517e+00 -1.30139625e+00 -3.50402333e-02 -2.57013232e-01
-7.11103499e-01 1.18821025e+00 9.26379979e-01 -1.70268625e-01
3.32728148e-01 -2.26304710e-01 1.12591619e-02 -3.64330679e-01
-9.62871730e-01 -3.23193163e-01 6.37339652e-01 1.21626687e+00
4.03631568e-01 2.37496614e-01 6.74113035e-02 7.63522804e-01
-4.48347092e-01 9.18777883e-02 4.33715045e-01 1.04780912e+00
-6.64903998e-01 -8.12948227e-01 -3.53073359e-01 2.56212771e-01
3.19618523e-01 -2.45727509e-01 -6.88848078e-01 8.42760086e-01
3.77641559e-01 1.01484156e+00 1.81117043e-01 -1.26986668e-01
3.57573539e-01 1.41363099e-01 8.63346279e-01 -8.71043921e-01
-3.50248724e-01 -7.48885214e-01 -1.72286212e-01 -6.51155114e-01
-5.45897424e-01 -2.57801861e-01 -1.34062958e+00 -9.38195735e-02
4.72349226e-01 -1.74977720e-01 1.23166695e-01 1.24586368e+00
2.16231734e-01 5.72564483e-01 -2.53986657e-01 -7.22724020e-01
-6.91210926e-01 -6.96532428e-01 -4.74227965e-01 8.08302939e-01
2.16197982e-01 -7.62821972e-01 -2.06624269e-01 3.98662478e-01] | [10.552327156066895, 2.21075701713562] |
d55eaa3c-7e44-43fa-aa92-22b3055e7513 | estimating-the-direction-and-radius-of-pipe | 2201.10184 | null | https://arxiv.org/abs/2201.10184v1 | https://arxiv.org/pdf/2201.10184v1.pdf | Estimating the Direction and Radius of Pipe from GPR Image by Ellipse Inversion Model | Ground Penetrating Radar (GPR) is widely used as a non-destructive approach to estimate buried utilities. When the GPR's detecting direction is perpendicular to a pipeline, a hyperbolic characteristic would be formed on the GPR B-scan image. However, in real-world applications, the direction of pipelines on the existing pipeline map could be inaccurate, and it is hard to ensure the moving direction of GPR to be actually perpendicular to underground pipelines. In this paper, a novel model is proposed to estimate the direction and radius of pipeline and revise the existing pipeline map from GPR B-scan images. The model consists of two parts: GPR B-scan image processing and Ellipse Iterative Inversion Algorithm (EIIA). Firstly, the GPR B-scan image is processed with downward-opening point set extracted. The obtained point set is then iteratively inverted to the elliptical cross section of the buried pipeline, which is caused by the angle between the GPR's detecting direction and the pipeline's direction. By minimizing the sum of the algebraic distances from the extracted point set to the inverted ellipse, the most likely pipeline's direction and radius are determined. Experiments on real-world datasets are conducted, and the results demonstrate the effectiveness of the method. | ['Huanhuan Chen', 'Shengfei Lyu', 'Qiuju Chen', 'Xiren Zhou'] | 2022-01-25 | null | null | null | null | ['gpr', 'gpr'] | ['computer-vision', 'miscellaneous'] | [ 1.23947039e-01 1.21389374e-01 7.31486261e-01 -3.86827737e-01
-4.47416872e-01 -2.85044104e-01 -1.61911264e-01 -1.24199532e-01
-7.52938613e-02 2.20587835e-01 -8.87379572e-02 -5.49834907e-01
-4.53312933e-01 -1.12968445e+00 -2.27780417e-01 -6.87744260e-01
-1.19900495e-01 4.79615301e-01 4.99847621e-01 -3.07343960e-01
5.56566417e-01 9.25908148e-01 -5.71085811e-01 -9.27113965e-02
5.24130881e-01 9.98244703e-01 6.05956435e-01 3.76455277e-01
2.30731070e-01 1.32009432e-01 -4.95472103e-01 -1.73862219e-01
2.86292285e-01 -1.58354357e-01 -5.90931058e-01 6.14866316e-02
-6.44223690e-01 -7.57802963e-01 -3.96395147e-01 1.19043303e+00
3.77051860e-01 -2.37299889e-01 6.91636980e-01 -8.45121861e-01
-2.37188339e-02 3.33606929e-01 -1.40948117e+00 2.09757373e-01
5.96288681e-01 -4.72568542e-01 4.23267066e-01 -9.33143973e-01
1.00833423e-01 9.93117690e-01 1.24592936e+00 -4.51268375e-01
-5.87857664e-01 -6.80560648e-01 -3.61980081e-01 1.98850423e-01
-1.72440445e+00 -9.70766321e-02 1.01979804e+00 -4.40876514e-01
4.07340467e-01 6.07315078e-02 8.20151091e-01 5.02234558e-03
4.07629222e-01 4.16101247e-01 7.38123417e-01 -6.31819963e-01
-7.58867338e-02 -8.13788846e-02 1.92721546e-01 5.27601361e-01
4.54441488e-01 2.17470571e-01 1.45757899e-01 -2.87563175e-01
8.79075825e-01 -2.10297033e-01 -6.62874758e-01 -1.67306244e-01
-6.18945718e-01 9.18215752e-01 3.60610276e-01 2.89745659e-01
-7.41693199e-01 -5.01551688e-01 6.70994073e-02 -3.81538212e-01
1.85435757e-01 1.86088070e-01 -1.21120304e-01 -2.27062758e-02
-1.22304249e+00 2.02581465e-01 8.03513944e-01 9.16119039e-01
6.47223651e-01 -3.92764695e-02 4.39392895e-01 4.49857920e-01
7.30117738e-01 6.63352847e-01 -7.06276521e-02 -3.97684872e-01
4.61977810e-01 3.77609789e-01 1.55242383e-01 -1.54669750e+00
-7.35233843e-01 4.76789698e-02 -5.83428204e-01 -4.58530456e-01
4.32316095e-01 -4.42060381e-01 -3.27881336e-01 6.03849947e-01
4.09653932e-01 1.38291553e-01 -3.35254818e-02 6.92024529e-01
7.78778315e-01 8.07373405e-01 -4.04223680e-01 -3.35051805e-01
1.61651289e+00 -2.45961860e-01 -7.10493147e-01 -4.07946467e-01
5.42636335e-01 -8.43943357e-01 3.10085386e-01 4.08124655e-01
-7.60437250e-01 -1.17799714e-01 -1.05396557e+00 6.95582509e-01
4.23349738e-01 4.21116620e-01 5.56179464e-01 5.75965226e-01
-5.39605319e-01 2.63908803e-01 -9.81578529e-01 -8.56035426e-02
2.53850251e-01 -1.83714598e-01 -8.40946063e-02 -1.97296605e-01
-1.16829979e+00 8.74099016e-01 5.55504024e-01 8.61339152e-01
-3.64027590e-01 -5.34722030e-01 -7.85271883e-01 -4.96220626e-02
6.34485437e-03 -1.20575368e-01 1.30210531e+00 -5.69484867e-02
-1.09064841e+00 2.35402972e-01 4.83825654e-02 -1.40832737e-01
4.62985784e-01 -2.26772204e-01 -7.11020291e-01 5.13207793e-01
4.41292971e-01 -2.69664705e-01 6.58862174e-01 -1.40389574e+00
-8.83805394e-01 -8.49433482e-01 -4.42270249e-01 6.65122196e-02
1.20440789e-01 1.13456108e-01 -1.69944853e-01 1.02485724e-01
1.17219257e+00 -5.40062606e-01 -3.15512657e-01 -2.96761036e-01
-7.81399906e-01 5.68217412e-02 1.34751511e+00 -1.25130379e+00
1.32118964e+00 -2.14335489e+00 -8.67636323e-01 7.15511858e-01
-1.93728000e-01 3.29098962e-02 5.36546588e-01 4.76792544e-01
-9.76111591e-02 -4.38353568e-01 -6.21681869e-01 4.16143119e-01
-4.04506534e-01 -2.98488289e-01 -4.23007071e-01 1.02806985e+00
-3.85368288e-01 3.99720639e-01 -7.33362675e-01 -4.15389031e-01
-1.26385778e-01 2.06620172e-01 -4.19888832e-02 1.93003058e-01
6.62647963e-01 3.17305654e-01 -7.95641482e-01 7.95947015e-01
1.64033520e+00 5.16621649e-01 1.70788392e-02 -8.00900340e-01
-6.73305035e-01 -2.15501577e-01 -1.57912922e+00 8.09395850e-01
-3.44616234e-01 6.11871779e-01 1.51021451e-01 -1.14155495e+00
1.59842312e+00 2.44566381e-01 5.08294821e-01 -6.64700389e-01
1.30488230e-02 3.64945233e-01 -1.11174084e-01 -1.10741746e+00
8.19585800e-01 -2.83753961e-01 -1.02622565e-02 4.11420763e-01
-7.56943583e-01 -8.27887416e-01 -1.22732438e-01 -3.43068689e-02
8.20290148e-01 1.42427370e-01 3.51200908e-01 -3.09534967e-01
2.45550677e-01 2.41422996e-01 5.14254153e-01 3.88704270e-01
1.85824901e-01 4.30611998e-01 4.02049631e-01 -4.77056026e-01
-8.18947732e-01 -8.44525397e-01 -7.60700703e-01 -3.93458307e-02
7.30701208e-01 1.28943786e-01 -6.25394225e-01 -1.99561596e-01
-1.30319014e-01 9.87434089e-01 -1.79851800e-01 -3.11889276e-02
-4.88156974e-01 -9.69874263e-01 7.94275880e-01 5.83563626e-01
8.07071567e-01 -8.36051643e-01 -9.54251766e-01 4.15145934e-01
-5.39201796e-01 -1.07017732e+00 7.67737851e-02 -3.28702360e-01
-9.97568190e-01 -1.22604668e+00 -5.46700716e-01 -5.42334080e-01
8.68598819e-01 5.57237923e-01 5.08288980e-01 -1.84008539e-01
-2.39275590e-01 2.47943372e-01 -3.79630983e-01 -5.28620958e-01
-1.04190558e-01 -7.45589852e-01 -3.61233860e-01 -1.79397821e-01
3.33471030e-01 -1.12704240e-01 -7.16345191e-01 8.03604066e-01
-7.01473773e-01 -2.55114675e-01 3.71855140e-01 5.42640686e-01
4.05992627e-01 1.02091873e+00 3.68731201e-01 -3.66756111e-01
6.29265368e-01 -6.52678311e-01 -7.28669405e-01 -2.41520423e-02
-5.04360497e-02 -6.51189506e-01 8.35861340e-02 -1.29726902e-01
-1.31484616e+00 -1.54269964e-01 -4.02520031e-01 1.30580410e-01
-6.57892972e-02 8.21435153e-01 -3.56222957e-01 -3.71191800e-02
4.61978137e-01 4.16777819e-01 -2.61728108e-01 -4.43541378e-01
-2.28310078e-01 9.95840907e-01 4.67066735e-01 -1.94453210e-01
7.94295311e-01 8.52492690e-01 7.27255270e-02 -1.48860955e+00
-7.00405598e-01 -7.19183743e-01 -7.21005201e-01 -3.22865129e-01
6.31704330e-01 -6.18078828e-01 -4.42444384e-01 6.16826773e-01
-1.52515912e+00 -4.34056073e-02 8.10155421e-02 9.85259950e-01
-2.09258929e-01 9.71301496e-01 -4.41393971e-01 -1.10083246e+00
-4.92257297e-01 -1.04283023e+00 8.58164787e-01 4.19604927e-01
-7.48808160e-02 -8.56900513e-01 1.61352158e-02 1.10638671e-01
-3.63396928e-02 1.03362076e-01 6.38911307e-01 -2.19981015e-01
-1.00776600e-02 -7.24942207e-01 -5.93441069e-01 -4.16356139e-02
-2.03698548e-03 2.56235421e-01 -5.97198844e-01 -1.61460102e-01
6.42870724e-01 5.00792027e-01 2.98435271e-01 7.27768302e-01
8.66897881e-01 -1.04130134e-01 -7.94502079e-01 8.44341695e-01
1.60791802e+00 6.63042963e-01 1.10867631e+00 8.08588147e-01
2.76554406e-01 6.30066156e-01 1.18077075e+00 7.62742460e-01
4.34280276e-01 4.81597483e-01 5.58144689e-01 -9.26018599e-03
6.15064263e-01 -2.68621594e-01 -3.67530785e-03 5.26279092e-01
-3.04086417e-01 2.11413115e-01 -9.36580539e-01 5.94157696e-01
-1.30020928e+00 -9.28611696e-01 -1.00352776e+00 2.13201118e+00
-7.37329274e-02 6.39415830e-02 -2.08234206e-01 4.12700802e-01
8.64461064e-01 -3.76148552e-01 -1.16182879e-01 -3.29875857e-01
1.38210818e-01 -1.27553895e-01 7.43715405e-01 2.59445220e-01
-6.75073981e-01 1.97714418e-01 5.87654686e+00 5.38995624e-01
-1.17252767e+00 -2.64350653e-01 5.83676156e-03 1.01380539e+00
-4.23177928e-01 4.07635599e-01 -8.74075830e-01 4.61842746e-01
2.63380945e-01 1.00587234e-02 -2.98548579e-01 7.00316489e-01
3.88691485e-01 -6.71295166e-01 -4.11787540e-01 9.01677370e-01
-2.30002314e-01 -7.78566539e-01 -4.76400971e-01 1.64814979e-01
2.05777511e-01 -2.76541263e-01 -4.90821362e-01 7.68396363e-04
2.34365910e-02 -4.71809149e-01 1.03349888e+00 6.38824880e-01
8.06520045e-01 -8.13442171e-01 1.03726673e+00 6.00800455e-01
-1.61809707e+00 -2.06805095e-01 -6.49593949e-01 2.52584759e-02
9.94687080e-01 9.91122067e-01 -1.37833476e+00 8.81346107e-01
8.64134729e-01 1.94367185e-01 -1.09217726e-01 1.30050683e+00
-5.32632768e-01 4.91246581e-01 -6.66307688e-01 1.46662354e-01
2.31525272e-01 -8.51331174e-01 8.56604099e-01 1.17151213e+00
1.15421557e+00 4.60552990e-01 -1.57137379e-01 7.93109596e-01
6.17559075e-01 1.89694315e-02 -8.23466599e-01 4.47071344e-01
8.88968945e-01 1.36282229e+00 -9.06865180e-01 1.45107180e-01
-1.99601158e-01 4.81817067e-01 -4.85236794e-01 2.04136506e-01
-1.01013160e+00 -8.09100270e-01 -3.41534376e-01 6.67762935e-01
3.70815843e-01 -1.09059036e-01 -3.34494084e-01 -2.39648074e-01
7.55501315e-02 -6.85847998e-02 3.10792834e-01 -1.15537107e+00
-7.57617533e-01 6.10071063e-01 2.62436956e-01 -1.24625444e+00
1.46844640e-01 -2.37681180e-01 -1.13463902e+00 8.96713376e-01
-1.16870642e+00 -1.08632171e+00 -5.65168321e-01 2.92568296e-01
1.30740985e-01 1.75204083e-01 6.59138918e-01 7.61627480e-02
-3.48989367e-01 1.90614313e-01 2.97135741e-01 2.22924590e-01
6.51969165e-02 -5.80758452e-01 1.51330546e-01 8.80877435e-01
-4.76143807e-01 3.26941699e-01 8.77144456e-01 -9.63290811e-01
-1.19401145e+00 -7.99942613e-01 4.03985530e-01 1.72929257e-01
8.34165752e-01 1.58653468e-01 -8.64552736e-01 8.25578570e-01
-4.55502391e-01 -1.21986158e-01 2.96497434e-01 -3.49155039e-01
2.00835049e-01 -8.30301791e-02 -1.53482783e+00 1.21484451e-01
3.57889712e-01 -2.40783095e-02 -7.47590482e-01 1.16434805e-01
7.81877562e-02 -5.03549874e-01 -8.94778073e-01 6.19670749e-01
7.52140880e-01 -8.18558633e-01 8.91078770e-01 3.09009224e-01
2.26712734e-01 -2.26025268e-01 -1.98274538e-01 -1.38861382e+00
-1.27750859e-01 -2.73917556e-01 3.71959031e-01 9.79640126e-01
1.66043550e-01 -7.57486105e-01 6.58780158e-01 1.23943478e-01
-4.55197334e-01 -8.60745490e-01 -9.90646839e-01 -4.68419373e-01
-3.67101043e-01 -4.45683181e-01 7.05204725e-01 6.82265818e-01
1.53187990e-01 1.49141088e-01 3.14453870e-01 9.15521979e-01
9.36737835e-01 2.23603576e-01 5.98546505e-01 -1.23421407e+00
9.84942615e-02 5.95116131e-02 -3.22498679e-01 -1.51207840e+00
-4.44080472e-01 -3.77620697e-01 3.93739432e-01 -1.97397947e+00
3.92065085e-02 -1.04784036e+00 6.70375705e-01 5.90606555e-02
2.79922605e-01 -1.08054809e-01 -3.56317371e-01 5.09150326e-01
2.34808728e-01 3.83992940e-01 1.37109447e+00 4.26849276e-02
-4.89720851e-01 6.80347860e-01 -4.74157423e-01 1.40097535e+00
7.11379886e-01 -5.27121425e-01 -1.06371269e-01 -4.19865429e-01
3.48779559e-01 8.01554263e-01 1.45751461e-01 -7.19607353e-01
3.54460090e-01 5.48713468e-02 2.73869336e-01 -1.32419312e+00
3.61319214e-01 -9.14737523e-01 2.98535943e-01 5.87725580e-01
7.48271346e-01 -7.65742874e-03 1.77249938e-01 4.51196343e-01
-2.60054320e-02 -1.02582061e+00 7.89974332e-01 -6.23786636e-02
-5.55240810e-01 1.72992900e-01 -3.44784707e-01 -4.62579995e-01
1.06296384e+00 -8.17318976e-01 -5.55067435e-02 -2.94785947e-01
-4.44843531e-01 3.75845246e-02 2.25444824e-01 -4.86794710e-01
1.16481888e+00 -9.46791768e-01 -9.43847537e-01 2.64668018e-01
-1.10917084e-01 5.25103390e-01 6.52261376e-01 1.21288526e+00
-1.02417326e+00 6.29054829e-02 1.38065130e-01 -8.27141047e-01
-1.03937960e+00 8.77870023e-02 3.53441417e-01 -2.25533620e-01
-1.13513911e+00 7.27105319e-01 1.85176618e-02 -3.72205973e-01
-5.50940454e-01 7.54739298e-03 -6.49207830e-01 2.02080712e-01
7.05766380e-01 8.00941169e-01 2.97726840e-01 -1.02516627e+00
-2.13318214e-01 7.08610833e-01 1.91780373e-01 -3.40765901e-02
1.61714709e+00 -3.08210999e-01 -2.38276422e-01 -8.69226828e-02
9.23884988e-01 2.90189534e-01 -9.44954693e-01 -7.47050196e-02
-1.32403925e-01 -6.07029915e-01 9.40887630e-02 -1.52992681e-01
-1.10446489e+00 7.13834226e-01 3.75625968e-01 5.30523360e-01
9.15647447e-01 1.85540468e-02 9.11399901e-01 1.87090114e-01
6.04471922e-01 -9.39872980e-01 -2.09694713e-01 3.37547362e-01
9.24203396e-01 -4.70819771e-01 3.90562892e-01 -6.97996974e-01
-4.62357223e-01 1.53098738e+00 3.11643481e-01 3.68565209e-02
1.05613625e+00 5.66910148e-01 -2.26476625e-01 -6.47229493e-01
6.40067816e-01 5.28168440e-01 -4.46622491e-01 8.58920634e-01
-1.11010700e-01 1.80834532e-01 -2.94122547e-01 9.53432977e-01
-4.66396689e-01 -1.89250156e-01 1.01821566e+00 1.09312379e+00
-8.37204814e-01 -1.45008504e-01 -1.15085852e+00 5.05832016e-01
-2.65993029e-01 1.41073644e-01 6.00517690e-01 8.19065511e-01
-7.53897727e-02 9.01026845e-01 4.64282602e-01 3.55430841e-02
6.88791871e-01 -7.48792470e-01 4.93708611e-01 -2.56445020e-01
4.78877246e-01 1.17833711e-01 1.21175997e-01 2.74570882e-02
4.31847898e-03 -6.97162092e-01 -1.70832944e+00 1.79671142e-02
-1.14760745e+00 6.04531705e-01 1.08005106e+00 9.27418351e-01
-3.95368665e-01 1.45182744e-01 9.99495864e-01 -8.30396533e-01
-7.10119724e-01 -1.13526487e+00 -1.25767004e+00 -2.02774376e-01
-3.93330932e-01 -8.26348245e-01 -6.40988290e-01 -1.68483928e-01] | [6.833602428436279, 1.4079837799072266] |
5c7ff065-8564-484f-b01f-35d76bde7311 | mepnet-a-model-driven-equivariant-proximal | 2306.14274 | null | https://arxiv.org/abs/2306.14274v1 | https://arxiv.org/pdf/2306.14274v1.pdf | MEPNet: A Model-Driven Equivariant Proximal Network for Joint Sparse-View Reconstruction and Metal Artifact Reduction in CT Images | Sparse-view computed tomography (CT) has been adopted as an important technique for speeding up data acquisition and decreasing radiation dose. However, due to the lack of sufficient projection data, the reconstructed CT images often present severe artifacts, which will be further amplified when patients carry metallic implants. For this joint sparse-view reconstruction and metal artifact reduction task, most of the existing methods are generally confronted with two main limitations: 1) They are almost built based on common network modules without fully embedding the physical imaging geometry constraint of this specific task into the dual-domain learning; 2) Some important prior knowledge is not deeply explored and sufficiently utilized. Against these issues, we specifically construct a dual-domain reconstruction model and propose a model-driven equivariant proximal network, called MEPNet. The main characteristics of MEPNet are: 1) It is optimization-inspired and has a clear working mechanism; 2) The involved proximal operator is modeled via a rotation equivariant convolutional neural network, which finely represents the inherent rotational prior underlying the CT scanning that the same organ can be imaged at different angles. Extensive experiments conducted on several datasets comprehensively substantiate that compared with the conventional convolution-based proximal network, such a rotation equivariance mechanism enables our proposed method to achieve better reconstruction performance with fewer network parameters. We will release the code at \url{https://github.com/hongwang01/MEPNet}. | ['Yefeng Zheng', 'Yuexiang Li', 'Dong Wei', 'Minghao Zhou', 'Hong Wang'] | 2023-06-25 | null | null | null | null | ['metal-artifact-reduction', 'computed-tomography-ct'] | ['medical', 'methodology'] | [ 2.22529806e-02 2.63235569e-02 -1.93269268e-01 -2.02551261e-01
-4.56126690e-01 -1.03910707e-01 3.05861384e-01 -3.32579106e-01
-1.63661331e-01 2.74373412e-01 4.30731922e-01 -3.04262936e-01
-3.47146332e-01 -8.22431922e-01 -7.34746277e-01 -7.06288099e-01
3.14105153e-01 2.48854846e-01 1.20333634e-01 -2.52808601e-01
-2.00304791e-01 4.37294364e-01 -7.40742922e-01 7.27639273e-02
8.03682864e-01 9.78082597e-01 4.35015202e-01 -2.43018940e-02
1.47320077e-01 6.60718858e-01 1.47866458e-01 -8.34112763e-02
3.84896219e-01 -3.78980815e-01 -6.39238298e-01 5.95384426e-02
-3.14762324e-01 -4.62919861e-01 -7.35577106e-01 1.18288589e+00
6.55994654e-01 -2.13747546e-01 3.18351835e-01 -6.87755227e-01
-5.81747413e-01 6.75353408e-01 -5.83626330e-01 1.01902843e-01
8.33756700e-02 1.76932842e-01 4.65837896e-01 -9.98004317e-01
5.07303119e-01 7.89948881e-01 7.57282913e-01 4.09498781e-01
-9.19096828e-01 -7.00088263e-01 1.35999799e-01 2.35355110e-03
-1.39205313e+00 -1.52991086e-01 1.13346303e+00 -2.59216100e-01
2.64487445e-01 2.95432448e-01 8.29341710e-01 1.12378073e+00
3.88701558e-01 6.87012136e-01 9.84688401e-01 6.89185709e-02
-6.12524152e-02 -5.10967411e-02 -2.70189285e-01 8.29102218e-01
2.78233081e-01 5.62224090e-02 1.69242784e-01 3.46500091e-02
1.38550317e+00 6.00720346e-01 -8.59027803e-01 -6.69829190e-01
-1.44061196e+00 7.40162730e-01 8.53077412e-01 4.61354613e-01
-5.16395092e-01 6.09929562e-02 5.57972550e-01 -6.68118671e-02
1.81369498e-01 1.15208581e-01 -1.16392352e-01 2.61596829e-01
-5.31067669e-01 -6.99485838e-02 3.90373200e-01 8.90709817e-01
5.38162887e-01 2.15734378e-01 6.89107850e-02 8.24192882e-01
4.72792357e-01 1.97711244e-01 8.07202280e-01 -6.38230562e-01
5.36802173e-01 6.33014202e-01 -1.82915583e-01 -1.00715828e+00
-6.20446861e-01 -8.89536500e-01 -1.32663977e+00 -3.65796536e-02
9.48308129e-03 -1.55240178e-01 -7.61751831e-01 1.57902324e+00
3.02034050e-01 2.29561090e-01 -2.35728204e-01 1.24877357e+00
1.09503770e+00 3.09939861e-01 -6.18637912e-02 -3.49545419e-01
1.48421192e+00 -8.82541001e-01 -6.87605917e-01 -1.59278408e-01
2.97790229e-01 -7.84818292e-01 1.04900134e+00 3.50611955e-01
-1.11101699e+00 -3.87790442e-01 -1.36541903e+00 2.97265537e-02
2.39908874e-01 5.28735854e-02 7.65653312e-01 4.67159271e-01
-6.15395546e-01 4.45584625e-01 -1.02786160e+00 8.57621059e-02
5.72306275e-01 3.32853943e-01 -3.29502046e-01 -4.13194478e-01
-1.07648361e+00 7.14679778e-01 1.96737573e-01 4.55180228e-01
-9.27306831e-01 -7.87912548e-01 -7.92800665e-01 -4.68904749e-02
5.22287428e-01 -9.86157775e-01 1.14959645e+00 -6.03920758e-01
-1.60340691e+00 4.78519917e-01 1.68532386e-01 1.19601943e-01
8.17681968e-01 -9.20816585e-02 -4.74060744e-01 1.36975080e-01
1.11640625e-01 4.95461076e-02 5.83777905e-01 -1.21280634e+00
-8.30788016e-02 -4.84697074e-01 -3.71801294e-02 4.44991052e-01
-3.29805940e-01 -2.86237508e-01 -7.67907739e-01 -9.81981754e-01
8.51870000e-01 -8.78787279e-01 -5.07101893e-01 2.01943666e-01
-4.41064656e-01 2.18597338e-01 6.51164770e-01 -6.07157767e-01
1.13006473e+00 -2.03415656e+00 1.10808253e-01 1.26661256e-01
4.66809958e-01 6.93093911e-02 2.37872645e-01 1.87940180e-01
-4.91566420e-01 -6.94597438e-02 -3.25625896e-01 -1.22240847e-02
-4.63939041e-01 1.57835618e-01 -7.51390979e-02 8.20436835e-01
-2.95556933e-01 8.92866850e-01 -8.28295827e-01 -4.03867304e-01
4.88291383e-01 6.37096882e-01 -5.65566242e-01 1.41275808e-01
5.97477630e-02 9.83600140e-01 -9.10093307e-01 5.37978232e-01
1.10695744e+00 -5.06637990e-01 1.85827836e-01 -6.27612889e-01
-1.12658978e-01 1.22116022e-01 -1.26225138e+00 2.17949247e+00
-6.75803959e-01 -1.97789162e-01 2.56938457e-01 -1.11863685e+00
5.18433928e-01 5.13112485e-01 9.16898191e-01 -7.02937126e-01
4.34503615e-01 3.23036790e-01 -5.40675670e-02 -6.71294093e-01
-3.46399359e-02 -4.44437146e-01 1.76799342e-01 2.19091624e-01
-2.62399971e-01 -6.78623840e-02 -5.18672049e-01 3.78576741e-02
9.06708479e-01 1.00549534e-01 1.88276336e-01 -2.88422197e-01
6.15224957e-01 -2.35547632e-01 8.77168417e-01 3.11301261e-01
-5.83177209e-02 8.87741566e-01 2.15566620e-01 -6.09234571e-01
-8.72688830e-01 -1.16501474e+00 -5.57674289e-01 3.23827267e-01
5.20052016e-01 -1.00469254e-01 -4.48532850e-01 -5.41736722e-01
-4.62199479e-01 2.98698157e-01 -4.65956748e-01 -1.59056291e-01
-9.07493830e-01 -9.56543386e-01 2.36196339e-01 6.29155278e-01
7.86380172e-01 -7.39786327e-01 -4.41509008e-01 2.90241271e-01
-4.37700450e-01 -1.13076425e+00 -5.74619472e-01 6.78915977e-02
-1.31668997e+00 -9.70784009e-01 -1.02067840e+00 -7.27158904e-01
8.46049607e-01 5.47514498e-01 8.58875215e-01 1.56860858e-01
-1.22318782e-01 5.34568690e-02 -2.69826561e-01 -1.73538223e-01
-2.12024152e-01 -9.35059339e-02 -6.11227080e-02 -9.89860855e-03
-7.86932185e-02 -9.01314914e-01 -1.08740139e+00 3.92047435e-01
-1.09878552e+00 4.54197764e-01 9.05642271e-01 9.70494866e-01
7.46076226e-01 8.72697160e-02 3.42450947e-01 -1.02381086e+00
3.25227648e-01 -6.34729385e-01 -3.89714539e-01 3.33131775e-02
-5.13362706e-01 -5.98124452e-02 7.96228111e-01 -1.95316941e-01
-1.18143320e+00 2.20413551e-01 -5.75978935e-01 -5.64471722e-01
-3.12563963e-02 5.22259355e-01 -3.29097927e-01 -2.20765024e-01
4.45057809e-01 4.72164810e-01 1.83818728e-01 -4.39705700e-01
8.22120383e-02 2.00102359e-01 2.98666716e-01 -4.12934303e-01
7.99780309e-01 8.64428222e-01 1.42234474e-01 -4.86291021e-01
-6.56073749e-01 -2.46851042e-01 -3.62637907e-01 -1.07180439e-01
8.78611445e-01 -1.07362759e+00 -7.07460225e-01 3.72667372e-01
-9.16722894e-01 4.98433895e-02 -2.40142178e-02 9.73138809e-01
-4.27559704e-01 7.69544363e-01 -7.56399810e-01 -1.83129206e-01
-4.59361464e-01 -1.68800116e+00 7.12597430e-01 1.12316631e-01
3.31558287e-01 -8.84319484e-01 -1.98308200e-01 2.91406363e-01
5.28061926e-01 4.10410792e-01 1.01547205e+00 -1.54598936e-01
-6.88332200e-01 -5.88342734e-02 -2.20379278e-01 2.97037989e-01
1.69331908e-01 -5.06184697e-01 -7.15006769e-01 -4.18955088e-01
6.94706380e-01 -6.88249022e-02 5.08184910e-01 6.38171256e-01
1.70449400e+00 -9.85000357e-02 -2.92692780e-01 1.09794891e+00
1.64099729e+00 3.76704223e-02 6.59228981e-01 2.72982001e-01
9.08311784e-01 1.86004877e-01 2.18648806e-01 5.01585543e-01
3.25753123e-01 6.31248474e-01 8.97786498e-01 -3.28176588e-01
8.33196193e-03 -2.98950344e-01 -1.41259164e-01 1.22929597e+00
-2.48240039e-01 1.02139227e-01 -7.18826056e-01 3.20143133e-01
-1.72446764e+00 -6.03485227e-01 -3.16844791e-01 2.06608915e+00
4.65948820e-01 1.96709279e-02 -2.50765115e-01 -2.30166540e-02
6.21204436e-01 2.06455752e-01 -6.48871899e-01 1.94504842e-01
1.31069675e-01 2.10540548e-01 5.55993199e-01 1.18767962e-01
-8.94406557e-01 2.84911901e-01 5.26871204e+00 8.35160434e-01
-1.43231916e+00 2.97091782e-01 5.07819831e-01 1.53031098e-02
-5.43157160e-01 -9.04373750e-02 -2.16512337e-01 4.75648254e-01
2.05210283e-01 1.32349715e-01 2.07217544e-01 8.98729265e-01
3.52019459e-01 2.32692093e-01 -8.94645274e-01 1.11598229e+00
-1.39234841e-01 -1.40991342e+00 1.43837824e-01 1.41812757e-01
5.41098893e-01 1.59538716e-01 2.53384411e-01 1.33706138e-01
-1.68921947e-02 -1.02119827e+00 6.10161126e-01 3.52270544e-01
8.99249434e-01 -7.31823921e-01 6.51833475e-01 4.07341093e-01
-1.24504375e+00 2.71038953e-02 -5.07797003e-01 2.41073027e-01
3.69320095e-01 7.26292610e-01 -5.04687667e-01 1.19022620e+00
7.15922058e-01 8.05350840e-01 -2.00623661e-01 9.54086363e-01
-2.75597721e-01 3.26427937e-01 -1.78461954e-01 4.31613326e-01
2.65702993e-01 -3.40311497e-01 5.85409284e-01 7.05819964e-01
4.15836662e-01 2.54720688e-01 1.94320470e-01 9.63633418e-01
-8.38904977e-02 2.02576786e-01 -5.42170763e-01 6.13582432e-01
1.92470595e-01 1.37354946e+00 -7.50755072e-01 -7.22611547e-02
-7.24771678e-01 8.12570632e-01 -5.27311526e-02 2.37283364e-01
-1.01402247e+00 4.08344343e-02 2.07201779e-01 3.50636870e-01
1.33079469e-01 -7.75429681e-02 -4.74734664e-01 -1.40325701e+00
2.08584771e-01 -8.24326873e-01 3.09100777e-01 -7.49048829e-01
-1.22155321e+00 7.49238551e-01 -1.03256777e-01 -1.69584751e+00
1.79609120e-01 -4.98907179e-01 -7.37987459e-01 8.89283359e-01
-1.57451797e+00 -1.12707603e+00 -7.12063193e-01 8.10281992e-01
4.70022500e-01 2.11995766e-02 7.06938684e-01 7.15844095e-01
-5.70683718e-01 3.97438139e-01 1.03566162e-01 2.31677681e-01
5.69507599e-01 -8.64223897e-01 -2.39981264e-01 7.42927730e-01
-3.56218755e-01 8.41516733e-01 3.54806602e-01 -5.08488774e-01
-1.72675407e+00 -1.05065715e+00 2.14583091e-02 -2.08134919e-01
4.56193864e-01 -7.46455491e-02 -8.70625556e-01 7.96718419e-01
1.52738109e-01 4.12106514e-01 5.76370239e-01 -2.53587097e-01
1.20480452e-02 -1.58274084e-01 -1.00283086e+00 5.37666738e-01
9.82558072e-01 -2.38027081e-01 -3.97832572e-01 4.67550516e-01
7.49760211e-01 -8.26800048e-01 -1.08802414e+00 7.46563017e-01
4.43007529e-01 -1.08333993e+00 1.12182319e+00 -8.39504749e-02
7.03692913e-01 -3.67439985e-01 7.18139932e-02 -1.11633873e+00
-4.83974636e-01 -1.70009091e-01 1.46772042e-01 7.34777451e-01
6.55533150e-02 -8.56157124e-01 8.45640421e-01 2.95801967e-01
-6.04618669e-01 -1.19030082e+00 -1.01039064e+00 -6.03122115e-01
2.30637386e-01 -4.64981765e-01 5.88556528e-01 1.16989827e+00
-2.48527363e-01 1.40802160e-01 -3.87700170e-01 4.35081542e-01
5.85045099e-01 9.19498727e-02 4.31470126e-01 -9.28694546e-01
-4.71533746e-01 -3.54206324e-01 -1.08804636e-01 -1.28109848e+00
-4.83570129e-01 -1.08275485e+00 -1.29407346e-01 -1.53397667e+00
4.04339224e-01 -8.07969570e-01 -3.89665186e-01 1.33495271e-01
-1.67696960e-02 8.07325915e-02 -1.72012299e-01 5.44893563e-01
-2.03559712e-01 6.93986416e-01 1.88992596e+00 2.37147540e-01
7.19868615e-02 1.98290236e-02 -9.22897160e-01 1.01164651e+00
5.29146194e-01 -3.46430153e-01 -5.18053234e-01 -8.04317594e-01
2.41121069e-01 5.36712646e-01 4.81237113e-01 -9.24447775e-01
2.44554624e-01 -2.08267067e-02 4.94900405e-01 -5.47610819e-01
2.12904125e-01 -1.13879049e+00 4.79040235e-01 8.32387507e-01
1.42309695e-01 1.05614237e-01 3.31180952e-02 6.07243359e-01
-2.73686647e-01 -1.69845760e-01 9.17553961e-01 -5.07787526e-01
-4.45276588e-01 7.94523656e-01 -1.24838121e-01 -1.12802237e-01
8.59659851e-01 -1.18814632e-01 1.23903476e-01 -2.10470185e-01
-6.86132848e-01 -1.92571003e-02 4.00190204e-01 2.38139644e-01
7.49845207e-01 -1.51347959e+00 -6.31165683e-01 3.38108003e-01
7.41560338e-03 5.46297491e-01 8.50980341e-01 1.29808283e+00
-8.53412330e-01 2.75102019e-01 -2.01552927e-01 -7.57880807e-01
-6.42851532e-01 6.03966713e-01 5.53861916e-01 -4.30345207e-01
-1.11194670e+00 6.02236450e-01 7.36422002e-01 -5.45195997e-01
-1.51170388e-01 -3.22540224e-01 -1.77965626e-01 -5.74642718e-01
1.84968710e-01 1.32783651e-01 1.94356695e-01 -5.06377280e-01
-3.26796472e-01 6.80574477e-01 -8.82851481e-02 1.79713622e-01
1.62339890e+00 -4.10102978e-02 -3.48525345e-02 -8.43632668e-02
1.13134801e+00 -3.27899307e-02 -1.11804402e+00 -4.26398933e-01
-6.11605883e-01 -4.99055654e-01 3.25028807e-01 -4.00563508e-01
-1.46030164e+00 8.72337162e-01 6.70678496e-01 -1.80165067e-01
1.23802614e+00 -1.81043684e-01 1.03361678e+00 -9.38737467e-02
3.96556824e-01 -6.99579000e-01 2.20947340e-01 2.38133311e-01
1.01881278e+00 -1.22906518e+00 2.99836308e-01 -7.33460903e-01
-4.81869340e-01 1.15842104e+00 6.46445096e-01 -2.80431837e-01
9.05292451e-01 1.74524322e-01 -5.28918765e-02 -5.39905190e-01
-1.00597002e-01 3.95072281e-01 9.83591974e-02 2.05640957e-01
4.95768934e-01 -3.83472107e-02 -5.10693610e-01 8.72705221e-01
-1.86596978e-02 4.08812314e-02 4.52787519e-01 9.30336475e-01
-1.92128718e-02 -9.31000829e-01 -3.65652502e-01 2.68943071e-01
-6.09781444e-01 1.30482530e-02 4.09028590e-01 8.06785107e-01
1.64191067e-01 4.49933261e-01 -4.28157032e-01 -2.69046932e-01
4.51811522e-01 -5.04533172e-01 3.98522973e-01 -6.18755639e-01
-3.53271902e-01 3.10330600e-01 -4.67880309e-01 -5.93171537e-01
-3.63288581e-01 -4.52095002e-01 -1.33249211e+00 -2.18861118e-01
-2.42443025e-01 -6.55441508e-02 5.54794371e-01 8.76059353e-01
1.88461155e-01 9.82153237e-01 7.45872319e-01 -6.75945759e-01
-5.70142567e-01 -7.06705987e-01 -5.78804314e-01 3.52648079e-01
8.03557038e-02 -6.55442417e-01 -5.76792136e-02 -3.43003243e-01] | [13.528203964233398, -2.5448341369628906] |
94d964be-93d4-4f8b-9f31-926f67b402fc | blpnet-a-new-dnn-model-and-bengali-ocr-engine | 2202.12250 | null | https://arxiv.org/abs/2202.12250v1 | https://arxiv.org/pdf/2202.12250v1.pdf | BLPnet: A new DNN model and Bengali OCR engine for Automatic License Plate Recognition | The development of the Automatic License Plate Recognition (ALPR) system has received much attention for the English license plate. However, despite being the sixth largest population around the world, no significant progress can be tracked in the Bengali language countries or states for the ALPR system addressing their more alarming traffic management with inadequate road-safety measures. This paper reports a computationally efficient and reasonably accurate Automatic License Plate Recognition (ALPR) system for Bengali characters with a new end-to-end DNN model that we call Bengali License Plate Network(BLPnet). The cascaded architecture for detecting vehicle regions prior to vehicle license plate (VLP) in the model is proposed to eliminate false positives resulting in higher detection accuracy of VLP. Besides, a lower set of trainable parameters is considered for reducing the computational cost making the system faster and more compatible for a real-time application. With a Computational Neural Network (CNN)based new Bengali OCR engine and word-mapping process, the model is characters rotation invariant, and can readily extract, detect and output the complete license plate number of a vehicle. The model feeding with17 frames per second (fps) on real-time video footage can detect a vehicle with the Mean Squared Error (MSE) of 0.0152, and the mean license plate character recognition accuracy of 95%. While compared to the other models, an improvement of 5% and 20% were recorded for the BLPnetover the prominent YOLO-based ALPR model and the Tesseract model for the number-plate detection accuracy and time requirement, respectively. | ['Tareque Bashar Ovi', 'Md. Akiful Hoque Akif', 'Abtahi Ishmam', 'Mahmudul Hasan', 'Koushik Roy', 'Hussain Nyeem', 'Md. Saif Hassan Onim'] | 2022-02-18 | null | null | null | null | ['license-plate-recognition'] | ['computer-vision'] | [-6.73773736e-02 -5.29875934e-01 -8.80196597e-03 -5.28156497e-02
-8.53075862e-01 -6.84224725e-01 4.54485834e-01 -6.12836778e-01
-6.76216006e-01 4.25576001e-01 -5.66598356e-01 -4.86817986e-01
3.38185698e-01 -8.18487883e-01 -6.96779668e-01 -6.30629539e-01
5.24258018e-01 3.51664573e-01 7.87319362e-01 -3.20757240e-01
4.57720906e-01 1.10718238e+00 -1.35437381e+00 1.76998541e-01
6.29875243e-01 7.70069718e-01 3.27913165e-01 8.71538758e-01
1.26075387e-01 8.57945144e-01 -5.98491192e-01 -7.38095820e-01
5.93772471e-01 -9.97414906e-03 -2.74009351e-02 1.62657186e-01
3.58001411e-01 -5.90319932e-01 -9.76881623e-01 1.05853903e+00
3.27979088e-01 9.57249925e-02 8.33648205e-01 -1.11628246e+00
-5.64854920e-01 -1.95626438e-01 -6.05967522e-01 2.36973718e-01
-2.21423686e-01 1.85417384e-01 2.98615485e-01 -1.40671504e+00
3.49765867e-01 7.48341322e-01 1.06805253e+00 5.88190198e-01
-4.61834997e-01 -8.44652534e-01 -5.16503155e-01 4.83578593e-01
-1.87006760e+00 -4.75493670e-01 5.69567144e-01 -4.50443655e-01
1.02692914e+00 5.04801631e-01 4.28104550e-01 3.69024932e-01
4.00012702e-01 5.88300586e-01 8.06151807e-01 -3.58086556e-01
-8.73631686e-02 2.79650956e-01 3.21656801e-02 7.32026458e-01
5.12718499e-01 -2.48455256e-01 1.93229839e-02 6.56165838e-01
1.18751645e+00 3.43518406e-01 2.50961542e-01 4.58068430e-01
-6.10067725e-01 6.19585991e-01 -1.14182748e-01 8.16289783e-02
-2.87462056e-01 -1.44807622e-02 2.56684750e-01 -8.89584348e-02
3.21830571e-01 6.51777163e-02 -2.18601022e-02 -2.10836008e-01
-1.28288877e+00 5.03601171e-02 4.67993379e-01 1.18703163e+00
6.75517559e-01 8.19809437e-01 6.92967966e-04 1.10352325e+00
5.03836811e-01 1.02457428e+00 2.97324896e-01 -5.26594102e-01
6.95068538e-01 6.99073493e-01 5.88187836e-02 -1.19484222e+00
-5.04347309e-02 -5.46772331e-02 -6.26693845e-01 5.45178890e-01
3.38459492e-01 -1.84298754e-01 -1.27267313e+00 4.40478683e-01
-4.17224020e-01 9.90401432e-02 2.08553106e-01 8.66764724e-01
6.05561972e-01 1.39785039e+00 -2.38586888e-01 2.59955496e-01
1.54443598e+00 -1.10210943e+00 -5.96511841e-01 -4.55550671e-01
4.47369337e-01 -9.61494386e-01 6.13922060e-01 1.87321380e-01
-8.84836674e-01 -5.95269084e-01 -1.35421693e+00 -4.70776595e-02
-5.53921044e-01 7.43795395e-01 4.87026898e-03 9.51612711e-01
-1.04625678e+00 -3.36771905e-02 -5.26422858e-01 -2.53639966e-01
2.54228145e-01 5.17670870e-01 -4.33269978e-01 -4.69491929e-01
-9.06264603e-01 1.27276421e+00 2.55474567e-01 4.48820531e-01
-6.95218205e-01 -3.18004876e-01 -5.93626976e-01 -4.12547290e-02
1.51173487e-01 4.20710057e-01 8.29213500e-01 -1.04585183e+00
-1.63291478e+00 8.58239949e-01 -2.91979760e-02 -3.85242224e-01
6.47035301e-01 -9.88771170e-02 -1.02406681e+00 2.18320698e-01
-1.89487040e-01 5.73776901e-01 6.63687348e-01 -9.24694419e-01
-1.00977063e+00 -8.06734785e-02 -3.72400671e-01 2.79085040e-01
-2.19159443e-02 6.51608169e-01 -1.11544847e+00 -4.74525362e-01
-1.06207564e-01 -1.01069105e+00 8.58663395e-02 -2.03932732e-01
-1.50191933e-01 -2.14784548e-01 1.14770055e+00 -1.22890210e+00
1.03370166e+00 -2.40640283e+00 -1.02719367e+00 3.14054579e-01
-4.53864560e-02 9.93660033e-01 -9.21433717e-02 3.33885133e-01
-9.70450416e-02 1.12100497e-01 -5.33738993e-02 6.85782060e-02
-1.87880516e-01 -6.59497548e-03 -1.93547040e-01 8.35208476e-01
4.32549328e-01 8.08270693e-01 -1.02164261e-01 -3.79579633e-01
5.73621035e-01 6.17238343e-01 -1.52662113e-01 4.25688270e-03
5.30167818e-01 -5.13448596e-01 -1.08107470e-01 1.00189638e+00
1.08618271e+00 3.92687470e-01 -4.20053720e-01 3.50269079e-02
-6.41655862e-01 -3.07260543e-01 -1.09939897e+00 3.33927631e-01
-9.07928869e-02 1.49851227e+00 8.48076493e-02 -6.11720920e-01
1.46649170e+00 3.74471426e-01 -1.69791114e-02 -9.56321299e-01
1.46656886e-01 5.36574900e-01 -1.23755917e-01 -5.82040429e-01
8.89881372e-01 1.08157441e-01 2.36839782e-02 -5.10870516e-01
-1.89864054e-01 3.31038207e-01 1.73297018e-01 -2.05800295e-01
7.11553395e-01 -1.98536649e-01 -2.06285417e-01 -1.74084783e-01
8.40914667e-01 3.91169131e-01 5.07026076e-01 5.61627805e-01
-2.46402010e-01 6.70801997e-01 7.57476389e-02 -5.57550251e-01
-1.55010259e+00 -8.05301964e-01 -1.49084209e-02 7.39276528e-01
2.04069778e-01 3.45954835e-01 -7.51670599e-01 2.48463452e-02
-3.41299653e-01 6.24651492e-01 -1.12619027e-02 1.39839157e-01
-1.05452299e+00 -5.83028436e-01 1.16654682e+00 5.95876694e-01
1.14288926e+00 -8.27678382e-01 -7.00322241e-02 8.85236785e-02
1.14183791e-01 -1.41259253e+00 -5.47017276e-01 -3.27406794e-01
-6.27642214e-01 -7.29659379e-01 -1.09543478e+00 -1.38829494e+00
6.89431012e-01 4.31530625e-01 4.70039696e-01 -2.90784717e-01
-1.04724234e-02 8.23135450e-02 -8.13027844e-02 -5.42512774e-01
-6.73634470e-01 -3.62062842e-01 8.93471688e-02 1.59399062e-01
8.32701683e-01 3.14627111e-01 -3.08899641e-01 6.05846763e-01
-7.94106483e-01 1.39523472e-03 9.62797761e-01 1.76811114e-01
3.52980226e-01 1.40107483e-01 3.70383263e-01 -2.95919955e-01
4.78074372e-01 -1.02391012e-01 -1.19593394e+00 7.38916472e-02
-5.24841607e-01 -7.39925146e-01 7.24195123e-01 -2.09899694e-01
-9.49776411e-01 2.40784258e-01 -4.37738657e-01 -6.32156909e-01
-3.16861838e-01 1.17134444e-01 -7.94110596e-02 -3.77299398e-01
2.26289943e-01 8.55928600e-01 4.24795151e-02 -1.22404411e-01
-2.21773908e-01 1.11516917e+00 8.12729716e-01 1.57907307e-01
1.10284913e+00 2.01392010e-01 -8.03439617e-02 -1.44053304e+00
2.62801677e-01 -8.69392455e-01 -4.29148048e-01 -8.20117056e-01
1.11443114e+00 -1.16548336e+00 -8.07931066e-01 1.13966155e+00
-1.17374837e+00 1.90575823e-01 2.73212910e-01 6.96421325e-01
-9.26040784e-02 3.50597441e-01 -6.64663851e-01 -1.09657228e+00
-1.70339316e-01 -1.19347775e+00 5.30460179e-01 4.36476171e-01
4.25767630e-01 -6.87894106e-01 -1.95773035e-01 5.78195274e-01
5.23339748e-01 -1.00970417e-01 4.27677095e-01 -9.08343256e-01
-8.65770221e-01 -1.08185506e+00 -7.04436898e-01 8.21234047e-01
-2.65455663e-01 3.57039899e-01 -8.51964593e-01 1.28885239e-01
-2.65169621e-01 2.70337194e-01 7.34959245e-01 2.95331031e-01
5.91192305e-01 -4.22518998e-01 -5.94540946e-02 3.20394993e-01
1.80710292e+00 8.09954047e-01 1.40899217e+00 6.28778756e-01
8.22244465e-01 2.69637465e-01 4.51543689e-01 -2.43002009e-02
2.26905912e-01 5.22370756e-01 1.88731357e-01 -3.12694788e-01
-1.62148699e-01 -3.40534635e-02 8.86648715e-01 9.54989195e-01
-5.65579474e-01 -4.28257287e-01 -1.26077163e+00 4.55945969e-01
-1.22940421e+00 -1.10147715e+00 -5.16055048e-01 2.07186413e+00
2.98374623e-01 1.31424055e-01 9.88173485e-02 9.70309004e-02
1.26326263e+00 -1.07842728e-01 -2.13901356e-01 -7.19998598e-01
-1.94448709e-01 -3.26158077e-01 1.37482107e+00 3.96872491e-01
-1.13029194e+00 1.17879176e+00 5.78092527e+00 1.15464544e+00
-1.31704843e+00 -2.96451926e-01 6.55159950e-01 4.79512304e-01
4.13771868e-01 -5.67963004e-01 -1.22009778e+00 4.52673942e-01
1.33801997e+00 1.67747468e-01 3.33413750e-01 9.33116496e-01
5.00585258e-01 1.39550880e-01 -4.73782510e-01 1.27750695e+00
5.42591214e-01 -1.57040596e+00 1.17938302e-01 1.43352166e-01
3.52005422e-01 3.56377125e-01 1.04149729e-01 4.78810132e-01
-1.55985147e-01 -9.92773414e-01 8.34897161e-01 5.88701844e-01
1.05986500e+00 -1.27729809e+00 1.14098799e+00 3.23995411e-01
-1.13373363e+00 -6.52959049e-02 -6.50333047e-01 8.83803964e-02
-1.42488718e-01 -2.55685538e-01 -1.15923202e+00 -1.33633462e-03
3.77112806e-01 3.44937682e-01 -5.18080533e-01 1.20167351e+00
2.19938099e-01 9.81067419e-01 -1.89940050e-01 -3.38042736e-01
6.87977314e-01 -4.91986245e-01 6.51210308e-01 1.75321615e+00
4.91574913e-01 8.69500190e-02 -2.67784983e-01 5.53825140e-01
-2.77946234e-01 -1.48726795e-02 -4.73210424e-01 -6.85891788e-03
4.29155260e-01 1.24890232e+00 -9.98215735e-01 -2.20400289e-01
-6.34036124e-01 9.26636279e-01 -5.37755609e-01 3.80288303e-01
-1.24923325e+00 -7.82126427e-01 3.97433102e-01 5.23948371e-01
4.29392517e-01 -4.58015591e-01 -1.99008524e-01 -7.97403932e-01
2.34558675e-02 -6.49640799e-01 -2.67618477e-01 -7.95823276e-01
-8.30755532e-01 4.84209925e-01 -3.38807791e-01 -1.63947225e+00
1.20815650e-01 -1.32794869e+00 -5.74099839e-01 9.27501440e-01
-1.63987756e+00 -1.09648728e+00 5.50771207e-02 5.89168787e-01
1.03131866e+00 -7.06949234e-01 4.30781275e-01 5.86161613e-01
-9.60303664e-01 7.30636060e-01 7.27809131e-01 1.04144526e+00
4.28915381e-01 -6.92958057e-01 4.29670602e-01 1.43637323e+00
-3.76615882e-01 1.63410634e-01 3.73424172e-01 -6.30545616e-01
-1.56272221e+00 -1.56467760e+00 8.73438656e-01 -2.53757447e-01
3.65927428e-01 -4.36591715e-01 -7.33633578e-01 5.61627924e-01
8.59671459e-02 -1.53458208e-01 3.61499637e-01 -7.70549774e-01
-2.48293892e-01 -2.68249184e-01 -1.20579720e+00 7.28155613e-01
7.45991319e-02 -3.46111983e-01 -4.24829692e-01 3.74578424e-02
2.23214924e-01 -2.66275197e-01 -4.14433658e-01 -1.85888615e-02
6.50357127e-01 -4.99204904e-01 8.52214396e-01 4.16173935e-02
3.94021124e-01 -7.24391639e-01 -2.53884643e-01 -4.70741332e-01
-5.17177284e-01 -7.48566613e-02 3.95112395e-01 1.34107816e+00
6.87975347e-01 -7.95970142e-01 8.64519179e-01 8.85197639e-01
-5.65716207e-01 -3.45578492e-01 -1.02276492e+00 -9.65647519e-01
-8.58465359e-02 -4.90365148e-01 -5.97560219e-02 6.38002872e-01
-6.26566112e-01 -8.86725709e-02 -6.46267176e-01 6.35265410e-01
2.99825311e-01 -6.49545372e-01 5.21938086e-01 -8.92156482e-01
3.59322160e-01 -4.55513328e-01 -1.05572927e+00 -8.33156466e-01
-2.36322090e-01 -5.43013096e-01 2.62679011e-01 -1.57782018e+00
5.20911552e-02 7.54478574e-02 -1.26953095e-01 4.53582495e-01
2.43033022e-01 8.08704615e-01 5.29876053e-01 5.03263414e-01
-4.74818408e-01 -1.13744341e-01 7.63083220e-01 -2.92619854e-01
-5.34885153e-02 1.40288711e-01 -2.12031588e-01 9.11381245e-01
8.03189039e-01 -4.63636190e-01 5.84667288e-02 -4.22441542e-01
8.53600204e-02 -1.28302106e-03 2.91647851e-01 -1.17814398e+00
6.42879903e-01 6.62852004e-02 8.03843498e-01 -1.12262869e+00
4.27728385e-01 -9.20671999e-01 3.98405157e-02 4.00136232e-01
3.33508581e-01 3.26360703e-01 6.14774823e-01 2.97209442e-01
-3.46363753e-01 -3.19071025e-01 1.03876293e+00 4.44101542e-02
-1.46410608e+00 8.43536332e-02 -1.32817531e+00 -4.45487887e-01
1.31914091e+00 -1.22590327e+00 -2.47808322e-01 -1.63122118e-01
1.27018187e-02 -2.46432483e-01 3.55309159e-01 5.50129175e-01
1.01536429e+00 -1.10964680e+00 -1.10230041e+00 2.36560091e-01
4.06114105e-03 -3.55164945e-01 4.09465253e-01 4.09813195e-01
-1.64614046e+00 8.38156402e-01 -4.34794605e-01 -3.70798767e-01
-1.28252757e+00 1.11543059e-01 5.21199882e-01 6.19445741e-02
-5.74500620e-01 5.55623829e-01 -1.99079216e-01 -2.63472974e-01
-1.57770142e-01 1.20064504e-01 -4.60131586e-01 -1.12996317e-01
6.97053194e-01 9.80538011e-01 2.92068869e-01 -1.19532490e+00
-5.38808048e-01 7.20895588e-01 -1.92297876e-01 -3.50737199e-02
1.19464695e+00 8.38523209e-02 -6.76411623e-03 1.87111557e-01
1.07817745e+00 2.16949880e-01 -1.29869270e+00 3.78389359e-01
-2.09179908e-01 -3.46181899e-01 1.19033530e-01 -8.14730406e-01
-9.26258326e-01 6.67925060e-01 8.06434035e-01 -1.17870130e-01
8.41100216e-01 -5.17321348e-01 6.13195598e-01 5.89634359e-01
6.90405443e-02 -1.31351364e+00 -4.41102564e-01 8.51963460e-01
7.20951200e-01 -9.01546001e-01 -3.14688265e-01 -6.48947433e-02
-8.30694497e-01 1.63100910e+00 5.95661521e-01 -5.76348305e-01
2.02213481e-01 5.23105383e-01 2.20570356e-01 3.36228728e-01
-5.10436669e-02 2.28795856e-01 2.79582262e-01 5.87322235e-01
4.19840328e-02 7.67094046e-02 1.75395049e-02 4.63807851e-01
1.25879481e-01 -2.65007883e-01 8.86131346e-01 6.05088353e-01
-8.09232652e-01 -4.16226059e-01 -8.00950110e-01 4.05045956e-01
-7.24706531e-01 -2.36923218e-01 -1.62895173e-01 1.31814408e+00
5.42179868e-02 9.01959181e-01 4.08137143e-01 -4.00393993e-01
4.54810292e-01 7.17829242e-02 -1.08756572e-01 -1.24304518e-01
-1.77740932e-01 9.03962255e-02 6.48466423e-02 -1.70488004e-03
-1.39112338e-01 -4.76428449e-01 -1.20030022e+00 -7.27609754e-01
-4.01906669e-01 -2.23301709e-01 9.82898951e-01 9.02793169e-01
1.02986619e-01 3.35096568e-01 5.13014913e-01 -7.38840938e-01
-8.55088979e-02 -7.94722497e-01 -7.32547104e-01 -2.82473624e-01
1.32861897e-01 1.61026593e-03 -1.22300304e-01 3.72536749e-01] | [9.836042404174805, -4.941368579864502] |
2922def7-983c-4522-ba43-64c0f1b27889 | learning-a-fast-3d-spectral-approach-to | 2212.08058 | null | https://arxiv.org/abs/2212.08058v1 | https://arxiv.org/pdf/2212.08058v1.pdf | Learning a Fast 3D Spectral Approach to Object Segmentation and Tracking over Space and Time | We pose video object segmentation as spectral graph clustering in space and time, with one graph node for each pixel and edges forming local space-time neighborhoods. We claim that the strongest cluster in this video graph represents the salient object. We start by introducing a novel and efficient method based on 3D filtering for approximating the spectral solution, as the principal eigenvector of the graph's adjacency matrix, without explicitly building the matrix. This key property allows us to have a fast parallel implementation on GPU, orders of magnitude faster than classical approaches for computing the eigenvector. Our motivation for a spectral space-time clustering approach, unique in video semantic segmentation literature, is that such clustering is dedicated to preserving object consistency over time, which we evaluate using our novel segmentation consistency measure. Further on, we show how to efficiently learn the solution over multiple input feature channels. Finally, we extend the formulation of our approach beyond the segmentation task, into the realm of object tracking. In extensive experiments we show significant improvements over top methods, as well as over powerful ensembles that combine them, achieving state-of-the-art on multiple benchmarks, both for tracking and segmentation. | ['Marius Leordeanu', 'Elena Burceanu'] | 2022-12-15 | null | null | null | null | ['video-object-segmentation', 'graph-clustering', 'spectral-graph-clustering'] | ['computer-vision', 'graphs', 'graphs'] | [ 2.26735815e-01 -4.24098484e-02 6.67490959e-02 3.37344073e-02
-7.28659689e-01 -9.02261078e-01 2.24804938e-01 1.28155231e-01
-2.29964808e-01 -9.30727497e-02 -6.16741776e-02 -5.58075458e-02
-1.96097568e-01 -5.13953030e-01 -8.64100218e-01 -7.66862094e-01
-3.75720024e-01 4.42613840e-01 6.37019992e-01 2.52594888e-01
1.78842187e-01 4.47034597e-01 -1.39700222e+00 -5.07801364e-04
6.04058504e-01 1.02454817e+00 8.79135914e-03 8.66111159e-01
5.95876537e-02 5.64376414e-01 -1.03771165e-01 -2.60411650e-01
5.61427951e-01 -4.23800737e-01 -1.10727358e+00 8.14670861e-01
9.32394385e-01 1.10491842e-01 -4.26649332e-01 1.43559170e+00
4.04044352e-02 3.62049490e-01 5.44009209e-01 -1.38478374e+00
-4.56713736e-01 4.94364500e-01 -8.62369061e-01 1.15406856e-01
2.98832804e-01 -1.25484943e-01 1.20160770e+00 -6.22013569e-01
1.04342663e+00 1.04441202e+00 9.85081553e-01 3.91909689e-01
-1.61553824e+00 -3.66910100e-02 4.85125482e-01 1.89824432e-01
-1.46508658e+00 -6.60141604e-03 6.72625124e-01 -8.08423817e-01
7.48202205e-01 4.23543066e-01 1.00379944e+00 4.93508399e-01
-9.75738540e-02 1.08374858e+00 8.70571315e-01 -2.66199291e-01
2.37762496e-01 -3.90346825e-01 4.55830544e-01 9.72779274e-01
3.53678316e-02 -3.27626199e-01 -2.51939267e-01 -2.19155729e-01
7.08115757e-01 -2.59614345e-02 -2.24119872e-01 -1.14517105e+00
-1.25140941e+00 7.39298284e-01 4.88326490e-01 2.52817035e-01
-2.06317514e-01 5.69204032e-01 3.35947573e-01 7.06354855e-03
4.44429129e-01 -3.81332226e-02 -2.02495024e-01 1.64736822e-01
-1.26064277e+00 2.48730853e-01 8.83375823e-01 1.09272885e+00
8.62826884e-01 -2.95781851e-01 -1.31432220e-01 3.60932320e-01
2.55029529e-01 3.36759716e-01 -7.24405572e-02 -1.45188916e+00
-1.17803715e-01 4.05267835e-01 -1.48736641e-01 -1.14642715e+00
-4.72745210e-01 -3.85244578e-01 -6.47119641e-01 -4.80176844e-02
5.99417984e-01 8.23460594e-02 -9.88071263e-01 1.90352309e+00
6.22417212e-01 9.24625993e-01 -2.96340674e-01 9.32874501e-01
3.70409429e-01 4.31028068e-01 4.74061333e-02 -4.70068127e-01
1.15088081e+00 -9.76164460e-01 -3.58193815e-01 8.14022720e-02
5.80615759e-01 -6.81182206e-01 6.81683123e-01 3.92430723e-01
-1.04071498e+00 -3.66410375e-01 -7.92746544e-01 4.50001359e-02
-6.77564517e-02 -1.85690716e-01 8.66399646e-01 5.58917701e-01
-1.38939905e+00 1.02304137e+00 -1.23309708e+00 -6.19004250e-01
5.12717724e-01 3.00426900e-01 -2.48386502e-01 1.29921660e-01
-6.14785671e-01 3.18921626e-01 2.76783049e-01 -2.72733402e-02
-4.68887299e-01 -7.11557448e-01 -7.32357681e-01 -2.42095888e-01
4.49453264e-01 -9.26700771e-01 1.01645339e+00 -1.16024792e+00
-9.45691347e-01 1.12128329e+00 -2.73292661e-01 -6.35165334e-01
6.41724288e-01 -7.29801971e-03 3.94140370e-02 4.71862316e-01
3.51494372e-01 6.21834636e-01 9.29837406e-01 -1.34555805e+00
-6.99181318e-01 -6.28509581e-01 -1.17437497e-01 1.38837039e-01
-1.02186762e-01 -1.03576258e-01 -1.10725689e+00 -6.11910343e-01
6.55811489e-01 -1.29526663e+00 -5.14163613e-01 -8.80872924e-03
-4.74802017e-01 -2.52601713e-01 1.00404918e+00 -7.90668726e-01
1.25653219e+00 -2.19132113e+00 5.95508277e-01 6.42879426e-01
6.21566594e-01 -1.10458538e-01 1.95513126e-02 6.46562651e-02
-1.24203347e-01 -5.88365495e-02 -5.31066537e-01 -2.80554563e-01
-8.51202663e-03 7.51509368e-02 -4.20027860e-02 9.61331546e-01
-5.43306693e-02 1.08686984e+00 -1.04709065e+00 -5.55728078e-01
2.38998815e-01 3.57024014e-01 -6.70833945e-01 -2.59763867e-01
-2.78584391e-01 3.79123062e-01 -5.13054430e-01 4.39435393e-01
8.29286933e-01 -6.42193019e-01 2.75260717e-01 -4.29207951e-01
-9.54915509e-02 -3.58718544e-01 -1.59630358e+00 1.95104289e+00
2.63260484e-01 4.62306261e-01 4.37627912e-01 -1.33612168e+00
2.48556793e-01 -3.49572040e-02 1.32813549e+00 -1.56344235e-01
1.37677550e-01 2.18779780e-02 -3.99249136e-01 -2.17857495e-01
3.82405251e-01 2.66961485e-01 5.51279560e-02 4.20544147e-01
1.16518393e-01 4.53552790e-02 4.76272702e-01 7.45841205e-01
1.17765594e+00 2.60469645e-01 -7.08663389e-02 -6.81994200e-01
3.19843531e-01 1.66254550e-01 4.80835646e-01 7.17584610e-01
-3.20808381e-01 6.49327695e-01 5.72611451e-01 -1.47174373e-01
-8.44374061e-01 -1.27725339e+00 -9.42148417e-02 9.64857459e-01
5.13018370e-01 -5.92092872e-01 -1.42499316e+00 -6.29016042e-01
2.10268989e-01 5.34164719e-02 -6.83421254e-01 1.28815427e-01
-6.28856301e-01 -6.36882126e-01 1.67622358e-01 5.57747722e-01
2.95150489e-01 -4.95282769e-01 -5.10136724e-01 1.72277957e-01
-2.23710969e-01 -1.30946946e+00 -8.94529641e-01 -6.85495660e-02
-1.07971716e+00 -1.36522937e+00 -5.93612731e-01 -9.24542546e-01
5.64216197e-01 6.38121247e-01 1.06397200e+00 2.92355359e-01
-5.48906386e-01 1.09975767e+00 -1.26037449e-01 1.03413165e-01
-9.62345526e-02 -9.96619239e-02 -8.89580548e-02 3.61413538e-01
2.41446137e-01 -4.71624196e-01 -5.79355836e-01 5.33644259e-02
-7.93477535e-01 1.59416869e-02 3.31306942e-02 3.74972403e-01
1.01586187e+00 1.88896600e-02 -1.32601023e-01 -1.02068794e+00
8.57331827e-02 -2.72009045e-01 -7.79934466e-01 3.72025013e-01
-2.22059399e-01 -7.92643130e-02 2.26140857e-01 -1.56096846e-01
-4.08301860e-01 7.42329359e-01 1.67027935e-01 -7.78721452e-01
8.21470097e-02 2.24056721e-01 8.65589827e-02 -4.89688098e-01
4.21780616e-01 7.41237327e-02 1.31294638e-01 -2.94090152e-01
7.66067624e-01 2.70101931e-02 7.48076022e-01 -5.77057779e-01
8.73428941e-01 9.57689643e-01 3.72034609e-01 -1.01581717e+00
-8.92414093e-01 -1.14340174e+00 -9.10964251e-01 -4.90422338e-01
1.25722933e+00 -8.38301897e-01 -8.14562917e-01 3.76460582e-01
-9.30315197e-01 -3.43479127e-01 -3.41136545e-01 3.49711329e-01
-9.24978077e-01 7.84312665e-01 -7.26424217e-01 -6.04197264e-01
1.32673845e-01 -1.15062439e+00 1.24878871e+00 1.21495195e-01
-1.06557533e-01 -1.14946461e+00 3.73011082e-02 3.04295033e-01
-9.41862762e-02 5.03006101e-01 5.31261206e-01 -2.82600790e-01
-1.05867589e+00 -8.17000214e-03 -2.98618376e-01 7.88480267e-02
-2.26462081e-01 3.09176058e-01 -7.41137922e-01 -4.43893045e-01
1.86264925e-02 1.90849498e-01 1.22872198e+00 9.23409522e-01
1.18217373e+00 6.02577850e-02 -6.34643435e-01 8.98901522e-01
1.63880873e+00 -3.87521446e-01 3.19202840e-01 -4.10634233e-03
1.26902592e+00 5.09433150e-01 4.92810518e-01 1.78006366e-01
3.43503356e-01 7.36450255e-01 1.80532932e-01 -1.09957106e-01
-3.01757485e-01 -4.04997990e-02 2.17983261e-01 7.99632192e-01
-2.69870013e-01 1.10055752e-01 -8.26366425e-01 5.86318374e-01
-2.37599850e+00 -1.01852632e+00 -5.56737125e-01 2.30377674e+00
3.82247478e-01 -7.40176961e-02 6.37966096e-01 2.42812894e-02
9.42602992e-01 8.08237791e-02 -4.76206154e-01 1.10150784e-01
-1.79757535e-01 1.90042451e-01 9.42249417e-01 7.95137405e-01
-1.74117017e+00 1.09004307e+00 6.47459555e+00 7.53959596e-01
-8.12801898e-01 2.00298250e-01 5.22950232e-01 8.79909918e-02
-1.68455288e-01 6.79049417e-02 -4.60088938e-01 1.72653854e-01
5.75404882e-01 -1.63505152e-01 6.67756855e-01 8.52314651e-01
-6.67040274e-02 -1.55096175e-02 -1.20347440e+00 1.05570459e+00
4.36808728e-02 -1.44730186e+00 -1.84611559e-01 1.86727885e-02
1.06433260e+00 1.43961817e-01 -4.06753458e-02 -3.41935337e-01
2.69377917e-01 -6.71181381e-01 9.20103133e-01 3.48031253e-01
3.15309852e-01 -6.04695439e-01 4.81087491e-02 3.99712324e-02
-1.75968838e+00 1.64613247e-01 -1.43035382e-01 3.07785690e-01
1.56999052e-01 4.66530740e-01 -2.66557038e-01 5.51569641e-01
7.70884573e-01 1.15095365e+00 -6.48085773e-01 1.15829134e+00
2.04440698e-01 4.51478660e-01 -5.20065606e-01 2.16611251e-01
3.07369381e-01 -6.24582529e-01 8.29689682e-01 1.43977559e+00
1.61012441e-01 2.59614348e-01 6.95057094e-01 7.66886532e-01
-3.90189961e-02 2.73024421e-02 -4.84227657e-01 5.75349182e-02
2.20330700e-01 1.35138822e+00 -1.51334381e+00 -4.11374241e-01
-5.31745791e-01 1.15948951e+00 2.28491858e-01 5.91237724e-01
-7.95780361e-01 6.52602613e-02 6.38026595e-01 3.53583023e-02
6.05139196e-01 -5.47740579e-01 -2.56305039e-01 -1.12803209e+00
1.08881928e-01 -6.46149993e-01 5.13778806e-01 -4.21743929e-01
-1.26204312e+00 1.06029242e-01 -1.52857602e-01 -1.09074700e+00
8.72402936e-02 -6.13354683e-01 -2.74977058e-01 3.96486700e-01
-9.51781034e-01 -1.13958776e+00 -1.98879391e-01 8.68241727e-01
3.41051787e-01 4.31677401e-01 3.92027378e-01 2.53943294e-01
-4.65493739e-01 1.99071214e-01 7.57499188e-02 2.55968064e-01
2.61695892e-01 -1.53564203e+00 5.01698196e-01 1.25896287e+00
7.03907728e-01 5.66934526e-01 6.93551362e-01 -6.99241817e-01
-1.80145347e+00 -1.21118975e+00 4.16525692e-01 -6.94468141e-01
1.02871585e+00 -4.12349045e-01 -8.91334653e-01 9.34836566e-01
-1.71191022e-02 2.70742744e-01 3.70401502e-01 -1.67100057e-02
-3.31269681e-01 2.20325455e-01 -8.73245418e-01 5.75556099e-01
1.48355794e+00 -4.45709854e-01 -2.55828857e-01 6.57965839e-01
7.74415374e-01 -6.42156720e-01 -8.20384324e-01 2.01440960e-01
3.13806146e-01 -9.71743226e-01 1.16201520e+00 -6.40291452e-01
-7.45042190e-02 -6.68088436e-01 -2.41888955e-01 -8.58540237e-01
-5.57506621e-01 -1.01832724e+00 -1.91849887e-01 8.70109439e-01
1.39751881e-01 -2.51987994e-01 1.15392637e+00 5.82805574e-01
-2.13772565e-01 -5.82570493e-01 -8.74505281e-01 -8.32723320e-01
-1.31383821e-01 -7.03955531e-01 6.63995519e-02 1.00348783e+00
-2.36954898e-01 1.25716552e-01 2.69092619e-02 4.44691122e-01
1.26468921e+00 4.09646571e-01 5.41092336e-01 -1.31512606e+00
-3.50458324e-01 -6.91099346e-01 -8.98510396e-01 -1.27622104e+00
1.87179327e-01 -1.18057501e+00 6.37905151e-02 -1.49784482e+00
4.66713905e-01 -2.14748815e-01 -1.58998892e-01 -2.68094428e-02
-1.79957822e-01 6.42306685e-01 3.96148145e-01 1.45945802e-01
-1.21989655e+00 -1.58063229e-02 1.26248538e+00 -1.29293114e-01
-1.64584100e-01 -1.06649607e-01 -4.50135291e-01 9.08597350e-01
2.97501385e-01 -3.13935995e-01 -2.61200041e-01 -3.92546028e-01
-1.39407879e-02 -1.05021641e-01 6.97259605e-01 -9.63831186e-01
4.59658831e-01 -6.75996114e-03 1.69283405e-01 -4.74550664e-01
1.21711433e-01 -8.08806479e-01 3.09644133e-01 4.04453725e-01
-1.46186814e-01 -1.95296988e-01 8.94418731e-02 9.73155141e-01
-1.33119956e-01 3.38435918e-02 8.89271319e-01 -7.68287852e-02
-1.12475657e+00 7.37424850e-01 -1.37367710e-01 3.05075765e-01
1.30279219e+00 -5.23193479e-01 1.70251653e-01 -1.65627226e-01
-1.31001019e+00 3.65561098e-01 8.31272066e-01 1.72053501e-01
3.17021936e-01 -1.38634884e+00 -4.26793635e-01 -2.54172347e-02
-8.38408470e-02 -1.85891956e-01 4.46163386e-01 1.23787367e+00
-7.10638463e-01 1.39155477e-01 1.68004125e-01 -1.26388371e+00
-1.48938334e+00 7.06174374e-01 4.03122783e-01 1.38195127e-01
-8.83648455e-01 1.03368127e+00 2.44821593e-01 1.60636269e-02
3.61922085e-01 -2.56699502e-01 1.75281063e-01 1.13538891e-01
2.18523115e-01 4.87541169e-01 5.70557639e-03 -9.11801338e-01
-5.99476218e-01 1.18211639e+00 3.65595192e-01 -3.82170267e-02
1.07086670e+00 -3.16476613e-01 -4.00557518e-01 3.27540755e-01
1.39978564e+00 -2.04199832e-02 -1.37422252e+00 -2.20836520e-01
1.86579049e-01 -4.99257058e-01 1.30226268e-02 -1.63587164e-02
-1.28176558e+00 5.66241920e-01 5.90260386e-01 7.21098483e-01
1.09285355e+00 3.74568820e-01 8.03729832e-01 2.69019514e-01
3.55977744e-01 -1.28586733e+00 -3.77297141e-02 4.21294630e-01
3.97758693e-01 -1.01090539e+00 4.82345894e-02 -9.64277267e-01
-3.91278833e-01 1.01774800e+00 1.22507602e-01 -6.73673809e-01
9.03830171e-01 1.87859580e-01 -1.25358328e-01 -4.23682868e-01
-2.75397331e-01 -4.98730510e-01 4.96981084e-01 5.90491295e-01
4.26777065e-01 2.32665479e-01 -9.47264731e-02 1.54747665e-01
4.65969415e-03 -3.51648718e-01 2.35815674e-01 6.86844349e-01
-5.26400685e-01 -9.91429210e-01 -2.20677689e-01 4.55298603e-01
-4.13946956e-01 1.60412148e-01 -4.48002249e-01 8.05131376e-01
5.36928012e-04 7.47629762e-01 2.87329465e-01 -1.86236426e-01
2.31874898e-01 -5.58283776e-02 8.59697521e-01 -5.35266519e-01
-2.88846254e-01 4.56917107e-01 -2.71598160e-01 -1.10853148e+00
-9.08913374e-01 -1.21844685e+00 -1.48399758e+00 -2.83224821e-01
-2.21712589e-01 5.89846745e-02 4.35129285e-01 8.40295374e-01
1.44893765e-01 3.64344448e-01 5.06911814e-01 -9.70942378e-01
-1.92662895e-01 -2.30203316e-01 -7.46321976e-01 7.89687097e-01
2.16006219e-01 -5.34826279e-01 -2.18834087e-01 4.07764614e-01] | [9.09134578704834, -0.25005388259887695] |
8f152fbd-d083-47ee-9b7c-6535dde7d3a8 | scale-invariant-fully-convolutional-network | 1906.04634 | null | https://arxiv.org/abs/1906.04634v1 | https://arxiv.org/pdf/1906.04634v1.pdf | Scale Invariant Fully Convolutional Network: Detecting Hands Efficiently | Existing hand detection methods usually follow the pipeline of multiple stages with high computation cost, i.e., feature extraction, region proposal, bounding box regression, and additional layers for rotated region detection. In this paper, we propose a new Scale Invariant Fully Convolutional Network (SIFCN) trained in an end-to-end fashion to detect hands efficiently. Specifically, we merge the feature maps from high to low layers in an iterative way, which handles different scales of hands better with less time overhead comparing to concatenating them simply. Moreover, we develop the Complementary Weighted Fusion (CWF) block to make full use of the distinctive features among multiple layers to achieve scale invariance. To deal with rotated hand detection, we present the rotation map to get rid of complex rotation and derotation layers. Besides, we design the multi-scale loss scheme to accelerate the training process significantly by adding supervision to the intermediate layers of the network. Compared with the state-of-the-art methods, our algorithm shows comparable accuracy and runs a 4.23 times faster speed on the VIVA dataset and achieves better average precision on Oxford hand detection dataset at a speed of 62.5 fps. | ['Tiejian Luo', 'Siwei Lyu', 'Dan Liu', 'Yanjun Wu', 'Libo Zhang', 'Dawei Du', 'Feiyue Huang'] | 2019-06-11 | null | null | null | null | ['hand-detection'] | ['computer-vision'] | [-1.36415064e-01 -5.35591364e-01 9.99490395e-02 -2.68793166e-01
-4.52780873e-01 -4.60873306e-01 1.44499123e-01 -2.69390315e-01
-1.01493716e+00 3.70771885e-01 -1.34316767e-02 -6.22199886e-02
3.40977371e-01 -5.68721890e-01 -5.07546186e-01 -6.44704580e-01
1.19166441e-01 9.99537259e-02 7.76436269e-01 -6.49863929e-02
8.11016783e-02 9.00712073e-01 -1.18244100e+00 1.35191098e-01
4.00584668e-01 1.00609004e+00 2.03778163e-01 7.51754701e-01
3.06284368e-01 5.90786636e-01 -5.35782337e-01 -3.69759321e-01
3.86023492e-01 4.18153852e-02 -7.44957209e-01 -3.23759727e-02
4.49906379e-01 -1.03952515e+00 -7.76222646e-01 7.64227688e-01
1.19137216e+00 2.15425305e-02 3.82200509e-01 -8.80204737e-01
-1.62695482e-01 2.72524387e-01 -1.09236181e+00 2.84242868e-01
1.58342779e-01 2.18450040e-01 6.82467699e-01 -1.12554717e+00
5.59376478e-01 1.32500994e+00 5.50988317e-01 4.36501801e-01
-8.87748241e-01 -7.49671519e-01 3.41646343e-01 2.40344480e-02
-1.68998456e+00 -1.23612463e-01 6.02488995e-01 -2.47417077e-01
9.29720998e-01 1.55558562e-04 6.71875298e-01 1.06753361e+00
-8.72669592e-02 1.14337826e+00 7.20384955e-01 -2.66684622e-01
-1.84266225e-01 -1.35891363e-01 -5.25081903e-02 1.00232375e+00
2.21092477e-01 -1.23600163e-01 -2.33611032e-01 1.57064125e-01
1.20551109e+00 3.57442617e-01 -8.30381885e-02 -3.24090183e-01
-1.20794153e+00 5.76656461e-01 8.63369346e-01 1.66400313e-01
-4.15518463e-01 2.92530656e-01 5.66751838e-01 -8.73570517e-02
1.54077813e-01 -5.95335551e-02 -6.00383222e-01 2.50802249e-01
-9.63999152e-01 3.42784107e-01 3.27192098e-01 9.56186295e-01
2.41483346e-01 -3.04527730e-01 -6.67499483e-01 6.99175358e-01
3.59533578e-01 4.15277660e-01 1.00668393e-01 -5.84134281e-01
6.83930099e-01 6.24850750e-01 1.47208810e-01 -6.32415771e-01
-7.53203690e-01 -5.62676072e-01 -8.52674186e-01 2.76946962e-01
7.50725389e-01 -2.11657599e-01 -1.13392663e+00 1.44594479e+00
5.14372110e-01 -2.78125614e-01 -4.09343988e-01 1.38151097e+00
5.72941363e-01 2.28903204e-01 9.55484509e-02 3.16816628e-01
1.73435020e+00 -1.29696596e+00 -4.22514707e-01 -1.41015157e-01
4.46527988e-01 -9.54601288e-01 1.12953556e+00 3.94306809e-01
-9.17965829e-01 -6.95447564e-01 -1.01106870e+00 -4.48867351e-01
-4.78055030e-01 7.98236847e-01 7.81776607e-01 5.23632288e-01
-8.20620775e-01 4.94039387e-01 -9.84269559e-01 -2.97766268e-01
6.57107949e-01 4.30409998e-01 -2.59001732e-01 9.29199085e-02
-8.99064720e-01 7.37511218e-01 3.56526166e-01 4.45735753e-01
-4.76887435e-01 -3.33813667e-01 -8.22512865e-01 2.10619881e-03
5.57783067e-01 -6.32788777e-01 1.05668998e+00 -5.47433972e-01
-1.47257805e+00 5.76040149e-01 -1.25134736e-01 -9.02612973e-03
1.01975381e+00 -5.88037252e-01 1.18642800e-01 2.06001043e-01
-6.42791837e-02 7.71929145e-01 9.54902589e-01 -6.90068483e-01
-8.11510324e-01 -5.47005653e-01 1.46013185e-01 -1.05571626e-02
-2.45394394e-01 4.68394369e-01 -1.07663786e+00 -9.85860169e-01
8.80414918e-02 -9.00378406e-01 -1.55440435e-01 4.58000422e-01
-4.48786616e-01 -5.84270120e-01 7.94008315e-01 -9.79770064e-01
1.28913867e+00 -2.09320092e+00 2.82031655e-01 2.24416360e-01
3.60174209e-01 4.24161434e-01 -9.89757553e-02 -1.44774944e-01
2.67698318e-02 -3.99712414e-01 -8.88943374e-02 -6.13580823e-01
-7.33825937e-02 -2.99249053e-01 2.84310728e-02 6.92074537e-01
4.17976230e-01 9.13051844e-01 -7.90748298e-01 -7.41124868e-01
4.26413566e-01 8.66179109e-01 -5.11804879e-01 3.62218730e-02
2.15364903e-01 2.68412113e-01 -4.28012967e-01 9.13815022e-01
7.59419262e-01 -2.79429942e-01 -1.29418671e-02 -2.80242562e-01
-3.13717812e-01 1.12025470e-01 -1.43955100e+00 1.91809225e+00
-4.38147515e-01 4.56203669e-01 9.73864421e-02 -3.46508652e-01
5.96288621e-01 2.11844549e-01 5.19594960e-02 -3.91817600e-01
5.83877861e-01 1.43983290e-01 5.32557257e-02 -1.95187420e-01
3.35367829e-01 3.98232073e-01 2.21675858e-01 4.05996472e-01
3.36100943e-02 3.70084345e-01 6.84594959e-02 1.41842246e-01
7.90151358e-01 3.72708201e-01 1.68113574e-01 1.47521541e-01
6.86981082e-01 -5.04155397e-01 3.39360416e-01 5.24800658e-01
-3.89500409e-01 7.21589565e-01 4.11311179e-01 -5.85896671e-01
-1.07242739e+00 -8.61588657e-01 1.05693005e-02 1.31483698e+00
1.08927712e-02 -3.70616347e-01 -8.76082540e-01 -9.40417171e-01
7.03391582e-02 -1.23759031e-01 -6.54304802e-01 1.95524096e-01
-1.06583416e+00 -5.37590027e-01 5.99648476e-01 1.32660019e+00
7.85768986e-01 -1.21320558e+00 -9.44093108e-01 2.99152553e-01
1.05289906e-01 -1.06637335e+00 -9.40515697e-01 1.02630414e-01
-4.68311459e-01 -9.95440662e-01 -1.28905714e+00 -9.65504825e-01
6.00074112e-01 4.19891685e-01 4.74215180e-01 7.33709545e-04
-9.37096596e-01 -2.24103719e-01 -2.64561951e-01 -2.49573976e-01
4.37977612e-01 5.09535909e-01 -2.70299800e-02 -2.32581541e-01
1.31592840e-01 -1.94074020e-01 -1.12507522e+00 4.94498223e-01
-5.09865463e-01 -1.01685829e-01 9.25768614e-01 8.06956887e-01
4.58133429e-01 -2.06759468e-01 2.67528266e-01 -1.77398622e-01
2.62409538e-01 3.16674083e-01 -6.59605443e-01 5.12920260e-01
-2.28479683e-01 1.59076899e-01 5.66111743e-01 -6.06623769e-01
-1.06138444e+00 6.82241380e-01 -2.08852068e-01 -4.81018782e-01
7.41434619e-02 -3.14184487e-01 -8.48160088e-02 -2.41215080e-01
4.44961816e-01 2.32327938e-01 -8.22580531e-02 -6.17929220e-01
5.91023982e-01 8.79173875e-01 5.01461744e-01 -2.26780176e-01
8.24983776e-01 6.38597846e-01 -1.26434490e-01 -5.40446460e-01
-7.84906685e-01 -6.54744208e-01 -1.00751352e+00 -2.12456003e-01
9.28943396e-01 -8.76643181e-01 -9.41992879e-01 8.86151016e-01
-1.42460775e+00 -2.50301272e-01 1.06374256e-01 5.78360617e-01
-1.08271405e-01 4.67726767e-01 -8.15778375e-01 -8.89581978e-01
-8.93396556e-01 -1.19815099e+00 1.48840773e+00 3.35004836e-01
-3.18037234e-02 -3.34741533e-01 -3.29002470e-01 4.07603271e-02
4.58914906e-01 9.00510922e-02 2.83443749e-01 -1.84247896e-01
-4.18787509e-01 -5.22997797e-01 -9.46977258e-01 2.83603042e-01
1.67477891e-01 -3.31138410e-02 -1.05916417e+00 -7.57911921e-01
-5.19515097e-01 -2.84876764e-01 1.20825040e+00 3.42229337e-01
1.31523979e+00 2.93589495e-02 -3.76259834e-01 6.94197774e-01
1.04799211e+00 -7.15317205e-02 3.67418379e-01 3.77057105e-01
8.66939902e-01 6.17284060e-01 5.37089109e-01 5.98493934e-01
2.57449716e-01 7.89854527e-01 1.31869435e-01 -4.41406012e-01
-3.54962111e-01 2.11663693e-02 1.21553794e-01 8.39008689e-02
-5.30374527e-01 1.83949351e-01 -5.22196352e-01 2.40180418e-01
-1.77071750e+00 -6.39384568e-01 2.49673828e-01 2.05342698e+00
7.27611840e-01 1.82680637e-01 7.56240726e-01 2.14621440e-01
8.99361908e-01 2.24550605e-01 -6.47302687e-01 1.13416493e-01
1.99562788e-01 4.60738450e-01 7.22490788e-01 2.89004505e-01
-1.62715745e+00 1.26453543e+00 5.36591530e+00 7.47396111e-01
-1.22567546e+00 4.67571579e-02 2.52901167e-01 -3.23373914e-01
6.72877312e-01 -3.07832807e-01 -9.28688526e-01 2.16309652e-01
4.93596159e-02 6.72488689e-01 6.16572022e-01 1.08467221e+00
3.86470482e-02 1.67546093e-01 -7.80006349e-01 1.09534371e+00
-9.67281684e-03 -7.53816903e-01 -1.70487896e-01 -2.46224366e-02
1.64572448e-01 -8.09639096e-02 -8.51303339e-02 1.35857269e-01
-1.92782834e-01 -7.60447741e-01 9.47761655e-01 2.92292625e-01
8.68116856e-01 -1.01479447e+00 8.43214691e-01 2.17474978e-02
-1.75217271e+00 -1.26915663e-01 -4.17930752e-01 6.80992901e-02
4.90198433e-02 1.73460498e-01 -5.97704291e-01 3.26602846e-01
8.78573418e-01 2.86141217e-01 -6.78442478e-01 9.75149333e-01
-4.99102294e-01 -7.77092800e-02 -4.44705606e-01 -5.79992384e-02
1.86567128e-01 3.06597143e-01 2.19287455e-01 1.43439877e+00
-8.25680140e-03 -4.29698676e-02 1.14902832e-01 6.16451383e-01
2.70069037e-02 2.64211539e-02 2.06035785e-02 3.21804374e-01
3.60929847e-01 1.49102998e+00 -1.06286824e+00 -3.70857269e-01
-4.74581301e-01 1.57097232e+00 5.48983395e-01 8.17124099e-02
-1.01541650e+00 -1.18135238e+00 3.68415594e-01 1.56665444e-01
6.69592261e-01 -3.75646651e-01 -1.13194026e-01 -1.11992979e+00
4.59140569e-01 -4.83761460e-01 3.17986399e-01 -3.21504623e-01
-1.00227845e+00 6.21626794e-01 -3.77403021e-01 -9.47470546e-01
1.53112844e-01 -8.56199026e-01 -5.43340087e-01 1.00822580e+00
-1.66880012e+00 -1.34784067e+00 -6.83575571e-01 8.39296222e-01
5.17900705e-01 1.26912102e-01 4.26943958e-01 4.49343801e-01
-8.96716297e-01 1.12004018e+00 -3.68182272e-01 8.35799873e-01
1.03780007e+00 -1.29149973e+00 8.92583251e-01 9.61173415e-01
-3.16007137e-01 9.32445645e-01 7.06382245e-02 -5.19130468e-01
-1.14134967e+00 -1.14794397e+00 5.97813606e-01 -2.46557459e-01
3.59769702e-01 -5.36284268e-01 -6.51085377e-01 4.28782195e-01
-2.56870806e-01 3.52127761e-01 1.76271319e-01 6.36643022e-02
-7.89316595e-01 -1.34072229e-01 -1.16474032e+00 4.42840576e-01
1.25179148e+00 -4.37692195e-01 -3.18987280e-01 3.23288590e-01
5.25165021e-01 -5.21996439e-01 -6.19093537e-01 3.01998585e-01
1.17125618e+00 -6.43932819e-01 1.10381079e+00 -4.49946731e-01
-2.36535985e-02 -5.17664671e-01 1.28022090e-01 -5.81326067e-01
-4.87756133e-01 -5.39608717e-01 -2.72527963e-01 9.47589874e-01
9.02628973e-02 -5.49180448e-01 8.23703229e-01 2.22163007e-01
1.79127991e-01 -9.27671671e-01 -9.34066594e-01 -7.21383095e-01
-2.60126084e-01 -1.32813692e-01 4.90865976e-01 3.45254958e-01
-1.14021525e-01 1.98580101e-01 -3.28898102e-01 2.12304041e-01
8.09678912e-01 1.36503920e-01 7.62680054e-01 -9.50871646e-01
-1.73183709e-01 -5.30720770e-01 -3.72240424e-01 -1.18273151e+00
-2.81007081e-01 -4.76142645e-01 3.17813121e-02 -1.40226626e+00
3.73666704e-01 -2.25173205e-01 -3.44509959e-01 7.18862951e-01
-5.27197242e-01 4.69487756e-01 3.60746354e-01 1.31383061e-01
-5.96392274e-01 1.53608903e-01 1.41976953e+00 3.00850458e-02
-4.71684068e-01 5.72534762e-02 -3.33628207e-01 8.23178709e-01
7.99412191e-01 -1.75521806e-01 8.28447714e-02 -3.69452506e-01
-3.08595031e-01 -1.88373506e-01 6.09416902e-01 -1.04650819e+00
3.14433873e-01 2.99596727e-01 9.40703988e-01 -9.11005080e-01
1.97589830e-01 -5.86154580e-01 -6.97200775e-01 8.40371549e-01
-9.58381221e-02 -9.67017561e-02 3.37505862e-02 2.61143237e-01
2.39125997e-01 2.76502460e-01 1.16808617e+00 1.78199708e-02
-4.94971812e-01 5.19712508e-01 2.32839566e-02 -4.84036744e-01
1.02224064e+00 -1.01552848e-02 -1.60449833e-01 1.68997750e-01
-5.85268557e-01 3.19164157e-01 2.51496941e-01 5.09233713e-01
6.33757234e-01 -1.03064835e+00 -6.49253309e-01 2.69151092e-01
-1.07704982e-01 3.69363636e-01 3.59108895e-01 9.17787313e-01
-8.01087260e-01 5.26554167e-01 -8.85254592e-02 -4.18828785e-01
-1.38482201e+00 6.47477567e-01 2.86678821e-01 -2.32744664e-01
-8.75294447e-01 1.10141158e+00 -3.14154737e-02 -1.76459402e-01
8.19631338e-01 -5.30617833e-01 -1.32670184e-03 2.45721593e-01
8.09922695e-01 5.61740041e-01 -4.94108088e-02 -4.95679468e-01
-7.43864119e-01 8.48425388e-01 -4.53583449e-01 7.41112009e-02
1.05022156e+00 2.96658814e-01 8.00998434e-02 -2.76003689e-01
1.18269384e+00 -1.10703029e-01 -1.59701347e+00 -3.24418306e-01
-2.97353387e-01 -4.49561685e-01 9.03481245e-02 -8.58039320e-01
-1.25157070e+00 9.99040782e-01 1.05729103e+00 -4.06623155e-01
9.74779785e-01 1.58104748e-01 9.48983252e-01 4.20514971e-01
2.37879172e-01 -1.20926332e+00 1.03146821e-01 2.84067154e-01
9.38019454e-01 -1.21498251e+00 2.91413128e-01 -4.45059597e-01
-4.41374362e-01 1.29621005e+00 7.16245651e-01 -2.00367898e-01
4.05275524e-01 3.58604938e-01 -6.49550185e-02 6.95881993e-02
-2.45324280e-02 -5.42735219e-01 3.45890641e-01 2.75126547e-01
3.36400151e-01 2.72096433e-02 -3.77925783e-01 5.31005681e-01
1.40233025e-01 1.09518610e-01 -2.80108660e-01 1.22137606e+00
-4.61687535e-01 -9.55790997e-01 -3.47939163e-01 2.64856368e-01
-4.33100700e-01 -6.71505509e-03 -3.77597034e-01 8.90201747e-01
2.27444857e-01 5.58062613e-01 -7.37055093e-02 -4.93585467e-01
5.83817542e-01 -5.88738397e-02 7.63061285e-01 -3.97124112e-01
-6.06653273e-01 3.81923109e-01 -4.40669686e-01 -8.13798547e-01
-2.63253987e-01 -5.02445698e-01 -1.28980553e+00 -2.89863884e-01
-6.46518350e-01 -5.30285835e-01 6.54600918e-01 9.24385428e-01
2.11650506e-01 7.60765910e-01 5.73982716e-01 -1.36532986e+00
-7.14001119e-01 -1.20913148e+00 -4.98838454e-01 -9.97795463e-02
3.55389118e-01 -9.03294146e-01 -8.27383846e-02 -4.98234481e-01] | [6.567039966583252, -0.6388445496559143] |
cdd50d02-1f4e-4dcd-85d9-c280f0945a8b | document-shadow-removal-with-foreground | null | null | https://ieeexplore.ieee.org/document/9897217 | https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9897217 | Document Shadow Removal with Foreground Detection Learning From Fully Synthetic Images | Shadow removal for document images is a major task for digitized document applications. Recent shadow removal models have been trained on pairs of shadow images and shadow-free images. However, obtaining a large-scale and diverse dataset is laborious and remains a great challenge. Thus, only small real datasets are available. To create relatively large datasets, a graphic renderer has been used to synthesize shadows, nonetheless, it is still necessary to capture real documents. Thus, the number of unique documents is limited, which negatively affects a network’s performance. In this paper, we present a large-scale and diverse dataset called fully synthetic document shadow removal dataset (FSDSRD) that does not require capturing documents. The experiments showed that the networks (pre-)trained on FSDSRD provided better results than networks trained only on real datasets. Additionally, because foreground maps are available in our dataset, we leveraged them during training for multitask learning, which provided noticeable improvements. The code is available at: https://github.com/IsHYuhi/DSRFGD. | ['Yoshimitsu Aoki', 'Naofumi Akimoto', 'Yuhi Matsuo'] | 2022-10-18 | null | null | null | 2022-2022-10 | ['shadow-removal'] | ['computer-vision'] | [ 4.68726516e-01 -2.58163601e-01 2.46320903e-01 -3.79994839e-01
-6.73948526e-01 -5.37538111e-01 5.65280974e-01 -6.04619563e-01
-1.29758179e-01 8.45625103e-01 2.27066111e-02 -3.52831423e-01
3.27191502e-01 -6.25556827e-01 -7.59784520e-01 -8.55178475e-01
3.42877448e-01 2.58724838e-01 7.68877029e-01 5.67522310e-02
9.54873860e-02 5.10227978e-01 -1.42708445e+00 4.17869538e-01
1.04225028e+00 6.88706100e-01 7.75899410e-01 7.88231850e-01
-2.10188359e-01 5.87178946e-01 -9.02909219e-01 -1.27077490e-01
3.02315086e-01 -3.97139758e-01 -3.18258137e-01 -1.05824256e-02
6.96095884e-01 -7.78030038e-01 -4.20431465e-01 6.62858963e-01
5.25710106e-01 1.39103219e-01 3.78939211e-01 -1.42524481e+00
-7.20996201e-01 3.98541838e-01 -7.18399525e-01 7.93896839e-02
1.09969489e-01 3.10809864e-03 6.85937166e-01 -9.30150986e-01
5.79541206e-01 1.08175755e+00 6.04663312e-01 5.15103579e-01
-9.28238213e-01 -8.91373754e-01 2.95314670e-01 -6.39632419e-02
-1.06972134e+00 -4.93633091e-01 8.62787187e-01 1.26113873e-02
5.80147684e-01 4.99827981e-01 7.27752566e-01 1.61173236e+00
8.66400674e-02 1.16719115e+00 1.48457897e+00 -4.33936685e-01
-7.33721778e-02 2.07865193e-01 1.08929761e-01 6.16858244e-01
3.30272198e-01 -1.80712447e-01 -5.79338551e-01 4.46085297e-02
8.70496750e-01 2.13892430e-01 -6.58011079e-01 -2.05424324e-01
-1.15188456e+00 4.99739707e-01 3.35496873e-01 2.20289394e-01
-2.82413922e-02 1.33049339e-01 1.05636820e-01 -5.89492060e-02
5.43147027e-01 -5.80620766e-03 -2.46444404e-01 -3.02492222e-03
-1.01812518e+00 1.73477069e-01 7.36098409e-01 9.04268682e-01
6.01340234e-01 1.78104520e-01 -1.41348869e-01 8.99691701e-01
-4.55706865e-02 9.91309583e-01 2.47179106e-01 -7.61038840e-01
4.95977670e-01 3.09963137e-01 1.02084607e-01 -1.18623948e+00
-2.53882557e-01 -2.03302458e-01 -8.12638581e-01 1.49054676e-01
4.98290807e-01 -1.11121118e-01 -1.26792681e+00 1.43194091e+00
2.30156153e-01 3.29455256e-01 -5.50342984e-02 1.09272015e+00
8.98211360e-01 7.53348708e-01 -2.83347785e-01 -3.37654166e-03
1.04446900e+00 -1.32744718e+00 -8.30064476e-01 -5.44364452e-01
2.80817777e-01 -9.79120970e-01 1.57607377e+00 6.00481570e-01
-6.90172434e-01 -3.78988504e-01 -1.08630610e+00 -8.57924744e-02
-4.61595505e-01 2.64960170e-01 8.82498324e-01 6.68712318e-01
-7.60292530e-01 3.17632794e-01 -7.17538416e-01 -2.14589193e-01
6.37716472e-01 -9.08113420e-02 -7.97814950e-02 -3.31173241e-01
-1.02092266e+00 6.40664876e-01 8.45514089e-02 2.54060954e-01
-1.04828942e+00 -4.17671889e-01 -2.97727019e-01 -1.18117139e-01
6.90956056e-01 -9.49764699e-02 1.15033317e+00 -8.86723697e-01
-1.34359252e+00 4.46793705e-01 -1.52248889e-02 -1.40920114e-02
9.51388240e-01 -5.06101012e-01 -3.25368524e-01 1.11454777e-01
-7.57904025e-03 3.32681268e-01 1.02110231e+00 -1.85730672e+00
-3.84055853e-01 -2.01207727e-01 7.58982301e-02 2.02475429e-01
-5.49929380e-01 -1.20731965e-01 -9.10524547e-01 -8.78721893e-01
-8.02181512e-02 -1.10644293e+00 2.16283098e-01 8.59693587e-02
-7.55769610e-01 2.24405780e-01 1.30965531e+00 -8.64474356e-01
9.09902036e-01 -2.02977514e+00 -2.91856647e-01 1.43797640e-02
1.23333044e-01 4.98303920e-01 -1.70784861e-01 3.63083929e-01
2.23923936e-01 2.26154104e-02 -4.57314909e-01 -5.44495702e-01
-1.33354366e-01 1.61292002e-01 -6.12385571e-01 2.38818794e-01
-1.70248598e-01 7.08565891e-01 -5.93331397e-01 -5.47464252e-01
2.81426072e-01 5.87551057e-01 3.45807038e-02 2.30447307e-01
-4.42136019e-01 2.09742993e-01 -5.59823751e-01 8.65570068e-01
1.07079434e+00 -2.45680943e-01 2.14901820e-01 -2.14684188e-01
6.83910176e-02 -2.18583763e-01 -1.22468412e+00 1.61411321e+00
-7.15537429e-01 1.03520310e+00 3.18946652e-02 -3.98886621e-01
8.00347567e-01 2.39946898e-02 2.25639239e-01 -8.93914223e-01
6.81990013e-02 1.69300377e-01 -3.03263068e-01 -3.03775460e-01
6.75776184e-01 2.85155892e-01 2.08455786e-01 9.34758484e-01
-6.80308938e-01 -2.36616105e-01 1.80183858e-01 3.80895674e-01
1.18480265e+00 4.82040763e-01 -2.82123238e-01 2.82724835e-02
1.00857094e-01 1.20889083e-01 5.13264477e-01 7.52023101e-01
-4.29269439e-03 1.08644485e+00 2.55066544e-01 -2.96177447e-01
-9.48746622e-01 -1.07612586e+00 -5.84172793e-02 1.13316917e+00
4.33251381e-01 -2.45956033e-01 -7.90240765e-01 -6.44053698e-01
7.38375261e-02 7.32136905e-01 -4.61986244e-01 2.11393908e-01
-7.32969940e-01 -9.40513551e-01 6.44217849e-01 5.54278851e-01
7.25174725e-01 -1.11657465e+00 -7.86197305e-01 -1.73365071e-01
-2.63352066e-01 -1.19762123e+00 -5.46075940e-01 1.67181805e-01
-5.28136969e-01 -1.30700266e+00 -9.00833488e-01 -5.58312356e-01
6.74872160e-01 1.14995062e+00 9.15849030e-01 2.78179914e-01
-5.32251179e-01 7.38775730e-02 -4.05431181e-01 -6.70052171e-01
-1.70517385e-01 -4.12536114e-02 -1.76783100e-01 -1.59396797e-01
3.91322449e-02 -4.39508229e-01 -6.15949392e-01 3.58326703e-01
-1.15708780e+00 6.11426353e-01 7.69373596e-01 6.60342932e-01
2.88612545e-01 4.20384221e-02 3.49303633e-01 -1.36547005e+00
7.27747381e-01 1.96012333e-02 -6.63983881e-01 4.20857191e-01
-5.41073859e-01 -3.41196686e-01 7.72768021e-01 -4.82916296e-01
-1.69403374e+00 -1.17445213e-03 3.51670772e-01 -4.27122712e-01
-1.12921782e-01 -4.48681749e-02 -1.66611344e-01 6.62053376e-02
5.69190383e-01 1.92298770e-01 -3.11957985e-01 -4.68668967e-01
2.63020664e-01 8.71981800e-01 4.79035288e-01 -5.23738623e-01
9.31441069e-01 7.35063732e-01 -3.46808106e-01 -8.37555647e-01
-9.40911412e-01 -1.80504858e-01 -5.94122291e-01 -3.48964423e-01
6.05140030e-01 -8.27797651e-01 -2.99316466e-01 9.68579829e-01
-1.05200934e+00 -8.71023118e-01 3.17312241e-01 8.78831595e-02
-9.10928822e-04 2.90238410e-01 -3.10613841e-01 -7.73883998e-01
-3.98620337e-01 -8.68777871e-01 1.06842065e+00 3.68293434e-01
1.97596163e-01 -7.29835629e-01 -6.73820749e-02 5.47582328e-01
4.99333709e-01 3.89565766e-01 7.13862538e-01 -1.61607161e-01
-1.01011872e+00 -1.49311751e-01 -7.44023502e-01 3.04248840e-01
5.46921134e-01 3.17934543e-01 -1.38320708e+00 -2.21653417e-01
-2.38494009e-01 -4.29213792e-01 1.12485087e+00 3.38426918e-01
1.45266652e+00 -3.81630403e-03 -5.04399419e-01 5.33635259e-01
1.34334850e+00 1.91928849e-01 5.49480259e-01 4.13897902e-01
1.13182604e+00 3.92460942e-01 7.54965723e-01 4.70322281e-01
3.85754526e-01 4.34314340e-01 1.71165913e-01 -4.98742729e-01
-6.54648662e-01 -1.12790495e-01 1.63559571e-01 6.48735702e-01
-1.34963199e-01 -7.71506071e-01 -1.10872734e+00 3.62178355e-01
-1.84710371e+00 -6.76062167e-01 -3.10036361e-01 1.89493680e+00
8.48850310e-01 2.16579974e-01 -3.05128306e-01 -3.02367983e-03
7.54117429e-01 5.95926702e-01 -6.71576679e-01 9.13638100e-02
-3.98509264e-01 -4.75768838e-03 4.56180423e-01 3.71694148e-01
-9.77206409e-01 1.20965719e+00 5.47210407e+00 7.36314595e-01
-1.24159169e+00 1.66629463e-01 4.19907391e-01 -2.45867267e-01
-4.42308605e-01 -7.56968334e-02 -6.60003662e-01 6.05184495e-01
5.93883216e-01 2.01828957e-01 3.80253762e-01 7.43990064e-01
2.82144338e-01 -5.98731875e-01 -6.49626493e-01 8.11856151e-01
2.91173488e-01 -1.17112029e+00 -1.58612505e-01 -1.08299978e-01
8.84772122e-01 -9.33327060e-03 -7.37523660e-02 9.06296670e-02
4.21049625e-01 -7.83222497e-01 6.17577553e-01 4.14355755e-01
8.77532303e-01 -6.13055646e-01 5.23830056e-01 3.40392530e-01
-1.04659688e+00 1.23139888e-01 -3.37644964e-01 3.06173801e-01
2.40090325e-01 9.66855526e-01 -9.91376460e-01 3.87103438e-01
8.91906381e-01 4.63227093e-01 -8.10250819e-01 7.66102076e-01
-4.00890440e-01 4.68079478e-01 -2.35862717e-01 -8.31052288e-02
-1.12523019e-01 -2.79998600e-01 1.43733770e-01 1.35696793e+00
1.68436408e-01 -1.00992039e-01 1.67916417e-01 5.30264854e-01
-3.03228289e-01 -2.39874646e-01 -5.50854266e-01 -6.12156279e-02
5.55617034e-01 1.57444143e+00 -1.18969166e+00 -3.02893221e-01
-3.98314357e-01 1.21186864e+00 1.56947151e-01 4.64509040e-01
-1.25342894e+00 -4.57963467e-01 3.13868195e-01 -5.33870943e-02
8.29219893e-02 -2.57282168e-01 -3.85391235e-01 -1.21499419e+00
1.40444279e-01 -9.43112612e-01 -2.08900832e-02 -1.10482395e+00
-8.84170175e-01 5.72914779e-01 -1.23891562e-01 -1.02469397e+00
1.95882127e-01 -4.37417030e-01 -6.56628013e-01 6.58873260e-01
-1.62365055e+00 -1.43444264e+00 -1.20462894e+00 5.28806269e-01
7.35962570e-01 4.86314371e-02 7.58572698e-01 3.46500456e-01
-8.49103272e-01 3.96880001e-01 3.57995927e-01 7.10314363e-02
1.11861455e+00 -1.21580517e+00 5.84075034e-01 9.48447227e-01
1.05415449e-01 4.70072925e-01 5.65987706e-01 -8.71701241e-01
-1.50301111e+00 -1.07588494e+00 2.27226630e-01 -4.32898939e-01
3.26233268e-01 -8.05591941e-01 -1.17521966e+00 6.16438568e-01
2.37999275e-01 -3.49394865e-02 4.67953295e-01 -1.05566576e-01
-2.59542882e-01 -2.26004854e-01 -8.11951101e-01 7.72562087e-01
1.25223207e+00 -5.15752792e-01 -1.51880547e-01 7.03806996e-01
5.62565267e-01 -7.11608112e-01 -3.37393075e-01 7.66206682e-02
7.48033524e-01 -1.23339093e+00 8.90643418e-01 2.43125986e-02
5.13944924e-01 -3.09918582e-01 -8.44511241e-02 -1.16030490e+00
1.64559960e-01 -3.29419315e-01 -3.41993779e-01 1.40007913e+00
3.57749701e-01 -6.82141721e-01 9.09074545e-01 6.24322295e-01
-1.61574557e-01 -6.67873144e-01 -3.39388132e-01 -7.54069924e-01
-3.16866010e-01 -2.44428650e-01 6.31603420e-01 8.08972657e-01
-8.39572847e-01 3.87496948e-02 -6.46728098e-01 2.97004819e-01
6.09539866e-01 7.10744202e-01 1.17722130e+00 -1.02422571e+00
-2.08740473e-01 2.05529835e-02 4.57904071e-01 -8.46603751e-01
1.33447945e-01 -5.75891614e-01 3.44926208e-01 -1.72875333e+00
4.55344886e-01 -8.00136089e-01 2.23574601e-02 6.73884511e-01
-3.31561863e-01 4.49444950e-01 3.40002686e-01 3.95489544e-01
-4.04740572e-01 5.16477287e-01 1.57480991e+00 -3.04160789e-02
-1.51558653e-01 5.53002916e-02 -5.35274982e-01 7.34284997e-01
1.03175509e+00 -4.70076650e-01 -5.57898700e-01 -7.35897303e-01
-1.63117081e-01 -2.66340703e-01 4.31743771e-01 -9.14126456e-01
1.02667540e-01 -3.40329438e-01 8.01860809e-01 -8.66927266e-01
8.13786387e-01 -5.90303481e-01 1.60107195e-01 3.44588280e-01
6.76227883e-02 -1.75930932e-01 1.79749772e-01 5.43610752e-01
1.42423049e-01 -6.05042614e-02 6.23865843e-01 -2.08958596e-01
-6.98769212e-01 1.83232903e-01 1.01016983e-02 -1.39837146e-01
1.00232208e+00 -2.27829084e-01 -7.90966868e-01 -2.80006021e-01
2.57600192e-02 2.51355711e-02 7.16627717e-01 5.46776533e-01
7.47010648e-01 -1.03462267e+00 -3.66400778e-01 4.56686802e-02
-1.25171363e-01 4.25388694e-01 3.12571913e-01 4.97810960e-01
-7.09655225e-01 3.95860791e-01 -1.99604690e-01 -4.57036555e-01
-1.47353780e+00 1.49723217e-01 -1.22904249e-01 8.89876708e-02
-8.66122246e-01 7.38252819e-01 2.85517186e-01 -4.91758376e-01
2.83135653e-01 -3.12887698e-01 2.33929902e-01 1.79732159e-01
4.39364672e-01 4.15676802e-01 1.78978592e-01 -2.01700643e-01
-2.80735791e-01 3.54160905e-01 -1.77732319e-01 2.51430608e-02
1.57050526e+00 -6.32529333e-02 8.34723189e-02 3.35813761e-01
8.72924507e-01 2.28147492e-01 -1.71904993e+00 -2.49246404e-01
-1.81450322e-01 -9.69065845e-01 1.54075801e-01 -8.64592552e-01
-1.25637805e+00 7.60421991e-01 5.10608912e-01 5.33581823e-02
1.16556728e+00 -3.35929900e-01 1.02647686e+00 5.27600586e-01
4.45801586e-01 -1.35846782e+00 4.35293019e-01 3.13198060e-01
9.79260862e-01 -1.50308740e+00 3.05816531e-01 -5.43526471e-01
-8.28063607e-01 1.10916078e+00 7.78234780e-01 1.48067564e-01
2.33617544e-01 5.44724166e-01 4.64415818e-01 2.31504031e-02
-5.38633764e-01 3.15331598e-03 -8.57340842e-02 6.15061641e-01
2.23320127e-01 1.33930100e-02 1.97504416e-01 2.90922821e-01
-2.20817700e-01 -8.36638734e-02 8.23293388e-01 1.23465276e+00
-2.75211543e-01 -1.04233479e+00 -5.66525161e-01 5.89772463e-01
-1.12079397e-01 -1.26673073e-01 -6.41154885e-01 9.92357075e-01
-9.08859000e-02 8.12636375e-01 -2.46199161e-01 -1.23583667e-01
6.02708571e-02 -2.56019890e-01 4.43275183e-01 -5.67433298e-01
-8.80224481e-02 -3.07039097e-02 1.82466656e-01 -4.81345475e-01
-2.68167496e-01 -5.13275802e-01 -1.34725928e+00 -5.18677235e-01
-5.66681325e-01 -3.26207012e-01 8.10851753e-01 7.02866077e-01
1.81492776e-01 8.66648972e-01 5.69719970e-01 -1.02469277e+00
-9.46022123e-02 -9.81039941e-01 -6.27229035e-01 1.53611675e-01
3.18035603e-01 -7.50818193e-01 -2.83814251e-01 1.98648080e-01] | [10.852837562561035, -4.083192825317383] |
b7758c11-7927-4b61-aa53-15cea27a327a | facial-information-analysis-technology-for | 2111.09303 | null | https://arxiv.org/abs/2111.09303v1 | https://arxiv.org/pdf/2111.09303v1.pdf | Facial Information Analysis Technology for Gender and Age Estimation | This is a study on facial information analysis technology for estimating gender and age, and poses are estimated using a transformation relationship matrix between the camera coordinate system and the world coordinate system for estimating the pose of a face image. Gender classification was relatively simple compared to age estimation, and age estimation was made possible using deep learning-based facial recognition technology. A comparative CNN was proposed to calculate the experimental results using the purchased database and the public database, and deep learning-based gender classification and age estimation performed at a significant level and was more robust to environmental changes compared to the existing machine learning techniques. | ['Sua Jung', 'Gilheum Park'] | 2021-11-17 | null | null | null | null | ['age-estimation', 'age-estimation'] | ['computer-vision', 'miscellaneous'] | [-6.95648134e-01 1.83205456e-01 -5.47199324e-02 -7.96488523e-01
2.18817100e-01 8.51498246e-02 3.96320492e-01 -1.48983940e-01
-6.11740351e-01 4.79758561e-01 -1.01278303e-02 5.04370272e-01
2.39942759e-01 -9.08117354e-01 -2.03610197e-01 -7.96795428e-01
-2.31153414e-01 3.60418707e-01 -5.13264835e-01 3.55615839e-02
4.84953433e-01 7.28903234e-01 -1.80084622e+00 -4.18458462e-01
8.46153721e-02 1.48690617e+00 -5.75385749e-01 4.47805524e-01
1.11556411e-01 4.49863851e-01 -7.62329161e-01 -8.75688553e-01
2.71959931e-01 1.24260619e-01 -4.88128722e-01 -1.35157779e-01
8.57160926e-01 -6.56578600e-01 -3.98066044e-01 8.33578706e-01
8.60532224e-01 -2.47988865e-01 7.54582644e-01 -1.76824880e+00
-5.04464328e-01 1.94488361e-01 -7.73873150e-01 -2.58139789e-01
7.32820928e-01 -6.95886731e-01 3.67738232e-02 -9.80343878e-01
3.95486474e-01 1.74090385e+00 1.09833944e+00 8.45852017e-01
-1.80188611e-01 -1.57764316e+00 -1.63972661e-01 2.93367505e-01
-1.80603826e+00 -1.92189604e-01 5.65209985e-01 -3.82725120e-01
6.26077831e-01 -1.15493640e-01 1.09344459e+00 1.09028924e+00
5.73326290e-01 5.31237945e-02 1.00681400e+00 -2.92205095e-01
-1.05315596e-01 -6.72704726e-02 -2.12186992e-01 1.16012836e+00
2.57646620e-01 9.50877443e-02 -8.79888535e-01 -7.60771632e-02
9.20041084e-01 -4.24329080e-02 7.01429725e-01 -3.60494368e-02
-6.98146403e-01 7.61914670e-01 2.11061984e-01 -2.01513916e-01
-1.83228433e-01 6.70269355e-02 5.37043393e-01 1.83976039e-01
7.92315483e-01 2.29551300e-01 -6.17167056e-01 -2.72164822e-01
-8.33149076e-01 1.44669890e-01 6.96437538e-01 8.31446350e-01
5.67911565e-01 4.43751067e-01 3.64872634e-01 6.34138942e-01
6.20267451e-01 7.78525829e-01 4.96280402e-01 -1.17410684e+00
-3.68761718e-02 7.51279831e-01 -4.16948497e-01 -1.44595790e+00
-8.34392428e-01 1.97924569e-01 -8.06963384e-01 3.68865639e-01
6.92773104e-01 -4.81831849e-01 -9.58358526e-01 1.59104788e+00
5.90304792e-01 -6.27455786e-02 -2.40390807e-01 4.02519614e-01
1.10537720e+00 2.84496456e-01 2.01670602e-01 -3.10590655e-01
1.52326179e+00 -3.09137434e-01 -9.45746243e-01 -1.46620050e-01
3.88290167e-01 -6.48319483e-01 5.61184958e-02 5.09406030e-01
-8.61902297e-01 -5.60697973e-01 -1.07321727e+00 4.52592038e-02
-6.61126792e-01 4.70356286e-01 1.11093926e+00 1.43702459e+00
-1.13200593e+00 4.74669874e-01 -6.96668327e-01 -7.45210946e-01
4.40560222e-01 1.01105940e+00 -1.06505656e+00 4.69718546e-01
-1.08527946e+00 1.20911956e+00 1.39041126e-01 3.04556042e-01
-4.02499855e-01 -5.78004658e-01 -1.23036993e+00 -4.66720313e-01
-4.81017947e-01 -3.71180117e-01 9.72431242e-01 -9.53565180e-01
-1.52475822e+00 1.31410277e+00 -2.37744465e-01 -1.28487796e-02
4.23292220e-01 -2.80244470e-01 -6.18112266e-01 7.76383877e-02
-1.88827068e-01 9.89343762e-01 1.23456705e+00 -5.50401270e-01
-5.15190482e-01 -1.30116951e+00 -3.08779925e-01 1.89245090e-01
-8.42511475e-01 4.76610899e-01 -1.66809171e-01 -2.11753353e-01
5.14735401e-01 -9.02763844e-01 2.40355387e-01 3.64731103e-01
1.86583921e-01 -4.18718159e-01 1.15830076e+00 -1.22731626e+00
1.06933355e+00 -1.90953362e+00 -8.80095512e-02 2.40696967e-01
3.88893068e-01 -2.99823303e-02 3.14787984e-01 7.84737691e-02
-3.14135641e-01 -1.19545288e-01 7.16200948e-01 -2.97196746e-01
-2.07504392e-01 -6.25780821e-02 4.53040183e-01 8.10296237e-01
-1.62244588e-01 3.20560604e-01 -4.72440034e-01 -8.95292878e-01
4.90039587e-02 6.43423557e-01 -2.84505248e-01 2.60209918e-01
7.24284828e-01 3.12267005e-01 -1.86774001e-01 1.24951708e+00
1.03716242e+00 8.10487509e-01 1.59066111e-01 -5.01263440e-01
1.17251739e-01 -6.51950181e-01 -1.05511117e+00 1.14143634e+00
-5.81723392e-01 6.99660897e-01 1.16390340e-01 -6.20274007e-01
1.68934798e+00 2.93223351e-01 6.68960690e-01 -3.93185914e-01
8.33769381e-01 4.87214923e-02 -9.11071822e-02 -4.83194739e-01
5.28489709e-01 1.48257017e-01 -1.59389853e-01 1.15804948e-01
5.12207329e-01 -1.09627709e-01 2.18063928e-02 -3.13196063e-01
1.54965222e-01 2.18699098e-01 3.31765383e-01 -2.89048493e-01
6.60192847e-01 -8.10103059e-01 7.85743356e-01 -6.68998435e-02
-6.40526533e-01 4.79734749e-01 6.42472327e-01 -1.08853793e+00
-8.22230995e-01 -1.03103137e+00 -2.09027112e-01 1.19960928e+00
-1.77977175e-01 -4.85917717e-01 -1.18980205e+00 -4.92665768e-01
2.57166237e-01 -3.24677020e-01 -1.08096075e+00 -3.55375916e-01
-3.68843526e-01 -7.42189884e-01 7.34749317e-01 8.90924275e-01
1.03625298e+00 -6.32759035e-01 -8.58037770e-02 -5.18598795e-01
2.39751846e-01 -9.64971960e-01 -2.02000082e-01 -3.76522571e-01
-9.29372847e-01 -1.29306543e+00 -6.22865200e-01 -1.03428626e+00
9.54413831e-01 -7.49011219e-01 9.19616461e-01 2.62955818e-02
-2.97046840e-01 4.91212219e-01 1.65549502e-01 -9.96853352e-01
-2.31349245e-02 2.60755688e-01 9.36336458e-01 5.86903393e-02
1.05827892e+00 -5.51933467e-01 -8.23917806e-01 2.88213640e-01
4.44757044e-02 -3.52852970e-01 3.36283088e-01 4.54192966e-01
-3.03246766e-01 -6.65762126e-02 5.93655288e-01 -3.71196181e-01
3.09928119e-01 -1.06421277e-01 -4.67187285e-01 -1.53758138e-01
-6.52024627e-01 -2.67729938e-01 -5.34432195e-02 -3.62890542e-01
-1.05418050e+00 4.28302079e-01 -3.53290081e-01 -1.10150255e-01
3.46838534e-02 7.02502280e-02 -3.35964292e-01 -4.15705502e-01
3.86015445e-01 -1.13771357e-01 7.02021956e-01 -2.17280373e-01
-1.18628986e-01 9.77393210e-01 4.12202448e-01 -6.04212701e-01
6.59649789e-01 2.85613179e-01 4.78079021e-01 -9.01469231e-01
-3.55612040e-01 1.38422862e-01 -1.38510656e+00 -8.42887878e-01
8.46695483e-01 -1.22111142e+00 -1.41428697e+00 1.53184319e+00
-8.73412728e-01 4.73803520e-01 5.82305670e-01 6.32489443e-01
-2.21402749e-01 3.38988826e-02 -6.31502271e-01 -9.28275883e-01
-5.56709766e-01 -8.93140972e-01 1.08878791e+00 5.60743809e-01
-5.11593044e-01 -1.02164745e+00 -2.30476528e-01 3.35300595e-01
2.08052859e-01 5.00211060e-01 5.05871296e-01 -2.52221584e-01
2.87177861e-01 -7.96633422e-01 -6.74509183e-02 2.79464185e-01
1.46212846e-01 7.86119282e-01 -1.18114924e+00 -2.59638399e-01
-1.64753154e-01 -2.73250997e-01 1.68118834e-01 3.96681190e-01
1.42844808e+00 -1.45814911e-01 -2.66813427e-01 7.83520758e-01
9.26611900e-01 4.18424457e-01 7.66797483e-01 2.80926555e-01
8.07165086e-01 7.13833332e-01 5.88960052e-01 8.58591914e-01
6.98301315e-01 4.19716716e-01 3.77940953e-01 2.20027212e-02
2.60917604e-01 -1.35930225e-01 7.31370598e-02 7.16806710e-01
-7.28710651e-01 7.52209961e-01 -9.47503507e-01 1.44627690e-01
-1.28610134e+00 -7.68378735e-01 4.83901575e-02 1.99364960e+00
7.29088724e-01 -2.78846800e-01 3.90005022e-01 1.74045920e-01
9.08715606e-01 8.53994489e-03 -4.43729043e-01 -9.42503870e-01
2.99761415e-01 3.62355471e-01 5.52242637e-01 5.17334640e-02
-1.22992790e+00 7.65563607e-01 8.17534351e+00 4.00167525e-01
-1.45810425e+00 -2.07893431e-01 9.94729161e-01 -9.82408822e-02
8.04696620e-01 -6.86518967e-01 -1.11428154e+00 3.39224517e-01
9.00930047e-01 -3.57074976e-01 1.63008332e-01 1.26088250e+00
-2.33577088e-01 -2.63803363e-01 -1.10994470e+00 1.64316952e+00
4.36148852e-01 -7.07679272e-01 -1.26199573e-01 1.52647480e-01
4.07823831e-01 -7.15130091e-01 4.61925536e-01 5.10232329e-01
-3.57622743e-01 -1.15843320e+00 5.88739336e-01 5.15562475e-01
9.79973912e-01 -1.25365055e+00 9.23678279e-01 -2.23557651e-01
-1.28813624e+00 -3.67671281e-01 -3.00828308e-01 -7.99487472e-01
-7.10660517e-01 -2.61492189e-02 -6.97558880e-01 1.65742755e-01
1.14244020e+00 7.88403153e-01 -1.11871207e+00 5.10683954e-01
-2.73002889e-02 -5.78280911e-02 -1.38438672e-01 -1.33925304e-01
-4.94323790e-01 -2.19527915e-01 -2.41266713e-01 6.76438749e-01
7.54695117e-01 -3.87596697e-01 -3.48718196e-01 -1.29584083e-02
-3.44953053e-02 1.91632405e-01 -5.72475016e-01 6.81648329e-02
5.93323767e-01 1.68673491e+00 -6.13911450e-01 9.18774977e-02
-2.59456754e-01 5.25069296e-01 -1.47006270e-02 -1.47944823e-01
-7.17827439e-01 -2.82830805e-01 1.13578296e+00 1.89564690e-01
-3.44833374e-01 -2.96951175e-01 -4.35465842e-01 -6.53071582e-01
-2.66091853e-01 -4.88250852e-01 4.07334834e-01 -8.21857214e-01
-9.25488412e-01 3.61658692e-01 6.70624152e-02 -1.07467723e+00
-5.78796566e-01 -1.15268970e+00 -6.72323942e-01 7.38613248e-01
-4.55533206e-01 -1.62218559e+00 -4.63714182e-01 4.38379318e-01
-8.72732624e-02 -8.71246994e-01 1.21524847e+00 3.69445056e-01
-7.67345309e-01 1.11438572e+00 -3.29974264e-01 5.76487184e-01
8.75723779e-01 -1.15366805e+00 2.90644243e-02 4.73727316e-01
-7.47843146e-01 6.66875899e-01 4.62820023e-01 -5.29960036e-01
-1.49580085e+00 -8.37519944e-01 1.00389087e+00 -6.95348561e-01
3.11954379e-01 -7.05745637e-01 -4.81339055e-04 6.81974947e-01
7.19736964e-02 -3.56576741e-02 7.98851192e-01 4.78355050e-01
-5.29935420e-01 -7.36744463e-01 -1.75222909e+00 3.90667647e-01
8.08328390e-01 -4.53420550e-01 -1.34768009e-01 -7.83613473e-02
4.70534824e-02 -6.90559864e-01 -1.40881455e+00 5.26917100e-01
1.58876598e+00 -7.59173930e-01 1.00705361e+00 -2.64224440e-01
4.67162579e-01 1.85844138e-01 1.53396308e-01 -9.29875851e-01
-1.40181348e-01 -1.80789053e-01 -3.86241168e-01 1.56887972e+00
-3.00628264e-02 -5.43604851e-01 1.29661572e+00 8.55033398e-01
5.09515941e-01 -8.19797397e-01 -1.13483107e+00 -4.92265046e-01
1.96139798e-01 6.56112432e-02 1.00213742e+00 7.59617209e-01
-1.31585971e-01 1.58144683e-02 -4.34233606e-01 2.37511486e-01
6.54472291e-01 -6.44717693e-01 8.10084760e-01 -1.54829860e+00
7.93418527e-01 -2.52570629e-01 -1.39510477e+00 3.95289838e-01
5.18883944e-01 -1.36306778e-01 -5.70450962e-01 -1.11522079e+00
4.37396504e-02 2.78138608e-01 -1.34631857e-01 3.99894297e-01
1.78957105e-01 7.83123255e-01 -2.05526605e-01 -4.48261708e-01
1.30584883e-02 3.28743964e-01 1.04728758e+00 -1.42076045e-01
2.43349120e-01 2.27734134e-01 -6.08337522e-01 1.12229872e+00
7.53145099e-01 -5.70879728e-02 -2.44575143e-01 -2.11687937e-01
4.88166481e-01 -1.94974333e-01 -1.95597127e-01 -1.20652986e+00
2.39472732e-01 1.46853909e-01 1.48721600e+00 -7.70818353e-01
7.69645631e-01 -6.84059620e-01 2.00926922e-02 5.90885699e-01
3.34786326e-01 3.36154640e-01 1.41800046e-01 -1.72479868e-01
-1.87128887e-01 1.27346307e-01 8.62621129e-01 2.66491193e-02
-9.00425732e-01 7.79610515e-01 -3.68946820e-01 -6.42221391e-01
1.21408415e+00 -7.53467441e-01 -7.91537687e-02 -4.61525619e-01
-9.99233961e-01 -1.74714357e-01 3.10216844e-01 7.04338551e-01
6.39453530e-01 -1.79298460e+00 -5.53059042e-01 5.46896160e-01
1.37259513e-01 -5.34234464e-01 1.13582484e-01 5.89581966e-01
-1.01023579e+00 9.41714123e-02 -1.06857085e+00 -4.19711262e-01
-1.95799196e+00 5.45731410e-02 5.06340683e-01 4.49837059e-01
3.47478747e-01 1.08979392e+00 -2.35522225e-01 -8.39218438e-01
5.08399725e-01 1.04695536e-01 -8.85151684e-01 7.28857219e-01
6.61226928e-01 7.25958586e-01 1.84057638e-01 -1.09775853e+00
-6.11752927e-01 8.68835330e-01 7.95862228e-02 2.48756528e-01
1.30674291e+00 -2.77654618e-01 -6.14688277e-01 2.20669940e-01
1.35439932e+00 -3.46341521e-01 -8.72679114e-01 2.22120658e-01
-2.60335535e-01 -7.20875740e-01 -2.28570718e-02 -4.17524219e-01
-1.68214929e+00 6.96650028e-01 1.38365531e+00 -2.21334279e-01
1.16037452e+00 -4.52707469e-01 4.18016076e-01 4.75226313e-01
4.86863226e-01 -1.69029295e+00 1.30800366e-01 6.02176189e-01
7.01658726e-01 -1.39776278e+00 4.73068506e-01 -4.30921346e-01
-2.96448022e-01 1.57451928e+00 1.27597702e+00 3.13207246e-02
1.21945775e+00 1.59928799e-01 3.72909278e-01 -6.09085858e-02
-8.84148479e-02 3.70984316e-01 1.93284690e-01 1.08469760e+00
7.17698932e-01 2.50189118e-02 -2.28346244e-01 5.86328149e-01
-9.02603745e-01 -2.22131088e-02 2.87248671e-01 7.48175561e-01
-1.73604339e-01 -1.08846748e+00 -6.71955466e-01 4.91939306e-01
-7.45836556e-01 3.31563562e-01 -4.39439654e-01 9.26811397e-01
4.88591284e-01 6.82851315e-01 5.81340730e-01 -7.82576442e-01
-1.08970948e-01 1.99646443e-01 1.06309116e+00 -2.81470388e-01
-3.47173095e-01 -6.52128398e-01 3.45932692e-02 -6.25543177e-01
-4.48731363e-01 -7.63122201e-01 -7.14360237e-01 -7.89178431e-01
8.20834935e-03 -2.64265597e-01 1.26412344e+00 7.60294914e-01
-5.17876111e-02 -3.27699482e-02 1.06750476e+00 -1.07553661e+00
4.07798961e-02 -1.15287614e+00 -1.14738333e+00 1.57819018e-01
-1.91583365e-01 -1.03383684e+00 -2.21401811e-01 7.99707845e-02] | [13.550028800964355, 0.9527871012687683] |
2afca225-324e-4471-83ae-1ffbf6af15be | invariant-collaborative-filtering-to | 2302.05328 | null | https://arxiv.org/abs/2302.05328v3 | https://arxiv.org/pdf/2302.05328v3.pdf | Invariant Collaborative Filtering to Popularity Distribution Shift | Collaborative Filtering (CF) models, despite their great success, suffer from severe performance drops due to popularity distribution shifts, where these changes are ubiquitous and inevitable in real-world scenarios. Unfortunately, most leading popularity debiasing strategies, rather than tackling the vulnerability of CF models to varying popularity distributions, require prior knowledge of the test distribution to identify the degree of bias and further learn the popularity-entangled representations to mitigate the bias. Consequently, these models result in significant performance benefits in the target test set, while dramatically deviating the recommendation from users' true interests without knowing the popularity distribution in advance. In this work, we propose a novel learning framework, Invariant Collaborative Filtering (InvCF), to discover disentangled representations that faithfully reveal the latent preference and popularity semantics without making any assumption about the popularity distribution. At its core is the distillation of unbiased preference representations (i.e., user preference on item property), which are invariant to the change of popularity semantics, while filtering out the popularity feature that is unstable or outdated. Extensive experiments on five benchmark datasets and four evaluation settings (i.e., synthetic long-tail, unbiased, temporal split, and out-of-distribution evaluations) demonstrate that InvCF outperforms the state-of-the-art baselines in terms of popularity generalization ability on real recommendations. Visualization studies shed light on the advantages of InvCF for disentangled representation learning. Our codes are available at https://github.com/anzhang314/InvCF. | ['Tat-Seng Chua', 'Yancheng Yuan', 'Xiang Wang', 'Jingnan Zheng', 'An Zhang'] | 2023-02-10 | null | null | null | null | ['collaborative-filtering'] | ['miscellaneous'] | [-1.92291856e-01 -6.72444105e-01 -5.68532884e-01 -1.93316624e-01
-5.09189904e-01 -7.27679908e-01 4.66137320e-01 -2.92585697e-03
2.91199377e-03 7.12664843e-01 5.34331381e-01 -2.00041518e-01
-3.32535356e-01 -8.35163057e-01 -7.51637757e-01 -9.08772230e-01
-1.44838676e-01 5.40171742e-01 -5.93156591e-02 -2.03951687e-01
2.61543363e-01 -6.26578704e-02 -1.57112324e+00 2.17288300e-01
9.17284012e-01 9.20202136e-01 -1.24153383e-01 2.68520206e-01
1.58710465e-01 6.19046330e-01 -2.21764550e-01 -2.94274509e-01
3.54375213e-01 -8.33807290e-02 -3.57311636e-01 -1.42485738e-01
4.37253088e-01 -4.99375612e-01 -7.69187152e-01 1.11685312e+00
4.02287066e-01 1.93148926e-01 6.98965609e-01 -1.29396176e+00
-1.17852283e+00 9.99085248e-01 -7.49315798e-01 3.78805786e-01
2.55133569e-01 7.34080970e-02 1.41048431e+00 -8.80567193e-01
2.92932481e-01 1.09539139e+00 5.96140325e-01 2.37315878e-01
-1.50611556e+00 -1.21332693e+00 6.51984334e-01 2.88318276e-01
-1.52581263e+00 -3.64322394e-01 7.31113613e-01 -5.37607431e-01
1.60627574e-01 2.55304843e-01 4.81178075e-01 1.66080976e+00
8.25338811e-02 9.84692395e-01 1.09968328e+00 1.91272691e-01
2.95979917e-01 5.62379621e-02 4.64938402e-01 2.11300567e-01
7.38247991e-01 2.22600833e-01 -6.52964950e-01 -7.11946905e-01
7.02785552e-01 5.26470065e-01 -6.83408320e-01 -8.14474225e-01
-9.14381623e-01 8.95888925e-01 2.55456895e-01 -1.03703633e-01
-3.93200189e-01 5.00193648e-02 2.68396616e-01 6.49323165e-01
6.77161157e-01 3.42084706e-01 -7.63501048e-01 -1.66324720e-01
-1.04856384e+00 5.49613357e-01 7.92682946e-01 7.92107880e-01
6.24595463e-01 4.29150052e-02 -2.69811243e-01 6.01398408e-01
3.65838677e-01 8.54365647e-01 4.59321946e-01 -6.30791068e-01
2.46508360e-01 1.58952832e-01 4.87321973e-01 -9.27122295e-01
4.03593592e-02 -9.85059977e-01 -7.91890085e-01 -2.85923213e-01
6.33885145e-01 5.01906723e-02 -6.42709732e-01 2.01817799e+00
1.02106512e-01 7.44248331e-01 -3.38579267e-01 8.11191976e-01
5.11721313e-01 4.40049589e-01 8.00278224e-03 -4.83855903e-02
1.03811407e+00 -6.50840402e-01 -3.48277152e-01 3.07757378e-01
3.27439696e-01 -5.43516219e-01 1.22154748e+00 6.88985169e-01
-9.43891823e-01 -3.28857869e-01 -8.29587877e-01 3.42304021e-01
-5.28429672e-02 -3.99881214e-01 7.70131052e-01 5.08522749e-01
-5.96957445e-01 7.62934387e-01 -6.52733624e-01 4.08368595e-02
5.69888353e-01 3.48713070e-01 -1.34234622e-01 -2.96561748e-01
-1.42363262e+00 3.01231802e-01 -9.81953219e-02 -2.59033114e-01
-9.41538513e-01 -1.31118679e+00 -3.19169879e-01 5.00857890e-01
4.36601609e-01 -8.76273096e-01 1.22054660e+00 -1.18694556e+00
-1.29951251e+00 1.92363530e-01 1.33959457e-01 -3.43015552e-01
5.95114112e-01 -6.04908884e-01 -6.16762459e-01 -3.13314915e-01
-6.48618937e-02 1.71084311e-02 1.18352604e+00 -1.21748245e+00
-5.39441884e-01 -1.62243351e-01 -3.66474167e-02 -8.67204517e-02
-3.91345203e-01 -5.73982477e-01 -2.41213411e-01 -8.85407388e-01
-4.32948433e-02 -1.01229060e+00 -7.08002225e-03 -1.28058210e-01
-2.50475794e-01 -9.17731002e-02 5.77270925e-01 -3.90051425e-01
1.46451497e+00 -2.27550435e+00 1.01759143e-01 3.73353124e-01
4.09620792e-01 3.39420862e-03 -3.60574841e-01 6.48502827e-01
-5.38425259e-02 1.39623031e-01 3.52437913e-01 -1.97200447e-01
1.66004017e-01 1.89318582e-01 -9.72263575e-01 8.09244573e-01
-4.61074114e-01 7.76702106e-01 -1.13746655e+00 1.51349172e-01
-1.27242237e-01 4.09825414e-01 -1.11744022e+00 1.03601366e-01
-2.90868938e-01 6.39813364e-01 -4.92235273e-01 5.54636240e-01
9.10648882e-01 -7.08325803e-01 2.28898793e-01 -1.19342141e-01
2.67892063e-01 4.07387346e-01 -1.19522917e+00 1.58958447e+00
-5.35303175e-01 2.01368704e-02 -6.12411015e-02 -6.82291865e-01
6.10066593e-01 2.39357650e-01 5.20222425e-01 -5.87754786e-01
1.66757390e-01 7.19618052e-02 4.67320941e-02 -7.51165226e-02
6.08943224e-01 -1.51218763e-02 -5.60262948e-02 7.75980353e-01
-3.57852876e-02 5.02999485e-01 -1.11931890e-01 4.38569039e-01
8.23523700e-01 -4.11268771e-02 9.53796059e-02 -3.25046659e-01
9.82987881e-02 -6.00676000e-01 7.41586268e-01 1.15129721e+00
4.94984016e-02 5.62443495e-01 3.79554003e-01 -3.08250636e-01
-8.71448338e-01 -1.55720603e+00 -3.11355311e-02 1.35270274e+00
4.58870649e-01 -4.77424294e-01 7.65947625e-02 -6.96436942e-01
4.59897727e-01 8.24603558e-01 -6.17885649e-01 -4.11245674e-01
-1.71896502e-01 -6.36149526e-01 1.51868373e-01 3.45557362e-01
1.08943149e-01 -4.48832333e-01 1.51159868e-01 4.10541110e-02
-1.50390700e-01 -5.78841031e-01 -7.25192785e-01 -3.37400556e-01
-8.16205025e-01 -1.10986137e+00 -6.32990241e-01 -1.41599506e-01
5.05075574e-01 7.74861991e-01 1.18569303e+00 2.78766397e-02
2.27383301e-01 5.13986230e-01 -4.87641156e-01 9.39683337e-03
3.19632255e-02 1.17106326e-01 3.58406067e-01 2.27031067e-01
3.63798052e-01 -1.14091754e+00 -1.00668597e+00 4.85963941e-01
-8.89462888e-01 -2.05651879e-01 4.73628998e-01 1.02836370e+00
4.13016379e-01 9.75744575e-02 6.80378497e-01 -1.37764430e+00
7.04082310e-01 -1.26728439e+00 -3.71257484e-01 1.02107590e-02
-9.15438771e-01 5.34603894e-02 6.51534379e-01 -9.30286825e-01
-8.75898004e-01 -6.21360064e-01 1.55052319e-01 -8.13304484e-01
1.51810870e-01 5.70872128e-01 -7.71168545e-02 4.46088523e-01
6.43006861e-01 2.93939471e-01 -1.04024567e-01 -9.20245647e-01
6.76925421e-01 5.47995508e-01 3.28552932e-01 -9.45633948e-01
9.53264117e-01 5.65906882e-01 -4.89565909e-01 -4.24896955e-01
-1.06976449e+00 -6.32068276e-01 -2.37523466e-01 2.98701704e-01
-1.20337762e-01 -1.34837699e+00 -4.00981396e-01 3.85851651e-01
-4.85439718e-01 -1.45416185e-01 -4.36483532e-01 5.47474742e-01
-3.19503099e-01 5.30263841e-01 -4.91067708e-01 -5.44206500e-01
-2.91492462e-01 -7.71099865e-01 6.77941680e-01 1.17384724e-01
-2.61726946e-01 -9.51393127e-01 2.36801118e-01 2.03967094e-01
7.60155797e-01 -2.93292195e-01 1.06602573e+00 -8.75896037e-01
-5.45993507e-01 -2.89647430e-01 -1.05705939e-01 3.26157659e-01
2.13209733e-01 -1.50642702e-02 -7.99328983e-01 -8.35509658e-01
-2.37271681e-01 -2.81170577e-01 1.00060058e+00 3.15096140e-01
1.40613782e+00 -5.58243155e-01 -1.97994635e-01 6.85235322e-01
1.05384707e+00 -4.55945313e-01 4.37954664e-01 -5.17547503e-02
4.49101537e-01 6.63795322e-02 3.88017714e-01 8.64160359e-01
3.63360852e-01 6.18626595e-01 4.61672336e-01 3.41988176e-01
-6.18253686e-02 -8.09454918e-01 4.69274133e-01 9.53149199e-01
-1.09352078e-02 -5.43818474e-01 -1.98535725e-01 3.20883423e-01
-1.89971900e+00 -1.19640625e+00 1.15706392e-01 2.63230634e+00
7.71267176e-01 9.52384248e-02 2.82641560e-01 -5.67517430e-02
6.70851231e-01 2.62769341e-01 -9.32121038e-01 8.41987282e-02
-5.84621504e-02 1.97294399e-01 5.65062225e-01 3.86610299e-01
-1.00095522e+00 6.47243798e-01 5.38460922e+00 8.62489641e-01
-1.22228360e+00 2.34570548e-01 3.85174453e-01 -4.53863055e-01
-8.21462095e-01 7.83933420e-03 -5.93997717e-01 8.67405891e-01
7.15073049e-01 -4.55750763e-01 7.27686048e-01 9.34320569e-01
8.20071846e-02 3.99243355e-01 -1.04056537e+00 1.08362985e+00
-1.55162618e-01 -1.20732164e+00 2.60771930e-01 1.52161106e-01
9.10884500e-01 3.95320356e-01 6.96834385e-01 5.95139444e-01
7.39789426e-01 -8.10597718e-01 8.79898310e-01 6.61634862e-01
7.05579877e-01 -4.46000904e-01 5.56567371e-01 3.64275873e-01
-8.84645104e-01 -3.08158666e-01 -5.11604309e-01 -3.23420316e-01
-7.52078788e-03 8.71423900e-01 -2.84090728e-01 4.82071072e-01
6.04403853e-01 1.08532703e+00 -3.88523489e-01 1.13104105e+00
-3.21908206e-01 1.13462770e+00 -1.10396758e-01 2.66235739e-01
-1.91656366e-01 -2.06069171e-01 6.01994812e-01 7.83006191e-01
4.33068514e-01 -8.73794407e-02 2.47850835e-01 8.77627790e-01
-4.18982238e-01 -5.73976059e-03 -2.69104809e-01 -2.25464642e-01
7.67441511e-01 9.80269372e-01 -2.14716688e-01 -1.67163387e-01
-5.54055512e-01 7.56218433e-01 2.31767923e-01 7.03500032e-01
-1.00093710e+00 1.93121597e-01 1.18830395e+00 4.20980006e-01
6.52530074e-01 -1.37715578e-01 2.01903984e-01 -1.83254433e+00
-3.07944447e-01 -1.05848408e+00 4.59952891e-01 -2.80786574e-01
-2.21553349e+00 1.00935206e-01 -6.26712516e-02 -1.33679008e+00
1.64755121e-01 -2.24993482e-01 -6.48653090e-01 8.56668115e-01
-1.49140763e+00 -9.98530507e-01 -1.14837185e-01 7.75311947e-01
4.57074791e-01 -2.40979958e-02 5.65993547e-01 4.51884061e-01
-4.52224106e-01 9.24340427e-01 7.35822678e-01 -2.60686398e-01
9.48176503e-01 -9.71512377e-01 1.35052517e-01 3.94518912e-01
3.64442140e-01 1.03094149e+00 8.24093401e-01 -5.38206995e-01
-1.65279400e+00 -9.93619502e-01 4.36347306e-01 -5.81089079e-01
1.09909773e+00 -4.69323546e-01 -1.09008360e+00 6.38539851e-01
-1.86742350e-01 3.23956609e-01 1.02638912e+00 6.75565124e-01
-1.05367553e+00 -6.96780607e-02 -8.30831468e-01 5.26923716e-01
1.15030992e+00 -5.13933539e-01 -4.30924714e-01 2.35185772e-01
6.65080547e-01 -1.41911000e-01 -8.30431461e-01 1.84514388e-01
1.06336617e+00 -8.79464209e-01 1.10322702e+00 -9.78994966e-01
3.07331383e-01 -1.79397941e-01 -3.98881704e-01 -1.32828081e+00
-8.94952893e-01 -6.46372259e-01 -4.87106651e-01 1.06841326e+00
3.08067918e-01 -8.74310493e-01 7.09806383e-01 3.34605753e-01
3.12342435e-01 -8.45605075e-01 -7.40686238e-01 -7.06553936e-01
3.20128798e-01 -1.63077235e-01 8.50567043e-01 1.09374690e+00
-2.02738076e-01 2.41630137e-01 -9.40741181e-01 4.51779038e-01
5.71946740e-01 6.36663795e-01 8.80453646e-01 -1.45424068e+00
-6.96935236e-01 -4.23214197e-01 -9.81390178e-02 -1.42607367e+00
1.44942895e-01 -9.05995965e-01 -5.75578213e-01 -9.81948495e-01
4.07501608e-01 -7.29079485e-01 -1.04165721e+00 2.20546857e-01
-2.14016989e-01 -2.62976624e-02 1.13897726e-01 7.23316789e-01
-7.71284878e-01 8.51524830e-01 1.12749648e+00 -2.75591433e-01
-9.72210169e-02 4.14448768e-01 -1.25697529e+00 5.69634140e-01
7.80248523e-01 -6.80519044e-01 -8.04202259e-01 -2.82250792e-01
5.81324816e-01 -2.40986362e-01 2.50131607e-01 -5.87014079e-01
4.80476618e-02 -3.19789886e-01 2.99796343e-01 -3.62865001e-01
2.02354401e-01 -6.25497639e-01 2.86670357e-01 2.94721305e-01
-4.84666407e-01 4.20970023e-02 -1.87397420e-01 1.21262610e+00
6.19830340e-02 2.06704691e-01 6.14580810e-01 1.00077860e-01
-3.66827428e-01 7.76815176e-01 -6.19411729e-02 3.37956518e-01
5.34413159e-01 1.47935480e-01 -6.58621669e-01 -6.95567310e-01
-7.04083741e-01 2.17346042e-01 7.31458426e-01 7.27255464e-01
2.82721639e-01 -1.17671716e+00 -7.04419255e-01 3.48291308e-01
7.70481452e-02 -4.80036646e-01 5.46911716e-01 9.39217389e-01
1.16976224e-01 7.07088560e-02 -5.42388782e-02 -2.56446600e-01
-8.70324910e-01 8.22500169e-01 -2.43565515e-02 -2.74057627e-01
-5.11642754e-01 8.11148107e-01 5.11961341e-01 -2.24320039e-01
2.41155863e-01 -1.95193946e-01 -5.19606620e-02 3.75359654e-02
5.07421792e-01 4.32732671e-01 -2.16396943e-01 -3.96247268e-01
-5.27851060e-02 1.89588636e-01 -5.19528985e-01 3.54219854e-01
1.26482975e+00 -2.53231943e-01 1.48448974e-01 4.23535258e-01
9.88799572e-01 5.79631627e-01 -1.28471112e+00 -6.18580818e-01
-3.41354758e-01 -1.07342243e+00 9.02712643e-02 -8.48185897e-01
-1.28387237e+00 6.90829396e-01 5.78109860e-01 2.52412140e-01
8.15747678e-01 -1.57153346e-02 8.39903235e-01 1.28513604e-01
5.53425133e-01 -5.55320144e-01 4.94695529e-02 3.09701353e-01
6.00350678e-01 -9.56690311e-01 3.27811867e-01 -1.87035695e-01
-5.67912340e-01 7.26432681e-01 4.00872111e-01 -3.97523344e-01
1.16341186e+00 -2.83270270e-01 -2.72300273e-01 -9.38970670e-02
-1.03648627e+00 1.27110884e-01 2.47426018e-01 2.75769591e-01
3.18645984e-01 2.90870994e-01 -1.57958493e-01 1.10462999e+00
-6.40997887e-02 -4.47487198e-02 3.93930078e-01 5.64013422e-01
-3.65208238e-01 -1.18566382e+00 1.24667436e-02 7.67075896e-01
-3.98517609e-01 -1.94909617e-01 -2.16195686e-03 5.85066617e-01
1.47733048e-01 7.24289477e-01 -1.68914441e-02 -5.22225082e-01
2.70743489e-01 -9.99256894e-02 3.55691969e-01 -6.70643330e-01
-2.91328222e-01 -4.41914499e-02 -2.34318271e-01 -6.39232039e-01
1.74220800e-02 -8.17177832e-01 -6.52023196e-01 -6.60657585e-01
-6.30146027e-01 4.50079292e-01 5.66806495e-02 5.33175707e-01
7.21718550e-01 1.03538610e-01 8.53689432e-01 -7.03224599e-01
-1.33108389e+00 -7.45807290e-01 -9.87975001e-01 8.53220999e-01
3.02335024e-01 -1.10271358e+00 -7.33428836e-01 -4.20858562e-01] | [9.89201831817627, 5.501735687255859] |
711eda0d-349f-474c-8a8b-124c6546a496 | rarely-a-problem-language-models-exhibit | 2212.08700 | null | https://arxiv.org/abs/2212.08700v2 | https://arxiv.org/pdf/2212.08700v2.pdf | Rarely a problem? Language models exhibit inverse scaling in their predictions following few-type quantifiers | How well do language models deal with quantification? In this study, we focus on 'few'-type quantifiers, as in 'few children like toys', which might pose a particular challenge for language models because the sentence components with out the quantifier are likely to co-occur, and 'few'-type quantifiers are rare. We present 960 English sentence stimuli from two human neurolinguistic experiments to 22 autoregressive transformer models of differing sizes. Not only do all the models perform poorly on 'few'-type quantifiers, but overall the larger the model, the worse its performance. This inverse scaling is consistent with previous work suggesting that larger models increasingly reflect online rather than offline human processing, and we argue that the decreasing performance of larger models may challenge uses of language models as the basis for natural language systems. | ['Benjamin K. Bergen', 'James A. Michaelov'] | 2022-12-16 | null | null | null | null | ['type'] | ['speech'] | [-1.98843837e-01 2.70057350e-01 1.95307970e-01 -2.97053248e-01
-6.35058403e-01 -5.99649608e-01 3.88380289e-01 4.72005516e-01
-7.37691164e-01 4.85816509e-01 5.63009977e-01 -6.88998044e-01
-1.90249421e-02 -9.27944958e-01 -4.82664227e-01 -1.57911509e-01
2.84194686e-02 6.40634358e-01 2.75309265e-01 -6.97374225e-01
1.17618114e-01 1.78021029e-01 -1.15742850e+00 4.93413448e-01
6.69384003e-01 4.42268044e-01 5.93884706e-01 6.15496278e-01
-2.39105016e-01 1.39399993e+00 -8.27552319e-01 -8.58921885e-01
-1.10960163e-01 -1.47620857e-01 -8.11279118e-01 -4.97988701e-01
5.77681720e-01 -3.89609426e-01 -2.43633002e-01 9.88610685e-01
2.22494081e-01 2.19081879e-01 7.87071347e-01 -5.78055799e-01
-1.08123732e+00 1.29020584e+00 -3.34032804e-01 8.20992708e-01
4.81013745e-01 3.95861179e-01 1.47630203e+00 -6.76576674e-01
6.28600597e-01 1.72991300e+00 8.68969619e-01 5.60533702e-01
-1.41254914e+00 -4.63712007e-01 3.11622679e-01 -1.41822025e-01
-1.29801774e+00 -4.55478251e-01 3.83357406e-01 -4.94268209e-01
1.62843788e+00 2.05979377e-01 8.43446434e-01 9.04455721e-01
4.49267834e-01 6.60036445e-01 1.34906423e+00 -2.60245711e-01
-1.56399190e-01 1.33058891e-01 3.21315706e-01 5.85210562e-01
2.86318451e-01 1.47959381e-01 -5.71181715e-01 -5.23467585e-02
7.73706675e-01 -3.04706603e-01 1.30871445e-01 3.78072798e-01
-1.02812433e+00 8.75652432e-01 1.21990740e-01 6.63457513e-01
-5.84666729e-01 1.73555076e-01 4.33840483e-01 4.49108750e-01
4.45086390e-01 6.43314540e-01 -6.58601284e-01 -4.20443982e-01
-8.72145355e-01 6.52300298e-01 6.66962326e-01 7.78790653e-01
4.47516859e-01 1.65911689e-01 -1.67724490e-01 1.04696870e+00
3.13398212e-01 4.87096578e-01 4.25543308e-01 -9.53915060e-01
8.05734158e-01 3.74826372e-01 9.86736268e-02 -8.92407238e-01
-7.23414600e-01 -1.42996475e-01 -1.47388667e-01 -2.04228789e-01
8.47724617e-01 1.46637306e-01 -4.06791866e-01 2.22672582e+00
-1.25108600e-01 -7.89049327e-01 -9.41573903e-02 7.77370751e-01
6.60521090e-01 6.42833829e-01 9.36121523e-01 -2.98327267e-01
1.49290395e+00 -3.62928420e-01 -6.17942810e-01 -7.23325968e-01
8.27627301e-01 -7.66205728e-01 1.57867301e+00 3.61490756e-01
-1.51830375e+00 -4.17680144e-01 -7.56815314e-01 -5.91599882e-01
-3.89857918e-01 -3.04866880e-01 1.22713053e+00 7.80361950e-01
-1.13327992e+00 6.62858903e-01 -8.09707284e-01 -6.93786368e-02
4.76379972e-03 1.53403142e-02 -2.18536332e-01 2.13161781e-01
-1.46989262e+00 1.62604177e+00 1.87032491e-01 2.96229473e-03
-7.67799377e-01 -6.00479245e-01 -9.85723078e-01 2.45769814e-01
1.81331430e-02 -4.08508569e-01 1.68853474e+00 -7.62477398e-01
-1.17405462e+00 1.14271259e+00 -3.35078686e-01 -4.10021216e-01
5.46538644e-02 -1.03797801e-01 -3.50744486e-01 2.48131417e-02
2.34178752e-01 2.70735741e-01 4.28036690e-01 -6.14797890e-01
-5.37223518e-01 -4.03798908e-01 4.37919199e-01 2.82835335e-01
-7.65281916e-02 8.42610180e-01 5.63889503e-01 -4.57868338e-01
1.85628265e-01 -2.94055343e-01 -3.02227810e-02 -3.19812089e-01
2.90763795e-01 -6.66633785e-01 -2.00563550e-01 -6.88945830e-01
1.56108296e+00 -1.88752687e+00 -1.26330972e-01 -1.38362542e-01
5.20971298e-01 -1.45130962e-01 -5.91092445e-02 5.75822473e-01
-1.38505220e-01 3.36593837e-01 2.74970204e-01 -2.78937221e-01
4.64607000e-01 2.29433894e-01 -3.77521873e-01 3.86931598e-01
4.39266533e-01 9.29774940e-01 -1.04918361e+00 -6.20007038e-01
-5.17207272e-02 1.51413098e-01 -6.49820864e-01 -7.75095299e-02
-2.41216213e-01 -2.45826945e-01 -1.15112551e-01 4.85545695e-01
3.87978941e-01 -1.14289127e-01 1.18912078e-01 2.44268149e-01
-4.82491970e-01 1.19821060e+00 -8.02843690e-01 1.00811458e+00
-3.92318666e-01 6.07628167e-01 2.87753612e-01 -6.34441912e-01
4.45297003e-01 2.85309553e-01 -3.42853308e-01 -9.17524278e-01
2.50124276e-01 9.21910927e-02 9.07925665e-01 -4.28557634e-01
7.51473427e-01 -8.03084791e-01 -3.48199695e-01 4.11714226e-01
-6.00102693e-02 -7.49472678e-01 6.20279849e-01 4.50075209e-01
1.08459044e+00 -1.69975653e-01 4.01365042e-01 -6.32005870e-01
1.59686163e-01 -8.37971494e-02 7.08356977e-01 9.01306033e-01
-3.87339413e-01 2.02411518e-01 8.26804280e-01 -2.41979524e-01
-1.05987453e+00 -1.30400836e+00 -2.64652908e-01 1.66049933e+00
-4.14732993e-01 -5.15689731e-01 -7.60493040e-01 1.50258303e-01
-1.40678257e-01 1.23576486e+00 -3.99852246e-01 -1.95627436e-01
-5.23459017e-01 -5.16546071e-01 6.61453247e-01 7.43971348e-01
-1.57699674e-01 -1.32656169e+00 -8.92039955e-01 6.07727349e-01
-2.30805516e-01 -9.41347897e-01 -4.20206577e-01 1.46366596e-01
-8.01682591e-01 -5.75691044e-01 -1.52319849e-01 -4.47000295e-01
7.10286200e-02 -1.20133661e-01 1.67493606e+00 3.70695084e-01
-3.97499315e-02 3.46066117e-01 -2.29206622e-01 -8.40617120e-01
-7.42900252e-01 -3.05719882e-01 1.39043286e-01 -8.96601498e-01
7.95372128e-01 -4.79154617e-01 -2.52663732e-01 -4.23198372e-01
-5.49318492e-01 -1.99721992e-01 3.18889081e-01 9.19801950e-01
5.90098314e-02 5.71271731e-03 1.37852117e-01 -8.48120391e-01
1.10471141e+00 -6.08251572e-01 -6.18877053e-01 -3.45566683e-02
-3.06200624e-01 -1.99450374e-01 9.39911246e-01 -8.24924946e-01
-8.68109941e-01 -7.49994040e-01 -2.12443009e-01 -5.84731027e-02
1.41169026e-01 6.21718943e-01 6.64573014e-02 5.13832808e-01
7.54000127e-01 -1.63176507e-02 -1.10141069e-01 -2.26327643e-01
6.81217089e-02 1.68513134e-01 1.79372549e-01 -1.03883743e+00
5.31694770e-01 3.69701460e-02 -2.02126041e-01 -9.40016270e-01
-8.04590642e-01 -1.76493078e-01 -1.24616578e-01 -8.88874531e-02
8.82522762e-01 -9.29004550e-01 -1.04073167e+00 2.92150468e-01
-1.39764547e+00 -6.07843578e-01 -2.00831473e-01 5.28676271e-01
-7.13954091e-01 2.17896044e-01 -1.29462373e+00 -1.38135338e+00
-3.23054314e-01 -1.11265218e+00 8.85291576e-01 2.21212476e-01
-8.94862711e-01 -1.02668822e+00 -1.78931266e-01 1.75173044e-01
4.83415186e-01 -3.64271462e-01 1.36056340e+00 -6.78489208e-01
-3.26606512e-01 6.38992563e-02 -2.02190146e-01 3.88175339e-01
-2.77607471e-01 1.62227541e-01 -7.49614835e-01 2.31000826e-01
4.07062978e-01 -4.85295922e-01 3.67166370e-01 2.62689501e-01
5.61718643e-01 -3.10441941e-01 1.70535997e-01 2.54534464e-02
1.20128310e+00 3.49534415e-02 5.00671744e-01 7.17010768e-03
4.15761173e-01 6.62665963e-01 6.06958747e-01 1.99854419e-01
1.10387182e+00 2.28981465e-01 -1.84603393e-01 3.56685191e-01
5.93486577e-02 -6.29127204e-01 5.97038090e-01 1.22510552e+00
7.00715408e-02 -2.40200125e-02 -1.24047506e+00 8.31314266e-01
-1.29895222e+00 -1.21799839e+00 -2.87054598e-01 1.94820523e+00
1.27287078e+00 5.45544684e-01 3.70748453e-02 -3.33739556e-02
3.95122766e-01 2.94523180e-01 7.26322271e-03 -1.10974503e+00
-1.32244080e-01 7.73895442e-01 2.07785100e-01 7.22046733e-01
-5.91836691e-01 1.37137747e+00 7.76295900e+00 6.64952159e-01
-8.62657785e-01 3.46057713e-01 3.55283260e-01 -4.25603129e-02
-5.01228213e-01 8.43925923e-02 -6.75054848e-01 5.27696669e-01
1.50932062e+00 -2.04565883e-01 6.61843181e-01 6.22313201e-01
3.85913700e-01 -4.86441761e-01 -1.36294210e+00 5.68354309e-01
-2.55184919e-01 -7.64042556e-01 -5.82835041e-02 -1.81938916e-01
1.54257134e-01 2.01711562e-02 4.31767516e-02 5.90823293e-01
5.78235507e-01 -1.00740504e+00 1.35011005e+00 6.28073514e-01
4.35443819e-01 -3.99412721e-01 2.64090538e-01 7.24096298e-01
-9.63720322e-01 -1.95288181e-01 -6.58218026e-01 -9.59555566e-01
3.26975822e-01 3.53221089e-01 -4.96544212e-01 -4.31936234e-01
6.15845561e-01 2.79642944e-03 -4.91806895e-01 4.75979805e-01
-4.62896645e-01 8.14356744e-01 -6.24559224e-01 -6.23041332e-01
1.41228870e-01 -2.88303822e-01 4.82579947e-01 1.29462266e+00
-1.24201670e-01 6.00527048e-01 -1.63125753e-01 1.11983454e+00
2.21127033e-01 2.87814230e-01 -5.70332944e-01 -4.57617193e-01
7.25349903e-01 9.65274870e-01 -7.37389803e-01 -5.57077467e-01
-6.68384314e-01 4.56224144e-01 8.58769298e-01 1.10494971e-01
-7.55304694e-01 5.87302893e-02 6.38866127e-01 2.59584814e-01
-1.83430836e-01 -5.36403477e-01 -4.18834955e-01 -1.13593423e+00
-5.52900042e-03 -1.01814532e+00 2.77911246e-01 -1.04478419e+00
-1.41099715e+00 2.20235437e-01 1.43081322e-01 -2.48947009e-01
-4.61500734e-01 -9.05804634e-01 -4.48093832e-01 1.12549651e+00
-8.47803831e-01 -9.90039885e-01 5.92476249e-01 2.11093858e-01
4.16374117e-01 4.28222656e-01 9.21608448e-01 1.46356329e-01
-7.89525062e-02 5.68603158e-01 -4.01881039e-01 6.71893731e-02
3.32852572e-01 -1.41076195e+00 5.59661567e-01 8.57086718e-01
1.56545386e-01 1.43383563e+00 9.73540902e-01 -7.23519921e-01
-1.09011245e+00 -2.62192726e-01 1.43448520e+00 -1.15205204e+00
1.29765844e+00 -4.75582629e-01 -9.72868264e-01 8.90951157e-01
1.03040390e-01 -5.30270338e-01 4.51168329e-01 5.80043137e-01
-6.12155855e-01 1.63043752e-01 -1.14951658e+00 8.12893569e-01
1.19134545e+00 -1.02536070e+00 -1.17544806e+00 4.41233367e-01
8.36783707e-01 -4.30774927e-01 -1.08062649e+00 -1.60460640e-02
5.67674577e-01 -8.78782988e-01 7.39509523e-01 -7.30260015e-01
6.44474328e-01 8.46497342e-03 -1.15883142e-01 -1.13756943e+00
-6.88816607e-01 -3.64153713e-01 2.04680897e-02 1.12986577e+00
6.61846101e-01 -7.23999381e-01 1.89340711e-01 1.17888308e+00
-3.25923413e-02 -5.22404552e-01 -1.09639263e+00 -5.85768402e-01
8.59412670e-01 -7.69621730e-01 5.52191556e-01 7.32288659e-01
2.93119282e-01 4.66443241e-01 2.64088124e-01 -2.16154605e-02
3.55001390e-01 -3.24024826e-01 8.25669095e-02 -1.00995934e+00
-4.73906338e-01 -6.89441502e-01 -4.16924059e-01 -8.42603981e-01
2.87554532e-01 -7.13397384e-01 8.22784677e-02 -1.19100010e+00
2.95278162e-01 -4.48189050e-01 1.68484524e-01 3.94391060e-01
-3.87290150e-01 -2.45356057e-02 3.56360257e-01 -2.87721545e-01
-2.95538157e-01 2.38852337e-01 9.78912890e-01 -4.39262809e-03
1.89356908e-01 -2.67644703e-01 -1.05298269e+00 1.00944138e+00
4.95562345e-01 -3.61659586e-01 -4.22674835e-01 -7.92557240e-01
8.38539243e-01 1.30783468e-01 2.10205197e-01 -6.43383324e-01
-2.38061156e-02 -3.70387226e-01 3.84520710e-01 -2.37946033e-01
3.55472505e-01 -5.79462588e-01 -9.86964926e-02 2.32432708e-01
-2.03755274e-01 7.48142421e-01 4.43967223e-01 -1.10654183e-01
1.69360891e-01 -4.97810990e-01 8.54546249e-01 -6.36366189e-01
-4.07785207e-01 -2.53680408e-01 -8.97980928e-01 4.46684450e-01
3.53395075e-01 -1.28988087e-01 -3.29125315e-01 -2.89304227e-01
-7.90874898e-01 2.55251769e-02 3.35533231e-01 3.03927094e-01
4.00325626e-01 -7.94517696e-01 -6.04222119e-01 -9.81146768e-02
-1.44459799e-01 -2.20522508e-01 9.03549343e-02 7.28447855e-01
-5.95572710e-01 3.71103913e-01 2.04942882e-01 -8.21206123e-02
-7.68686831e-01 5.66545486e-01 3.72658044e-01 -2.29471505e-01
-4.42053139e-01 1.23283434e+00 2.44277239e-01 -3.96168202e-01
-2.22650662e-01 -7.73646295e-01 -9.46439058e-02 2.84707189e-01
6.35113001e-01 2.07716912e-01 -1.72687978e-01 -8.31687570e-01
-3.62345129e-01 -1.39961380e-03 -3.88838023e-01 -2.88116664e-01
1.21597993e+00 -1.74912751e-01 -6.08175337e-01 1.13328767e+00
7.26245642e-01 3.10002744e-01 -5.38550436e-01 1.16295725e-01
-6.03885576e-02 -2.19710946e-01 -2.56994367e-01 -6.58499718e-01
-3.61354411e-01 8.45592380e-01 -3.49971233e-03 4.39157635e-01
7.24278271e-01 3.11701417e-01 3.67079914e-01 1.15700431e-01
7.58790672e-01 -1.47467637e+00 -4.86880362e-01 9.52980876e-01
8.02847266e-01 -6.78196609e-01 -9.46305972e-03 -2.40926713e-01
-5.13587356e-01 7.65500784e-01 6.51330948e-01 -2.36593857e-01
2.42468178e-01 4.35173184e-01 -8.33033323e-02 -3.40137571e-01
-1.29620016e+00 -2.64181256e-01 -2.54928112e-01 4.78328377e-01
1.05592036e+00 4.38195616e-01 -8.54622006e-01 9.63264167e-01
-1.12592161e+00 -2.72602260e-01 7.10087836e-01 5.56779683e-01
-4.12707925e-01 -7.69968748e-01 -3.93254369e-01 8.00774992e-01
-7.12034941e-01 -9.08505976e-01 -1.87649518e-01 8.29655647e-01
9.00625959e-02 1.10173142e+00 2.96409160e-01 1.53181612e-01
3.79047722e-01 1.35439143e-01 8.89112413e-01 -9.75315750e-01
-8.96773934e-01 -2.99566567e-01 3.49350274e-01 -6.08135104e-01
2.87995487e-02 -1.17274821e+00 -1.34639323e+00 -5.86790323e-01
-3.73809844e-01 -6.66785538e-02 2.63939619e-01 1.00727952e+00
-3.79163980e-01 3.09395790e-01 -6.21005222e-02 -3.67487252e-01
-1.06758320e+00 -1.39044249e+00 -6.32408977e-01 4.04162824e-01
1.36997327e-01 -6.75189197e-01 -5.87294281e-01 -1.58764526e-01] | [10.52440357208252, 9.08178424835205] |
7e273516-a248-452b-83a3-3ef055a78225 | paddleseg-a-high-efficient-development | 2101.06175 | null | https://arxiv.org/abs/2101.06175v1 | https://arxiv.org/pdf/2101.06175v1.pdf | PaddleSeg: A High-Efficient Development Toolkit for Image Segmentation | Image Segmentation plays an essential role in computer vision and image processing with various applications from medical diagnosis to autonomous car driving. A lot of segmentation algorithms have been proposed for addressing specific problems. In recent years, the success of deep learning techniques has tremendously influenced a wide range of computer vision areas, and the modern approaches of image segmentation based on deep learning are becoming prevalent. In this article, we introduce a high-efficient development toolkit for image segmentation, named PaddleSeg. The toolkit aims to help both developers and researchers in the whole process of designing segmentation models, training models, optimizing performance and inference speed, and deploying models. Currently, PaddleSeg supports around 20 popular segmentation models and more than 50 pre-trained models from real-time and high-accuracy levels. With modular components and backbone networks, users can easily build over one hundred models for different requirements. Furthermore, we provide comprehensive benchmarks and evaluations to show that these segmentation algorithms trained on our toolkit have more competitive accuracy. Also, we provide various real industrial applications and practical cases based on PaddleSeg. All codes and examples of PaddleSeg are available at https://github.com/PaddlePaddle/PaddleSeg. | ['Yuying Hao', 'Baohua Lai', 'Zeyu Chen', 'Zewu Wu', 'Guowei Chen', 'Lutao Chu', 'Yi Liu'] | 2021-01-15 | null | null | null | null | ['human-part-segmentation'] | ['computer-vision'] | [-2.02843934e-01 -2.08201885e-01 -3.00521165e-01 -4.52399701e-01
-5.92787981e-01 -2.63898373e-01 -5.62904775e-02 -1.23597980e-01
-3.23720694e-01 8.06083679e-02 -5.49099386e-01 -5.54742694e-01
3.11721057e-01 -8.11983705e-01 -4.38339025e-01 -6.05031848e-01
7.29706809e-02 4.96078849e-01 6.52606845e-01 1.17335897e-02
2.24859029e-01 4.73136604e-01 -1.37161088e+00 1.41910929e-02
1.00698102e+00 9.91652310e-01 4.66798335e-01 6.83571696e-01
-3.78815144e-01 3.03256124e-01 -6.59958005e-01 -2.24598184e-01
3.83975118e-01 -2.36109197e-01 -7.28481770e-01 1.47987977e-01
3.33509669e-02 -2.86579341e-01 -8.98801312e-02 1.06407738e+00
4.52893674e-01 -2.44323984e-01 3.24159831e-01 -1.35042965e+00
-4.54647392e-01 5.65154791e-01 -8.63075256e-01 7.95375109e-02
-2.42711723e-01 2.62439102e-01 5.18505633e-01 -6.11318171e-01
1.63206682e-01 1.00124288e+00 7.44493783e-01 5.61963975e-01
-5.63789248e-01 -8.62053096e-01 1.81824580e-01 2.48460159e-01
-1.11989689e+00 -1.76962882e-01 5.67076921e-01 -5.05376399e-01
7.49115109e-01 1.36981264e-01 6.87258184e-01 4.79021579e-01
3.50179404e-01 1.18601489e+00 7.50047922e-01 -2.47788355e-01
1.38525620e-01 2.33569294e-02 4.86881673e-01 8.05695176e-01
1.06145427e-01 -1.82524532e-01 1.89534932e-01 2.41361678e-01
1.02449858e+00 1.23537414e-01 -1.57119299e-03 -2.03800693e-01
-1.14468467e+00 7.89988101e-01 4.43025887e-01 2.94042617e-01
-1.34606838e-01 1.80434361e-01 5.22246838e-01 -1.01746306e-01
2.67402112e-01 1.75239787e-01 -5.86466610e-01 -7.31793195e-02
-8.16155076e-01 2.01172560e-01 7.10259438e-01 1.17153549e+00
9.11922991e-01 4.23495695e-02 6.41519725e-02 1.10184288e+00
2.45516002e-01 3.83222878e-01 6.91626787e-01 -1.13600862e+00
2.48170152e-01 6.53422773e-01 -2.57732660e-01 -8.99421573e-01
-5.29922962e-01 -2.98523694e-01 -9.17763412e-01 1.80733278e-01
2.30467826e-01 -1.97423980e-01 -1.27041125e+00 1.12551522e+00
5.09105384e-01 3.13817471e-01 -3.11841816e-01 8.66018176e-01
9.65866506e-01 8.63964856e-01 4.88786660e-02 1.55309513e-01
1.44598949e+00 -1.47579694e+00 -4.44228351e-01 -4.66711462e-01
5.68775952e-01 -8.52687299e-01 1.24745262e+00 6.21823192e-01
-9.72925961e-01 -8.49330366e-01 -1.05218136e+00 -1.59939125e-01
-3.38116646e-01 2.78237641e-01 8.24827969e-01 6.83645487e-01
-9.57754016e-01 4.61089432e-01 -1.04960966e+00 -2.54881382e-01
7.19500303e-01 4.23550159e-01 9.98556092e-02 -9.16909724e-02
-8.60348344e-01 4.27745551e-01 4.02784169e-01 1.80119112e-01
-7.45918810e-01 -4.02221829e-01 -7.38201559e-01 -2.30537519e-01
4.65522200e-01 -5.59513986e-01 1.61627674e+00 -6.82379544e-01
-1.45803428e+00 8.19262862e-01 5.98913841e-02 -4.75399554e-01
4.25978810e-01 -3.85346055e-01 -3.74631792e-01 -2.99659576e-02
2.00789198e-01 8.70910168e-01 5.76744497e-01 -9.88505781e-01
-7.48757362e-01 -2.45812729e-01 -2.00407188e-02 1.77063588e-02
-2.69725949e-01 9.99629349e-02 -1.17408419e+00 -4.51385230e-01
-6.54018447e-02 -7.64344692e-01 -6.54884458e-01 -1.68909300e-02
-5.64196348e-01 -3.14834148e-01 1.19602704e+00 -5.82711995e-01
1.13550854e+00 -2.18300414e+00 -3.32384467e-01 -1.52018825e-02
2.35164985e-01 6.43405855e-01 7.99389381e-04 4.04059179e-02
2.35589251e-01 1.51175454e-01 -5.88157237e-01 -1.37630001e-01
-6.47760257e-02 3.00507218e-01 -4.13921252e-02 2.38172457e-01
2.70648077e-02 9.76288319e-01 -6.32640660e-01 -6.90006912e-01
6.52807713e-01 2.87934273e-01 -4.43471581e-01 2.23500848e-01
-3.01235795e-01 4.23196316e-01 -6.46178305e-01 8.40335131e-01
8.20241809e-01 -3.87830168e-01 -6.78809956e-02 -2.30867453e-02
-1.24364018e-01 -1.34112118e-02 -1.05094731e+00 1.67064428e+00
-3.50119919e-01 4.39410239e-01 2.04964742e-01 -1.29197383e+00
9.97605503e-01 -2.44513024e-02 4.71212745e-01 -8.11066031e-01
4.73557860e-01 2.23714709e-01 -2.92322114e-02 -5.89134395e-01
3.84359032e-01 3.74098033e-01 -1.30284518e-01 4.15914118e-01
-7.93149546e-02 -2.81330794e-01 5.04152179e-01 -3.45953517e-02
6.39858365e-01 1.19690716e-01 2.67778486e-01 -1.58368796e-01
5.55474997e-01 2.78747112e-01 8.62645090e-01 4.88244176e-01
-4.72170264e-01 6.32864833e-01 3.43028784e-01 -5.34358919e-01
-9.45888221e-01 -8.12531233e-01 -2.65032947e-01 9.94585216e-01
5.77865362e-01 -2.22017020e-01 -1.29973519e+00 -4.78205115e-01
-3.09950292e-01 4.08704609e-01 -2.11329415e-01 1.06890336e-01
-6.57717943e-01 -8.13068151e-01 5.39229631e-01 6.95926428e-01
1.12507081e+00 -1.27820158e+00 -7.79960454e-01 2.79200524e-01
-9.01170895e-02 -1.32343352e+00 -4.77411747e-01 -1.88587725e-01
-1.03980577e+00 -1.09196770e+00 -8.23628604e-01 -1.27570105e+00
5.39683342e-01 4.49507475e-01 1.20844638e+00 3.34928066e-01
-6.86541796e-01 5.75315803e-02 -2.58837879e-01 -5.68823457e-01
-3.05731058e-01 3.86992097e-01 -4.12713200e-01 -4.56824064e-01
3.46138597e-01 -3.55210811e-01 -7.61618733e-01 5.87181211e-01
-8.65839124e-01 4.59436446e-01 7.05920756e-01 3.61950248e-01
7.99347281e-01 3.84282395e-02 5.39888561e-01 -1.08926737e+00
5.08575618e-01 -3.09313387e-01 -8.72181177e-01 1.71007589e-01
-6.95984602e-01 -3.62324715e-01 5.11067867e-01 -2.12455198e-01
-8.12278450e-01 1.16403393e-01 -8.17820966e-01 -2.64801592e-01
-5.48190355e-01 4.03842151e-01 -3.60385239e-01 1.94753662e-01
3.89137149e-01 1.12520821e-01 4.93964218e-02 -5.92000723e-01
4.68307048e-01 8.82451892e-01 6.09404504e-01 -5.65752327e-01
3.22447032e-01 2.37732530e-01 -5.05884945e-01 -7.54839897e-01
-6.42245233e-01 -4.41515237e-01 -4.66503352e-01 -2.84146011e-01
1.06851232e+00 -7.00542808e-01 -4.06526476e-01 9.98093665e-01
-1.01820326e+00 -7.31420219e-01 6.31679296e-02 1.40394375e-01
-3.66152048e-01 3.52469593e-01 -7.94902265e-01 -3.71147156e-01
-5.75463295e-01 -1.73691761e+00 1.03412521e+00 7.49954522e-01
1.74104929e-01 -9.74512756e-01 -2.43918315e-01 3.58402044e-01
3.69925797e-01 2.54606694e-01 6.83402598e-01 -4.49827164e-01
-7.03744352e-01 -1.97560713e-01 -3.27297717e-01 5.03583372e-01
1.12821423e-01 2.89483756e-01 -7.22275317e-01 -7.44794980e-02
-9.53615904e-02 -1.86862260e-01 7.09131777e-01 6.85812235e-01
1.77154636e+00 8.04182491e-04 -7.07644582e-01 8.11719596e-01
1.32880187e+00 5.94821334e-01 6.20982468e-01 3.63817632e-01
8.11417341e-01 4.11065578e-01 6.67973101e-01 8.24643075e-02
5.32731473e-01 4.35802609e-01 4.87957776e-01 -5.08174837e-01
8.90994594e-02 1.57738715e-01 5.51550090e-02 9.61788535e-01
1.30316943e-01 -7.42901936e-02 -1.23032391e+00 5.47704458e-01
-1.78036284e+00 -5.70742249e-01 -3.83535057e-01 1.79716861e+00
7.03243613e-01 3.54315609e-01 2.30814263e-01 4.66204472e-02
1.06231737e+00 -5.67861274e-02 -9.21362579e-01 -3.88497204e-01
3.75023007e-01 2.74017811e-01 4.00877953e-01 3.52188706e-01
-1.33260226e+00 1.14801931e+00 6.51774979e+00 9.78918254e-01
-1.40563095e+00 4.45399806e-02 1.06085503e+00 3.35834593e-01
1.77381530e-01 -2.35886827e-01 -9.20270205e-01 5.15464902e-01
6.92578375e-01 -5.02026714e-02 4.79894318e-02 1.23476875e+00
2.55649894e-01 -2.04084069e-01 -6.40308022e-01 9.96834099e-01
-2.17317030e-01 -1.47751379e+00 -2.24755064e-01 -5.34190275e-02
6.16465569e-01 3.24512750e-01 5.74925868e-03 3.14217597e-01
3.71688336e-01 -9.03035581e-01 4.58147198e-01 6.85855001e-02
7.48311102e-01 -8.26874077e-01 7.72463083e-01 5.45092881e-01
-1.25449443e+00 1.33526638e-01 -4.80974466e-01 4.48906347e-02
3.01511675e-01 7.81469226e-01 -6.53326333e-01 4.67986196e-01
8.62756610e-01 7.47010767e-01 -5.43475330e-01 1.26230824e+00
-2.73505867e-01 7.40031719e-01 -1.87974080e-01 8.25279132e-02
3.25934321e-01 -3.95788014e-01 -6.76565841e-02 1.28571737e+00
2.44176373e-01 5.54015040e-02 3.79079193e-01 8.02858233e-01
2.70898286e-02 7.03913793e-02 -2.09295675e-01 -1.06422007e-02
4.80097353e-01 1.61873269e+00 -1.34694886e+00 -5.43676972e-01
-4.72913951e-01 8.33830118e-01 -8.66720751e-02 1.55377507e-01
-1.29095435e+00 -6.61189020e-01 7.57533371e-01 -9.02454462e-03
8.74122903e-02 -4.14335102e-01 -5.82788467e-01 -7.81947315e-01
-2.04588637e-01 -9.07195449e-01 1.70516998e-01 -5.61089039e-01
-8.79743576e-01 7.83608437e-01 -7.64199644e-02 -1.02766347e+00
-2.58808560e-03 -7.39700139e-01 -9.01157558e-01 6.56466842e-01
-1.37634456e+00 -8.83659780e-01 -6.43336892e-01 5.17013907e-01
1.00274229e+00 -5.90124913e-02 4.90720063e-01 4.71842051e-01
-1.10697150e+00 4.58313018e-01 2.70513352e-02 6.13707006e-01
4.67856288e-01 -1.06884420e+00 1.01548362e+00 9.28023040e-01
5.90408444e-02 5.36454141e-01 3.93552750e-01 -3.14726412e-01
-1.06102383e+00 -1.34920275e+00 1.81162983e-01 1.19603202e-02
2.60034293e-01 -3.31160754e-01 -9.13185656e-01 6.37757719e-01
2.51328170e-01 9.21347365e-03 4.31430519e-01 -1.80899560e-01
8.62577409e-02 -1.79798558e-01 -1.07876849e+00 5.11238456e-01
8.15327942e-01 -5.66215058e-05 -1.96477413e-01 3.77519786e-01
7.93267965e-01 -7.94032872e-01 -6.23212993e-01 4.14069057e-01
3.99986714e-01 -1.27405655e+00 7.26722062e-01 -1.30720763e-02
4.14284736e-01 -3.69231552e-01 3.99093598e-01 -8.89351070e-01
-1.56168833e-01 -2.54567087e-01 6.41833395e-02 1.11163497e+00
2.47467905e-01 -7.70293593e-01 9.88447130e-01 5.13708949e-01
-6.01137161e-01 -1.23181713e+00 -3.49236727e-01 -5.77556312e-01
2.29789600e-01 -7.39060819e-01 7.65584171e-01 6.55845582e-01
-4.98187006e-01 1.62467405e-01 1.55151589e-02 1.08632348e-01
4.78975207e-01 3.13983023e-01 9.76274788e-01 -9.83711839e-01
-2.34513804e-01 -6.41192079e-01 -3.35060120e-01 -1.67379999e+00
-1.89470902e-01 -6.31228328e-01 2.23992169e-01 -2.03264093e+00
3.68407294e-02 -7.52611399e-01 -5.26665635e-02 6.75439477e-01
-1.52605385e-01 4.04792428e-01 9.59480703e-02 2.46623471e-01
-5.71538031e-01 1.46924138e-01 1.50659251e+00 -2.07685143e-01
-1.14289522e-01 2.28652596e-01 -9.16881919e-01 1.02654958e+00
1.23163640e+00 -1.99627042e-01 -4.49844211e-01 -7.97872961e-01
-3.07612360e-01 -1.97660118e-01 2.10490376e-01 -1.23363006e+00
2.53318459e-01 -2.02941045e-01 2.29395986e-01 -7.53170133e-01
-1.70126297e-02 -5.64122438e-01 -3.77492085e-02 6.12682760e-01
1.80867910e-01 1.48195148e-01 3.49330068e-01 3.28059830e-02
-3.97012472e-01 -2.48788610e-01 1.00391161e+00 -2.70481288e-01
-1.32635963e+00 7.04995036e-01 -1.93582892e-01 3.86896580e-02
1.30421197e+00 -4.23545659e-01 -1.41549766e-01 -1.97107159e-02
-6.06126845e-01 4.87525672e-01 3.88805360e-01 4.55646873e-01
6.26583934e-01 -9.98495817e-01 -4.46570992e-01 4.01404411e-01
-8.32628161e-02 6.50396049e-01 2.82484442e-01 7.75130868e-01
-1.10125506e+00 3.20107698e-01 -2.03534693e-01 -9.01617289e-01
-1.14058483e+00 3.92480373e-01 4.75737214e-01 -1.41113967e-01
-7.84552217e-01 9.76503313e-01 4.84279037e-01 -4.77972925e-01
1.26131117e-01 -6.21255755e-01 -6.65500015e-02 -4.77337122e-01
6.23462081e-01 4.53150600e-01 1.35558039e-01 -3.24554473e-01
-2.61795729e-01 6.90783799e-01 -2.39529416e-01 2.74757922e-01
1.10328567e+00 -5.06296083e-02 -1.15894318e-01 2.65982032e-01
9.45530295e-01 -3.98482502e-01 -1.40577805e+00 1.75785437e-01
-1.02430567e-01 -2.13571414e-01 -1.83758195e-02 -6.36251986e-01
-1.59684217e+00 1.05763662e+00 6.80192411e-01 2.64017701e-01
1.35390854e+00 -2.31621303e-02 1.36516225e+00 1.01290755e-01
6.66873991e-01 -1.08135581e+00 -9.13829580e-02 6.04005158e-01
4.39499646e-01 -1.17992198e+00 -1.08091004e-01 -6.38290167e-01
-5.96163750e-01 1.04016054e+00 9.30383205e-01 -3.32620829e-01
6.82388425e-01 6.21196568e-01 5.16312182e-01 -8.49808306e-02
-1.62142739e-01 -1.88306142e-02 -4.44018021e-02 6.13660336e-01
5.50967932e-01 1.21764168e-01 -2.88947105e-01 6.94780469e-01
-2.58904427e-01 -1.34123245e-03 2.91208148e-01 7.58401811e-01
-7.14445770e-01 -1.32442296e+00 -3.75418872e-01 5.10998726e-01
-4.08091605e-01 8.88446122e-02 -9.54751484e-03 9.32475328e-01
4.00232613e-01 8.24613690e-01 1.47650480e-01 -3.94136131e-01
1.44334957e-01 -3.34229767e-01 2.68073827e-01 -6.68810606e-01
-3.28676313e-01 1.85603827e-01 -3.02773684e-01 -6.42348886e-01
-2.33217031e-01 -3.77965659e-01 -1.69902122e+00 -3.97731543e-01
-3.43697876e-01 1.85848117e-01 7.21368134e-01 9.30173337e-01
3.61898631e-01 6.93692029e-01 3.92902315e-01 -9.36854541e-01
-1.39964074e-01 -8.11544418e-01 -5.45881331e-01 -1.35171533e-01
-1.44362658e-01 -3.77800345e-01 1.77386865e-01 2.65770018e-01] | [9.523313522338867, 0.0024700062349438667] |
399e0261-f8ce-4217-b8fd-3a3f89d5881e | predicting-team-performance-with-spatial | 2206.10720 | null | https://arxiv.org/abs/2206.10720v1 | https://arxiv.org/pdf/2206.10720v1.pdf | Predicting Team Performance with Spatial Temporal Graph Convolutional Networks | This paper presents a new approach for predicting team performance from the behavioral traces of a set of agents. This spatiotemporal forecasting problem is very relevant to sports analytics challenges such as coaching and opponent modeling. We demonstrate that our proposed model, Spatial Temporal Graph Convolutional Networks (ST-GCN), outperforms other classification techniques at predicting game score from a short segment of player movement and game features. Our proposed architecture uses a graph convolutional network to capture the spatial relationships between team members and Gated Recurrent Units to analyze dynamic motion information. An ablative evaluation was performed to demonstrate the contributions of different aspects of our architecture. | ['Gita Sukthankar', 'Shengnan Hu'] | 2022-06-21 | null | null | null | null | ['sports-analytics'] | ['computer-vision'] | [-3.09594512e-01 -3.43246162e-01 -3.00704181e-01 -2.92594619e-02
-2.12549657e-01 -5.38254857e-01 4.23182040e-01 3.68430950e-02
-5.13713479e-01 3.29169005e-01 4.65781957e-01 -2.20202729e-01
-3.97949040e-01 -1.07816172e+00 -4.41411614e-01 -2.64778048e-01
-8.76406908e-01 2.97363043e-01 6.72161996e-01 -9.64656293e-01
5.61825275e-01 5.07536113e-01 -1.43532419e+00 5.60139894e-01
2.30636373e-01 8.50799561e-01 -1.97810411e-01 1.28148234e+00
1.05479598e-01 1.98427725e+00 -8.19640279e-01 -4.04139161e-01
3.17080200e-01 -7.38954902e-01 -8.01666021e-01 -2.84177631e-01
1.08480431e-01 -7.71205798e-02 -1.11998606e+00 4.75357920e-01
3.96946400e-01 8.29579413e-01 3.00912917e-01 -1.31442153e+00
-2.59160697e-01 7.90884793e-01 -2.87511289e-01 1.17893803e+00
5.12252688e-01 4.07223791e-01 1.02951717e+00 -1.08739540e-01
8.32005620e-01 8.50013852e-01 1.14269543e+00 3.02429080e-01
-4.57083195e-01 -5.03470302e-01 3.68862540e-01 7.86783159e-01
-9.68411803e-01 8.06177706e-02 8.72782111e-01 -5.88266790e-01
1.19065547e+00 1.01978280e-01 1.35052216e+00 1.20923257e+00
5.41287899e-01 9.01395202e-01 5.62382877e-01 1.32651284e-01
5.09253480e-02 -1.13961697e+00 3.33064646e-01 9.33460295e-01
-3.97845089e-01 3.75584722e-01 -1.33551073e+00 8.74288753e-02
8.84276450e-01 1.27199262e-01 2.53670782e-01 6.74466789e-02
-8.39827716e-01 9.54902172e-01 8.40674996e-01 2.89295882e-01
-3.93932462e-01 9.17671680e-01 7.00690329e-01 5.19878507e-01
6.82106137e-01 3.53066355e-01 2.82801777e-01 -1.06265855e+00
-9.20990109e-01 6.95134640e-01 5.37248552e-01 4.33404952e-01
2.33910009e-01 6.60502672e-01 -3.92637223e-01 3.27396005e-01
-1.82476014e-01 -7.02958405e-02 6.53278172e-01 -8.13567996e-01
5.27025402e-01 7.53809094e-01 -1.73063904e-01 -1.53334987e+00
-9.80276227e-01 -6.88053668e-01 -3.81721944e-01 1.47272199e-01
4.70568925e-01 -2.74909109e-01 -7.34109104e-01 1.50557816e+00
-9.63249728e-02 1.00050557e+00 -1.17902741e-01 1.16992092e+00
1.37919879e+00 5.51675975e-01 8.57876688e-02 3.33478272e-01
8.07502627e-01 -1.10349417e+00 -5.45821548e-01 -2.94769078e-01
9.63189125e-01 -2.70744637e-02 6.97053552e-01 -7.86162913e-03
-1.16243994e+00 -8.51625025e-01 -8.35166156e-01 2.64313549e-01
-2.94716150e-01 -2.20832393e-01 9.83536959e-01 2.84384608e-01
-1.11506772e+00 9.81240451e-01 -1.18608713e+00 -3.44266862e-01
3.14408094e-01 4.38650757e-01 -1.06317841e-01 5.35498083e-01
-1.25684118e+00 7.17477620e-01 1.90140024e-01 4.20456737e-01
-1.14980030e+00 -2.61466950e-01 -1.04064631e+00 4.98935990e-02
3.28666687e-01 -3.00761789e-01 1.10030770e+00 -5.51144660e-01
-1.60885596e+00 7.00757027e-01 2.82564074e-01 -9.24487889e-01
6.22619867e-01 -2.09455118e-01 -3.66611689e-01 -1.98875383e-01
1.93596035e-01 -5.10736667e-02 3.00643981e-01 -2.92146921e-01
-7.83864141e-01 -3.55719805e-01 1.27625138e-01 5.74105561e-01
-2.63764746e-02 1.20113092e-02 -5.82569957e-01 -7.48616397e-01
-6.21022806e-02 -1.05315840e+00 -6.17563844e-01 -9.77116227e-01
-3.07830304e-01 -4.27103966e-01 4.78075236e-01 -5.75556159e-01
1.59456801e+00 -2.03584242e+00 4.77984190e-01 3.00569117e-01
4.89730775e-01 5.36162481e-02 -1.40462875e-01 8.43607605e-01
3.08729649e-01 -2.45922312e-01 4.26264018e-01 -8.17977637e-02
-1.60561115e-01 -1.94367412e-02 -1.68879807e-01 5.42953610e-01
-3.53149116e-01 1.53915560e+00 -8.19572151e-01 -9.35063511e-02
2.46816009e-01 3.67315337e-02 -3.64471197e-01 7.61616081e-02
-2.82638203e-02 3.78001004e-01 -6.68192923e-01 3.68902683e-01
-1.59149125e-01 -1.60081863e-01 3.62384260e-01 4.62176442e-01
-1.40057698e-01 3.02099586e-01 -8.71347845e-01 1.97678614e+00
1.09435894e-01 1.15553606e+00 -4.59073484e-01 -1.11081028e+00
1.06291306e+00 5.85041791e-02 8.54711115e-01 -8.54272783e-01
3.58377963e-01 -2.88401127e-01 2.94225425e-01 -5.02281249e-01
8.25147212e-01 2.99594223e-01 -6.89538062e-01 6.14568233e-01
1.94107309e-01 3.76947373e-01 3.60524982e-01 3.29850048e-01
1.71205986e+00 3.37411612e-01 -4.52211238e-02 9.90869030e-02
1.83749124e-02 6.70343101e-01 6.84157550e-01 1.05462050e+00
-6.03097856e-01 2.31483549e-01 6.82008564e-01 -1.37558532e+00
-6.57301545e-01 -7.23346353e-01 1.00319326e+00 1.64516377e+00
3.58154893e-01 -7.85495698e-01 -5.24512112e-01 -3.52747232e-01
-9.97545049e-02 -1.24344952e-01 -1.28608680e+00 -3.30893219e-01
-1.24199271e+00 -5.55884302e-01 8.35984766e-01 1.01889920e+00
5.19127071e-01 -1.39266181e+00 -9.33968067e-01 5.03258348e-01
-2.00696573e-01 -1.00048602e+00 -4.99562234e-01 8.68728105e-03
-6.42599463e-01 -1.37515008e+00 -1.36322081e-01 -8.68628681e-01
-3.97054553e-01 2.60995120e-01 1.34195971e+00 2.64212042e-01
-3.07734370e-01 2.36065969e-01 -5.83630979e-01 -4.01140153e-01
9.35311094e-02 4.48062003e-01 -5.68828098e-02 6.43769056e-02
6.06646121e-01 -7.19410896e-01 -5.48529983e-01 1.58518821e-01
-2.95187414e-01 9.10375044e-02 -4.08384427e-02 4.58076477e-01
1.82288080e-01 -1.79715469e-01 1.50647074e-01 -7.15073645e-01
1.21938908e+00 -5.14801323e-01 -3.48638952e-01 -8.22028145e-02
1.01255484e-01 -3.39722872e-01 3.71254086e-01 -6.27287567e-01
-3.97437155e-01 -3.23661119e-02 3.36326957e-02 -5.81034184e-01
1.57904714e-01 8.11699927e-01 7.29020774e-01 -1.29018769e-01
8.85399759e-01 3.42199028e-01 1.38963327e-01 -1.37824655e-01
2.39432871e-01 6.35774583e-02 9.77687657e-01 -1.39141545e-01
3.88652802e-01 5.55872798e-01 1.62878290e-01 -6.31264210e-01
-7.57932544e-01 -6.45389259e-01 -7.99612105e-01 -1.20360863e+00
1.01352119e+00 -9.48996067e-01 -1.60493422e+00 5.90618610e-01
-7.28752911e-01 -8.02611947e-01 -2.97978580e-01 6.28450811e-01
-8.51047754e-01 -9.90190953e-02 -1.18052471e+00 -8.03069592e-01
-1.08471103e-01 -8.03597450e-01 7.12333381e-01 3.48867774e-01
-2.22700238e-01 -1.24517775e+00 7.01508462e-01 5.22494316e-01
2.65827209e-01 9.09181714e-01 1.76946312e-01 -7.35779405e-01
-4.04616028e-01 -3.10127705e-01 5.86967230e-01 -5.98920882e-01
-2.66812176e-01 -1.96590483e-01 -3.73390585e-01 -1.33684084e-01
-6.04932308e-01 -3.34955871e-01 1.06392312e+00 9.93766785e-01
9.31435525e-01 1.32618025e-01 -2.48929039e-01 8.84247780e-01
9.46182489e-01 2.81434089e-01 4.62819546e-01 7.12730050e-01
1.06084359e+00 4.82149601e-01 5.14838457e-01 5.39863288e-01
5.57206690e-01 9.11213875e-01 5.78998744e-01 -1.35897370e-02
-1.13796204e-01 -3.80554646e-01 3.98116797e-01 6.79477751e-01
-1.03576756e+00 -4.39914733e-01 -1.20148182e+00 5.35251498e-01
-2.49931216e+00 -1.66144335e+00 -4.56656754e-01 1.40049565e+00
-2.66925067e-01 2.42926404e-01 1.13605762e+00 -5.63649647e-03
6.03961647e-01 6.67936862e-01 -5.28791606e-01 -6.23821318e-01
-1.71267375e-01 2.40336522e-01 5.33597112e-01 4.78691399e-01
-1.22493291e+00 1.60305691e+00 7.28868628e+00 6.40357196e-01
-8.15710843e-01 1.37421191e-01 3.47260952e-01 -6.49981499e-01
3.12874883e-01 -3.89927447e-01 -3.65203738e-01 2.33213663e-01
9.72878098e-01 -2.21138030e-01 3.05856168e-01 6.28675997e-01
3.91222388e-01 3.08057576e-01 -5.81445336e-01 9.20296490e-01
2.07790792e-01 -1.91143858e+00 -1.89736575e-01 2.51243860e-01
3.22154343e-01 2.91312277e-01 2.84163415e-01 5.91455102e-01
1.11211097e+00 -1.42723048e+00 8.04199457e-01 7.04585254e-01
1.84628755e-01 -1.00771809e+00 7.05367446e-01 3.49367976e-01
-1.72088981e+00 -5.46445251e-01 -2.34221533e-01 -1.14756858e+00
3.12835455e-01 -5.26988566e-01 -6.48378968e-01 5.01526475e-01
1.00277400e+00 1.41421223e+00 -5.62908411e-01 8.03664088e-01
8.05218294e-02 7.29923785e-01 3.62704769e-02 -3.76973480e-01
9.41965222e-01 -2.85519272e-01 4.99620587e-01 9.70120907e-01
1.92148790e-01 3.03105146e-01 4.56450701e-01 4.17189479e-01
1.81083962e-01 1.39130885e-02 -9.80681956e-01 -8.62969309e-02
9.70233902e-02 8.81483614e-01 -9.52417493e-01 -1.03718311e-01
-2.32809454e-01 7.52567232e-01 8.67868662e-01 2.06100717e-01
-9.68412757e-01 3.07055898e-02 9.69838798e-01 6.95062056e-02
1.55970782e-01 -7.03616798e-01 -1.99031204e-01 -9.95423436e-01
-2.24462420e-01 -4.69743937e-01 1.02854657e+00 -5.62376618e-01
-8.77226055e-01 7.97481656e-01 -3.03227156e-01 -1.49730456e+00
-6.55841887e-01 -4.69541818e-01 -1.18071675e+00 5.23090363e-01
-8.14846635e-01 -1.28793132e+00 -5.73329508e-01 8.28553140e-01
7.32066154e-01 -7.77262449e-01 3.35740685e-01 -1.80548728e-01
-6.31875813e-01 2.19415233e-01 -2.63458967e-01 8.91329110e-01
-5.76643497e-02 -1.01379907e+00 8.81483018e-01 9.16566670e-01
6.66942656e-01 -1.54092908e-01 5.47186434e-01 -7.61858106e-01
-1.16161811e+00 -9.44753289e-01 4.32597876e-01 -6.92946374e-01
8.66784573e-01 -2.85016447e-01 -5.01740813e-01 1.02907598e+00
1.29859924e-01 7.70912811e-05 9.76123393e-01 4.22783822e-01
4.70703803e-02 2.62884617e-01 -1.02659538e-01 7.31017768e-01
1.59022021e+00 -3.23195338e-01 -3.66718411e-01 2.24790335e-01
1.92529336e-01 -7.38998055e-01 -7.75778234e-01 2.02447608e-01
5.05424678e-01 -1.20191443e+00 9.86229479e-01 -1.43307495e+00
5.37599862e-01 7.82136992e-03 1.68148965e-01 -1.39013302e+00
-6.63634121e-01 -7.20120788e-01 -1.97612166e-01 2.39778012e-01
1.45247862e-01 9.88055244e-02 1.59769654e+00 1.78763345e-01
-3.15409303e-01 -6.02333665e-01 -9.54355836e-01 -6.99340224e-01
3.16379778e-02 -7.27524638e-01 5.15500665e-01 8.66095006e-01
3.58973771e-01 2.58495539e-01 -9.72202778e-01 -7.69966617e-02
3.30861956e-01 3.78101170e-01 1.11296988e+00 -1.13138139e+00
-3.59628886e-01 -8.16173017e-01 -1.33647346e+00 -9.65460598e-01
2.56373823e-01 -7.66796947e-01 -3.31039250e-01 -1.58754981e+00
-5.24423737e-03 1.19648939e-02 -2.69385993e-01 1.50905922e-01
-6.34245388e-03 7.17375159e-01 3.88370633e-01 2.79596478e-01
-1.16914952e+00 4.53565389e-01 1.33025432e+00 -2.09853142e-01
-6.12226129e-01 1.90680578e-01 -1.89551279e-01 5.96395433e-01
8.04121256e-01 -2.60709554e-01 -3.73823196e-01 -5.79732418e-01
4.50719208e-01 6.45310342e-01 4.32468057e-01 -1.37975848e+00
7.15420723e-01 -3.51379514e-01 1.48677975e-01 -5.07123590e-01
6.48575425e-01 3.71198766e-02 1.27681017e-01 8.16629350e-01
-4.52150702e-01 6.26547217e-01 8.43771026e-02 8.74894679e-01
-5.83737493e-01 4.49753940e-01 -5.04850643e-03 -2.90693641e-01
-1.44129491e+00 7.62938201e-01 -8.20157111e-01 1.55795768e-01
1.30053735e+00 -7.05201805e-01 -2.26352289e-01 -1.04880369e+00
-1.31210911e+00 4.18440253e-01 -4.31888774e-02 6.54582918e-01
7.41696596e-01 -1.42375958e+00 -9.48586643e-01 -2.41253003e-01
1.62501689e-02 -5.94338477e-01 7.79211223e-01 5.97043395e-01
-1.11339509e+00 1.90944374e-01 -7.02992678e-01 -5.87327182e-01
-1.31845415e+00 2.09333599e-01 7.03384817e-01 -5.52946270e-01
-1.10840940e+00 1.12982488e+00 -2.13242352e-01 -3.10986549e-01
1.38818890e-01 -9.94991362e-02 -9.77156699e-01 -1.04946382e-01
2.92096645e-01 4.99450207e-01 -9.27795470e-02 -9.45393920e-01
-3.81886959e-01 4.08871919e-01 2.63636678e-01 -1.24695279e-01
1.64468503e+00 2.80573934e-01 4.47104812e-01 7.76998699e-01
4.28841203e-01 -5.03685951e-01 -1.41996920e+00 -1.96924984e-01
1.43044606e-01 -4.14084315e-01 -4.71164025e-02 -1.61575824e-01
-1.31027901e+00 8.44122350e-01 4.26759243e-01 5.05175591e-01
7.17404008e-01 3.50048356e-02 7.66699731e-01 1.13390565e-01
4.37486798e-01 -1.20664132e+00 5.82337916e-01 8.71381342e-01
4.57376122e-01 -1.07117105e+00 -2.89200872e-01 -1.86249171e-03
-1.10214472e+00 1.10194445e+00 8.85235548e-01 -7.73406744e-01
7.35864103e-01 2.37574384e-01 2.73565203e-01 -9.98103738e-01
-1.01809072e+00 -6.94808066e-01 4.08008277e-01 5.71157277e-01
3.94273728e-01 2.17653915e-01 -3.02873589e-02 6.58390999e-01
-7.57321596e-01 -2.55634248e-01 6.14875317e-01 9.53280807e-01
-2.70770967e-01 -5.49728453e-01 1.29473563e-02 4.84028071e-01
-4.78015870e-01 2.32616767e-01 -7.29634643e-01 9.96257842e-01
-1.05992675e-01 9.81746376e-01 5.40122271e-01 -1.22633576e+00
4.36239123e-01 -2.96366632e-01 2.78779447e-01 -5.67627370e-01
-1.29517102e+00 -2.72131473e-01 9.23737884e-02 -1.18683541e+00
-5.78205228e-01 -6.13691032e-01 -9.13447559e-01 -1.00122738e+00
1.91859692e-01 2.27937996e-01 1.11775286e-01 9.10917461e-01
2.41558045e-01 1.05486035e+00 3.26435149e-01 -9.77603078e-01
3.54878008e-01 -1.00797606e+00 -1.02082002e+00 6.20731771e-01
1.15287984e-02 -8.71495128e-01 1.81418106e-01 -2.66110331e-01] | [6.759116172790527, 0.3353029787540436] |
9361ed17-eaf4-46c8-99a8-87e836d8f7bf | on-automatic-data-augmentation-for-3d-point | 2112.06029 | null | https://arxiv.org/abs/2112.06029v1 | https://arxiv.org/pdf/2112.06029v1.pdf | On Automatic Data Augmentation for 3D Point Cloud Classification | Data augmentation is an important technique to reduce overfitting and improve learning performance, but existing works on data augmentation for 3D point cloud data are based on heuristics. In this work, we instead propose to automatically learn a data augmentation strategy using bilevel optimization. An augmentor is designed in a similar fashion to a conditional generator and is optimized by minimizing a base model's loss on a validation set when the augmented input is used for training the model. This formulation provides a more principled way to learn data augmentation on 3D point clouds. We evaluate our approach on standard point cloud classification tasks and a more challenging setting with pose misalignment between training and validation/test sets. The proposed strategy achieves competitive performance on both tasks and we provide further insight into the augmentor's ability to learn the validation set distribution. | ['Chuan-Sheng Foo', 'Fayao Liu', 'Xun Xu', 'Wanyue Zhang'] | 2021-12-11 | null | null | null | null | ['3d-object-classification', '3d-object-recognition', 'point-cloud-classification'] | ['computer-vision', 'computer-vision', 'computer-vision'] | [ 2.67553777e-01 4.12535280e-01 -3.42573404e-01 -6.18457913e-01
-8.55581403e-01 -5.04029274e-01 5.40890217e-01 2.85325468e-01
-4.64325815e-01 5.15905142e-01 -2.43865684e-01 -5.41390240e-01
2.68484533e-01 -6.41584039e-01 -9.69177246e-01 -7.03915238e-01
-1.19493626e-01 9.22003567e-01 -1.38837993e-01 1.56669673e-02
1.55282840e-01 9.83498156e-01 -1.31813407e+00 -2.16242131e-02
8.51330578e-01 9.02172863e-01 -2.10425481e-02 6.05518878e-01
-2.87681520e-01 1.62123993e-01 -5.36750793e-01 -1.23295918e-01
6.51683390e-01 6.26406670e-02 -7.41445303e-01 3.28147948e-01
6.67969942e-01 8.70687068e-02 -1.07262500e-01 7.38684475e-01
4.01668489e-01 5.41179848e-04 8.12489986e-01 -1.44472301e+00
-3.35999131e-01 1.27349153e-01 -5.17273843e-01 1.23151459e-01
-3.29865590e-02 1.61885858e-01 7.12091625e-01 -1.01511931e+00
2.28110418e-01 9.44341183e-01 8.47270966e-01 6.71755910e-01
-1.57204533e+00 -6.98038936e-01 3.33766967e-01 -4.71518040e-01
-1.24092698e+00 -1.64053962e-01 1.02985573e+00 -6.00753248e-01
6.59909725e-01 3.71177346e-01 7.93074012e-01 8.41977000e-01
-3.80116582e-01 9.03643250e-01 1.17443800e+00 -4.90768105e-01
2.52408952e-01 4.77753758e-01 1.62391245e-01 4.83823359e-01
2.88471937e-01 2.94121832e-01 -3.01373571e-01 -3.91655296e-01
8.75628829e-01 -1.60553157e-01 -1.46836609e-01 -9.91222203e-01
-1.02480996e+00 9.11883175e-01 7.48043835e-01 -2.54766852e-01
-3.55353773e-01 7.70087168e-02 1.86556205e-01 1.73151210e-01
8.45580101e-01 7.96693444e-01 -7.47096837e-01 2.72007287e-01
-6.22232735e-01 6.16986275e-01 5.88914156e-01 1.16853726e+00
7.85867512e-01 1.31332153e-03 -1.35731995e-01 8.84321749e-01
4.56456363e-01 5.69835722e-01 1.03705272e-01 -5.95478058e-01
6.02011561e-01 7.20388651e-01 8.76084417e-02 -4.28566307e-01
-4.69453394e-01 -7.32834995e-01 -7.51552761e-01 6.00488603e-01
2.45995864e-01 -1.71814129e-01 -1.32910717e+00 1.65808880e+00
6.21460915e-01 2.00756535e-01 -2.40801927e-02 7.95493126e-01
6.45498216e-01 4.80551600e-01 -1.01029664e-01 -3.11528780e-02
9.11642551e-01 -5.61418235e-01 -3.08382452e-01 -2.89745599e-01
7.35323787e-01 -5.79496741e-01 1.26461554e+00 2.34944478e-01
-9.95208621e-01 -6.52565300e-01 -1.13312864e+00 1.60366908e-01
-2.52303451e-01 1.15978405e-01 7.75239408e-01 7.35414982e-01
-6.40744865e-01 5.52674890e-01 -8.92063498e-01 -6.97133094e-02
8.21439624e-01 6.61889970e-01 -4.30007875e-01 -1.71847120e-02
-6.06829941e-01 9.71185088e-01 3.60225648e-01 -1.51952311e-01
-7.07511067e-01 -1.08192396e+00 -8.95765364e-01 -4.01904285e-01
1.42093182e-01 -7.89091885e-01 1.37348437e+00 -3.85482043e-01
-1.29714191e+00 1.17187643e+00 6.51011243e-02 -4.85497117e-01
4.00286585e-01 -2.92895079e-01 1.57937005e-01 -3.92249286e-01
1.36380484e-02 9.47783470e-01 1.00323510e+00 -1.71656311e+00
-4.73586559e-01 -6.80558741e-01 -2.19137594e-01 3.01385909e-01
-2.20849290e-01 -5.37044704e-01 -5.11810958e-01 -7.68899977e-01
4.86055464e-01 -1.33255196e+00 -5.82794130e-01 -3.85871087e-03
-5.39682269e-01 -1.59128398e-01 9.33859825e-01 -9.07507762e-02
7.10293293e-01 -1.90745759e+00 5.89127373e-03 5.56294620e-01
1.24648273e-01 3.65452796e-01 -1.29769564e-01 1.58632934e-01
-4.18844610e-01 2.47363210e-01 -6.85399652e-01 -7.19482899e-01
-1.59481362e-01 4.56247419e-01 -4.56956983e-01 4.60299581e-01
6.58243477e-01 7.06987798e-01 -6.70690179e-01 -1.44326121e-01
4.17912245e-01 3.47545356e-01 -7.78946936e-01 5.25022686e-01
-3.67220134e-01 8.90962183e-01 -3.60275269e-01 6.08308315e-01
9.51559842e-01 -1.90877974e-01 -3.89649451e-01 3.39301266e-02
9.28481445e-02 3.11117768e-01 -1.12640667e+00 1.66034603e+00
-6.24671817e-01 4.51511651e-01 -1.91780254e-01 -9.45991457e-01
1.31182837e+00 1.30219728e-01 5.58020949e-01 -8.30103755e-02
9.21514481e-02 6.20350130e-02 -1.05542846e-01 -2.58585155e-01
2.60178447e-01 -4.90048021e-01 -5.62100783e-02 2.85552979e-01
6.16854802e-02 -6.74329102e-01 -3.79768461e-01 -8.31495002e-02
7.68534780e-01 2.57782817e-01 8.72344077e-02 -1.00565314e-01
3.54603648e-01 3.16102415e-01 3.91649246e-01 7.52041221e-01
2.38411829e-01 7.83676803e-01 2.92540103e-01 -6.22785509e-01
-1.40072048e+00 -1.08194542e+00 -6.10697508e-01 5.18021464e-01
-1.43263340e-01 -2.77993232e-01 -3.87188196e-01 -1.16712761e+00
4.95904773e-01 8.19812298e-01 -7.17656195e-01 -1.99300915e-01
-4.12992209e-01 -9.16457832e-01 2.85641074e-01 7.49407589e-01
2.74878025e-01 -5.81806660e-01 -5.60368150e-02 -1.49838954e-01
1.05088495e-01 -1.19374609e+00 -1.79026708e-01 4.38442260e-01
-1.53855133e+00 -9.89298880e-01 -3.77893925e-01 -6.24718606e-01
1.16373658e+00 1.53309688e-01 1.14067364e+00 7.64082000e-02
3.02841589e-02 2.62507796e-01 -3.24193507e-01 -1.15274692e+00
-4.22507554e-01 3.67590159e-01 2.30081245e-01 -1.99784666e-01
5.03501952e-01 -6.12393081e-01 -2.09698856e-01 2.08217964e-01
-7.46988535e-01 4.74299975e-02 6.25869870e-01 9.05912817e-01
8.32942724e-01 -2.44068116e-01 3.88993412e-01 -1.13938761e+00
3.84169996e-01 -4.40362930e-01 -7.99582720e-01 -3.40912282e-01
-6.90625310e-01 2.84542114e-01 3.20248306e-01 -2.59040147e-01
-6.78475380e-01 6.07925057e-01 -4.82621223e-01 -9.11348641e-01
-4.18580502e-01 3.20647836e-01 -2.18780488e-01 -5.17949343e-01
8.38871360e-01 -7.11228931e-03 2.48181969e-01 -8.74493897e-01
3.06850791e-01 4.88687724e-01 4.33261573e-01 -7.32847810e-01
1.29115427e+00 4.99902070e-01 3.13094884e-01 -7.97182322e-01
-8.50054920e-01 -5.30237257e-01 -1.00059700e+00 1.11693114e-01
5.78039348e-01 -9.01403248e-01 -6.68020427e-01 1.26769736e-01
-1.10066307e+00 -3.18394214e-01 -7.20963717e-01 6.40636623e-01
-6.54184997e-01 2.51329225e-02 3.86028811e-02 -8.41474950e-01
-1.72312990e-01 -9.76984859e-01 1.29809272e+00 -7.29036182e-02
-2.70684133e-03 -8.84873271e-01 3.01681668e-01 2.36226961e-01
5.46444915e-02 5.60217083e-01 8.95445287e-01 -9.29782569e-01
-6.16503537e-01 -6.72934532e-01 3.81414667e-02 7.47273862e-01
9.74604413e-02 -2.46941537e-01 -1.24122953e+00 -3.67764235e-01
1.27840936e-01 -2.85294056e-01 7.24693179e-01 3.62520128e-01
1.51538885e+00 -2.69501925e-01 -3.49568009e-01 9.41991985e-01
1.19665468e+00 -1.74698830e-01 4.75569278e-01 3.05088073e-01
7.87386239e-01 4.09479082e-01 8.05588782e-01 2.78864622e-01
1.42568320e-01 7.38812268e-01 7.23035336e-01 -3.01049203e-01
1.45675331e-01 -4.77196157e-01 -2.26123229e-01 4.03546602e-01
-1.16146132e-01 1.35728002e-01 -1.21075344e+00 3.73833865e-01
-1.61719990e+00 -3.40962648e-01 -2.69803643e-01 2.47866201e+00
6.46279454e-01 2.88856357e-01 1.96282864e-01 3.49105775e-01
3.30372840e-01 -1.00747608e-01 -5.46963274e-01 -8.50533880e-03
9.62189026e-03 4.02572632e-01 5.96852303e-01 5.07448971e-01
-1.41608548e+00 8.38844538e-01 6.84287024e+00 2.95426011e-01
-1.05509675e+00 -1.65581346e-01 5.04989624e-01 -1.57802954e-01
-1.27498552e-01 -9.63687673e-02 -8.62107694e-01 2.33258933e-01
3.70519340e-01 2.67232984e-01 -7.31962621e-02 1.09442198e+00
-2.25368422e-02 2.03150854e-01 -1.42634988e+00 1.05457342e+00
-1.44321516e-01 -1.50169802e+00 5.08906841e-02 3.39763969e-01
8.27452898e-01 2.27304772e-01 2.33595029e-01 3.67581010e-01
2.41429925e-01 -9.79611874e-01 4.25564378e-01 1.89774573e-01
7.69994915e-01 -7.36273408e-01 7.90001333e-01 5.40486276e-01
-6.20786250e-01 1.06010728e-01 -2.00852484e-01 -1.41267076e-01
4.06648917e-03 5.41372895e-01 -1.48718750e+00 2.23080739e-01
5.18801451e-01 4.74145979e-01 -5.71242809e-01 1.25005198e+00
-3.30090433e-01 5.79617381e-01 -5.93632936e-01 2.83688664e-01
4.18126062e-02 -2.18521357e-01 9.30440545e-01 9.64466155e-01
7.62302503e-02 1.26954868e-01 3.60758036e-01 7.77901947e-01
-2.32216895e-01 -9.99851227e-02 -1.06337571e+00 2.58802116e-01
4.10810471e-01 1.07035303e+00 -4.16712672e-01 -5.77391498e-02
-2.61718392e-01 5.42506754e-01 3.39786321e-01 2.22501218e-01
-4.74608958e-01 1.12303391e-01 9.34712887e-01 2.08897650e-01
1.17710359e-01 -4.54153210e-01 -8.78176153e-01 -9.40762520e-01
1.12393163e-01 -5.71411908e-01 2.75430381e-01 -6.74833357e-01
-1.39959860e+00 4.95947182e-01 1.93402931e-01 -1.53246224e+00
-4.19864729e-02 -7.47026861e-01 -5.46226799e-01 1.06169236e+00
-1.49756682e+00 -1.30003750e+00 -4.51298028e-01 3.03225607e-01
2.43307725e-01 -2.81614929e-01 8.38320434e-01 1.69509351e-01
-2.86202461e-01 6.31222188e-01 -2.57886261e-01 1.88457087e-01
3.17701668e-01 -1.55084908e+00 8.17204416e-01 6.46630406e-01
2.68466651e-01 3.28340024e-01 7.20694244e-01 -6.31523073e-01
-1.27819991e+00 -1.29888523e+00 5.03509223e-01 -9.97288346e-01
8.89176801e-02 -5.81477821e-01 -1.15248132e+00 8.58681798e-01
-3.77127945e-01 1.89805001e-01 8.81134331e-01 5.95140755e-01
-2.14987874e-01 -5.56210130e-02 -1.25256205e+00 3.12251866e-01
8.97542894e-01 -2.38574728e-01 -6.18525088e-01 5.88385642e-01
7.62613952e-01 -8.72545481e-01 -9.52847481e-01 8.39720488e-01
4.19304110e-02 -4.25479770e-01 1.14787233e+00 -1.04743731e+00
2.56465733e-01 -4.28893387e-01 -2.38871947e-01 -1.58946037e+00
-1.57906324e-01 -3.88693333e-01 -1.95001751e-01 9.48187351e-01
5.16954243e-01 -4.95735586e-01 1.43161988e+00 5.23416996e-01
-4.75784034e-01 -1.09219956e+00 -7.09237456e-01 -8.19568515e-01
2.25856408e-01 -8.06652844e-01 7.41105497e-01 9.98521030e-01
-4.43654001e-01 4.87112254e-01 -1.63643941e-01 6.37803435e-01
8.38834941e-01 -6.90541491e-02 1.25963688e+00 -1.52270603e+00
-2.56857034e-02 -1.98870599e-01 -4.81942564e-01 -1.04001367e+00
1.05951115e-01 -1.18346024e+00 1.09042544e-02 -1.23195231e+00
-1.18669681e-01 -1.25140655e+00 8.43097176e-03 4.67970192e-01
-2.39312887e-01 3.96558076e-01 3.68212253e-01 2.21256122e-01
5.70234917e-02 8.02368522e-01 1.34528434e+00 -3.26146148e-02
-5.44599295e-01 6.77626193e-01 -6.05846226e-01 7.37914503e-01
8.20443451e-01 -5.77879012e-01 -4.32912230e-01 -4.84225005e-01
9.35293287e-02 -2.46765777e-01 5.29052258e-01 -9.24245596e-01
-5.74287288e-02 -4.16183807e-02 6.75376475e-01 -8.54770184e-01
5.16673625e-01 -1.07520413e+00 -1.17766112e-01 2.82825202e-01
-3.86999875e-01 -2.40021627e-02 6.08347178e-01 6.10532820e-01
-6.15359060e-02 -2.57654309e-01 9.59736943e-01 1.40200689e-01
-1.93647996e-01 8.09463620e-01 1.91129044e-01 -5.33604585e-02
1.06373262e+00 -1.85482532e-01 1.42936543e-01 -1.11771695e-01
-9.61253107e-01 3.59758526e-01 5.60226858e-01 5.27072430e-01
7.43457377e-01 -1.63987076e+00 -8.37320030e-01 6.74116790e-01
4.12878782e-01 6.78400636e-01 -2.37692267e-01 5.23802161e-01
-3.47134739e-01 2.91903049e-01 -1.22309756e-02 -1.15557718e+00
-1.14433575e+00 5.82481980e-01 3.86716992e-01 -1.52581051e-01
-6.13103926e-01 9.87020433e-01 9.71818864e-02 -1.05893826e+00
2.82893151e-01 -3.53286445e-01 -4.80695963e-02 -2.58527637e-01
2.00771406e-01 4.61871317e-03 3.30453783e-01 -3.68715584e-01
-2.35346571e-01 5.01477361e-01 -1.16179600e-01 -1.34725366e-02
1.43164670e+00 3.37824881e-01 3.02252352e-01 5.33885956e-01
1.34832120e+00 -2.04629973e-01 -1.28911448e+00 -4.77439463e-01
-9.48541239e-02 -8.46311986e-01 8.42215717e-02 -5.18370569e-01
-9.14027154e-01 9.22900021e-01 7.69576371e-01 1.57540783e-01
8.60926986e-01 2.58604348e-01 2.24455610e-01 3.92659694e-01
1.79264382e-01 -7.84113646e-01 1.22399509e-01 4.66364264e-01
1.10918128e+00 -1.70498490e+00 9.99364480e-02 -6.65356457e-01
-6.42695367e-01 8.61299276e-01 8.38229656e-01 -2.77338028e-01
8.43542039e-01 2.59585947e-01 4.99738753e-02 -3.32394719e-01
-5.45006335e-01 -2.98421323e-01 5.28778017e-01 9.14071679e-01
3.14478517e-01 -7.33841658e-02 2.58834451e-01 2.97464192e-01
-4.31249678e-01 -1.45169690e-01 1.77103341e-01 9.45066392e-01
-2.31131375e-01 -1.24278438e+00 -5.19444942e-01 6.66026175e-01
5.46992570e-02 2.29284480e-01 -4.17504221e-01 1.01314008e+00
5.72553910e-02 3.85724008e-01 3.97091776e-01 -5.33772588e-01
7.04373240e-01 1.85079813e-01 5.84363699e-01 -1.05046773e+00
-4.61691648e-01 -1.37518615e-01 -1.65082961e-01 -2.68795550e-01
-2.84303010e-01 -8.56717587e-01 -1.08996069e+00 2.88758124e-03
-4.16413128e-01 2.94391930e-01 1.02483952e+00 7.29763865e-01
3.20618957e-01 4.48843241e-01 9.40467298e-01 -1.00246346e+00
-6.60267234e-01 -9.65946078e-01 -2.56116778e-01 4.06828344e-01
6.70943379e-01 -7.85867572e-01 -2.41927177e-01 -2.13751033e-01] | [8.020560264587402, -3.315605401992798] |
f8e5c28f-0f76-4d67-a144-92f967694374 | yolo-and-k-means-based-3d-object-detection | 2004.11465 | null | https://arxiv.org/abs/2004.11465v1 | https://arxiv.org/pdf/2004.11465v1.pdf | YOLO and K-Means Based 3D Object Detection Method on Image and Point Cloud | Lidar based 3D object detection and classification tasks are essential for automated driving(AD). A Lidar sensor can provide the 3D point coud data reconstruction of the surrounding environment. But the detection in 3D point cloud still needs a strong algorithmic challenge. This paper consists of three parts.(1)Lidar-camera calib. (2)YOLO, based detection and PointCloud extraction, (3) k-means based point cloud segmentation. In our research, Camera can capture the image to make the Real-time 2D Object Detection by using YOLO, I transfer the bounding box to node whose function is making 3d object detection on point cloud data from Lidar. By comparing whether 2D coordinate transferred from the 3D point is in the object bounding box or not, and doing a k-means clustering can achieve High-speed 3D object recognition function in GPU. | ['Yoko SASAKI', 'Weimin WANG', 'Kentaro SHIMIZU', 'Xuanyu YIN'] | 2020-04-21 | null | null | null | null | ['3d-object-recognition'] | ['computer-vision'] | [-1.87198147e-01 -6.92239821e-01 5.97909987e-02 -3.39442402e-01
-3.69767666e-01 -4.56715643e-01 2.04956189e-01 2.59153754e-01
-4.71621275e-01 -4.99440357e-02 -7.40082026e-01 -5.99712014e-01
1.67517275e-01 -1.06917787e+00 -5.60348034e-01 -4.13470954e-01
8.65168422e-02 1.11639535e+00 9.32698905e-01 -1.95530728e-01
8.11713934e-01 1.44998300e+00 -1.97994959e+00 -2.03557611e-01
5.48960567e-01 1.22799969e+00 5.04980922e-01 7.96875060e-01
-7.50896990e-01 -3.40814799e-01 -5.22696257e-01 3.99679005e-01
4.89403963e-01 2.50357360e-01 -5.15636541e-02 4.07691021e-03
3.83469999e-01 -4.02189374e-01 3.73068333e-01 1.07420993e+00
4.68032420e-01 -1.31615072e-01 9.12121773e-01 -1.80058289e+00
1.15096815e-01 -3.80568385e-01 -9.67569113e-01 3.22427362e-01
4.78310585e-02 -6.44075405e-03 -7.76744857e-02 -1.22899723e+00
4.19421762e-01 1.59028387e+00 6.84497178e-01 1.48202196e-01
-6.46586359e-01 -1.17654324e+00 -3.53407145e-01 3.14072132e-01
-1.70136654e+00 -2.64107175e-02 9.94121730e-01 -5.84258378e-01
1.03516400e+00 3.82974207e-01 7.96365976e-01 -9.60378870e-02
3.29019427e-01 4.23091561e-01 1.12635028e+00 -3.12690586e-01
2.16130316e-01 2.23941356e-01 6.02490485e-01 4.01933461e-01
2.68440247e-01 1.73843801e-01 -2.79757917e-01 -1.62963629e-01
7.48129606e-01 3.67941797e-01 2.68348902e-01 -3.72151136e-01
-9.45843816e-01 7.40311861e-01 4.94870931e-01 -1.92897692e-01
-1.54807903e-02 2.48840317e-01 2.81996995e-01 2.90986784e-02
1.51844710e-01 -3.48312110e-01 -2.84559369e-01 -1.50775298e-01
-7.18098938e-01 1.77716419e-01 1.67990267e-01 1.50077343e+00
1.19728887e+00 -2.56157637e-01 5.76655686e-01 2.62631327e-01
4.43202436e-01 9.72120583e-01 2.33395770e-01 -8.59553039e-01
4.10020024e-01 1.07867277e+00 5.43424822e-02 -9.45642591e-01
-3.26934129e-01 3.04204017e-01 -4.44634110e-01 1.07570338e+00
-1.53222352e-01 2.07104146e-01 -1.02710617e+00 2.93176711e-01
1.16790271e+00 2.81497359e-01 -1.55407622e-01 9.67909396e-01
1.25931692e+00 8.65535796e-01 -5.00638783e-01 4.97550480e-02
1.52098763e+00 -3.82638246e-01 -4.06640530e-01 -2.05529526e-01
6.99275017e-01 -9.02712703e-01 6.77922666e-01 2.74444044e-01
-7.45292068e-01 -9.64060962e-01 -1.10247350e+00 -4.14702088e-01
-5.03461659e-01 5.83260320e-02 1.91957936e-01 6.05593026e-01
-7.47785628e-01 -7.48440251e-02 -8.05366516e-01 -2.88082302e-01
5.33126771e-01 5.91653168e-01 -3.18506986e-01 -1.43722266e-01
-3.66183847e-01 9.34589207e-01 5.38857996e-01 -2.56158561e-02
-4.56299156e-01 -6.14632607e-01 -5.14545619e-01 -4.50589240e-01
2.95722783e-01 -4.91109639e-01 7.47304797e-01 -1.08172290e-01
-1.00256348e+00 1.22010076e+00 -7.10972667e-01 -3.60309064e-01
4.85212415e-01 -1.96397826e-01 1.41328480e-02 6.01607226e-02
4.00257647e-01 8.64350379e-01 8.12419832e-01 -1.40586364e+00
-1.03938663e+00 -9.78701532e-01 -8.37983429e-01 3.49732041e-01
4.33189720e-01 9.48357284e-02 -4.56346929e-01 4.04312730e-01
7.01190770e-01 -7.59853125e-01 -2.62782961e-01 3.62895697e-01
-3.15464258e-01 -6.92999721e-01 1.80488038e+00 1.53083682e-01
5.09365916e-01 -2.38820887e+00 -8.18187952e-01 3.20690840e-01
2.17962563e-01 3.42851013e-01 4.81932312e-01 8.37071687e-02
1.34758413e-01 6.58895522e-02 1.49415173e-02 -2.40763515e-01
-2.48973057e-01 2.97993988e-01 -3.27977449e-01 5.30017138e-01
2.80771375e-01 6.07173860e-01 -4.26742047e-01 -9.00944173e-01
6.55107379e-01 4.09129709e-01 -1.85136408e-01 -7.54422694e-02
1.96592093e-01 2.01725319e-01 -7.95174241e-01 7.61903405e-01
1.37289560e+00 5.90033710e-01 -8.70130599e-01 -6.43838495e-02
-6.78391159e-01 -3.87510508e-02 -1.59610057e+00 1.13308799e+00
9.09062475e-02 7.21133113e-01 2.73490787e-01 -6.36559725e-01
1.70298052e+00 -2.79906541e-01 3.95436853e-01 -4.38216984e-01
4.61639613e-01 3.70116264e-01 -2.97200620e-01 -2.87349075e-01
6.40673518e-01 -6.08169707e-04 1.19759059e-02 2.07340077e-01
-7.98884690e-01 -9.77824092e-01 -4.35795248e-01 -5.87599427e-02
6.87968612e-01 -1.62718102e-01 6.22115005e-03 -2.21688464e-01
4.98031825e-01 1.01027346e+00 3.80570561e-01 5.76696873e-01
-3.24859470e-01 4.90755409e-01 -2.00225916e-02 -3.57149392e-01
-1.06869757e+00 -8.76797259e-01 -7.13725626e-01 1.09080292e-01
7.97587752e-01 -1.88262071e-02 -4.34935331e-01 -1.55261979e-01
6.58096671e-01 5.64887345e-01 -8.07341859e-02 4.59932610e-02
-8.54207039e-01 -1.62216201e-01 6.31439611e-02 5.83051145e-01
5.88491917e-01 -6.23564720e-01 -1.31644917e+00 3.87331694e-02
4.90056992e-01 -1.02284634e+00 -1.36372760e-01 4.44513202e-01
-1.33884501e+00 -9.54834819e-01 6.79430738e-02 -8.84829879e-01
6.46582961e-01 9.62627769e-01 7.62645483e-01 1.64093032e-01
-6.09321892e-01 2.98350543e-01 -1.69173300e-01 -1.38804221e+00
4.24331650e-02 -3.82903457e-01 1.84715077e-01 -6.13400578e-01
1.12334263e+00 -3.82215232e-01 -4.31761503e-01 6.07602179e-01
-2.18763545e-01 -1.92414641e-01 2.36173838e-01 -1.89056590e-01
1.19671082e+00 4.73158807e-01 -2.06465110e-01 -3.21466625e-01
1.57696858e-01 -1.89607829e-01 -1.23291409e+00 -6.88315272e-01
-3.86262685e-01 -6.15776002e-01 -3.79753262e-02 -1.73435315e-01
-3.24521333e-01 5.30157030e-01 1.12402905e-02 -9.58053231e-01
-6.64461374e-01 -1.21228524e-01 -9.31256264e-02 -1.82594389e-01
4.93751526e-01 2.01543048e-01 1.30531579e-01 -5.85264862e-01
3.25069100e-01 1.15349340e+00 6.08887255e-01 -2.79714137e-01
1.19630623e+00 9.62662756e-01 4.74563003e-01 -1.12118840e+00
-7.40685835e-02 -1.01180589e+00 -1.42789292e+00 -4.11576629e-01
1.13958907e+00 -9.28197384e-01 -9.73772287e-01 1.29075348e-01
-1.54694319e+00 5.71562499e-02 -2.37357154e-01 5.15283644e-01
-3.09876174e-01 2.40232468e-01 2.97968555e-02 -1.28023541e+00
-3.07196498e-01 -1.17713165e+00 1.37754869e+00 8.02689120e-02
2.48264477e-01 -2.41381407e-01 -5.42381294e-02 3.68377656e-01
-3.63196850e-01 4.26627994e-01 6.68137550e-01 -1.19184367e-01
-1.00901580e+00 -6.95589423e-01 -4.97514635e-01 -1.24687806e-01
-2.95482695e-01 3.59344572e-01 -9.40514565e-01 1.47828102e-01
2.49787062e-01 2.77372152e-01 5.09955227e-01 3.60154629e-01
1.00409031e+00 4.71496195e-01 -1.00398290e+00 6.22309506e-01
1.51134932e+00 4.74997967e-01 4.49518830e-01 3.19445193e-01
6.73257172e-01 3.08193207e-01 1.35999513e+00 2.49991175e-02
5.36421657e-01 6.04087472e-01 8.05869460e-01 -1.36319932e-03
-9.07034203e-02 -1.12231962e-01 2.52939552e-01 7.32758641e-01
6.37073442e-02 2.74330169e-01 -1.29566336e+00 4.25292134e-01
-1.61355555e+00 -6.08921468e-01 -1.41447139e+00 2.16664410e+00
2.49745637e-01 5.18902838e-01 3.27372819e-01 4.99860793e-01
8.44913721e-01 -5.93460858e-01 -6.47229910e-01 -4.80360419e-01
3.54405969e-01 1.71373546e-01 8.95934403e-01 4.78173405e-01
-7.40544617e-01 1.02004957e+00 5.47609186e+00 8.62669528e-01
-1.12531340e+00 1.31936505e-01 1.21587664e-02 8.37130565e-03
1.92288846e-01 9.90845561e-02 -1.70590568e+00 4.65909928e-01
4.83376086e-01 -1.48742897e-02 -3.12800169e-01 1.24315429e+00
4.01193917e-01 -6.11817241e-01 -8.11628103e-01 1.49120975e+00
-2.80884635e-02 -1.04118848e+00 -1.22623868e-01 4.06362742e-01
2.19596416e-01 3.93376201e-01 -3.72891277e-01 1.44883385e-02
3.86509560e-02 -7.75793314e-01 9.68861818e-01 2.73625493e-01
9.51673210e-01 -8.75984728e-01 5.36858678e-01 1.14905369e+00
-1.57856417e+00 1.72461823e-01 -9.68443990e-01 -2.24346414e-01
1.04233183e-01 8.42853963e-01 -1.20866466e+00 2.41928056e-01
1.04680347e+00 4.77853656e-01 -3.65541071e-01 1.28615081e+00
1.88168481e-01 2.11856127e-01 -8.87496173e-01 -3.29939902e-01
2.32089147e-01 -6.79367065e-01 7.88911939e-01 9.66652572e-01
5.34599066e-01 4.34010148e-01 3.83785069e-01 9.24403548e-01
4.01784092e-01 4.74691913e-02 -9.02343929e-01 6.53124869e-01
9.42988336e-01 1.16116083e+00 -1.01726401e+00 -2.86604613e-01
-1.65969729e-02 3.69093835e-01 -1.52684838e-01 -3.13563079e-01
-6.87831521e-01 -7.27265418e-01 6.77066386e-01 6.50516152e-01
3.89726669e-01 -1.08406699e+00 -8.49027932e-01 -3.11152935e-01
-2.10740753e-02 1.33821726e-01 -4.42439988e-02 -1.25631595e+00
-6.71670318e-01 1.98407829e-01 2.32047275e-01 -1.42692757e+00
3.65394384e-01 -8.99627686e-01 -7.22086012e-01 1.03921235e+00
-1.76600575e+00 -8.38028908e-01 -7.12867081e-01 7.61206806e-01
5.21713912e-01 2.77757317e-01 3.14612120e-01 3.61933142e-01
-2.02152669e-01 -3.06715310e-01 -1.40656764e-02 7.74375116e-03
2.98146099e-01 -9.83463824e-01 2.99518347e-01 4.22338277e-01
-2.63690174e-01 1.75781503e-01 2.70617634e-01 -1.16844773e+00
-1.69960988e+00 -1.09811282e+00 5.75397432e-01 -9.99338508e-01
1.38109654e-01 -4.86699045e-01 -7.49825358e-01 4.81021881e-01
-4.28394645e-01 1.31194815e-01 2.64354289e-01 -5.64633608e-01
1.73293099e-01 -2.71331161e-01 -1.41784835e+00 1.19407445e-01
1.12780452e+00 -2.02732891e-01 -5.27466297e-01 4.22412097e-01
6.05833590e-01 -8.72841477e-01 -5.73137879e-01 3.69413853e-01
1.91590160e-01 -8.04114640e-01 1.28271961e+00 -2.96966702e-01
-1.37434155e-01 -1.05068052e+00 -1.17959090e-01 -3.33162308e-01
3.81370075e-02 -1.32801622e-01 3.95688295e-01 1.08799100e+00
-1.24943808e-01 -5.99855959e-01 1.08119643e+00 3.06609839e-01
-5.58754027e-01 -2.89186716e-01 -1.29196882e+00 -7.67745554e-01
-1.51224928e-02 -1.18291390e+00 8.61582696e-01 7.00133502e-01
-6.47331953e-01 4.66439903e-01 5.62227666e-01 5.78176737e-01
7.76352286e-01 4.59516048e-01 1.51964617e+00 -1.78382099e+00
8.96380603e-01 -4.04572815e-01 -1.02957749e+00 -1.36543548e+00
-1.35611475e-01 -9.74473834e-01 -3.06430720e-02 -1.57088041e+00
-2.44786873e-01 -1.09445703e+00 6.55416548e-01 1.23860739e-01
2.95357287e-01 3.97936940e-01 8.42635855e-02 6.93967700e-01
-8.66683722e-02 2.84974128e-01 1.19041193e+00 1.37853011e-01
-2.96714514e-01 1.64137244e-01 1.35619417e-01 7.82434106e-01
6.99674666e-01 -7.94686556e-01 -1.02596812e-01 -5.17131865e-01
2.55169924e-02 7.71446899e-02 4.29788619e-01 -1.21351957e+00
4.96098846e-01 -3.37588817e-01 5.89480162e-01 -2.05141068e+00
8.26012015e-01 -1.49445128e+00 -2.00524088e-02 7.00171828e-01
7.21019685e-01 1.84590876e-01 4.09508049e-01 4.22311306e-01
9.22632143e-02 -4.95956093e-01 9.64732051e-01 -3.82666230e-01
-7.11148322e-01 3.56313676e-01 -3.83960038e-01 -4.54298407e-01
1.57285154e+00 -1.29121935e+00 -2.12317500e-02 1.33717105e-01
-3.97459418e-01 5.27348816e-01 5.81101954e-01 6.94316253e-02
1.04391897e+00 -1.41035831e+00 -6.08208656e-01 5.03594279e-01
1.20967142e-01 9.03086543e-01 -3.00497770e-01 6.44568980e-01
-1.07165074e+00 4.35629368e-01 -3.03465202e-02 -1.64313221e+00
-1.69126523e+00 4.91574526e-01 2.71429032e-01 8.45711589e-01
-7.60202289e-01 8.36861312e-01 -2.14309990e-01 -6.25936031e-01
-1.25809804e-01 -6.24297857e-01 -1.48682326e-01 -2.83257943e-02
2.98189342e-01 6.31413341e-01 1.72315210e-01 -1.02031374e+00
-6.86994016e-01 1.47349215e+00 3.43895286e-01 -4.54554781e-02
9.76389408e-01 -1.30367815e-01 -1.43506423e-01 6.43776298e-01
1.08821940e+00 -1.18982382e-01 -7.72021890e-01 6.10584579e-02
-1.37538552e-01 -8.86002243e-01 3.58714283e-01 -2.58215796e-02
-6.04446173e-01 1.43196142e+00 1.07866251e+00 1.35263294e-01
6.31124258e-01 9.63667333e-02 8.71258140e-01 4.27038491e-01
6.14473701e-01 -9.68661010e-01 -3.52304369e-01 6.36121452e-01
4.99997675e-01 -1.20211518e+00 2.88547635e-01 -8.04985106e-01
-2.53875226e-01 1.46270025e+00 5.80612361e-01 -4.38482821e-01
8.79535437e-01 6.17920101e-01 2.21159324e-01 -7.10532308e-01
-1.13549098e-01 -3.81167352e-01 -1.08756544e-02 9.49816585e-01
-4.54405218e-01 9.45212245e-02 1.36944607e-01 1.23900898e-01
-7.14673162e-01 -1.36603460e-01 3.06378543e-01 1.15987432e+00
-1.33575439e+00 -6.82686567e-01 -1.26766336e+00 4.73728865e-01
3.89918894e-01 4.05528247e-01 -4.24976736e-01 8.29402864e-01
4.70675141e-01 8.13724101e-01 8.42224479e-01 -1.95480764e-01
5.03396571e-01 4.61051092e-02 2.37607971e-01 -7.35728264e-01
-1.83983997e-01 1.41263843e-01 -5.28108358e-01 -3.78686517e-01
1.35453850e-01 -7.66282499e-01 -1.91979790e+00 -2.51746118e-01
-6.10904336e-01 1.00955009e-01 1.49977160e+00 4.83643591e-01
4.55760866e-01 -2.21784294e-01 8.45341682e-01 -1.03598261e+00
-3.25049497e-02 -4.61178243e-01 -5.69928288e-01 -1.16367064e-01
2.91406900e-01 -7.46109366e-01 -2.91447699e-01 -1.75360262e-01] | [7.735830307006836, -2.6410229206085205] |
38f61d9c-8bc3-423a-a190-5df9a1d9dbb2 | efficient-multi-scale-3d-cnn-with-fully | 1603.05959 | null | http://arxiv.org/abs/1603.05959v3 | http://arxiv.org/pdf/1603.05959v3.pdf | Efficient Multi-Scale 3D CNN with Fully Connected CRF for Accurate Brain Lesion Segmentation | We propose a dual pathway, 11-layers deep, three-dimensional Convolutional
Neural Network for the challenging task of brain lesion segmentation. The
devised architecture is the result of an in-depth analysis of the limitations
of current networks proposed for similar applications. To overcome the
computational burden of processing 3D medical scans, we have devised an
efficient and effective dense training scheme which joins the processing of
adjacent image patches into one pass through the network while automatically
adapting to the inherent class imbalance present in the data. Further, we
analyze the development of deeper, thus more discriminative 3D CNNs. In order
to incorporate both local and larger contextual information, we employ a dual
pathway architecture that processes the input images at multiple scales
simultaneously. For post-processing of the network's soft segmentation, we use
a 3D fully connected Conditional Random Field which effectively removes false
positives. Our pipeline is extensively evaluated on three challenging tasks of
lesion segmentation in multi-channel MRI patient data with traumatic brain
injuries, brain tumors, and ischemic stroke. We improve on the state-of-the-art
for all three applications, with top ranking performance on the public
benchmarks BRATS 2015 and ISLES 2015. Our method is computationally efficient,
which allows its adoption in a variety of research and clinical settings. The
source code of our implementation is made publicly available. | ['Ben Glocker', 'Daniel Rueckert', 'David K. Menon', 'Andrew D. Kane', 'Joanna P. Simpson', 'Virginia F. J. Newcombe', 'Christian Ledig', 'Konstantinos Kamnitsas'] | 2016-03-18 | null | null | null | null | ['3d-medical-imaging-segmentation', 'brain-lesion-segmentation-from-mri'] | ['medical', 'medical'] | [ 5.45964062e-01 2.28609100e-01 1.63504221e-02 -4.37213361e-01
-9.56124485e-01 -2.36831307e-01 3.55405122e-01 2.14719251e-01
-8.13453972e-01 2.87484705e-01 1.74492136e-01 -5.38365185e-01
-1.74019188e-01 -4.60997909e-01 -5.91535330e-01 -6.91147923e-01
-3.45348835e-01 6.10983908e-01 6.85795724e-01 1.16500564e-01
1.70848176e-01 9.15401340e-01 -1.15625393e+00 7.39937961e-01
4.99557316e-01 1.13799191e+00 2.29200318e-01 5.58180332e-01
1.01771869e-01 5.63756645e-01 -2.40415260e-01 -2.79239900e-02
4.24813271e-01 -1.61656827e-01 -1.17828298e+00 4.98965643e-02
3.01196754e-01 -3.60706210e-01 -2.71485507e-01 8.97226453e-01
9.32296813e-01 -1.77884638e-01 7.69736350e-01 -5.25379360e-01
-6.99962005e-02 4.65990275e-01 -5.89527845e-01 8.42725515e-01
-1.65663838e-01 3.96901101e-01 5.08659065e-01 -9.29740846e-01
6.33085489e-01 8.33826959e-01 9.27790821e-01 4.94622558e-01
-1.12375331e+00 -3.95802379e-01 -5.12320697e-02 2.05036089e-01
-1.12604403e+00 -2.28694871e-01 4.60355073e-01 -8.16971660e-01
1.21165979e+00 1.32273436e-01 7.66090393e-01 1.06715453e+00
4.65472907e-01 9.21274900e-01 1.22544265e+00 -2.22052053e-01
2.73939133e-01 -3.50700706e-01 2.84925789e-01 6.08352125e-01
1.44120112e-01 -8.16349685e-02 -6.18591672e-03 -1.23835146e-01
7.94699490e-01 2.74847075e-02 -1.88865751e-01 -4.30459917e-01
-1.30086029e+00 6.85558736e-01 6.47126973e-01 5.17985344e-01
-7.75777340e-01 -9.82662812e-02 4.80063617e-01 -1.91837072e-01
6.54048383e-01 1.66849598e-01 -4.28584307e-01 2.24516019e-01
-1.21238303e+00 7.09274858e-02 4.19745564e-01 3.61078799e-01
2.54106134e-01 -3.29788595e-01 -4.28384513e-01 7.72809327e-01
2.65564591e-01 -9.88313928e-02 6.85191929e-01 -4.61110055e-01
2.23348856e-01 7.81940877e-01 -5.24404585e-01 -4.63203847e-01
-1.11750019e+00 -7.67232299e-01 -9.88001406e-01 4.79715437e-01
5.15840352e-01 -1.96501091e-01 -1.58175480e+00 1.24539316e+00
4.24376160e-01 7.26363882e-02 -2.46483892e-01 9.29287553e-01
7.85356045e-01 1.04664497e-01 2.38229290e-01 1.54100493e-01
1.47771883e+00 -8.42899442e-01 -3.42716575e-01 -2.92442620e-01
7.87329435e-01 -5.51494598e-01 7.32595742e-01 2.87939787e-01
-1.23701811e+00 -1.73829436e-01 -8.34942639e-01 -2.17618663e-02
-4.36211765e-01 2.33869895e-01 5.37335157e-01 6.20847106e-01
-1.28285062e+00 4.51734126e-01 -1.21450639e+00 -1.83931664e-01
1.18413520e+00 6.77147865e-01 -4.48881894e-01 -2.90964618e-02
-9.91456926e-01 1.23952913e+00 4.52098340e-01 3.20129633e-01
-8.44633281e-01 -1.02543139e+00 -6.12349987e-01 -2.91362733e-01
1.04099363e-01 -6.97407067e-01 1.06453192e+00 -8.94598186e-01
-1.21346450e+00 1.47255468e+00 1.34922147e-01 -6.22074664e-01
8.66430342e-01 5.63366599e-02 -9.11173970e-02 4.91813123e-01
1.58407688e-01 6.95796847e-01 7.04074860e-01 -8.49320292e-01
-4.27768975e-01 -6.96492136e-01 -1.53106377e-01 -4.49612327e-02
6.16566055e-02 1.72395214e-01 -3.69353443e-01 -7.15048373e-01
1.72651067e-01 -8.11395466e-01 -7.62138486e-01 -2.19251420e-02
-3.11294377e-01 4.21601459e-02 5.33356190e-01 -8.03921342e-01
8.00995588e-01 -2.13733363e+00 1.43213719e-01 3.36944580e-01
5.13837457e-01 3.28067601e-01 1.08880967e-01 -3.36371988e-01
-4.88578826e-01 -1.48292156e-02 -7.45846033e-01 -3.96426171e-01
-3.81394237e-01 -3.90789993e-02 2.37227544e-01 7.43510365e-01
3.53597671e-01 1.08003294e+00 -5.81781805e-01 -4.01326030e-01
2.12610707e-01 5.52778900e-01 -6.41347528e-01 6.83773383e-02
2.44147614e-01 7.12512493e-01 -3.31543475e-01 4.87461179e-01
6.75496340e-01 -2.80673146e-01 -1.51686072e-01 -2.50604957e-01
-1.50861502e-01 2.00350583e-01 -9.17263627e-01 2.06476068e+00
-2.42547557e-01 3.70688140e-01 1.84699327e-01 -1.43578732e+00
6.11306727e-01 4.09125865e-01 9.30223048e-01 -7.08157182e-01
4.96178329e-01 4.08514410e-01 1.74866930e-01 -7.13463545e-01
-2.03137666e-01 -1.53881401e-01 1.60036087e-01 4.46463138e-01
2.65780061e-01 -6.45741597e-02 1.28820702e-01 2.22535022e-02
1.23495126e+00 1.05497297e-02 1.24732494e-01 -5.67096770e-01
4.76227224e-01 1.47287786e-01 3.14202160e-01 7.87643492e-01
-5.45410633e-01 9.11142230e-01 6.75073266e-01 -6.68689668e-01
-8.58272970e-01 -1.03821790e+00 -4.51390266e-01 6.15956783e-01
-2.47240901e-01 7.72181749e-02 -1.04438877e+00 -8.91155064e-01
-2.34845966e-01 2.24163651e-01 -1.03974044e+00 -1.26286253e-01
-8.93852711e-01 -1.35635543e+00 6.29196703e-01 6.08844459e-01
3.67943972e-01 -1.05673540e+00 -1.13709116e+00 4.66669947e-01
-4.20284905e-02 -1.16575241e+00 -1.63246065e-01 4.97777402e-01
-1.10119736e+00 -1.29207695e+00 -9.96183813e-01 -8.97940814e-01
6.87822044e-01 -5.47441514e-03 1.01127839e+00 -3.47255841e-02
-8.69600415e-01 2.97932297e-01 -1.95293188e-01 -4.15110916e-01
-2.82836795e-01 3.60032052e-01 -3.50890130e-01 -9.90119502e-02
2.66867012e-01 -5.59085488e-01 -9.25904155e-01 3.86010222e-02
-1.11049414e+00 2.02635109e-01 9.85013902e-01 8.29738617e-01
7.20922709e-01 -2.58129805e-01 5.46324790e-01 -8.54774475e-01
4.38078791e-01 -5.68840384e-01 -2.37793878e-01 6.04102500e-02
-1.89858317e-01 -6.92791939e-02 2.19835281e-01 -2.04357967e-01
-8.20356488e-01 5.38505673e-01 -6.86739802e-01 -2.64022611e-02
-5.97348332e-01 4.46980327e-01 2.37373561e-01 -1.46534562e-01
7.33253658e-01 -3.73275131e-02 2.09456310e-01 -4.61119950e-01
1.58406556e-01 5.07269919e-01 6.12094700e-01 -2.24817008e-01
3.78972292e-01 7.40896106e-01 1.52812570e-01 -5.42931616e-01
-7.89698303e-01 -5.94534993e-01 -1.19105256e+00 -2.72454381e-01
1.28082252e+00 -6.60779178e-01 -2.11237282e-01 7.96392143e-01
-1.07767606e+00 -5.43382466e-01 -3.52045268e-01 5.48907816e-01
-4.10068750e-01 2.12160274e-01 -6.43655121e-01 -1.05535217e-01
-6.25061929e-01 -1.71662581e+00 1.04373753e+00 5.95078394e-02
-8.47832412e-02 -9.42806602e-01 -3.61982062e-02 3.03701729e-01
5.82491994e-01 5.50657928e-01 1.17432404e+00 -8.66254747e-01
-3.01116943e-01 -2.60079890e-01 -3.65417242e-01 4.14248109e-01
-2.90886518e-02 -3.28410178e-01 -9.51628625e-01 -1.88374668e-01
-1.27734229e-01 -2.30244979e-01 1.20989275e+00 9.19338286e-01
1.41946530e+00 3.83599728e-01 -3.22172374e-01 7.38668144e-01
1.19446003e+00 -1.32457495e-01 5.47821760e-01 4.87041652e-01
5.98325074e-01 6.42708838e-01 2.67482400e-02 2.15792120e-01
3.16819102e-01 4.04880613e-01 7.24903047e-01 -7.75753975e-01
-4.25958425e-01 5.04099250e-01 -8.48749280e-02 4.53698218e-01
-1.04236603e-01 2.54313052e-01 -1.28789043e+00 7.96480417e-01
-1.55034804e+00 -6.15725279e-01 -3.87270302e-01 1.85954237e+00
7.51721859e-01 3.88251781e-01 1.50824815e-01 1.72121152e-01
4.93281752e-01 -1.81165002e-02 -4.88829881e-01 -1.86155021e-01
-6.44236133e-02 7.80391276e-01 5.95919967e-01 2.69582659e-01
-1.48281109e+00 7.31059253e-01 6.17460346e+00 6.28070474e-01
-1.28013217e+00 4.74911749e-01 9.63929415e-01 -1.18809171e-01
2.18431816e-01 -4.37892169e-01 -4.13061053e-01 2.17458680e-01
7.07907736e-01 5.39479136e-01 -9.10623148e-02 5.04192114e-01
2.83836216e-01 -3.05408061e-01 -1.04314339e+00 6.60404265e-01
-8.36649761e-02 -1.54425418e+00 -1.51310697e-01 -9.64849517e-02
6.56493127e-01 6.63498878e-01 8.30263346e-02 -6.31135190e-03
-1.07452512e-01 -1.13553977e+00 7.36872792e-01 3.59175652e-01
7.90239275e-01 -6.08371794e-01 8.52721155e-01 2.32578844e-01
-7.99868643e-01 -1.76891237e-02 8.74477848e-02 2.29631156e-01
1.05741195e-01 8.63400519e-01 -8.42025638e-01 5.40421009e-01
1.02487242e+00 5.94411373e-01 -7.04399586e-01 1.30804896e+00
-4.72841933e-02 5.30959547e-01 -2.52086133e-01 4.81268644e-01
5.76520979e-01 2.30522826e-01 5.18008828e-01 1.71340954e+00
-4.62768413e-02 1.04978882e-01 1.84087187e-01 7.58279026e-01
1.42554656e-01 1.39451131e-01 -2.94941038e-01 5.28257310e-01
-3.44703376e-01 1.31745327e+00 -1.24635613e+00 -3.80550623e-01
-2.28877544e-01 8.25578690e-01 2.31553942e-01 1.43085346e-01
-6.32225156e-01 -1.54608190e-01 4.47768986e-01 1.05170496e-01
2.97942907e-01 -2.25046158e-01 -8.22962821e-01 -9.29683328e-01
7.13591650e-02 -6.58508956e-01 5.49671769e-01 -2.83920974e-01
-1.12148464e+00 8.49883616e-01 -7.52771497e-02 -8.01787138e-01
-9.08376500e-02 -8.08152139e-01 -6.68668091e-01 1.01229870e+00
-1.78856254e+00 -1.02178490e+00 -2.17402190e-01 7.41319299e-01
4.00765002e-01 -4.60537896e-02 7.49033809e-01 5.76346219e-01
-5.71726322e-01 2.62220681e-01 -2.05606461e-01 1.89639851e-01
5.54182053e-01 -1.19015861e+00 5.08729398e-01 9.79149699e-01
-4.16302204e-01 2.08039597e-01 1.86892822e-01 -5.46209931e-01
-8.71245503e-01 -1.16127861e+00 6.17680848e-01 -3.06298196e-01
6.85748577e-01 -3.21601599e-01 -1.01836336e+00 5.45468390e-01
1.31804431e-02 4.66973364e-01 6.53862834e-01 -4.47836012e-01
-4.66821976e-02 2.55174905e-01 -1.33854890e+00 2.10714608e-01
9.78750527e-01 -2.87887573e-01 -7.10236847e-01 4.73733187e-01
3.39523137e-01 -7.12845027e-01 -9.31236506e-01 7.35379219e-01
3.89217615e-01 -1.01783550e+00 1.18972766e+00 -5.79519033e-01
4.90604490e-01 -3.32535803e-02 2.87683219e-01 -1.19046223e+00
-3.97291660e-01 -1.97051853e-01 2.84244478e-01 4.08857942e-01
4.67980683e-01 -5.40620804e-01 8.15848470e-01 4.85443860e-01
-7.02548981e-01 -1.29145718e+00 -1.18866658e+00 -2.08662301e-01
4.06946987e-01 -5.31864107e-01 2.09699437e-01 6.05168581e-01
-3.35090756e-01 -1.70945570e-01 2.21681386e-01 1.95791632e-01
6.26742125e-01 -2.32688278e-01 1.56734392e-01 -1.22857976e+00
-4.23374698e-02 -8.48551154e-01 -5.23892224e-01 -7.80960500e-01
8.16308782e-02 -1.32022965e+00 -6.43296614e-02 -1.75172853e+00
2.23762393e-01 -4.33817625e-01 -4.97459739e-01 6.13697231e-01
-1.55150414e-01 6.76353455e-01 -1.48316547e-01 1.70397341e-01
-3.12435150e-01 9.30080637e-02 1.36156631e+00 -1.84397459e-01
-1.80492967e-01 1.31450027e-01 -4.97254401e-01 7.98757434e-01
8.40450287e-01 -5.37418723e-01 -5.31522855e-02 -6.64960384e-01
-3.12617749e-01 -2.99505383e-01 7.35579133e-01 -1.05881596e+00
1.72098607e-01 4.20579910e-01 6.89936757e-01 -6.68999255e-01
6.33811876e-02 -7.84492850e-01 -2.82001048e-01 7.70793259e-01
-3.29898179e-01 -6.68560416e-02 4.47149456e-01 8.37745816e-02
-6.91117942e-02 -9.82867107e-02 1.23869419e+00 -3.24618101e-01
-4.89775330e-01 6.41315401e-01 -6.49778306e-01 -1.21600224e-05
1.08163619e+00 -3.11476678e-01 3.16842906e-02 3.49302083e-01
-1.06141186e+00 8.60760063e-02 -4.37067300e-02 2.56032884e-01
5.76386273e-01 -9.59347427e-01 -9.06599462e-01 3.31472218e-01
-1.04549766e-01 2.51892298e-01 5.93209028e-01 1.53975034e+00
-7.14415848e-01 3.94490093e-01 -6.05973244e-01 -1.00448680e+00
-9.50830221e-01 5.85332885e-02 8.36243808e-01 -6.34896517e-01
-1.11326909e+00 8.69988203e-01 1.86478868e-01 -4.04466450e-01
3.55855346e-01 -6.97583973e-01 -4.89078611e-01 -2.07381342e-02
5.90030193e-01 9.93777737e-02 7.93026388e-01 -6.71994567e-01
-6.04028761e-01 4.38284814e-01 -2.35995814e-01 6.86263293e-02
1.69153857e+00 9.00292471e-02 -2.63332039e-01 -2.98699401e-02
1.33217323e+00 -5.90205789e-01 -1.28767467e+00 -2.51399845e-01
1.24836244e-01 4.28517759e-02 4.57791895e-01 -9.73056316e-01
-1.35308385e+00 1.02735412e+00 1.06989539e+00 -3.61587219e-02
1.18317342e+00 7.80182183e-02 8.15210640e-01 -1.10635810e-01
-1.05091140e-01 -8.85563910e-01 -2.24875718e-01 5.89845300e-01
7.27265060e-01 -1.29254913e+00 -9.89359096e-02 -1.91401064e-01
-3.23036462e-01 1.27836871e+00 4.34247285e-01 -2.13304937e-01
8.89723241e-01 4.71248269e-01 8.71997699e-02 -5.84952056e-01
-2.85130680e-01 -2.26721913e-01 4.81062412e-01 5.56950867e-01
3.68741423e-01 4.76858839e-02 -2.95389742e-01 6.68448389e-01
2.30722919e-01 5.49202040e-02 3.29482853e-01 9.58465338e-01
-3.54708523e-01 -9.13016200e-01 -2.00293750e-01 6.48958683e-01
-8.49471986e-01 -1.54908672e-01 -2.30758607e-01 6.39648080e-01
3.58688772e-01 3.80245805e-01 -2.57317536e-02 4.66778018e-02
5.73709965e-01 5.07398658e-02 5.40072560e-01 -5.72180510e-01
-1.00514853e+00 4.84220795e-02 -2.37873673e-01 -7.13940799e-01
-5.67343056e-01 -8.25302303e-01 -1.41544020e+00 4.22292709e-01
1.10371947e-01 -4.18064505e-01 8.13921750e-01 1.26163208e+00
3.11069697e-01 8.90010715e-01 3.35117906e-01 -1.22566867e+00
-3.17233741e-01 -8.71526957e-01 -2.01777950e-01 3.27487618e-01
3.82474482e-01 -7.25092232e-01 1.89761687e-02 -1.90991275e-02] | [14.327427864074707, -2.2741119861602783] |
3dcad040-7a24-45d3-9a8c-5701c7695fe7 | multi-task-knowledge-enhancement-for-zero | 2306.06302 | null | https://arxiv.org/abs/2306.06302v1 | https://arxiv.org/pdf/2306.06302v1.pdf | Multi-Task Knowledge Enhancement for Zero-Shot and Multi-Domain Recommendation in an AI Assistant Application | Recommender systems have found significant commercial success but still struggle with integrating new users. Since users often interact with content in different domains, it is possible to leverage a user's interactions in previous domains to improve that user's recommendations in a new one (multi-domain recommendation). A separate research thread on knowledge graph enhancement uses external knowledge graphs to improve single domain recommendations (knowledge graph enhancement). Both research threads incorporate related information to improve predictions in a new domain. We propose in this work to unify these approaches: Using information from interactions in other domains as well as external knowledge graphs to make predictions in a new domain that would be impossible with either information source alone. We apply these ideas to a dataset derived from millions of users' requests for content across three domains (videos, music, and books) in a live virtual assistant application. We demonstrate the advantage of combining knowledge graph enhancement with previous multi-domain recommendation techniques to provide better overall recommendations as well as for better recommendations on new users of a domain. | ['Aram Galstyan', 'Greg Ver Steeg', 'Tony Chen', 'Xing Fan', 'Fan Yang', 'Ziyan Jiang', 'Elan Markowitz'] | 2023-06-09 | null | null | null | null | ['knowledge-graphs'] | ['knowledge-base'] | [ 1.20914117e-01 2.58797556e-01 -5.17913878e-01 -2.97757775e-01
-1.88466281e-01 -8.40420723e-01 5.05906522e-01 2.94316262e-01
2.61310134e-02 6.76252186e-01 6.94126070e-01 -2.21228540e-01
-6.00612462e-01 -8.71604919e-01 -3.57107311e-01 8.86393636e-02
1.58894271e-01 8.50358605e-01 7.00048983e-01 -8.08470190e-01
6.03762388e-01 4.14613754e-01 -1.60279965e+00 6.53684199e-01
1.04751897e+00 3.26470852e-01 3.97187501e-01 5.32747984e-01
-4.08763289e-01 5.76269388e-01 -3.68286550e-01 -6.38986230e-01
3.71029347e-01 -7.90875852e-02 -9.15153205e-01 1.34127736e-01
5.39400160e-01 -4.85218197e-01 -5.52903593e-01 6.11843646e-01
2.20848665e-01 8.69924784e-01 6.28086865e-01 -1.13852739e+00
-1.07779932e+00 8.34528446e-01 -3.74319822e-01 3.06751013e-01
8.01053941e-01 -4.82657820e-01 1.09321856e+00 -6.55514598e-01
1.08727193e+00 1.05157685e+00 6.60008311e-01 3.14368755e-01
-1.12948215e+00 -4.41262484e-01 5.76972604e-01 2.82507569e-01
-7.95515895e-01 -2.44541824e-01 6.78650558e-01 -5.21697521e-01
1.03808689e+00 5.86560555e-02 5.80256462e-01 9.55113292e-01
-5.10779262e-01 6.29569411e-01 4.81408596e-01 -3.24642748e-01
-4.84109595e-02 7.54440308e-01 5.42449892e-01 2.79257208e-01
3.19524497e-01 -7.43011758e-02 -6.50181115e-01 -4.75317001e-01
7.12553442e-01 3.52426261e-01 -2.14518398e-01 -5.80491662e-01
-8.80436897e-01 7.82587528e-01 2.08512232e-01 5.22969186e-01
-5.04397810e-01 -7.18829632e-01 7.67873898e-02 6.49264097e-01
4.95894849e-01 1.22185743e+00 -8.31983209e-01 -3.00109804e-01
-6.63233280e-01 4.99127716e-01 1.37987328e+00 1.11403751e+00
7.92344928e-01 -3.06012362e-01 -2.47173961e-02 1.10688102e+00
2.78159350e-01 6.48065582e-02 5.25076985e-01 -1.01980925e+00
2.66075283e-01 9.17552352e-01 3.28022867e-01 -1.30811560e+00
-3.14248145e-01 -6.19323909e-01 -3.98113616e-02 -6.90739453e-02
5.04049480e-01 -2.91884840e-01 -6.56584918e-01 1.31163204e+00
3.55414718e-01 2.48830944e-01 -3.73135135e-02 7.20195115e-01
1.05339801e+00 4.47819769e-01 -1.28627032e-01 -1.46641076e-01
8.86684477e-01 -1.18288291e+00 -4.80930239e-01 -1.63052112e-01
7.14418590e-01 -1.03251576e+00 8.92762303e-01 8.90369058e-01
-7.98902333e-01 -8.05575192e-01 -8.18888485e-01 3.65848616e-02
-7.20166206e-01 -1.89492032e-01 6.40129864e-01 5.99881053e-01
-1.00795424e+00 1.11018240e+00 -1.30257398e-01 -8.71679366e-01
9.87169221e-02 5.26417315e-01 -2.07320228e-01 -3.85126859e-01
-1.24193966e+00 1.01045895e+00 6.61858320e-02 -9.47222710e-01
-2.35288039e-01 -1.04417157e+00 -3.08897346e-01 4.44191396e-02
7.26118505e-01 -5.84249258e-01 1.26631761e+00 -1.08008158e+00
-1.29172671e+00 1.73993841e-01 7.77061209e-02 1.64828345e-03
4.17039432e-02 -6.13189936e-01 -6.69203937e-01 -3.67119968e-01
-5.87812401e-02 2.13914756e-02 6.87549829e-01 -1.22375655e+00
-1.01432860e+00 -4.64397699e-01 5.79521358e-01 5.71702659e-01
-8.31577957e-01 -2.30503350e-01 -7.56040335e-01 -6.41630113e-01
-1.83178365e-01 -8.37870717e-01 -2.45902047e-01 -5.60180426e-01
1.90299407e-01 -2.59875387e-01 1.02481675e+00 -8.30560267e-01
1.37822771e+00 -1.85187602e+00 4.06307966e-01 6.96032941e-01
6.02345586e-01 5.15754700e-01 -4.39337552e-01 6.86722100e-01
1.89902425e-01 1.03364922e-01 7.11599469e-01 1.07266828e-01
-1.04201622e-01 3.07326436e-01 -4.32339087e-02 -2.54104137e-01
-3.27730358e-01 4.71662134e-01 -1.15184653e+00 2.67992280e-02
7.47891292e-02 4.61481541e-01 -1.04504645e+00 2.78848615e-02
-3.30159307e-01 3.44130009e-01 -6.21572077e-01 3.04767996e-01
5.56387007e-01 -5.37196040e-01 7.04954088e-01 -2.90258676e-01
2.17729494e-01 3.85351658e-01 -1.41867065e+00 1.61311686e+00
-3.95499557e-01 2.94737786e-01 -1.78783461e-01 -8.78024757e-01
1.00572479e+00 1.30969258e-02 6.73206806e-01 -5.21830738e-01
-8.21934938e-02 -4.10191528e-02 -5.05244322e-02 -4.71541226e-01
1.01134837e+00 2.42055982e-01 3.38344485e-01 7.36923993e-01
4.08725768e-01 2.78708279e-01 4.64014083e-01 7.60479867e-01
1.51431489e+00 2.83526480e-01 3.85790437e-01 3.11832246e-03
2.13299021e-01 1.87047318e-01 2.17599511e-01 1.02970755e+00
2.47804195e-01 1.92624673e-01 1.05635419e-01 -1.71297297e-01
-8.11570346e-01 -8.30898643e-01 3.50301147e-01 1.92795312e+00
1.98883638e-01 -9.54893827e-01 -3.89074460e-02 -1.14600503e+00
4.85979944e-01 7.58693635e-01 -4.34729367e-01 -2.03419849e-01
-2.99732685e-01 -1.54037938e-01 -1.67802289e-01 5.65604031e-01
6.42091706e-02 -6.24269903e-01 3.38520974e-01 4.01659787e-01
-1.57446578e-01 -9.06260610e-01 -3.57018709e-01 -8.70035291e-02
-1.00761557e+00 -1.09679389e+00 -7.24023998e-01 -5.32901525e-01
5.42007923e-01 7.85864890e-01 1.41347098e+00 1.97336242e-01
2.33340919e-01 9.97005224e-01 -1.09494233e+00 -5.96711077e-02
-4.59726661e-01 2.94835269e-01 3.66991431e-01 -3.53083372e-01
7.06416249e-01 -9.47861373e-01 -3.53570849e-01 8.82018149e-01
-6.28833115e-01 -2.20213905e-01 3.35113287e-01 6.43747687e-01
7.61259422e-02 2.30036452e-01 7.68532693e-01 -1.67338908e+00
1.04424012e+00 -1.01476169e+00 -2.56053865e-01 3.21861953e-01
-1.13258123e+00 -5.86732142e-02 5.04593730e-01 -7.24754393e-01
-1.45277619e+00 -9.93593261e-02 1.54177263e-01 -4.62075025e-02
-1.31293252e-01 8.43121827e-01 2.56994426e-01 -1.79112867e-01
1.28345084e+00 -3.07654023e-01 -3.09196085e-01 -8.80966961e-01
7.60728002e-01 6.82755053e-01 2.96206415e-01 -6.55491054e-01
7.55451560e-01 -4.08590548e-02 -4.88383651e-01 -5.67393720e-01
-8.18478107e-01 -1.14110506e+00 -8.34580660e-01 -2.45796859e-01
7.81885237e-02 -6.45028353e-01 -5.14457762e-01 -2.98086852e-01
-7.83791542e-01 -1.27406746e-01 -3.98930192e-01 5.18035650e-01
-1.24938041e-01 5.98106146e-01 -2.12633550e-01 -3.72019112e-01
1.09589333e-02 -6.32637262e-01 3.91396910e-01 5.85270882e-01
-4.47637141e-01 -1.30206239e+00 3.60420227e-01 7.09336102e-01
6.41652703e-01 -4.52337176e-01 7.19085217e-01 -1.46571529e+00
-2.71308511e-01 -4.14260447e-01 -3.53120834e-01 2.59387553e-01
3.82141411e-01 -1.07355289e-01 -6.90130949e-01 -2.97609627e-01
-5.18807530e-01 1.69562533e-01 4.99542534e-01 1.06132757e-02
6.26547098e-01 -2.46219561e-01 -8.02875042e-01 -1.28130704e-01
1.07866085e+00 3.48472297e-01 5.41972995e-01 2.45026052e-01
7.13029981e-01 5.79429626e-01 8.73218298e-01 8.55207026e-01
4.33743805e-01 9.45871115e-01 -9.57121029e-02 7.54035413e-02
-2.92651594e-01 -2.20747784e-01 3.22735161e-01 6.06540084e-01
-6.21321380e-01 -4.71973956e-01 -7.62583256e-01 5.20212412e-01
-2.10677266e+00 -9.30856466e-01 -1.18157983e-01 2.24660039e+00
4.92450804e-01 -1.45648569e-02 5.54516196e-01 -1.87401474e-01
5.95889032e-01 -4.40764993e-01 -6.45804763e-01 -3.85463238e-01
2.42644548e-01 2.83388704e-01 3.98027003e-01 3.00486952e-01
-6.99243963e-01 8.77859950e-01 6.38342094e+00 5.03150582e-01
-4.25348103e-01 -1.64443124e-02 -7.78238699e-02 -5.06844111e-02
-4.12060380e-01 2.28247657e-01 -1.02064562e+00 8.61870050e-02
8.69911432e-01 -6.19117558e-01 8.01388323e-01 1.12099385e+00
-2.03315929e-01 -8.84537250e-02 -1.07663655e+00 7.18279362e-01
2.08265185e-01 -1.44903505e+00 9.38296616e-02 1.15313344e-01
9.38308716e-01 1.13174438e-01 4.93402872e-03 7.49934912e-01
1.04570413e+00 -3.24036986e-01 -4.36397642e-01 6.86515629e-01
6.75021484e-02 -5.54042459e-01 6.97938263e-01 3.26839983e-01
-9.20629740e-01 -2.11104691e-01 -2.99840808e-01 -3.53207618e-01
3.61012332e-02 4.84895825e-01 -1.43476295e+00 7.31749654e-01
7.50516117e-01 1.28687775e+00 -5.02138197e-01 1.07207787e+00
1.27711490e-01 4.65266764e-01 -2.88303971e-01 9.72575396e-02
-2.85221845e-01 -1.57770649e-01 6.10833466e-01 1.05484211e+00
5.12311399e-01 3.80173117e-01 3.81038964e-01 3.87234807e-01
-2.47483492e-01 4.88447011e-01 -7.74995029e-01 -3.78742754e-01
4.34212714e-01 1.38869834e+00 -2.99824029e-01 -6.13429546e-01
-9.51323509e-01 9.95945811e-01 3.03455621e-01 4.14212704e-01
-3.76204252e-01 -2.84203708e-01 8.16520870e-01 7.15525091e-01
3.31259936e-01 -2.07040071e-01 9.01433825e-02 -1.17666805e+00
-4.42351967e-01 -9.76576269e-01 7.55118430e-01 -7.41188407e-01
-1.83200014e+00 2.58766741e-01 4.96946312e-02 -1.28258467e+00
-1.95412204e-01 -5.28561711e-01 -1.77795291e-01 7.04381108e-01
-1.38011587e+00 -8.38170767e-01 -3.67908955e-01 7.83656836e-01
4.34903026e-01 -5.69240332e-01 7.40109622e-01 6.66312873e-01
-4.13894840e-02 5.08794606e-01 4.70078081e-01 -3.56617987e-01
1.33505309e+00 -1.08448541e+00 2.43626833e-01 4.80297327e-01
3.44021440e-01 8.74967754e-01 5.76748848e-01 -1.12962806e+00
-1.26241672e+00 -7.51301706e-01 7.52520680e-01 -7.63005733e-01
6.77271068e-01 3.52947325e-01 -1.01911163e+00 8.13031197e-01
1.23979062e-01 -5.28985798e-01 1.25298345e+00 1.00355172e+00
-4.78771508e-01 1.09648265e-01 -1.15430450e+00 4.94362563e-01
1.44126594e+00 -9.17151049e-02 -8.46854687e-01 5.10877848e-01
4.90700632e-01 -2.89847553e-01 -1.32977843e+00 2.77925789e-01
6.58911526e-01 -7.68480957e-01 1.21752763e+00 -1.01647830e+00
2.24704131e-01 -1.47679523e-01 -2.08446845e-01 -1.66470909e+00
-8.37252021e-01 -7.15261877e-01 -4.45182055e-01 1.26809537e+00
5.36895275e-01 -5.81198454e-01 8.93160045e-01 1.04411077e+00
-1.56985134e-01 -1.44210141e-02 -4.66628671e-02 -6.27204061e-01
-3.18659276e-01 -2.37752333e-01 6.20627403e-01 1.25275528e+00
7.78279781e-01 6.31535709e-01 -6.35901511e-01 5.30909188e-02
2.22108379e-01 1.62914917e-01 1.26814675e+00 -2.01759028e+00
-7.25476444e-01 -2.22450629e-01 -1.78919777e-01 -1.32952344e+00
-2.71061510e-01 -1.06756067e+00 -4.83116388e-01 -1.92634881e+00
2.83317834e-01 -6.56545699e-01 -7.13390648e-01 5.99574983e-01
-2.10230872e-02 9.15073529e-02 1.55742943e-01 3.07561308e-01
-8.86076987e-01 -8.36348608e-02 1.42276752e+00 5.09639159e-02
-7.59027541e-01 2.42540628e-01 -1.28254664e+00 7.26660311e-01
5.78572392e-01 -3.95396560e-01 -8.12920213e-01 -1.27251580e-01
5.83257139e-01 -4.09203246e-02 -3.42586756e-01 -9.51278687e-01
3.21349323e-01 -1.87617093e-01 4.89246458e-01 -3.67419153e-01
3.92823249e-01 -9.35386598e-01 2.69749910e-01 -1.99567586e-01
-3.36707979e-01 -2.61914074e-01 2.91139007e-01 9.26570058e-01
1.28029510e-01 -1.90694854e-01 1.79244414e-01 -8.57867450e-02
-1.26023793e+00 1.31160274e-01 -4.04566109e-01 -2.00899556e-01
8.99583697e-01 -4.38688874e-01 -5.88203549e-01 -7.48273969e-01
-1.36678267e+00 6.44914657e-02 4.16875154e-01 9.45160508e-01
5.97025275e-01 -1.14485073e+00 -3.90614301e-01 4.90118153e-02
3.17464799e-01 -9.07064676e-01 1.64082170e-01 5.88398159e-01
1.97257474e-01 4.44683507e-02 -5.15460312e-01 -4.55194265e-02
-1.56897306e+00 6.06004953e-01 -2.48680174e-01 -4.54467863e-01
-3.51778358e-01 7.49234200e-01 -6.95927590e-02 -7.30262816e-01
1.40087202e-01 1.19319916e-01 -8.90148938e-01 3.05689842e-01
5.00405252e-01 6.90413058e-01 3.45018953e-02 -2.34282345e-01
-4.40805554e-02 3.64903778e-01 -7.32400358e-01 1.28256992e-01
1.61762404e+00 -3.68006349e-01 2.82000273e-01 1.05380841e-01
6.03228867e-01 4.44426924e-01 -7.40055025e-01 -9.59176600e-01
-9.57754068e-03 -9.86142039e-01 6.96453229e-02 -1.32178903e+00
-1.04903018e+00 2.50287175e-01 2.80946702e-01 4.40350920e-01
8.84723008e-01 1.22308522e-01 7.93842554e-01 6.95828319e-01
3.47150534e-01 -1.36661077e+00 3.96024674e-01 5.59406459e-01
7.30725586e-01 -1.18914282e+00 4.95696545e-01 -4.17359769e-01
-1.04534519e+00 1.34210300e+00 7.15863407e-01 1.39099941e-01
1.15647602e+00 -1.59017920e-01 -2.08741948e-01 -2.10155278e-01
-7.43362963e-01 -4.52549666e-01 7.22017229e-01 1.03601599e+00
7.50700951e-01 -7.27187544e-02 -3.09616029e-01 9.31431353e-01
2.68579513e-01 4.69920248e-01 6.06877029e-01 9.30427849e-01
-5.73863328e-01 -1.78754461e+00 4.31290977e-02 1.16053581e+00
-1.75836333e-03 -4.48840782e-02 -6.23839855e-01 4.50784951e-01
6.50910288e-02 1.25186980e+00 -4.23368424e-01 -9.86706138e-01
5.63529611e-01 1.70293245e-02 3.84460598e-01 -1.14171517e+00
-7.24878430e-01 6.21268712e-02 6.03551030e-01 -5.11489570e-01
-4.29728299e-01 -7.37571478e-01 -1.00535679e+00 -6.05306864e-01
-6.02891982e-01 2.79437929e-01 5.94139457e-01 7.32632041e-01
9.33881700e-01 5.60438514e-01 2.34646752e-01 -6.82405055e-01
-2.07658842e-01 -8.18872929e-01 -7.09264934e-01 6.40230596e-01
-2.97823220e-01 -1.07851017e+00 1.62160043e-02 2.02324435e-01] | [10.02224063873291, 5.759999752044678] |
2455bac0-3d78-4c8e-b2b7-1264645fc8c2 | personalized-federated-learning-via-gradient | 2304.11524 | null | https://arxiv.org/abs/2304.11524v1 | https://arxiv.org/pdf/2304.11524v1.pdf | Personalized Federated Learning via Gradient Modulation for Heterogeneous Text Summarization | Text summarization is essential for information aggregation and demands large amounts of training data. However, concerns about data privacy and security limit data collection and model training. To eliminate this concern, we propose a federated learning text summarization scheme, which allows users to share the global model in a cooperative learning manner without sharing raw data. Personalized federated learning (PFL) balances personalization and generalization in the process of optimizing the global model, to guide the training of local models. However, multiple local data have different distributions of semantics and context, which may cause the local model to learn deviated semantic and context information. In this paper, we propose FedSUMM, a dynamic gradient adapter to provide more appropriate local parameters for local model. Simultaneously, FedSUMM uses differential privacy to prevent parameter leakage during distributed training. Experimental evidence verifies FedSUMM can achieve faster model convergence on PFL algorithm for task-specific text summarization, and the method achieves superior performance for different optimization metrics for text summarization. | ['Jing Xiao', 'Zhangcheng Huang', 'Lingwei Kong', 'Jianzong Wang', 'Rongfeng Pan'] | 2023-04-23 | null | null | null | null | ['personalized-federated-learning', 'text-summarization'] | ['methodology', 'natural-language-processing'] | [ 2.11373270e-02 8.96882564e-02 -5.70267379e-01 -6.04961872e-01
-1.17599678e+00 -5.54864168e-01 2.03052431e-01 3.28338236e-01
-4.15702403e-01 9.26545203e-01 7.55956531e-01 7.69903213e-02
-8.38637501e-02 -5.88154376e-01 -5.35851955e-01 -6.38202369e-01
5.83909154e-02 2.91089028e-01 -1.06284223e-01 8.66794661e-02
2.87966758e-01 -1.23086445e-01 -1.07288384e+00 6.31970108e-01
1.54960215e+00 8.64750266e-01 3.37259889e-01 7.60894418e-01
-3.67950052e-01 7.79702723e-01 -9.82547522e-01 -5.62643945e-01
1.99929133e-01 -4.16245013e-01 -9.03195381e-01 -2.90438116e-01
7.75876045e-01 -5.99327147e-01 -3.64752024e-01 1.17018449e+00
8.57150078e-01 2.82236844e-01 2.48102039e-01 -1.18682492e+00
-6.08229935e-01 1.02826190e+00 -3.52779835e-01 -1.13257557e-01
7.96618462e-02 -2.76964437e-02 1.16606939e+00 -2.52417088e-01
4.42279428e-01 9.82106507e-01 5.34815788e-01 7.13209331e-01
-9.57377851e-01 -5.49056053e-01 4.92329836e-01 -1.45425845e-04
-9.64596331e-01 -4.31638181e-01 5.41626453e-01 2.69488335e-01
6.98383093e-01 9.67691898e-01 5.60737252e-01 7.79960513e-01
3.60891312e-01 1.52443802e+00 6.09837234e-01 1.28948331e-01
6.60422623e-01 2.42440104e-01 3.30384284e-01 3.72457772e-01
7.12261498e-01 -6.63909018e-01 -1.08994734e+00 -8.59386683e-01
-1.42883152e-01 3.39986026e-01 -4.25734788e-01 -2.72394389e-01
-8.31070006e-01 6.24898672e-01 3.65922719e-01 -7.92411342e-02
-3.21549535e-01 3.77017818e-03 7.32637167e-01 4.31837767e-01
8.81028950e-01 3.02166760e-01 -8.77985001e-01 -3.80439870e-02
-1.03277993e+00 4.05750722e-01 1.08907044e+00 1.09955168e+00
9.55035925e-01 -3.17725331e-01 -5.35532355e-01 8.88525486e-01
1.49853975e-01 5.53086340e-01 1.08198488e+00 -1.11472297e+00
9.03202951e-01 8.50037217e-01 -1.65353268e-01 -8.95467758e-01
-1.08618855e-01 -3.49818468e-01 -1.11978388e+00 -7.06912816e-01
-9.29046050e-02 -7.00593412e-01 -2.93681175e-01 1.52391303e+00
5.45367360e-01 -1.19837858e-02 4.03708309e-01 6.89850867e-01
6.86494589e-01 7.85980999e-01 -6.12110272e-02 -4.63601559e-01
1.00527275e+00 -1.01978517e+00 -7.69383013e-01 -2.01940730e-01
1.12793779e+00 -2.47263610e-01 1.03741574e+00 1.55860916e-01
-1.11261714e+00 1.88327089e-01 -8.04436743e-01 -2.40353867e-01
-1.00987457e-01 -1.21457040e-01 7.59571135e-01 6.41128004e-01
-9.68976676e-01 5.96684217e-01 -7.67211020e-01 -3.30548495e-01
9.17581379e-01 4.90334272e-01 -9.53789130e-02 -2.22547352e-02
-1.00485599e+00 1.59435943e-01 4.70880091e-01 -3.09318781e-01
-3.20157886e-01 -9.99411941e-01 -5.30630469e-01 4.07527179e-01
2.32054338e-01 -1.22646439e+00 1.46630585e+00 -6.43228173e-01
-1.46455538e+00 4.16892499e-01 -3.08602035e-01 -7.48185635e-01
6.24101043e-01 -3.76034260e-01 -5.87813072e-02 -9.09409225e-02
7.42329867e-04 1.81213588e-01 6.39890313e-01 -1.11672711e+00
-9.42229390e-01 -8.08083832e-01 -4.54998732e-01 5.77908278e-01
-8.86309862e-01 -2.58225322e-01 -3.16222638e-01 -5.25023103e-01
-6.39913529e-02 -4.30101693e-01 -3.87379348e-01 -5.27341485e-01
-6.48472786e-01 -2.24217921e-01 1.06789458e+00 -8.16325307e-01
1.67505455e+00 -1.89709258e+00 -1.51324332e-01 2.26742044e-01
3.85135174e-01 3.64124089e-01 1.70312244e-02 5.03301203e-01
7.30791986e-01 2.87593186e-01 -4.73458827e-01 -7.98422456e-01
1.15612999e-01 1.78046122e-01 -4.22488779e-01 2.32759088e-01
-7.45802999e-01 1.01352036e+00 -7.18245447e-01 -7.76959300e-01
-2.09522828e-01 -1.84005145e-02 -7.24778235e-01 2.37213090e-01
-4.24926072e-01 9.19896662e-02 -1.07748044e+00 3.56272906e-01
8.45017374e-01 -2.03371078e-01 4.09750491e-01 -1.56597421e-02
1.98933691e-01 1.86359152e-01 -9.79306221e-01 2.00067329e+00
-4.50335085e-01 1.67066202e-01 3.38910043e-01 -8.38664651e-01
7.08161175e-01 1.35208458e-01 6.42510355e-01 -5.57846367e-01
1.00008816e-01 1.67739943e-01 -7.77665913e-01 -5.11682034e-01
9.47093427e-01 5.81217110e-01 -2.80027509e-01 1.09938729e+00
-1.13001041e-01 1.50338352e-01 -3.73452723e-01 7.59106755e-01
9.06148911e-01 -4.57553685e-01 2.53898084e-01 -2.83044130e-01
3.59827995e-01 -1.57682598e-01 7.48230577e-01 9.54184890e-01
-1.24817103e-01 3.24555784e-01 3.22441548e-01 -3.81321460e-01
-5.22382319e-01 -5.78233480e-01 2.22804591e-01 1.38593316e+00
2.67117143e-01 -7.85139263e-01 -1.10522950e+00 -1.11471772e+00
1.74698412e-01 9.08776045e-01 -1.23339675e-01 -5.07400811e-01
-5.07045031e-01 -9.59825933e-01 4.30995822e-01 1.15117781e-01
8.90111506e-01 -7.43008614e-01 -4.11778092e-01 1.32595822e-01
-5.40228188e-01 -4.96079326e-01 -1.24306023e+00 -2.19691947e-01
-1.15585172e+00 -8.31366420e-01 -3.40134919e-01 -5.60339272e-01
6.90610111e-01 5.75547338e-01 7.39582419e-01 -7.94524699e-02
4.43501957e-03 5.63341320e-01 -7.13468492e-02 -5.32974839e-01
-3.68713528e-01 6.70085907e-01 -1.44906387e-01 2.85578281e-01
3.36222678e-01 -5.67151785e-01 -9.50251281e-01 1.29386159e-02
-1.02313340e+00 6.02647625e-02 3.00628394e-01 7.69230962e-01
4.69957024e-01 -2.21848711e-01 8.44124436e-01 -1.32075989e+00
1.28183210e+00 -6.20209217e-01 -2.30575144e-01 8.41029704e-01
-1.02046847e+00 1.48161724e-01 9.70352769e-01 -1.53271481e-01
-1.39685059e+00 -7.80266300e-02 3.86408955e-01 6.37877807e-02
1.91509992e-01 4.27004606e-01 -6.63450956e-01 5.95158599e-02
6.91363275e-01 4.91099507e-01 2.72077739e-01 -6.17088497e-01
4.28019047e-01 1.25813103e+00 3.46240550e-01 -5.28687835e-01
1.66947260e-01 4.44045722e-01 -4.75395322e-01 -6.40967250e-01
-8.26620579e-01 -5.58793604e-01 -1.23816401e-01 2.62777030e-01
3.43645871e-01 -9.79269385e-01 -6.53411686e-01 6.18120313e-01
-9.20633316e-01 -2.74639338e-01 -6.37266397e-01 2.65263110e-01
-4.96297657e-01 6.55483127e-01 -3.87288779e-01 -5.47363102e-01
-1.35438335e+00 -5.60250759e-01 8.88776660e-01 4.34312493e-01
-1.96984738e-01 -1.12313747e+00 1.53716907e-01 6.12264276e-01
4.76057887e-01 -2.33366892e-01 5.76734066e-01 -1.09134126e+00
-6.37714386e-01 -5.26733935e-01 1.51585385e-01 4.42103744e-01
3.64192605e-01 -3.94920170e-01 -9.28187132e-01 -6.19809687e-01
3.24886829e-01 -3.76737446e-01 9.47931647e-01 4.06460971e-01
1.48213542e+00 -1.39975977e+00 -3.34776402e-01 9.48579252e-01
1.14220405e+00 -4.63920981e-01 2.28793979e-01 9.07012969e-02
8.31370175e-01 5.22374153e-01 6.02345467e-01 9.24316883e-01
7.42596507e-01 1.48597211e-01 2.25698724e-01 3.95018250e-01
2.70922959e-01 -6.25693619e-01 4.34336156e-01 9.01943982e-01
6.06456339e-01 -4.12679642e-01 -2.28150710e-01 3.04070473e-01
-2.38110328e+00 -9.33579028e-01 1.60446122e-01 2.46703720e+00
1.19217873e+00 -6.75789535e-01 3.99772823e-02 -5.07040024e-01
6.17065787e-01 3.90433937e-01 -1.13635051e+00 -2.98328191e-01
-2.84907430e-01 -4.04232115e-01 9.84204650e-01 3.75055730e-01
-6.94174230e-01 9.51325238e-01 5.96693420e+00 1.09027851e+00
-1.07729495e+00 2.53138691e-01 7.86312103e-01 -6.32824481e-01
-7.76648700e-01 3.16917077e-02 -7.42468715e-01 8.48415792e-01
7.46175110e-01 -9.88826454e-01 4.60233092e-01 8.79835606e-01
1.69985175e-01 -5.77135831e-02 -7.75973618e-01 9.82681215e-01
1.79125279e-01 -1.50328350e+00 4.68397230e-01 1.24276027e-01
1.12567294e+00 2.20831379e-01 -7.10643083e-02 1.08373575e-01
6.64352179e-01 -5.62158585e-01 1.55706078e-01 5.79712808e-01
3.39122593e-01 -8.94560456e-01 4.14946109e-01 8.76296341e-01
-6.94909215e-01 -3.51638794e-01 -6.47868693e-01 4.13829893e-01
-1.61943123e-01 5.06104946e-01 -7.25088418e-01 8.21335852e-01
5.32332242e-01 6.13247991e-01 -6.08264148e-01 9.34537530e-01
2.59933770e-01 7.17012763e-01 -4.39610958e-01 -2.59266227e-01
-7.23691285e-02 -3.16690177e-01 5.91442645e-01 1.06064916e+00
2.19391704e-01 -1.09961852e-01 3.16416055e-01 3.30901533e-01
-5.41239619e-01 6.38368726e-01 -4.14070845e-01 3.65821123e-01
8.74621332e-01 1.23552763e+00 1.64716139e-01 -3.84649932e-01
-9.53494385e-02 1.08104300e+00 4.67030287e-01 3.64690155e-01
-3.11684728e-01 -2.06842303e-01 7.07993925e-01 -2.57747918e-01
-2.42274761e-01 1.57061189e-01 -5.47301412e-01 -1.48453164e+00
3.10520947e-01 -9.71896827e-01 1.14367509e+00 -1.61293402e-01
-1.35782361e+00 5.44368625e-02 -6.02137782e-02 -8.30341995e-01
-1.33901298e-01 3.86563808e-01 -1.14400041e+00 6.48028910e-01
-1.17164075e+00 -1.26448584e+00 -2.25542232e-01 9.35212672e-01
4.61354315e-01 -3.82861912e-01 6.80920303e-01 -1.58265427e-01
-7.03766644e-01 1.26262438e+00 8.68008614e-01 -3.22347522e-01
1.06312442e+00 -1.12260747e+00 2.69823998e-01 6.34874761e-01
-1.41832471e-01 5.82580864e-01 4.52019066e-01 -8.32742691e-01
-1.55982006e+00 -1.46803737e+00 9.19710100e-01 -2.04424649e-01
1.89890802e-01 -2.48951852e-01 -1.02440631e+00 6.57255709e-01
1.92697912e-01 -2.53038168e-01 1.05055201e+00 1.91089988e-01
-5.51648736e-02 -5.46130538e-01 -1.58662677e+00 5.83233297e-01
9.33685482e-01 -3.34995627e-01 -2.20467508e-01 7.61543214e-01
1.06714296e+00 -1.90937743e-01 -8.50379884e-01 -1.04007587e-01
2.99067706e-01 -7.98529625e-01 5.29527664e-01 -7.53558576e-01
-1.80764154e-01 6.82855546e-02 -1.19794562e-01 -1.53096092e+00
-3.23316902e-02 -1.25645387e+00 -7.43263006e-01 1.45976961e+00
2.84568191e-01 -1.18767154e+00 1.11973619e+00 1.11357880e+00
-1.26677051e-01 -7.14615464e-01 -8.99382949e-01 -4.87550914e-01
1.79598987e-01 -5.15742041e-02 1.24312985e+00 9.23184693e-01
3.56354415e-01 2.60162532e-01 -4.32771534e-01 2.28355937e-02
9.50397849e-01 2.30109110e-01 1.09799933e+00 -9.84452605e-01
-4.27466491e-03 -3.57593834e-01 3.78414318e-02 -1.26401389e+00
3.27739775e-01 -1.09788632e+00 -3.78370970e-01 -1.67819059e+00
4.63002801e-01 -3.89576912e-01 -2.20863417e-01 4.06907827e-01
-5.31160712e-01 -7.31294572e-01 7.01091662e-02 6.52414799e-01
-1.19011128e+00 9.00219083e-01 1.16910744e+00 -2.03575432e-01
-4.43710417e-01 5.15519679e-01 -1.27230465e+00 3.22777718e-01
1.17446971e+00 -4.24924642e-01 -7.23634303e-01 -5.70170641e-01
1.05306998e-01 -8.49029922e-04 3.99550907e-02 -6.19483948e-01
9.46180582e-01 -2.49231666e-01 7.34224468e-02 -7.07564652e-01
-3.14000607e-01 -6.86002076e-01 -1.71658948e-01 2.11404398e-01
-8.21690500e-01 -3.88854802e-01 -2.77682304e-01 8.43376637e-01
-7.41460696e-02 -9.29503292e-02 3.99902731e-01 -9.36536342e-02
-1.45571113e-01 8.36370885e-01 -4.18194160e-02 2.86356866e-01
8.43619108e-01 -1.01427183e-01 -5.09552538e-01 -6.13206804e-01
-2.86333352e-01 1.15788579e+00 6.10686302e-01 2.64387161e-01
4.49092418e-01 -1.05850565e+00 -9.19889212e-01 1.91982642e-01
-7.38575757e-02 5.77049136e-01 7.90623307e-01 5.43311059e-01
-1.51857838e-01 2.90508568e-01 3.39391947e-01 -2.71595359e-01
-1.38367760e+00 2.98911124e-01 3.04219663e-01 -4.71613139e-01
-6.54451489e-01 6.80917144e-01 6.48511425e-02 -7.59348869e-01
4.59121197e-01 3.49865705e-02 1.99054092e-01 7.09134489e-02
8.13641727e-01 8.08562875e-01 9.47180092e-02 -9.14328843e-02
5.54030389e-02 -1.23688124e-01 -5.17046809e-01 1.29363179e-01
1.15538430e+00 -6.26707971e-01 -4.97521043e-01 2.36905754e-01
1.57204735e+00 2.33042717e-01 -1.30709958e+00 -6.97705090e-01
-1.87443271e-01 -5.05041957e-01 1.48148060e-01 -7.39934504e-01
-1.15120912e+00 2.46185765e-01 2.52956480e-01 1.21558063e-01
1.03895187e+00 -2.73128331e-01 1.40768218e+00 7.25337386e-01
2.68870026e-01 -1.50633860e+00 -3.15940589e-01 3.80905718e-01
5.04114509e-01 -1.05816948e+00 2.67083973e-01 1.29661739e-01
-1.04126513e+00 7.03827143e-01 6.82522774e-01 2.53215313e-01
3.86071503e-01 -1.07842110e-01 -5.96881397e-02 -5.56003638e-02
-1.04341435e+00 6.56116366e-01 4.27334532e-02 4.36918408e-01
-3.12953107e-02 1.68863446e-01 -3.05314690e-01 1.01036584e+00
-3.66285324e-01 -1.60724834e-01 2.41211683e-01 1.04632115e+00
-7.03404129e-01 -1.00735688e+00 -1.73815280e-01 9.58471656e-01
-5.53416848e-01 -2.99923699e-02 -6.47754550e-01 -7.09997416e-02
-4.35644954e-01 9.75434244e-01 -6.52058721e-02 -2.75389284e-01
1.26067281e-01 1.49575591e-01 -4.73315008e-02 -3.92399788e-01
-9.69842494e-01 -2.84547865e-01 -4.52641435e-02 -6.27850950e-01
9.88298506e-02 -7.46142864e-01 -1.37332451e+00 -6.72706723e-01
-3.81237477e-01 6.86520696e-01 7.24590242e-01 6.10383272e-01
1.08312261e+00 7.84890503e-02 1.18840778e+00 -2.98913777e-01
-1.31746209e+00 -7.16139019e-01 -7.29553521e-01 3.13063473e-01
3.41788411e-01 6.02349639e-01 -3.81672263e-01 -1.79774567e-01] | [5.836994647979736, 6.292841911315918] |
6b63f936-25fc-4460-89e7-d3fe7b366857 | occuseg-occupancy-aware-3d-instance | 2003.06537 | null | https://arxiv.org/abs/2003.06537v3 | https://arxiv.org/pdf/2003.06537v3.pdf | OccuSeg: Occupancy-aware 3D Instance Segmentation | 3D instance segmentation, with a variety of applications in robotics and augmented reality, is in large demands these days. Unlike 2D images that are projective observations of the environment, 3D models provide metric reconstruction of the scenes without occlusion or scale ambiguity. In this paper, we define "3D occupancy size", as the number of voxels occupied by each instance. It owns advantages of robustness in prediction, on which basis, OccuSeg, an occupancy-aware 3D instance segmentation scheme is proposed. Our multi-task learning produces both occupancy signal and embedding representations, where the training of spatial and feature embeddings varies with their difference in scale-aware. Our clustering scheme benefits from the reliable comparison between the predicted occupancy size and the clustered occupancy size, which encourages hard samples being correctly clustered and avoids over segmentation. The proposed approach achieves state-of-the-art performance on 3 real-world datasets, i.e. ScanNetV2, S3DIS and SceneNN, while maintaining high efficiency. | ['Lan Xu', 'Lu Fang', 'Tian Zheng', 'Lei Han'] | 2020-03-14 | occuseg-occupancy-aware-3d-instance-1 | http://openaccess.thecvf.com/content_CVPR_2020/html/Han_OccuSeg_Occupancy-Aware_3D_Instance_Segmentation_CVPR_2020_paper.html | http://openaccess.thecvf.com/content_CVPR_2020/papers/Han_OccuSeg_Occupancy-Aware_3D_Instance_Segmentation_CVPR_2020_paper.pdf | cvpr-2020-6 | ['3d-instance-segmentation-1'] | ['computer-vision'] | [ 1.18200086e-01 3.69783938e-01 7.48975854e-03 -4.35508609e-01
-6.03391647e-01 -2.54765391e-01 5.00314653e-01 1.63861796e-01
-4.72890794e-01 4.60623175e-01 -1.75555665e-02 2.49661077e-02
-2.86283195e-01 -7.15989709e-01 -7.19387770e-01 -8.36561680e-01
-1.80685282e-01 8.56496096e-01 4.31881011e-01 2.71597832e-01
3.79820883e-01 6.44433022e-01 -1.67914176e+00 -2.87599057e-01
5.42133272e-01 1.20894170e+00 6.63899481e-01 3.64065140e-01
-2.74873167e-01 3.33644211e-01 -2.80048281e-01 3.68147910e-01
5.38673759e-01 1.76367596e-01 -4.41037744e-01 4.85342592e-01
3.41156960e-01 -1.11529656e-01 -5.45388579e-01 7.88940489e-01
6.15455210e-01 2.00354755e-01 8.97531927e-01 -1.24549937e+00
-3.30417067e-01 2.05388770e-01 -9.02986825e-01 3.89943533e-02
2.97641486e-01 -1.50475278e-01 9.02893066e-01 -1.12254953e+00
5.10529995e-01 1.12450695e+00 6.27342582e-01 1.15247525e-01
-1.26755965e+00 -3.05858493e-01 2.40091205e-01 4.44586165e-02
-1.85288787e+00 -3.51508819e-02 7.38355517e-01 -6.27488673e-01
1.02498376e+00 2.25757211e-01 9.09083068e-01 7.04053938e-01
4.13932323e-01 7.67128348e-01 1.25589168e+00 -2.02074409e-01
6.43494844e-01 2.49092981e-01 -8.20624679e-02 5.69749951e-01
3.07599187e-01 -1.52912065e-01 -4.08405572e-01 5.09334244e-02
1.18667889e+00 2.08408266e-01 -1.65769890e-01 -1.06854475e+00
-1.35387027e+00 7.22591698e-01 5.20810544e-01 6.80347439e-03
-4.53636885e-01 2.01500013e-01 1.45834461e-01 -2.18898118e-01
4.71855342e-01 2.95366079e-01 -5.60340762e-01 1.11300156e-01
-8.08069348e-01 1.53234065e-01 3.69795203e-01 1.26361763e+00
1.12774682e+00 -6.97962865e-02 3.10145050e-01 8.70023608e-01
5.00220001e-01 6.72182262e-01 4.76570249e-01 -9.67651367e-01
4.40222658e-02 7.43567765e-01 5.69852218e-02 -7.65583515e-01
-6.98186040e-01 -4.63561714e-01 -8.13586771e-01 1.97432250e-01
5.05210049e-02 3.09277862e-01 -1.27152145e+00 1.26451147e+00
7.62380600e-01 5.84613919e-01 3.35241742e-02 9.57968950e-01
6.24732196e-01 5.97747743e-01 -2.97510386e-01 -5.44648878e-02
1.12522185e+00 -8.32312703e-01 -4.73908484e-01 -5.03335178e-01
4.77225959e-01 -5.54268122e-01 8.40976596e-01 3.46935809e-01
-6.51148558e-01 -6.03543043e-01 -1.09638882e+00 -1.47734702e-01
-4.31314349e-01 -2.15027481e-01 7.85616398e-01 5.45192242e-01
-9.66417551e-01 3.39427590e-01 -1.07115757e+00 -3.82715583e-01
3.32219183e-01 4.20604825e-01 -4.14889961e-01 -3.56821530e-02
-5.47140062e-01 1.01232100e+00 4.82908219e-01 -3.36112715e-02
-7.73700833e-01 -6.15538538e-01 -1.06503844e+00 -3.20189685e-01
3.32916558e-01 -3.43763053e-01 8.50487113e-01 -7.74588138e-02
-1.12197685e+00 9.82285738e-01 -9.60364863e-02 -3.66698027e-01
4.88868237e-01 -9.04635340e-02 -7.41301524e-03 -5.20804599e-02
2.64209956e-01 1.03334796e+00 6.12681150e-01 -1.48229969e+00
-7.15664208e-01 -7.11623907e-01 -6.40533566e-02 5.82912266e-01
1.91269532e-01 -8.03037405e-01 -5.17088711e-01 9.11323205e-02
1.14608502e+00 -9.74215329e-01 -8.55001152e-01 1.44328892e-01
-3.57865155e-01 -2.59171575e-02 9.22514856e-01 -2.50045300e-01
6.49884343e-01 -2.38709259e+00 1.58788592e-01 4.28654671e-01
2.81987846e-01 -5.21884143e-01 8.72519538e-02 1.30528986e-01
2.14135349e-01 -2.53344271e-02 -2.56947219e-01 -5.32083452e-01
1.97926790e-01 6.61977351e-01 -5.00737615e-02 9.82278049e-01
1.00462735e-01 6.45124853e-01 -8.33556712e-01 -7.54092455e-01
9.66613472e-01 5.51298916e-01 -5.25639296e-01 1.30596310e-01
-1.04381546e-01 4.72574741e-01 -5.87328136e-01 7.19327152e-01
9.27321553e-01 -1.42345682e-01 6.01012968e-02 -1.53119877e-01
-3.32759887e-01 9.73819643e-02 -1.57408369e+00 2.01963043e+00
-5.37042856e-01 3.65268081e-01 -2.07511529e-01 -7.91604102e-01
1.05943823e+00 4.25982960e-02 8.07807088e-01 -7.06958175e-01
4.44656238e-02 3.05115938e-01 -4.08908069e-01 -1.76053867e-01
8.18787754e-01 2.79622018e-01 -3.65675271e-01 1.06072932e-01
-7.81619772e-02 -7.70568252e-01 -2.10431442e-01 -1.18051969e-01
9.93717670e-01 2.86242694e-01 6.71250582e-01 -4.15761501e-01
6.30748197e-02 3.96496020e-02 6.25030220e-01 6.42818093e-01
-3.05537522e-01 7.50277936e-01 1.94473416e-01 -4.46934253e-01
-1.26418221e+00 -1.43385923e+00 -6.17445827e-01 5.09616971e-01
8.44349027e-01 1.04513533e-01 -4.05045211e-01 -2.53066033e-01
2.11251363e-01 3.71477157e-01 -6.54557824e-01 2.18076229e-01
-4.59690958e-01 -5.74610651e-01 -6.35175919e-03 3.94177854e-01
3.44654351e-01 -7.79732704e-01 -1.05175495e+00 1.94375828e-01
-2.81849783e-02 -1.27912509e+00 -1.02713630e-01 6.77661896e-01
-9.29700911e-01 -9.56649244e-01 -3.56953740e-01 -8.84338677e-01
6.96822762e-01 5.98958671e-01 8.41227293e-01 -2.71063656e-01
-4.80140865e-01 4.32825834e-01 -3.91364276e-01 -3.70450139e-01
3.25980932e-01 1.38302997e-01 2.81819344e-01 -3.70381415e-01
3.83382499e-01 -7.81317890e-01 -5.64709485e-01 3.61970842e-01
-7.59558976e-01 2.08439469e-01 6.85919762e-01 5.76010525e-01
1.50904870e+00 -2.73843855e-02 7.54447561e-03 -6.39320493e-01
-1.18101865e-01 -5.88333488e-01 -6.25059485e-01 -5.18130735e-02
-4.32808548e-01 -1.61712423e-01 1.83553234e-01 -2.55759180e-01
-4.75204974e-01 3.94584715e-01 8.09993669e-02 -6.49731457e-01
-3.60507697e-01 2.66044110e-01 -1.13232814e-01 -2.55810861e-02
6.11658335e-01 2.73403645e-01 -3.09343427e-01 -3.99575144e-01
3.89624834e-01 7.38057435e-01 4.20146883e-01 -3.20364743e-01
6.76452518e-01 6.35056674e-01 8.09500366e-02 -1.24995434e+00
-6.51625931e-01 -8.94227087e-01 -1.04000449e+00 -3.11679423e-01
9.77970541e-01 -1.14452505e+00 -2.32415065e-01 2.32055441e-01
-9.93842721e-01 -3.20781291e-01 -5.65762937e-01 7.64885783e-01
-9.68144476e-01 9.52285007e-02 -1.96289137e-01 -1.12088215e+00
2.54036844e-01 -1.23419249e+00 1.52585852e+00 4.51003201e-02
-7.23575428e-02 -7.66742229e-01 6.74873367e-02 1.44777417e-01
-5.94829731e-02 5.52252769e-01 6.91458642e-01 -3.69536549e-01
-1.13411832e+00 -1.84143577e-02 -2.09481552e-01 -5.64199463e-02
3.45306396e-02 -2.59409100e-01 -1.22272170e+00 -1.97277054e-01
-7.04702958e-02 -2.60991514e-01 5.17730415e-01 7.85314262e-01
1.37735868e+00 3.47200125e-01 -4.31156665e-01 7.20104992e-01
1.63565719e+00 1.20414540e-01 7.16416359e-01 2.41060168e-01
7.00132191e-01 4.54321295e-01 9.98031259e-01 7.38015234e-01
4.61927474e-01 6.32571638e-01 8.28752160e-01 -1.96752518e-01
6.31980076e-02 -2.04638630e-01 -1.80914864e-01 8.52022111e-01
3.29943091e-01 -1.23242773e-01 -9.51077640e-01 6.65154874e-01
-1.93470848e+00 -5.10542810e-01 -1.94650412e-01 2.19748783e+00
3.05261761e-01 1.36743784e-01 -2.71752447e-01 8.08062330e-02
5.75799942e-01 4.21300739e-01 -9.05001402e-01 -1.22489132e-01
-9.45673659e-02 1.52109787e-01 1.04414356e+00 6.02929592e-01
-1.19383383e+00 9.18487549e-01 6.15349770e+00 7.11778283e-01
-8.30428958e-01 1.23198651e-01 5.00787079e-01 -1.21539757e-01
-3.16819727e-01 -2.85238117e-01 -7.67331362e-01 1.71131074e-01
6.34022534e-01 2.08593160e-01 3.11562449e-01 1.12335706e+00
1.90741122e-01 -4.33215857e-01 -9.98362243e-01 1.23558354e+00
8.59773159e-02 -1.10438740e+00 -3.04805040e-01 5.00402093e-01
7.76243031e-01 4.31158334e-01 4.27040607e-02 1.82006806e-01
1.29184484e-01 -9.01886404e-01 1.09776831e+00 2.61912763e-01
8.21658373e-01 -6.14897549e-01 7.46711254e-01 4.90935922e-01
-1.43304729e+00 7.06598461e-02 -6.90862536e-01 2.62446497e-02
3.62882614e-01 7.92155147e-01 -9.77613747e-01 4.42525208e-01
6.99159503e-01 5.79322577e-01 -3.16849709e-01 1.18486035e+00
-2.25017127e-02 -1.61823437e-01 -6.50378168e-01 4.19633910e-02
2.40425929e-01 -3.96704108e-01 2.24964425e-01 8.80209863e-01
4.39798445e-01 -1.69307850e-02 6.91826224e-01 6.37217760e-01
2.78501332e-01 1.35144200e-02 -8.70541334e-01 3.09355170e-01
8.63312960e-01 1.11507559e+00 -1.02464485e+00 -2.01850668e-01
-2.86536157e-01 9.30521905e-01 1.27530187e-01 1.68529321e-02
-9.91842151e-01 2.79544946e-02 7.93155491e-01 7.19173774e-02
4.24522519e-01 -7.30022013e-01 -6.29068494e-01 -7.86243916e-01
3.92031893e-02 -1.22577943e-01 -1.78865537e-01 -6.22542620e-01
-8.92268419e-01 4.23318475e-01 6.33582175e-02 -1.39069033e+00
-1.18413763e-02 -6.82839572e-01 -8.10634810e-03 5.44985831e-01
-1.53490078e+00 -9.59322870e-01 -5.34833610e-01 2.47038454e-01
8.06059778e-01 3.93791229e-01 8.36111963e-01 3.02611530e-01
-3.83289218e-01 7.18915388e-02 1.96152985e-01 -3.26590002e-01
1.90196618e-01 -1.42718196e+00 3.76066625e-01 4.95371222e-01
2.01192141e-01 6.93392381e-02 7.70210207e-01 -5.48474669e-01
-1.35083628e+00 -1.30530643e+00 5.00753224e-01 -4.19362515e-01
4.76298928e-01 -6.11271501e-01 -6.19318128e-01 6.45303130e-01
-3.75354499e-01 2.04123393e-01 6.50338471e-01 3.48138846e-02
-1.77920148e-01 1.54405862e-01 -1.37552941e+00 3.89336556e-01
1.47287452e+00 -5.30903578e-01 -4.97014701e-01 3.81074518e-01
8.67414534e-01 -8.22857082e-01 -9.54634070e-01 3.92442346e-01
5.80123246e-01 -1.08268702e+00 1.09648645e+00 1.34962603e-01
1.40228979e-02 -5.60532689e-01 -8.68012905e-01 -1.09369028e+00
-3.81320447e-01 2.00403646e-01 -6.97503313e-02 5.88782370e-01
3.92547488e-01 -4.26411510e-01 1.08979821e+00 3.73774171e-01
-6.56880617e-01 -8.50794375e-01 -1.35695148e+00 -7.61730492e-01
-3.07894289e-01 -6.53722823e-01 6.09987080e-01 8.90423954e-01
-4.07508641e-01 -7.16976076e-02 -1.85320988e-01 7.22953260e-01
8.03715765e-01 7.97356516e-02 1.04332900e+00 -1.42598903e+00
-3.16117033e-02 3.76014188e-02 -9.64019835e-01 -1.43658710e+00
-3.37542146e-02 -6.37329996e-01 3.03534299e-01 -1.72831941e+00
8.48124083e-03 -7.27843225e-01 -2.70100713e-01 2.43683442e-01
4.45913553e-01 3.23796123e-01 -8.46155584e-02 3.25296462e-01
-7.65663981e-01 7.67216802e-01 1.01069248e+00 -1.66005999e-01
-4.99802351e-01 -4.12183583e-01 -1.80014893e-01 7.27264941e-01
6.90269649e-01 -3.37275177e-01 -5.60483575e-01 -5.39257348e-01
-1.42049387e-01 -1.09841181e-04 1.81304559e-01 -1.31662774e+00
-1.05471507e-01 -2.36005649e-01 4.40079063e-01 -1.14934218e+00
9.55947459e-01 -1.11959910e+00 3.70132744e-01 2.37634197e-01
7.22234920e-02 -1.85685351e-01 -2.85900999e-02 7.61629999e-01
7.20277801e-02 -4.64398004e-02 7.95116663e-01 -3.45597923e-01
-1.08248377e+00 6.44628286e-01 -2.03872591e-01 -3.40454578e-01
1.27517331e+00 -8.97801101e-01 6.49798438e-02 1.31363243e-01
-6.67945147e-01 4.79084581e-01 7.73267150e-01 3.48971397e-01
9.05179203e-01 -1.32509899e+00 -2.72299826e-01 2.28698403e-01
3.92452270e-01 7.31579602e-01 4.83551532e-01 8.84809077e-01
-7.52691448e-01 3.20419103e-01 -4.06604856e-02 -1.17891681e+00
-8.35702360e-01 3.17164361e-01 1.62611023e-01 9.12404656e-02
-8.36413264e-01 7.30473161e-01 2.69678384e-01 -8.71672034e-01
3.79991561e-01 -5.57448387e-01 1.35320034e-02 -4.35745269e-02
2.82834053e-01 2.16441661e-01 -1.24983259e-01 -7.83875108e-01
-3.75611603e-01 6.84435368e-01 1.54067308e-01 -5.30254729e-02
1.43555641e+00 -3.72846603e-01 1.03629790e-01 9.55772340e-01
1.07500780e+00 -3.45932037e-01 -1.46860647e+00 -1.78990155e-01
-2.52963863e-02 -8.74612629e-01 3.26279759e-01 -1.60817757e-01
-7.24440873e-01 7.67306268e-01 1.09175634e+00 2.07671866e-01
6.33232176e-01 3.32480162e-01 5.25390744e-01 3.25403541e-01
9.79164839e-01 -1.05804813e+00 -6.56653717e-02 4.97176707e-01
6.94156349e-01 -1.29091382e+00 9.97399688e-02 -5.58322787e-01
-4.22449857e-01 6.49835706e-01 7.35344708e-01 -2.63384461e-01
6.13027036e-01 1.19247049e-01 6.14414774e-02 -4.43588287e-01
-2.15664089e-01 -2.93108046e-01 -2.08947107e-01 8.34318161e-01
3.07196695e-02 4.51175570e-01 5.32792807e-02 4.75488044e-02
-3.15859616e-01 -5.88496506e-01 5.06742239e-01 9.31698143e-01
-9.70278800e-01 -6.61621034e-01 -4.79873538e-01 6.49083555e-01
2.85656512e-01 2.29875654e-01 5.94791397e-02 9.84407306e-01
4.65320319e-01 5.66409528e-01 4.91190434e-01 -5.35813630e-01
4.44048554e-01 -1.65985614e-01 5.39292097e-01 -8.50622654e-01
3.69275987e-01 2.27816105e-01 -2.12755829e-01 -7.19137549e-01
-4.58539724e-01 -8.33368301e-01 -1.52714670e+00 1.98622961e-02
-5.06356418e-01 -1.03245161e-01 1.16424489e+00 7.78579831e-01
2.20846877e-01 4.51459646e-01 8.00830066e-01 -1.43677557e+00
-2.62276173e-01 -8.09864163e-01 -1.23235250e+00 -6.48783892e-02
2.09187388e-01 -1.01662898e+00 -3.56466144e-01 -3.78004879e-01] | [8.075126647949219, -2.6271541118621826] |
31dd3c45-6607-4ee1-abbe-1da93e26d661 | detection-of-3d-bounding-boxes-of-vehicles | 2003.13137 | null | https://arxiv.org/abs/2003.13137v2 | https://arxiv.org/pdf/2003.13137v2.pdf | Detection of 3D Bounding Boxes of Vehicles Using Perspective Transformation for Accurate Speed Measurement | Detection and tracking of vehicles captured by traffic surveillance cameras is a key component of intelligent transportation systems. We present an improved version of our algorithm for detection of 3D bounding boxes of vehicles, their tracking and subsequent speed estimation. Our algorithm utilizes the known geometry of vanishing points in the surveilled scene to construct a perspective transformation. The transformation enables an intuitive simplification of the problem of detecting 3D bounding boxes to detection of 2D bounding boxes with one additional parameter using a standard 2D object detector. Main contribution of this paper is an improved construction of the perspective transformation which is more robust and fully automatic and an extended experimental evaluation of speed estimation. We test our algorithm on the speed estimation task of the BrnoCompSpeed dataset. We evaluate our approach with different configurations to gauge the relationship between accuracy and computational costs and benefits of 3D bounding box detection over 2D detection. All of the tested configurations run in real-time and are fully automatic. Compared to other published state-of-the-art fully automatic results our algorithm reduces the mean absolute speed measurement error by 32% (1.10 km/h to 0.75 km/h) and the absolute median error by 40% (0.97 km/h to 0.58 km/h). | ['Milan Ftáčnik', 'Viktor Kocur'] | 2020-03-29 | null | null | null | null | ['vehicle-speed-estimation'] | ['computer-vision'] | [-1.19431578e-01 -2.30644166e-01 7.97663406e-02 -1.13289267e-01
-5.04103422e-01 -8.51870298e-01 8.33917975e-01 1.06445104e-01
-9.08081770e-01 3.66263509e-01 -4.55417365e-01 -6.20032191e-01
1.45163685e-01 -8.43270361e-01 -7.09962070e-01 -6.57079697e-01
-1.22031219e-01 7.51463175e-01 1.14484990e+00 -1.36170998e-01
2.16199040e-01 1.19103765e+00 -1.55894339e+00 -3.38451117e-01
5.10041296e-01 7.94049859e-01 -6.32131845e-02 1.27803314e+00
1.79478958e-01 2.06354305e-01 -4.47285801e-01 -1.65005013e-01
5.84782660e-01 1.47044539e-01 -2.77619809e-01 2.64217794e-01
8.38844359e-01 -6.94884241e-01 -2.66815990e-01 8.43109846e-01
1.31765351e-01 1.13608547e-01 8.15991700e-01 -1.52273583e+00
4.75804538e-01 -4.00635839e-01 -6.54400229e-01 5.88381290e-01
2.79571712e-01 1.26222774e-01 2.26902321e-01 -5.36780953e-01
7.20243573e-01 1.04667819e+00 6.91892624e-01 3.18722486e-01
-1.18249321e+00 -6.06145382e-01 -3.13425273e-01 2.44676396e-01
-1.77086210e+00 -5.78263462e-01 4.98003423e-01 -7.46851921e-01
8.49667013e-01 2.40183905e-01 5.01628101e-01 4.08789754e-01
-6.91884607e-02 3.39643776e-01 9.05228972e-01 -3.47584218e-01
1.47219390e-01 6.50120854e-01 3.31069767e-01 8.50643635e-01
7.55055189e-01 3.44309300e-01 4.07664597e-01 -9.82916565e-04
6.31304383e-01 -7.26233274e-02 2.21162587e-01 -5.05504310e-01
-9.42388892e-01 8.66838396e-01 1.05411686e-01 -1.14769675e-02
-1.08662486e-01 1.71615705e-01 3.43455285e-01 -6.10792451e-02
4.22061384e-01 -1.54201731e-01 -2.40607262e-01 -3.10538352e-01
-9.16724205e-01 5.02706587e-01 5.29351473e-01 1.21997190e+00
6.84838057e-01 -2.44592264e-01 1.10780485e-01 3.02334040e-01
1.37582794e-01 1.06036139e+00 -3.54076594e-01 -9.21851873e-01
6.68484151e-01 6.37172461e-01 4.36613202e-01 -9.20294344e-01
-6.52113259e-01 -9.66588333e-02 -2.59821802e-01 4.96435553e-01
9.10246134e-01 -3.96543115e-01 -5.93395114e-01 1.19164002e+00
8.20324898e-01 6.45578876e-02 -2.09159091e-01 8.48059297e-01
3.58487844e-01 7.42565095e-01 -1.13926932e-01 -1.60930306e-01
1.30334044e+00 -6.40446723e-01 -2.66007721e-01 2.48176977e-01
8.16072643e-01 -8.84171903e-01 3.07435095e-01 1.67143866e-01
-8.57736468e-01 -5.22943616e-01 -8.06704283e-01 3.01684916e-01
-6.30251169e-01 2.76016444e-01 1.16302967e-01 1.08441257e+00
-8.24310243e-01 2.06949905e-01 -6.71058536e-01 -6.11250103e-01
3.43733758e-01 3.10528576e-01 -3.73462468e-01 2.26872757e-01
-7.23323822e-01 1.15253282e+00 1.73389152e-01 -1.71366617e-01
-6.65318370e-01 -5.65020740e-01 -8.50886106e-01 -2.64218748e-01
6.29212439e-01 -3.92137170e-01 1.12164593e+00 -3.05755973e-01
-1.21654296e+00 1.01849914e+00 -4.11932945e-01 -6.31113589e-01
1.07991087e+00 -2.54556388e-01 -5.21030009e-01 3.71698350e-01
1.21557564e-01 5.91407061e-01 5.06426036e-01 -1.04080570e+00
-1.00967109e+00 -3.92130792e-01 -1.15921877e-01 -1.36718303e-01
2.34459043e-01 2.39413068e-01 -4.70362991e-01 -2.28279755e-02
-2.60862619e-01 -1.13835406e+00 -1.38388291e-01 1.20053522e-01
-1.82311296e-01 -2.40279615e-01 1.29358459e+00 -4.01962250e-01
8.65550935e-01 -1.78053069e+00 -4.02186066e-01 2.77829170e-01
2.07637757e-01 6.05478227e-01 1.64110228e-01 3.34668756e-01
3.07812572e-01 -1.39723286e-01 6.25737980e-02 -1.61841482e-01
-2.56368756e-01 -1.47043252e-02 -7.01680360e-03 1.09340537e+00
8.15993845e-02 4.23371553e-01 -8.20211947e-01 -5.64248979e-01
8.87405336e-01 7.27172792e-01 -2.35151544e-01 1.68499630e-02
3.11514229e-01 1.36766672e-01 -4.35133636e-01 6.04081213e-01
1.09151018e+00 3.84763569e-01 -4.53498632e-01 -3.66807450e-04
-9.06045198e-01 9.76589546e-02 -1.43523455e+00 7.17369914e-01
-1.67837933e-01 1.13340807e+00 -1.63037870e-02 -6.25098944e-01
9.05563176e-01 2.16807127e-01 3.94552678e-01 -5.37849486e-01
3.60390276e-01 1.14283316e-01 -2.87281126e-01 -4.96154159e-01
6.11548305e-01 5.71645536e-02 2.00089261e-01 1.25790238e-01
-2.60545671e-01 -1.31726593e-01 6.40132487e-01 1.45634934e-01
9.39513564e-01 -1.17365204e-01 3.31517458e-01 -4.07308429e-01
8.54083121e-01 2.71872431e-01 -1.97296869e-02 4.69286263e-01
-5.89434445e-01 3.58822227e-01 5.41128814e-01 -6.16952121e-01
-1.25239170e+00 -9.20165181e-01 -3.48017961e-01 6.31229520e-01
2.25282282e-01 -2.36718655e-01 -9.63582218e-01 -7.79924691e-01
5.93243502e-02 5.84533989e-01 -6.94233596e-01 3.21484864e-01
-7.84748733e-01 -4.79880005e-01 3.90407681e-01 4.58835751e-01
3.65196913e-01 -1.89406335e-01 -1.12713361e+00 -3.28571573e-02
3.42267513e-01 -1.66783214e+00 -2.77093500e-01 -5.07047951e-01
-7.86359251e-01 -1.31063759e+00 -6.16573989e-01 -2.11754203e-01
8.19716871e-01 7.84173191e-01 8.35118055e-01 -3.04719317e-03
-3.06424320e-01 3.85798961e-01 -2.96988100e-01 -6.40358150e-01
-6.00966275e-01 -1.54728338e-01 9.85874832e-02 -5.76804299e-03
4.34812427e-01 1.63441971e-01 -6.04216874e-01 8.81690025e-01
-4.73701298e-01 -5.98096251e-02 2.47679621e-01 -5.87704740e-02
3.39027315e-01 6.76812977e-02 -1.92329705e-01 -6.30152643e-01
-4.10561934e-02 -3.43803614e-01 -1.61961389e+00 -3.66613030e-01
-4.18787390e-01 -2.01567367e-01 4.52179730e-01 -2.64106482e-01
-6.09437644e-01 4.46566641e-01 -1.06644876e-01 -3.81825596e-01
-5.69236100e-01 -6.19435966e-01 1.43988296e-01 -3.19291800e-01
4.48527545e-01 -2.76692901e-02 7.37790614e-02 -3.57558131e-01
3.39537591e-01 4.80459481e-01 3.38261068e-01 5.58132157e-02
1.09473217e+00 8.93832922e-01 6.09112144e-01 -1.23707557e+00
-2.78809130e-01 -9.46012139e-01 -9.51572061e-01 -7.44176388e-01
8.35041821e-01 -5.92594087e-01 -9.21095371e-01 2.94218659e-01
-1.35542989e+00 -8.44364688e-02 6.19133748e-02 5.32996595e-01
-2.72922754e-01 4.19342905e-01 -1.12322204e-01 -1.10091126e+00
-1.56550750e-01 -1.17412901e+00 1.32519460e+00 1.99533686e-01
-1.37070278e-02 -8.90071213e-01 1.13401748e-01 3.31205696e-01
4.84449565e-01 4.36118364e-01 1.25181809e-01 -3.32138866e-01
-5.25388420e-01 -8.38157713e-01 -4.09797817e-01 1.89319417e-01
-2.84673989e-01 4.26843613e-01 -8.51860881e-01 -8.91202092e-02
-4.21325356e-01 6.91433966e-01 7.60532081e-01 5.19966364e-01
4.48968798e-01 -6.65654764e-02 -7.52943754e-01 2.50514835e-01
1.53011191e+00 3.48462254e-01 5.03886759e-01 4.30270433e-01
4.28348184e-01 6.64873838e-01 6.43367767e-01 2.90032178e-01
4.04674709e-01 1.08524692e+00 5.36618948e-01 -5.73793091e-02
-1.55021876e-01 6.14168122e-02 4.30835813e-01 -3.76934335e-02
-3.57647628e-01 -2.28287145e-01 -1.09778428e+00 6.07791543e-01
-1.55543280e+00 -1.03740597e+00 -8.67115974e-01 2.57271218e+00
-1.77927777e-01 4.26383615e-01 9.00682271e-01 3.37040633e-01
1.00751972e+00 -3.00000370e-01 1.72156155e-01 -4.06541109e-01
5.11604846e-01 -3.60959649e-01 1.29634130e+00 8.75748217e-01
-1.41362464e+00 7.24254608e-01 6.17468548e+00 4.24835175e-01
-9.50867772e-01 1.31796420e-01 2.78050989e-01 -1.48922667e-01
5.67283094e-01 -1.43875316e-01 -1.47818065e+00 6.04016483e-01
1.12269986e+00 5.34352250e-02 2.49367282e-02 9.07198548e-01
5.29392242e-01 -5.08441985e-01 -8.97736311e-01 7.90146887e-01
1.03303500e-01 -1.25629485e+00 -1.59923807e-01 3.03354293e-01
4.58361834e-01 1.81006685e-01 -3.65696371e-01 1.50450403e-02
-1.25589341e-01 -5.17597139e-01 7.92412221e-01 2.88342834e-01
6.23731494e-01 -9.46651697e-01 8.80761623e-01 3.88468176e-01
-1.47987914e+00 2.25940645e-01 -2.62646317e-01 -1.78098708e-01
5.04687488e-01 3.51622343e-01 -1.31058204e+00 1.35544330e-01
3.36980671e-01 2.61799872e-01 -7.39040017e-01 1.36841261e+00
-1.85990930e-02 4.23180282e-01 -7.90410340e-01 -2.03873515e-01
5.40978312e-01 -3.25992435e-01 1.06764352e+00 1.73585200e+00
2.14293942e-01 -3.91299240e-02 -3.50479260e-02 6.23974800e-01
3.66961032e-01 -1.09827481e-01 -8.33274424e-01 6.52356446e-01
3.82649630e-01 1.27274537e+00 -8.73740852e-01 -5.04359722e-01
-4.57790911e-01 3.30797911e-01 -5.70624582e-02 6.07358292e-02
-1.20355988e+00 -5.59548199e-01 6.91429436e-01 8.42572927e-01
5.57729900e-01 -5.17330825e-01 1.08828753e-01 -6.09905779e-01
-5.89434151e-03 1.42360479e-01 7.90588483e-02 -4.39496845e-01
-4.51490521e-01 4.58105475e-01 5.55671871e-01 -1.33739352e+00
-1.60384685e-01 -9.22958195e-01 -6.37113690e-01 5.16006291e-01
-1.36562490e+00 -9.06327903e-01 -5.17846465e-01 3.18412870e-01
5.32670856e-01 -2.87197791e-02 2.79821038e-01 5.77374160e-01
-6.32184505e-01 3.58330578e-01 -3.27742733e-02 2.44885176e-01
1.58708245e-01 -9.29893196e-01 5.76141238e-01 1.20235646e+00
-3.18799138e-01 1.90762445e-01 9.68352437e-01 -3.51631850e-01
-1.34262300e+00 -1.27677822e+00 9.59885120e-01 -9.18019414e-01
6.74473643e-01 -4.98654753e-01 -4.73330021e-01 4.99475002e-01
-1.49241611e-01 2.38092393e-01 1.55627459e-01 -4.67485875e-01
-1.64504692e-01 -4.12640214e-01 -1.24596047e+00 3.40588212e-01
7.12920427e-01 1.50494706e-02 -3.29050332e-01 3.74099910e-01
3.32177371e-01 -3.23273182e-01 -4.37737972e-01 2.03079253e-01
7.73103237e-01 -1.02478611e+00 9.15787339e-01 -2.52051115e-01
-3.69469315e-01 -5.65770745e-01 1.01629347e-01 -6.65936887e-01
9.30086989e-03 -6.74469709e-01 9.40450951e-02 1.00255775e+00
2.44301975e-01 -7.46854126e-01 6.86338425e-01 4.61573333e-01
5.39371483e-02 -4.49629962e-01 -1.29306829e+00 -8.77717435e-01
-2.08719715e-01 -8.22878659e-01 1.95334136e-01 2.55743355e-01
-4.15618211e-01 2.80411661e-01 -1.09527595e-01 4.05719280e-01
9.52287376e-01 -1.41442150e-01 1.23311627e+00 -1.29942083e+00
3.08766693e-01 -6.98450625e-01 -1.03881836e+00 -1.04319406e+00
-2.85261929e-01 -2.89176017e-01 -3.27610046e-01 -1.23472321e+00
1.31466076e-01 -2.60986477e-01 2.62754560e-01 -1.82751462e-01
3.13433111e-01 3.21990252e-01 2.13191584e-01 -2.26516545e-01
-6.67041123e-01 -1.86430186e-01 6.88754261e-01 2.79979378e-01
-8.66848975e-02 3.77091706e-01 4.70159575e-02 8.73059571e-01
8.16375911e-01 -6.19116247e-01 -5.68489283e-02 -1.87332839e-01
-9.47849825e-02 -1.46353573e-01 7.18463480e-01 -1.06084204e+00
1.78780526e-01 -4.44814712e-02 1.45672202e-01 -1.05470622e+00
3.11051846e-01 -1.10919571e+00 -6.57630488e-02 6.79200351e-01
2.23972797e-01 1.74022883e-01 5.01038015e-01 3.85659903e-01
3.69878143e-01 -2.02879399e-01 1.05227828e+00 5.61541021e-02
-7.62775898e-01 1.53365746e-01 -7.61067808e-01 -2.15324670e-01
1.61857986e+00 -6.15414143e-01 -3.67631674e-01 -2.05761224e-01
-3.60724866e-01 6.91762241e-03 6.31555974e-01 3.61391008e-01
3.49917650e-01 -9.67481315e-01 -8.10530603e-01 3.11374962e-01
-7.89630115e-02 -3.74488980e-01 -6.54044896e-02 1.01415634e+00
-1.09418917e+00 9.53698397e-01 -1.00820586e-01 -8.27535927e-01
-1.86430776e+00 6.94366395e-01 4.67993975e-01 1.53999180e-01
-5.15818655e-01 2.81766355e-01 -4.05034684e-02 -2.46919077e-02
1.48595959e-01 -3.75918835e-01 -2.33223855e-01 1.19516999e-02
7.63408005e-01 1.25074720e+00 1.13090500e-01 -1.18968010e+00
-7.15760946e-01 1.06891525e+00 3.57734226e-02 -6.03268016e-03
9.72873867e-01 -2.26626053e-01 2.69649863e-01 4.87809628e-02
1.33424509e+00 2.07556114e-01 -1.28530395e+00 2.59748280e-01
-1.89940676e-01 -6.43492699e-01 1.35955006e-01 -2.71385461e-01
-8.44436467e-01 8.06197226e-01 8.94994199e-01 4.98827398e-01
6.62707925e-01 8.47328678e-02 6.93221331e-01 2.99819887e-01
2.68376261e-01 -8.55723619e-01 -5.64388394e-01 4.80195075e-01
5.20641506e-01 -1.31027162e+00 2.07094356e-01 -6.81334019e-01
-2.97758758e-01 1.19837165e+00 5.00574350e-01 -2.57384151e-01
4.39445287e-01 2.79058009e-01 -8.53001177e-02 -1.16516620e-01
-3.89451295e-01 -5.46915710e-01 3.68499577e-01 4.81793374e-01
1.81578919e-02 1.83152914e-01 -3.65219206e-01 -3.26116443e-01
5.63977845e-02 -2.61239499e-01 6.31174982e-01 7.82813787e-01
-6.89507246e-01 -6.79540515e-01 -7.00735211e-01 4.71166149e-02
-1.32997558e-01 4.57426429e-01 -2.31499359e-01 1.39885974e+00
2.52514005e-01 1.13466084e+00 3.70015323e-01 -1.82809025e-01
6.68856800e-01 -1.52474493e-01 3.81106019e-01 -1.39235586e-01
-6.87201023e-02 -4.82963994e-02 2.97916204e-01 -4.93570417e-01
-4.84626383e-01 -8.83422852e-01 -1.11894417e+00 -5.26404798e-01
-4.36866790e-01 5.88887222e-02 1.05494821e+00 7.50527501e-01
2.67359942e-01 -9.71991383e-03 6.54290020e-01 -1.12010920e+00
-1.56001434e-01 -6.76265419e-01 -2.87870944e-01 2.34805241e-01
6.29402399e-01 -9.50153053e-01 -3.57713550e-01 4.68257926e-02] | [7.975043296813965, -1.1930713653564453] |
e39bf389-3f24-4e9e-b3b7-3111ffe03d10 | direct-cortical-thickness-estimation-using | null | null | https://doi.org/10.1002/hbm.25159 | https://onlinelibrary.wiley.com/doi/epdf/10.1002/hbm.25159 | Direct cortical thickness estimation using deep learning‐based anatomy segmentation and cortex parcellation | Accurate and reliable measures of cortical thickness from magnetic resonance imaging are an important biomarker to study neurodegenerative and neurological disorders. Diffeomorphic registration‐based cortical thickness (DiReCT) is a known technique to derive such measures from non‐surface‐based volumetric tissue maps. ANTs provides an open‐source method for estimating cortical thickness, derived by applying DiReCT to an atlas‐based segmentation. In this paper, we propose DL+DiReCT, a method using high‐quality deep learning‐based neuroanatomy segmentations followed by DiReCT, yielding accurate and reliable cortical thickness measures in a short time. We evaluate the methods on two independent datasets and compare the results against surface‐based measures from FreeSurfer. Good correlation of DL+DiReCT with FreeSurfer was observed (r = .887) for global mean cortical thickness compared to ANTs versus FreeSurfer (r = .608). Experiments suggest that both DiReCT‐based methods had higher sensitivity to changes in cortical thickness than Freesurfer. However, while ANTs showed low scan‐rescan robustness, DL+DiReCT showed similar robustness to Freesurfer. Effect‐sizes for group‐wise differences of healthy controls compared to individuals with dementia were highest with the deep learning‐based segmentation. DL+DiReCT is a promising combination of a deep learning‐based method with a traditional registration technique to detect subtle changes in cortical thickness. | ['Richard McKinley', 'Roland Wiest', 'Mauricio Reyes', 'Christian Rummel', 'Michael Rebsamen'] | 2020-11-05 | null | null | null | null | ['3d-medical-imaging-segmentation', 'diffeomorphic-medical-image-registration', 'brain-morphometry'] | ['medical', 'medical', 'medical'] | [-3.09085339e-01 7.28597343e-02 3.33011180e-01 -4.94271278e-01
-7.77502298e-01 -1.19761400e-01 4.57379580e-01 -1.49771973e-01
-9.27437246e-01 1.08585191e+00 -4.13656011e-02 1.23712316e-01
-1.83201283e-01 -1.06808913e+00 -4.50388908e-01 -6.64464951e-01
-5.16058326e-01 1.12571073e+00 6.56552851e-01 -9.23765451e-03
3.23531270e-01 5.25473714e-01 -7.72013545e-01 -2.22688690e-02
9.48238194e-01 6.82183087e-01 6.78478301e-01 1.31374657e-01
-3.47296327e-01 9.58753005e-02 -1.88600674e-01 -1.50086701e-01
2.15820983e-01 -2.78037041e-01 -1.00879705e+00 -2.81778574e-01
6.76872015e-01 -6.66272342e-01 1.90520138e-01 9.23149347e-01
4.70829517e-01 -2.52843171e-01 7.87003696e-01 -5.45721173e-01
-5.42258680e-01 7.30823576e-01 -8.83138418e-01 7.71492660e-01
-9.55505893e-02 2.09740609e-01 8.85183811e-02 -6.43799067e-01
9.48125839e-01 1.22008777e+00 1.03623152e+00 4.92861480e-01
-1.35127437e+00 -9.50348318e-01 -4.16343451e-01 4.20740634e-01
-1.23095286e+00 -3.49208772e-01 3.16816390e-01 -7.82334089e-01
8.38885784e-01 1.34483203e-01 9.59950447e-01 6.07881188e-01
8.77472818e-01 1.59166772e-02 1.91595566e+00 2.03179801e-03
3.26132059e-01 -4.14319217e-01 3.03759426e-01 6.40085399e-01
6.28599703e-01 1.41545814e-02 5.98172774e-04 -1.41472533e-01
1.09593499e+00 -3.98741692e-01 -8.02697092e-02 -8.29694420e-02
-1.31071770e+00 6.99373186e-01 6.45727754e-01 8.07398021e-01
-4.62707013e-01 3.13675851e-02 5.70333183e-01 1.75395206e-01
8.22306991e-01 2.12755367e-01 -1.60135269e-01 2.07861617e-01
-1.45839787e+00 3.88807207e-01 4.05531824e-02 2.45526686e-01
3.81098360e-01 1.71105012e-01 -3.67158242e-02 9.57414925e-01
5.42860270e-01 8.18543613e-01 8.25409532e-01 -1.03015959e+00
2.82608211e-01 5.11755764e-01 -5.82044959e-01 -5.47011614e-01
-1.00982869e+00 4.00357507e-02 -9.77739334e-01 8.88689995e-01
7.79923201e-01 2.37588858e-04 -8.97661567e-01 1.43664968e+00
1.52563199e-01 -2.70434618e-01 -7.31192827e-01 1.15484583e+00
5.32398760e-01 -2.41073713e-01 3.21751595e-01 -1.37179181e-01
1.37328506e+00 -2.72558868e-01 -3.49245936e-01 -2.52875388e-01
6.40572608e-01 -3.32522452e-01 8.17023933e-01 8.93118456e-02
-1.23464060e+00 -6.52997941e-02 -8.25316191e-01 -1.11722484e-01
-2.76428342e-01 -2.04601005e-01 3.58138740e-01 7.59242296e-01
-1.50166678e+00 8.72775674e-01 -1.27961636e+00 -4.91542608e-01
7.91932583e-01 4.49428558e-01 -8.58639419e-01 6.41627237e-02
-6.97491467e-01 1.46424949e+00 8.01694095e-02 -1.96912587e-01
-3.68526518e-01 -8.40673447e-01 -3.68833393e-01 -5.30305624e-01
-4.47804332e-01 -6.67379618e-01 7.32701838e-01 -4.92105573e-01
-1.19651258e+00 1.39627588e+00 -2.26185977e-01 -6.04777157e-01
7.13110387e-01 1.17404565e-01 -6.08795397e-02 4.45508212e-01
5.19299626e-01 5.77703476e-01 4.01125759e-01 -7.30711043e-01
2.52820075e-01 -1.28445053e+00 -5.97179711e-01 -1.41422376e-01
1.84065357e-01 3.75665665e-01 5.51036537e-01 -2.37113282e-01
6.98289514e-01 -8.08831334e-01 -1.30335718e-01 6.24593675e-01
-3.53766263e-01 -2.50509113e-01 6.45537019e-01 -1.34162629e+00
3.36523533e-01 -1.32668519e+00 -1.83421046e-01 3.26162040e-01
8.07490170e-01 8.67572203e-02 5.60033508e-02 -3.71449351e-01
-6.22050464e-02 4.12112415e-01 -8.01741898e-01 -1.05526380e-01
-2.19469607e-01 -2.15734988e-01 6.34001493e-01 1.02975130e+00
-2.38141134e-01 1.11350930e+00 -6.56480670e-01 -7.16419697e-01
1.39605477e-01 6.04070246e-01 -7.90078193e-03 -1.43334806e-01
5.72689235e-01 4.88684386e-01 -2.28253409e-01 3.09743732e-01
1.14389682e+00 2.60953218e-01 2.91279554e-01 -3.48730236e-02
-3.05141240e-01 -5.64235263e-02 -3.51269305e-01 1.44803679e+00
-1.10365786e-01 6.91883743e-01 4.58487660e-01 -9.08517838e-01
1.32198405e+00 3.12384903e-01 6.43079281e-01 -9.78182018e-01
4.22526658e-01 4.43622828e-01 4.49679494e-01 -2.71338761e-01
-4.96272981e-01 -6.34578645e-01 5.07407129e-01 6.40760958e-01
8.93340912e-03 -1.50256231e-01 1.12361938e-01 -8.86115506e-02
1.28462493e+00 1.29923254e-01 2.62539953e-01 -9.96632576e-01
4.30415839e-01 -3.39635670e-01 3.46439093e-01 1.77562684e-02
-6.54465497e-01 6.48567319e-01 4.44526970e-01 -4.48067635e-01
-1.18591547e+00 -1.51109695e+00 -8.32928777e-01 2.90475041e-01
-4.45693195e-01 7.37339631e-02 -1.17390263e+00 -4.54088032e-01
1.37943432e-01 2.94624299e-01 -7.72233188e-01 1.33121744e-01
-6.28641546e-01 -1.23546255e+00 7.02181160e-01 5.20697415e-01
8.74616683e-01 -1.00381052e+00 -5.96294761e-01 1.77182764e-01
-1.49695396e-01 -7.81975448e-01 -2.48558819e-01 -2.29657903e-01
-1.54582357e+00 -1.42867470e+00 -1.47633719e+00 -5.93430459e-01
6.50173783e-01 -1.58199549e-01 9.18025136e-01 1.20874971e-01
-4.07080680e-01 -6.06145943e-03 -4.75060642e-02 -5.95348887e-02
-3.16502988e-01 -1.11816414e-01 3.03230464e-01 -6.19818568e-01
4.84724283e-01 -1.16174316e+00 -9.58703995e-01 1.92425519e-01
-5.47410190e-01 -1.37618497e-01 7.86640525e-01 -1.10438585e-01
9.07246709e-01 -8.12375844e-01 7.94517398e-01 -6.11048937e-01
8.87935996e-01 -4.86804396e-01 -2.77524173e-01 -1.53972462e-01
-9.84592795e-01 -1.43472776e-01 1.12680420e-01 -2.00947806e-01
-6.86411679e-01 -3.16522568e-01 -1.86540633e-01 4.11618315e-02
-2.73219287e-01 1.35430992e-01 1.89183354e-01 -4.66091663e-01
7.17624128e-01 3.98445688e-02 7.91533053e-01 -6.83830142e-01
-8.57277587e-02 3.15210521e-01 4.51708406e-01 -4.78035539e-01
4.05236691e-01 6.79846644e-01 1.79968193e-01 -7.38855362e-01
1.01997092e-01 -1.32080078e-01 -1.42123485e+00 -3.29150110e-01
1.23824036e+00 -2.29379341e-01 -3.42477888e-01 7.58515418e-01
-8.98566067e-01 -7.49592900e-01 -2.54018735e-02 7.52774596e-01
-5.59808493e-01 5.56302190e-01 -6.10918343e-01 -2.87056059e-01
-1.10279882e+00 -1.22872281e+00 6.71260297e-01 -1.13746755e-01
-4.29782152e-01 -1.21869874e+00 4.86853451e-01 2.72009671e-01
8.03871751e-01 6.88672364e-01 8.81893158e-01 -3.99904609e-01
2.78699640e-02 -4.15237881e-02 -6.49236619e-01 1.61300808e-01
3.30939531e-01 -3.88068944e-01 -4.42369133e-01 -2.05616616e-02
6.26766160e-02 2.69127972e-02 7.76684284e-01 8.14938664e-01
8.71163487e-01 -1.13343596e-02 -2.84439087e-01 3.51181030e-01
1.39206994e+00 1.64008230e-01 9.63072598e-01 9.13757741e-01
5.01405060e-01 5.34198284e-01 -2.29552016e-01 -3.42504270e-02
4.88696456e-01 7.14253604e-01 3.40834297e-02 -1.19156614e-01
-7.28596210e-01 6.74886286e-01 2.95830637e-01 7.68292546e-01
-7.41590321e-01 1.02109647e+00 -1.22464454e+00 4.25014883e-01
-1.12302935e+00 -1.12504423e+00 -1.03255749e+00 2.25248694e+00
8.20219815e-01 1.70524232e-02 4.67734903e-01 1.18095931e-02
8.55346441e-01 -2.53216654e-01 -7.33080685e-01 -4.75637585e-01
-1.70056939e-01 6.63901925e-01 6.26347542e-01 2.97938108e-01
-7.23345339e-01 4.55313802e-01 6.85632467e+00 1.54787198e-01
-1.23090601e+00 7.21968412e-01 4.95220751e-01 -2.46488988e-01
-2.22667027e-02 -2.66490608e-01 -1.19504459e-01 5.28455913e-01
1.17628098e+00 -5.12029767e-01 3.61264497e-01 3.48113060e-01
6.12385154e-01 -5.19132316e-01 -5.90561330e-01 5.55971801e-01
-2.10853681e-01 -1.17954051e+00 -3.20719123e-01 4.66310024e-01
2.93993473e-01 6.45755827e-01 -3.68811578e-01 -1.97571084e-01
-7.48285651e-02 -9.71928239e-01 3.91907841e-01 1.15578246e+00
1.26893628e+00 -4.50331718e-01 7.66510189e-01 -1.25742182e-01
-1.29406595e+00 6.23986483e-01 -5.94166160e-01 -8.10772777e-02
3.64644170e-01 1.01292419e+00 -7.92900920e-01 3.80490609e-02
9.15094793e-01 2.78989106e-01 -6.97116911e-01 1.41872215e+00
-7.69248838e-03 4.42386448e-01 -2.61365473e-01 4.00639951e-01
-1.11781478e-01 -5.31448603e-01 4.54181939e-01 1.11912715e+00
2.96894968e-01 1.53143806e-02 -2.43609890e-01 1.62488019e+00
2.19664648e-01 1.91026270e-01 -3.88375223e-01 3.48803371e-01
3.62131357e-01 1.41657627e+00 -1.30710673e+00 -1.49026632e-01
-2.93485105e-01 5.59878170e-01 3.12660128e-01 5.59290163e-02
-4.42650020e-01 -2.61380404e-01 2.47551247e-01 9.16818738e-01
-3.76343220e-01 -6.06747925e-01 -8.02326381e-01 -5.48359096e-01
-1.29387109e-02 -3.27100635e-01 -5.94440810e-02 -8.13823998e-01
-1.21435308e+00 6.17938280e-01 2.89340049e-01 -6.52777731e-01
8.45288560e-02 -4.83690113e-01 -9.21620786e-01 1.30532146e+00
-1.14661777e+00 -8.34889293e-01 -2.40916327e-01 5.48589945e-01
2.35983387e-01 2.69021690e-02 9.28156316e-01 -5.30200638e-02
-3.76066804e-01 1.24072805e-01 1.03504464e-01 4.08523619e-01
6.91227913e-01 -1.45691156e+00 4.15629566e-01 5.49338579e-01
-6.10269845e-01 8.51628482e-01 1.84454635e-01 -1.13328469e+00
-8.25768173e-01 -1.06526840e+00 6.85363352e-01 -1.12482920e-01
6.87897742e-01 1.40476972e-01 -1.17233717e+00 6.30219042e-01
-3.76565345e-02 -3.45507562e-02 5.61518013e-01 -1.33352220e-01
-2.38046423e-01 8.97154212e-02 -1.88590932e+00 -3.37594338e-02
8.90451968e-01 -2.84484178e-01 -9.55184937e-01 2.29845658e-01
-1.37404546e-01 3.31639424e-02 -1.44367480e+00 2.19107807e-01
8.28428745e-01 -1.07524049e+00 9.93589759e-01 1.25197515e-01
2.10874721e-01 1.92378327e-01 3.92060548e-01 -1.16846871e+00
-6.79802001e-01 3.82232547e-01 4.90842909e-02 9.11827922e-01
1.39360800e-01 -9.55971777e-01 7.64563859e-01 8.62058342e-01
-2.89412618e-01 -6.00585401e-01 -1.50774992e+00 -1.12533832e+00
1.07727814e+00 -6.79136738e-02 2.13905185e-01 6.75781429e-01
1.84872061e-01 -2.99496740e-01 6.43311858e-01 -3.14543933e-01
1.10316968e+00 -4.52155739e-01 1.42368181e-02 -1.86982799e+00
6.62907481e-01 -7.27484167e-01 -6.92298234e-01 3.13372403e-01
3.26679707e-01 -1.54333258e+00 -3.09925735e-01 -2.21242213e+00
5.72448492e-01 -4.85088944e-01 -6.50132820e-02 5.26025057e-01
3.30458671e-01 5.86581707e-01 9.48587507e-02 4.92803663e-01
2.39142433e-01 3.23874891e-01 1.60403359e+00 -1.92435950e-01
-2.67426908e-01 -3.88536811e-01 -3.79634678e-01 8.77306640e-01
1.30093277e+00 -5.01380265e-01 -1.88226730e-01 -3.31525236e-01
-3.99906009e-01 -2.84705102e-01 6.60394430e-01 -1.38418782e+00
-2.95101613e-01 6.81049600e-02 7.98929095e-01 -1.27349377e-01
1.63643405e-01 -2.83887476e-01 2.05603074e-02 7.61312485e-01
2.11392120e-01 1.27427634e-02 1.84714794e-01 -2.23790839e-01
5.57772994e-01 -6.65751919e-02 1.16954279e+00 -5.89494050e-01
-3.25178862e-01 5.41105390e-01 -7.06263721e-01 -1.56219766e-01
7.11981356e-01 -5.45129836e-01 -4.97894168e-01 2.02581495e-01
-1.02076364e+00 -4.00680661e-01 3.47618490e-01 1.35931432e-01
7.83520341e-01 -1.31129086e+00 -1.04104054e+00 -3.30939293e-01
-6.32099926e-01 -3.30112934e-01 1.28402784e-01 1.81913519e+00
-7.74653435e-01 4.72981721e-01 -1.13661909e+00 -4.42245901e-01
-1.25766587e+00 4.88041528e-02 7.48541236e-01 -2.82750845e-01
-1.22456121e+00 6.06910169e-01 1.59365050e-02 -3.22278321e-01
-5.78643143e-01 -3.26920331e-01 -2.41072580e-01 -2.88392361e-02
8.08640420e-01 1.03904736e+00 1.94575593e-01 -1.00955355e+00
-7.60705829e-01 7.90527403e-01 1.52878312e-03 -3.53023440e-01
1.51379287e+00 -3.47869694e-01 -7.16815472e-01 4.70979005e-01
1.13616502e+00 -4.80431944e-01 -8.61552060e-01 1.54856637e-01
1.68299645e-01 5.36844768e-02 6.01767242e-01 -9.68953669e-01
-1.44019806e+00 8.55684757e-01 1.64880633e+00 -3.23089547e-02
6.80143297e-01 1.34255320e-01 9.61658657e-01 -2.75026560e-02
9.77343559e-01 -9.28816617e-01 -5.71458995e-01 1.86281756e-01
1.28759229e+00 -8.58996332e-01 2.26046413e-01 -9.11917090e-02
-1.21072922e-02 1.29987955e+00 6.11245155e-01 -7.09564388e-01
8.03938806e-01 4.82700616e-01 9.73736271e-02 -5.01825333e-01
2.06239015e-01 -4.09880653e-02 4.14426565e-01 1.11093616e+00
8.08875859e-01 2.11933434e-01 -1.16163862e+00 7.62904048e-01
-5.29455066e-01 3.11321884e-01 3.39405864e-01 6.15656853e-01
-9.27028060e-01 -1.10753655e+00 -7.18334675e-01 1.13860226e+00
-4.25157934e-01 1.43888146e-01 -6.51877761e-01 9.57489252e-01
2.19103470e-01 4.83011335e-01 2.18243480e-01 1.39971764e-03
-2.60178056e-02 2.00722873e-01 8.56048644e-01 -4.26046789e-01
-2.61974394e-01 -2.74961054e-01 -8.52572247e-02 -8.82837474e-01
-4.51068789e-01 -1.08654642e+00 -1.78082168e+00 -4.85154957e-01
9.71421748e-02 -4.36464995e-01 6.94680750e-01 1.36328673e+00
9.08603370e-02 3.96796167e-01 -6.02146648e-02 -1.22855461e+00
2.18484595e-01 -1.23960948e+00 -1.08878779e+00 2.53257871e-01
-1.42927036e-01 -1.04286778e+00 -1.44190997e-01 -1.48419105e-02] | [14.054495811462402, -2.119762420654297] |
44136e45-7430-474a-9cdd-bf60ebac0a9c | temporal-feature-warping-for-video-shadow | 2107.14287 | null | https://arxiv.org/abs/2107.14287v1 | https://arxiv.org/pdf/2107.14287v1.pdf | Temporal Feature Warping for Video Shadow Detection | While single image shadow detection has been improving rapidly in recent years, video shadow detection remains a challenging task due to data scarcity and the difficulty in modelling temporal consistency. The current video shadow detection method achieves this goal via co-attention, which mostly exploits information that is temporally coherent but is not robust in detecting moving shadows and small shadow regions. In this paper, we propose a simple but powerful method to better aggregate information temporally. We use an optical flow based warping module to align and then combine features between frames. We apply this warping module across multiple deep-network layers to retrieve information from neighboring frames including both local details and high-level semantic information. We train and test our framework on the ViSha dataset. Experimental results show that our model outperforms the state-of-the-art video shadow detection method by 28%, reducing BER from 16.7 to 12.0. | ['Dimitris Samaras', 'Hieu Le', 'Shilin Hu'] | 2021-07-29 | null | null | null | null | ['shadow-detection'] | ['computer-vision'] | [ 4.78141069e-01 -5.23577988e-01 -8.61267820e-02 -2.90796548e-01
-4.14896131e-01 -3.36221427e-01 5.37024021e-01 -4.63743925e-01
-4.04191285e-01 9.21509147e-01 4.40636456e-01 -2.28573009e-01
3.21270525e-01 -3.51062834e-01 -5.50593972e-01 -8.40272188e-01
-2.17383489e-01 -1.20254323e-01 1.06172431e+00 -2.37235101e-03
1.17558606e-01 2.87742823e-01 -1.43644822e+00 2.98541784e-01
9.52743709e-01 9.11590338e-01 7.14424491e-01 8.79202783e-01
1.83297113e-01 1.18030596e+00 -6.05991721e-01 2.84995347e-01
2.26677492e-01 -5.75470626e-01 -2.72709966e-01 1.51674850e-02
7.80409217e-01 -1.06339979e+00 -8.12109470e-01 5.61600089e-01
5.55643022e-01 1.91915765e-01 7.08879977e-02 -1.28176129e+00
-3.27087790e-01 7.51119480e-02 -5.67154527e-01 6.64052725e-01
4.52985674e-01 1.75272375e-01 7.52166092e-01 -8.69668484e-01
7.38027811e-01 1.24559569e+00 5.37982523e-01 3.66233200e-01
-9.09161270e-01 -7.78333366e-01 2.36877531e-01 6.42688811e-01
-1.01261854e+00 -5.60056686e-01 6.60135806e-01 -6.94053713e-04
7.67847776e-01 1.07355475e-01 7.08187759e-01 1.26425934e+00
4.18089718e-01 1.17110777e+00 1.22834134e+00 -2.68988851e-02
-3.68546583e-02 -4.18572843e-01 -1.20748401e-01 9.79094863e-01
7.33801872e-02 2.22178966e-01 -1.14845359e+00 7.29179606e-02
6.23436391e-01 9.30349752e-02 -8.25916350e-01 -2.55860597e-01
-1.31780457e+00 4.05129582e-01 5.32295763e-01 2.60843396e-01
-3.33745390e-01 4.89850968e-01 2.59499401e-01 -2.20962148e-02
4.49449509e-01 4.97812368e-02 -3.41977477e-01 -2.15223178e-01
-1.36126637e+00 1.19980581e-01 7.53749669e-01 6.27675533e-01
6.02124631e-01 2.75313795e-01 -5.80450714e-01 4.80028838e-01
2.36650258e-01 9.64856148e-01 1.01206422e-01 -1.31172872e+00
2.37025976e-01 -1.41298622e-01 2.76523709e-01 -1.02678430e+00
-4.16156262e-01 -2.27987483e-01 -5.53806245e-01 2.94143647e-01
2.35270306e-01 -9.98764783e-02 -1.21373355e+00 1.59225667e+00
1.17934987e-01 8.85945439e-01 -7.90439323e-02 1.28834045e+00
8.00380766e-01 8.63170803e-01 -1.96060047e-01 -4.04989451e-01
1.05500352e+00 -1.31186306e+00 -1.02108014e+00 -3.97641331e-01
3.68502624e-02 -9.46962833e-01 7.03968704e-01 2.22419530e-01
-7.98997641e-01 -4.63023216e-01 -1.26673520e+00 -1.14263490e-01
-1.01213954e-01 -1.54187739e-01 7.13492811e-01 4.27931041e-01
-1.16638589e+00 3.94068688e-01 -9.32796836e-01 -4.83415872e-01
6.35411859e-01 2.21368790e-01 -5.23940921e-02 -4.82231319e-01
-1.10459328e+00 7.27177203e-01 1.65177956e-01 1.48651615e-01
-1.33153248e+00 -6.91912234e-01 -7.76547670e-01 8.69948417e-02
7.54354775e-01 -6.68479204e-01 1.03993094e+00 -9.90721285e-01
-1.48470962e+00 1.14614092e-01 -8.18439126e-01 -6.85678959e-01
5.56446314e-01 -6.66462779e-01 -4.43034619e-01 2.88076907e-01
1.21629462e-01 6.54465914e-01 1.03216958e+00 -1.45619011e+00
-8.74644458e-01 1.14253893e-01 5.67026138e-02 3.15607399e-01
-2.29354858e-01 1.27888262e-01 -9.86433387e-01 -7.19542623e-01
-1.55533493e-01 -1.03422594e+00 -1.12823565e-02 2.60082752e-01
-3.11838657e-01 1.60526097e-01 1.73289490e+00 -7.61811852e-01
1.20368803e+00 -1.78660917e+00 -9.04754773e-02 -5.18634096e-02
3.96353900e-01 5.60029089e-01 -1.30226031e-01 2.07186893e-01
5.78650534e-01 -4.50990558e-01 -5.86091757e-01 -5.34340680e-01
-2.52485186e-01 3.75900090e-01 -5.12822032e-01 5.67441106e-01
5.12801632e-02 9.14637089e-01 -1.16878641e+00 -6.49300933e-01
5.75182796e-01 9.54995275e-01 -2.43884832e-01 3.73706102e-01
-1.49471924e-01 3.54997456e-01 -2.99226224e-01 1.02451718e+00
8.42008293e-01 -1.78412586e-01 2.77482450e-01 -2.49112546e-01
-2.24014968e-01 2.12280229e-01 -7.63519108e-01 1.72944069e+00
-4.49419409e-01 1.61264372e+00 3.39482427e-02 -4.14694160e-01
3.06940228e-01 1.00063100e-01 7.09271193e-01 -7.98042715e-01
4.60371971e-02 4.36059237e-02 -5.19251414e-02 -5.20485103e-01
7.13012874e-01 3.18301678e-01 2.82109141e-01 3.12991261e-01
-2.62084961e-01 2.12565213e-01 1.49311423e-01 3.25805962e-01
1.57492197e+00 4.58670795e-01 7.59784281e-02 -1.47221029e-01
3.86110693e-01 -2.76955485e-01 8.53902519e-01 7.93047786e-01
-5.33622086e-01 7.92057991e-01 1.34690523e-01 -2.64529109e-01
-6.04195714e-01 -1.27206218e+00 1.44441038e-01 1.06652820e+00
6.24174654e-01 -4.84425426e-01 -3.74705940e-01 -6.98727548e-01
8.36082697e-02 5.75744808e-01 -5.95207155e-01 1.20912619e-01
-8.69261146e-01 -6.09746099e-01 2.86093086e-01 5.82880259e-01
8.73475969e-01 -1.01481462e+00 -1.18243825e+00 2.43911654e-01
-4.58173722e-01 -1.58537364e+00 -5.95323026e-01 -4.56499904e-02
-3.91800612e-01 -1.04401028e+00 -8.04229915e-01 -4.28419501e-01
2.06059188e-01 1.24674380e+00 1.19256318e+00 2.98407882e-01
-3.44099522e-01 2.14621738e-01 -3.54768932e-01 -3.52042526e-01
-9.79534015e-02 -3.22345644e-01 -1.26396224e-01 1.06804319e-01
2.91561950e-02 -4.18824017e-01 -9.26936090e-01 2.85052180e-01
-9.39769506e-01 3.53741825e-01 7.70829916e-01 8.21701825e-01
1.43547162e-01 -1.59775779e-01 -7.43077463e-03 -6.28991663e-01
1.70983970e-01 -2.80963331e-01 -6.02905035e-01 1.33746296e-01
-6.11359417e-01 -1.13028526e-01 4.46322292e-01 -7.21878707e-02
-1.39990830e+00 4.91187945e-02 4.39311475e-01 -6.31938100e-01
3.80456983e-03 1.45067126e-01 -2.15969793e-02 -3.06201249e-01
1.67577848e-01 1.60367414e-01 -1.95875108e-01 -3.28115039e-02
3.29796582e-01 3.56797457e-01 7.51001596e-01 -3.75764109e-02
1.07748532e+00 1.13175154e+00 5.40344752e-02 -9.31744277e-01
-9.85399544e-01 -6.34700835e-01 -4.70741361e-01 -6.59544945e-01
8.03771615e-01 -1.02933824e+00 -5.52433431e-01 5.95555723e-01
-1.21514297e+00 -7.72379637e-01 3.81351888e-01 2.12735668e-01
-3.45202744e-01 7.24156439e-01 -4.71110880e-01 -6.66191697e-01
-2.73385376e-01 -1.11657345e+00 1.33762920e+00 1.75706595e-01
1.56018689e-01 -7.98619568e-01 -7.07493871e-02 3.22178900e-01
7.27338552e-01 3.61378253e-01 2.31652260e-02 1.80730909e-01
-1.35879326e+00 2.41503850e-01 -6.60755515e-01 2.05235586e-01
3.04122150e-01 -9.21867937e-02 -1.14364636e+00 -2.66270906e-01
-2.34545976e-01 -6.74113184e-02 1.65589297e+00 5.65892696e-01
1.06568754e+00 2.14101542e-02 -5.77288389e-01 7.08422542e-01
1.36524844e+00 1.04159966e-01 6.65250719e-01 2.60484308e-01
1.12668622e+00 2.68721163e-01 9.23691571e-01 3.73965681e-01
4.31158781e-01 7.96673775e-01 3.99588019e-01 -3.15943062e-01
-7.07170844e-01 1.53175816e-01 6.78818703e-01 3.31447810e-01
-1.80697635e-01 -7.04113007e-01 -9.07486975e-01 4.75395828e-01
-2.23409009e+00 -1.23729265e+00 4.61078174e-02 1.84284699e+00
4.86235261e-01 2.73343474e-01 -3.43713552e-01 -1.82803616e-01
3.99855644e-01 9.57772374e-01 -5.44573665e-01 2.21598238e-01
-4.90410537e-01 -2.23068669e-02 8.65835249e-01 8.35764587e-01
-1.29425824e+00 1.41420233e+00 6.02009678e+00 5.39667189e-01
-1.10038352e+00 1.05708718e-01 2.72417992e-01 -3.03915650e-01
-1.47903532e-01 1.45967588e-01 -4.01118815e-01 6.02491438e-01
7.06600249e-01 2.13621736e-01 5.02794385e-01 3.15773994e-01
5.81865847e-01 -7.38360286e-01 -8.07251930e-01 8.65488291e-01
5.62643886e-01 -1.57806063e+00 -2.25876108e-01 -1.40490225e-02
9.47681248e-01 3.82779241e-01 -4.87728007e-02 7.75079057e-02
1.87364787e-01 -8.98470044e-01 7.24582613e-01 7.16865897e-01
5.50076604e-01 -4.86467779e-01 7.34987020e-01 -1.63410217e-01
-1.49817610e+00 -1.58174530e-01 -9.81114618e-03 -8.66775587e-02
5.19985855e-01 7.01179564e-01 -9.51172531e-01 3.25892985e-01
1.04053783e+00 1.01574540e+00 -4.49724168e-01 1.05301380e+00
-2.60613501e-01 6.26223147e-01 -2.81045228e-01 1.10771060e-01
2.40537703e-01 -1.83448177e-02 5.60723603e-01 1.43377960e+00
2.09034562e-01 2.23840997e-02 4.15376127e-01 3.90282184e-01
-4.08709422e-02 -5.44174850e-01 -5.95100760e-01 3.00247163e-01
4.80006874e-01 1.19659531e+00 -7.01925755e-01 -5.79050779e-01
-6.54110432e-01 1.52649498e+00 -6.34013414e-02 6.73654020e-01
-1.27051437e+00 -9.02890116e-02 9.48356211e-01 -2.72649586e-01
4.37987357e-01 -5.83534777e-01 7.88178593e-02 -1.09077144e+00
5.50403707e-02 -5.19165695e-01 6.50108904e-02 -1.07753778e+00
-5.97260952e-01 3.75760913e-01 -3.31297964e-01 -1.16494310e+00
-1.24538526e-01 -4.27071989e-01 -5.52400947e-01 4.31732088e-01
-2.07171249e+00 -1.17584515e+00 -9.93155777e-01 5.40921092e-01
8.09582174e-01 4.74873818e-02 3.02637279e-01 2.59058625e-01
-4.75565463e-01 2.53931016e-01 1.92021415e-01 5.58211431e-02
1.15300119e+00 -1.10608709e+00 3.79164547e-01 1.40115070e+00
5.36395013e-02 1.24005660e-01 9.92612362e-01 -7.33557224e-01
-1.55531883e+00 -1.14951694e+00 7.19241500e-01 -4.02402788e-01
6.61933839e-01 -3.46095204e-01 -9.37492192e-01 3.00028116e-01
5.32102346e-01 3.97941172e-01 2.14273840e-01 -3.51646394e-01
-4.30864215e-01 -2.77314812e-01 -6.20154202e-01 7.27361977e-01
1.32525885e+00 -6.14925802e-01 -1.32419243e-01 3.25607210e-01
8.37032378e-01 -5.73957860e-01 -3.75894010e-01 5.78956962e-01
7.80319393e-01 -1.51552129e+00 9.88179028e-01 7.11701959e-02
2.28175402e-01 -6.16798043e-01 -3.63400102e-01 -1.00472450e+00
-1.84595600e-01 -9.51153874e-01 -6.33453012e-01 8.44685853e-01
-2.79692616e-02 -3.56956571e-01 8.66620660e-01 1.21475652e-01
-3.35814863e-01 -7.24891484e-01 -8.33880424e-01 -6.89023018e-01
-7.29061484e-01 -4.54353005e-01 6.78675547e-02 6.40312254e-01
-4.53419060e-01 2.06396103e-01 -8.19103241e-01 3.48758310e-01
7.38790035e-01 5.61669469e-01 8.76904964e-01 -8.81430745e-01
-1.78006917e-01 -3.44590157e-01 -1.81067407e-01 -1.19662941e+00
4.32370663e-01 -3.08721840e-01 5.83357811e-01 -1.71837425e+00
4.05315399e-01 -1.10460119e-02 -4.77967143e-01 3.18213493e-01
-4.68029618e-01 3.88554543e-01 5.17409325e-01 2.23564893e-01
-9.97768402e-01 7.81940520e-01 1.08126175e+00 -6.34292588e-02
-4.35902327e-02 -3.75364095e-01 -1.00646414e-01 5.75945735e-01
6.68651342e-01 -1.67203978e-01 -3.05822045e-01 -4.57460672e-01
-4.64507043e-01 3.07794809e-02 7.36628234e-01 -1.34507203e+00
2.42728218e-01 -2.39901647e-01 5.54736972e-01 -8.10125887e-01
7.61725366e-01 -8.53021681e-01 -3.50693077e-01 5.52000523e-01
1.13855712e-01 -1.27801999e-01 1.31194770e-01 1.02984571e+00
-3.87859456e-02 6.99999511e-01 6.28214717e-01 2.62622863e-01
-1.36226594e+00 3.74981195e-01 -4.76586580e-01 -2.24215269e-01
8.45008850e-01 -6.76683411e-02 -6.42509520e-01 -6.36136949e-01
8.61648992e-02 3.69288474e-01 6.35742247e-01 6.91899717e-01
9.07236397e-01 -9.72215593e-01 -5.83348691e-01 -1.59756616e-02
-7.74929225e-02 -3.47501576e-01 1.75031751e-01 1.05139244e+00
-5.13178706e-01 6.04330838e-01 -2.01180458e-01 -7.65164375e-01
-1.54172301e+00 2.80533373e-01 2.52360195e-01 1.14830412e-01
-7.90407121e-01 7.47709334e-01 3.13812524e-01 5.17919302e-01
3.95559013e-01 -3.31242263e-01 1.04393288e-01 -1.37161359e-01
5.49656034e-01 3.95867318e-01 -2.80937791e-01 -7.08318055e-01
-5.72654724e-01 4.02261198e-01 1.97283402e-01 -1.09394751e-01
9.97320056e-01 -3.49400789e-01 -1.10854721e-02 2.85571128e-01
1.23557115e+00 -4.60362108e-03 -2.06463289e+00 -3.95720065e-01
-1.41365066e-01 -8.68522823e-01 3.63997489e-01 -7.13504434e-01
-1.17639685e+00 5.76164246e-01 7.68422246e-01 -1.19005637e-02
1.34939456e+00 -1.46503439e-02 1.26396537e+00 3.83311719e-01
3.13850790e-01 -1.08376229e+00 3.28530461e-01 6.60203874e-01
7.06260383e-01 -1.67237747e+00 3.21923196e-01 -4.47753131e-01
-4.40151244e-01 1.00319123e+00 5.38530648e-01 -5.41265346e-02
6.11710310e-01 4.47378963e-01 2.13636696e-01 -2.43061986e-02
-9.03130412e-01 -5.67829728e-01 3.89369488e-01 5.06158173e-01
2.87044168e-01 -5.64694516e-02 -7.83892870e-02 -1.43149287e-01
3.11316818e-01 -3.00869998e-02 5.97705781e-01 1.07353520e+00
-6.67025864e-01 -7.64468193e-01 -4.33780998e-01 3.01657200e-01
-1.94917664e-01 -2.58708328e-01 -3.69122058e-01 8.18635404e-01
-5.43899275e-03 1.18741202e+00 3.61450650e-02 -5.43215990e-01
-2.10021317e-01 -3.90225977e-01 3.85117501e-01 -8.41711983e-02
7.79123977e-02 2.30679467e-01 3.45449984e-01 -1.33312881e+00
-9.69095528e-01 -8.53426635e-01 -1.34188402e+00 -2.86499113e-01
-9.15788785e-02 -4.14875835e-01 5.13093412e-01 1.18361223e+00
3.27728689e-01 8.96419466e-01 4.81711447e-01 -1.33868754e+00
2.64255136e-01 -6.17821515e-01 -2.10066527e-01 9.56580937e-02
9.76587594e-01 -8.86051416e-01 -2.82761693e-01 8.36604983e-02] | [10.835538864135742, -4.1108622550964355] |
9ea56af6-3b27-4302-8134-e3725c837441 | adding-guardrails-to-advanced-chatbots | 2306.07500 | null | https://arxiv.org/abs/2306.07500v1 | https://arxiv.org/pdf/2306.07500v1.pdf | Adding guardrails to advanced chatbots | Generative AI models continue to become more powerful. The launch of ChatGPT in November 2022 has ushered in a new era of AI. ChatGPT and other similar chatbots have a range of capabilities, from answering student homework questions to creating music and art. There are already concerns that humans may be replaced by chatbots for a variety of jobs. Because of the wide spectrum of data chatbots are built on, we know that they will have human errors and human biases built into them. These biases may cause significant harm and/or inequity toward different subpopulations. To understand the strengths and weakness of chatbot responses, we present a position paper that explores different use cases of ChatGPT to determine the types of questions that are answered fairly and the types that still need improvement. We find that ChatGPT is a fair search engine for the tasks we tested; however, it has biases on both text generation and code generation. We find that ChatGPT is very sensitive to changes in the prompt, where small changes lead to different levels of fairness. This suggests that we need to immediately implement "corrections" or mitigation strategies in order to improve fairness of these systems. We suggest different strategies to improve chatbots and also advocate for an impartial review panel that has access to the model parameters to measure the levels of different types of biases and then recommends safeguards that move toward responses that are less discriminatory and more accurate. | ['Lisa Singh', 'Yanchen Wang'] | 2023-06-13 | null | null | null | null | ['code-generation', 'chatbot', 'chatbot'] | ['computer-code', 'methodology', 'natural-language-processing'] | [-2.90413618e-01 4.33961719e-01 -3.79722901e-02 -4.25691217e-01
-6.91642582e-01 -6.60297513e-01 5.98473728e-01 3.23029272e-02
-3.60807568e-01 9.57356453e-01 6.89140856e-01 -6.60017073e-01
-1.64336562e-01 -7.51588225e-01 -1.84104726e-01 -3.73747766e-01
6.86758995e-01 7.86403477e-01 2.59822682e-02 -5.80022216e-01
6.89397454e-01 3.77417430e-02 -1.14313471e+00 3.21978480e-01
1.15457845e+00 6.27356321e-02 -1.35864437e-01 8.18165779e-01
-3.66981000e-01 1.51250064e+00 -1.45924294e+00 -9.06425655e-01
2.47740418e-01 -7.48460531e-01 -1.23762929e+00 -5.76380789e-01
2.44010404e-01 -3.30244988e-01 -4.53324094e-02 8.37884486e-01
8.90818298e-01 1.49958372e-01 4.83958155e-01 -1.50953114e+00
-1.01464164e+00 9.82447028e-01 -1.71380237e-01 3.38845670e-01
5.29896438e-01 6.34721935e-01 7.62145936e-01 -1.58026740e-01
6.33520067e-01 1.51836514e+00 9.11823153e-01 7.94818759e-01
-1.10250604e+00 -1.14290202e+00 -3.00361425e-01 -1.57514997e-02
-9.60307121e-01 -5.16404927e-01 2.08017454e-01 -6.03272676e-01
7.95155466e-01 7.22638488e-01 5.93110085e-01 1.17250395e+00
3.40606511e-01 1.33797899e-01 1.38677180e+00 -2.91814744e-01
1.65439188e-01 5.40646553e-01 3.24859358e-02 4.05662686e-01
4.18560803e-01 -1.51829407e-01 -5.85261464e-01 -6.86528385e-01
4.56993759e-01 -3.45489383e-01 5.09882383e-02 3.41296047e-01
-8.67606997e-01 1.17732120e+00 2.87188500e-01 5.21037817e-01
-1.45858616e-01 2.70036489e-01 2.51951873e-01 3.42701346e-01
5.50216258e-01 1.17650974e+00 -2.88733859e-02 -8.20444882e-01
-6.90200031e-01 5.94067514e-01 1.20560288e+00 7.96628356e-01
4.49875861e-01 -2.06366330e-01 -5.59309661e-01 1.02700889e+00
1.35507450e-01 2.56223589e-01 4.30378348e-01 -1.45664465e+00
4.65509266e-01 6.95155263e-01 1.95949122e-01 -1.02623832e+00
-1.31728649e-01 -8.70424137e-02 -2.27835685e-01 2.98387170e-01
8.72737586e-01 -5.19866467e-01 -4.83847439e-01 1.55820966e+00
6.56625479e-02 -4.98167038e-01 -4.25578833e-01 7.46423125e-01
6.77206278e-01 4.37568516e-01 2.85345852e-01 3.30971837e-01
1.01890147e+00 -4.47065413e-01 -6.56589806e-01 -3.66569608e-01
6.40483558e-01 -1.06129718e+00 1.24961150e+00 -3.52707915e-02
-1.08044803e+00 -1.28863789e-02 -5.67158222e-01 -1.73659429e-01
-2.49846891e-01 -6.51874840e-01 5.50192118e-01 1.36948538e+00
-8.35858047e-01 6.81625485e-01 -4.59900111e-01 -6.02103531e-01
4.17059541e-01 9.02998224e-02 1.06278047e-01 -9.23281312e-02
-1.23242557e+00 1.33153450e+00 -2.79032588e-01 -8.17790180e-02
-4.42342103e-01 -7.25921273e-01 -3.14214438e-01 1.65375948e-01
2.76772380e-01 -6.11677349e-01 1.39036155e+00 -8.11142981e-01
-1.36476409e+00 6.99903429e-01 1.50998190e-01 -1.92974433e-01
7.90397942e-01 1.16067111e-01 4.41639014e-02 -3.51294190e-01
4.75584686e-01 4.79021937e-01 2.73152262e-01 -9.48184788e-01
-5.99646688e-01 -1.50612295e-01 3.04561049e-01 1.94356143e-01
-1.62743226e-01 6.10571146e-01 4.30972278e-01 -3.18810433e-01
-5.56200027e-01 -1.08519328e+00 -8.00419599e-02 -4.52998072e-01
-2.60221332e-01 -3.69747698e-01 5.24975240e-01 -6.48924053e-01
1.08447683e+00 -1.56391454e+00 -4.40755904e-01 -4.00933698e-02
2.70524353e-01 1.83377072e-01 -1.53559223e-02 6.50785685e-01
4.71459985e-01 9.60980058e-01 2.48673618e-01 -3.52373049e-02
2.33411252e-01 1.73633248e-01 -7.85218328e-02 1.15256384e-01
-4.59001251e-02 8.34432721e-01 -9.77404416e-01 -5.20338833e-01
-2.47967511e-01 2.09810719e-01 -7.92798579e-01 8.88284296e-02
1.36298807e-02 3.80819261e-01 -3.06565464e-01 4.04276758e-01
2.40080476e-01 -1.70802295e-01 -5.21414028e-03 8.92552972e-01
-3.93675923e-01 9.42312419e-01 -4.20686394e-01 9.55204546e-01
-3.25052917e-01 8.78765106e-01 7.35059977e-02 -5.29244721e-01
9.13955688e-01 3.56335938e-01 3.90428603e-02 -5.39763808e-01
3.05830151e-01 2.80389696e-01 5.14582634e-01 -5.67950964e-01
7.25229502e-01 -3.64491016e-01 -2.53761441e-01 9.87936676e-01
-2.80972421e-01 -6.83970809e-01 7.30228573e-02 3.62630635e-01
1.34450734e+00 -3.91495675e-01 -2.43724864e-02 -4.44159806e-01
-2.17223734e-01 4.84987110e-01 4.59716886e-01 1.24263000e+00
-5.16788065e-01 3.83431196e-01 7.75610149e-01 -3.40104640e-01
-1.01094782e+00 -4.85060602e-01 1.03816673e-01 1.43679047e+00
-2.41622850e-01 -2.68627435e-01 -7.57998824e-01 -6.14038825e-01
-4.21110094e-02 1.29404747e+00 -5.99456906e-01 -3.74427140e-01
-3.27699512e-01 -6.65776372e-01 1.07320130e+00 8.13905969e-02
3.68935406e-01 -1.25255275e+00 -8.60564590e-01 -4.13493887e-02
-5.41272879e-01 -6.26615822e-01 -6.01823926e-01 -1.01985209e-01
-6.43007100e-01 -9.33666348e-01 -6.80794716e-01 -3.33312035e-01
3.17922652e-01 1.93852991e-01 1.11657977e+00 4.67420876e-01
-1.37115523e-01 4.30616111e-01 -4.82173443e-01 -9.80081737e-01
-1.12522840e+00 2.88116813e-01 -5.28424263e-01 -7.55633652e-01
4.75914925e-01 -4.35287774e-01 -3.60943675e-01 3.95838678e-01
-4.36201721e-01 -1.81242451e-01 3.78132492e-01 6.99824154e-01
-6.32686436e-01 -2.94839710e-01 8.46914768e-01 -1.22768712e+00
1.43077326e+00 -6.82341099e-01 -1.15719393e-01 1.80467650e-01
-8.27980042e-01 -2.03830838e-01 2.98758954e-01 -3.66046041e-01
-1.18302798e+00 -6.72772706e-01 -1.09050116e-02 2.12044135e-01
-1.31053388e-01 3.58000308e-01 2.78509408e-01 -5.12396917e-02
1.28744030e+00 -4.55147862e-01 2.81861663e-01 -2.38007620e-01
9.39452425e-02 9.90871966e-01 -1.80170722e-02 -6.99735105e-01
6.39220655e-01 3.76698636e-02 -6.93597555e-01 -6.05957210e-01
-4.41746891e-01 -4.87405322e-02 2.22250164e-01 -5.38004518e-01
7.84174979e-01 -4.99155402e-01 -9.70381737e-01 1.60593256e-01
-1.07788384e+00 -8.94674659e-01 -2.57795602e-01 2.82689631e-01
-1.43164203e-01 5.37748151e-02 -8.05809557e-01 -9.98245239e-01
-3.42782378e-01 -8.63989055e-01 4.77987349e-01 5.21732926e-01
-1.05118823e+00 -7.39190161e-01 3.53647709e-01 1.06766713e+00
9.13941681e-01 -6.84028566e-02 1.04216385e+00 -9.24671888e-01
-3.77058595e-01 -3.51491451e-01 -6.22530431e-02 1.41652152e-01
1.23029143e-01 2.15102375e-01 -7.75335371e-01 8.45229328e-02
7.94544220e-02 -3.64459068e-01 2.05433205e-01 2.24172771e-01
6.06367469e-01 -8.12292278e-01 -7.63448775e-02 -9.14371759e-02
7.93525696e-01 5.81614196e-01 7.77454555e-01 4.48560923e-01
4.34014291e-01 8.42951775e-01 3.01480204e-01 3.42726529e-01
5.89294553e-01 4.00801688e-01 1.00158177e-01 3.72597575e-01
5.92682958e-02 -3.89088184e-01 4.29298401e-01 6.80704772e-01
-2.84277320e-01 -2.98326284e-01 -1.17139626e+00 6.15992546e-01
-1.75024724e+00 -1.34724104e+00 -2.45118946e-01 2.17214727e+00
9.37508047e-01 1.47289768e-01 5.22462189e-01 -2.99858004e-01
8.92292976e-01 -8.78881887e-02 -2.57889658e-01 -1.10272717e+00
3.90706271e-01 2.87820965e-01 4.30533051e-01 5.25268972e-01
-2.64781237e-01 7.30879307e-01 7.78335094e+00 4.49979275e-01
-9.24194634e-01 2.16733903e-01 8.22527826e-01 -2.17659667e-01
-5.17482638e-01 2.59469509e-01 -3.56919795e-01 7.41476893e-01
1.04421580e+00 -8.27416658e-01 6.66019559e-01 7.80750811e-01
2.07094654e-01 -3.04682434e-01 -6.53609931e-01 3.30130488e-01
-2.76337415e-02 -1.28983295e+00 -3.54473919e-01 7.39418045e-02
8.31464469e-01 -4.84748967e-02 -1.51375026e-01 5.64594448e-01
1.26369107e+00 -1.27724993e+00 7.07974434e-01 4.03113306e-01
2.41289839e-01 -6.96422338e-01 8.38397145e-01 5.91819227e-01
-1.96793064e-01 -8.93485695e-02 -4.28367704e-01 -7.98083246e-01
-6.46527782e-02 3.44912082e-01 -1.17615938e+00 1.78243667e-02
7.46017277e-01 -1.48560842e-02 -6.33380592e-01 1.06698143e+00
-2.00083107e-01 8.14282179e-01 -3.22327353e-02 -7.25803673e-01
-8.01685974e-02 3.07247583e-02 4.38780665e-01 9.78404462e-01
3.97623062e-01 1.94460228e-01 -3.78208727e-01 1.20179975e+00
-3.52716632e-02 -5.60034402e-02 -6.54890954e-01 -4.04591024e-01
9.71589148e-01 1.07773781e+00 -6.03650212e-01 -2.93627560e-01
-3.65953818e-02 4.76063818e-01 9.28280652e-02 8.15310702e-02
-7.27696180e-01 -4.89131451e-01 8.58798385e-01 2.27422193e-01
-4.04275894e-01 1.19199730e-01 -6.89305127e-01 -7.39442885e-01
-5.40699005e-01 -1.34738314e+00 1.60582498e-01 -9.59381700e-01
-1.36566782e+00 2.74376273e-01 -1.34231433e-01 -4.45336312e-01
-4.28332329e-01 3.88904917e-03 -9.27617431e-01 1.04178059e+00
-6.80928707e-01 -5.04403353e-01 -3.65923345e-01 2.32421771e-01
2.33954221e-01 6.10289536e-03 6.14595234e-01 1.91620439e-01
-2.65290976e-01 7.34623253e-01 -2.26132110e-01 3.06960225e-01
1.08483100e+00 -9.67995942e-01 3.91216099e-01 4.24511522e-01
-6.48140758e-02 8.77557695e-01 9.36171472e-01 -7.37969816e-01
-7.58365035e-01 -6.13531649e-01 1.27292311e+00 -1.12033737e+00
4.87928480e-01 -2.63205022e-01 -6.75978065e-01 8.70637000e-01
5.05565822e-01 -1.06673205e+00 7.67589986e-01 4.61593539e-01
-1.68021560e-01 3.08342814e-01 -1.24368846e+00 8.01786602e-01
8.79280508e-01 -6.26371205e-01 -6.53578699e-01 4.27958310e-01
6.27085686e-01 -3.29454839e-01 -5.08481383e-01 -2.82387048e-01
6.52421772e-01 -1.27633047e+00 4.71297175e-01 -6.49497271e-01
7.70496786e-01 2.44584575e-01 4.00426507e-01 -1.62056386e+00
-5.74927092e-01 -9.13379133e-01 9.63582814e-01 1.48482275e+00
5.14535367e-01 -9.99998868e-01 8.39884698e-01 1.46615100e+00
-1.22662984e-01 -3.15381110e-01 -7.70822048e-01 -6.04100823e-01
5.60311735e-01 -1.57924369e-01 8.27138603e-01 1.40455461e+00
5.54573119e-01 4.61856753e-01 -4.77515548e-01 -3.54149699e-01
1.73145831e-01 -2.35072464e-01 9.77745652e-01 -1.11743772e+00
-6.48172572e-02 -7.07766771e-01 -1.35256588e-01 -2.20566779e-01
-2.00881720e-01 -7.67261088e-01 2.49196693e-01 -1.70272481e+00
3.57212901e-01 -6.42620265e-01 3.59965235e-01 6.19666815e-01
-4.40757126e-01 5.33355623e-02 5.13253570e-01 2.69600004e-01
-1.98427334e-01 7.95292482e-02 1.12811780e+00 -2.19132006e-02
-3.62007678e-01 8.58250409e-02 -1.53384936e+00 4.30202007e-01
1.03495765e+00 -9.29658294e-01 -2.10281581e-01 -3.23771238e-01
5.25509477e-01 -8.00309330e-02 3.96531492e-01 -9.28747833e-01
2.79865652e-01 -5.47557056e-01 1.10680507e-02 2.66208738e-01
-5.85055239e-02 -3.77385229e-01 3.79487842e-01 5.31374872e-01
-7.52912223e-01 1.37721837e-01 7.69858360e-02 -3.30837592e-02
2.33295441e-01 -5.32266915e-01 5.77969193e-01 -3.32506478e-01
3.09733957e-01 -3.16798955e-01 -1.05972660e+00 5.02153277e-01
6.66146874e-01 -8.12632442e-02 -9.40590739e-01 -1.07772577e+00
-1.59430385e-01 2.39278838e-01 5.92497051e-01 4.91343111e-01
-1.14548802e-01 -1.02177799e+00 -8.38066280e-01 -4.23250973e-01
-1.37193009e-01 -3.55083734e-01 1.03119805e-01 7.19884098e-01
-6.87951863e-01 3.73286188e-01 -4.49819207e-01 6.11508079e-02
-1.27072966e+00 1.02095768e-01 3.47503036e-01 -1.53422713e-01
-2.86080927e-01 1.02140462e+00 -1.62976325e-01 -5.57733893e-01
-9.38155316e-03 -7.01375231e-02 -2.62767263e-02 4.39541601e-02
3.63308430e-01 7.65387535e-01 -1.39378250e-01 -2.61622339e-01
-2.45308846e-01 -3.04762572e-01 4.51879241e-02 -2.84268111e-01
1.15758622e+00 1.27421886e-01 -2.81042784e-01 4.40737963e-01
5.04423857e-01 4.17569071e-01 -8.43871236e-01 4.73125100e-01
-3.83332223e-02 -9.31536376e-01 -5.00085652e-01 -1.21777475e+00
-6.64357364e-01 7.26621985e-01 9.27995369e-02 7.09765673e-01
4.86548662e-01 -4.18007597e-02 5.26786625e-01 2.28994370e-01
3.99629533e-01 -1.24699295e+00 1.07452847e-01 5.17278194e-01
9.05995727e-01 -8.92603695e-01 -3.89830917e-02 -2.68136915e-02
-9.04335022e-01 6.42295659e-01 6.79649115e-01 -1.40487440e-02
6.92920834e-02 1.50798559e-01 3.86642277e-01 -3.52935880e-01
-1.01269877e+00 1.96476892e-01 -3.14925820e-01 7.17350841e-01
8.49544168e-01 2.33848810e-01 -8.25413942e-01 4.90720481e-01
-8.72727573e-01 4.68134061e-02 1.01865911e+00 6.87594771e-01
-4.76975411e-01 -1.26677358e+00 -8.72637987e-01 6.66399181e-01
-4.73588169e-01 -3.71026397e-02 -1.23941457e+00 8.50549281e-01
1.00035816e-01 1.50008273e+00 -7.67062679e-02 -5.91656268e-01
2.32258409e-01 5.67765415e-01 1.84190363e-01 -7.88421571e-01
-1.28113270e+00 -4.67790663e-01 6.09166324e-01 -2.37095699e-01
-1.13636777e-01 -6.84375882e-01 -8.58698487e-01 -1.16788566e+00
-4.28031623e-01 5.07065237e-01 4.72097039e-01 8.47039402e-01
4.74320292e-01 3.39545518e-01 2.08667368e-01 -4.46044981e-01
-5.99814594e-01 -1.18420339e+00 -2.31109753e-01 2.91320324e-01
-2.52707303e-01 -3.21157038e-01 -4.83040333e-01 -4.79261041e-01] | [10.36596965789795, 7.534711837768555] |
94ee2b1b-722c-4f24-92c8-c16407074d3a | modetr-moving-object-detection-with | 2106.11422 | null | https://arxiv.org/abs/2106.11422v1 | https://arxiv.org/pdf/2106.11422v1.pdf | MODETR: Moving Object Detection with Transformers | Moving Object Detection (MOD) is a crucial task for the Autonomous Driving pipeline. MOD is usually handled via 2-stream convolutional architectures that incorporates both appearance and motion cues, without considering the inter-relations between the spatial or motion features. In this paper, we tackle this problem through multi-head attention mechanisms, both across the spatial and motion streams. We propose MODETR; a Moving Object DEtection TRansformer network, comprised of multi-stream transformer encoders for both spatial and motion modalities, and an object transformer decoder that produces the moving objects bounding boxes using set predictions. The whole architecture is trained end-to-end using bi-partite loss. Several methods of incorporating motion cues with the Transformer model are explored, including two-stream RGB and Optical Flow (OF) methods, and multi-stream architectures that take advantage of sequence information. To incorporate the temporal information, we propose a new Temporal Positional Encoding (TPE) approach to extend the Spatial Positional Encoding(SPE) in DETR. We explore two architectural choices for that, balancing between speed and time. To evaluate the our network, we perform the MOD task on the KITTI MOD [6] data set. Results show significant 5% mAP of the Transformer network for MOD over the state-of-the art methods. Moreover, the proposed TPE encoding provides 10% mAP improvement over the SPE baseline. | ['Ahmad El-Sallab', 'Eslam Mohamed'] | 2021-06-21 | null | null | null | null | ['moving-object-detection'] | ['computer-vision'] | [ 2.33909905e-01 -2.77571887e-01 -4.56571579e-02 -4.05956060e-01
-7.09813118e-01 -3.44486952e-01 9.28485692e-01 -2.92880923e-01
-8.57238770e-01 2.58662164e-01 9.35656875e-02 -2.82592714e-01
8.81625339e-02 -6.90725207e-01 -9.54357684e-01 -7.42073417e-01
-1.14571638e-01 2.22813860e-01 9.05054867e-01 -3.04977983e-01
2.93466836e-01 3.75368416e-01 -1.88322091e+00 5.42986929e-01
6.44926310e-01 1.37146592e+00 4.46089149e-01 1.09617269e+00
-1.04783371e-01 1.22931802e+00 -8.42333958e-02 -3.17358226e-01
4.35559630e-01 -2.31362820e-01 -5.46421826e-01 -8.79271924e-02
7.40512192e-01 -5.29414237e-01 -4.92371708e-01 6.35535538e-01
4.59768951e-01 2.99961835e-01 4.42201972e-01 -1.47438109e+00
-1.94665313e-01 2.96994299e-01 -6.48973823e-01 5.44520736e-01
-1.44137681e-01 6.71124518e-01 9.06889856e-01 -1.00653028e+00
5.15077889e-01 1.40039861e+00 5.08686066e-01 4.70161229e-01
-9.39996123e-01 -6.51238263e-01 3.19019824e-01 8.35679352e-01
-1.07223141e+00 -7.03740954e-01 5.71068823e-01 -6.56858325e-01
1.19595373e+00 -3.84003222e-02 4.43239242e-01 1.00999427e+00
4.40890670e-01 1.02722490e+00 6.52643681e-01 -1.13921642e-01
1.38747692e-01 -1.22347241e-02 9.77835804e-02 5.74179828e-01
-1.29523143e-01 4.81468350e-01 -9.00435150e-01 4.67784494e-01
4.62908953e-01 -5.20784035e-03 -1.14139654e-01 -4.06743824e-01
-1.22975516e+00 4.96420294e-01 5.98226070e-01 -4.53667220e-04
-3.45065236e-01 6.27899230e-01 3.95838886e-01 -1.18678339e-01
3.54663312e-01 3.06565166e-02 -3.88714850e-01 -3.24570060e-01
-1.23666036e+00 3.63736302e-01 2.39091158e-01 1.08972847e+00
8.04595590e-01 -2.04643384e-02 -5.30891776e-01 2.85534948e-01
6.48936868e-01 5.11780560e-01 4.86533105e-01 -9.30898011e-01
8.29263330e-01 3.41356099e-01 1.61165565e-01 -6.24703825e-01
-5.00083983e-01 -3.47466111e-01 -5.17520189e-01 5.14524579e-01
3.74125749e-01 9.15073231e-02 -1.09976208e+00 1.70272887e+00
4.37031299e-01 5.27762473e-01 7.91876465e-02 1.21696246e+00
6.89486384e-01 7.47177541e-01 2.70249277e-01 1.69216290e-01
1.43916082e+00 -1.51875973e+00 -6.23566091e-01 -5.23727536e-01
7.11467922e-01 -6.14749551e-01 7.27968097e-01 2.05602244e-01
-1.23469567e+00 -9.39700127e-01 -1.19967067e+00 -5.56102157e-01
-4.16734785e-01 2.77009457e-01 2.23354459e-01 3.54929030e-01
-1.16101503e+00 8.00285339e-01 -1.30384791e+00 -2.08691329e-01
4.47985411e-01 3.39209557e-01 -2.35653907e-01 -1.72696352e-01
-1.03188229e+00 9.22031224e-01 2.88826704e-01 3.99152398e-01
-1.06166530e+00 -6.37289166e-01 -1.06595194e+00 2.91157495e-02
2.95592286e-02 -8.41204226e-01 1.25025117e+00 -9.07647848e-01
-1.41078460e+00 5.53641379e-01 -6.03696704e-01 -7.47676253e-01
8.19501758e-01 -4.92624223e-01 -1.97225928e-01 2.55575329e-01
1.60990357e-01 1.04519844e+00 1.02107728e+00 -8.53247643e-01
-1.13780475e+00 -2.25703716e-01 -2.59179939e-02 1.79123163e-01
-6.62218928e-02 -5.07503711e-02 -5.46380401e-01 -4.28695709e-01
-1.94920182e-01 -8.93798053e-01 -1.07966542e-01 4.10480022e-01
-2.22658962e-01 -9.13763866e-02 1.16821957e+00 -5.62009513e-01
1.08619797e+00 -2.37806511e+00 2.53793329e-01 -4.44125712e-01
2.21753530e-02 3.89116883e-01 -3.05723637e-01 1.85182974e-01
1.12947516e-01 -2.53715456e-01 -3.93709004e-01 -1.07537818e+00
2.39391066e-02 4.91256863e-02 -3.83575231e-01 5.13711929e-01
7.70073891e-01 9.90654051e-01 -8.51047456e-01 -3.64722222e-01
4.27445471e-01 6.13290548e-01 -6.58866644e-01 2.43313789e-01
-1.34536237e-01 4.11064953e-01 -1.36416763e-01 4.67396915e-01
8.15796375e-01 -2.39514951e-02 -3.51061523e-01 -1.95793897e-01
-4.45531756e-01 4.27387863e-01 -1.06278467e+00 1.72781670e+00
-3.87306571e-01 9.31316137e-01 5.49014732e-02 -5.59398651e-01
5.84528446e-01 1.76399067e-01 2.07769588e-01 -9.43838239e-01
1.81740060e-01 1.73829019e-01 1.26628757e-01 -6.25892758e-01
7.92398572e-01 -1.42381698e-01 1.72653362e-01 2.51784697e-02
2.64169067e-01 3.57156783e-01 3.42878819e-01 1.23446167e-01
1.17080033e+00 6.93055570e-01 -2.19997287e-01 -8.59058872e-02
6.04245782e-01 5.95914535e-02 4.79038984e-01 4.84193236e-01
-4.70577627e-01 8.51398587e-01 3.41850072e-01 -4.52045053e-01
-1.02162087e+00 -8.08638990e-01 5.22314347e-02 1.11767995e+00
5.15103340e-01 -2.23624215e-01 -3.59357804e-01 -5.26111722e-01
-2.73882542e-02 6.41458631e-01 -6.94869161e-01 -2.55260229e-01
-9.78715837e-01 -6.11699641e-01 3.88051391e-01 9.91621554e-01
6.73095584e-01 -9.42719281e-01 -1.27688050e+00 4.15091485e-01
-1.97701812e-01 -1.48677158e+00 -5.36591828e-01 2.44603157e-01
-5.89101672e-01 -9.81585383e-01 -6.04447901e-01 -4.48539495e-01
1.21534951e-01 4.64414537e-01 7.77585506e-01 -1.65095687e-01
-1.70449615e-01 2.93789227e-02 -5.14473736e-01 -2.42195651e-01
-1.06281780e-01 7.50737125e-03 -1.53644279e-01 4.45587426e-01
2.67617911e-01 -4.21630293e-01 -1.00190508e+00 3.97693485e-01
-8.17620635e-01 4.18801218e-01 7.31947362e-01 5.92253864e-01
3.50212932e-01 -4.97071981e-01 1.37605876e-01 -3.66043866e-01
-3.55929166e-01 -5.38384020e-01 -6.15860820e-01 -2.15471864e-01
-4.06234413e-01 2.61440158e-01 4.38843369e-01 -1.94166079e-01
-1.02812529e+00 3.55791807e-01 -3.54687780e-01 -8.65566015e-01
-4.17629518e-02 -3.11823059e-02 -1.65132388e-01 6.51442409e-02
3.47570717e-01 3.24842185e-01 -1.60108536e-01 -4.14448410e-01
4.80779588e-01 4.97315019e-01 6.88527584e-01 -9.51007605e-02
6.54651523e-01 8.52613628e-01 7.99937397e-02 -5.84739566e-01
-5.21342576e-01 -8.08501780e-01 -8.59231412e-01 -3.01326692e-01
1.29279470e+00 -1.07224941e+00 -4.96422797e-01 6.75379217e-01
-1.44604409e+00 -5.64111650e-01 -1.04841933e-01 6.12489164e-01
-6.42458260e-01 5.84532097e-02 -6.16118550e-01 -8.76304269e-01
-2.22213104e-01 -1.50727272e+00 1.48914778e+00 6.29518256e-02
1.58340678e-01 -5.95973969e-01 -1.08228132e-01 3.38010669e-01
5.79105198e-01 2.99156785e-01 4.96260643e-01 -3.71003658e-01
-1.11091459e+00 -2.53320411e-02 -5.01168311e-01 5.68056107e-02
-3.97104830e-01 -1.28162969e-02 -1.41569293e+00 -2.61359751e-01
-2.08547220e-01 9.01913643e-03 1.55488217e+00 3.76417518e-01
7.44658411e-01 8.89836438e-03 -4.24888045e-01 8.54770005e-01
1.44591784e+00 2.95432746e-01 7.92820334e-01 5.61978400e-01
1.05914605e+00 6.61042094e-01 7.40364313e-01 3.77734810e-01
7.60674834e-01 8.80823255e-01 8.01048160e-01 1.21625751e-01
-4.65769589e-01 -1.76012412e-01 7.68358290e-01 2.93946326e-01
-5.32528087e-02 -4.70165402e-01 -9.62722301e-01 7.92419255e-01
-2.05563259e+00 -1.12308419e+00 -4.66120273e-01 2.02682090e+00
3.45687240e-01 4.00959700e-01 2.01741308e-01 2.96154022e-01
5.19547880e-01 2.85595238e-01 -5.57758689e-01 -2.84971595e-01
-1.15285605e-01 -5.18205352e-02 7.03395188e-01 5.93276918e-01
-1.30223382e+00 9.29292202e-01 4.92643976e+00 6.55577779e-01
-1.37910438e+00 3.04656655e-01 4.44066733e-01 -3.63002717e-01
4.90929969e-02 -2.77918461e-03 -1.22752953e+00 5.69123089e-01
1.16988063e+00 2.68981218e-01 2.42279638e-02 6.55230582e-01
4.32369053e-01 -2.02182308e-01 -1.28877747e+00 7.64409304e-01
2.47311126e-02 -1.27563047e+00 -2.17767745e-01 -1.00194924e-02
3.90367478e-01 4.65063930e-01 2.92218365e-02 4.48251337e-01
-1.27159253e-01 -8.09639215e-01 1.56948507e+00 5.21141052e-01
6.71474040e-01 -5.16961217e-01 6.76048994e-01 4.29404676e-01
-1.66649127e+00 -3.27022851e-01 -8.83775502e-02 -9.87767056e-02
5.66681743e-01 2.67948240e-01 -5.46288729e-01 5.50465822e-01
7.79156804e-01 1.05491650e+00 -6.46369576e-01 1.25616109e+00
-8.03959221e-02 4.35926944e-01 -2.57068098e-01 2.82992929e-01
6.15243495e-01 1.83240980e-01 7.35142350e-01 1.35256147e+00
3.86807442e-01 -1.21923193e-01 -1.03022039e-01 8.90556753e-01
2.19772726e-01 -4.34329599e-01 -3.64312202e-01 2.80570358e-01
2.63888031e-01 1.25315273e+00 -5.54348886e-01 -3.95805418e-01
-4.52709168e-01 1.06121099e+00 2.29027539e-01 2.60053933e-01
-1.10731959e+00 -3.34276497e-01 8.56467605e-01 3.86189133e-01
8.47709179e-01 -3.33641142e-01 -1.65097788e-01 -8.43236446e-01
1.89690083e-01 -1.76504537e-01 2.62185991e-01 -8.86034250e-01
-7.60696828e-01 7.46081948e-01 -2.94883866e-02 -1.50816762e+00
-2.78893113e-01 -8.69187653e-01 -6.81165218e-01 8.20472896e-01
-2.22203493e+00 -1.24925268e+00 -4.48856711e-01 3.54307771e-01
9.25699830e-01 2.99143314e-01 2.39731893e-01 6.02285802e-01
-6.63724363e-01 5.34106016e-01 -2.32361943e-01 -8.47143978e-02
7.11786985e-01 -1.17167091e+00 6.84229076e-01 1.17235720e+00
-1.93175614e-01 7.31099620e-02 5.34614444e-01 -2.49533355e-01
-1.48945403e+00 -1.66598654e+00 9.22243536e-01 -6.39250278e-01
4.77080405e-01 -3.98768276e-01 -8.69630814e-01 7.00498581e-01
2.40352333e-01 3.64354283e-01 1.82226837e-01 -5.23120821e-01
-3.22303712e-01 -5.31551652e-02 -8.69897485e-01 3.11713368e-01
1.10766602e+00 -1.85396954e-01 -2.42765829e-01 -1.63510233e-01
9.50038493e-01 -5.07576585e-01 -3.49444360e-01 4.85929608e-01
4.67944324e-01 -1.22795630e+00 8.12091947e-01 -2.63685256e-01
8.19691181e-01 -7.54271030e-01 -2.22761184e-01 -9.56509650e-01
-3.69256049e-01 -4.90035415e-01 -3.05300474e-01 1.00690496e+00
3.74213606e-01 -3.04571837e-01 6.53301775e-01 3.95811826e-01
-5.54926991e-01 -8.67726982e-01 -1.13932264e+00 -7.45814085e-01
-5.63848838e-02 -8.06537330e-01 3.62362176e-01 4.61581767e-01
-1.39581740e-01 5.28367519e-01 -4.49322104e-01 2.83443302e-01
4.48039025e-01 -7.62395710e-02 7.32261240e-01 -7.71101654e-01
-3.86268824e-01 -6.71505451e-01 -6.58955216e-01 -1.42195177e+00
-9.16276947e-02 -7.22359478e-01 4.43332791e-01 -1.49656940e+00
-6.99470192e-02 -2.20258653e-01 -2.64963269e-01 3.39289874e-01
-2.92313784e-01 3.07822764e-01 4.87874091e-01 1.89479381e-01
-8.06521654e-01 8.66124153e-01 1.08241749e+00 -8.08153898e-02
-7.41662458e-02 -2.78806657e-01 -1.18711434e-01 5.15491962e-01
4.70610917e-01 -3.47248882e-01 -2.82325894e-01 -8.51282716e-01
-8.88307989e-02 1.40431628e-01 7.74469852e-01 -1.39524364e+00
5.94196916e-01 6.12142943e-02 3.64120185e-01 -1.01254570e+00
6.70915186e-01 -6.29785657e-01 -2.21356615e-01 4.70643133e-01
-1.90161079e-01 1.91495523e-01 5.00829041e-01 7.08351076e-01
-2.11049363e-01 1.55867442e-01 9.71521020e-01 2.15320885e-01
-1.18779325e+00 4.57378894e-01 -3.24628681e-01 -1.44661456e-01
1.06727719e+00 -3.07176501e-01 -4.44217831e-01 6.83325231e-02
-6.01353467e-01 5.85719645e-01 1.86800703e-01 6.92934096e-01
6.29284441e-01 -1.28486776e+00 -7.60191262e-01 2.76938766e-01
2.04112604e-01 2.41223887e-01 3.75332534e-01 1.33968067e+00
-5.23946166e-01 4.69208211e-01 -2.12473288e-01 -9.53231215e-01
-9.20024276e-01 5.90447009e-01 5.05758762e-01 -1.21434867e-01
-6.83767378e-01 1.15918839e+00 4.30470198e-01 -2.82958057e-02
1.51868626e-01 -5.25566518e-01 -2.87793756e-01 8.09350759e-02
8.43759418e-01 5.45272768e-01 1.66598007e-01 -9.52651143e-01
-5.00211835e-01 7.43502557e-01 2.77572591e-02 -3.77293706e-01
1.07576108e+00 -2.71141291e-01 2.74826497e-01 5.51147878e-01
1.28177667e+00 -4.91927832e-01 -1.87562406e+00 -2.03831226e-01
-5.10910228e-02 -5.29818535e-01 1.36415452e-01 -6.02806926e-01
-1.02856791e+00 1.38690424e+00 8.16608369e-01 -1.26167879e-01
1.10812271e+00 -2.23214060e-01 9.49991226e-01 -1.87480360e-01
1.45346731e-01 -6.26063585e-01 -1.80992205e-02 6.56056106e-01
7.46295452e-01 -1.41442227e+00 -4.29677963e-01 -3.34793389e-01
-6.44825876e-01 1.02194297e+00 8.27175081e-01 -1.71093568e-01
4.65104550e-01 2.00146183e-01 -3.89712788e-02 -7.14078024e-02
-1.13975847e+00 -6.21246397e-01 4.07297015e-01 2.95618057e-01
2.59590834e-01 -2.82005399e-01 -6.10014424e-02 5.04919648e-01
1.37992231e-02 3.52022909e-02 3.13460171e-01 1.05055499e+00
-4.72395033e-01 -7.92692006e-01 -1.72619700e-01 7.79178143e-02
-2.64765292e-01 -1.64623007e-01 5.72554581e-02 6.55720770e-01
6.02932692e-01 8.88914227e-01 3.79142374e-01 -5.77507079e-01
5.20745873e-01 2.10444294e-02 3.28963339e-01 -3.23555410e-01
-4.77865100e-01 1.00847855e-02 2.88849045e-03 -9.28583324e-01
-4.19166595e-01 -7.55553484e-01 -1.34008193e+00 -1.42098576e-01
-1.63149968e-01 -4.11610067e-01 7.10996926e-01 1.08668041e+00
5.07509232e-01 8.72849584e-01 4.66871679e-01 -1.49125981e+00
-2.12482110e-01 -9.61320817e-01 -1.73970163e-01 2.09464341e-01
8.75049710e-01 -7.03238785e-01 -3.46750915e-01 1.52640581e-01] | [8.234055519104004, -1.2922035455703735] |
0506c779-1fe8-4efd-bd15-c10298636744 | semi-automatic-sign-language-corpora | null | null | https://aclanthology.org/L12-1433 | https://aclanthology.org/L12-1433.pdf | Semi-Automatic Sign Language Corpora Annotation using Lexical Representations of Signs | Nowadays many researches focus on the automatic recognition of sign language. High recognition rates are achieved using lot of training data. This data is, generally, collected by manual annotating SL video corpus. However this is time consuming and the results depend on the annotators knowledge. In this work we intend to assist the annotation in terms of glosses which consist on writing down the sign meaning sign for sign thanks to automatic video processing techniques. In this case using learning data is not suitable since at the first step it will be needed to manually annotate the corpus. Also the context dependency of signs and the co-articulation effect in continuous SL make the collection of learning data very difficult. Here we present a novel approach which uses lexical representations of sign to overcome these problems and image processing techniques to match sign performances to sign representations. Signs are described using Zeebede (ZBD) which is a descriptor of signs that considers the high variability of signs. A ZBD database is used to stock signs and can be queried using several characteristics. From a video corpus sequence features are extracted using a robust body part tracking approach and a semi-automatic sign segmentation algorithm. Evaluation has shown the performances and limitation of the proposed approach. | ['Michael Filhol', 'Christophe Collet', 'Matilde Gonzalez'] | 2012-05-01 | null | null | null | lrec-2012-5 | ['hand-segmentation'] | ['computer-vision'] | [ 2.82720506e-01 -2.33696029e-01 -1.27737358e-01 -5.52719831e-01
-3.94699544e-01 -5.25185764e-01 6.35566950e-01 -3.35986130e-02
-9.37262118e-01 7.35552907e-01 2.09744707e-01 4.19386327e-02
-4.60173339e-01 -3.51429969e-01 -1.99943736e-01 -6.86533332e-01
1.23834744e-01 7.16134489e-01 5.96129298e-01 -2.88451046e-01
4.50607181e-01 6.04587495e-01 -2.09979486e+00 3.18165839e-01
7.01922715e-01 5.47567606e-01 4.16189820e-01 8.17006409e-01
-4.59045857e-01 5.75619161e-01 -7.02290535e-01 1.19389713e-01
3.24809551e-01 -8.17219555e-01 -5.93586564e-01 2.79256761e-01
4.76354063e-01 -2.03562260e-01 4.05971408e-01 9.93386745e-01
5.59687257e-01 7.57657588e-02 1.05696750e+00 -1.06955731e+00
2.49912336e-01 4.28927958e-01 -2.53979675e-02 1.26662493e-01
4.53490943e-01 2.95848716e-02 6.84948683e-01 -5.46695948e-01
1.17823148e+00 8.99683774e-01 4.82914180e-01 5.62929273e-01
-7.13205218e-01 -3.87685895e-01 -2.03622833e-01 6.34189129e-01
-1.11323428e+00 -2.00089261e-01 6.97953641e-01 -7.83423960e-01
6.83942676e-01 2.26609781e-01 8.92742276e-01 6.33514047e-01
-3.36093217e-01 7.49636889e-01 1.50921583e+00 -9.29657161e-01
7.67715797e-02 1.64560854e-01 5.32759011e-01 5.65063715e-01
2.86591321e-01 -1.96926951e-01 -2.79129267e-01 4.29802716e-01
5.12155890e-01 -4.36398804e-01 -2.94839770e-01 -5.17540038e-01
-9.01584983e-01 4.20835555e-01 -4.16249037e-02 1.03271317e+00
-5.56588650e-01 -6.71225339e-02 6.92552745e-01 3.29559386e-01
-3.25920910e-01 1.22205772e-01 -3.32663745e-01 -5.61293662e-01
-1.19993246e+00 1.16336837e-01 7.94268191e-01 5.97943068e-01
1.24201961e-01 -5.53902872e-02 1.56651184e-01 6.62092268e-01
4.50845897e-01 5.63323021e-01 7.41185427e-01 -2.98842043e-01
1.73919514e-01 7.60864556e-01 -4.46941927e-02 -7.73745775e-01
-4.98123407e-01 1.86811566e-01 -3.03766489e-01 8.36760759e-01
1.12281859e+00 1.06543273e-01 -1.43828726e+00 1.33928478e+00
3.50645214e-01 -3.99846405e-01 1.20324649e-01 1.06931484e+00
8.40819240e-01 3.52210909e-01 4.42986548e-01 -3.00643057e-01
1.37037563e+00 -4.43053186e-01 -1.14655042e+00 5.96797824e-01
7.60047197e-01 -1.13694036e+00 1.04072833e+00 9.04508591e-01
-7.95609355e-01 -4.91583198e-01 -1.09090722e+00 1.11225747e-01
-6.66341543e-01 7.21552908e-01 2.00356603e-01 8.38872194e-01
-7.09545434e-01 3.98970932e-01 -8.02384198e-01 -9.01332796e-01
1.98289694e-04 6.07273221e-01 -5.24303079e-01 1.07978553e-01
-7.14359462e-01 1.24693823e+00 8.19927633e-01 4.21586215e-01
-1.24926232e-01 1.13941822e-02 -6.56624615e-01 -4.14110214e-01
1.94077700e-01 -2.34015077e-01 1.00562620e+00 -1.43185842e+00
-1.64116979e+00 1.16196620e+00 4.79122531e-03 -2.84442931e-01
9.56351340e-01 -1.06521755e-01 -5.43925822e-01 4.67079431e-01
-2.79346704e-01 2.40497336e-01 8.53737533e-01 -1.13864827e+00
-7.85542667e-01 -2.51985371e-01 -3.30239922e-01 1.53896526e-01
-2.37206779e-02 3.71304661e-01 -4.01188880e-01 -4.89291936e-01
2.77982742e-01 -8.71428668e-01 4.08124983e-01 -1.40036225e-01
6.14788532e-02 -3.06870311e-01 1.12473702e+00 -1.12360549e+00
1.17750025e+00 -1.74301016e+00 1.95410043e-01 6.90501273e-01
-1.87143937e-01 8.99247885e-01 1.12196416e-01 3.08757097e-01
3.53712477e-02 -2.71839023e-01 -1.23800628e-01 1.74049258e-01
7.89516121e-02 5.80866694e-01 3.24808449e-01 3.91625017e-01
-1.20796546e-01 2.42016211e-01 -6.24676108e-01 -1.26632929e+00
7.09002614e-01 5.68315148e-01 -2.72203684e-01 7.90368542e-02
-1.51715100e-01 6.77295208e-01 -4.06867057e-01 6.27637923e-01
4.44796413e-01 4.43125039e-01 3.45446736e-01 -4.74009335e-01
-2.82138616e-01 -1.50674105e-01 -1.52946651e+00 1.42939746e+00
-4.08696353e-01 6.12816632e-01 -2.93057766e-02 -1.27338898e+00
1.01163411e+00 5.90067148e-01 5.47855318e-01 -4.54962969e-01
6.29440367e-01 8.80207658e-01 1.79669648e-01 -1.23032558e+00
1.72751710e-01 9.46459174e-02 4.02616113e-01 3.25265348e-01
2.27570862e-01 -1.63797587e-01 9.40141976e-01 -3.33999604e-01
5.88667333e-01 7.49783874e-01 5.56207895e-01 2.67110318e-02
1.13769078e+00 2.20194057e-01 1.96226299e-01 2.38518819e-01
-6.26299679e-02 1.90457225e-01 2.91303992e-01 -2.86577135e-01
-1.06301701e+00 -5.30573726e-01 -2.59842515e-01 5.87477684e-01
-3.85940433e-01 -2.11606771e-01 -7.16243923e-01 -7.00798631e-01
-3.32920223e-01 3.00284714e-01 -3.29998195e-01 4.21488494e-01
-8.99315536e-01 -4.36203510e-01 3.41404408e-01 1.78631514e-01
5.70540965e-01 -1.39307618e+00 -1.17086351e+00 1.23133086e-01
1.05039090e-01 -8.25519621e-01 1.50768742e-01 -1.80476964e-01
-1.00587845e+00 -1.31621563e+00 -1.10708535e+00 -8.94302487e-01
8.46537650e-01 -4.81100380e-01 5.89337826e-01 4.06610608e-01
-5.48316538e-01 4.88153070e-01 -9.47285354e-01 -4.77530390e-01
-8.34845006e-01 3.73851173e-02 -2.81898499e-01 -1.99960638e-02
5.84530115e-01 -4.86522526e-01 -2.27900118e-01 3.54742646e-01
-1.08164966e+00 -2.24947020e-01 9.04462755e-01 6.50261700e-01
4.07377362e-01 -3.58817071e-01 1.12117797e-01 -6.36810184e-01
4.70607936e-01 2.38486052e-01 -8.18807602e-01 4.73349661e-01
-4.20430839e-01 3.42086226e-01 1.87333316e-01 -4.16513115e-01
-9.03247714e-01 4.55140084e-01 -3.57899606e-01 3.72963585e-02
-6.54409409e-01 2.91389674e-01 -2.29669362e-01 -3.53155315e-01
4.94923741e-01 2.09153399e-01 2.42219701e-01 -7.24445403e-01
1.54579356e-01 9.10906136e-01 4.47262168e-01 -2.35814154e-01
6.11223161e-01 1.29092187e-01 3.48424017e-01 -1.17937934e+00
-1.41199619e-01 -7.37424850e-01 -1.22352612e+00 -8.86876047e-01
1.02161014e+00 -1.13745118e-02 -5.59684396e-01 5.82612574e-01
-1.08471298e+00 -8.95063803e-02 -2.37809852e-01 9.92224395e-01
-6.85190976e-01 6.52255654e-01 -4.61801887e-02 -1.10874319e+00
-1.78986847e-01 -9.96510684e-01 6.88697696e-01 2.01016814e-01
-2.62920797e-01 -6.77138984e-01 1.63331836e-01 3.40879530e-01
1.62046775e-01 3.82508069e-01 6.98483169e-01 -6.70991123e-01
-4.24315959e-01 -5.65442383e-01 -7.92849585e-02 5.73997080e-01
8.65369961e-02 1.60209298e-01 -5.05700529e-01 3.08675736e-01
-4.02265459e-01 -3.12063992e-01 6.75549388e-01 4.80880111e-01
3.75861138e-01 6.06403425e-02 -6.69285655e-02 2.59207815e-01
1.68462026e+00 5.41503131e-01 6.47569716e-01 4.55605835e-01
4.43537086e-01 8.21715176e-01 9.81960475e-01 2.79619694e-01
-1.55589446e-01 1.02306449e+00 -6.22804184e-03 2.03610018e-01
-4.57007647e-01 1.49813676e-02 3.10411751e-01 8.61963332e-01
-8.92684042e-01 -4.51929905e-02 -9.93221939e-01 4.20229495e-01
-1.70671690e+00 -1.21827006e+00 -6.49968207e-01 2.20707321e+00
6.29764080e-01 1.66333377e-01 4.76425409e-01 8.75525534e-01
5.03421485e-01 -2.65590787e-01 1.87723726e-01 -5.13696373e-01
-1.15642279e-01 3.73750359e-01 5.19646764e-01 6.70544386e-01
-1.03183222e+00 8.69653106e-01 4.92882395e+00 6.58770442e-01
-1.37298155e+00 1.83973759e-01 -4.96907860e-01 2.00793132e-01
3.23612511e-01 -1.70063317e-01 -8.63944471e-01 4.41007525e-01
4.80367333e-01 1.84062824e-01 -1.07557476e-01 6.39221132e-01
4.66632545e-01 -6.09564245e-01 -7.39680350e-01 1.14171100e+00
6.09761655e-01 -6.85737669e-01 8.57869834e-02 -3.93396728e-02
4.66488063e-01 -3.53861421e-01 -4.87064749e-01 -7.18414411e-02
-3.38948548e-01 -6.64962828e-01 6.97231650e-01 9.50518072e-01
6.30619228e-01 -3.85426253e-01 9.93406475e-01 7.19161183e-02
-1.23142564e+00 1.98517889e-02 -6.62089065e-02 2.02694952e-01
6.44554734e-01 -8.72343704e-02 -1.03813541e+00 3.52198184e-01
2.43348405e-01 4.01234627e-01 -5.26053667e-01 1.65997052e+00
-3.82480204e-01 4.68026012e-01 -7.26816475e-01 -4.70587641e-01
2.07197711e-01 -3.58584613e-01 7.66793370e-01 1.44125342e+00
4.83280450e-01 7.35584600e-03 -8.71153846e-02 4.36193436e-01
6.82967067e-01 9.33316767e-01 -6.80676103e-01 -1.06174029e-01
-2.59442031e-01 9.47836518e-01 -1.17245996e+00 -4.86904979e-01
-1.06564321e-01 1.01802433e+00 -5.05860329e-01 -1.53560415e-02
-3.39694053e-01 -4.84726936e-01 -6.41334057e-02 2.48236850e-01
2.83905268e-01 -4.73191410e-01 1.78231820e-01 -9.07191396e-01
8.21748748e-02 -7.76660800e-01 5.06425023e-01 -6.42487466e-01
-5.86046338e-01 5.78663051e-01 3.48554373e-01 -1.55720842e+00
-6.77392125e-01 -1.01028490e+00 -2.37465248e-01 5.51735878e-01
-1.19255328e+00 -1.35564148e+00 -4.75730747e-01 6.36670768e-01
6.85069621e-01 -1.74287707e-01 7.16821551e-01 5.69439173e-01
-2.03468665e-01 1.77774966e-01 -6.80800378e-02 2.80027777e-01
7.53596783e-01 -1.24178600e+00 -7.28072226e-01 8.98812652e-01
3.98892522e-01 1.27883658e-01 9.03161824e-01 -7.48871863e-01
-6.61109507e-01 -1.29108215e-02 1.36671114e+00 -1.93980858e-01
3.39287460e-01 2.10955918e-01 -5.78640401e-01 3.07031631e-01
3.90291698e-02 -2.16958404e-01 4.65822011e-01 -3.24534059e-01
6.41677007e-02 -1.50896385e-02 -9.62688565e-01 2.84366727e-01
8.74384165e-01 -1.14150040e-01 -1.12498748e+00 2.33442172e-01
-6.02356076e-01 -2.03757629e-01 -6.38400972e-01 2.17250302e-01
9.99181747e-01 -9.07296240e-01 5.76681256e-01 -6.37864232e-01
-6.25270009e-02 -5.13742030e-01 2.08098412e-01 -7.70257473e-01
6.53500259e-01 -2.69121081e-01 4.14748728e-01 1.16300189e+00
2.80182064e-01 -2.04182252e-01 8.05067718e-01 4.81517404e-01
1.19537897e-01 -3.74248065e-02 -9.14107442e-01 -8.60974967e-01
-6.71229243e-01 -3.43304873e-01 3.52279283e-02 7.25904047e-01
1.74988180e-01 -1.14299774e-01 -2.96315491e-01 -2.17250556e-01
6.11323118e-01 -5.92009462e-02 8.80315185e-01 -1.54496479e+00
-1.59149751e-01 -7.13106811e-01 -1.10083508e+00 -4.04534787e-01
-5.59355728e-02 -7.06717134e-01 -3.78713645e-02 -1.59423721e+00
-4.63725984e-01 -2.99833834e-01 -7.40681961e-02 2.35499457e-01
4.05036420e-01 3.20302129e-01 5.03790796e-01 1.66131154e-01
-3.81143689e-01 -7.40144849e-02 1.17337179e+00 2.06382245e-01
-4.05291140e-01 1.45335913e-01 6.60235703e-01 9.08150136e-01
8.14914048e-01 -2.49247327e-01 -2.97214568e-01 -6.22222992e-03
1.03529409e-01 -2.67366529e-01 2.56359428e-01 -1.28524268e+00
1.94506168e-01 1.72618568e-01 9.69949290e-02 -8.21618855e-01
2.50092477e-01 -1.26388836e+00 1.67702317e-01 8.31125259e-01
-1.25918731e-01 1.85749587e-02 -1.89051643e-01 1.33895606e-01
-4.16604131e-01 -8.91187727e-01 7.84490585e-01 -2.58530676e-01
-1.28004134e+00 -2.66270459e-01 -6.38007164e-01 -2.16928020e-01
1.07357252e+00 -8.96296501e-01 4.08579260e-01 -4.44071859e-01
-1.15040648e+00 -1.32066101e-01 1.92487329e-01 2.89130807e-02
3.41537058e-01 -1.10163939e+00 -4.58830655e-01 1.79312766e-01
2.24707365e-01 -4.29948837e-01 3.12456965e-01 1.22483015e+00
-1.27033567e+00 4.05741751e-01 -1.00413597e+00 -3.70500982e-01
-2.17947292e+00 2.31896952e-01 2.55769283e-01 -2.35447288e-01
-6.96385443e-01 3.46878380e-01 -8.31797659e-01 1.89841419e-01
4.03824806e-01 -7.43304849e-01 -1.08963215e+00 4.55369115e-01
2.89069593e-01 3.21659595e-01 -1.23754114e-01 -1.14455795e+00
-1.08233213e-01 1.34225392e+00 4.30972487e-01 -5.13079762e-01
1.11652577e+00 9.30963829e-02 6.63045719e-02 4.83243585e-01
8.85375321e-01 1.71988636e-01 -6.36390567e-01 1.06628038e-01
4.80231076e-01 -4.60887194e-01 -7.22828805e-02 -1.01358569e+00
-7.28167117e-01 8.23724270e-01 9.43154514e-01 -1.22015677e-01
1.17890060e+00 -2.71094114e-01 4.50006813e-01 4.70725775e-01
2.80491680e-01 -1.70885825e+00 -4.61031765e-01 1.37972102e-01
1.00720859e+00 -1.14458323e+00 1.12629877e-02 -4.15777743e-01
-6.26786590e-01 1.63437068e+00 1.65414691e-01 -1.11029461e-01
8.32282662e-01 1.17852576e-01 4.28279817e-01 -1.79735690e-01
1.72285885e-01 -9.54443336e-01 4.72545743e-01 7.07895100e-01
5.29583514e-01 -1.09442517e-01 -1.81278503e+00 3.21041465e-01
-1.09312885e-01 6.87747896e-01 3.83850515e-01 1.13957870e+00
-4.48631883e-01 -1.77960706e+00 -7.10494459e-01 2.56584942e-01
-4.45256919e-01 3.71992826e-01 -4.70001847e-01 1.42616177e+00
6.73737943e-01 6.32118404e-01 -3.54304373e-01 -6.49068654e-02
5.98371089e-01 6.05232298e-01 9.41285670e-01 -2.00940490e-01
-5.48507452e-01 1.22358769e-01 4.02800083e-01 -2.92274386e-01
-1.13855207e+00 -9.45722401e-01 -1.24636650e+00 3.64473253e-01
-2.86944985e-01 1.37884289e-01 1.34449458e+00 1.13148177e+00
-5.03005147e-01 2.01446459e-01 -1.98901623e-01 -8.53778601e-01
-4.31084633e-01 -1.08245146e+00 -6.23145163e-01 9.07742500e-01
-1.16631836e-02 -9.49314713e-01 -1.11350559e-01 5.62856019e-01] | [9.103132247924805, -6.345076084136963] |
94a54897-a035-4410-858b-914684476cd9 | deep-learning-based-instance-segmentation-in | 1806.11137 | null | http://arxiv.org/abs/1806.11137v1 | http://arxiv.org/pdf/1806.11137v1.pdf | Deep Learning Based Instance Segmentation in 3D Biomedical Images Using Weak Annotation | Instance segmentation in 3D images is a fundamental task in biomedical image
analysis. While deep learning models often work well for 2D instance
segmentation, 3D instance segmentation still faces critical challenges, such as
insufficient training data due to various annotation difficulties in 3D
biomedical images. Common 3D annotation methods (e.g., full voxel annotation)
incur high workloads and costs for labeling enough instances for training deep
learning 3D instance segmentation models. In this paper, we propose a new weak
annotation approach for training a fast deep learning 3D instance segmentation
model without using full voxel mask annotation. Our approach needs only 3D
bounding boxes for all instances and full voxel annotation for a small fraction
of the instances, and uses a novel two-stage 3D instance segmentation model
utilizing these two kinds of annotation, respectively. We evaluate our approach
on several biomedical image datasets, and the experimental results show that
(1) with full annotated boxes and a small amount of masks, our approach can
achieve similar performance as the best known methods using full annotation,
and (2) with similar annotation time, our approach outperforms the best known
methods that use full annotation. | ['Si-Yuan Zhang', 'Zhuo Zhao', 'Danny Z. Chen', 'Lin Yang', 'Ian H. Guldner', 'Hao Zheng'] | 2018-06-28 | null | null | null | null | ['3d-instance-segmentation-1'] | ['computer-vision'] | [ 2.15594411e-01 5.99726737e-01 -3.12423587e-01 -5.63251555e-01
-9.71989095e-01 -3.40187162e-01 2.77374871e-02 3.74837577e-01
-6.44267142e-01 5.84822834e-01 -4.83023494e-01 -4.01084810e-01
2.72825837e-01 -5.43281734e-01 -8.72102797e-01 -5.54781973e-01
1.14202768e-01 8.92392278e-01 7.94613898e-01 2.93366879e-01
5.78788742e-02 5.23258209e-01 -1.01113737e+00 3.37096974e-02
7.89018571e-01 1.17351973e+00 3.01882774e-01 4.07486439e-01
-7.12523460e-01 3.14342260e-01 -7.69124568e-01 -2.89667696e-02
5.00532269e-01 -2.84310728e-01 -1.11169076e+00 5.22907734e-01
3.36292744e-01 -5.59545577e-01 6.78009689e-02 1.05873799e+00
6.71886623e-01 -1.54637158e-01 6.72151387e-01 -8.54373991e-01
-6.67728558e-02 3.50825518e-01 -9.66933668e-01 1.40867963e-01
-1.55916587e-01 -5.38325347e-02 4.86001670e-01 -6.34277761e-01
4.64463562e-01 9.29113150e-01 8.48485529e-01 7.75616050e-01
-1.14272070e+00 -5.63748837e-01 2.25607648e-01 -2.61372745e-01
-1.49139833e+00 -9.87699851e-02 7.00323045e-01 -5.79277575e-01
8.73433769e-01 9.91715565e-02 7.40950227e-01 5.11398375e-01
5.73613383e-02 9.54249084e-01 1.17026389e+00 -2.23148435e-01
3.22627157e-01 -9.26270708e-02 5.43132603e-01 7.64844656e-01
2.73022026e-01 -4.93486047e-01 2.79635340e-01 -1.21284552e-01
1.13095975e+00 1.78773239e-01 -1.18821360e-01 -2.23241523e-01
-1.20542109e+00 5.33578455e-01 2.89327979e-01 2.63810784e-01
-1.33843914e-01 1.65839389e-01 6.93544745e-01 -2.42118478e-01
8.28344941e-01 1.34984851e-01 -6.99022412e-01 8.85021165e-02
-1.10369921e+00 2.52475720e-02 5.54346025e-01 1.20411325e+00
8.81385803e-01 -3.35418522e-01 -3.26271027e-01 9.98215973e-01
-5.21664554e-03 1.85661122e-01 5.31047702e-01 -6.61366224e-01
2.42147580e-01 9.07501459e-01 3.98179442e-02 -5.49649000e-01
-8.39455724e-01 -4.00135845e-01 -9.63361263e-01 -4.05042432e-02
6.69498324e-01 -3.25128883e-02 -1.32501864e+00 1.34605145e+00
6.78700209e-01 4.13863331e-01 -2.91826069e-01 9.07502413e-01
1.12853277e+00 3.81480902e-01 -4.44091819e-02 -3.05506468e-01
1.47207439e+00 -9.27682161e-01 -7.36145496e-01 -1.36666313e-01
1.14486909e+00 -4.80486095e-01 1.11831677e+00 2.73184955e-01
-1.15951025e+00 -5.54323554e-01 -8.85352373e-01 -3.33043724e-01
-2.75908232e-01 2.08853766e-01 6.52447402e-01 8.27585042e-01
-8.82894754e-01 3.56250197e-01 -1.04463816e+00 -1.10733755e-01
1.00334513e+00 7.33292162e-01 -3.88969302e-01 -1.34221524e-01
-8.16457093e-01 5.77453196e-01 5.75779319e-01 2.45143160e-01
-7.89785087e-01 -7.26893365e-01 -8.28414679e-01 -1.72739714e-01
5.90741992e-01 -3.80906820e-01 1.30543864e+00 -5.69883406e-01
-1.20682538e+00 1.33862257e+00 -1.74910769e-01 -2.81880766e-01
6.69729114e-01 -1.97721019e-01 3.77549201e-01 2.41811663e-01
1.09944358e-01 8.26743722e-01 4.70650971e-01 -1.19704986e+00
-4.82544690e-01 -7.38124073e-01 2.30235785e-01 6.99646845e-02
-2.72717178e-01 -3.65189940e-01 -9.20548022e-01 -3.91313106e-01
4.88999933e-01 -7.13269711e-01 -7.18929529e-01 1.55867189e-01
-6.16463244e-01 -4.64658827e-01 7.99896479e-01 -5.16334534e-01
9.40883815e-01 -1.98639131e+00 -1.74143493e-01 1.53729051e-01
5.05762935e-01 3.05350453e-01 1.71391606e-01 -3.49657625e-01
1.38151661e-01 4.26856101e-01 -5.54217517e-01 -4.26471502e-01
-6.86905906e-02 4.54740554e-01 1.97044075e-01 4.64541137e-01
1.75348416e-01 7.70940959e-01 -8.05130184e-01 -9.70860243e-01
6.11309260e-02 3.36613595e-01 -6.79568350e-01 2.24801451e-01
-3.04377317e-01 6.41749918e-01 -6.55058026e-01 6.31633401e-01
8.37257385e-01 -5.38230002e-01 2.08866820e-01 -3.12907517e-01
1.31637812e-01 4.03838158e-02 -1.04289567e+00 1.94452775e+00
-2.76109636e-01 1.75410300e-01 5.65410256e-02 -1.50115323e+00
9.76919889e-01 4.34074283e-01 7.08412826e-01 -5.11699021e-01
3.08703035e-01 6.14659250e-01 -2.11420715e-01 -6.39411688e-01
-6.56249598e-02 -1.39131442e-01 -1.37886375e-01 4.47116047e-01
1.21674379e-02 -3.80501807e-01 1.26159191e-01 -2.34754477e-02
1.04086375e+00 2.29565307e-01 3.61864716e-01 -2.86200374e-01
5.36099255e-01 8.37712735e-02 8.88304591e-01 6.03197515e-01
-4.50987220e-01 6.47915781e-01 8.96057904e-01 -7.67487109e-01
-1.10557139e+00 -4.21270549e-01 -3.65974456e-01 4.74659353e-01
4.55291808e-01 -1.92048296e-01 -1.17696381e+00 -1.12456739e+00
-2.81272978e-01 7.04190955e-02 -6.09805703e-01 2.25587815e-01
-7.57630646e-01 -1.09212542e+00 5.67298472e-01 6.86034083e-01
6.89141273e-01 -6.37559593e-01 -6.50127292e-01 1.80180863e-01
7.32103959e-02 -1.37205935e+00 -4.29161072e-01 2.28725865e-01
-1.38906109e+00 -1.18498075e+00 -1.12610590e+00 -1.02668226e+00
1.15069258e+00 1.51514351e-01 8.80942643e-01 5.95328093e-01
-5.72886527e-01 1.52918519e-02 -1.83039889e-01 -4.61098313e-01
-1.43244296e-01 2.17244774e-01 -2.25410730e-01 -4.57542360e-01
3.87611300e-01 -2.09146634e-01 -7.58173227e-01 4.84018236e-01
-9.40387845e-01 1.88847601e-01 7.36772835e-01 1.04271376e+00
1.28327000e+00 6.45834133e-02 6.13936067e-01 -1.43266296e+00
8.76229778e-02 -1.14623390e-01 -6.42270863e-01 8.71653296e-03
-3.55142951e-01 -2.70207465e-01 3.71265978e-01 -4.83012766e-01
-7.40099728e-01 3.34662586e-01 -5.53798497e-01 -4.73148823e-01
-4.59708393e-01 1.76035181e-01 -2.81241804e-01 -9.18217674e-02
4.84837711e-01 5.92764542e-02 1.25751510e-01 -7.99850702e-01
3.47116147e-03 6.83437169e-01 1.43053457e-01 -5.47719777e-01
3.58103842e-01 6.30550623e-01 6.10732958e-02 -6.24653101e-01
-1.10205984e+00 -5.15372396e-01 -1.15628302e+00 -6.28909376e-03
1.11693561e+00 -5.98658979e-01 -5.49702466e-01 4.60410297e-01
-1.18597984e+00 -7.01531470e-01 -2.58307487e-01 4.85981733e-01
-5.14354706e-01 5.75223505e-01 -6.95588470e-01 -5.16494691e-01
-4.58393365e-01 -1.66448092e+00 1.55090916e+00 2.42987997e-03
-2.78820004e-02 -9.45309937e-01 -2.86687613e-01 4.32198405e-01
-5.75309619e-02 3.72404724e-01 1.15872037e+00 -1.01890302e+00
-5.41292906e-01 -4.47863460e-01 -4.22455966e-01 2.57041901e-01
1.24514721e-01 -5.43862700e-01 -8.57388258e-01 -7.64757022e-02
-2.22714171e-02 -3.11543107e-01 6.58614874e-01 6.57977045e-01
1.91246891e+00 2.27940716e-02 -7.38794088e-01 5.65130889e-01
1.13001287e+00 2.78140962e-01 4.35763299e-01 -8.85394122e-03
8.56629908e-01 5.58205843e-01 6.06171548e-01 2.13515803e-01
1.82176918e-01 5.91976643e-01 4.47551936e-01 -6.66191816e-01
3.69985646e-04 2.36725628e-01 -3.41042638e-01 8.01281154e-01
4.37539853e-02 4.15261947e-02 -1.03690541e+00 5.40861070e-01
-1.70384300e+00 -3.17421347e-01 -2.51771986e-01 2.14063573e+00
1.04465055e+00 4.17569935e-01 1.29412010e-01 3.61546338e-01
7.30488837e-01 -1.09793529e-01 -7.57218719e-01 -1.85225643e-02
3.44637990e-01 4.13575411e-01 4.55772370e-01 3.20939124e-01
-1.25995815e+00 8.62910211e-01 6.32133198e+00 1.04944324e+00
-8.00688803e-01 3.85655999e-01 1.09232116e+00 -1.22191027e-01
6.08678125e-02 -1.92848712e-01 -9.56219256e-01 3.44265908e-01
2.93812454e-01 4.50289875e-01 -3.53542000e-01 9.08434093e-01
2.37219289e-01 -1.29667342e-01 -1.17545629e+00 1.03042579e+00
-4.16966863e-02 -1.33959877e+00 -4.03413996e-02 1.32693708e-01
5.62749147e-01 -3.43036294e-01 -4.69704598e-01 2.76109397e-01
-2.48667881e-01 -9.92039084e-01 4.18610811e-01 5.84269986e-02
9.82594669e-01 -6.42777562e-01 1.07009649e+00 5.77847660e-01
-9.19397771e-01 3.60673100e-01 -5.87374508e-01 2.61581182e-01
1.63211763e-01 9.96512830e-01 -8.01129699e-01 3.38201702e-01
6.35001361e-01 4.51356560e-01 -2.29398668e-01 1.19586957e+00
-5.53581081e-02 6.03480577e-01 -4.07958388e-01 -8.88941716e-03
3.59837443e-01 -7.43257701e-02 2.21098319e-01 1.12055779e+00
1.59227073e-01 4.51301843e-01 7.03961968e-01 7.28798330e-01
-2.93563724e-01 2.90794551e-01 -2.40002304e-01 1.96064353e-01
1.91653132e-01 1.19937575e+00 -1.36234319e+00 -5.65621495e-01
-4.05075461e-01 9.65496838e-01 2.16230750e-01 1.14025682e-01
-8.93578112e-01 -4.00169104e-01 2.67116845e-01 1.94154188e-01
8.45421255e-02 -2.02817649e-01 -4.35096204e-01 -7.32055783e-01
-3.53877321e-02 -3.09282035e-01 3.55977654e-01 -4.45597947e-01
-1.09084952e+00 4.79254156e-01 1.08877001e-02 -9.94548440e-01
2.91236579e-01 -8.91589940e-01 -3.05019736e-01 7.29045272e-01
-1.55214167e+00 -1.08570755e+00 -4.83205974e-01 3.66660565e-01
7.03802168e-01 2.81085849e-01 6.52164042e-01 5.59751511e-01
-7.20232248e-01 5.52967429e-01 -3.18854839e-01 4.06571001e-01
5.11277258e-01 -1.30699801e+00 2.45654121e-01 2.35238999e-01
-1.11599132e-01 3.98569971e-01 1.00857854e-01 -7.13871479e-01
-1.20367908e+00 -1.06908262e+00 5.74820161e-01 -2.21708670e-01
6.51066750e-02 -4.65297371e-01 -9.84328508e-01 6.25314951e-01
-2.56057382e-01 4.89729494e-01 9.47966576e-01 -5.24041019e-02
8.56887698e-02 7.54689723e-02 -1.37482762e+00 3.62604439e-01
1.20399940e+00 -2.28573587e-02 -3.99628043e-01 8.71873677e-01
1.01038325e+00 -1.01024985e+00 -1.05028462e+00 5.73008597e-01
3.36266041e-01 -6.48343563e-01 8.39494467e-01 -5.27648926e-01
1.31823495e-01 -3.64240736e-01 1.27373695e-01 -6.92956507e-01
2.99796816e-02 -3.70292544e-01 3.46818380e-02 7.61140645e-01
4.78539348e-01 -4.82091874e-01 1.01972044e+00 5.88589728e-01
-6.16012812e-01 -1.35771334e+00 -1.00999248e+00 -7.29588449e-01
1.77730784e-01 -5.41863739e-01 6.95984542e-01 9.20728862e-01
-3.63431126e-01 -1.19164931e-02 5.31144207e-03 7.79504925e-02
5.49809754e-01 1.50403246e-01 7.85782397e-01 -1.29006934e+00
-6.76534027e-02 -3.11135262e-01 -4.74296182e-01 -1.41337502e+00
1.29376337e-01 -9.57621276e-01 4.71373089e-02 -1.76296163e+00
4.11701947e-01 -7.69203305e-01 -1.62635028e-01 7.71188557e-01
-2.48085663e-01 6.50058568e-01 -3.75501394e-01 2.99115241e-01
-6.02737427e-01 1.47704512e-01 1.57724500e+00 -2.31918901e-01
-2.10794449e-01 1.09878138e-01 -3.51864785e-01 9.30379927e-01
6.69342101e-01 -4.09574479e-01 -4.36133415e-01 -5.80706298e-01
-6.35728389e-02 5.21372072e-03 1.94495320e-01 -8.04912329e-01
1.92336831e-02 1.02792203e-01 6.19095504e-01 -1.00051522e+00
2.56324053e-01 -8.30348849e-01 -2.99560666e-01 4.78788048e-01
-1.77316561e-01 -5.27520835e-01 3.71312737e-01 4.68977004e-01
-9.86363739e-03 -4.67421442e-01 8.66964042e-01 -4.83303279e-01
-5.42113483e-01 7.72403955e-01 -1.40385807e-01 9.47661549e-02
1.17952192e+00 -6.17900908e-01 1.57477357e-03 3.12841713e-01
-1.09444416e+00 3.73624831e-01 3.95513564e-01 2.13619508e-02
4.73515838e-01 -1.18645573e+00 -3.74527305e-01 7.54211023e-02
-4.23185937e-02 9.39526677e-01 3.91751707e-01 1.07655370e+00
-7.17446864e-01 3.92318666e-01 1.37820348e-01 -1.16104317e+00
-1.15299642e+00 4.69844073e-01 4.74440604e-01 -3.49151492e-01
-8.93259406e-01 9.31978285e-01 6.99157476e-01 -6.72360718e-01
2.43113756e-01 -5.83866119e-01 -1.07841760e-01 -1.01050280e-01
4.24785703e-01 7.71508738e-02 2.17707306e-01 -3.24800938e-01
-3.88744056e-01 7.23298073e-01 -2.70187885e-01 3.13133478e-01
1.16237092e+00 1.71624586e-01 -8.78134593e-02 4.09962952e-01
1.20415783e+00 -4.24398243e-01 -1.23447025e+00 -1.69465557e-01
5.59746325e-02 -3.88633132e-01 1.20468371e-01 -4.12373990e-01
-1.29469872e+00 1.27805638e+00 6.27036095e-01 1.34566754e-01
1.08515000e+00 1.47228479e-01 1.14269698e+00 2.29190975e-01
4.51559067e-01 -1.04105330e+00 6.10874966e-02 2.59320378e-01
3.24058294e-01 -1.21543646e+00 1.27552688e-01 -7.92264462e-01
-2.28621602e-01 9.00615275e-01 8.32622707e-01 2.38725487e-02
6.94138587e-01 3.79678637e-01 -1.67511739e-02 -3.55371773e-01
-2.25659281e-01 -3.77917551e-02 1.84936225e-01 3.92428517e-01
6.19793952e-01 -1.37692645e-01 -5.52705824e-01 8.19034219e-01
3.93624067e-01 -5.00606038e-02 2.08991230e-01 9.56770360e-01
-5.19546032e-01 -1.08756149e+00 -2.97390282e-01 7.98520923e-01
-8.05848777e-01 1.30772293e-01 -1.86547518e-01 9.86052513e-01
2.98236609e-01 3.20028096e-01 3.01311195e-01 2.08901111e-02
3.63597602e-01 1.00420177e-01 5.55880368e-01 -9.30846751e-01
-3.71483058e-01 5.44424295e-01 -2.71040529e-01 -6.00296140e-01
-5.16343653e-01 -3.37034523e-01 -1.85793340e+00 2.57455379e-01
-5.81282914e-01 6.58440143e-02 6.28707707e-01 1.12077212e+00
2.05787599e-01 5.39580047e-01 3.10907274e-01 -9.58970964e-01
-1.89097673e-01 -9.22736526e-01 -6.82345986e-01 2.69934565e-01
6.63001090e-02 -7.97743440e-01 -5.77830337e-03 2.71227330e-01] | [14.668722152709961, -2.2716152667999268] |
10cac1a4-4d13-403f-bc57-ff5e1b00d044 | pseudo-convolutional-policy-gradient-for | 2003.03983 | null | https://arxiv.org/abs/2003.03983v1 | https://arxiv.org/pdf/2003.03983v1.pdf | Pseudo-Convolutional Policy Gradient for Sequence-to-Sequence Lip-Reading | Lip-reading aims to infer the speech content from the lip movement sequence and can be seen as a typical sequence-to-sequence (seq2seq) problem which translates the input image sequence of lip movements to the text sequence of the speech content. However, the traditional learning process of seq2seq models always suffers from two problems: the exposure bias resulted from the strategy of "teacher-forcing", and the inconsistency between the discriminative optimization target (usually the cross-entropy loss) and the final evaluation metric (usually the character/word error rate). In this paper, we propose a novel pseudo-convolutional policy gradient (PCPG) based method to address these two problems. On the one hand, we introduce the evaluation metric (refers to the character error rate in this paper) as a form of reward to optimize the model together with the original discriminative target. On the other hand, inspired by the local perception property of convolutional operation, we perform a pseudo-convolutional operation on the reward and loss dimension, so as to take more context around each time step into account to generate a robust reward and loss for the whole optimization. Finally, we perform a thorough comparison and evaluation on both the word-level and sentence-level benchmarks. The results show a significant improvement over other related methods, and report either a new state-of-the-art performance or a competitive accuracy on all these challenging benchmarks, which clearly proves the advantages of our approach. | ['Mingshuang Luo', 'Xilin Chen', 'Shuang Yang', 'Shiguang Shan'] | 2020-03-09 | null | null | null | null | ['lipreading'] | ['computer-vision'] | [ 3.86200756e-01 -1.53789014e-01 -2.70137459e-01 -3.09165597e-01
-1.23591673e+00 -1.35075301e-01 5.34180641e-01 -1.78496063e-01
-6.91640675e-01 7.29695201e-01 5.00964522e-01 -1.88481390e-01
2.11750537e-01 -2.18383566e-01 -8.59328568e-01 -1.00875890e+00
2.00202838e-01 -2.36805640e-02 1.15460210e-01 9.04466435e-02
4.63606596e-01 1.11326851e-01 -1.56286502e+00 1.70676470e-01
1.00442040e+00 1.21032405e+00 6.13210440e-01 5.29389262e-01
-3.27522367e-01 6.93421543e-01 -5.19911945e-01 -4.18362707e-01
8.11502114e-02 -7.34335721e-01 -6.28982902e-01 -8.02151114e-02
2.77741909e-01 -2.02235729e-01 -1.29278451e-01 1.43891001e+00
9.88073051e-01 2.43646100e-01 5.13012767e-01 -1.03043962e+00
-4.51471955e-01 5.18264949e-01 -6.81060195e-01 -9.44098756e-02
2.78815925e-01 6.50336802e-01 1.10062349e+00 -9.22171235e-01
4.61826533e-01 1.37294710e+00 3.36487830e-01 7.31519341e-01
-8.81043315e-01 -5.96313596e-01 2.62121171e-01 5.51429033e-01
-1.19128835e+00 -8.21550071e-01 7.91991651e-01 -3.05686444e-01
6.23636425e-01 2.43467942e-01 3.57003182e-01 1.18045926e+00
1.83489099e-01 1.16732335e+00 1.11423886e+00 -4.16271836e-01
2.37305775e-01 1.99922562e-01 -4.78175670e-01 5.03334045e-01
-5.16685843e-01 1.42297059e-01 -5.53825736e-01 3.56159210e-01
2.59299070e-01 -3.96640927e-01 -5.39959252e-01 -1.25615433e-01
-1.06825507e+00 6.08988523e-01 2.82872796e-01 5.89991510e-02
-2.61874974e-01 2.21537828e-01 6.65354550e-01 -1.49668343e-02
3.01070809e-01 1.01914726e-01 -4.22183841e-01 -5.62541783e-01
-9.04685974e-01 1.23393610e-01 5.77226758e-01 7.05054820e-01
5.10816753e-01 4.17991206e-02 -7.26647615e-01 1.06562293e+00
4.34027195e-01 4.80055988e-01 7.89009094e-01 -6.15731955e-01
8.72334361e-01 8.67620390e-03 -1.44146189e-01 -7.38742888e-01
-3.57484445e-02 -2.42034808e-01 -7.70426154e-01 3.21412683e-01
4.18876261e-01 -1.91078976e-01 -6.58686757e-01 2.15907574e+00
2.46377990e-01 2.48192281e-01 1.56610250e-01 1.04694748e+00
6.58486903e-01 7.47757971e-01 2.33281419e-01 -5.21601498e-01
1.23797178e+00 -1.41995037e+00 -9.38920677e-01 -2.46139213e-01
3.93946767e-01 -7.43773162e-01 1.47135997e+00 1.57874584e-01
-1.17015088e+00 -5.74468017e-01 -8.17259371e-01 1.23346141e-02
-1.34232536e-01 3.02808851e-01 -7.03915162e-03 2.21026093e-01
-9.21710670e-01 5.28195024e-01 -3.82917374e-01 1.83883385e-05
3.17501754e-01 1.02662876e-01 -8.95480812e-02 1.54399008e-01
-1.21422434e+00 7.73262024e-01 3.26629937e-01 1.65345430e-01
-7.93636084e-01 -7.18528092e-01 -7.19579816e-01 2.08622754e-01
5.81081033e-01 -2.88106620e-01 1.16978145e+00 -1.17361403e+00
-2.18583632e+00 6.18260026e-01 -2.75190026e-01 -4.01809186e-01
9.90988672e-01 -2.71298051e-01 -1.28665045e-01 -2.38096453e-02
-1.93645328e-01 9.24472630e-01 1.04771221e+00 -9.33589816e-01
-6.46813571e-01 -7.70585760e-02 -2.00593784e-01 4.33714598e-01
-2.22794041e-01 7.15931365e-03 -6.26443505e-01 -8.32864940e-01
-4.35426444e-01 -7.15352535e-01 1.20032728e-01 3.37593853e-02
-4.88641322e-01 -5.80998123e-01 4.67961997e-01 -8.79950523e-01
1.38956845e+00 -2.44912076e+00 2.65594542e-01 -4.33231890e-01
-1.95265874e-01 4.39872146e-01 -3.77664626e-01 2.74363101e-01
-1.63241699e-01 2.15541780e-01 -4.30786520e-01 -7.48474956e-01
1.01216666e-01 -1.59002721e-01 -2.91020453e-01 3.77844274e-01
4.06705320e-01 9.51256335e-01 -1.03867245e+00 -7.76060224e-01
8.14382657e-02 4.59920019e-01 -5.43664515e-01 4.58759248e-01
-4.50650722e-01 5.37573397e-01 -2.34745532e-01 2.23558336e-01
8.01304936e-01 2.16228679e-01 -6.72545955e-02 -1.20902732e-01
-2.43185103e-01 6.03841484e-01 -9.43607271e-01 1.91069782e+00
-6.68619871e-01 6.44999623e-01 6.95633981e-03 -9.98823464e-01
7.87830114e-01 3.34280550e-01 3.58394295e-01 -8.34526479e-01
-4.82086837e-03 1.51817560e-01 -7.12545663e-02 -9.14335608e-01
3.04884046e-01 -4.66085561e-02 2.55227506e-01 2.08855242e-01
-9.16821137e-02 -1.80841479e-02 -8.98661390e-02 -3.06378812e-01
6.25714719e-01 5.08951366e-01 8.98302346e-02 -1.13362364e-01
9.64376092e-01 -5.51446617e-01 6.05760276e-01 3.76600802e-01
-3.88015687e-01 6.36141002e-01 8.19010735e-01 9.68138278e-02
-1.01422155e+00 -8.49753976e-01 2.79461243e-03 9.96083736e-01
4.45330031e-02 -2.30510876e-01 -1.05058920e+00 -9.75350320e-01
-1.92788869e-01 7.63045669e-01 -5.27375221e-01 -2.71715820e-01
-5.62272668e-01 -4.86991018e-01 5.23620725e-01 3.93014580e-01
6.64829493e-01 -1.52529407e+00 -3.70305747e-01 1.74053863e-01
-3.33312273e-01 -1.22091627e+00 -1.06092870e+00 1.33820381e-02
-3.87746543e-01 -5.75137377e-01 -9.73705947e-01 -8.50370049e-01
3.02567691e-01 -2.49448135e-01 6.96729243e-01 -6.00986108e-02
-1.40305366e-02 -1.91633746e-01 -3.05528045e-01 -2.96118975e-01
-5.13479710e-01 2.58076712e-02 -5.65840490e-02 4.34983432e-01
5.70098944e-02 -3.42873067e-01 -6.77467108e-01 6.70561567e-02
-7.23815680e-01 1.11630656e-01 8.46207678e-01 8.97552609e-01
6.56346679e-01 -2.67818719e-01 8.12336862e-01 -1.48996532e-01
7.71531582e-01 -2.69385368e-01 -6.48005784e-01 3.47080320e-01
-5.97676158e-01 3.19812715e-01 8.04078519e-01 -7.29680300e-01
-8.79700840e-01 7.67137334e-02 -6.15461528e-01 -6.18996561e-01
-9.40608159e-02 2.20980182e-01 -5.25761187e-01 2.02650040e-01
2.45708376e-02 8.01188290e-01 3.55868608e-01 -6.53208673e-01
3.66735309e-01 9.68138099e-01 5.11674404e-01 -4.63749588e-01
4.68378693e-01 1.10364661e-01 -1.38282165e-01 -7.29285240e-01
-6.58189714e-01 -2.21199527e-01 -2.56829739e-01 -2.28126600e-01
1.05683053e+00 -5.91625094e-01 -1.04964447e+00 6.86128139e-01
-1.28709447e+00 -5.24427533e-01 -3.68426442e-01 5.44576406e-01
-9.10200715e-01 5.94448805e-01 -2.88071573e-01 -1.01393867e+00
-4.51294512e-01 -1.58583403e+00 9.54695225e-01 2.99142629e-01
2.26390168e-01 -8.10869157e-01 2.45990045e-03 1.38933584e-01
3.00327182e-01 -1.66539978e-02 9.21880007e-01 -5.42770982e-01
-4.74751145e-01 1.63362935e-01 -3.59548330e-01 8.81862819e-01
-8.97581223e-03 -1.84851795e-01 -1.05932891e+00 -1.56808332e-01
9.85607132e-02 -4.23278481e-01 9.28710103e-01 3.30918044e-01
1.49325943e+00 -6.13158822e-01 4.84436601e-02 5.71475208e-01
1.46826339e+00 5.77183999e-02 8.25980842e-01 1.27873853e-01
5.52316248e-01 6.43735349e-01 6.48464918e-01 3.36409569e-01
2.88618237e-01 1.07646334e+00 3.12386721e-01 1.23812117e-01
-5.61904728e-01 -6.65542781e-01 7.49952614e-01 8.76717925e-01
2.68768221e-01 -3.08203787e-01 -5.05843103e-01 4.43640441e-01
-1.86447906e+00 -9.14189577e-01 2.72334456e-01 2.41140246e+00
1.11053407e+00 1.70536965e-01 2.04443008e-01 3.32970731e-02
7.97824562e-01 4.15864289e-01 -7.20942795e-01 -5.72404385e-01
-2.14676932e-02 -2.41307486e-02 2.83563286e-01 7.61483073e-01
-8.39881897e-01 1.11515808e+00 5.44129801e+00 1.45211315e+00
-1.51712334e+00 8.05580020e-02 6.40372813e-01 -2.44834032e-02
-1.52300775e-01 -3.31110001e-01 -9.05403733e-01 9.26063418e-01
7.51143098e-01 -3.95750813e-02 7.54989266e-01 5.48906207e-01
4.80307937e-01 -1.49830114e-02 -1.01623058e+00 1.07258129e+00
8.60128701e-02 -9.82637048e-01 3.83663140e-02 6.94728866e-02
5.10794759e-01 8.53072293e-03 2.86219746e-01 3.14813465e-01
-3.10609192e-01 -1.03829980e+00 1.12966871e+00 6.27457976e-01
1.21787798e+00 -6.92259312e-01 5.90312600e-01 5.54499626e-01
-1.20284474e+00 -1.21019095e-01 -1.71579927e-01 2.69578665e-01
2.11228937e-01 2.98670799e-01 -7.56944180e-01 4.17658627e-01
4.93471622e-01 6.12652600e-01 -1.34041294e-01 1.08062398e+00
-4.07816380e-01 5.89865506e-01 -9.57700238e-02 -5.09330094e-01
5.07902026e-01 -2.10685223e-01 7.75755882e-01 1.42478919e+00
1.40236691e-01 -3.99232894e-01 -1.22550473e-01 1.00532377e+00
-2.73913503e-01 3.65586042e-01 -2.65021980e-01 2.20425390e-02
2.25252584e-01 1.06108260e+00 -5.18260561e-02 -1.64211497e-01
-2.98017085e-01 9.48281348e-01 4.87526447e-01 4.17917103e-01
-9.21309114e-01 -5.74254870e-01 8.69136572e-01 -1.41632438e-01
6.21516645e-01 1.43122971e-01 -2.21988410e-01 -9.78258789e-01
3.35085243e-01 -1.04163933e+00 -1.95773691e-01 -6.51136875e-01
-1.01443183e+00 5.56439281e-01 -2.10321039e-01 -1.28845918e+00
-3.68700892e-01 -3.80296916e-01 -6.03338122e-01 1.06823587e+00
-2.02179527e+00 -6.66637540e-01 2.77754385e-02 4.00666624e-01
9.66187119e-01 -3.99031118e-02 5.26452363e-01 2.59896994e-01
-6.85060084e-01 1.12355328e+00 2.05787808e-01 1.31893307e-01
7.03315914e-01 -1.05445039e+00 2.87490964e-01 6.41633630e-01
-1.56464547e-01 2.21622679e-02 5.25822878e-01 -2.83106208e-01
-1.15208161e+00 -8.57407510e-01 1.03342164e+00 -7.70457536e-02
6.05332553e-01 -3.72140139e-01 -8.85630131e-01 6.74998015e-02
2.85969585e-01 -1.14089847e-01 1.79961845e-01 -4.58232373e-01
-1.25311494e-01 -3.77157003e-01 -1.00596619e+00 8.29121470e-01
1.07033587e+00 -5.22312820e-01 -4.31723118e-01 1.09827429e-01
9.31309640e-01 -3.26695830e-01 -5.04412115e-01 5.49656034e-01
6.46153808e-01 -8.63784730e-01 7.16739058e-01 -5.89218497e-01
7.13962793e-01 -1.64577737e-01 -1.94949672e-01 -1.26818073e+00
1.98789090e-02 -8.59430492e-01 -1.36910204e-03 1.52531481e+00
4.63495135e-01 -5.34398735e-01 4.45038348e-01 -3.02074421e-02
-1.75113276e-01 -1.48290396e+00 -1.19457757e+00 -8.75482321e-01
2.67833054e-01 -4.02440906e-01 6.40927076e-01 3.76065969e-01
-2.32769419e-02 3.02389830e-01 -4.56988752e-01 -1.29532114e-01
4.34890032e-01 -1.05068222e-01 5.66347122e-01 -5.86363375e-01
-3.12244445e-01 -8.93263280e-01 -2.21763656e-01 -1.69238222e+00
5.27615488e-01 -7.52530754e-01 6.83566272e-01 -1.14758420e+00
1.35997131e-01 -3.78901660e-01 -3.73595774e-01 2.01056004e-01
-5.53638697e-01 -2.08164990e-01 5.27263939e-01 7.76854008e-02
-4.68644381e-01 9.06743050e-01 1.57964289e+00 -1.83397710e-01
-1.98704764e-01 1.44161746e-01 -5.13226867e-01 5.12810290e-01
7.15471447e-01 -3.57570976e-01 -2.70487100e-01 -4.91458505e-01
-3.63374390e-02 1.36815920e-01 2.75974512e-01 -8.01165998e-01
2.10542455e-01 -2.15977252e-01 -1.99034885e-01 -4.22949076e-01
4.17492747e-01 -5.36752105e-01 -6.44709229e-01 5.17655909e-01
-7.42456496e-01 -2.13677004e-01 9.28013548e-02 5.93557835e-01
-2.32593268e-01 -3.83797377e-01 1.08811152e+00 1.24351166e-01
-5.47511995e-01 3.51411581e-01 1.06278807e-01 4.14470434e-01
8.72220635e-01 7.34032094e-02 -6.16719648e-02 -4.43019241e-01
-3.36846769e-01 2.65500486e-01 2.29793563e-01 5.77121854e-01
5.88922620e-01 -1.36497295e+00 -9.58619297e-01 3.12302351e-01
6.26320578e-03 -1.55038133e-01 1.18716285e-01 1.10686398e+00
-1.44705996e-01 3.26213807e-01 1.35282623e-02 -6.10289097e-01
-1.05129397e+00 6.71496630e-01 4.79016870e-01 -3.78443420e-01
-6.21055126e-01 9.02152121e-01 2.66583800e-01 -1.95735708e-01
8.53319049e-01 -2.88349569e-01 -1.94730118e-01 8.51337891e-03
5.52206218e-01 1.61357626e-01 -1.34657174e-01 -5.18960476e-01
-3.72744143e-01 7.12984681e-01 1.98402271e-01 -3.21799099e-01
1.06873178e+00 -2.34295949e-01 1.92467660e-01 4.60481614e-01
1.57058454e+00 1.49695545e-01 -1.70475030e+00 -1.40547976e-01
-1.66987553e-01 -2.93964326e-01 -1.38499752e-01 -8.73912394e-01
-9.92993891e-01 1.40734184e+00 6.63289666e-01 2.11919844e-02
1.10299313e+00 -4.62913364e-02 9.67180490e-01 -9.24602300e-02
-1.80494562e-01 -1.23789585e+00 9.16893035e-02 4.61661428e-01
1.18538642e+00 -1.15877926e+00 -4.80007976e-01 -1.29714206e-01
-8.93227339e-01 9.92778778e-01 4.85684603e-01 1.51463866e-01
4.49128121e-01 1.54842630e-01 1.16202489e-01 6.18033886e-01
-7.68431365e-01 -3.31275702e-01 2.78889179e-01 4.93316025e-01
2.75353462e-01 1.79044586e-02 -6.07237339e-01 4.04451311e-01
-1.57159507e-01 1.19311020e-01 -1.26324683e-01 2.54784167e-01
-3.57871532e-01 -1.06091595e+00 -8.28720331e-02 2.83112582e-02
-4.87669021e-01 -3.39117825e-01 -7.75694996e-02 3.41073304e-01
1.23416394e-01 7.73161292e-01 3.52319069e-02 -4.61059868e-01
2.41028145e-01 2.65848905e-01 2.25403696e-01 -2.83476353e-01
-3.34897667e-01 1.46745905e-01 -1.38778389e-01 -5.93022108e-01
-9.33284983e-02 -6.37639761e-01 -1.18845451e+00 -8.30806419e-02
-3.24167818e-01 3.68141048e-02 9.93390501e-01 1.04591739e+00
2.13505775e-01 5.51564217e-01 1.01231480e+00 -7.65941978e-01
-1.16849995e+00 -1.23598647e+00 -2.55910367e-01 5.19291818e-01
6.23593509e-01 -4.05454874e-01 -5.25814593e-01 -1.32096142e-01] | [14.334310531616211, 5.007132053375244] |
c1377006-8359-4106-af44-15a8ccf8e787 | dual-memory-units-with-uncertainty-regulation | 2302.05160 | null | https://arxiv.org/abs/2302.05160v1 | https://arxiv.org/pdf/2302.05160v1.pdf | Dual Memory Units with Uncertainty Regulation for Weakly Supervised Video Anomaly Detection | Learning discriminative features for effectively separating abnormal events from normality is crucial for weakly supervised video anomaly detection (WS-VAD) tasks. Existing approaches, both video and segment-level label oriented, mainly focus on extracting representations for anomaly data while neglecting the implication of normal data. We observe that such a scheme is sub-optimal, i.e., for better distinguishing anomaly one needs to understand what is a normal state, and may yield a higher false alarm rate. To address this issue, we propose an Uncertainty Regulated Dual Memory Units (UR-DMU) model to learn both the representations of normal data and discriminative features of abnormal data. To be specific, inspired by the traditional global and local structure on graph convolutional networks, we introduce a Global and Local Multi-Head Self Attention (GL-MHSA) module for the Transformer network to obtain more expressive embeddings for capturing associations in videos. Then, we use two memory banks, one additional abnormal memory for tackling hard samples, to store and separate abnormal and normal prototypes and maximize the margins between the two representations. Finally, we propose an uncertainty learning scheme to learn the normal data latent space, that is robust to noise from camera switching, object changing, scene transforming, etc. Extensive experiments on XD-Violence and UCF-Crime datasets demonstrate that our method outperforms the state-of-the-art methods by a sizable margin. | ['Wei Yang', 'Junqing Yu', 'Hang Zhou'] | 2023-02-10 | null | null | null | null | ['video-anomaly-detection'] | ['computer-vision'] | [ 6.45658225e-02 -5.81237264e-02 -2.86014408e-01 -4.24536943e-01
-4.48061764e-01 -1.81524470e-01 5.54956079e-01 2.41757810e-01
-2.05781981e-01 2.06370771e-01 2.67312467e-01 -2.35376433e-01
3.91732603e-02 -6.81207657e-01 -8.52579832e-01 -7.30958581e-01
-3.48651588e-01 1.98795706e-01 1.80579320e-01 2.01666772e-01
7.80636743e-02 4.27349478e-01 -1.52474546e+00 3.78301442e-01
7.97197044e-01 1.32600319e+00 -2.81287432e-01 4.76233214e-01
-1.15046531e-01 1.05140960e+00 -3.91035825e-01 -3.05795968e-01
1.52543887e-01 -4.43178028e-01 -6.62038147e-01 4.03158545e-01
4.22393858e-01 -7.38950670e-01 -8.90158772e-01 1.24497616e+00
2.86090553e-01 2.80350953e-01 9.14236486e-01 -1.56783330e+00
-9.33840036e-01 2.98626214e-01 -6.86054587e-01 7.88657844e-01
3.08933526e-01 3.03450674e-01 1.03645515e+00 -7.42905140e-01
2.54589796e-01 1.24261403e+00 3.34354877e-01 6.41181052e-01
-9.07636046e-01 -6.32695019e-01 6.91172123e-01 7.04783738e-01
-1.25782406e+00 -4.30701911e-01 8.76907647e-01 -6.14263594e-01
8.18875432e-01 1.29022747e-01 5.99587619e-01 1.74808443e+00
2.57377774e-01 1.04716539e+00 5.46198189e-01 3.56496386e-02
2.03151092e-01 -1.84940085e-01 2.85250574e-01 9.32419002e-01
3.53195041e-01 -1.87558278e-01 -3.69997203e-01 -2.50768751e-01
6.41496658e-01 4.92023498e-01 -5.41991889e-01 -5.37725031e-01
-8.83143008e-01 8.04244637e-01 8.30485523e-02 1.97857901e-01
-4.35180545e-01 8.12276229e-02 8.48032176e-01 4.79803056e-01
5.36372662e-01 -7.10384399e-02 -2.57268518e-01 -2.05148309e-01
-5.38179576e-01 2.16275989e-03 3.17231387e-01 7.57075131e-01
4.50375259e-01 2.87732989e-01 -5.00856102e-01 5.89623988e-01
4.84900057e-01 1.93249077e-01 8.33263159e-01 -4.75522846e-01
4.48274612e-01 7.18244553e-01 -2.95973688e-01 -1.28459084e+00
-2.03273654e-01 -2.21415669e-01 -1.04624295e+00 7.77912736e-02
1.41761705e-01 6.51955977e-02 -1.11891007e+00 1.72227561e+00
1.37168899e-01 1.00294864e+00 -1.69930011e-02 8.79700899e-01
6.31315291e-01 6.90480709e-01 1.29154801e-01 -3.14575642e-01
1.08254874e+00 -7.77042270e-01 -8.75613153e-01 -2.59040266e-01
8.77195299e-01 -1.52703628e-01 1.01403058e+00 3.42780858e-01
-7.86615789e-01 -4.62832570e-01 -1.01979518e+00 -7.20720878e-03
-3.85667622e-01 -9.62605849e-02 4.01947409e-01 3.55784506e-01
-6.35810077e-01 5.96718431e-01 -1.02065206e+00 -1.18345365e-01
6.73907280e-01 1.62159771e-01 -5.57708919e-01 -2.23920807e-01
-1.31182921e+00 5.37955046e-01 5.02532959e-01 2.43373826e-01
-1.16756320e+00 -3.78092259e-01 -1.30240703e+00 5.18689826e-02
5.35016775e-01 -3.38126123e-01 6.03182793e-01 -1.09664476e+00
-1.15814900e+00 8.24094117e-01 -7.00673088e-02 -5.39519370e-01
3.35673928e-01 -3.87216032e-01 -7.49054074e-01 1.57170802e-01
-8.73811394e-02 2.61433572e-01 1.34950137e+00 -8.56275439e-01
-3.82429630e-01 -5.93805850e-01 -7.71825062e-03 1.09010808e-01
-6.29586160e-01 -4.41947728e-02 -5.69409668e-01 -9.01706159e-01
7.48380050e-02 -6.25225425e-01 -8.37136582e-02 -7.71303009e-03
-4.41977113e-01 -4.43309963e-01 1.26415408e+00 -9.10965383e-01
1.54611540e+00 -2.50233650e+00 3.71941179e-01 3.35812658e-01
4.52769548e-01 3.84055734e-01 -1.90730974e-01 -2.68136561e-01
-4.36544389e-01 -1.17221307e-02 -1.94339126e-01 -3.53835702e-01
-2.50460841e-02 5.52576900e-01 -3.12537640e-01 7.62854397e-01
6.35272503e-01 7.25796402e-01 -8.95604551e-01 -3.64778250e-01
1.45324543e-01 3.06454480e-01 -6.56574726e-01 5.08912265e-01
4.91322540e-02 3.12500060e-01 -3.74747068e-01 9.10815656e-01
5.14354527e-01 -1.49477556e-01 -1.37183398e-01 -1.85905918e-01
3.96771550e-01 -3.48719209e-02 -1.27687848e+00 1.51029623e+00
-6.91964775e-02 4.21722025e-01 -1.21782325e-01 -1.46123958e+00
7.11904347e-01 3.43820363e-01 4.16198254e-01 -5.19386172e-01
2.95684040e-01 -3.45107391e-02 -5.69056906e-02 -7.91238248e-01
2.21996889e-01 2.26579294e-01 3.72645296e-02 1.69065937e-01
3.35914969e-01 7.58877277e-01 -6.25882298e-02 3.21352810e-01
1.27550685e+00 -8.91983584e-02 1.76449880e-01 -1.40493875e-02
7.44863272e-01 -7.26195872e-01 8.07052910e-01 4.55335408e-01
-6.04385853e-01 5.56418121e-01 1.07950377e+00 -4.66373146e-01
-7.18283653e-01 -1.12760735e+00 4.24471945e-02 8.71864259e-01
1.71876982e-01 -4.77560192e-01 -6.45698726e-01 -1.21395802e+00
-8.48296359e-02 6.98521197e-01 -7.61244178e-01 -8.15786481e-01
-4.74340528e-01 -6.57328188e-01 4.43938971e-01 7.72314489e-01
4.30683941e-01 -9.15877938e-01 -2.26908505e-01 -4.65691797e-02
-1.82525471e-01 -1.22582793e+00 -5.10723472e-01 1.80984028e-02
-6.49975896e-01 -1.12916231e+00 -3.30764323e-01 -4.13553923e-01
7.10567057e-01 -5.67330979e-02 7.84652770e-01 7.73595870e-02
-2.29196027e-01 5.83922684e-01 -5.09799361e-01 -5.19864894e-02
-2.09905326e-01 -3.21072787e-01 4.06150401e-01 6.07514858e-01
6.89468443e-01 -5.88321209e-01 -4.93717492e-01 6.65201321e-02
-1.17580736e+00 -3.45423132e-01 5.68226218e-01 8.67163241e-01
5.95573306e-01 3.71809714e-02 4.01243478e-01 -6.79976404e-01
5.05424678e-01 -1.04001784e+00 -2.47821033e-01 -1.16628688e-02
-5.08768737e-01 9.21796039e-02 7.52601206e-01 -5.25291920e-01
-7.62359917e-01 -1.46198824e-01 -2.20023215e-01 -1.23314273e+00
-4.34466749e-01 1.80106550e-01 -4.08106595e-01 2.62896866e-01
2.63950050e-01 3.35186869e-01 -3.52391005e-02 -2.66902238e-01
1.71651900e-01 6.33327425e-01 5.07749259e-01 -4.05014753e-01
6.73280001e-01 4.47984666e-01 -1.57205224e-01 -7.47759163e-01
-8.27760994e-01 -5.44287503e-01 -5.99017084e-01 -1.93371236e-01
1.16251385e+00 -8.43506157e-01 -2.45498762e-01 5.71351171e-01
-9.00371253e-01 -9.09672603e-02 -3.82904261e-01 5.16356885e-01
-5.34196496e-01 7.70417035e-01 -6.87098920e-01 -6.59512877e-01
3.01906001e-02 -1.32608390e+00 1.11197829e+00 6.03475841e-03
1.82470903e-02 -9.09149587e-01 6.11935407e-02 1.94166794e-01
5.87729663e-02 4.12039280e-01 1.07032204e+00 -9.95561898e-01
-4.32274908e-01 -3.42442513e-01 -3.18525732e-01 7.17283845e-01
3.82240340e-02 -3.89338471e-02 -9.02629018e-01 -3.85444611e-01
2.96097621e-02 -3.08873147e-01 1.11672986e+00 1.98192120e-01
1.82726097e+00 -4.67127681e-01 -2.33496889e-01 7.96869397e-01
1.00694621e+00 7.79858008e-02 8.19816649e-01 1.62733167e-01
1.08941960e+00 2.84196287e-01 4.74793404e-01 6.28588617e-01
3.58641535e-01 4.75230753e-01 7.83717394e-01 1.18616529e-01
1.83133528e-01 -1.26069233e-01 9.27003503e-01 6.98746681e-01
-2.06480566e-02 -3.71740162e-01 -6.96093619e-01 6.52863264e-01
-1.99965072e+00 -1.02654803e+00 1.02933630e-01 2.07969141e+00
3.36907148e-01 2.48643085e-01 1.25316173e-01 3.03650051e-01
7.48039246e-01 6.05045497e-01 -6.03448749e-01 -5.32615304e-01
1.36182662e-02 -1.18762158e-01 1.24416232e-01 1.48060530e-01
-1.48499095e+00 6.38741434e-01 5.12823296e+00 9.48625445e-01
-1.02585113e+00 1.59431323e-01 7.87683785e-01 -2.14785188e-01
-1.46963149e-01 -2.96378404e-01 -5.05533934e-01 7.60586441e-01
1.10468960e+00 1.92655280e-01 2.19798028e-01 9.05531764e-01
2.52348334e-02 3.99907082e-01 -1.29016232e+00 1.08746707e+00
3.64145666e-01 -9.74315941e-01 4.72416937e-01 1.12739634e-02
3.06816339e-01 -1.08359545e-01 3.30840908e-02 6.26834631e-01
-1.84417024e-01 -9.75396633e-01 4.55839753e-01 6.42133057e-01
6.25698686e-01 -8.07807684e-01 8.78110945e-01 2.30479226e-01
-1.09913266e+00 -3.27042669e-01 -3.31260741e-01 1.14140585e-01
3.54869217e-02 5.93013465e-01 -1.72587767e-01 4.95214999e-01
8.68030727e-01 1.04550517e+00 -7.69755781e-01 6.49240375e-01
-1.78374909e-02 6.57691538e-01 -4.23621014e-02 4.51272935e-01
4.40960974e-01 -2.33582363e-01 8.49273384e-01 1.20609868e+00
2.66782492e-01 5.03531620e-02 2.69414037e-01 6.65013492e-01
-5.54927066e-02 3.67454737e-02 -9.45499539e-01 -1.23861998e-01
-2.98375953e-02 1.09093869e+00 -3.45114172e-01 -3.11441600e-01
-7.70741642e-01 1.45638609e+00 5.28262496e-01 3.97702068e-01
-1.12837887e+00 -2.48463482e-01 8.53665471e-01 -2.66187005e-02
3.06890368e-01 -1.49642929e-01 2.86971956e-01 -1.62191200e+00
2.45853871e-01 -8.44927251e-01 8.70329142e-01 -3.01638395e-01
-1.44075501e+00 5.37127316e-01 -5.48972897e-02 -1.34020281e+00
-3.15954179e-01 -7.62218714e-01 -8.32531929e-01 3.34213138e-01
-1.50693977e+00 -1.01168311e+00 -2.94521034e-01 9.86371756e-01
5.76520264e-01 -2.60933638e-01 6.05885446e-01 5.91248095e-01
-1.18272865e+00 8.07752967e-01 -1.10461280e-01 5.67848265e-01
6.05418682e-01 -1.22073936e+00 1.32556990e-01 1.20273829e+00
1.12671211e-01 3.02891374e-01 3.73514473e-01 -6.40406072e-01
-1.26640022e+00 -1.51659846e+00 3.00928593e-01 -4.15497392e-01
8.35947156e-01 -3.50051105e-01 -1.35928750e+00 1.02503204e+00
-1.40560523e-01 5.18901944e-01 7.67577350e-01 -2.66208174e-03
-5.34434378e-01 9.18997526e-02 -1.05014741e+00 5.84519565e-01
1.19781709e+00 -5.92522264e-01 -6.45070255e-01 1.58326715e-01
7.02553272e-01 -3.20871085e-01 -6.84227705e-01 5.79742968e-01
1.55288681e-01 -1.04105639e+00 8.03041995e-01 -1.10536945e+00
4.11388308e-01 -2.57819504e-01 -3.02240223e-01 -1.25878108e+00
-5.38669884e-01 -2.77354717e-01 -1.06636810e+00 1.25314367e+00
2.33533117e-03 -5.95828354e-01 5.96848130e-01 4.44060445e-01
-3.86693835e-01 -1.09124410e+00 -1.09616554e+00 -7.80257583e-01
-3.13698173e-01 -7.04742730e-01 3.42655808e-01 1.19274545e+00
-7.98910633e-02 1.81428149e-01 -6.55175924e-01 5.80636978e-01
4.79120016e-01 -2.47305989e-01 3.71620417e-01 -1.03660297e+00
-2.13483155e-01 -4.53213900e-01 -1.13983035e+00 -9.57452178e-01
5.75099647e-01 -7.35036314e-01 -2.66934782e-01 -9.69950259e-01
1.37905329e-01 1.92713533e-02 -7.79180050e-01 3.64064664e-01
-3.84163588e-01 1.06272757e-01 -2.66903728e-01 -4.63982448e-02
-9.56418514e-01 8.29262912e-01 7.69510925e-01 -1.82245836e-01
2.57117786e-02 -8.89072046e-02 -4.86004174e-01 1.04627538e+00
5.76162815e-01 -2.16412306e-01 -3.87543976e-01 -4.03279573e-01
-6.96837157e-02 -2.64715821e-01 5.38017213e-01 -9.17838871e-01
1.07817603e-02 -5.90209216e-02 4.59350139e-01 -5.04556596e-01
2.33975887e-01 -9.00376737e-01 -4.97734100e-01 3.19222391e-01
-1.80690050e-01 1.10568181e-01 -1.81749631e-02 1.14298368e+00
-3.31311762e-01 -2.07346857e-01 6.52794659e-01 9.42575261e-02
-1.03603077e+00 9.31843400e-01 -4.98214722e-01 1.78754464e-01
1.29952753e+00 -1.40511617e-01 -1.30447298e-01 -3.24631602e-01
-7.53820598e-01 5.10074794e-01 1.83322221e-01 7.70291030e-01
8.86361003e-01 -1.67457092e+00 -5.55738986e-01 5.52226543e-01
3.82436395e-01 -2.50297189e-02 6.10938191e-01 9.47303653e-01
-1.58372194e-01 1.10558085e-01 -1.85570374e-01 -6.68748915e-01
-1.05905902e+00 7.69501925e-01 3.07511896e-01 -2.86100596e-01
-7.90868998e-01 6.83123887e-01 1.88509598e-01 -7.07838088e-02
4.85403687e-01 -3.71070743e-01 -4.53111738e-01 1.86755687e-01
6.93706810e-01 4.91536856e-01 -5.94975613e-02 -7.84410834e-01
-3.73640358e-01 2.69510061e-01 -3.06514829e-01 5.20896733e-01
1.05480433e+00 -1.03288390e-01 1.36993136e-02 5.44048607e-01
1.32940257e+00 -2.08900437e-01 -1.25059032e+00 -3.21807623e-01
-4.66390103e-02 -5.34133017e-01 1.91573620e-01 -2.97983922e-02
-1.27424121e+00 9.68823612e-01 8.52654517e-01 3.23463589e-01
1.27859926e+00 9.05539468e-02 8.89217377e-01 2.70357072e-01
-2.45111704e-01 -9.82130826e-01 3.47513288e-01 4.26436186e-01
7.05466807e-01 -1.47921324e+00 -3.30149710e-01 -9.12594423e-02
-7.48093247e-01 1.11077106e+00 1.10822082e+00 -2.48007506e-01
6.41640484e-01 -2.46867184e-02 -3.45354348e-01 -3.68140817e-01
-6.69327497e-01 -9.60492045e-02 5.18937707e-01 5.05973101e-01
9.79109555e-02 -5.53063564e-02 1.94551036e-01 8.20204616e-01
5.02173841e-01 -3.76416206e-01 4.32249188e-01 7.07415581e-01
-4.07133937e-01 -6.88630998e-01 -1.67803943e-01 9.57344472e-01
-6.52988672e-01 6.50575310e-02 -1.25069842e-01 5.52833438e-01
2.71294951e-01 4.94093388e-01 5.42266369e-01 -4.99926597e-01
2.88213372e-01 2.89900869e-01 2.53858179e-01 -5.52620769e-01
4.83040325e-02 -1.41528070e-01 -1.32177845e-01 -9.52954710e-01
-1.70598447e-01 -7.37387896e-01 -1.12255943e+00 -6.11034557e-02
-2.81377017e-01 -1.66100205e-03 -7.80386552e-02 1.08884442e+00
4.07893628e-01 7.20273018e-01 6.53004766e-01 -6.94351256e-01
-4.79649335e-01 -8.15033615e-01 -6.69417977e-01 7.04778075e-01
6.04070127e-01 -9.15778875e-01 -6.46299303e-01 -1.34701341e-01] | [7.843392372131348, 1.6237285137176514] |
1b122463-182a-44a5-81d3-2d20676657b8 | semantics-ontology-and-explanation | 2304.11124 | null | https://arxiv.org/abs/2304.11124v1 | https://arxiv.org/pdf/2304.11124v1.pdf | Semantics, Ontology and Explanation | The terms 'semantics' and 'ontology' are increasingly appearing together with 'explanation', not only in the scientific literature, but also in organizational communication. However, all of these terms are also being significantly overloaded. In this paper, we discuss their strong relation under particular interpretations. Specifically, we discuss a notion of explanation termed ontological unpacking, which aims at explaining symbolic domain descriptions (conceptual models, knowledge graphs, logical specifications) by revealing their ontological commitment in terms of their assumed truthmakers, i.e., the entities in one's ontology that make the propositions in those descriptions true. To illustrate this idea, we employ an ontological theory of relations to explain (by revealing the hidden semantics of) a very simple symbolic model encoded in the standard modeling language UML. We also discuss the essential role played by ontology-driven conceptual models (resulting from this form of explanation processes) in properly supporting semantic interoperability tasks. Finally, we discuss the relation between ontological unpacking and other forms of explanation in philosophy and science, as well as in the area of Artificial Intelligence. | ['Nicola Guarino', 'Giancarlo Guizzardi'] | 2023-04-21 | null | null | null | null | ['philosophy'] | ['miscellaneous'] | [ 3.23732883e-01 1.20101190e+00 -1.78980842e-01 -4.44386214e-01
3.12315017e-01 -4.73622024e-01 1.00514030e+00 4.09849465e-01
3.22859645e-01 4.37371284e-01 5.02762020e-01 -6.48449838e-01
-9.42108691e-01 -9.02515411e-01 -3.38022202e-01 -1.48486167e-01
1.33064076e-01 6.07987225e-01 1.98495328e-01 -6.13275766e-01
3.88997316e-01 3.23442668e-01 -2.02353740e+00 1.87680572e-01
8.97202849e-01 7.25123167e-01 2.95480508e-02 1.20789550e-01
-7.56869793e-01 1.22656870e+00 -4.50569004e-01 -7.18768954e-01
-5.92624843e-01 -4.32378322e-01 -1.49149036e+00 4.64544684e-01
-2.48054549e-01 2.49280125e-01 9.80357919e-03 1.13387728e+00
-4.79642451e-01 -1.18529253e-01 3.93498659e-01 -1.69848704e+00
-1.18808258e+00 7.36207843e-01 3.53899598e-01 -1.07555494e-01
8.69657457e-01 -3.22432846e-01 9.98756111e-01 -4.56298470e-01
8.99142802e-01 1.43151820e+00 4.81745005e-01 5.43461204e-01
-1.08955860e+00 -7.36797228e-02 1.40137319e-02 5.12250543e-01
-1.29652154e+00 -3.72212887e-01 4.11914349e-01 -6.32341683e-01
8.75774682e-01 5.32691061e-01 6.46362662e-01 6.34986579e-01
1.46934301e-01 6.37111813e-02 8.48048449e-01 -1.08991230e+00
2.26057544e-01 8.20515931e-01 7.33882546e-01 6.15010679e-01
8.70415926e-01 -8.52443725e-02 -5.87273955e-01 -9.71164182e-02
7.71563709e-01 -1.22268692e-01 -1.01123877e-01 -6.31619513e-01
-9.41026926e-01 8.32964301e-01 1.16167106e-01 8.82780433e-01
-4.23422962e-01 1.62551776e-01 1.02931283e-01 1.18208453e-01
6.47135079e-02 4.84854192e-01 -4.03653324e-01 4.15285416e-02
9.71916169e-02 2.74065912e-01 1.17694283e+00 1.02425241e+00
7.68228710e-01 -5.68927899e-02 5.94789982e-01 2.45602041e-01
9.53583598e-01 -7.81498030e-02 3.95755142e-01 -1.41793287e+00
-2.18951508e-01 1.27247334e+00 4.24360931e-01 -9.84040022e-01
-2.53449947e-01 -4.44269478e-01 -1.13645621e-01 -6.10329444e-03
2.63631046e-01 4.83974904e-01 -2.97848523e-01 1.64820170e+00
3.76127213e-01 7.28586167e-02 6.66946113e-01 7.87802517e-01
7.87048221e-01 1.00934938e-01 4.40642208e-01 -1.56677440e-01
1.84660983e+00 -5.49167275e-01 -1.23168468e+00 -2.75039315e-01
7.50976443e-01 -2.28152633e-01 8.68212879e-01 2.88931578e-02
-9.81379867e-01 -1.12643085e-01 -1.00498128e+00 -2.60551006e-01
-8.83462191e-01 -3.85514975e-01 1.23105025e+00 5.61522126e-01
-9.84308124e-01 4.83875841e-01 -4.19973195e-01 -7.22658277e-01
1.47100314e-02 1.60597026e-01 -3.38434726e-01 3.14873755e-02
-1.36555791e+00 1.24075282e+00 6.95771873e-01 7.50948265e-02
-9.80741829e-02 -2.83842713e-01 -1.12009275e+00 2.86595970e-01
6.09092236e-01 -8.80095303e-01 1.05835700e+00 -1.21627045e+00
-1.03716588e+00 1.24680865e+00 -2.14813322e-01 -5.60485840e-01
8.06852244e-03 -4.30278443e-02 -1.10780382e+00 -3.48287635e-02
4.15205300e-01 2.90634334e-01 1.47710502e-01 -1.53971505e+00
-5.26713312e-01 -6.09685123e-01 9.30648208e-01 -2.13836059e-01
-1.49662187e-02 7.44607598e-02 -2.94729844e-02 6.55628219e-02
7.59873152e-01 -5.78577995e-01 -7.44653866e-02 -1.71987817e-01
-3.24940652e-01 -5.00649810e-01 5.55300415e-01 -3.44354808e-01
1.41626835e+00 -2.04016066e+00 1.12503901e-01 4.41185474e-01
5.77514946e-01 -2.23720029e-01 5.49242616e-01 7.97953546e-01
-2.35185757e-01 5.24242580e-01 -2.06958488e-01 1.02176711e-01
7.91827619e-01 9.16943073e-01 -4.60826099e-01 1.50769308e-01
-1.04807289e-02 8.20808947e-01 -7.75113463e-01 -3.76944721e-01
3.51390690e-01 4.98407394e-01 -2.31624782e-01 -2.84624279e-01
-4.40999180e-01 3.49671453e-01 -6.73957705e-01 5.53380787e-01
7.32344091e-02 -6.05737925e-01 8.95901978e-01 1.25534520e-01
-3.98035496e-01 4.80892450e-01 -1.07926965e+00 1.63307858e+00
-3.35825413e-01 5.66464484e-01 -3.59941542e-01 -9.02228415e-01
7.94671118e-01 6.65765107e-01 -1.60804056e-02 -4.16390538e-01
1.62917316e-01 5.11341035e-01 2.14796185e-01 -8.48746061e-01
4.15516943e-01 -5.79610884e-01 -3.60709243e-02 5.09665549e-01
-2.85245091e-01 -5.60336001e-03 1.78759351e-01 2.59038568e-01
4.78898376e-01 4.43089813e-01 9.10599172e-01 -4.66564268e-01
9.37668145e-01 3.66926134e-01 2.44900033e-01 4.91878927e-01
1.62382409e-01 -1.59884155e-01 7.62866795e-01 -7.63195515e-01
-9.96735990e-01 -6.30097628e-01 -5.09085119e-01 5.91413796e-01
5.25258243e-01 -7.43341088e-01 -8.20498228e-01 -1.13144249e-01
-1.11031011e-01 1.40877628e+00 -6.66021883e-01 -1.03288002e-01
-3.31645012e-02 -1.09349884e-01 3.25639993e-01 2.24822536e-01
5.02000377e-02 -1.10049784e+00 -1.23156929e+00 2.12235153e-01
-1.68054223e-01 -1.32325745e+00 8.95250618e-01 -5.27079636e-03
-7.38662541e-01 -1.29051459e+00 4.78267461e-01 -2.61153698e-01
6.52525187e-01 1.61467213e-02 1.13506496e+00 8.58637631e-01
6.00444749e-02 6.55481815e-01 -7.15826571e-01 -6.93614781e-01
-7.34339118e-01 -4.03806120e-01 -5.87665029e-02 -2.37112343e-02
6.68496966e-01 -7.14226365e-01 1.09056696e-01 1.09776899e-01
-1.41825223e+00 3.52696538e-01 7.22924396e-02 1.12322576e-01
1.54529169e-01 4.64618474e-01 3.40765864e-01 -1.13417709e+00
5.39733112e-01 -6.15525663e-01 -2.03487694e-01 4.25759345e-01
-8.41607332e-01 4.24128354e-01 7.42749721e-02 3.67515273e-02
-1.15396655e+00 -7.90738761e-01 2.78260678e-01 4.29044142e-02
-4.85912532e-01 7.55681694e-01 -3.90391290e-01 2.08524153e-01
4.96417701e-01 1.45124137e-01 2.59513050e-01 -4.08533603e-01
3.89234036e-01 4.58255559e-01 5.21663904e-01 -6.89736187e-01
7.57590473e-01 7.04346001e-01 4.26286519e-01 -6.12909198e-01
-9.75717723e-01 -1.81012407e-01 -6.20034814e-01 -1.31673589e-01
9.27656829e-01 -3.42317790e-01 -9.44093585e-01 -5.11359394e-01
-1.26591814e+00 3.58235419e-01 -6.06862545e-01 4.86259252e-01
-8.12298715e-01 3.06056082e-01 2.61564348e-02 -1.25186074e+00
2.98110902e-01 -8.56941700e-01 7.00518966e-01 1.82211474e-01
-7.99905717e-01 -1.39407063e+00 -3.98632199e-01 4.83298570e-01
2.50717402e-01 3.80296737e-01 1.56510818e+00 -1.06300640e+00
-3.92202884e-01 -2.39109352e-01 -1.87246561e-01 -1.84499830e-01
1.74611077e-01 -2.49410138e-01 -8.84478688e-01 6.69350386e-01
1.65710986e-01 2.53346890e-01 -3.79420847e-01 9.51024797e-03
5.76246023e-01 -5.30895948e-01 -3.54982793e-01 -1.55519135e-02
1.66744578e+00 4.72158253e-01 9.53675270e-01 7.60115206e-01
2.52189606e-01 1.46579587e+00 5.39282799e-01 2.53173977e-01
5.60898244e-01 8.94667864e-01 5.36475837e-01 2.87805825e-01
7.72846416e-02 -1.97894946e-01 -2.46914089e-01 3.85982931e-01
-6.11824512e-01 1.94675758e-01 -1.09217286e+00 1.53083578e-01
-2.18706393e+00 -1.08866525e+00 -8.49308908e-01 1.91651857e+00
3.52435082e-01 4.68714051e-02 -3.33410561e-01 2.14281261e-01
6.50981188e-01 -2.54053682e-01 -9.75799114e-02 -6.27227902e-01
-1.95315510e-01 -3.62902731e-02 -2.39107475e-01 9.20566976e-01
-4.25061703e-01 8.76356363e-01 5.93144989e+00 2.65583694e-01
-3.57749730e-01 4.21868354e-01 -1.36602074e-01 7.21633196e-01
-8.91233802e-01 7.63685942e-01 -3.42126012e-01 7.13058263e-02
1.05369341e+00 -5.44230640e-01 4.24451113e-01 7.96003878e-01
2.64123857e-01 -1.64847583e-01 -1.11658835e+00 4.28155810e-01
-4.47979681e-02 -1.55077136e+00 4.25295770e-01 4.10739541e-01
2.72879124e-01 -9.45812643e-01 -5.04341543e-01 -1.62379786e-01
2.14359000e-01 -1.03207159e+00 1.31827283e+00 6.41273558e-01
3.75276268e-01 -3.80206168e-01 1.03316998e+00 3.31498653e-01
-9.62690115e-01 -1.80476934e-01 -1.47590145e-01 -5.75119734e-01
3.24125856e-01 1.98129058e-01 -3.84229273e-01 1.11188090e+00
2.75424451e-01 3.86160284e-01 -1.00847848e-01 4.07038540e-01
-5.78939736e-01 5.87593422e-05 1.85590032e-02 1.20836914e-01
1.69119582e-01 -4.97778535e-01 4.08358127e-01 6.68416500e-01
3.70349884e-01 4.68714237e-01 -3.61323655e-01 1.22819579e+00
5.23204207e-01 -1.67450294e-01 -7.32540309e-01 -2.02590138e-01
5.16347170e-01 6.64924443e-01 -7.90070176e-01 -5.72554708e-01
-4.61163431e-01 5.87425411e-01 -2.89635509e-01 3.17488372e-01
-8.79009843e-01 1.65428817e-02 7.67215252e-01 3.67743164e-01
-4.22156394e-01 2.76830811e-02 -2.73822218e-01 -1.00860310e+00
-1.97095558e-01 -5.28065503e-01 3.35953571e-02 -1.39637005e+00
-8.09525847e-01 5.22850811e-01 3.77076596e-01 -8.35941911e-01
-4.80740815e-01 -7.36860514e-01 -2.54128337e-01 7.34454989e-01
-1.26883268e+00 -1.27162504e+00 -3.02857339e-01 2.54085898e-01
1.95097819e-01 4.17884767e-01 1.32273948e+00 -4.18349057e-02
-9.88903418e-02 -4.40031052e-01 -5.11755586e-01 -4.92387384e-01
-2.19345942e-01 -1.07617366e+00 9.58066434e-02 5.02930939e-01
2.17179626e-01 1.07316732e+00 1.28766513e+00 -4.95956570e-01
-1.31109285e+00 -2.84937859e-01 1.85517061e+00 -7.11408436e-01
1.09083021e+00 1.24288887e-01 -1.10192394e+00 1.30049026e+00
2.57518858e-01 -4.68406975e-01 8.86473298e-01 1.79646894e-01
-3.73472661e-01 4.01536077e-01 -1.13181019e+00 7.06485391e-01
1.38732767e+00 -6.00632131e-01 -1.23778605e+00 3.54697704e-01
1.04551446e+00 -9.82561857e-02 -1.00814176e+00 2.72494797e-02
7.92223573e-01 -1.32714677e+00 1.05476260e+00 -1.08317769e+00
3.18071425e-01 -5.22556424e-01 -4.00263906e-01 -5.44735968e-01
-3.36741835e-01 -3.77278894e-01 2.21048272e-03 1.12798548e+00
3.84185553e-01 -1.23912895e+00 3.55198473e-01 1.09733093e+00
-3.42082828e-01 -2.59810418e-01 -9.30961311e-01 -6.20983481e-01
-4.34537590e-01 -7.89714038e-01 1.14134037e+00 1.34604883e+00
8.30844343e-01 3.20055902e-01 1.90842777e-01 3.65472496e-01
6.05853558e-01 2.68107150e-02 5.17003119e-01 -1.97797191e+00
-1.56697948e-02 -2.60496348e-01 -7.08528221e-01 -3.18141371e-01
3.72482240e-01 -8.18636060e-01 -3.87998879e-01 -2.00972652e+00
-1.44686848e-01 -2.56240904e-01 2.13385865e-01 4.26657796e-01
4.68025684e-01 -1.89732850e-01 1.29931942e-01 6.79567575e-01
-3.63809079e-01 1.97051261e-02 9.18451607e-01 4.97488618e-01
1.17641576e-01 -5.35216153e-01 -1.14240205e+00 1.17611885e+00
6.28077149e-01 -5.67271888e-01 -5.80795944e-01 -2.81328440e-01
7.42897451e-01 2.38334253e-01 9.33897376e-01 -7.46415555e-01
1.64902493e-01 -4.97048199e-01 -4.07596976e-01 2.99904019e-01
2.83283830e-01 -1.38971138e+00 9.96255755e-01 6.47629380e-01
-3.17656010e-01 -6.41458258e-02 9.10482090e-03 2.17174038e-01
-1.95074841e-01 -8.65450978e-01 1.76549390e-01 -1.57147735e-01
-1.10627496e+00 -5.80733716e-01 -4.33881253e-01 -4.35111493e-01
9.62811887e-01 -7.94330955e-01 -4.72274274e-01 -2.91762471e-01
-1.22940147e+00 -1.46358386e-01 5.89182198e-01 1.34929031e-01
3.32525045e-01 -1.06773973e+00 -3.65512748e-03 -1.16095496e-02
3.24643284e-01 -3.55461985e-01 2.51642168e-01 8.75706613e-01
-5.74945331e-01 9.61129904e-01 -1.27852395e-01 -6.64259940e-02
-8.24346006e-01 5.80282390e-01 3.12978148e-01 1.19989753e-01
-7.24968255e-01 2.20040366e-01 4.28988367e-01 -3.20243150e-01
-1.54602438e-01 -2.18915030e-01 -5.21832228e-01 -2.98410177e-01
3.27961653e-01 3.11272591e-01 -4.74192172e-01 -1.00011444e+00
-6.15746081e-01 4.68986869e-01 6.46953523e-01 -1.09136984e-01
1.06242454e+00 -5.49226999e-01 -7.39554524e-01 8.09592247e-01
3.64400417e-01 2.04518493e-02 -3.52532715e-01 -1.28295317e-01
7.05893934e-01 -2.60566413e-01 -4.84531224e-01 -7.82745481e-01
-1.62372559e-01 5.78425050e-01 4.57792878e-02 1.24043119e+00
5.67313969e-01 6.05276763e-01 7.22587332e-02 4.58428264e-02
6.30376577e-01 -5.92137575e-01 -5.14135301e-01 1.23754464e-01
9.20441508e-01 -6.42736971e-01 1.01063689e-02 -1.28366697e+00
-4.71782237e-01 1.42727184e+00 2.23117977e-01 3.46080631e-01
2.88002491e-01 1.85862673e-03 2.90064048e-02 -9.59532857e-01
-6.80721104e-01 -5.76328337e-01 2.65346579e-02 7.72176564e-01
4.17979181e-01 2.15804428e-01 -6.17798686e-01 8.78816545e-01
-3.39576364e-01 3.08334261e-01 7.64263868e-01 9.07147646e-01
-8.24960887e-01 -1.21843517e+00 -6.74977660e-01 -2.06009045e-01
-5.02408803e-01 -5.16700745e-02 -5.53341448e-01 1.32655907e+00
5.31354308e-01 1.18421900e+00 2.68889725e-01 -6.39417320e-02
4.32201177e-01 3.42688382e-01 3.46895218e-01 -6.28228128e-01
-2.98778743e-01 -3.48108441e-01 6.29145145e-01 -3.69778544e-01
-9.11723793e-01 -3.50960255e-01 -1.88920677e+00 -4.87635583e-01
-1.84052929e-01 8.03249002e-01 8.89336169e-01 1.62621188e+00
4.66910116e-02 7.09729671e-01 -2.38445923e-01 -2.33820766e-01
-1.59579709e-01 -5.82630634e-01 -7.15062380e-01 3.31099540e-01
-2.40532681e-01 -9.29759383e-01 -5.67548394e-01 2.90073097e-01] | [8.860614776611328, 6.820672035217285] |
992f0c06-7739-4467-9879-5c12e37f0156 | expunations-augmenting-puns-with-keywords-and | 2210.13513 | null | https://arxiv.org/abs/2210.13513v1 | https://arxiv.org/pdf/2210.13513v1.pdf | ExPUNations: Augmenting Puns with Keywords and Explanations | The tasks of humor understanding and generation are challenging and subjective even for humans, requiring commonsense and real-world knowledge to master. Puns, in particular, add the challenge of fusing that knowledge with the ability to interpret lexical-semantic ambiguity. In this paper, we present the ExPUNations (ExPUN) dataset, in which we augment an existing dataset of puns with detailed crowdsourced annotations of keywords denoting the most distinctive words that make the text funny, pun explanations describing why the text is funny, and fine-grained funniness ratings. This is the first humor dataset with such extensive and fine-grained annotations specifically for puns. Based on these annotations, we propose two tasks: explanation generation to aid with pun classification and keyword-conditioned pun generation, to challenge the current state-of-the-art natural language understanding and generation models' ability to understand and generate humor. We showcase that the annotated keywords we collect are helpful for generating better novel humorous texts in human evaluation, and that our natural language explanations can be leveraged to improve both the accuracy and robustness of humor classifiers. | ['Nanyun Peng', 'Yang Liu', 'Jing Huang', 'Tagyoung Chung', 'Alessandra Cervone', 'Shereen Oraby', 'Anjali Narayan-Chen', 'Jiao Sun'] | 2022-10-24 | null | null | null | null | ['explanation-generation'] | ['natural-language-processing'] | [ 4.65767048e-02 2.14445144e-01 6.54549003e-02 -8.94789770e-02
-4.66798961e-01 -7.54394352e-01 8.08702230e-01 2.09180161e-01
-2.58200858e-02 9.11881268e-01 9.99449968e-01 -1.01415262e-01
3.69633406e-01 -6.04050875e-01 -3.65262687e-01 -1.47106936e-02
4.18144494e-01 6.80363297e-01 -2.66723245e-01 -8.05539668e-01
5.73454738e-01 -4.61121649e-02 -1.52757692e+00 8.32381308e-01
1.21374059e+00 5.96848667e-01 -2.24181451e-02 7.49479890e-01
-1.53359860e-01 1.55380464e+00 -1.01061428e+00 -8.49471390e-01
-2.13416889e-02 -6.84294820e-01 -1.22520721e+00 2.16696076e-02
3.25460196e-01 -4.43123102e-01 -1.09511256e-01 6.91036522e-01
5.99301577e-01 2.56264359e-01 8.72884572e-01 -1.37445402e+00
-1.62142420e+00 1.37251008e+00 -7.37029687e-02 3.61006474e-04
8.07879448e-01 6.48092389e-01 1.41254342e+00 -1.16831338e+00
6.95355237e-01 1.49802434e+00 6.67544365e-01 9.79442656e-01
-1.08021009e+00 -4.49579954e-01 -5.07629931e-01 7.09588408e-01
-8.68163645e-01 -2.12352544e-01 9.24553871e-01 -6.43080890e-01
1.16159177e+00 2.78083354e-01 3.55881274e-01 1.75662529e+00
-3.07563156e-01 9.77559149e-01 1.16343391e+00 -4.70279098e-01
2.36171171e-01 1.56186372e-01 2.08702192e-01 6.36825979e-01
6.37691617e-02 -4.74876054e-02 -1.08776605e+00 -2.63792694e-01
4.13038820e-01 -2.12130517e-01 -4.28852856e-01 4.24814343e-01
-1.16400480e+00 1.30284965e+00 6.13212466e-01 1.73914343e-01
-3.04533064e-01 8.65520388e-02 6.10652387e-01 -3.59494612e-02
2.41177246e-01 1.36577272e+00 -1.24444470e-01 -3.61994505e-01
-6.91883683e-01 8.02814603e-01 1.14577079e+00 9.63170886e-01
4.20096427e-01 1.16303168e-01 -5.86417317e-01 9.34923470e-01
-2.13127598e-01 5.05844235e-01 8.27021122e-01 -9.72270608e-01
2.75639921e-01 8.74027610e-01 4.04430956e-01 -1.05194366e+00
-5.23375332e-01 -1.60039768e-01 -4.47105289e-01 4.31765690e-02
4.54554051e-01 5.16633838e-02 -4.63649005e-01 1.60576856e+00
-2.03823194e-01 -2.11765617e-01 1.05566792e-01 1.30589175e+00
1.30435491e+00 3.48702073e-01 1.64943337e-01 6.48689419e-02
1.56081843e+00 -1.28931677e+00 -5.78626633e-01 -4.49044377e-01
6.33796871e-01 -9.10946429e-01 1.92359078e+00 3.00527066e-01
-8.63203883e-01 -3.21932405e-01 -9.69506741e-01 -6.35001659e-01
-4.88071740e-01 -1.44060329e-02 6.30398870e-01 2.52232254e-01
-3.76472354e-01 7.53638506e-01 4.97930460e-02 -2.97614396e-01
4.00398254e-01 -4.63773906e-01 -1.03690602e-01 9.12736356e-02
-1.69370651e+00 1.40907252e+00 4.83961046e-01 -4.15960580e-01
-8.75592053e-01 -8.51069808e-01 -9.85421062e-01 -1.16133327e-02
2.37524003e-01 -9.42744076e-01 1.55212355e+00 -8.57867599e-01
-9.11275744e-01 9.49914932e-01 -4.21336778e-02 -6.45550132e-01
4.03981596e-01 -4.18965727e-01 -4.46484685e-02 -9.42143425e-03
6.10197902e-01 6.83180690e-01 7.69298434e-01 -1.38452017e+00
-4.18364862e-03 -3.52296904e-02 2.97508657e-01 6.25231788e-02
-5.56825042e-01 5.52727431e-02 6.31507754e-01 -8.27072799e-01
-5.66511393e-01 -5.75002789e-01 6.46196529e-02 -5.00582039e-01
-7.72687733e-01 -4.23186600e-01 5.75804889e-01 -8.49967062e-01
1.23201013e+00 -1.67953873e+00 1.01291932e-01 -3.93396586e-01
5.39554834e-01 2.62955297e-02 -2.11500317e-01 6.26554310e-01
1.81712613e-01 3.23199242e-01 -1.05627306e-01 -4.28225338e-01
5.17622948e-01 2.26448447e-01 -8.24794888e-01 -2.67316610e-01
2.93417305e-01 1.33398819e+00 -1.41214693e+00 -4.12694514e-01
-4.27084938e-02 4.19943966e-02 -5.09773433e-01 5.80370069e-01
-5.67285895e-01 1.57596976e-01 -1.72861069e-01 6.41953707e-01
8.39381739e-02 -4.41320539e-01 -3.25747281e-01 7.70700574e-02
3.33924979e-01 7.10111976e-01 -5.87486982e-01 1.20707929e+00
-6.26623154e-01 7.28849471e-01 -7.73957908e-01 1.01975940e-01
1.07498538e+00 1.64979175e-01 -1.83432743e-01 -5.01962960e-01
2.19068721e-01 3.96624446e-01 -2.43037358e-01 -6.27326667e-01
1.16148627e+00 -6.28608286e-01 -4.19558018e-01 7.25230575e-01
-4.36560027e-02 -6.12720072e-01 3.94094259e-01 5.65613866e-01
1.19221544e+00 -1.54700741e-01 7.79294789e-01 -1.23483330e-01
3.92234206e-01 3.49018931e-01 1.06661238e-01 8.66375327e-01
-1.48983285e-01 6.36652172e-01 5.43271005e-01 -5.36537170e-01
-1.23408961e+00 -1.02343810e+00 1.85982138e-01 1.27831364e+00
-1.60274863e-01 -6.92542195e-01 -6.50755405e-01 -7.09664345e-01
2.34780803e-01 1.62515366e+00 -7.70720959e-01 -2.89115938e-03
-3.55953872e-01 -3.51924598e-01 8.97780180e-01 7.00672388e-01
2.93295354e-01 -1.66138029e+00 -9.46276963e-01 3.25475156e-01
-7.06382096e-01 -1.17785168e+00 -6.32244945e-01 2.20874861e-01
-3.06027651e-01 -1.34400499e+00 -1.94094002e-01 -3.35548192e-01
3.02100092e-01 2.16799885e-01 1.77753937e+00 6.04289651e-01
-3.68878186e-01 2.24449232e-01 -9.39251840e-01 -6.89014852e-01
-7.99821496e-01 -1.91959858e-01 -4.86126840e-02 -7.33833909e-01
7.60941625e-01 -6.92933202e-01 -2.30138958e-01 1.47882968e-01
-8.09877574e-01 2.97486782e-01 4.97885011e-02 1.22638464e+00
-2.08011791e-01 -4.68778223e-01 8.59394431e-01 -9.94812846e-01
1.33361065e+00 -5.81739783e-01 2.89994389e-01 -1.63047746e-01
-4.45464939e-01 5.58889844e-02 1.21268892e+00 -3.90908331e-01
-8.28545034e-01 -2.52811402e-01 3.60444218e-01 -3.28176379e-01
-5.16634285e-02 3.22436392e-01 1.20421775e-01 3.94976050e-01
1.69031632e+00 6.48614671e-03 -2.63161987e-01 -2.58486718e-01
1.14198136e+00 7.61608720e-01 1.01321554e+00 -8.88323247e-01
8.91586363e-01 9.88733992e-02 -1.65847540e-01 -2.90651053e-01
-1.49756038e+00 -6.13558769e-01 -6.20367885e-01 -1.81711316e-01
7.59326279e-01 -7.68401861e-01 -5.50831020e-01 1.84894621e-01
-1.95707607e+00 -2.91235775e-01 -4.51191962e-01 -2.91709840e-01
-9.04573500e-01 4.36105549e-01 -7.62201190e-01 -9.03970361e-01
-7.11257160e-01 -5.20955086e-01 9.72016335e-01 2.40858808e-01
-1.21018553e+00 -1.03964603e+00 -4.20550331e-02 9.90576208e-01
2.08098948e-01 4.07432884e-01 1.22008073e+00 -1.09599137e+00
-1.23404779e-01 9.78153720e-02 -4.06035364e-01 3.38492155e-01
-1.85807243e-01 -4.44288701e-01 -1.14051235e+00 5.67310631e-01
-1.04240239e-01 -1.25766206e+00 9.37989593e-01 -3.37541550e-01
9.74393487e-01 -1.00609565e+00 2.76553065e-01 -5.60874864e-02
9.40854907e-01 -5.43593705e-01 6.44704819e-01 4.25907433e-01
6.67579532e-01 6.45652533e-01 5.40743053e-01 1.05692613e+00
5.04737198e-01 4.15183693e-01 5.36993504e-01 1.50008559e-01
-2.34641135e-01 -7.22828090e-01 4.83284295e-01 3.88291508e-01
-6.74926564e-02 -1.98357806e-01 -8.40064168e-01 7.25880980e-01
-2.05976272e+00 -1.42213500e+00 -5.14265895e-01 1.59127545e+00
1.40784705e+00 -1.07580185e-01 1.20204680e-01 8.31085593e-02
3.18911165e-01 2.80275464e-01 -2.83743829e-01 -9.37812626e-01
-3.53850991e-01 1.28528863e-01 -4.93496656e-02 7.83550739e-01
-7.52623260e-01 1.20867741e+00 5.78013277e+00 7.24562407e-01
-3.46948087e-01 7.33666644e-02 2.64662802e-01 -8.83762315e-02
-8.15977514e-01 -4.79953783e-03 -4.42643702e-01 3.02488029e-01
5.78379452e-01 -2.65980035e-01 8.81107152e-01 1.13949323e+00
3.93251002e-01 6.55094162e-02 -1.47798181e+00 9.53263998e-01
7.71802604e-01 -1.29137933e+00 5.48875511e-01 -5.34715772e-01
8.88444126e-01 -6.76618397e-01 -2.23439902e-01 5.19896567e-01
5.33390343e-01 -1.69962800e+00 9.83989358e-01 4.43203688e-01
3.50960463e-01 -4.34546173e-01 8.89903665e-01 5.14278352e-01
-5.61396420e-01 -2.67322868e-01 -4.13632095e-01 -7.57377923e-01
3.75281930e-01 6.47929788e-01 -1.56416702e+00 3.16230766e-03
2.02458084e-01 6.35951877e-01 -1.02909613e+00 4.61486965e-01
-1.10352397e+00 4.60398376e-01 3.65131170e-01 -6.59558594e-01
6.78730160e-02 4.54879463e-01 7.16776669e-01 1.46082950e+00
1.04567826e-01 3.75413746e-01 4.01900470e-01 1.72470033e+00
-4.31167006e-01 1.27986461e-01 -4.62428570e-01 -3.41616571e-01
6.20487273e-01 1.48585296e+00 -1.20069973e-01 -4.30949360e-01
1.27044186e-01 1.32615769e+00 6.31069839e-01 -4.99467179e-03
-8.42472613e-01 -1.78556487e-01 4.37224329e-01 2.33859256e-01
-3.29942495e-01 3.37901227e-02 -1.19198012e+00 -1.06285572e+00
-2.33279034e-01 -1.09569311e+00 3.75172615e-01 -1.45391238e+00
-2.01520228e+00 4.15391177e-01 -2.96238303e-01 -6.86673045e-01
-5.83490670e-01 -8.51792455e-01 -9.61623311e-01 1.00429296e+00
-1.20142698e+00 -1.59784591e+00 -5.98724604e-01 2.96349883e-01
9.34900641e-01 -1.80823177e-01 8.83964062e-01 -4.06926662e-01
1.98667973e-01 3.88154894e-01 -7.84385085e-01 1.66187674e-01
9.52863395e-01 -1.68348038e+00 4.07544136e-01 6.46041811e-01
7.88822323e-02 7.63733149e-01 1.41641390e+00 -8.13157260e-01
-7.73147583e-01 -9.44783151e-01 1.43678892e+00 -1.31739402e+00
1.12952316e+00 -1.85312435e-01 -8.88696551e-01 4.06465828e-01
3.45185161e-01 -6.22999907e-01 8.15592825e-01 3.24121445e-01
-1.12484980e+00 8.27733219e-01 -1.04424775e+00 7.37707019e-01
1.10611773e+00 -7.59869754e-01 -1.23581564e+00 7.99651504e-01
8.94024551e-01 -2.75566041e-01 -2.71532953e-01 1.34148309e-02
2.14763641e-01 -9.25328314e-01 6.66928172e-01 -1.23987126e+00
1.72476459e+00 -4.01398540e-01 -1.96894795e-01 -1.64407539e+00
-3.45003843e-01 -5.53372264e-01 -1.75963238e-01 1.23277926e+00
5.14733076e-01 1.89057961e-01 3.94117057e-01 6.97813511e-01
-2.78436899e-01 -5.87072134e-01 -5.62941730e-01 -6.96882784e-01
1.50021940e-01 -4.09282595e-01 5.67767143e-01 1.02460098e+00
8.88308287e-01 8.89122725e-01 -6.26903772e-01 -3.46979380e-01
4.52621728e-01 1.17037758e-01 9.32860732e-01 -9.56885457e-01
-4.65963662e-01 -5.02988756e-01 -2.67693490e-01 -7.13240206e-01
5.59704244e-01 -1.21141529e+00 3.49915743e-01 -1.52097762e+00
9.08131003e-01 3.37873250e-01 3.56032640e-01 8.95339072e-01
-6.61588371e-01 3.48974735e-01 5.57847619e-01 6.56744838e-02
-7.59205222e-01 6.95827544e-01 1.22585726e+00 -7.29013905e-02
-1.55755132e-01 -7.20515370e-01 -1.16126955e+00 8.41571987e-01
6.75019681e-01 -2.14309156e-01 -2.46299297e-01 -2.16891244e-01
3.98021817e-01 -2.37451836e-01 1.12401795e+00 -6.34116948e-01
6.01959676e-02 -4.93580967e-01 3.78384322e-01 -1.83449298e-01
2.36847967e-01 -1.69593349e-01 -4.48820770e-01 2.11967841e-01
-6.01874530e-01 -5.17902970e-02 -2.48054806e-02 3.47258240e-01
1.08341418e-01 -5.72179437e-01 7.82374799e-01 -3.78834456e-01
-7.05689907e-01 -4.06176120e-01 -1.83740303e-01 5.91421247e-01
5.96344531e-01 -2.78186314e-02 -1.07986605e+00 -8.99481416e-01
-3.69207412e-01 9.76662412e-02 6.66758895e-01 6.50809526e-01
8.12193930e-01 -1.47691119e+00 -1.11665893e+00 -4.82540429e-01
5.61813891e-01 -5.27125061e-01 1.29306346e-01 2.61498332e-01
-1.71239361e-01 1.16148025e-01 -3.41648847e-01 6.19790070e-02
-1.05908561e+00 7.45245755e-01 4.26159389e-02 -2.21950397e-01
-5.15763581e-01 8.43110025e-01 -8.17777216e-02 -4.92588460e-01
-2.29972869e-01 1.84571385e-01 -2.65759230e-01 4.28676568e-02
7.62219787e-01 6.12323463e-01 -3.11942667e-01 -5.22024035e-01
-8.08270834e-03 -2.59905428e-01 4.22556475e-02 -1.65346831e-01
9.56558943e-01 1.28921971e-01 -4.55356956e-01 5.70270479e-01
6.08464837e-01 2.33380184e-01 -6.69171870e-01 3.77786346e-02
1.66637942e-01 -6.26014113e-01 -6.65147364e-01 -1.49226284e+00
1.85832812e-03 8.21319044e-01 -3.90103549e-01 1.35271713e-01
5.72614491e-01 2.26592466e-01 1.17834997e+00 5.11547267e-01
4.11249131e-01 -1.10759425e+00 8.24253440e-01 1.14313006e+00
1.56591928e+00 -1.08651006e+00 -1.34036645e-01 -4.05321091e-01
-1.19611764e+00 1.45552158e+00 1.00177717e+00 -1.34012122e-02
-3.03159893e-01 -7.54508972e-02 1.24914140e-01 -3.05739433e-01
-9.62227702e-01 -1.73127562e-01 5.26522934e-01 6.24665976e-01
9.13325131e-01 3.98084372e-01 -6.09540820e-01 1.56633234e+00
-1.15992963e+00 -2.20662475e-01 1.12069654e+00 1.04060836e-01
-9.50905919e-01 -6.80714309e-01 -4.92179513e-01 3.99589568e-01
1.13160066e-01 -5.86637914e-01 -1.49771512e+00 3.23680729e-01
2.37765744e-01 1.17785561e+00 -7.26817846e-01 -4.91725504e-01
2.82713801e-01 4.24117237e-01 2.61012375e-01 -9.17079329e-01
-8.95838022e-01 -9.80290353e-01 6.49773479e-01 -3.54222387e-01
1.74419925e-01 -2.91771561e-01 -1.59242630e+00 -6.76972985e-01
-3.97999026e-02 4.66606691e-02 9.33593735e-02 1.21459913e+00
2.55054329e-02 2.65505105e-01 3.43851656e-01 -7.19853520e-01
-1.14101338e+00 -1.19715464e+00 -4.51177984e-01 1.30712938e+00
-1.22315302e-01 -5.30292153e-01 -5.89971840e-01 2.88982034e-01] | [8.925348281860352, 10.949594497680664] |
1119d2bb-bf92-46fd-b89f-f63bd66d7c7b | principled-paraphrase-generation-with-1 | 2205.12213 | null | https://arxiv.org/abs/2205.12213v3 | https://arxiv.org/pdf/2205.12213v3.pdf | Principled Paraphrase Generation with Parallel Corpora | Round-trip Machine Translation (MT) is a popular choice for paraphrase generation, which leverages readily available parallel corpora for supervision. In this paper, we formalize the implicit similarity function induced by this approach, and show that it is susceptible to non-paraphrase pairs sharing a single ambiguous translation. Based on these insights, we design an alternative similarity metric that mitigates this issue by requiring the entire translation distribution to match, and implement a relaxation of it through the Information Bottleneck method. Our approach incorporates an adversarial term into MT training in order to learn representations that encode as much information about the reference translation as possible, while keeping as little information about the input as possible. Paraphrases can be generated by decoding back to the source from this representation, without having to generate pivot translations. In addition to being more principled and efficient than round-trip MT, our approach offers an adjustable parameter to control the fidelity-diversity trade-off, and obtains better results in our experiments. | ['Gorka Labaka', 'Aitor Soroa', 'Eneko Agirre', 'Mikel Artetxe', 'Aitor Ormazabal'] | 2022-05-24 | principled-paraphrase-generation-with | https://aclanthology.org/2022.acl-long.114 | https://aclanthology.org/2022.acl-long.114.pdf | acl-2022-5 | ['paraphrase-generation', 'paraphrase-generation'] | ['computer-code', 'natural-language-processing'] | [ 4.80050892e-01 4.21721816e-01 -4.78867561e-01 -3.40549260e-01
-1.20342529e+00 -1.01033926e+00 8.95292103e-01 -1.48640852e-02
-3.11485261e-01 9.90489185e-01 4.61973459e-01 -5.99015236e-01
1.77245006e-01 -7.22999811e-01 -1.08676267e+00 -5.24331391e-01
4.43031937e-01 5.76234400e-01 -1.49695069e-01 -4.96239543e-01
4.03092623e-01 3.15736294e-01 -8.97154391e-01 3.48370820e-01
1.13512385e+00 3.34065229e-01 2.45443597e-01 4.58893269e-01
-3.94016504e-02 5.06962359e-01 -6.43242121e-01 -9.28338349e-01
6.35676205e-01 -8.26622486e-01 -1.08482444e+00 -1.23335995e-01
2.57388145e-01 -2.02076375e-01 -3.82406414e-01 9.12682950e-01
5.13781190e-01 -7.31513277e-02 7.76970983e-01 -1.03792310e+00
-8.43012929e-01 7.54997969e-01 -3.71115178e-01 2.63526529e-01
4.83111709e-01 1.99824899e-01 1.32542503e+00 -9.59796011e-01
8.24390411e-01 9.39772964e-01 6.21320963e-01 5.49226105e-01
-1.56763375e+00 -5.04950821e-01 -2.24185079e-01 -4.81073335e-02
-1.21333432e+00 -7.08526134e-01 7.60980487e-01 -1.76284954e-01
8.17435503e-01 2.90420443e-01 4.21096832e-01 1.39634061e+00
4.38801795e-01 7.43243217e-01 9.73407030e-01 -6.98396444e-01
1.58110932e-01 1.98305890e-01 -2.67598867e-01 4.10405815e-01
1.60330281e-01 5.24180382e-02 -5.94725192e-01 -4.21725482e-01
7.08216310e-01 -2.51342684e-01 -4.24922556e-01 -6.00035369e-01
-1.31250238e+00 9.35463428e-01 4.61377263e-01 1.96126744e-01
-1.81892991e-01 -3.64150293e-02 3.41066182e-01 8.69333565e-01
4.08237427e-01 8.65569174e-01 -2.59274364e-01 -1.48493975e-01
-1.08870816e+00 3.60132456e-01 8.57956767e-01 1.13883054e+00
9.45395112e-01 -3.09566915e-01 -2.38592848e-01 8.62220347e-01
-2.16405988e-02 4.30276424e-01 7.00572491e-01 -9.05304313e-01
1.06963837e+00 2.48516873e-01 2.15615302e-01 -6.71573579e-01
1.18952289e-01 -3.98246497e-01 -6.41844749e-01 -9.05938298e-02
3.72110009e-01 -3.71651612e-02 -7.33619153e-01 2.06634068e+00
5.68276225e-03 -2.58387953e-01 9.17465910e-02 8.75076771e-01
1.15918741e-02 6.55067801e-01 -4.24990803e-01 -1.98601216e-01
1.02088594e+00 -1.06539297e+00 -2.83687323e-01 -4.48916942e-01
5.47907770e-01 -9.96133387e-01 1.27203155e+00 -1.25130981e-01
-1.35060084e+00 -1.84390366e-01 -1.25608635e+00 -2.17886314e-01
1.20614037e-01 -2.55041718e-01 3.81931633e-01 5.09190619e-01
-1.22912812e+00 1.01620710e+00 -8.29195440e-01 -3.55326474e-01
1.08245425e-01 4.11297351e-01 -3.93306553e-01 -1.75905049e-01
-1.27761555e+00 1.20810401e+00 1.46244273e-01 -1.79899827e-01
-5.79182565e-01 -6.06692374e-01 -6.65075898e-01 -4.85028364e-02
3.24847884e-02 -1.25915313e+00 1.35183847e+00 -1.38184583e+00
-1.88274515e+00 8.11817169e-01 -3.51225048e-01 -4.42237526e-01
7.39394009e-01 3.07749901e-02 1.42186448e-01 5.78883514e-02
2.70374328e-01 6.08868241e-01 9.44703519e-01 -1.02157319e+00
-8.07536170e-02 -1.48573264e-01 1.47487059e-01 4.76585865e-01
-1.04915358e-01 1.18597828e-01 -2.06955671e-01 -9.02792573e-01
1.83638886e-01 -1.20967662e+00 -2.85243362e-01 -1.60748616e-01
-4.66993600e-01 2.37776920e-01 8.10665861e-02 -5.67410767e-01
8.97721767e-01 -1.77707958e+00 6.47328436e-01 1.61873996e-01
-5.34040108e-02 5.38631603e-02 -4.86060232e-01 9.37933803e-01
6.94769472e-02 2.07503557e-01 -6.71420753e-01 -4.74346131e-01
-6.76142201e-02 3.07595462e-01 -6.89399183e-01 3.04419130e-01
4.44060743e-01 1.08076179e+00 -1.02628994e+00 -3.25326413e-01
-2.55482048e-01 1.71334654e-01 -7.15709746e-01 3.41488332e-01
-1.79563358e-01 4.76321518e-01 -3.57689917e-01 2.92421490e-01
5.52428961e-01 -1.52701244e-01 3.63450974e-01 1.35713056e-01
1.16938524e-01 8.89393032e-01 -6.36017025e-01 1.94569755e+00
-7.40651846e-01 5.82715571e-01 -8.64397064e-02 -1.03318655e+00
9.47368145e-01 3.34103078e-01 1.72670603e-01 -6.09808564e-01
-1.24854028e-01 4.87358868e-01 -8.24957788e-02 -1.13743827e-01
4.80609000e-01 -3.71419549e-01 -1.06326891e-02 8.73212039e-01
-4.25996780e-02 -3.04567248e-01 -4.15563621e-02 3.11878264e-01
1.12360013e+00 2.62755305e-01 3.04920018e-01 -2.81278521e-01
3.28976989e-01 9.47331861e-02 5.98556161e-01 8.03550184e-01
3.03929765e-02 9.65098858e-01 4.11462039e-01 -2.22404331e-01
-1.63923669e+00 -1.32147098e+00 1.62097812e-01 7.59639323e-01
1.45292491e-01 -3.62584621e-01 -7.72809803e-01 -7.40169525e-01
-8.42175633e-02 7.25462675e-01 -5.14751792e-01 -2.79099345e-01
-9.54423249e-01 -5.83816946e-01 7.08744884e-01 3.70000273e-01
2.38393843e-01 -8.44846725e-01 -5.39249629e-02 2.95850903e-01
-6.29950643e-01 -6.15449965e-01 -9.17619348e-01 1.36720896e-01
-1.07035875e+00 -7.63620913e-01 -9.30259526e-01 -7.60644794e-01
7.42072642e-01 3.79869968e-01 1.29555357e+00 -1.49961486e-01
1.55486718e-01 -2.03006431e-01 -3.01430076e-01 4.44475748e-02
-9.49836850e-01 6.15190029e-01 -4.03245538e-02 -1.73347920e-01
1.55116186e-01 -9.54125643e-01 -5.59689581e-01 3.39349270e-01
-8.70892167e-01 2.41932869e-01 6.92387521e-01 1.11579227e+00
3.48945022e-01 -6.76187992e-01 6.89874530e-01 -6.77532017e-01
8.67245138e-01 -6.29978299e-01 -2.65147477e-01 2.36487255e-01
-6.59460664e-01 5.24165809e-01 1.03352833e+00 -4.17574137e-01
-7.68163681e-01 -1.91796213e-01 -1.72303822e-02 -4.43418831e-01
3.36228698e-01 2.94434845e-01 -1.07323691e-01 -5.79471067e-02
8.22855592e-01 5.31957269e-01 2.29954556e-01 -5.31458914e-01
6.72066212e-01 7.68521488e-01 4.30119962e-01 -8.96790028e-01
1.08735061e+00 2.42772743e-01 -1.92714170e-01 -2.76608407e-01
-5.69422960e-01 -4.10443060e-02 -6.15147948e-01 3.81263703e-01
3.23592514e-01 -8.57122719e-01 -4.76327054e-02 2.22091600e-02
-1.23597050e+00 -1.71544462e-01 -3.02004457e-01 2.87874132e-01
-9.51080680e-01 4.81845170e-01 -7.51171827e-01 -2.29071036e-01
-4.95582759e-01 -1.21873212e+00 9.16062355e-01 -1.83616117e-01
-5.12218237e-01 -7.36137152e-01 3.19789410e-01 3.18140596e-01
5.92379987e-01 -1.27698019e-01 1.16013336e+00 -7.18754828e-01
-7.45096564e-01 9.10681020e-03 -8.89518782e-02 3.81541252e-01
3.37542534e-01 -2.82415718e-01 -5.85534990e-01 -4.93496686e-01
1.58772320e-01 -3.34290087e-01 7.18216002e-01 -1.85570151e-01
5.74658871e-01 -7.14052856e-01 -8.37970003e-02 7.53425658e-01
1.16206086e+00 -1.48495898e-01 6.39387608e-01 5.59649289e-01
6.10512555e-01 5.53673208e-01 3.21397126e-01 6.59063309e-02
3.40932250e-01 9.55939472e-01 2.29194183e-02 1.68197915e-01
-9.44079310e-02 -6.75122023e-01 5.52951217e-01 1.08889925e+00
1.21351972e-01 -1.17159866e-01 -6.66497052e-01 5.26185155e-01
-1.72248542e+00 -1.01705611e+00 3.92545283e-01 2.40106249e+00
1.29459178e+00 -2.89294589e-02 1.29124641e-01 2.11109146e-02
6.99054003e-01 2.77506113e-01 -4.14964408e-01 -7.57555723e-01
-1.71549827e-01 3.63810360e-01 5.74381888e-01 6.84635997e-01
-6.23811007e-01 1.04385340e+00 6.95370340e+00 6.68729126e-01
-1.01912403e+00 3.56408767e-02 4.45451260e-01 -1.92529082e-01
-9.26720858e-01 3.35808843e-01 -4.16542292e-01 6.14265800e-01
9.23431456e-01 -5.33631921e-01 8.96472037e-01 4.99115080e-01
1.70218915e-01 3.58840317e-01 -1.44627905e+00 5.36246479e-01
-1.44802360e-02 -1.34915411e+00 3.54150742e-01 6.11881074e-03
7.82659054e-01 8.90789777e-02 6.08125748e-03 1.04595244e-01
3.82970840e-01 -9.00266349e-01 7.70398378e-01 8.81598964e-02
7.75398374e-01 -8.62004161e-01 5.32990992e-01 4.87527519e-01
-7.72223949e-01 1.04518786e-01 -5.26121616e-01 -9.02258977e-02
1.42705262e-01 4.50769722e-01 -9.99188006e-01 7.67749786e-01
-7.82283992e-02 4.08320546e-01 -3.63151938e-01 6.71313524e-01
-5.09218514e-01 6.55369818e-01 -2.09918335e-01 2.28779778e-01
1.57320723e-01 -5.03108382e-01 6.50844157e-01 1.15661931e+00
4.76250708e-01 -3.68599653e-01 8.83791670e-02 1.06444597e+00
-4.62257504e-01 1.86476231e-01 -9.45501924e-01 -5.96020138e-03
8.10790777e-01 7.14991689e-01 -2.15576380e-01 -2.27622122e-01
-2.49223053e-01 1.60788131e+00 6.87882245e-01 4.27779973e-01
-6.81354523e-01 -4.87982690e-01 8.39821219e-01 4.32804413e-02
9.45908502e-02 -1.96264520e-01 -4.27219182e-01 -1.42586517e+00
4.29119706e-01 -1.20820141e+00 -3.53552029e-02 -5.94955921e-01
-1.33585393e+00 6.70164347e-01 -1.15327403e-01 -1.28227687e+00
-7.74269819e-01 -1.25529170e-01 -7.11656213e-01 1.28579271e+00
-1.38133717e+00 -1.12703562e+00 3.55597496e-01 2.67722696e-01
5.61450958e-01 -7.68612325e-02 7.15823531e-01 9.90401804e-02
-4.24552023e-01 1.06622100e+00 3.24573576e-01 -4.55968361e-03
8.75543773e-01 -1.09425676e+00 1.11450827e+00 8.15749347e-01
2.51203090e-01 1.01038885e+00 7.53031313e-01 -4.07543302e-01
-1.45449364e+00 -9.61110175e-01 1.38020563e+00 -6.16487086e-01
7.58417010e-01 -4.30648118e-01 -8.79514337e-01 8.64981353e-01
2.72895992e-01 -3.67154539e-01 4.89230067e-01 3.01685953e-03
-7.23465264e-01 7.93799534e-02 -1.03725958e+00 8.82990181e-01
1.28174424e+00 -9.10916388e-01 -1.01143849e+00 3.56790751e-01
9.57433224e-01 -3.08109641e-01 -7.04912066e-01 1.70934558e-01
3.89485329e-01 -8.72383773e-01 9.87749577e-01 -6.65639043e-01
8.59603941e-01 -1.04220457e-01 -1.92990795e-01 -1.70267117e+00
-4.38771546e-01 -1.11354995e+00 1.04657233e-01 1.20630622e+00
7.18672872e-01 -8.07134748e-01 8.67668569e-01 3.90858084e-01
-2.39580959e-01 -8.55446935e-01 -1.07371080e+00 -1.14567745e+00
6.36660516e-01 2.15538070e-01 7.69127905e-01 8.74994755e-01
2.71431178e-01 4.82785940e-01 -5.81516445e-01 4.26777489e-02
4.86240834e-01 3.31374228e-01 8.72722328e-01 -5.81877410e-01
-6.99665844e-01 -3.84823442e-01 -1.29911661e-01 -1.40473616e+00
3.97290021e-01 -1.37272060e+00 -2.70446297e-02 -1.18267584e+00
3.14697146e-01 -4.59612101e-01 -1.72092646e-01 2.67230958e-01
-2.87505329e-01 3.42791766e-01 3.46723586e-01 7.67960787e-01
1.22690983e-01 6.51838005e-01 1.24189115e+00 -7.07942694e-02
-1.51598468e-01 9.60720256e-02 -8.94930780e-01 2.61422545e-01
8.97837818e-01 -6.80270195e-01 -5.09522915e-01 -8.43831837e-01
1.88026264e-01 1.93560183e-01 8.73100013e-02 -5.33427477e-01
2.27436963e-02 -3.34767938e-01 1.06672563e-01 -5.62172458e-02
2.49779761e-01 -3.43486577e-01 2.01857388e-01 2.73172706e-01
-7.19061911e-01 4.87815440e-01 -1.61798447e-01 4.11992073e-01
-9.62626189e-02 -3.83678794e-01 8.17353427e-01 -1.75822631e-01
1.26559392e-01 8.85044690e-03 -2.99685895e-01 3.04286063e-01
5.64739347e-01 -1.22168593e-01 -2.44848266e-01 -4.32600796e-01
-2.12935805e-01 -2.95194648e-02 1.12865257e+00 4.14374143e-01
2.13663116e-01 -1.40361679e+00 -9.43268418e-01 3.67536485e-01
1.79042965e-01 -2.43065223e-01 -3.83901507e-01 4.90015447e-01
-4.72067028e-01 4.18993533e-01 -2.09287629e-01 -3.32846135e-01
-8.21750045e-01 5.25064409e-01 2.98805922e-01 -4.02847111e-01
-6.73779488e-01 6.01475894e-01 7.66047835e-02 -6.26853585e-01
-2.06420809e-01 3.77676375e-02 4.75603193e-01 -3.43043208e-01
3.98688972e-01 7.07283691e-02 2.07188383e-01 -4.57757384e-01
-2.92296678e-01 4.47819442e-01 -3.55335265e-01 -5.76676190e-01
9.15621161e-01 -2.77777165e-01 -2.19477573e-03 1.53878078e-01
1.32338059e+00 2.57534653e-01 -1.15103877e+00 -3.24741840e-01
8.96781236e-02 -6.78856313e-01 -4.70337003e-01 -6.93113863e-01
-6.82834566e-01 7.79323876e-01 -1.32742181e-01 -1.33303434e-01
8.95419538e-01 -1.46574184e-01 1.00885546e+00 5.02419591e-01
6.55548871e-01 -7.91335821e-01 4.38896641e-02 5.26840985e-01
8.64640117e-01 -9.96829152e-01 -1.83071688e-01 -3.82310987e-01
-6.44172192e-01 8.87154400e-01 1.85459778e-01 -3.31346482e-01
-9.58631337e-02 8.67125243e-02 1.79227635e-01 3.96912307e-01
-8.54448378e-01 2.42177650e-01 2.30429508e-02 6.31285667e-01
4.70076948e-01 -6.42104968e-02 -5.38853467e-01 1.90001950e-01
-5.02549946e-01 -8.19694623e-02 5.57111442e-01 9.43780482e-01
-2.30901241e-01 -1.74875224e+00 -2.29547203e-01 1.19381346e-01
-5.03441155e-01 -4.40009654e-01 -7.20144987e-01 4.93904382e-01
-4.29385871e-01 9.12589848e-01 4.07895558e-02 -4.23549622e-01
2.23464295e-01 2.68845648e-01 6.80723429e-01 -6.37894928e-01
-7.54028618e-01 -3.61756057e-01 2.16978312e-01 -5.27745008e-01
1.54058844e-01 -3.67506385e-01 -8.05705726e-01 -5.83303273e-01
-1.76865533e-01 4.09396917e-01 6.84673131e-01 9.44743335e-01
7.31991172e-01 -5.20563461e-02 1.16474485e+00 -9.47971761e-01
-1.32792354e+00 -7.83014536e-01 -7.96720386e-02 6.77147627e-01
3.10573459e-01 -3.15948784e-01 -4.87560034e-01 -1.37661770e-01] | [11.686931610107422, 9.847975730895996] |
31dede66-97c5-476f-acdc-61f0d714c975 | attentive-history-selection-for | 1908.09456 | null | https://arxiv.org/abs/1908.09456v1 | https://arxiv.org/pdf/1908.09456v1.pdf | Attentive History Selection for Conversational Question Answering | Conversational question answering (ConvQA) is a simplified but concrete setting of conversational search. One of its major challenges is to leverage the conversation history to understand and answer the current question. In this work, we propose a novel solution for ConvQA that involves three aspects. First, we propose a positional history answer embedding method to encode conversation history with position information using BERT in a natural way. BERT is a powerful technique for text representation. Second, we design a history attention mechanism (HAM) to conduct a "soft selection" for conversation histories. This method attends to history turns with different weights based on how helpful they are on answering the current question. Third, in addition to handling conversation history, we take advantage of multi-task learning (MTL) to do answer prediction along with another essential conversation task (dialog act prediction) using a uniform model architecture. MTL is able to learn more expressive and generic representations to improve the performance of ConvQA. We demonstrate the effectiveness of our model with extensive experimental evaluations on QuAC, a large-scale ConvQA dataset. We show that position information plays an important role in conversation history modeling. We also visualize the history attention and provide new insights into conversation history understanding. | ['W. Bruce Croft', 'Cen Chen', 'Yongfeng Zhang', 'Liu Yang', 'Mohit Iyyer', 'Minghui Qiu', 'Chen Qu'] | 2019-08-26 | null | null | null | null | ['conversational-search'] | ['natural-language-processing'] | [ 9.32688266e-03 1.39766410e-01 4.00627851e-02 -6.67923093e-01
-7.49763191e-01 -6.20416343e-01 8.43819678e-01 -1.15590781e-01
-3.18485081e-01 4.79829550e-01 1.11346877e+00 -4.73725736e-01
-3.18014674e-04 -7.19920218e-01 -2.64061451e-01 -4.09250677e-01
3.65229920e-02 7.32437074e-01 2.82917351e-01 -7.80508339e-01
1.37598172e-01 2.85772439e-02 -1.22453403e+00 7.68588245e-01
7.77539611e-01 9.39009964e-01 3.46802294e-01 1.08805895e+00
-4.63166654e-01 1.50622737e+00 -9.19240952e-01 -6.43970370e-01
-1.96521819e-01 -8.26946318e-01 -1.59586024e+00 -2.53992267e-02
3.24612223e-02 -5.81124842e-01 -5.81923723e-01 2.64346957e-01
5.52085340e-01 5.67933142e-01 3.50322783e-01 -1.08589303e+00
-5.25395453e-01 1.09313285e+00 1.05677554e-02 3.86273175e-01
6.24908566e-01 4.01237309e-01 1.56008029e+00 -5.94145477e-01
4.92172062e-01 1.55983436e+00 3.79543215e-01 7.64644384e-01
-7.75329947e-01 -2.89841950e-01 5.34946084e-01 8.41601014e-01
-5.41184425e-01 -3.41864854e-01 1.13570511e+00 -2.36886859e-01
9.66128945e-01 5.98258793e-01 5.80784857e-01 1.34560752e+00
-3.37219983e-02 1.21342146e+00 6.46125972e-01 -2.07631692e-01
6.41133040e-02 -6.47854209e-02 7.09592104e-01 6.05941355e-01
-1.05009162e+00 -3.76605034e-01 -6.10369086e-01 -4.09821719e-01
1.39406905e-01 -6.00150134e-03 -5.04405499e-01 2.23261103e-01
-9.58767653e-01 1.23509574e+00 6.16142511e-01 4.10076618e-01
-1.97143361e-01 1.97504386e-01 6.39789581e-01 5.82711518e-01
2.87491679e-01 6.03525162e-01 -4.24338639e-01 -5.39583445e-01
-6.19902052e-02 2.83814400e-01 1.30344307e+00 6.86191022e-01
6.60243094e-01 -3.12554121e-01 -7.59137809e-01 1.14976740e+00
-2.61043981e-02 -1.86203122e-02 4.98434335e-01 -1.15498376e+00
4.99675065e-01 8.76743019e-01 -1.11491652e-02 -7.43694425e-01
-4.40776497e-01 -1.87128872e-01 -6.81930304e-01 -4.97967005e-01
3.10948342e-01 -1.12231508e-01 -5.14281631e-01 1.85307586e+00
2.20242307e-01 1.14377506e-01 1.40219420e-01 7.69144535e-01
1.16677415e+00 9.51665282e-01 1.13336304e-02 -1.25387743e-01
1.63983333e+00 -1.62723911e+00 -8.62542212e-01 -2.75900990e-01
4.77493912e-01 -4.55663174e-01 1.49173617e+00 1.40238881e-01
-9.44427729e-01 -5.28277457e-01 -7.46701717e-01 -5.59831381e-01
-1.26487300e-01 -3.61175425e-02 8.22636425e-01 2.48600245e-01
-9.23458457e-01 2.45060131e-01 -5.61198235e-01 -2.60570645e-01
-3.37429233e-02 2.20677540e-01 1.74399614e-01 3.07589746e-03
-1.57558918e+00 9.01994944e-01 9.78087857e-02 1.99317068e-01
-7.61340380e-01 -4.10672396e-01 -1.02191830e+00 4.02676225e-01
5.96288621e-01 -8.17018211e-01 1.87863314e+00 -6.32342160e-01
-1.94058931e+00 4.80296016e-01 -6.13564670e-01 -6.31706893e-01
2.24987537e-01 -3.61140706e-02 -1.94343403e-01 3.67868513e-01
-2.08588511e-01 3.99022162e-01 4.66603458e-01 -1.04165435e+00
-4.72961694e-01 -2.43732199e-01 8.21594357e-01 3.10809910e-01
-4.69232827e-01 3.14606838e-02 -7.09137261e-01 -3.29467803e-01
-2.86768198e-01 -8.78542364e-01 -1.87086567e-01 -3.70404303e-01
-1.44619852e-01 -1.03526616e+00 7.96144545e-01 -8.21123481e-01
1.65743816e+00 -1.80707669e+00 4.85208362e-01 -4.27601665e-01
4.58936930e-01 2.97204524e-01 -2.21460819e-01 1.01845205e+00
5.47665358e-01 -2.52381414e-02 -2.93115765e-01 -8.49956930e-01
1.64658412e-01 5.04269123e-01 -7.44960546e-01 -2.22220659e-01
2.45093286e-01 1.33601451e+00 -8.16946030e-01 -3.51647139e-01
-4.68143150e-02 2.05608845e-01 -6.36934102e-01 8.19696665e-01
-7.64241338e-01 6.52921736e-01 -6.32519662e-01 2.05867931e-01
2.25389719e-01 -6.29818499e-01 2.85699993e-01 -3.92228104e-02
1.63859174e-01 7.88901865e-01 -5.48250318e-01 1.76785541e+00
-8.63206267e-01 6.85165763e-01 8.34704190e-02 -7.89689779e-01
7.43407845e-01 4.20385808e-01 1.41498730e-01 -6.28117800e-01
2.47686833e-01 -1.96942642e-01 1.04571804e-01 -7.94654131e-01
6.52307391e-01 8.56237039e-02 -3.31521988e-01 9.76933300e-01
4.37810197e-02 -5.50117716e-02 6.40181527e-02 5.48009336e-01
1.31130505e+00 -3.07923913e-01 1.42707273e-01 2.26058841e-01
9.20777261e-01 -1.77301422e-01 4.27457511e-01 6.79076552e-01
-3.21251869e-01 4.41454202e-01 7.75433242e-01 -5.03234327e-01
-5.56964219e-01 -6.34497941e-01 3.10191959e-01 1.64517570e+00
1.85781419e-01 -6.41984403e-01 -4.84306127e-01 -9.46832478e-01
-1.61690876e-01 7.61703670e-01 -8.36960316e-01 -2.71667778e-01
-1.15195358e+00 -4.00194705e-01 3.40137571e-01 6.55127585e-01
6.55328393e-01 -1.47473383e+00 -3.53086352e-01 3.33621383e-01
-6.95796311e-01 -1.06969345e+00 -8.83527637e-01 -9.75626633e-02
-4.02233720e-01 -1.06475735e+00 -5.45610428e-01 -7.90939033e-01
-1.10849470e-01 2.75042564e-01 1.31057894e+00 3.84120256e-01
9.77391377e-02 6.89236701e-01 -6.43348336e-01 -1.39813006e-01
-5.45207858e-01 4.50153589e-01 -5.79016566e-01 1.71908230e-01
3.29802334e-01 -6.92567766e-01 -9.19992924e-01 4.41485554e-01
-6.71634257e-01 4.37792651e-02 1.49340808e-01 1.07782352e+00
-1.60804644e-01 -5.42727888e-01 9.32224214e-01 -1.00419140e+00
1.20309186e+00 -6.08332336e-01 -4.54887822e-02 5.33693373e-01
-2.36667365e-01 1.40399575e-01 7.23218799e-01 -5.18363476e-01
-1.26892650e+00 -4.55770254e-01 -5.42200327e-01 -1.40255645e-01
2.02136755e-01 5.16428530e-01 -1.18020929e-01 4.57013577e-01
5.33728123e-01 3.18706065e-01 5.70677370e-02 -5.85095823e-01
5.67512095e-01 8.54625225e-01 3.85090113e-01 -6.14326000e-01
2.51434654e-01 3.83892596e-01 -6.92850947e-01 -7.32975543e-01
-1.07149529e+00 -8.22034955e-01 -1.87564254e-01 -2.23447412e-01
1.01624346e+00 -5.43279111e-01 -1.47639549e+00 3.64574671e-01
-1.44883418e+00 -6.08860254e-01 3.05552804e-03 7.31187835e-02
-4.35494304e-01 5.80264270e-01 -1.03109324e+00 -1.07778502e+00
-8.15617979e-01 -1.07556403e+00 9.36666310e-01 2.08923087e-01
-3.01575392e-01 -1.11941338e+00 1.66745052e-01 9.48178232e-01
5.68227828e-01 -4.16533127e-02 1.05567896e+00 -8.65531027e-01
-6.19226396e-01 2.25760818e-01 -2.73646303e-02 1.79473713e-01
-6.51166514e-02 -4.50680643e-01 -1.10861564e+00 -1.63579732e-01
4.00059193e-01 -6.16580784e-01 1.11273909e+00 -1.40068635e-01
1.20183885e+00 -5.86562932e-01 -1.26894653e-01 2.32051343e-01
6.60734057e-01 3.07684481e-01 4.25188899e-01 1.29477782e-02
5.26916981e-01 9.23298001e-01 3.88207912e-01 4.18456465e-01
9.78074551e-01 8.35451365e-01 2.56196409e-01 4.40830916e-01
-2.67944306e-01 -2.60151595e-01 3.18109810e-01 1.25992632e+00
3.03501874e-01 -6.46565676e-01 -8.30824733e-01 5.45770288e-01
-2.20151234e+00 -1.02716255e+00 -5.16362786e-02 1.40710580e+00
1.06354129e+00 -1.03965169e-02 2.11120293e-01 -1.89635307e-01
4.13798302e-01 5.36089003e-01 -6.65022731e-01 -4.40935105e-01
9.13866162e-02 1.47845984e-01 -5.26791215e-01 8.60872507e-01
-8.10406268e-01 1.01430547e+00 5.47637606e+00 5.67734182e-01
-9.45381463e-01 2.72409290e-01 5.04879057e-01 9.49176252e-02
-6.16839826e-01 2.46169493e-02 -7.20251560e-01 4.42954987e-01
9.03870225e-01 -3.18895817e-01 4.90102619e-01 7.01512158e-01
-1.51226461e-01 8.29381049e-02 -1.29052043e+00 8.66817176e-01
2.24979371e-01 -1.60568821e+00 1.20194428e-01 -3.10648441e-01
3.43655676e-01 -2.34918788e-01 -3.30748886e-01 1.00368559e+00
4.29447800e-01 -9.35114145e-01 2.40229413e-01 5.00869036e-01
2.99400359e-01 -6.15831316e-01 6.96994543e-01 5.78556359e-01
-1.26158857e+00 -5.49908698e-01 -2.84900934e-01 -2.54992127e-01
6.02373898e-01 7.99234957e-02 -1.10493755e+00 5.18278182e-01
5.22970676e-01 5.15613675e-01 -4.13957596e-01 5.99551797e-01
-4.03394759e-01 7.54614055e-01 -1.10457223e-02 -4.35668170e-01
4.02316183e-01 -5.11905253e-02 4.38899785e-01 1.07594216e+00
-2.57282734e-01 4.52635914e-01 3.30974877e-01 7.10382462e-01
-2.83707798e-01 -1.07957743e-01 -3.14135969e-01 -9.97607559e-02
4.63699162e-01 1.11325192e+00 -7.31601119e-02 -2.63089001e-01
-3.38970602e-01 1.19068718e+00 6.91971242e-01 3.11798841e-01
-6.68792367e-01 -1.97650433e-01 6.66426599e-01 -3.00209075e-01
2.27692306e-01 -2.18358487e-01 1.69332147e-01 -1.21804619e+00
6.80306628e-02 -1.10000598e+00 7.75328577e-01 -6.07205868e-01
-1.48866963e+00 1.04880774e+00 -1.20404355e-01 -6.76914215e-01
-7.54348636e-01 -3.73127669e-01 -1.10373855e+00 9.27638471e-01
-1.45980680e+00 -1.40230501e+00 -4.66302156e-01 6.92475796e-01
1.20927572e+00 -8.41347054e-02 9.80056822e-01 1.45141616e-01
-5.10715246e-01 6.13549531e-01 -3.77279133e-01 3.05636257e-01
5.57168663e-01 -1.39261198e+00 6.12297177e-01 3.34909290e-01
1.12983398e-01 7.72768557e-01 4.24763054e-01 -2.30433166e-01
-1.41947997e+00 -6.90999866e-01 1.04313207e+00 -7.13175714e-01
7.27132797e-01 -6.84213042e-01 -1.12441111e+00 7.66792893e-01
7.31978178e-01 -7.79635906e-01 9.27023470e-01 5.26438534e-01
-2.43754268e-01 -1.02591194e-01 -5.36157429e-01 7.87920952e-01
1.03968227e+00 -9.53109086e-01 -1.06952906e+00 4.77006674e-01
1.59271586e+00 -2.25658774e-01 -5.91053605e-01 2.68516272e-01
3.40334266e-01 -8.41408312e-01 7.08122075e-01 -8.39767158e-01
4.50907350e-01 1.83852315e-01 1.09878965e-01 -1.14933395e+00
-7.45840669e-02 -1.12111354e+00 -5.28922975e-01 1.27385354e+00
3.21834415e-01 -5.08916438e-01 7.06269205e-01 6.00437522e-01
-3.60896558e-01 -1.12446404e+00 -9.06243801e-01 -2.85148352e-01
2.03662291e-02 -3.48834664e-01 7.48804748e-01 7.49774158e-01
2.61401534e-01 1.14345694e+00 -5.78933299e-01 -1.19149573e-01
-1.13528170e-01 7.04874396e-01 9.70219791e-01 -1.03556001e+00
-6.45354331e-01 -5.13955295e-01 2.08303541e-01 -1.87679899e+00
4.79915917e-01 -7.68936455e-01 -3.84872109e-02 -1.69017446e+00
1.40039757e-01 -1.55806884e-01 -6.31017908e-02 1.90721661e-01
-4.74109501e-01 -3.27334315e-01 2.05023810e-01 2.01477066e-01
-7.74827063e-01 1.11857867e+00 1.44503474e+00 -3.25584203e-01
-3.18456292e-01 3.78043801e-01 -7.23486304e-01 4.05465007e-01
7.34489679e-01 -5.15236259e-02 -7.78766036e-01 -6.16260469e-01
1.83197603e-01 6.16470516e-01 1.31900653e-01 -4.52767193e-01
5.63799381e-01 8.69215280e-03 -3.62080932e-01 -6.29499435e-01
8.68377328e-01 -3.95718336e-01 -5.41829348e-01 2.47221738e-01
-7.39091158e-01 1.41578674e-01 8.17231685e-02 7.76456773e-01
-4.07870620e-01 -2.42491677e-01 2.66256034e-01 -1.67654157e-01
-7.50647008e-01 2.73407549e-01 -3.75057936e-01 1.92366317e-01
7.01459289e-01 3.31446350e-01 -4.86248702e-01 -1.14486802e+00
-8.27046633e-01 1.04707825e+00 -1.56935662e-01 6.67760611e-01
5.75789094e-01 -1.05475318e+00 -6.96425438e-01 -1.55324191e-01
2.65006304e-01 -1.42993450e-01 4.79299158e-01 5.30124485e-01
-1.82993680e-01 4.31829751e-01 3.45397890e-01 -5.04792988e-01
-1.24172473e+00 4.30967540e-01 4.03911769e-01 -6.59253657e-01
-6.14336789e-01 1.08720958e+00 3.20031464e-01 -6.59814000e-01
6.93418324e-01 -3.01676333e-01 -6.99659824e-01 2.68235296e-01
7.50616670e-01 1.14680238e-01 -1.19796343e-01 -7.72237256e-02
-1.20697893e-01 2.02455372e-01 -4.06356633e-01 -2.51536161e-01
1.23830116e+00 -2.72747099e-01 -1.54660344e-01 6.03535712e-01
1.25961828e+00 2.62061432e-02 -1.21963656e+00 -5.72601557e-01
1.29011780e-01 -1.22690767e-01 -4.11802948e-01 -9.33145106e-01
-6.63768589e-01 1.12769091e+00 2.14687642e-02 5.16900957e-01
1.02581179e+00 4.10436094e-01 1.38686419e+00 8.10812652e-01
1.11570783e-01 -7.94182241e-01 7.09030092e-01 1.02588928e+00
1.45776784e+00 -1.25612402e+00 -5.17611086e-01 -2.59219825e-01
-1.00884759e+00 1.07180870e+00 8.01859260e-01 2.82986104e-01
4.63549286e-01 -2.41650537e-01 2.54132867e-01 -4.33167428e-01
-1.29096353e+00 -4.49248910e-01 2.59147912e-01 1.12129845e-01
2.98870504e-01 -2.78093249e-01 -4.20383483e-01 8.44964623e-01
-3.08325678e-01 -4.76438761e-01 3.22190791e-01 7.03089893e-01
-3.44191343e-01 -1.31993151e+00 2.70648807e-01 1.23426773e-01
-9.94993374e-02 -1.57222584e-01 -8.82291913e-01 4.71640170e-01
-4.01077390e-01 1.53951550e+00 9.98893194e-03 -6.89857423e-01
2.95625091e-01 5.33825576e-01 2.78783944e-02 -8.46147537e-01
-1.22159636e+00 -3.88691097e-01 4.60897952e-01 -4.93237317e-01
-1.19662233e-01 -1.74899802e-01 -1.15475237e+00 -3.84463787e-01
-1.87627152e-01 5.66530287e-01 2.91935414e-01 9.39756751e-01
4.95865524e-01 6.94478750e-01 9.06586170e-01 -2.77196616e-01
-9.12912786e-01 -1.35101819e+00 -2.56378241e-02 4.98838037e-01
4.29674327e-01 -3.31261456e-01 -4.42008585e-01 -2.88850307e-01] | [12.221796035766602, 7.874264717102051] |
f344f3d7-ef56-4e42-9e31-c0aad8f6aa49 | balf-simple-and-efficient-blur-aware-local | 2211.14731 | null | https://arxiv.org/abs/2211.14731v2 | https://arxiv.org/pdf/2211.14731v2.pdf | BALF: Simple and Efficient Blur Aware Local Feature Detector | Local feature detection is a key ingredient of many image processing and computer vision applications, such as visual odometry and localization. Most existing algorithms focus on feature detection from a sharp image. They would thus have degraded performance once the image is blurred, which could happen easily under low-lighting conditions. To address this issue, we propose a simple yet both efficient and effective keypoint detection method that is able to accurately localize the salient keypoints in a blurred image. Our method takes advantages of a novel multi-layer perceptron (MLP) based architecture that significantly improve the detection repeatability for a blurred image. The network is also light-weight and able to run in real-time, which enables its deployment for time-constrained applications. Extensive experimental results demonstrate that our detector is able to improve the detection repeatability with blurred images, while keeping comparable performance as existing state-of-the-art detectors for sharp images. | ['Peidong Liu', 'Ben M. Chen', 'Yu Zhai', 'Zhenjun Zhao'] | 2022-11-27 | null | null | null | null | ['keypoint-detection'] | ['computer-vision'] | [ 6.54740110e-02 -3.67655456e-01 7.13013932e-02 -5.43001816e-02
-2.86081702e-01 -2.82844692e-01 5.69256961e-01 1.10856451e-01
-7.44831085e-01 3.63034666e-01 -2.39361241e-01 -1.05069116e-01
-1.16389342e-01 -4.50726509e-01 -7.17392385e-01 -5.05502403e-01
-5.13870716e-02 -8.54808912e-02 7.26304591e-01 2.75969077e-02
3.82820010e-01 7.25570619e-01 -1.75515926e+00 -2.25628957e-01
6.82052374e-01 9.47830856e-01 8.53966236e-01 7.72799253e-01
3.45310837e-01 7.37659991e-01 -5.56038320e-01 1.58654347e-01
2.65384823e-01 1.27474695e-01 -4.82796162e-01 3.67450267e-02
7.54554272e-01 -5.79632759e-01 -3.34494859e-01 1.31742406e+00
3.59150946e-01 5.27945571e-02 2.53550231e-01 -9.03960943e-01
-6.59713686e-01 8.33508447e-02 -8.30096066e-01 4.31994081e-01
3.53740275e-01 7.71745071e-02 6.06065810e-01 -9.97541785e-01
3.84750783e-01 1.12074244e+00 6.77733660e-01 3.10474366e-01
-1.14425600e+00 -2.50624478e-01 -4.83211465e-02 4.68181640e-01
-1.25267863e+00 -4.41687614e-01 7.67566204e-01 -2.56689519e-01
7.68175364e-01 1.54935464e-01 3.78012598e-01 6.50420368e-01
5.46034396e-01 7.26706743e-01 1.02881050e+00 -4.25470471e-01
2.11933687e-01 4.37486507e-02 -1.60054296e-01 8.39775383e-01
5.33708155e-01 1.87954947e-01 -5.20594776e-01 1.66586399e-01
1.15040541e+00 3.41351062e-01 -6.15939856e-01 -5.92871249e-01
-1.40937614e+00 5.56261420e-01 1.04161370e+00 2.07010686e-01
-6.43728018e-01 3.69732499e-01 6.52036592e-02 2.01234251e-01
2.49114051e-01 5.87125182e-01 -2.76432127e-01 -1.71395913e-01
-9.90393400e-01 6.42808229e-02 4.89081234e-01 6.17482722e-01
7.44226933e-01 -1.99943781e-01 -2.46675126e-02 7.11793542e-01
3.17827165e-01 5.14358282e-01 5.98673284e-01 -7.56219685e-01
1.12929158e-01 5.87221265e-01 5.41019976e-01 -1.10544693e+00
-6.24693751e-01 -3.08070242e-01 -6.76052630e-01 7.67178595e-01
4.11479324e-01 -1.85551140e-02 -9.96372819e-01 1.23633862e+00
2.22923428e-01 3.01577717e-01 -1.06417790e-01 1.42131317e+00
5.99647880e-01 4.92869794e-01 -3.92795086e-01 -3.27745825e-02
1.57230353e+00 -9.84678864e-01 -4.26411867e-01 -7.63994455e-01
-1.61561817e-01 -9.59532082e-01 9.01311219e-01 2.78339356e-01
-8.25726926e-01 -7.24336088e-01 -1.20421934e+00 -1.11859187e-01
-3.40718865e-01 3.71394843e-01 5.77437639e-01 3.79237294e-01
-1.06942940e+00 4.81481880e-01 -9.40497577e-01 -4.59979862e-01
5.05814850e-02 3.75965059e-01 -4.28864062e-01 -1.24703879e-02
-8.54214191e-01 1.22100174e+00 3.75803262e-01 2.86120057e-01
-5.89813471e-01 -4.37761366e-01 -1.00382698e+00 2.01383457e-01
3.59175980e-01 -7.73645401e-01 1.33612216e+00 -6.31751537e-01
-1.68035412e+00 6.98608220e-01 -3.20044309e-01 -7.07452238e-01
6.10831738e-01 -6.83931589e-01 -1.83997825e-01 3.73355329e-01
-7.76275471e-02 5.86798847e-01 1.51981485e+00 -1.03289390e+00
-7.13275433e-01 -1.63080767e-01 1.87936574e-02 2.29043692e-01
-1.05123721e-01 7.60224164e-02 -4.12543803e-01 -3.60258132e-01
1.42835543e-01 -8.50095391e-01 -2.52455443e-01 2.04551131e-01
-5.07777371e-02 4.48009819e-02 1.02947748e+00 -3.60439837e-01
7.48955309e-01 -2.14584112e+00 -1.21711448e-01 -3.19244772e-01
2.51903653e-01 6.49355173e-01 1.98955655e-01 2.90581942e-01
1.62757486e-01 -5.77736676e-01 3.33180390e-02 -2.64414072e-01
-3.20305139e-01 -8.48422199e-02 -2.79554099e-01 7.40363002e-01
2.82830030e-01 7.57279038e-01 -9.39155757e-01 -9.74691436e-02
7.58771539e-01 4.98832405e-01 -2.34722406e-01 2.21386686e-01
6.71221912e-02 3.65516424e-01 -1.97327226e-01 5.11685848e-01
7.99644649e-01 -1.96360752e-01 -3.60900402e-01 -2.44596168e-01
-4.55814183e-01 2.01849174e-02 -1.34082651e+00 1.60767531e+00
-7.12285519e-01 9.86147523e-01 2.07466736e-01 -6.56005919e-01
1.03880918e+00 -5.27189523e-02 4.41827113e-03 -5.15381038e-01
6.41584992e-02 3.23583513e-01 -1.83117092e-01 -5.23993492e-01
9.56977248e-01 1.26940772e-01 1.67054459e-01 6.88471496e-02
-8.97437483e-02 -1.74568996e-01 2.72770151e-02 -1.77126139e-01
1.16617692e+00 1.84958447e-02 4.73080516e-01 -2.43944958e-01
6.84574485e-01 -3.75132471e-01 2.99984068e-01 9.53706801e-01
-4.24456954e-01 7.53668666e-01 -9.19495225e-02 -6.44707441e-01
-9.17130113e-01 -1.00833261e+00 -1.08305149e-01 7.58513153e-01
8.62996042e-01 -6.77059740e-02 -5.82789123e-01 -2.81843692e-01
1.40332028e-01 1.17091313e-01 -2.87756026e-01 -1.21403620e-01
-4.83953416e-01 -5.55983782e-01 7.73642510e-02 4.33830887e-01
7.11896300e-01 -1.00025773e+00 -1.45817137e+00 2.61762679e-01
1.32462204e-01 -1.25378549e+00 -3.41557026e-01 1.86141893e-01
-6.74789608e-01 -1.02796221e+00 -9.36110556e-01 -9.51075733e-01
7.47252762e-01 7.65546024e-01 6.67655349e-01 -3.11923832e-01
-5.20018935e-01 2.20959499e-01 -7.91348293e-02 -3.56727630e-01
-1.05658658e-01 -1.70567557e-01 3.46685916e-01 2.01287374e-01
3.71232301e-01 -3.03679645e-01 -9.60893214e-01 2.79039174e-01
-7.51557291e-01 -5.47251515e-02 9.95701432e-01 9.02440846e-01
1.28055975e-01 5.35089709e-02 2.17312813e-01 -2.19763815e-01
7.48673081e-01 3.30300517e-02 -1.09945989e+00 1.36183590e-01
-5.79127371e-01 2.88821816e-01 6.85518622e-01 -4.25060004e-01
-9.52741265e-01 3.53437752e-01 1.18197069e-01 -4.29058045e-01
-3.71623576e-01 2.85897940e-01 5.71558475e-01 -6.41438544e-01
7.87043452e-01 2.27042779e-01 4.16917019e-02 -6.27351582e-01
3.46319675e-01 9.00938153e-01 8.93933535e-01 1.00715145e-01
9.05780673e-01 6.47038519e-01 1.27781257e-01 -1.14120185e+00
-4.00897235e-01 -9.98487413e-01 -6.52474165e-01 -1.26673773e-01
4.00217474e-01 -1.10715139e+00 -1.00209248e+00 7.83512473e-01
-1.28443015e+00 3.78389098e-02 2.20616832e-01 7.15650499e-01
-3.95092815e-01 5.61228693e-01 -7.34163642e-01 -8.03143859e-01
-4.47213113e-01 -1.18510020e+00 1.15169728e+00 8.83112669e-01
1.21448681e-01 -8.30170631e-01 -1.14417709e-02 -5.37953340e-02
6.33992910e-01 1.22408003e-01 1.59388795e-01 1.40485451e-01
-8.60093772e-01 -3.44824344e-01 -5.97801089e-01 2.35285342e-01
2.43368059e-01 -9.30072740e-02 -1.11839831e+00 -4.47026461e-01
2.25348193e-02 -5.57726808e-02 1.13129675e+00 5.89270949e-01
6.29592180e-01 -1.38587147e-01 -4.09316927e-01 4.97607768e-01
1.58344042e+00 -8.98710936e-02 3.13394278e-01 5.38450897e-01
5.27193010e-01 2.48484358e-01 6.74320757e-01 1.74108252e-01
2.31375083e-01 9.26494062e-01 6.71353161e-01 -4.16037977e-01
-1.59021452e-01 -1.64934099e-02 4.38417077e-01 2.29125977e-01
1.82046294e-02 2.92683899e-01 -7.06952095e-01 8.76859963e-01
-2.10281968e+00 -8.51757765e-01 -3.43289860e-02 2.23221254e+00
6.70377374e-01 2.27983981e-01 -1.24203853e-01 -4.31354623e-03
7.27092206e-01 2.13752136e-01 -4.14541900e-01 -3.52862179e-01
2.35676318e-01 1.61778145e-02 8.70231092e-01 5.21898210e-01
-1.38873339e+00 8.74335170e-01 6.31920671e+00 3.13596874e-01
-1.60483062e+00 -8.78702700e-02 -6.55882731e-02 6.42955005e-02
4.89046216e-01 -2.20747963e-01 -6.79536581e-01 5.98364413e-01
6.55787408e-01 2.47578081e-02 3.84924650e-01 1.13087630e+00
4.21380103e-01 -6.03889406e-01 -8.63590240e-01 1.38605797e+00
1.32223263e-01 -1.21255529e+00 -5.14765918e-01 -2.08391666e-01
5.60179472e-01 4.36745226e-01 1.67114496e-01 -2.77320981e-01
-1.88743323e-02 -7.63953984e-01 6.31216228e-01 4.76249456e-01
5.45578599e-01 -7.16784894e-01 9.19382453e-01 5.18134296e-01
-1.03166091e+00 -2.49710321e-01 -6.55462921e-01 -4.75593418e-01
1.60408616e-01 9.14806485e-01 -1.27696025e+00 2.92660564e-01
9.04605150e-01 4.79508102e-01 -7.25831866e-01 1.67019486e+00
-5.64389288e-01 -1.22468360e-03 -4.27299708e-01 -1.86481923e-01
2.84055918e-01 7.32669830e-02 7.11575270e-01 1.18197823e+00
3.54240060e-01 -4.73772883e-01 2.29306370e-01 6.63916409e-01
8.41397494e-02 -4.13007766e-01 -7.57559061e-01 5.35079360e-01
3.46941859e-01 1.51498735e+00 -7.23883331e-01 -1.48101524e-01
-4.11065131e-01 1.32748342e+00 2.85096794e-01 2.04219326e-01
-5.85619628e-01 -8.29368711e-01 7.91542411e-01 -4.97841947e-02
6.39189899e-01 -5.84037721e-01 -3.75292003e-02 -1.25268853e+00
2.85862118e-01 -3.36735398e-01 -9.92348325e-03 -8.51543725e-01
-8.05625081e-01 6.11733794e-01 -1.94720551e-01 -1.25561368e+00
-2.78011620e-01 -9.89355922e-01 -5.12682676e-01 7.37780094e-01
-2.02765250e+00 -1.08896446e+00 -6.84828877e-01 3.63837183e-01
4.93515253e-01 1.78451046e-01 5.72061002e-01 1.33194536e-01
-3.63857806e-01 1.64870337e-01 2.82285839e-01 1.47481356e-02
9.91599858e-01 -1.38034308e+00 5.39135516e-01 1.50107777e+00
3.21755528e-01 7.23522961e-01 9.85171080e-01 -1.98326558e-01
-1.44434130e+00 -8.80474806e-01 6.68412387e-01 -3.03106964e-01
5.65715551e-01 -3.55833501e-01 -8.06272626e-01 3.21187705e-01
2.01348782e-01 3.62359613e-01 -1.54786944e-01 -1.09375753e-01
-1.31076440e-01 -4.40422088e-01 -9.87527966e-01 5.37375271e-01
6.04908168e-01 -6.49955451e-01 -9.10347342e-01 9.76615772e-02
4.48148519e-01 -7.52785265e-01 -4.60153580e-01 2.77947813e-01
5.30340791e-01 -1.24454761e+00 1.12963581e+00 2.34096915e-01
1.99539005e-03 -8.48868847e-01 4.42054957e-01 -1.47071207e+00
-5.90120614e-01 -8.23787451e-01 -3.64550799e-01 6.90869510e-01
2.41796244e-02 -8.68234396e-01 5.08742869e-01 1.81813538e-01
1.03876777e-01 -3.55498642e-01 -9.82755601e-01 -8.74399126e-01
-8.03009748e-01 -1.68851279e-02 1.43248245e-01 4.83353734e-01
1.63339257e-01 3.30073237e-01 -5.92176616e-01 6.20095789e-01
6.87127411e-01 4.59876925e-01 7.02233851e-01 -1.21084106e+00
-1.67722940e-01 -2.73165792e-01 -8.78834009e-01 -1.41996837e+00
-3.20841253e-01 -2.14237481e-01 4.32001382e-01 -1.33081782e+00
-4.44237925e-02 -1.29436061e-01 -3.52661610e-01 3.46015245e-01
-3.50336075e-01 6.22171998e-01 2.34083146e-01 2.77478397e-01
-5.65398753e-01 2.56429136e-01 8.76379848e-01 -1.81663558e-02
-3.13864350e-01 1.72768787e-01 -4.19245601e-01 8.53011191e-01
6.30977690e-01 -3.37341100e-01 -1.12938449e-01 -3.93933058e-01
-1.08524852e-01 -4.03694361e-01 7.17959583e-01 -1.32603085e+00
7.30972350e-01 -6.32546246e-02 5.03832757e-01 -4.90081638e-01
3.86837810e-01 -8.63562047e-01 -2.11569473e-01 8.28541160e-01
2.27068201e-01 9.38106850e-02 2.35722631e-01 6.28582001e-01
-3.34374607e-01 -3.05437654e-01 1.04500782e+00 -1.26759887e-01
-1.23115206e+00 -1.22137964e-01 -1.99237719e-01 -5.87538540e-01
1.16510403e+00 -1.94487259e-01 -6.02602959e-01 -4.00547713e-01
-6.64172992e-02 2.36259345e-02 7.90480077e-01 6.25064552e-01
9.29013193e-01 -9.67450976e-01 -5.01259565e-01 3.90369713e-01
1.92531079e-01 -4.05457430e-02 -2.25875694e-02 9.82398689e-01
-1.00452864e+00 6.61948979e-01 -3.81044447e-01 -8.64644110e-01
-1.27391922e+00 8.26936424e-01 3.06731164e-01 1.45386517e-01
-8.18869531e-01 7.89832234e-01 1.30625023e-02 1.46267608e-01
8.60467628e-02 -8.29876065e-01 -1.41402766e-01 -2.53281265e-01
9.01316106e-01 2.68389672e-01 1.51235834e-01 -4.64828789e-01
-5.75139284e-01 8.98973525e-01 -1.64318293e-01 1.10903293e-01
1.18181133e+00 -2.99784571e-01 -1.54674845e-02 2.61472851e-01
8.92232060e-01 2.16611817e-01 -1.59329653e+00 -3.85687858e-01
5.51569201e-02 -7.44804978e-01 2.89652258e-01 -5.19190371e-01
-5.54452240e-01 6.84074581e-01 7.46394992e-01 3.67164165e-01
1.13684082e+00 -8.62698779e-02 6.62990689e-01 4.79894549e-01
4.31431979e-01 -9.57731366e-01 6.63133338e-02 4.17875409e-01
7.49082863e-01 -1.44855750e+00 1.55640379e-01 -2.55703270e-01
-3.37122262e-01 1.41641414e+00 3.32930207e-01 -3.81126493e-01
2.79543996e-01 1.33347258e-01 2.63242781e-01 -4.73453961e-02
-4.39824104e-01 -5.29589474e-01 4.62612987e-01 5.87325633e-01
-4.97115552e-02 -1.04756840e-01 -1.73716545e-02 -3.41014713e-01
1.59447923e-01 -7.67188147e-02 7.74825573e-01 9.05239999e-01
-7.89116979e-01 -6.55903220e-01 -6.45775795e-01 -7.00153112e-02
-5.06340981e-01 2.19577141e-02 -1.16536289e-01 5.11364818e-01
-4.35212180e-02 8.81745338e-01 1.24992065e-01 -1.09549031e-01
3.24845940e-01 -2.87238777e-01 4.90197986e-01 -4.01035666e-01
-1.89793423e-01 -5.82074672e-02 -3.06397349e-01 -6.85462475e-01
-4.47664231e-01 -3.93289357e-01 -9.44574833e-01 -6.50280714e-02
-4.88486320e-01 -1.82402089e-01 8.81355226e-01 8.08490038e-01
5.76280415e-01 3.01561683e-01 5.05109966e-01 -1.23811615e+00
-5.16363442e-01 -9.42889273e-01 -4.59657967e-01 2.24039674e-01
1.14723611e+00 -7.01355934e-01 -1.94738925e-01 7.43548200e-02] | [8.363675117492676, -1.4841102361679077] |
d9ab7284-1260-49a9-b740-a2e5c64fef0a | self-supervised-learning-of-3d-objects-from | 1911.08850 | null | https://arxiv.org/abs/1911.08850v1 | https://arxiv.org/pdf/1911.08850v1.pdf | Self-supervised Learning of 3D Objects from Natural Images | We present a method to learn single-view reconstruction of the 3D shape, pose, and texture of objects from categorized natural images in a self-supervised manner. Since this is a severely ill-posed problem, carefully designing a training method and introducing constraints are essential. To avoid the difficulty of training all elements at the same time, we propose training category-specific base shapes with fixed pose distribution and simple textures first, and subsequently training poses and textures using the obtained shapes. Another difficulty is that shapes and backgrounds sometimes become excessively complicated to mistakenly reconstruct textures on object surfaces. To suppress it, we propose using strong regularization and constraints on object surfaces and background images. With these two techniques, we demonstrate that we can use natural image collections such as CIFAR-10 and PASCAL objects for training, which indicates the possibility to realize 3D object reconstruction on diverse object categories beyond synthetic datasets. | ['Tatsuya Harada', 'Hiroharu Kato'] | 2019-11-20 | null | null | null | null | ['3d-object-reconstruction'] | ['computer-vision'] | [ 2.68109679e-01 -1.61665305e-02 4.45194066e-01 -6.10210240e-01
-3.89988661e-01 -7.01850355e-01 4.73750114e-01 -4.34001118e-01
-4.75118756e-02 4.81255174e-01 -2.77343154e-01 3.53545733e-02
8.10533538e-02 -6.60419643e-01 -9.58234608e-01 -7.08510995e-01
2.65350431e-01 8.55574906e-01 3.61009777e-01 1.08437181e-01
2.06054121e-01 7.40610659e-01 -1.86479437e+00 3.32055181e-01
6.75983906e-01 9.07336652e-01 3.61129731e-01 3.99541616e-01
-2.88684458e-01 4.59031671e-01 -4.50801224e-01 -4.39845324e-01
4.66366529e-01 -2.04045981e-01 -5.49300134e-01 1.01662588e+00
8.35258245e-01 -3.30933839e-01 1.20591961e-01 1.13055301e+00
1.61208734e-01 3.88130844e-02 1.07917535e+00 -9.33629274e-01
-5.57875514e-01 -1.18718140e-01 -6.86945260e-01 -3.98425937e-01
1.95341840e-01 1.50448678e-03 4.95466381e-01 -1.15209270e+00
5.35348952e-01 1.31333053e+00 5.24748504e-01 7.75838315e-01
-1.50410986e+00 -3.65961432e-01 3.80002022e-01 -1.77711368e-01
-1.38160622e+00 -4.48054224e-01 1.03974509e+00 -5.41404784e-01
2.83466905e-01 3.92641842e-01 3.89835864e-01 1.26218438e+00
-2.17987984e-01 7.44408906e-01 1.31260395e+00 -4.12510425e-01
3.05172563e-01 5.04676163e-01 -9.78855491e-02 7.00963914e-01
3.12713325e-01 -1.77390680e-01 -1.69460535e-01 -1.19256645e-01
1.09072828e+00 -1.28194001e-02 -2.50992775e-01 -1.00468850e+00
-1.26973546e+00 3.27543169e-01 1.13594681e-01 -1.24505334e-01
-1.06977828e-01 -1.29745662e-01 4.38445024e-02 8.95331129e-02
4.88163322e-01 3.01806867e-01 -5.12091696e-01 3.23499501e-01
-4.94358510e-01 1.81904599e-01 6.86625242e-01 1.13745248e+00
8.99287224e-01 1.04422331e-01 2.03098506e-01 9.34310317e-01
2.01423272e-01 8.06685150e-01 -7.53645077e-02 -9.47250724e-01
2.87584066e-01 5.20202279e-01 4.44739789e-01 -9.76943493e-01
-1.32813558e-01 -3.49266142e-01 -8.03752005e-01 4.98928696e-01
7.91312397e-01 2.59108543e-01 -1.28059113e+00 1.45175076e+00
5.71556330e-01 -5.14634214e-02 -1.16771750e-01 9.73768294e-01
7.59118915e-01 4.25691068e-01 -2.59714454e-01 -1.49380147e-01
1.08699107e+00 -7.23920524e-01 -4.13389385e-01 -2.15647295e-01
2.30348751e-01 -8.07734907e-01 1.45612037e+00 6.33706570e-01
-9.89723980e-01 -6.45875871e-01 -7.00455546e-01 1.20646864e-01
-7.49714896e-02 3.88325840e-01 6.67634666e-01 5.88937938e-01
-4.48436320e-01 4.77167308e-01 -8.14707816e-01 -8.18116218e-02
4.06421542e-01 1.30157247e-01 -5.19135952e-01 -2.28987008e-01
-3.44652623e-01 6.29291058e-01 1.79892242e-01 1.43264621e-01
-9.22983348e-01 -5.76616347e-01 -7.63074756e-01 -1.34617627e-01
4.42596763e-01 -5.12445748e-01 8.38148236e-01 -1.09230173e+00
-1.45855427e+00 1.11253989e+00 5.78689529e-03 2.33150750e-01
6.74943686e-01 -8.95559490e-02 -1.16704248e-01 -1.57332048e-02
4.78664588e-04 5.41793704e-01 1.31177533e+00 -1.89456952e+00
-2.19049305e-01 -6.08439505e-01 -5.26513485e-03 2.75523961e-01
-2.10579038e-01 -3.86393040e-01 -5.64910591e-01 -5.80879569e-01
6.68006837e-01 -8.59836698e-01 -2.37796023e-01 3.65121782e-01
-4.10546362e-01 9.84484702e-02 8.79236519e-01 -4.71469253e-01
2.91150302e-01 -2.39767933e+00 2.21264228e-01 2.27005526e-01
-2.92916372e-02 5.25495410e-02 -2.43970752e-01 -1.92259610e-01
3.52174863e-02 -2.41411570e-02 -1.86768129e-01 -5.70058107e-01
-2.05120295e-01 3.85826111e-01 -3.04998666e-01 4.39917088e-01
2.06404626e-01 5.00364423e-01 -7.06631839e-01 -2.92537719e-01
2.69650608e-01 4.43714678e-01 -7.36610830e-01 4.55800831e-01
-6.08430922e-01 8.57892036e-01 -5.85795820e-01 6.27958477e-01
9.34394062e-01 -4.50848222e-01 1.19356491e-01 -3.75222355e-01
2.20209464e-01 -5.20674232e-03 -1.42462361e+00 1.43741286e+00
-4.59558845e-01 2.07196325e-01 2.71159440e-01 -1.17897844e+00
1.06037652e+00 7.53227770e-02 3.33109587e-01 -3.65080625e-01
1.64366364e-02 1.29963666e-01 -4.09564435e-01 -6.15068376e-01
-4.03598649e-03 -2.68506289e-01 3.47605795e-01 3.49179894e-01
-6.08878285e-02 -6.39570296e-01 -2.32602656e-01 -1.86163977e-01
5.05639911e-01 4.04858142e-01 -2.27586344e-01 -2.90224016e-01
5.03621697e-01 -1.34036571e-01 4.19041634e-01 4.25681740e-01
3.15143138e-01 1.04812980e+00 2.52161413e-01 -6.32058442e-01
-1.17267203e+00 -1.28605044e+00 -3.11900884e-01 6.71189189e-01
2.67942011e-01 9.33335871e-02 -6.27953708e-01 -8.67930710e-01
-2.98197251e-02 5.23350000e-01 -5.32580376e-01 1.72699556e-01
-4.87756073e-01 -6.55250311e-01 2.44802590e-02 3.36618245e-01
5.16288400e-01 -8.62551987e-01 -2.13255912e-01 -7.10415766e-02
-1.63143426e-01 -1.23564231e+00 -4.34768766e-01 1.33026555e-01
-9.34934914e-01 -1.12681413e+00 -7.35660434e-01 -9.49929357e-01
1.37066376e+00 5.79713702e-01 1.07052827e+00 2.10755430e-02
-1.78215250e-01 4.71292317e-01 -1.40886500e-01 -3.10079634e-01
-4.44187880e-01 -4.92347836e-01 1.33560836e-01 3.58211637e-01
-1.44127235e-01 -7.97419965e-01 -3.68794739e-01 7.42485106e-01
-7.80904472e-01 3.80199015e-01 2.94182837e-01 7.87551284e-01
9.32306468e-01 1.60309926e-01 3.55186224e-01 -1.01313639e+00
-1.16512358e-01 7.34606385e-02 -8.43717277e-01 4.21269983e-01
-1.49188980e-01 -5.09978607e-02 6.60984159e-01 -8.01761985e-01
-1.23999321e+00 5.53072989e-01 8.57530832e-02 -8.43359470e-01
-6.04115963e-01 3.77266365e-03 -4.96210694e-01 -1.75328702e-01
8.54864240e-01 3.86710018e-01 2.21666351e-01 -7.16646612e-01
2.55555689e-01 4.37216878e-01 4.73296374e-01 -8.91785502e-01
9.73718405e-01 7.64674366e-01 -1.19470805e-02 -9.75505769e-01
-1.04743958e+00 -2.87829101e-01 -7.39876211e-01 -2.48432249e-01
7.37760186e-01 -9.65550601e-01 -5.80070019e-01 5.41213155e-01
-1.14626384e+00 -3.30924958e-01 -1.17909297e-01 3.31955940e-01
-7.32054472e-01 4.55468535e-01 -2.88290262e-01 -6.96050763e-01
1.71380714e-01 -1.10782444e+00 1.25532174e+00 -1.13461643e-01
1.92226306e-01 -8.92051876e-01 -3.71023208e-01 6.80125952e-01
1.93035692e-01 2.52553970e-01 9.93816137e-01 -1.50947332e-01
-9.20021057e-01 -3.04516479e-02 -2.60644734e-01 5.34006894e-01
4.81219620e-01 3.88101046e-03 -1.16785610e+00 -3.19267660e-01
3.48078579e-01 -5.57555795e-01 5.21334052e-01 2.21370682e-01
1.55373740e+00 -1.16351031e-01 -7.84340501e-02 8.10844898e-01
1.33036602e+00 9.26578939e-02 4.85867679e-01 -1.40290171e-01
8.29180598e-01 8.53623450e-01 4.75972772e-01 2.53681570e-01
1.80644929e-01 7.06471384e-01 3.83852601e-01 -7.93170929e-02
-1.23646997e-01 -3.15295070e-01 7.28854612e-02 6.14680171e-01
-2.52064347e-01 -1.12595581e-01 -6.89955294e-01 2.70918906e-01
-1.41512024e+00 -7.49195337e-01 -6.78411499e-02 2.37624979e+00
7.65222847e-01 2.01662511e-01 -2.84976494e-02 4.46741767e-02
5.44570923e-01 -1.47702590e-01 -5.30118823e-01 2.32631981e-01
-7.49155432e-02 -1.52118176e-01 1.47216782e-01 3.39630008e-01
-1.00999629e+00 8.93758774e-01 6.27052546e+00 7.69672453e-01
-1.31236315e+00 -2.08484605e-01 7.98169672e-01 1.70374036e-01
-6.15565538e-01 -1.04943693e-01 -7.95969188e-01 1.37645364e-01
5.64298742e-02 4.22132969e-01 7.24918067e-01 8.75268638e-01
-9.02700946e-02 5.38133644e-03 -1.14214814e+00 1.20137107e+00
1.38755187e-01 -1.08702934e+00 2.45573029e-01 -1.63499802e-01
7.26136923e-01 -3.83691460e-01 9.07984525e-02 6.12988845e-02
4.52093147e-02 -8.43565047e-01 9.48350251e-01 3.66291553e-01
9.30549264e-01 -3.58113557e-01 1.76963359e-01 7.09969342e-01
-7.82052159e-01 3.06629986e-01 -6.41545296e-01 4.68928367e-02
-4.13352065e-02 7.25590110e-01 -7.13980436e-01 3.33677024e-01
6.80875182e-01 6.38607621e-01 -5.59861958e-01 9.05032575e-01
-1.33580357e-01 4.36764717e-01 -4.19460624e-01 1.44581422e-01
-2.56742328e-01 -4.21094447e-01 5.17116845e-01 6.83560073e-01
2.17009410e-01 2.45147586e-01 2.45695531e-01 1.16928685e+00
2.87715495e-01 -3.99694405e-02 -8.01924467e-01 1.26972079e-01
1.06261596e-01 1.23266327e+00 -9.73262429e-01 -1.50116056e-01
-5.04961014e-01 9.57629621e-01 4.08423543e-01 5.49077034e-01
-7.72945762e-01 3.16259325e-01 4.01010245e-01 3.89250070e-01
4.09348458e-01 -4.87069339e-01 -4.20620501e-01 -1.59933996e+00
2.53790438e-01 -9.03015077e-01 -1.56025589e-01 -8.98920894e-01
-1.58371425e+00 5.34616649e-01 7.73698688e-02 -1.60566890e+00
1.83134526e-01 -8.57125401e-01 -3.09830815e-01 6.06318712e-01
-1.08979142e+00 -1.02151358e+00 -3.57107639e-01 7.31227577e-01
6.29654586e-01 -1.17929719e-01 8.24680686e-01 2.15086102e-01
-2.28869930e-01 2.91741520e-01 -4.69902791e-02 -1.31073847e-01
5.36561430e-01 -1.03054070e+00 2.04215109e-01 2.98461020e-01
4.04961050e-01 4.92424399e-01 6.19664192e-01 -4.56896842e-01
-1.50741112e+00 -1.00549591e+00 2.27835983e-01 -6.69728875e-01
1.32812127e-01 -8.63587797e-01 -1.02218127e+00 4.95297819e-01
-2.93163687e-01 3.16263795e-01 2.99060613e-01 9.52940956e-02
-4.03158069e-01 -5.68786785e-02 -1.07003510e+00 6.69257939e-01
1.30580652e+00 -4.10793841e-01 -3.67634535e-01 6.88895404e-01
4.10344273e-01 -6.60537243e-01 -5.07330358e-01 7.05435634e-01
4.30675954e-01 -1.07787192e+00 1.12634909e+00 -5.81441879e-01
1.24484934e-01 -5.21874130e-01 -3.99731636e-01 -1.14501905e+00
-1.27785370e-01 -2.86410898e-01 3.00168633e-01 1.12776649e+00
2.93466032e-01 -3.46171886e-01 1.09590757e+00 6.44114077e-01
-4.42100987e-02 -5.84225237e-01 -6.20958686e-01 -8.49638879e-01
-9.51702744e-02 -3.05671066e-01 3.26664656e-01 9.80644166e-01
-6.72037840e-01 3.48879188e-01 -3.83859575e-01 4.02628213e-01
8.98844898e-01 5.24289250e-01 1.17525089e+00 -1.18362057e+00
-4.03554887e-01 -2.83886939e-02 -1.31724015e-01 -1.18520451e+00
1.40706211e-01 -5.61906159e-01 2.01768517e-01 -9.90678549e-01
2.27849185e-01 -9.49712873e-01 3.16519022e-01 2.65312284e-01
-3.09360167e-03 2.37799197e-01 -9.18273628e-02 2.07936630e-01
-3.87629241e-01 7.10370183e-01 1.77373552e+00 -1.54153526e-01
-4.38227430e-02 4.78211164e-01 -3.48885328e-01 1.05552745e+00
4.99253422e-01 -3.29082996e-01 -7.02534854e-01 -6.81471527e-01
2.38890443e-02 1.13752052e-01 6.35628343e-01 -7.46012866e-01
-2.15730727e-01 -3.95462036e-01 6.51332557e-01 -6.82774782e-01
5.43195903e-01 -1.20864713e+00 3.96882445e-01 -3.01904995e-02
-1.38980448e-01 -3.38937581e-01 1.29395306e-01 6.99086905e-01
-5.58760092e-02 -2.10841924e-01 9.62842226e-01 -4.19714987e-01
-4.30370539e-01 3.82122278e-01 -3.78494114e-02 -6.38759658e-02
8.67684186e-01 -3.53009850e-01 3.31887417e-02 -2.66986310e-01
-1.00214577e+00 3.26392166e-02 8.51530910e-01 2.86438167e-01
7.26997972e-01 -1.18771982e+00 -4.45594996e-01 8.88092220e-01
1.07797772e-01 5.36981940e-01 2.38580137e-01 3.58809918e-01
-5.21045744e-01 -8.90021026e-02 -2.80179411e-01 -9.62418675e-01
-1.15648222e+00 7.69222915e-01 4.06982750e-01 2.37872511e-01
-7.31355786e-01 8.06191266e-01 7.50275314e-01 -7.64202058e-01
5.10931969e-01 -2.86105990e-01 -1.17798178e-02 -3.93427670e-01
3.25176060e-01 -1.14784069e-01 4.80851717e-02 -4.07367736e-01
-1.68456420e-01 8.28016996e-01 3.84501666e-02 2.51282379e-02
1.40527070e+00 -2.09743842e-01 3.27830836e-02 3.02998930e-01
8.97050381e-01 2.26349294e-01 -1.62584102e+00 -2.60800153e-01
-2.35663027e-01 -7.82328010e-01 -3.28877658e-01 -4.89937216e-01
-1.11724102e+00 1.02652895e+00 4.07593578e-01 1.43927053e-01
9.55055237e-01 2.04928759e-02 3.10700625e-01 6.19125545e-01
7.99098253e-01 -8.02370191e-01 5.13567269e-01 5.09817779e-01
1.12109399e+00 -1.26117301e+00 -1.26468778e-01 -9.19405103e-01
-6.53812706e-01 1.11040008e+00 8.15756738e-01 -1.61845028e-01
7.56313086e-01 2.49785870e-01 7.29058161e-02 -1.73788249e-01
-5.34832478e-01 -2.53855661e-02 4.79466528e-01 6.55934632e-01
1.36196256e-01 -9.68510211e-02 2.28577241e-01 3.84248704e-01
8.08455795e-02 -3.60763967e-01 3.83293986e-01 6.31457865e-01
-3.52966338e-01 -1.10891974e+00 -4.74057704e-01 3.65005404e-01
-2.69697160e-01 1.97069839e-01 -3.17110002e-01 6.03997588e-01
9.18793082e-02 4.52488989e-01 -1.93179213e-02 -3.61806065e-01
4.60301161e-01 -6.25147000e-02 9.44854736e-01 -8.40717077e-01
7.98280388e-02 3.66204977e-01 1.00320086e-01 -1.94674268e-01
-5.34125209e-01 -5.42234480e-01 -9.49425519e-01 2.53543854e-02
-4.34048831e-01 -1.78808361e-01 5.15866816e-01 8.08982909e-01
1.43014729e-01 1.14657484e-01 7.70587564e-01 -1.18568385e+00
-7.74868309e-01 -7.25411057e-01 -7.83514500e-01 7.94151664e-01
1.72957793e-01 -1.07700455e+00 -5.33458829e-01 4.02374595e-01] | [8.536688804626465, -3.0149855613708496] |
6b3586fa-c707-4e54-812e-3f4126f217b9 | an-end-to-end-neural-network-framework-for | 1903.09424 | null | http://arxiv.org/abs/1903.09424v1 | http://arxiv.org/pdf/1903.09424v1.pdf | An end-to-end Neural Network Framework for Text Clustering | The unsupervised text clustering is one of the major tasks in natural
language processing (NLP) and remains a difficult and complex problem.
Conventional \mbox{methods} generally treat this task using separated steps,
including text representation learning and clustering the representations. As
an improvement, neural methods have also been introduced for continuous
representation learning to address the sparsity problem. However, the
multi-step process still deviates from the unified optimization target.
Especially the second step of cluster is generally performed with conventional
methods such as k-Means. We propose a pure neural framework for text clustering
in an end-to-end manner. It jointly learns the text representation and the
clustering model. Our model works well when the context can be obtained, which
is nearly always the case in the field of NLP. We have our method
\mbox{evaluated} on two widely used benchmarks: IMDB movie reviews for
sentiment classification and $20$-Newsgroup for topic categorization. Despite
its simplicity, experiments show the model outperforms previous clustering
methods by a large margin. Furthermore, the model is also verified on English
wiki dataset as a large corpus. | ['Jinchao Zhang', 'Xingyi Cheng', 'Jie Zhou'] | 2019-03-22 | null | null | null | null | ['text-clustering'] | ['natural-language-processing'] | [-1.56193338e-02 1.80712715e-02 -2.35746101e-01 -8.02063167e-01
-8.77599180e-01 -3.15991431e-01 5.55384040e-01 4.09214377e-01
-7.73113191e-01 2.96343565e-01 3.53894383e-01 -9.59556699e-02
-9.57154259e-02 -5.36238194e-01 -5.06998301e-01 -8.46526146e-01
3.14341813e-01 6.46260262e-01 -1.32627770e-01 -4.80354168e-02
4.18937504e-01 -1.51747346e-01 -1.57157326e+00 3.58905375e-01
8.87869358e-01 1.04277527e+00 4.06845838e-01 2.29376003e-01
-5.96707642e-01 7.93611169e-01 -5.63472390e-01 -3.54366869e-01
1.43392578e-01 -2.28475690e-01 -8.41675103e-01 2.59359211e-01
4.88540903e-02 2.64827847e-01 3.13491863e-03 1.18724513e+00
3.83545876e-01 5.77055812e-01 1.02465272e+00 -1.10740328e+00
-6.69974864e-01 1.07998002e+00 -9.91399825e-01 -2.60460168e-01
-1.39749527e-01 -6.62624657e-01 1.30695903e+00 -9.63120759e-01
3.55549306e-01 1.30144513e+00 3.63244146e-01 4.62114453e-01
-7.73173213e-01 -5.70804358e-01 7.49950111e-01 8.26915279e-02
-1.40551507e+00 -1.33660629e-01 7.70127833e-01 -5.55964172e-01
7.09209144e-01 2.57928818e-02 -1.88201871e-02 7.98500478e-01
-2.52583534e-01 1.15810108e+00 7.04424620e-01 -5.60060203e-01
3.21365505e-01 4.19773787e-01 8.47473621e-01 2.87541062e-01
1.91178426e-01 -8.13971877e-01 -1.77475020e-01 1.36069119e-01
8.86593014e-02 2.98047304e-01 -3.79677504e-01 -1.97436005e-01
-1.21299195e+00 1.08818376e+00 2.69260079e-01 3.69121969e-01
-2.00743750e-01 3.65678668e-02 6.78430319e-01 8.00886452e-02
6.84209704e-01 3.72876674e-01 -6.15072429e-01 4.04331647e-03
-9.93426502e-01 4.81728166e-02 9.52246845e-01 1.02467072e+00
6.13919437e-01 -2.35828571e-02 5.16652800e-02 1.26949298e+00
4.74283904e-01 8.89217705e-02 9.81305897e-01 -5.97593963e-01
6.99563026e-01 8.05941939e-01 -6.27791928e-03 -1.20883310e+00
-5.36903560e-01 -4.87484097e-01 -1.20978928e+00 -8.80448595e-02
1.71835437e-01 -3.50774139e-01 -7.62744069e-01 1.55507541e+00
1.84821531e-01 -1.61670402e-01 1.75158381e-01 8.05341184e-01
1.00991273e+00 1.11582696e+00 5.52768596e-02 -5.22782326e-01
1.26846516e+00 -1.35358393e+00 -9.09817994e-01 -1.96687639e-01
7.07803309e-01 -6.98220551e-01 1.18355167e+00 7.78953791e-01
-6.64419949e-01 -4.50777411e-01 -9.61053073e-01 -1.58219427e-01
-7.27406263e-01 5.08096576e-01 6.84824526e-01 6.72433436e-01
-7.86393821e-01 3.54350418e-01 -6.55563056e-01 -4.18296218e-01
2.91659236e-01 4.45370793e-01 -1.64899543e-01 -1.66743562e-01
-9.97718334e-01 4.18158382e-01 5.16072452e-01 2.73491621e-01
-4.20990855e-01 -3.25132161e-01 -8.00672233e-01 1.48628637e-01
5.08302629e-01 -2.38305509e-01 1.04930317e+00 -1.18529308e+00
-1.66799688e+00 8.13973665e-01 -2.58069903e-01 -3.88182908e-01
2.11402178e-01 -4.44627732e-01 -2.99338698e-01 -3.73526812e-02
9.43501666e-02 5.44835985e-01 7.32751667e-01 -1.46439326e+00
-6.12064421e-01 -4.18239981e-01 -1.08544625e-01 5.14538527e-01
-9.51845467e-01 2.36983463e-01 -8.94623816e-01 -8.01965117e-01
1.28134757e-01 -7.50640869e-01 -5.26536226e-01 -3.77740115e-01
-5.04775286e-01 -6.24462903e-01 6.30591393e-01 -3.48823100e-01
1.40055478e+00 -2.20590591e+00 2.30634883e-01 1.53822973e-01
1.21088840e-01 4.93027531e-02 2.05354705e-01 3.74578804e-01
-1.69708371e-01 2.70606309e-01 -3.14302027e-01 -9.23406124e-01
2.47233555e-01 -7.23299682e-02 -4.23641652e-01 4.68517274e-01
-2.14573085e-01 5.19788325e-01 -6.09351695e-01 -6.04668021e-01
-1.05469070e-01 3.87742847e-01 -5.23341537e-01 1.05080292e-01
-3.86443228e-01 9.30753648e-02 -3.81038100e-01 3.31028998e-01
4.42396164e-01 -5.84523857e-01 2.43978426e-01 -2.27005705e-02
-8.24123174e-02 1.11317590e-01 -1.45683098e+00 1.89239335e+00
-2.60736138e-01 5.44321120e-01 1.47397384e-01 -1.66027296e+00
9.33366835e-01 3.47940713e-01 6.34702682e-01 -1.89254060e-01
3.23534489e-01 -9.52706039e-02 -1.34735823e-01 -4.73271251e-01
8.29251885e-01 -4.65633385e-02 -2.82740146e-01 6.27202749e-01
4.59218584e-02 6.24053478e-02 3.51272136e-01 3.98895532e-01
5.85502803e-01 -4.20694277e-02 2.55821526e-01 -5.34196436e-01
5.29722452e-01 2.17423096e-01 5.85482001e-01 5.99460244e-01
1.13777250e-01 9.69163656e-01 5.78525245e-01 -1.77327797e-01
-5.49362898e-01 -3.92326385e-01 -1.91095993e-01 1.46871531e+00
1.74258530e-01 -8.18032146e-01 -7.29262948e-01 -8.41483772e-01
-2.89215446e-01 5.12436450e-01 -7.36986339e-01 1.39092198e-02
-2.84590453e-01 -1.26487327e+00 1.40894547e-01 4.58832324e-01
3.37124944e-01 -9.56832290e-01 7.12796077e-02 1.53196365e-01
-3.32455248e-01 -1.13030255e+00 -4.35205668e-01 5.37468612e-01
-6.16800964e-01 -7.83039272e-01 -6.40605509e-01 -1.24390447e+00
8.68710101e-01 4.70520020e-01 9.43141997e-01 1.34722877e-03
-9.14391503e-02 2.97590554e-01 -7.48217762e-01 -4.82842833e-01
5.52679822e-02 3.25785697e-01 5.64099699e-02 2.99562126e-01
5.55580378e-01 -2.40262911e-01 -6.58350766e-01 3.77912998e-01
-1.07047975e+00 -1.49814650e-01 4.06377584e-01 8.49223554e-01
6.11627996e-01 6.22974277e-01 8.66335034e-01 -1.28805768e+00
8.31813037e-01 -6.85983777e-01 -3.29885751e-01 2.19459027e-01
-6.84214771e-01 -1.71165571e-01 1.00384545e+00 -3.71808499e-01
-1.15471816e+00 1.98775098e-01 -2.56884359e-02 -2.71605879e-01
-2.50641078e-01 1.02956808e+00 -3.48066896e-01 7.01233327e-01
5.55537045e-01 1.77080959e-01 -8.68711621e-02 -6.28476143e-01
3.55273813e-01 1.06840312e+00 1.97861195e-01 -7.15771675e-01
4.57649052e-01 5.63231409e-01 -5.80539942e-01 -9.44549263e-01
-1.03462434e+00 -9.99180555e-01 -6.29390776e-01 2.62281168e-02
9.39875841e-01 -1.15071607e+00 -6.73568726e-01 3.19881499e-01
-9.91931379e-01 -1.23481527e-01 -6.36353344e-02 6.53717220e-01
-2.35988006e-01 6.55148566e-01 -5.57934165e-01 -7.45336771e-01
-4.68675584e-01 -1.13721049e+00 9.57270682e-01 1.61526706e-02
5.42749930e-03 -9.81208980e-01 -8.91942903e-02 4.28526163e-01
1.14770278e-01 -7.44021386e-02 1.06795561e+00 -1.02824152e+00
-1.30825732e-02 -2.49615029e-01 -1.63051724e-01 6.33870244e-01
6.54735863e-02 8.43803491e-03 -9.22452033e-01 -3.07759702e-01
2.57350296e-01 -5.69917500e-01 1.23066580e+00 4.27950293e-01
1.60012722e+00 -2.03759179e-01 -3.12063038e-01 3.67223620e-01
1.51691580e+00 1.54236719e-01 5.09126723e-01 2.12820947e-01
7.84453690e-01 1.02625060e+00 7.11922050e-01 4.29431409e-01
4.55495030e-01 3.34522277e-01 2.29610860e-01 2.91934684e-02
5.22804976e-01 1.91184714e-01 4.52727050e-01 1.38278890e+00
1.63142234e-01 -5.81263185e-01 -1.00610924e+00 4.41200614e-01
-2.31045341e+00 -7.45643377e-01 -8.06390345e-02 1.96366858e+00
7.87209809e-01 1.05322018e-01 5.62685877e-02 2.76242793e-01
9.15426314e-01 1.17759481e-01 -3.16976666e-01 -1.36407927e-01
9.64888483e-02 -1.36704683e-01 1.27916336e-01 2.89566010e-01
-1.42030036e+00 1.08698106e+00 5.07960081e+00 1.01613319e+00
-1.03114927e+00 -5.72286826e-03 8.19130540e-01 -7.95454979e-02
-4.76945117e-02 -3.11223090e-01 -1.03049004e+00 4.89542633e-01
5.94274819e-01 -4.45617549e-02 7.93381929e-02 1.08352411e+00
1.01651445e-01 -3.74013628e-03 -9.07476306e-01 1.13293433e+00
4.13964927e-01 -1.10333347e+00 1.58899456e-01 -1.20400399e-01
7.42500186e-01 -9.10591409e-02 -3.96072585e-03 6.79806888e-01
4.11443710e-01 -1.06768060e+00 3.60332817e-01 7.90961981e-02
4.85245913e-01 -8.77982616e-01 7.21656919e-01 5.25232852e-01
-1.09906197e+00 -1.29112676e-01 -6.43842399e-01 1.42449141e-01
2.14688219e-02 8.49345982e-01 -3.90765399e-01 5.48121333e-01
9.23226357e-01 1.10550308e+00 -4.03229505e-01 8.37912917e-01
7.64008053e-03 8.59993756e-01 -2.71805823e-01 -3.69214833e-01
4.20495450e-01 -4.33690667e-01 1.48514450e-01 1.45724130e+00
2.14092806e-02 -8.15464333e-02 4.75431681e-01 4.62322414e-01
-3.45816582e-01 7.29813874e-01 -3.35037649e-01 -1.86023578e-01
1.00928135e-01 1.53629291e+00 -1.09688532e+00 -3.76734555e-01
-4.11557496e-01 6.93280458e-01 5.14949918e-01 4.99134630e-01
-6.89142108e-01 -6.69386685e-01 2.28158742e-01 -3.07803214e-01
3.52257907e-01 -2.18689337e-01 -2.82272846e-01 -1.46003890e+00
1.31400660e-01 -8.14482272e-01 5.18437266e-01 -3.48788232e-01
-1.52651072e+00 7.36001551e-01 -5.25400564e-02 -1.37136960e+00
-3.53266001e-02 -7.43201196e-01 -4.55310464e-01 3.47447395e-01
-1.59545445e+00 -9.24607754e-01 -2.04630539e-01 8.03449750e-01
9.86019373e-01 -2.91300356e-01 7.78029680e-01 4.78395432e-01
-1.01669502e+00 4.75849450e-01 6.49449110e-01 4.07462984e-01
8.89328122e-01 -1.48494637e+00 -1.89020857e-01 7.03589559e-01
1.13560341e-01 8.54226410e-01 3.33458960e-01 -4.47600991e-01
-1.19360220e+00 -1.18407559e+00 5.79385638e-01 -1.17739454e-01
6.90015912e-01 -7.71418929e-01 -8.50431859e-01 6.52356744e-01
5.32313943e-01 -3.18048209e-01 1.06009781e+00 4.02744502e-01
-3.48968357e-01 -1.94437236e-01 -5.70922673e-01 5.13836086e-01
6.61796510e-01 -1.05340414e-01 -5.81606448e-01 7.14391887e-01
9.73353267e-01 1.77810173e-02 -6.12088740e-01 2.04561949e-01
8.62005129e-02 -4.77012604e-01 7.30441332e-01 -6.05602741e-01
7.57308245e-01 -3.97453874e-01 -3.20738733e-01 -1.26523781e+00
-8.61234441e-02 -5.24414539e-01 1.49850339e-01 1.59393215e+00
6.11702442e-01 -4.22564059e-01 7.79887319e-01 4.77658927e-01
-1.68046787e-01 -7.79122710e-01 -4.78578955e-01 -4.24441159e-01
3.56859595e-01 -5.83529651e-01 8.20332244e-02 1.29909098e+00
4.28181976e-01 7.92836666e-01 -4.21983838e-01 1.22378068e-02
5.79335213e-01 2.91422784e-01 7.22102761e-01 -1.38834858e+00
-3.36399376e-01 -6.06412530e-01 -5.80445193e-02 -1.40938270e+00
4.75781411e-01 -1.02164614e+00 3.97630453e-01 -1.69231379e+00
2.69673020e-01 -4.23041999e-01 -3.66697788e-01 1.91182718e-01
-3.24338794e-01 -2.16694266e-01 -3.71335186e-02 5.02799094e-01
-1.06580102e+00 6.74102783e-01 8.05959105e-01 -2.58494675e-01
-2.45765790e-01 2.41175573e-02 -1.10557675e+00 8.34959030e-01
9.21032727e-01 -5.32215238e-01 -6.58217907e-01 -4.75982308e-01
4.54385102e-01 -3.70163381e-01 -3.54671061e-01 -7.55911410e-01
7.43050039e-01 -4.43283506e-02 1.50566399e-01 -7.80833900e-01
2.73652911e-01 -1.05084550e+00 -5.17688870e-01 -1.04883805e-01
-6.86113954e-01 -1.03054322e-01 -2.06974611e-01 8.34358513e-01
-5.94473898e-01 -3.98252577e-01 6.65685713e-01 -2.25493595e-01
-5.29608011e-01 3.71531516e-01 -5.28376043e-01 1.57417029e-01
8.24072182e-01 1.52610540e-01 -2.88670570e-01 -3.63902301e-01
-6.74385130e-01 6.54869437e-01 7.44415522e-02 4.90549505e-01
4.66810226e-01 -8.66879880e-01 -6.34247661e-01 -3.40147108e-01
1.65028885e-01 5.84801912e-01 1.10334389e-01 7.86313653e-01
-2.76504189e-01 3.22993606e-01 4.37507659e-01 -5.86830080e-01
-9.44468498e-01 8.09552193e-01 1.16122060e-01 -3.98737937e-01
-5.23436368e-01 6.02302492e-01 5.07005274e-01 -6.49277747e-01
9.39840198e-01 -2.56019741e-01 -8.98804963e-01 3.99503410e-01
6.11142933e-01 -4.74178558e-03 9.09574255e-02 -4.58678961e-01
3.99407843e-04 5.56462705e-01 -5.98221540e-01 -2.13039443e-02
1.63212502e+00 -3.40746552e-01 -3.96530956e-01 8.08028102e-01
1.53453147e+00 -1.84019580e-01 -7.28883624e-01 -4.11899716e-01
2.17331588e-01 9.65733975e-02 -5.17730601e-02 -4.75571364e-01
-1.19099081e+00 1.00928092e+00 1.76877633e-01 3.67168665e-01
8.43872905e-01 -8.25403035e-02 6.04819298e-01 1.00840926e+00
3.66818644e-02 -1.56784046e+00 2.18162611e-01 6.19943082e-01
8.32358479e-01 -1.50556576e+00 9.23360437e-02 -5.17572284e-01
-9.85670209e-01 9.42957342e-01 8.12159717e-01 -1.33639470e-01
1.20069683e+00 8.70148763e-02 2.29018688e-01 -2.84670025e-01
-6.16004288e-01 -1.11030251e-01 2.00413749e-01 2.83309788e-01
6.76079452e-01 -1.50900424e-01 -5.17941773e-01 1.13833332e+00
-1.19050525e-01 -4.02833790e-01 4.02269751e-01 9.36461329e-01
-5.09116173e-01 -9.84480679e-01 -1.75746351e-01 5.16781926e-01
-7.68151104e-01 -2.05592826e-01 -4.47026461e-01 6.66845441e-01
-2.25130424e-01 1.15840530e+00 2.61672512e-02 -2.18083054e-01
-4.27651778e-02 1.30944356e-01 -3.38695437e-01 -9.40597057e-01
-7.43521690e-01 3.41346443e-01 -1.03435770e-01 -1.96230844e-01
-7.68745184e-01 -6.07362926e-01 -1.56070876e+00 8.79501551e-02
-5.57050467e-01 6.17113113e-01 8.69478226e-01 8.33936810e-01
2.91463703e-01 4.08716291e-01 8.50846469e-01 -6.13164663e-01
-3.85917425e-01 -1.02217567e+00 -8.36925209e-01 6.61278605e-01
-4.81879860e-02 -4.49149758e-01 -6.01108849e-01 3.08736593e-01] | [10.428473472595215, 6.730825901031494] |
9a15436d-31bc-4481-a08d-db63bbd968f2 | nuclear-segmentation-and-classification-on | 2301.03418 | null | https://arxiv.org/abs/2301.03418v1 | https://arxiv.org/pdf/2301.03418v1.pdf | Nuclear Segmentation and Classification: On Color & Compression Generalization | Since the introduction of digital and computational pathology as a field, one of the major problems in the clinical application of algorithms has been the struggle to generalize well to examples outside the distribution of the training data. Existing work to address this in both pathology and natural images has focused almost exclusively on classification tasks. We explore and evaluate the robustness of the 7 best performing nuclear segmentation and classification models from the largest computational pathology challenge for this problem to date, the CoNIC challenge. We demonstrate that existing state-of-the-art (SoTA) models are robust towards compression artifacts but suffer substantial performance reduction when subjected to shifts in the color domain. We find that using stain normalization to address the domain shift problem can be detrimental to the model performance. On the other hand, neural style transfer is more consistent in improving test performance when presented with large color variations in the wild. | ['Nasir Rajpoot', 'Abhir Bhalerao', 'Fayyaz Minhas', 'Shan E Ahmed Raza', 'Mostafa Jahanifar', 'Simon Graham', 'Robert Jewsbury', 'Quoc Dang Vu'] | 2023-01-09 | null | null | null | null | ['nuclear-segmentation'] | ['medical'] | [ 7.84815669e-01 -1.37636170e-01 1.53240770e-01 -2.47997761e-01
-1.21918142e+00 -5.80569804e-01 3.97915244e-01 1.01174161e-01
-6.76812887e-01 7.86669493e-01 -3.65942940e-02 -1.81549221e-01
-5.94780296e-02 -1.77978486e-01 -5.23636818e-01 -9.49441254e-01
1.46983922e-01 5.57900369e-01 4.54656631e-01 -1.10400081e-01
2.31286570e-01 6.87062860e-01 -9.36872363e-01 5.52553892e-01
7.41306603e-01 6.28560364e-01 9.78385210e-02 7.59311736e-01
5.84041551e-02 4.33160365e-01 -7.28952825e-01 -4.84222800e-01
2.14462221e-01 -6.39513493e-01 -1.05104709e+00 1.05176292e-01
8.64764869e-01 -2.42213026e-01 -2.00144455e-01 1.25333786e+00
8.80890787e-01 -2.04283774e-01 8.09034824e-01 -9.66520131e-01
-7.44067311e-01 1.65461674e-01 -6.18481040e-01 7.56922901e-01
-1.10158436e-01 4.74713236e-01 4.65353251e-01 -4.02956307e-01
1.29594266e+00 1.08463168e+00 8.45358729e-01 7.86905050e-01
-1.86707282e+00 -4.50402260e-01 -2.95665294e-01 2.29341269e-01
-1.11992562e+00 -3.21110427e-01 4.98232782e-01 -4.80315536e-01
9.03977334e-01 2.62082189e-01 5.03742337e-01 9.93410230e-01
4.32622284e-01 4.42303717e-01 1.49786854e+00 -5.02136171e-01
2.69367933e-01 1.44188954e-02 -8.31696093e-02 3.28640282e-01
4.24396425e-01 -1.32161140e-01 -2.54767269e-01 -1.23676039e-01
6.63774490e-01 -5.20582974e-01 -5.31666338e-01 -2.70897925e-01
-1.21565449e+00 4.55163985e-01 3.12435657e-01 5.63653767e-01
2.35631522e-02 1.52229354e-01 6.44519091e-01 3.27133149e-01
5.95690489e-01 6.52082622e-01 -4.03377742e-01 -1.33088216e-01
-1.07039487e+00 2.08612993e-01 5.18068314e-01 4.35539246e-01
2.29319349e-01 -1.71243727e-01 -2.28212401e-01 8.71814787e-01
-1.09483607e-01 -2.82202121e-02 8.00675035e-01 -1.00205922e+00
1.82059228e-01 5.28329194e-01 -2.95069426e-01 -7.00935423e-01
-5.48239768e-01 -8.36824596e-01 -5.06535649e-01 5.40542960e-01
9.89196956e-01 1.68926846e-02 -1.29588509e+00 1.58629096e+00
2.52163887e-01 1.78428274e-02 1.12715542e-01 7.03232229e-01
5.28556466e-01 -4.67608534e-02 4.12552983e-01 -1.75491527e-01
1.01401770e+00 -5.87780774e-01 -5.42934835e-01 -3.61084074e-01
1.04024351e+00 -9.02349174e-01 1.13929832e+00 4.40919638e-01
-1.02817070e+00 -2.24766955e-01 -1.19670451e+00 -2.48347670e-01
-4.12391037e-01 -6.76401183e-02 2.06884608e-01 8.21700692e-01
-1.17545760e+00 7.01009095e-01 -1.05388367e+00 -8.95018101e-01
7.98813760e-01 3.66806805e-01 -6.16411209e-01 -6.02569222e-01
-4.86434370e-01 1.24130511e+00 3.51683110e-01 -1.44759774e-01
-3.26751381e-01 -1.15511203e+00 -3.95943642e-01 -1.99997500e-01
-2.44407058e-02 -5.53325593e-01 1.11058629e+00 -1.20807421e+00
-8.97009373e-01 1.37429690e+00 1.96911141e-01 -4.86107051e-01
8.86398852e-01 1.93713114e-01 -3.60904187e-01 3.96496475e-01
5.25037982e-02 9.34455693e-01 4.16699469e-01 -1.27683425e+00
-2.98261762e-01 -5.52932680e-01 -6.04990244e-01 2.38246173e-01
-3.68293405e-01 -2.04085242e-02 -4.79991734e-01 -8.44393313e-01
5.74969575e-02 -1.06771517e+00 -1.52780429e-01 4.83726621e-01
-1.32708624e-01 2.79651076e-01 8.76666248e-01 -1.05878568e+00
7.55175173e-01 -2.51136208e+00 -2.98071397e-03 3.88064459e-02
-7.04799518e-02 3.70700836e-01 -4.11353223e-02 4.18742485e-02
-4.27564502e-01 3.04847240e-01 -3.98187697e-01 -1.79637015e-01
-1.94517002e-01 3.06059152e-01 1.25382632e-01 9.38859940e-01
1.60539299e-01 7.58923769e-01 -7.21423030e-01 -8.05305719e-01
-1.83738828e-01 3.63248855e-01 -5.55269897e-01 -1.76415429e-01
-5.20814136e-02 5.99828184e-01 4.50009815e-02 4.64641452e-01
8.30940485e-01 -4.26020205e-01 3.58431578e-01 -3.07976037e-01
4.08702016e-01 -7.97283500e-02 -6.21064782e-01 1.91412389e+00
5.43436743e-02 7.12073326e-01 2.24250227e-01 -9.37226117e-01
1.55694589e-01 2.82372922e-01 5.01073539e-01 -6.44936502e-01
2.67363749e-02 4.89935875e-01 8.06249857e-01 -4.88059998e-01
8.03398266e-02 -5.71116984e-01 4.76928085e-01 2.47146174e-01
2.31990457e-01 -2.38723069e-01 4.63176072e-01 1.05723642e-01
1.31395257e+00 1.57678604e-01 9.71790850e-02 -3.53260547e-01
2.58913726e-01 4.16280299e-01 5.75552404e-01 4.56723034e-01
-7.83561647e-01 1.13724065e+00 6.26238525e-01 1.38159059e-02
-1.21289492e+00 -1.14035118e+00 -3.97581339e-01 7.59222090e-01
-1.89192057e-01 8.02683458e-02 -9.80545402e-01 -8.31024706e-01
-3.05070747e-02 6.48249686e-01 -9.23802495e-01 -1.46561712e-01
-4.39948767e-01 -1.15859628e+00 7.48240769e-01 6.35646582e-01
3.51523250e-01 -7.78146267e-01 -6.30697250e-01 1.39715195e-01
5.43236136e-02 -1.25236452e+00 -4.11479384e-01 2.16038257e-01
-1.08023381e+00 -1.05594313e+00 -1.09815991e+00 -7.88970053e-01
8.83034289e-01 -1.90110013e-01 1.05556250e+00 3.77163142e-01
-9.40266669e-01 6.01927102e-01 -2.64808297e-01 -3.71006578e-01
-6.91744685e-01 -5.43695204e-02 -5.53970397e-01 -3.30516040e-01
1.61580861e-01 -3.38597953e-01 -6.17049038e-01 1.22717015e-01
-1.35696077e+00 3.32725607e-02 6.75387323e-01 1.02914345e+00
6.13422811e-01 -1.11151353e-01 3.38346541e-01 -1.19108546e+00
4.67323780e-01 -4.78419699e-02 3.09003461e-02 3.08522165e-01
-4.82542098e-01 -1.40052438e-01 4.28380340e-01 -4.10489261e-01
-9.92780387e-01 4.89120260e-02 -2.30327360e-02 -1.67150244e-01
-3.57730180e-01 3.71298432e-01 1.26809984e-01 -5.74391842e-01
1.09846842e+00 2.15357155e-01 3.00879389e-01 -2.43221506e-01
5.90038970e-02 5.75395711e-02 6.17539465e-01 -4.35131043e-01
6.64684176e-01 8.63517165e-01 2.77251184e-01 -8.72497201e-01
-4.96919125e-01 -4.44182038e-01 -7.97701240e-01 -1.98169559e-01
9.31519806e-01 -5.40833592e-01 1.50341049e-01 5.45726180e-01
-7.70334542e-01 -5.90205371e-01 -3.65701228e-01 3.58954251e-01
-6.85149312e-01 5.35150707e-01 -6.69275343e-01 -2.09410354e-01
-1.54975995e-01 -1.16941798e+00 6.71942234e-01 1.57525122e-01
-3.48656207e-01 -1.08905149e+00 1.11369625e-01 3.98062915e-01
3.57553691e-01 5.55904865e-01 1.55426991e+00 -6.82239473e-01
-2.06743300e-01 -2.08302021e-01 -3.58249694e-01 4.24157858e-01
1.67054400e-01 -8.67141858e-02 -9.39752340e-01 -4.39485520e-01
-1.27837449e-01 -4.66521859e-01 1.02913916e+00 4.54778910e-01
1.06428742e+00 3.66613567e-01 -4.27383453e-01 7.78469026e-01
1.48167264e+00 1.09143324e-01 9.00503397e-01 4.79959071e-01
2.03818753e-01 6.59473479e-01 3.80403221e-01 -3.74864370e-01
-1.72761962e-01 4.41689074e-01 1.47265658e-01 -5.86154878e-01
-7.93641865e-01 2.29749113e-01 -8.86886492e-02 2.68951625e-01
4.37208302e-02 -3.73367593e-02 -1.17091906e+00 8.15160275e-01
-1.42491901e+00 -7.42917895e-01 1.13971174e-01 1.86463559e+00
1.12533033e+00 2.99208671e-01 -2.16333326e-02 2.26109996e-01
5.35198092e-01 -3.50950092e-01 -5.74050367e-01 -9.99239981e-02
-1.80633709e-01 2.44569227e-01 6.59241021e-01 7.69761875e-02
-1.02217948e+00 6.70403302e-01 6.94934368e+00 8.99931610e-01
-1.30410874e+00 2.27302909e-01 1.13504291e+00 -3.01109612e-01
1.53416380e-01 -4.30055499e-01 -2.44148389e-01 3.26881677e-01
6.61253333e-01 1.78792059e-01 2.02927336e-01 4.29431736e-01
-4.77151293e-03 -3.34424108e-01 -1.26709592e+00 8.32428038e-01
3.72279465e-01 -1.12752426e+00 -9.01453495e-02 1.58253744e-01
8.39173794e-01 3.20915841e-02 5.59696078e-01 -6.51116446e-02
5.74136823e-02 -1.31476140e+00 2.44649142e-01 3.43280256e-01
1.09789622e+00 -4.17148888e-01 7.97513664e-01 -1.20303288e-01
-3.59877229e-01 1.19247615e-01 -1.89437971e-01 4.54189658e-01
-1.99009553e-01 2.36743748e-01 -1.17649221e+00 3.24666739e-01
7.15014935e-01 4.63840008e-01 -1.05392492e+00 1.53605044e+00
2.16615111e-01 7.62046874e-01 -1.90881208e-01 3.68462473e-01
2.15500519e-01 1.65268779e-01 3.63853782e-01 1.51204526e+00
2.57865936e-01 -6.84186369e-02 -2.25380939e-02 5.24187982e-01
5.02654426e-02 1.72622591e-01 -1.72726050e-01 -1.00679182e-01
-2.50170708e-01 1.16587579e+00 -1.34509492e+00 -2.05982432e-01
-3.79295170e-01 9.57009137e-01 2.85506099e-01 5.68980455e-01
-4.65131819e-01 -1.21829771e-01 3.96671027e-01 2.81388998e-01
2.67156899e-01 -8.21265131e-02 -7.79742718e-01 -7.41444409e-01
-2.12281793e-01 -1.06014645e+00 6.89047217e-01 -8.52027953e-01
-1.46515703e+00 3.26869160e-01 -3.28748316e-01 -1.02654350e+00
8.88685063e-02 -8.94953132e-01 -3.91126871e-01 9.06677246e-01
-1.50734913e+00 -1.26392841e+00 -1.77817151e-01 5.56074321e-01
2.71666020e-01 -1.61638111e-02 8.25275302e-01 3.41629505e-01
-1.87651008e-01 7.21527994e-01 4.52053368e-01 1.58414185e-01
1.34957409e+00 -1.47957885e+00 5.99391432e-03 8.98939550e-01
-1.26288563e-01 4.17849779e-01 8.58728230e-01 -6.88492060e-01
-9.36770201e-01 -9.40951407e-01 4.94654238e-01 -4.75034863e-01
4.95646685e-01 -5.99106215e-02 -1.14967740e+00 6.87809408e-01
2.20586494e-01 3.35967124e-01 8.81024301e-01 -2.04963401e-01
-3.22687745e-01 7.62350112e-02 -1.49954963e+00 5.97559094e-01
6.35751545e-01 -4.36091810e-01 -3.33327413e-01 2.86011755e-01
2.96296477e-01 -6.34751081e-01 -8.72824311e-01 1.82476059e-01
3.88376921e-01 -7.37924695e-01 8.75501812e-01 -7.69560218e-01
6.00505590e-01 -2.51359403e-01 -5.12483679e-02 -1.42556584e+00
-3.88924122e-01 -1.40365928e-01 3.91091794e-01 9.18906391e-01
4.76784140e-01 -3.91667157e-01 1.06748962e+00 5.58357835e-01
-2.33011022e-01 -5.45301676e-01 -1.21008980e+00 -6.52479649e-01
7.21367657e-01 -1.54497027e-01 -1.35858685e-01 9.48584378e-01
-1.00044301e-02 -1.23685725e-01 2.79844075e-01 -8.68528038e-02
5.93251467e-01 -1.99822351e-01 4.17578518e-01 -9.55952942e-01
-4.54821974e-01 -5.57126343e-01 -8.08158278e-01 -2.03027412e-01
-1.57068789e-01 -1.19793129e+00 2.02120151e-02 -1.52522731e+00
4.02409971e-01 -2.13562801e-01 -4.11269277e-01 3.18637222e-01
-6.17290251e-02 8.29390407e-01 3.33057135e-01 1.32486716e-01
-3.58193159e-01 -6.40657917e-02 1.65035069e+00 -1.83866188e-01
1.37085408e-01 -4.94028211e-01 -7.97870874e-01 5.82892478e-01
6.90031230e-01 -5.25597751e-01 -3.33858758e-01 -5.11823833e-01
-1.26946792e-02 -4.66091424e-01 4.88323092e-01 -1.30631101e+00
1.41371921e-01 1.10040136e-01 9.14191067e-01 -2.93505311e-01
3.33526075e-01 -7.13919282e-01 5.40246144e-02 7.16096282e-01
-3.90695602e-01 -3.28261815e-02 6.36500537e-01 5.34501851e-01
1.92438532e-02 -2.32511804e-01 1.38646030e+00 -3.02814960e-01
-4.94782627e-01 -5.47851920e-02 -2.87895739e-01 3.98399532e-01
9.90741074e-01 -4.20695126e-01 -6.38091803e-01 7.55036017e-03
-1.03745425e+00 -1.56848133e-02 7.02121615e-01 -9.31531787e-02
2.15804577e-01 -9.30355728e-01 -8.05904090e-01 -1.63355485e-01
4.31798808e-02 -4.00765747e-01 5.17973781e-01 1.19405913e+00
-9.77623940e-01 3.02612275e-01 -5.78137457e-01 -8.01800072e-01
-1.52739143e+00 2.94980884e-01 7.34407961e-01 -5.13818622e-01
-5.50454319e-01 9.23245072e-01 1.27703473e-01 -1.59173027e-01
1.39733374e-01 -1.87537417e-01 1.01348922e-01 -1.53539374e-01
4.41697001e-01 2.82679141e-01 6.34962618e-01 -4.51011300e-01
-3.59543383e-01 3.80499303e-01 -4.80515897e-01 -2.25062311e-01
1.40803730e+00 1.77598059e-01 1.14841640e-01 1.69202238e-01
1.20722687e+00 -4.65670824e-01 -1.22756839e+00 -4.63378504e-02
2.18730122e-02 -4.35968608e-01 1.87409088e-01 -1.55177355e+00
-1.03476369e+00 9.25606370e-01 1.25305331e+00 -6.74992055e-02
1.15743136e+00 -1.83457941e-01 5.57359159e-01 -1.77040845e-02
-2.29632780e-02 -1.27628076e+00 2.32018605e-02 1.64882943e-01
6.66509271e-01 -1.09743035e+00 2.49374568e-01 -5.31890750e-01
-5.62849224e-01 1.14958823e+00 6.30982697e-01 -6.03339002e-02
4.26888049e-01 3.31832170e-01 3.47734332e-01 -7.17668086e-02
-5.89206457e-01 2.71761492e-02 2.31432021e-01 1.11627162e+00
5.95524192e-01 -3.26237857e-01 -4.69963938e-01 2.25986466e-01
6.77790046e-02 1.21389024e-01 6.49313986e-01 1.09528077e+00
-1.05246250e-03 -1.02968991e+00 -5.59415996e-01 6.21929169e-01
-9.57673550e-01 9.84100699e-02 -4.58604515e-01 9.91973519e-01
2.62262195e-01 3.28682035e-01 4.23371941e-02 1.13384649e-01
1.08693071e-01 3.37953985e-01 9.60244477e-01 -7.30926573e-01
-7.70130754e-01 3.04500759e-01 -5.94093651e-02 -2.61436462e-01
-4.54921603e-01 -1.03236437e+00 -1.32913589e+00 -1.02269180e-01
-3.50239947e-02 -3.89320940e-01 6.26187205e-01 9.07020390e-01
4.31833304e-02 8.78572106e-01 -3.56954545e-01 -5.75989366e-01
-5.28891802e-01 -8.45829666e-01 -7.92615354e-01 9.05809164e-01
2.20135659e-01 -5.04523039e-01 -9.87307653e-02 6.37436271e-01] | [15.06659984588623, -2.8944811820983887] |
0fcba34b-13b1-467d-a6ff-cf9e9dbfa82e | game-of-tones-faculty-detection-of-gpt-4 | 2305.18081 | null | https://arxiv.org/abs/2305.18081v1 | https://arxiv.org/pdf/2305.18081v1.pdf | Game of Tones: Faculty detection of GPT-4 generated content in university assessments | This study explores the robustness of university assessments against the use of Open AI's Generative Pre-Trained Transformer 4 (GPT-4) generated content and evaluates the ability of academic staff to detect its use when supported by the Turnitin Artificial Intelligence (AI) detection tool. The research involved twenty-two GPT-4 generated submissions being created and included in the assessment process to be marked by fifteen different faculty members. The study reveals that although the detection tool identified 91% of the experimental submissions as containing some AI-generated content, the total detected content was only 54.8%. This suggests that the use of adversarial techniques regarding prompt engineering is an effective method in evading AI detection tools and highlights that improvements to AI detection software are needed. Using the Turnitin AI detect tool, faculty reported 54.5% of the experimental submissions to the academic misconduct process, suggesting the need for increased awareness and training into these tools. Genuine submissions received a mean score of 54.4, whereas AI-generated content scored 52.3, indicating the comparable performance of GPT-4 in real-life situations. Recommendations include adjusting assessment strategies to make them more resistant to the use of AI tools, using AI-inclusive assessment where possible, and providing comprehensive training programs for faculty and students. This research contributes to understanding the relationship between AI-generated content and academic assessment, urging further investigation to preserve academic integrity. | ['Don Hickerson', 'James McGaughran', 'Darius Postma', 'Jasper Roe', 'Mike Perkins'] | 2023-05-29 | null | null | null | null | ['prompt-engineering'] | ['natural-language-processing'] | [-3.54088061e-02 4.45810795e-01 1.93190172e-01 2.24457443e-01
-7.96008229e-01 -1.14649880e+00 4.19099748e-01 3.25490564e-01
-8.98483098e-02 2.88582623e-01 2.18893252e-02 -8.78982008e-01
-1.79092482e-01 -9.56135035e-01 -8.76868010e-01 -3.22008312e-01
5.02040207e-01 4.12746221e-01 -1.31214634e-01 -3.40568185e-01
9.06943381e-01 7.62632966e-01 -1.36628413e+00 3.90761383e-02
1.13853991e+00 2.30993241e-01 -5.93771577e-01 9.93405461e-01
1.31897600e-02 1.27653110e+00 -1.76759362e+00 -9.87605155e-01
3.00915867e-01 -3.21913064e-01 -9.49434459e-01 -3.92066717e-01
1.04656410e+00 -4.07808065e-01 1.31239817e-01 1.08379126e+00
4.16456193e-01 4.88200821e-02 4.76539642e-01 -1.66091263e+00
-1.12926543e+00 6.80849433e-01 5.60641736e-02 5.73797822e-01
7.84500003e-01 6.65682018e-01 6.32593334e-01 -3.77032876e-01
5.30501544e-01 1.10693705e+00 5.74488401e-01 3.36339206e-01
-1.21651518e+00 -1.61086154e+00 -5.79097867e-01 4.83180955e-02
-1.07550490e+00 -3.05741787e-01 7.88185596e-01 -8.86381686e-01
6.14570737e-01 3.15682024e-01 8.37137759e-01 1.21414995e+00
6.69180512e-01 1.14658237e-01 1.40898836e+00 -5.61878324e-01
3.74926120e-01 6.99373841e-01 2.62450010e-01 6.25526845e-01
8.30338895e-01 -5.91896884e-02 -1.27284288e-01 -6.07237935e-01
4.91495341e-01 -6.80501640e-01 2.62314230e-01 4.26512152e-01
-8.46733809e-01 7.33604312e-01 1.12197742e-01 4.62670922e-01
-2.07630575e-01 -5.04419729e-02 4.38191444e-01 4.22569871e-01
1.16522536e-01 1.39416337e+00 2.60633677e-01 -6.65708780e-01
-5.95472336e-01 2.81272709e-01 9.50779676e-01 7.99465418e-01
2.09631011e-01 5.49913764e-01 1.31198943e-01 2.49550313e-01
2.05030859e-01 2.49431074e-01 2.58376390e-01 -1.42701912e+00
7.02323392e-02 1.04063177e+00 9.38529149e-02 -1.39665735e+00
3.06828260e-01 -7.40148723e-01 1.51436061e-01 6.37196600e-01
6.46298885e-01 -1.19050249e-01 -5.40167749e-01 1.24424636e+00
5.85876480e-02 1.76605254e-01 7.94899985e-02 4.69783187e-01
5.97597122e-01 5.13760746e-01 5.39748490e-01 3.69298130e-01
1.33562922e+00 -2.09527522e-01 -7.08652377e-01 -9.88053009e-02
8.77591133e-01 -9.93527770e-01 1.18782198e+00 6.88630402e-01
-1.21169722e+00 -5.43026090e-01 -1.46720207e+00 3.44169945e-01
-4.84604985e-01 -4.87701207e-01 1.11668080e-01 1.56825531e+00
-7.58750916e-01 3.72001201e-01 -3.16302538e-01 -1.58175439e-01
4.84868944e-01 2.25524932e-01 -1.46666661e-01 -1.36098638e-01
-1.18397462e+00 9.75929379e-01 -1.37154549e-01 -2.61638820e-01
-8.43469977e-01 -1.13679338e+00 -6.19293630e-01 3.02233063e-02
-5.07513657e-02 -2.84201026e-01 1.25985813e+00 -7.03556478e-01
-1.11325002e+00 8.69809628e-01 4.59843069e-01 -1.73209324e-01
5.53219318e-01 -1.29158735e-01 -5.87162793e-01 4.40612108e-01
4.63074476e-01 1.08922467e-01 3.23886991e-01 -1.30667913e+00
-1.73287585e-01 -2.69777179e-01 2.37058029e-01 -2.92027920e-01
-6.76726818e-01 4.41807479e-01 6.77053213e-01 -5.54604769e-01
-3.76594305e-01 -9.35105562e-01 2.43230313e-01 -4.60770190e-01
-3.21506947e-01 -5.17518044e-01 8.09624493e-01 -6.19337082e-01
1.19556940e+00 -1.77216125e+00 -8.57304096e-01 4.80574191e-01
2.67880946e-01 3.51879388e-01 -2.31393185e-02 5.75541377e-01
-1.30098239e-01 9.65348184e-01 3.49455535e-01 3.73889297e-01
2.11193904e-01 -2.97848374e-01 -4.70717609e-01 3.95941913e-01
2.79120058e-01 4.77430463e-01 -1.13818765e+00 -3.79689217e-01
1.00191049e-01 1.72025621e-01 -2.49915689e-01 1.32409140e-01
1.82795987e-01 7.56419301e-02 -4.36279416e-01 9.40425515e-01
3.15171301e-01 1.56971619e-01 -4.09407586e-01 5.36485732e-01
-2.04962611e-01 3.91734779e-01 -7.85220623e-01 5.52635193e-01
-1.76131174e-01 9.86391068e-01 -1.02390468e-01 -6.11942708e-01
1.35725427e+00 6.18565381e-01 3.09149418e-02 -5.42465806e-01
3.74140173e-01 2.84444273e-01 4.37163681e-01 -3.87825996e-01
5.41805804e-01 -5.87074040e-03 -2.74689019e-01 7.95035779e-01
-1.64335683e-01 -6.06569588e-01 7.32577685e-03 5.49273491e-01
1.35242796e+00 -1.13717906e-01 -3.74002755e-01 -3.29207301e-01
3.15571755e-01 4.78716463e-01 3.09527695e-01 1.06800842e+00
-7.19022572e-01 6.14320077e-02 4.99962032e-01 -1.47451058e-01
-1.20124173e+00 -5.77548206e-01 -1.89031929e-01 8.75183225e-01
-3.79266858e-01 -3.20552826e-01 -1.05876112e+00 -6.60305917e-01
-9.21428874e-02 1.67766285e+00 -5.26608467e-01 -7.52672434e-01
-1.46833703e-01 3.00141126e-02 1.28996587e+00 3.13415110e-01
3.23589325e-01 -1.08464491e+00 -7.65821457e-01 2.96919137e-01
-5.55365197e-02 -9.65532660e-01 -7.10127279e-02 -3.74296904e-01
-4.84685212e-01 -9.77882922e-01 -2.40989134e-01 -5.53127527e-01
6.63602173e-01 -4.39029261e-02 8.98587048e-01 3.85322571e-01
-4.06531066e-01 1.01474535e+00 -6.99079856e-02 -1.04683006e+00
-1.39991784e+00 -3.51975918e-01 -7.55069405e-02 -7.24653959e-01
9.06796157e-01 -4.06719029e-01 -6.38311580e-02 2.58253843e-01
-8.69128585e-01 -3.61234277e-01 3.29153210e-01 3.68942827e-01
-2.60248065e-01 5.42084202e-02 9.03559625e-01 -7.22127199e-01
1.02835202e+00 -4.03201401e-01 -3.20296347e-01 1.24113597e-01
-8.89916182e-01 -4.03160542e-01 8.73358846e-01 -7.21130610e-01
-6.76992297e-01 -8.13583612e-01 2.75622189e-01 -4.22259837e-01
-4.62381274e-01 2.74907678e-01 -4.84642684e-02 -7.03374267e-01
1.17831016e+00 -2.11431786e-01 1.19496085e-01 2.90748358e-01
-4.89624023e-01 5.98119974e-01 4.21873808e-01 -7.78443038e-01
1.10123050e+00 -3.41849685e-01 -6.04434907e-02 -5.67319989e-01
-6.26128435e-01 9.79451835e-02 -2.55614836e-02 -7.82069206e-01
5.38528264e-01 -6.93430662e-01 -1.05155373e+00 2.19098225e-01
-7.20689237e-01 -2.73473412e-01 -2.82452349e-02 5.54572165e-01
-2.67906059e-02 1.96348205e-01 -6.72922015e-01 -1.00359988e+00
-3.53065938e-01 -1.30227184e+00 3.48524898e-01 4.06928301e-01
-8.23078275e-01 -7.03690648e-01 8.82618502e-02 1.18286562e+00
2.12680817e-01 4.16627347e-01 8.88781905e-01 -1.24952257e+00
-1.74835071e-01 -8.39493871e-01 2.63319075e-01 3.71038735e-01
-1.52790114e-01 5.98938823e-01 -1.05538440e+00 -4.05941457e-02
8.21968447e-03 -4.58563149e-01 -3.84573966e-01 -2.96472073e-01
6.53962135e-01 -7.06586361e-01 1.47870183e-01 -2.93131799e-01
1.20230889e+00 7.17032075e-01 6.09781504e-01 1.05466592e+00
5.52037716e-01 6.66642725e-01 4.75630522e-01 4.51312549e-02
-4.29102555e-02 8.83018002e-02 3.57715011e-01 6.28608465e-01
1.39022097e-01 -2.72573322e-01 7.43708014e-01 6.79773629e-01
1.00157283e-01 -1.42893016e-01 -1.31358683e+00 4.82211202e-01
-9.38129961e-01 -1.36156201e+00 -2.85051376e-01 1.95729864e+00
5.95950127e-01 7.44835794e-01 5.59114553e-02 2.80562013e-01
6.39380336e-01 -5.12659371e-01 -3.87048483e-01 -1.41134691e+00
3.03665757e-01 4.55042034e-01 2.85851717e-01 2.86155790e-01
-4.97966111e-01 4.06216949e-01 6.73445654e+00 3.96020830e-01
-6.44795239e-01 -3.25543076e-01 8.29563439e-01 2.21968979e-01
-6.48610651e-01 8.98062959e-02 -6.37866020e-01 6.92159772e-01
1.70240998e+00 -8.62703085e-01 -7.56966919e-02 9.40817177e-01
1.49496526e-01 4.32658568e-02 -6.65424049e-01 -2.53178477e-02
2.45243713e-01 -1.09493542e+00 1.48817867e-01 3.68201673e-01
6.50915802e-01 -9.07951951e-01 1.68433115e-01 5.73548913e-01
9.66173947e-01 -1.14608634e+00 7.58292019e-01 2.64908403e-01
2.14292869e-01 -1.11911631e+00 8.47262800e-01 9.44986418e-02
-3.64959836e-01 -1.16464227e-01 -9.65101570e-02 -4.11850929e-01
-7.39663064e-01 1.16765209e-01 -1.55378902e+00 2.69060344e-01
5.83470166e-01 -1.13762833e-01 -9.32644665e-01 8.73722672e-01
-1.59354210e-01 1.55771196e+00 -2.04635128e-01 -2.05560356e-01
2.34972343e-01 -1.80279523e-01 8.89352202e-01 1.04801142e+00
2.27842093e-01 2.90681180e-02 -7.26487786e-02 1.14074576e+00
-7.78935775e-02 -1.86629355e-01 -8.94388855e-01 -5.84389210e-01
1.01198840e+00 1.08368695e+00 -5.58807135e-01 -2.61812657e-01
-1.08228453e-01 4.06725436e-01 -1.27335325e-01 2.12037452e-02
-6.70414746e-01 -7.66997516e-01 4.88185883e-01 3.25046390e-01
-4.40787494e-01 5.61682321e-02 -7.71978438e-01 -1.18305951e-01
-2.53303617e-01 -1.21839237e+00 1.68817312e-01 -1.05805802e+00
-1.12311435e+00 1.91290870e-01 -5.95947877e-02 -1.09099972e+00
-2.41549671e-01 -4.50893968e-01 -9.91197944e-01 8.77391517e-01
-4.80045766e-01 -1.00283992e+00 -4.87072617e-01 3.45468111e-02
2.08766267e-01 -1.85018450e-01 8.75784934e-01 -1.47113517e-01
-7.46078014e-01 1.15309048e+00 -3.16550702e-01 4.47996318e-01
8.16653550e-01 -1.13671899e+00 6.41638115e-02 9.35892522e-01
-4.40460086e-01 1.03512359e+00 1.03332734e+00 -9.72027659e-01
-1.39809871e+00 -7.57474840e-01 6.73098445e-01 -8.16120148e-01
9.29329038e-01 1.07773520e-01 -1.06369972e+00 8.21405947e-01
4.93478984e-01 -6.73046649e-01 1.24265659e+00 -2.83047169e-01
-2.79347092e-01 3.80995154e-01 -1.54045844e+00 8.57559979e-01
3.65459234e-01 -6.70913577e-01 -8.83918703e-01 1.87324911e-01
5.99875391e-01 -1.92037046e-01 -1.46073830e+00 4.03130576e-02
5.75307727e-01 -5.23201644e-01 6.75381243e-01 -4.51598376e-01
8.68269801e-01 8.73660892e-02 5.44483960e-01 -1.06172526e+00
-5.01718819e-01 -7.29352415e-01 2.15578303e-01 1.65242589e+00
1.99403122e-01 -6.22285068e-01 6.80432081e-01 1.60610974e+00
-4.60351884e-01 -3.17675203e-01 -6.45106077e-01 -5.07538080e-01
5.60311735e-01 -3.28618944e-01 3.27365160e-01 1.53806329e+00
2.17730016e-01 -4.04752372e-03 3.35279077e-01 2.26195768e-01
6.86421812e-01 -4.87393081e-01 8.90421212e-01 -1.40786183e+00
1.85590252e-01 -8.07674348e-01 -8.70172679e-01 4.28883463e-01
1.20653197e-01 -7.38915026e-01 -4.64345723e-01 -1.15884793e+00
-2.59450246e-02 -1.89812630e-01 1.01695880e-02 6.48275673e-01
-2.89864361e-01 2.70141542e-01 3.44272643e-01 1.92653745e-01
2.15858385e-01 -8.27615857e-02 1.18510675e+00 -7.29867220e-02
-1.44638464e-01 -2.56580621e-01 -1.37210870e+00 4.96585816e-01
1.06592631e+00 -5.59882462e-01 -1.43936813e-01 2.41268098e-01
2.46732444e-01 -5.31564295e-01 6.78651810e-01 -1.46440732e+00
3.91433090e-01 -3.36457461e-01 6.14296019e-01 -1.21297933e-01
-8.92342106e-02 -8.63523960e-01 1.50729641e-01 5.33856809e-01
-4.46980596e-01 2.37591535e-01 8.88182938e-01 -1.58937946e-01
2.76407570e-01 -7.67836452e-01 5.04490674e-01 -1.45581812e-01
-1.39023382e-02 -5.80925584e-01 -1.12945867e+00 1.93286799e-02
1.36813831e+00 -7.55107582e-01 -6.73572779e-01 -5.25467932e-01
-1.86178371e-01 -3.81570086e-02 6.91219389e-01 1.81308225e-01
4.46757793e-01 -9.82410848e-01 -7.67196715e-01 9.09239873e-02
3.59777547e-02 -3.79069239e-01 1.04720019e-01 3.84778470e-01
-7.53100753e-01 2.19260201e-01 -5.77239275e-01 3.33308652e-02
-1.21181107e+00 3.65634322e-01 5.93472756e-02 1.55720219e-01
-6.41919136e-01 5.96382737e-01 -3.97259593e-01 -2.80514121e-01
1.75276265e-01 5.23642004e-01 -3.04706782e-01 -1.29113480e-01
5.64263046e-01 7.59624779e-01 5.66128194e-02 -4.28817183e-01
-2.92569846e-02 -7.68314525e-02 -3.48696560e-01 -2.34763175e-01
8.82471025e-01 4.98336941e-01 1.47623882e-01 6.33055270e-01
6.28457725e-01 5.16601443e-01 -3.93433154e-01 4.29890484e-01
5.42324185e-02 -5.21660268e-01 5.47389835e-02 -1.13663280e+00
-4.49320048e-01 4.91668016e-01 3.86200219e-01 3.75276208e-01
5.95507324e-01 -4.75650132e-01 4.73010510e-01 2.67421901e-01
9.82195884e-02 -9.82285142e-01 3.36306006e-01 1.39394358e-01
8.23149264e-01 -5.74162483e-01 3.65689211e-02 -1.11491226e-01
-5.21654546e-01 1.26794136e+00 1.25552571e+00 -2.68723309e-01
3.62547711e-02 2.29155019e-01 5.71658798e-02 -3.31637174e-01
-5.62454700e-01 8.53169024e-01 2.32323259e-02 5.12067735e-01
6.18735552e-01 1.86933741e-01 -2.37444848e-01 5.03466368e-01
-8.70194077e-01 6.12226278e-02 1.48922217e+00 1.11992526e+00
-2.82491088e-01 -8.27277362e-01 -1.14216268e+00 5.85314691e-01
-8.98244977e-01 7.77014866e-02 -1.19211757e+00 8.15101981e-01
-1.26582608e-01 1.27430105e+00 1.89061239e-01 -6.30627632e-01
3.01220894e-01 6.30887985e-01 1.02648981e-01 -3.53484333e-01
-1.45748138e+00 -5.17221749e-01 2.51525879e-01 -6.13402128e-02
1.59956753e-01 -6.84388459e-01 -8.65266442e-01 -1.01011169e+00
-3.88073742e-01 3.98365676e-01 6.14119470e-01 6.20825052e-01
6.29867196e-01 7.17562139e-01 3.39671254e-01 -1.55846179e-01
-5.78913510e-01 -1.09160554e+00 -1.36633161e-02 3.66191417e-01
-1.35328189e-01 -5.26731074e-01 -4.85782206e-01 -1.76385269e-01] | [10.019489288330078, 7.294789791107178] |
bfa1e721-9c21-4342-bd38-7604917c67dd | it-s-raw-audio-generation-with-state-space | 2202.09729 | null | https://arxiv.org/abs/2202.09729v1 | https://arxiv.org/pdf/2202.09729v1.pdf | It's Raw! Audio Generation with State-Space Models | Developing architectures suitable for modeling raw audio is a challenging problem due to the high sampling rates of audio waveforms. Standard sequence modeling approaches like RNNs and CNNs have previously been tailored to fit the demands of audio, but the resultant architectures make undesirable computational tradeoffs and struggle to model waveforms effectively. We propose SaShiMi, a new multi-scale architecture for waveform modeling built around the recently introduced S4 model for long sequence modeling. We identify that S4 can be unstable during autoregressive generation, and provide a simple improvement to its parameterization by drawing connections to Hurwitz matrices. SaShiMi yields state-of-the-art performance for unconditional waveform generation in the autoregressive setting. Additionally, SaShiMi improves non-autoregressive generation performance when used as the backbone architecture for a diffusion model. Compared to prior architectures in the autoregressive generation setting, SaShiMi generates piano and speech waveforms which humans find more musical and coherent respectively, e.g. 2x better mean opinion scores than WaveNet on an unconditional speech generation task. On a music generation task, SaShiMi outperforms WaveNet on density estimation and speed at both training and inference even when using 3x fewer parameters. Code can be found at https://github.com/HazyResearch/state-spaces and samples at https://hazyresearch.stanford.edu/sashimi-examples. | ['Christopher Ré', 'Chris Donahue', 'Albert Gu', 'Karan Goel'] | 2022-02-20 | null | null | null | null | ['audio-generation', 'music-generation', 'music-generation'] | ['audio', 'audio', 'music'] | [-2.90505420e-02 3.53932083e-02 1.20442830e-01 8.80333111e-02
-1.17881119e+00 -7.60410190e-01 5.59245527e-01 -5.27242184e-01
1.97147429e-01 5.24049520e-01 5.98325372e-01 -3.88055295e-01
-1.72879845e-02 -4.47123885e-01 -7.34909952e-01 -5.84760904e-01
-3.07266593e-01 4.54455912e-01 -3.01439762e-01 -2.09032595e-01
-7.25604668e-02 1.34824693e-01 -1.24591315e+00 1.42486915e-01
6.30210876e-01 9.58650291e-01 1.21670313e-01 1.34824038e+00
2.36019075e-01 7.91199267e-01 -8.82710874e-01 -3.70318532e-01
7.66602233e-02 -5.98539114e-01 -5.35053849e-01 -3.31470549e-01
4.25471991e-01 -3.63034576e-01 -6.05437160e-01 8.02236617e-01
9.23838317e-01 2.66830087e-01 8.69819164e-01 -1.18803108e+00
-1.00656354e+00 1.38446915e+00 -3.29254150e-01 2.61656821e-01
6.25910908e-02 2.57588178e-01 1.18426836e+00 -1.05194438e+00
1.45394757e-01 1.36523604e+00 8.08304727e-01 6.66920960e-01
-1.36603403e+00 -1.00077152e+00 -8.57510418e-02 1.32870302e-01
-1.63397360e+00 -8.09228361e-01 8.86961639e-01 -3.61544997e-01
1.03089654e+00 2.60183573e-01 4.89979118e-01 1.61846685e+00
-8.52299761e-03 9.83408868e-01 3.56862843e-01 -1.39130652e-01
2.14539856e-01 -2.94853777e-01 -1.53562471e-01 1.09395757e-01
-1.92585796e-01 2.51727045e-01 -8.36788177e-01 -3.56365144e-01
1.10034144e+00 -6.59162223e-01 -3.19251627e-01 2.64562488e-01
-1.08462560e+00 7.16264725e-01 6.16157800e-02 1.93021595e-01
-5.08634329e-01 8.90686750e-01 3.01178336e-01 4.52437937e-01
5.12795329e-01 5.03437340e-01 -3.68184596e-01 -9.12632942e-01
-1.25431979e+00 4.61553752e-01 8.94610763e-01 1.00171983e+00
1.35620445e-01 1.20527589e+00 -2.25569960e-02 9.39179003e-01
6.17408305e-02 7.09046662e-01 6.21041179e-01 -1.20172238e+00
3.31029922e-01 -4.61272717e-01 3.37931216e-02 -6.54047728e-01
-2.59154081e-01 -9.18238640e-01 -1.03050268e+00 -1.05603434e-01
3.30593675e-01 -5.80607355e-01 -8.79920244e-01 1.90588915e+00
-1.79356039e-01 7.85396457e-01 7.31559545e-02 6.53885841e-01
5.42104065e-01 1.09082818e+00 -1.53353304e-01 1.15808733e-01
1.05815721e+00 -7.78366327e-01 -6.32481277e-01 -1.83265328e-01
2.84693539e-01 -9.69422221e-01 1.09085333e+00 8.57146919e-01
-1.55701005e+00 -6.87624276e-01 -9.57408726e-01 1.79547697e-01
1.40410095e-01 2.99350858e-01 4.29858118e-01 8.02480876e-01
-1.44771028e+00 5.72712064e-01 -8.98444295e-01 1.07900009e-01
1.28766209e-01 7.87135586e-02 2.29302600e-01 4.84309644e-01
-1.33890462e+00 6.52526200e-01 2.42369384e-01 1.58652756e-02
-1.28997612e+00 -1.26246297e+00 -7.78106391e-01 4.31158870e-01
2.96117365e-02 -6.14347577e-01 1.82969320e+00 -6.17381215e-01
-1.90648186e+00 -1.79982267e-03 -6.46658167e-02 -9.07887220e-01
3.04815859e-01 -5.58933079e-01 -6.72299922e-01 8.83308575e-02
-5.22028685e-01 7.07449436e-01 1.10450733e+00 -8.73747051e-01
-1.76830485e-01 3.64163220e-01 -3.64952475e-01 -4.00534198e-02
-2.87419230e-01 -6.34537265e-02 -5.75541370e-02 -1.16661298e+00
-4.25880730e-01 -9.70384955e-01 -1.92912459e-01 -6.19922042e-01
-3.95742506e-01 -2.05016658e-01 5.50616980e-01 -8.61901462e-01
1.44764936e+00 -2.06230545e+00 1.91451684e-01 1.43362656e-01
2.52246261e-02 3.28693777e-01 -5.66232383e-01 6.86936498e-01
-3.80825102e-01 2.84642756e-01 7.15869144e-02 -4.80987400e-01
4.11105365e-01 -3.24758589e-01 -9.46973085e-01 3.44566107e-02
3.53885263e-01 1.01067877e+00 -5.46217561e-01 1.06537290e-01
2.90023796e-02 8.15363765e-01 -7.71057308e-01 9.33141541e-03
-3.73233378e-01 2.90457040e-01 1.16979472e-01 3.19973081e-01
3.79640013e-01 -2.54091114e-01 5.34456735e-03 -4.61759754e-02
1.52147397e-01 6.08212769e-01 -1.04643476e+00 1.79838657e+00
-7.92198241e-01 8.20553958e-01 4.99681607e-02 -6.85451508e-01
1.06920385e+00 7.68743098e-01 2.69420385e-01 -5.59294745e-02
1.53427929e-01 2.97514170e-01 2.55449206e-01 1.05453432e-02
6.27072453e-01 -2.36391976e-01 1.08455859e-01 5.31799197e-01
3.56234878e-01 -4.90336984e-01 1.26074582e-01 3.42661858e-01
1.00885391e+00 1.12042554e-01 -3.22411135e-02 -9.47623476e-02
-3.87329869e-02 -5.97691774e-01 2.68408626e-01 8.65943253e-01
3.70807171e-01 9.93846297e-01 3.30923468e-01 1.06192760e-01
-1.15566289e+00 -1.36940515e+00 1.18570812e-01 1.16929531e+00
-6.38810098e-01 -6.85175896e-01 -7.13181794e-01 1.60378218e-01
-1.64536834e-01 1.22495759e+00 -2.25883931e-01 -3.31294835e-01
-5.24417579e-01 -6.52820230e-01 1.30776525e+00 7.65624046e-01
-1.70739949e-01 -1.11264837e+00 -1.70268148e-01 4.88609672e-01
-2.21260875e-01 -8.33664298e-01 -7.64448404e-01 5.03011905e-02
-5.79062343e-01 -2.53093213e-01 -8.99274886e-01 -4.72412527e-01
-1.62118644e-01 -1.67867377e-01 1.31178844e+00 -2.25844011e-01
-1.00214824e-01 3.67924243e-01 -2.78154701e-01 -6.52529299e-01
-7.30677903e-01 3.27041537e-01 3.54527026e-01 -1.69928506e-01
-7.45185018e-02 -1.21365559e+00 -6.26879513e-01 -7.68257007e-02
-9.24528539e-01 -1.64341360e-01 5.56100428e-01 7.71367610e-01
2.93854978e-02 -5.51281720e-02 1.23699403e+00 -4.06570613e-01
1.02940190e+00 -6.89334631e-01 -4.48870927e-01 -1.15605481e-01
-3.96694005e-01 -8.93834978e-02 7.03579545e-01 -8.74486208e-01
-9.30828691e-01 -4.44905758e-01 -5.87875664e-01 -8.21446121e-01
1.18833721e-01 7.00217187e-01 2.22376779e-01 4.97912884e-01
7.95545936e-01 4.11572993e-01 -3.77436890e-03 -5.64394116e-01
6.68156564e-01 6.28515482e-01 7.22280860e-01 -5.74468315e-01
8.86510372e-01 -7.56993592e-02 -4.10260499e-01 -9.62417424e-01
-4.63329792e-01 -1.10608406e-01 -1.12943046e-01 -1.17450021e-01
4.34436768e-01 -1.07726240e+00 -6.78456485e-01 6.02742255e-01
-1.15913296e+00 -6.80372179e-01 -4.03737336e-01 6.07282877e-01
-8.22178185e-01 2.37443671e-01 -1.04766273e+00 -1.25375187e+00
-6.87360585e-01 -7.06860065e-01 9.67274427e-01 -1.06431246e-02
-7.46862233e-01 -1.09034526e+00 2.23760843e-01 -1.67481214e-01
8.84091198e-01 -2.84582138e-01 9.19026554e-01 -6.21652663e-01
-3.69484127e-01 -1.92495361e-01 2.67102242e-01 5.87659001e-01
-1.48386270e-01 1.98095575e-01 -1.18172038e+00 -1.92554086e-01
-3.44866067e-02 -4.47046518e-01 7.93066919e-01 7.15746224e-01
1.09058404e+00 -3.77467453e-01 2.82246470e-01 6.10367239e-01
8.03911090e-01 3.11233521e-01 6.77793205e-01 -3.39189231e-01
4.74036336e-01 1.00479215e-01 1.12906642e-01 6.81689620e-01
2.36038417e-01 5.93275189e-01 1.15750730e-01 1.12496749e-01
-4.37657237e-01 -5.98054886e-01 7.65578389e-01 1.47406626e+00
7.12848306e-02 -5.40547192e-01 -8.24745953e-01 6.25213504e-01
-1.63391900e+00 -1.18587732e+00 5.53645240e-03 1.99902296e+00
1.13502049e+00 1.85031831e-01 3.19074929e-01 5.78447059e-02
3.45122635e-01 2.32246637e-01 -5.63623309e-01 -3.65571320e-01
-1.81255162e-01 6.94152832e-01 5.91990501e-02 5.76959670e-01
-7.71770418e-01 8.75503600e-01 6.60888243e+00 1.37932026e+00
-1.12802815e+00 1.07217208e-01 6.85936451e-01 -5.60245931e-01
-5.25174856e-01 -2.36377403e-01 -7.38670945e-01 4.39244479e-01
1.74074960e+00 -4.73930329e-01 6.21741772e-01 6.59750640e-01
3.31272393e-01 5.44033647e-01 -1.08000302e+00 1.17712498e+00
-1.91843119e-02 -1.52933991e+00 1.60600603e-01 -7.94739351e-02
6.35231376e-01 2.06963256e-01 6.07785523e-01 5.78456998e-01
6.31229281e-01 -1.32223070e+00 1.05148065e+00 4.42063391e-01
9.64556098e-01 -7.98120618e-01 3.16073090e-01 2.89487898e-01
-1.20562422e+00 4.69357632e-02 -4.55093116e-01 -5.17780706e-03
5.16279578e-01 4.62895721e-01 -1.00651097e+00 3.91835362e-01
2.89004654e-01 5.42039633e-01 -1.55160561e-01 1.04844368e+00
-2.55378317e-02 1.47282982e+00 -3.79667193e-01 6.65229633e-02
2.13830337e-01 -1.68221936e-01 7.87640214e-01 1.31824458e+00
9.24808919e-01 -2.32603014e-01 -1.70565724e-01 1.04267800e+00
-1.07066832e-01 -1.22801043e-01 -5.12334406e-01 -3.48735303e-01
7.65387654e-01 1.05586028e+00 -1.64564580e-01 -4.13897425e-01
2.99415290e-02 6.05798542e-01 1.06167309e-02 7.57885396e-01
-1.01653993e+00 -3.40630919e-01 7.30279565e-01 -1.53791299e-02
4.39745486e-01 -5.07511973e-01 -2.03708813e-01 -1.07057071e+00
-3.34504038e-01 -1.10854435e+00 3.53539698e-02 -1.14121819e+00
-1.29031074e+00 8.71509314e-01 -8.38160068e-02 -1.30855966e+00
-1.14292991e+00 -3.74222755e-01 -8.98696542e-01 1.05196869e+00
-1.07342041e+00 -1.07609916e+00 1.80087268e-01 2.59308279e-01
8.56421828e-01 -3.28527629e-01 9.43941116e-01 3.01003397e-01
-2.10673720e-01 7.28265405e-01 7.21592382e-02 -5.54750673e-02
5.93243241e-01 -1.30003095e+00 1.17168272e+00 8.61204028e-01
6.58717513e-01 6.37739539e-01 9.56717014e-01 -5.05547166e-01
-1.19127464e+00 -1.09298456e+00 4.66361552e-01 -4.54100072e-01
1.19404173e+00 -4.42135811e-01 -1.00121641e+00 7.99464047e-01
6.20282829e-01 -7.00386703e-01 8.31857324e-01 1.92437381e-01
-4.65224862e-01 1.24443717e-01 -4.06347781e-01 9.23362374e-01
9.26612556e-01 -6.57187045e-01 -2.70890146e-01 1.56256005e-01
9.88397181e-01 -3.25830013e-01 -9.49614286e-01 9.00301188e-02
6.01925552e-01 -6.16469383e-01 1.04320753e+00 -5.25125921e-01
5.37063956e-01 -1.13805458e-01 -1.20443385e-02 -1.73103237e+00
-3.98599505e-01 -1.53871787e+00 -5.57115495e-01 1.35512471e+00
7.24556744e-01 -3.82879287e-01 6.29978597e-01 2.17378229e-01
-3.57181311e-01 -5.33942461e-01 -7.81701446e-01 -1.03807080e+00
3.97894859e-01 -8.98088396e-01 7.05028117e-01 5.22541881e-01
-4.15053256e-02 4.64804202e-01 -9.18320298e-01 9.24601927e-02
4.06494409e-01 -1.30743071e-01 7.41874576e-01 -8.62048984e-01
-7.45203733e-01 -8.34099114e-01 -5.95576316e-02 -1.39750111e+00
1.56810209e-01 -9.15570319e-01 4.57618982e-02 -1.13274610e+00
-3.88762534e-01 -3.26414853e-01 -1.13902062e-01 1.35647953e-01
3.76404612e-03 4.16483104e-01 4.45376784e-01 -1.01193346e-01
-6.50674552e-02 8.45072687e-01 7.66751826e-01 8.63003135e-02
-2.31730655e-01 3.24960202e-01 -7.94862330e-01 6.38439834e-01
1.20636606e+00 -1.13598324e-01 -8.32543433e-01 -4.53957170e-01
2.86576658e-01 4.20782149e-01 2.92443484e-01 -1.19271028e+00
1.17226206e-01 2.26977304e-01 3.08744818e-01 -6.59427106e-01
9.29935575e-01 -1.02618881e-01 6.22829854e-01 1.82636872e-01
-4.93914396e-01 2.45876595e-01 3.84389460e-01 4.96553630e-01
-2.52618343e-01 -2.33189255e-01 4.84317899e-01 1.62349418e-01
-1.93361163e-01 3.01189899e-01 -7.05779433e-01 2.47999504e-01
2.41355866e-01 1.44336149e-01 -4.02232379e-01 -1.26157153e+00
-8.29873562e-01 -1.54640526e-01 -3.10557354e-02 5.57785094e-01
7.65953064e-01 -1.43868792e+00 -1.04157817e+00 1.30421922e-01
-5.19294679e-01 -3.35395515e-01 5.13422728e-01 8.25877011e-01
-2.57300496e-01 4.08134282e-01 1.84442982e-01 -3.61944824e-01
-9.77616906e-01 1.73584357e-01 2.12008968e-01 -2.18534455e-01
-4.30827856e-01 1.07664156e+00 3.85057867e-01 -1.71774566e-01
3.28187317e-01 -5.16475439e-01 1.18963301e-01 -2.27561798e-02
7.03692377e-01 4.59416300e-01 1.22443829e-02 -2.39321023e-01
5.05597703e-02 1.46634161e-01 7.70475939e-02 -5.93727052e-01
1.27430487e+00 -1.54283820e-02 2.14345574e-01 7.25178897e-01
9.42760706e-01 2.58585036e-01 -1.26637125e+00 -2.48131137e-02
-1.98790088e-01 -1.48479054e-02 -1.19095683e-01 -8.24902892e-01
-9.44351673e-01 1.17858386e+00 1.34066060e-01 4.32077855e-01
9.71807182e-01 -2.00310305e-01 9.22569156e-01 1.33227840e-01
5.37418388e-02 -6.56540155e-01 2.91325063e-01 7.90267348e-01
1.26844192e+00 -4.46828544e-01 -5.09879589e-01 3.34731527e-02
-9.63257790e-01 1.01858544e+00 3.93607974e-01 -2.44156331e-01
7.77531743e-01 6.56281948e-01 1.16746418e-01 1.22450456e-01
-1.35288942e+00 1.63235635e-01 3.49189788e-01 6.34276688e-01
6.46278739e-01 1.27072215e-01 3.40031236e-01 8.39925945e-01
-1.06293452e+00 -1.93498999e-01 8.00671756e-01 2.95831650e-01
-2.36005291e-01 -9.00891483e-01 -3.74109894e-01 4.49841708e-01
-5.18042505e-01 -6.31097555e-01 -1.50027364e-01 4.30726260e-01
-6.77238345e-01 1.11819339e+00 2.19178975e-01 -4.60626155e-01
-3.05032432e-02 4.09515649e-01 3.02046239e-01 -5.08737624e-01
-5.39605558e-01 6.93282843e-01 2.59621531e-01 -1.17885232e-01
5.75244054e-02 -5.79516888e-01 -1.06310117e+00 -3.68740648e-01
-1.85115874e-01 2.88783729e-01 4.38855171e-01 4.09674674e-01
5.60443044e-01 8.23377311e-01 3.38499755e-01 -1.08985484e+00
-1.10033166e+00 -1.51458883e+00 -7.14654326e-01 -3.64621729e-02
3.79756570e-01 -2.54125744e-01 -2.89201558e-01 2.40477711e-01] | [15.569672584533691, 5.882933139801025] |
d548c746-0e46-4f1c-b033-e3702ca52330 | directional-diffusion-models-for-graph | 2306.13210 | null | https://arxiv.org/abs/2306.13210v1 | https://arxiv.org/pdf/2306.13210v1.pdf | Directional diffusion models for graph representation learning | In recent years, diffusion models have achieved remarkable success in various domains of artificial intelligence, such as image synthesis, super-resolution, and 3D molecule generation. However, the application of diffusion models in graph learning has received relatively little attention. In this paper, we address this gap by investigating the use of diffusion models for unsupervised graph representation learning. We begin by identifying the anisotropic structures of graphs and a crucial limitation of the vanilla forward diffusion process in learning anisotropic structures. This process relies on continuously adding an isotropic Gaussian noise to the data, which may convert the anisotropic signals to noise too quickly. This rapid conversion hampers the training of denoising neural networks and impedes the acquisition of semantically meaningful representations in the reverse process. To address this challenge, we propose a new class of models called {\it directional diffusion models}. These models incorporate data-dependent, anisotropic, and directional noises in the forward diffusion process. To assess the efficacy of our proposed models, we conduct extensive experiments on 12 publicly available datasets, focusing on two distinct graph representation learning tasks. The experimental results demonstrate the superiority of our models over state-of-the-art baselines, indicating their effectiveness in capturing meaningful graph representations. Our studies not only provide valuable insights into the forward process of diffusion models but also highlight the wide-ranging potential of these models for various graph-related tasks. | ['Qiang Sun', 'Fan Zhou', 'Yuling Yang', 'Run Yang'] | 2023-06-22 | null | null | null | null | ['super-resolution', 'graph-learning', '3d-molecule-generation', 'graph-representation-learning'] | ['computer-vision', 'graphs', 'medical', 'methodology'] | [ 4.34431255e-01 8.35595652e-02 -1.11358322e-01 -3.68423201e-02
-3.56245041e-01 -4.59984273e-01 7.26065159e-01 1.56209975e-01
-1.99971527e-01 3.02428693e-01 5.35334587e-01 -2.47107074e-01
-1.25018567e-01 -9.90778327e-01 -4.02832121e-01 -9.16403890e-01
2.38556657e-02 3.19463253e-01 2.63630152e-01 -2.08568409e-01
1.79122075e-01 6.00020945e-01 -8.70067716e-01 2.63557136e-01
1.00323772e+00 7.68211544e-01 2.20849529e-01 4.55603480e-01
-4.40445155e-01 9.63906169e-01 -5.65639973e-01 -3.31597537e-01
1.18264221e-01 -4.98288244e-01 -6.01626337e-01 1.88566744e-01
2.82659531e-01 -2.75554866e-01 -7.62152374e-01 1.28821623e+00
6.44757330e-01 2.48871937e-01 9.09101069e-01 -8.51954341e-01
-1.46668553e+00 6.11418366e-01 -8.12372267e-01 4.34146643e-01
1.29105002e-01 1.44416481e-01 1.04569829e+00 -7.15842247e-01
1.03690755e+00 1.30857611e+00 5.95864713e-01 6.12954021e-01
-1.38395917e+00 -5.84201276e-01 4.12647218e-01 1.06876634e-01
-1.12900901e+00 -3.50474626e-01 1.33360338e+00 -5.55182219e-01
8.29039097e-01 -1.10026166e-01 5.10405898e-01 1.32255936e+00
1.33491620e-01 8.52042079e-01 1.21452928e+00 -1.94728658e-01
1.63586169e-01 -4.79709655e-01 3.74162495e-02 8.51835608e-01
4.06901151e-01 9.25132912e-03 -4.81119573e-01 -4.53870445e-02
1.07941926e+00 -9.42221060e-02 -3.02744567e-01 -3.94291341e-01
-1.11526346e+00 9.73715246e-01 8.98448765e-01 4.36354727e-01
-5.42374671e-01 2.91031837e-01 2.05151349e-01 1.58387646e-01
9.39029872e-01 2.92795748e-01 2.23017231e-01 2.16889620e-01
-7.01427102e-01 1.33956801e-02 4.11782593e-01 8.63285363e-01
3.99372935e-01 3.25549543e-01 -2.10728422e-01 8.22585344e-01
3.96781325e-01 2.86461502e-01 3.74284118e-01 -7.04880297e-01
5.82836390e-01 7.11286306e-01 -3.82056743e-01 -1.31774819e+00
-4.10611331e-01 -5.60102701e-01 -1.36261415e+00 -1.24384247e-01
3.70689005e-01 1.73629597e-01 -1.18785739e+00 1.58147788e+00
2.68862993e-01 3.37537438e-01 1.31500224e-02 8.87498081e-01
1.13672459e+00 6.24658823e-01 2.37079337e-01 -9.45915505e-02
8.79120290e-01 -1.00088668e+00 -9.13036346e-01 -1.52906731e-01
5.90542436e-01 -7.27875590e-01 9.22534943e-01 2.74246573e-01
-1.08498478e+00 -4.80590254e-01 -9.72610593e-01 -3.36337447e-01
-3.01385880e-01 -4.54834253e-02 1.02508235e+00 4.22428966e-01
-1.08769834e+00 7.09309042e-01 -8.78502309e-01 -2.83752769e-01
7.32721388e-01 7.94411153e-02 -3.07431757e-01 -4.87687647e-01
-1.00703323e+00 5.07560015e-01 -1.16700053e-01 2.01377392e-01
-7.84901917e-01 -4.17016298e-01 -7.76097834e-01 -4.94308658e-02
2.90742636e-01 -9.17372108e-01 5.57832122e-01 -8.06523800e-01
-1.41747642e+00 6.76751256e-01 -1.57684416e-01 -3.90990824e-01
5.00922978e-01 2.17167716e-02 -3.42222184e-01 2.86659867e-01
3.67881805e-02 3.95320654e-01 1.06795490e+00 -1.32540429e+00
2.10572388e-02 -5.38871169e-01 6.08274229e-02 1.34401336e-01
-5.19401848e-01 -2.53974468e-01 -5.61971962e-01 -1.34608173e+00
4.57028508e-01 -8.71728122e-01 -5.32226264e-01 5.90622276e-02
-3.38082552e-01 -1.58056706e-01 6.71931386e-01 -5.94099104e-01
1.24608409e+00 -2.11373067e+00 5.35228968e-01 2.57701933e-01
7.84261882e-01 4.50133160e-02 -5.35430372e-01 3.92400086e-01
7.83998817e-02 3.45222682e-01 -4.12066340e-01 -3.98289353e-01
-3.24694663e-01 2.61393636e-01 -2.78622985e-01 3.17994803e-01
2.00856894e-01 1.13754690e+00 -9.99499798e-01 -3.32587808e-01
-1.71232764e-02 9.04641271e-01 -4.62107062e-01 1.83056742e-01
-1.40986562e-01 6.59775794e-01 -7.04552174e-01 4.81886476e-01
6.86440349e-01 -4.97336179e-01 2.41170093e-01 -4.37485069e-01
2.92583108e-01 2.02780038e-01 -9.83315587e-01 1.83823812e+00
-1.08819723e-01 3.16068858e-01 4.70471121e-02 -1.20253563e+00
9.64715004e-01 -2.01375913e-02 5.50698698e-01 -8.22855055e-01
-1.11055531e-01 1.13104038e-01 3.93511960e-03 -3.80311549e-01
3.61871421e-01 -1.89690247e-01 4.46600407e-01 5.49072325e-01
1.69530641e-02 -2.26229534e-01 2.30760455e-01 5.14250934e-01
1.15025032e+00 1.01324350e-01 -5.03218621e-02 -1.84658721e-01
3.16205859e-01 -2.78321713e-01 3.09436202e-01 6.97360992e-01
-1.39882475e-01 7.61744976e-01 4.59238976e-01 -3.54959071e-01
-6.99169874e-01 -1.01913190e+00 4.01453614e-01 8.57160330e-01
1.96204245e-01 -5.40109694e-01 -6.22218013e-01 -9.44707692e-01
-1.42396972e-01 3.73363644e-01 -8.10769618e-01 -4.19728249e-01
-6.00006461e-01 -1.03126001e+00 5.06650150e-01 6.71938062e-01
6.18460953e-01 -8.07216227e-01 3.11793089e-01 1.54427797e-01
-2.07314044e-01 -1.25757170e+00 -6.48331940e-01 -6.30423278e-02
-1.31125104e+00 -8.98678005e-01 -8.87272477e-01 -8.12509239e-01
1.00398493e+00 8.27863634e-01 1.03566110e+00 2.68236816e-01
-5.98740391e-02 3.76046449e-01 -3.08544755e-01 -4.60995026e-02
-4.43430871e-01 1.82924017e-01 -2.36990497e-01 1.60371393e-01
3.78883481e-02 -7.05067337e-01 -7.52977431e-01 1.45597190e-01
-1.20339203e+00 2.68796645e-02 7.35149086e-01 7.15139031e-01
8.61352801e-01 1.49799854e-01 5.64348340e-01 -1.20188928e+00
1.24962282e+00 -4.66722071e-01 -3.22564960e-01 7.96930939e-02
-7.01478601e-01 3.05787623e-01 6.44137084e-01 -5.57464778e-01
-9.68022704e-01 -1.70477703e-01 -1.92490384e-01 -4.58269149e-01
2.07057565e-01 8.47939014e-01 -1.18741862e-01 -3.58644754e-01
6.19902730e-01 2.87563562e-01 1.16427876e-01 -5.59889615e-01
6.77234173e-01 1.20459199e-01 1.76545888e-01 -6.01451337e-01
8.90142977e-01 7.47273326e-01 2.72792071e-01 -1.09543741e+00
-7.41775751e-01 -3.64205569e-01 -5.04058301e-01 -8.10426176e-02
8.23814929e-01 -7.78256059e-01 -9.64726955e-02 5.38124740e-01
-9.84956086e-01 -3.50066811e-01 -3.20197642e-01 1.86247528e-01
-3.26202780e-01 8.53432953e-01 -8.33114743e-01 -4.98450726e-01
-3.47687155e-01 -1.33215141e+00 9.50707436e-01 -1.25968335e-02
-3.19048092e-02 -1.31597674e+00 -3.91378738e-02 4.44267422e-01
6.08689010e-01 2.57226497e-01 1.15989387e+00 -2.87424713e-01
-7.06048012e-01 -1.27608716e-01 -3.39955956e-01 2.91379213e-01
2.01954037e-01 -1.89939708e-01 -7.71400332e-01 -2.69636363e-01
-9.16334018e-02 -2.16971487e-01 1.40535700e+00 5.24925411e-01
1.25974536e+00 -6.98209181e-02 -3.39180946e-01 8.19052756e-01
1.02085686e+00 -1.54332206e-01 6.09777927e-01 9.75070894e-02
1.08874679e+00 5.39548397e-01 5.77006042e-02 -2.30669621e-02
5.18243968e-01 5.06948352e-01 3.18348169e-01 -3.11740190e-01
-7.40591764e-01 -4.32106167e-01 2.18686074e-01 1.26345265e+00
-5.36369443e-01 -3.84701818e-01 -6.45219266e-01 3.66263121e-01
-1.61018276e+00 -7.97676384e-01 -1.80653065e-01 1.88678384e+00
5.44723928e-01 1.79597363e-01 -5.85356727e-02 4.38305326e-02
5.38533032e-01 7.09851146e-01 -5.38186789e-01 1.74572261e-03
-3.95460308e-01 1.79207146e-01 2.92370081e-01 4.70798463e-01
-1.04396713e+00 1.14104521e+00 6.30514097e+00 7.22979248e-01
-1.11918676e+00 -5.04802503e-02 7.21590161e-01 3.35618019e-01
-7.03452647e-01 -1.11248940e-01 -3.69017452e-01 2.25446403e-01
4.76847559e-01 4.41581272e-02 6.23753369e-01 5.69455862e-01
2.04795301e-01 4.22674209e-01 -7.95490921e-01 9.29341137e-01
1.11993842e-01 -1.51790786e+00 6.22307599e-01 2.79012322e-01
9.52630699e-01 5.93505763e-02 2.69208431e-01 3.31596397e-02
5.21954060e-01 -1.08226442e+00 3.82064849e-01 4.85895157e-01
5.22068560e-01 -4.65927243e-01 3.68321657e-01 6.16055317e-02
-1.36923063e+00 1.79500446e-01 -5.73371828e-01 1.53395995e-01
1.34506255e-01 9.69436884e-01 -5.16585052e-01 7.11587787e-01
3.21439147e-01 1.07849371e+00 -6.03993475e-01 7.49926686e-01
-4.71605182e-01 7.86737680e-01 -1.74655497e-01 1.66189268e-01
1.76803872e-01 -5.90431809e-01 5.33062220e-01 1.09909165e+00
3.47755492e-01 1.47400707e-01 1.06287844e-01 1.04426336e+00
-5.81034839e-01 1.60532743e-01 -8.38723719e-01 -5.36071181e-01
7.18200430e-02 1.12531209e+00 -1.04695272e+00 -1.83456957e-01
-4.52426523e-01 1.12572420e+00 5.96194088e-01 8.40659440e-01
-6.45370543e-01 -6.65962622e-02 3.53549719e-01 9.22127217e-02
1.73409566e-01 -7.64434636e-01 -3.88614029e-01 -1.28336835e+00
-3.62843238e-02 -8.50044310e-01 4.90652531e-01 -4.47605610e-01
-1.62822533e+00 5.60086250e-01 -3.18806678e-01 -7.64897764e-01
2.07116649e-01 -5.31048417e-01 -5.66339850e-01 6.80411637e-01
-1.71501863e+00 -1.35917294e+00 -5.14440715e-01 4.92873937e-01
5.30529857e-01 2.04174258e-02 5.21306515e-01 2.98956960e-01
-4.64721054e-01 3.54898363e-01 8.10240582e-02 1.15768924e-01
5.39328277e-01 -1.18861246e+00 8.47306848e-01 8.67124438e-01
3.71369570e-01 8.01002920e-01 3.37202042e-01 -9.87199783e-01
-1.64770031e+00 -1.16202998e+00 5.63502550e-01 -3.33120316e-01
7.83298016e-01 -3.16457391e-01 -1.13200331e+00 4.66885656e-01
-1.85542807e-01 2.30635270e-01 5.17547071e-01 -5.15309088e-02
-4.81031090e-01 9.41765606e-02 -7.78909624e-01 7.05846667e-01
1.57434177e+00 -5.54194093e-01 -1.62102282e-01 3.80155236e-01
6.17288947e-01 -1.66768581e-01 -8.42695713e-01 3.48951489e-01
3.05020452e-01 -8.24956059e-01 1.30508828e+00 -6.86500371e-01
4.89503801e-01 -1.24886684e-01 5.98874949e-02 -1.49843919e+00
-6.17534935e-01 -5.99080801e-01 -4.85021561e-01 1.13082564e+00
3.20227712e-01 -6.17402375e-01 8.07303607e-01 3.08382869e-01
-7.06015378e-02 -7.71544278e-01 -6.78445458e-01 -6.46959960e-01
3.38040978e-01 -2.85412520e-01 4.02816176e-01 1.06880176e+00
-5.70584238e-01 5.30002117e-01 -1.67357951e-01 -6.66337507e-03
7.34787107e-01 7.00341538e-02 7.03832328e-01 -1.24796355e+00
-2.71577656e-01 -6.86021090e-01 -3.15832257e-01 -1.50411570e+00
1.53735757e-01 -1.16266489e+00 -3.31235230e-01 -2.08435822e+00
8.38784128e-02 -3.37994576e-01 -2.55652487e-01 7.81885833e-02
-4.71278936e-01 2.40048096e-01 2.96154507e-02 4.88653153e-01
-4.52848792e-01 8.19801688e-01 1.74481833e+00 -5.45028806e-01
-9.28585529e-02 -1.77807406e-01 -9.06855464e-01 7.27155149e-01
6.37114346e-01 -2.98410028e-01 -7.40099311e-01 -8.34156513e-01
2.11856335e-01 -3.56565267e-01 4.01431136e-02 -6.81097865e-01
1.13592036e-01 -8.83160233e-02 3.24615806e-01 6.07069805e-02
1.62428156e-01 -6.75987780e-01 -1.62104353e-01 2.94895649e-01
-3.19572568e-01 1.16929494e-01 -1.05749927e-01 1.03546286e+00
-9.05330703e-02 1.80547424e-02 6.88336909e-01 -2.88467139e-01
-5.88283062e-01 5.69161892e-01 -1.97716206e-01 1.82008445e-01
6.91673040e-01 -2.26828143e-01 -3.84113669e-01 -5.63919008e-01
-7.19030023e-01 -1.39127880e-01 4.92337912e-01 4.74646121e-01
9.21705782e-01 -1.24664092e+00 -6.42174840e-01 2.19190806e-01
-6.63943449e-03 1.37659088e-01 2.06537291e-01 8.45730484e-01
-4.80340540e-01 -2.37915944e-02 1.07960202e-01 -3.78763974e-01
-8.47152710e-01 7.41612375e-01 2.11731270e-01 -5.30666471e-01
-9.32002664e-01 9.02752757e-01 5.35540819e-01 -1.34180367e-01
1.50960207e-01 -4.64639187e-01 -3.18915188e-01 -7.27270544e-02
2.79635996e-01 4.10070032e-01 -8.18457082e-02 -6.50999725e-01
-8.30675066e-02 6.92884147e-01 -2.03306943e-01 2.01469317e-01
1.46168721e+00 -1.38867095e-01 -2.22929984e-01 1.24850281e-01
1.00020564e+00 1.26202151e-01 -1.17060375e+00 -2.58783758e-01
2.72332933e-02 -2.68977821e-01 3.25601727e-01 -4.65808839e-01
-1.43685651e+00 1.02334535e+00 3.89050990e-01 3.29449058e-01
1.05669403e+00 3.54516469e-02 9.34095740e-01 1.76725566e-01
1.52791694e-01 -9.00772512e-01 5.11076152e-01 4.26075429e-01
9.96410489e-01 -1.15320742e+00 1.84398875e-01 -7.76137114e-01
-6.32255793e-01 9.77568269e-01 3.67481858e-01 -2.04799995e-01
5.66809773e-01 6.40805960e-02 6.90434352e-02 -6.00582063e-01
-3.95528942e-01 -3.37253422e-01 6.01360559e-01 8.93442273e-01
5.94475746e-01 -1.34416476e-01 -2.77536452e-01 4.39498693e-01
1.96594343e-01 -1.10356227e-01 4.30529416e-01 8.77075016e-01
-1.83052838e-01 -1.06399691e+00 3.83587293e-02 3.54015291e-01
-3.03050220e-01 -1.70762971e-01 -8.61159742e-01 4.22905564e-01
-3.36243331e-01 8.95578623e-01 -4.19962853e-01 -2.01504916e-01
3.65327507e-01 -1.67297617e-01 6.04921937e-01 -3.99417341e-01
-2.31257424e-01 2.26740941e-01 -6.38527423e-03 -4.70584840e-01
-5.93196273e-01 -2.75931984e-01 -1.21990895e+00 -1.47594005e-01
-2.80142128e-01 9.04063229e-04 4.03191954e-01 6.69461012e-01
6.11770272e-01 8.13618243e-01 3.34943056e-01 -8.64905715e-01
-4.94957030e-01 -9.36851919e-01 -4.33165610e-01 8.20542276e-01
1.44081205e-01 -8.62983167e-01 -3.43862563e-01 4.76641767e-02] | [6.983514785766602, 6.094307899475098] |
5cfc52a0-09c9-49f0-8243-1ab1a1d201df | self-supervised-learning-by-view-synthesis | 2304.11330 | null | https://arxiv.org/abs/2304.11330v1 | https://arxiv.org/pdf/2304.11330v1.pdf | Self-supervised Learning by View Synthesis | We present view-synthesis autoencoders (VSA) in this paper, which is a self-supervised learning framework designed for vision transformers. Different from traditional 2D pretraining methods, VSA can be pre-trained with multi-view data. In each iteration, the input to VSA is one view (or multiple views) of a 3D object and the output is a synthesized image in another target pose. The decoder of VSA has several cross-attention blocks, which use the source view as value, source pose as key, and target pose as query. They achieve cross-attention to synthesize the target view. This simple approach realizes large-angle view synthesis and learns spatial invariant representation, where the latter is decent initialization for transformers on downstream tasks, such as 3D classification on ModelNet40, ShapeNet Core55, and ScanObjectNN. VSA outperforms existing methods significantly for linear probing and is competitive for fine-tuning. The code will be made publicly available. | ['Jiaya Jia', 'Tao Hu', 'Xiangyu Zhang', 'Shaoteng Liu'] | 2023-04-22 | null | null | null | null | ['3d-classification'] | ['computer-vision'] | [-1.55115426e-01 2.45447263e-01 -4.53772809e-04 -4.71466959e-01
-7.47650921e-01 -6.63593292e-01 6.41358614e-01 -9.01338458e-01
2.90597863e-02 1.91107348e-01 3.15191984e-01 -4.33651656e-01
3.96901011e-01 -8.88196349e-01 -1.14444721e+00 -6.87639952e-01
5.53915858e-01 8.90749395e-01 2.81817436e-01 7.43965432e-02
-8.57709497e-02 7.88466156e-01 -1.21376717e+00 3.93243015e-01
1.50725543e-01 1.26391327e+00 6.43469214e-01 7.60110021e-01
9.51065272e-02 6.33673072e-01 -5.48286617e-01 -3.62394959e-01
4.14148331e-01 -2.40295529e-02 -7.64049590e-01 3.02692443e-01
8.26215982e-01 -9.26632464e-01 -5.53329170e-01 7.70888686e-01
9.77480888e-01 -2.02055976e-01 6.84466124e-01 -1.17844319e+00
-8.68297338e-01 5.17332613e-01 -6.39187217e-01 1.01384278e-02
2.46021062e-01 3.82767260e-01 1.06376469e+00 -1.46782470e+00
7.18706131e-01 1.59535861e+00 2.79534489e-01 6.99222267e-01
-1.01657760e+00 -6.51602745e-01 3.20867151e-01 1.34899244e-01
-7.84961820e-01 -4.94853616e-01 6.83733582e-01 -2.84803748e-01
1.38873327e+00 -1.89535737e-01 9.86823082e-01 1.59967375e+00
1.48563772e-01 1.18100679e+00 8.68036866e-01 -3.73132415e-02
3.39476168e-02 -2.46050403e-01 -3.52231622e-01 8.08797002e-01
-1.90370262e-01 3.48801792e-01 -5.37376165e-01 2.45059893e-01
1.28874695e+00 1.08629249e-01 -2.91040540e-01 -1.03090370e+00
-1.49188614e+00 7.25170970e-01 7.34734297e-01 -2.07354203e-01
-3.70043337e-01 2.33152494e-01 1.56311259e-01 4.82254863e-01
3.07394028e-01 2.85258770e-01 -6.64767683e-01 1.00907639e-01
-3.68982017e-01 8.66453424e-02 5.22586465e-01 1.25698507e+00
5.10679185e-01 4.48924035e-01 -5.88384308e-02 6.69044614e-01
3.93564165e-01 1.07016826e+00 2.90026605e-01 -9.33797061e-01
6.69559479e-01 6.20791972e-01 -1.09623305e-01 -4.36676025e-01
-2.66326666e-01 -5.79967201e-01 -7.27223814e-01 4.51279879e-01
1.09703638e-01 -8.56424645e-02 -1.32662439e+00 1.64306819e+00
5.36242008e-01 3.69385295e-02 2.02722296e-01 1.02535856e+00
1.46382725e+00 6.81284249e-01 -5.37393272e-01 2.65180379e-01
1.14737618e+00 -1.42559564e+00 -1.98649004e-01 -3.33948910e-01
-9.45942476e-03 -7.44916856e-01 1.12510693e+00 3.77458811e-01
-1.48355842e+00 -7.45071054e-01 -8.72930229e-01 -4.83658880e-01
-1.52409703e-01 2.97076821e-01 5.74968696e-01 3.00209299e-02
-1.29888403e+00 9.99027938e-02 -9.61469173e-01 -3.38542849e-01
5.30398309e-01 2.36763671e-01 -5.02528727e-01 -1.83065176e-01
-6.84840858e-01 7.95107663e-01 1.32789332e-02 -1.80644938e-03
-1.74276543e+00 -8.64249945e-01 -1.08065152e+00 1.49670944e-01
2.53913701e-01 -1.17195070e+00 1.52695262e+00 -6.00629151e-01
-1.78952277e+00 1.32534432e+00 -3.44659649e-02 -2.40119740e-01
4.57128227e-01 -1.50037527e-01 -1.44489065e-01 1.89798966e-01
3.35391641e-01 1.07341683e+00 1.32557535e+00 -1.41916239e+00
-2.96740174e-01 -5.91365874e-01 2.99141318e-01 5.00725150e-01
1.39940798e-01 -4.03052002e-01 -9.04519677e-01 -4.91655856e-01
4.37731177e-01 -9.40129340e-01 2.55485363e-02 3.16726297e-01
-6.01376534e-01 -9.98020470e-02 9.72503364e-01 -2.47863814e-01
3.04405451e-01 -2.17575622e+00 5.98065495e-01 -3.57760005e-02
3.72512251e-01 6.00361228e-02 -2.15216428e-01 3.02777380e-01
-4.08672512e-01 -5.50638914e-01 1.82233647e-01 -3.75800073e-01
-1.34436399e-01 2.03054845e-01 -5.98015308e-01 3.74178648e-01
8.60298723e-02 1.23847759e+00 -6.67125404e-01 -1.23464808e-01
5.17972767e-01 3.61664653e-01 -6.97069049e-01 6.25195563e-01
-4.14418340e-01 3.95034373e-01 -5.59314609e-01 7.89528430e-01
7.23396361e-01 -7.70595849e-01 -1.84848741e-01 -6.57770395e-01
1.56368595e-02 3.28015983e-01 -8.16567838e-01 2.07549858e+00
-7.49458492e-01 4.36299145e-01 5.95149957e-03 -8.34713042e-01
8.27698946e-01 1.43206343e-01 3.02477747e-01 -6.47814095e-01
1.41902909e-01 -1.35411888e-01 -2.83579439e-01 -5.49105823e-01
-4.31880215e-03 1.62785858e-01 9.27619189e-02 6.10484898e-01
4.88513917e-01 -5.61720014e-01 -4.08207566e-01 1.29710838e-01
8.13095450e-01 4.42414165e-01 2.93045878e-01 1.54547334e-01
3.11964929e-01 -2.33610064e-01 8.16395357e-02 3.93457413e-01
1.39328599e-01 9.39445376e-01 4.30284351e-01 -5.72843194e-01
-1.10987461e+00 -1.59296405e+00 3.44364308e-02 8.76693368e-01
1.54335529e-01 -1.26782835e-01 -4.41711575e-01 -8.41722667e-01
2.29436904e-01 5.45768380e-01 -5.87251425e-01 -1.30859941e-01
-5.02896309e-01 4.79658358e-02 1.82206005e-01 9.61215913e-01
7.26344466e-01 -1.07706201e+00 -7.45594442e-01 -1.23561963e-01
-9.11690071e-02 -1.24756956e+00 -6.48736537e-01 4.13249433e-01
-1.02130532e+00 -1.11442876e+00 -1.00286961e+00 -9.99446929e-01
8.24019313e-01 6.52092338e-01 1.39958215e+00 -4.17478979e-01
1.53173000e-01 6.81937218e-01 -2.48489007e-02 -4.57792073e-01
-5.59904911e-02 9.16893557e-02 -8.04984476e-03 -1.68433353e-01
1.51921466e-01 -8.68566275e-01 -7.64199674e-01 4.59418952e-01
-4.25726891e-01 3.27055007e-01 8.43246222e-01 9.94600952e-01
8.60822260e-01 -8.28668773e-01 7.86917135e-02 -5.78601062e-01
1.64099708e-01 -1.05405726e-01 -6.84747994e-01 2.07838535e-01
-1.64399996e-01 4.52285111e-02 5.54142892e-01 -2.93149680e-01
-8.09150577e-01 2.43386924e-01 -4.47368294e-01 -1.29344833e+00
-1.44491404e-01 8.63546215e-04 -5.36868513e-01 -2.69181803e-02
4.82812107e-01 4.79191154e-01 1.93035036e-01 -4.72118497e-01
5.19391894e-01 2.64155328e-01 4.08516556e-01 -3.25457305e-01
1.01046860e+00 6.25970900e-01 -2.08285809e-01 -5.77669024e-01
-9.44156468e-01 -1.15678497e-01 -6.96924508e-01 -2.31440648e-01
1.08567750e+00 -1.38257742e+00 -9.07727778e-01 6.49600685e-01
-1.09810030e+00 -5.88195860e-01 -3.15650374e-01 4.62205261e-01
-8.24222982e-01 -1.78217128e-01 -3.41818988e-01 -2.95983814e-02
-3.79788041e-01 -1.53931952e+00 1.72334194e+00 1.34388879e-01
2.37057731e-01 -8.25456977e-01 -2.47617185e-01 4.62261647e-01
1.90889776e-01 -2.31950413e-02 8.70214581e-01 -3.41498166e-01
-1.00215602e+00 1.46928474e-01 -2.83410728e-01 3.44436884e-01
2.18069069e-02 -2.06621215e-01 -1.24287486e+00 -6.40356183e-01
3.24394852e-02 -8.68104994e-01 8.33060026e-01 7.62329996e-01
1.46473706e+00 -2.25680117e-02 -3.93679678e-01 1.13512051e+00
1.14041901e+00 1.38158992e-01 3.43630999e-01 -2.40516234e-02
8.23036313e-01 6.59625083e-02 1.82880893e-01 2.70222843e-01
7.38541543e-01 6.34882569e-01 9.96535957e-01 -3.39925170e-01
-3.26972425e-01 -5.12269318e-01 4.69349295e-01 8.52745771e-01
-3.33941244e-02 -2.30941355e-01 -7.26720929e-01 2.98791677e-01
-1.32341111e+00 -1.07703376e+00 3.91361147e-01 1.83652985e+00
4.04347152e-01 3.40644270e-01 9.71272364e-02 -2.79574215e-01
2.14854345e-01 4.65740025e-01 -1.15782785e+00 -1.74457490e-01
2.45854631e-02 1.84109256e-01 2.19701096e-01 5.00726879e-01
-9.06253755e-01 1.03747725e+00 6.19920731e+00 1.97098494e-01
-1.51108491e+00 -8.66816565e-02 4.54915017e-01 -2.46563390e-01
-5.13655305e-01 -3.39332908e-01 -8.07429433e-01 4.10092175e-02
2.51090169e-01 3.61367553e-01 4.67827260e-01 1.19638860e+00
-1.76689774e-01 2.36008331e-01 -1.47359037e+00 1.22256422e+00
4.65326369e-01 -1.31406498e+00 3.76569211e-01 -3.09224278e-02
6.25916064e-01 4.93079364e-01 4.99826580e-01 4.25079018e-01
6.01349354e-01 -9.16519523e-01 7.00432181e-01 3.00208569e-01
1.18255198e+00 -4.95646894e-01 1.45806849e-01 2.71826565e-01
-1.15331841e+00 -5.71356788e-02 -4.58117038e-01 2.69843012e-01
1.96288198e-01 1.55284658e-01 -8.64313006e-01 4.02928442e-01
9.42611933e-01 1.07572091e+00 -4.07039464e-01 6.19811058e-01
-4.66599286e-01 2.64337271e-01 -2.21392378e-01 9.25425738e-02
4.70588088e-01 3.69783342e-02 6.82119071e-01 4.79992896e-01
2.45424747e-01 -7.56005049e-02 5.85457645e-02 9.05594409e-01
-2.20876873e-01 -4.45298880e-01 -9.79490936e-01 2.85042256e-01
3.09676379e-01 1.10135937e+00 -3.56282920e-01 -3.63649338e-01
-6.53512120e-01 1.20554984e+00 4.01598573e-01 4.84472036e-01
-9.09463942e-01 -8.84852111e-02 6.22994006e-01 7.39397155e-03
9.07339156e-01 -1.35758072e-01 4.05437034e-03 -1.44556081e+00
1.23473704e-02 -1.02088308e+00 2.08237633e-01 -1.56229162e+00
-1.13264978e+00 7.28772998e-01 -8.19470175e-03 -1.63988411e+00
-2.94072449e-01 -9.58832264e-01 -6.19132757e-01 7.10738003e-01
-1.28610933e+00 -1.59721220e+00 -5.82008421e-01 8.81657660e-01
9.24865246e-01 -6.04276776e-01 9.01634991e-01 9.53530446e-02
-2.35139355e-01 6.53860450e-01 -2.87474334e-01 1.09860539e-01
6.91051364e-01 -1.40248656e+00 9.62818742e-01 4.62845981e-01
3.97170693e-01 3.87763560e-01 2.67101049e-01 -3.01158369e-01
-1.67154908e+00 -1.07149971e+00 5.48926771e-01 -7.01690972e-01
2.84016997e-01 -5.70233583e-01 -3.89786512e-01 1.17754531e+00
4.62727636e-01 2.91810662e-01 3.42754960e-01 -5.33554666e-02
-5.87265670e-01 -2.21572936e-01 -8.51626992e-01 6.54474735e-01
1.34393764e+00 -7.30113566e-01 -6.77066922e-01 3.51661205e-01
9.38726366e-01 -1.22368813e+00 -6.41978204e-01 2.56578982e-01
6.85310721e-01 -1.14112318e+00 1.41101718e+00 -7.73540795e-01
7.43441105e-01 -1.82389945e-01 -2.76092291e-01 -1.57485461e+00
-3.37756634e-01 -1.84451148e-01 -2.79240698e-01 5.57649910e-01
4.03159827e-01 -4.25727725e-01 7.49962628e-01 -2.30825752e-01
-5.67315280e-01 -1.12198699e+00 -7.15062618e-01 -4.77864355e-01
-4.29784134e-03 -1.79679096e-01 7.71604776e-01 6.39537692e-01
-7.20228434e-01 8.48130643e-01 -2.45730877e-01 4.15129721e-01
6.66661739e-01 6.92197800e-01 1.11109674e+00 -9.57073748e-01
-5.92857599e-01 -2.73762435e-01 -1.21501230e-01 -1.85803187e+00
5.87048642e-02 -1.02973843e+00 -3.21552813e-01 -1.65013361e+00
-4.60561663e-02 1.96912602e-04 2.28817418e-01 4.83297944e-01
2.61323750e-01 9.50063542e-02 1.39940068e-01 -4.23646113e-03
-2.46825293e-01 6.66188717e-01 2.03921556e+00 -3.62670213e-01
-1.00902930e-01 3.38297248e-01 -4.76662099e-01 7.18523681e-01
4.06369895e-01 1.81197152e-02 -8.56744707e-01 -1.03089499e+00
-5.29524125e-02 4.70150828e-01 6.40082538e-01 -8.19751143e-01
7.16492906e-02 8.94741267e-02 1.02934396e+00 -1.18847835e+00
6.53087080e-01 -9.98277783e-01 3.32227238e-02 3.44946265e-01
-2.34001517e-01 4.19829279e-01 2.61007738e-03 5.28422892e-01
-7.22995996e-02 2.74444282e-01 7.83044577e-01 -4.87304926e-01
-6.37833834e-01 8.86744142e-01 1.20957896e-01 2.93813776e-02
9.36164439e-01 -3.11851174e-01 -2.24757835e-01 -5.74951649e-01
-9.82648551e-01 4.38425660e-01 4.51197594e-01 5.37452579e-01
9.51729178e-01 -1.59168172e+00 -4.05939639e-01 5.57277799e-01
1.69441566e-01 6.58040285e-01 2.99910039e-01 5.33359647e-01
-4.10514712e-01 3.50748152e-01 -4.09008116e-01 -1.05541670e+00
-9.81558800e-01 7.04760373e-01 6.24363124e-01 9.34526622e-02
-9.55605805e-01 1.28349555e+00 7.07211792e-01 -6.86486304e-01
4.88297939e-01 -5.48042357e-01 -1.08130038e-01 -7.51801953e-02
3.33715200e-01 -1.95608586e-01 -9.49496776e-02 -3.71367693e-01
-2.48032779e-01 8.07328999e-01 -1.00573458e-01 -7.43954331e-02
1.63738573e+00 2.88209673e-02 1.74453393e-01 4.55450356e-01
1.32970774e+00 -2.28796244e-01 -1.67400944e+00 -3.91959667e-01
-8.48706007e-01 -2.27852792e-01 -4.58779819e-02 -7.77764082e-01
-1.48933566e+00 1.22847426e+00 4.89124626e-01 -2.03602195e-01
1.03903198e+00 3.05565536e-01 6.62315905e-01 6.65576398e-01
3.25036526e-01 -5.98303378e-01 6.23199642e-01 7.07141161e-01
1.42118466e+00 -1.23416436e+00 -2.04396293e-01 -4.75328378e-02
-8.04791391e-01 1.11193395e+00 1.20506680e+00 -2.27565929e-01
7.74572670e-01 3.82514924e-01 1.30886957e-01 -5.36613584e-01
-1.02947903e+00 -1.60162866e-01 1.59905404e-01 8.35253060e-01
1.24613550e-02 -1.96412906e-01 8.27590585e-01 2.87939429e-01
-3.53157759e-01 -4.77288455e-01 8.47804621e-02 5.58651984e-01
-7.60387853e-02 -7.00724840e-01 -6.02926873e-02 3.69089812e-01
8.24256837e-02 -2.41349172e-02 -4.25764233e-01 8.77289653e-01
-2.07814142e-01 1.83257788e-01 2.88625062e-01 -5.10163009e-01
7.49165773e-01 -9.35021937e-02 8.26839983e-01 -7.37793446e-01
-3.73205692e-01 1.74566865e-01 -1.41559020e-01 -8.00165355e-01
-3.81590664e-01 -5.04625738e-01 -8.88620555e-01 -1.73999771e-01
-2.06934512e-01 -3.80440086e-01 3.24900270e-01 7.18661308e-01
4.16151464e-01 6.40538931e-01 9.90131378e-01 -1.11976171e+00
-7.55855978e-01 -8.32010150e-01 -9.73762423e-02 5.19749746e-02
5.71854651e-01 -9.31271255e-01 -9.89463329e-02 5.32725919e-03] | [8.229355812072754, -3.3811769485473633] |
4a4369a4-5634-4214-89b8-a458b2cca0ab | graph-neural-networks-for-cross-camera-data | 2201.06311 | null | https://arxiv.org/abs/2201.06311v1 | https://arxiv.org/pdf/2201.06311v1.pdf | Graph Neural Networks for Cross-Camera Data Association | Cross-camera image data association is essential for many multi-camera computer vision tasks, such as multi-camera pedestrian detection, multi-camera multi-target tracking, 3D pose estimation, etc. This association task is typically stated as a bipartite graph matching problem and often solved by applying minimum-cost flow techniques, which may be computationally inefficient with large data. Furthermore, cameras are usually treated by pairs, obtaining local solutions, rather than finding a global solution at once. Other key issue is that of the affinity measurement: the widespread usage of non-learnable pre-defined distances, such as the Euclidean and Cosine ones. This paper proposes an efficient approach for cross-cameras data-association focused on a global solution, instead of processing cameras by pairs. To avoid the usage of fixed distances, we leverage the connectivity of Graph Neural Networks, previously unused in this scope, using a Message Passing Network to jointly learn features and similarity. We validate the proposal for pedestrian multi-view association, showing results over the EPFL multi-camera pedestrian dataset. Our approach considerably outperforms the literature data association techniques, without requiring to be trained in the same scenario in which it is tested. Our code is available at \url{http://www-vpu.eps.uam.es/publications/gnn_cca}. | ['Pablo Carballeira', 'José M. Martínez', 'Juan C. SanMiguel', 'Elena Luna'] | 2022-01-17 | null | null | null | null | ['3d-pose-estimation'] | ['computer-vision'] | [-8.34757984e-02 -3.81858915e-01 1.97423995e-01 -1.08421758e-01
-6.80793762e-01 -7.05716074e-01 5.70037723e-01 5.28948843e-01
-8.07830751e-01 6.19157553e-01 -2.32127219e-01 -1.50281608e-01
-1.63730800e-01 -6.68337107e-01 -8.30713987e-01 -6.42739534e-01
-1.13694198e-01 5.48959851e-01 2.38356471e-01 -1.16966786e-02
1.49909660e-01 6.71845496e-01 -1.60205197e+00 -4.49137837e-02
3.42617422e-01 7.56141961e-01 8.23399425e-02 9.09607530e-01
2.86162585e-01 2.58549452e-01 -2.94848233e-01 -8.90059590e-01
4.01415795e-01 -3.60114016e-02 -5.36663175e-01 3.05673748e-01
9.62429702e-01 -4.14676853e-02 -2.26631269e-01 1.20116043e+00
3.78545612e-01 2.97404498e-01 4.80892241e-01 -1.48937559e+00
-2.29289994e-01 8.51524100e-02 -7.11143851e-01 3.51052225e-01
6.44915164e-01 4.84375749e-03 9.74205971e-01 -7.38740504e-01
5.19048631e-01 1.07163930e+00 8.32719505e-01 2.49228016e-01
-1.28095639e+00 -3.40596616e-01 1.29816439e-02 5.87004125e-01
-1.66735888e+00 -3.59627873e-01 6.71267629e-01 -5.27336419e-01
6.36329830e-01 3.52125525e-01 7.89123893e-01 1.06881940e+00
-6.41893446e-02 4.38228905e-01 9.47484910e-01 -4.20160711e-01
7.70770609e-02 1.48656458e-01 1.60542190e-01 7.01902509e-01
7.02504635e-01 -7.29272515e-02 -1.98546171e-01 -1.01328619e-01
5.55879772e-01 1.27070680e-01 -2.30816618e-01 -8.11765075e-01
-1.04849672e+00 9.21546936e-01 4.86453325e-01 3.91518176e-01
-3.82663570e-02 1.34064823e-01 4.69359249e-01 2.76713639e-01
2.75220126e-01 1.80066049e-01 -2.89308757e-01 1.41299561e-01
-7.95337558e-01 1.85672611e-01 7.93939352e-01 1.04853868e+00
9.91523385e-01 -4.01217252e-01 4.47886467e-01 4.87908125e-01
3.50862354e-01 4.15145487e-01 -1.62337665e-02 -7.12861538e-01
6.37698710e-01 5.04816115e-01 9.22117829e-02 -1.43721473e+00
-4.28365082e-01 -2.80715823e-01 -1.04456508e+00 1.86771840e-01
9.97842014e-01 -1.77534342e-01 -8.09268206e-02 1.55267715e+00
5.27715862e-01 1.83669537e-01 -1.05383106e-01 1.12657869e+00
4.16673303e-01 2.61359781e-01 -5.90331033e-02 -1.15756549e-01
1.48631859e+00 -8.42269480e-01 -1.55208722e-01 -2.48685554e-01
4.88940775e-01 -8.36590350e-01 3.67080510e-01 5.12943566e-01
-8.57893467e-01 -5.95588684e-01 -8.03404391e-01 8.56286958e-02
-7.01197207e-01 2.30471045e-01 4.29618359e-01 9.17324662e-01
-1.11205816e+00 4.44199145e-01 -4.66277540e-01 -8.58983338e-01
2.03471839e-01 6.05869114e-01 -8.47221732e-01 -1.70206100e-01
-8.31215799e-01 1.00176227e+00 5.33143878e-01 2.11598501e-01
-4.66857761e-01 -2.66542435e-01 -8.68623197e-01 9.42506175e-03
5.10098279e-01 -8.72006357e-01 6.16055250e-01 -9.82660472e-01
-1.12968755e+00 8.52999866e-01 1.98507842e-04 -5.51278830e-01
7.21432209e-01 -2.41624564e-01 -3.24620962e-01 1.96742311e-01
9.66475904e-02 6.75760925e-01 7.51473427e-01 -1.12545633e+00
-5.33311129e-01 -4.94465947e-01 2.90983945e-01 1.32660031e-01
-3.69471967e-01 -5.34571409e-02 -6.37660801e-01 -2.55487591e-01
-1.08010665e-01 -1.16795290e+00 -4.71418828e-01 1.11965016e-01
-2.77532071e-01 -7.59479105e-02 6.74385071e-01 -4.72903937e-01
7.48040199e-01 -1.96271455e+00 4.47243392e-01 4.48177487e-01
3.57722044e-01 1.70862600e-01 -3.98018956e-02 3.31713021e-01
-1.91096663e-01 -3.65617037e-01 -1.13679737e-01 -5.58973372e-01
-1.71805099e-01 4.20397706e-02 4.52581465e-01 1.04503787e+00
1.29537545e-02 4.57467377e-01 -9.30324614e-01 -7.18458176e-01
6.75998986e-01 5.94026566e-01 -4.57144260e-01 -2.11251713e-02
1.69248730e-01 4.24482793e-01 -6.69192523e-02 3.76771748e-01
7.75453746e-01 -3.01755309e-01 3.28179866e-01 -5.71006179e-01
-2.83324689e-01 -6.45040095e-01 -1.82918751e+00 1.76523972e+00
-3.23367268e-01 6.19835019e-01 -3.31962481e-02 -1.28179979e+00
9.04788911e-01 1.75517768e-01 6.77976251e-01 -3.10393006e-01
3.89676094e-01 1.60583794e-01 -3.32343221e-01 -3.35704327e-01
5.50529420e-01 3.76300514e-01 3.47080715e-02 1.64676264e-01
1.66380897e-01 3.37749153e-01 5.80043197e-01 1.23403668e-01
9.06305373e-01 -7.99757317e-02 5.13129354e-01 -2.52152473e-01
1.10819042e+00 -1.49827123e-01 1.21000156e-01 5.24236381e-01
-1.51882410e-01 5.84122121e-01 3.68535250e-01 -4.96969908e-01
-1.04522300e+00 -1.00235534e+00 8.63950029e-02 7.43493259e-01
4.05892909e-01 -5.50965846e-01 -6.12691581e-01 -8.09723020e-01
-1.02067828e-01 1.15929745e-01 -4.58023190e-01 1.47571608e-01
-6.55292690e-01 -7.08662629e-01 3.17371070e-01 2.44841456e-01
3.05072904e-01 -3.30857754e-01 -7.44604290e-01 2.04348072e-01
-1.87753782e-01 -1.36739314e+00 -4.86180127e-01 6.97024688e-02
-7.01416254e-01 -1.50077212e+00 -8.20137322e-01 -4.09768879e-01
7.60107338e-01 6.59045279e-01 1.02069485e+00 1.17973536e-01
-4.44247037e-01 9.39087152e-01 -2.52215087e-01 2.33459324e-01
-8.45167264e-02 1.18560955e-01 2.24256858e-01 3.97779375e-01
5.28808951e-01 -5.21744609e-01 -6.19510949e-01 4.72146422e-01
-5.96334517e-01 -2.52155840e-01 5.85461438e-01 7.36111462e-01
3.74712735e-01 -2.36741245e-01 5.01835570e-02 -4.39317256e-01
2.46213868e-01 -3.68745029e-01 -9.57323968e-01 4.43022579e-01
-1.83136493e-01 -1.63227886e-01 6.68126702e-01 -4.10746932e-01
-5.31240582e-01 7.01271713e-01 3.09407860e-02 -6.34025991e-01
-6.39684021e-01 -4.43537952e-03 -4.46364790e-01 -6.35778069e-01
4.60150480e-01 4.16573808e-02 -3.96585800e-02 -2.56259620e-01
5.90697885e-01 3.26682270e-01 4.55823243e-01 -4.59838390e-01
8.22668195e-01 4.84177262e-01 5.21532357e-01 -1.09479654e+00
-4.21962231e-01 -9.99577165e-01 -1.06089020e+00 -7.39648163e-01
1.10089707e+00 -8.09693992e-01 -1.09921706e+00 1.88003376e-01
-1.49414694e+00 3.39540571e-01 1.09371550e-01 6.84246182e-01
-5.02693176e-01 8.59035313e-01 -2.92169154e-01 -7.51090169e-01
-2.55621560e-02 -1.24933016e+00 1.06561911e+00 1.89165592e-01
-9.66296270e-02 -1.16869259e+00 1.12215884e-01 4.68432695e-01
3.11555155e-02 2.29274824e-01 2.63426274e-01 -5.74395299e-01
-7.73705304e-01 -3.40130329e-01 -4.40937191e-01 4.43056449e-02
-5.09260967e-02 -2.05427408e-02 -8.96697402e-01 -5.50234437e-01
-4.15507406e-01 1.13538280e-02 6.89409792e-01 3.21107358e-01
8.14193130e-01 -3.54616381e-02 -4.17309552e-01 5.29681861e-01
1.67302620e+00 -2.11755484e-01 2.26361692e-01 3.97900790e-01
8.33575308e-01 7.51447558e-01 2.76101649e-01 5.09645283e-01
5.50710082e-01 1.07373488e+00 6.65363848e-01 3.40369269e-02
6.71788007e-02 2.69774258e-01 2.54709929e-01 5.59840560e-01
-3.34712595e-01 -2.77938396e-01 -8.87448967e-01 3.39582920e-01
-1.96719396e+00 -1.11489892e+00 -5.98774731e-01 2.24947977e+00
9.71905049e-03 4.89601158e-02 5.16860545e-01 1.86413839e-01
1.05773354e+00 -1.10102333e-01 -2.29470003e-02 -7.17967004e-02
-1.38198435e-01 -1.51827887e-01 7.08474338e-01 7.64383554e-01
-1.56339788e+00 4.69894260e-01 4.71781397e+00 5.64294875e-01
-7.82758534e-01 2.56969005e-01 3.41880679e-01 -4.75287698e-02
3.22549194e-01 -2.47209687e-02 -9.61586177e-01 4.55654263e-01
6.21601403e-01 1.06394343e-01 2.13899225e-01 7.54513085e-01
-6.98847026e-02 -3.45065027e-01 -1.23268986e+00 1.45114791e+00
4.49831247e-01 -1.18910253e+00 -1.11226968e-01 2.25900531e-01
2.25901917e-01 -1.78946480e-01 -1.87159196e-01 -2.57900566e-01
1.30144004e-02 -5.24170518e-01 6.03160441e-01 4.31576520e-01
3.62883419e-01 -6.71532750e-01 6.44834340e-01 7.92918280e-02
-1.73945403e+00 -3.50572206e-02 -3.22424114e-01 1.37577057e-01
2.42439210e-01 6.36981606e-01 -5.01474261e-01 1.09730208e+00
6.33828819e-01 8.02244067e-01 -8.56869698e-01 1.42640579e+00
1.46809265e-01 -1.12940818e-01 -4.69431370e-01 -1.57146335e-01
2.33344808e-01 -5.97115219e-01 7.15988576e-01 1.39529514e+00
4.91317809e-01 -3.26884866e-01 3.18475068e-01 3.90501171e-01
2.83447802e-01 2.05175743e-01 -8.84035230e-01 7.63939500e-01
3.20503004e-02 1.48443174e+00 -1.01128948e+00 -1.53873429e-01
-7.83295870e-01 1.02164471e+00 3.58013004e-01 7.17813522e-02
-9.46844161e-01 -9.56333578e-02 4.64589566e-01 -7.09533691e-02
4.47033674e-01 -5.43148160e-01 2.14570597e-01 -1.26319242e+00
7.25169629e-02 -5.45978725e-01 6.54212654e-01 -4.80816931e-01
-1.41372836e+00 5.34947872e-01 2.06087485e-01 -1.46155000e+00
-1.35546893e-01 -8.66862953e-01 -4.42869037e-01 5.27657449e-01
-1.26124263e+00 -1.26065493e+00 -4.42848295e-01 1.07537460e+00
2.82107502e-01 -5.54233342e-02 5.93570352e-01 9.33670282e-01
-6.17083907e-01 4.75949079e-01 -6.07973598e-02 1.83062837e-01
7.67231762e-01 -1.33535028e+00 -1.58221517e-02 1.09306431e+00
3.33799899e-01 2.35657111e-01 6.66919947e-01 -3.19791645e-01
-1.45728111e+00 -1.01545846e+00 7.88983524e-01 -4.45771903e-01
8.50181699e-01 -4.23002422e-01 -5.70439577e-01 4.90441144e-01
2.81990826e-01 2.09623083e-01 7.58124053e-01 6.99118748e-02
-2.54387110e-01 -3.55135053e-01 -8.81768107e-01 4.88477975e-01
9.66038465e-01 -3.29198837e-01 -2.37498701e-01 4.96366411e-01
1.36431381e-01 -6.82375208e-02 -8.36237431e-01 -1.39006063e-01
5.20824432e-01 -1.14273024e+00 1.28141320e+00 -3.70799422e-01
-1.20352864e-01 -5.06712496e-01 -4.17951852e-01 -1.21317112e+00
-2.06566572e-01 -3.01888376e-01 4.55850735e-02 1.20419025e+00
2.27110967e-01 -6.04409158e-01 8.35058093e-01 2.55943269e-01
1.42980963e-01 -2.08200648e-01 -1.32108676e+00 -8.64693344e-01
-3.76358837e-01 -2.74566263e-01 2.89521396e-01 8.38605165e-01
-3.50237191e-01 4.37712938e-01 -4.90998954e-01 5.61016977e-01
1.26456904e+00 -3.70487794e-02 1.09296477e+00 -1.42930925e+00
-4.53736544e-01 -5.32774866e-01 -8.78851116e-01 -7.69281387e-01
1.92901999e-01 -7.35754371e-01 -3.29735637e-01 -1.07354093e+00
1.68408807e-02 -3.60956550e-01 7.94242099e-02 -6.60805553e-02
1.03661224e-01 3.76490980e-01 5.24895787e-01 -8.25887620e-02
-9.43462491e-01 2.05642924e-01 7.35863209e-01 -2.95584679e-01
1.35300845e-01 1.53868601e-01 -1.01884030e-01 6.82252765e-01
9.06946838e-01 -4.82452393e-01 -7.01158941e-02 -4.28186357e-01
2.32277885e-01 9.60342214e-02 9.67822015e-01 -1.47141039e+00
8.17459226e-01 1.47426024e-01 4.28327262e-01 -4.55335110e-01
7.40548551e-01 -1.37829781e+00 4.76524055e-01 5.40416121e-01
1.42327115e-01 6.21053755e-01 3.88761759e-02 6.61750555e-01
-8.60224739e-02 -4.42095667e-01 8.41562688e-01 -3.31526726e-01
-8.16823781e-01 3.30649048e-01 -3.85794938e-01 -2.65451759e-01
1.44735122e+00 -3.23101580e-01 -1.38244659e-01 -9.04871002e-02
-9.44067180e-01 7.78383166e-02 4.09217685e-01 1.75484031e-01
5.36134183e-01 -1.31603944e+00 -5.75984657e-01 5.53975292e-02
2.11775914e-01 -3.01737666e-01 3.04584801e-01 1.13379407e+00
-4.66336995e-01 3.28990787e-01 -3.61797810e-01 -9.45859969e-01
-1.73295653e+00 8.97062421e-01 4.33016568e-01 -2.60208964e-01
-4.34329569e-01 3.56320381e-01 -2.48947263e-01 -1.36454523e-01
1.41656876e-01 6.99260607e-02 -3.49725723e-01 4.72702593e-01
3.09132755e-01 6.35654390e-01 9.50761288e-02 -8.50086451e-01
-4.88730967e-01 1.09123552e+00 2.45910019e-01 7.13262334e-02
1.17596543e+00 -3.28209877e-01 -1.76285014e-01 1.12784289e-01
1.36267960e+00 -1.34242922e-01 -9.11537707e-01 -9.69411433e-03
1.31406769e-01 -5.94770133e-01 -2.31434599e-01 -2.97301188e-02
-1.13056052e+00 8.19578052e-01 8.29554677e-01 3.91651750e-01
8.50852489e-01 -5.35830893e-02 4.06066597e-01 6.14784241e-01
4.33409333e-01 -7.87484527e-01 6.62275255e-02 2.43404403e-01
5.01319647e-01 -1.43068433e+00 1.24626979e-01 -4.04653996e-01
-2.50761777e-01 1.47503614e+00 6.37450159e-01 -1.41633973e-01
7.59854019e-01 1.42350763e-01 -2.03552082e-01 -4.74016406e-02
-2.55941272e-01 -5.73319852e-01 1.53599784e-01 6.17968500e-01
1.68646887e-01 4.95060347e-02 -1.38196468e-01 -1.84614450e-01
1.23923734e-01 -3.15685153e-01 4.75317210e-01 5.49775064e-01
-1.51327312e-01 -1.33648098e+00 -7.39207447e-01 1.73600808e-01
-1.09519377e-01 7.08838105e-02 -3.46979022e-01 9.29306328e-01
3.57527256e-01 9.46564972e-01 8.00304487e-02 -3.33812773e-01
4.06458467e-01 -2.38053277e-01 7.42875516e-01 -9.43466350e-02
-6.49734080e-01 -1.55793399e-01 1.14962552e-02 -5.09297788e-01
-1.06612265e+00 -9.67038989e-01 -2.61730433e-01 -5.83220422e-01
-3.27976376e-01 1.52014107e-01 5.74292481e-01 7.99517691e-01
6.82421625e-02 1.22744054e-01 5.50603628e-01 -1.24031019e+00
-1.03971832e-01 -4.23116267e-01 -2.54259229e-01 4.85402942e-01
3.11601430e-01 -8.41004431e-01 -2.93997467e-01 5.10092117e-02] | [8.231216430664062, -1.7834471464157104] |
270c6f43-08dd-458a-98e1-2771e62e9f60 | ptse-a-multi-model-ensemble-method-for | 2305.11304 | null | https://arxiv.org/abs/2305.11304v1 | https://arxiv.org/pdf/2305.11304v1.pdf | pTSE: A Multi-model Ensemble Method for Probabilistic Time Series Forecasting | Various probabilistic time series forecasting models have sprung up and shown remarkably good performance. However, the choice of model highly relies on the characteristics of the input time series and the fixed distribution that the model is based on. Due to the fact that the probability distributions cannot be averaged over different models straightforwardly, the current time series model ensemble methods cannot be directly applied to improve the robustness and accuracy of forecasting. To address this issue, we propose pTSE, a multi-model distribution ensemble method for probabilistic forecasting based on Hidden Markov Model (HMM). pTSE only takes off-the-shelf outputs from member models without requiring further information about each model. Besides, we provide a complete theoretical analysis of pTSE to prove that the empirical distribution of time series subject to an HMM will converge to the stationary distribution almost surely. Experiments on benchmarks show the superiority of pTSE overall member models and competitive ensemble methods. | ['Sheng Li', 'Yuchen Huang', 'Ge Jin', 'Yijia Ruan', 'Zhixuan Chu', 'Yunyi Zhou'] | 2023-05-16 | null | null | null | null | ['probabilistic-time-series-forecasting'] | ['time-series'] | [-8.82963017e-02 -5.47289252e-01 1.76894560e-01 -4.48026717e-01
-8.74102592e-01 -7.33230770e-01 4.54347283e-01 -2.71205425e-01
1.67526618e-01 5.82469404e-01 -1.24185331e-01 -5.78335285e-01
-3.19497079e-01 -6.74613535e-01 -6.47926927e-01 -1.29014754e+00
-1.36053249e-01 6.05839729e-01 2.71849521e-02 -4.31648418e-02
-4.36516525e-03 3.09370428e-01 -1.65415537e+00 1.36181489e-01
8.03935945e-01 1.07063794e+00 1.40960403e-02 7.47271001e-01
-1.16146833e-01 4.74066049e-01 -7.11841047e-01 -2.42217645e-01
1.39304504e-01 -4.30422693e-01 -2.26907451e-02 -3.57247531e-01
-1.93391681e-01 -1.20034367e-01 -2.26127833e-01 8.81933331e-01
3.88271958e-01 3.30679834e-01 8.97983432e-01 -1.78509450e+00
-2.43034020e-01 8.08817744e-01 -3.14305276e-01 1.35593817e-01
3.15227197e-04 -1.39721066e-01 7.04690635e-01 -6.30112529e-01
1.23212382e-01 1.03219068e+00 8.96605968e-01 3.29500973e-01
-1.15949893e+00 -7.77373791e-01 3.79344642e-01 1.53349236e-01
-1.38810837e+00 -8.90796259e-02 7.08868563e-01 -3.61917853e-01
8.52061510e-01 3.25572342e-01 2.09812120e-01 1.25119483e+00
6.09464467e-01 5.84404588e-01 1.32033646e+00 -1.64863169e-01
4.40452993e-01 -8.69388878e-02 4.07199889e-01 6.74681365e-02
-1.55971304e-01 2.77987093e-01 -3.55845481e-01 -7.04214573e-01
2.29795232e-01 3.00467581e-01 -1.63105950e-01 9.59705412e-02
-9.81173992e-01 5.30887902e-01 -2.55595207e-01 3.04503590e-01
-5.76334953e-01 2.36834809e-01 3.33586782e-01 1.65523514e-01
6.11801207e-01 -1.77515641e-01 -8.34002912e-01 -4.76238281e-01
-1.12835419e+00 3.37913394e-01 1.00651395e+00 8.39523554e-01
3.44820499e-01 1.31633863e-01 -5.25431894e-02 3.73042732e-01
2.09560439e-01 1.08681977e+00 3.34663838e-01 -8.31476331e-01
2.50550270e-01 1.27744675e-01 4.02476728e-01 -1.07800651e+00
-1.85379699e-01 -4.93531942e-01 -1.02250445e+00 -1.33095190e-01
5.43411732e-01 -2.36695588e-01 -7.59958446e-01 1.69740415e+00
1.35023192e-01 7.35351562e-01 6.96989195e-03 4.59004790e-01
2.21797585e-01 1.20420671e+00 9.12535097e-03 -5.98803699e-01
8.44650328e-01 -5.47281384e-01 -6.58742070e-01 4.01602119e-01
4.75099832e-01 -7.98027217e-01 4.55488771e-01 5.81580341e-01
-5.61165631e-01 -4.25983936e-01 -6.99347258e-01 6.79538071e-01
-3.12707722e-01 2.23437715e-02 2.74172366e-01 7.62872159e-01
-7.33427465e-01 6.11808300e-01 -1.24910593e+00 -3.45410779e-02
-5.68451099e-02 2.12069988e-01 2.47162720e-03 1.17107101e-01
-1.17858112e+00 8.68225574e-01 3.70719403e-01 2.50766784e-01
-8.09253395e-01 -6.18975520e-01 -3.85209024e-01 2.30294243e-01
4.99281362e-02 -4.88916278e-01 1.40586913e+00 -7.81363845e-01
-1.69024587e+00 -1.08139277e-01 -6.88769996e-01 -6.17717505e-01
5.19496441e-01 -1.19072653e-01 -7.47790933e-01 -3.85343313e-01
-2.97682971e-01 -1.23669751e-01 8.71354938e-01 -1.16238701e+00
-7.51128674e-01 -2.14347482e-01 -3.85281295e-01 -3.77951890e-01
-1.68944269e-01 1.12554140e-01 -1.38059735e-01 -7.10074604e-01
2.06089877e-02 -1.25725007e+00 -4.57511663e-01 -7.65584648e-01
-3.48035336e-01 -5.95694780e-01 1.06341100e+00 -6.61650300e-01
1.64047813e+00 -2.22501516e+00 -1.52730897e-01 5.06965041e-01
-1.76526621e-01 -2.29299098e-01 8.75896066e-02 9.23411667e-01
-4.91977148e-02 -3.14729251e-02 -2.73382515e-01 -3.56834501e-01
3.08848828e-01 4.45428789e-01 -1.29020894e+00 4.75284457e-01
-3.33949745e-01 3.99026483e-01 -7.54318476e-01 -3.26333463e-01
3.31664503e-01 5.15295804e-01 4.41679433e-02 6.91282228e-02
-3.40781122e-01 4.62180167e-01 -4.11154240e-01 4.07044709e-01
9.10057306e-01 -2.61816412e-01 3.01891536e-01 -1.44309867e-02
-8.42340365e-02 1.80149645e-01 -1.17071664e+00 9.63056445e-01
-4.66056734e-01 2.98874319e-01 -6.07736826e-01 -9.47903335e-01
1.10031450e+00 5.71032584e-01 7.89551020e-01 -3.10858469e-02
7.59928897e-02 3.02769959e-01 -5.90432435e-02 8.25112462e-02
3.95372003e-01 -7.83852711e-02 -2.49673694e-01 7.53435075e-01
-1.11064404e-01 9.12336111e-02 3.60640362e-02 -1.85219318e-01
7.22074091e-01 3.46135348e-01 9.76474062e-02 -1.28381506e-01
1.41976342e-01 -9.33636650e-02 7.39427030e-01 8.89532149e-01
-3.56287621e-02 4.38570708e-01 3.03527832e-01 -4.28797722e-01
-9.53057766e-01 -1.10265756e+00 -3.10759723e-01 1.21788836e+00
-1.85445592e-01 -4.59749550e-01 -5.22524834e-01 -5.67270935e-01
-1.31149873e-01 1.18878233e+00 -6.08659506e-01 1.77971184e-01
-3.84787202e-01 -1.08666813e+00 3.39559853e-01 6.20188773e-01
1.53181525e-02 -7.00763762e-01 -4.91182923e-01 6.70014203e-01
-3.47770572e-01 -1.03558981e+00 -3.92265022e-01 1.60587922e-01
-1.02031815e+00 -5.60660839e-01 -5.74089468e-01 -1.61858588e-01
3.00009787e-01 1.99434698e-01 1.05337846e+00 -4.68614936e-01
6.69551551e-01 4.21023011e-01 -3.56874138e-01 -8.95109296e-01
-6.00936711e-01 1.00434363e-01 1.60188287e-01 2.56915271e-01
4.40377384e-01 -1.05650687e+00 -3.60647768e-01 6.52292430e-01
-6.28536224e-01 -7.43526220e-02 2.46779978e-01 6.78333879e-01
5.65156460e-01 6.66734695e-01 7.07952619e-01 -4.34372067e-01
3.24305385e-01 -7.13612676e-01 -8.32091212e-01 6.82716727e-01
-5.68803906e-01 2.82057356e-02 7.03033030e-01 -5.23851216e-01
-9.75574195e-01 1.14963911e-01 1.15159608e-01 -6.63551569e-01
-3.86177488e-02 7.68406451e-01 4.71981578e-02 5.62474787e-01
8.36111605e-02 7.11316347e-01 -1.53753772e-01 -4.41882372e-01
2.36625761e-01 6.44970059e-01 6.41527355e-01 -6.65268421e-01
5.25411069e-01 4.74539369e-01 6.36373758e-02 -4.42078263e-01
-7.94258535e-01 -3.69578153e-01 -3.46907526e-01 -3.79290074e-01
4.07249629e-01 -8.04085851e-01 -8.77789497e-01 6.76900983e-01
-1.23297071e+00 -1.20446593e-01 2.20299751e-01 6.00809216e-01
-4.87778723e-01 2.83156663e-01 -2.73543388e-01 -1.60499656e+00
-3.36460322e-01 -8.89444292e-01 9.95093644e-01 -1.07642193e-03
-1.99992776e-01 -1.11973655e+00 2.54361719e-01 -2.33112082e-01
5.49985588e-01 5.66424310e-01 7.57280231e-01 -8.51550579e-01
-3.22263449e-01 -7.03769624e-01 2.97121465e-01 4.10705179e-01
-1.02799132e-01 7.14022875e-01 -9.53699946e-01 -1.73598215e-01
5.92924170e-02 4.76789296e-01 7.29258835e-01 7.15944707e-01
1.41984153e+00 -1.63737580e-01 -6.05851591e-01 2.17409238e-01
1.25079274e+00 3.42790484e-01 5.90248346e-01 -1.42765846e-02
3.67025703e-01 4.62205648e-01 4.48310465e-01 7.99499035e-01
5.52572608e-01 6.06112540e-01 1.72888175e-01 6.16498411e-01
6.39206827e-01 -2.89417535e-01 8.49172652e-01 1.22545731e+00
-3.43779534e-01 -5.83470166e-01 -1.06432974e+00 4.18139279e-01
-1.97603464e+00 -1.36855459e+00 -3.90060425e-01 2.52784276e+00
5.15837133e-01 1.16834812e-01 2.77308524e-01 2.34875187e-01
8.41741025e-01 -3.18345875e-02 -5.69775283e-01 -2.44490653e-01
-2.74646163e-01 3.13102826e-02 5.65283000e-01 1.36913300e-01
-1.13617373e+00 3.00908238e-01 6.96628428e+00 1.13802242e+00
-1.29083467e+00 1.02010980e-01 6.40746176e-01 1.75908163e-01
-2.75087178e-01 -6.84696250e-04 -9.54087079e-01 1.08393204e+00
1.51143789e+00 -6.82573974e-01 2.86894977e-01 8.76888931e-01
1.45966157e-01 7.41043389e-02 -9.69713092e-01 9.72524464e-01
-3.94382656e-01 -1.03388286e+00 -1.62967682e-01 1.26698613e-01
1.00102103e+00 2.64293909e-01 2.47551262e-01 5.16457617e-01
6.88813210e-01 -1.00870943e+00 7.39419639e-01 1.04199910e+00
2.85748422e-01 -9.69326556e-01 1.02223051e+00 6.76022351e-01
-1.21912098e+00 -5.63528202e-02 -2.54621148e-01 -2.06702575e-02
4.24324632e-01 9.94424522e-01 -5.59329033e-01 9.51542974e-01
7.64899969e-01 5.57914376e-01 5.22394739e-02 9.08137143e-01
1.46440446e-01 1.29654217e+00 -9.74462092e-01 -7.64708221e-02
1.34887502e-01 -1.95942268e-01 6.04697526e-01 1.05627894e+00
1.01852953e+00 4.13134657e-02 2.78118581e-01 3.41280222e-01
4.61054623e-01 -1.15031213e-01 -3.86374980e-01 -9.07892510e-02
7.36912251e-01 9.78002369e-01 -5.81634283e-01 -4.25745398e-01
-3.06692481e-01 4.54733223e-01 -2.70422965e-01 5.09787321e-01
-1.12457633e+00 1.36681255e-02 3.75624359e-01 -3.98312241e-01
6.58479810e-01 -4.83184010e-01 -3.89856786e-01 -1.02725792e+00
1.25745302e-02 -8.57183278e-01 4.88370538e-01 -6.32021248e-01
-1.82821620e+00 9.24286127e-01 1.96476907e-01 -1.67342615e+00
-6.10426784e-01 -4.02326256e-01 -8.21582139e-01 9.88766015e-01
-8.73344779e-01 -9.46054041e-01 6.00962378e-02 4.54522312e-01
2.00128570e-01 -3.18294317e-02 1.01669669e+00 -1.03657462e-01
-6.13162160e-01 5.27663410e-01 8.52580965e-01 -3.05056095e-01
4.59412724e-01 -1.08956587e+00 3.78992110e-01 8.22886109e-01
-4.80663553e-02 4.24689233e-01 1.12691891e+00 -4.53454494e-01
-1.25733674e+00 -1.10885239e+00 9.66983557e-01 -8.50597501e-01
8.47630739e-01 -1.45088760e-02 -1.01403880e+00 6.11633539e-01
8.86663422e-02 -1.80396065e-01 8.57894957e-01 3.41342926e-01
-3.65633547e-01 -1.98912993e-01 -7.64697254e-01 3.09502989e-01
4.62558120e-01 -3.50913316e-01 -4.55469161e-01 2.09544107e-01
3.33914042e-01 -2.76800126e-01 -1.18821704e+00 4.58337456e-01
8.65560949e-01 -8.59180987e-01 6.86182439e-01 -5.12613475e-01
2.46895969e-01 -3.62656534e-01 -5.72880685e-01 -1.31459641e+00
-1.14153273e-01 -8.48726809e-01 -4.92964953e-01 1.35501587e+00
4.14433569e-01 -1.10552323e+00 1.91433668e-01 7.84398615e-01
1.37540093e-02 -7.21442938e-01 -1.17210090e+00 -1.26177001e+00
2.14422330e-01 -9.70073998e-01 1.26405513e+00 8.18585992e-01
-2.26814821e-01 -8.73988345e-02 -6.40930355e-01 6.18488371e-01
6.85018957e-01 7.29406178e-01 8.47454011e-01 -1.33801091e+00
-3.65083814e-01 -5.00247657e-01 -1.32797346e-01 -9.83812273e-01
2.54489094e-01 -5.20754576e-01 1.88467488e-01 -1.05256438e+00
1.02528334e-01 -6.30062997e-01 -8.41098845e-01 1.99957550e-01
-3.03984642e-01 -2.81134754e-01 1.62309557e-01 5.87423503e-01
-6.60824060e-01 7.27694809e-01 6.79825068e-01 2.34122828e-01
-7.42054880e-02 5.26568830e-01 -2.79775262e-01 6.83863103e-01
9.36820805e-01 -7.38436162e-01 -3.83931845e-01 -1.32152007e-03
1.31253302e-01 4.64966923e-01 3.72923166e-01 -8.40725899e-01
3.08424562e-01 -2.44373828e-01 1.03471249e-01 -1.25038290e+00
1.18237600e-01 -8.89362633e-01 6.97943032e-01 1.82334438e-01
-9.60207060e-02 3.53872448e-01 1.56531096e-01 8.86214852e-01
-3.47177535e-01 1.16856314e-01 2.34840944e-01 4.75398630e-01
-4.05952215e-01 4.26031083e-01 -4.80207562e-01 -3.99021983e-01
8.45207989e-01 -4.44453768e-02 -4.10516784e-02 -7.47677088e-01
-9.43018496e-01 1.73498735e-01 1.61189333e-01 2.39441693e-01
5.53924330e-02 -1.25420654e+00 -9.15502548e-01 -1.82768911e-01
-7.53552020e-02 -3.15342396e-01 5.69671631e-01 1.03937304e+00
1.66428909e-01 4.11859572e-01 2.62706071e-01 -8.64420950e-01
-9.23810005e-01 7.12736368e-01 3.42637420e-01 -5.52676499e-01
-3.69119257e-01 2.91059732e-01 8.54876414e-02 -5.66466689e-01
9.56532732e-02 -4.85304534e-01 1.23413935e-01 -1.96665958e-01
5.88097036e-01 5.62581658e-01 1.61765128e-01 -4.49869633e-01
-4.09261465e-01 3.46857876e-01 2.28446603e-01 -2.41039023e-01
1.41442347e+00 -2.03191534e-01 -1.35373875e-01 1.08403385e+00
9.44134533e-01 -2.96906888e-01 -1.20710206e+00 -2.26215452e-01
-4.36360836e-02 -2.39958033e-01 -3.20696682e-02 -8.47265303e-01
-7.43566811e-01 7.99381495e-01 5.25693059e-01 6.07075810e-01
1.27201724e+00 -3.69988739e-01 9.63750422e-01 3.11408430e-01
7.40926802e-01 -9.05274093e-01 -7.84993351e-01 6.22469246e-01
8.48065972e-01 -9.20433223e-01 -2.62449652e-01 -1.25956982e-01
-6.25553787e-01 1.29543912e+00 1.21691334e-03 -1.49366513e-01
1.22891688e+00 4.48460221e-01 3.58156599e-02 3.58182669e-01
-1.22060978e+00 9.66282636e-02 4.03063923e-01 5.28118789e-01
2.88025022e-01 5.45500815e-01 -6.94988295e-02 9.82719600e-01
-4.59083766e-01 1.65773854e-01 1.58010826e-01 5.15935242e-01
-4.28346135e-02 -9.80415821e-01 -6.11044466e-01 4.18337733e-01
-4.34438944e-01 3.86925675e-02 3.42847079e-01 4.83111501e-01
-1.87001765e-01 1.19961381e+00 1.47924468e-01 -4.29786056e-01
1.32424697e-01 3.70673001e-01 1.92767486e-01 1.25456110e-01
-3.19484413e-01 2.64450580e-01 -1.81828976e-01 -1.31570175e-01
-4.29764330e-01 -1.05714262e+00 -9.25289094e-01 -8.23024035e-01
-3.37143183e-01 3.75317663e-01 6.55020773e-01 1.15311038e+00
6.72236443e-01 3.55272830e-01 9.66786325e-01 -7.90089011e-01
-9.82847631e-01 -1.09674978e+00 -6.44769371e-01 -2.01301016e-02
2.85913125e-02 -6.83521032e-01 -5.20197213e-01 1.67031318e-01] | [6.9586029052734375, 3.2743217945098877] |
ed0c2692-2a27-4dec-9796-e4bef1c8f8c6 | restricted-generative-projection-for-one | 2307.04097 | null | https://arxiv.org/abs/2307.04097v1 | https://arxiv.org/pdf/2307.04097v1.pdf | Restricted Generative Projection for One-Class Classification and Anomaly Detection | We present a simple framework for one-class classification and anomaly detection. The core idea is to learn a mapping to transform the unknown distribution of training (normal) data to a known target distribution. Crucially, the target distribution should be sufficiently simple, compact, and informative. The simplicity is to ensure that we can sample from the distribution easily, the compactness is to ensure that the decision boundary between normal data and abnormal data is clear and reliable, and the informativeness is to ensure that the transformed data preserve the important information of the original data. Therefore, we propose to use truncated Gaussian, uniform in hypersphere, uniform on hypersphere, or uniform between hyperspheres, as the target distribution. We then minimize the distance between the transformed data distribution and the target distribution while keeping the reconstruction error for the original data small enough. Comparative studies on multiple benchmark datasets verify the effectiveness of our methods in comparison to baselines. | ['Jicong Fan', 'Ruoyu Sun', 'Feng Xiao'] | 2023-07-09 | null | null | null | null | ['anomaly-detection', 'one-class-classification'] | ['methodology', 'miscellaneous'] | [-9.60999131e-02 1.51019931e-01 5.06322794e-02 -4.50009763e-01
-1.28645211e-01 -2.78216332e-01 3.26520950e-01 5.96933179e-02
-1.30827069e-01 7.67748833e-01 -9.92904603e-02 -3.08536272e-02
-2.71530509e-01 -9.27111208e-01 -5.18964648e-01 -1.06469667e+00
3.66923288e-02 7.26590931e-01 3.19485277e-01 9.53137651e-02
3.00864935e-01 6.99173212e-01 -1.52697623e+00 -1.73631743e-01
1.18732643e+00 1.25535464e+00 -1.15307078e-01 2.36603290e-01
-7.89291114e-02 4.24538970e-01 -8.41960311e-01 -7.33065456e-02
5.23231328e-01 -3.84150565e-01 -5.81887901e-01 2.21921265e-01
-5.47853438e-03 -3.44113648e-01 -1.37449428e-01 1.51201487e+00
2.76303291e-01 3.67326766e-01 1.04452717e+00 -1.47680295e+00
-5.40187776e-01 4.93296720e-02 -7.52350092e-01 6.48121834e-02
-1.76055431e-02 -2.65121847e-01 6.60635769e-01 -5.07910192e-01
3.34019870e-01 9.12733674e-01 3.44195276e-01 5.16657293e-01
-9.73495007e-01 -5.62116683e-01 1.09831609e-01 -5.08629531e-02
-1.44881940e+00 -3.18911463e-01 5.38610399e-01 -3.40796292e-01
1.94517985e-01 4.65321571e-01 5.70883095e-01 8.67579818e-01
4.01666284e-01 4.64517564e-01 5.62855482e-01 -2.98633903e-01
4.50857282e-01 4.24673736e-01 2.78954171e-02 5.64032912e-01
4.93576795e-01 1.03609972e-02 -6.09564036e-03 -4.07660544e-01
4.05972004e-01 3.87005061e-01 -7.43415415e-01 -6.26891911e-01
-8.10234547e-01 7.54040599e-01 7.61279166e-02 2.50100315e-01
-2.76261747e-01 -4.98825550e-01 2.87789494e-01 3.78715426e-01
3.32205713e-01 2.96459049e-01 -1.30309463e-01 -1.72435135e-01
-7.28306711e-01 1.17746063e-01 7.56480336e-01 8.39340150e-01
6.43968999e-01 1.23063475e-01 -5.86224124e-02 7.18039572e-01
3.55696499e-01 7.56972790e-01 7.09758043e-01 -4.97859150e-01
3.32433820e-01 7.52510071e-01 1.90843210e-01 -1.05296469e+00
-6.24719262e-02 -3.01647276e-01 -1.09609520e+00 2.20072001e-01
5.38727999e-01 -1.18816435e-01 -7.08894312e-01 1.37391210e+00
4.05114979e-01 1.51595041e-01 7.39840567e-02 8.19438040e-01
3.53592962e-01 7.61543751e-01 -7.00077295e-01 -2.29159653e-01
6.47633910e-01 -3.86597782e-01 -5.69075644e-01 -7.37058595e-02
4.05816555e-01 -2.84281194e-01 1.21015835e+00 5.12729406e-01
-6.01305425e-01 -2.68059313e-01 -1.06424713e+00 4.59855735e-01
-2.00052977e-01 1.55320063e-01 8.74135867e-02 6.17106557e-01
-4.57679600e-01 5.68285406e-01 -8.72988582e-01 -1.44423023e-01
3.91246885e-01 6.83663562e-02 -6.18323624e-01 1.82624608e-01
-9.49110627e-01 6.59542382e-01 6.74834371e-01 -1.62875295e-01
-5.51791847e-01 -5.90012908e-01 -7.94842064e-01 1.36091605e-01
1.09789139e-02 -1.84031948e-01 7.63704360e-01 -7.70261347e-01
-1.15128493e+00 5.11556983e-01 -9.71035741e-04 -5.69108069e-01
5.16927898e-01 3.72754708e-02 -4.62907434e-01 7.54882917e-02
2.47681141e-03 2.57153008e-02 1.04838800e+00 -1.15588689e+00
-7.93071449e-01 -3.94788623e-01 -5.60558498e-01 1.01960286e-01
-6.98357940e-01 -3.22539866e-01 -3.09592605e-01 -4.44761246e-01
4.02709544e-01 -6.53299510e-01 9.89250764e-02 3.71563286e-02
-5.74159086e-01 -1.40186191e-01 1.30174327e+00 -6.37983441e-01
1.16506636e+00 -2.68410158e+00 -1.62544265e-01 8.25486541e-01
3.37587208e-01 1.27673810e-02 2.23283559e-01 1.86570585e-02
-2.15189666e-01 -2.23966837e-01 -5.69194973e-01 -1.48045346e-01
-5.40979058e-02 3.38278294e-01 -8.31773818e-01 9.56001282e-01
-9.35441926e-02 4.95224923e-01 -7.35047996e-01 -1.12639166e-01
4.01656777e-02 2.27767423e-01 -3.41644853e-01 4.80399281e-01
6.44178316e-02 4.28625911e-01 -5.57008803e-01 6.36309862e-01
9.24396098e-01 3.97742614e-02 -4.40268546e-01 1.60521179e-01
3.10874194e-01 1.83261912e-02 -1.40776241e+00 8.38003516e-01
5.78572564e-02 4.83623356e-01 1.72976717e-01 -1.22922337e+00
1.48081422e+00 5.76041304e-02 5.93553245e-01 -4.68134731e-01
1.91286981e-01 1.90612674e-01 6.24673367e-02 -2.58578658e-01
1.06165044e-01 -3.80437315e-01 2.91189075e-01 6.93383396e-01
-3.99489135e-01 -2.38689259e-01 -1.87640451e-02 -1.16305493e-01
6.43997252e-01 -3.63963723e-01 3.12108666e-01 -3.96743298e-01
3.97144169e-01 -4.30255860e-01 7.94584870e-01 3.48446250e-01
-1.78457022e-01 8.66475224e-01 8.10987055e-01 -2.13779628e-01
-7.91438222e-01 -1.09979486e+00 -4.55130517e-01 4.64398652e-01
3.56953800e-01 -9.27068219e-02 -6.06930852e-01 -1.11213839e+00
1.31394878e-01 8.71407270e-01 -7.35140860e-01 -7.28571475e-01
-2.21620798e-01 -8.17032695e-01 3.52753907e-01 3.71569932e-01
5.40211439e-01 -9.27232444e-01 -3.52323979e-01 -2.78479725e-01
-1.58536613e-01 -7.31054902e-01 -4.88029003e-01 5.19449450e-02
-8.36714864e-01 -1.27307034e+00 -4.26308572e-01 -5.55926085e-01
1.06100810e+00 5.94855882e-02 6.70094311e-01 1.00497995e-03
-1.15090266e-01 2.01066043e-02 -2.87376553e-01 -4.95291412e-01
-4.22040641e-01 -4.25639421e-01 3.19491416e-01 3.90695214e-01
3.87040496e-01 -5.45421302e-01 -2.06786454e-01 4.51716483e-01
-1.04485846e+00 -5.09751022e-01 1.63027108e-01 6.61323309e-01
7.05050766e-01 8.98149788e-01 4.60058570e-01 -7.27561772e-01
5.88758409e-01 -6.63214207e-01 -5.42602062e-01 1.84135363e-01
-7.71032631e-01 6.12678975e-02 9.66049016e-01 -4.21934575e-01
-7.59285927e-01 -2.64364541e-01 -1.41898571e-02 -6.78174675e-01
-2.59244591e-01 1.45894095e-01 -5.13414979e-01 9.84518975e-02
7.23429263e-01 6.00225151e-01 4.16001737e-01 -3.90597969e-01
-7.29796337e-03 9.42615986e-01 6.37868583e-01 -7.10485637e-01
8.94664824e-01 7.39595532e-01 3.45645919e-02 -9.22008574e-01
-6.07023537e-01 -4.65225369e-01 -7.33636081e-01 1.90397471e-01
3.35211098e-01 -2.76425868e-01 -3.10470343e-01 3.98941189e-01
-5.74734807e-01 -4.12036926e-02 -8.30703437e-01 4.19346005e-01
-4.23060775e-01 7.24219322e-01 -8.65501612e-02 -9.67091501e-01
-4.09738511e-01 -7.91111231e-01 8.49119127e-01 2.46765941e-01
-9.16946009e-02 -1.16565275e+00 2.78250817e-02 -2.58690476e-01
3.13339949e-01 5.84768772e-01 1.10014153e+00 -1.15914035e+00
-4.81040664e-02 -6.81152940e-01 -1.14995509e-01 8.74520004e-01
6.62127256e-01 2.56000668e-01 -5.96558988e-01 -6.00650072e-01
6.09182894e-01 -1.54783905e-01 7.13212252e-01 2.86613762e-01
1.47310460e+00 -4.11055744e-01 -3.54146272e-01 7.86146939e-01
9.40904915e-01 3.38688672e-01 7.82636762e-01 3.56361091e-01
3.69289875e-01 5.03936172e-01 7.69410133e-01 6.09177947e-01
1.04702502e-01 1.86307788e-01 4.55614179e-01 -3.74650657e-02
3.01186174e-01 -5.30508272e-02 3.33009750e-01 2.46030688e-01
2.76738465e-01 -2.98720319e-02 -7.49884725e-01 4.46156681e-01
-1.60015810e+00 -8.95544350e-01 1.20239839e-01 2.80695629e+00
7.21825540e-01 3.08314532e-01 1.72271118e-01 3.67697924e-01
7.40855873e-01 -2.41269454e-01 -6.46408916e-01 -1.38929263e-01
-1.43193692e-01 -1.77496284e-01 7.90249780e-02 3.24105293e-01
-1.14766324e+00 1.94101632e-01 6.06539583e+00 8.04068327e-01
-1.37109244e+00 -4.27162170e-01 5.75938463e-01 -1.19780600e-01
-2.14041412e-01 -2.51859367e-01 -7.91179657e-01 8.49931836e-01
6.42683208e-01 -4.93928611e-01 2.22197682e-01 9.74509954e-01
1.48587162e-02 3.03571834e-03 -1.17483640e+00 9.72568810e-01
3.21714915e-02 -7.61058867e-01 4.19705547e-02 2.40097940e-01
4.95366514e-01 -1.16861043e-02 2.86155920e-02 3.19550037e-01
-9.59589705e-02 -1.02950299e+00 4.96447027e-01 7.20779240e-01
7.39270985e-01 -9.72993076e-01 8.76085460e-01 7.54796922e-01
-8.74449968e-01 -2.57919766e-02 -8.38095009e-01 2.45764554e-01
-2.52331197e-01 9.86309469e-01 -7.54422724e-01 5.46044469e-01
6.95784807e-01 8.66382122e-01 -5.95346391e-01 1.19576967e+00
1.38887214e-02 4.78244662e-01 -6.25446558e-01 2.82677621e-01
-1.49698883e-01 -7.96484113e-01 7.54489005e-01 8.79628778e-01
3.90493542e-01 -2.17786059e-01 3.57984781e-01 9.16158974e-01
5.52344210e-02 1.60324559e-01 -7.52676785e-01 -2.72675278e-03
5.50109446e-01 1.04227364e+00 -6.82378411e-01 -1.72786862e-01
-3.12975854e-01 8.31359386e-01 1.01897836e-01 4.13113892e-01
-7.64860928e-01 -8.96554232e-01 5.35409510e-01 2.99251348e-01
1.83735967e-01 1.68351039e-01 -2.91925222e-01 -1.18382859e+00
5.10302126e-01 -7.18424201e-01 6.73533797e-01 -2.22182289e-01
-1.58025408e+00 7.46563435e-01 7.98623860e-02 -1.45535100e+00
-2.24311873e-01 -6.52174175e-01 -9.75373209e-01 8.11389327e-01
-1.14608371e+00 -5.28175831e-01 -3.85310918e-01 7.67928779e-01
-3.02439202e-02 -5.77871144e-01 6.49824440e-01 9.46036726e-02
-5.93324304e-01 6.27485693e-01 5.55865765e-01 2.47815758e-01
6.49600267e-01 -1.31936955e+00 -2.64849793e-02 9.90006804e-01
-9.63376462e-02 7.20364749e-01 5.18744946e-01 -7.21037507e-01
-7.86928236e-01 -1.27492917e+00 2.08118498e-01 -4.06562954e-01
5.04830897e-01 -3.16952407e-01 -1.32825530e+00 6.28080368e-01
-3.26196998e-01 3.42256904e-01 5.50372601e-01 -2.24993855e-01
-2.58717090e-01 -3.00262570e-01 -1.62311113e+00 2.89117604e-01
3.05045366e-01 -1.45841479e-01 -6.62768304e-01 3.43153447e-01
1.87710613e-01 -4.33116466e-01 -7.50241041e-01 5.69601476e-01
2.44917974e-01 -1.04365253e+00 7.71206498e-01 -5.46946287e-01
2.37262761e-03 -5.16506433e-01 -2.19262406e-01 -1.47650564e+00
-2.31984351e-02 -4.43746448e-01 -5.45340180e-01 1.11346960e+00
1.62044227e-01 -1.13871801e+00 7.62441456e-01 4.68807787e-01
-3.34969489e-03 -9.33259189e-01 -9.75441635e-01 -9.21298563e-01
1.69434786e-01 -6.22901618e-02 8.28946888e-01 8.86570454e-01
-6.12222068e-02 1.03879916e-02 -1.85676739e-01 3.50203097e-01
5.63434541e-01 4.23817724e-01 6.31616294e-01 -1.69326639e+00
-1.11046121e-01 -3.72326374e-01 -5.37491977e-01 -8.99670124e-01
2.59164780e-01 -7.48414040e-01 2.29143328e-03 -1.38998330e+00
2.00845510e-01 -5.34493923e-01 -2.20561877e-01 5.16767681e-01
-1.20060049e-01 -7.30468100e-03 -3.23295265e-01 4.51413512e-01
-1.85712367e-01 1.00437832e+00 1.25367248e+00 6.33203462e-02
-3.73491168e-01 5.70089519e-01 -8.35615754e-01 7.46344566e-01
8.35751474e-01 -3.32167357e-01 -7.11106181e-01 1.94713637e-01
-2.20169455e-01 -3.77136976e-01 -9.07858554e-03 -8.52940321e-01
-3.12033519e-02 -4.48298097e-01 5.37607014e-01 -5.18358469e-01
1.53425381e-01 -9.43650186e-01 -3.57464850e-01 2.91937411e-01
5.51101305e-02 -3.61033738e-01 5.26583642e-02 5.80120504e-01
-2.97770411e-01 -3.84739071e-01 1.23901153e+00 4.05696303e-01
-2.89737433e-01 5.34108520e-01 1.29697651e-01 2.34256953e-01
1.29194987e+00 -3.74685735e-01 -1.18790172e-01 -4.91775274e-01
-5.01072884e-01 3.02774280e-01 7.17952967e-01 3.85303169e-01
8.32859635e-01 -1.29923069e+00 -6.62022471e-01 8.37843716e-01
2.57983625e-01 2.94862330e-01 -3.07065487e-01 8.42273295e-01
-4.25522268e-01 2.02951714e-01 -1.16007209e-01 -6.42547011e-01
-9.96273279e-01 4.66604024e-01 6.15639985e-01 2.43079856e-01
-1.04976261e+00 5.63287675e-01 3.14832658e-01 -7.33617246e-01
4.23925281e-01 -4.18968856e-01 -2.49587223e-01 -2.95340061e-01
8.88873398e-01 5.17348945e-01 2.86498636e-01 -3.60613883e-01
-2.50828356e-01 2.22046673e-01 -6.07555434e-02 2.96058118e-01
1.19345987e+00 -8.98296386e-02 -4.08572376e-01 4.11288232e-01
1.11794281e+00 2.65272379e-01 -1.21673453e+00 -2.67594904e-02
-2.09690988e-01 -8.27900350e-01 -7.73964673e-02 -3.67437214e-01
-1.06705844e+00 6.79133058e-01 4.38945979e-01 5.98675549e-01
1.20356297e+00 -6.50435388e-02 5.65305591e-01 4.59411144e-01
1.67059094e-01 -9.19199526e-01 -5.48651516e-02 4.03402448e-01
9.67142165e-01 -8.33436549e-01 -7.08018094e-02 -1.47493139e-01
-9.55992818e-01 1.18120205e+00 8.58742177e-01 -1.95908368e-01
8.73602033e-01 2.44857177e-01 -1.21275619e-01 -8.91497657e-02
-4.09401298e-01 2.13426366e-01 4.84705418e-01 8.83693278e-01
-3.56478505e-02 -1.09803587e-01 3.22167546e-01 6.81510985e-01
-4.97540623e-01 -3.99444222e-01 6.16350889e-01 6.12396657e-01
-7.40422010e-01 -8.35595906e-01 -5.98008335e-01 8.91727626e-01
-2.98548043e-01 3.13504875e-01 -3.11094910e-01 7.08041489e-01
2.31162179e-02 6.98047161e-01 4.16174531e-01 -7.17072785e-02
3.69980514e-01 9.60115492e-02 4.34752032e-02 -6.91119552e-01
4.30024296e-01 -5.17959185e-02 -5.08946061e-01 -3.94403905e-01
3.19049269e-01 -6.38447106e-01 -1.59554410e+00 -1.81627557e-01
-4.34339672e-01 6.62572026e-01 3.19949508e-01 1.01898837e+00
2.85270452e-01 1.52206942e-01 9.71307635e-01 -1.95028678e-01
-9.96246994e-01 -8.91900063e-01 -1.13017941e+00 3.72520775e-01
5.75460017e-01 -7.08766043e-01 -7.22694099e-01 -2.29375541e-01] | [7.588520526885986, 2.4209837913513184] |
c6fd0f54-fa3b-422a-8c91-e067a32e0f40 | sketch-me-that-shoe | null | null | http://openaccess.thecvf.com/content_cvpr_2016/html/Yu_Sketch_Me_That_CVPR_2016_paper.html | http://openaccess.thecvf.com/content_cvpr_2016/papers/Yu_Sketch_Me_That_CVPR_2016_paper.pdf | Sketch Me That Shoe | We investigate the problem of fine-grained sketch-based image retrieval (SBIR), where free-hand human sketches are used as queries to perform instance-level retrieval of images. This is an extremely challenging task because (i) visual comparisons not only need to be fine-grained but also executed cross-domain, (ii) free-hand (finger) sketches are highly abstract, making fine-grained matching harder, and most importantly (iii) annotated cross-domain sketch-photo datasets required for training are scarce, challenging many state-of-the-art machine learning techniques. In this paper, for the first time, we address all these challenges, providing a step towards the capabilities that would underpin a commercial sketch-based image retrieval application. We introduce a new database of 1,432 sketch-photo pairs from two categories with 32,000 fine-grained triplet ranking annotations. We then develop a deep triplet-ranking model for instance-level SBIR with a novel data augmentation and staged pre-training strategy to alleviate the issue of insufficient fine-grained training data. Extensive experiments are carried out to contribute a variety of insights into the challenges of data sufficiency and over-fitting avoidance when training deep networks for fine-grained cross-domain ranking tasks. | ['Chen-Change Loy', 'Yi-Zhe Song', 'Qian Yu', 'Timothy M. Hospedales', 'Tao Xiang', 'Feng Liu'] | 2016-06-01 | null | null | null | cvpr-2016-6 | ['sketch-based-image-retrieval'] | ['computer-vision'] | [ 2.86235362e-01 -7.92054474e-01 -3.13531280e-01 -4.14122671e-01
-1.26595914e+00 -9.63590205e-01 9.09951329e-01 -4.22747098e-02
-3.00093681e-01 5.19912064e-01 -3.94559912e-02 -6.07829317e-02
-4.66648579e-01 -6.23809099e-01 -6.92206204e-01 -3.49622220e-01
9.47883353e-02 7.94047356e-01 1.48460045e-01 -3.87552053e-01
4.92590159e-01 1.07056952e+00 -1.88494515e+00 7.29268312e-01
3.96616340e-01 1.42545724e+00 1.79580346e-01 5.74237645e-01
-2.65933275e-01 3.57064813e-01 -6.67479753e-01 -6.75361037e-01
5.68973184e-01 2.16343716e-01 -6.59597933e-01 -1.64829701e-01
1.43736160e+00 -6.64811134e-01 -4.31018859e-01 7.42117345e-01
6.58454955e-01 1.14803009e-01 7.98625529e-01 -1.54350960e+00
-8.41702461e-01 -1.19984038e-01 -4.38759893e-01 -1.47274375e-01
2.63172835e-01 5.33066280e-02 1.28842318e+00 -1.26395154e+00
8.15641463e-01 1.36232245e+00 3.90458465e-01 6.00546539e-01
-1.00506210e+00 -8.81233990e-01 2.36854199e-02 8.79659951e-02
-1.63539159e+00 -4.86212492e-01 9.21959519e-01 -3.92256141e-01
8.43654335e-01 3.54881525e-01 4.98064965e-01 1.12823796e+00
-3.34221363e-01 7.19884038e-01 9.52442467e-01 -3.04437190e-01
-2.34841675e-01 -1.14807241e-01 -2.29397267e-01 7.97261178e-01
1.13298982e-01 9.00343508e-02 -6.56430006e-01 -4.46620852e-01
1.20389867e+00 3.69777590e-01 2.66941100e-01 -6.64868295e-01
-1.44475377e+00 6.40951395e-01 6.39964342e-01 4.20613706e-01
-2.55097657e-01 4.11498964e-01 5.59559882e-01 6.74441814e-01
1.77533686e-01 6.37239397e-01 -3.08896571e-01 1.02110349e-01
-1.20066726e+00 5.80491424e-01 4.85568285e-01 1.11908126e+00
8.17769945e-01 -3.11640084e-01 -4.24624175e-01 1.34483111e+00
-2.20917072e-03 6.17497146e-01 1.56543493e-01 -8.85052502e-01
6.08270168e-01 5.93500555e-01 1.80295527e-01 -1.02287686e+00
2.42230996e-01 -2.62313820e-02 -9.53612030e-01 2.55744964e-01
4.55524445e-01 7.13932633e-01 -1.02843809e+00 1.40183568e+00
-7.78975859e-02 -3.55395228e-01 -5.20197272e-01 1.12365818e+00
9.70774889e-01 3.00703734e-01 1.39955491e-01 3.91551733e-01
1.48653543e+00 -9.47671354e-01 -1.78679854e-01 3.95801328e-02
-4.98590991e-03 -1.13989854e+00 1.43481243e+00 2.32907996e-01
-1.17192817e+00 -8.07313502e-01 -1.06549299e+00 -5.29138982e-01
-8.45758379e-01 3.27626020e-01 7.05840647e-01 2.29695916e-01
-1.11694193e+00 6.87619090e-01 -1.26689196e-01 -3.94956023e-01
6.52687907e-01 3.22906047e-01 -7.51519918e-01 -5.06364644e-01
-1.04963279e+00 6.95480406e-01 1.65168662e-02 -7.96023384e-02
-8.30899835e-01 -8.72924864e-01 -4.56402123e-01 8.64868462e-02
2.04879284e-01 -6.63302720e-01 8.61903906e-01 -6.89328849e-01
-9.22344208e-01 1.32786238e+00 1.95919964e-02 1.91837713e-01
5.40385365e-01 4.55100127e-02 -2.88159639e-01 1.90398514e-01
2.84162998e-01 1.10019672e+00 1.06792295e+00 -1.17526627e+00
-4.37060624e-01 -3.64184260e-01 1.76972717e-01 3.81548889e-02
-2.70424604e-01 1.01729162e-01 -7.92382419e-01 -9.89643574e-01
-1.36939928e-01 -8.41663659e-01 1.31775826e-01 6.82131767e-01
-4.88721542e-02 -6.66694701e-01 7.76089609e-01 -3.19164455e-01
9.66363072e-01 -1.99303508e+00 -1.02191165e-01 2.80636668e-01
2.18438879e-01 4.55518872e-01 -7.51114547e-01 8.86733592e-01
-3.32186483e-02 1.70997128e-01 1.18095674e-01 -2.76129842e-01
3.46850544e-01 4.55192477e-02 -5.95525324e-01 -3.61450426e-02
5.58198214e-01 1.24633837e+00 -8.02937746e-01 -6.80561662e-01
3.05122972e-01 4.10677135e-01 -2.20590949e-01 4.18886900e-01
-1.95875257e-01 1.57756135e-01 -5.53539157e-01 1.17058218e+00
6.38485193e-01 -4.26078707e-01 -1.06004793e-02 -7.07528591e-01
6.03681915e-02 2.03983318e-02 -9.16767240e-01 1.99916530e+00
-5.28458714e-01 6.26339376e-01 -1.88054368e-01 -7.18200743e-01
9.50443923e-01 -6.02245517e-03 4.34649438e-01 -1.15177500e+00
-3.40618759e-01 5.19936621e-01 -4.88568753e-01 -7.90880322e-02
7.91201174e-01 -1.55783370e-01 -2.66679555e-01 5.24669886e-01
9.09310579e-02 -3.63746911e-01 3.14854503e-01 2.94238627e-01
9.25625622e-01 6.89493716e-02 6.97265416e-02 -1.32601663e-01
3.90554845e-01 -1.87038211e-03 1.91304199e-02 1.02427948e+00
-3.83361690e-02 8.37771833e-01 1.44783482e-01 -9.16708708e-01
-1.42415535e+00 -1.17945254e+00 -6.06116615e-02 1.49625254e+00
1.80435002e-01 -3.01713407e-01 -1.57405943e-01 -6.29831314e-01
4.33996290e-01 -3.04714322e-01 -5.77265918e-01 3.08626473e-01
-5.62157035e-01 -3.77861559e-02 7.19615698e-01 5.20929515e-01
4.43701893e-01 -1.15911376e+00 -1.35718003e-01 -4.75925021e-02
-1.65107012e-01 -1.07378483e+00 -8.22366834e-01 -2.29061708e-01
-6.73943460e-01 -1.21568656e+00 -1.24742043e+00 -8.66197228e-01
6.72441483e-01 6.91999972e-01 1.68697786e+00 6.01871073e-01
-6.69929802e-01 5.58733344e-01 -1.82111695e-01 -1.26861453e-01
1.47476355e-02 -1.66164916e-02 -1.59549236e-01 -2.93159962e-01
2.66008079e-01 -4.11949426e-01 -9.50150609e-01 6.65328145e-01
-1.09230149e+00 -2.66995460e-01 9.86250520e-01 1.20284760e+00
7.61309147e-01 -3.37120265e-01 4.84194189e-01 -4.94726837e-01
7.76720703e-01 8.12129602e-02 -7.92342484e-01 8.64712238e-01
-4.16122854e-01 7.60391802e-02 4.62345004e-01 -5.90520799e-01
-6.85130060e-01 -3.49256583e-02 5.31797931e-02 -9.40275311e-01
-6.72841519e-02 1.66675895e-01 9.84956697e-02 -5.35923779e-01
5.75077772e-01 7.36627355e-02 -1.39087692e-01 -5.40068567e-01
5.27909517e-01 6.48794949e-01 5.32678008e-01 -1.10672426e+00
9.26148772e-01 4.25765336e-01 1.13794692e-01 -7.06874073e-01
-6.78857446e-01 -6.87569320e-01 -7.41791189e-01 -6.70688087e-03
5.59829354e-01 -9.59884226e-01 -8.87109280e-01 3.49352360e-01
-1.12670052e+00 -3.31214249e-01 -1.60856947e-01 -1.48966908e-01
-4.75605756e-01 4.22809899e-01 -4.92524803e-01 -6.70112908e-01
-6.19895160e-01 -1.09385037e+00 1.86683559e+00 -9.45944339e-02
1.01703197e-01 -6.02533102e-01 1.20347673e-02 5.19212484e-01
4.57337528e-01 2.05958560e-02 8.47801447e-01 -2.94311762e-01
-1.07347333e+00 -3.30528468e-01 -1.02847004e+00 7.78361037e-02
5.11213280e-02 -2.54416019e-02 -8.64250183e-01 -4.58765984e-01
-9.49810505e-01 -9.34658706e-01 8.11334491e-01 -7.05564767e-02
1.52688575e+00 -8.13785493e-02 -2.55873501e-01 4.52286810e-01
1.47489631e+00 -2.13135540e-01 6.35230720e-01 1.47764266e-01
7.77938902e-01 6.21758103e-01 9.45986211e-01 3.58438045e-01
2.90405452e-01 1.07006693e+00 2.16533557e-01 -2.45084599e-01
-4.37152505e-01 -4.73232538e-01 -3.84606421e-01 5.29988885e-01
-1.47313207e-01 -9.34283957e-02 -6.73768878e-01 7.56757736e-01
-1.69244874e+00 -1.11288476e+00 2.93481231e-01 2.30170465e+00
7.58403540e-01 -3.81080210e-01 2.51159787e-01 -1.03146516e-01
6.29801989e-01 2.87763864e-01 -5.28227925e-01 -1.18521027e-01
-9.17510986e-02 4.96206611e-01 2.72773504e-01 9.25735459e-02
-1.19675040e+00 1.09281683e+00 5.79132986e+00 1.29200149e+00
-1.11181867e+00 -1.79826856e-01 4.79850858e-01 -2.28690673e-02
-3.12216192e-01 -2.92761445e-01 -4.68374848e-01 2.81927884e-01
1.28525719e-01 4.91754025e-01 7.80592382e-01 7.50187159e-01
-5.39283276e-01 2.46338934e-01 -1.40850377e+00 1.64229512e+00
5.84787596e-03 -1.56992185e+00 4.78299022e-01 -1.64569225e-02
7.25711763e-01 1.34675041e-01 1.25490651e-01 2.79320180e-01
1.79658383e-01 -1.16107595e+00 5.85386813e-01 5.86904109e-01
1.70958078e+00 -4.94580090e-01 2.87100673e-01 -1.51481688e-01
-1.50882137e+00 -1.80517361e-02 -6.63316488e-01 3.08159351e-01
-2.26563737e-01 3.96039039e-01 -6.00247741e-01 5.77597201e-01
7.93942988e-01 4.75841522e-01 -5.31080127e-01 8.34130049e-01
2.00475797e-01 -3.23065460e-01 -2.26524815e-01 -9.29257460e-03
2.85866052e-01 3.02275568e-02 6.46416694e-02 1.14704907e+00
3.24444890e-01 7.14141726e-02 2.61688009e-02 8.63582611e-01
-4.40660715e-01 -1.77112967e-01 -8.16057503e-01 -3.08599919e-01
7.45690823e-01 1.45878589e+00 -4.79797244e-01 -3.81961048e-01
-2.29293615e-01 1.16707957e+00 5.75303733e-01 3.82371396e-01
-3.32017839e-01 -5.13553381e-01 9.80703890e-01 -2.41938769e-03
3.48875344e-01 -2.08640710e-01 3.79926385e-03 -1.18185043e+00
2.64453769e-01 -8.66273284e-01 4.32343960e-01 -1.02953529e+00
-2.07885408e+00 4.97546792e-01 -2.38397971e-01 -1.33928907e+00
-3.31125975e-01 -8.65412652e-01 -3.28548044e-01 9.86663222e-01
-1.83715415e+00 -1.79667604e+00 -6.34439945e-01 8.60870063e-01
5.03810763e-01 -2.63234705e-01 8.81768465e-01 8.11337948e-01
1.48679391e-01 1.06143522e+00 -4.03394438e-02 2.97792405e-01
1.19705856e+00 -9.38689232e-01 5.36528945e-01 1.33960783e-01
4.36933607e-01 8.91496778e-01 5.10491841e-02 -5.00601530e-01
-1.79335642e+00 -9.54021156e-01 8.49336088e-01 -7.40664244e-01
6.25026226e-01 -6.14363015e-01 -7.01604187e-01 9.72018614e-02
-2.69707322e-01 4.43374008e-01 4.05430555e-01 2.07180440e-01
-9.14609849e-01 -3.70285213e-01 -1.15594316e+00 4.95194972e-01
1.33805847e+00 -1.18865609e+00 -3.56248319e-01 4.19743985e-01
3.46059948e-01 -2.62554169e-01 -1.17780626e+00 3.18490267e-01
1.48216856e+00 -8.48370552e-01 1.58854210e+00 -7.23458946e-01
4.15012807e-01 -1.26010612e-01 -3.94710958e-01 -8.25431883e-01
-3.21193814e-01 -3.10995370e-01 2.43721262e-01 1.12694097e+00
-4.02461365e-02 -2.94644147e-01 8.76702845e-01 5.64249754e-01
1.97938725e-01 -6.43339694e-01 -7.35073447e-01 -8.38999033e-01
6.93526939e-02 -9.78738517e-02 9.23937201e-01 9.99195695e-01
-5.24756908e-01 4.62912656e-02 -4.53222394e-01 -2.32649669e-01
6.97416842e-01 5.53401589e-01 1.04112554e+00 -1.43992615e+00
8.04673508e-02 -6.72291219e-01 -4.21361446e-01 -1.12417054e+00
-7.81354588e-03 -5.70774555e-01 -1.01630174e-01 -1.46231413e+00
3.47883165e-01 -1.04384208e+00 -4.79424179e-01 5.25711477e-01
-7.05423579e-03 9.24021125e-01 4.98589277e-01 6.10850871e-01
-8.17986012e-01 2.95616180e-01 1.40860689e+00 -4.52672005e-01
3.80128056e-01 -3.03831369e-01 -4.33615237e-01 7.70749822e-02
2.62445986e-01 -1.82030588e-01 -4.56448078e-01 -5.77418566e-01
2.42402792e-01 2.31916413e-01 8.45284641e-01 -5.84732592e-01
3.01972598e-01 -2.28625253e-01 5.96115768e-01 -9.25196707e-01
7.17633307e-01 -9.62632120e-01 7.44689405e-02 -2.48977803e-02
-4.42164898e-01 1.41547635e-01 7.58739859e-02 4.16868389e-01
-3.81671876e-01 5.91682792e-02 5.23391426e-01 -3.64746422e-01
-8.43331993e-01 7.71086097e-01 3.25392991e-01 1.13958724e-01
5.40672302e-01 -3.28417093e-01 -5.09703398e-01 -2.72969663e-01
-2.67981291e-01 -1.48590291e-02 7.51518965e-01 6.53056741e-01
7.09194958e-01 -1.82133579e+00 -6.03911996e-01 2.97090024e-01
7.85459518e-01 -2.44889259e-01 4.85841453e-01 2.61864722e-01
-2.91336685e-01 7.76427388e-01 -5.95445931e-01 -5.37932754e-01
-1.28404403e+00 6.03501022e-01 -7.19157383e-02 -4.87283498e-01
-1.65419206e-01 9.17256355e-01 3.95391554e-01 -7.15715826e-01
2.10883707e-01 5.59667610e-02 3.29513222e-01 1.06091470e-01
4.24558192e-01 2.37234727e-01 3.42032373e-01 -4.01206315e-01
-4.66147482e-01 9.24887478e-01 -1.23049550e-01 3.61122820e-03
1.25903523e+00 2.08168894e-01 -1.79153442e-01 3.70655432e-02
1.45521426e+00 -2.80171961e-01 -1.05320919e+00 -3.29376966e-01
-1.10406399e-01 -9.07181561e-01 -2.86856204e-01 -1.15774608e+00
-8.21869731e-01 1.12265158e+00 6.34409666e-01 -4.08887342e-02
1.02464449e+00 7.33495280e-02 8.57341170e-01 7.90581822e-01
5.96690476e-01 -1.11654317e+00 4.79075670e-01 2.05720633e-01
1.35694289e+00 -1.48149347e+00 7.22809285e-02 -1.71244182e-02
-4.15371746e-01 1.17676401e+00 4.36381459e-01 -2.95734316e-01
2.91557074e-01 -1.39084205e-01 3.44101177e-03 -4.60325658e-01
-6.75347447e-01 -1.33847401e-01 9.09828842e-01 4.97099668e-01
4.12630588e-01 -7.13263974e-02 4.35456336e-02 8.62444341e-02
2.29115754e-01 1.52606919e-01 -3.35212022e-01 8.22757363e-01
-7.28014931e-02 -1.51695633e+00 -2.92099416e-01 7.09448516e-01
-1.79019734e-01 -2.62421131e-01 -7.37798393e-01 8.01441014e-01
-2.13533595e-01 5.59686422e-01 1.06344894e-01 -1.89071804e-01
3.80254656e-01 -1.31477401e-01 8.67709517e-01 -3.04938436e-01
-5.50854266e-01 -2.60790199e-01 2.16434943e-03 -7.40063906e-01
-3.29623193e-01 -2.01152906e-01 -4.47301626e-01 -4.23669934e-01
-1.76975250e-01 -1.15747996e-01 8.39734852e-01 6.68399394e-01
7.28551447e-01 6.81798086e-02 5.55451155e-01 -1.26512015e+00
-5.95070064e-01 -7.24816322e-01 -4.36969668e-01 7.16804326e-01
3.65872294e-01 -9.24039304e-01 7.92710036e-02 -3.17457616e-01] | [11.606618881225586, 0.5889549851417542] |
9b892554-c427-4f08-98aa-e7944b293375 | impact-of-experiencing-misrecognition-by | 2306.07302 | null | https://arxiv.org/abs/2306.07302v1 | https://arxiv.org/pdf/2306.07302v1.pdf | Impact of Experiencing Misrecognition by Teachable Agents on Learning and Rapport | While speech-enabled teachable agents have some advantages over typing-based ones, they are vulnerable to errors stemming from misrecognition by automatic speech recognition (ASR). These errors may propagate, resulting in unexpected changes in the flow of conversation. We analyzed how such changes are linked with learning gains and learners' rapport with the agents. Our results show they are not related to learning gains or rapport, regardless of the types of responses the agents should have returned given the correct input from learners without ASR errors. We also discuss the implications for optimal error-recovery policies for teachable agents that can be drawn from these findings. | ['Erin Walker', 'Adriana Kovashka', 'Timothy Nokes-Malach', 'Nikki Lobczowski', 'Mingzhi Yu', 'Diane Litman', 'Yuya Asano'] | 2023-06-11 | null | null | null | null | ['automatic-speech-recognition'] | ['speech'] | [-9.03465450e-02 4.54700232e-01 5.02313673e-02 -4.63204116e-01
-5.53751349e-01 -8.04632485e-01 6.89613163e-01 3.89342844e-01
-5.87991059e-01 9.56594288e-01 4.22526151e-01 -6.93017364e-01
-2.72010893e-01 -7.46684134e-01 -7.23703384e-01 -1.87605441e-01
1.27185866e-01 4.40880120e-01 3.52059364e-01 -6.97184265e-01
3.53533059e-01 5.15920579e-01 -1.82528841e+00 4.46203500e-01
1.19743276e+00 -5.57193533e-02 1.37117922e-01 1.01160979e+00
-4.82566357e-01 1.60810041e+00 -1.35524893e+00 -3.97893637e-01
-2.94126421e-01 -2.97060758e-01 -1.14418614e+00 -1.91985052e-02
3.73688400e-01 -7.97579467e-01 -7.63488173e-01 8.87461662e-01
1.59254208e-01 5.53122163e-01 3.58041853e-01 -1.44823396e+00
-6.22870684e-01 9.47668076e-01 6.14705324e-01 2.46873528e-01
1.02572370e+00 3.90903383e-01 3.81832421e-01 -3.68738204e-01
5.99940658e-01 1.47606444e+00 5.45760691e-01 1.03659904e+00
-8.36660981e-01 -3.66753519e-01 1.08749039e-01 1.37480110e-01
-8.40406299e-01 -9.32738602e-01 3.87652703e-02 -3.70532125e-01
1.15298223e+00 5.80129504e-01 4.84229028e-01 1.30769658e+00
-1.13532208e-01 6.19134903e-01 1.29846680e+00 -4.86226112e-01
4.52189371e-02 7.38111019e-01 6.47792161e-01 6.34031296e-01
1.61611006e-01 3.62954259e-01 -9.57231820e-01 -8.40377808e-02
5.82620144e-01 -2.39436463e-01 -4.31410789e-01 3.65600169e-01
-9.20245767e-01 5.25319755e-01 -6.59412667e-02 4.00352657e-01
-4.02827650e-01 -8.11623335e-02 1.64341792e-01 1.17982507e+00
1.98656201e-01 8.49204481e-01 -4.47747976e-01 -1.00316072e+00
-5.98480143e-02 1.34481698e-01 1.27413654e+00 1.10947514e+00
6.47573948e-01 3.13328952e-01 -1.55415423e-02 1.00228810e+00
3.87904555e-01 6.63194835e-01 4.67122972e-01 -9.23903406e-01
4.01007831e-01 7.43303835e-01 3.00189435e-01 -7.61108398e-01
-2.74966210e-01 -1.29283301e-03 3.41400266e-01 3.39998692e-01
9.33752894e-01 -3.47800910e-01 -5.65046549e-01 1.43270028e+00
8.82938504e-02 -1.41263008e-01 3.79000425e-01 4.99781162e-01
9.78020728e-01 5.36723793e-01 -7.27117136e-02 -3.11798453e-01
9.59918082e-01 -7.94557333e-01 -9.91455793e-01 -1.82550877e-01
1.07365882e+00 -1.05795693e+00 1.17053008e+00 2.26683542e-01
-1.05603051e+00 -1.26136526e-01 -6.34899497e-01 3.30730200e-01
-3.15304875e-01 -6.74126625e-01 4.57863510e-01 1.06455612e+00
-1.12179685e+00 6.66728139e-01 -8.35214794e-01 -5.60815096e-01
-2.55331576e-01 2.97557592e-01 -7.63289630e-02 4.17300552e-01
-8.96602571e-01 1.23192942e+00 -3.54247689e-01 -1.86461210e-01
-9.64010537e-01 -4.13715243e-01 -4.96086240e-01 -1.16334468e-01
4.49388474e-01 -1.35820672e-01 1.92281282e+00 -1.20246756e+00
-2.20360160e+00 5.31424880e-01 -1.26286581e-01 -1.15305349e-01
5.38594961e-01 -1.40589520e-01 -3.54935706e-01 -1.69123076e-02
-2.48260081e-01 6.61978777e-03 2.85276026e-01 -1.10634899e+00
-8.51270020e-01 -2.57297546e-01 3.46369445e-01 2.99916476e-01
-2.13896394e-01 2.24419057e-01 5.24049640e-01 -3.23440015e-01
-2.49104127e-01 -9.18282926e-01 2.98787981e-01 -5.11874080e-01
4.57727723e-02 -7.34800041e-01 5.30266583e-01 -5.05287647e-01
1.10011518e+00 -1.95784009e+00 -2.60725290e-01 5.99523410e-02
1.02832712e-01 6.90185666e-01 -5.05967855e-01 9.47533488e-01
3.09090316e-01 2.83419281e-01 6.22646213e-01 1.66017096e-02
2.30136104e-02 3.68894964e-01 -3.50417674e-01 6.80779219e-02
-1.32742658e-01 4.34702098e-01 -9.64804411e-01 8.18581805e-02
2.41939411e-01 3.48939389e-01 -3.93749952e-01 9.47865665e-01
-7.92690143e-02 2.81430572e-01 -3.67973506e-01 2.27452412e-01
1.76402956e-01 8.62901360e-02 5.20211235e-02 8.21842849e-01
-4.25720334e-01 1.39044142e+00 -5.23579359e-01 7.43219793e-01
-5.08734584e-01 8.60903859e-01 2.65731156e-01 -5.54709792e-01
1.04724002e+00 6.34391487e-01 -3.16790253e-01 -6.54640019e-01
-1.16375647e-01 2.42641181e-01 5.48332572e-01 -7.97650158e-01
5.94691575e-01 2.21441418e-01 4.33052778e-01 1.01588321e+00
-1.23520233e-01 -5.37126437e-02 -1.64094660e-02 4.08426791e-01
1.11579490e+00 -4.83916819e-01 -9.08603296e-02 3.58925238e-02
3.55669290e-01 1.93198577e-01 2.49786496e-01 1.08594263e+00
-3.54436904e-01 -3.31988543e-01 3.15796345e-01 -2.73473382e-01
-7.07125843e-01 -7.38481700e-01 1.80352941e-01 1.52742517e+00
-4.40890081e-02 -4.22414541e-01 -9.35641825e-01 -7.09136188e-01
-7.45203570e-02 1.19155371e+00 -1.00713506e-01 -4.02123839e-01
-5.87920964e-01 9.46535841e-02 9.10748482e-01 3.84707510e-01
1.24618262e-02 -1.14731443e+00 -5.60258627e-01 3.26996773e-01
-2.92254627e-01 -7.64292657e-01 -5.49537539e-01 -1.10850170e-01
-7.82050729e-01 -1.13335645e+00 -1.59759715e-01 -5.46844721e-01
7.39081681e-01 6.97338820e-01 7.67211080e-01 1.16016948e+00
2.72813410e-01 1.20544136e+00 -7.92309701e-01 -4.03481007e-01
-1.51747179e+00 2.23693699e-02 4.89972144e-01 -4.46003318e-01
3.93335313e-01 -2.41890252e-01 5.98971674e-04 6.50152326e-01
-4.06393081e-01 -1.85136586e-01 3.14745642e-02 7.34161258e-01
-6.61128342e-01 -3.09488952e-01 8.50976348e-01 -8.84078324e-01
1.20356989e+00 -1.78428322e-01 -3.05868119e-01 4.97098118e-01
-9.73238289e-01 1.50463693e-02 5.11429846e-01 -6.75591946e-01
-1.18892610e+00 -6.30045712e-01 -2.51704752e-01 1.39551073e-01
-7.13467300e-01 2.54968643e-01 4.94077235e-01 -4.46831465e-01
8.49332750e-01 3.84570032e-01 5.72680652e-01 -3.08876932e-01
-1.07913293e-01 1.33070457e+00 -3.62789296e-02 -7.48447776e-01
5.30087173e-01 -3.83690834e-01 -1.02495277e+00 -1.37092507e+00
-2.30332002e-01 4.16330760e-03 -8.63506347e-02 -6.92685544e-01
3.25923681e-01 -5.93404591e-01 -1.34037924e+00 7.86816657e-01
-1.27321720e+00 -8.17513466e-01 -5.32062985e-02 7.10434318e-01
-4.38932091e-01 1.65791914e-01 -9.77909029e-01 -1.41576231e+00
2.92416997e-02 -1.38542390e+00 7.41055831e-02 5.91253638e-01
-8.36430848e-01 -1.01280308e+00 -1.47901580e-01 8.03556442e-01
7.50745296e-01 -9.31224823e-01 9.86598790e-01 -1.38692653e+00
-7.47041047e-01 1.04540825e-01 4.77745712e-01 1.50230080e-01
3.06150705e-01 3.56247425e-01 -8.34600031e-01 -2.58352071e-01
-5.44922948e-02 -5.90121508e-01 -8.36106017e-03 -1.89148366e-01
2.40838647e-01 -1.16761827e+00 -3.65717299e-02 -5.20455651e-02
6.50814295e-01 8.67099702e-01 2.94027895e-01 4.41769153e-01
2.94506967e-01 9.92363572e-01 3.88587862e-01 -3.29263769e-02
4.08426106e-01 7.03629136e-01 -4.86686565e-02 6.39302313e-01
-1.98971286e-01 -6.65006876e-01 9.84507382e-01 1.19503081e+00
-1.23956904e-01 -2.73191839e-01 -9.68500912e-01 4.11877632e-01
-1.64348710e+00 -1.11111093e+00 -3.69725794e-01 2.27162838e+00
1.04406941e+00 -1.38762906e-01 4.91348982e-01 -1.38367236e-01
7.40552247e-01 -2.91827023e-02 -2.08444417e-01 -9.11170006e-01
2.42042571e-01 -3.68678756e-02 2.71515429e-01 1.29495203e+00
1.07385203e-01 9.04706419e-01 7.84801865e+00 1.50546253e-01
-1.09881449e+00 4.39170301e-02 1.32246897e-01 8.49560350e-02
-6.66728914e-01 -2.20811740e-01 -7.11597621e-01 3.41557533e-01
1.29086840e+00 -4.35018301e-01 9.34493661e-01 6.38514221e-01
2.09451437e-01 -3.22738551e-02 -1.18545532e+00 2.54912198e-01
-1.23352036e-01 -1.07039797e+00 -1.25660136e-01 -2.60444045e-01
1.83714747e-01 -2.19847202e-01 -1.68005899e-01 4.65307325e-01
9.05402124e-01 -9.11414504e-01 5.65852642e-01 3.02052259e-01
1.70536488e-01 -6.19061232e-01 4.25076693e-01 6.60203576e-01
-3.82644743e-01 -7.66766071e-02 -1.64170191e-01 -6.02017343e-01
-3.85395914e-01 -3.37519884e-01 -1.53953016e+00 -1.86363220e-01
3.08908343e-01 -6.58806562e-02 -1.47694200e-01 6.67876184e-01
-5.84958971e-01 1.01040626e+00 -5.17019965e-02 -8.19344878e-01
-1.10479929e-01 -1.13735348e-02 8.72939408e-01 9.45771694e-01
-1.06958924e-02 4.52589184e-01 -1.37977645e-01 5.92965603e-01
1.99011639e-01 -4.18393724e-02 -8.12620997e-01 -5.03215730e-01
1.33089960e+00 6.78925812e-01 -3.40327352e-01 -4.13451046e-01
-3.80122364e-01 6.53334975e-01 5.14836490e-01 5.23210049e-01
-1.74370572e-01 -3.48291010e-01 1.09850490e+00 2.28770167e-01
-4.83377606e-01 -1.83589354e-01 -2.06145039e-03 -9.10102963e-01
-2.63327658e-01 -1.50816464e+00 -5.99034727e-02 -6.00097656e-01
-1.21762145e+00 5.12267828e-01 -3.02035153e-01 -5.69819629e-01
-3.44430596e-01 -3.09266776e-01 -6.73781753e-01 6.00084066e-01
-1.02178764e+00 -2.28680477e-01 -4.54499898e-03 4.09805268e-01
7.85540640e-01 -3.13072503e-01 1.04859853e+00 6.55165315e-02
-6.14384294e-01 1.05924916e+00 -3.02257925e-01 1.40977681e-01
7.83855379e-01 -1.01545477e+00 1.99713290e-01 5.42291045e-01
-4.76353429e-02 1.06074691e+00 8.02939236e-01 -6.38066232e-01
-1.60254347e+00 -5.11255264e-01 1.09440148e+00 -6.78733230e-01
6.66804254e-01 2.68293712e-02 -1.25603592e+00 1.15198374e+00
4.74936545e-01 -8.74422967e-01 6.88185096e-01 3.44240129e-01
-3.71295780e-01 1.71070546e-01 -1.09391713e+00 7.26069212e-01
1.14689624e+00 -9.10077512e-01 -8.74874473e-01 3.19840968e-01
1.02613521e+00 -4.98477042e-01 -8.82951915e-01 -1.74912468e-01
3.91508102e-01 -8.94153178e-01 6.13548279e-01 -9.86229062e-01
1.91240102e-01 1.52655706e-01 2.23998383e-01 -1.74455512e+00
-1.74620256e-01 -9.52752769e-01 3.61887485e-01 1.30396175e+00
7.26130903e-01 -1.32922673e+00 4.04581606e-01 1.29100645e+00
-4.01705265e-01 -1.18665121e-01 -7.17734158e-01 -7.55872607e-01
5.47918379e-02 -2.11686894e-01 7.36590922e-01 1.27048421e+00
7.99999297e-01 1.96536928e-01 -2.21685886e-01 3.51051807e-01
1.24416456e-01 -6.52132094e-01 9.04391468e-01 -8.79539847e-01
-8.02348331e-02 -5.15589714e-01 -1.68189302e-01 -1.12074351e+00
2.31700808e-01 -6.41949177e-01 3.27385545e-01 -1.10304022e+00
-3.57742399e-01 -7.84455717e-01 3.84340107e-01 6.22039199e-01
-3.20754021e-01 -8.27756226e-01 4.85122591e-01 2.22725958e-01
-4.14521903e-01 3.43197763e-01 1.03824019e+00 1.88211426e-01
-5.22205532e-01 4.18239176e-01 -5.63478529e-01 7.03752100e-01
6.63693845e-01 -6.33187413e-01 -4.81214494e-01 -5.53031087e-01
1.26316339e-01 6.96164787e-01 6.32472411e-02 -5.83260298e-01
6.35262191e-01 -6.34186745e-01 -4.83844489e-01 9.31222662e-02
-2.15802304e-02 -7.31337428e-01 -1.18611590e-03 6.25305295e-01
-9.13570642e-01 3.02642077e-01 2.88179725e-01 1.44976214e-01
3.70391607e-02 -5.90711474e-01 5.00944018e-01 1.48218036e-01
-2.37854913e-01 -3.27714890e-01 -1.60523379e+00 -9.08271968e-02
9.77674484e-01 -2.16182262e-01 -1.08599591e+00 -8.26177835e-01
-4.86518621e-01 2.84797817e-01 5.43067217e-01 7.57356226e-01
6.74138188e-01 -9.03321862e-01 -5.91966391e-01 2.65856743e-01
3.91893461e-03 -3.93919200e-01 -1.13908224e-01 5.71848094e-01
-5.17536700e-01 5.45669436e-01 -1.47069395e-01 -3.27746749e-01
-1.74273109e+00 5.11035845e-02 5.95038891e-01 4.35952723e-01
-3.72382134e-01 8.46095502e-01 -3.30315202e-01 -9.29900169e-01
8.67326796e-01 -2.00712994e-01 -6.42840862e-02 -2.00501427e-01
8.71205032e-01 8.52551341e-01 1.01661719e-01 -2.69104928e-01
-3.33029866e-01 -1.94633573e-01 -4.25745636e-01 -2.37544864e-01
1.00204086e+00 -2.48748139e-01 4.01010439e-02 7.81718969e-01
4.88295794e-01 3.01089585e-01 -7.46598840e-01 -1.81005672e-01
-9.83809382e-02 -7.48329818e-01 -3.89528602e-01 -9.38109100e-01
-6.09556675e-01 6.78187311e-01 2.04656839e-01 7.03050375e-01
1.55247286e-01 -1.37057915e-01 5.22232175e-01 7.85833776e-01
5.10978401e-01 -1.12863815e+00 3.34507018e-01 7.54800677e-01
6.74791336e-01 -9.45158958e-01 -6.05966985e-01 -4.65798140e-01
-5.51159441e-01 1.18359709e+00 1.04037428e+00 4.18912441e-01
1.29881382e-01 3.56041938e-01 7.01985598e-01 -1.45606026e-01
-1.40977168e+00 1.54636368e-01 -3.45425725e-01 1.07618761e+00
9.06274319e-01 3.05619001e-01 -4.26711321e-01 1.48933738e-01
-4.70586210e-01 -1.69147193e-01 1.24963593e+00 1.00569367e+00
-9.20981884e-01 -1.10204279e+00 -6.62796199e-01 2.46201977e-01
-5.58578335e-02 8.84341374e-02 -8.71579349e-01 5.53525865e-01
-6.62658393e-01 1.77598214e+00 1.01973496e-01 -6.95978165e-01
6.52990818e-01 4.56412882e-01 3.56377363e-01 -9.51576293e-01
-1.14764094e+00 -6.56027555e-01 8.28024626e-01 -3.34576070e-01
-1.85552076e-01 -6.69299960e-01 -1.37391281e+00 -1.00610960e+00
-6.14395142e-01 5.14256775e-01 3.69202048e-01 1.02379453e+00
2.21981138e-01 4.00166959e-01 6.75237775e-01 7.45717734e-02
-9.98363078e-01 -9.90801990e-01 -1.70526028e-01 3.79928261e-01
4.63943243e-01 -2.42087469e-01 -6.18541420e-01 -4.33610290e-01] | [12.359941482543945, 7.957873821258545] |
2f9a01a7-8bf2-47f7-89ff-99f8f3cd4ca7 | docee-a-large-scale-dataset-for-document | null | null | https://openreview.net/forum?id=t5zJTjgwngX | https://openreview.net/pdf?id=t5zJTjgwngX | DocEE: A Large-Scale Dataset for Document-level Event Extraction | Event extraction (EE) is the task of identifying events and their types, along with the involved arguments. Despite the great success in sentence-level event extraction, events are more naturally presented in the form of document, with event arguments scattering in multiple sentences. However, a major barrier to promote document-level event extraction has been the lack of large-scale and practical training and evaluation datasets. In this paper, we present DocEE, a new document-level EE dataset including 20,000+ events, 100,000+ arguments. We highlight three features: large-scale annotations, fine-grained event arguments and application-oriented settings. Experiments show that even SOTA models show inferior performance on DocEE, especially in cross-domain settings, indicating that DocEE is still a challenging task. We will publish DocEE upon acceptance. | ['Anonymous'] | 2021-09-17 | null | null | null | acl-arr-september-2021-9 | ['document-level-event-extraction'] | ['natural-language-processing'] | [ 3.08532357e-01 2.31341153e-01 -1.41604379e-01 -2.84196705e-01
-1.38543987e+00 -9.90037978e-01 9.79055405e-01 7.82733083e-01
-7.22634852e-01 1.03088355e+00 8.00996482e-01 -2.76579142e-01
-1.51266098e-01 -6.66998625e-01 -7.83570945e-01 -1.35390118e-01
-3.56267780e-01 4.52557772e-01 4.47411805e-01 4.99589294e-02
-4.95518968e-02 3.21158022e-01 -1.29866135e+00 9.76749659e-01
4.40917909e-01 8.99341822e-01 -1.59819812e-01 6.26793087e-01
-3.29199702e-01 1.27224183e+00 -1.02972281e+00 -7.65565217e-01
-3.19540858e-01 -4.25758451e-01 -1.05058920e+00 -3.98525000e-01
2.13703457e-02 -9.77263972e-02 -4.44590189e-02 4.86059219e-01
6.44710660e-01 -4.82066385e-02 6.26176059e-01 -1.20468926e+00
-3.03330477e-02 1.23640418e+00 -2.16242760e-01 8.28190684e-01
6.25022352e-01 -4.11258161e-01 1.41468835e+00 -1.03001332e+00
1.17821586e+00 1.13046288e+00 6.96339309e-01 1.01520583e-01
-9.33506489e-01 -4.90254670e-01 1.43019736e-01 1.08972237e-01
-9.15462792e-01 -3.56839389e-01 5.84875047e-01 -3.00143421e-01
1.61085343e+00 5.16507804e-01 5.35676301e-01 1.57793903e+00
1.97857574e-01 1.14158082e+00 9.41436231e-01 -4.98073518e-01
1.72151640e-01 -1.16374373e-01 3.57838005e-01 1.72089338e-02
3.08043033e-01 -1.42641887e-01 -7.89106667e-01 -5.88480532e-01
1.98573440e-01 -1.60078660e-01 7.36904591e-02 4.26036596e-01
-1.47530329e+00 7.19391167e-01 -7.67141208e-02 4.00325924e-01
-6.53437495e-01 -4.72783409e-02 9.94551897e-01 1.51953086e-01
5.71202278e-01 3.18391800e-01 -1.07924485e+00 -5.86928785e-01
-7.32138753e-01 7.39110947e-01 1.01163936e+00 8.62600327e-01
-2.34128069e-02 -3.47924173e-01 -5.31592667e-01 8.08048904e-01
-4.71301712e-02 2.17499495e-01 1.80850729e-01 -5.96687794e-01
1.17237508e+00 6.42725587e-01 2.19891027e-01 -8.25648844e-01
-6.28021181e-01 -2.00382307e-01 -3.42326343e-01 -3.12172621e-01
4.35489297e-01 -5.01480401e-01 -3.52430046e-01 1.76964116e+00
3.69785905e-01 -2.43904646e-02 2.10723385e-01 4.98341560e-01
1.22100914e+00 7.72244513e-01 5.52260637e-01 -2.95233220e-01
1.79076326e+00 -5.32200873e-01 -9.32941318e-01 -5.68900108e-01
5.55652916e-01 -9.53318536e-01 6.67987585e-01 2.92301238e-01
-1.33279693e+00 1.12237059e-01 -6.23250902e-01 -8.88108611e-02
-5.24605274e-01 1.88826308e-01 9.40363824e-01 2.36074015e-01
-1.64630190e-01 2.33518720e-01 -7.07488179e-01 -3.73511046e-01
4.22568828e-01 1.06505631e-02 -3.03865463e-01 4.36334044e-01
-1.57978153e+00 9.17403460e-01 7.93786287e-01 -3.00402075e-01
-4.98372614e-01 -7.97681808e-01 -8.14424396e-01 5.13636656e-02
6.56139076e-01 -1.30169615e-01 1.53477871e+00 -1.77653059e-01
-7.79157043e-01 9.28788304e-01 -1.62242055e-01 -6.39087915e-01
4.23288405e-01 -4.13080633e-01 -9.20544088e-01 1.93676159e-01
3.41357321e-01 2.41571948e-01 3.13512564e-01 -6.17127538e-01
-1.15849948e+00 -4.97221872e-02 1.00153059e-01 1.41927339e-02
-2.71971941e-01 8.96789670e-01 -3.14538807e-01 -8.42309773e-01
-2.70267427e-01 -5.06448388e-01 -4.62520644e-02 -6.20377302e-01
-5.21798074e-01 -8.70654047e-01 5.82681775e-01 -5.87145329e-01
1.51333046e+00 -2.24323726e+00 -3.16052258e-01 -3.25335830e-01
1.08507805e-01 -1.00927241e-01 2.26883695e-01 9.02986169e-01
-3.84456187e-01 7.68582821e-02 6.78730384e-02 -2.36164942e-01
2.50501782e-01 1.75673902e-01 -7.15207696e-01 -1.92839764e-02
7.13293076e-01 1.01508558e+00 -1.16635239e+00 -8.08788955e-01
-8.66321251e-02 1.97993964e-01 -4.28913176e-01 2.34461315e-02
-2.38363698e-01 -1.28528690e-02 -6.81578636e-01 6.20965064e-01
3.72275352e-01 -5.11330485e-01 2.46283948e-01 -3.41472417e-01
-1.74155638e-01 1.16254401e+00 -1.37726569e+00 1.27566957e+00
-2.17196763e-01 7.21273184e-01 -7.18398765e-02 -8.41981590e-01
3.00571382e-01 7.03005970e-01 6.19101942e-01 -5.20162940e-01
1.27250224e-01 2.75899470e-01 -1.62016913e-01 -3.09682190e-01
6.98809564e-01 -4.86480258e-02 -8.31912756e-01 3.56531739e-01
1.24548428e-01 1.04536675e-01 8.71232629e-01 5.91122329e-01
1.42196000e+00 -1.98254570e-01 6.64804578e-01 -4.52829152e-02
1.92956589e-02 3.59492540e-01 6.93280458e-01 7.96019077e-01
9.20460448e-02 6.75438702e-01 8.83886993e-01 -3.59617352e-01
-7.83814371e-01 -9.93962288e-01 -5.89719057e-01 9.66776788e-01
-4.30347770e-01 -1.04023170e+00 -4.00681078e-01 -1.11135340e+00
-2.29136318e-01 7.33113229e-01 -5.11345148e-01 3.26576650e-01
-9.49086964e-01 -1.42751110e+00 9.18666542e-01 7.18898833e-01
1.70272738e-01 -1.37916529e+00 -7.63217092e-01 9.25729632e-01
-6.99905932e-01 -1.45034885e+00 -3.12770993e-01 7.72293091e-01
-4.42089528e-01 -1.25919795e+00 -2.91034251e-01 -4.99709010e-01
9.44555402e-02 -4.19339627e-01 1.73500419e+00 -5.16024709e-01
-2.67053455e-01 1.67775434e-02 -6.14446044e-01 -9.53143835e-01
-4.61139351e-01 -6.74728751e-02 -4.10681248e-01 -4.92918491e-01
5.54419875e-01 -3.13452572e-01 -3.95318657e-01 1.35873826e-02
-1.04674375e+00 -2.28370398e-01 6.06367648e-01 7.26899028e-01
5.32315075e-01 2.89943993e-01 9.09003139e-01 -1.24751079e+00
7.04012871e-01 -6.75540149e-01 -2.44019046e-01 1.04175098e-01
-3.40421826e-01 -2.35709846e-01 4.83595908e-01 -5.74238241e-01
-1.29506886e+00 -2.15902984e-01 -4.31612670e-01 5.27222693e-01
-3.40565324e-01 8.04383039e-01 -6.18562698e-02 9.20055628e-01
8.00297797e-01 -1.02643207e-01 -7.07816958e-01 -3.92532945e-01
1.52129412e-01 4.54189032e-01 5.41765392e-01 -8.00890207e-01
3.06545556e-01 4.72617030e-01 -3.05757284e-01 -5.13817191e-01
-1.31092858e+00 -2.97250479e-01 -1.48768157e-01 1.37450546e-01
8.47826421e-01 -1.11732757e+00 -3.55856776e-01 2.29090273e-01
-1.37303829e+00 -1.39290944e-01 -6.52034402e-01 5.89743376e-01
-7.73262829e-02 2.62823831e-02 -1.00763977e+00 -5.90855122e-01
-3.74802977e-01 -6.08520508e-01 1.19921947e+00 3.24847326e-02
-7.06661284e-01 -8.84877443e-01 1.57376483e-01 4.70427088e-02
-1.18466973e-01 5.78564227e-01 7.51262605e-01 -1.03325057e+00
-2.03102782e-01 -3.86444032e-01 -2.22288266e-01 -1.80144429e-01
-2.08897926e-02 -1.21485665e-02 -7.37686098e-01 2.57621557e-01
-2.89635867e-01 -5.22629619e-01 7.70722687e-01 3.75053704e-01
1.02708149e+00 -2.37607732e-01 -5.94123065e-01 -1.42964320e-02
9.82499421e-01 2.13067129e-01 4.78906393e-01 5.40693223e-01
1.60400793e-01 5.14315784e-01 7.96976686e-01 7.24195242e-01
4.38533366e-01 5.18985271e-01 6.14878088e-02 -1.38769895e-01
-1.92261711e-01 -2.60751337e-01 4.57936376e-01 6.34985507e-01
8.36459920e-02 -7.80373156e-01 -7.88757324e-01 8.22435975e-01
-1.74694538e+00 -1.32587242e+00 -6.55335605e-01 1.60507798e+00
1.17401719e+00 6.58919156e-01 2.14181945e-01 4.65533525e-01
4.16188449e-01 2.95649767e-01 -5.48788980e-02 -2.71094918e-01
-5.66732764e-01 3.35875720e-01 2.45270446e-01 -1.33744970e-01
-1.42260969e+00 6.44225478e-01 7.09230232e+00 1.05114067e+00
-7.28286922e-01 2.51135796e-01 4.13419932e-01 -2.01618269e-01
-2.33694300e-01 1.16209006e-02 -1.28597379e+00 6.82374179e-01
1.26313722e+00 -6.92901760e-02 -3.16112846e-01 5.69949448e-01
1.40438631e-01 -7.64198303e-02 -1.17200327e+00 5.86286008e-01
-2.33414799e-01 -1.81598639e+00 -1.66056558e-01 -6.10929616e-02
4.21329349e-01 2.65774012e-01 -4.28939730e-01 5.69670260e-01
4.50203121e-01 -4.95389342e-01 1.13821828e+00 -1.36980355e-01
5.09686768e-01 -5.50909519e-01 6.55078650e-01 2.00894967e-01
-1.32228398e+00 6.96970001e-02 3.16502191e-02 2.16059741e-02
7.47408271e-01 1.08589351e+00 -7.34879076e-01 7.78578639e-01
9.47458148e-01 8.09996307e-01 -4.26758707e-01 7.97433197e-01
-1.67029813e-01 1.14507282e+00 -6.32713437e-01 -2.13697404e-01
1.81841969e-01 4.21276152e-01 5.84455490e-01 1.85567057e+00
1.88550144e-01 3.60497922e-01 9.96830687e-02 5.48270702e-01
-2.24920020e-01 4.59358171e-02 -3.25550914e-01 -3.97743225e-01
5.20819724e-01 1.32502711e+00 -9.76435959e-01 -5.81357598e-01
-6.86984956e-01 6.71128035e-01 4.01965946e-01 8.62090886e-02
-1.14506483e+00 -3.30126494e-01 3.30846936e-01 -6.09037802e-02
3.81805837e-01 1.85757473e-01 -8.37777257e-02 -1.26689756e+00
1.71889961e-01 -9.16036367e-01 1.13821471e+00 -4.87008393e-01
-1.53366876e+00 7.11102366e-01 3.04677933e-01 -1.05189514e+00
-4.99398440e-01 -5.93086541e-01 -6.53420329e-01 3.76594722e-01
-1.26966071e+00 -9.36267436e-01 1.83908403e-01 4.36475694e-01
6.77518308e-01 1.93271682e-01 8.36462557e-01 7.80545056e-01
-5.46932101e-01 4.47124243e-01 -6.94151372e-02 4.71021712e-01
8.18109572e-01 -1.49439216e+00 4.08317745e-01 6.63328767e-01
3.42804521e-01 3.22145939e-01 8.80545139e-01 -8.61654103e-01
-1.15390587e+00 -1.01778257e+00 1.64652777e+00 -1.03782022e+00
9.78472888e-01 -6.03225172e-01 -6.28920794e-01 8.93293202e-01
2.26668999e-01 -5.63160628e-02 6.97209001e-01 4.56073523e-01
-1.94310263e-01 1.80127621e-01 -9.33295548e-01 3.99230808e-01
1.07397485e+00 -5.88125527e-01 -1.10671473e+00 7.07748890e-01
6.59767866e-01 -6.94528580e-01 -1.05749023e+00 3.90832394e-01
1.01637878e-01 -5.22767007e-01 1.21414673e+00 -6.25470757e-01
7.04612255e-01 -5.14459722e-02 -3.25017497e-02 -8.79748523e-01
-5.42198792e-02 -4.19521481e-01 -4.25074846e-01 1.78348196e+00
8.48729670e-01 -4.33623552e-01 4.99106735e-01 4.70164388e-01
-1.56261146e-01 -4.97349113e-01 -8.07123184e-01 -7.23083794e-01
-1.71886031e-02 -8.99878561e-01 5.17805636e-01 1.03926408e+00
1.57833949e-01 6.27313614e-01 -2.09033698e-01 2.28518248e-01
3.67987067e-01 3.27789456e-01 1.27933428e-01 -1.21649170e+00
-3.82736921e-01 -2.72965193e-01 9.72691178e-02 -5.41729987e-01
9.04955193e-02 -8.32992256e-01 -2.68304907e-03 -1.56259477e+00
4.92160350e-01 -3.19073498e-01 -3.09385419e-01 5.15228331e-01
-4.29572254e-01 1.08560920e-01 -3.16122204e-01 -1.11324433e-02
-8.60361814e-01 1.65910065e-01 7.15440035e-01 7.65491799e-02
-1.11362286e-01 -8.54815096e-02 -4.70887303e-01 8.37552249e-01
5.84173501e-01 -8.52438152e-01 -8.47420171e-02 -2.93632090e-01
5.74466765e-01 1.58344314e-01 3.85770738e-01 -6.55860245e-01
1.82105508e-02 -1.81698501e-01 5.73473155e-01 -1.19891524e+00
1.86350763e-01 -5.22116721e-01 3.28471243e-01 1.36277586e-01
-4.68938500e-01 2.73211569e-01 4.11868811e-01 5.73429048e-01
-5.09958327e-01 -2.86662430e-01 8.36943537e-02 -7.46638551e-02
-6.65862739e-01 -2.03169286e-02 -5.96514702e-01 7.85178125e-01
8.36852133e-01 3.06459993e-01 -4.51672316e-01 5.32102697e-02
-8.32599401e-01 1.75622329e-02 -2.52551526e-01 3.93828899e-01
2.48292446e-01 -1.19652247e+00 -1.04976106e+00 -3.81629795e-01
3.33967060e-01 -1.60377156e-02 4.43123430e-02 6.86589837e-01
-1.02793453e-02 4.34967935e-01 3.58154297e-01 -2.16257483e-01
-1.23222136e+00 3.01200509e-01 -2.34296128e-01 -8.73164654e-01
-1.00934911e+00 7.44946301e-01 -2.34279712e-03 -3.07217866e-01
1.15405217e-01 -5.11895895e-01 -2.46507108e-01 5.15483260e-01
7.88253784e-01 1.70198530e-01 3.55866343e-01 -1.60746053e-01
-7.08916903e-01 -2.51986682e-01 -1.95248097e-01 -2.77332366e-01
1.52045965e+00 1.65236145e-01 -7.35612437e-02 2.82734275e-01
8.61511409e-01 4.30561602e-01 -1.05681705e+00 -1.17530905e-01
5.47395349e-01 4.85247262e-02 -1.25496864e-01 -1.08366013e+00
-4.65054005e-01 3.19000006e-01 -5.76108322e-02 5.50015390e-01
8.90863597e-01 6.42503440e-01 9.22564626e-01 3.11757982e-01
2.02139333e-01 -1.17231858e+00 -1.28582373e-01 6.70551598e-01
8.41520429e-01 -1.08888769e+00 3.55247594e-02 -6.68707073e-01
-4.83764023e-01 7.95089066e-01 3.05933774e-01 5.17105944e-02
5.61705530e-01 7.36305654e-01 -2.18984827e-01 -5.41362882e-01
-1.06222117e+00 -1.19564995e-01 3.22284281e-01 2.84347795e-02
6.91726506e-01 -2.63471510e-02 -6.54767513e-01 1.22243917e+00
-1.85663059e-01 -8.18526968e-02 3.43857110e-01 1.27306628e+00
-2.42954306e-02 -1.34910285e+00 -2.73500651e-01 6.90790534e-01
-1.33596587e+00 -3.13997298e-01 -3.83812308e-01 9.75674272e-01
2.11262047e-01 1.08571780e+00 -5.47221340e-02 3.27204138e-01
6.60999775e-01 2.30597734e-01 3.23647976e-01 -6.10690534e-01
-1.01413488e+00 3.66144627e-02 9.87264812e-01 -4.92971122e-01
-4.85079557e-01 -1.08896840e+00 -1.60630691e+00 2.39521284e-02
-2.49129936e-01 5.37937641e-01 3.78284454e-01 1.15812683e+00
4.59771186e-01 8.62268448e-01 4.64247726e-02 -2.78684199e-01
-2.43992150e-01 -8.14191937e-01 -4.18513566e-01 7.80233681e-01
2.62764911e-03 -5.99223375e-01 -4.13382411e-01 2.01076984e-01] | [9.04712963104248, 9.244243621826172] |
c36d6702-749f-43c2-a39a-778d69ea3d03 | incremental-and-iterative-learning-of-answer | 1802.07966 | null | http://arxiv.org/abs/1802.07966v2 | http://arxiv.org/pdf/1802.07966v2.pdf | Incremental and Iterative Learning of Answer Set Programs from Mutually Distinct Examples | Over the years the Artificial Intelligence (AI) community has produced
several datasets which have given the machine learning algorithms the
opportunity to learn various skills across various domains. However, a subclass
of these machine learning algorithms that aimed at learning logic programs,
namely the Inductive Logic Programming algorithms, have often failed at the
task due to the vastness of these datasets. This has impacted the usability of
knowledge representation and reasoning techniques in the development of AI
systems. In this research, we try to address this scalability issue for the
algorithms that learn answer set programs. We present a sound and complete
algorithm which takes the input in a slightly different manner and performs an
efficient and more user controlled search for a solution. We show via
experiments that our algorithm can learn from two popular datasets from machine
learning community, namely bAbl (a question answering dataset) and MNIST (a
dataset for handwritten digit recognition), which to the best of our knowledge
was not previously possible. The system is publicly available at
https://goo.gl/KdWAcV. This paper is under consideration for acceptance in
TPLP. | ['Arindam Mitra', 'Chitta Baral'] | 2018-02-22 | null | null | null | null | ['handwritten-digit-recognition'] | ['computer-vision'] | [ 1.56772926e-01 2.85439879e-01 -8.31605718e-02 -5.47812223e-01
-4.11453247e-01 -8.16040576e-01 5.27036428e-01 2.97484338e-01
-1.67676225e-01 7.97268212e-01 -3.45013380e-01 -8.19388628e-01
-4.31827039e-01 -1.16244578e+00 -7.59131610e-01 -2.05156147e-01
5.58988824e-02 7.84193695e-01 4.99717236e-01 -1.90502673e-01
4.58538353e-01 3.19185346e-01 -1.84472072e+00 6.99083447e-01
7.45649099e-01 1.05267334e+00 -6.22723140e-02 6.80985332e-01
-3.36104035e-01 1.48918986e+00 -7.60906711e-02 -5.81702173e-01
1.83606401e-01 -2.76223093e-01 -1.42996716e+00 -3.46267432e-01
6.05410695e-01 7.52220526e-02 -1.47452906e-01 9.36908543e-01
1.12318151e-01 -1.35592386e-01 3.57045054e-01 -1.43836391e+00
-5.34310997e-01 8.90567899e-01 5.54671548e-02 5.16543053e-02
5.55567265e-01 -3.54382582e-02 1.42799652e+00 -7.18525708e-01
6.48066342e-01 1.16480017e+00 5.16825378e-01 5.21536648e-01
-1.13238573e+00 -4.08820242e-01 -3.27059865e-01 7.76693285e-01
-1.17413676e+00 -2.35320449e-01 8.13249528e-01 -3.31789225e-01
9.85629082e-01 4.34161514e-01 4.18053448e-01 6.14888906e-01
-2.53268689e-01 9.97960925e-01 1.32266033e+00 -9.74680245e-01
3.59895676e-01 5.22631586e-01 5.62034905e-01 1.04978669e+00
1.85303122e-01 3.77898756e-03 -3.87672633e-01 -1.61227033e-01
5.07099688e-01 -3.62750769e-01 -5.90906627e-02 -4.56360906e-01
-9.07888293e-01 1.07754230e+00 2.51522541e-01 4.81985152e-01
-2.88878996e-02 9.45152789e-02 3.86627197e-01 7.52808869e-01
-1.66264072e-01 6.63849056e-01 -7.22104847e-01 -2.24486038e-01
-5.73778391e-01 5.09941638e-01 1.60743117e+00 8.90161812e-01
7.46841967e-01 -4.36762959e-01 2.29312330e-01 6.05522513e-01
2.91957855e-01 2.44905263e-01 2.49058291e-01 -9.09484029e-01
2.19199210e-01 1.12568593e+00 -1.55259386e-01 -1.11830699e+00
-1.86991498e-01 1.92337021e-01 -4.33719844e-01 1.15525499e-01
8.32556725e-01 1.98808480e-02 -6.52235270e-01 1.42672217e+00
1.74510151e-01 8.12853277e-02 2.23057017e-01 7.18433440e-01
9.87522066e-01 5.77012718e-01 -7.08105937e-02 5.74179180e-02
1.20406449e+00 -8.15695584e-01 -4.06263381e-01 -1.97310358e-01
8.12408090e-01 -5.98672926e-01 1.13138258e+00 9.91800129e-01
-9.15274858e-01 -2.78665632e-01 -9.27777112e-01 -8.58989507e-02
-6.32435620e-01 6.45391922e-03 1.12646663e+00 6.01645291e-01
-8.86722445e-01 3.36632133e-01 -4.94328439e-01 -5.62977612e-01
4.59590226e-01 5.10036349e-01 -3.23598832e-01 -5.60525954e-01
-1.23937106e+00 1.12006855e+00 7.62250721e-01 -1.51718497e-01
-3.76951605e-01 -4.29939240e-01 -6.44459128e-01 -1.41745746e-01
7.97444284e-01 -3.49554777e-01 1.35501158e+00 -1.32527959e+00
-1.41310179e+00 1.10470438e+00 7.32846782e-02 -8.23067307e-01
2.88008660e-01 -2.14816965e-02 -1.80900350e-01 6.05002269e-02
-4.10730004e-01 5.05722284e-01 4.86476600e-01 -9.40843642e-01
-5.35363019e-01 -4.79511112e-01 5.91332734e-01 -3.02361310e-01
-7.98185319e-02 9.37384516e-02 -1.65350065e-01 -2.42341191e-01
-2.05877051e-03 -8.67729723e-01 5.90075962e-02 -2.73118336e-02
-1.39994966e-02 -6.33220613e-01 7.34552443e-01 -2.89902449e-01
1.13294661e+00 -1.72766113e+00 2.36631498e-01 3.31937164e-01
-5.02774343e-02 4.17917222e-01 2.44060054e-01 5.18871725e-01
-2.73370035e-02 1.33192182e-01 -4.11467940e-01 4.16084498e-01
5.65769933e-02 6.95047915e-01 -6.06000841e-01 4.07202840e-02
3.57476383e-01 9.62965667e-01 -8.88626099e-01 -7.03450680e-01
9.71926078e-02 -4.65534553e-02 -7.07546949e-01 2.00220108e-01
-9.94875848e-01 -1.33031365e-02 -5.87784767e-01 8.16284060e-01
3.78000230e-01 -2.12857887e-01 3.60969394e-01 3.72806354e-03
-1.39828682e-01 1.07418492e-01 -1.26969719e+00 1.48322332e+00
-2.10593864e-01 7.77920604e-01 -3.43061000e-01 -1.47805047e+00
1.03941381e+00 3.79599720e-01 1.37333304e-01 -6.23242199e-01
-1.04264626e-02 5.40694177e-01 3.21313828e-01 -1.02449107e+00
2.25357652e-01 -2.96207070e-01 -2.60017980e-02 4.06533808e-01
1.37240618e-01 -4.21911240e-01 4.77137893e-01 -6.51504332e-03
1.15640616e+00 2.77511865e-01 3.54159385e-01 -2.59370714e-01
9.26159739e-01 6.61338806e-01 3.31146598e-01 8.20053518e-01
1.02440901e-02 1.70871839e-01 6.88148677e-01 -8.40113878e-01
-9.93293822e-01 -7.27885306e-01 -2.30517924e-01 1.00134897e+00
-2.37581402e-01 -4.72368956e-01 -4.97622639e-01 -5.16424239e-01
-1.04136718e-02 7.38998294e-01 -3.44767243e-01 -3.40683162e-02
-7.27341890e-01 -3.94944012e-01 7.36679852e-01 4.08947796e-01
5.54935455e-01 -1.43194306e+00 -6.41869545e-01 6.92589208e-02
-6.14224002e-03 -9.81610239e-01 6.17715240e-01 2.80920595e-01
-1.00008845e+00 -1.45273304e+00 -6.10142341e-03 -9.56712186e-01
5.98370731e-01 -5.28218150e-01 1.33301342e+00 2.68490136e-01
-5.24454296e-01 5.12565255e-01 -5.83067298e-01 -7.78616309e-01
-5.59772432e-01 1.97569206e-01 -2.78124958e-01 -2.03092456e-01
8.59982073e-01 -3.86141390e-01 8.07369426e-02 1.02608368e-01
-1.15817225e+00 3.58613022e-02 5.20343721e-01 6.97644591e-01
3.95520031e-01 2.45905071e-01 5.27441204e-01 -1.27488399e+00
5.28573036e-01 -5.51344097e-01 -8.24964702e-01 4.91724223e-01
-5.31502306e-01 4.48388457e-01 7.53055453e-01 -2.21485227e-01
-6.35203600e-01 1.03309594e-01 -9.28786173e-02 -9.88191087e-03
-4.67414588e-01 9.04529393e-01 -1.10303424e-01 -2.16072291e-01
8.31287801e-01 1.34495169e-01 5.81954680e-02 -3.09180528e-01
2.77953297e-01 6.87848866e-01 4.64375138e-01 -9.28365231e-01
7.49610901e-01 1.63562015e-01 2.17775896e-01 -6.69518769e-01
-8.99246573e-01 -2.18933389e-01 -5.15739143e-01 -6.13083951e-02
5.28008580e-01 -2.77530283e-01 -6.82782054e-01 4.59421813e-01
-9.60738719e-01 -3.27066928e-01 -1.61537513e-01 2.39097401e-01
-6.59540832e-01 1.37406573e-01 -2.14037493e-01 -8.25469613e-01
-1.85831621e-01 -8.60530674e-01 3.35544944e-01 2.38944307e-01
-3.26380223e-01 -1.05875671e+00 1.44793540e-01 5.17064035e-01
2.89990485e-01 3.56552124e-01 1.36686790e+00 -9.20519590e-01
-7.44930148e-01 -2.24433064e-01 -1.73454508e-02 4.58009809e-01
-1.26448184e-01 1.28904656e-01 -9.87674832e-01 1.79740131e-01
-1.99469164e-01 -8.18696856e-01 6.42068684e-01 -5.49067603e-03
1.26402771e+00 -3.17907542e-01 -5.81003278e-02 1.02396756e-01
1.62370980e+00 1.80715382e-01 7.83715427e-01 6.56714261e-01
2.67164379e-01 5.95629632e-01 5.98690510e-01 7.64313117e-02
3.52745950e-01 4.31178898e-01 3.91596228e-01 4.11306888e-01
2.77777016e-01 -4.91130836e-02 -1.08439447e-02 4.76821065e-01
-1.30634487e-01 -6.55217022e-02 -1.59244287e+00 5.35364032e-01
-2.09335899e+00 -9.56694245e-01 -2.12812662e-01 1.91646945e+00
1.26638293e+00 1.84861273e-01 3.75100188e-02 5.72131336e-01
1.88277781e-01 -2.67211854e-01 -3.44488233e-01 -7.35757053e-01
-5.87471090e-02 6.78189754e-01 1.00897454e-01 5.81155002e-01
-1.07615221e+00 1.03163791e+00 5.35772371e+00 5.46339571e-01
-9.62262750e-01 -2.52040416e-01 4.03371722e-01 2.47121707e-01
-2.32963249e-01 2.43763566e-01 -5.15094221e-01 9.68629718e-02
1.05125952e+00 -1.38319656e-01 5.93945026e-01 1.01576114e+00
-4.61597592e-01 -2.55422294e-01 -1.54286730e+00 9.01389658e-01
8.11063275e-02 -1.34196496e+00 -1.13772284e-02 -4.36599374e-01
4.32914764e-01 -1.75557792e-01 -7.64375776e-02 5.66879392e-01
3.00635815e-01 -1.36448681e+00 5.32786608e-01 7.49467731e-01
1.74248174e-01 -8.69512737e-01 7.22968221e-01 5.86753845e-01
-5.85239589e-01 -5.02609730e-01 -3.33989322e-01 -5.95672131e-01
-4.24877644e-01 4.52674359e-01 -1.05682659e+00 3.94998103e-01
6.04889512e-01 4.78792280e-01 -7.10306346e-01 1.07165730e+00
-3.17730248e-01 8.97210419e-01 -3.87970388e-01 -4.24734622e-01
1.98121712e-01 -8.76788199e-02 3.29778910e-01 9.20209944e-01
-3.71416248e-02 2.58645475e-01 7.56813213e-02 9.24194634e-01
-2.04045102e-02 4.61908616e-02 -6.77532792e-01 -2.67601192e-01
1.55319765e-01 1.03550041e+00 -4.64260995e-01 -1.20345019e-01
-4.22875762e-01 4.25227076e-01 5.01065671e-01 1.30761713e-01
-6.40280843e-01 -4.11322027e-01 1.48407578e-01 9.85969752e-02
1.05898097e-01 -2.16854095e-01 -4.09055114e-01 -1.03791082e+00
2.01016128e-01 -1.31284523e+00 6.81779623e-01 -7.30737507e-01
-1.30106175e+00 4.99281704e-01 2.05798492e-01 -6.02835774e-01
-3.64810497e-01 -1.16270959e+00 -3.85024667e-01 4.85852569e-01
-1.50042665e+00 -1.12875664e+00 -2.73918927e-01 4.98534679e-01
2.44390592e-01 -3.50021660e-01 1.04122758e+00 3.03398371e-01
-2.56082565e-01 2.57710189e-01 -3.49797219e-01 3.15945297e-01
5.61398029e-01 -1.27159286e+00 -3.07400227e-01 5.62642217e-01
3.24200094e-01 5.05764782e-01 6.59228683e-01 -6.84600249e-02
-2.02716398e+00 -7.98415720e-01 1.23867023e+00 -7.33064294e-01
8.21030021e-01 -9.96797904e-02 -9.39598799e-01 8.84259641e-01
3.39121744e-02 -9.54933017e-02 5.90958178e-01 1.33816199e-02
-4.27230150e-01 -2.07956702e-01 -1.18410003e+00 3.53769273e-01
6.05426967e-01 -5.31831920e-01 -1.10932255e+00 4.34032381e-01
3.47546816e-01 -3.39062989e-01 -6.75377607e-01 5.25817513e-01
4.73273665e-01 -8.57607484e-01 7.81630218e-01 -1.00851679e+00
7.29772389e-01 -2.90769517e-01 -3.92555088e-01 -7.18503296e-01
1.56961784e-01 -2.68514693e-01 -4.60640162e-01 1.15005696e+00
5.31477869e-01 -6.54963672e-01 7.16698945e-01 1.01040387e+00
2.54651785e-01 -1.05540407e+00 -7.94309258e-01 -4.61618841e-01
3.23513299e-01 -6.69636130e-01 4.41136926e-01 8.76309037e-01
2.20444664e-01 3.45053226e-01 6.16205186e-02 6.10321723e-02
3.79379213e-01 6.49437785e-01 6.71521127e-01 -1.52053630e+00
-4.09375370e-01 -4.01908815e-01 -6.57901645e-01 -6.35181069e-01
2.59190500e-01 -1.20335996e+00 -2.61917472e-01 -1.37519693e+00
8.29257369e-02 -7.19483376e-01 -1.49826030e-03 8.89658689e-01
2.59335041e-01 1.06952891e-01 1.65999848e-02 -7.56362900e-02
-4.72440422e-01 -1.57821015e-01 9.33008909e-01 -2.15127081e-01
3.86557057e-02 -6.20756336e-02 -5.93182564e-01 8.68621290e-01
1.15050399e+00 -4.28249508e-01 -4.14260447e-01 -4.48114753e-01
7.26452649e-01 -2.04366446e-01 6.06650591e-01 -1.02551198e+00
4.00978297e-01 -2.86391556e-01 -2.12997608e-02 -2.35236183e-01
-1.10789932e-01 -1.07670414e+00 -4.09712009e-02 5.69391131e-01
-4.48891461e-01 -1.41694099e-01 2.89696842e-01 1.10890724e-01
-3.94231617e-01 -7.74987996e-01 6.82563841e-01 -3.62228394e-01
-1.10014260e+00 1.23249382e-01 -1.01976484e-01 2.99827933e-01
1.04025912e+00 3.45685817e-02 -2.45008722e-01 -1.84441566e-01
-4.67714697e-01 3.23401809e-01 2.03015238e-01 3.60408843e-01
5.44592500e-01 -9.54087496e-01 -6.61876261e-01 8.72204527e-02
1.35186225e-01 1.11758195e-01 -3.62733603e-01 7.01409280e-01
-1.02298880e+00 7.29773760e-01 -1.75434038e-01 -3.88727486e-01
-1.34504437e+00 5.54468036e-01 4.33767021e-01 -3.35978270e-01
-3.29159796e-01 7.73347139e-01 -5.42588413e-01 -6.72595620e-01
3.82691681e-01 -4.31409746e-01 -2.76887536e-01 -2.26166159e-01
4.97645319e-01 2.14564502e-01 -5.98728098e-02 -8.03676620e-02
-4.22549725e-01 2.75378913e-01 -6.99837133e-02 1.72195628e-01
1.61362147e+00 5.22872090e-01 -6.15086913e-01 5.61353564e-01
8.91984344e-01 -3.53461742e-01 -4.92972553e-01 -3.08063924e-01
5.76881230e-01 -3.67969722e-01 -4.63327095e-02 -1.09664977e+00
-6.95384800e-01 6.94539011e-01 3.94155055e-01 5.21786869e-01
1.13888824e+00 1.72650740e-01 4.95231777e-01 1.13904989e+00
6.02288663e-01 -9.25247073e-01 -3.16518918e-02 9.92132843e-01
7.82547355e-01 -1.42848039e+00 -3.43773700e-02 -3.94398123e-01
-3.92667681e-01 1.46988225e+00 6.25292122e-01 -1.05229050e-01
4.84078705e-01 5.12887657e-01 -6.22566566e-02 -2.71904707e-01
-9.16678846e-01 -1.41218781e-01 1.37796685e-01 4.09103602e-01
5.23890138e-01 -8.76276419e-02 -3.70467097e-01 4.29782629e-01
-2.74857670e-01 6.78122640e-01 3.32623303e-01 1.26560569e+00
-5.50993562e-01 -1.53567648e+00 -2.10780859e-01 3.58322769e-01
-4.49102759e-01 -1.71197161e-01 -7.77528286e-01 8.11773956e-01
2.45095044e-01 8.80972028e-01 -2.03287244e-01 -1.33365333e-01
2.39747331e-01 5.63094139e-01 8.41609597e-01 -5.24850130e-01
-4.64112937e-01 -1.02382886e+00 2.90740043e-01 -2.74526596e-01
-5.90825975e-01 -4.17438239e-01 -1.47464490e+00 -2.95024097e-01
-1.56944320e-01 2.73977667e-01 5.14411330e-01 9.79393721e-01
5.26150130e-02 6.70067519e-02 1.80772990e-01 -1.26843169e-01
-7.00129867e-01 -7.42713392e-01 -2.37648964e-01 3.58909965e-01
-5.03786877e-02 -3.71733516e-01 -1.60392344e-01 3.07210624e-01] | [8.935501098632812, 7.10919713973999] |
d8a8df04-b8fc-44d1-a518-4622d72c0f6c | robust-small-object-detection-on-the-water | null | null | http://openaccess.thecvf.com//content/ICCV2021/html/Cheng_Robust_Small_Object_Detection_on_the_Water_Surface_Through_Fusion_ICCV_2021_paper.html | http://openaccess.thecvf.com//content/ICCV2021/papers/Cheng_Robust_Small_Object_Detection_on_the_Water_Surface_Through_Fusion_ICCV_2021_paper.pdf | Robust Small Object Detection on the Water Surface Through Fusion of Camera and Millimeter Wave Radar | In recent years, unmanned surface vehicles (USVs) have been experiencing growth in various applications. With the expansion of USVs' application scenes from the typical marine areas to inland waters, new challenges arise for the object detection task, which is an essential part of the perception system of USVs. In our work, we focus on a relatively unexplored task for USVs in inland waters: small object detection on water surfaces, which is of vital importance for safe autonomous navigation and USVs' certain missions such as floating waste cleaning. Considering the limitations of vision-based object detection, we propose a novel vision-radar fusion based method for robust small object detection on water surfaces. By using a novel representation format of millimeter wave radar point clouds and applying a deep-level multi-scale fusion of RGB images and radar data, the proposed method can efficiently utilize the characteristics of radar data and improve the accuracy and robustness for small object detection on water surfaces. We test the method on the real-world floating bottle dataset that we collected and released. The result shows that, our method improves the average detection accuracy significantly compared to the vision-based methods and achieves state-of-the-art performance. Besides, the proposed method performs robustly when single sensor degrades. | ['Yimin Liu', 'Hu Xu', 'Yuwei Cheng'] | 2021-01-01 | null | null | null | iccv-2021-1 | ['small-object-detection'] | ['computer-vision'] | [ 1.94966137e-01 -4.10488665e-01 5.87796092e-01 -2.37795830e-01
-5.02955675e-01 -5.40063858e-01 4.16458398e-01 -1.16634354e-01
-6.06413007e-01 2.85203487e-01 -2.11045474e-01 -1.67639837e-01
-4.33695093e-02 -1.07294190e+00 -5.48029900e-01 -8.61694217e-01
-1.43385828e-01 -1.04685500e-02 7.25186586e-01 -5.48182011e-01
2.50681221e-01 5.52452981e-01 -1.84814405e+00 -3.55775431e-02
1.00909257e+00 1.37492204e+00 7.21871793e-01 4.35340255e-01
-1.46931082e-01 -1.78712029e-02 -2.19317779e-01 1.54455453e-01
6.85949504e-01 1.11023635e-01 1.01490960e-01 6.17716238e-02
6.45856202e-01 -5.19173861e-01 -2.93774158e-01 1.53329992e+00
4.98728842e-01 6.33276403e-02 5.80487430e-01 -9.12512720e-01
-4.13985968e-01 -7.79167414e-02 -7.11962938e-01 3.12559098e-01
1.35052994e-01 2.83542395e-01 6.47166133e-01 -8.37761104e-01
2.44792148e-01 1.37216794e+00 5.98721802e-01 3.09518993e-01
-5.41189373e-01 -1.06239092e+00 1.65340498e-01 3.32278192e-01
-1.32416987e+00 -5.64018250e-01 1.75946429e-01 -4.77502227e-01
6.83875561e-01 1.09760985e-01 6.26470804e-01 3.10997933e-01
3.61682057e-01 5.12033045e-01 1.07767379e+00 -1.06602721e-01
2.60013789e-01 -1.60090074e-01 2.03974679e-01 8.10972452e-01
1.22042501e+00 2.16096058e-01 -1.24889091e-01 -8.05385038e-02
5.45327008e-01 5.10843217e-01 -6.30569875e-01 -3.50671560e-01
-9.49080527e-01 7.72052407e-01 6.40407681e-01 4.60983515e-02
-3.29759777e-01 1.22546457e-01 -1.41516924e-01 2.16797054e-01
9.92070511e-02 -1.83755811e-02 -4.03894216e-01 2.47005492e-01
-3.87929112e-01 3.83828133e-01 6.76869750e-01 1.15696859e+00
8.24329555e-01 1.23330101e-01 1.83933958e-01 4.55883920e-01
1.00126529e+00 1.63456929e+00 3.38987887e-01 -3.74733359e-01
5.80107450e-01 8.67246926e-01 6.01892173e-01 -1.03616667e+00
-5.75187504e-01 -2.27858216e-01 -5.38895786e-01 6.42480016e-01
4.53543402e-02 -8.16221237e-02 -1.17141736e+00 1.09668362e+00
4.39383596e-01 9.89434943e-02 7.97650993e-01 1.15739238e+00
1.26859486e+00 8.04974377e-01 -1.80116296e-01 -2.33436152e-01
1.58366418e+00 -4.36930001e-01 -7.34431505e-01 -5.80294847e-01
2.42143512e-01 -6.42619073e-01 5.65189123e-01 4.49877560e-01
-2.91619569e-01 -3.69268805e-01 -1.46120059e+00 4.51652318e-01
-3.40350747e-01 1.06141791e-01 6.38112009e-01 5.57134986e-01
-5.19732058e-01 7.03049302e-02 -1.02727723e+00 -6.76152349e-01
3.82634252e-01 2.37989068e-01 -4.32991236e-01 -5.20704389e-01
-8.87301445e-01 9.74510312e-01 2.44368389e-02 4.98447180e-01
-9.21599627e-01 -1.33579627e-01 -9.25216198e-01 -1.14542872e-01
4.22702581e-01 -9.40630734e-02 1.07398152e+00 -4.20082420e-01
-1.00854719e+00 3.37657541e-01 -4.75716591e-02 -3.96597892e-01
3.22893620e-01 -4.83208120e-01 -4.42003906e-01 5.70364036e-02
2.17949655e-02 1.40616400e-02 6.74421966e-01 -1.41889048e+00
-1.13798940e+00 -8.16224515e-01 6.97137564e-02 3.22282851e-01
-9.40504223e-02 -2.93171912e-01 -2.31633380e-01 -8.60315934e-02
7.02994466e-01 -9.86747324e-01 -1.70066178e-01 4.37123626e-01
1.55339897e-01 -1.40913606e-01 1.11520743e+00 -1.43304914e-01
5.19345939e-01 -2.24314809e+00 -1.50615752e-01 -7.59288967e-02
-8.89343396e-02 4.31040019e-01 -1.53972134e-01 3.64183187e-01
6.94818854e-01 -4.02962714e-01 -3.86394769e-01 -9.99194309e-02
-1.85049266e-01 2.99018234e-01 -3.56642365e-01 8.23671579e-01
-1.84829250e-01 4.66884583e-01 -9.02122438e-01 -2.74421155e-01
2.40294144e-01 2.07539663e-01 -9.22590122e-02 4.09264594e-01
-2.76598707e-03 2.24415287e-01 -7.84493804e-01 1.30045450e+00
1.38131416e+00 3.17851037e-01 -1.89352408e-01 -2.89814979e-01
-6.36398852e-01 -5.37015736e-01 -1.43742394e+00 1.34506226e+00
-3.08588177e-01 5.16881108e-01 6.00370705e-01 -6.40092313e-01
1.25304389e+00 -1.47396982e-01 -2.24947538e-02 -7.15477586e-01
8.29168409e-02 4.08037186e-01 4.50741872e-02 -1.01207078e+00
3.94930363e-01 -3.24448347e-01 3.55545431e-02 -2.18466684e-01
-3.91144872e-01 -4.39597726e-01 -1.84157982e-01 -1.27455175e-01
9.74436522e-01 1.43580884e-01 2.98399210e-01 -3.93644780e-01
4.46806312e-01 2.06888333e-01 7.21447170e-01 5.97019732e-01
-2.26748869e-01 3.51511151e-01 -3.96095723e-01 -4.80963856e-01
-2.14513123e-01 -1.27828538e+00 -4.41401780e-01 6.67631924e-01
9.54662144e-01 3.21836054e-01 -2.65053481e-01 -3.42700899e-01
5.19331992e-01 2.27513075e-01 -5.39729953e-01 -1.12530448e-01
-2.58231789e-01 -9.41506565e-01 2.98051298e-01 3.27598780e-01
9.39312100e-01 -8.71423185e-01 -1.06574237e+00 1.66567490e-01
3.90036516e-02 -1.30611539e+00 2.50755072e-01 -6.57926202e-02
-9.65533257e-01 -1.47858894e+00 -4.89444613e-01 -7.54735589e-01
5.95418692e-01 1.16908121e+00 4.23806220e-01 2.29533110e-02
-3.76239240e-01 3.43781590e-01 -7.83300936e-01 -1.08423853e+00
1.58808723e-01 -6.57930017e-01 4.36591417e-01 2.16182411e-01
5.95920861e-01 -1.58577979e-01 -8.64935517e-01 3.54964703e-01
-8.30062568e-01 -4.58517015e-01 9.29392457e-01 2.96321273e-01
3.00110370e-01 -2.19546601e-01 3.05801511e-01 -2.61817575e-01
6.48036748e-02 -5.42572856e-01 -1.29499054e+00 6.25686347e-03
-3.06844771e-01 -1.17149740e-01 2.53331721e-01 -9.56868604e-02
-6.78636730e-01 2.88450986e-01 1.95059866e-01 -1.86884657e-01
-3.22591066e-02 3.14330310e-01 -1.17922582e-01 -6.23354912e-01
5.15621364e-01 5.44405818e-01 1.23325013e-01 -5.15254676e-01
-1.35301128e-01 1.10827947e+00 3.82647067e-01 1.04167536e-01
1.07905245e+00 9.59698617e-01 2.51064539e-01 -1.37666249e+00
-4.63062435e-01 -6.98079705e-01 -2.38255590e-01 -2.20407024e-01
8.29313934e-01 -1.33091605e+00 -8.97410810e-01 6.70239687e-01
-1.18593228e+00 1.22199103e-01 3.74802053e-01 7.84183979e-01
6.84569106e-02 5.89979827e-01 7.53120333e-03 -1.13357723e+00
-6.19208753e-01 -1.26385760e+00 1.13943303e+00 5.39563894e-01
6.93198621e-01 -4.02772814e-01 4.70545627e-02 1.19555578e-01
5.27940571e-01 3.00699681e-01 6.76038787e-02 -3.70372355e-01
-1.11162806e+00 -3.10713291e-01 -5.09622812e-01 1.78813800e-01
4.87067670e-01 -2.22338110e-01 -8.38675439e-01 -5.36955476e-01
2.52819359e-02 2.01070588e-02 1.21293294e+00 2.59957701e-01
1.57181561e-01 4.41154018e-02 -6.30467832e-01 9.07996416e-01
1.77781796e+00 5.70982039e-01 4.86461639e-01 6.00886226e-01
5.39884090e-01 4.51940268e-01 1.23190868e+00 6.99367762e-01
5.04628122e-01 2.24780068e-01 1.18556464e+00 1.36544079e-01
3.02404612e-01 3.64597887e-01 3.67301971e-01 3.60395014e-01
-3.98403466e-01 -2.71968469e-02 -7.24424005e-01 5.24306953e-01
-1.89457345e+00 -6.22280002e-01 -3.85401040e-01 2.11714005e+00
-5.01320057e-04 -1.21658131e-01 -7.27696180e-01 -2.62239575e-01
6.02334559e-01 6.12033829e-02 -6.53728962e-01 3.51577044e-01
-2.98882499e-02 -4.03994992e-02 1.07368672e+00 3.10400277e-01
-1.22612476e+00 8.23229849e-01 5.39969635e+00 1.64026394e-01
-1.09452176e+00 -1.01361461e-02 -4.52787489e-01 3.27446461e-01
-9.04575828e-03 -2.47984797e-01 -1.20603466e+00 2.76345581e-01
2.11965367e-01 2.39726961e-01 4.70124697e-03 9.32738543e-01
2.79980838e-01 -4.14931357e-01 -5.78237355e-01 1.16180837e+00
9.84751508e-02 -9.89843607e-01 1.92668423e-01 -9.38161556e-03
4.49871123e-01 4.01095569e-01 -2.83393651e-01 9.74721983e-02
1.55950874e-01 -6.20731711e-01 6.81522369e-01 5.30860662e-01
4.91578072e-01 -5.67438185e-01 1.20268798e+00 3.12578022e-01
-1.52478123e+00 -2.95327276e-01 -8.93330634e-01 -4.34402436e-01
2.92792637e-02 3.87948006e-01 -5.59120715e-01 4.03046072e-01
1.08321595e+00 6.58194005e-01 -2.44959027e-01 1.34065175e+00
-2.00384974e-01 1.37349263e-01 -6.03861809e-01 -5.45085728e-01
4.30849701e-01 -5.22783816e-01 6.12193406e-01 1.05411673e+00
8.38791907e-01 7.69160748e-01 2.69006789e-01 4.02203679e-01
3.04289162e-01 -4.99270335e-02 -9.84682202e-01 2.09694415e-01
5.24884880e-01 1.42986751e+00 -4.29658741e-01 -2.48223081e-01
-6.48370981e-01 5.48211217e-01 -3.30203533e-01 2.61455238e-01
-5.85017264e-01 -7.85074055e-01 1.07407379e+00 -8.27209651e-03
6.14331543e-01 -6.28372014e-01 6.92742616e-02 -1.07744026e+00
-6.86085448e-02 -3.42881024e-01 5.63062429e-02 -7.60648906e-01
-8.18126857e-01 5.77921331e-01 -3.00393492e-01 -1.81393671e+00
5.59073389e-01 -9.51975107e-01 -5.62718451e-01 5.83378792e-01
-2.05424356e+00 -1.20832288e+00 -1.16852963e+00 3.85705203e-01
5.17410755e-01 -1.76288679e-01 5.41543663e-01 7.56673217e-02
-1.22510664e-01 2.79030297e-02 3.06247175e-01 -3.21052931e-02
6.27984881e-01 -7.97833860e-01 2.16688320e-01 1.11161721e+00
-3.67840469e-01 4.60902989e-01 8.82188678e-01 -8.19933355e-01
-2.35873199e+00 -1.19121885e+00 -2.96090040e-02 -6.06995113e-02
5.51158488e-01 -2.57813424e-01 -6.26087964e-01 6.30160809e-01
-5.92418164e-02 3.05069745e-01 3.16705614e-01 -6.48418963e-01
3.46968845e-02 -5.51969588e-01 -1.33594370e+00 1.72052070e-01
9.32535350e-01 2.93019086e-01 -9.72758830e-01 3.12908530e-01
7.67581224e-01 -5.28595924e-01 -3.07978451e-01 8.50658000e-01
8.20666611e-01 -7.87928343e-01 8.32478344e-01 -7.45270699e-02
-1.46251231e-01 -8.54941070e-01 -7.72891045e-01 -1.21665883e+00
-3.26485097e-01 -1.98207989e-01 2.89321154e-01 7.13896573e-01
-4.09101211e-02 -9.54333603e-01 5.30670404e-01 -4.47609648e-03
-5.32990694e-01 -1.14490956e-01 -1.12652969e+00 -7.71868527e-01
-6.25539720e-01 -2.04883620e-01 4.38124895e-01 4.20525700e-01
-2.04737276e-01 1.12139946e-02 -2.05576256e-01 1.16280031e+00
1.10450971e+00 5.11870503e-01 1.03505325e+00 -1.62070596e+00
1.88698858e-01 1.59145594e-01 -8.35251868e-01 -1.12357986e+00
-4.41317081e-01 -4.04303014e-01 6.37671411e-01 -2.05639625e+00
-5.99798448e-02 -3.08173150e-01 -7.30847344e-02 3.88746858e-01
-8.60677287e-02 5.09800911e-01 1.98661834e-01 2.03535467e-01
-5.45130730e-01 6.70743346e-01 1.13148844e+00 -3.79425108e-01
-1.36045635e-01 1.25384942e-01 -5.28155506e-01 8.45383227e-01
3.57554227e-01 -3.35551292e-01 1.10231191e-02 -6.35238886e-01
-8.30616057e-02 -3.02994847e-02 1.44315481e-01 -1.37174106e+00
4.19329822e-01 -4.54008460e-01 1.57351896e-01 -7.34583974e-01
4.94356692e-01 -1.17415178e+00 -2.11227372e-01 1.17934346e+00
8.54409456e-01 -2.90385246e-01 3.30769330e-01 8.83869112e-01
-1.73950657e-01 -1.99436650e-01 9.91151810e-01 -3.06040555e-01
-1.45505559e+00 3.43309760e-01 -4.37287360e-01 -3.93371910e-01
1.24221706e+00 -2.70073444e-01 -5.61309099e-01 -1.49387112e-02
-2.40390062e-01 5.82628071e-01 3.52759004e-01 4.91916090e-01
1.23127961e+00 -9.61840093e-01 -8.35803449e-01 4.35454220e-01
6.12159491e-01 3.93545479e-01 1.82931721e-01 5.42171717e-01
-9.13968086e-01 1.38878793e-01 -1.94832653e-01 -8.19414735e-01
-1.35116386e+00 1.78295359e-01 2.20390141e-01 6.47628367e-01
-6.39561057e-01 7.86593854e-01 4.88920152e-01 -2.01565996e-01
-4.69845422e-02 -8.91041636e-01 -3.87425512e-01 -1.88782495e-02
8.95358503e-01 4.00663167e-01 3.23463939e-02 -5.87730110e-01
-6.94880486e-01 1.21513546e+00 2.71186680e-01 3.08650792e-01
1.46535385e+00 -2.48344705e-01 -3.29492763e-02 2.53331482e-01
6.81744993e-01 1.14289403e-01 -1.32458830e+00 -2.36608624e-01
-4.54455584e-01 -5.98503947e-01 1.79095522e-01 -4.21703130e-01
-9.03547287e-01 7.94949114e-01 1.19003356e+00 -4.71564606e-02
8.16488564e-01 -2.42225640e-02 7.83435106e-01 9.70455885e-01
8.94315183e-01 -6.58356428e-01 -2.03881145e-01 7.07918525e-01
8.74942362e-01 -1.71364486e+00 3.46146733e-01 -3.69961023e-01
-3.44530165e-01 1.18465781e+00 6.27161026e-01 -4.25865889e-01
5.20200133e-01 2.86087215e-01 5.30531108e-01 -2.87025541e-01
-2.44394183e-01 -6.24412894e-01 -9.82264429e-02 7.04135060e-01
-2.96755672e-01 1.14167266e-01 -2.02490538e-01 6.38689637e-01
1.28641143e-01 -1.74462408e-01 7.10999012e-01 1.25390530e+00
-1.62392557e+00 -3.69899362e-01 -8.68123114e-01 4.38074619e-01
-9.12499055e-02 -9.04219374e-02 1.89625785e-01 8.57497931e-01
2.01653615e-01 1.10872090e+00 1.98847160e-01 -4.27069455e-01
8.23534012e-01 -4.86220419e-01 4.22355831e-01 -5.60694933e-01
1.80039704e-01 -1.29881203e-01 -1.50603741e-01 -4.35297161e-01
-6.03814960e-01 -6.32798672e-01 -1.55271804e+00 3.73471558e-01
-5.89273989e-01 2.30430111e-01 1.06190765e+00 9.83018458e-01
1.37423828e-01 1.93157390e-01 6.87036276e-01 -1.18641639e+00
-8.07772160e-01 -1.20253241e+00 -9.69415605e-01 1.64370909e-02
6.58435106e-01 -1.16766191e+00 -6.39532208e-01 -2.55695432e-01] | [7.973013877868652, -1.6344987154006958] |
342f1a1a-8a1a-47aa-a63d-e7278e3812e7 | cnn-lstm-models-for-multi-speaker-source | 1912.09254 | null | https://arxiv.org/abs/1912.09254v1 | https://arxiv.org/pdf/1912.09254v1.pdf | CNN-LSTM models for Multi-Speaker Source Separation using Bayesian Hyper Parameter Optimization | In recent years there have been many deep learning approaches towards the multi-speaker source separation problem. Most use Long Short-Term Memory - Recurrent Neural Networks (LSTM-RNN) or Convolutional Neural Networks (CNN) to model the sequential behavior of speech. In this paper we propose a novel network for source separation using an encoder-decoder CNN and LSTM in parallel. Hyper parameters have to be chosen for both parts of the network and they are potentially mutually dependent. Since hyper parameter grid search has a high computational burden, random search is often preferred. However, when sampling a new point in the hyper parameter space, it can potentially be very close to a previously evaluated point and thus give little additional information. Furthermore, random sampling is as likely to sample in a promising area as in an hyper space area dominated with poor performing models. Therefore, we use a Bayesian hyper parameter optimization technique and find that the parallel CNN-LSTM outperforms the LSTM-only and CNN-only model. | ['Hugo Van hamme', 'Jeroen Zegers'] | 2019-12-19 | null | null | null | null | ['multi-speaker-source-separation'] | ['speech'] | [-6.84330473e-04 5.34250624e-02 -1.76740527e-01 -6.85374737e-02
-1.35829687e+00 -3.00238490e-01 3.06071639e-01 -1.15480654e-01
-4.74992841e-01 7.92484045e-01 3.53488922e-01 -1.80594027e-01
-1.85197651e-01 -5.19056439e-01 -5.95839739e-01 -1.09965956e+00
-1.12548940e-01 6.62793636e-01 3.07453603e-01 -2.96402648e-02
-3.49105224e-02 3.42621803e-01 -1.36898243e+00 2.01725304e-01
5.22246182e-01 8.39046299e-01 4.04565722e-01 8.39145660e-01
-1.35404333e-01 6.83642268e-01 -8.18669140e-01 -4.68152799e-02
1.41349375e-01 -5.01779079e-01 -8.49655747e-01 -5.01169264e-01
5.43049201e-02 -2.94046625e-02 -1.17163181e-01 1.25314009e+00
9.28893447e-01 3.47476840e-01 3.74809951e-01 -8.59891295e-01
1.37989670e-01 1.11955011e+00 -4.18861985e-01 3.52653742e-01
9.17625353e-02 -1.23751499e-01 7.90039241e-01 -7.55854726e-01
1.57557905e-01 1.27168798e+00 7.83481419e-01 3.57840896e-01
-1.06331301e+00 -7.26221859e-01 6.18670732e-02 1.61966890e-01
-1.54680634e+00 -6.88470900e-01 8.83093715e-01 -1.79210365e-01
1.08335674e+00 -5.99734113e-02 5.07128894e-01 1.29236507e+00
2.46964525e-02 8.53352189e-01 6.64763570e-01 -5.14629543e-01
3.09564620e-01 2.80770600e-01 -1.21389218e-01 4.83108491e-01
-1.75561860e-01 1.35873199e-01 -7.25677252e-01 -4.23055112e-01
6.79019749e-01 -3.33834827e-01 -2.12498128e-01 -7.26066902e-02
-1.05387115e+00 9.05803502e-01 4.27294195e-01 5.35617292e-01
-4.46705669e-01 2.77712226e-01 5.14599621e-01 2.31238663e-01
3.72886807e-01 4.44472194e-01 -5.26816428e-01 -4.94558036e-01
-1.36278045e+00 2.80110389e-01 8.80969465e-01 5.55450439e-01
6.08752310e-01 3.35328281e-01 2.21725963e-02 1.29256856e+00
4.15706068e-01 1.84533879e-01 8.23303223e-01 -7.23058462e-01
5.47890425e-01 1.65649831e-01 -2.80713111e-01 -6.56742156e-01
-3.87346953e-01 -6.36556506e-01 -9.14228678e-01 9.22383964e-02
4.20962781e-01 -5.13783634e-01 -8.46654296e-01 1.64606428e+00
2.27088779e-01 4.00505155e-01 1.26114279e-01 7.69815803e-01
6.00168467e-01 9.14589405e-01 -1.71781346e-01 -2.11212113e-01
1.15923417e+00 -1.07185733e+00 -6.50605321e-01 -4.06451315e-01
4.49142754e-01 -7.41030276e-01 6.03501678e-01 4.52785939e-01
-1.41457975e+00 -4.69521016e-01 -1.11051774e+00 2.29788437e-01
-3.92466843e-01 6.67386726e-02 6.53982460e-02 6.19473159e-01
-1.22768748e+00 6.84168339e-01 -7.48842478e-01 -5.08213677e-02
2.19989479e-01 6.78164661e-01 1.59583211e-01 2.43828908e-01
-1.48231781e+00 8.19432378e-01 5.84445894e-01 3.59630704e-01
-1.04491711e+00 -3.97978872e-01 -5.63970864e-01 2.97031134e-01
4.11566615e-01 -4.44314420e-01 1.44850683e+00 -9.75233138e-01
-1.98357153e+00 5.08916378e-02 -2.78501183e-01 -7.31465459e-01
3.39038163e-01 -1.74225152e-01 -3.28428984e-01 -2.65377155e-03
-3.45734596e-01 5.58592737e-01 1.10254538e+00 -1.05811048e+00
-4.65872496e-01 -1.49208739e-01 -2.62547612e-01 3.67686570e-01
-3.12332302e-01 3.22373003e-01 -3.68085712e-01 -7.22035289e-01
1.98695332e-01 -1.03802431e+00 -3.37819904e-01 -5.98643422e-01
-6.28556907e-01 -1.55805051e-01 6.38049126e-01 -6.07073843e-01
1.27812970e+00 -2.10851192e+00 1.68383941e-01 3.98651093e-01
-1.52237257e-02 5.25724113e-01 1.50625845e-02 2.12974429e-01
-1.49283260e-01 2.97804862e-01 -1.70735061e-01 -4.63709801e-01
-2.84108669e-01 3.58407609e-02 -2.07938865e-01 2.92643785e-01
-1.68270588e-01 5.79387844e-01 -7.11897254e-01 -5.07216871e-01
4.26811427e-02 8.08604062e-01 -2.11047724e-01 8.48553106e-02
-1.25732705e-01 2.44686827e-01 -1.97688371e-01 3.92396688e-01
5.22913158e-01 -7.31712505e-02 3.57864276e-02 4.42272983e-02
-4.33184132e-02 6.78079844e-01 -1.34180307e+00 1.52448797e+00
-7.07378209e-01 1.01345944e+00 1.91531107e-01 -8.30580056e-01
8.70793760e-01 8.91729116e-01 1.72963291e-01 -3.60439003e-01
2.01218143e-01 4.55023944e-01 1.81652591e-01 -3.20904374e-01
4.20323789e-01 -2.01350227e-01 4.87193912e-01 4.79054958e-01
9.41363126e-02 1.56979337e-01 -1.73608512e-01 -3.03738117e-01
8.99940014e-01 -1.77171782e-01 5.78297675e-02 -1.43985182e-01
4.70618933e-01 -4.74410415e-01 6.62338734e-01 7.84446716e-01
-1.24632381e-01 7.54892170e-01 5.64617515e-01 -7.15698749e-02
-8.39725316e-01 -6.06047273e-01 -1.99262902e-01 1.01856565e+00
-3.03814262e-01 -2.01557443e-01 -8.26024294e-01 -5.11840880e-01
-4.87942725e-01 6.74559355e-01 -4.01379555e-01 -1.13763198e-01
-8.86232316e-01 -8.18267226e-01 8.71060789e-01 4.47981238e-01
4.58931923e-01 -1.15683949e+00 -6.91862583e-01 5.60138345e-01
-1.31094873e-01 -9.19999719e-01 -3.31343740e-01 7.39029229e-01
-9.39121425e-01 -6.27458453e-01 -1.10703349e+00 -7.37399280e-01
1.42838180e-01 4.82730307e-02 8.73418212e-01 -2.57275969e-01
2.08595276e-01 -3.17056090e-01 -2.13596687e-01 -2.49611765e-01
-4.97892290e-01 4.87813830e-01 -7.67439837e-03 7.53713921e-02
3.35365206e-01 -5.31112194e-01 -4.29770052e-01 2.37261564e-01
-4.92985904e-01 -1.97950631e-01 4.66298670e-01 8.25157225e-01
1.65934607e-01 4.14249957e-01 7.08248615e-01 -5.77322960e-01
7.28286266e-01 -5.73157072e-01 -5.76743305e-01 1.59387931e-01
-5.16643047e-01 4.23750907e-01 5.06064594e-01 -7.30211794e-01
-1.12226880e+00 2.69137267e-02 -2.58019507e-01 -7.25873232e-01
-1.94248632e-01 7.28485465e-01 -1.29109636e-01 1.07768230e-01
5.26724160e-01 7.31639341e-02 -1.99556619e-01 -7.44782805e-01
-1.02339193e-01 7.42455602e-01 9.39524099e-02 -2.54971206e-01
3.18285286e-01 6.85242936e-02 -3.43940884e-01 -1.07730448e+00
-5.88499248e-01 -3.34260345e-01 -5.45627654e-01 -2.68983155e-01
7.93193579e-01 -8.40709507e-01 -4.65184361e-01 5.28881073e-01
-1.37370145e+00 -4.23705131e-01 -1.18238308e-01 7.81552196e-01
-2.38013789e-01 -5.12246042e-02 -4.36746299e-01 -1.10434628e+00
-1.45570129e-01 -1.52181041e+00 9.90703940e-01 3.06518108e-01
-2.77352929e-01 -1.31991363e+00 2.93756366e-01 -3.63812111e-02
4.95603651e-01 -2.35654145e-01 7.83927023e-01 -1.00990438e+00
-2.42454320e-01 -2.41439417e-01 1.47171527e-01 2.70572692e-01
-1.25503048e-01 4.84619141e-02 -1.41387486e+00 -2.76231766e-01
2.60187536e-01 -7.00436831e-02 1.08173120e+00 8.65663886e-01
1.13920760e+00 -4.45422679e-01 -3.88982683e-01 6.72176540e-01
1.08358884e+00 3.79402936e-01 4.82208401e-01 2.86037743e-01
6.63626730e-01 6.38165116e-01 2.16082595e-02 2.09801719e-01
1.28730565e-01 7.76134968e-01 1.77973256e-01 -1.12847067e-01
-8.87985229e-02 -1.28419682e-01 7.05112696e-01 8.81348968e-01
1.80322036e-01 -4.67538267e-01 -1.09506488e+00 5.92913091e-01
-1.75705838e+00 -8.13962877e-01 1.10604994e-01 2.46646094e+00
8.26585233e-01 3.33009034e-01 2.98048139e-01 3.25512707e-01
9.07078862e-01 2.96680808e-01 -4.53064829e-01 -3.84085208e-01
-1.50509804e-01 -3.11912522e-02 5.06331325e-01 5.89834571e-01
-8.84813368e-01 7.60653675e-01 6.52912235e+00 1.02571559e+00
-1.20715952e+00 1.45760849e-01 5.37443340e-01 -5.30613244e-01
-1.14023551e-01 -3.60622704e-02 -1.13690448e+00 5.26904881e-01
1.57572711e+00 8.43325630e-02 2.94156671e-01 6.03146076e-01
1.48985758e-01 -3.07282567e-01 -9.09730136e-01 1.17167282e+00
1.87316053e-02 -1.04948282e+00 -2.20277250e-01 1.91839486e-01
4.38413024e-01 4.49441046e-01 1.39482230e-01 1.24250658e-01
3.54741633e-01 -1.14013338e+00 6.14487588e-01 3.26104522e-01
3.54445338e-01 -1.07654965e+00 8.45918834e-01 5.84778011e-01
-1.08159912e+00 -2.15152889e-01 -4.69493032e-01 4.87003773e-01
2.95782268e-01 7.42320538e-01 -1.04548216e+00 2.89492458e-01
8.11764598e-01 2.75396883e-01 -4.01944011e-01 1.26144421e+00
-2.97964197e-02 8.57264936e-01 -5.38474202e-01 -8.21499601e-02
3.57105225e-01 -1.11184716e-02 8.48772347e-01 1.26039171e+00
4.24081951e-01 -3.28631133e-01 4.75750677e-02 5.92691123e-01
-6.13567866e-02 -1.41038507e-01 -4.78168130e-01 5.36980331e-02
4.59813148e-01 9.67002809e-01 -7.24706113e-01 -3.53348315e-01
-2.02303126e-01 6.20984316e-01 1.95643023e-01 6.31052315e-01
-6.73511326e-01 -3.63232851e-01 4.73696172e-01 -9.79375392e-02
4.12781477e-01 -1.11252040e-01 -2.51237750e-01 -9.02203381e-01
-2.67797887e-01 -8.84405196e-01 3.25931638e-01 -5.73078394e-01
-8.94837022e-01 1.02797091e+00 7.20936731e-02 -1.19294322e+00
-7.26211488e-01 -2.47221991e-01 -6.24942243e-01 9.87795711e-01
-1.52150762e+00 -5.92013121e-01 2.70326167e-01 4.84736472e-01
9.41474617e-01 -4.40242112e-01 7.25581050e-01 2.12294504e-01
-7.56955564e-01 5.73578358e-01 2.90960222e-01 7.51869008e-02
5.24058104e-01 -1.11464858e+00 1.61299974e-01 7.36428976e-01
2.73721546e-01 5.26850045e-01 5.77516615e-01 -5.38442075e-01
-8.42040241e-01 -8.38052392e-01 1.00346029e+00 4.46884930e-02
3.80350798e-01 -4.92304534e-01 -1.21005487e+00 3.46744269e-01
4.62409377e-01 -3.10661793e-01 5.70213854e-01 3.19838762e-01
-1.99212596e-01 -1.07986582e-02 -8.15787494e-01 5.76524734e-01
3.51735026e-01 -7.35994101e-01 -3.21030885e-01 2.56935537e-01
4.86371547e-01 -3.45289826e-01 -5.57609558e-01 1.86644539e-01
3.93316448e-01 -1.04298937e+00 9.44195092e-01 -1.77275851e-01
-4.12860587e-02 4.60382514e-02 4.21607159e-02 -1.60884190e+00
-6.42889887e-02 -8.73414218e-01 -3.45639646e-01 1.24152815e+00
9.12969351e-01 -5.18427789e-01 7.17718363e-01 3.45139652e-01
5.32399900e-02 -7.37042367e-01 -1.23381364e+00 -8.51068258e-01
1.38573155e-01 -6.57964647e-01 7.28038609e-01 4.86401975e-01
-1.19420968e-01 4.66748416e-01 -3.95586103e-01 2.31829837e-01
5.51110208e-01 -2.25869223e-01 2.23055080e-01 -1.24835193e+00
-3.16294849e-01 -7.82367885e-01 1.30294099e-01 -1.20971203e+00
3.57041121e-01 -6.90664351e-01 4.25438523e-01 -1.29192662e+00
7.56468475e-02 -4.14572626e-01 -5.15181720e-01 3.56947094e-01
3.32436264e-02 -1.91708490e-01 -8.67116153e-02 2.27056414e-01
-3.46743971e-01 5.30193686e-01 7.30454683e-01 -4.87441272e-02
-7.16338813e-01 4.77204710e-01 -1.53141811e-01 9.31123972e-01
1.01619005e+00 -7.59288788e-01 -3.79634470e-01 -4.19719398e-01
4.55577284e-01 4.87327307e-01 3.05298418e-01 -1.08522046e+00
6.39171004e-01 1.90273777e-01 3.42092931e-01 -7.05157340e-01
6.04778171e-01 -6.08441412e-01 -2.81589404e-02 2.45042890e-01
-6.23287261e-01 -2.48299558e-02 2.34472200e-01 4.38502431e-01
-4.12434757e-01 -6.97111189e-01 9.11571324e-01 -1.83519706e-01
-6.07729964e-02 2.03880459e-01 -8.03422213e-01 -8.63255337e-02
4.76234943e-01 -2.44984090e-01 3.04035723e-01 -5.90609908e-01
-6.97001696e-01 1.18279837e-01 1.55574372e-02 4.57810104e-01
5.92400908e-01 -1.05320585e+00 -5.05709231e-01 2.98137754e-01
-3.96270484e-01 8.92774388e-02 1.72567800e-01 7.01461256e-01
-1.07088372e-01 4.53393131e-01 3.04480404e-01 -5.93243122e-01
-1.23304772e+00 2.40485117e-01 6.18625045e-01 -2.13516638e-01
-4.03386682e-01 1.17521620e+00 -1.34563699e-01 -3.89088035e-01
6.60258949e-01 -1.80822760e-01 -4.64248538e-01 4.32260424e-01
5.25192440e-01 6.01230204e-01 1.68164521e-01 -6.29004598e-01
-2.98208266e-01 3.69074672e-01 -3.53139564e-02 -6.99799478e-01
1.35646224e+00 -1.82863012e-01 1.08297877e-02 9.32974100e-01
1.44401336e+00 -1.26391709e-01 -1.25485981e+00 -4.29336905e-01
2.22070694e-01 -9.02326554e-02 7.19710708e-01 -4.79767412e-01
-1.07006907e+00 1.19166112e+00 5.88603020e-01 3.62324536e-01
9.83966112e-01 -1.06228858e-01 7.68175066e-01 4.16955382e-01
-9.05149654e-02 -1.42669129e+00 -8.49349145e-03 7.60478795e-01
7.57120192e-01 -1.17004144e+00 -2.67831206e-01 1.24674760e-01
-6.39544129e-01 1.43902588e+00 3.67249906e-01 1.92654565e-01
9.44740474e-01 4.57573503e-01 3.89040262e-02 -9.27880257e-02
-9.43217337e-01 -1.22911237e-01 2.26114929e-01 3.45966995e-01
5.01355052e-01 -2.24023491e-01 5.95674276e-01 1.97327673e-01
-3.03060919e-01 -2.44966209e-01 3.64971966e-01 5.29440403e-01
-4.63467091e-01 -1.07632685e+00 -5.90137839e-01 2.70485997e-01
-5.59868753e-01 -3.31749499e-01 -8.59471485e-02 3.76967460e-01
-1.11089729e-01 1.07360113e+00 9.01888087e-02 -1.84730530e-01
-1.29290774e-01 4.16706204e-01 1.67731062e-01 -4.72317874e-01
-8.99029136e-01 7.85241187e-01 1.36402711e-01 -3.68760198e-01
-2.65688807e-01 -9.35764313e-01 -1.21264374e+00 -3.21967229e-02
-7.58701324e-01 3.38496804e-01 9.64891076e-01 1.01920927e+00
2.23221868e-01 6.92050219e-01 6.60945117e-01 -9.05744910e-01
-5.67673922e-01 -1.07876921e+00 -5.41212380e-01 -5.88145018e-01
8.22885215e-01 -4.08151984e-01 -4.87566143e-01 -3.02822977e-01] | [14.680815696716309, 6.139276504516602] |
1007e2f0-2d4d-496b-847d-56e6c3b05d39 | improving-prosody-for-unseen-texts-in-speech | 2111.07549 | null | https://arxiv.org/abs/2111.07549v1 | https://arxiv.org/pdf/2111.07549v1.pdf | Improving Prosody for Unseen Texts in Speech Synthesis by Utilizing Linguistic Information and Noisy Data | Recent advancements in end-to-end speech synthesis have made it possible to generate highly natural speech. However, training these models typically requires a large amount of high-fidelity speech data, and for unseen texts, the prosody of synthesized speech is relatively unnatural. To address these issues, we propose to combine a fine-tuned BERT-based front-end with a pre-trained FastSpeech2-based acoustic model to improve prosody modeling. The pre-trained BERT is fine-tuned on the polyphone disambiguation task, the joint Chinese word segmentation (CWS) and part-of-speech (POS) tagging task, and the prosody structure prediction (PSP) task in a multi-task learning framework. FastSpeech 2 is pre-trained on large-scale external data that are noisy but easier to obtain. Experimental results show that both the fine-tuned BERT model and the pre-trained FastSpeech 2 can improve prosody, especially for those structurally complex sentences. | ['Caixia Gong', 'Ruixiong Zhang', 'Mengnan He', 'Ming Yan', 'Mengxi Nie', 'Yuqing Zhang', 'Zhu Li'] | 2021-11-15 | null | null | null | null | ['polyphone-disambiguation', 'chinese-word-segmentation'] | ['natural-language-processing', 'natural-language-processing'] | [ 1.47875458e-01 9.50012431e-02 -1.86127990e-01 -4.15318578e-01
-1.47571588e+00 -4.58125085e-01 6.72946796e-02 -2.26094931e-01
-3.17631304e-01 5.66852450e-01 6.24039352e-01 -4.07641023e-01
7.30000734e-01 -5.34972608e-01 -6.08436525e-01 -4.81396973e-01
4.54227686e-01 4.08666611e-01 4.53290284e-01 -2.77090788e-01
-3.13158274e-01 5.83235919e-02 -1.25774169e+00 5.26285052e-01
9.14162159e-01 7.76525140e-01 1.03290844e+00 9.85125661e-01
-3.94629031e-01 4.09748733e-01 -7.73060322e-01 -2.58619368e-01
8.40177312e-02 -6.62158966e-01 -5.20170152e-01 -1.10838845e-01
3.25501598e-02 -1.14746809e-01 9.13330391e-02 1.03024447e+00
7.78782010e-01 3.63872647e-01 3.42567265e-01 -6.40550733e-01
-3.45286995e-01 1.25107718e+00 -1.19732544e-01 2.72773921e-01
1.70139253e-01 1.23103529e-01 1.23513758e+00 -1.02614784e+00
2.91367918e-01 1.56393421e+00 4.24269795e-01 8.49593878e-01
-1.02789688e+00 -6.98024809e-01 1.81614846e-01 -1.36660904e-01
-1.22128820e+00 -7.80626595e-01 1.03716922e+00 -1.77794129e-01
1.18411112e+00 1.78077340e-01 4.18187648e-01 1.31453657e+00
-1.71695620e-01 1.05631828e+00 6.67877197e-01 -6.24612689e-01
2.14299172e-01 -9.03334245e-02 -2.92070478e-01 2.40597799e-01
-6.48399234e-01 3.14805210e-01 -4.59477574e-01 1.54033884e-01
6.16387844e-01 -5.90996742e-01 -2.50245184e-01 3.45205098e-01
-1.04983187e+00 7.67464995e-01 -2.87401956e-02 5.19479334e-01
-4.36890185e-01 -2.88519077e-02 5.64009666e-01 8.60916600e-02
4.14702892e-01 3.89483422e-01 -7.24058092e-01 -4.92659032e-01
-1.12290561e+00 -4.10053395e-02 8.13468277e-01 1.10165215e+00
4.29489970e-01 6.63145781e-01 -2.45663866e-01 1.56571627e+00
4.61509705e-01 6.67758882e-01 7.82667637e-01 -6.36304259e-01
8.30319107e-01 -1.85251653e-01 -2.22681358e-01 -3.16954166e-01
-7.10538924e-02 -5.04769027e-01 -5.16011894e-01 -2.31996730e-01
3.21095437e-01 -5.97572744e-01 -1.06813407e+00 1.91904366e+00
2.44118690e-01 1.29119739e-01 3.28323454e-01 9.19012785e-01
6.56279325e-01 1.41791022e+00 2.72828937e-01 -4.46875513e-01
1.20268083e+00 -1.59241199e+00 -1.00388396e+00 -5.47990322e-01
3.70917261e-01 -1.11474133e+00 1.49386764e+00 3.33694220e-01
-1.48195267e+00 -1.14650786e+00 -9.48205292e-01 -4.35548313e-02
-3.40746567e-02 2.77007431e-01 -1.42364159e-01 6.50240898e-01
-7.62748063e-01 2.73803115e-01 -8.25610459e-01 2.65840948e-01
-4.29842845e-02 6.90722689e-02 4.77488562e-02 1.33727774e-01
-1.44863558e+00 5.93020260e-01 5.62112808e-01 -1.14898674e-01
-9.84684825e-01 -6.15046620e-01 -9.87727404e-01 3.49356592e-01
4.33734864e-01 -3.57953757e-01 1.72032487e+00 -1.17285669e+00
-2.25083470e+00 3.26303840e-01 -1.73988357e-01 -2.58151591e-01
9.27057862e-02 -1.73652634e-01 -6.69375718e-01 1.43620327e-01
-1.53707549e-01 7.18908012e-01 1.03388381e+00 -1.04289186e+00
-7.03157902e-01 7.18565062e-02 -6.55975163e-01 4.40170258e-01
-2.20985726e-01 4.94581878e-01 -5.07800519e-01 -1.06001568e+00
-3.14223081e-01 -7.71462858e-01 -3.68832082e-01 -6.80401742e-01
-3.91678989e-01 -2.52882451e-01 7.27268815e-01 -1.01320088e+00
1.38412857e+00 -2.38325548e+00 1.04651608e-01 -1.14294380e-01
-6.08624935e-01 7.72061050e-01 -5.24022162e-01 3.56757939e-01
-1.41295213e-02 1.43687278e-01 -1.52084336e-01 -6.85790837e-01
-1.16905812e-02 3.32560837e-01 -3.61003190e-01 -4.12530690e-01
4.31522995e-01 6.95416987e-01 -9.52113569e-01 -5.47062457e-01
3.73636335e-01 3.67639661e-01 -5.33206701e-01 5.69472134e-01
-4.65456724e-01 6.55162632e-01 -3.08566958e-01 3.90299380e-01
3.08508188e-01 3.92511368e-01 1.54212385e-01 1.60240889e-01
-3.08221191e-01 1.02491212e+00 -1.09064686e+00 1.74925041e+00
-8.43572915e-01 6.47975132e-02 3.63461316e-01 -7.56960630e-01
1.05519283e+00 8.14454496e-01 7.25636780e-02 -6.42125785e-01
9.96016711e-02 4.06658411e-01 3.54654849e-01 -3.51008892e-01
4.15968716e-01 -5.45800865e-01 -1.58220559e-01 7.85607696e-02
3.46836627e-01 -7.13666320e-01 5.85529953e-02 -2.95891434e-01
9.09391820e-01 3.82606909e-02 2.40018904e-01 -6.38103709e-02
6.54650748e-01 -1.58621311e-01 8.73998284e-01 2.98854560e-01
-2.31612459e-01 9.81885016e-01 -1.57075766e-02 1.95092022e-01
-1.20015693e+00 -1.14855909e+00 1.35284498e-01 1.41595650e+00
-2.87770689e-01 -4.65919644e-01 -1.06265366e+00 -4.95328218e-01
-3.86902153e-01 9.52025890e-01 2.92933136e-01 3.52727948e-04
-1.02944672e+00 -2.48921663e-01 6.90336585e-01 6.37448907e-01
1.23662844e-01 -1.50274539e+00 2.38288939e-01 8.58826756e-01
-4.34446216e-01 -1.43987918e+00 -9.76039886e-01 2.83886105e-01
-5.26075184e-01 -3.83444786e-01 -9.03821886e-01 -1.20547140e+00
8.77076685e-02 -2.46811621e-02 9.36296523e-01 -1.54293507e-01
4.26922500e-01 -2.93345511e-01 -6.61365390e-01 -3.98905188e-01
-1.18155098e+00 2.31987193e-01 9.72078890e-02 -2.58388906e-03
1.09932229e-01 -5.20986617e-01 -3.17723304e-02 3.61699134e-01
-7.47995257e-01 1.55782178e-01 6.43872201e-01 9.61661935e-01
9.24739420e-01 1.13172807e-01 1.11857045e+00 -5.76053441e-01
5.98973453e-01 -1.61214814e-01 -5.59103727e-01 1.28535733e-01
-5.61475940e-02 -9.09097418e-02 1.22982717e+00 -8.05900514e-01
-1.44412887e+00 3.64162058e-01 -1.18082988e+00 -4.86404657e-01
-2.63659626e-01 6.66511893e-01 -6.91489995e-01 5.87435007e-01
4.82197315e-01 2.70921201e-01 -2.49177262e-01 -7.51279771e-01
4.26617175e-01 1.19537580e+00 7.32607067e-01 -6.17946029e-01
6.42143011e-01 -4.77242529e-01 -6.61838174e-01 -1.27823007e+00
-9.62154448e-01 -5.51487148e-01 -5.24673164e-01 3.74051072e-02
8.90814960e-01 -1.19335914e+00 3.02673448e-02 4.95425582e-01
-1.31891346e+00 -7.65304089e-01 -3.27531755e-01 7.36296356e-01
-7.40354896e-01 4.87075955e-01 -8.56279790e-01 -9.85707402e-01
-5.22292793e-01 -1.36637032e+00 1.11849701e+00 1.47484303e-01
-2.34234571e-01 -7.59775758e-01 6.96048513e-02 7.30134308e-01
4.19016451e-01 -3.34608406e-01 9.76608515e-01 -7.54677832e-01
-3.57764423e-01 1.51219398e-01 2.24420667e-01 1.07714534e+00
1.67003006e-01 6.57455474e-02 -9.65958178e-01 -3.24415811e-03
1.09688461e-01 -4.18643504e-01 5.80820024e-01 4.45204496e-01
9.30589736e-01 -1.97077483e-01 7.40590617e-02 3.11072737e-01
7.91196942e-01 5.41604638e-01 4.14848298e-01 -5.04001260e-01
6.26625955e-01 5.20548105e-01 8.25478077e-01 1.10110417e-01
3.57942998e-01 7.15838790e-01 4.96370569e-02 5.41615412e-02
-7.29243696e-01 -5.91444135e-01 7.89050341e-01 1.91615057e+00
3.75998884e-01 -4.25366551e-01 -9.65014815e-01 7.27336407e-01
-1.53639901e+00 -6.81378007e-01 8.66510253e-03 1.94925153e+00
1.48865771e+00 3.93354505e-01 2.15114206e-01 8.13614428e-02
9.77993548e-01 2.77605504e-01 -1.30109757e-01 -5.39390028e-01
-8.75347108e-02 4.82696563e-01 1.24517102e-02 8.23965013e-01
-9.06502724e-01 1.63730204e+00 5.88725996e+00 1.39599359e+00
-1.24738276e+00 2.92524159e-01 3.18847120e-01 1.15515269e-01
-4.10967648e-01 3.78223099e-02 -1.08437526e+00 6.38176501e-01
1.57223618e+00 4.95009534e-02 5.42993009e-01 1.05232227e+00
4.52824414e-01 1.39175475e-01 -9.20526505e-01 8.51245165e-01
-8.52993056e-02 -1.03951585e+00 -1.33786410e-01 -3.54810596e-01
5.74363351e-01 2.33153060e-01 -1.18648566e-01 7.78555751e-01
2.30050281e-01 -7.31499910e-01 1.00747097e+00 2.51307506e-02
7.82189608e-01 -7.60636806e-01 5.05203724e-01 8.60699713e-01
-1.40900946e+00 9.93120521e-02 -2.42766470e-01 2.13806465e-01
7.26317704e-01 4.90310758e-01 -1.06281316e+00 2.73228496e-01
1.66883722e-01 1.80760100e-01 9.79253277e-02 8.16311836e-01
-6.99733019e-01 1.23983264e+00 -3.79400700e-01 -2.20381483e-01
3.37247789e-01 -4.34856687e-04 6.76799178e-01 1.47532642e+00
3.02751839e-01 2.18832999e-01 5.16030133e-01 7.95895636e-01
-7.50649720e-02 3.96265984e-01 3.44787887e-03 -5.92723668e-01
6.32157505e-01 1.07209134e+00 -4.51882690e-01 -3.53895903e-01
-5.05168796e-01 8.64395082e-01 1.84426352e-01 2.38150761e-01
-6.68679237e-01 -3.14622670e-01 7.19823837e-01 -1.54864445e-01
4.91966635e-01 -4.90776896e-01 -1.20599702e-01 -1.00646305e+00
-2.54667014e-01 -1.13020742e+00 -4.70350012e-02 -8.03271055e-01
-1.24083996e+00 9.16746557e-01 -3.53216916e-01 -8.44359756e-01
-7.32474864e-01 -4.40994173e-01 -7.59372890e-01 1.17181385e+00
-1.67663157e+00 -1.06630731e+00 4.13680434e-01 2.72372276e-01
1.41674578e+00 -2.32769459e-01 9.85692680e-01 3.94973069e-01
-6.97912931e-01 5.20990312e-01 -1.48152202e-01 2.53958732e-01
5.75539708e-01 -1.20091367e+00 7.78151512e-01 1.00641429e+00
1.92247063e-01 8.15839171e-02 5.33924520e-01 -7.97546208e-01
-9.79764462e-01 -1.31489956e+00 1.03940892e+00 -3.37990224e-02
5.62125921e-01 -7.12854564e-01 -1.06380236e+00 3.28000665e-01
1.19177572e-01 -5.36604039e-02 8.98064613e-01 8.20091516e-02
7.87430108e-02 -9.45018418e-03 -6.67138457e-01 6.54864609e-01
9.22707021e-01 -7.08204806e-01 -9.50337529e-01 8.64082873e-02
1.24506378e+00 -5.33823371e-01 -7.68170714e-01 2.54499584e-01
2.63066649e-01 -4.67388123e-01 7.87674427e-01 -3.89541388e-01
2.26241827e-01 -1.07960030e-01 -4.37235236e-01 -1.87402534e+00
-1.79000482e-01 -9.45034564e-01 1.25131518e-01 1.51248872e+00
8.93482208e-01 -5.38032949e-02 4.55257624e-01 2.69747585e-01
-9.58783329e-01 -3.59538347e-01 -8.97958517e-01 -1.02739215e+00
1.72484711e-01 -7.14004278e-01 6.01046920e-01 5.24632096e-01
9.01308507e-02 9.71351981e-01 -3.55685264e-01 2.93708771e-01
1.13786627e-02 -9.79501009e-02 6.30116582e-01 -9.02425408e-01
-6.73348248e-01 -2.67317772e-01 2.58034438e-01 -1.36926270e+00
3.90612215e-01 -6.40067577e-01 7.37118959e-01 -1.10848343e+00
-5.32872975e-01 -5.48014581e-01 -1.40041873e-01 4.84737784e-01
-4.28709745e-01 -2.98928052e-01 4.38098371e-01 -1.04625404e-01
-1.75958887e-01 1.03526986e+00 1.46699345e+00 8.84254724e-02
-5.59598207e-01 3.72534305e-01 -2.51157165e-01 6.87765181e-01
7.37397313e-01 -6.53954268e-01 -4.08435136e-01 -4.30371135e-01
-3.70700955e-01 6.60717845e-01 -3.95608455e-01 -1.00299251e+00
1.58807442e-01 -2.98965186e-01 -1.41398966e-01 -7.23312736e-01
8.12966883e-01 -5.74929655e-01 -2.65663803e-01 8.92874599e-02
-3.41857016e-01 -3.76126558e-01 2.31944457e-01 2.71143615e-01
-6.37133837e-01 -5.35890043e-01 1.03789139e+00 -1.55503526e-01
-4.49887455e-01 1.45125166e-01 -5.20452321e-01 3.22151452e-01
6.22170806e-01 5.67942895e-02 5.65846823e-02 -4.72321004e-01
-7.63683081e-01 9.12861973e-02 -1.79224342e-01 6.33340180e-01
5.28227568e-01 -1.21248722e+00 -7.53422081e-01 6.01215363e-01
-1.19481020e-01 2.27825880e-01 2.15782791e-01 2.93079019e-01
-9.44985822e-02 3.47987771e-01 9.57829133e-02 -4.38427866e-01
-1.14322639e+00 5.19251645e-01 1.89225480e-01 -2.19324961e-01
-3.37233603e-01 1.05850673e+00 8.28599334e-02 -6.00345075e-01
4.06238794e-01 -6.42887115e-01 -3.50856856e-02 -1.74117595e-01
5.57644725e-01 2.89304573e-02 -3.41284275e-02 -7.87128329e-01
-1.03081211e-01 1.14775166e-01 6.80841282e-02 -5.78390360e-01
1.28238666e+00 -1.96754590e-01 4.14778262e-01 4.57478404e-01
9.25000131e-01 4.41352934e-01 -1.49141860e+00 -3.02214682e-01
-3.15329581e-02 -2.65679806e-02 2.22232863e-01 -8.37779105e-01
-7.38506615e-01 1.19259119e+00 -1.14895768e-01 2.14621559e-01
1.00270700e+00 -1.58587173e-01 1.60917330e+00 2.00887069e-01
2.76727974e-01 -1.63610041e+00 3.09012890e-01 1.01933956e+00
1.03503251e+00 -1.02137601e+00 -8.88502896e-01 -8.05124819e-01
-9.83652115e-01 1.05125880e+00 6.59165561e-01 -3.26413708e-03
7.31914699e-01 5.69149733e-01 3.69749874e-01 5.84858418e-01
-7.89672554e-01 -3.82518739e-01 3.16188633e-01 5.71947277e-01
3.17662805e-01 2.37338021e-01 -1.13525786e-01 1.11293757e+00
-6.97985709e-01 -3.62954497e-01 2.63034523e-01 4.93282348e-01
-7.37806797e-01 -1.52442062e+00 -3.52461308e-01 -1.56345353e-01
-4.71079528e-01 -4.97517705e-01 -2.65060157e-01 2.38100171e-01
3.15052122e-02 1.21480882e+00 1.74742173e-02 -3.97090673e-01
3.42322350e-01 5.26590705e-01 1.16984963e-01 -1.26272130e+00
-6.57586992e-01 8.53681684e-01 5.03105223e-01 -2.13803038e-01
-2.08046399e-02 -5.39781451e-01 -1.51275337e+00 3.85532618e-01
-5.98506033e-01 2.92251676e-01 8.43003154e-01 1.14820206e+00
7.30112642e-02 7.56605864e-01 8.91343415e-01 -6.95722163e-01
-8.76117527e-01 -1.31744480e+00 -4.81498092e-01 2.77162297e-03
3.30273002e-01 -2.08456367e-01 -1.78770021e-01 1.33758202e-01] | [14.68235969543457, 6.7828049659729] |
497067c4-d8e9-4c91-bd00-f9e7f9be7d11 | linguistic-cues-to-deception-assessed-by | null | null | https://aclanthology.org/W12-0401 | https://aclanthology.org/W12-0401.pdf | Linguistic Cues to Deception Assessed by Computer Programs: A Meta-Analysis | null | ["Iris {\\'o}n-Gitlin", 'Bl', 'Siegfried Ludwig Sporer', 'Valerie Hauch', 'Jaume Masip'] | 2012-04-01 | null | null | null | ws-2012-4 | ['deception-detection'] | ['miscellaneous'] | [-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01
-8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01
-2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00
-3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01
-9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01
7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01
7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00
-1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01
3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01
9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01
5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00
-5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01
1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00
7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01
-1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02
-1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01
1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00
1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01
3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01
8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01
9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01
-9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01
-1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01
-2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01
-8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00
1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01
8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00
-6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01
-6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01
3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01
4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02
4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01
1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01
9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01
-1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01
-7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00
-1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02
1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01
-1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01
-1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01
-4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01
-1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01
6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01
-1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00
-7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01
-1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01
3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01
-1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01
1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01
3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00
-2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01
2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01
-1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01
4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01
2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00
1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01
-3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00
7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01
3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01
-2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01
7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01
-2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01
6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00
7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00
-3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01
-6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00
3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01
1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01
-5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01
-5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01
-2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01
-1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01
5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01
-2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01
-4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01
5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01
-6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00
1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01
8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02
-6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01
-1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01
1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01
8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03
3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01
8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01
-1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01
6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01
-9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01
-1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00
-5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02
-5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01
1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01
-4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01
1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01
6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01
5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01
1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01
1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01
1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01
1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00
-8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01
-1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01
6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01
4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00
-1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00
-7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00
2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01
-4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02
1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01
3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01
9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01
7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01
6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01
1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01
-1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00
3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01
9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01
-4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01
4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00
2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01
5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01
2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01
-2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01
-6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01
-1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01
-5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01
-1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00
-6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01
-1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01
1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02
1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01
5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01
-4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01
-1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01
6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01
7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01
1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01
1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01
8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01
8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00
-4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01
-5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02
5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01
-2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00
-3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01
6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01
5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00
3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01
-8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00
-8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01
3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01
-3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05
-1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01
8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02
6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01
5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00
1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01
2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02
-1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01
-4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02
9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01
-8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02
7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01
-3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01
-1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01
-1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01
-2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01
1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03
6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01
-4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01
9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01
8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01
1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01
-1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01
8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01
5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01
9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01
-4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01
1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01
-9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00
5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01
-7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00
-6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01
-5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01
-5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01
4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01
-9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00
-8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00
-5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01
-2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00
-3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02
2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01] | [-7.283079624176025, 3.8269400596618652] |
e201e8f8-3dbc-438e-b03c-a2c5d875ee35 | exploring-effectiveness-of-gpt-3-in | 2305.18156 | null | https://arxiv.org/abs/2305.18156v1 | https://arxiv.org/pdf/2305.18156v1.pdf | Exploring Effectiveness of GPT-3 in Grammatical Error Correction: A Study on Performance and Controllability in Prompt-Based Methods | Large-scale pre-trained language models such as GPT-3 have shown remarkable performance across various natural language processing tasks. However, applying prompt-based methods with GPT-3 for Grammatical Error Correction (GEC) tasks and their controllability remains underexplored. Controllability in GEC is crucial for real-world applications, particularly in educational settings, where the ability to tailor feedback according to learner levels and specific error types can significantly enhance the learning process. This paper investigates the performance and controllability of prompt-based methods with GPT-3 for GEC tasks using zero-shot and few-shot setting. We explore the impact of task instructions and examples on GPT-3's output, focusing on controlling aspects such as minimal edits, fluency edits, and learner levels. Our findings demonstrate that GPT-3 could effectively perform GEC tasks, outperforming existing supervised and unsupervised approaches. We also showed that GPT-3 could achieve controllability when appropriate task instructions and examples are given. | ['Naoaki Okazaki', 'Sho Takase', 'Masahiro Kaneko', 'Mengsay Loem'] | 2023-05-29 | null | null | null | null | ['grammatical-error-correction'] | ['natural-language-processing'] | [ 2.29696244e-01 1.66901767e-01 1.69262171e-01 -6.42941117e-01
-6.16788149e-01 -3.51979375e-01 4.14257735e-01 7.86801934e-01
-7.18882740e-01 4.32818413e-01 2.58744121e-01 -4.29750204e-01
-1.70416191e-01 -6.29090905e-01 -7.74376631e-01 -1.92430854e-01
-7.07723647e-02 4.92598981e-01 2.54286435e-02 -5.60431540e-01
6.07331276e-01 -1.32863969e-02 -1.71085894e+00 2.49602675e-01
1.86190343e+00 1.17211431e-01 6.67780995e-01 1.05727112e+00
-2.48209506e-01 5.35296798e-01 -4.97990668e-01 -5.16153157e-01
-1.48500264e-01 -4.13106561e-01 -4.82020169e-01 -4.49892104e-01
7.29550540e-01 1.14167772e-01 9.01891515e-02 1.13512003e+00
8.70019734e-01 5.93061090e-01 5.70738196e-01 -9.54044461e-01
-1.03317654e+00 1.03522789e+00 1.06446780e-01 2.15199634e-01
3.81030858e-01 6.25397623e-01 5.91813564e-01 -7.87696064e-01
6.33613527e-01 1.43303573e+00 6.08696699e-01 8.89098346e-01
-9.90609407e-01 -5.57258725e-01 3.55312318e-01 1.24490052e-01
-8.96768749e-01 -5.46557784e-01 2.23860636e-01 -5.19317806e-01
1.45671523e+00 6.09967001e-02 5.58659852e-01 1.04450548e+00
4.94379073e-01 8.45267832e-01 1.01125908e+00 -1.06909227e+00
1.37969792e-01 1.90371901e-01 4.57325846e-01 1.05212545e+00
6.30989820e-02 9.04839635e-02 -8.84162128e-01 4.47237372e-01
3.39115649e-01 -3.79210263e-01 -3.47293228e-01 7.38390116e-03
-1.04427862e+00 7.07823396e-01 -9.45328102e-02 2.32607014e-02
4.59140819e-03 2.28894353e-01 6.31999314e-01 7.21462250e-01
5.32174885e-01 1.06293261e+00 -7.15582013e-01 -8.92693460e-01
-7.39150465e-01 2.33549103e-01 5.78238368e-01 1.71690464e+00
1.51220918e-01 1.58753335e-01 -1.18170488e+00 1.06249797e+00
1.06723085e-01 5.30943215e-01 7.56579399e-01 -5.98286986e-01
9.20307577e-01 7.45868564e-01 -1.09572820e-01 -6.15795016e-01
-3.38705182e-01 -3.80859733e-01 -4.20828849e-01 -5.84012941e-02
2.95322508e-01 -3.39765191e-01 -8.54391277e-01 1.86553705e+00
4.87867296e-02 -1.97835863e-01 -1.99700773e-01 5.18842578e-01
9.00099456e-01 3.58875901e-01 5.94463050e-01 -2.00752765e-01
1.12490451e+00 -9.95372653e-01 -9.87483561e-01 -2.92074800e-01
1.28305006e+00 -7.68828630e-01 1.89142227e+00 4.11376357e-01
-1.16899669e+00 -5.84538937e-01 -6.50128543e-01 -3.81437153e-01
-3.18800420e-01 3.00741136e-01 3.55567783e-01 6.56666756e-01
-9.65388417e-01 9.86373365e-01 -8.67135465e-01 -5.53640723e-01
1.89270034e-01 1.85031727e-01 -9.92571190e-02 -2.03789964e-01
-1.11747587e+00 1.15062094e+00 2.74187207e-01 -1.48908153e-01
-9.45107698e-01 -1.03225458e+00 -9.39593792e-01 3.72426927e-01
3.83056343e-01 -4.62143749e-01 1.62059140e+00 -3.69108677e-01
-1.89889348e+00 6.37630939e-01 7.01423921e-03 -2.37238765e-01
5.63394487e-01 -5.24591982e-01 -1.23312518e-01 -4.50141966e-01
4.72056456e-02 7.72821307e-01 5.28137565e-01 -3.40904206e-01
-4.90552843e-01 -1.03470705e-01 -1.19994305e-01 4.58590090e-01
-5.94421387e-01 2.08733335e-01 -2.31852844e-01 -6.54631793e-01
-3.88084769e-01 -8.91565502e-01 -1.61574900e-01 -2.95275927e-01
-1.81870788e-01 -7.17443228e-01 1.83855116e-01 -4.62301224e-01
1.70084643e+00 -1.88534820e+00 2.21875936e-01 -3.63956273e-01
-5.07607162e-02 7.67135322e-01 -4.84665215e-01 4.02382582e-01
1.80893347e-01 3.90090346e-01 7.43283108e-02 -4.09358412e-01
1.12837702e-01 -7.85651356e-02 3.61322135e-01 -1.19511727e-02
1.83564946e-01 9.39849436e-01 -1.39739752e+00 -4.94086385e-01
1.77118912e-01 1.45516574e-01 -9.73786056e-01 5.46903849e-01
-4.15834844e-01 3.42970997e-01 -1.64374873e-01 4.27636623e-01
2.24298239e-02 1.94659218e-01 -2.27073103e-01 5.61258316e-01
-3.07851940e-01 5.57219088e-01 -7.83603668e-01 1.73445356e+00
-8.95631909e-01 6.56157434e-01 -2.37192661e-01 -6.35843933e-01
1.12326074e+00 2.56313235e-01 -3.60614508e-01 -9.32947159e-01
1.59186438e-01 3.60685326e-02 2.39723265e-01 -8.21564376e-01
8.78096461e-01 1.23921625e-01 -1.64374709e-01 5.16988993e-01
5.69630027e-01 -3.96527380e-01 4.90088403e-01 4.13637578e-01
1.13911772e+00 1.67640626e-01 2.85403997e-01 -6.74512744e-01
2.93454491e-02 -8.04947838e-02 3.57566148e-01 1.17193615e+00
-5.78101754e-01 4.15084034e-01 4.15179968e-01 1.33122101e-01
-7.68147469e-01 -5.28217673e-01 4.35065329e-02 1.75020015e+00
-4.06398714e-01 -7.96934783e-01 -1.06386626e+00 -7.06603944e-01
-1.09374784e-01 1.33380842e+00 -4.33999836e-01 -7.09062099e-01
-3.40994656e-01 -3.43407184e-01 3.77429992e-01 3.58402610e-01
3.81936371e-01 -1.22574306e+00 -6.67648613e-01 4.36078310e-01
-4.54908572e-02 -8.48953128e-01 -9.89830911e-01 2.68791258e-01
-1.06009269e+00 -8.04909647e-01 -2.87011623e-01 -8.02260816e-01
9.25158143e-01 1.84880849e-02 1.37047958e+00 2.43945807e-01
-1.67040125e-01 6.38451755e-01 -5.66622019e-01 -8.07178676e-01
-5.87302148e-01 2.07474634e-01 1.98905215e-01 -6.66379511e-01
4.68869299e-01 -3.51369172e-01 -1.28578454e-01 -6.89759031e-02
-3.62164646e-01 3.40559661e-01 3.38960499e-01 7.82488346e-01
2.76630610e-01 -1.73951104e-01 5.56879878e-01 -1.48072326e+00
1.38086379e+00 -7.47138634e-02 -3.88708442e-01 7.38653362e-01
-1.06056106e+00 2.54885465e-01 6.08723640e-01 -6.11668408e-01
-1.30587757e+00 -3.51298064e-01 -3.45875532e-03 -1.98401019e-01
-3.12220063e-02 4.86799866e-01 -9.57353860e-02 -1.35615870e-01
8.90800357e-01 8.19264576e-02 -2.24224329e-01 -3.00342858e-01
4.69087213e-01 7.22990096e-01 3.21375847e-01 -1.00751376e+00
2.65273750e-01 -9.07760561e-01 -4.81118232e-01 -5.71539938e-01
-9.76479113e-01 -1.46406204e-01 -6.45686984e-01 -3.43474507e-01
6.79079950e-01 -1.07789683e+00 -6.07245505e-01 4.18067396e-01
-1.18214440e+00 -7.58156300e-01 -1.54958546e-01 4.08496797e-01
-3.92950118e-01 -2.22147778e-01 -7.54972935e-01 -7.47105360e-01
-6.23841703e-01 -1.31586337e+00 7.46489227e-01 5.45500278e-01
-6.36363029e-01 -1.21551061e+00 -8.89526680e-02 3.84386420e-01
7.31649876e-01 -2.04143539e-01 1.17485607e+00 -9.26248670e-01
-3.40370476e-01 -9.55713987e-02 2.13337511e-01 2.51578718e-01
-1.30888209e-01 3.99399251e-02 -5.82274735e-01 -2.82060951e-01
-3.18086177e-01 -4.15948123e-01 4.94156569e-01 2.13018164e-01
1.24233556e+00 -5.10434985e-01 2.91125048e-02 5.28368652e-01
1.16966236e+00 3.26641910e-02 8.20670575e-02 2.68522143e-01
8.78336430e-01 4.81352359e-01 8.72938097e-01 1.94462821e-01
3.81426364e-01 6.51357532e-01 -3.63653116e-02 6.66181087e-01
-4.68620479e-01 -6.85540915e-01 5.01768649e-01 1.42050874e+00
4.58229631e-02 -4.59905446e-01 -1.07740700e+00 5.99383473e-01
-1.71523643e+00 -6.46554291e-01 -3.44589442e-01 2.26063848e+00
1.35388231e+00 1.27850026e-01 -4.83894467e-01 -4.30328816e-01
6.78542376e-01 -3.40199262e-01 -4.21415716e-01 -1.00951290e+00
2.09287107e-01 3.52794677e-01 2.26336390e-01 5.96053720e-01
-6.56203270e-01 1.34410167e+00 6.59278297e+00 8.58152449e-01
-8.35136592e-01 3.26720476e-01 3.17734867e-01 -2.96469122e-01
-3.94749433e-01 -2.35533684e-01 -1.23623192e+00 6.86097920e-01
1.27216637e+00 -5.19762695e-01 2.99324900e-01 7.53740549e-01
6.04568422e-01 1.41020566e-02 -1.21758521e+00 6.25000715e-01
7.78250303e-03 -7.88764894e-01 4.62744124e-02 -4.07699645e-01
1.02295923e+00 -2.78087795e-01 -1.18303142e-01 1.19102955e+00
7.06905544e-01 -9.96518672e-01 6.32099807e-01 5.78262448e-01
8.10712874e-01 -6.45505846e-01 3.69564325e-01 5.25458395e-01
-6.24405503e-01 -4.61035818e-02 -4.87564683e-01 -3.03472579e-01
-1.98918693e-02 5.65370023e-01 -9.94621158e-01 9.51772556e-02
5.07827699e-01 5.72374105e-01 -9.22440767e-01 9.77667570e-01
-8.41774523e-01 9.29994702e-01 2.47468073e-02 -5.51776111e-01
-1.04606226e-01 -1.21020347e-01 6.00562871e-01 1.58733737e+00
5.19707203e-01 2.72762537e-01 1.17370412e-02 7.36028910e-01
-2.05869287e-01 4.30545419e-01 -5.27909040e-01 -2.79656172e-01
1.03856933e+00 9.87388849e-01 -2.29666203e-01 -4.38540488e-01
-2.10151393e-02 8.06115985e-01 1.01013458e+00 1.93672150e-01
-4.41338927e-01 -6.43871546e-01 5.25573730e-01 2.20761389e-01
1.33184576e-02 -3.62015754e-01 -5.82407236e-01 -1.17721236e+00
-2.24270537e-01 -1.06656754e+00 6.70668036e-02 -7.93420792e-01
-1.00709224e+00 2.62049675e-01 7.57936537e-02 -9.74990129e-01
6.52347924e-03 -5.73102713e-01 -1.08985889e+00 7.66788960e-01
-1.22599781e+00 -6.25738263e-01 -1.93297863e-01 2.69918859e-01
1.01145911e+00 -4.18323636e-01 8.05668533e-01 6.69836551e-02
-1.00686240e+00 1.12721658e+00 1.70150578e-01 -2.21272007e-01
1.07454789e+00 -1.55397105e+00 5.27463615e-01 1.17865598e+00
-1.01516284e-01 9.04545009e-01 8.90621066e-01 -8.18576872e-01
-1.31703615e+00 -1.29936969e+00 1.46582150e+00 -6.06823087e-01
3.42419863e-01 -4.65731472e-01 -1.02822697e+00 6.24095738e-01
1.96374938e-01 -1.27430290e-01 5.58610916e-01 5.85003734e-01
5.04606143e-02 1.77019149e-01 -9.91161585e-01 7.49487698e-01
1.44048989e+00 -4.55679208e-01 -6.68134630e-01 5.59145689e-01
1.08810282e+00 -9.09946203e-01 -8.62848163e-01 7.89238214e-02
5.78359477e-02 -5.63751221e-01 5.12141228e-01 -9.02368784e-01
7.71508336e-01 3.27440768e-01 3.86991501e-01 -2.17351913e+00
-5.06061614e-01 -6.26351655e-01 -8.53464007e-02 1.53886950e+00
4.78580445e-01 -2.31508106e-01 2.46969119e-01 1.10309458e+00
-7.48603642e-01 -7.06143200e-01 -4.36468393e-01 -7.87593305e-01
3.24979633e-01 -3.06505084e-01 4.69072670e-01 9.25278723e-01
3.00170451e-01 4.05459315e-01 -1.91390768e-01 -1.86001241e-01
3.46646070e-01 -5.11007786e-01 6.09421134e-01 -1.13108492e+00
-1.04530424e-01 -3.86515141e-01 6.95540532e-02 -8.06332111e-01
2.00339213e-01 -1.12066603e+00 4.70823258e-01 -1.33817649e+00
-1.60675738e-02 -5.96319735e-01 -4.82315905e-02 7.80609548e-01
-9.74810660e-01 -6.03324890e-01 3.69920462e-01 -2.41462603e-01
-8.05027962e-01 8.33976209e-01 1.46515095e+00 8.45752731e-02
-5.38627386e-01 -2.01923400e-01 -7.07188964e-01 2.68790245e-01
7.46610641e-01 -5.05687237e-01 -5.96655369e-01 -8.28284442e-01
2.30984300e-01 -1.04216345e-01 -2.97533453e-01 -8.90985966e-01
5.54312825e-01 -2.97445595e-01 -1.58336744e-01 2.80890465e-01
-4.68979388e-01 -1.20159827e-01 -5.02243400e-01 6.02597445e-02
-9.55241144e-01 2.02227145e-01 4.47891265e-01 3.63546193e-01
-2.48024203e-02 -6.46564662e-01 6.22821450e-01 -2.78068870e-01
-6.06263220e-01 2.10989043e-01 -5.74974954e-01 4.59674269e-01
7.78223336e-01 9.21811312e-02 -4.46527094e-01 -2.39855573e-01
-4.42773372e-01 7.68066943e-01 1.38213083e-01 7.30464280e-01
5.99269569e-01 -1.15352964e+00 -8.33090484e-01 3.00820589e-01
5.08509278e-01 1.81647584e-01 3.02743286e-01 7.39749610e-01
-2.72088081e-01 4.55678463e-01 -7.67065883e-02 -5.67031384e-01
-1.53476822e+00 2.69056410e-01 2.26822928e-01 -3.88652891e-01
-4.91075367e-01 1.22838044e+00 -2.81422883e-01 -9.84803498e-01
4.83070433e-01 -3.81036848e-01 -2.46985301e-01 -1.44182831e-01
5.19901454e-01 1.59472108e-01 4.08344895e-01 1.68634295e-01
1.83671936e-01 2.46027216e-01 -4.19034749e-01 1.73696548e-01
1.28640902e+00 -2.24382207e-01 3.26734871e-01 4.48505670e-01
7.13998795e-01 -3.51793826e-01 -1.22344446e+00 -2.24925905e-01
2.05959484e-01 -4.80781823e-01 2.50220388e-01 -1.31988382e+00
-5.44672191e-01 1.18864584e+00 4.26161706e-01 -4.61353421e-01
8.89862299e-01 -4.80720490e-01 3.59678745e-01 5.61570585e-01
4.72984195e-01 -1.34838891e+00 1.46426439e-01 1.07498479e+00
6.34280264e-01 -1.41616225e+00 -3.41155797e-01 -2.07954511e-01
-8.12375069e-01 1.09618795e+00 1.33214080e+00 3.20177794e-01
1.54206306e-01 -1.56598434e-01 -1.53423110e-02 -2.31084779e-01
-1.34762859e+00 1.56437457e-01 2.84381658e-01 5.69610298e-01
1.23979855e+00 2.87463635e-01 -6.88398421e-01 6.98467791e-01
-5.74139476e-01 -3.92359942e-02 7.98879504e-01 9.89092827e-01
-6.45007432e-01 -1.20636821e+00 9.25042853e-02 6.07466221e-01
-1.17778793e-01 -5.28614461e-01 7.11150374e-03 5.04270554e-01
-3.41300070e-02 8.83982182e-01 -1.00697696e-01 -1.75205216e-01
7.04818666e-01 4.41393703e-01 6.55393302e-01 -1.42125607e+00
-1.09412277e+00 -5.91560900e-01 2.86894828e-01 -4.22658443e-01
1.89676315e-01 -5.69186449e-01 -1.10140872e+00 -1.96174651e-01
-4.43897039e-01 2.27825314e-01 6.45869792e-01 9.99548495e-01
5.81147492e-01 8.36522579e-01 6.71675354e-02 -2.79856920e-01
-7.54231930e-01 -1.47750723e+00 -2.26902500e-01 3.62392873e-01
-9.43302140e-02 -5.86701870e-01 -1.61323443e-01 -5.66654699e-03] | [11.236106872558594, 9.212557792663574] |
ad197b89-9eca-4b6a-955c-1d7cf21c8dde | composable-text-control-operations-in-latent | 2208.00638 | null | https://arxiv.org/abs/2208.00638v2 | https://arxiv.org/pdf/2208.00638v2.pdf | Composable Text Controls in Latent Space with ODEs | Real-world text applications often involve composing a wide range of text control operations, such as editing the text w.r.t. an attribute, manipulating keywords and structure, and generating new text of desired properties. Prior work typically learns/finetunes a language model (LM) to perform individual or specific subsets of operations. Recent research has studied combining operations in a plug-and-play manner, often with costly search or optimization in the complex sequence space. This paper proposes a new efficient approach for composable text operations in the compact latent space of text. The low-dimensionality and differentiability of the text latent vector allow us to develop an efficient sampler based on ordinary differential equations (ODEs) given arbitrary plug-in operators (e.g., attribute classifiers). By connecting pretrained LMs (e.g., GPT2) to the latent space through efficient adaption, we then decode the sampled vectors into desired text sequences. The flexible approach permits diverse control operators (sentiment, tense, formality, keywords, etc.) acquired using any relevant data from different domains. Experiments show that composing those operators within our approach manages to generate or edit high-quality text, substantially improving over previous methods in terms of generation quality and efficiency. | ['Zhiting Hu', 'Zhen Li', 'Shuguang Cui', 'Xiaodong He', 'Junwei Bao', 'Xiaodan Liang', 'Zichao Yang', 'Yuan Gao', 'Zeyu Feng', 'Guangyi Liu'] | 2022-08-01 | null | null | null | null | ['text-style-transfoer'] | ['natural-language-processing'] | [ 5.57657182e-01 -1.36689439e-01 -5.62195219e-02 -1.99159399e-01
-6.48990452e-01 -1.03653479e+00 1.03909564e+00 2.67809242e-01
-4.77511585e-01 6.36217058e-01 7.40076974e-02 -2.96164781e-01
-1.49065843e-02 -8.24883282e-01 -7.42067099e-01 -6.58196390e-01
2.94321150e-01 9.85441566e-01 -9.47268754e-02 -3.99716526e-01
6.15216970e-01 4.15594339e-01 -1.78856289e+00 2.61416972e-01
1.41237366e+00 6.63956285e-01 3.63249660e-01 9.24706399e-01
-8.48020673e-01 2.14600071e-01 -5.43934405e-01 -6.18492842e-01
3.74322951e-01 -4.73700643e-01 -6.29834890e-01 1.91233426e-01
2.78699994e-01 9.20891017e-03 2.36267298e-01 1.00271678e+00
3.02462310e-01 3.75047505e-01 1.01509786e+00 -1.15074825e+00
-6.92562401e-01 1.07123315e+00 -1.05806463e-03 -2.72507250e-01
5.51204860e-01 1.22477666e-01 1.27290070e+00 -9.74263608e-01
6.28437281e-01 1.29080153e+00 4.84509319e-01 5.83056152e-01
-1.76193261e+00 -4.72608835e-01 1.35627449e-01 -2.49077499e-01
-1.11352074e+00 -3.40421349e-01 5.15522361e-01 -5.33557296e-01
9.79623556e-01 6.16565526e-01 7.06612408e-01 1.31267953e+00
3.37993801e-01 9.67893958e-01 9.19863224e-01 -5.79877317e-01
2.85812557e-01 4.69896078e-01 -2.50371158e-01 7.83913672e-01
8.58537555e-02 -3.58051687e-01 -5.82190931e-01 -3.40805441e-01
4.70323533e-01 1.64594632e-02 -1.90114807e-02 -4.04466212e-01
-1.51066673e+00 8.91798377e-01 -2.48446971e-01 1.46490052e-01
-1.78251624e-01 -5.04584052e-02 4.35298175e-01 6.13874018e-01
4.81055856e-01 8.72319579e-01 -7.32892156e-01 -4.14660126e-01
-1.14352226e+00 4.93319541e-01 1.14428449e+00 1.21273291e+00
8.45690966e-01 -9.95605662e-02 -5.11706531e-01 8.51965368e-01
6.72245175e-02 5.43639660e-01 7.90115654e-01 -5.23556173e-01
6.43298686e-01 5.77676356e-01 1.48576841e-01 -5.35212934e-01
-6.91481903e-02 -1.74981549e-01 -9.08061683e-01 -1.56998724e-01
3.83857042e-01 -2.33239964e-01 -9.43498969e-01 1.63499546e+00
2.43920147e-01 -3.00151676e-01 -1.76753998e-01 1.32662281e-01
-9.63815898e-02 9.01728332e-01 -1.27582401e-01 -4.76404309e-01
1.20144987e+00 -8.25274348e-01 -7.41179049e-01 -1.03675164e-01
7.81125426e-01 -8.68838251e-01 1.77601707e+00 6.09475553e-01
-1.31178474e+00 -4.19340312e-01 -8.00162017e-01 -2.12042049e-01
-6.18888021e-01 2.15786055e-01 3.83011401e-01 5.74576259e-01
-9.06728208e-01 8.27771783e-01 -6.35960400e-01 -1.86219186e-01
8.88362452e-02 5.08313417e-01 -7.75283575e-02 5.09668767e-01
-1.21281064e+00 6.43173337e-01 5.63889503e-01 -1.93222269e-01
-6.38878286e-01 -8.65434527e-01 -6.77337885e-01 5.00289276e-02
5.34097552e-01 -1.03564405e+00 1.28491378e+00 -9.63302851e-01
-2.16745472e+00 6.44041300e-01 -1.33788049e-01 -4.20409352e-01
9.15068746e-01 -3.00552845e-01 -1.08594157e-01 -2.75817335e-01
6.71080798e-02 5.10793149e-01 1.42754889e+00 -7.32588947e-01
-6.13947153e-01 -7.71675706e-02 -2.01768652e-01 2.80174434e-01
-7.16744661e-01 -5.37529364e-02 -3.01146448e-01 -1.09549618e+00
-7.58889541e-02 -1.17489564e+00 -1.64219588e-01 -6.70899972e-02
-6.61801994e-01 -4.44646925e-01 5.55549324e-01 -3.95439267e-01
1.53659666e+00 -1.85192466e+00 7.55063593e-01 2.75105506e-01
-1.14619546e-02 -3.10366731e-02 1.35437017e-02 7.57332325e-01
1.01776436e-01 4.46474403e-01 -5.48812032e-01 -5.42829096e-01
3.81084085e-01 2.19374895e-01 -5.26023388e-01 1.69225693e-01
7.74598420e-02 1.06559336e+00 -7.66492546e-01 -5.80543697e-01
-6.93463385e-02 5.89763410e-02 -9.30539668e-01 1.51974678e-01
-9.37044978e-01 1.86856449e-01 -4.97782379e-01 1.93739057e-01
7.71772712e-02 -1.12725116e-01 1.48367330e-01 -4.22651991e-02
-2.52503604e-01 5.07505298e-01 -1.45639658e+00 1.88260245e+00
-8.76811147e-01 4.40616578e-01 -3.02171916e-01 -9.21986818e-01
1.02292156e+00 1.50240093e-01 2.31546685e-01 -8.97193700e-02
3.01151127e-02 3.65840137e-01 -2.94766098e-01 -3.68684322e-01
7.57479072e-01 -7.94396624e-02 -3.61074686e-01 8.41852367e-01
1.65297225e-01 -5.34175992e-01 5.72824895e-01 2.19414249e-01
7.88803458e-01 1.80777714e-01 3.19365621e-01 -3.60590905e-01
5.72597504e-01 -2.01841235e-01 -1.18569568e-01 8.70785713e-01
6.23362958e-01 1.88933060e-01 6.80988491e-01 -8.02177340e-02
-1.36422133e+00 -9.11334157e-01 -6.22814894e-02 1.41730630e+00
-4.56228167e-01 -7.24703550e-01 -8.20866823e-01 -6.01891875e-01
-6.83307974e-03 1.13743687e+00 -5.82630217e-01 -2.76175171e-01
-6.47198439e-01 -5.99516332e-01 5.15660942e-01 2.00325906e-01
2.88529634e-01 -1.11649406e+00 -4.08416092e-01 3.36936057e-01
-1.39713228e-01 -6.38401806e-01 -9.90860939e-01 3.97925735e-01
-9.25730228e-01 -3.02680612e-01 -5.60821772e-01 -7.28678644e-01
7.25122273e-01 -3.85312915e-01 9.53652084e-01 -3.65740180e-01
-1.46444827e-01 1.79261923e-01 -2.69757301e-01 -3.04331571e-01
-8.37456644e-01 6.23282135e-01 1.38523132e-01 3.40071797e-01
-6.23181127e-02 -5.77039778e-01 -5.09927236e-02 1.06964251e-02
-1.27138948e+00 2.65435100e-01 6.17532849e-01 9.83947456e-01
5.97267568e-01 1.95167214e-01 1.05159625e-01 -1.16906929e+00
1.16701424e+00 -2.90868014e-01 -7.84705639e-01 4.48243052e-01
-8.65137696e-01 8.80107045e-01 9.15625751e-01 -9.62379515e-01
-1.05229819e+00 1.11456878e-01 2.06488326e-01 -1.82742491e-01
1.09993011e-01 5.13504863e-01 -2.89564610e-01 4.98280436e-01
7.07875073e-01 5.77652574e-01 3.82315181e-02 -4.93111819e-01
7.13736355e-01 7.09982216e-01 2.14215860e-01 -8.74350309e-01
1.03717244e+00 2.23143816e-01 -1.78860947e-01 -7.39538848e-01
-6.69575393e-01 -9.63044539e-02 -7.51827300e-01 2.01042145e-01
6.84391677e-01 -3.13949764e-01 -6.02783382e-01 4.25326943e-01
-9.84458387e-01 -6.01787388e-01 -6.17722988e-01 1.93619013e-01
-7.95659840e-01 2.78530687e-01 -5.25548279e-01 -5.84530711e-01
-5.31036317e-01 -1.22072279e+00 1.29894471e+00 -2.00533435e-01
-6.75122738e-01 -1.12364364e+00 -2.35548685e-03 -1.07468866e-01
3.27447057e-01 -1.46128401e-01 1.22150016e+00 -8.30257535e-01
-5.37454069e-01 -2.58490741e-01 4.25532669e-01 3.62038136e-01
1.96641758e-01 3.77441496e-01 -4.63409007e-01 -1.93255946e-01
-3.92981693e-02 -2.81015281e-02 6.34905040e-01 5.66089079e-02
1.29165053e+00 -8.43786120e-01 -2.45787650e-01 8.68407369e-01
9.67307866e-01 3.83160226e-02 3.16864759e-01 2.87310690e-01
5.53758025e-01 4.56482112e-01 4.91979033e-01 7.00011849e-01
7.69395381e-02 7.12677181e-01 -2.17372477e-01 5.05164623e-01
4.80849952e-01 -5.54110289e-01 7.86163449e-01 8.68885219e-01
1.59146011e-01 -3.91466826e-01 -8.14580500e-01 1.87638417e-01
-1.66540408e+00 -9.22849417e-01 1.68371305e-01 2.02766967e+00
1.48343360e+00 3.82977873e-01 2.61830483e-02 2.01646686e-01
5.30826569e-01 -9.73681826e-03 -5.82427621e-01 -5.87943554e-01
-2.31551751e-02 3.85480732e-01 4.62482035e-01 5.90416610e-01
-8.54882181e-01 1.08401465e+00 5.56074858e+00 1.15567887e+00
-1.20371580e+00 -2.03892723e-01 1.93257064e-01 -1.96611762e-01
-9.14764047e-01 8.15695673e-02 -1.10628843e+00 5.12751043e-01
9.68000948e-01 -5.69868624e-01 8.95183265e-01 6.28788173e-01
1.60504445e-01 2.72943377e-01 -1.30051386e+00 8.75478089e-01
1.12707831e-01 -1.44456923e+00 6.05400622e-01 -2.64835674e-02
7.11741507e-01 -4.91449803e-01 2.42571339e-01 3.58404249e-01
4.25460577e-01 -8.65287125e-01 1.05815554e+00 7.14680731e-01
9.59232330e-01 -5.60102046e-01 -1.93722788e-02 6.13034487e-01
-8.43205154e-01 -6.50099963e-02 -9.00945216e-02 2.72828877e-01
7.39400610e-02 5.08913159e-01 -8.60975385e-01 1.85122639e-01
3.24485481e-01 6.88865960e-01 -5.37639081e-01 2.82185584e-01
-2.49115616e-01 4.92452890e-01 -5.14562964e-01 -5.42198420e-01
1.91589579e-01 -5.49364507e-01 8.29559624e-01 1.33106184e+00
5.40112913e-01 -2.69209027e-01 1.08385041e-01 1.15752161e+00
-1.36339754e-01 4.36994344e-01 -4.02546704e-01 -4.93094891e-01
3.34499449e-01 1.25533402e+00 -7.07926691e-01 -6.20073557e-01
1.03508331e-01 1.34875607e+00 1.11263804e-01 3.63449991e-01
-7.30809212e-01 -5.72706163e-01 5.59708118e-01 1.80443317e-01
5.01282811e-01 -4.66334581e-01 -4.07533377e-01 -1.45883679e+00
2.60844946e-01 -1.25702572e+00 1.58657581e-01 -6.48249686e-01
-1.12880147e+00 5.55524945e-01 2.51196146e-01 -1.01423585e+00
-4.78112310e-01 -4.40012127e-01 -3.22457492e-01 9.68508780e-01
-8.73892307e-01 -7.74273634e-01 9.45168138e-02 6.53818429e-01
9.43617642e-01 -3.45160842e-01 7.98649251e-01 -9.62766781e-02
-5.61416090e-01 8.28506172e-01 2.44456246e-01 -2.26165667e-01
6.90889060e-01 -1.52258539e+00 6.51892602e-01 6.03830516e-01
3.49049807e-01 9.79076862e-01 8.38365793e-01 -7.75822699e-01
-1.56773150e+00 -1.01672065e+00 9.95338380e-01 -6.04790390e-01
1.00983059e+00 -9.61100638e-01 -7.76799023e-01 8.45478117e-01
1.06400378e-01 -5.66894412e-01 4.07283694e-01 -4.83600907e-02
-2.20162734e-01 -6.62884191e-02 -7.72058070e-01 1.15353870e+00
1.06171560e+00 -6.13627374e-01 -5.34946859e-01 4.05095786e-01
9.03841078e-01 -5.12939751e-01 -7.34932125e-01 -7.45750815e-02
5.14406443e-01 -4.73028362e-01 7.45956302e-01 -8.17505717e-01
3.77650201e-01 -1.24382533e-01 6.16087615e-02 -1.58297229e+00
-8.31983387e-02 -1.35601056e+00 -1.98880047e-01 1.27122557e+00
6.31865978e-01 -6.78600252e-01 4.73784775e-01 4.50900584e-01
-4.65961099e-02 -7.40520597e-01 -4.07687157e-01 -6.51199818e-01
1.62252337e-01 -5.08848310e-01 8.41933608e-01 7.10590005e-01
-5.59441894e-02 6.15520298e-01 -1.97163016e-01 -1.97820261e-01
3.43454152e-01 1.57999203e-01 8.20973814e-01 -1.25622296e+00
-4.93693501e-01 -6.75050795e-01 5.56919575e-02 -1.32701409e+00
3.39490294e-01 -1.40031648e+00 -2.14278605e-02 -1.26457489e+00
-1.85863107e-01 -6.01882815e-01 3.06311309e-01 4.55269396e-01
-2.35091805e-01 -4.34398592e-01 1.59578964e-01 3.06151360e-01
-1.71014711e-01 6.00996137e-01 1.21414876e+00 -2.16031566e-01
-5.11937916e-01 1.88233823e-01 -4.86346394e-01 5.02797782e-01
1.00398588e+00 -4.96447295e-01 -5.48995972e-01 -2.37147823e-01
7.64876783e-01 -3.69417183e-02 2.72473320e-02 -6.06302381e-01
2.90944964e-01 -3.67256969e-01 5.68947606e-02 -3.84374321e-01
1.66134655e-01 -5.53459585e-01 1.40207246e-01 3.37447762e-01
-9.31008518e-01 3.58295321e-01 3.94444391e-02 4.70614702e-01
-2.03100196e-03 -4.49564248e-01 5.71218908e-01 -1.55425817e-01
-1.10262290e-01 2.78207928e-01 -5.74067950e-01 3.48650903e-01
8.27351093e-01 -1.96363628e-01 1.68874055e-01 -2.04641327e-01
-7.72014856e-01 -3.70589197e-02 4.44690883e-01 4.84626949e-01
1.92994431e-01 -1.03281534e+00 -6.76257610e-01 5.67624092e-01
-5.37707955e-02 -3.64660658e-03 -2.93068141e-01 6.07706070e-01
-3.53372216e-01 3.56592208e-01 5.03488369e-02 -5.48835337e-01
-9.66797948e-01 5.22422254e-01 2.69458622e-01 -5.76240540e-01
-4.81953412e-01 7.74934888e-01 -1.51670337e-01 -5.50841570e-01
1.42086238e-01 -7.66073525e-01 9.02686939e-02 5.83474934e-01
1.12485677e-01 2.76866913e-01 1.06502742e-01 -1.96685959e-02
1.97591320e-01 3.88728648e-01 -1.97976604e-01 -5.17272115e-01
1.11834955e+00 -1.43566052e-03 -4.01683480e-01 8.73341143e-01
1.24393880e+00 2.58045346e-01 -1.03438830e+00 -2.54951358e-01
2.08849534e-01 -1.17098503e-01 -2.66243398e-01 -4.96972322e-01
-5.08007646e-01 7.20797479e-01 3.71518806e-02 4.45059568e-01
8.96572232e-01 -1.39421657e-01 7.06897736e-01 7.60962069e-01
1.09655358e-01 -1.37961137e+00 2.71019638e-01 6.79966033e-01
9.31221366e-01 -6.91176772e-01 -2.34075814e-01 -2.92651076e-02
-7.69048691e-01 1.35039806e+00 1.69486225e-01 6.55171052e-02
6.99076235e-01 4.68119651e-01 -2.76641965e-01 1.74095392e-01
-1.12362158e+00 1.37498066e-01 3.29354584e-01 4.37775664e-02
3.74890387e-01 1.11397304e-01 -3.43835115e-01 1.94580704e-01
-7.98047423e-01 -2.01426148e-01 5.04487872e-01 7.21483827e-01
-3.32736284e-01 -1.53975821e+00 -2.73192704e-01 8.83824885e-01
-2.17896715e-01 -4.55949873e-01 -2.42952615e-01 4.48717386e-01
5.68523854e-02 4.65953082e-01 1.45498395e-01 -9.96670723e-02
3.67609262e-01 5.60633421e-01 4.26862687e-01 -8.91504884e-01
-6.96970522e-01 1.08337365e-01 -1.24667035e-02 -2.72351533e-01
1.45665556e-02 -1.24214852e+00 -1.14835358e+00 -1.63762927e-01
-3.43696773e-01 3.02226394e-01 7.70707309e-01 9.43306565e-01
2.01910168e-01 4.53215271e-01 4.86306727e-01 -6.90973163e-01
-1.14659035e+00 -8.94916177e-01 -5.11253476e-01 4.42051679e-01
2.53585637e-01 -3.56108129e-01 -3.99112612e-01 5.24788499e-01] | [11.797492027282715, 9.168097496032715] |
79b8e1e6-2a66-437d-bfc6-d26a849bc01e | improving-facial-attribute-prediction-using | 1704.08740 | null | http://arxiv.org/abs/1704.08740v1 | http://arxiv.org/pdf/1704.08740v1.pdf | Improving Facial Attribute Prediction using Semantic Segmentation | Attributes are semantically meaningful characteristics whose applicability
widely crosses category boundaries. They are particularly important in
describing and recognizing concepts where no explicit training example is
given, \textit{e.g., zero-shot learning}. Additionally, since attributes are
human describable, they can be used for efficient human-computer interaction.
In this paper, we propose to employ semantic segmentation to improve facial
attribute prediction. The core idea lies in the fact that many facial
attributes describe local properties. In other words, the probability of an
attribute to appear in a face image is far from being uniform in the spatial
domain. We build our facial attribute prediction model jointly with a deep
semantic segmentation network. This harnesses the localization cues learned by
the semantic segmentation to guide the attention of the attribute prediction to
the regions where different attributes naturally show up. As a result of this
approach, in addition to recognition, we are able to localize the attributes,
despite merely having access to image level labels (weak supervision) during
training. We evaluate our proposed method on CelebA and LFWA datasets and
achieve superior results to the prior arts. Furthermore, we show that in the
reverse problem, semantic face parsing improves when facial attributes are
available. That reaffirms the need to jointly model these two interconnected
tasks. | ['Mahdi M. Kalayeh', 'Boqing Gong', 'Mubarak Shah'] | 2017-04-27 | improving-facial-attribute-prediction-using-1 | http://openaccess.thecvf.com/content_cvpr_2017/html/Kalayeh_Improving_Facial_Attribute_CVPR_2017_paper.html | http://openaccess.thecvf.com/content_cvpr_2017/papers/Kalayeh_Improving_Facial_Attribute_CVPR_2017_paper.pdf | cvpr-2017-7 | ['face-parsing'] | ['computer-vision'] | [ 3.16623867e-01 4.16619956e-01 -3.06674480e-01 -8.41119289e-01
-4.11622792e-01 -4.85701472e-01 6.64947987e-01 1.45106837e-01
-3.60188127e-01 5.47693253e-01 -5.00546992e-02 1.50211066e-01
-1.34514287e-01 -9.38404024e-01 -7.98981786e-01 -7.54013002e-01
2.14115262e-01 7.54034817e-01 -1.08897030e-01 -4.75191846e-02
2.20890008e-02 5.92729628e-01 -1.99683213e+00 2.37997100e-01
7.29559541e-01 1.46092737e+00 3.47080767e-01 1.88366711e-01
-4.43402648e-01 6.78631186e-01 -2.61764586e-01 -5.28437018e-01
5.24492189e-03 -4.42715228e-01 -9.46314871e-01 4.73603129e-01
5.91739535e-01 -1.88593209e-01 1.31286532e-01 1.06837511e+00
6.88724518e-02 2.00495437e-01 9.36285973e-01 -1.29408598e+00
-6.20688558e-01 2.72408605e-01 -5.47453642e-01 -1.33940756e-01
2.23946944e-01 -1.22381181e-01 1.23257422e+00 -9.10248101e-01
6.31342411e-01 1.18741894e+00 3.10172886e-01 8.01318228e-01
-1.20874870e+00 -5.23810863e-01 4.69635457e-01 2.82072157e-01
-1.47362304e+00 -7.18601704e-01 8.27656865e-01 -4.27861422e-01
4.41323578e-01 -2.90462896e-02 4.42241400e-01 1.08631527e+00
-3.36929500e-01 9.20091629e-01 8.93294811e-01 -3.69650543e-01
4.23777878e-01 2.88617611e-01 -3.63906696e-02 7.06972897e-01
-7.35317022e-02 -2.76262343e-01 -6.08722925e-01 2.71370653e-02
4.57564682e-01 1.20910220e-01 5.76782506e-03 -5.34085989e-01
-7.19593346e-01 6.90902829e-01 3.83448541e-01 1.82967946e-01
-5.61111987e-01 2.43051589e-01 1.44737184e-01 -4.41954359e-02
5.63197136e-01 4.79029566e-02 -3.72988999e-01 1.21275395e-01
-8.29932332e-01 -5.58691695e-02 5.38680017e-01 1.03322589e+00
1.16926455e+00 -1.24113701e-01 -3.15518398e-03 1.08326077e+00
2.82878339e-01 5.16791999e-01 2.23482996e-01 -9.52910244e-01
1.18105605e-01 5.21338880e-01 -1.62672013e-01 -8.82421970e-01
-2.35962659e-01 -2.23955482e-01 -6.78557158e-01 1.03665948e-01
5.09873271e-01 8.39321613e-02 -1.08816993e+00 2.04139996e+00
4.77228999e-01 3.68803412e-01 -3.78966928e-02 9.22400057e-01
6.12051010e-01 4.92145598e-01 7.61840343e-01 7.95880854e-02
1.67754579e+00 -5.44944286e-01 -6.75097346e-01 -4.51455176e-01
5.51550567e-01 -5.11265278e-01 1.18773150e+00 1.41347662e-01
-6.80836916e-01 -5.10973394e-01 -7.68346429e-01 -5.68244644e-02
-5.19360304e-01 1.77177280e-01 9.22929645e-01 5.87702990e-01
-8.73839498e-01 6.32346630e-01 -6.65007770e-01 -6.83761001e-01
9.85093772e-01 4.50508386e-01 -5.46871781e-01 -1.70487717e-01
-1.06608927e+00 6.15104079e-01 2.81212449e-01 -6.98137060e-02
-7.80463636e-01 -4.35729116e-01 -9.83593941e-01 2.56150782e-01
5.13499856e-01 -5.55686235e-01 1.00421953e+00 -1.55148566e+00
-1.25290096e+00 1.14710689e+00 -5.42696476e-01 -1.92029327e-01
8.76585841e-02 -1.09306596e-01 -1.48538917e-01 4.24872220e-01
2.21392393e-01 1.02852154e+00 1.01021528e+00 -1.29593039e+00
-7.81459928e-01 -7.90342450e-01 1.51138738e-01 1.17096923e-01
-5.01831293e-01 -9.42622274e-02 -5.10245979e-01 -4.33627546e-01
2.48487145e-01 -8.66914153e-01 -6.62615672e-02 2.06483334e-01
-3.90352398e-01 -4.94718939e-01 7.85490215e-01 -3.43811631e-01
6.57538652e-01 -2.32749891e+00 5.53917363e-02 3.58258486e-01
1.08342253e-01 -2.56013162e-02 -1.38162762e-01 -3.89100686e-02
-2.52281316e-04 2.09199131e-01 -3.82574648e-01 -3.26552570e-01
1.12554794e-02 4.53196168e-01 -1.47200301e-01 5.07678509e-01
5.72551668e-01 8.70769143e-01 -6.08345985e-01 -7.54950345e-01
2.94735897e-02 5.19755483e-01 -5.60905576e-01 1.31319493e-01
-5.32616019e-01 5.91805458e-01 -8.45052540e-01 8.48227620e-01
6.08948410e-01 -2.20454305e-01 4.69472781e-02 -8.36612135e-02
2.85020351e-01 -1.34503776e-02 -9.40976679e-01 1.71604633e+00
-5.69809914e-01 3.82471621e-01 1.02513030e-01 -1.25201225e+00
7.79183149e-01 2.97614396e-01 5.35012364e-01 -7.29390204e-01
2.07625106e-01 3.75156403e-02 -2.00614333e-01 -4.71748799e-01
7.50317499e-02 -3.46866250e-01 -2.79786270e-02 4.21057045e-01
3.85629624e-01 1.10853307e-01 7.76066631e-02 1.67061836e-01
5.30323744e-01 3.56403112e-01 3.06463778e-01 -3.20852041e-01
5.94651937e-01 -2.25346774e-01 7.25821793e-01 4.17796850e-01
-1.11766830e-01 5.18517971e-01 8.34839344e-01 -3.88862610e-01
-9.07991886e-01 -8.80882740e-01 -2.87943691e-01 1.49295664e+00
2.16514885e-01 -1.67546973e-01 -1.04551911e+00 -7.91535735e-01
-1.22475460e-01 6.88202679e-01 -8.50886226e-01 -5.47528453e-03
-3.22527230e-01 -4.20637876e-01 1.89451337e-01 5.70951581e-01
4.33573157e-01 -1.15434027e+00 -5.72313368e-01 -3.21339667e-02
-1.57686681e-01 -1.45712066e+00 -1.41395420e-01 1.53263867e-01
-6.16482973e-01 -9.64806437e-01 -5.14528036e-01 -7.47312665e-01
7.89822757e-01 -1.24163972e-02 1.12864780e+00 2.18911469e-01
-2.58500516e-01 6.04901493e-01 -3.80040318e-01 -4.24928337e-01
-4.09651995e-02 5.20215780e-02 -1.15309656e-03 5.66710770e-01
7.34954953e-01 -6.37051702e-01 -5.45091450e-01 2.75225341e-01
-6.88965678e-01 -8.14169422e-02 6.71137631e-01 7.47517228e-01
7.93516994e-01 -2.96714395e-01 5.41553974e-01 -1.21776867e+00
5.32749183e-02 -5.60574353e-01 -3.12668771e-01 3.37526619e-01
-3.25575948e-01 1.50065869e-01 5.52354693e-01 -3.68999094e-01
-1.16142154e+00 5.11663556e-01 -1.91980541e-01 -3.80149484e-01
-7.94716060e-01 1.89129889e-01 -7.53487468e-01 2.17419520e-01
3.03602874e-01 8.68967362e-03 1.15472853e-01 -3.57178628e-01
4.43429917e-01 5.89464128e-01 3.34633261e-01 -8.31926227e-01
5.38056135e-01 6.95121288e-01 1.37827188e-01 -1.08755910e+00
-1.06063902e+00 -3.75023365e-01 -9.41638708e-01 -1.50197893e-01
1.19554067e+00 -7.65713930e-01 -8.27340603e-01 1.67059779e-01
-1.08455813e+00 -8.06272030e-02 -3.03129733e-01 2.62709439e-01
-8.48002434e-01 1.58199400e-01 -3.52069318e-01 -8.91857028e-01
1.32476628e-01 -9.70787525e-01 1.26348484e+00 2.93033153e-01
-2.40914077e-01 -7.97797799e-01 -6.35597467e-01 3.23899537e-01
1.08460620e-01 1.15234196e-01 1.10864067e+00 -9.00701761e-01
-5.90696990e-01 2.75225043e-02 -2.98955560e-01 2.05107510e-01
2.48611137e-01 -9.29667056e-02 -1.53949952e+00 1.42251194e-01
-1.10280372e-01 -4.22916412e-01 9.13375080e-01 2.70926893e-01
1.61477494e+00 -2.11280942e-01 -3.06877077e-01 5.75737953e-01
1.12104130e+00 4.58284244e-02 4.96588320e-01 5.73421409e-03
7.83689141e-01 1.20743072e+00 8.53167892e-01 4.51503456e-01
2.56082743e-01 7.83991218e-01 5.10094047e-01 -1.65240616e-01
-1.68685522e-02 -1.94952101e-01 3.56646292e-02 4.64926027e-02
-2.02202529e-01 -1.30604848e-01 -8.33440304e-01 4.87138450e-01
-1.64263058e+00 -9.48725820e-01 8.37297440e-02 2.17994189e+00
6.80756927e-01 -1.57204539e-01 -5.21000288e-02 -1.44658238e-01
7.54588664e-01 8.80424976e-02 -5.74947715e-01 -3.51366103e-01
-1.35550976e-01 3.48279297e-01 2.73941040e-01 2.63528347e-01
-1.34555900e+00 1.30425191e+00 4.83166409e+00 9.21125531e-01
-8.79846931e-01 2.29614615e-01 1.14650822e+00 3.97796370e-02
-2.85796583e-01 -1.10717990e-01 -7.98226953e-01 3.72247398e-01
7.58099318e-01 1.75771728e-01 2.54933894e-01 9.94619668e-01
-9.55558121e-02 -2.14007199e-01 -1.36149311e+00 8.86132717e-01
9.02290046e-02 -8.90935898e-01 1.52319461e-01 1.18931435e-01
2.95863956e-01 -5.12287438e-01 2.47427344e-01 1.50781721e-01
-1.19763687e-02 -1.32259309e+00 6.64539576e-01 4.37506914e-01
8.40823233e-01 -8.98425758e-01 5.96274436e-01 2.58936971e-01
-1.12243688e+00 -2.08544210e-02 -5.39940119e-01 2.08634108e-01
-5.02602085e-02 3.77206504e-01 -7.40271032e-01 1.97014511e-01
6.30852818e-01 5.05350530e-01 -3.59739006e-01 4.64558065e-01
-4.71287608e-01 4.01431322e-01 -2.73553133e-01 1.18642718e-01
3.21473271e-01 -2.73333311e-01 4.90225405e-02 9.47840869e-01
3.25144053e-01 4.57469702e-01 1.79647028e-01 9.71513391e-01
-2.12973773e-01 3.07203412e-01 -6.96814179e-01 -1.02481693e-01
4.03262943e-01 1.25136006e+00 -8.00792575e-01 -1.76464394e-01
-5.97855985e-01 1.20155084e+00 4.13125843e-01 4.21260446e-01
-6.57650650e-01 -1.66107938e-01 9.61108923e-01 2.99811810e-01
4.38679934e-01 7.52758682e-02 -2.76321650e-01 -7.58635044e-01
-7.39698559e-02 -5.11909723e-01 3.59002322e-01 -5.34317613e-01
-1.26465416e+00 6.18480563e-01 -1.82440832e-01 -1.01063430e+00
-4.00194049e-01 -7.75871217e-01 -3.80569637e-01 9.03134584e-01
-1.60557556e+00 -1.50223649e+00 -3.26024711e-01 6.51264846e-01
5.44798076e-01 -9.89553779e-02 9.64856863e-01 2.57617474e-01
-3.66510123e-01 5.56280971e-01 -3.11091334e-01 2.66510308e-01
6.63552761e-01 -1.11679614e+00 1.48334473e-01 5.57446778e-01
4.71886456e-01 6.90427661e-01 5.49448252e-01 -4.91895407e-01
-1.06608343e+00 -9.93619025e-01 9.18552637e-01 -4.01936829e-01
4.71954882e-01 -6.15664482e-01 -1.06630659e+00 5.47811627e-01
-1.70930587e-02 3.52143884e-01 8.27899456e-01 3.67092788e-01
-3.95753175e-01 -1.73564732e-01 -1.15472984e+00 2.71681666e-01
1.07677197e+00 -6.57785475e-01 -2.49006629e-01 2.18917459e-01
4.27446723e-01 1.20448068e-01 -6.34287238e-01 2.98227191e-01
3.34654659e-01 -8.97877932e-01 8.98711562e-01 -7.72307575e-01
5.12803793e-01 -2.42840499e-04 -3.17437708e-01 -1.05915487e+00
-4.72234525e-02 -9.77417231e-02 8.12980980e-02 1.58754587e+00
5.43324649e-01 -4.55322206e-01 1.04602623e+00 9.32297289e-01
1.47856027e-01 -7.05624104e-01 -9.86325145e-01 -6.60725117e-01
-5.95581159e-02 -4.61286664e-01 7.57687211e-01 1.01819932e+00
-2.62665689e-01 4.29150105e-01 -1.57284349e-01 2.33444497e-01
7.38157153e-01 1.64054245e-01 4.71335292e-01 -1.59752297e+00
-1.34085774e-01 -3.05887818e-01 -6.32781506e-01 -9.63988066e-01
7.35771835e-01 -8.61926556e-01 1.41009077e-01 -1.27710581e+00
2.63342321e-01 -7.91115642e-01 -3.27884555e-01 7.92355061e-01
-7.61942118e-02 3.52356613e-01 1.47544578e-01 1.33430198e-01
-5.35967052e-01 6.73306525e-01 1.06063604e+00 -1.29983306e-01
2.15604722e-01 9.57325660e-03 -7.09876597e-01 9.59248722e-01
8.71326745e-01 -4.23447996e-01 -4.83327508e-01 -2.92194217e-01
-5.47895506e-02 -1.69075668e-01 4.60815966e-01 -7.73020148e-01
2.32021600e-01 -2.22796500e-01 5.54994702e-01 -1.91845447e-01
5.58458745e-01 -1.05371439e+00 -3.27006280e-01 -7.43855536e-02
-2.03303814e-01 -3.50520730e-01 3.59771326e-02 5.87360919e-01
-2.65257359e-01 -3.84233952e-01 9.04991150e-01 -1.66270018e-01
-1.21188557e+00 5.75559497e-01 -1.91890135e-01 3.32701840e-02
1.18298113e+00 -4.56271052e-01 -4.31547835e-02 -4.57603872e-01
-1.02779806e+00 1.57566875e-01 4.77412671e-01 2.56631702e-01
6.42307878e-01 -1.22431636e+00 -4.10210758e-01 2.76400626e-01
4.23873514e-01 -8.89807940e-02 2.66547859e-01 7.43242025e-01
-9.75781828e-02 1.94878116e-01 -3.02756906e-01 -7.20354497e-01
-1.17948544e+00 5.67392766e-01 1.89332217e-01 5.00344038e-01
-4.62438226e-01 1.00049353e+00 7.48414874e-01 -2.81268805e-01
2.78984964e-01 2.85271645e-01 -4.18048680e-01 3.82660866e-01
4.23089325e-01 -1.66571975e-01 -4.50374745e-02 -9.29204762e-01
-4.20021892e-01 7.86735415e-01 -2.10790131e-02 -2.94039324e-02
1.36411691e+00 -1.74442738e-01 -1.82030380e-01 3.18772644e-01
1.15026391e+00 -9.48808566e-02 -1.37469172e+00 -2.82161713e-01
3.16058218e-01 -4.34206188e-01 -1.45913780e-01 -5.26404142e-01
-1.32297873e+00 1.17819738e+00 6.38581872e-01 -1.09247953e-01
1.24932694e+00 5.52423358e-01 4.67919409e-01 3.17853063e-01
3.56314063e-01 -1.21800017e+00 -2.51327138e-02 2.57819146e-01
3.80099535e-01 -1.41023242e+00 -2.46223360e-01 -8.39739203e-01
-1.00329232e+00 1.00622261e+00 7.45372176e-01 2.07117677e-01
5.00823498e-01 1.16265789e-01 -2.45781876e-02 -2.84999192e-01
-5.19414723e-01 -5.51200271e-01 2.70368218e-01 7.14059711e-01
5.23260236e-01 4.86504585e-02 2.53269989e-02 7.22504795e-01
-2.12471914e-02 -4.09817159e-01 1.58693939e-01 5.65678477e-01
-5.96718431e-01 -1.23794174e+00 -1.61827296e-01 4.13053930e-01
-5.57330608e-01 -3.51609997e-02 -4.43224072e-01 7.14574337e-01
3.03383291e-01 8.30869675e-01 3.01497161e-01 -3.70016461e-03
1.03160016e-01 3.55857611e-01 3.71403247e-01 -7.45481670e-01
-3.78594920e-02 -4.05107513e-02 2.45216228e-02 -6.82044685e-01
-4.29997772e-01 -6.83964372e-01 -1.44952273e+00 -8.35914686e-02
5.62419854e-02 1.66350782e-01 7.08338499e-01 9.96299624e-01
2.75738806e-01 2.60402024e-01 5.32145500e-01 -5.38812101e-01
-1.24852285e-01 -6.51813745e-01 -7.07398295e-01 5.94313502e-01
1.22191109e-01 -8.59971404e-01 -1.47142410e-01 2.13343352e-01] | [9.949938774108887, 1.9494751691818237] |
6d344b9c-7399-4dba-9260-ffc4b1c03f2d | generalized-local-optimality-for-video | 2112.11729 | null | https://arxiv.org/abs/2112.11729v1 | https://arxiv.org/pdf/2112.11729v1.pdf | Generalized Local Optimality for Video Steganalysis in Motion Vector Domain | The local optimality of motion vectors (MVs) is an intrinsic property in video coding, and any modifications to the MVs will inevitably destroy this optimality, making it a sensitive indicator of steganography in the MV domain. Thus the local optimality is commonly used to design steganalytic features, and the estimation for local optimality has become a top priority in video steganalysis. However, the local optimality in existing works is often estimated inaccurately or using an unreasonable assumption, limiting its capability in steganalysis. In this paper, we propose to estimate the local optimality in a more reasonable and comprehensive fashion, and generalize the concept of local optimality in two aspects. First, the local optimality measured in a rate-distortion sense is jointly determined by MV and predicted motion vector (PMV), and the variability of PMV will affect the estimation for local optimality. Hence we generalize the local optimality from a static estimation to a dynamic one. Second, the PMV is a special case of MV, and can also reflect the embedding traces in MVs. So we generalize the local optimality from the MV domain to the PMV domain. Based on the two generalizations of local optimality, we construct new types of steganalytic features and also propose feature symmetrization rules to reduce feature dimension. Extensive experiments performed on three databases demonstrate the effectiveness of the proposed features, which achieve state-of-the-art in both accuracy and robustness in various conditions, including cover source mismatch, video prediction methods, video codecs, and video resolutions. | ['Yang Liu', 'Yanzhen Ren', 'Lina Wang', 'Liming Zhai'] | 2021-12-22 | null | null | null | null | ['steganalysis', 'video-prediction'] | ['computer-vision', 'computer-vision'] | [ 0.25582975 -0.58184713 -0.32169345 0.10262836 -0.18599166 -0.3707903
0.44545087 -0.3322485 0.02472542 0.27636814 0.21733116 -0.23882647
0.13714644 -0.77912945 -0.63554716 -1.1978058 -0.22086197 -0.51945263
0.48178896 -0.38229838 0.5668113 0.27028176 -1.3176374 0.124423
0.73989016 1.1249127 0.4796169 0.6211393 0.20933662 0.66735363
-0.55379975 -0.21541502 0.46483773 -0.87416786 -0.24031211 0.22460368
-0.2386194 -0.3713858 -0.5958665 1.42634 0.3289569 -0.27981672
0.5064883 -1.287033 -0.46979544 0.44644117 -0.4395238 0.4522056
0.25184712 0.40537563 0.8740777 -0.61153656 0.5272098 1.1227635
0.66646546 0.24179229 -0.61549085 -0.76484096 -0.02548037 0.69896895
-1.6200852 -0.52204967 0.92631584 -0.48193103 0.2711884 0.47613436
0.8180891 1.0051267 0.74687624 0.6550654 1.084854 -0.27735296
-0.07182296 0.08018378 -0.58799124 0.6562176 0.38773495 0.25682497
-0.09017707 -0.11251874 0.94368035 0.07028835 -0.7903567 -0.4799743
-1.422613 0.96568173 -0.09267253 0.45223567 -0.03034739 0.14832355
0.40300688 0.6479771 0.11254228 0.15155292 -0.18691118 -0.22224882
-0.6961162 -0.05886947 0.5693612 0.80988353 0.78509784 0.3752581
-0.12375778 0.48911175 0.38322112 0.7263504 0.70946693 -0.64628756
0.49046728 0.19909804 -0.09701769 -1.9977949 0.00649179 -0.44508082
-1.0280565 -0.02391534 -0.09677372 0.09738338 -0.57072866 1.6139013
0.10336228 0.50371873 0.04240636 0.78741246 0.54457086 0.95792776
-0.45290735 -0.80497813 0.9112992 -0.67488277 -0.83433545 0.05080573
0.8025153 -0.9126175 0.6283686 0.34472728 -0.59011877 -0.5064849
-1.3435667 0.67923266 0.05787747 -0.19929071 0.09187154 0.8577278
-0.8432269 0.7207259 -0.54217446 -0.07035176 -0.0364719 0.1819749
-0.3128849 -0.12226036 -1.5621842 0.6375704 0.69472367 0.05314496
-0.95866686 -0.12831035 -0.7699688 0.1009946 0.4916689 -0.1340165
0.54618144 -0.9035731 -1.3997756 0.2660164 -0.02353383 -0.1815536
0.63916653 0.5140355 -1.0761497 0.21020295 -0.16503736 -0.01634792
1.5481676 -1.0787636 -0.7396242 0.12921202 -0.08850113 -0.01250663
-0.46501112 -0.11609502 -0.6087897 -0.9263116 0.3787572 -0.8310035
0.01565189 -0.28517032 -0.19832912 0.22820544 1.0926117 -0.96892494
1.7805284 -2.2869005 0.1158434 0.51887584 0.27905598 0.3628885
0.03047316 0.38627493 -0.05488425 0.30579233 -0.2600837 0.25992706
-0.24919759 0.11087279 -0.08923896 0.8087117 -0.19884813 0.71260947
-0.866935 -0.73054713 0.35225323 0.29069215 -0.5397128 0.03063625
0.24362364 0.56781524 -0.5975211 0.58728266 0.9442304 -0.19399482
0.27069348 -0.27647096 -0.07225576 -0.19170111 -1.2350713 1.1207983
-0.2426568 0.7435754 -0.20464483 -1.296127 1.0552399 0.4172459
0.6845294 -0.54149586 0.2960943 0.3193837 0.04831002 -0.7768124
0.28346175 0.16224237 0.06583975 0.20728049 -0.2379724 0.38706362
-0.1197086 -0.17456116 1.0673053 -0.21295154 0.7567747 -0.3986007
0.9604524 -0.42341667 0.99549294 0.5831378 -0.20331167 0.6019109
0.61096954 -0.21548639 -1.2077907 -0.46497387 -0.08742795 0.35372946
0.7603528 -0.5567794 -0.52506477 -0.7765459 -0.20968243 0.3641658
-0.36990243 -0.59994644 -0.7778624 -0.5617814 0.5636895 -0.05889928
0.9699602 -0.7165776 -0.43943682 0.2501983 -0.5083498 -1.092332
-0.6522366 -0.5934905 -0.7478312 -1.1658845 -0.85391754 -0.60960513
0.48484045 0.643206 0.5479586 0.37599954 0.20251648 0.03211834
-0.77259064 0.17980433 -0.90622604 -0.2505918 0.09754769 0.5790997
-0.0083502 -0.5385272 -0.73373073 0.704198 -1.0808624 0.04018132
0.7902234 0.8481749 0.36827856 0.53643364 0.09612104 -0.5404221
0.30200562 -0.7329498 -0.52771544 0.23213032 -0.8362465 0.132511
0.774149 -0.46898186 -0.6906786 -0.44195214 -0.17922689 -0.79540217
0.23311529 0.5423796 -0.623388 -0.5122813 0.1662428 0.85335124
0.1411685 -0.31656167 -0.20720544 0.60521716 0.18998398 0.01690847
1.1078718 0.42060456 0.26384118 -0.8651398 0.1668446 -0.27790675
-0.12126253 -0.17464678 0.46648502 -0.746611 -0.5790101 0.6602021
-1.0038942 0.27237046 0.07110798 0.6422741 -0.4849402 1.2059506
-0.2764135 -0.44519863 0.00725346 -1.641114 0.47391263 -0.20718096
0.2595656 -1.1848286 -0.08529071 -0.20821467 0.27613935 0.4942991
0.8355277 -0.43113685 -0.63209313 -0.23041701 -0.09348965 0.6342631
0.32643953 -0.01695492 -0.4977286 -0.67425215 0.4626607 0.4114554
0.8171214 0.3080566 0.91981494 -0.82181364 -0.34582055 1.0240741
1.6175559 0.37981492 0.9110429 0.40336254 0.67512476 0.24220043
0.8478851 0.77541405 -0.02149716 0.8234822 0.7151086 0.4266169
0.039799 -0.16347529 0.7720229 1.2452972 -0.36499974 -0.43647093
-0.5362476 0.44543 -1.8613309 -1.1635745 -0.01889165 2.384403
0.5581579 -0.09404565 -0.24658199 0.37984923 1.1800808 0.8513773
-0.18552403 -0.22346207 -0.31451827 -0.46132764 0.8309858 0.3849917
-1.0104166 0.6021895 5.909527 1.5689191 -1.3374159 0.34439158
0.3219568 0.4299018 -0.5706581 0.27954376 -0.56725544 1.0680075
0.4928898 -0.07028767 0.4372986 0.7384785 0.22952232 0.29262888
-0.5855047 1.3866861 0.22759995 -1.3067466 0.20731391 0.44519004
0.7842199 -0.45080855 0.32044512 -0.1435505 -0.20418274 -0.71858317
0.6446025 0.33127427 1.0967174 -0.7520954 0.9008601 0.34537748
-1.470058 -0.19844612 -0.62095726 0.3016871 0.15236133 0.5053807
-0.46434703 0.68771064 0.4808488 1.0553063 -0.3684988 0.89466864
-0.11753809 0.5310046 -0.131284 0.02167278 0.2473025 -0.21579595
1.112568 1.0219897 0.90847886 0.01915236 -0.03586641 0.5118822
0.23200473 0.29763386 -0.64563376 -0.01942075 0.50539744 0.8088833
-0.40759945 -0.16115804 -0.43824744 1.025987 -0.51657206 0.32850182
-0.9101282 -0.36659163 0.7864319 0.03910013 0.78349304 -0.17834328
0.11623909 -1.5655125 0.10139366 -1.119956 0.11351006 -0.18563761
-0.6946296 0.3398965 -0.05145453 -2.0072744 -0.4513902 -0.33059123
-0.57266635 0.358482 -1.5519036 -0.8489337 -0.44638857 0.78360593
0.48186627 -0.45456785 0.42383012 0.17340723 -0.44489223 0.8121374
0.69647366 0.08572648 0.47236624 -0.33892748 0.23647939 1.2654356
-0.10423382 0.2083544 0.9094393 -0.70894 -1.4382001 -0.9504944
0.53300184 0.20139183 0.49256623 0.03287559 -0.72992355 0.5596742
-0.05630026 0.05803866 0.24783742 -0.720127 -0.34785417 -0.08309226
-1.1729894 0.6579242 1.114161 -0.29823327 -0.16104297 0.11774442
0.6478415 -0.16432843 -0.91106194 0.62550884 0.6877893 -1.3000163
0.8602668 -0.02648014 0.32362115 -0.33709183 -0.4378883 -1.1159648
-0.6397837 -0.91468924 -0.41181126 1.0623354 -0.13216536 -0.8439917
0.5526896 -0.22449706 -0.10587154 -0.58792204 -1.1172196 -1.0598541
-0.31570524 -0.34694996 0.8752404 1.1322857 -0.12232026 -0.5231025
-1.031537 0.26663965 0.59792715 -0.0311245 0.6609094 -0.8232055
-0.48243582 -0.44886705 -1.0434546 -1.4129846 -0.11509906 -0.5737332
-0.05058859 -0.89541614 0.02785675 -0.48598164 -0.40983126 -0.10474647
-0.11758997 0.23080356 0.15576567 0.8670617 -0.27025738 0.5624129
1.3026719 -0.17183246 -0.01789874 0.05121073 -0.36388862 0.71035373
0.76860225 -0.4803775 -0.3443056 -0.23310529 0.16773449 0.07901415
0.14675172 -1.2091125 -0.03138451 -0.4892429 0.0704082 -0.26099536
0.2739308 -1.0264345 0.3612301 0.82537115 0.19562998 -0.14576653
-0.43475 0.818255 -0.33995283 -0.3343156 0.81962246 -0.1899346
-1.21508 0.5338466 -0.33244836 -0.11802044 1.2055191 -0.7543586
-0.08379133 -0.7144762 -0.28125784 -0.27674097 0.79973733 0.3895364
1.0262849 -1.5435606 -0.7566587 0.59530294 0.17841342 -0.7347067
0.31366402 1.1798723 -0.807178 0.40431568 -0.20262192 -0.7402617
-1.2843715 0.8185387 0.09982114 -0.275499 -0.61195606 0.3985132
0.35573652 0.3607603 -0.25173306 0.03106584 -0.5126622 -0.1969914
0.55833614 0.50810146 -0.24850069 -1.0921073 -0.15698947 0.9168077
0.20485643 0.10308963 0.8134337 -0.7481295 -0.05278391 0.21158285
1.708624 0.24662317 -1.0668325 -0.43952107 -0.46710813 -1.0668687
-0.06635526 0.24049866 -1.3155824 0.72353476 0.60503614 0.38295922
1.1291535 -0.27694035 1.0902394 0.0840025 0.55924535 -0.8445113
0.02213627 0.30761874 0.6335317 -1.153537 0.08308521 -0.5755562
-0.55613804 1.1609874 0.11339413 -0.09307403 0.7835569 -0.2612387
-0.17970827 0.20898372 -0.33457232 0.169823 0.22424385 0.5806953
-0.18757306 -0.07431413 -0.6344867 0.22225751 0.04397124 -0.37848502
0.5613188 0.81473875 -0.70145345 -0.9668148 -0.60719854 0.25364318
-0.5761966 -0.2227029 0.25966164 0.7195625 0.35362715 1.0032862
-0.12530768 -1.108502 -0.2592297 -0.41715243 0.09592534 -0.03713511
0.12070289 0.1296725 -0.15479226 -0.65465194 -0.4168953 -0.5958395
-0.8526774 -0.86485195 -0.57674605 0.20075116 0.4659001 1.1039563
0.2855299 0.21367095 1.3383741 -0.42141753 -0.68975735 -0.83740914
-0.869889 0.49376038 0.47087908 -0.5577986 -0.79896396 0.07256956] | [4.297179698944092, 8.056249618530273] |
1ae9029c-d079-4ef4-9fb8-f96b518512b2 | neural-object-descriptors-for-multi-view | 2004.04485 | null | https://arxiv.org/abs/2004.04485v2 | https://arxiv.org/pdf/2004.04485v2.pdf | NodeSLAM: Neural Object Descriptors for Multi-View Shape Reconstruction | The choice of scene representation is crucial in both the shape inference algorithms it requires and the smart applications it enables. We present efficient and optimisable multi-class learned object descriptors together with a novel probabilistic and differential rendering engine, for principled full object shape inference from one or more RGB-D images. Our framework allows for accurate and robust 3D object reconstruction which enables multiple applications including robot grasping and placing, augmented reality, and the first object-level SLAM system capable of optimising object poses and shapes jointly with camera trajectory. | ['Kentaro Wada', 'Edgar Sucar', 'Andrew Davison'] | 2020-04-09 | null | null | null | null | ['3d-object-reconstruction'] | ['computer-vision'] | [ 1.98216468e-01 -1.40023381e-01 1.55569851e-01 -3.51785213e-01
-7.00359643e-01 -9.28847313e-01 7.74649501e-01 1.69897050e-01
-2.14450821e-01 2.20658287e-01 -4.68850106e-01 -1.56016961e-01
-5.31540275e-01 -6.95187211e-01 -8.21742654e-01 -5.92354774e-01
1.54590011e-01 1.37336946e+00 3.79998505e-01 1.35802314e-01
4.79386061e-01 1.36501169e+00 -1.96347308e+00 -2.63404518e-01
2.23087028e-01 1.30664718e+00 6.12039030e-01 8.97441208e-01
2.69616861e-02 1.47547588e-01 7.49309063e-02 -1.54323548e-01
5.33012927e-01 2.59040684e-01 -3.42552960e-01 2.32625425e-01
4.52610135e-01 -4.94397342e-01 -1.29065573e-01 7.64561713e-01
4.62333828e-01 2.94193804e-01 7.85049319e-01 -1.22500062e+00
-4.25045550e-01 -2.91138500e-01 -2.89272934e-01 -4.87840354e-01
8.64630878e-01 3.13925773e-01 6.21167183e-01 -9.52010155e-01
6.96921527e-01 1.45924032e+00 5.14803708e-01 4.40518558e-01
-1.14560974e+00 -1.18394621e-01 -2.96755344e-01 -3.48427966e-02
-1.50694811e+00 -4.11444277e-01 8.43685627e-01 -4.82302189e-01
9.12458777e-01 3.51473182e-01 6.75523937e-01 7.92361259e-01
2.96677887e-01 6.56533778e-01 9.38212037e-01 -2.45108470e-01
4.95271802e-01 1.75887153e-01 -5.97820520e-01 6.58764243e-01
1.62736878e-01 1.31675228e-01 -4.13532674e-01 -4.63761181e-01
1.28631890e+00 1.57317385e-01 7.93261230e-02 -1.40570045e+00
-1.42624474e+00 4.92884308e-01 2.86782652e-01 -3.37326825e-01
-4.97564405e-01 5.23301244e-01 -3.23362299e-03 -2.90277779e-01
-2.93241944e-02 3.82085383e-01 -6.20561779e-01 -4.06895965e-01
-2.90643841e-01 4.20770824e-01 8.01422775e-01 1.66228688e+00
7.29311466e-01 -1.53717816e-01 3.66258472e-01 3.48602921e-01
8.64600658e-01 1.07710135e+00 -2.41306037e-01 -1.51194382e+00
-9.69415307e-02 4.90682840e-01 4.04234082e-01 -9.12357986e-01
-3.74265015e-01 2.62770772e-01 -3.06614965e-01 5.64819038e-01
-6.24264181e-02 5.55293858e-01 -7.92726338e-01 1.00830150e+00
8.12409282e-01 -4.37940620e-02 -3.16332467e-02 1.00278723e+00
8.09025168e-01 1.95575640e-01 -2.05834523e-01 1.89591974e-01
1.28908682e+00 -1.76591441e-01 -2.80420721e-01 -1.24525264e-01
2.07168058e-01 -8.32488477e-01 7.19505787e-01 7.07064033e-01
-1.15674341e+00 -1.20762110e-01 -8.45895767e-01 -4.78794873e-01
-1.83889702e-01 -5.79763502e-02 9.52943146e-01 4.20955241e-01
-8.46238494e-01 4.77452457e-01 -1.09024704e+00 -2.13616461e-01
5.40186524e-01 6.49829209e-01 -5.41001916e-01 -3.00073504e-01
-1.29861876e-01 1.31942856e+00 6.48384571e-01 1.47814184e-01
-9.33218181e-01 -6.51420653e-01 -9.51341987e-01 -4.90592122e-01
2.46310204e-01 -7.61665106e-01 1.18586230e+00 -4.95805293e-02
-1.89039397e+00 1.25460351e+00 4.18377444e-02 4.08248454e-02
3.65169555e-01 -3.17889512e-01 3.14889580e-01 2.67120302e-01
-3.35726112e-01 6.56385481e-01 7.51882315e-01 -1.65736544e+00
-2.38847136e-01 -1.03427947e+00 4.43739556e-02 5.64154863e-01
6.68554723e-01 -1.81058481e-01 -3.08651775e-01 2.21737064e-02
7.65138388e-01 -8.67924690e-01 -4.31621373e-01 7.43519545e-01
-1.55560270e-01 -1.00133136e-01 1.32763445e+00 -3.64991009e-01
-3.18747908e-01 -1.93561256e+00 5.22845924e-01 1.93613872e-01
-7.93561563e-02 -2.32919455e-01 9.16739479e-02 1.34199247e-01
6.71837628e-01 -4.25008267e-01 -7.69398436e-02 -5.54369092e-01
4.34365809e-01 6.33152127e-01 -4.69610512e-01 9.73010540e-01
8.28838497e-02 8.67664576e-01 -9.53479528e-01 -4.07273889e-01
9.96815860e-01 9.09040332e-01 -4.25641865e-01 5.46892405e-01
-5.66269815e-01 9.27898049e-01 -6.24315619e-01 9.87703800e-01
9.38355565e-01 2.09612027e-01 -1.69296429e-01 -3.68596882e-01
-1.08734414e-01 1.63438022e-01 -1.49674976e+00 2.20763564e+00
-5.20651639e-01 2.44692355e-01 4.54903632e-01 -3.05751204e-01
1.14638352e+00 1.58041388e-01 7.33449161e-01 -3.03422421e-01
2.58443207e-01 1.83932543e-01 -6.80163801e-01 -3.77148271e-01
8.09169233e-01 -2.11262479e-01 8.06668482e-04 4.20431823e-01
8.84651989e-02 -1.43740857e+00 -6.25750601e-01 2.84443540e-03
8.08786929e-01 9.75581586e-01 3.81988108e-01 -2.81891137e-01
1.60658762e-01 1.39239654e-01 2.06647888e-01 3.83271247e-01
1.15414537e-01 6.64304674e-01 -3.41503263e-01 -3.49010348e-01
-1.19107199e+00 -1.52423930e+00 -6.69791520e-01 6.55670524e-01
7.46287942e-01 2.13721633e-01 6.36733919e-02 9.36497934e-03
6.64552033e-01 8.93198252e-01 -1.10947877e-01 -1.03718489e-02
-4.88102198e-01 -3.98554541e-02 -2.03253943e-02 6.44797325e-01
-1.78010702e-01 -9.59264934e-01 -1.12414050e+00 2.21855298e-01
3.25271964e-01 -1.20726085e+00 9.20811594e-02 3.12067121e-01
-1.01763010e+00 -1.09714448e+00 -7.20540807e-02 -2.98851073e-01
6.89901233e-01 3.19271833e-01 8.86513531e-01 -3.70498806e-01
-6.38802350e-01 1.26275492e+00 -2.48257101e-01 -5.33712983e-01
-3.37920368e-01 -4.87333119e-01 2.90784270e-01 -3.83548290e-01
1.99206308e-01 -8.41195226e-01 -3.52093607e-01 3.68609637e-01
-5.89245200e-01 -3.64963859e-02 4.28053141e-01 5.67089655e-02
1.03868794e+00 -2.53312320e-01 -2.41566330e-01 -4.08769436e-02
3.50306705e-02 -1.34610191e-01 -1.19480002e+00 2.28597835e-01
-2.21829802e-01 6.37226775e-02 1.87164396e-02 -4.37831283e-01
-8.60619783e-01 6.76763892e-01 -1.06849059e-01 -6.91564739e-01
-5.13650596e-01 -2.58198529e-01 -4.81369853e-01 -5.89720547e-01
4.20039326e-01 1.68403938e-01 6.76466152e-02 -5.62258780e-01
7.57425010e-01 4.81648922e-01 8.34533811e-01 -9.05811608e-01
7.27094889e-01 8.29109490e-01 5.89376628e-01 -8.38956177e-01
-2.78476030e-01 -5.94759881e-01 -1.34876382e+00 -3.18104029e-01
8.29778016e-01 -6.87372386e-01 -1.23444307e+00 3.56360555e-01
-1.52504563e+00 -2.17898414e-01 -5.75281620e-01 5.77257216e-01
-1.35400522e+00 2.02240258e-01 2.30002925e-02 -1.33103347e+00
8.44520610e-03 -1.41394317e+00 1.78669190e+00 -6.32751659e-02
2.44134124e-02 -8.42770159e-01 -5.23736514e-02 1.43010512e-01
1.76562950e-01 5.70282698e-01 4.75051939e-01 -7.61223286e-02
-1.23984170e+00 -3.77590507e-01 -2.14373603e-01 -2.22644299e-01
7.32693970e-02 1.87501907e-01 -9.70330536e-01 -2.31178954e-01
3.10696550e-02 -5.58318257e-01 2.02317998e-01 3.10373664e-01
1.00943053e+00 2.48103775e-02 -4.54173267e-01 8.15851450e-01
1.61534524e+00 -2.69039702e-02 3.25384408e-01 5.73775657e-02
8.42004538e-01 4.26459581e-01 8.45348954e-01 5.56069493e-01
5.88668704e-01 1.01583874e+00 1.14446950e+00 6.25892401e-01
2.20325455e-01 -7.03755170e-02 -1.87600534e-02 5.81008673e-01
-8.96325782e-02 1.23762392e-01 -9.03155565e-01 2.44986624e-01
-1.68247390e+00 -4.78342205e-01 -2.64456630e-01 2.43111753e+00
5.09517550e-01 -3.25999975e-01 -2.38734782e-01 -1.65815547e-01
2.74798721e-01 -2.11561739e-01 -8.59705627e-01 -2.93943226e-01
4.78739999e-02 1.47657409e-01 6.35902166e-01 6.18029118e-01
-8.38098943e-01 6.92183912e-01 6.62436390e+00 4.35976446e-01
-6.81302905e-01 7.61454254e-02 -3.97159994e-01 6.29447252e-02
-5.90006769e-01 2.14684516e-01 -8.97740245e-01 -1.83006376e-01
5.69040716e-01 1.66000694e-01 4.54239875e-01 1.07147050e+00
-2.66401976e-01 -4.64712501e-01 -1.26595938e+00 1.24543452e+00
9.09897164e-02 -1.15812790e+00 -1.66490391e-01 1.30944610e-01
3.11249703e-01 2.92410970e-01 -2.89909333e-01 -2.63926834e-01
5.16631126e-01 -6.39424503e-01 1.22968757e+00 7.91432977e-01
6.34988070e-01 -4.98242915e-01 6.07493594e-02 7.05837846e-01
-9.75495100e-01 2.03177631e-01 -4.29796964e-01 2.26547107e-01
5.03800571e-01 3.86779636e-01 -9.05166566e-01 5.00770688e-01
5.20082593e-01 2.99626410e-01 -6.67315200e-02 1.13344574e+00
-1.08584292e-01 -2.92753160e-01 -9.32109654e-01 3.53541225e-02
-2.59520084e-01 -3.56846511e-01 9.93518353e-01 7.01476216e-01
3.68331075e-01 4.45949733e-01 3.11111033e-01 1.00902832e+00
2.73697108e-01 -3.74205172e-01 -7.04332352e-01 2.87462473e-01
5.94831526e-01 1.41458929e+00 -8.66674423e-01 1.45307377e-01
1.08296253e-01 8.48890901e-01 1.15640238e-01 -9.66887027e-02
-5.16854644e-01 6.89729825e-02 6.78107917e-01 -3.65850367e-02
3.77166331e-01 -1.07498300e+00 -3.54327738e-01 -1.00866151e+00
1.09688587e-01 -1.98812947e-01 -1.24862164e-01 -1.19188750e+00
-9.66181874e-01 2.12251469e-01 3.77482325e-01 -1.06378603e+00
-2.86231458e-01 -1.19804120e+00 2.70737521e-02 6.58120394e-01
-1.38944829e+00 -1.59151709e+00 -2.77983040e-01 4.48208541e-01
2.23518610e-01 3.30290496e-01 9.96974409e-01 -2.05591008e-01
2.00235695e-01 -2.88609177e-01 -5.07085249e-02 -4.91205901e-01
2.62661815e-01 -1.25309515e+00 1.14575870e-01 3.76673728e-01
1.99362442e-01 3.39757949e-01 8.11594963e-01 -5.76733947e-01
-2.63391066e+00 -7.13852048e-01 -6.94958195e-02 -1.08229947e+00
3.77655566e-01 -4.65128124e-01 -5.99485219e-01 7.28965044e-01
-5.40470839e-01 3.92140269e-01 1.24742709e-01 -2.87074596e-01
-2.22811341e-01 5.91693260e-02 -1.63536251e+00 9.15311873e-02
1.00509918e+00 -6.69901550e-01 -5.64118505e-01 5.61195076e-01
6.40579700e-01 -1.24053860e+00 -1.14491248e+00 6.70391440e-01
8.16614151e-01 -8.13538671e-01 1.29319215e+00 -3.28974515e-01
-1.72173142e-01 -3.98854524e-01 -8.42709422e-01 -8.32692146e-01
-1.37224704e-01 -5.71158469e-01 -6.61938965e-01 9.26593244e-01
-2.75168240e-01 -4.78274822e-01 7.37389803e-01 1.13943100e+00
-2.65217870e-01 -5.63443601e-01 -1.22548258e+00 -7.94088364e-01
-4.04768288e-01 -9.63576555e-01 7.43445694e-01 4.32122737e-01
-6.98729098e-01 -4.67583537e-01 1.78063530e-02 7.02308953e-01
1.23371661e+00 5.90710640e-01 1.06008291e+00 -1.40115297e+00
-1.32660732e-01 -3.50068390e-01 -8.59187543e-01 -1.25013280e+00
8.23270455e-02 -7.39305854e-01 5.15656233e-01 -1.50373793e+00
-7.28358924e-02 -9.84158576e-01 3.43746841e-01 2.57528037e-01
4.80871022e-01 1.50298461e-01 3.05544864e-02 1.99925587e-01
-6.31356061e-01 7.03508019e-01 1.22479475e+00 2.17033118e-01
-5.72817139e-02 5.21857589e-02 -1.34618701e-02 8.14569831e-01
2.73395717e-01 -3.78730208e-01 7.03258626e-03 -7.38070607e-01
2.59921789e-01 2.38091778e-02 1.02154219e+00 -7.24694073e-01
3.40886742e-01 -5.26241004e-01 3.47977847e-01 -1.05622208e+00
1.07457101e+00 -1.27143610e+00 4.17035341e-01 1.97719976e-01
-1.27029745e-03 -2.70836093e-02 2.52418131e-01 4.91594166e-01
4.48653191e-01 -2.45254233e-01 7.02894747e-01 -3.07405829e-01
-8.58016431e-01 6.13409042e-01 2.02653661e-01 -3.75956327e-01
1.24431002e+00 -5.76161504e-01 1.78432852e-01 -1.61230624e-01
-5.29712677e-01 -3.05500161e-02 1.03157902e+00 5.00371277e-01
1.08932269e+00 -1.58760643e+00 -5.08987784e-01 4.15575981e-01
9.30503905e-02 7.77092040e-01 -4.57522944e-02 6.54711187e-01
-7.56241798e-01 2.71146834e-01 -1.14271514e-01 -1.25968301e+00
-9.06060517e-01 3.97237271e-01 1.80040240e-01 5.90257704e-01
-7.31859386e-01 7.73219764e-01 -2.91770905e-01 -1.12934220e+00
5.61433844e-02 -3.34354222e-01 5.57786167e-01 -6.25184357e-01
2.72295117e-01 5.22390783e-01 1.81339875e-01 -1.12430322e+00
-5.86997271e-01 8.13605905e-01 5.77295721e-01 -1.12069838e-01
1.48160279e+00 -2.99962819e-01 -4.19145852e-01 7.39254475e-01
9.95719016e-01 -3.02485406e-01 -1.71016908e+00 -1.60692453e-01
-1.42730385e-01 -7.78349578e-01 3.64423037e-01 -4.62750494e-01
-4.15487915e-01 9.07176435e-01 4.88735974e-01 -1.26439437e-01
5.86592674e-01 4.44061339e-01 4.35174853e-01 6.01195157e-01
1.48515332e+00 -6.54860139e-01 -2.39485398e-01 5.42370617e-01
1.16735530e+00 -1.21993995e+00 5.63681841e-01 -3.20523232e-01
-2.75693536e-01 1.20487237e+00 3.18682998e-01 -2.61456698e-01
6.64963424e-01 6.98819280e-01 -2.74503857e-01 -5.15970826e-01
-2.27730572e-01 -9.57171340e-03 6.13172054e-01 1.06244528e+00
-3.02323371e-01 1.71411812e-01 6.89173281e-01 -1.98385328e-01
1.68359112e-02 -2.88001329e-01 -8.46867710e-02 1.11866224e+00
-5.79278409e-01 -9.66845155e-01 -5.87253630e-01 2.28657201e-01
1.68677300e-01 5.14315605e-01 -1.44060701e-01 6.72267199e-01
-3.85757834e-02 3.56998414e-01 2.21647933e-01 -2.42642984e-02
4.39028889e-01 -1.97345614e-01 1.47569966e+00 -6.38898671e-01
3.15401666e-02 -1.17953293e-01 -2.50133276e-01 -1.06763840e+00
-5.75141191e-01 -9.70489502e-01 -1.53588367e+00 5.60877994e-02
-6.05232239e-01 -3.51920515e-01 1.38856292e+00 9.77468908e-01
4.45075989e-01 -1.93918124e-01 4.92075622e-01 -1.83653128e+00
-6.35747075e-01 -5.86237311e-01 -6.32763386e-01 1.23571627e-01
2.99716353e-01 -1.04358983e+00 -1.41392037e-01 -1.42396882e-01] | [7.407236576080322, -2.528921604156494] |
48136f20-12cd-4406-8efc-2c1cfcb489bb | bifsmn-binary-neural-network-for-keyword | 2202.06483 | null | https://arxiv.org/abs/2202.06483v5 | https://arxiv.org/pdf/2202.06483v5.pdf | BiFSMN: Binary Neural Network for Keyword Spotting | The deep neural networks, such as the Deep-FSMN, have been widely studied for keyword spotting (KWS) applications. However, computational resources for these networks are significantly constrained since they usually run on-call on edge devices. In this paper, we present BiFSMN, an accurate and extreme-efficient binary neural network for KWS. We first construct a High-frequency Enhancement Distillation scheme for the binarization-aware training, which emphasizes the high-frequency information from the full-precision network's representation that is more crucial for the optimization of the binarized network. Then, to allow the instant and adaptive accuracy-efficiency trade-offs at runtime, we also propose a Thinnable Binarization Architecture to further liberate the acceleration potential of the binarized network from the topology perspective. Moreover, we implement a Fast Bitwise Computation Kernel for BiFSMN on ARMv8 devices which fully utilizes registers and increases instruction throughput to push the limit of deployment efficiency. Extensive experiments show that BiFSMN outperforms existing binarization methods by convincing margins on various datasets and is even comparable with the full-precision counterpart (e.g., less than 3% drop on Speech Commands V1-12). We highlight that benefiting from the thinnable architecture and the optimized 1-bit implementation, BiFSMN can achieve an impressive 22.3x speedup and 15.5x storage-saving on real-world edge hardware. Our code is released at https://github.com/htqin/BiFSMN. | ['Xianglong Liu', 'Jie Luo', 'Zejun Ma', 'Yao Tian', 'Yang Zhang', 'Xiaoyang Li', 'Yifu Ding', 'Xudong Ma', 'Haotong Qin'] | 2022-02-14 | null | null | null | null | ['keyword-spotting'] | ['speech'] | [-8.33949074e-02 -3.46896142e-01 -5.90459108e-01 -3.79049957e-01
-4.58656549e-01 -2.08294153e-01 6.63991868e-02 -7.53371269e-02
-6.13428950e-01 4.03819621e-01 -1.05545372e-01 -1.08529222e+00
-9.41923708e-02 -9.27378356e-01 -9.88920808e-01 -6.86543643e-01
1.00252740e-01 -7.39889964e-02 2.19489634e-01 -2.69831061e-01
-9.55953076e-02 4.84911889e-01 -1.46051943e+00 1.93444282e-01
4.74074006e-01 1.49388218e+00 2.94159621e-01 5.66894174e-01
-5.68884164e-02 5.69258034e-01 -6.58735096e-01 -4.74361360e-01
3.96404505e-01 2.70307720e-01 -3.10084820e-01 -7.46557176e-01
4.66865629e-01 -8.60258877e-01 -8.89611244e-01 1.13968730e+00
7.79403269e-01 -1.06940657e-01 1.08893149e-01 -1.15184319e+00
-2.79675484e-01 1.18915629e+00 -4.57969993e-01 4.29633439e-01
-5.35594404e-01 1.58422917e-01 9.98379171e-01 -7.91907251e-01
4.73098420e-02 1.02889454e+00 7.37106204e-01 3.12411219e-01
-7.02301979e-01 -1.16027176e+00 -9.53191146e-02 6.76149487e-01
-1.72940576e+00 -9.07721698e-01 6.18031323e-01 2.46872365e-01
1.33043015e+00 4.53673571e-01 6.27172589e-01 8.60618114e-01
-8.54563192e-02 9.41800177e-01 5.70907772e-01 -2.65630543e-01
2.29746833e-01 -2.67662674e-01 5.64388990e-01 6.17355049e-01
6.28180683e-01 4.37686779e-02 -8.21561456e-01 8.97048712e-02
6.30774856e-01 1.44063637e-01 -5.44200361e-01 1.61004007e-01
-7.63175428e-01 5.37355900e-01 5.36044002e-01 1.98132455e-01
-2.45223865e-01 8.21790159e-01 8.09693635e-01 1.68721989e-01
4.50638905e-02 -1.22583918e-01 -6.95096970e-01 -5.74379027e-01
-1.21072412e+00 -1.44167587e-01 6.39979005e-01 1.18145597e+00
5.66267073e-01 4.92788017e-01 -2.25329086e-01 7.61742294e-01
1.04724906e-01 6.48593485e-01 6.96845174e-01 -4.63784784e-01
6.54287040e-01 4.51022685e-01 -5.49870729e-01 -9.51607704e-01
-3.80571336e-01 -5.67153394e-01 -1.27997184e+00 -1.68584630e-01
1.87618732e-01 -7.42598772e-02 -9.82846320e-01 1.54432845e+00
2.22972631e-01 2.09076792e-01 -5.59832491e-02 8.88762414e-01
1.03013062e+00 8.46477091e-01 -2.58969277e-01 1.78706467e-01
1.53152084e+00 -1.22198379e+00 -7.43561804e-01 -1.58972502e-01
7.61094332e-01 -5.57786703e-01 1.20465338e+00 4.06096458e-01
-1.08788407e+00 -5.06045580e-01 -1.62374616e+00 -2.44086459e-01
-3.84847432e-01 5.19051552e-01 7.80139506e-01 1.00407088e+00
-1.11641121e+00 7.01326072e-01 -1.11254108e+00 2.92165935e-01
4.66785491e-01 7.78419316e-01 1.02170398e-02 2.43335113e-01
-1.34912586e+00 3.55347037e-01 7.04295099e-01 1.84326261e-01
-5.65698206e-01 -8.17816079e-01 -6.66194499e-01 6.61265731e-01
4.91237819e-01 -2.61485696e-01 1.33600831e+00 -6.34296596e-01
-1.74968767e+00 1.54157564e-01 8.34230632e-02 -8.15961659e-01
4.52315137e-02 -4.21505943e-02 -4.41112846e-01 1.30840495e-01
-6.18965626e-01 5.53471208e-01 7.30999768e-01 -3.60008925e-01
-5.81079602e-01 -1.22950293e-01 2.03217342e-02 1.41777406e-02
-1.08674729e+00 -1.71336532e-01 -8.38837206e-01 -9.10087883e-01
-6.90613613e-02 -7.29220688e-01 6.71521500e-02 -9.44957063e-02
-5.74486613e-01 -1.05680466e-01 8.29373181e-01 -6.61415517e-01
1.70640206e+00 -2.30795622e+00 -4.70047176e-01 2.17149809e-01
3.99877846e-01 8.61665785e-01 1.10377699e-01 -1.03503272e-01
-8.03985540e-03 -6.03788439e-03 2.45908335e-01 -3.08014840e-01
2.03430504e-01 1.04976697e-02 -2.31478602e-01 4.37462986e-01
-1.86852515e-01 9.76931155e-01 -3.27932566e-01 -1.84277698e-01
-7.02577457e-03 5.79455316e-01 -5.67250669e-01 -1.68067873e-01
6.85040355e-02 -5.01999617e-01 -1.04411140e-01 8.09871435e-01
9.14867043e-01 -3.76228899e-01 3.04560214e-01 -7.19246924e-01
-1.15493588e-01 7.33664870e-01 -1.03338420e+00 1.51992440e+00
-5.71272016e-01 8.58800411e-01 1.56882554e-01 -9.63306785e-01
7.20610738e-01 1.22824624e-01 1.66341607e-02 -1.03335094e+00
5.74822307e-01 5.28354228e-01 -2.19642408e-02 1.83493465e-01
8.82802129e-01 5.54922104e-01 -7.54566398e-03 3.00742686e-01
6.51664361e-02 4.83939230e-01 6.27709404e-02 1.48546308e-01
9.68648672e-01 -4.99662757e-01 4.80518416e-02 -3.90165389e-01
3.03982437e-01 -3.99694085e-01 4.88367856e-01 7.31484234e-01
-1.28037319e-01 6.44326732e-02 3.51723671e-01 -3.69145602e-01
-1.01329589e+00 -6.46032810e-01 -3.11802328e-01 1.11592495e+00
1.69795021e-01 -7.48800755e-01 -9.11201954e-01 -3.22970420e-01
-1.11076735e-01 5.31583905e-01 -2.44334359e-02 -4.75577563e-01
-6.44158244e-01 -8.42041194e-01 1.09386373e+00 7.96333671e-01
8.38756025e-01 -4.94782537e-01 -9.67834651e-01 -5.71595728e-02
2.43300915e-01 -1.11041737e+00 -6.56017184e-01 6.42400682e-01
-7.89680004e-01 -5.09652436e-01 -6.41825259e-01 -6.46181822e-01
2.39471346e-01 3.57126206e-01 8.90837014e-01 3.07697654e-01
-1.80648610e-01 -4.14958447e-01 -2.71277308e-01 -1.78272337e-01
-3.27704810e-02 4.52263206e-01 2.97802061e-01 -3.42750043e-01
4.93375897e-01 -7.23980546e-01 -7.90677309e-01 1.34674028e-01
-8.63618851e-01 3.64659429e-01 7.90700018e-01 1.03776443e+00
4.94964331e-01 2.38094702e-01 3.20666224e-01 -3.98788720e-01
3.41989487e-01 -1.75325722e-01 -8.71241450e-01 1.28451034e-01
-9.25240397e-01 1.62038222e-01 7.29691982e-01 -7.34310806e-01
-5.38998246e-01 -1.48546860e-01 -4.80421782e-01 -6.03607833e-01
4.15227830e-01 3.35240394e-01 -3.42090219e-01 -3.84753913e-01
4.23848391e-01 2.67398655e-01 -1.21509761e-01 -4.36403573e-01
2.94753343e-01 1.09901500e+00 5.77025533e-01 -4.66728836e-01
4.69536215e-01 1.40971169e-01 -1.42973453e-01 -7.26161718e-01
-2.14280114e-01 -1.12180464e-01 1.83487505e-01 1.25047520e-01
2.04609454e-01 -1.15252888e+00 -1.30727541e+00 7.00360596e-01
-1.14401114e+00 -6.62883818e-01 -3.09774000e-02 2.88642049e-01
8.22691694e-02 3.82903337e-01 -9.06294286e-01 -6.00684881e-01
-1.09453177e+00 -1.50016272e+00 8.74341071e-01 4.53323752e-01
1.52678534e-01 -4.88212675e-01 -7.66642749e-01 2.19943430e-02
7.26972401e-01 -5.05625129e-01 7.61642992e-01 -6.94361210e-01
-7.13632286e-01 -4.12656367e-02 -8.08616281e-01 4.31562811e-01
-2.31283784e-01 -1.54938728e-01 -9.99820650e-01 -3.98141980e-01
4.76821559e-03 -2.63958603e-01 9.73801434e-01 3.33024710e-01
1.73376143e+00 -4.17527229e-01 -2.93775916e-01 1.21409607e+00
1.35986757e+00 3.56187105e-01 5.07873833e-01 2.87703186e-01
7.42244840e-01 -2.31179461e-01 2.55897909e-01 5.20296633e-01
2.73668617e-01 8.44558060e-01 5.19796789e-01 -1.01904221e-01
-3.31958950e-01 -5.68872280e-02 2.57979006e-01 1.15240371e+00
3.84360522e-01 -4.33019340e-01 -8.31776023e-01 3.57857466e-01
-1.45522010e+00 -3.93926650e-01 8.01126566e-03 2.14511776e+00
1.25493014e+00 6.28734469e-01 -2.14733332e-01 5.61899364e-01
6.72403812e-01 2.97979742e-01 -7.74660170e-01 -5.47319412e-01
-3.08149476e-02 6.24523103e-01 1.22559690e+00 3.66642803e-01
-9.08268094e-01 9.27025855e-01 5.04809427e+00 1.82545567e+00
-1.48215771e+00 3.47980112e-01 9.53183591e-01 -4.04695958e-01
-9.00328159e-02 -3.42548937e-01 -1.41106081e+00 7.27797151e-01
1.38858616e+00 -2.66406890e-02 4.92123634e-01 1.08101082e+00
-1.21560626e-01 7.28206262e-02 -9.57627773e-01 1.32973957e+00
-2.42631018e-01 -1.58200371e+00 -1.78937018e-01 7.04244077e-02
3.10398281e-01 1.05239540e-01 1.83509991e-01 3.19647670e-01
-7.15909153e-02 -9.20819402e-01 9.78942871e-01 -9.61004272e-02
1.39583862e+00 -1.18548286e+00 6.98826611e-01 1.65768400e-01
-1.20612371e+00 -5.25734276e-02 -4.98954058e-01 -4.67837416e-02
8.06474537e-02 1.05855381e+00 -7.48326123e-01 2.29084492e-01
1.06161749e+00 1.18173130e-01 -3.29720497e-01 7.51997650e-01
-1.15207639e-02 8.20440233e-01 -7.77701080e-01 -4.67697620e-01
2.06790447e-01 1.77579507e-01 1.12497211e-01 1.42503858e+00
4.10748810e-01 1.72092598e-02 -4.06669229e-01 5.03553510e-01
-4.49287146e-01 -5.99961504e-02 -7.65481144e-02 -1.56087786e-01
9.81982946e-01 1.39718425e+00 -6.98713243e-01 -5.16540170e-01
-2.47186095e-01 8.69576335e-01 2.74275869e-01 -8.43939930e-03
-1.21269190e+00 -9.55460608e-01 9.00224030e-01 -1.54400960e-01
5.22606730e-01 -2.45128185e-01 -5.55125177e-01 -9.58282173e-01
1.69891104e-01 -1.03777874e+00 -8.37118030e-02 -4.76479888e-01
-4.18474883e-01 8.26030672e-01 -2.54956841e-01 -6.06249094e-01
9.53217521e-02 -8.88116717e-01 -2.53453135e-01 8.54516685e-01
-1.55585039e+00 -7.61014283e-01 -3.52225810e-01 4.40800190e-01
3.71080160e-01 -1.83083549e-01 4.80362684e-01 9.51114714e-01
-9.48949158e-01 1.55766678e+00 2.82161295e-01 1.56534150e-01
4.13976401e-01 -6.65147364e-01 9.31714356e-01 9.82185185e-01
3.38107646e-02 7.68050730e-01 3.65177900e-01 -4.64782953e-01
-1.82042193e+00 -1.00058722e+00 4.83315825e-01 4.58017558e-01
7.42025256e-01 -5.96288025e-01 -7.58284569e-01 2.87292063e-01
5.42059019e-02 5.18561304e-02 5.18336236e-01 -1.56640485e-01
-4.67376769e-01 -5.41062117e-01 -8.01869154e-01 8.30803752e-01
1.11332905e+00 -4.72223163e-01 2.94860989e-01 2.83965349e-01
1.17069495e+00 -9.05703127e-01 -5.66871345e-01 5.97763181e-01
6.18789852e-01 -8.84901345e-01 1.06550539e+00 -1.58829510e-01
2.56731510e-01 2.09024623e-02 -4.86783922e-01 -7.75419235e-01
-3.98412235e-02 -6.88521445e-01 -7.18996763e-01 1.07098567e+00
2.26322368e-01 -6.65721536e-01 1.12396312e+00 1.58112943e-01
-3.77482414e-01 -1.17946064e+00 -1.13928175e+00 -8.53297830e-01
-1.86936975e-01 -7.35642672e-01 9.76639152e-01 5.81661105e-01
-1.54669806e-01 1.11055471e-01 -3.82929653e-01 1.09474167e-01
3.01486582e-01 -1.91449538e-01 5.32478809e-01 -5.80233753e-01
-5.98303437e-01 -7.10584164e-01 -3.77063930e-01 -1.72351372e+00
-2.47495323e-02 -7.32497215e-01 6.69441074e-02 -7.41278350e-01
-5.58308847e-02 -6.36558175e-01 -2.67546147e-01 7.95245588e-01
4.14274633e-02 4.20016736e-01 1.79732874e-01 -2.82360557e-02
-3.83755505e-01 4.60503429e-01 7.19846189e-01 -1.79449409e-01
-3.08215264e-02 -2.35610440e-01 -6.92842364e-01 6.11003220e-01
9.42632437e-01 -3.25121641e-01 -3.83897871e-01 -7.36838043e-01
2.57427692e-01 -1.42000660e-01 1.37962088e-01 -1.17577291e+00
5.41191638e-01 1.36692241e-01 6.49905205e-02 -6.64462149e-01
4.87949848e-01 -7.16182172e-01 7.48293772e-02 6.86661720e-01
8.64056870e-02 2.05841795e-01 4.64232564e-01 1.67322919e-01
7.07498267e-02 -2.93355286e-01 7.20961630e-01 5.22564709e-01
-6.76083386e-01 4.36167687e-01 -2.05849454e-01 -2.22539544e-01
6.53845489e-01 -1.03720397e-01 -7.07214594e-01 -2.13984951e-01
-4.88213710e-02 -1.02660786e-02 9.09964442e-02 1.85601600e-02
6.33000851e-01 -1.35927951e+00 -1.48472250e-01 4.76207882e-01
-3.79572898e-01 1.92890927e-01 5.06314456e-01 7.88203120e-01
-8.94658983e-01 8.68493617e-01 2.18456034e-02 -4.42329168e-01
-1.24462914e+00 4.35068160e-01 3.24903846e-01 -2.03862548e-01
-6.03640437e-01 1.16769290e+00 -7.53111299e-03 7.73665980e-02
6.39820933e-01 -6.36399567e-01 2.65055329e-01 -2.80832887e-01
6.92123830e-01 5.74443936e-01 5.90255916e-01 -2.89263695e-01
-3.95972937e-01 1.50712907e-01 -1.84294477e-01 2.13216528e-01
1.02472150e+00 1.27395377e-01 -1.18819013e-01 -1.73700884e-01
1.34315050e+00 -1.51208848e-01 -1.03369629e+00 -3.13911915e-01
-2.72259533e-01 -1.22696862e-01 6.82252288e-01 -5.46618164e-01
-1.69928110e+00 9.68089640e-01 7.27194726e-01 -3.21000218e-02
1.57734728e+00 -3.82043362e-01 1.55207109e+00 5.61593592e-01
2.41230041e-01 -1.08942711e+00 -1.61056772e-01 5.40620148e-01
2.93661624e-01 -7.71116734e-01 2.91672943e-04 -3.73414367e-01
9.89119411e-02 1.14228570e+00 6.04923964e-01 1.65097505e-01
7.89602041e-01 9.71433103e-01 -3.02157968e-01 3.81881036e-02
-6.01889729e-01 2.66053081e-01 2.10673183e-01 8.88321325e-02
7.31325448e-02 1.66373491e-01 -3.57246459e-01 1.07437432e+00
-4.87014502e-01 -9.32165459e-02 2.64851898e-01 7.41254568e-01
-3.49858046e-01 -9.72023666e-01 -2.40522549e-01 5.72774887e-01
-5.70071638e-01 -7.43644357e-01 2.39824206e-01 4.56154138e-01
2.70425458e-03 5.39019763e-01 2.15184018e-01 -9.82026279e-01
6.51184097e-02 -1.77613616e-01 1.96894065e-01 -9.38476026e-02
-5.37300348e-01 1.07090073e-02 1.17932528e-01 -6.98821306e-01
3.27986389e-01 -1.50242886e-02 -1.29238665e+00 -1.03081286e+00
-6.74359739e-01 -1.46688238e-01 1.17000592e+00 6.01834416e-01
6.87440872e-01 8.92699063e-01 3.07844132e-01 -9.01633859e-01
-7.96711981e-01 -6.86237097e-01 -4.50516969e-01 -4.31603134e-01
3.08722734e-01 -4.92748410e-01 -3.36755127e-01 -5.57947874e-01] | [8.514071464538574, 2.985804319381714] |
96a76384-aa40-4b86-8eb3-0b272528862a | learning-spatial-temporal-graphs-for-active | 2112.01479 | null | https://arxiv.org/abs/2112.01479v2 | https://arxiv.org/pdf/2112.01479v2.pdf | Learning Spatial-Temporal Graphs for Active Speaker Detection | We address the problem of active speaker detection through a new framework, called SPELL, that learns long-range multimodal graphs to encode the inter-modal relationship between audio and visual data. We cast active speaker detection as a node classification task that is aware of longer-term dependencies. We first construct a graph from a video so that each node corresponds to one person. Nodes representing the same identity share edges between them within a defined temporal window. Nodes within the same video frame are also connected to encode inter-person interactions. Through extensive experiments on the Ava-ActiveSpeaker dataset, we demonstrate that learning graph-based representation, owing to its explicit spatial and temporal structure, significantly improves the overall performance. SPELL outperforms several relevant baselines and performs at par with state of the art models while requiring an order of magnitude lower computation cost. | ['Somdeb Majumdar', 'Tanaya Guha', 'Subarna Tripathi', 'Kyle Min', 'Sourya Roy'] | 2021-12-02 | null | null | null | null | ['audio-visual-active-speaker-detection'] | ['computer-vision'] | [ 2.27537081e-01 1.81054130e-01 -2.84661382e-01 -4.69239414e-01
-9.69459653e-01 -8.05644572e-01 6.82822883e-01 2.72273511e-01
-2.47598529e-01 7.35314516e-03 6.94016039e-01 1.61806583e-01
4.01718579e-02 -3.68392259e-01 -5.42833209e-01 -6.18295848e-01
-8.18872988e-01 3.39532793e-01 2.75712013e-01 6.00967519e-02
-6.76629245e-02 3.20123553e-01 -1.28093743e+00 5.48966467e-01
3.44441682e-02 9.79285002e-01 -1.21583976e-01 1.04690063e+00
-6.36384776e-03 1.07811594e+00 -5.32162547e-01 -3.88156831e-01
-1.45024076e-01 -4.37226027e-01 -7.73277164e-01 3.84493291e-01
8.65874767e-01 -6.92507029e-02 -6.81770027e-01 7.64822185e-01
4.78045732e-01 3.52701634e-01 4.42652881e-01 -1.63763332e+00
-5.93619764e-01 8.02838385e-01 -7.50326037e-01 4.61566865e-01
9.08170700e-01 -1.20125420e-01 1.50985038e+00 -8.61006796e-01
5.68241119e-01 1.59414208e+00 5.89534938e-01 5.50772607e-01
-1.40206766e+00 -5.28639197e-01 6.62875473e-01 3.62986892e-01
-1.63720179e+00 -9.11320210e-01 9.12079215e-01 -3.94092500e-01
8.06631267e-01 3.55380058e-01 7.95258105e-01 1.22715330e+00
-1.29880026e-01 9.40535784e-01 3.82503688e-01 -3.61694306e-01
3.58613208e-02 -2.34598696e-01 3.91711324e-01 9.73792195e-01
-4.37710404e-01 -2.21360907e-01 -1.55019069e+00 -3.52766752e-01
4.56433058e-01 -2.70117462e-01 -4.79250431e-01 -6.97321653e-01
-1.34435010e+00 8.54663968e-01 4.65390652e-01 6.46591634e-02
-1.25058770e-01 4.53713596e-01 3.01038265e-01 3.43957335e-01
4.23081487e-01 -3.91453430e-02 2.72305429e-01 -9.69798118e-02
-8.62526476e-01 -1.25102520e-01 9.49866712e-01 8.06484342e-01
6.50237024e-01 -7.34113455e-02 -1.97589874e-01 7.05232978e-01
7.05340147e-01 3.76489699e-01 -5.27230315e-02 -1.00799108e+00
4.65113103e-01 5.95572889e-01 -1.49363860e-01 -1.08821917e+00
-2.86676109e-01 -2.91858971e-01 -4.10921514e-01 -5.38042523e-02
2.84379661e-01 3.55142131e-02 -7.48197019e-01 1.85835969e+00
4.12179917e-01 5.93009472e-01 -1.06436022e-01 7.93397188e-01
1.13685393e+00 6.47722781e-01 1.03160255e-01 -4.00049359e-01
1.39233649e+00 -1.04221869e+00 -8.17224920e-01 -3.68818104e-01
2.53622562e-01 -4.41271514e-01 7.71306634e-01 2.03197882e-01
-9.57750022e-01 -3.93672049e-01 -9.48001623e-01 -2.09791549e-02
-8.38799849e-02 -3.32861692e-01 6.81617200e-01 5.62130868e-01
-1.54323947e+00 1.24585181e-02 -9.51191604e-01 -5.36835074e-01
1.87289611e-01 5.05088806e-01 -6.32786810e-01 1.88921809e-01
-1.03345156e+00 1.60918891e-01 -2.08418906e-01 8.59126374e-02
-1.19536901e+00 -5.33464968e-01 -1.24534845e+00 7.00944215e-02
3.49754155e-01 -6.29385769e-01 1.30811417e+00 -1.12273085e+00
-1.32928300e+00 1.00060475e+00 -5.86005330e-01 -4.76701856e-01
1.27900228e-01 9.62274894e-02 -7.68838406e-01 6.30681038e-01
-1.73376337e-01 7.93902755e-01 1.19792676e+00 -1.32736063e+00
-5.47341585e-01 -5.73292077e-01 2.84846127e-01 5.21345675e-01
-5.07357478e-01 3.43639791e-01 -1.04512215e+00 -5.09631872e-01
2.04521701e-01 -1.04974222e+00 2.44305015e-01 1.34004459e-01
-5.43605268e-01 -4.25939232e-01 1.09300625e+00 -5.01262963e-01
1.29760802e+00 -2.41605306e+00 5.70697010e-01 3.08555186e-01
6.54543996e-01 -5.15709043e-01 -3.89267772e-01 4.94950056e-01
-4.52170745e-02 -8.48175138e-02 5.78638315e-02 -9.53687608e-01
3.33011076e-02 -7.42666945e-02 -1.06463917e-01 6.76301956e-01
-1.72014505e-01 9.03713644e-01 -8.59482169e-01 -6.53912425e-01
-7.41699338e-02 8.75514388e-01 -2.47268721e-01 2.61838019e-01
9.39493626e-02 3.77229661e-01 -1.20262519e-01 7.11385965e-01
2.52520561e-01 -4.66446906e-01 3.12602997e-01 -1.93847224e-01
1.59274831e-01 3.05631429e-01 -1.23551011e+00 2.06608438e+00
-2.41810605e-01 1.14259303e+00 6.36012912e-01 -8.18846047e-01
4.87969071e-01 5.41003585e-01 5.18395185e-01 -3.94457191e-01
-3.05727214e-01 -3.99187773e-01 -6.17371835e-02 -4.01455075e-01
3.44242364e-01 4.11287338e-01 -1.11386761e-01 4.32156652e-01
1.05089635e-01 4.00797099e-01 1.17517211e-01 1.03961384e+00
1.24286091e+00 -2.67942369e-01 6.64245104e-03 -6.28260747e-02
6.13808155e-01 -4.71271276e-01 2.87815362e-01 8.15654635e-01
-4.70766395e-01 2.87113994e-01 5.57747781e-01 -4.15902399e-03
-2.01357290e-01 -1.34202373e+00 2.73431927e-01 1.77409041e+00
2.72361964e-01 -8.30675304e-01 -4.71938550e-01 -6.35880470e-01
3.38248089e-02 2.42171407e-01 -7.34206140e-01 1.81193873e-01
-7.32232869e-01 -1.12686627e-01 4.66355950e-01 5.38875699e-01
1.84144765e-01 -8.17997456e-01 -1.87397927e-01 3.79722640e-02
-3.02818567e-01 -1.37891400e+00 -1.02806771e+00 -1.20873645e-01
-4.64978427e-01 -9.40555453e-01 -8.05597305e-01 -1.17659438e+00
5.22743464e-01 3.85337651e-01 1.45962906e+00 1.12256594e-02
-1.78445891e-01 1.20247376e+00 -2.72473812e-01 2.92850472e-02
-3.23560268e-01 1.18673490e-02 -8.53502285e-03 5.56885958e-01
2.31554657e-01 -4.94301915e-01 -7.06702590e-01 1.02734759e-01
-2.99660504e-01 -1.47241205e-01 4.08142135e-02 5.40548682e-01
4.85066980e-01 -3.66403699e-01 4.98343885e-01 -7.18476713e-01
3.82742614e-01 -1.40952229e-01 -3.12117934e-01 4.40054595e-01
-2.07421035e-02 -1.83443651e-01 -1.23487942e-01 -5.05315721e-01
-7.82532454e-01 6.05932057e-01 3.40276212e-01 -4.45997566e-01
-8.89640115e-03 4.38724846e-01 -2.10688740e-01 -2.50656813e-01
4.11310524e-01 1.26274273e-01 2.55900528e-02 -1.32639632e-01
5.09316504e-01 5.43769538e-01 7.61278927e-01 -1.40995368e-01
6.40280008e-01 6.50610685e-01 -3.67082804e-02 -1.13676488e+00
-6.56355858e-01 -7.89947152e-01 -6.81159675e-01 -5.88685811e-01
9.86825824e-01 -1.29884732e+00 -1.20542908e+00 3.88521612e-01
-1.06374991e+00 -3.58618259e-01 -6.44291788e-02 4.56405699e-01
-3.48495066e-01 5.59801280e-01 -7.42432415e-01 -1.11931741e+00
-1.16907209e-01 -8.90092731e-01 1.39805436e+00 -6.54668808e-02
-2.75452077e-01 -1.21927810e+00 1.97437376e-01 4.91722286e-01
-3.38744670e-02 8.14862847e-02 5.59567630e-01 -5.48015893e-01
-7.31065035e-01 -7.13009089e-02 1.16375394e-01 -1.94630966e-01
2.06267700e-01 -4.41691130e-02 -1.07045364e+00 -6.51464939e-01
-4.39280748e-01 -2.50461251e-01 1.15788949e+00 3.26630592e-01
7.97418952e-01 -2.65871316e-01 -6.26744092e-01 3.98564875e-01
6.89491391e-01 2.64835842e-02 2.42120400e-01 -1.93957046e-01
1.02552414e+00 6.50566936e-01 3.64237018e-02 3.18216175e-01
5.38274765e-01 8.61544073e-01 4.04290766e-01 -1.09369189e-01
-5.40632904e-01 -2.07309887e-01 7.76818752e-01 7.05037832e-01
7.01807886e-02 -7.08549857e-01 -9.30717170e-01 5.55214882e-01
-1.91698074e+00 -1.16777241e+00 -1.93183184e-01 2.12703538e+00
4.87578094e-01 -4.03527021e-02 6.01741910e-01 1.61128029e-01
9.51390147e-01 5.96267343e-01 -4.11556005e-01 7.41761029e-02
-2.08989337e-01 -2.27943391e-01 -1.03702433e-01 9.99106526e-01
-1.28336072e+00 8.89301598e-01 6.75336933e+00 2.73792714e-01
-1.01195967e+00 3.98763791e-02 3.88350368e-01 -2.84579426e-01
-1.55479923e-01 -2.67671824e-01 -6.42439783e-01 5.63024580e-02
9.20316637e-01 -8.31211656e-02 6.15080714e-01 3.93771350e-01
6.37218878e-02 1.09701961e-01 -1.62435913e+00 1.33281553e+00
6.51608229e-01 -1.24419892e+00 -2.35224441e-02 9.26154777e-02
3.81470799e-01 2.28114966e-02 1.37771115e-01 -1.68052167e-02
2.91259497e-01 -1.05851376e+00 8.39909792e-01 3.57236415e-01
7.55354226e-01 -5.90690911e-01 9.81000587e-02 1.10607542e-01
-1.97967291e+00 3.49109583e-02 3.22684020e-01 2.27690175e-01
3.33996356e-01 4.29934561e-02 -7.40111113e-01 2.57011920e-01
7.82527268e-01 9.41367090e-01 -7.12145269e-01 8.00018132e-01
-2.50132650e-01 7.19093621e-01 -2.68925488e-01 1.71839476e-01
-4.98075187e-02 2.08036393e-01 9.23782527e-01 1.36972070e+00
4.78961952e-02 -9.71297547e-02 3.46550107e-01 4.41727042e-01
-3.68681282e-01 -1.38355391e-02 -6.22104406e-01 -1.27670124e-01
6.17721617e-01 1.26718497e+00 -6.52176976e-01 -2.67064840e-01
-7.98683465e-01 1.33295774e+00 4.48858380e-01 7.15843678e-01
-7.65836358e-01 -8.87741968e-02 5.31071424e-01 9.99020711e-02
3.81548434e-01 -3.80859822e-01 2.84952641e-01 -1.04952168e+00
-1.39571652e-01 -7.30580807e-01 9.83351469e-01 -7.31278479e-01
-1.25991225e+00 7.23841906e-01 -2.67182857e-01 -9.90933239e-01
-5.13943017e-01 -2.61955053e-01 -5.66663921e-01 6.18313253e-01
-1.11545622e+00 -1.45037985e+00 -5.00361979e-01 1.15796542e+00
4.75493670e-01 -1.56266853e-01 9.81358171e-01 2.17363071e-02
-4.64631826e-01 8.24559629e-01 -4.04389501e-01 5.34279644e-01
6.62547112e-01 -1.19313741e+00 4.39111769e-01 8.28200579e-01
1.04618132e+00 5.58640718e-01 5.19374669e-01 -4.70665067e-01
-1.59720421e+00 -6.85845792e-01 9.91571963e-01 -5.42991340e-01
8.71509135e-01 -1.04867887e+00 -7.72515714e-01 9.55153525e-01
6.72144294e-01 3.71343046e-01 8.17413151e-01 5.46827257e-01
-8.74699831e-01 -1.67837769e-01 -7.42103457e-01 2.38976777e-01
1.37540925e+00 -1.17003119e+00 -3.67369831e-01 3.48324239e-01
6.26235783e-01 -3.02315354e-01 -6.51798844e-01 8.56458098e-02
5.37240982e-01 -8.88441741e-01 1.14225030e+00 -6.09931350e-01
-2.14398757e-01 -1.40063643e-01 -1.63644403e-01 -1.17415452e+00
-3.88277292e-01 -1.07191920e+00 -6.62263453e-01 1.49828541e+00
5.86471796e-01 -2.73602217e-01 7.77656913e-01 4.82073516e-01
-5.58942556e-02 -2.55473822e-01 -1.29347539e+00 -6.49269879e-01
-5.62028348e-01 -3.89372021e-01 2.66894668e-01 1.01735556e+00
4.25047487e-01 7.82674611e-01 -3.35230261e-01 5.45304358e-01
8.58661234e-01 1.10080875e-01 6.99794650e-01 -1.41651249e+00
-5.71650207e-01 -1.88064963e-01 -8.34791005e-01 -1.31184518e+00
6.13875687e-01 -8.49843621e-01 2.61113588e-02 -1.61542964e+00
4.71576750e-01 1.23448499e-01 -3.96589130e-01 5.69190443e-01
-4.02637199e-02 5.48620999e-01 1.73145235e-01 3.23827446e-01
-1.10463810e+00 2.25549966e-01 8.67406845e-01 -5.34833014e-01
-3.75004768e-01 -1.35085896e-01 -3.67169410e-01 7.04121292e-01
1.64616928e-01 -3.33196998e-01 -5.21527231e-01 -6.13667309e-01
2.52403796e-01 5.28634489e-01 5.68048298e-01 -7.47819602e-01
7.01417863e-01 2.44243219e-01 1.61965325e-01 -5.06780624e-01
8.92473578e-01 -7.29384661e-01 -6.71898760e-03 5.89690655e-02
-8.28116715e-01 1.57069743e-01 -4.98602092e-02 1.00749505e+00
-4.61829990e-01 4.08848077e-01 4.89972919e-01 2.62391031e-01
-9.87076283e-01 4.04115736e-01 -2.79418230e-01 2.64441827e-03
8.41510177e-01 -3.42486352e-02 -2.74996191e-01 -1.06057405e+00
-1.18885386e+00 2.77636260e-01 1.41267434e-01 5.94400764e-01
5.43529630e-01 -1.34185553e+00 -6.44026339e-01 -2.73735798e-03
4.09697682e-01 -4.95475680e-01 3.71849716e-01 7.66962290e-01
-2.44545937e-02 1.36529267e-01 4.47720706e-01 -1.07455337e+00
-1.99949622e+00 4.50018883e-01 4.26241785e-01 1.63574770e-01
-7.18403637e-01 1.41820884e+00 4.97996926e-01 1.26128718e-01
1.02243567e+00 6.16356207e-04 -3.26619178e-01 3.35434765e-01
6.96469724e-01 1.96927890e-01 -1.18549101e-01 -1.33554924e+00
-9.03258204e-01 4.92041826e-01 9.11752582e-02 -5.20409405e-01
9.96281624e-01 -4.84492004e-01 4.06824052e-02 8.04592550e-01
1.18990278e+00 3.71105373e-01 -1.36110163e+00 -4.00323540e-01
-3.10863666e-02 -3.73754472e-01 2.72487253e-01 -5.23776710e-01
-1.29727435e+00 8.36509883e-01 6.77406311e-01 5.24033844e-01
9.22453284e-01 7.82683611e-01 2.27044180e-01 3.45228046e-01
7.86188319e-02 -7.23719418e-01 5.91883063e-01 2.93071896e-01
9.37862098e-01 -1.24018204e+00 -9.09811035e-02 -7.27063596e-01
-8.12226057e-01 9.42694247e-01 2.52318799e-01 2.33251259e-01
6.40349805e-01 3.61019999e-01 2.81576216e-01 -3.93505901e-01
-7.53787756e-01 -3.42014492e-01 7.76134372e-01 7.06257343e-01
6.22941673e-01 5.32036945e-02 5.48841655e-01 3.21000010e-01
5.03064655e-02 -5.19066393e-01 5.67227751e-02 5.87674141e-01
-1.95504442e-01 -8.70180845e-01 -2.31006622e-01 -1.90932468e-01
-2.31129497e-01 -7.03501850e-02 -7.95970678e-01 6.97250664e-01
-3.87701631e-01 1.54866338e+00 5.72729051e-01 -3.87986749e-01
2.27042094e-01 1.61718667e-01 5.28970301e-01 -6.50935888e-01
-5.84288478e-01 5.20161688e-01 3.98605675e-01 -6.77002251e-01
-8.60555828e-01 -8.89830112e-01 -1.46092927e+00 -2.81811923e-01
-1.73286527e-01 2.62226254e-01 3.30501825e-01 5.50162911e-01
4.29402709e-01 4.70547348e-01 6.64153636e-01 -6.60339057e-01
-1.38621610e-02 -7.36928642e-01 -7.95967400e-01 5.80425203e-01
7.22663224e-01 -4.66694385e-01 -6.22957230e-01 2.59336621e-01] | [14.464519500732422, 5.102067470550537] |
f3af10aa-a9ab-478c-8419-89164a864667 | self-supervised-predictive-coding-models | 2305.12464 | null | https://arxiv.org/abs/2305.12464v2 | https://arxiv.org/pdf/2305.12464v2.pdf | Self-supervised Predictive Coding Models Encode Speaker and Phonetic Information in Orthogonal Subspaces | Self-supervised speech representations are known to encode both speaker and phonetic information, but how they are distributed in the high-dimensional space remains largely unexplored. We hypothesize that they are encoded in orthogonal subspaces, a property that lends itself to simple disentanglement. Applying principal component analysis to representations of two predictive coding models, we identify two subspaces that capture speaker and phonetic variances, and confirm that they are nearly orthogonal. Based on this property, we propose a new speaker normalization method which collapses the subspace that encodes speaker information, without requiring transcriptions. Probing experiments show that our method effectively eliminates speaker information and outperforms a previous baseline in phone discrimination tasks. Moreover, the approach generalizes and can be used to remove information of unseen speakers. | ['Sharon Goldwater', 'Hao Tang', 'Oli Liu'] | 2023-05-21 | null | null | null | null | ['disentanglement'] | ['methodology'] | [ 2.62307167e-01 6.50694370e-02 -4.00349647e-01 -4.79990870e-01
-7.54444778e-01 -9.79224384e-01 6.10308170e-01 -2.55996615e-01
-1.45024676e-02 3.41094345e-01 9.40093815e-01 -3.45070332e-01
2.61161197e-03 -2.16267884e-01 -3.53193998e-01 -8.49021316e-01
2.34049022e-01 4.77804691e-01 -2.25993410e-01 -1.74815785e-02
-4.33003530e-02 4.93653864e-01 -1.76245582e+00 3.37310255e-01
6.46121144e-01 5.32786846e-01 -1.09168880e-01 5.49793959e-01
-1.75507292e-01 4.74062562e-01 -5.90161860e-01 -8.37317109e-02
2.18041301e-01 -5.15497506e-01 -7.02035844e-01 2.52052009e-01
7.02299893e-01 -2.08757166e-02 -5.44461429e-01 1.09787285e+00
4.78329808e-01 1.93493411e-01 1.09457779e+00 -1.06210911e+00
-8.40975821e-01 6.14632368e-01 -2.22498193e-01 2.29483888e-01
2.22376972e-01 -4.49960053e-01 1.15242112e+00 -8.58992219e-01
2.92032331e-01 1.45350695e+00 4.89182115e-01 6.71306193e-01
-1.83680809e+00 -6.95164323e-01 1.17640644e-01 9.04799253e-02
-1.49233902e+00 -1.11059332e+00 1.04923952e+00 -7.09132135e-01
9.10749614e-01 5.90929866e-01 4.41157699e-01 1.45818102e+00
-1.43937737e-01 8.55264664e-01 1.00957727e+00 -5.51607311e-01
2.79572845e-01 3.95161539e-01 3.54827791e-01 3.19292217e-01
4.51499671e-02 1.65167853e-01 -7.55908787e-01 -2.51946539e-01
4.44785416e-01 -1.11986183e-01 -4.67524976e-01 -8.27706277e-01
-1.11226118e+00 9.60680306e-01 -1.76109612e-01 4.61652666e-01
-1.73956484e-01 -3.44193578e-01 2.17417061e-01 1.46924391e-01
3.88098806e-01 1.23169132e-01 -3.88401240e-01 -1.38223201e-01
-1.01992273e+00 -1.05354398e-01 9.05850649e-01 9.42620754e-01
6.42253399e-01 4.54157978e-01 7.29912985e-03 1.21892726e+00
3.68512869e-01 8.53296161e-01 9.49094296e-01 -8.55073452e-01
2.82003105e-01 2.34107062e-01 -2.67582178e-01 -8.93858075e-01
-2.65733510e-01 -3.94364089e-01 -8.60160232e-01 -2.31671840e-01
1.45823047e-01 1.38372675e-01 -8.67908776e-01 2.21358538e+00
-2.78411359e-02 2.42538750e-01 3.97852004e-01 6.50225997e-01
5.86011887e-01 5.15956163e-01 -1.25692561e-01 -3.80660623e-01
1.27582240e+00 -6.98633850e-01 -8.45914125e-01 -3.65548283e-01
2.02965185e-01 -5.76282680e-01 9.07703400e-01 2.89255142e-01
-8.74954224e-01 -6.39356017e-01 -9.43422318e-01 1.06764443e-01
-2.36138791e-01 2.15308383e-01 5.34928918e-01 1.12873495e+00
-9.99002337e-01 4.07119185e-01 -8.42747867e-01 -3.89820844e-01
1.96938798e-01 2.68437177e-01 -6.08848274e-01 1.37429640e-01
-1.01652420e+00 6.02234364e-01 3.26245874e-01 -7.25308061e-02
-7.82832325e-01 -5.08433163e-01 -1.04481602e+00 2.28542030e-01
-1.43686175e-01 -5.08876979e-01 1.06569314e+00 -7.13115871e-01
-1.69234622e+00 5.67978203e-01 -8.18720162e-01 -2.39567950e-01
-1.17279790e-01 1.54709043e-02 -7.53974736e-01 3.05092055e-02
4.52029593e-02 4.50422883e-01 1.25796449e+00 -1.18199193e+00
-2.94746995e-01 -6.72372103e-01 -6.43569589e-01 3.38320643e-01
-9.53744113e-01 5.09975851e-02 -3.75271767e-01 -7.90687740e-01
7.72579908e-01 -1.04170370e+00 8.23092312e-02 -5.59081674e-01
-6.36408985e-01 2.63105874e-04 5.78974009e-01 -5.89050651e-01
1.15475988e+00 -2.46025443e+00 5.05787194e-01 3.52218390e-01
1.42089978e-01 -4.83997865e-03 -2.65931785e-01 3.05084765e-01
-4.43997592e-01 1.20722868e-01 -2.56683975e-01 -5.66263855e-01
1.83579594e-01 4.77679908e-01 -8.30440044e-01 5.31251907e-01
3.43420729e-02 5.89414358e-01 -5.56309164e-01 -2.93921083e-01
1.40015870e-01 6.53404117e-01 -6.84370220e-01 1.71701133e-01
4.39799070e-01 4.65621710e-01 -1.59958955e-02 4.37822133e-01
7.44460464e-01 3.58503938e-01 4.40385014e-01 -1.21530771e-01
1.30880103e-01 7.69802570e-01 -1.28282702e+00 1.51790583e+00
-1.11646906e-01 7.54472017e-01 1.68967575e-01 -1.22936189e+00
9.95746017e-01 4.73048866e-01 2.54312187e-01 -3.80585492e-01
-3.03114001e-02 -7.49557242e-02 5.45219257e-02 -2.52887607e-01
2.75095791e-01 -5.32927573e-01 -1.13469124e-01 3.33971679e-01
4.61795419e-01 -8.09279084e-02 -1.95288911e-01 9.30821672e-02
8.22230697e-01 -2.79165179e-01 3.54903877e-01 -4.12041754e-01
4.72304702e-01 -5.27264953e-01 7.31578410e-01 7.01417506e-01
-3.90132874e-01 6.47213519e-01 3.27891797e-01 1.49788074e-02
-7.42836177e-01 -1.57101202e+00 -4.82945949e-01 1.35677814e+00
-1.67715222e-01 -5.17927349e-01 -6.70749903e-01 -4.86921817e-01
2.84250341e-02 8.14209342e-01 -6.62167609e-01 -3.80722374e-01
-3.81243855e-01 -5.37008464e-01 5.98859608e-01 7.08973885e-01
-3.22492629e-01 -5.58749378e-01 1.68923229e-01 -2.15043128e-03
-2.82407850e-01 -1.05036688e+00 -5.78499019e-01 5.89726090e-01
-9.64708626e-01 -6.96138263e-01 -5.24315894e-01 -6.96248591e-01
5.17382979e-01 5.79748571e-01 8.38804483e-01 -6.66621268e-01
1.22180514e-01 5.61058521e-01 -5.00013679e-02 -3.46597880e-01
-6.71066940e-01 1.57274321e-01 7.56522119e-01 1.77386582e-01
7.32584178e-01 -8.80860686e-01 -4.42140587e-02 3.84639889e-01
-5.24326503e-01 -2.58530915e-01 4.27573442e-01 8.32786858e-01
4.53657538e-01 1.16490878e-01 5.28307319e-01 -6.35657966e-01
5.88901997e-01 -3.51837665e-01 -2.08729148e-01 3.63263078e-02
-3.05226505e-01 3.23886722e-01 4.90237445e-01 -5.28539777e-01
-1.00461638e+00 3.92344326e-01 -8.38755369e-02 -6.74539447e-01
-7.03775883e-01 1.67187080e-01 -5.53129137e-01 2.43957669e-01
5.34282148e-01 5.42566001e-01 1.15334310e-01 -8.01434040e-01
6.15535915e-01 8.52815807e-01 6.08115017e-01 -4.53691989e-01
8.86488855e-01 4.86533910e-01 -4.72598284e-01 -1.31332850e+00
-7.18645096e-01 -7.71967113e-01 -9.57312763e-01 3.11384499e-01
6.23238504e-01 -1.05066121e+00 -5.38815498e-01 -1.21505462e-01
-1.08604133e+00 2.96046019e-01 -3.55934709e-01 9.14840877e-01
-6.09238148e-01 5.15470743e-01 -5.40428281e-01 -1.05907977e+00
4.63703275e-02 -9.72847700e-01 1.04721928e+00 -2.27782205e-01
-6.30702078e-01 -7.89734185e-01 4.33515310e-01 4.58604813e-01
3.01402748e-01 -6.09264195e-01 8.97806764e-01 -1.15381324e+00
-1.16111942e-01 -1.36690095e-01 2.59407401e-01 5.61370671e-01
3.41513455e-01 -1.94332674e-01 -1.46868074e+00 -3.72813314e-01
3.03765237e-01 2.27507744e-02 1.19416523e+00 3.16774160e-01
9.69952822e-01 -4.57270563e-01 -3.36072177e-01 7.23499954e-01
6.38704598e-01 6.77434579e-02 3.87966365e-01 -1.50158107e-01
8.27117085e-01 9.33963478e-01 -3.36127579e-01 1.19661383e-01
2.55642891e-01 8.12568903e-01 1.66690815e-02 2.11319223e-01
-8.37128162e-02 -4.15013343e-01 6.21179402e-01 1.48630488e+00
1.55130193e-01 -2.28575971e-02 -6.65413499e-01 5.96730173e-01
-1.55271506e+00 -1.24625278e+00 2.02734426e-01 2.23748207e+00
7.35522628e-01 -9.17910784e-02 2.29520679e-01 3.91568124e-01
6.92760110e-01 2.43447185e-01 -5.31137168e-01 -3.71144861e-01
-4.59089667e-01 6.31087795e-02 1.54643133e-01 5.03475845e-01
-1.25476003e+00 9.72514093e-01 8.00874996e+00 6.43218577e-01
-9.62225378e-01 2.92105582e-02 5.83832264e-02 -1.88467339e-01
-4.96831596e-01 -1.25504330e-01 -8.34570944e-01 2.14794666e-01
1.19581878e+00 -1.10247776e-01 5.21695673e-01 1.02300870e+00
3.44646089e-02 6.59965575e-01 -1.34903765e+00 1.10277712e+00
5.62998593e-01 -7.95331776e-01 3.25310111e-01 2.14090645e-01
5.65346241e-01 -3.56372725e-03 3.13225865e-01 4.33746636e-01
3.35203737e-01 -9.95542228e-01 7.35080838e-01 2.32155293e-01
3.93293142e-01 -5.37568152e-01 3.01069558e-01 4.52345937e-01
-1.11940253e+00 -2.57515341e-01 -5.56163788e-01 3.78671996e-02
7.62469247e-02 3.84678036e-01 -9.91074204e-01 1.61880672e-01
5.02325237e-01 5.66180706e-01 -4.51971620e-01 5.94943583e-01
-3.41373652e-01 1.00811398e+00 -9.25637335e-02 3.75936270e-01
-3.30762267e-01 -2.21339032e-01 8.22860360e-01 1.35850048e+00
3.65896434e-01 -4.66281287e-02 4.56147045e-02 9.08338845e-01
-1.73288733e-02 2.49917358e-01 -7.83785343e-01 -1.53945014e-01
6.80926859e-01 8.92638266e-01 -4.88293856e-01 -4.41718310e-01
-3.33112568e-01 1.18114436e+00 4.49460238e-01 6.26727879e-01
-3.89175355e-01 -1.10227950e-01 1.31079197e+00 -2.90196210e-01
5.38273871e-01 -3.22261870e-01 -1.75081432e-01 -1.56224036e+00
-9.02900472e-02 -9.01025176e-01 1.17291115e-01 -3.75237614e-01
-1.36025941e+00 7.27762997e-01 -1.68365955e-01 -1.24639833e+00
-6.26078963e-01 -5.99965394e-01 -3.48641664e-01 8.77251565e-01
-1.15119267e+00 -9.05312896e-01 1.90336972e-01 7.43000448e-01
3.75056088e-01 -5.94053090e-01 1.32270181e+00 9.41855311e-02
-8.36894870e-01 8.34873021e-01 5.00021875e-01 1.71602562e-01
6.79479599e-01 -1.27667308e+00 2.48875916e-01 7.30366528e-01
9.03523684e-01 1.13205838e+00 5.93501925e-01 -2.91769981e-01
-1.28002584e+00 -7.92230248e-01 9.28743184e-01 -7.46448696e-01
5.71779728e-01 -7.77779937e-01 -1.21158814e+00 7.65325427e-01
5.87664656e-02 -1.51095137e-01 1.45589948e+00 5.48098803e-01
-1.02210069e+00 -9.44220424e-02 -6.51092649e-01 4.37354088e-01
9.31696832e-01 -1.17487979e+00 -1.12626231e+00 1.20642088e-01
7.67530680e-01 -1.59462839e-02 -5.02489328e-01 7.97436237e-02
6.91290379e-01 -8.39624763e-01 1.16656852e+00 -9.98483777e-01
-1.62818149e-01 -1.89904064e-01 -7.92854726e-01 -1.53593493e+00
-5.90688229e-01 -5.18257320e-01 -2.32470155e-01 1.45202720e+00
4.08907831e-01 -6.91268802e-01 6.89880133e-01 4.52166408e-01
-1.11078538e-01 -8.83409306e-02 -1.23064280e+00 -1.11677432e+00
2.54597396e-01 -6.62414968e-01 6.08206272e-01 1.28798473e+00
3.34614575e-01 8.02237391e-01 -4.09974694e-01 3.64412695e-01
6.49771035e-01 1.35651514e-01 5.29604077e-01 -1.62408149e+00
-3.93545449e-01 -6.17260456e-01 -6.80055439e-01 -1.24418318e+00
8.18477392e-01 -1.16259289e+00 9.51987207e-02 -1.09556282e+00
3.86863470e-01 1.47020156e-02 -4.93841887e-01 5.60348928e-01
1.40607674e-02 1.87849462e-01 6.14751689e-02 4.87547725e-01
-1.65708840e-01 8.65874052e-01 4.13770288e-01 -1.97354361e-01
-4.27570194e-01 3.82645056e-02 -1.03533065e+00 9.23103333e-01
7.59403706e-01 -3.87773424e-01 -4.35417622e-01 -3.51357192e-01
-4.64645952e-01 -2.05622748e-01 6.50617480e-02 -9.23275113e-01
7.35146701e-02 7.42315641e-03 4.16919559e-01 -4.68026519e-01
6.56546533e-01 -7.67374933e-01 -8.80646259e-02 5.71135543e-02
-5.66593468e-01 -4.23951536e-01 3.16210210e-01 7.36732483e-01
-3.88109267e-01 -9.45604071e-02 5.41614473e-01 3.61328930e-01
-2.86676198e-01 5.30003868e-02 -6.86304748e-01 -6.41194880e-02
5.61534286e-01 -1.86914951e-01 -5.94082940e-03 -5.21537900e-01
-8.08472157e-01 -3.51675421e-01 1.19438998e-01 6.76045001e-01
3.56848031e-01 -1.60358346e+00 -6.23240352e-01 7.95424521e-01
2.11176574e-01 -6.49409056e-01 3.39373022e-01 5.90805709e-01
3.44622999e-01 8.03157985e-01 8.14162418e-02 -6.81898832e-01
-1.35458553e+00 8.21514904e-01 9.23456252e-02 3.73758316e-01
-3.84855449e-01 7.92382300e-01 5.13452888e-01 -7.97074735e-01
4.02684212e-01 -1.37858689e-01 -4.09135073e-01 3.72413576e-01
7.09413171e-01 2.39823937e-01 1.57184936e-02 -1.34215248e+00
-5.47433197e-01 6.52233601e-01 -6.31620064e-02 -3.51232082e-01
1.33912444e+00 -3.03360552e-01 -4.06996794e-02 6.90202713e-01
1.51556718e+00 4.65966016e-01 -8.67042542e-01 -3.78862590e-01
1.64885968e-02 -4.85395193e-01 6.62825117e-03 -2.79054642e-01
-7.53938496e-01 1.20931852e+00 5.70996702e-01 4.07928973e-01
9.21498537e-01 2.94914871e-01 2.35342830e-01 3.53209138e-01
-1.55368805e-01 -8.40159774e-01 -1.42188028e-01 6.54750526e-01
9.03999507e-01 -1.00421345e+00 -3.75013113e-01 -5.50923944e-01
-6.47464573e-01 1.01235735e+00 2.25186542e-01 3.15707445e-01
6.80556953e-01 2.41326883e-01 1.18807890e-01 2.14280531e-01
-6.60868883e-01 -3.15737486e-01 6.64695859e-01 9.69166577e-01
6.40106499e-01 3.30141693e-01 2.11027652e-01 8.92911494e-01
-7.57474720e-01 -6.59206688e-01 2.48938560e-01 3.58390421e-01
-4.74269360e-01 -1.04607809e+00 -6.93308115e-01 1.52033478e-01
-1.20161459e-01 -2.29810745e-01 -5.62731504e-01 2.27635235e-01
-1.84164807e-01 9.92262781e-01 1.47555292e-01 -5.12642264e-01
4.01725918e-01 7.18458176e-01 1.21023476e-01 -7.40113676e-01
1.44308895e-01 4.97056365e-01 -1.68338746e-01 -4.10647303e-01
-1.77403897e-01 -8.79506886e-01 -8.27335060e-01 2.21124273e-02
-3.89500678e-01 3.36103201e-01 7.32919991e-01 7.71748066e-01
4.62890863e-01 4.73547488e-01 6.47643328e-01 -8.68531704e-01
-8.88445795e-01 -9.63083088e-01 -8.07320535e-01 4.32173967e-01
4.95075315e-01 -6.53943539e-01 -6.53815985e-01 1.86256111e-01] | [14.498603820800781, 6.366350173950195] |
dc7357e0-ea0c-4624-bd22-08c5ca109b5d | semantic-guided-zero-shot-learning-for-low | 2110.00970 | null | https://arxiv.org/abs/2110.00970v4 | https://arxiv.org/pdf/2110.00970v4.pdf | Semantic-Guided Zero-Shot Learning for Low-Light Image/Video Enhancement | Low-light images challenge both human perceptions and computer vision algorithms. It is crucial to make algorithms robust to enlighten low-light images for computational photography and computer vision applications such as real-time detection and segmentation. This paper proposes a semantic-guided zero-shot low-light enhancement network (SGZ) which is trained in the absence of paired images, unpaired datasets, and segmentation annotation. Firstly, we design an enhancement factor extraction network using depthwise separable convolution for an efficient estimate of the pixel-wise light deficiency of an low-light image. Secondly, we propose a recurrent image enhancement network to progressively enhance the low-light image with affordable model size. Finally, we introduce an unsupervised semantic segmentation network for preserving the semantic information during intensive enhancement. Extensive experiments on benchmark datasets and a low-light video demonstrate that our model outperforms the previous state-of-the-art. We further discuss the benefits of the proposed method for low-light detection and segmentation. Code is available at https://github.com/ShenZheng2000/Semantic-Guided-Low-Light-Image-Enhancement | ['Gaurav Gupta', 'Shen Zheng'] | 2021-10-03 | null | null | null | null | ['unsupervised-semantic-segmentation', 'video-enhancement'] | ['computer-vision', 'computer-vision'] | [ 7.94441283e-01 -2.29929611e-01 8.85223001e-02 -6.56866729e-01
-7.23144948e-01 -3.05422336e-01 2.08552852e-01 -7.83329830e-02
-6.00206733e-01 4.87736672e-01 -1.94985837e-01 -2.63160855e-01
2.17287660e-01 -9.15716052e-01 -8.06008875e-01 -8.30836594e-01
5.18441856e-01 -3.85581911e-01 5.62332153e-01 -2.43970007e-01
4.72472399e-01 3.44618350e-01 -1.78296793e+00 2.04483405e-01
8.53547096e-01 9.75621939e-01 6.48553967e-01 9.18730199e-01
7.26382509e-02 8.05594325e-01 9.07376334e-02 -1.71633199e-01
7.13326573e-01 -4.87042487e-01 -5.63741565e-01 3.63078535e-01
7.56761253e-01 -7.51944423e-01 -2.56768107e-01 1.38980401e+00
6.56225622e-01 2.93502241e-01 2.52060860e-01 -1.01336515e+00
-5.85281491e-01 -1.71407983e-02 -8.42492104e-01 5.03871262e-01
4.66289669e-02 5.82312405e-01 5.67817330e-01 -9.87334013e-01
5.92522383e-01 7.43872344e-01 6.43094659e-01 4.37611103e-01
-1.01502681e+00 -6.01189852e-01 -2.23641157e-01 3.68214399e-01
-9.89418387e-01 -6.26170158e-01 9.66296136e-01 -3.52125950e-02
8.25618207e-01 -4.31724712e-02 5.64026296e-01 2.38722622e-01
1.25535727e-01 5.55658579e-01 1.39715576e+00 -6.27188146e-01
1.67788967e-01 -1.17233440e-01 -2.39153448e-02 9.59321260e-01
1.61976025e-01 5.00700951e-01 -5.00382066e-01 4.73058939e-01
6.52768970e-01 2.73209244e-01 -3.72594446e-01 1.74458757e-01
-7.97345102e-01 2.82705277e-01 6.26143098e-01 9.94316712e-02
-4.48536754e-01 2.76156873e-01 7.96833932e-02 -5.57177328e-03
8.07031274e-01 1.78122342e-01 -3.68053168e-01 1.97299957e-01
-1.03864133e+00 -1.63503379e-01 -5.05248606e-02 5.75128257e-01
1.19294071e+00 -2.91536395e-02 -5.35178967e-02 9.08164561e-01
7.89656416e-02 5.21242261e-01 -4.80757691e-02 -1.28174520e+00
1.47790059e-01 4.15993065e-01 1.71328709e-01 -6.51382565e-01
-4.67689127e-01 -2.06811056e-01 -7.79334366e-01 6.56075835e-01
3.77973735e-01 -2.00441889e-02 -1.11130881e+00 1.48602474e+00
5.44343114e-01 4.87889141e-01 2.78392844e-02 1.30003107e+00
9.41287458e-01 8.12772274e-01 -3.57627198e-02 -4.22176033e-01
1.26581776e+00 -1.20123374e+00 -5.70559382e-01 -3.19775701e-01
1.80181772e-01 -9.64231074e-01 1.18556619e+00 2.55044281e-01
-1.70611525e+00 -6.54821634e-01 -1.05546725e+00 -7.77508020e-01
-1.67135194e-01 2.54830867e-01 7.05009341e-01 6.87895775e-01
-1.11950111e+00 4.58922684e-01 -6.82080448e-01 -1.40683413e-01
6.18385971e-01 6.20580167e-02 3.38823162e-02 -5.20382881e-01
-9.44953263e-01 4.33551073e-01 4.64953601e-01 2.59135753e-01
-9.45307195e-01 -9.52558041e-01 -8.42730343e-01 -2.05691114e-01
5.78282356e-01 -8.09682727e-01 1.16340196e+00 -8.16425502e-01
-1.44309902e+00 1.14877236e+00 -4.55915660e-01 -2.78937638e-01
2.92816550e-01 1.51282400e-01 -6.90565072e-03 5.64516127e-01
8.58909488e-02 6.95511401e-01 9.47674692e-01 -1.47712040e+00
-9.59962070e-01 -2.55479068e-01 2.37791598e-01 4.23469067e-01
-9.11975652e-02 2.12319419e-01 -7.94562697e-01 -2.69539386e-01
9.74180400e-02 -5.74894190e-01 -3.65467399e-01 3.38596016e-01
-3.20404649e-01 4.10096869e-02 8.28556418e-01 -6.51228845e-01
8.18336070e-01 -2.12005377e+00 -6.38923585e-01 -2.56501049e-01
3.30606550e-01 3.60807627e-01 -2.55979121e-01 -1.39132783e-01
-1.15296606e-03 -3.76436830e-01 -6.47120535e-01 -5.61701119e-01
-4.43081558e-01 1.42768584e-02 -1.78725515e-02 7.35908508e-01
1.46238014e-01 9.46597993e-01 -1.06787837e+00 -5.71269453e-01
8.06916535e-01 7.14992583e-01 -5.53298533e-01 2.56582081e-01
-1.83112159e-01 3.73570949e-01 -3.12347282e-02 1.03700352e+00
1.00223589e+00 -1.03433952e-01 -2.47335941e-01 -6.42145991e-01
-4.96932983e-01 -1.90538466e-01 -1.03407967e+00 1.72626567e+00
-6.92659080e-01 8.35344434e-01 2.16470912e-01 -5.66161871e-01
6.57478988e-01 -2.61249691e-01 5.03359020e-01 -1.08912182e+00
5.16270876e-01 2.16012254e-01 -6.17801607e-01 -5.62729537e-01
7.27443933e-01 -2.85163879e-01 4.92704958e-01 6.34440124e-01
-2.63460219e-01 -5.35442352e-01 4.69891906e-01 8.29972699e-02
6.86313570e-01 4.00069840e-02 -5.01098344e-03 8.24425370e-02
6.36780441e-01 -3.94861519e-01 6.45895183e-01 6.32451534e-01
-4.51522261e-01 7.81535387e-01 -2.63620645e-01 -4.08133268e-01
-1.09739554e+00 -9.38193560e-01 -1.86333030e-01 1.26685202e+00
6.96936965e-01 -4.41741198e-02 -8.29585016e-01 -1.53426498e-01
-4.91038144e-01 7.15182662e-01 -2.72060037e-01 -5.94731756e-02
-3.53845507e-01 -9.38704014e-01 -1.08238775e-02 3.55124414e-01
1.07010150e+00 -8.75599146e-01 -9.80537653e-01 -9.92166102e-02
-4.37103987e-01 -1.47454488e+00 -4.40390378e-01 2.45169729e-01
-6.27610326e-01 -1.09241104e+00 -4.93168354e-01 -9.84943151e-01
7.91207433e-01 8.25533926e-01 1.00243855e+00 2.03138888e-01
-7.17887819e-01 3.93005341e-01 -1.60236478e-01 -3.69408458e-01
-2.29182422e-01 -3.83190274e-01 -3.70444506e-01 1.52736157e-01
2.26481542e-01 -4.94991779e-01 -1.29187810e+00 2.65691400e-01
-1.18348908e+00 5.01822114e-01 6.85798228e-01 8.19499910e-01
8.86035264e-01 5.55008113e-01 1.48490667e-01 -7.80653775e-01
4.42592889e-01 1.59295768e-01 -7.31706560e-01 2.43144050e-01
-8.27084243e-01 -3.83593321e-01 5.52969992e-01 -4.07587700e-02
-1.71474814e+00 2.16729879e-01 -2.12544024e-01 -3.21629167e-01
-3.81186396e-01 -2.43402213e-01 -2.97508426e-02 -4.82545674e-01
6.15464509e-01 4.54641312e-01 -2.47465655e-01 -1.72930971e-01
7.02341855e-01 6.16506338e-01 9.59348261e-01 -2.87903965e-01
5.39970815e-01 1.12168956e+00 2.96933949e-01 -1.05835509e+00
-1.11227930e+00 -7.48003185e-01 -5.37460268e-01 -5.36874175e-01
1.12228453e+00 -1.02281237e+00 -8.38098109e-01 7.44868815e-01
-1.13716686e+00 -7.86859334e-01 -4.98958379e-01 2.08241314e-01
-7.55525172e-01 5.39659381e-01 -9.28354919e-01 -7.92982399e-01
-5.71359754e-01 -1.19054735e+00 1.30313909e+00 6.24029338e-01
6.36900544e-01 -8.72399330e-01 -2.18406141e-01 7.76737690e-01
3.68371308e-01 -9.16669331e-03 5.41173995e-01 6.90274477e-01
-9.32481885e-01 1.23922981e-01 -9.28955495e-01 7.73180902e-01
-1.13490252e-02 -6.53287321e-02 -1.25930595e+00 -8.28475058e-02
4.27703932e-02 -4.43099648e-01 1.12426031e+00 9.84586000e-01
1.33059216e+00 1.85729146e-01 9.18792561e-02 1.28271723e+00
1.78654218e+00 9.54097584e-02 8.46682727e-01 3.23299944e-01
6.97275996e-01 5.55995286e-01 7.94294417e-01 3.85207474e-01
4.38941419e-01 3.94477516e-01 5.99879801e-01 -9.00630236e-01
-6.70728087e-01 5.49269840e-02 1.15767248e-01 3.60784709e-01
-2.26611361e-01 -5.63364662e-02 -5.24816930e-01 4.86713022e-01
-1.30055583e+00 -1.00885713e+00 -4.83101875e-01 1.90266609e+00
9.90719378e-01 -1.12666257e-01 -2.13455871e-01 1.18219860e-01
7.93160260e-01 1.37540489e-01 -7.63921022e-01 -1.84777156e-01
-3.72019112e-01 2.49741435e-01 8.48874271e-01 8.88439357e-01
-1.01869977e+00 1.10017645e+00 5.25038195e+00 9.44365561e-01
-1.06425107e+00 4.79177505e-01 1.03763759e+00 -1.37191817e-01
-3.95479947e-02 5.19247428e-02 -8.12685907e-01 5.50968766e-01
3.78714710e-01 1.38112148e-02 5.48314035e-01 4.44322765e-01
6.63683236e-01 -6.22701228e-01 -5.56890607e-01 1.29940426e+00
3.41179013e-01 -1.27696466e+00 -2.49954462e-01 -3.84543002e-01
1.18062878e+00 2.14246452e-01 1.14769213e-01 -3.92721653e-01
-3.42240073e-02 -3.73802483e-01 5.57122052e-01 6.04267180e-01
1.06063497e+00 -7.01551259e-01 3.16651940e-01 1.04005530e-01
-1.23149300e+00 -1.24209791e-01 -5.40674150e-01 1.27412617e-01
5.39481401e-01 8.17859590e-01 -5.97898424e-01 3.52569550e-01
8.34379494e-01 9.43801224e-01 -6.56055391e-01 9.50328171e-01
-5.02034724e-01 3.59626681e-01 -2.50544786e-01 3.60713899e-01
1.69905707e-01 -4.18544978e-01 3.29518259e-01 1.20028579e+00
-5.83964260e-03 4.98951077e-01 1.21451743e-01 9.86737370e-01
-1.10744625e-01 -1.30501434e-01 -3.30790341e-01 2.92319447e-01
1.60195548e-02 1.50967753e+00 -1.06760418e+00 -2.32269794e-01
-5.20252407e-01 1.23309362e+00 -2.66690165e-01 5.86311758e-01
-6.04850948e-01 -5.98866701e-01 3.24851185e-01 3.93026918e-01
-1.53383687e-01 -1.71163976e-01 -5.51716030e-01 -1.01647902e+00
1.16533358e-02 -2.67900437e-01 2.51715153e-01 -1.46081567e+00
-9.81361926e-01 5.36216497e-01 -2.52898097e-01 -1.07223880e+00
2.18123451e-01 -5.06708741e-01 -6.33492172e-01 7.70224571e-01
-2.39866829e+00 -1.39779592e+00 -8.11601758e-01 9.02884960e-01
9.10700619e-01 3.18674058e-01 7.40984548e-03 5.59738398e-01
-4.60723549e-01 1.09355859e-01 6.91766441e-02 -1.25627324e-01
7.38100708e-01 -8.64365101e-01 1.58484086e-01 1.62310100e+00
-3.22823226e-01 1.73333302e-01 5.70315301e-01 -5.27339995e-01
-1.17744875e+00 -1.35356438e+00 5.27713299e-01 -9.18915868e-02
2.39996850e-01 -1.23668082e-01 -6.31823063e-01 3.34732711e-01
2.70456076e-01 4.04606760e-01 3.21274608e-01 -6.16413116e-01
1.37879133e-01 -3.32658708e-01 -1.29167223e+00 5.52852750e-01
1.06769085e+00 -5.51767588e-01 -1.23248838e-01 4.04705793e-01
7.40905344e-01 -4.07415867e-01 -3.69240671e-01 4.11196619e-01
3.98537785e-01 -1.47121823e+00 1.29454041e+00 2.02121481e-01
7.49008775e-01 -5.08773983e-01 3.54543072e-03 -9.81546998e-01
-6.06794469e-02 -5.61453760e-01 2.80223578e-01 1.02679837e+00
-1.93037353e-02 -6.32215738e-01 9.14248645e-01 5.85227609e-01
-3.25877517e-01 -6.93806291e-01 -4.99966472e-01 -4.23417985e-01
-4.88100141e-01 -7.21945941e-01 1.66244373e-01 8.23982000e-01
-7.88090169e-01 1.22542746e-01 -3.43524665e-01 2.19962269e-01
1.26301205e+00 3.52261066e-01 6.52751148e-01 -7.26385534e-01
-1.73516378e-01 -3.12260836e-01 -2.64010310e-01 -9.60244179e-01
5.90193234e-02 -8.70631278e-01 3.17175359e-01 -1.79341257e+00
5.20771801e-01 -1.85106188e-01 -1.61720291e-01 4.97134835e-01
-3.03060025e-01 8.69603515e-01 1.18544430e-01 -8.95622820e-02
-7.88412273e-01 5.15194416e-01 1.46400964e+00 -1.25121534e-01
-4.44378108e-01 -2.44848087e-01 -6.32542253e-01 8.89546454e-01
7.80309558e-01 -2.52390683e-01 -5.28854430e-01 -3.74014646e-01
2.99599022e-01 -2.50640512e-01 6.80086911e-01 -8.66733551e-01
6.43758953e-01 -2.12981343e-01 4.08886820e-01 -6.53889656e-01
3.90747577e-01 -6.64363205e-01 -2.96514630e-01 3.62701952e-01
-2.04351291e-01 -5.43524384e-01 7.24193156e-02 4.33431655e-01
-4.81660403e-02 -2.21587762e-01 1.46016788e+00 -2.10569888e-01
-1.23236668e+00 4.76311117e-01 -1.44314868e-02 8.96502361e-02
1.12685955e+00 -3.74508023e-01 -6.97187304e-01 -2.30759755e-01
-3.41075301e-01 1.61287323e-01 6.94248378e-01 -9.24248025e-02
1.16071844e+00 -8.09512913e-01 -6.73394442e-01 4.40900445e-01
2.43107840e-01 -1.06333857e-02 8.36757004e-01 8.90867293e-01
-7.72788882e-01 -3.92993307e-03 -3.96645367e-01 -5.25973499e-01
-1.59101772e+00 3.19672793e-01 4.35934514e-01 2.71607101e-01
-7.65180945e-01 9.56802130e-01 4.35524553e-01 5.69676422e-02
-1.03396684e-01 -4.41386610e-01 3.41516584e-01 -4.35857207e-01
7.17344999e-01 5.89990854e-01 1.40206292e-01 -4.75633115e-01
-1.05317179e-02 9.13039625e-01 6.74841776e-02 2.11792916e-01
1.52446306e+00 -8.14828217e-01 -2.59173036e-01 1.73753634e-01
1.13772547e+00 -3.54188174e-01 -1.50974619e+00 -2.11792454e-01
-8.32044125e-01 -1.00165689e+00 7.47637808e-01 -7.01878369e-01
-1.35167396e+00 1.02670479e+00 8.74103487e-01 -9.31962356e-02
1.92916322e+00 -1.24136373e-01 1.07900429e+00 1.83695406e-01
1.70102805e-01 -1.27312362e+00 1.96209237e-01 1.92127451e-01
3.77132475e-01 -1.72526395e+00 2.66211014e-02 -6.21921420e-01
-3.79460454e-01 1.04798484e+00 5.65723836e-01 1.09071448e-01
5.99625528e-01 4.09891635e-01 1.84991241e-01 -2.48891130e-01
-6.16747260e-01 -7.81776428e-01 1.27548769e-01 7.34482884e-01
1.73776850e-01 -3.39233428e-01 -9.39180404e-02 8.33726078e-02
3.58926803e-01 2.06309751e-01 7.98586130e-01 6.89406574e-01
-6.81059897e-01 -6.53765023e-01 -3.16265881e-01 2.51411378e-01
-4.73361492e-01 -3.89963776e-01 -1.53478133e-02 2.16410249e-01
4.53603417e-01 1.16212606e+00 -7.16236830e-02 -4.09012437e-02
6.98745772e-02 -3.97376955e-01 5.55070996e-01 -4.37770963e-01
-2.05727041e-01 2.16038540e-01 -3.97564203e-01 -8.55325520e-01
-7.97819436e-01 -2.94758499e-01 -1.14545345e+00 -2.02097997e-01
-3.02770227e-01 -5.03084302e-01 9.07339454e-01 4.99211252e-01
1.35359332e-01 4.48592156e-01 7.93707073e-01 -1.21813107e+00
2.56171048e-01 -4.64588910e-01 -9.09236252e-01 5.48132777e-01
4.15820301e-01 -5.00358462e-01 -5.08305848e-01 4.92343813e-01] | [10.668170928955078, -2.567002296447754] |
38432f57-7dec-473c-8966-ae54ca63e36e | rethinking-cnn-models-for-audio | 2007.11154 | null | https://arxiv.org/abs/2007.11154v2 | https://arxiv.org/pdf/2007.11154v2.pdf | Rethinking CNN Models for Audio Classification | In this paper, we show that ImageNet-Pretrained standard deep CNN models can be used as strong baseline networks for audio classification. Even though there is a significant difference between audio Spectrogram and standard ImageNet image samples, transfer learning assumptions still hold firmly. To understand what enables the ImageNet pretrained models to learn useful audio representations, we systematically study how much of pretrained weights is useful for learning spectrograms. We show (1) that for a given standard model using pretrained weights is better than using randomly initialized weights (2) qualitative results of what the CNNs learn from the spectrograms by visualizing the gradients. Besides, we show that even though we use the pretrained model weights for initialization, there is variance in performance in various output runs of the same model. This variance in performance is due to the random initialization of linear classification layer and random mini-batch orderings in multiple runs. This brings significant diversity to build stronger ensemble models with an overall improvement in accuracy. An ensemble of ImageNet pretrained DenseNet achieves 92.89% validation accuracy on the ESC-50 dataset and 87.42% validation accuracy on the UrbanSound8K dataset which is the current state-of-the-art on both of these datasets. | ['Dipika Singhania', 'Angela Yao', 'Kamalesh Palanisamy'] | 2020-07-22 | null | null | null | null | ['environmental-sound-classification'] | ['audio'] | [ 4.17179130e-02 -1.59650221e-01 1.03692360e-01 -4.02991444e-01
-7.48656511e-01 -4.56017166e-01 4.12255973e-01 -2.76896656e-01
-5.83324492e-01 4.56322849e-01 4.46606666e-01 4.97510750e-03
-3.37298401e-02 -7.89185345e-01 -8.14651966e-01 -6.78279638e-01
-5.59734643e-01 -2.83147907e-03 2.03390613e-01 -3.80427390e-01
-2.07998708e-01 1.46991521e-01 -1.61963117e+00 8.15242767e-01
1.09255217e-01 1.27738440e+00 -1.47209793e-01 9.99920726e-01
3.70342553e-01 8.49311709e-01 -8.76508296e-01 -1.17608435e-01
2.09242329e-01 -3.76346678e-01 -8.98640335e-01 -1.71209082e-01
8.18258703e-01 -2.94481456e-01 -4.19297278e-01 8.50033879e-01
6.39761090e-01 1.54641867e-01 6.37638509e-01 -1.29210925e+00
-4.98020262e-01 1.04965127e+00 -2.93624401e-01 4.78792250e-01
3.46755534e-02 4.89247411e-01 1.39897478e+00 -7.02436209e-01
3.05933923e-01 1.20927107e+00 9.55102861e-01 4.05861676e-01
-1.33748639e+00 -1.07279623e+00 6.97896034e-02 2.00096488e-01
-1.35010207e+00 -5.06099522e-01 5.87975323e-01 -6.65332496e-01
1.02604818e+00 6.24598749e-02 8.40707421e-01 1.24986696e+00
1.43894311e-02 5.13120413e-01 8.89670074e-01 -2.87000090e-01
-2.06540320e-02 1.96710348e-01 2.39016026e-01 4.13293213e-01
-4.40975651e-02 3.02590281e-01 -7.26851046e-01 3.08505818e-02
6.29899085e-01 -4.03225213e-01 -3.95236075e-01 6.27431646e-02
-1.17019451e+00 7.43562698e-01 8.07028115e-01 4.65836585e-01
-2.72106919e-02 7.73572683e-01 6.55317128e-01 5.48263013e-01
4.76164848e-01 5.22896707e-01 -5.53201020e-01 -3.60085517e-01
-1.00402141e+00 1.77879900e-01 5.23274124e-01 4.24828738e-01
6.57620370e-01 7.69851983e-01 2.58953929e-01 1.00545704e+00
7.16638193e-02 2.14632511e-01 5.75322986e-01 -1.03612351e+00
3.25280666e-01 7.96732530e-02 -4.49666560e-01 -9.16624069e-01
-3.47505599e-01 -9.82354045e-01 -7.23178029e-01 3.42291534e-01
5.23161650e-01 -4.56080288e-01 -9.07375455e-01 1.87892127e+00
-4.32488948e-01 4.86875713e-01 8.42067599e-02 8.47932100e-01
9.26065624e-01 7.86309958e-01 1.10400811e-01 5.11825144e-01
1.00235450e+00 -8.85408401e-01 -3.01564991e-01 -1.62349209e-01
5.11681020e-01 -6.72129810e-01 1.21565151e+00 4.98936057e-01
-9.37366068e-01 -9.80935156e-01 -1.39580226e+00 1.01552740e-01
-3.86266857e-01 2.11009845e-01 5.81564367e-01 6.33634925e-01
-1.28846538e+00 9.79707956e-01 -6.19500756e-01 -3.45234603e-01
5.08019447e-01 4.91632938e-01 -1.86872661e-01 3.58650476e-01
-1.17795682e+00 6.55789614e-01 5.61801672e-01 -1.69310361e-01
-1.40525150e+00 -1.09397590e+00 -7.03100562e-01 3.08864713e-01
-3.23728994e-02 -4.46793258e-01 1.51472449e+00 -1.29589701e+00
-1.37380505e+00 6.14865482e-01 3.25806201e-01 -9.29577172e-01
3.25047284e-01 -3.67688715e-01 -5.01873195e-01 8.70589763e-02
-2.35531434e-01 1.18353629e+00 8.14322293e-01 -1.12561917e+00
-6.91087902e-01 1.76386341e-01 5.62002659e-02 -5.18139675e-02
-4.46558267e-01 -4.73629236e-02 3.77714634e-02 -6.44688368e-01
-2.50091642e-01 -9.04202938e-01 1.33956790e-01 -3.62085432e-01
-9.54863057e-02 -1.15408324e-01 7.14135110e-01 -5.23951054e-01
1.16952944e+00 -2.42959714e+00 -1.42213553e-01 1.79319322e-01
2.12646484e-01 3.70756909e-02 -4.38334405e-01 2.01079637e-01
-5.87541580e-01 3.74696791e-01 2.67523639e-02 -1.49589926e-01
-7.24808350e-02 1.12916157e-02 -5.41020870e-01 2.36825824e-01
3.76511663e-01 7.36858308e-01 -8.89355600e-01 9.60735008e-02
1.36942923e-01 6.28547192e-01 -8.07707608e-01 -2.09514990e-01
2.77147666e-02 1.85994014e-01 2.59658158e-01 2.14334339e-01
3.50369841e-01 -6.85929954e-02 -1.01302080e-01 -2.77719766e-01
4.25375402e-02 6.52027488e-01 -1.01663136e+00 1.67004597e+00
-4.23363805e-01 1.12206995e+00 -1.25111684e-01 -1.08417368e+00
7.41448283e-01 4.55695689e-01 4.67022479e-01 -4.94290411e-01
2.07122445e-01 2.07158759e-01 6.70541227e-01 -2.12541103e-01
3.58876586e-01 -1.44945636e-01 1.36083707e-01 4.52205777e-01
8.03105474e-01 -2.54849285e-01 7.68377632e-02 1.32847384e-01
9.75155354e-01 -1.79750055e-01 -2.94252813e-01 -4.25172895e-01
4.95330617e-02 -1.30747274e-01 2.82608032e-01 7.20946789e-01
-6.55145720e-02 8.01681399e-01 4.48965311e-01 -5.69510758e-01
-1.11651027e+00 -9.23032284e-01 -2.01088905e-01 1.49493265e+00
-5.31243682e-01 -7.40910292e-01 -6.59017861e-01 -4.54553753e-01
-6.93378001e-02 5.05293071e-01 -8.14318359e-01 -2.91515708e-01
-2.41206065e-01 -8.01314771e-01 1.26561856e+00 7.34040439e-01
6.67829931e-01 -1.21111321e+00 -5.68584025e-01 2.57212967e-01
3.19856480e-02 -1.01851141e+00 -8.16948935e-02 6.62544310e-01
-8.04023027e-01 -8.01848292e-01 -6.41530573e-01 -6.89519644e-01
4.93694991e-02 1.01058155e-01 1.34893465e+00 1.34624451e-01
-1.29276782e-01 4.58850682e-01 -1.82401448e-01 -8.49728346e-01
-2.11830691e-01 4.89091635e-01 7.85307214e-02 -9.46586132e-02
2.95392990e-01 -9.14077818e-01 -5.02597928e-01 2.02870462e-02
-7.10130095e-01 -1.21656209e-01 1.40352860e-01 7.24853516e-01
2.82332778e-01 1.62714049e-01 5.39037049e-01 -5.69340587e-01
7.27696955e-01 -4.95086610e-01 -2.45414034e-01 -1.97213113e-01
-2.46653691e-01 -2.27668807e-02 4.29463118e-01 -5.10693967e-01
-5.87211251e-01 -1.22316368e-01 -2.66494811e-01 -4.26574796e-01
-2.78002143e-01 5.18597662e-01 4.02731866e-01 1.48460060e-01
1.04223871e+00 -1.38360634e-01 -1.26071885e-01 -4.88201529e-01
1.56756356e-01 4.58399564e-01 5.61590195e-01 -6.53016686e-01
5.69169998e-01 2.91470885e-01 -4.40962940e-01 -9.67897236e-01
-1.01526368e+00 -2.66559541e-01 -4.05493975e-01 -3.41086507e-01
1.02454376e+00 -1.30036402e+00 -6.15585387e-01 4.81371939e-01
-9.55649912e-01 -9.31575477e-01 -4.74886537e-01 5.55254698e-01
-3.53919357e-01 -3.85874778e-01 -7.58576870e-01 -6.70547783e-01
-7.08818063e-03 -1.22119379e+00 8.66942942e-01 2.85525504e-03
-5.50141573e-01 -1.11069119e+00 1.78550720e-01 -1.96987078e-01
5.71610868e-01 1.92309633e-01 8.80548954e-01 -6.48551524e-01
-2.63572365e-01 1.58110231e-01 -1.61226586e-01 7.48619974e-01
1.26949111e-02 2.87654221e-01 -1.73696804e+00 -2.19953403e-01
-3.35760385e-01 -6.26494110e-01 1.37878823e+00 4.56353515e-01
1.41101933e+00 -7.19873831e-02 1.67436704e-01 7.07952082e-01
1.21222830e+00 2.45207936e-01 6.95057154e-01 5.64058781e-01
4.17437673e-01 3.01076919e-01 -1.80110365e-01 4.91751507e-02
1.08529620e-01 5.79265118e-01 2.98962861e-01 -2.74885535e-01
-3.82012665e-01 -3.16721112e-01 6.46138966e-01 9.57878530e-01
-2.47817189e-01 1.15683533e-01 -1.08661258e+00 5.71856499e-01
-1.62092316e+00 -1.04748154e+00 7.28016794e-02 1.91439307e+00
8.03401530e-01 3.61672848e-01 3.71541113e-01 6.77534401e-01
3.36069733e-01 2.60850608e-01 -1.75795853e-01 -5.01682699e-01
-1.52388841e-01 7.14543462e-01 3.01206976e-01 3.47224027e-01
-1.27763140e+00 7.42848217e-01 7.29340076e+00 8.08662176e-01
-1.62559259e+00 1.03939295e-01 7.84268141e-01 -3.33016545e-01
1.67528823e-01 -3.51978987e-01 -6.45890713e-01 2.20066458e-01
1.50612128e+00 2.44878419e-03 5.23790538e-01 8.88744295e-01
-1.19055204e-01 5.25470227e-02 -1.18148768e+00 1.16386545e+00
-2.84307390e-01 -1.50703287e+00 4.76175211e-02 5.93881980e-02
8.93469810e-01 3.98250937e-01 4.08371955e-01 6.54824555e-01
5.45620203e-01 -1.31913698e+00 9.91962850e-01 1.67181358e-01
7.61980593e-01 -9.03975725e-01 7.38996267e-01 -1.15939833e-01
-1.14210200e+00 -1.56706095e-01 -4.94016826e-01 -2.92173982e-01
-3.67811263e-01 4.79218006e-01 -9.29696977e-01 3.00719470e-01
1.16878021e+00 8.50260496e-01 -7.08436549e-01 1.04705834e+00
-1.69323578e-01 1.21521664e+00 -4.39329743e-01 1.84029087e-01
5.74358523e-01 2.89155602e-01 1.61917761e-01 1.70229733e+00
2.15054289e-01 -3.57698649e-01 -7.13282637e-03 6.65475667e-01
-3.14848304e-01 -1.48075685e-01 -6.58273518e-01 -8.87198672e-02
1.65315568e-01 1.17509353e+00 -6.37005985e-01 -4.95811969e-01
-2.98180789e-01 4.22950566e-01 2.73635119e-01 5.12709379e-01
-7.54787445e-01 -3.87010306e-01 8.67461503e-01 1.50116086e-01
3.39453012e-01 -2.47992933e-01 -4.88026530e-01 -8.77978742e-01
-3.69442880e-01 -9.89889085e-01 3.65539014e-01 -1.07037818e+00
-9.90934789e-01 7.23985732e-01 2.89327279e-02 -1.17200828e+00
-2.38077298e-01 -6.90905273e-01 -9.06810701e-01 6.24830127e-01
-1.27066553e+00 -8.17136407e-01 -3.75995874e-01 5.52680254e-01
5.36260188e-01 -3.65679860e-01 9.80735004e-01 4.27106321e-01
-4.18805242e-01 6.57840729e-01 -1.28960788e-01 6.14431083e-01
7.68757105e-01 -1.41835332e+00 2.87525117e-01 5.48492134e-01
7.40044236e-01 6.72804475e-01 6.12349451e-01 -2.42837034e-02
-6.34967327e-01 -1.14442563e+00 2.96078116e-01 -3.08193177e-01
8.79046619e-01 -4.14490521e-01 -9.31723654e-01 8.23536694e-01
5.91217995e-01 -1.21858291e-01 7.82235324e-01 4.26735342e-01
-6.40748262e-01 -3.84274215e-01 -6.25487030e-01 3.78836751e-01
8.52286041e-01 -7.51588941e-01 -4.13785309e-01 9.80012417e-02
7.67910659e-01 -2.93297499e-01 -7.76763797e-01 4.15597469e-01
7.64586687e-01 -1.08009160e+00 9.18061793e-01 -7.88724005e-01
6.97959721e-01 -1.91361219e-01 -2.37179115e-01 -1.76574254e+00
-2.63769507e-01 -2.97624767e-01 2.36815184e-01 1.15386498e+00
7.52575576e-01 -3.71459007e-01 5.58892608e-01 2.53785849e-01
-2.59280950e-01 -6.59021080e-01 -6.35691226e-01 -8.52135539e-01
3.01573962e-01 -1.10914350e+00 5.80234170e-01 9.32548225e-01
-1.71844453e-01 6.15879059e-01 -1.13998517e-01 -1.54845089e-01
2.00485349e-01 -5.48604906e-01 9.63302970e-01 -1.23979306e+00
-3.38096708e-01 -5.36832809e-01 -7.09520996e-01 -6.36297643e-01
8.51454362e-02 -1.01590419e+00 -1.09432042e-01 -1.09473968e+00
5.21921888e-02 -4.16089296e-01 -7.42835224e-01 7.22086072e-01
2.77192712e-01 6.99406624e-01 4.67129588e-01 -7.20613599e-02
-4.06836241e-01 2.06655249e-01 8.51156533e-01 -3.05882990e-01
-3.12251508e-01 -2.47047752e-01 -7.34254122e-01 7.95261383e-01
1.08266640e+00 -4.63063240e-01 -5.49277961e-01 -6.87740386e-01
2.62265146e-01 -3.82474422e-01 6.80843830e-01 -1.71152985e+00
-5.33555411e-02 3.79705697e-01 7.12416828e-01 -2.76038378e-01
4.54935730e-01 -5.81440508e-01 9.79397520e-02 3.39341104e-01
-7.40645647e-01 -9.59290192e-02 7.30297863e-01 2.52605170e-01
-5.07815778e-01 -1.54549807e-01 8.71910453e-01 -1.36305720e-01
-6.28664017e-01 1.16640456e-01 -5.04779220e-01 1.30695000e-01
2.88262069e-01 -1.46364883e-01 -2.43006498e-01 -7.57574141e-01
-1.01038504e+00 -2.04017729e-01 -1.96905993e-02 4.96150881e-01
2.94636965e-01 -1.48779154e+00 -7.92131901e-01 1.01697803e-01
-6.44007698e-02 -1.61254361e-01 5.96915521e-02 7.36366570e-01
-2.06171945e-01 4.48824286e-01 -3.50182444e-01 -9.49637711e-01
-1.18042517e+00 -6.87121674e-02 5.89486599e-01 -5.35947606e-02
-3.54331285e-01 1.07027543e+00 3.13057095e-01 -2.22552136e-01
5.00290930e-01 -7.20640063e-01 -1.23580158e-01 2.23919854e-01
5.63723445e-01 1.21053301e-01 3.35330456e-01 -4.12602216e-01
-1.29990101e-01 4.32155401e-01 3.00909905e-03 -4.06372100e-01
1.69366407e+00 4.92101997e-01 2.74423808e-01 9.00577307e-01
1.44080579e+00 -1.50840849e-01 -1.35524523e+00 2.71376595e-02
-1.26899868e-01 -5.41990437e-03 2.62514114e-01 -7.67871797e-01
-1.46983051e+00 1.52541411e+00 8.67710292e-01 3.50090176e-01
9.30136025e-01 -2.78153390e-01 3.31484348e-01 4.54352409e-01
1.26119733e-01 -9.08397973e-01 3.01866829e-01 8.84290516e-01
1.06839669e+00 -9.35852528e-01 -1.29114732e-01 4.24532406e-02
-7.73331463e-01 1.34701765e+00 6.65259302e-01 -5.75181246e-01
7.95159876e-01 3.62364531e-01 1.36404365e-01 -1.07134297e-01
-8.53342354e-01 -2.32263327e-01 3.99565041e-01 6.76075041e-01
7.63462961e-01 -6.31962940e-02 4.33628172e-01 6.99425161e-01
-9.14451480e-01 -2.83662640e-02 3.41524154e-01 5.88610232e-01
-4.26807761e-01 -6.88673198e-01 -1.81911692e-01 3.75233501e-01
-5.91046393e-01 -2.22459197e-01 -5.10363877e-01 1.07026780e+00
2.28507519e-01 7.84402132e-01 2.86269903e-01 -8.88731956e-01
2.57848024e-01 4.68365610e-01 4.13556397e-01 -6.72906697e-01
-9.91774857e-01 1.62198409e-01 1.38124704e-01 -2.60311037e-01
-4.37758029e-01 -2.23706022e-01 -1.14825392e+00 -2.49451727e-01
-2.26614147e-01 2.06982344e-01 5.78052580e-01 6.52998626e-01
1.98396608e-01 1.08402145e+00 3.44756514e-01 -1.12490773e+00
-2.62359321e-01 -1.28296196e+00 -4.73612279e-01 2.68023729e-01
4.92877871e-01 -6.02935672e-01 -5.50683975e-01 4.02430207e-01] | [15.383883476257324, 5.267778396606445] |
e61baf12-0c9c-465c-b361-3017ded76590 | analysis-of-cardiovascular-changes-caused-by | 1912.05083 | null | http://arxiv.org/abs/1912.05083v1 | http://arxiv.org/pdf/1912.05083v1.pdf | Analysis of Cardiovascular Changes Caused by Epileptic Seizures in Human Photoplethysmogram Signal | Objectives: This study examines human Photoplethysmogram (PPG) along with
Electrocardiogram (ECG) signals to study cardiac autonomic imbalance in
epileptic seizures. The significance and the prevalence of changes in PPG
morphological parameters have been investigated to find common patterns among
subjects. Alterations in cardiovascular parameters measured by PPG/ECG signals
are used to train a neural network based on LSTM for automatic seizure
detection. Methods: Electroencephalogram (EEG), ECG, and PPG signals from 12
different subjects ( 8 males;4 females;age 34.3$\pm$ 13.8) were recorded
including 57 seizures and 101 hours of inter-ictal data. 12 PPG features
significantly changing due to epileptic seizures were extracted and normalized
based on a proposed z-score metric. 7 feature are heart rate variability
related and 5 features hemodynamic related. Results: A consistent pattern of
ictal change was observed for all the features across the subjects/seziures.
The proposed seizure detector is subject independent and works for both
nocturnal and diurnal seizures. With an average of 0.52 false alarms per hour,
positive predictive value of $43\%$ and sensitivity of $92\%$, the new proposed
hemodynamic based seizure detector shows improvement over the the heart rate
variability based detector. Conclusion: The cardiac autonomic imbalance due to
seizure manifests itself in variations of peripheral hemodynamics measured by
PPG signal, suggesting vasoconstriction in limbs. These variations can be used
on a consumer seizure detecting devices with optical sensors for seizure
detection. Significance: The stereotyped pattern is common among all the
subjects which can help understand the mechanism of cardiac autonomic imbalance
induced by epileptic seizures. | [] | 2019-12-11 | null | null | null | null | ['heart-rate-variability'] | ['medical'] | [-1.72716647e-01 -5.24814904e-01 4.02622670e-01 -2.77357191e-01
1.53289974e-01 -2.28942782e-01 -5.16128652e-02 5.68134040e-02
-3.43630522e-01 9.54205155e-01 -8.88250694e-02 5.15400246e-02
-1.86545849e-01 -3.44701022e-01 -5.23893908e-02 -8.42939734e-01
-8.55923057e-01 -2.00767830e-01 -6.61332667e-01 -1.20314568e-01
3.13359052e-01 5.28709292e-01 -1.06674337e+00 -8.17114636e-02
5.81537783e-01 1.31750393e+00 -2.91187435e-01 6.56896532e-01
2.66320467e-01 1.82478115e-01 -1.14442253e+00 5.18045485e-01
3.82551998e-01 -9.42719758e-01 6.71835542e-02 -2.19789907e-01
2.60993280e-02 -7.70290419e-02 -6.75925985e-02 8.35285485e-01
1.28683031e+00 -9.49755311e-02 5.26588440e-01 -1.29321420e+00
-9.13917795e-02 2.94455349e-01 -5.83157599e-01 9.79059100e-01
2.38183394e-01 3.80432367e-01 -3.09939757e-02 -4.37627912e-01
-1.29593894e-01 5.28471172e-01 8.77138972e-01 5.67585945e-01
-1.12277794e+00 -1.25812984e+00 -5.36122739e-01 4.65407640e-01
-1.60510552e+00 -4.04495150e-01 8.81726861e-01 -6.02739394e-01
1.31830227e+00 2.60017157e-01 1.31008697e+00 9.26234782e-01
9.55298305e-01 -2.85923213e-01 1.56664705e+00 -1.43829420e-01
2.63304651e-01 2.59371579e-01 3.03078562e-01 3.20220619e-01
3.38289261e-01 3.03025275e-01 -9.11103308e-01 -1.91055045e-01
5.48014045e-01 -1.57114312e-01 -6.06986225e-01 3.34907949e-01
-8.84285867e-01 4.27080214e-01 -2.05445930e-01 6.09837472e-01
-7.87778735e-01 2.39316523e-01 6.83305502e-01 7.71728694e-01
1.51888922e-01 7.38833189e-01 -8.14331591e-01 -7.16304779e-01
-8.02818239e-01 -1.81172669e-01 7.89482236e-01 3.15415561e-01
3.44938993e-01 5.91741800e-01 -4.66296487e-02 6.84063435e-01
1.26488328e-01 6.88239813e-01 1.11900413e+00 -5.20230532e-01
-1.40030131e-01 6.59830511e-01 -2.90090919e-01 -1.07970881e+00
-1.01162648e+00 -4.53054547e-01 -9.73324716e-01 4.43491787e-02
5.24740294e-02 -4.91928279e-01 -5.51801383e-01 1.70687735e+00
-2.76013166e-01 2.83927053e-01 1.45706743e-01 6.98738277e-01
9.84549522e-01 9.15024132e-02 4.01531518e-01 -8.19643497e-01
1.43616343e+00 1.63600147e-01 -8.44125688e-01 7.87293166e-02
2.92987764e-01 -4.23455805e-01 5.24694979e-01 4.74803597e-01
-6.52821004e-01 -3.16768050e-01 -1.11093962e+00 9.10814583e-01
-3.59205514e-01 -6.90783635e-02 1.95952371e-01 1.13053524e+00
-1.16911066e+00 8.52005482e-01 -9.44347978e-01 -6.24131739e-01
1.37300104e-01 6.48700535e-01 -5.79131283e-02 9.38873291e-01
-1.37255895e+00 9.94933784e-01 2.74583817e-01 1.73006907e-01
-3.28528941e-01 -6.14751577e-01 -5.25703549e-01 3.82861979e-02
-4.96234357e-01 -5.34543633e-01 4.25165325e-01 -6.99719965e-01
-1.64931536e+00 6.54510081e-01 -9.59538296e-02 -3.54548514e-01
2.49515221e-01 2.50926256e-01 -9.65149224e-01 1.00934602e-01
-2.15145752e-01 1.01940393e-01 5.89258015e-01 -2.05881000e-01
-1.25792161e-01 -7.43458569e-01 -8.95118773e-01 3.42180073e-01
-8.50084648e-02 1.80697024e-01 5.78041196e-01 -3.56542706e-01
1.89058959e-01 -5.48518062e-01 1.76273897e-01 -6.08404219e-01
-3.11374843e-01 4.59649265e-02 8.11768830e-01 -6.83669746e-01
1.27913046e+00 -1.89093816e+00 -5.49986959e-01 4.57744390e-01
3.60834569e-01 3.60461213e-02 1.89568818e-01 2.86232829e-01
-5.56539655e-01 3.12300678e-02 -2.85137713e-01 3.40186745e-01
-9.75836441e-02 -2.05246061e-01 1.72454014e-03 8.15907001e-01
-2.69349813e-01 8.88599515e-01 -3.12470317e-01 8.56552720e-02
5.00121713e-01 5.74927270e-01 3.09053123e-01 9.95316431e-02
5.60021579e-01 8.16850007e-01 -1.61841318e-01 7.44533062e-01
5.09593785e-01 2.86907077e-01 -1.92058638e-01 -1.24977283e-01
-2.44764447e-01 -1.26451824e-03 -9.29425538e-01 1.27944028e+00
-4.01066877e-02 9.75519061e-01 -4.92047071e-01 -1.05286026e+00
1.52948797e+00 8.59260082e-01 6.68335259e-01 -1.01112247e+00
5.82124531e-01 2.62318105e-01 6.21671915e-01 -8.68700564e-01
-4.84047204e-01 -4.23466295e-01 7.69078285e-02 6.18873537e-01
-1.23656236e-01 6.36939183e-02 -4.57688682e-02 -4.11703587e-01
1.24286139e+00 -1.45308167e-01 8.15694690e-01 -7.53434420e-01
6.70460761e-01 -4.84240949e-01 7.89501429e-01 6.49904966e-01
-6.00196660e-01 3.49140391e-02 5.02501488e-01 -8.28429103e-01
-4.58773553e-01 -7.22626984e-01 -5.56968749e-01 3.47089291e-01
-6.68856800e-02 7.44994879e-02 -5.70422709e-01 1.25734597e-01
-7.34587535e-02 3.99962246e-01 -4.94317323e-01 -2.54868805e-01
-2.72766173e-01 -1.44808054e+00 7.50243485e-01 4.63542551e-01
6.09875619e-01 -1.61233699e+00 -1.57338071e+00 5.64281046e-01
1.80996746e-01 -7.71198213e-01 1.69525832e-01 4.60619897e-01
-9.74021435e-01 -1.07153380e+00 -3.82932663e-01 -2.65671462e-01
1.74976468e-01 -7.95648634e-01 9.11102951e-01 -3.48089427e-01
-9.10914004e-01 4.47170973e-01 -1.71215221e-01 -8.31583560e-01
-8.26930553e-02 -5.31801343e-01 7.41529524e-01 -1.00497968e-01
1.18357980e+00 -1.47606301e+00 -9.23065007e-01 8.77202898e-02
-5.01989089e-02 -6.26498759e-01 4.96887952e-01 2.29588285e-01
4.32393432e-01 -2.86919087e-01 1.13583231e+00 -8.19468200e-02
1.07541192e+00 -3.93823177e-01 -4.44807142e-01 -2.89990276e-01
-9.97539759e-01 -5.11941671e-01 7.57122934e-01 -3.40686560e-01
-1.87795222e-01 -3.36082637e-01 4.61063772e-01 -3.34349632e-01
-5.35047114e-01 1.91589922e-01 6.36117905e-02 -1.00006491e-01
8.78225386e-01 5.97977877e-01 -7.05499500e-02 -9.00793225e-02
-6.98970318e-01 5.53942084e-01 4.67333585e-01 -2.11308412e-02
3.49701524e-01 -2.28262339e-02 3.52351189e-01 -1.16113329e+00
2.09694758e-01 -3.57755631e-01 -4.15475547e-01 -1.50261283e-01
8.86404097e-01 -8.67711604e-01 -1.08363438e+00 6.77539229e-01
-7.79302716e-01 -2.50933111e-01 -8.99784043e-02 1.06885314e+00
-4.10585880e-01 1.83327824e-01 -3.84985507e-01 -1.03194261e+00
-1.20943975e+00 -8.27641010e-01 1.37643993e-01 3.32459450e-01
-7.14071333e-01 -8.18758130e-01 3.88129503e-01 -4.72176909e-01
9.32532370e-01 8.48151386e-01 6.02747381e-01 -8.60151291e-01
1.78098455e-01 -2.49127328e-01 2.63473243e-01 2.69331723e-01
6.52130604e-01 -2.65773624e-01 -1.10191119e+00 -2.02002063e-01
6.29101813e-01 1.57722846e-01 2.66894817e-01 9.07534122e-01
9.02545154e-01 -7.27301389e-02 -4.07256819e-02 7.99075902e-01
1.59926176e+00 1.30210841e+00 6.16969645e-01 1.87209308e-01
2.46610865e-01 9.21153799e-02 -4.77880836e-01 6.84610963e-01
-3.88927311e-02 2.11008266e-02 -4.14970517e-02 1.25458650e-02
3.03986460e-01 5.27241766e-01 5.65782785e-01 6.80991888e-01
-2.90707171e-01 2.02801824e-01 -8.78325343e-01 3.26460719e-01
-1.10199523e+00 -7.90194750e-01 -2.47892365e-01 2.34682298e+00
6.57390594e-01 -3.80998611e-01 2.16868103e-01 2.20870838e-01
8.19698870e-01 -3.27199042e-01 -9.42026496e-01 -1.03016233e+00
-2.37854660e-01 8.34698677e-01 3.95486057e-01 -2.33210232e-02
-6.15514636e-01 2.85370737e-01 6.10789680e+00 -3.07017058e-01
-1.86942244e+00 -2.06791565e-01 6.23376608e-01 -4.33135539e-01
4.74409550e-01 -3.65380853e-01 -3.55711877e-01 8.92687798e-01
1.47947156e+00 -7.17971087e-01 3.58853966e-01 3.53164375e-01
7.79216588e-01 -2.45533705e-01 -1.05453479e+00 1.75656033e+00
2.36867756e-01 -7.95704901e-01 -6.12277985e-01 3.17721963e-02
1.67544350e-01 2.87072480e-01 -1.30677864e-01 1.13507815e-01
-6.76547647e-01 -1.23689175e+00 -1.14459284e-01 8.38052690e-01
1.21394408e+00 -6.12177134e-01 8.15425456e-01 -5.51776774e-02
-1.07299459e+00 -1.06064358e-03 -4.57327031e-02 -3.90223205e-01
-2.16853708e-01 5.07368982e-01 -8.96915913e-01 -2.72697657e-02
9.15437639e-01 7.84466743e-01 -4.03985083e-01 1.11982203e+00
-8.65324810e-02 7.86496282e-01 -6.45195007e-01 -2.07255587e-01
-2.54269719e-01 -1.72561601e-01 9.13885057e-01 9.50289071e-01
5.85757971e-01 4.88745362e-01 -3.74239206e-01 1.22851026e+00
3.36234421e-01 3.29956919e-01 -7.03784943e-01 1.09942228e-01
4.54914391e-01 1.29648626e+00 -7.90984929e-01 -2.72502035e-01
-2.16009945e-01 4.33543235e-01 -7.58910656e-01 2.81005710e-01
-5.17749429e-01 -8.87236536e-01 4.56604838e-01 -7.26346150e-02
-4.84026194e-01 4.17321861e-01 -7.83319056e-01 -9.80510712e-01
2.23017454e-01 -6.82347953e-01 2.62875825e-01 -5.24469793e-01
-9.12544012e-01 8.91154110e-01 -8.92451704e-02 -1.20802867e+00
-5.13216376e-01 -1.77399457e-01 -1.32180750e+00 1.38920987e+00
-1.12338936e+00 -7.96320587e-02 -5.35079658e-01 7.76384711e-01
1.76612243e-01 -4.02060628e-01 1.21866488e+00 -3.41821760e-02
-8.44924808e-01 5.57967126e-01 -5.16461849e-01 -1.42322332e-01
6.07534826e-01 -1.22245860e+00 -4.75293696e-02 8.30433547e-01
-4.63518053e-01 5.67871749e-01 5.78966737e-01 -4.92087543e-01
-9.30092096e-01 -8.00493240e-01 8.65599096e-01 -2.23919868e-01
3.53775799e-01 -9.17829424e-02 -7.43915141e-01 1.77894130e-01
5.02514601e-01 -6.70308918e-02 1.04227889e+00 -4.18949932e-01
4.07849103e-01 -3.92725676e-01 -1.50921261e+00 1.05515577e-01
2.15576679e-01 -1.71851218e-01 -8.86659741e-01 -4.25264724e-02
-1.92207888e-01 -9.62144807e-02 -1.28687763e+00 6.34819448e-01
7.10769176e-01 -1.12016654e+00 1.57389954e-01 -8.69493634e-02
-1.82821080e-01 -6.93459213e-02 2.25298941e-01 -1.50712669e+00
2.34039221e-02 -1.21111345e+00 1.43002346e-01 5.64415455e-01
3.12110662e-01 -1.32731855e+00 5.12403131e-01 9.28611517e-01
-2.39278525e-01 -7.75565028e-01 -8.28459203e-01 -6.34174764e-01
-6.47267923e-02 -3.64517838e-01 4.20336157e-01 1.03824437e+00
7.85863161e-01 2.24217072e-01 -2.19016131e-02 -1.55497510e-02
5.71460664e-01 -1.92330778e-01 1.19456261e-01 -1.36514878e+00
3.59479263e-02 -3.78270537e-01 -9.59227800e-01 4.08766240e-01
-3.38888794e-01 -7.75562108e-01 -3.68060231e-01 -1.13871181e+00
7.87737817e-02 -4.52371351e-02 -8.65248084e-01 8.01593363e-01
3.00466686e-01 4.76576626e-01 -3.24082285e-01 -2.29217216e-01
3.49148780e-01 1.65287673e-01 5.40583849e-01 2.40103170e-01
-1.09833622e+00 -1.79070964e-01 -4.99855638e-01 6.15622163e-01
1.32480717e+00 -4.37540293e-01 -3.35964948e-01 3.45728099e-01
9.09741446e-02 4.07473683e-01 1.71190873e-01 -1.47580922e+00
1.81853399e-01 2.92344034e-01 1.10632193e+00 -4.15913969e-01
1.29945427e-01 -5.02678037e-01 3.94022673e-01 8.85961115e-01
1.82882234e-01 4.42760020e-01 5.73670805e-01 5.22004105e-02
-2.42719233e-01 3.35331947e-01 8.58768284e-01 -2.47328073e-01
-4.58581150e-01 4.41464633e-01 -7.43096292e-01 9.68834311e-02
1.14091539e+00 -8.89083207e-01 -2.46655449e-01 -2.74848819e-01
-1.02990448e+00 -1.19796202e-01 -7.07643703e-02 2.68471390e-01
5.70003331e-01 -9.30193663e-01 -7.63317704e-01 6.94328427e-01
5.31268083e-02 -8.13659787e-01 4.04306829e-01 1.56819022e+00
-4.05487508e-01 5.47018766e-01 -9.41513360e-01 -6.68237805e-01
-1.13390601e+00 -2.57958561e-01 1.04547203e+00 4.59152579e-01
-7.72750020e-01 7.87477612e-01 -1.64563566e-01 5.16044796e-01
8.92419145e-02 -3.26886892e-01 -6.58138692e-01 2.04570055e-01
4.81304795e-01 5.57702184e-01 2.89772451e-01 -2.21993595e-01
-5.88942230e-01 7.26342916e-01 3.29092920e-01 1.32406399e-01
1.31890178e+00 -4.69832383e-02 -5.20819783e-01 5.77480137e-01
1.07581723e+00 -3.27844888e-01 -5.67930937e-01 4.61091906e-01
-3.67854416e-01 2.79582542e-04 -1.50995061e-01 -1.35763419e+00
-1.37177467e+00 7.55181670e-01 1.60053349e+00 2.12010145e-01
1.51189661e+00 -5.15217960e-01 4.81417686e-01 4.47248816e-01
2.66691685e-01 -1.14078200e+00 -5.24085462e-01 1.23082967e-02
7.51990139e-01 -5.74466884e-01 -1.29500151e-01 5.31800151e-01
-4.24882442e-01 1.37542915e+00 5.53264141e-01 -3.33337337e-01
7.98963368e-01 2.87110388e-01 3.90929043e-01 -4.79097962e-01
-6.73328221e-01 2.94636726e-01 2.49597188e-02 6.12865508e-01
5.32617271e-01 1.84782684e-01 -9.32901740e-01 5.96275151e-01
-5.43273449e-01 1.09627739e-01 9.36794698e-01 4.32834685e-01
-2.62521535e-01 -5.80448031e-01 -1.73729792e-01 9.56438065e-01
-7.58486509e-01 -2.96348512e-01 -3.23973000e-01 8.82297158e-01
2.39406496e-01 9.99833226e-01 2.39722252e-01 -3.21538687e-01
4.04297650e-01 7.30820537e-01 2.63832867e-01 -3.73060584e-01
-8.47265363e-01 1.87004745e-01 -3.85589957e-01 -7.30703950e-01
-4.14967716e-01 -5.87063253e-01 -1.19075656e+00 2.31270343e-01
4.76283841e-02 1.76759690e-01 9.11243916e-01 7.46450663e-01
5.64589560e-01 4.90749896e-01 7.00766861e-01 -2.40752921e-01
-7.10770264e-02 -1.57622921e+00 -1.19476342e+00 6.93426877e-02
4.79358912e-01 -5.47449052e-01 -9.11686420e-01 9.48276464e-03] | [13.533238410949707, 3.2873823642730713] |
5c512fee-7bc2-4101-844f-c9637a758515 | improving-generalization-in-meta-learning-via | 2306.08460 | null | https://arxiv.org/abs/2306.08460v1 | https://arxiv.org/pdf/2306.08460v1.pdf | Improving Generalization in Meta-Learning via Meta-Gradient Augmentation | Meta-learning methods typically follow a two-loop framework, where each loop potentially suffers from notorious overfitting, hindering rapid adaptation and generalization to new tasks. Existing schemes solve it by enhancing the mutual-exclusivity or diversity of training samples, but these data manipulation strategies are data-dependent and insufficiently flexible. This work alleviates overfitting in meta-learning from the perspective of gradient regularization and proposes a data-independent \textbf{M}eta-\textbf{G}radient \textbf{Aug}mentation (\textbf{MGAug}) method. The key idea is to first break the rote memories by network pruning to address memorization overfitting in the inner loop, and then the gradients of pruned sub-networks naturally form the high-quality augmentation of the meta-gradient to alleviate learner overfitting in the outer loop. Specifically, we explore three pruning strategies, including \textit{random width pruning}, \textit{random parameter pruning}, and a newly proposed \textit{catfish pruning} that measures a Meta-Memorization Carrying Amount (MMCA) score for each parameter and prunes high-score ones to break rote memories as much as possible. The proposed MGAug is theoretically guaranteed by the generalization bound from the PAC-Bayes framework. In addition, we extend a lightweight version, called MGAug-MaxUp, as a trade-off between performance gains and resource overhead. Extensive experiments on multiple few-shot learning benchmarks validate MGAug's effectiveness and significant improvement over various meta-baselines. The code is publicly available at \url{https://github.com/xxLifeLover/Meta-Gradient-Augmentation}. | ['Yilong Yin', 'Yuling Ma', 'Xiushan Nie', 'Qi Wei', 'Haoliang Sun', 'Ren Wang'] | 2023-06-14 | null | null | null | null | ['meta-learning', 'network-pruning', 'memorization'] | ['methodology', 'methodology', 'natural-language-processing'] | [ 9.86662805e-02 -6.30174438e-03 -4.80838865e-01 -3.44142258e-01
-6.77783012e-01 -2.41857156e-01 3.54710311e-01 1.96610734e-01
-8.18996251e-01 8.45435441e-01 1.15260236e-01 -3.10089231e-01
-2.71335185e-01 -9.26522017e-01 -9.36391175e-01 -8.54751289e-01
1.97447717e-01 4.70964648e-02 3.57504576e-01 -3.21619719e-01
2.86335140e-01 7.73332417e-02 -1.72758651e+00 3.26496184e-01
1.25806725e+00 9.54203486e-01 2.95747012e-01 5.29215455e-01
-2.23850831e-01 6.03043139e-01 -4.29091007e-01 -4.80373710e-01
2.72415251e-01 -3.24007332e-01 -6.36147916e-01 -4.60210592e-01
5.18494427e-01 -2.50814617e-01 -3.79893899e-01 1.10517764e+00
6.16043925e-01 6.34145319e-01 4.80585426e-01 -9.88931537e-01
-5.90126634e-01 8.49909782e-01 -7.26683199e-01 5.60960889e-01
-2.91390777e-01 8.12733397e-02 8.74030590e-01 -9.31814194e-01
2.53151804e-01 9.68664289e-01 7.66654670e-01 9.10533786e-01
-1.02241337e+00 -8.79271448e-01 5.57188511e-01 5.22555560e-02
-1.19995487e+00 -3.52412790e-01 4.75168645e-01 -6.67903423e-02
9.44260538e-01 4.51568902e-01 4.69576061e-01 1.16280699e+00
2.64287069e-02 1.08979499e+00 9.58028018e-01 -5.33047795e-01
3.85962695e-01 6.08098395e-02 6.59091473e-01 8.98011267e-01
4.90981787e-01 1.25292018e-01 -7.44014978e-01 -1.46385789e-01
6.73073888e-01 3.11490655e-01 -3.37914497e-01 -1.96576878e-01
-7.70229399e-01 7.87566602e-01 4.58263606e-01 1.94136262e-01
-4.83775176e-02 9.43470895e-02 4.74220634e-01 2.69594759e-01
5.44381082e-01 4.33865011e-01 -5.84742665e-01 -2.29535222e-01
-1.03572357e+00 1.40791938e-01 4.92767900e-01 8.88208926e-01
9.14822102e-01 1.73280627e-01 -4.72549170e-01 1.30635071e+00
1.84504203e-02 1.12552933e-01 1.14889038e+00 -6.26726389e-01
6.89089537e-01 6.43643677e-01 -2.51476973e-01 -5.49256682e-01
-3.32616389e-01 -7.86282957e-01 -9.96966422e-01 2.66351253e-02
9.03594047e-02 -2.38174394e-01 -1.14305794e+00 2.01856804e+00
3.22485238e-01 4.31125790e-01 -8.16115960e-02 5.93264103e-01
9.75183010e-01 5.27645588e-01 3.39286536e-01 -3.42473477e-01
1.01542878e+00 -1.31381238e+00 -3.33800167e-01 -4.12679613e-01
8.70374739e-01 -2.24216029e-01 1.51009917e+00 4.56982374e-01
-1.10918152e+00 -6.35749400e-01 -1.36805189e+00 6.76639229e-02
-4.91308987e-01 -6.15567937e-02 6.88036263e-01 8.09308052e-01
-7.34559238e-01 9.51257408e-01 -8.05776000e-01 1.51553750e-02
5.95333397e-01 3.44687045e-01 9.94107570e-04 2.69197971e-02
-1.11898100e+00 6.27250493e-01 7.99549639e-01 -6.57061487e-02
-7.90946782e-01 -8.56734753e-01 -7.86547124e-01 1.22562975e-01
5.07986844e-01 -7.43474185e-01 1.08994257e+00 -7.28439987e-01
-1.51432943e+00 6.47984982e-01 4.27487008e-02 -5.08782148e-01
5.18238842e-01 -5.91091931e-01 -2.00913116e-01 -3.14305842e-01
-3.39310259e-01 7.55620837e-01 8.55468929e-01 -1.12730598e+00
-6.69297755e-01 -4.62381423e-01 -1.28357545e-01 4.55674887e-01
-8.93864810e-01 -5.17846525e-01 -6.13941550e-01 -9.81433094e-01
1.87435880e-01 -6.89278841e-01 -1.68348670e-01 -6.23749912e-01
-2.82567441e-01 -1.68049127e-01 5.27093768e-01 -2.66701400e-01
1.84169626e+00 -2.02211213e+00 9.82310921e-02 7.74976984e-02
1.94290504e-01 7.13961482e-01 -4.73780125e-01 4.69655357e-03
-1.41126998e-02 3.07454258e-01 -2.53912807e-01 -4.00311828e-01
-2.95692921e-01 3.29519600e-01 -4.16107059e-01 1.51679486e-01
-2.11740673e-01 7.88887382e-01 -9.75302696e-01 -2.94731200e-01
-6.95033297e-02 2.54547596e-01 -6.93092108e-01 1.90176323e-01
-2.00112820e-01 -9.44618955e-02 -4.28103477e-01 6.72520101e-01
6.27263308e-01 -1.38168931e-01 -6.09892718e-02 7.51012117e-02
-3.16019170e-04 2.86575705e-01 -1.24035716e+00 1.75059843e+00
-4.66609687e-01 -5.42037981e-03 -2.68345028e-01 -8.69346082e-01
9.85406160e-01 -1.53874293e-01 2.67194092e-01 -6.73536479e-01
2.26148561e-01 1.32853389e-01 -2.42850393e-01 -2.11174250e-01
7.90586412e-01 5.08929370e-04 1.57063738e-01 5.30612171e-01
3.43905181e-01 2.14486524e-01 2.12265477e-01 1.31437480e-01
1.04501998e+00 3.00944835e-01 1.30031928e-01 -1.57398984e-01
3.88400912e-01 -2.73113102e-01 7.84467101e-01 1.07700813e+00
-1.69167787e-01 4.82256234e-01 1.11721389e-01 -3.25916588e-01
-7.79187560e-01 -9.62634921e-01 -1.41442657e-01 1.80569446e+00
9.35262740e-02 -6.31619692e-01 -9.03872967e-01 -8.53261888e-01
3.27783562e-02 8.93936872e-01 -7.55395412e-01 -7.89983273e-01
-5.48030496e-01 -1.35681069e+00 7.35486746e-01 6.68042779e-01
7.80584574e-01 -9.84966993e-01 -5.92049956e-01 9.81425419e-02
-2.16744989e-02 -3.81850094e-01 -4.85009372e-01 6.24970019e-01
-1.37370491e+00 -8.42263877e-01 -5.96643388e-01 -5.09813249e-01
5.72818756e-01 4.46664453e-01 1.02305603e+00 3.40131074e-01
-2.63395518e-01 1.38571441e-01 -3.55105847e-01 -5.44871151e-01
-6.86020777e-02 3.68089437e-01 1.82149842e-01 -4.14721489e-01
6.51120663e-01 -7.26769745e-01 -7.22104549e-01 3.00492793e-01
-9.59838688e-01 1.82815976e-02 7.47093797e-01 1.13993907e+00
8.73076200e-01 -4.39624004e-02 7.60498106e-01 -1.07994258e+00
8.76845002e-01 -5.69443226e-01 -2.06114277e-01 4.09199059e-01
-1.03818941e+00 1.76654965e-01 5.12164354e-01 -7.78687477e-01
-1.14845133e+00 -3.61573130e-01 -5.49101867e-02 -6.16859972e-01
1.06415801e-01 3.49718064e-01 -9.30853039e-02 1.13003492e-01
9.42089438e-01 4.02193546e-01 -2.39296570e-01 -5.47846854e-01
4.08134133e-01 4.63978976e-01 4.74100083e-01 -8.77240121e-01
4.22908694e-01 2.11689755e-01 -3.97237957e-01 -7.11162865e-01
-1.40661240e+00 -5.24852931e-01 -4.45100665e-01 -1.91271026e-02
4.00771797e-01 -8.66997182e-01 -4.98887986e-01 3.52786094e-01
-5.31830490e-01 -6.01146042e-01 -5.18506765e-01 3.97159427e-01
-3.91868740e-01 3.48682612e-01 -6.13658011e-01 -7.74411440e-01
-8.10542226e-01 -8.43670249e-01 5.25893867e-01 5.59534669e-01
7.61228129e-02 -8.02122235e-01 1.88295498e-01 3.01766813e-01
3.41545254e-01 -7.72310942e-02 9.74929333e-01 -8.24323475e-01
-8.54211226e-02 2.56866701e-02 -1.56243458e-01 6.71463132e-01
-2.02297240e-01 -3.10286999e-01 -1.15020180e+00 -5.06862819e-01
9.59657654e-02 -7.36021578e-01 1.45550418e+00 3.24914187e-01
1.45512807e+00 -4.46011633e-01 -2.13148579e-01 9.74589288e-01
1.34827435e+00 -1.88622642e-02 6.59265399e-01 6.88808382e-01
5.92646241e-01 7.46957436e-02 5.23509264e-01 5.43681681e-01
1.30539134e-01 1.75179169e-01 4.18233246e-01 3.02575588e-01
-6.67783394e-02 -4.52700466e-01 3.73054981e-01 9.62188661e-01
-2.36813605e-01 -1.82839081e-01 -6.83333814e-01 3.03937823e-01
-1.96077979e+00 -9.95639622e-01 1.90867946e-01 2.48131680e+00
1.20017350e+00 3.92218888e-01 3.82105410e-02 9.12952274e-02
5.58743119e-01 2.25504473e-01 -7.79925346e-01 -4.71188515e-01
-6.66206032e-02 3.19609106e-01 4.81493354e-01 2.51986980e-01
-1.13143837e+00 9.83092129e-01 5.38520813e+00 1.20636237e+00
-8.14312756e-01 3.76964778e-01 7.88239300e-01 -4.92028326e-01
-2.21459866e-01 -8.20601508e-02 -1.43559766e+00 4.04473484e-01
8.98527920e-01 1.25212133e-01 4.41288084e-01 1.11058545e+00
-2.83688098e-01 -2.63016969e-01 -8.26052547e-01 7.96094239e-01
1.00533634e-01 -1.27117705e+00 4.14790303e-01 -2.58580446e-01
8.28332067e-01 5.71830347e-02 2.66717970e-01 1.07445455e+00
3.81724507e-01 -8.38065326e-01 5.58873832e-01 5.81410170e-01
6.32004023e-01 -9.36262786e-01 4.01563138e-01 6.11951172e-01
-1.01071572e+00 -3.86470914e-01 -8.00552845e-01 2.05136403e-01
-3.21137816e-01 5.58986783e-01 -3.98539752e-01 3.76149029e-01
8.00277948e-01 3.68270993e-01 -8.50270033e-01 9.99217927e-01
-2.00095639e-01 8.00669909e-01 -2.36268178e-01 -1.33443192e-01
3.23001713e-01 -1.50530651e-01 5.95015049e-01 1.34098518e+00
2.19810843e-01 2.39448443e-01 1.23817548e-01 6.15381241e-01
-3.14221233e-01 1.64638385e-01 -1.71242729e-01 1.94363281e-01
7.13135779e-01 1.16690552e+00 -4.51296151e-01 -3.36286098e-01
-1.13653861e-01 7.92670131e-01 8.55377793e-01 3.55863154e-01
-7.71380842e-01 -5.55894732e-01 3.01787138e-01 1.98914837e-02
1.60075784e-01 2.34270897e-02 -5.03856540e-01 -1.22387505e+00
6.16591983e-02 -8.37724507e-01 8.55141461e-01 -3.53227526e-01
-1.07214570e+00 4.65742111e-01 1.42453268e-01 -1.01400828e+00
1.53119698e-01 -3.91886264e-01 -7.62783051e-01 6.48142219e-01
-1.41877210e+00 -8.82694185e-01 -4.66891736e-01 5.40777802e-01
7.33422577e-01 -3.89629900e-01 6.63417697e-01 2.23174796e-01
-1.03167534e+00 1.15831232e+00 2.85442859e-01 -2.08149597e-01
5.35363972e-01 -1.19676983e+00 1.99257508e-02 7.70954251e-01
2.43592039e-02 9.19437528e-01 4.56424564e-01 -7.59103775e-01
-1.18310475e+00 -1.22892141e+00 3.11281115e-01 -1.54170692e-01
4.17433023e-01 -1.37272671e-01 -1.32230842e+00 6.89445615e-01
-2.20912904e-01 -9.08090249e-02 7.22715318e-01 4.46843594e-01
-5.55180550e-01 -1.88969195e-01 -9.78034139e-01 7.44779170e-01
1.15394568e+00 -6.47051260e-02 -8.66648734e-01 1.75500602e-01
7.06428528e-01 -4.29815650e-01 -6.49193168e-01 6.21439695e-01
3.98439705e-01 -1.13589191e+00 1.05395234e+00 -7.85880446e-01
3.64190727e-01 8.65527168e-02 -4.11503725e-02 -1.19089282e+00
-3.64657849e-01 -5.88470161e-01 -7.77716160e-01 1.33271873e+00
4.50777531e-01 -3.51674527e-01 1.05748224e+00 5.16609311e-01
-2.81416923e-01 -1.19618797e+00 -8.46744061e-01 -9.56625760e-01
3.53109300e-01 -4.58524823e-01 5.29167295e-01 9.32509720e-01
-2.95597594e-03 3.04273903e-01 -3.03465575e-01 -1.72574475e-01
5.37581444e-01 -2.12561265e-01 6.81069314e-01 -1.06059325e+00
-5.07830858e-01 -6.99959755e-01 1.09668687e-01 -1.12908614e+00
-1.60156548e-01 -1.05907726e+00 -4.38074470e-02 -1.22227895e+00
4.81219918e-01 -6.93113029e-01 -8.30567539e-01 7.04552293e-01
-6.30404472e-01 -2.93549448e-02 2.70251874e-02 2.80720204e-01
-7.78672814e-01 8.41895163e-01 1.06853282e+00 4.78010625e-02
-7.43202507e-01 4.63062897e-02 -7.51046121e-01 8.82715940e-01
9.61220205e-01 -5.85852027e-01 -6.05356574e-01 -4.52754110e-01
2.55970091e-01 -3.92318606e-01 1.53104946e-01 -1.25891161e+00
3.69097024e-01 -1.06999815e-01 6.08721673e-01 -5.64999700e-01
3.21753234e-01 -2.11119264e-01 -3.44146460e-01 3.97149801e-01
-4.83609200e-01 3.47263739e-02 3.98813784e-01 8.39239538e-01
6.46477416e-02 -7.10563660e-01 9.50507998e-01 -2.81361610e-01
-5.97575307e-01 3.89789551e-01 5.48122786e-02 3.70191604e-01
8.21053445e-01 -2.71337450e-01 -5.57400107e-01 1.92034885e-01
-5.75977147e-01 3.17177832e-01 1.92184582e-01 4.08265263e-01
7.17255890e-01 -1.21564889e+00 -4.16933477e-01 2.74004489e-01
1.02082603e-01 1.70175284e-01 5.29524982e-01 6.99582875e-01
-8.30696374e-02 1.84938367e-02 -1.11860231e-01 -3.63698900e-01
-1.16352022e+00 5.05672634e-01 3.68193805e-01 -4.84274954e-01
-6.74906671e-01 1.36845767e+00 5.50846122e-02 -4.28132713e-01
7.21858859e-01 -1.68063506e-01 -9.08401832e-02 4.58455198e-02
7.85332203e-01 6.58783257e-01 1.03554666e-01 -3.51859219e-02
1.06207862e-01 1.52719378e-01 -7.25305259e-01 9.60031077e-02
1.36974299e+00 7.76976570e-02 2.08015934e-01 4.34601218e-01
9.56842542e-01 -2.63887972e-01 -1.39296377e+00 -3.92365873e-01
5.80485314e-02 -2.95637816e-01 1.87695220e-01 -8.53546560e-01
-9.88121331e-01 7.65135705e-01 6.04350030e-01 -1.61584944e-01
1.13450539e+00 -4.40922141e-01 9.12361860e-01 7.96505630e-01
1.77111357e-01 -1.55814254e+00 2.81751335e-01 7.11785138e-01
6.65928423e-01 -1.10143912e+00 3.22210491e-01 1.36481598e-01
-6.25418663e-01 9.35420752e-01 1.17675066e+00 -1.56541467e-02
6.01758838e-01 2.89327577e-02 -4.04657394e-01 -1.26096634e-02
-8.20980906e-01 8.16417579e-03 3.38985085e-01 1.86441168e-01
3.67602289e-01 -9.17354375e-02 -4.66911882e-01 1.10354316e+00
-1.15754887e-01 6.52536051e-03 1.33846194e-01 1.29677331e+00
-9.84564543e-01 -1.00331986e+00 -1.31623521e-01 7.78359592e-01
-3.53894800e-01 -2.43510887e-01 -1.69472724e-01 8.02116632e-01
2.95868337e-01 6.08444333e-01 -1.78366080e-01 -5.60830116e-01
3.61872256e-01 3.05370867e-01 4.76126969e-01 -7.16166317e-01
-8.81956875e-01 9.00748745e-03 -1.96427241e-01 -3.67326736e-01
-5.44486307e-02 -3.88019204e-01 -1.30886137e+00 -2.66200632e-01
-5.27622163e-01 2.34243363e-01 3.20802778e-01 8.30156684e-01
3.40181321e-01 5.65354884e-01 3.60572755e-01 -6.16487503e-01
-9.85942006e-01 -1.13709569e+00 -4.73453015e-01 2.99027473e-01
-4.03084494e-02 -7.22020030e-01 -3.62734854e-01 -2.66467899e-01] | [9.498403549194336, 3.429732084274292] |
d2123c54-9fb1-4cae-81a2-8cc165953770 | proof-artifact-co-training-for-theorem | 2102.06203 | null | https://arxiv.org/abs/2102.06203v2 | https://arxiv.org/pdf/2102.06203v2.pdf | Proof Artifact Co-training for Theorem Proving with Language Models | Labeled data for imitation learning of theorem proving in large libraries of formalized mathematics is scarce as such libraries require years of concentrated effort by human specialists to be built. This is particularly challenging when applying large Transformer language models to tactic prediction, because the scaling of performance with respect to model size is quickly disrupted in the data-scarce, easily-overfitted regime. We propose PACT ({\bf P}roof {\bf A}rtifact {\bf C}o-{\bf T}raining), a general methodology for extracting abundant self-supervised data from kernel-level proof terms for co-training alongside the usual tactic prediction objective. We apply this methodology to Lean, an interactive proof assistant which hosts some of the most sophisticated formalized mathematics to date. We instrument Lean with a neural theorem prover driven by a Transformer language model and show that PACT improves theorem proving success rate on a held-out suite of test theorems from 32\% to 48\%. | ['Stanislas Polu', 'Edward W. Ayers', 'Yuhuai Wu', 'Jason Rute', 'Jesse Michael Han'] | 2021-02-11 | proof-artifact-co-training-for-theorem-1 | https://openreview.net/forum?id=rpxJc9j04U | https://openreview.net/pdf?id=rpxJc9j04U | iclr-2022-4 | ['automated-theorem-proving', 'automated-theorem-proving'] | ['miscellaneous', 'reasoning'] | [ 3.03425133e-01 3.48259419e-01 -3.63467067e-01 -1.32773951e-01
-9.60289061e-01 -9.73393679e-01 5.59186041e-01 1.37942001e-01
-3.99481952e-02 8.55191350e-01 -4.18160856e-01 -1.25124395e+00
-5.00790358e-01 -6.29218161e-01 -1.26206291e+00 -2.04863355e-01
-4.21138287e-01 7.78266966e-01 -5.60263358e-02 -3.39334607e-01
2.95321256e-01 1.45225078e-01 -1.33984160e+00 4.33854729e-01
1.16092467e+00 5.76383054e-01 1.42546773e-01 1.10976601e+00
5.41190542e-02 1.70345938e+00 -2.71484554e-01 -4.81211007e-01
2.17705354e-01 -4.17670965e-01 -1.32003272e+00 -5.19293845e-01
6.30338728e-01 -3.32343102e-01 -2.43767932e-01 1.20849335e+00
-1.95007369e-01 -4.80649203e-01 6.50100887e-01 -1.85087168e+00
-2.24677950e-01 1.45521426e+00 -1.86925083e-01 1.86546966e-01
3.04392755e-01 1.62516564e-01 1.26053524e+00 -4.22871649e-01
8.94617498e-01 1.15369594e+00 5.91730773e-01 5.78276098e-01
-1.65269458e+00 -8.47611308e-01 -1.53434411e-01 3.17428291e-01
-1.07506931e+00 -4.67528164e-01 8.39595854e-01 -5.82088292e-01
1.07804525e+00 3.13914001e-01 5.00500679e-01 1.13224173e+00
3.41216803e-01 1.12713206e+00 1.07121909e+00 -6.30703926e-01
6.86927233e-03 3.54305595e-01 3.38843912e-01 1.37594187e+00
7.41719129e-03 1.42533541e-01 -4.81546372e-01 -2.69246310e-01
4.49632615e-01 -3.94797415e-01 7.37054870e-02 -8.97497714e-01
-1.61937928e+00 8.99441540e-01 -1.66041870e-02 2.87379295e-01
1.89314127e-01 6.11493051e-01 6.60938680e-01 1.02052939e+00
2.95672089e-01 9.88186538e-01 -9.73755360e-01 -3.70971113e-01
-1.09561741e+00 8.04152608e-01 1.13886571e+00 1.23372436e+00
3.80691618e-01 2.73550451e-02 1.94003329e-01 2.30211765e-01
-6.98086293e-03 4.77007002e-01 -4.97781448e-02 -1.46588826e+00
6.35204136e-01 5.93499124e-01 -2.47062985e-02 -4.34435666e-01
-2.93054849e-01 -2.64775604e-01 -6.14018679e-01 2.31962785e-01
7.74327934e-01 -6.82548657e-02 -1.94496840e-01 1.83193612e+00
-8.56526755e-03 -1.90737136e-02 2.20959544e-01 2.25070909e-01
5.05256653e-01 5.10757983e-01 -2.28594199e-01 -3.91077906e-01
1.01809084e+00 -7.68711209e-01 -1.95927948e-01 9.45845470e-02
1.31146932e+00 -4.30315226e-01 1.02716780e+00 8.80985439e-01
-1.16574669e+00 -9.43006203e-02 -1.02132916e+00 -3.99500467e-02
-3.15196663e-01 -2.24842504e-01 1.25686407e+00 1.59125268e-01
-1.05865300e+00 9.68863606e-01 -6.62577987e-01 3.01528759e-02
6.64950848e-01 5.85421920e-01 -2.68782884e-01 -1.84053198e-01
-1.03214586e+00 9.17219877e-01 6.78966403e-01 -2.08706707e-01
-1.45519018e+00 -1.20897567e+00 -9.40398276e-01 -3.59549560e-02
7.54745245e-01 -4.89148289e-01 1.64810359e+00 -8.11183035e-01
-1.27981234e+00 8.88837099e-01 1.24807388e-01 -7.82639027e-01
8.85714531e-01 -1.25784606e-01 9.10978615e-02 2.03978010e-02
3.22650164e-01 5.26715517e-01 9.22412694e-01 -9.48003650e-01
-5.83952725e-01 -1.81297228e-01 3.67209792e-01 -3.40933204e-01
-3.74320522e-02 2.23007217e-01 5.08478165e-01 -2.78022081e-01
-1.54480875e-01 -7.09617436e-01 2.14614809e-01 -2.23098025e-01
-5.39536476e-01 -7.08599985e-01 9.16503727e-01 -4.45773959e-01
9.62332010e-01 -1.59167206e+00 4.83623266e-01 1.56349540e-01
8.25718939e-01 4.54154946e-02 1.25476554e-01 2.99434841e-01
-4.24550086e-01 1.59151495e-01 -1.43426266e-02 2.62340844e-01
5.31104803e-01 9.97222513e-02 -6.02770865e-01 4.03510243e-01
1.95580631e-01 1.00451088e+00 -1.39447546e+00 -8.96320999e-01
1.33421361e-01 -5.30282199e-01 -8.21223199e-01 2.07472980e-01
-1.08099103e+00 -4.17467281e-02 -5.01948714e-01 7.48417497e-01
1.91339016e-01 -5.06632447e-01 2.06042692e-01 4.54104632e-01
-3.26280929e-02 5.23588061e-01 -7.02385485e-01 1.97413528e+00
-3.37177604e-01 8.47106099e-01 1.96804807e-01 -1.10933161e+00
4.35147792e-01 3.77585530e-01 3.85723114e-01 -2.40077823e-01
2.55554676e-01 3.48707974e-01 4.03322160e-01 -4.45421576e-01
1.88972816e-01 -4.57517087e-01 -1.04819477e-01 7.87166774e-01
2.77500302e-01 -6.58414423e-01 4.45343107e-01 6.42355442e-01
1.55249131e+00 5.63193679e-01 4.96626273e-02 -5.29965222e-01
5.44423699e-01 5.25723755e-01 2.74040788e-01 9.77762878e-01
-6.08513765e-02 -3.59148800e-01 1.14320087e+00 -5.19451439e-01
-1.40854001e+00 -8.54506016e-01 2.59179296e-03 1.36186767e+00
-3.31192911e-01 -7.66778648e-01 -7.90163159e-01 -9.63309586e-01
1.30665511e-01 9.71339524e-01 -2.55352706e-01 -2.72529721e-01
-5.13902426e-01 1.02753192e-02 1.16962218e+00 2.00837806e-01
3.07514787e-01 -1.17910707e+00 -4.81866360e-01 1.53422117e-01
-3.96002829e-02 -9.09786463e-01 7.98241869e-02 7.96765625e-01
-8.33827794e-01 -1.48019314e+00 -7.05479756e-02 -9.22267497e-01
5.01863837e-01 -3.10063511e-01 1.29567540e+00 2.31920943e-01
-1.74265265e-01 1.92668304e-01 -7.15965033e-02 -6.90317333e-01
-1.16550994e+00 2.24597737e-01 3.16474557e-01 -7.86688924e-01
2.84870267e-01 -6.93338096e-01 4.58126068e-01 -1.40485153e-01
-5.34373164e-01 5.93497872e-01 3.70243877e-01 1.11432457e+00
-2.55629897e-01 2.23187283e-01 1.61614239e-01 -9.22398865e-01
5.95519543e-01 -1.56231895e-01 -1.25090373e+00 3.73120487e-01
-7.88808823e-01 4.30958748e-01 1.13644040e+00 -5.38829446e-01
-5.06742060e-01 -3.54574233e-01 7.34184802e-01 -8.26666057e-01
1.54634351e-02 3.48225445e-01 1.63151413e-01 -1.55565431e-02
8.12942207e-01 2.81115234e-01 1.50628105e-01 1.91985294e-02
3.02439511e-01 2.17363000e-01 6.26538217e-01 -1.33589339e+00
1.55206001e+00 -2.91736692e-01 2.76759535e-01 -3.20233911e-01
-9.20024276e-01 1.07599519e-01 -5.64521730e-01 -3.48059535e-02
2.50044107e-01 -5.10053933e-01 -1.31870472e+00 1.20256610e-01
-1.25090981e+00 -1.00817919e+00 -1.30941272e-01 3.43021393e-01
-1.11711788e+00 2.00505361e-01 -7.61899114e-01 -9.45020318e-01
-4.03955907e-01 -1.17642140e+00 7.77536988e-01 -3.51162672e-01
-4.68905449e-01 -9.52192485e-01 4.95290942e-02 4.70603257e-01
3.60108130e-02 9.15314108e-02 1.71324086e+00 -9.90335584e-01
-7.43308485e-01 -1.68377310e-02 -2.98377723e-01 4.33908761e-01
-2.50795662e-01 -2.55713053e-02 -8.49229515e-01 -1.67347193e-01
-6.34479225e-02 -1.02610934e+00 3.33575755e-01 3.81894992e-03
1.21025157e+00 -5.58403850e-01 -2.49445722e-01 4.34173137e-01
9.57054794e-01 -2.82850862e-01 1.29364729e-01 4.66929525e-01
6.22513473e-01 1.09718472e-01 4.57961321e-01 -2.90122554e-02
1.89910650e-01 1.64725363e-01 2.03257993e-01 2.94348776e-01
3.91615599e-01 -4.59922791e-01 6.77055895e-01 4.65438813e-01
-6.72878250e-02 3.18479061e-01 -1.31619275e+00 4.02573913e-01
-1.71261322e+00 -1.38432086e+00 2.08164826e-01 1.81737423e+00
1.45155692e+00 6.57185614e-01 9.85495895e-02 5.24089396e-01
2.52753198e-01 -2.40599945e-01 -3.92622203e-01 -4.93080139e-01
2.96066970e-01 5.66120505e-01 3.30839396e-01 6.30575180e-01
-9.97128308e-01 1.11938608e+00 6.05387068e+00 9.50731039e-01
-8.16437244e-01 -2.16843754e-01 2.53666252e-01 -2.43969653e-02
-2.12227628e-01 3.94578308e-01 -3.37584406e-01 8.79955292e-02
1.22845912e+00 -4.64779437e-01 1.09842575e+00 1.22816277e+00
-5.16369045e-02 -1.05348840e-01 -2.13796020e+00 8.00578654e-01
-9.70373079e-02 -1.55786395e+00 -2.92975217e-01 1.10736787e-01
4.14279819e-01 2.77280007e-02 -7.34625384e-02 9.12724912e-01
7.96891689e-01 -1.30541778e+00 1.03933311e+00 1.43945351e-01
7.82665014e-01 -7.72091568e-01 4.58004236e-01 9.58004951e-01
-6.57676518e-01 -2.60074288e-01 -1.22036189e-02 -3.94878060e-01
-7.41052926e-01 9.75758359e-02 -1.36652386e+00 4.26083505e-01
3.00953656e-01 8.49050999e-01 -6.16921186e-01 2.10859135e-01
-2.63442129e-01 6.60383165e-01 -4.26222473e-01 -3.26665223e-01
3.69561225e-01 -4.39951792e-02 6.63719118e-01 1.07268381e+00
-2.94449717e-01 -1.18213624e-01 2.34872490e-01 1.49184823e+00
-2.27222383e-01 -3.88077468e-01 -1.18605411e+00 -3.23388964e-01
9.80605632e-02 1.20923460e+00 -5.10654867e-01 -5.40712833e-01
8.62514079e-02 4.89086002e-01 5.16951978e-01 -5.77066354e-02
-1.08047378e+00 -4.26698476e-01 1.96676835e-01 2.13905368e-02
1.33017465e-01 -8.37690905e-02 -4.25823554e-02 -1.40622663e+00
1.51870713e-01 -1.62296557e+00 1.85576126e-01 -9.32274163e-01
-1.01050389e+00 -4.63203453e-02 5.26808679e-01 -9.24154460e-01
-5.61199367e-01 -1.02072608e+00 -3.66134375e-01 6.19708836e-01
-1.24270618e+00 -1.32247305e+00 3.00293833e-01 6.68458402e-01
4.97828603e-01 -5.43009758e-01 1.02903891e+00 -1.33777380e-01
-3.22903931e-01 6.03124321e-01 -1.60027981e-01 5.14101684e-01
2.43015096e-01 -1.62328041e+00 -1.33337006e-01 7.47793913e-01
2.05461562e-01 8.56550932e-01 1.03206205e+00 -3.84193748e-01
-2.24120855e+00 -8.98765922e-01 9.65755820e-01 -9.23788905e-01
1.54695141e+00 -6.42834961e-01 -4.72827822e-01 1.05126071e+00
-9.36115310e-02 -1.15138769e-01 2.01658607e-01 4.11993563e-01
-8.71484935e-01 -1.64203003e-01 -9.48371649e-01 8.79140258e-01
9.76548612e-01 -1.04112244e+00 -1.02208745e+00 8.34319174e-01
6.46819711e-01 -5.42551994e-01 -9.40572321e-01 1.61687657e-01
4.71045554e-01 -6.04384065e-01 6.20630980e-01 -1.10726178e+00
9.50520813e-01 -6.68335035e-02 -1.04523897e-01 -8.41236711e-01
1.32302433e-01 -1.33690763e+00 -5.03939211e-01 9.44341481e-01
5.46481192e-01 -3.05280387e-01 7.25597858e-01 4.32038933e-01
5.14766052e-02 -5.02027452e-01 -8.89845669e-01 -7.68805146e-01
6.28307104e-01 -9.29602146e-01 1.13518968e-01 1.13995779e+00
8.96619797e-01 7.12465167e-01 5.08748628e-02 -1.22398905e-01
9.46346402e-01 3.73491049e-01 1.09162378e+00 -1.43632472e+00
-3.62510681e-01 -7.17637479e-01 -1.78787708e-01 -5.50689459e-01
8.59251142e-01 -1.68818676e+00 2.12799432e-03 -8.50774050e-01
5.36027789e-01 -4.82768893e-01 -1.37607881e-03 8.22822511e-01
3.45832974e-01 -4.07223612e-01 1.60171576e-02 -1.18754342e-01
-9.44857657e-01 1.99419990e-01 1.15312541e+00 -5.43315172e-01
1.43879414e-01 -1.22110143e-01 -4.82571810e-01 9.49006200e-01
4.59562480e-01 -5.46902895e-01 -4.52019066e-01 -1.61589712e-01
5.81113279e-01 3.51643860e-01 8.13075244e-01 -8.82109940e-01
2.87557632e-01 -4.21307623e-01 9.60352644e-02 -3.86195451e-01
-2.02131391e-01 -7.79797018e-01 -2.51369655e-01 6.50546551e-01
-8.27463746e-01 1.57208636e-01 2.88282573e-01 1.31991580e-01
2.78467864e-01 -7.83188567e-02 4.16353106e-01 -4.05062050e-01
-3.11756015e-01 -3.00591290e-02 -2.68702596e-01 3.10816288e-01
8.77020061e-01 2.10241407e-01 -2.60439277e-01 -1.72937036e-01
-2.46995941e-01 5.76674759e-01 2.82664299e-01 1.84038565e-01
4.09490049e-01 -8.87398720e-01 -7.02371597e-01 9.60767940e-02
-4.28496016e-04 2.82265306e-01 -4.72317308e-01 1.19161570e+00
-6.25655174e-01 6.42249227e-01 -4.79526035e-02 -6.49650514e-01
-1.41968107e+00 8.88480783e-01 4.63937461e-01 -7.13990629e-01
-8.11846137e-01 6.80268705e-01 -2.03188568e-01 -7.64300942e-01
4.14586425e-01 -7.62099326e-01 4.57846224e-01 -4.79415804e-01
2.97613680e-01 1.93985507e-01 -1.54221803e-01 2.27625057e-01
-1.68127358e-01 -7.67086595e-02 -3.45280409e-01 -3.78690958e-02
1.58721173e+00 7.76598155e-01 -6.08237088e-01 6.59932137e-01
8.40560138e-01 -2.76611596e-01 -6.97454333e-01 -1.63972989e-01
4.01064038e-01 2.13251844e-01 -2.22080220e-02 -8.09885323e-01
-2.82978058e-01 6.95640326e-01 -1.88582048e-01 3.50123882e-01
5.60371220e-01 1.08852573e-01 1.92729309e-01 1.35826850e+00
7.80882180e-01 -1.04342473e+00 -3.94496247e-02 8.99431884e-01
6.66082382e-01 -1.07096899e+00 2.92347789e-01 1.20916404e-01
-1.09960072e-01 1.34111643e+00 3.60998839e-01 -2.12302968e-01
2.29157180e-01 4.77519810e-01 -3.20391893e-01 -3.58622044e-01
-1.22364366e+00 4.25881445e-01 -1.02570899e-01 3.11317056e-01
2.59038687e-01 -1.07056402e-01 4.58601177e-01 3.40487152e-01
-6.77431762e-01 3.87014866e-01 4.64004934e-01 1.11379790e+00
-1.83354780e-01 -1.04963124e+00 -3.75084251e-01 4.75143999e-01
-3.46701890e-01 -3.50460708e-01 -3.35809827e-01 1.32080770e+00
6.22803718e-02 6.59942865e-01 -4.38784838e-01 -2.66672105e-01
-2.99384475e-01 5.67467213e-01 1.10372305e+00 -5.75000584e-01
-6.71027303e-01 -1.88881904e-01 2.90316194e-01 -3.67621571e-01
-2.79282331e-01 -7.12387443e-01 -1.23108101e+00 -8.13865125e-01
-2.98420101e-01 4.11344737e-01 2.03951359e-01 1.30240023e+00
-2.58509487e-01 4.88945305e-01 5.20954788e-01 -6.32108033e-01
-1.11562920e+00 -1.07291365e+00 -5.61034858e-01 3.80539745e-02
6.21645153e-01 -4.33870286e-01 -6.05928779e-01 1.54990166e-01] | [8.937151908874512, 7.057676792144775] |
9889adb6-31d6-4370-b44c-bf037d5b38a6 | one-shot-fine-grained-instance-retrieval | 1707.00811 | null | http://arxiv.org/abs/1707.00811v1 | http://arxiv.org/pdf/1707.00811v1.pdf | One-Shot Fine-Grained Instance Retrieval | Fine-Grained Visual Categorization (FGVC) has achieved significant progress
recently. However, the number of fine-grained species could be huge and
dynamically increasing in real scenarios, making it difficult to recognize
unseen objects under the current FGVC framework. This raises an open issue to
perform large-scale fine-grained identification without a complete training
set. Aiming to conquer this issue, we propose a retrieval task named One-Shot
Fine-Grained Instance Retrieval (OSFGIR). "One-Shot" denotes the ability of
identifying unseen objects through a fine-grained retrieval task assisted with
an incomplete auxiliary training set. This paper first presents the detailed
description to OSFGIR task and our collected OSFGIR-378K dataset. Next, we
propose the Convolutional and Normalization Networks (CN-Nets) learned on the
auxiliary dataset to generate a concise and discriminative representation.
Finally, we present a coarse-to-fine retrieval framework consisting of three
components, i.e., coarse retrieval, fine-grained retrieval, and query
expansion, respectively. The framework progressively retrieves images with
similar semantics, and performs fine-grained identification. Experiments show
our OSFGIR framework achieves significantly better accuracy and efficiency than
existing FGVC and image retrieval methods, thus could be a better solution for
large-scale fine-grained object identification. | ['Hantao Yao', 'Qi Tian', 'Yongdong Zhang', 'Shiliang Zhang', 'Jintao Li'] | 2017-07-04 | null | null | null | null | ['fine-grained-visual-categorization'] | ['computer-vision'] | [ 7.27137104e-02 -8.06767344e-01 -6.64016157e-02 -3.98836166e-01
-1.03527117e+00 -9.43654895e-01 7.45794594e-01 1.36160642e-01
-5.66005468e-01 6.14830077e-01 -8.73601902e-03 2.44822994e-01
-3.73121560e-01 -7.86170900e-01 -4.75686640e-01 -7.06838608e-01
2.53282070e-01 5.08600175e-01 3.06711465e-01 7.87799954e-02
4.05464292e-01 7.33643889e-01 -1.98731709e+00 3.09637874e-01
5.47556818e-01 1.47361171e+00 4.59743291e-01 5.68085849e-01
-2.45060921e-01 3.76905829e-01 -8.77080560e-01 -3.33371699e-01
2.12008804e-01 1.78796485e-01 -9.12076354e-01 -1.45876825e-01
6.09094858e-01 -7.54523575e-01 -2.62033761e-01 1.07468593e+00
5.51977098e-01 1.28526866e-01 8.40724945e-01 -1.35890055e+00
-1.06854773e+00 2.51164138e-01 -4.20077682e-01 4.23185974e-01
-7.61900377e-03 -3.52402069e-02 8.45954239e-01 -8.59303117e-01
4.74058509e-01 1.44946146e+00 4.46931869e-01 5.73097467e-01
-9.08560216e-01 -1.04673970e+00 2.83926725e-01 2.99605995e-01
-2.04294515e+00 -2.74859995e-01 2.63679713e-01 -4.34041500e-01
8.18235815e-01 2.45001689e-01 2.57421017e-01 7.74258137e-01
-3.91214639e-01 6.93479359e-01 8.07313263e-01 -7.05598667e-02
1.10082231e-01 -1.50089422e-02 3.15625697e-01 3.89926106e-01
3.77063274e-01 1.84503838e-01 -3.11415613e-01 -2.90651470e-01
7.98972964e-01 4.47002769e-01 -9.29503962e-02 -2.20407903e-01
-1.24615312e+00 7.52124846e-01 7.77071178e-01 3.08176786e-01
-2.11107239e-01 2.18629345e-01 6.44143522e-01 3.25594217e-01
3.74728054e-01 4.11014944e-01 -3.42119485e-01 1.25455499e-01
-9.08712268e-01 8.51122215e-02 4.07923311e-01 9.70337152e-01
9.43477869e-01 -7.90593252e-02 -4.41760868e-01 1.20930803e+00
1.58397526e-01 7.56769061e-01 8.64181280e-01 -6.12159550e-01
6.03140257e-02 6.45721436e-01 1.13749921e-01 -9.67381477e-01
-1.04977705e-01 -3.99722546e-01 -9.40918982e-01 -2.68347174e-01
2.63258778e-02 5.91886640e-01 -1.09355974e+00 1.49606419e+00
2.07532957e-01 8.36200193e-02 -1.07981078e-01 1.19350922e+00
1.34639454e+00 6.49529338e-01 4.37913269e-01 1.02141380e-01
1.44727767e+00 -1.04207015e+00 -3.32583934e-01 1.26483268e-03
1.93721011e-01 -6.83337748e-01 1.05149150e+00 -1.82759073e-02
-4.84584361e-01 -8.26781571e-01 -9.58827019e-01 -8.07162970e-02
-9.57193494e-01 3.11411351e-01 7.00934052e-01 3.28415185e-01
-9.88988161e-01 3.42929423e-01 -3.36771339e-01 -4.78890210e-01
4.54742432e-01 4.75359112e-01 -6.96524143e-01 -2.29008466e-01
-1.16735339e+00 4.02330935e-01 7.74204850e-01 1.39105439e-01
-1.22081923e+00 -6.08948171e-01 -6.79594636e-01 3.10827315e-01
3.49158496e-01 -5.40423274e-01 1.09920943e+00 -5.68009734e-01
-7.67166615e-01 1.29243183e+00 2.95997299e-02 -1.84677765e-01
1.46729022e-01 -4.64462414e-02 -4.59072530e-01 2.45684758e-01
5.64134777e-01 9.39327955e-01 1.12773764e+00 -1.17906845e+00
-8.04798484e-01 -5.69150269e-01 1.92657366e-01 7.86859728e-03
-6.23632908e-01 4.04584855e-02 -8.89250517e-01 -9.03540134e-01
-2.09234253e-01 -6.75724089e-01 1.76603511e-01 1.42222539e-01
-1.96715802e-01 -6.50194943e-01 7.15018690e-01 -2.19620019e-01
1.33094800e+00 -2.29192710e+00 -9.74489152e-02 -1.88057050e-01
2.91236281e-01 5.73916018e-01 -5.79954863e-01 3.97548109e-01
2.18598843e-02 2.86136538e-01 1.02583200e-01 -8.91727805e-02
2.22969532e-01 4.67840619e-02 -5.26649237e-01 2.25784913e-01
2.58015424e-01 1.24867141e+00 -9.19040859e-01 -6.11621380e-01
3.15984905e-01 3.26219976e-01 -1.98253930e-01 4.47385490e-01
2.40483205e-03 9.27413106e-02 -8.26967776e-01 1.17970765e+00
8.12475026e-01 -5.98142743e-01 -2.33995944e-01 -5.98646581e-01
-9.64373350e-03 -7.04191148e-01 -9.86126363e-01 1.58442545e+00
-3.26343745e-01 2.06853613e-01 -1.08973406e-01 -6.95199668e-01
9.08041418e-01 -2.30952293e-01 5.35742827e-02 -9.86461759e-01
9.33813304e-02 4.85767722e-01 -5.99314392e-01 -2.30792627e-01
7.86406457e-01 -1.28570572e-01 -4.86513019e-01 4.36595380e-01
2.76347727e-01 2.44375058e-02 3.04739058e-01 2.67022252e-01
7.64458001e-01 -4.82681617e-02 3.92309755e-01 -2.05614895e-01
6.96412444e-01 5.09774275e-02 2.65749633e-01 1.04635251e+00
-4.57638323e-01 6.94653630e-01 -3.21206033e-01 -6.36378765e-01
-7.66897380e-01 -1.04262495e+00 -1.81217879e-01 1.57005239e+00
7.24140942e-01 -3.48103762e-01 -5.77436209e-01 -6.45667493e-01
3.52618009e-01 -1.54749110e-01 -7.75356889e-01 -3.39188069e-01
-2.09300682e-01 -6.89268231e-01 8.58843207e-01 5.15429676e-01
8.81777644e-01 -1.31157935e+00 -3.07694376e-01 -3.89243439e-02
-2.95585543e-01 -9.86288607e-01 -7.76361108e-01 3.75875793e-02
-3.57156664e-01 -1.14974880e+00 -1.05389273e+00 -9.06057477e-01
4.94136423e-01 8.26479614e-01 1.07652056e+00 3.33003789e-01
-8.62718165e-01 3.29193562e-01 -6.17088616e-01 -1.32264867e-01
9.94980335e-02 1.16760142e-01 1.28128782e-01 8.28579739e-02
7.14991987e-01 -3.78360488e-02 -8.97486806e-01 4.78058249e-01
-1.34691012e+00 -4.56877649e-01 6.20124280e-01 1.08582258e+00
1.00035465e+00 7.47492015e-02 6.98236823e-01 -4.75256681e-01
6.98674440e-01 -3.10620725e-01 -6.88397765e-01 7.60035634e-01
-4.40786839e-01 1.03507146e-01 7.48654306e-01 -5.53186297e-01
-7.16815829e-01 -1.57483026e-01 -7.76426122e-02 -9.29567456e-01
-3.55881244e-01 7.71833807e-02 -6.47554696e-02 -3.78229588e-01
5.07743061e-01 4.56787169e-01 -6.32934690e-01 -7.69072413e-01
3.63971382e-01 1.16083896e+00 7.66490877e-01 -5.10622442e-01
8.75742197e-01 4.55832273e-01 -3.83059800e-01 -5.34282386e-01
-1.06069338e+00 -1.02586198e+00 -6.47168279e-01 6.54317662e-02
8.32109809e-01 -1.12960160e+00 -1.00946331e+00 6.70918643e-01
-8.77952278e-01 -3.66114751e-02 -2.97312379e-01 8.83485675e-02
-2.69056231e-01 3.48608494e-01 -5.13431728e-01 -5.97892761e-01
-7.92787552e-01 -1.04060209e+00 1.88577187e+00 4.11914945e-01
2.36043334e-01 -5.68281054e-01 -8.43348280e-02 3.25155109e-01
5.75256586e-01 -2.41128042e-01 6.61033034e-01 -7.00714052e-01
-7.09012806e-01 -3.78649116e-01 -9.79057848e-01 1.58614725e-01
1.72575623e-01 -2.44614974e-01 -1.14558876e+00 -6.13437057e-01
-5.08654237e-01 -8.55050504e-01 1.18915212e+00 -4.49186303e-02
1.54513705e+00 -1.04603052e-01 -5.24879217e-01 8.64249885e-01
1.62404609e+00 1.58863962e-01 2.80461401e-01 4.62861985e-01
6.65047467e-01 3.62668186e-01 1.04618633e+00 5.01687169e-01
2.67365128e-01 6.56166971e-01 3.46663594e-01 1.06585763e-01
-3.50053310e-01 -4.21391457e-01 -4.32560205e-01 4.48899686e-01
1.20146453e-01 -4.02978182e-01 -6.28160417e-01 6.57801628e-01
-1.68296969e+00 -8.26011956e-01 5.61750293e-01 2.04181647e+00
3.88197392e-01 -3.42716068e-01 -7.99792185e-02 -7.91716352e-02
1.01586783e+00 1.00876622e-01 -9.89546359e-01 -3.87965590e-02
-2.77627140e-01 6.34380579e-02 3.74843597e-01 2.34241299e-02
-1.48749578e+00 1.29293168e+00 5.54057455e+00 1.56119633e+00
-1.29524374e+00 8.70195553e-02 5.25687039e-01 1.53919488e-01
-2.46027410e-02 -4.28402662e-01 -1.07913804e+00 3.60250175e-01
4.25744474e-01 -1.70620903e-01 4.86742496e-01 9.59730685e-01
-4.43806350e-01 2.47850671e-01 -7.66729772e-01 1.48778582e+00
2.43754938e-01 -1.23916924e+00 6.84558988e-01 -2.01400369e-01
4.27218974e-01 3.74364667e-02 -8.56775492e-02 5.51780105e-01
2.50437349e-01 -9.72482026e-01 6.01122916e-01 3.13083053e-01
1.54268622e+00 -5.37085116e-01 7.35828280e-01 3.23904574e-01
-1.68390918e+00 -3.16826642e-01 -8.58126581e-01 3.67694527e-01
-2.96104223e-01 1.76957697e-01 -3.66646498e-01 6.06904268e-01
1.17859769e+00 6.19066656e-01 -8.55574965e-01 1.18802619e+00
2.51332998e-01 4.63826070e-03 -1.22953713e-01 -9.75840017e-02
2.63681948e-01 2.26767048e-01 4.62790541e-02 1.15738344e+00
2.02598080e-01 1.54319286e-01 4.06209111e-01 5.51021755e-01
-3.65305871e-01 8.20457488e-02 -2.72764176e-01 -4.01159942e-01
6.74066186e-01 1.60531354e+00 -8.53432298e-01 -5.07639945e-01
-8.40250477e-02 1.20341837e+00 5.28377771e-01 2.86626726e-01
-5.06529212e-01 -5.90928614e-01 7.01192141e-01 -2.93773949e-01
4.58681256e-01 3.36906254e-01 5.17442048e-01 -1.48106420e+00
-1.82730794e-01 -7.46724546e-01 9.44007695e-01 -9.37744498e-01
-1.85031462e+00 1.01865959e+00 -1.45707726e-01 -1.28768778e+00
-1.49183795e-01 -5.39589703e-01 -2.78740041e-02 8.26237619e-01
-1.85114622e+00 -1.56418240e+00 -7.78972626e-01 8.42297554e-01
6.90979242e-01 -7.18494058e-02 1.12185657e+00 6.08557403e-01
-3.81898344e-01 8.08045089e-01 2.39580855e-01 1.14921249e-01
9.32664931e-01 -9.45207834e-01 1.98814526e-01 4.08868194e-01
-8.25617462e-02 6.82696223e-01 9.23780426e-02 -5.65611899e-01
-1.38223207e+00 -1.51651740e+00 5.38211107e-01 -2.02546328e-01
7.09721684e-01 -3.45247835e-01 -7.58467495e-01 2.46502578e-01
-5.06529272e-01 4.01650667e-01 5.52898705e-01 -2.16861874e-01
-8.82198155e-01 -3.96055967e-01 -1.26262259e+00 1.95927367e-01
1.02816212e+00 -8.99157882e-01 -5.64277291e-01 3.29050899e-01
9.41079199e-01 8.01693499e-02 -9.09202576e-01 6.14862382e-01
8.60712528e-01 -4.66137081e-01 1.35298705e+00 -5.28194129e-01
-1.73273049e-02 -5.24011552e-01 -5.53699493e-01 -1.02879870e+00
-6.68681681e-01 5.72813898e-02 1.49441496e-01 1.23471653e+00
-2.37155423e-01 -5.01147926e-01 6.50967240e-01 1.45476595e-01
7.43028224e-02 -3.22598606e-01 -6.08624876e-01 -1.03608596e+00
2.44095884e-02 -1.34024590e-01 1.00741136e+00 6.94703043e-01
-5.32214999e-01 1.81502461e-01 -3.20901006e-01 1.85713440e-01
7.63328612e-01 9.52077866e-01 6.32812858e-01 -1.48430550e+00
-1.06225446e-01 -4.11181301e-01 -6.49762571e-01 -1.08414686e+00
1.48883998e-01 -8.77084196e-01 1.92270532e-01 -1.45631993e+00
7.43146241e-01 -5.22828817e-01 -5.92272282e-01 5.64839482e-01
-2.71242321e-01 1.15309381e+00 3.01616311e-01 7.42259264e-01
-1.20430291e+00 5.67148983e-01 1.16996348e+00 -4.99673188e-01
2.03635886e-01 -3.84231240e-01 -7.81325042e-01 3.33938032e-01
4.70902085e-01 -1.72930345e-01 -2.37721086e-01 -3.85796130e-01
-1.45270094e-01 -1.30028322e-01 5.03007472e-01 -7.33301103e-01
2.97631711e-01 -5.77366985e-02 6.17356956e-01 -7.79625237e-01
4.40512031e-01 -7.46653676e-01 2.35771149e-01 3.68554145e-01
-2.00385988e-01 -2.80320168e-01 3.04538906e-01 5.48072159e-01
-7.20542431e-01 -1.62709579e-01 6.73162222e-01 -3.34432393e-01
-1.43494380e+00 7.84329295e-01 1.04176819e-01 -2.10087225e-02
9.60169435e-01 -1.88611984e-01 -6.53140843e-01 6.33991137e-02
-6.47646308e-01 3.07767212e-01 5.05971253e-01 7.59497941e-01
6.65001690e-01 -1.42507970e+00 -5.87609768e-01 3.05526912e-01
9.63227093e-01 -2.27030620e-01 7.22104132e-01 7.31484145e-02
-4.21712816e-01 9.52703834e-01 -2.16019809e-01 -5.30417740e-01
-1.30095255e+00 1.15824485e+00 1.78551733e-01 -2.13925093e-01
-1.55715436e-01 9.91378844e-01 9.25499558e-01 -6.13385618e-01
3.12612176e-01 2.42750570e-01 -3.49894345e-01 2.80247360e-01
8.98337185e-01 1.28657237e-01 1.20414458e-01 -8.81444991e-01
-5.88819444e-01 1.06762111e+00 -2.18042925e-01 6.62580967e-01
1.15503490e+00 -4.24053639e-01 -1.83000281e-01 1.34669602e-01
1.42231560e+00 -4.76783544e-01 -8.79651546e-01 -3.53117645e-01
-2.32624054e-01 -5.93449116e-01 -2.24971399e-01 -1.01091135e+00
-9.79943871e-01 1.02365768e+00 8.94010007e-01 2.27222845e-01
1.36517751e+00 3.51051211e-01 7.23970294e-01 6.48509920e-01
7.32912481e-01 -8.04170609e-01 2.90939957e-02 4.08187300e-01
8.63572299e-01 -1.50765932e+00 -2.13971555e-01 -3.46600711e-01
-1.22178286e-01 1.01585162e+00 7.18642771e-01 7.82289915e-03
5.81961989e-01 -1.34102613e-01 7.24447444e-02 -2.50839472e-01
-5.30152142e-01 -5.18875778e-01 6.06793046e-01 5.23303926e-01
-7.17286393e-02 2.13306472e-01 4.92927618e-02 6.65596366e-01
1.30024493e-01 1.51672915e-01 -2.31920838e-01 6.41486049e-01
-6.58410966e-01 -8.81157160e-01 -4.43244964e-01 6.97469890e-01
-4.19851393e-01 -2.55344152e-01 -6.00228250e-01 5.56159437e-01
1.03402235e-01 9.10138965e-01 1.72780409e-01 -4.28889781e-01
3.07332784e-01 -3.20647299e-01 1.94885656e-01 -5.89637041e-01
-5.20777166e-01 -8.94485856e-05 -3.35924596e-01 -6.22643292e-01
-3.49144578e-01 2.25691330e-02 -8.87947798e-01 -2.82641649e-01
-5.25800049e-01 2.78129250e-01 5.16842723e-01 7.79849827e-01
4.85467017e-01 2.12227762e-01 5.82993627e-01 -8.66055131e-01
-5.00199020e-01 -8.53019476e-01 -8.35401654e-01 5.94812930e-01
3.04682732e-01 -8.00096989e-01 -4.29407090e-01 -2.06810027e-01] | [9.714996337890625, 1.927155613899231] |
ae5f620c-c47b-4959-9af3-4ccfbe0ded90 | semantic-space-grounded-weighted-decoding-for | 2305.02820 | null | https://arxiv.org/abs/2305.02820v1 | https://arxiv.org/pdf/2305.02820v1.pdf | Semantic Space Grounded Weighted Decoding for Multi-Attribute Controllable Dialogue Generation | Controlling chatbot utterance generation with multiple attributes such as personalities, emotions and dialogue acts is a practically useful but under-studied problem. We propose a novel controllable generation framework called DASC that possesses strong controllability with weighted decoding paradigm, while improving generation quality with the grounding in an attribute semantics space. Generation with multiple attributes is then intuitively implemented with an interpolation of multiple attribute embeddings. Experiments show that DASC can achieve state-of-the-art control accuracy in 3-aspect controllable generation tasks while also producing interesting and reasonably sensible responses, even if in an out-of-distribution robustness test. Visualization of the meaningful representations learned in the attribute semantic space also supports its effectiveness. | ['Kenny Q. Zhu', 'Mengyue Wu', 'Zhiling Zhang'] | 2023-05-04 | null | null | null | null | ['dialogue-generation', 'dialogue-generation'] | ['natural-language-processing', 'speech'] | [ 1.84833437e-01 6.79937720e-01 -1.04930729e-01 -6.82266831e-01
-7.04078496e-01 -4.71966207e-01 8.82709205e-01 3.15390006e-02
3.82155813e-05 9.68070030e-01 7.11654246e-01 3.49629074e-01
-1.83993295e-01 -9.93733346e-01 -2.53223568e-01 -7.43688464e-01
1.32646963e-01 7.89555609e-01 -4.28912610e-01 -1.02212608e+00
8.34550261e-02 -1.69654742e-01 -1.47040927e+00 3.37336481e-01
1.19146943e+00 8.48897934e-01 -2.28504941e-01 7.83675909e-01
-6.49856254e-02 9.32385206e-01 -9.50250506e-01 -8.64874780e-01
-9.54441503e-02 -4.88126367e-01 -8.42483103e-01 -1.20804764e-01
5.61341569e-02 7.46363476e-02 1.61043599e-01 8.49110901e-01
7.14500070e-01 2.79024482e-01 7.63567150e-01 -1.57009578e+00
-1.46767461e+00 9.55858529e-01 3.50318760e-01 -7.09949732e-01
7.87611246e-01 6.27700031e-01 1.39456010e+00 -4.20402706e-01
6.17533147e-01 1.71606445e+00 3.47391069e-01 1.24131405e+00
-1.37048709e+00 -5.15155911e-01 1.13083452e-01 -3.13689321e-01
-8.62977564e-01 -3.65237100e-03 6.59856081e-01 -1.51639163e-01
9.77972329e-01 3.96500021e-01 5.49527049e-01 1.85967386e+00
-9.49342176e-03 8.53039742e-01 1.14816189e+00 -3.00077528e-01
4.24615741e-01 4.49758232e-01 1.56240225e-01 6.04926646e-01
-8.42570961e-02 -1.00813163e-02 -6.83350801e-01 -3.60415995e-01
5.14168382e-01 -2.96816379e-01 -1.96852967e-01 -1.58652782e-01
-1.45333183e+00 1.39481509e+00 2.43239582e-01 4.41259257e-02
-3.99294287e-01 4.46966618e-01 4.35520351e-01 7.00606465e-01
7.03260779e-01 1.23921418e+00 -5.55890977e-01 -7.18132138e-01
-1.28595129e-01 7.03768492e-01 9.35539424e-01 1.61026216e+00
3.60545695e-01 5.89242578e-01 -9.88219798e-01 8.52778256e-01
-3.16119231e-02 6.45419419e-01 5.79235137e-01 -1.07786500e+00
4.63266343e-01 8.83418977e-01 2.27893040e-01 -6.92790091e-01
-3.85495424e-01 -7.62499124e-02 -1.05264390e+00 1.03487261e-01
2.93319166e-01 -5.05034626e-01 -4.59991127e-01 2.23233724e+00
4.26622391e-01 -3.14301401e-01 3.83276969e-01 8.45090687e-01
6.69947267e-01 4.61947441e-01 3.75307262e-01 -1.12303726e-01
1.34277844e+00 -8.86291504e-01 -1.13698876e+00 1.70936584e-01
5.04637241e-01 -3.78148675e-01 1.67881739e+00 3.50262642e-01
-8.78218949e-01 -4.69140798e-01 -8.09895933e-01 -2.24865526e-01
-4.47191209e-01 -1.10094272e-01 1.03875732e+00 7.03044951e-01
-9.37728763e-01 5.56920290e-01 -1.15433820e-01 -3.17004293e-01
1.26768440e-01 3.11745226e-01 -4.10512745e-01 3.27102602e-01
-1.82400906e+00 1.28214955e+00 2.61318624e-01 -5.30552626e-01
-7.81204879e-01 -7.57758856e-01 -1.00802338e+00 -5.68381231e-03
2.62378275e-01 -8.50373864e-01 1.15473032e+00 -6.93290472e-01
-2.09885049e+00 3.52817953e-01 4.34778571e-01 -3.03641617e-01
5.31617999e-01 -3.20292145e-01 -4.81845766e-01 -1.55228049e-01
-5.33388592e-02 8.33680511e-01 1.01940787e+00 -9.23813522e-01
-4.00537133e-01 -2.66148299e-01 3.60130697e-01 3.18227023e-01
-7.28473246e-01 -1.12290367e-01 3.67286444e-01 -6.87492013e-01
-5.43283880e-01 -9.73088980e-01 -3.66570354e-01 -2.88075477e-01
-4.18349683e-01 -8.29242945e-01 6.12421274e-01 -1.93242773e-01
1.22001266e+00 -1.59129524e+00 6.43388391e-01 -2.18001336e-01
2.63585210e-01 -7.74512067e-02 -3.93196553e-01 5.39071441e-01
1.66627109e-01 2.97640562e-01 2.06999958e-01 -2.07935259e-01
6.36700690e-01 -6.26329184e-02 -2.65432000e-01 1.64613217e-01
7.06414163e-01 9.22078431e-01 -1.13398397e+00 -2.48482406e-01
1.16109453e-01 3.53857815e-01 -9.00117040e-01 7.42943108e-01
-5.24270713e-01 3.03304225e-01 -7.34622538e-01 5.32222927e-01
8.96058530e-02 1.40415221e-01 5.42821474e-02 2.17641234e-01
2.47187197e-01 1.42745584e-01 -9.13198471e-01 1.73885441e+00
-7.15659201e-01 1.67632699e-01 -2.06779152e-01 -6.20801985e-01
1.46296287e+00 6.98026896e-01 2.35000387e-01 -1.86985627e-01
1.49255991e-01 -1.22825250e-01 2.00379249e-02 -6.44782662e-01
8.97451580e-01 -4.92310017e-01 -8.71053874e-01 6.16936624e-01
6.47270560e-01 -8.89608800e-01 1.53032588e-02 2.25089774e-01
8.60397100e-01 2.42653303e-02 3.58043462e-01 -5.35280287e-01
1.17605925e-01 -2.66304195e-01 2.87587941e-01 7.15888083e-01
-2.58663476e-01 6.69024467e-01 9.10763860e-01 -2.91440874e-01
-1.01506805e+00 -8.80808294e-01 2.15005353e-01 1.41335440e+00
-2.89681911e-01 -5.07636547e-01 -9.03880298e-01 -8.14499080e-01
1.32783026e-01 1.35684204e+00 -9.74019527e-01 -5.93138337e-01
-1.36441723e-01 -4.38799948e-01 8.49602759e-01 6.83233798e-01
1.32084295e-01 -1.32416081e+00 -4.56645101e-01 3.26762915e-01
-2.69959569e-01 -7.37835169e-01 -4.91030365e-01 1.93023652e-01
-4.59665477e-01 -7.36943245e-01 -3.81254226e-01 -4.09515679e-01
2.48258799e-01 -4.15845692e-01 1.30688763e+00 -1.84421763e-01
4.23657931e-02 4.09649044e-01 -6.95425987e-01 -6.01747394e-01
-6.11407876e-01 2.03938082e-01 3.53405088e-01 3.98308598e-02
3.88637781e-01 -4.34999675e-01 -1.25470862e-01 7.18677863e-02
-7.52768099e-01 -4.29728031e-02 5.02665415e-02 1.38314199e+00
-9.61224213e-02 -5.73159039e-01 1.28218615e+00 -1.05643439e+00
1.65996790e+00 -5.12594938e-01 -7.98596293e-02 3.46763432e-01
-6.92795575e-01 4.53307152e-01 9.28378105e-01 -7.35939741e-01
-1.16642249e+00 -2.36215904e-01 1.30413279e-01 -2.48832971e-01
-2.45350018e-01 4.66444343e-03 -3.32961708e-01 6.62254035e-01
9.22806323e-01 -4.15517278e-02 3.56479913e-01 -1.42005878e-03
1.01365721e+00 8.70923579e-01 -3.00347619e-02 -1.06321180e+00
5.74643672e-01 -2.00350016e-01 -2.67858207e-01 -4.61040020e-01
-7.30341434e-01 -1.73318788e-01 -3.70382249e-01 -3.35547954e-01
1.08347046e+00 -6.38328612e-01 -1.06535614e+00 2.55665720e-01
-1.27626348e+00 -3.41669649e-01 -5.18102646e-01 2.46018887e-01
-1.07155418e+00 -4.03271794e-01 -6.90637887e-01 -1.09801948e+00
-6.43100441e-01 -1.13379157e+00 1.21028268e+00 7.69536644e-02
-9.37034547e-01 -1.02422750e+00 -9.68697295e-03 4.60152388e-01
8.11555028e-01 4.91767049e-01 1.19520080e+00 -1.03553545e+00
-1.85988963e-01 -4.56829488e-01 3.16602319e-01 3.89360070e-01
2.51466632e-01 4.47905809e-02 -1.15916920e+00 2.01181602e-02
-5.10070659e-02 -1.01092482e+00 4.64237630e-01 -8.83482769e-02
8.82659912e-01 -6.14550054e-01 3.55974615e-01 2.19286606e-01
9.92580056e-01 6.98075369e-02 5.29752791e-01 -1.65502429e-01
5.25151849e-01 9.30907309e-01 8.50015819e-01 9.16329384e-01
2.58289218e-01 7.37323999e-01 4.25408095e-01 1.86935589e-01
2.23243803e-01 -5.88169932e-01 4.84837264e-01 8.72120500e-01
-3.46590042e-01 -3.66722822e-01 -2.50854790e-01 2.88999885e-01
-1.96911204e+00 -1.13530552e+00 -8.04842729e-03 1.92831552e+00
1.33130276e+00 -2.28932664e-01 9.53587666e-02 -1.75255254e-01
7.43409455e-01 2.23420456e-01 -3.38456750e-01 -1.07812548e+00
-1.96735486e-01 4.62189049e-01 2.31969613e-03 5.21777153e-01
-8.93200219e-01 1.10363603e+00 6.59639835e+00 1.02610552e+00
-6.27264321e-01 1.07589215e-01 8.26669276e-01 7.54665118e-04
-7.65337348e-01 -4.44333822e-01 -6.12001181e-01 3.80475551e-01
8.13566089e-01 -3.55974704e-01 6.81147039e-01 1.27354062e+00
2.63931036e-01 5.34582317e-01 -1.31917512e+00 5.52509189e-01
1.76417738e-01 -1.19900930e+00 2.17805088e-01 -2.72232056e-01
7.74149299e-01 -9.09472704e-01 4.06702042e-01 9.59664524e-01
8.76863301e-01 -1.39921081e+00 7.01560259e-01 4.47304100e-01
1.02427912e+00 -8.94119620e-01 6.39557362e-01 2.09972844e-01
-6.31517947e-01 8.40025675e-03 -4.13255185e-01 -2.82092571e-01
-3.12269293e-02 1.02743737e-01 -9.79232013e-01 2.86146045e-01
1.66985422e-01 3.21649998e-01 -3.25390339e-01 3.66456024e-02
-4.99644995e-01 5.07226288e-01 1.78928107e-01 -9.40005839e-01
3.28667700e-01 -3.85581702e-01 3.91668916e-01 1.21275580e+00
3.14203978e-01 3.17395598e-01 -6.45608902e-02 1.25249982e+00
-6.10188469e-02 1.38248056e-01 -9.83691454e-01 -1.42300382e-01
6.07849479e-01 1.29065859e+00 3.92179489e-02 -3.12917084e-01
1.38382122e-01 8.89567971e-01 5.15682101e-01 1.92767829e-01
-1.06991291e+00 -4.73815709e-01 1.20450568e+00 -3.49268079e-01
-1.15550883e-01 2.20392898e-01 -5.23438394e-01 -1.02017260e+00
-3.39640826e-01 -1.14199746e+00 1.37473851e-01 -6.93341017e-01
-1.58094978e+00 6.92909598e-01 -1.50397904e-02 -1.23871911e+00
-6.43988609e-01 -5.18355012e-01 -6.81999564e-01 6.88817322e-01
-7.48973846e-01 -1.28666782e+00 -8.46556947e-02 7.68633842e-01
8.30883920e-01 -5.37944555e-01 1.66613197e+00 -3.72790039e-01
-3.21844786e-01 9.96662140e-01 -1.95998937e-01 -1.50964916e-01
7.97009706e-01 -1.79085207e+00 2.22441509e-01 3.37949172e-02
-2.57216752e-01 8.55761468e-01 9.23554242e-01 -3.19196939e-01
-1.51322770e+00 -1.17981589e+00 9.99175012e-01 -6.52238131e-01
6.81457937e-01 -7.26923704e-01 -4.49519813e-01 4.02570486e-01
6.80925310e-01 -5.09597003e-01 7.26311088e-01 5.52790344e-01
-5.29487967e-01 8.99049118e-02 -1.51904833e+00 9.14203465e-01
8.73289704e-01 -3.87078971e-01 -6.03977323e-01 5.99369049e-01
1.44685495e+00 -2.04829335e-01 -1.12651753e+00 2.81960275e-02
3.08178753e-01 -5.73012471e-01 6.65693343e-01 -1.28361547e+00
7.31489539e-01 1.96480989e-01 -3.15144569e-01 -2.02156997e+00
-4.34646398e-01 -1.26375318e+00 1.35153949e-01 1.62581086e+00
5.41016281e-01 -5.20412982e-01 5.00555873e-01 1.00416696e+00
-1.50181860e-01 -7.19274402e-01 -7.87076414e-01 -6.17045462e-01
3.50562930e-01 -3.87544751e-01 1.01902342e+00 1.07823968e+00
8.94943118e-01 9.03018296e-01 -8.82708848e-01 -4.29914713e-01
4.16482300e-01 -1.10484816e-01 9.25026417e-01 -9.67179775e-01
-1.77280411e-01 -2.68435299e-01 -3.89353335e-01 -5.93555152e-01
5.13197660e-01 -7.81279087e-01 1.16549596e-01 -1.19020867e+00
-8.88537019e-02 -4.36668456e-01 -7.18938513e-03 4.10148114e-01
-6.01156533e-01 -2.58865952e-01 3.89725834e-01 -5.77606082e-01
-6.46218359e-01 1.22393036e+00 1.50779629e+00 -3.59064013e-01
1.00505657e-01 1.05729222e-01 -1.05181158e+00 2.41717279e-01
1.08489180e+00 -2.72091776e-01 -6.69988394e-01 -5.08600846e-02
-1.81818604e-01 2.84757614e-01 -8.36794898e-02 -5.80832124e-01
-2.19982207e-01 -6.18238151e-01 -1.24989189e-01 2.23236516e-01
6.87215328e-01 -4.22044426e-01 -1.91723228e-01 1.00633837e-01
-1.23673069e+00 1.81218252e-01 -2.78358787e-01 6.12968147e-01
-1.60272062e-01 -1.18627902e-02 7.13500381e-01 -1.23584196e-01
-1.53243795e-01 -7.95615241e-02 -5.50942361e-01 4.11791712e-01
1.00364852e+00 1.89469740e-01 -5.09691477e-01 -9.76331294e-01
-4.09932256e-01 3.49394888e-01 3.88825610e-02 8.36335123e-01
6.92275286e-01 -1.72918522e+00 -9.68409061e-01 1.45611152e-01
3.87824893e-01 -3.82045180e-01 3.71874794e-02 2.64824092e-01
-2.01857999e-01 3.01672578e-01 -1.99989736e-01 -3.58806193e-01
-1.00463104e+00 5.62249064e-01 -1.92413349e-02 -4.09863800e-01
-2.03219488e-01 1.11942339e+00 -4.11109030e-02 -9.66769516e-01
2.61238426e-01 -4.03493106e-01 -1.62298292e-01 1.64571419e-01
3.31220061e-01 4.47055787e-01 -1.43845886e-01 -2.90628016e-01
1.40466318e-01 -2.22563297e-02 2.56396353e-01 -1.92886457e-01
1.12837541e+00 4.27794486e-01 -1.49226014e-06 4.05316681e-01
9.90584016e-01 -3.34060550e-01 -1.10338926e+00 1.89338461e-01
-1.26186922e-01 -3.76238734e-01 -5.41037917e-01 -1.18609691e+00
-8.52337241e-01 8.49866331e-01 2.65031695e-01 4.08732027e-01
6.49382532e-01 -3.31628054e-01 4.55099910e-01 4.67224240e-01
5.89525759e-01 -1.54435503e+00 4.91847187e-01 6.99470162e-01
1.30076468e+00 -1.19498706e+00 -4.90924984e-01 -1.66039333e-01
-1.77562916e+00 1.05319583e+00 9.40110266e-01 -1.80188954e-01
2.29397863e-01 2.77601779e-01 2.34049901e-01 -1.38731048e-01
-1.56640649e+00 -3.72659415e-02 -8.70909616e-02 1.09734619e+00
7.15888619e-01 6.70526862e-01 -5.11295557e-01 9.97349799e-01
-5.43370247e-01 -5.16117692e-01 5.89134276e-01 2.63455719e-01
-3.30164373e-01 -1.12677491e+00 1.57180894e-02 3.54156613e-01
-1.92874208e-01 -1.47017822e-01 -7.05591261e-01 9.40484822e-01
-1.07952014e-01 1.46653008e+00 -1.78186789e-01 -8.47865820e-01
5.23114204e-01 3.47124934e-01 1.43699899e-01 -8.07111263e-01
-1.24216831e+00 -3.63652945e-01 5.28185427e-01 -4.96101975e-01
-1.08396642e-01 -4.00766194e-01 -1.02608120e+00 -4.18441325e-01
-4.96010214e-01 4.35324073e-01 2.46829405e-01 5.74425578e-01
3.03446859e-01 5.69189548e-01 9.03255761e-01 -3.92827988e-01
-1.37139869e+00 -1.60008526e+00 -6.62974894e-01 1.03801191e+00
-1.29227951e-01 -6.55944049e-01 -3.10778767e-01 -2.11348340e-01] | [12.780756950378418, 8.12799072265625] |
add22b24-7941-48cd-82a8-77057963d61c | toward-unlimited-self-learning-monte-carlo | 2211.14024 | null | https://arxiv.org/abs/2211.14024v1 | https://arxiv.org/pdf/2211.14024v1.pdf | Toward Unlimited Self-Learning Monte Carlo with Annealing Process Using VAE's Implicit Isometricity | Self-learning Monte Carlo (SLMC) methods are recently proposed to accelerate Markov chain Monte Carlo (MCMC) methods by using a machine learning model.With generative models having latent variables, SLMC methods realize efficient Monte Carlo updates with less autocorrelation. However, SLMC methods are difficult to directly apply to multimodal distributions for which training data are difficult to obtain. In this paper, we propose a novel SLMC method called the ``annealing VAE-SLMC" to drastically expand the range of applications. Our VAE-SLMC utilizes a variational autoencoder (VAE) as a generative model to make efficient parallel proposals independent of any previous state by applying the theoretically derived implicit isometricity of the VAE. We combine an adaptive annealing process to the VAE-SLMC, making our method applicable to the cases where obtaining unbiased training data is difficult in practical sense due to slow mixing. We also propose a parallel annealing process and an exchange process between chains to make the annealing operation more precise and efficient. Experiments validate that our method can proficiently obtain unbiased samples from multiple multimodal toy distributions and practical multimodal posterior distributions, which is difficult to achieve with the existing SLMC methods. | ['Yuhei Umeda', 'Hiromoto Masayuki', 'Akira Nakagawa', 'Yuma Ichikawa'] | 2022-11-25 | null | null | null | null | ['self-learning'] | ['natural-language-processing'] | [ 4.84861135e-02 -2.11271048e-01 -8.76258388e-02 -2.68871814e-01
-9.18761313e-01 -2.35138580e-01 9.67071652e-01 -4.31548715e-01
-4.73024487e-01 1.20692205e+00 -1.42352963e-02 -3.16592097e-01
2.16071263e-01 -9.49222565e-01 -8.57080400e-01 -1.25434959e+00
4.47055519e-01 1.18942034e+00 3.04291159e-01 1.30954534e-01
1.09160870e-01 4.27965432e-01 -1.38698745e+00 -3.11701857e-02
1.16348886e+00 3.73095691e-01 3.08984756e-01 8.02578509e-01
-1.74660459e-01 4.26296502e-01 -3.91626269e-01 -4.00398076e-01
-1.25157669e-01 -9.69811559e-01 -3.68477523e-01 8.35651234e-02
-1.35898829e-01 -4.95778233e-01 2.23261043e-01 1.09062183e+00
5.55051863e-01 3.65309209e-01 1.27501369e+00 -1.03457367e+00
-2.78754801e-01 6.53999507e-01 -6.71029270e-01 -1.36874124e-01
1.38876457e-02 1.26500785e-01 6.35697424e-01 -7.74538159e-01
6.30909860e-01 1.31872261e+00 4.74899858e-01 7.03301072e-01
-1.44523215e+00 -6.47586524e-01 -1.60839766e-01 2.27457210e-01
-1.23522031e+00 -2.81156152e-01 8.23436975e-01 -1.97067425e-01
6.59596860e-01 1.21192068e-01 1.03384614e+00 1.42855048e+00
3.14180225e-01 1.02636790e+00 1.28709364e+00 -4.95405138e-01
9.86509919e-01 -2.17693374e-02 -1.24428041e-01 7.07961380e-01
4.08404022e-01 5.36245033e-02 -2.50892043e-01 -4.68378901e-01
1.07927072e+00 -2.47022999e-03 8.21167007e-02 -6.07644737e-01
-1.11714983e+00 1.24288952e+00 -7.96991214e-02 1.20588824e-01
-4.65041608e-01 4.62553889e-01 1.26654014e-01 -1.03260264e-01
3.34237516e-01 -1.59369372e-02 7.21913353e-02 -3.48797470e-01
-1.31125617e+00 3.47485363e-01 9.71039176e-01 7.94139147e-01
7.76305318e-01 2.68217444e-01 2.22920284e-01 4.96706635e-01
6.95636809e-01 1.07995737e+00 4.65109706e-01 -1.08259034e+00
1.83218971e-01 -1.08996339e-01 3.52233559e-01 -2.93413311e-01
4.59851734e-02 -3.34631234e-01 -1.07584119e+00 1.56125978e-01
4.40948933e-01 -1.93632051e-01 -8.27241123e-01 1.60835958e+00
6.22149050e-01 3.12301636e-01 8.49366412e-02 5.64222991e-01
2.59293586e-01 1.12772143e+00 -1.52538389e-01 -7.08444536e-01
1.01490164e+00 -7.81679690e-01 -7.65632629e-01 6.36240542e-02
4.31084692e-01 -7.59397507e-01 1.09117103e+00 4.66930866e-01
-1.11108172e+00 -2.45544016e-01 -1.00651705e+00 3.34935516e-01
1.22877680e-01 1.32107893e-02 7.55273283e-01 8.91215980e-01
-7.09585786e-01 6.26671970e-01 -1.54248416e+00 -1.08527347e-01
2.34173387e-01 1.44164309e-01 1.06124274e-01 5.90450987e-02
-9.79662478e-01 7.21768320e-01 4.25356686e-01 7.22277686e-02
-1.21187389e+00 -9.25409719e-02 -7.07227349e-01 -5.26460744e-02
1.98442385e-01 -8.64522636e-01 1.28880870e+00 -6.68025732e-01
-2.22763371e+00 1.08961463e-01 -6.14453495e-01 -3.25415790e-01
7.80843675e-01 6.58782423e-02 -2.03579843e-01 1.67107403e-01
-9.33643430e-02 5.16305625e-01 1.23544157e+00 -1.38253963e+00
-8.89359936e-02 2.00841967e-02 -5.27852178e-01 7.48504475e-02
-1.33195221e-02 -4.36582625e-01 -2.88897336e-01 -3.67426693e-01
2.34194443e-01 -1.02683020e+00 -3.83514076e-01 -6.01337373e-01
-1.38322160e-01 -2.95659065e-01 7.09143460e-01 -3.87972981e-01
1.10475457e+00 -1.76788485e+00 3.98647964e-01 4.90903914e-01
-1.01620212e-01 5.91681339e-02 1.12102292e-01 6.66477621e-01
3.74176204e-01 -2.52983630e-01 -5.52444577e-01 -7.43038297e-01
2.22909361e-01 6.50526762e-01 -4.85276043e-01 6.46636188e-01
-3.62785012e-02 8.89052093e-01 -8.86705756e-01 -8.70821178e-01
1.04988270e-01 4.51800466e-01 -9.11496401e-01 1.92962676e-01
-4.19698745e-01 5.59246600e-01 -3.88899028e-01 4.52378124e-01
7.99572170e-01 -5.11920333e-01 4.47130650e-01 6.86443076e-02
2.64910888e-02 -9.83308926e-02 -1.46530676e+00 1.43157959e+00
-3.73803586e-01 2.27684081e-01 -7.81257600e-02 -7.76835859e-01
6.71268106e-01 1.39483988e-01 1.32551165e-02 -1.87636223e-02
1.60204828e-01 5.05404294e-01 -4.94303629e-02 -1.03973404e-01
4.64952558e-01 -6.88910246e-01 -5.44850975e-02 8.57598901e-01
1.10865712e-01 -5.35495222e-01 3.99996370e-01 3.78366172e-01
5.20841599e-01 3.78964484e-01 4.71199453e-01 -2.16824353e-01
4.25101221e-01 -2.24343389e-01 5.74744642e-01 1.02700651e+00
-5.41666970e-02 4.34765697e-01 2.72800535e-01 -1.32615924e-01
-1.33221102e+00 -1.48041689e+00 -1.86878651e-01 6.25243485e-01
9.48937088e-02 -3.01342338e-01 -9.73809421e-01 -5.11911690e-01
-4.65254098e-01 7.53051162e-01 -3.22138011e-01 7.01986775e-02
-6.47173643e-01 -1.22551322e+00 1.87089190e-01 4.78554279e-01
5.46089232e-01 -8.79051626e-01 -3.85236114e-01 4.65962887e-01
-2.82008827e-01 -5.77826440e-01 -2.62994647e-01 -1.51952684e-01
-1.26179099e+00 -5.13124406e-01 -9.42423880e-01 -3.55268598e-01
6.87849820e-01 -2.04542190e-01 7.18635798e-01 -4.08854783e-01
1.44748077e-01 1.33464426e-01 -2.09588513e-01 -1.92087181e-02
-9.29133654e-01 2.09175929e-01 2.37276345e-01 -2.50554085e-02
9.20829922e-02 -8.22595656e-01 -6.04213476e-01 3.01862270e-01
-1.02906549e+00 3.15918326e-01 6.65809512e-01 1.13927817e+00
6.28256619e-01 6.52059317e-02 3.81410420e-01 -5.78713357e-01
5.58982491e-01 -3.29157889e-01 -8.74297321e-01 1.75097302e-01
-6.15080059e-01 5.59122026e-01 5.24365962e-01 -6.72417760e-01
-1.45844376e+00 -1.24541901e-01 -5.02005637e-01 -2.48263896e-01
5.09689003e-02 3.51256102e-01 -2.01668486e-01 3.03975672e-01
3.36969018e-01 4.43297356e-01 2.27876797e-01 -3.95702183e-01
4.58907694e-01 5.47509134e-01 3.93568933e-01 -8.20055842e-01
6.59540832e-01 8.20811093e-01 1.96364462e-01 -6.64795339e-01
-5.18595278e-01 -1.22063845e-01 -4.01432097e-01 -2.59730041e-01
9.20778930e-01 -7.16496289e-01 -8.57290626e-01 6.14052176e-01
-1.05654860e+00 -3.18226129e-01 -2.78359801e-01 1.08973551e+00
-9.16115165e-01 8.50831389e-01 -8.91584635e-01 -1.11805403e+00
-1.98359296e-01 -1.42702615e+00 9.85671699e-01 2.92945594e-01
-1.13848850e-01 -1.10515714e+00 4.62302268e-01 2.00567290e-01
2.98118234e-01 -7.33957589e-02 7.45390177e-01 -3.08860958e-01
-7.49321103e-01 -1.25262827e-01 2.14560643e-01 2.29838163e-01
-2.25898266e-01 3.09254497e-01 -6.34697735e-01 -5.56560934e-01
2.48950765e-01 -2.38219172e-01 1.01936483e+00 6.13567472e-01
7.71902084e-01 -1.34648770e-01 -5.19711852e-01 3.51530671e-01
1.34959364e+00 1.75344408e-01 7.85084307e-01 1.85313728e-02
4.33040291e-01 1.67340249e-01 3.87014329e-01 6.88035846e-01
1.72007516e-01 4.51238781e-01 3.02515566e-01 3.09157223e-01
2.73671359e-01 -4.85961765e-01 7.45575070e-01 1.57791924e+00
-4.00397629e-01 -3.26183736e-01 -7.27474928e-01 3.47141385e-01
-1.95141602e+00 -1.26274121e+00 -2.47364521e-01 2.14598083e+00
1.08997905e+00 6.92731440e-02 -4.64228764e-02 -3.59860733e-02
8.80841255e-01 1.62966862e-01 -4.98420358e-01 -1.98435947e-01
1.13627568e-01 4.05869216e-01 2.62802660e-01 7.53954768e-01
-8.91237974e-01 8.32332730e-01 6.27135801e+00 1.47508955e+00
-7.37309873e-01 5.91797471e-01 2.65451729e-01 -5.46621419e-02
-7.60519564e-01 3.52039397e-01 -1.07010019e+00 6.54956043e-01
1.06454873e+00 2.95715958e-01 3.69322151e-01 8.54624331e-01
7.92063121e-03 -4.48597282e-01 -7.95663714e-01 9.39086556e-01
-4.12042364e-02 -1.33989513e+00 1.01472601e-01 1.99801639e-01
1.09539342e+00 -2.37125829e-01 -5.60268201e-02 2.31243193e-01
7.12999225e-01 -6.57232940e-01 6.19153082e-01 6.49370968e-01
4.36498702e-01 -9.36025739e-01 6.41688406e-01 7.53479362e-01
-9.51654375e-01 2.34610543e-01 -4.91442055e-01 1.60141215e-01
8.46638978e-01 9.44039166e-01 -7.85816073e-01 3.41846347e-01
2.35067189e-01 3.03389281e-01 -1.06377244e-01 7.36252069e-01
-4.33595985e-01 1.03494084e+00 -6.15005136e-01 -4.61630285e-01
3.20232064e-01 -6.96897209e-01 7.01424420e-01 9.07228351e-01
5.81968427e-01 -3.49989057e-01 -1.83151573e-01 1.08128524e+00
3.28208476e-01 -5.67162521e-02 -1.70592397e-01 -2.65255421e-01
4.66851681e-01 8.83722246e-01 -1.03927386e+00 -7.64458060e-01
2.69115269e-02 1.06481934e+00 9.47196186e-02 4.74364787e-01
-1.13933671e+00 -1.90123528e-01 7.00996071e-02 -2.08914191e-01
5.65075219e-01 -5.83418667e-01 9.64622647e-02 -1.69902086e+00
-4.35624748e-01 -7.97168851e-01 1.58859566e-01 -5.69176972e-01
-1.37106729e+00 3.95538658e-01 3.24923664e-01 -1.10837090e+00
-9.41349030e-01 -2.96133727e-01 -5.85862160e-01 6.49640739e-01
-1.04510319e+00 -1.01916158e+00 1.40490770e-01 4.95079160e-01
4.89694208e-01 -1.65133029e-01 6.52343392e-01 -9.76536348e-02
-4.80942845e-01 2.67415911e-01 8.03080916e-01 -3.73593986e-01
5.63196540e-01 -1.05328834e+00 2.13496044e-01 9.05109763e-01
2.04934895e-01 8.03978562e-01 9.64929283e-01 -9.38266754e-01
-1.24465835e+00 -5.90066254e-01 5.27529299e-01 -1.85234293e-01
5.84067941e-01 -4.86589044e-01 -7.82046497e-01 5.73103309e-01
1.75245285e-01 -5.04722416e-01 8.00998867e-01 -5.69053739e-02
-8.33656564e-02 1.04893208e-01 -9.55036163e-01 6.91060066e-01
6.60463750e-01 -4.58622187e-01 -6.28834069e-01 1.49923339e-01
3.16626281e-01 -1.38520867e-01 -7.21809626e-01 1.61964357e-01
6.32986069e-01 -8.71317625e-01 8.27702820e-01 -2.00225115e-01
3.67489636e-01 -4.97843206e-01 -1.49248838e-01 -1.23290074e+00
9.60240047e-03 -6.84393227e-01 -5.70066810e-01 1.04803586e+00
2.04668283e-01 -9.32737350e-01 7.29804397e-01 8.53961185e-02
1.95805639e-01 -4.14156407e-01 -1.14724040e+00 -8.84631217e-01
2.26795226e-01 -5.48368096e-01 6.96427166e-01 6.36297584e-01
-2.54782945e-01 1.38237566e-01 -7.50195682e-01 2.38157623e-02
1.13369930e+00 4.45864588e-01 8.39592218e-01 -1.04616523e+00
-1.00904059e+00 -2.60234565e-01 2.57037073e-01 -1.43051863e+00
1.65944085e-01 -6.58818841e-01 1.70841277e-01 -1.30891037e+00
6.52677119e-01 -2.00159639e-01 2.16634050e-01 -1.55197158e-01
-1.72005191e-01 1.86155230e-01 -8.23094975e-03 5.52292764e-01
-5.07172167e-01 1.10219204e+00 1.35289240e+00 2.03161463e-01
-1.32378712e-01 1.29149988e-01 -1.27725340e-02 7.26481140e-01
6.99414015e-01 -5.78160167e-01 -4.23527896e-01 4.55064215e-02
5.01833737e-01 2.16552839e-01 2.37130538e-01 -9.32581604e-01
1.79464430e-01 -2.18242094e-01 2.99447685e-01 -9.81582880e-01
5.17033279e-01 -3.27926248e-01 6.33306324e-01 6.91978574e-01
7.43427202e-02 -2.52965927e-01 -2.76400238e-01 8.62638354e-01
-2.64379513e-02 -7.28047311e-01 8.31248164e-01 -1.19507782e-01
-2.16406867e-01 4.31767367e-02 -8.58248055e-01 -3.20673257e-01
8.17156553e-01 4.36132438e-02 -1.92623269e-02 -6.60643637e-01
-9.29036319e-01 -7.26020336e-02 6.25519812e-01 -4.80837345e-01
6.15324020e-01 -1.50471950e+00 -3.41246277e-01 2.00411797e-01
-3.99443984e-01 5.85998893e-02 3.82570058e-01 1.08278978e+00
-6.79453552e-01 1.30350515e-01 -3.85732204e-02 -7.57300198e-01
-9.15926516e-01 4.14365202e-01 -2.34911796e-02 -6.39301956e-01
-4.15923566e-01 6.36064649e-01 -7.33000934e-02 -5.10293067e-01
-2.24066287e-01 -1.75970390e-01 2.51965404e-01 8.25295523e-02
3.26748043e-01 5.36206424e-01 -3.82363707e-01 -2.38026962e-01
-5.25874868e-02 4.22915578e-01 -1.06027700e-01 -8.09690714e-01
1.17887342e+00 -2.60981739e-01 -1.03493504e-01 5.84612429e-01
9.48505402e-01 1.95201874e-01 -1.48157609e+00 -2.38972232e-02
-4.53089386e-01 -2.74925530e-01 -1.10100977e-01 -1.10596389e-01
-6.02758527e-01 1.15742028e+00 4.12071407e-01 -2.42302809e-02
7.17460215e-01 1.43449754e-01 9.66228247e-01 4.06301111e-01
5.07380247e-01 -1.20509124e+00 2.26195782e-01 3.43070328e-01
4.18101698e-01 -1.10589933e+00 8.14119354e-02 -9.81321633e-02
-7.18435466e-01 1.07393467e+00 1.37153342e-01 2.00693551e-02
5.65607071e-01 1.90783516e-01 -4.27496910e-01 1.00164644e-01
-6.13061965e-01 -9.77176800e-02 3.77181661e-03 2.10905239e-01
-1.15843406e-02 1.69461787e-01 -5.91991663e-01 4.47323650e-01
-5.90668898e-03 3.02808210e-02 5.44754982e-01 9.76202011e-01
-5.17071545e-01 -1.44927776e+00 -4.51099008e-01 1.74714819e-01
-1.06949650e-01 -9.30460468e-02 2.86515623e-01 6.92721307e-01
-1.63977996e-01 5.36384404e-01 3.24151874e-01 1.55093193e-01
-5.31889379e-01 3.02804470e-01 8.36136878e-01 -2.00495809e-01
1.00225091e-01 7.73824036e-01 -1.06024727e-01 -1.62365168e-01
-6.63120210e-01 -1.03971291e+00 -1.24102068e+00 -6.15593731e-01
-5.48941374e-01 5.02221406e-01 7.61961102e-01 1.05465770e+00
6.77153990e-02 1.13166764e-01 3.33862692e-01 -1.05569959e+00
-9.04428720e-01 -9.54382479e-01 -7.10996091e-01 6.24310598e-02
-9.76013243e-02 -7.55571604e-01 -4.97954518e-01 5.47175221e-02] | [6.936182022094727, 3.895627498626709] |
49a0e396-e17d-4817-ba59-2d2688a40c9d | open-long-tailed-recognition-in-a-dynamic | 2208.08349 | null | https://arxiv.org/abs/2208.08349v1 | https://arxiv.org/pdf/2208.08349v1.pdf | Open Long-Tailed Recognition in a Dynamic World | Real world data often exhibits a long-tailed and open-ended (with unseen classes) distribution. A practical recognition system must balance between majority (head) and minority (tail) classes, generalize across the distribution, and acknowledge novelty upon the instances of unseen classes (open classes). We define Open Long-Tailed Recognition++ (OLTR++) as learning from such naturally distributed data and optimizing for the classification accuracy over a balanced test set which includes both known and open classes. OLTR++ handles imbalanced classification, few-shot learning, open-set recognition, and active learning in one integrated algorithm, whereas existing classification approaches often focus only on one or two aspects and deliver poorly over the entire spectrum. The key challenges are: 1) how to share visual knowledge between head and tail classes, 2) how to reduce confusion between tail and open classes, and 3) how to actively explore open classes with learned knowledge. Our algorithm, OLTR++, maps images to a feature space such that visual concepts can relate to each other through a memory association mechanism and a learned metric (dynamic meta-embedding) that both respects the closed world classification of seen classes and acknowledges the novelty of open classes. Additionally, we propose an active learning scheme based on visual memory, which learns to recognize open classes in a data-efficient manner for future expansions. On three large-scale open long-tailed datasets we curated from ImageNet (object-centric), Places (scene-centric), and MS1M (face-centric) data, as well as three standard benchmarks (CIFAR-10-LT, CIFAR-100-LT, and iNaturalist-18), our approach, as a unified framework, consistently demonstrates competitive performance. Notably, our approach also shows strong potential for the active exploration of open classes and the fairness analysis of minority groups. | ['Stella X. Yu', 'Boqing Gong', 'Jiayun Wang', 'Xiaohang Zhan', 'Zhongqi Miao', 'Ziwei Liu'] | 2022-08-17 | null | null | null | null | ['imbalanced-classification', 'open-set-learning'] | ['miscellaneous', 'miscellaneous'] | [ 2.08159029e-01 1.27517313e-01 -6.02561057e-01 -4.40262109e-01
-9.62387145e-01 -6.02761865e-01 5.56268275e-01 2.43619859e-01
-3.85891885e-01 4.87954885e-01 1.53258711e-01 -4.77555096e-02
-3.66934419e-01 -8.93418014e-01 -7.28536963e-01 -7.50753462e-01
-2.99013734e-01 8.08870554e-01 1.51918111e-02 1.04606412e-01
1.01480223e-01 3.75932872e-01 -1.97926342e+00 5.08165658e-01
7.91429162e-01 1.29370749e+00 -3.51018876e-01 5.43354750e-01
-3.62279773e-01 1.05034530e+00 -3.93441945e-01 -4.05223817e-01
5.21372199e-01 -1.72269698e-02 -8.82546067e-01 9.24624205e-02
9.42927599e-01 -1.04307942e-02 -2.39565283e-01 7.05710471e-01
6.34578466e-01 3.37819546e-01 8.32206666e-01 -1.55201733e+00
-1.12024009e+00 1.93239063e-01 -7.39868820e-01 4.31925595e-01
1.35655582e-01 5.33981919e-01 1.08154082e+00 -1.41118109e+00
8.01778316e-01 1.15761697e+00 4.27692801e-01 6.00543439e-01
-1.59294176e+00 -8.24022174e-01 5.63562095e-01 5.67192256e-01
-1.51920223e+00 -6.08628273e-01 5.16889632e-01 -6.75850928e-01
1.06761706e+00 4.01794761e-01 5.39431691e-01 1.25020087e+00
-5.70236240e-03 7.71942079e-01 1.01286685e+00 -3.04976434e-01
7.51188934e-01 2.78122902e-01 6.66513741e-01 4.70081151e-01
1.15796261e-01 1.75798297e-01 -8.53588283e-01 -2.76475251e-01
-1.08916730e-01 4.44245458e-01 -3.00668389e-01 -1.01910675e+00
-9.46273386e-01 8.73948574e-01 7.01448500e-01 -9.53591764e-02
-8.99819359e-02 -3.40089589e-01 3.27878147e-01 6.26278996e-01
7.73352683e-01 4.85711277e-01 -4.45693344e-01 8.22829232e-02
-9.00938392e-01 -1.28945217e-01 9.97741699e-01 7.52918601e-01
1.23016095e+00 -1.61975950e-01 -2.06160054e-01 9.56884325e-01
2.17504904e-01 4.42811638e-01 5.05767822e-01 -5.18408895e-01
3.42796177e-01 9.47531879e-01 -3.94431740e-01 -8.54646921e-01
-5.06776199e-03 -7.62136877e-01 -7.16990650e-01 5.78898132e-01
1.16345949e-01 1.63193613e-01 -1.30909276e+00 1.83756149e+00
3.80570173e-01 2.09120020e-01 1.80087075e-01 5.31716764e-01
1.14359879e+00 6.79426730e-01 9.38378647e-02 -2.81721264e-01
8.09920847e-01 -1.07813358e+00 -2.32401982e-01 -4.02241766e-01
5.98832846e-01 -2.86532760e-01 1.14267671e+00 2.65721232e-01
-7.38399029e-01 -5.52695394e-01 -9.31189895e-01 -8.77250824e-03
-8.78825128e-01 -4.27584529e-01 4.11849648e-01 5.17492115e-01
-1.16608071e+00 1.47372767e-01 -4.19088811e-01 -3.57272476e-01
1.16757631e+00 3.83600593e-01 -7.50015676e-01 -2.88382590e-01
-7.22140193e-01 7.00780332e-01 1.87133476e-01 -1.78803712e-01
-1.31845260e+00 -1.18286693e+00 -9.86171544e-01 1.73482835e-01
4.69606459e-01 -3.31584066e-01 6.99046493e-01 -1.41009569e+00
-7.71636724e-01 1.18086720e+00 8.31889138e-02 -3.29900861e-01
3.61409873e-01 -1.22488113e-02 -5.76956093e-01 -1.10574096e-01
-5.86095219e-03 9.41885591e-01 9.70009685e-01 -1.32274830e+00
-3.50298584e-01 -8.27136934e-01 5.74979037e-02 2.92097360e-01
-7.85063207e-01 -5.49023092e-01 -8.85015354e-02 -5.09770572e-01
1.89827845e-01 -7.04699695e-01 9.34927091e-02 6.91028059e-01
-2.29856864e-01 -1.49210945e-01 1.11493909e+00 -1.68050259e-01
1.10086823e+00 -2.48796606e+00 2.02762634e-01 2.18732879e-01
6.99370682e-01 3.18482369e-01 -5.11383235e-01 2.11647764e-01
-5.26498497e-01 -7.05376863e-02 -2.58897334e-01 -3.11553836e-01
-1.69966333e-02 4.06605691e-01 -8.21473360e-01 5.34721732e-01
1.76761657e-01 1.02060711e+00 -8.47457469e-01 -3.52127105e-01
5.29286265e-02 2.06571206e-01 -6.50316834e-01 2.64984637e-01
-5.78783043e-02 -4.69055660e-02 2.13103026e-01 1.07265377e+00
6.95867479e-01 -3.71616751e-01 -2.83871323e-01 4.98811565e-02
2.11125743e-02 -3.35714102e-01 -1.09058905e+00 1.45596373e+00
-3.77387792e-01 6.48208857e-01 -1.39558658e-01 -8.34842682e-01
1.05996346e+00 -2.03477129e-01 2.44029805e-01 -9.75240052e-01
-2.68938839e-01 1.43349275e-01 -5.66822141e-02 -4.50262547e-01
1.57691106e-01 9.83568206e-02 2.16204554e-01 4.09848481e-01
4.81322140e-01 3.75166744e-01 -6.82288874e-03 4.18294430e-01
1.25739276e+00 -2.59820193e-01 3.24385047e-01 -1.88351095e-01
2.62794718e-02 -1.42083675e-01 5.93643248e-01 1.00242388e+00
-6.49607360e-01 6.74709201e-01 4.07119542e-01 -7.14734793e-01
-5.33359408e-01 -1.68137324e+00 -3.43336761e-01 1.52096975e+00
2.98838884e-01 -1.75836712e-01 -2.40708575e-01 -1.05506170e+00
3.19646716e-01 5.73250294e-01 -1.13151383e+00 -5.00795603e-01
3.86711322e-02 -7.90471971e-01 2.51932561e-01 5.49654424e-01
2.26721227e-01 -9.82918501e-01 -3.42163950e-01 -1.63392872e-01
2.29134843e-01 -5.69176078e-01 -3.04245532e-01 6.69562817e-01
-6.02446318e-01 -1.29255056e+00 -5.73286593e-01 -8.39451194e-01
7.29629934e-01 2.63965964e-01 1.28917873e+00 -1.02991492e-01
-7.31715560e-01 7.50248432e-01 -3.33093673e-01 -4.62263644e-01
2.07322553e-01 -7.56704062e-02 7.46966805e-03 6.04018033e-01
5.74071169e-01 -5.81565738e-01 -5.70091426e-01 4.22519863e-01
-7.63948798e-01 -2.92506874e-01 5.10527313e-01 8.61727297e-01
5.74085712e-01 -4.30483311e-01 6.25019908e-01 -9.08421218e-01
-2.36734701e-03 -8.86747956e-01 -2.38998249e-01 5.87960422e-01
-8.24296594e-01 -2.56611377e-01 2.06171542e-01 -8.03157151e-01
-7.76616871e-01 -4.07631099e-02 1.38915405e-01 -6.95515990e-01
-1.43241420e-01 -6.02685148e-03 -4.19257849e-01 -3.08261752e-01
1.15528774e+00 3.41230601e-01 -3.35307866e-02 -2.77742505e-01
4.30565834e-01 7.48000562e-01 4.67172593e-01 -4.63732809e-01
9.59451020e-01 7.58511961e-01 -5.18135130e-01 -8.29013646e-01
-1.03783321e+00 -6.65314674e-01 -7.49500334e-01 -1.51598558e-01
5.23739159e-01 -1.13167214e+00 -3.88551146e-01 5.35329223e-01
-6.59928739e-01 -4.27524000e-01 -1.07333434e+00 1.40331209e-01
-4.19134408e-01 1.12963818e-01 -6.46887943e-02 -7.15450466e-01
-3.71101528e-01 -9.75381494e-01 9.96085405e-01 2.80244529e-01
-3.68679613e-01 -9.86444294e-01 4.54566032e-01 4.20313686e-01
5.37426353e-01 1.98256090e-01 1.06302190e+00 -1.03318238e+00
-6.96666420e-01 -6.98103607e-02 -3.12631041e-01 3.77120554e-01
-1.38157949e-01 -7.46631920e-02 -1.38725519e+00 -7.06158221e-01
-2.77105063e-01 -1.04524374e+00 1.31548226e+00 -4.83695418e-02
1.38552284e+00 -3.40058476e-01 -2.29431152e-01 8.66900980e-01
1.30175245e+00 2.29918081e-02 7.72621751e-01 4.67195027e-02
5.63446403e-01 4.82375592e-01 2.82784462e-01 2.87018955e-01
3.70885938e-01 3.07258338e-01 6.43417418e-01 -2.89885819e-01
-2.34106928e-01 -1.07817523e-01 3.57284695e-01 5.10984898e-01
4.37924355e-01 -1.12884700e-01 -1.07497096e+00 6.87715173e-01
-1.66875029e+00 -1.11556816e+00 3.51058453e-01 2.33174324e+00
8.75794709e-01 1.62963942e-01 -9.65607464e-02 1.85219660e-01
6.15076005e-01 3.88899654e-01 -9.95490134e-01 -4.16291654e-01
-3.91800910e-01 3.30030859e-01 2.95590814e-02 2.33881027e-01
-1.15486193e+00 4.12233800e-01 5.90725470e+00 9.95031416e-01
-1.18411863e+00 3.15719068e-01 1.06930017e+00 -5.18602908e-01
-2.95051366e-01 7.20065832e-02 -8.03218901e-01 3.77293378e-01
4.89610672e-01 -5.55315055e-02 2.55299360e-01 1.14280438e+00
-6.94805682e-01 2.37777773e-02 -1.51334476e+00 1.47370470e+00
4.85510826e-01 -1.45368636e+00 2.79389471e-01 1.58178195e-01
9.09540236e-01 2.80017674e-01 4.21094894e-01 8.71670842e-01
5.76396883e-02 -1.10020792e+00 6.78441703e-01 5.14073730e-01
9.72585201e-01 -3.54951531e-01 2.48090774e-01 4.01134938e-01
-1.00524271e+00 -6.85166717e-01 -3.03070545e-01 -5.57080284e-02
-5.01472831e-01 4.82012630e-01 -2.81080157e-01 3.59396003e-02
8.87193799e-01 9.17436123e-01 -8.66091549e-01 1.19178736e+00
1.98878914e-01 4.45272952e-01 -2.64234208e-02 3.71604115e-01
-4.95015690e-03 2.30826139e-01 5.31803966e-01 8.54976237e-01
-2.48384804e-01 -1.50234073e-01 4.73891318e-01 8.39204848e-01
-3.26204538e-01 4.71755676e-02 -6.98843122e-01 -1.29508525e-01
3.35467815e-01 1.20873344e+00 -5.29495478e-01 -3.68172079e-01
-4.86838579e-01 7.75768459e-01 7.44292021e-01 4.69334304e-01
-5.66043139e-01 -3.40215802e-01 8.00647080e-01 2.29158148e-01
2.31599510e-01 2.97479182e-01 -1.94281891e-01 -1.44148278e+00
4.32615615e-02 -9.48796391e-01 1.02139211e+00 -4.55653459e-01
-1.70057940e+00 6.51800573e-01 -2.08565667e-01 -1.18805563e+00
1.68256372e-01 -6.45137191e-01 -5.57807267e-01 5.75975835e-01
-1.54355240e+00 -1.03471506e+00 -5.45136213e-01 7.68962622e-01
6.15891457e-01 -4.71146464e-01 9.30405617e-01 3.67415518e-01
-5.91080427e-01 9.35573697e-01 1.77409843e-01 2.91438907e-01
7.74187624e-01 -1.14977622e+00 1.55033380e-01 5.00135660e-01
4.90991563e-01 4.02325124e-01 9.65603068e-02 -5.28770328e-01
-1.34863746e+00 -1.31039512e+00 6.93332553e-01 -8.78951252e-01
5.93026698e-01 -6.67476714e-01 -1.06135201e+00 6.45297527e-01
-2.31179401e-01 8.70915771e-01 1.01799858e+00 3.37393939e-01
-1.14683723e+00 -5.11716008e-01 -1.21390212e+00 2.50703663e-01
1.22367620e+00 -7.90490448e-01 -7.76350498e-01 2.47459948e-01
5.74405611e-01 -2.10321069e-01 -5.65209687e-01 5.64122915e-01
6.06920660e-01 -1.03229964e+00 9.80335832e-01 -9.56813931e-01
2.51528740e-01 1.72618069e-02 -5.42519987e-01 -1.17722392e+00
-4.61079836e-01 -2.07281724e-01 -6.53974354e-01 1.08941007e+00
4.05124903e-01 -8.07766974e-01 8.76995623e-01 4.22363281e-01
2.18370426e-02 -1.16340971e+00 -1.07037938e+00 -8.20281148e-01
3.14245075e-02 -1.74683005e-01 2.59410590e-01 1.10537398e+00
-3.99521261e-01 2.40330398e-01 -1.07911015e-02 -4.78510149e-02
8.11332822e-01 4.29437816e-01 8.04196715e-01 -1.51178837e+00
-3.31294656e-01 -1.78430766e-01 -8.73239458e-01 -5.54428756e-01
2.50858516e-01 -1.17369103e+00 -3.30711544e-01 -1.14635801e+00
7.75299788e-01 -4.91462290e-01 -6.03269815e-01 8.25050831e-01
6.04795553e-02 5.21932304e-01 1.30802110e-01 3.43542337e-01
-1.19399023e+00 7.42918491e-01 7.33211696e-01 -5.82910717e-01
-1.39965087e-01 -2.59887725e-01 -8.18300843e-01 5.90071917e-01
2.00914964e-01 -4.83340859e-01 -6.16611898e-01 -3.65328491e-01
1.73319608e-01 -4.83317286e-01 7.73042560e-01 -1.03882861e+00
4.07663226e-01 -1.43923670e-01 7.86447704e-01 -4.94915396e-01
3.81748885e-01 -7.98487663e-01 -1.56158313e-01 3.97457093e-01
-6.71964288e-01 -2.22936079e-01 4.47797850e-02 8.92250597e-01
-1.94924504e-01 2.37998754e-01 1.16032243e+00 1.07191282e-03
-1.11823368e+00 9.06112134e-01 1.75153553e-01 8.15929472e-01
1.35178995e+00 -6.54283762e-01 -8.10990453e-01 -1.15990914e-01
-1.00755715e+00 4.20248777e-01 5.27172863e-01 8.24353993e-01
7.68206060e-01 -1.23237872e+00 -4.98136938e-01 7.10496783e-01
1.01756299e+00 -1.29825681e-01 5.97455621e-01 6.72706902e-01
-1.07048191e-01 -1.25011683e-01 -2.54327655e-01 -9.46149588e-01
-1.02032912e+00 7.45306969e-01 3.95668566e-01 -1.89313143e-02
-3.82098526e-01 1.12124550e+00 6.84190512e-01 -8.50069642e-01
7.07409143e-01 2.96837121e-01 -1.04685407e-02 5.64333975e-01
8.88260067e-01 4.45661992e-01 2.77418882e-01 -7.93130040e-01
-5.64818501e-01 3.99598330e-01 -3.39117676e-01 4.23271954e-01
1.43723750e+00 -9.74476803e-03 -4.69333008e-02 7.54068792e-01
1.61306405e+00 -2.50307053e-01 -1.23422313e+00 -6.07762873e-01
-5.17518558e-02 -5.51806331e-01 -1.03870220e-02 -9.73769486e-01
-1.08015895e+00 9.52472150e-01 1.09195817e+00 -1.57076970e-01
9.81565237e-01 1.45047218e-01 3.46620470e-01 4.59099948e-01
3.75274122e-01 -1.09739769e+00 4.66497749e-01 4.05019253e-01
8.09989870e-01 -1.49708319e+00 -2.67377675e-01 -1.39990196e-01
-4.63072330e-01 8.73530865e-01 1.02295685e+00 1.99027568e-01
8.91031086e-01 2.56348014e-01 2.27745906e-01 -2.56301463e-01
-1.22392166e+00 -2.20167950e-01 4.33715731e-01 7.39395559e-01
-1.31331056e-01 -5.46124857e-03 6.42709076e-01 2.90603906e-01
2.58332372e-01 -1.75965354e-01 1.37569591e-01 9.52916920e-01
-6.05197787e-01 -5.25825024e-01 -2.55065233e-01 7.67391086e-01
1.42217919e-01 -8.61644000e-02 -5.02380610e-01 5.20991445e-01
5.01396954e-01 7.75095820e-01 5.47554016e-01 -3.42601627e-01
2.82086253e-01 2.10846424e-01 2.76296288e-01 -1.01618469e+00
-4.20455039e-01 -6.37792110e-01 -2.92263538e-01 -7.31400728e-01
4.88491580e-02 -2.64609814e-01 -7.28045523e-01 -1.11640833e-01
-2.14944288e-01 1.49704590e-02 2.11487815e-01 7.32318342e-01
6.44527733e-01 7.69001469e-02 8.85649920e-01 -4.93931085e-01
-8.68290603e-01 -4.15207148e-01 -7.16336310e-01 6.68291450e-01
5.59099734e-01 -8.06866109e-01 -8.07666838e-01 -3.16071749e-01] | [9.636489868164062, 2.8284690380096436] |
9f360d60-06da-4e38-9a5a-9c3e2f60f016 | multi-relational-classification-via-bayesian | null | null | https://dl.acm.org/doi/pdf/10.1145/3292500.3330863 | https://dl.acm.org/doi/pdf/10.1145/3292500.3330863 | Multi-Relational Classification via Bayesian Ranked Non-Linear Embeddings | The task of classifying multi-relational data spans a wide range of domains such as document classification in citation networks, classification of emails, and protein labeling in proteins interaction graphs. Current state-of-the-art classification models rely on learning per-entity latent representations by mining the whole structure of the relations' graph, however, they still face two major problems. Firstly, it is very challenging to generate expressive latent representations in sparse multi-relational settings with implicit feedback relations as there is very little information per-entity. Secondly, for entities with structured properties such as titles and abstracts (text) in documents, models have to be modified ad-hoc. In this paper, we aim to overcome these two main drawbacks by proposing a flexible nonlinear latent embedding model (BRNLE) for the classification of multi-relational data. The proposed model can be applied to entities with structured properties such as text by utilizing the numerical vector representations of those properties. To address the sparsity problem of implicit feedback relations, the model is optimized via a sparsely-regularized multi-relational pair-wise Bayesian personalized ranking loss (BPR). Experiments on four different real-world datasets show that the proposed model significantly outperforms state-of-the-art models for multi-relational classification. | ['Ahmed Rashed; Josif Grabocka; Lars Schmidt-Thieme'] | 2019-08-06 | null | null | null | the-25th-acm-sigkdd-conference-on-knowledge | ['heterogeneous-node-classification'] | ['graphs'] | [ 1.91254154e-01 3.35765004e-01 -7.50514746e-01 -3.84516835e-01
-6.12377942e-01 -1.68408424e-01 4.62807298e-01 6.21301293e-01
-1.15673564e-01 7.53431201e-01 3.10407221e-01 -1.67342484e-01
-8.42005730e-01 -9.49640393e-01 -6.33050799e-01 -5.01487374e-01
-1.87494084e-01 6.64349854e-01 1.26968414e-01 -1.53934479e-01
2.39247650e-01 5.74149370e-01 -1.39433920e+00 5.30239582e-01
1.04549849e+00 8.04032385e-01 -1.34604722e-01 1.58859000e-01
-3.78085315e-01 7.85611212e-01 -2.41360053e-01 -5.75055957e-01
-3.90312560e-02 1.33151844e-01 -9.44958210e-01 -3.46561931e-02
2.83301473e-01 1.72412276e-01 -6.53679729e-01 9.94615316e-01
2.83952236e-01 1.72523513e-01 1.11211383e+00 -1.41464019e+00
-8.71008515e-01 6.41541064e-01 -7.05350399e-01 6.34727702e-02
3.30262542e-01 -6.70825481e-01 1.77217364e+00 -1.16013324e+00
7.57673502e-01 1.50572920e+00 6.05700850e-01 4.10378128e-02
-1.48910213e+00 -5.02352476e-01 4.14634645e-01 2.17224464e-01
-1.53931248e+00 -1.20105460e-01 8.05409670e-01 -6.55770898e-01
1.09130454e+00 3.11603606e-01 1.95374116e-01 9.40706015e-01
1.48475096e-01 6.30246878e-01 7.99640059e-01 -3.15617979e-01
-1.68851346e-01 4.03146088e-01 5.49157143e-01 7.26718187e-01
3.83266091e-01 -5.56641221e-01 -7.33172953e-01 -6.81895792e-01
5.04808426e-01 3.08928609e-01 -2.52490044e-01 -7.99859285e-01
-1.07056320e+00 1.09134400e+00 4.63098556e-01 5.62893331e-01
-3.37925196e-01 -2.04067290e-01 4.09486175e-01 4.38718766e-01
8.34176838e-01 2.92134851e-01 -6.07055902e-01 3.82814974e-01
-7.87034273e-01 2.15608343e-01 9.06647205e-01 1.01094568e+00
8.14518690e-01 -4.12257046e-01 -1.44494250e-01 1.27701223e+00
6.09421313e-01 -1.73785657e-01 5.58125496e-01 -4.18933481e-01
9.94288206e-01 1.16146231e+00 -1.26058906e-01 -1.42216635e+00
-3.76094311e-01 -5.32931089e-01 -1.28813684e+00 -4.53466862e-01
1.84882991e-02 2.81922996e-01 -5.32337546e-01 1.44932866e+00
4.18658912e-01 -3.55820321e-02 2.17788652e-01 5.28282225e-01
1.12454748e+00 8.64972413e-01 1.10392533e-02 -4.63009864e-01
1.30132461e+00 -9.20592070e-01 -9.05126572e-01 1.11947127e-01
8.84697378e-01 -4.84018534e-01 6.71066523e-01 3.12522203e-02
-7.96216965e-01 -3.55301559e-01 -8.96127939e-01 -2.00944215e-01
-7.01006770e-01 3.38152289e-01 8.70789051e-01 2.17491791e-01
-6.78529024e-01 5.94381213e-01 -3.69788706e-01 -2.29729831e-01
2.85655975e-01 5.79322040e-01 -6.42974973e-01 -2.13147447e-01
-1.56871629e+00 7.24959612e-01 3.94078612e-01 3.11241187e-02
-4.71442603e-02 -7.58343518e-01 -9.16458011e-01 3.92694741e-01
6.04557037e-01 -5.45357823e-01 4.23333734e-01 -3.53159547e-01
-1.17972350e+00 7.07518697e-01 -1.76290914e-01 -1.78777277e-01
3.10135663e-01 -4.53427374e-01 -4.62063760e-01 8.36771801e-02
3.09154361e-01 -2.72912141e-02 5.62379360e-01 -1.10917568e+00
-9.40300077e-02 -4.37809825e-01 2.62371171e-02 2.85470307e-01
-8.07304204e-01 -9.67255682e-02 -5.10051787e-01 -7.66891360e-01
4.09994423e-01 -7.70587921e-01 -3.48506749e-01 -1.55031979e-01
-5.66572785e-01 -7.65424967e-01 8.37441564e-01 -3.06867272e-01
1.27464032e+00 -1.95407844e+00 4.86059308e-01 4.24720228e-01
4.09509540e-01 1.64262429e-01 -7.95305371e-02 6.59532130e-01
-3.05273086e-01 1.61791205e-01 1.06186524e-01 -1.78246051e-01
-1.72420859e-01 6.98180348e-02 -5.03450871e-01 4.40190226e-01
2.71679819e-01 8.20807695e-01 -9.50488567e-01 -7.69423187e-01
-1.09272853e-01 5.33974767e-01 -4.28060353e-01 2.67783761e-01
-6.23114556e-02 -8.33308231e-03 -7.86381006e-01 6.53307140e-01
3.92129928e-01 -9.79589581e-01 5.69142699e-01 -4.10293251e-01
8.02590474e-02 3.62297177e-01 -1.41469002e+00 1.42214084e+00
-2.84023881e-01 1.83337569e-01 -1.99081391e-01 -1.55514789e+00
9.71361518e-01 4.39954698e-01 9.23407197e-01 -9.86281484e-02
-2.33848408e-01 1.60388172e-01 -2.81621218e-01 -2.93859810e-01
5.50368786e-01 1.10082597e-01 -2.46259309e-02 4.00738627e-01
2.66067177e-01 3.07455182e-01 1.99587166e-01 6.17735863e-01
1.19548655e+00 -1.91011369e-01 4.42107439e-01 -4.47167382e-02
6.78386688e-01 -4.07281071e-01 6.42834663e-01 5.96702337e-01
5.23929834e-01 1.74215838e-01 8.08925986e-01 -3.46154988e-01
-6.46810114e-01 -8.25542450e-01 -2.83466280e-01 1.16302347e+00
2.53939070e-05 -8.03667784e-01 1.15531296e-01 -7.82847881e-01
3.29636216e-01 1.52985573e-01 -5.68929315e-01 -1.65383160e-01
-2.88256407e-01 -1.16266930e+00 1.52811706e-01 2.88731158e-01
-2.12625302e-02 -7.85339892e-01 1.73088506e-01 3.60515833e-01
-4.35403660e-02 -1.24449730e+00 -9.76275653e-02 4.37394321e-01
-9.54772115e-01 -9.27994728e-01 -7.50087261e-01 -6.48689747e-01
7.75349021e-01 2.47619092e-01 9.95757222e-01 -4.60997298e-02
-2.70129979e-01 2.73120642e-01 -4.45585787e-01 2.31841415e-01
-3.91291045e-02 4.69110012e-01 2.41967469e-01 3.57237369e-01
2.87329078e-01 -5.19190609e-01 -2.26832807e-01 3.86711836e-01
-8.78655255e-01 -4.16950323e-02 7.78781891e-01 1.21396327e+00
7.77606845e-01 2.04298347e-01 8.59885395e-01 -1.67992938e+00
8.08405459e-01 -8.69076848e-01 -3.39144915e-01 6.84113383e-01
-1.02728808e+00 2.73751676e-01 5.70850432e-01 -6.39836431e-01
-8.48322272e-01 -7.80312344e-02 2.82361865e-01 -4.12018478e-01
2.68082500e-01 9.89522219e-01 -2.02165723e-01 3.70626301e-02
3.93702447e-01 -2.34277137e-02 -4.13547665e-01 -5.25777698e-01
5.65664828e-01 8.30076337e-01 -7.86004812e-02 -6.19075298e-01
9.12415981e-01 1.36449635e-01 4.39076185e-01 -9.17414963e-01
-1.16789389e+00 -9.30616975e-01 -7.80769348e-01 2.09654480e-01
5.31426787e-01 -1.07720900e+00 -6.79286480e-01 5.99371269e-02
-1.08263588e+00 3.85834247e-01 -3.60958651e-02 5.96905470e-01
-4.19074446e-01 5.57070553e-01 -8.43351185e-01 -7.80895531e-01
-3.15954804e-01 -9.16058183e-01 8.91239524e-01 -1.09122790e-01
-2.60542870e-01 -9.83237684e-01 2.34056890e-01 5.32032728e-01
1.53597414e-01 -4.48260829e-02 1.41493177e+00 -8.70344996e-01
-7.34331727e-01 -3.18418294e-01 -4.23157394e-01 7.25935102e-02
2.92417288e-01 -2.47107461e-01 -7.21556187e-01 -1.76589981e-01
-4.88565952e-01 -7.34700978e-01 1.09617341e+00 1.57947198e-01
1.25757670e+00 -5.29348731e-01 -7.49893546e-01 4.94239181e-01
1.20269489e+00 -3.18865031e-01 2.94789612e-01 7.04012811e-02
1.04949093e+00 8.94233942e-01 8.55882287e-01 4.70217168e-01
5.18206477e-01 8.42582941e-01 1.41189694e-01 3.38615216e-02
5.31537294e-01 -3.25300992e-01 1.22048192e-01 1.32655501e+00
-1.18331999e-01 -2.79696107e-01 -6.86672330e-01 4.45951313e-01
-2.30282593e+00 -8.79290581e-01 -1.89918756e-01 1.93766117e+00
1.12478030e+00 1.11529335e-01 -2.46396467e-01 1.95056170e-01
7.09942698e-01 3.58485043e-01 -5.30268371e-01 -1.61100179e-01
-1.48346767e-01 -8.64367783e-02 3.49234521e-01 3.01056057e-01
-1.26529646e+00 9.52495873e-01 5.34552336e+00 9.27164495e-01
-6.60484195e-01 2.52000708e-02 5.54186642e-01 2.03851759e-01
-4.37982440e-01 -5.62140532e-03 -1.28365290e+00 1.20693788e-01
9.42801178e-01 -4.56039235e-02 6.48554564e-02 8.35792840e-01
-2.31714547e-01 2.78502643e-01 -1.15953648e+00 1.07350814e+00
1.38108339e-02 -1.35748231e+00 3.15227300e-01 1.93900838e-01
8.41929853e-01 -1.85370728e-01 9.95552242e-02 4.17064250e-01
2.75515765e-01 -1.03176928e+00 -7.47449845e-02 6.06407285e-01
9.20415163e-01 -5.64721704e-01 6.23314321e-01 3.21742892e-01
-1.34167933e+00 -9.40280966e-04 -6.35220587e-01 3.29763830e-01
-8.78806785e-02 9.73107755e-01 -6.28758967e-01 9.28607225e-01
5.88731766e-01 1.15828645e+00 -3.58569801e-01 6.92154288e-01
4.69805188e-02 2.65297621e-01 -1.73472047e-01 -7.75188953e-02
3.08109391e-02 -2.97578782e-01 2.53493160e-01 1.14471245e+00
1.26338392e-01 1.78954244e-01 1.12087347e-01 6.55328989e-01
-5.59170306e-01 6.27960324e-01 -8.89032125e-01 -4.36450064e-01
3.29446852e-01 1.39765453e+00 -5.82925081e-01 -2.51834810e-01
-6.39694154e-01 6.28876388e-01 9.29618776e-01 4.23395574e-01
-3.73809546e-01 -3.54841888e-01 3.50756109e-01 9.01687220e-02
1.87129736e-01 -1.05185792e-01 -7.52210617e-02 -1.43891561e+00
-1.76228005e-02 -5.71480393e-01 6.50401711e-01 -3.88625085e-01
-1.80888939e+00 5.12058258e-01 1.58016562e-01 -1.23313630e+00
-1.64810121e-01 -6.19048834e-01 1.80199482e-02 7.52456129e-01
-1.55871224e+00 -1.25816178e+00 3.84369679e-02 5.65956056e-01
1.79393291e-01 -4.67967808e-01 1.01890898e+00 4.95398700e-01
-4.41946208e-01 6.85589731e-01 4.05256897e-01 1.52221039e-01
6.84508502e-01 -1.21101141e+00 -1.01416163e-01 1.26386419e-01
3.69955301e-01 1.05556870e+00 3.40700507e-01 -7.06591845e-01
-1.38821030e+00 -1.17932236e+00 1.15674174e+00 -2.18586564e-01
9.15641606e-01 -3.92900348e-01 -1.26659417e+00 6.92909718e-01
-2.81474054e-01 3.95774275e-01 1.16488671e+00 8.25186014e-01
-5.55965304e-01 -1.26893729e-01 -8.14173281e-01 4.26343977e-01
9.60730076e-01 -7.57504582e-01 -5.77832103e-01 7.49867260e-01
7.03243673e-01 9.18584876e-03 -1.36394501e+00 6.96359992e-01
5.40758848e-01 -3.12044382e-01 1.32097113e+00 -9.69060063e-01
4.95345742e-01 -3.98946404e-02 -1.91040665e-01 -1.23364925e+00
-4.78849769e-01 -3.58887464e-01 -7.01777101e-01 1.35623002e+00
4.51134950e-01 -6.89463675e-01 7.75456190e-01 4.52788651e-01
3.00824940e-01 -1.28090942e+00 -9.14838374e-01 -4.93120790e-01
-1.37548327e-01 7.41565824e-02 2.15846598e-01 1.36428332e+00
1.80063307e-01 9.53301609e-01 -5.49246788e-01 2.32931584e-01
6.67752862e-01 4.59555864e-01 6.46511078e-01 -1.77429104e+00
-3.19426090e-01 -1.12580478e-01 -6.23955369e-01 -1.12413526e+00
5.43817997e-01 -1.13539529e+00 -6.06608868e-01 -1.56082451e+00
4.78293151e-01 -9.07150924e-01 -6.28653646e-01 3.82646084e-01
-2.48592287e-01 -2.05347523e-01 -2.27922261e-01 4.63684022e-01
-8.40579152e-01 9.50795054e-01 9.47973967e-01 -4.05120492e-01
-1.13099411e-01 1.85341194e-01 -6.56260133e-01 5.35936892e-01
3.85862708e-01 -7.13894427e-01 -5.47353625e-01 1.64196998e-01
6.97243869e-01 4.04401660e-01 1.00033162e-02 -3.86268556e-01
3.38493407e-01 -1.68726578e-01 2.90733814e-01 -9.42684770e-01
6.30472958e-01 -9.26083028e-01 3.72440144e-02 1.04105309e-01
-7.34905064e-01 -1.26785487e-01 -3.39438558e-01 1.15223539e+00
-4.34346884e-01 -3.21951151e-01 4.75027621e-01 1.37601830e-02
-1.38240710e-01 5.56344926e-01 -1.36221558e-01 1.79020524e-01
8.49104643e-01 2.21998602e-01 -3.44360113e-01 -5.81097789e-02
-7.28662193e-01 3.46662223e-01 -4.32148352e-02 5.29682577e-01
8.70662034e-01 -1.50512612e+00 -5.86558104e-01 8.99751931e-02
5.13523698e-01 5.00618070e-02 1.18382007e-01 6.71409249e-01
5.13449982e-02 7.15544879e-01 1.32326514e-01 -3.98144543e-01
-1.46590734e+00 5.73601365e-01 -2.27595270e-01 -9.58243787e-01
-5.51562190e-01 7.49827504e-01 3.33090186e-01 -6.12316787e-01
2.23647684e-01 -1.32510945e-01 -7.32625961e-01 4.96852577e-01
1.30828887e-01 2.01378852e-01 8.85474980e-02 -7.95333922e-01
-2.49355853e-01 5.57469130e-01 -6.20271266e-01 3.05187285e-01
1.68053746e+00 1.24299049e-01 -4.43476051e-01 8.26284170e-01
1.39663863e+00 -7.86155984e-02 -5.43310046e-01 -8.06192875e-01
3.17734689e-01 -4.72426057e-01 -1.60665095e-01 -1.82201445e-01
-8.52952123e-01 7.71620154e-01 1.23118952e-01 2.27058649e-01
4.81740564e-01 2.03840971e-01 3.87908578e-01 8.83039176e-01
3.43301505e-01 -1.03747356e+00 2.78184474e-01 5.21655321e-01
8.26478243e-01 -1.16800201e+00 5.03295183e-01 -7.70996213e-01
-5.83808541e-01 1.02226186e+00 5.27182817e-01 9.08494741e-02
1.15834069e+00 -1.72417954e-01 -4.27363306e-01 -5.00638604e-01
-1.04193401e+00 6.09450638e-02 8.08943391e-01 1.08910829e-01
8.34559560e-01 7.22683873e-03 -4.28043425e-01 5.28753102e-01
2.29896963e-01 -2.38046139e-01 1.84866846e-01 7.67119408e-01
-1.74097344e-01 -1.32265425e+00 -7.47473687e-02 1.12453699e+00
-6.83701038e-01 -1.44531667e-01 -2.48739079e-01 3.24093670e-01
-2.40627214e-01 6.90071166e-01 -4.10019219e-01 -2.02137411e-01
6.85096979e-02 1.27878010e-01 9.11006555e-02 -1.02241945e+00
-3.18530440e-01 -9.80086401e-02 1.00144707e-01 -3.59535158e-01
-6.52252257e-01 -6.14931881e-01 -9.61501837e-01 3.33467275e-02
-8.24922621e-01 4.32177603e-01 3.46028447e-01 8.75474095e-01
4.09312904e-01 6.40586019e-01 5.60684919e-01 -4.79289860e-01
-7.08609104e-01 -1.06909490e+00 -8.63190055e-01 5.32012165e-01
1.47838444e-01 -1.18461990e+00 -3.08685035e-01 -2.15011090e-01] | [8.651556015014648, 7.782830715179443] |
4f59c38c-c857-45f3-86d1-2d2f18ebe085 | computable-stability-for-persistence-rank | 2307.02904 | null | https://arxiv.org/abs/2307.02904v1 | https://arxiv.org/pdf/2307.02904v1.pdf | Computable Stability for Persistence Rank Function Machine Learning | Persistent homology barcodes and diagrams are a cornerstone of topological data analysis. Widely used in many real data settings, they relate variation in topological information (as measured by cellular homology) with variation in data, however, they are challenging to use in statistical settings due to their complex geometric structure. In this paper, we revisit the persistent homology rank function -- an invariant measure of ``shape" that was introduced before barcodes and persistence diagrams and captures the same information in a form that is more amenable to data and computation. In particular, since they are functions, techniques from functional data analysis -- a domain of statistics adapted for functions -- apply directly to persistent homology when represented by rank functions. Rank functions, however, have been less popular than barcodes because they face the challenge that stability -- a property that is crucial to validate their use in data analysis -- is difficult to guarantee, mainly due to metric concerns on rank function space. However, rank functions extend more naturally to the increasingly popular and important case of multiparameter persistent homology. In this paper, we study the performance of rank functions in functional inferential statistics and machine learning on both simulated and real data, and in both single and multiparameter persistent homology. We find that the use of persistent homology captured by rank functions offers a clear improvement over existing approaches. We then provide theoretical justification for our numerical experiments and applications to data by deriving several stability results for single- and multiparameter persistence rank functions under various metrics with the underlying aim of computational feasibility and interpretability. | ['Gregory Henselman-Petrusek', 'Anthea Monod', 'Pierre Faugère', 'Inés García-Redondo', 'Qiquan Wang'] | 2023-07-06 | null | null | null | null | ['topological-data-analysis'] | ['graphs'] | [ 2.53187772e-02 -1.83630437e-01 -1.98949486e-01 6.83856979e-02
-5.17566442e-01 -8.03273797e-01 5.52173734e-01 5.48630416e-01
-2.61317223e-01 1.00133789e+00 1.05114445e-01 -2.84297734e-01
-7.41055250e-01 -8.70882273e-01 -7.97199130e-01 -1.00796568e+00
-8.09261322e-01 2.77365476e-01 4.69286501e-01 -4.92368370e-01
3.91198337e-01 5.72161078e-01 -1.58952081e+00 -3.68548542e-01
6.95500851e-01 8.27293754e-01 -2.45639756e-01 7.80460417e-01
2.41600841e-01 1.88184455e-01 -1.66937143e-01 -1.94449723e-01
2.72766888e-01 -4.58590269e-01 -9.21106815e-01 -4.45537060e-01
4.37728971e-01 3.26750159e-01 -2.62716413e-01 9.43711996e-01
4.31870967e-01 -3.63654667e-03 9.13887739e-01 -1.28606880e+00
-3.62629414e-01 7.60416910e-02 -2.14211434e-01 3.14079046e-01
4.77402031e-01 -1.23553284e-01 1.30607235e+00 -7.34808981e-01
7.71672606e-01 1.32098329e+00 9.41041589e-01 8.33205059e-02
-1.78878021e+00 -5.12144379e-02 -4.68137413e-01 4.73957248e-02
-1.37599158e+00 -1.88237932e-02 5.98676562e-01 -1.06513214e+00
1.79924086e-01 6.72164917e-01 7.89135695e-01 7.61351109e-01
5.26462913e-01 3.08993667e-01 1.41699851e+00 -3.07841212e-01
1.22333579e-01 -2.59137034e-01 5.01236916e-01 9.95049715e-01
6.01575792e-01 1.83274727e-02 -2.97957957e-01 -3.02321255e-01
1.15166259e+00 -2.60232538e-01 -5.26188254e-01 -8.44068170e-01
-1.29274464e+00 1.04865575e+00 3.07125956e-01 5.03769398e-01
7.94027895e-02 2.14691117e-01 4.34378445e-01 6.32379830e-01
3.44744384e-01 5.42436540e-01 -2.23142222e-01 -2.52124637e-01
-4.47169602e-01 4.71215814e-01 8.83270860e-01 5.09803712e-01
8.62561822e-01 -4.61175621e-01 9.32973400e-02 6.67518556e-01
9.46107656e-02 4.41453874e-01 2.44276732e-01 -7.29542732e-01
3.80381532e-02 6.56266034e-01 -1.39502436e-01 -1.55415297e+00
-6.47138953e-01 -3.63784373e-01 -1.08213890e+00 1.65548041e-01
9.89437997e-01 6.56966209e-01 -5.08130193e-01 2.02512383e+00
2.73777872e-01 5.87378629e-03 -1.64090440e-01 7.40504265e-01
2.49852389e-01 2.84969240e-01 -2.16117129e-01 -4.11662549e-01
1.23555195e+00 -1.48936421e-01 -3.52225095e-01 6.38896406e-01
7.44490206e-01 -5.01622856e-01 1.33859468e+00 2.46768832e-01
-9.52369034e-01 -1.78836752e-02 -1.04159701e+00 -1.00214966e-01
-6.52318597e-01 -4.10124600e-01 4.66668785e-01 4.60619539e-01
-1.29415810e+00 8.06762397e-01 -6.60694003e-01 -7.77295828e-01
1.23034045e-01 3.20262015e-01 -6.16056383e-01 3.63657326e-01
-1.18653572e+00 9.50104237e-01 -1.08355954e-02 -2.02302665e-01
-2.98130482e-01 -8.51885974e-01 -7.30054617e-01 -5.02806790e-02
6.98908046e-02 -6.01663888e-01 7.64685035e-01 -2.49476492e-01
-1.13565838e+00 8.78025413e-01 1.57686263e-01 -2.63024896e-01
7.09769547e-01 2.05233991e-01 5.07494435e-02 7.47951418e-02
1.23686336e-01 6.37227744e-02 3.26253623e-01 -9.47779000e-01
-1.10837340e-01 -3.80833656e-01 2.62427360e-01 -2.80636817e-01
-8.68713483e-02 -4.47933018e-01 -3.90375815e-02 -5.90822816e-01
3.79501343e-01 -8.79200339e-01 -5.08218408e-02 2.95768440e-01
-2.95488000e-01 -1.66735709e-01 7.18699992e-01 -3.36133868e-01
1.34568644e+00 -1.99790728e+00 6.65850043e-01 5.69114566e-01
4.64064777e-01 -2.29141675e-02 1.25382796e-01 4.62663531e-01
-1.76697001e-01 5.05505145e-01 -5.22419274e-01 3.14054102e-01
6.65554926e-02 1.35748848e-01 -1.38406321e-01 1.05242133e+00
1.94261640e-01 8.24052334e-01 -9.39226866e-01 -6.97696924e-01
1.61754027e-01 4.18827623e-01 -6.09022021e-01 -2.34146610e-01
-4.51211594e-02 4.88655388e-01 -3.40453476e-01 5.00312209e-01
4.08577651e-01 -2.96229333e-01 1.39211770e-03 -1.53521851e-01
-2.52026528e-01 -9.29176584e-02 -1.05076575e+00 1.29828572e+00
-3.37382764e-01 8.48882973e-01 7.17877671e-02 -1.31796288e+00
8.21857691e-01 -1.93582829e-02 6.34366810e-01 -5.17688215e-01
2.57991161e-02 3.46251160e-01 -1.87276844e-02 -4.17027950e-01
1.84170529e-01 -3.29886168e-01 -4.12691921e-01 5.46272062e-02
-2.32201129e-01 6.98653236e-03 3.88468295e-01 2.44677514e-02
1.36506522e+00 -3.12282503e-01 4.61120129e-01 -1.03095317e+00
7.17315793e-01 -1.38208166e-01 4.82464045e-01 4.54615027e-01
-1.24871731e-01 5.14784276e-01 1.20618141e+00 -3.05637747e-01
-1.32822287e+00 -1.37162125e+00 -7.70738661e-01 6.93703115e-01
3.85191709e-01 -3.67013186e-01 -3.84293914e-01 -2.50861913e-01
3.21750134e-01 -2.01902732e-01 -9.32558298e-01 -1.34801427e-02
-6.23403013e-01 -7.91033149e-01 7.19899058e-01 2.25468099e-01
1.02855586e-01 -4.76128340e-01 -4.85389411e-01 2.17885431e-03
7.67232552e-02 -6.00018561e-01 -2.59677082e-01 -9.96141974e-03
-9.07244742e-01 -1.62791955e+00 -7.46549785e-01 -6.26605749e-01
5.21328449e-01 -1.43328020e-02 8.87713194e-01 2.95292884e-01
-6.13735259e-01 5.34833550e-01 -1.95457295e-01 2.77086526e-01
-4.10976827e-01 8.55736434e-02 3.03174555e-01 -4.99421582e-02
-3.79123628e-01 -8.53816152e-01 -5.93406558e-01 7.10549057e-01
-1.17870474e+00 -2.43669942e-01 4.96670634e-01 1.02741134e+00
5.11472225e-01 -1.19218230e-01 4.39811230e-01 -7.26576984e-01
4.76073682e-01 -4.93791312e-01 -7.49366879e-01 1.62759453e-01
-6.41416013e-01 4.32799220e-01 5.63114583e-01 -2.28453398e-01
-7.96964467e-02 -3.46355110e-01 1.53058708e-01 -6.79176375e-02
4.45861220e-01 7.67942011e-01 -9.96289104e-02 -4.42682445e-01
7.26428747e-01 1.27318327e-03 4.08022732e-01 -4.65584040e-01
2.30141535e-01 2.33286068e-01 5.52792609e-01 -8.99951398e-01
8.06273103e-01 5.30861437e-01 8.75240266e-01 -1.16069043e+00
-3.56167167e-01 -5.58854878e-01 -7.25286543e-01 -6.15237169e-02
7.15335846e-01 -3.37136626e-01 -1.14436662e+00 3.23011667e-01
-9.08459127e-01 -2.35840335e-01 -1.82044938e-01 2.41822481e-01
-1.02698386e+00 4.84926999e-01 -5.28417706e-01 -7.81515777e-01
2.93053746e-01 -1.09338391e+00 9.65142787e-01 -2.11443812e-01
-1.21379144e-01 -1.58233702e+00 5.45855403e-01 -2.64335334e-01
5.54342151e-01 9.99695480e-01 1.42102540e+00 -3.12122405e-01
-4.41931814e-01 -1.47530571e-01 -1.85215622e-01 2.31465921e-02
-1.03357181e-01 1.09341078e-01 -4.52376008e-01 -3.38224620e-01
-2.71722376e-01 1.60235912e-02 8.21522892e-01 2.76060373e-01
9.92578030e-01 -4.12359536e-01 -2.98261225e-01 4.84133154e-01
1.65365887e+00 -3.60685885e-01 5.99598408e-01 1.99312001e-01
4.81810540e-01 8.43351722e-01 3.76892537e-01 2.07764819e-01
1.53901115e-01 9.82293367e-01 3.19476753e-01 3.44803967e-02
1.09337643e-01 -2.09122464e-01 2.82760561e-01 1.02200472e+00
-2.06513450e-01 1.92497984e-01 -9.18438077e-01 4.08756971e-01
-2.05024934e+00 -1.04369307e+00 -5.32974422e-01 2.70680737e+00
9.19960558e-01 -4.23871912e-02 6.71896100e-01 4.26632732e-01
8.88164639e-01 -1.81520209e-01 -1.99304223e-01 -1.27878010e-01
-4.67458546e-01 2.17015922e-01 5.57603598e-01 7.34191895e-01
-9.01582003e-01 4.58769888e-01 6.57855511e+00 6.64923608e-01
-1.21056664e+00 -5.07904813e-02 3.06880057e-01 3.29641700e-01
-2.05750778e-01 1.95908159e-01 -3.70128751e-01 5.54085910e-01
7.83505559e-01 -3.21510524e-01 1.60929367e-01 6.36888027e-01
1.45973474e-01 -2.21585006e-01 -1.29009831e+00 9.47136462e-01
-2.83483386e-01 -1.23558867e+00 -1.64660692e-01 5.35844743e-01
3.74413013e-01 -3.63867372e-01 2.02944696e-01 -1.16241343e-01
-7.83367008e-02 -1.20971799e+00 5.47344804e-01 6.83332920e-01
9.91105676e-01 -4.15457129e-01 6.29307866e-01 -6.86393678e-02
-1.30759168e+00 7.04219863e-02 -3.23505163e-01 -4.94411476e-02
-3.95919047e-02 8.33907187e-01 -4.59576368e-01 5.09309411e-01
2.93727905e-01 5.77668905e-01 -6.78944468e-01 1.40479624e+00
2.91644186e-01 3.26360941e-01 -4.34138834e-01 -7.73994923e-02
-1.38634413e-01 -2.46001884e-01 6.76633239e-01 1.22769654e+00
3.83540660e-01 -1.06544383e-01 1.17919117e-01 7.21767187e-01
3.20353031e-01 3.48137200e-01 -9.18376148e-01 3.08270156e-02
3.22932333e-01 1.08885825e+00 -8.22593033e-01 1.68031797e-01
-1.58240512e-01 3.06115180e-01 2.48732105e-01 7.36809373e-02
-6.18094802e-01 -5.05522609e-01 9.47469711e-01 7.69348979e-01
-5.55164963e-02 -6.28514886e-01 -2.56135017e-01 -1.13416243e+00
1.98171921e-02 -4.49225694e-01 4.11103427e-01 -1.37762964e-01
-1.46623623e+00 1.24810494e-01 3.28226328e-01 -1.12976205e+00
-2.07347572e-01 -1.05040514e+00 -3.64261210e-01 5.41529179e-01
-1.08942139e+00 -9.24832702e-01 -1.50835007e-01 4.36834544e-01
-1.05570287e-01 3.78043711e-01 7.50810027e-01 2.29342222e-01
-3.89438212e-01 5.13024271e-01 3.91874760e-01 6.53439714e-03
3.76493424e-01 -1.53812778e+00 3.49459574e-02 4.67215627e-01
-2.08968028e-01 7.56723106e-01 9.70041096e-01 -4.90027755e-01
-1.71415603e+00 -6.82257414e-01 5.34120381e-01 -7.48421609e-01
1.05162287e+00 -5.73274136e-01 -9.73508716e-01 3.02133888e-01
-6.62416816e-01 2.15680987e-01 5.04488111e-01 1.60866052e-01
-3.82630855e-01 7.06466986e-03 -1.14410305e+00 5.55382133e-01
1.22126210e+00 -7.07567155e-01 -2.79116392e-01 1.26256824e-01
5.20985723e-01 1.62206646e-02 -1.42319441e+00 6.40683651e-01
8.24264348e-01 -1.03406632e+00 1.03204131e+00 -4.18299913e-01
2.29855850e-02 -5.30054748e-01 -4.18427885e-01 -1.02894294e+00
-2.81298339e-01 -8.14520240e-01 2.38827597e-02 8.81277859e-01
2.46138260e-01 -8.91154170e-01 4.06573355e-01 1.46513745e-01
8.61377567e-02 -9.81367767e-01 -1.32449281e+00 -1.17352450e+00
5.40101230e-01 -1.22729942e-01 1.41409114e-01 1.00617170e+00
3.68088514e-01 3.35389256e-01 -7.54647255e-02 -2.04205513e-01
6.30012989e-01 -1.82124093e-01 7.86416054e-01 -1.84510052e+00
-2.19547972e-01 -8.25138152e-01 -1.33118415e+00 -7.01153219e-01
1.26110256e-01 -1.10099137e+00 5.43652363e-02 -1.01100039e+00
1.98981792e-01 -7.11063623e-01 -6.64669350e-02 1.18961677e-01
1.04255565e-01 3.39754105e-01 -4.48685735e-02 3.67193103e-01
-3.75617981e-01 3.95898730e-01 1.15753257e+00 2.40348130e-01
-1.72141522e-01 -1.75211906e-01 -3.94183457e-01 4.84669685e-01
5.61332047e-01 -2.00462304e-02 -2.04681203e-01 3.00094306e-01
5.74785411e-01 1.08757362e-01 8.27906787e-01 -1.07720292e+00
-3.01600266e-02 -6.17800057e-02 -3.47403884e-02 -1.91224158e-01
2.32444331e-01 -5.46267390e-01 2.59377837e-01 5.70883393e-01
-2.26160139e-01 4.37970728e-01 -9.01637003e-02 7.37087786e-01
2.41985898e-02 5.73096499e-02 9.84196842e-01 2.80899495e-01
-2.61392206e-01 2.63395756e-01 -1.23428784e-01 3.02100450e-01
9.04238880e-01 -2.52285600e-01 -7.06051826e-01 -4.25459653e-01
-7.30038285e-01 -2.04946584e-04 1.02655888e+00 7.74055794e-02
4.17006761e-01 -1.46653140e+00 -5.48304498e-01 1.18164852e-01
1.75942048e-01 -2.82828152e-01 -1.39862373e-01 1.54032815e+00
-7.88852453e-01 6.03523493e-01 -1.89447016e-01 -9.52726305e-01
-1.11171997e+00 5.13565361e-01 1.48565605e-01 -2.32359678e-01
-5.47851264e-01 2.06615835e-01 3.93914312e-01 -1.56784758e-01
5.44897132e-02 -4.46510315e-01 1.99099869e-01 1.52564004e-01
1.04609057e-01 6.37610257e-01 -7.06886351e-02 -8.35113645e-01
-3.78910005e-01 1.02325344e+00 4.53413606e-01 -2.15486303e-01
1.15863371e+00 -1.43472478e-01 -4.94303614e-01 8.90342951e-01
1.54156268e+00 -7.76601164e-03 -8.71197522e-01 4.53938656e-02
3.98566216e-01 -4.22932237e-01 -5.04182458e-01 -2.85917938e-01
-4.41165209e-01 8.53934407e-01 5.28890252e-01 8.47417951e-01
7.15482175e-01 3.41782093e-01 3.57391417e-01 1.53307766e-01
4.55654562e-01 -7.06080735e-01 1.95691928e-01 4.95508820e-01
9.78783548e-01 -7.58837104e-01 -1.97188765e-01 -6.27745032e-01
2.32775837e-01 1.02077544e+00 1.99394837e-01 -1.57814950e-01
8.24028075e-01 -2.69136187e-02 -3.70470315e-01 -2.32918099e-01
-5.24724782e-01 -2.94348985e-01 3.46733302e-01 4.44224864e-01
5.60384035e-01 -1.39522916e-02 -8.20820928e-01 2.04466507e-01
-4.45161372e-01 -3.82873386e-01 7.24988878e-01 8.26488912e-01
-6.50024533e-01 -1.11372161e+00 -3.77096713e-01 4.85993117e-01
-3.76160830e-01 2.00845361e-01 -4.37078059e-01 1.14956391e+00
-1.10297948e-01 4.91939873e-01 6.60279393e-02 -4.42350388e-01
1.49530023e-01 -2.38646660e-03 5.85191250e-01 -3.76484394e-01
-1.70558263e-02 -2.78893888e-01 -1.40826568e-01 -5.64515114e-01
-4.08297896e-01 -7.10330725e-01 -1.04429913e+00 -7.11016417e-01
-3.87002289e-01 4.31353718e-01 7.46320426e-01 7.71889150e-01
1.99374273e-01 8.98430273e-02 6.18881464e-01 -6.11769557e-01
-4.92205828e-01 -6.52655840e-01 -7.90342808e-01 4.86634374e-01
5.96576452e-01 -1.15539908e+00 -6.56681597e-01 -2.58853585e-01] | [7.481867790222168, 4.214142322540283] |
65870482-89fa-4e6a-9384-0bed89535895 | native-language-identification-using-phonetic | null | null | https://aclanthology.org/W17-5046 | https://aclanthology.org/W17-5046.pdf | Native Language Identification using Phonetic Algorithms | In this paper, we discuss the results of the IUCL system in the NLI Shared Task 2017. For our system, we explore a variety of phonetic algorithms to generate features for Native Language Identification. These features are contrasted with one of the most successful type of features in NLI, character n-grams. We find that although phonetic features do not perform as well as character n-grams alone, they do increase overall F1 score when used together with character n-grams. | ['ra', 'S K{\\"u}bler', 'Charese Smiley'] | 2017-09-01 | null | null | null | ws-2017-9 | ['native-language-identification'] | ['natural-language-processing'] | [ 5.43591864e-02 -4.84130085e-01 -5.41745186e-01 -4.49831307e-01
-1.02190936e+00 -1.00096452e+00 9.52526510e-01 2.68712342e-02
-7.90096700e-01 6.37570083e-01 6.82943821e-01 -4.40804452e-01
-9.97095183e-02 -1.62165046e-01 -1.69228286e-01 -4.23428297e-01
3.11097682e-01 6.38368130e-01 -1.90639347e-01 -1.03578471e-01
3.85993242e-01 5.21336138e-01 -1.73609364e+00 1.92344993e-01
9.80542660e-01 7.23164618e-01 -1.02883980e-01 6.32581532e-01
-3.09812069e-01 2.90739119e-01 -6.90934777e-01 -4.01054174e-01
-4.05886918e-02 -3.59915257e-01 -9.02088225e-01 -6.35478556e-01
6.34896755e-01 -4.81536649e-02 -1.91711187e-01 1.09413266e+00
6.41562104e-01 2.49464497e-01 1.10682046e+00 -8.86313677e-01
-6.40495420e-01 1.05868208e+00 1.75444111e-01 4.23206180e-01
9.04638469e-01 2.43330717e-01 1.31918776e+00 -1.16704798e+00
6.65317953e-01 1.54447091e+00 7.74858057e-01 3.67168307e-01
-1.15455961e+00 -7.89108276e-01 -3.18814963e-01 1.15036927e-01
-1.37719822e+00 -7.42902637e-01 6.20052516e-01 -5.07833660e-01
1.54210675e+00 5.07900953e-01 3.70184213e-01 1.33477998e+00
7.11136162e-02 9.17328656e-01 1.33022547e+00 -7.07323194e-01
-3.18876177e-01 3.25561225e-01 5.29377460e-01 2.44665325e-01
1.13490671e-01 3.08425158e-01 -6.03061020e-01 -3.65716636e-01
2.72162199e-01 -5.07434666e-01 1.42866537e-01 4.28560525e-01
-1.44551742e+00 1.08615124e+00 -1.08538449e-01 8.67935181e-01
9.33668390e-02 7.87460580e-02 4.56714094e-01 2.15993956e-01
4.33064997e-01 9.03025866e-01 -4.18427020e-01 -5.98213255e-01
-8.47133219e-01 3.70898396e-01 7.04826891e-01 6.47734404e-01
7.50452161e-01 1.59639210e-01 -5.34806907e-01 1.36101449e+00
6.99718371e-02 5.24777055e-01 8.25197697e-01 -9.17781234e-01
1.17450930e-01 1.35364801e-01 -9.35531259e-02 -7.41300702e-01
-3.57272834e-01 -1.92142382e-01 -2.57474601e-01 -1.33560285e-01
7.16475189e-01 -1.90240089e-02 -9.48946595e-01 1.73042810e+00
-3.16822827e-01 -3.41769367e-01 -6.34678975e-02 2.87617385e-01
8.58916461e-01 6.09302878e-01 2.49635309e-01 1.86952518e-03
1.16280937e+00 -9.63376105e-01 -7.41584718e-01 -2.78777063e-01
8.24515700e-01 -1.19025254e+00 1.03969431e+00 3.58421981e-01
-8.60952258e-01 -8.13133657e-01 -5.70252657e-01 -8.33081976e-02
-8.32777679e-01 3.38069767e-01 8.80442858e-01 9.64865208e-01
-1.19100165e+00 6.69235826e-01 -1.22565120e-01 -5.24461031e-01
-1.07198015e-01 3.26654941e-01 -5.41869819e-01 2.29492038e-01
-1.14412081e+00 1.07079828e+00 4.56266701e-01 -5.64000607e-01
-4.07299966e-01 -6.15079343e-01 -7.14604676e-01 -1.11048996e-01
-4.77897599e-02 4.29102592e-03 1.23457038e+00 -7.49617219e-01
-1.47850847e+00 1.09687161e+00 -4.48368043e-01 -2.57151067e-01
7.35470727e-02 -1.73784256e-01 -6.21019185e-01 -1.40069410e-01
2.43413925e-01 9.07606602e-01 3.81954044e-01 -9.35513854e-01
-7.81963110e-01 -2.09267717e-02 -3.63836020e-01 1.44215614e-01
-4.32524830e-01 9.41780865e-01 -3.21984552e-02 -6.41668141e-01
-2.45602354e-01 -9.88834381e-01 1.71078425e-02 -7.31939673e-01
-4.31944489e-01 -8.19651365e-01 3.70359451e-01 -8.77602577e-01
1.04457974e+00 -2.20102692e+00 -7.47051165e-02 2.03526050e-01
-2.67177135e-01 3.70448649e-01 -1.46657348e-01 2.97361493e-01
-6.86590746e-02 7.72719860e-01 1.39555275e-01 -2.37312943e-01
2.85444379e-01 6.02868982e-02 -1.85165584e-01 3.96239869e-02
-7.31800776e-03 9.90911365e-01 -8.05991650e-01 -2.84853280e-01
2.85127014e-01 4.36050028e-01 -5.33249855e-01 -2.74426430e-01
1.53051391e-01 4.25315738e-01 2.40069509e-01 6.27338886e-01
2.87834078e-01 3.77532750e-01 -1.03525355e-01 1.73748836e-01
-4.26745057e-01 1.24268281e+00 -7.88599849e-01 1.31311846e+00
-4.81837422e-01 5.92713058e-01 -2.42381558e-01 -7.80112028e-01
9.07621562e-01 3.61709237e-01 1.35973558e-01 -5.17228723e-01
8.98079947e-02 7.30662584e-01 7.17785418e-01 -1.47521207e-02
4.99557495e-01 1.85430482e-01 -4.91227299e-01 2.53511071e-01
2.40797743e-01 -2.59368420e-01 2.30869725e-01 -1.31844699e-01
6.33329988e-01 7.80987600e-03 5.36994755e-01 -7.57063568e-01
6.22841597e-01 4.79170829e-02 4.87968981e-01 1.00164747e+00
-2.36776114e-01 6.19069636e-01 1.58753291e-01 -1.36500346e-02
-1.09524179e+00 -9.77665961e-01 -4.72205251e-01 1.37189031e+00
-6.32311165e-01 -5.60757995e-01 -5.41501284e-01 -4.58006889e-01
1.96127564e-01 8.02580357e-01 -3.93598527e-01 -4.46105599e-02
-3.71099204e-01 -4.41245258e-01 1.05958760e+00 5.99042296e-01
-4.64015156e-02 -1.27631795e+00 -2.67165359e-02 1.47194266e-01
-1.79561287e-01 -1.01182616e+00 -6.96601629e-01 4.89185929e-01
-3.36803436e-01 -3.77429754e-01 -6.94115281e-01 -9.35317695e-01
-2.63259504e-02 -1.56088043e-02 9.14825141e-01 -1.32304370e-01
-4.73707169e-01 3.08931530e-01 -4.60896790e-01 -4.91766870e-01
-7.38355517e-01 4.80027884e-01 4.07801300e-01 -4.26053494e-01
8.33700776e-01 -2.66363382e-01 5.35401292e-02 1.01199374e-01
-1.93420470e-01 -4.63848501e-01 3.32734793e-01 8.16397667e-01
3.94663453e-01 -1.73524782e-01 4.69462425e-01 -8.32983851e-01
7.46408582e-01 -1.66691646e-01 -5.38867451e-02 4.28682789e-02
-4.70108628e-01 -2.28577822e-01 7.19676733e-01 -6.75525904e-01
-6.98101640e-01 -3.41485813e-02 -7.42308855e-01 -2.40110029e-02
-5.07178545e-01 4.77164358e-01 -2.65807986e-01 -3.65545779e-01
3.85924757e-01 2.06059262e-01 -2.93793082e-01 -9.10574019e-01
1.68892741e-01 1.01786137e+00 6.41822040e-01 -5.61732888e-01
6.16366744e-01 -3.11837614e-01 -6.02860391e-01 -1.01349509e+00
-5.78874528e-01 -6.82448268e-01 -7.18135774e-01 1.90430760e-01
9.57698882e-01 -8.09146821e-01 -5.41915119e-01 7.44834960e-01
-9.48514462e-01 -1.55368701e-01 -2.24684060e-01 7.51554370e-01
-4.23030376e-01 3.62680465e-01 -5.05673766e-01 -8.18474889e-01
-3.02878648e-01 -1.44593847e+00 8.04869533e-01 1.80452108e-01
-8.64070296e-01 -8.85231197e-01 1.29798174e-01 2.42076367e-01
7.04892755e-01 -1.50264382e-01 1.06897938e+00 -1.42287302e+00
1.89751506e-01 -1.74696550e-01 9.20130834e-02 4.77064252e-01
3.36460412e-01 1.91086248e-01 -1.21674395e+00 -1.78161725e-01
-3.87461841e-01 -4.08809036e-01 8.03308606e-01 3.89249206e-01
1.05540037e+00 -3.12440634e-01 -2.19137937e-01 9.45179582e-01
1.18540478e+00 4.90214199e-01 4.58717436e-01 2.88564593e-01
8.61610889e-01 4.84446466e-01 3.34080100e-01 5.31087369e-02
3.01933765e-01 8.40872467e-01 -2.76610613e-01 2.94427514e-01
-3.00536305e-01 -2.98470557e-01 3.36775601e-01 9.65859413e-01
-1.49229178e-02 -4.31781113e-02 -1.32605827e+00 7.84458220e-01
-1.25167346e+00 -8.30762506e-01 -2.16328174e-01 2.10813832e+00
8.41370583e-01 8.59588012e-02 4.96880472e-01 4.07220423e-01
5.84353924e-01 2.20911484e-03 -1.16869867e-01 -9.35881793e-01
-5.74632943e-01 4.62725729e-01 5.35886288e-01 7.15183020e-01
-1.06519008e+00 1.48895216e+00 8.17989349e+00 1.16205919e+00
-1.03580618e+00 5.32637239e-02 3.75958145e-01 1.02892563e-01
-5.39047062e-01 3.03825215e-02 -1.26610327e+00 7.09615290e-01
1.42192137e+00 -3.21078598e-01 6.56921685e-01 8.01158667e-01
-3.74484986e-01 -9.92027670e-02 -1.09701502e+00 9.30293620e-01
1.02470577e-01 -8.80028546e-01 1.48340821e-01 3.01038116e-01
5.56015730e-01 1.42046794e-01 2.95003265e-01 3.26223403e-01
3.20497394e-01 -1.46347916e+00 8.00905347e-01 1.29212782e-01
1.06633627e+00 -9.36115801e-01 5.03287733e-01 5.75208142e-02
-8.89205933e-01 5.39850034e-02 -3.94875765e-01 -1.41423032e-01
-1.02145061e-01 3.24078143e-01 -7.10030913e-01 3.50402355e-01
6.52425170e-01 5.20552456e-01 -8.13276649e-01 1.13390803e+00
-1.26628473e-01 1.01419735e+00 -5.49211383e-01 -1.63226843e-01
4.52737510e-01 -1.58084072e-02 6.78746879e-01 1.69728637e+00
1.97809070e-01 -4.72127885e-01 1.42714679e-01 4.83263999e-01
2.64026046e-01 5.18480241e-01 -1.01163220e+00 -4.69159096e-01
7.18326807e-01 1.08702362e+00 -4.67303097e-01 -5.68566442e-01
-4.22231019e-01 1.04718173e+00 3.05287182e-01 9.41539779e-02
-3.56894433e-01 -6.69799805e-01 1.03153062e+00 -2.99316913e-01
1.03297845e-01 -2.67464608e-01 -4.62731719e-01 -8.39269042e-01
-3.81990314e-01 -1.24283135e+00 2.00433865e-01 -4.90221232e-01
-1.32525170e+00 5.91138422e-01 1.57458372e-02 -7.31875956e-01
-8.11343729e-01 -9.58501995e-01 -5.62556565e-01 1.35145724e+00
-1.12245643e+00 -9.60013747e-01 3.23903054e-01 3.53837162e-01
4.14692849e-01 -5.15662849e-01 1.11946774e+00 2.05936551e-01
-3.77608597e-01 1.04787111e+00 3.76256198e-01 3.85913819e-01
1.00140417e+00 -1.27295947e+00 9.10213888e-01 6.89912736e-01
4.41236287e-01 9.01205420e-01 7.21516609e-01 -6.65877998e-01
-9.68101978e-01 -5.96306086e-01 1.55146670e+00 -4.46467429e-01
6.91450775e-01 -4.28622335e-01 -7.69079864e-01 8.66685808e-01
1.10855445e-01 -4.71204400e-01 8.44738543e-01 4.53086734e-01
-4.86353457e-01 3.71081293e-01 -1.02895606e+00 3.95283043e-01
1.26988602e+00 -9.82889831e-01 -8.27775896e-01 3.37622106e-01
6.16674900e-01 -1.08152814e-01 -8.91484082e-01 4.64353144e-01
7.48216271e-01 -6.62705421e-01 1.21299851e+00 -5.80017686e-01
1.29842579e-01 2.32498109e-01 -2.59538561e-01 -1.65772438e+00
-7.30827808e-01 -5.26885927e-01 7.35833168e-01 1.70188284e+00
8.48517835e-01 -1.04140258e+00 2.99649090e-01 3.75364304e-01
-2.13949844e-01 -3.19936275e-01 -1.12930024e+00 -1.21537137e+00
6.35828912e-01 -5.49902678e-01 5.50295472e-01 9.40350175e-01
3.30583423e-01 2.30054259e-01 -3.07162762e-01 -5.89341938e-01
1.30446121e-01 -2.74116307e-01 4.92778242e-01 -1.48116100e+00
-3.99929971e-01 -9.21979010e-01 -3.98611069e-01 -4.89706188e-01
6.10094547e-01 -1.46704721e+00 1.58277377e-02 -1.30277717e+00
3.54165643e-01 -5.35530269e-01 -1.53560638e-01 7.26431549e-01
-3.65812182e-01 4.30815041e-01 4.23408508e-01 1.59294903e-01
-4.37637838e-03 -8.77254456e-02 5.17043591e-01 -1.66512474e-01
-2.46517360e-01 -1.63176298e-01 -7.68319607e-01 7.26300120e-01
1.06508899e+00 -3.77310097e-01 -9.75938886e-02 -5.07657290e-01
-1.76040068e-01 -4.38965887e-01 -1.52605653e-01 -1.13604295e+00
-4.43106629e-02 -4.51947609e-03 5.44086158e-01 -5.19937575e-01
5.98428786e-01 -1.22789405e-01 8.91179144e-02 3.98781598e-01
-2.65614212e-01 3.68413568e-01 3.51703256e-01 -2.85154760e-01
-1.72361061e-01 -5.06375253e-01 6.72086298e-01 -1.21574052e-01
-7.07685828e-01 -3.89904082e-02 -9.20806587e-01 1.53591976e-01
4.74904597e-01 -3.09365779e-01 -2.13316858e-01 -2.84252971e-01
-4.42283332e-01 -1.58908680e-01 6.15841508e-01 7.32703686e-01
1.07454054e-01 -1.28943145e+00 -7.45760858e-01 3.90998542e-01
1.14441402e-01 -1.05422354e+00 -2.51337051e-01 6.25682533e-01
-1.99285761e-01 1.02668166e+00 -3.22871029e-01 -4.15338784e-01
-1.26026511e+00 4.48242962e-01 -9.46102291e-03 -1.18802220e-01
-3.05538535e-01 9.63415563e-01 -1.63507029e-01 -8.03261280e-01
4.56468940e-01 -9.97640714e-02 -4.34865296e-01 3.47330749e-01
7.20416248e-01 2.98295915e-01 -2.87673399e-02 -1.21838987e+00
-8.04672480e-01 7.91621745e-01 -1.80203706e-01 -6.65342510e-01
1.04396474e+00 1.93108656e-02 -1.09637283e-01 1.04205573e+00
1.30851710e+00 3.60994071e-01 -1.09544441e-01 -1.44759193e-01
2.83392072e-01 -5.05260408e-01 -7.37603903e-02 -9.49387312e-01
-4.79507178e-01 6.91770673e-01 3.70725751e-01 1.67312503e-01
2.76656061e-01 -1.48526624e-01 7.40976214e-01 1.93977714e-01
2.63283670e-01 -1.33759356e+00 -5.13575792e-01 9.71815586e-01
4.17494953e-01 -1.09157646e+00 -3.44457388e-01 -3.58671933e-01
-7.88987517e-01 9.25381184e-01 3.50120932e-01 -6.49953410e-02
5.19758344e-01 3.54315609e-01 1.95205092e-01 3.04008901e-01
-4.55331206e-01 -6.29695356e-01 4.81545627e-01 8.01327169e-01
9.23769414e-01 4.20060068e-01 -8.16109896e-01 4.84491467e-01
-9.07808602e-01 -5.18476188e-01 1.96796864e-01 4.03715938e-01
-5.60502052e-01 -1.35137856e+00 -4.94584292e-01 8.23921680e-01
-7.59568870e-01 -4.78010803e-01 -7.60593653e-01 6.69358253e-01
1.26263097e-01 1.18816447e+00 3.02390933e-01 -6.60102546e-01
-1.59810022e-01 9.01058674e-01 4.52323765e-01 -7.50336289e-01
-1.04454982e+00 -5.52535243e-02 5.26971519e-01 -3.23797315e-01
-4.50865254e-02 -1.16464341e+00 -9.10041213e-01 -3.54708433e-01
-8.26249197e-02 1.48058951e-01 9.64537084e-01 8.79427612e-01
-6.41530082e-02 1.39527082e-01 4.94499862e-01 -8.41505468e-01
-4.73768979e-01 -1.32866633e+00 -4.19419199e-01 2.60858804e-01
1.83930978e-01 -5.24719357e-01 -8.78455818e-01 -4.56538230e-01] | [10.489344596862793, 10.518136024475098] |
e4f40fb1-1235-4975-ad3b-4ccdfdd0e27c | a-weakly-supervised-learning-framework-for | 2209.02957 | null | https://arxiv.org/abs/2209.02957v1 | https://arxiv.org/pdf/2209.02957v1.pdf | A Weakly Supervised Learning Framework for Salient Object Detection via Hybrid Labels | Fully-supervised salient object detection (SOD) methods have made great progress, but such methods often rely on a large number of pixel-level annotations, which are time-consuming and labour-intensive. In this paper, we focus on a new weakly-supervised SOD task under hybrid labels, where the supervision labels include a large number of coarse labels generated by the traditional unsupervised method and a small number of real labels. To address the issues of label noise and quantity imbalance in this task, we design a new pipeline framework with three sophisticated training strategies. In terms of model framework, we decouple the task into label refinement sub-task and salient object detection sub-task, which cooperate with each other and train alternately. Specifically, the R-Net is designed as a two-stream encoder-decoder model equipped with Blender with Guidance and Aggregation Mechanisms (BGA), aiming to rectify the coarse labels for more reliable pseudo-labels, while the S-Net is a replaceable SOD network supervised by the pseudo labels generated by the current R-Net. Note that, we only need to use the trained S-Net for testing. Moreover, in order to guarantee the effectiveness and efficiency of network training, we design three training strategies, including alternate iteration mechanism, group-wise incremental mechanism, and credibility verification mechanism. Experiments on five SOD benchmarks show that our method achieves competitive performance against weakly-supervised/unsupervised methods both qualitatively and quantitatively. | ['Sam Kwong', 'Yao Zhao', 'Shiqi Wang', 'Qiuping Jiang', 'Chen Zhang', 'Qi Qin', 'Runmin Cong'] | 2022-09-07 | null | null | null | null | ['saliency-detection', 'salient-object-detection'] | ['computer-vision', 'computer-vision'] | [ 4.25171614e-01 3.19231391e-01 -1.71855703e-01 -3.53189796e-01
-8.16458285e-01 -1.08811080e-01 4.04599398e-01 4.06162404e-02
-4.93497252e-01 6.18949890e-01 1.38916727e-02 2.29167752e-02
4.68808353e-01 -6.05782926e-01 -6.02314055e-01 -1.05827653e+00
4.21128452e-01 1.96877360e-01 7.80509710e-01 8.24186429e-02
-7.91791975e-02 -2.40430664e-02 -1.57071495e+00 9.77125987e-02
8.82981777e-01 1.26042378e+00 3.48887295e-01 2.28255957e-01
-5.09907193e-02 1.13784134e+00 -3.05705667e-01 -2.68862844e-01
2.72073179e-01 -5.54773569e-01 -5.80525696e-01 3.65471244e-01
9.05799717e-02 -4.83071387e-01 -7.06945956e-02 1.32743490e+00
7.13746190e-01 -1.60764351e-01 4.42437619e-01 -1.33806169e+00
-4.11076963e-01 7.68652499e-01 -8.90620172e-01 -6.86147660e-02
-3.11564416e-01 3.46776664e-01 1.04777157e+00 -9.83809948e-01
4.35173333e-01 9.32833135e-01 7.00919569e-01 5.91511607e-01
-1.13851035e+00 -8.08433950e-01 4.25714552e-01 6.10657083e-03
-1.36998248e+00 -4.79635298e-01 1.04049313e+00 -3.76987040e-01
3.00821513e-01 1.73497964e-02 4.23947036e-01 8.66707444e-01
-3.66134316e-01 1.10760915e+00 1.02778327e+00 -3.91700715e-01
3.68596375e-01 3.73932928e-01 1.20792858e-01 8.70702565e-01
9.63330343e-02 -6.89142868e-02 -5.60846388e-01 1.21538274e-01
6.58498645e-01 8.54722857e-02 -2.59930521e-01 -4.28301632e-01
-1.13330102e+00 7.33214498e-01 5.46582699e-01 7.89927170e-02
-2.19950020e-01 4.25596759e-02 4.93088752e-01 -1.28739998e-02
6.28663659e-01 1.68815292e-02 -4.49924946e-01 2.75028050e-01
-1.04697466e+00 -1.22464292e-01 3.30521822e-01 1.21740937e+00
9.70025420e-01 -1.57065496e-01 -4.36724961e-01 8.54400277e-01
5.90199828e-01 1.66903421e-01 4.94695753e-01 -6.36927009e-01
4.32721496e-01 8.09291661e-01 4.46117707e-02 -7.13259041e-01
-4.24621105e-01 -8.14176798e-01 -1.00599742e+00 1.40991807e-01
2.88088471e-01 -1.33209333e-01 -1.03753448e+00 1.65163839e+00
6.01177335e-01 1.32997245e-01 -6.58730417e-02 1.16319394e+00
8.81511807e-01 5.85891306e-01 3.15954000e-01 -4.63675469e-01
1.36393380e+00 -1.48220015e+00 -7.55157590e-01 -2.63813734e-01
8.45091820e-01 -5.06883860e-01 1.18208146e+00 2.99239516e-01
-1.16738379e+00 -6.46770179e-01 -1.18501818e+00 -3.10584486e-01
-1.32688463e-01 7.35095918e-01 3.41218621e-01 3.57112944e-01
-7.68353343e-01 3.69771749e-01 -6.71543419e-01 -9.83350947e-02
7.73506939e-01 2.49055579e-01 -9.20559093e-02 1.44808874e-01
-1.12977111e+00 4.83852476e-01 5.26607871e-01 4.51153934e-01
-1.05914938e+00 -3.99903268e-01 -8.49343598e-01 1.36791870e-01
5.49481809e-01 -2.20741332e-01 1.33307004e+00 -1.26271951e+00
-1.39309037e+00 1.03096747e+00 -7.65984952e-02 -3.40569526e-01
8.07140291e-01 -6.39651269e-02 -2.37264335e-01 8.21106210e-02
4.30288523e-01 7.20445335e-01 7.69898593e-01 -1.47320497e+00
-1.01400745e+00 -1.55778140e-01 -4.20682728e-02 2.87898570e-01
-4.62584347e-01 1.06089756e-01 -6.32550657e-01 -6.83791935e-01
2.39372492e-01 -7.38932788e-01 -1.56488478e-01 3.56459826e-01
-6.27696574e-01 -2.19963416e-01 8.30328465e-01 -5.56669593e-01
1.22432649e+00 -2.39256930e+00 -1.18206315e-01 1.28691439e-02
4.22025293e-01 3.99167418e-01 6.00870140e-02 -1.01060338e-01
-1.03134867e-02 -9.00848731e-02 -5.62157273e-01 -8.51035893e-01
1.24899624e-02 1.20150879e-01 -2.77407646e-01 4.80564922e-01
5.22659421e-01 7.64616668e-01 -1.18571484e+00 -1.00961947e+00
-1.22149162e-01 1.02347046e-01 -1.97954029e-01 3.32486033e-01
-3.70550722e-01 3.47914815e-01 -4.72015023e-01 8.73750210e-01
7.28258193e-01 -5.67705274e-01 -1.06989473e-01 -2.96781570e-01
-2.56950706e-01 2.70585537e-01 -1.36485541e+00 1.58359396e+00
-1.79251820e-01 1.89696029e-01 1.41294956e-01 -9.44374502e-01
9.38369572e-01 1.26988053e-01 2.26019010e-01 -6.40927315e-01
3.84598702e-01 2.31863454e-01 -2.49913350e-01 -4.45299298e-01
2.09984615e-01 -2.17003196e-01 1.27487451e-01 5.69870353e-01
5.19458652e-02 1.84314564e-01 1.55759975e-01 2.54813194e-01
9.79573965e-01 3.23192269e-01 8.19328651e-02 -5.71049824e-02
7.79672027e-01 -2.51225810e-02 1.18981850e+00 4.60288823e-01
-4.36932385e-01 8.42750371e-01 7.12274730e-01 -1.36160359e-01
-9.01126325e-01 -7.59756863e-01 -4.45932299e-02 1.25649345e+00
6.16726875e-01 -2.77188748e-01 -6.93235159e-01 -1.08910882e+00
-3.41179460e-01 3.78513396e-01 -6.72414899e-01 -2.30007708e-01
-3.77315521e-01 -9.37261224e-01 4.37382489e-01 6.25393391e-01
9.36267972e-01 -1.19297516e+00 -5.57551026e-01 2.28510961e-01
-8.51366520e-02 -9.48986351e-01 -5.87719381e-01 4.93370861e-01
-5.75349152e-01 -9.45675850e-01 -7.51585186e-01 -1.08158898e+00
9.76507008e-01 4.01100785e-01 7.45629013e-01 3.66120160e-01
8.75537321e-02 -3.82604182e-01 -4.94825006e-01 -3.74654561e-01
-2.35166118e-01 1.47578463e-01 -2.68386304e-01 3.73084486e-01
1.69355392e-01 -3.46050978e-01 -6.84445322e-01 5.54152548e-01
-9.46956038e-01 5.47582150e-01 8.01802516e-01 1.06988084e+00
8.30810547e-01 1.05027609e-01 7.22794652e-01 -1.06269753e+00
9.49203596e-02 -4.67085481e-01 -7.12139249e-01 2.12618321e-01
-7.07885742e-01 -8.15018639e-02 6.90401912e-01 -3.88151437e-01
-1.30114579e+00 5.38842797e-01 -1.65533990e-01 -3.70697528e-01
3.69804129e-02 3.24482024e-01 -5.98682523e-01 1.09027803e-01
5.68859160e-01 3.05046856e-01 -1.06342189e-01 -4.88920182e-01
3.06455642e-01 9.92502093e-01 6.84036911e-01 -2.56975323e-01
8.55612874e-01 6.17150664e-01 -2.73865342e-01 -1.16663456e-01
-1.28417456e+00 -5.30998826e-01 -6.05438530e-01 -3.08964998e-01
7.89049625e-01 -1.16928828e+00 -2.25657672e-01 8.72603238e-01
-1.14140546e+00 -5.20781457e-01 -4.65905786e-01 2.20555305e-01
-2.56906450e-01 2.80377954e-01 -7.17254758e-01 -7.18297243e-01
-3.81439209e-01 -1.14529157e+00 1.32505393e+00 4.26799774e-01
2.93058455e-01 -5.49957395e-01 -1.94377854e-01 3.48416477e-01
1.08742572e-01 1.10603295e-01 5.55197358e-01 -6.17369950e-01
-4.74828094e-01 -2.89118066e-02 -7.55698383e-01 5.39905190e-01
7.00769946e-02 -1.23811282e-01 -1.37151384e+00 -1.52851269e-01
4.19711620e-02 -5.92854500e-01 1.00503731e+00 5.48857190e-02
1.19082403e+00 -1.09672129e-01 -3.15951169e-01 5.22580922e-01
1.25119591e+00 -1.70033216e-01 5.10337353e-01 3.49853456e-01
9.02315557e-01 6.88572407e-01 8.82407725e-01 3.44973475e-01
5.07565498e-01 3.81139636e-01 5.84611237e-01 -4.07328546e-01
-2.43283749e-01 -4.46187973e-01 3.11822534e-01 6.84263706e-01
1.55762404e-01 -1.20119728e-01 -5.95842183e-01 4.92111415e-01
-2.20198727e+00 -6.95317030e-01 -3.50126743e-01 2.07794213e+00
1.24118721e+00 4.78397340e-01 2.40474686e-01 2.76405841e-01
9.41098452e-01 2.04583570e-01 -7.05203891e-01 2.86408275e-01
-2.02884406e-01 -3.33797753e-01 5.48079133e-01 9.41446126e-02
-1.24817479e+00 9.49528754e-01 5.37713003e+00 9.86683011e-01
-1.08911264e+00 4.32313353e-01 9.33994770e-01 -7.01527447e-02
-1.62910372e-01 7.28589743e-02 -9.00282383e-01 7.62380481e-01
4.19265956e-01 3.67400080e-01 -6.55431971e-02 9.86625850e-01
2.48202428e-01 -3.57809275e-01 -9.18618143e-01 8.64433348e-01
1.68752670e-02 -1.15164304e+00 -2.49984011e-01 -1.98124424e-01
8.22277904e-01 7.69435689e-02 -2.23068178e-01 3.27427089e-01
2.25992993e-01 -4.76886749e-01 1.09372699e+00 2.26873159e-01
8.33039701e-01 -4.54207182e-01 9.31456268e-01 6.11659229e-01
-1.27025831e+00 -1.11616075e-01 -3.60729218e-01 -5.15291514e-03
2.54210651e-01 9.33789670e-01 -2.13766217e-01 4.10740852e-01
7.45582819e-01 8.76740873e-01 -5.80512583e-01 1.13045526e+00
-6.15000844e-01 6.99064374e-01 -2.26577595e-01 8.84460136e-02
1.42243907e-01 -2.24576220e-02 1.62139922e-01 1.19158220e+00
-1.27855912e-01 2.88071912e-02 2.11162373e-01 9.62394595e-01
-3.22539091e-01 -1.95742061e-04 -3.24159004e-02 3.55423719e-01
4.79576141e-01 1.73480105e+00 -1.03789532e+00 -5.98049283e-01
-4.47595984e-01 8.35549712e-01 4.91377056e-01 2.08735853e-01
-1.00867438e+00 -3.24862987e-01 3.32330912e-02 8.61675069e-02
2.50916451e-01 2.14765117e-01 -4.21524376e-01 -1.03662848e+00
1.64696500e-01 -5.82614422e-01 3.83933425e-01 -9.43420112e-01
-1.29530323e+00 3.96866471e-01 -2.45134488e-01 -1.41253746e+00
2.27730244e-01 -2.45366052e-01 -6.71817005e-01 8.13172758e-01
-1.98882914e+00 -1.28831220e+00 -6.07446074e-01 3.74068588e-01
5.69090247e-01 1.63408175e-01 2.11516127e-01 4.98045623e-01
-9.18549001e-01 5.68199098e-01 -6.38803095e-02 3.00686866e-01
7.76705086e-01 -1.21778083e+00 1.85538847e-02 9.83151555e-01
-1.76532850e-01 9.18976963e-02 4.28660959e-01 -6.36265635e-01
-6.99350893e-01 -1.52269959e+00 8.19297910e-01 4.61204126e-02
5.96892357e-01 -5.04845142e-01 -1.04898775e+00 4.66504395e-01
-1.88124582e-01 4.00156200e-01 4.00958180e-01 -3.08956444e-01
-1.50240481e-01 -2.93861270e-01 -1.10982239e+00 4.37027812e-01
1.12825823e+00 -4.39604789e-01 -3.73359293e-01 4.78351712e-01
8.63029957e-01 -4.58781451e-01 -4.22195286e-01 4.89621609e-01
1.58266723e-01 -9.69632089e-01 4.92322266e-01 4.38155159e-02
5.21367550e-01 -8.59675586e-01 2.87211925e-01 -8.80088151e-01
-2.67645597e-01 -4.75687027e-01 -6.15036152e-02 1.71402371e+00
2.79061407e-01 -4.90568489e-01 9.65040386e-01 3.95192474e-01
-3.45210642e-01 -9.25231099e-01 -7.33561814e-01 -5.46897471e-01
-5.05974948e-01 -1.80104181e-01 5.48806608e-01 9.15989220e-01
-1.53453276e-01 3.98691446e-01 -4.36427474e-01 2.00171426e-01
6.17416978e-01 2.46709511e-02 6.67952716e-01 -1.10958898e+00
-2.51023650e-01 -4.30444360e-01 -6.60248995e-02 -1.20337462e+00
-2.80980319e-02 -6.45913959e-01 6.25221908e-01 -1.40987980e+00
4.25989330e-01 -7.71928489e-01 -4.69770551e-01 8.42363894e-01
-4.14327294e-01 4.25943464e-01 -4.52690683e-02 5.25582552e-01
-9.59546685e-01 8.81714284e-01 1.14723647e+00 -2.05114514e-01
-2.85797656e-01 -1.07200429e-01 -7.92668521e-01 8.75464320e-01
6.08791530e-01 -6.53210223e-01 -4.85805124e-01 -3.44261497e-01
2.39497289e-01 -3.89078975e-01 4.63267833e-01 -8.53924513e-01
4.22615945e-01 4.22768621e-03 1.49233058e-01 -5.54114163e-01
-3.38324718e-02 -7.19216764e-01 -2.76005358e-01 1.90606102e-01
-4.58275348e-01 -4.44218159e-01 -1.87767416e-01 6.30740821e-01
-2.48439357e-01 -3.40850592e-01 9.74791825e-01 7.28462338e-02
-6.93532348e-01 4.01668042e-01 2.64193658e-02 5.28899319e-02
1.11368275e+00 -1.29371539e-01 -3.95302594e-01 3.53009775e-02
-5.52652180e-01 7.15612113e-01 4.18230951e-01 1.43732339e-01
3.82237434e-01 -1.32056844e+00 -5.89086771e-01 1.29709169e-01
3.44527215e-01 7.58296251e-01 1.87343672e-01 1.00566757e+00
-2.94157684e-01 -2.82827526e-01 1.48232803e-01 -5.33780217e-01
-9.96384442e-01 4.80828971e-01 3.38024735e-01 -3.18610340e-01
-5.96975505e-01 9.41971719e-01 3.95424426e-01 -2.63026029e-01
4.47218329e-01 -2.57474959e-01 -2.52517670e-01 2.08093211e-01
5.18146276e-01 1.92791700e-01 -7.11473301e-02 -7.24884272e-01
-2.02368915e-01 3.39205325e-01 -1.10593855e-01 1.12075835e-01
1.20670712e+00 -3.20748299e-01 -1.44385457e-01 5.09881794e-01
9.18848336e-01 -2.41541803e-01 -1.75648534e+00 -5.43504357e-01
-9.90994275e-02 -1.45853788e-01 3.03610206e-01 -6.52752697e-01
-1.29838729e+00 8.06120157e-01 5.43864906e-01 1.43764243e-01
1.32834601e+00 8.54279399e-02 8.32959592e-01 -3.42929289e-02
1.18035369e-01 -1.43716502e+00 8.36771131e-02 1.76546827e-01
4.83358890e-01 -1.44553697e+00 -3.86603107e-03 -5.52116871e-01
-7.12603509e-01 7.14187562e-01 8.68030429e-01 2.06008684e-02
5.66814125e-01 1.57890454e-01 1.15265258e-01 -5.95819280e-02
-5.45371652e-01 -4.70014840e-01 9.44419876e-02 4.07597870e-01
1.09484918e-01 -2.74511069e-01 -2.25146264e-01 8.89619887e-01
3.15257698e-01 2.03425467e-01 2.65430510e-01 8.80185902e-01
-6.27799094e-01 -8.69157791e-01 -1.41099930e-01 5.17655075e-01
-1.19360790e-01 -1.94976985e-01 -8.49324763e-02 5.07878423e-01
6.61682963e-01 9.00503039e-01 -1.23072252e-01 -3.88990730e-01
1.54662579e-01 -1.17057510e-01 -1.75262704e-01 -7.45595515e-01
-5.15904784e-01 3.22863191e-01 1.24804154e-02 -3.20309252e-01
-7.04670906e-01 -6.21420443e-01 -1.52172637e+00 1.83810785e-01
-8.75662327e-01 1.02181055e-01 3.11695486e-01 1.08631194e+00
1.47003591e-01 5.24127424e-01 7.79777050e-01 -8.11341643e-01
-5.75431466e-01 -1.03139162e+00 -7.95579731e-01 3.24368268e-01
3.79182428e-01 -6.99883163e-01 -4.62579459e-01 3.26154947e-01] | [9.286413192749023, 1.243064284324646] |
ab940292-7ef7-4f83-94b5-4a901dcebe3c | empowering-llm-based-machine-translation-with | 2305.14328 | null | https://arxiv.org/abs/2305.14328v1 | https://arxiv.org/pdf/2305.14328v1.pdf | Empowering LLM-based Machine Translation with Cultural Awareness | Traditional neural machine translation (NMT) systems often fail to translate sentences that contain culturally specific information. Most previous NMT methods have incorporated external cultural knowledge during training, which requires fine-tuning on low-frequency items specific to the culture. Recent in-context learning utilizes lightweight prompts to guide large language models (LLMs) to perform machine translation, however, whether such an approach works in terms of injecting culture awareness into machine translation remains unclear. To this end, we introduce a new data curation pipeline to construct a culturally relevant parallel corpus, enriched with annotations of cultural-specific entities. Additionally, we design simple but effective prompting strategies to assist this LLM-based translation. Extensive experiments show that our approaches can largely help incorporate cultural knowledge into LLM-based machine translation, outperforming traditional NMT systems in translating cultural-specific sentences. | ['Junjie Hu', 'Diyi Yang', 'Ming Jiang', 'Binwei Yao'] | 2023-05-23 | null | null | null | null | ['nmt'] | ['computer-code'] | [ 2.07882077e-01 -5.73653243e-02 -5.17067552e-01 -4.97169405e-01
-1.15408504e+00 -8.69419098e-01 7.27825165e-01 -6.21505268e-02
-6.30847633e-01 1.06774104e+00 6.72944427e-01 -6.05757594e-01
5.23744285e-01 -5.39502382e-01 -1.03015339e+00 -5.99467345e-02
5.71640790e-01 8.01875234e-01 -4.96956497e-01 -6.62107766e-01
1.62655577e-01 -2.35823467e-01 -6.42639935e-01 6.68351650e-01
1.51356351e+00 -1.52035803e-02 3.89941543e-01 3.83658469e-01
-3.64300489e-01 4.82887477e-01 -6.12408698e-01 -9.48194623e-01
1.28069252e-01 -8.88748109e-01 -8.56299579e-01 -4.60802674e-01
3.52827251e-01 -1.51424542e-01 4.58509594e-01 9.84780908e-01
3.96184266e-01 -2.94633955e-01 6.08601451e-01 -8.71874392e-01
-1.58778298e+00 1.20216250e+00 -1.71966121e-01 -8.00002366e-02
5.27620077e-01 2.41380241e-02 8.76533091e-01 -1.26087153e+00
1.08969545e+00 1.16975522e+00 9.28362191e-01 1.00446963e+00
-1.10178506e+00 -6.24469638e-01 1.76606059e-01 2.14327686e-02
-1.09685957e+00 -3.42550009e-01 7.29288340e-01 -3.14374059e-01
1.33336198e+00 2.73390889e-01 7.06721902e-01 1.89997029e+00
7.51886591e-02 7.00548887e-01 1.49238682e+00 -8.34368289e-01
-2.20078398e-02 4.87674236e-01 -3.17079186e-01 4.57644790e-01
2.91372597e-01 -1.82358235e-01 -8.72457147e-01 -8.36857781e-02
5.84247410e-01 -5.31914055e-01 -1.82821691e-01 2.41511643e-01
-1.74537086e+00 7.69626141e-01 -3.99499238e-02 4.53227758e-01
-2.59297490e-01 -1.29995495e-01 3.17351937e-01 6.93565845e-01
7.12764204e-01 8.91700149e-01 -7.13028610e-01 -3.08697253e-01
-6.58748746e-01 1.44816786e-02 8.41206968e-01 1.36545491e+00
7.36030638e-01 -2.38052741e-01 -2.00728968e-01 1.29221535e+00
-5.80120422e-02 8.95176172e-01 4.88622338e-01 -8.33169818e-01
9.01243091e-01 4.84502852e-01 1.78980112e-01 -4.56417620e-01
-1.71167269e-01 -4.83312964e-01 -4.99046355e-01 -7.60400474e-01
2.03152344e-01 -5.82012892e-01 -6.77577555e-01 1.99029613e+00
1.64694339e-01 -1.27882317e-01 3.86216760e-01 1.07749248e+00
3.12257767e-01 5.18709302e-01 1.97479948e-01 -2.18225211e-01
1.27999473e+00 -1.07949591e+00 -7.07270443e-01 -6.05370760e-01
1.14457464e+00 -1.21242583e+00 1.56909990e+00 -1.07842751e-01
-9.27172184e-01 -2.62911260e-01 -8.02060366e-01 -2.87631482e-01
-4.79329020e-01 2.05298498e-01 7.50404298e-01 8.44585240e-01
-1.16252887e+00 2.53560454e-01 -4.99993503e-01 -9.21007276e-01
-7.37558156e-02 1.48743257e-01 -2.75658727e-01 -5.00758946e-01
-1.49575365e+00 1.31270444e+00 2.73888290e-01 1.28668249e-01
-4.08656955e-01 -6.46014154e-01 -8.23270440e-01 -5.56126654e-01
1.00055173e-01 -1.31156898e+00 1.31667686e+00 -1.60888600e+00
-1.68668962e+00 8.59909594e-01 -4.29361433e-01 -4.42668647e-02
3.13547909e-01 -5.09197414e-01 -6.28353834e-01 -1.77790701e-01
3.43843848e-01 8.21539462e-01 5.29407859e-01 -1.12706518e+00
-5.25754631e-01 -2.30138144e-03 -1.65420532e-01 5.93030572e-01
-7.88442910e-01 7.11482227e-01 -4.53504533e-01 -8.33286464e-01
-2.01191694e-01 -1.08942056e+00 -2.85845637e-01 -6.30272567e-01
-3.03204268e-01 6.50480539e-02 1.71285272e-01 -9.76336896e-01
1.11069202e+00 -1.47808599e+00 3.45761120e-01 1.13858648e-01
-4.59571093e-01 1.12680137e-01 -5.59577286e-01 6.65070713e-01
5.90761364e-01 3.81796688e-01 -1.45291060e-01 -3.65397364e-01
7.33098164e-02 3.57139140e-01 -2.28884593e-01 -2.54659504e-01
6.27391815e-01 1.27285457e+00 -1.16222000e+00 -4.44525748e-01
-4.26780999e-01 3.99935961e-01 -7.64616132e-01 -3.18222027e-03
-5.13500690e-01 6.51847541e-01 -2.82462448e-01 8.09540153e-01
3.22319329e-01 7.49655813e-02 7.06724823e-01 3.28523368e-01
-8.92480165e-02 6.34094000e-01 -4.11558062e-01 2.02024674e+00
-6.58931732e-01 3.45185876e-01 -1.56656161e-01 -3.80583137e-01
8.17505658e-01 3.69534612e-01 6.14202097e-02 -9.19424772e-01
-1.54522851e-01 7.34541178e-01 -6.46514073e-02 -5.86079895e-01
8.25457692e-01 -2.69208312e-01 -2.45607629e-01 6.31040096e-01
-2.30501010e-03 -2.54412950e-03 1.18315414e-01 -2.34618820e-02
7.64061928e-01 6.33779824e-01 5.93618825e-02 -2.26405054e-01
2.57543653e-01 7.23929763e-01 8.80746126e-01 5.14048517e-01
-3.22938301e-02 4.96099263e-01 -1.11303888e-01 -1.86912626e-01
-1.32817316e+00 -5.47894359e-01 2.71072060e-01 1.30521595e+00
-1.98223904e-01 -6.16418481e-01 -1.11292636e+00 -7.98886538e-01
-2.85962045e-01 8.29957843e-01 -5.11515737e-01 -1.46013558e-01
-9.81426418e-01 -9.53625441e-01 6.88630819e-01 6.06506944e-01
1.08001530e-01 -7.60090232e-01 1.68939568e-02 3.88185501e-01
-8.16845596e-01 -1.32767248e+00 -8.52850080e-01 -1.21329509e-01
-9.61713314e-01 -4.08267021e-01 -8.39280307e-01 -1.13123357e+00
6.62495673e-01 2.32675999e-01 1.36961889e+00 -7.87556469e-02
5.09386897e-01 1.26922932e-02 -4.71278995e-01 -5.68633974e-01
-8.72300088e-01 7.04234123e-01 3.80726814e-01 -3.08694035e-01
1.10174632e+00 -4.42438036e-01 -2.58235067e-01 3.78527611e-01
-4.03212845e-01 4.59656328e-01 8.45138490e-01 7.81150877e-01
2.84287065e-01 -1.03380346e+00 7.09871113e-01 -1.05452991e+00
1.02102494e+00 -5.83794177e-01 -1.15603827e-01 6.72857225e-01
-5.67284584e-01 2.10219231e-02 8.95797968e-01 -7.99008489e-01
-1.14868808e+00 -2.06732556e-01 -4.21812721e-02 -8.00148994e-02
8.42869189e-03 8.58705223e-01 -3.43409181e-02 1.20630614e-01
8.95165920e-01 2.70588994e-01 -3.83245260e-01 -4.95201409e-01
5.42585313e-01 8.39343846e-01 4.39397901e-01 -1.22443080e+00
8.86708796e-01 -7.96807930e-02 -5.98964989e-01 -3.65438432e-01
-7.64862776e-01 -1.56398952e-01 -8.40631723e-01 -1.86107963e-01
8.71009409e-01 -1.16951191e+00 6.99937046e-02 4.39209826e-02
-1.12843847e+00 -4.41796154e-01 2.36324593e-01 8.72701049e-01
-4.20433521e-01 2.71623470e-02 -9.37142074e-01 -5.17556489e-01
-4.70116407e-01 -8.77121150e-01 1.17980278e+00 -6.14649504e-02
-7.85421371e-01 -1.08082283e+00 3.89536172e-01 8.28339875e-01
5.76670349e-01 -2.03129262e-01 1.29613090e+00 -4.59938228e-01
-5.40019095e-01 2.36930624e-01 9.20797661e-02 9.83376279e-02
1.03827648e-01 -1.47528663e-01 -6.48595572e-01 -1.30578727e-01
-3.80552441e-01 -4.83382642e-01 3.29861611e-01 -1.00258514e-01
1.17062345e-01 -4.44624454e-01 -6.19896948e-02 7.67573178e-01
1.16171706e+00 -5.54374531e-02 2.92491287e-01 7.10906863e-01
7.68783152e-01 6.27653837e-01 7.00105608e-01 -1.32156044e-01
8.75737071e-01 5.83763361e-01 -4.04695153e-01 -1.42249644e-01
-9.87743959e-02 -6.19526923e-01 8.78037632e-01 1.64990485e+00
-3.84176582e-01 1.07696071e-01 -9.67783391e-01 7.03518271e-01
-1.81605184e+00 -7.19885230e-01 -6.42396510e-02 1.75391757e+00
1.42447782e+00 -1.73890114e-01 -9.71885547e-02 -7.87451863e-01
4.75919247e-01 -3.74045253e-01 -4.31199253e-01 -7.04196274e-01
-5.81840575e-01 -4.07753652e-03 4.11467701e-01 4.71538126e-01
-6.36228025e-01 1.59712529e+00 6.59488440e+00 3.90620053e-01
-1.11781085e+00 4.39081490e-01 2.53154457e-01 -1.84651211e-01
-9.31120932e-01 2.32594416e-01 -8.13221812e-01 2.66449749e-01
1.21600831e+00 -5.14497980e-02 7.32389748e-01 6.61732912e-01
4.14461404e-01 3.41135710e-01 -1.19143355e+00 6.16260886e-01
3.24532598e-01 -1.33970070e+00 3.08415502e-01 -5.37910014e-02
1.11849642e+00 2.99052328e-01 1.34714907e-02 6.75517917e-01
6.85823798e-01 -7.32398212e-01 7.98560023e-01 3.41931313e-01
8.70393634e-01 -7.17314005e-01 6.07518792e-01 5.36024392e-01
-5.50516188e-01 1.48932949e-01 -5.67457676e-01 -1.53925896e-01
1.14170872e-01 3.41609985e-01 -1.24925351e+00 6.33358300e-01
4.45577353e-01 4.21145320e-01 -6.40547097e-01 2.75529593e-01
-4.56220418e-01 1.01305354e+00 -1.10393412e-01 -3.28038454e-01
2.66135097e-01 -4.37686622e-01 3.53855044e-01 1.51783252e+00
6.69067621e-01 -1.27338395e-01 1.75858706e-01 7.04670489e-01
-4.77039635e-01 8.08575332e-01 -7.26003408e-01 -3.22108537e-01
5.56758702e-01 9.01954591e-01 -2.02736586e-01 -4.69204456e-01
-7.37766802e-01 1.44700634e+00 7.53083169e-01 5.54629982e-01
-5.85596979e-01 1.45517498e-01 9.31002140e-01 -2.70765036e-01
-1.01741210e-01 -4.72975463e-01 -5.67482233e-01 -1.73297119e+00
3.35988522e-01 -1.35684252e+00 1.80644348e-01 -9.37125087e-01
-1.40561283e+00 5.01030803e-01 -3.57074350e-01 -1.05837047e+00
-4.69706804e-01 -5.42842269e-01 -1.78858206e-01 1.24101877e+00
-1.42281473e+00 -1.82077038e+00 3.75543237e-01 4.09250945e-01
5.98909259e-01 -2.54806489e-01 1.12021959e+00 3.07987899e-01
-6.69935882e-01 8.32944512e-01 1.83241248e-01 3.67791206e-01
1.47716665e+00 -1.03479064e+00 6.09224677e-01 1.20749354e+00
2.12327510e-01 1.40990448e+00 6.19016290e-01 -1.25097537e+00
-1.57194805e+00 -1.28860831e+00 1.75122845e+00 -1.10104918e+00
6.22623920e-01 -7.65799224e-01 -7.61895061e-01 8.98434877e-01
6.58617556e-01 -7.02003896e-01 1.07682097e+00 7.40527034e-01
-6.59841835e-01 3.00023288e-01 -6.84136510e-01 1.16396940e+00
1.40801489e+00 -9.26170945e-01 -7.11888790e-01 5.90187490e-01
1.01331568e+00 -1.68644309e-01 -1.04051888e+00 1.93329230e-01
5.27533472e-01 -1.26259029e-01 6.66089714e-01 -1.25382090e+00
7.11602688e-01 -2.55537957e-01 -3.25279176e-01 -1.81232321e+00
-4.93746310e-01 -8.80838990e-01 1.47547796e-01 1.24022114e+00
1.08062625e+00 -4.42037076e-01 3.16818684e-01 6.05901182e-01
-3.08545023e-01 -4.18377727e-01 -5.61888516e-01 -7.22729623e-01
5.88344395e-01 -4.03026342e-01 8.71274590e-01 1.87928057e+00
3.33316505e-01 8.30460191e-01 -7.05927610e-01 2.06414983e-01
2.81048596e-01 4.97557782e-02 7.57402360e-01 -6.86422646e-01
-4.02397037e-01 -4.06152457e-01 2.73068607e-01 -5.11073291e-01
3.10606718e-01 -1.22306812e+00 -1.36762321e-01 -1.59718239e+00
4.97288793e-01 -2.79399663e-01 -1.00401454e-01 6.14655435e-01
-7.10602045e-01 4.11643595e-01 4.24851030e-02 3.28331888e-01
-3.61528397e-01 3.03682595e-01 1.20972168e+00 -7.66705896e-04
-2.62193590e-01 -5.12363851e-01 -9.82350588e-01 3.82310539e-01
9.47733283e-01 -5.68706512e-01 -2.73029089e-01 -1.10620308e+00
6.55793905e-01 -3.70221615e-01 -2.09856510e-01 -4.65550691e-01
2.22302690e-01 -6.91534102e-01 3.86767954e-01 -8.56539980e-02
-1.34686427e-02 -4.33982015e-01 1.26765087e-01 6.66069314e-02
-5.78630745e-01 5.44226408e-01 4.56271581e-02 1.48075715e-01
-1.00070469e-01 3.74163352e-02 2.08595842e-01 -3.70253444e-01
-6.20556593e-01 -1.05639108e-01 -5.47219515e-01 1.65527746e-01
5.19441187e-01 -4.48846817e-02 -5.33445477e-01 -2.96281576e-01
-2.03409404e-01 2.19412595e-01 9.48350072e-01 7.82551348e-01
1.67288646e-01 -1.55459487e+00 -1.03241658e+00 -4.61803265e-02
5.16512573e-01 -5.39533079e-01 -1.16342381e-01 9.24117148e-01
-3.76361370e-01 5.66053391e-01 -3.53112400e-01 -3.04437667e-01
-1.00025690e+00 4.98114437e-01 1.23916052e-01 7.90640115e-06
-1.83820739e-01 7.35579014e-01 -2.88297057e-01 -1.12397468e+00
-2.44135946e-01 -3.51475894e-01 1.45515844e-01 -2.18128651e-01
2.96346098e-01 -2.57014066e-01 1.11376978e-01 -5.35936296e-01
-2.51899540e-01 4.55520451e-01 -1.27291098e-01 -7.17876434e-01
1.23920214e+00 -3.10621619e-01 -3.45036864e-01 4.92383927e-01
9.36444581e-01 4.67666268e-01 -7.57082701e-01 -4.83625650e-01
4.44082260e-01 -2.82838792e-01 -5.13270974e-01 -1.43725479e+00
-3.08993489e-01 6.79625988e-01 -3.52506014e-03 -6.78488731e-01
1.00899470e+00 -2.47190312e-01 1.07420075e+00 8.10817897e-01
6.50331199e-01 -1.72334814e+00 -3.54620963e-01 9.61816430e-01
6.57817066e-01 -1.31463289e+00 -4.85987723e-01 -3.12360585e-01
-9.01511848e-01 9.49786246e-01 8.46112311e-01 5.96000373e-01
1.64366271e-02 2.15562701e-01 9.36097085e-01 3.69776040e-01
-1.07141960e+00 -4.79674935e-02 3.50524604e-01 7.44910955e-01
9.50662673e-01 3.45318466e-01 -6.60335541e-01 5.52250087e-01
-6.01368487e-01 -1.00922221e-02 3.49370956e-01 9.11731601e-01
-1.24200106e-01 -1.70854795e+00 -4.43355352e-01 7.26113990e-02
-4.53644425e-01 -7.22259700e-01 -1.10698223e+00 5.45488954e-01
1.50477409e-01 8.40117335e-01 -3.51074785e-01 -6.14881814e-01
1.44595563e-01 6.59879804e-01 5.44286728e-01 -7.16311157e-01
-9.65125442e-01 2.01405451e-01 7.04297900e-01 -3.40209961e-01
-3.54049861e-01 -8.73862386e-01 -7.79876709e-01 -1.75049946e-01
1.03597596e-01 2.23436430e-01 8.70611489e-01 9.85742092e-01
4.98667985e-01 1.91760883e-02 3.55759740e-01 -2.58458823e-01
-3.84860694e-01 -1.16204119e+00 3.35071713e-01 5.11045277e-01
-5.57660684e-02 6.84030354e-02 3.62111688e-01 1.86601475e-01] | [11.616621971130371, 10.325987815856934] |
c140108f-4773-4fd8-a301-615b1f407dcb | novel-chapter-abstractive-summarization-using-1 | 2211.04903 | null | https://arxiv.org/abs/2211.04903v1 | https://arxiv.org/pdf/2211.04903v1.pdf | Novel Chapter Abstractive Summarization using Spinal Tree Aware Sub-Sentential Content Selection | Summarizing novel chapters is a difficult task due to the input length and the fact that sentences that appear in the desired summaries draw content from multiple places throughout the chapter. We present a pipelined extractive-abstractive approach where the extractive step filters the content that is passed to the abstractive component. Extremely lengthy input also results in a highly skewed dataset towards negative instances for extractive summarization; we thus adopt a margin ranking loss for extraction to encourage separation between positive and negative examples. Our extraction component operates at the constituent level; our approach to this problem enriches the text with spinal tree information which provides syntactic context (in the form of constituents) to the extraction model. We show an improvement of 3.71 Rouge-1 points over best results reported in prior work on an existing novel chapter dataset. | ['Kathleen McKeown', 'Vittorio Castelli', 'Muhammad Khalifa', 'Faisal Ladhak', 'Miguel Ballesteros', 'Hardy Hardy'] | 2022-11-09 | null | null | null | null | ['abstractive-text-summarization', 'extractive-summarization'] | ['natural-language-processing', 'natural-language-processing'] | [ 6.98675752e-01 6.22154832e-01 -2.57448405e-01 -2.65974134e-01
-1.33236980e+00 -1.02737367e+00 6.26757264e-01 8.01628888e-01
-5.25171876e-01 9.90915775e-01 9.22053277e-01 -3.17276299e-01
6.21746890e-02 -5.42741597e-01 -4.90397424e-01 -1.81059763e-01
1.87162310e-01 3.80239308e-01 2.59179235e-01 -1.06360681e-01
6.58027053e-01 2.84537405e-01 -1.24018061e+00 7.51977146e-01
9.88587976e-01 4.92310315e-01 -2.42862582e-01 1.15228236e+00
-5.65892637e-01 7.65638709e-01 -1.15173376e+00 -5.32358348e-01
1.46490291e-01 -5.20300865e-01 -9.44723070e-01 1.66690066e-01
7.69375682e-01 -1.56277031e-01 3.78509983e-02 8.86017144e-01
3.46749336e-01 1.43893316e-01 9.37070191e-01 -8.62469316e-01
-9.46328565e-02 1.20433533e+00 -7.32675493e-01 4.03132439e-01
5.54446936e-01 -9.77276415e-02 1.59145439e+00 -9.81406093e-01
8.70135128e-01 9.01528358e-01 4.73111808e-01 4.77416813e-01
-1.29324460e+00 -2.33024985e-01 4.02683616e-01 -2.63947606e-01
-7.09690690e-01 -5.68888068e-01 6.99385703e-01 -4.14991260e-01
1.22229326e+00 6.18449152e-01 7.38475978e-01 9.52850521e-01
2.25406259e-01 1.30070257e+00 6.88050866e-01 -5.20735204e-01
2.69254863e-01 2.52797961e-01 5.89701295e-01 1.63280711e-01
3.69611740e-01 -6.75501883e-01 -6.33719087e-01 -1.24461852e-01
-7.97108468e-03 -3.39755297e-01 -1.87521979e-01 1.13174193e-01
-7.83736050e-01 7.21369445e-01 1.47467941e-01 8.20541233e-02
-6.89228237e-01 -1.57787606e-01 7.46137083e-01 1.56794190e-01
5.20195186e-01 9.96932447e-01 -6.05063021e-01 -3.74271631e-01
-1.62080526e+00 6.85683489e-01 1.16173577e+00 1.00085640e+00
3.69359136e-01 -2.27324978e-01 -7.88413048e-01 6.63729489e-01
-9.72156748e-02 1.47447258e-01 2.17637032e-01 -8.22320580e-01
9.77251112e-01 9.44824934e-01 -1.01043634e-01 -6.80376351e-01
-2.42814362e-01 -7.89648354e-01 -2.33252466e-01 -5.66081889e-02
2.05373242e-01 -4.73177642e-01 -8.24799061e-01 1.30278027e+00
2.72624552e-01 -7.25180149e-01 2.68472105e-01 3.14412922e-01
1.26251984e+00 6.73330009e-01 2.40950167e-01 -3.77483815e-01
1.38473403e+00 -9.35835481e-01 -9.67013299e-01 -5.02046347e-01
6.98371589e-01 -9.07682478e-01 1.06373847e+00 5.04043579e-01
-1.38984621e+00 9.18722525e-02 -1.18737888e+00 -4.47185397e-01
-3.59241366e-01 3.81808877e-01 2.21203730e-01 2.49451891e-01
-6.62666559e-01 6.25979125e-01 -5.25424838e-01 -2.91887641e-01
6.85052574e-01 2.28783906e-01 -2.57242620e-01 5.88140264e-02
-8.53603840e-01 8.55529249e-01 3.92675221e-01 -2.25747108e-01
-7.23109767e-02 -9.76538837e-01 -8.17121446e-01 3.30096215e-01
6.39602602e-01 -5.98287702e-01 1.40480173e+00 -5.51511168e-01
-9.74796176e-01 6.78148270e-01 -3.18932056e-01 -4.27726924e-01
5.95464647e-01 -6.29701436e-01 1.31386966e-01 3.06210458e-01
2.44787350e-01 6.41086936e-01 5.06839335e-01 -1.16226101e+00
-8.68540347e-01 -2.15228528e-01 -7.46110603e-02 3.50058943e-01
-4.42970216e-01 5.34617156e-02 -5.14984965e-01 -8.92483771e-01
-1.43916216e-02 -6.28570676e-01 -1.24884218e-01 -5.59336245e-01
-8.39512289e-01 -4.25930172e-01 6.71294630e-01 -1.01810360e+00
1.77023089e+00 -1.81501949e+00 2.15678841e-01 2.21250504e-01
3.28566164e-01 -2.30851606e-03 -8.83415788e-02 7.82443345e-01
3.51285227e-02 5.12840986e-01 -3.37868392e-01 -4.41240162e-01
1.16619736e-01 -9.46321338e-02 -4.80521083e-01 -9.65554416e-02
4.86622483e-01 7.46782482e-01 -9.39524353e-01 -8.73350382e-01
-2.70609736e-01 5.55487275e-02 -6.22085631e-01 9.32142362e-02
-2.88961411e-01 -3.34160566e-01 -3.22273910e-01 3.45712364e-01
5.89741170e-01 5.12198880e-02 -6.82367533e-02 -2.46927857e-01
-3.17321301e-01 8.18424404e-01 -9.46103096e-01 1.32300425e+00
-1.70059130e-01 8.04329216e-01 -5.56078227e-03 -4.29442167e-01
6.39105499e-01 2.77079437e-02 2.38963023e-01 -5.78841567e-02
4.45322581e-02 1.43074542e-01 -1.01843607e-02 -4.71759230e-01
1.17040503e+00 -1.17078610e-01 -4.46085513e-01 3.97042841e-01
1.60865843e-01 -5.16894341e-01 9.82489645e-01 1.02352107e+00
1.44854224e+00 3.80645916e-02 3.30015928e-01 -2.49523655e-01
3.47537011e-01 3.42387408e-01 4.63627666e-01 8.58111739e-01
1.79154873e-01 6.46485329e-01 1.14773834e+00 6.95509687e-02
-9.99820530e-01 -8.53953600e-01 9.70389843e-02 8.98203969e-01
-5.73846221e-01 -1.07878828e+00 -7.72903383e-01 -1.24304473e+00
-6.22884445e-02 1.20669699e+00 -6.83003068e-01 5.01267053e-02
-7.22018063e-01 -4.35934901e-01 4.48830724e-01 6.77330136e-01
-1.11025147e-01 -1.01966977e+00 -5.30108809e-01 2.37510130e-01
-2.07508922e-01 -1.04591751e+00 -4.44467247e-01 5.72892785e-01
-8.87812555e-01 -9.57033575e-01 -5.61216712e-01 -4.91394967e-01
7.54412234e-01 -3.97909641e-01 1.30066192e+00 -4.97262785e-03
-1.21888295e-01 1.07560456e-01 -4.17418867e-01 -8.65204155e-01
-5.81764579e-01 6.81896269e-01 -5.60760796e-01 -5.90307593e-01
5.99034905e-01 -2.23598778e-01 -3.60216677e-01 -6.59829736e-01
-8.99531901e-01 1.37990341e-01 7.41782486e-01 7.08268583e-01
3.58838081e-01 -1.45963475e-01 6.45261168e-01 -1.29418230e+00
1.09760344e+00 -4.26928908e-01 -2.23512262e-01 1.66763738e-01
-5.34281492e-01 2.23532677e-01 6.51867449e-01 -1.24980301e-01
-1.00446987e+00 7.93316662e-02 -2.64197946e-01 2.44392425e-01
-6.04672246e-02 6.17307246e-01 -4.67667021e-02 7.67126262e-01
6.86203837e-01 -3.39951657e-04 -1.62773013e-01 -3.47659022e-01
2.61974543e-01 7.84528255e-01 5.83143950e-01 -3.40050936e-01
7.92279601e-01 -2.34399475e-02 -8.31196457e-02 -9.39332902e-01
-1.09188509e+00 -5.10063231e-01 -6.65649712e-01 -1.93715781e-01
5.19310951e-01 -6.99467123e-01 -1.73654824e-01 -1.56287670e-01
-1.20762372e+00 -1.45354167e-01 -9.04200137e-01 1.96516395e-01
-1.29210204e-01 4.37204987e-01 -7.79154837e-01 -8.37482035e-01
-8.20683360e-01 -6.91069841e-01 1.15202987e+00 3.92643690e-01
-1.10061717e+00 -7.44618297e-01 -7.41801187e-02 3.29782277e-01
3.66559401e-02 2.61123657e-01 1.00520349e+00 -1.14774942e+00
-2.88457662e-01 -5.40303528e-01 -9.01190862e-02 2.78694689e-01
-6.32958636e-02 4.41696167e-01 -8.29128921e-01 -1.89534947e-02
-1.84524894e-01 -3.17577094e-01 1.25838137e+00 2.40422204e-01
7.42883861e-01 -7.70087242e-01 -2.94281274e-01 5.93239767e-03
1.09654260e+00 -3.14965874e-01 5.10043383e-01 4.13645327e-01
5.67096889e-01 9.80538487e-01 6.76974833e-01 4.03937429e-01
2.42485046e-01 1.12040695e-02 3.23082507e-02 -7.04519004e-02
-1.27168894e-01 -2.66384184e-01 3.92169476e-01 7.31867552e-01
4.65760350e-01 -5.47232449e-01 -7.20166683e-01 8.01279128e-01
-1.60114264e+00 -1.05056822e+00 -3.96072865e-01 1.82004511e+00
1.22724843e+00 6.11738086e-01 1.74460143e-01 4.37225699e-01
2.50177264e-01 1.96030155e-01 -2.17601851e-01 -7.53405392e-01
-1.00504003e-01 2.23616332e-01 3.14289987e-01 6.49809599e-01
-1.03068233e+00 9.19047236e-01 6.67207336e+00 8.13641906e-01
-7.56572306e-01 -4.71337378e-01 4.03171927e-01 -6.86178446e-01
-6.23059273e-01 6.88280314e-02 -1.09722924e+00 3.04291040e-01
8.08758795e-01 -2.68602192e-01 -3.51350993e-01 6.98602080e-01
2.64938414e-01 -5.40667415e-01 -1.14832711e+00 4.14306790e-01
2.15185434e-01 -1.37592936e+00 3.14134836e-01 6.25676513e-02
6.16514683e-01 -1.27276063e-01 -1.57871738e-01 3.29216003e-01
3.57025802e-01 -7.45524764e-01 7.56982625e-01 2.93216556e-01
3.78379971e-01 -9.45355415e-01 8.29948664e-01 3.09145361e-01
-7.58523285e-01 4.81915139e-02 -1.09198101e-01 -4.28708941e-02
2.34437913e-01 8.81790698e-01 -1.17240119e+00 4.21830148e-01
3.84069741e-01 6.98746562e-01 -7.34623432e-01 9.30156112e-01
-5.32900751e-01 7.55545616e-01 -3.49305451e-01 -4.24722970e-01
3.93137753e-01 3.78100127e-02 9.36007440e-01 1.79381394e+00
6.37928024e-02 -2.49670465e-02 1.14705727e-01 7.88497448e-01
-3.98680359e-01 4.45622772e-01 -4.58251268e-01 -1.00004904e-01
3.89742017e-01 1.64984775e+00 -8.74111295e-01 -5.87156713e-01
-1.53857116e-02 7.29668617e-01 2.93494821e-01 1.89169124e-01
-2.27181539e-01 -8.60405564e-01 1.01753511e-01 3.84861454e-02
6.68292701e-01 1.37415633e-01 -7.50422060e-01 -1.00120294e+00
2.95116574e-01 -9.67713892e-01 5.82724810e-01 -4.05564040e-01
-1.11410034e+00 5.18731415e-01 -6.97863400e-02 -8.84243846e-01
-3.53819370e-01 -3.50792676e-01 -9.04330611e-01 8.27511787e-01
-1.27473438e+00 -8.95294905e-01 3.17978598e-02 -2.38939092e-01
1.05110061e+00 1.16992280e-01 5.65152526e-01 -1.61517426e-01
-6.56325996e-01 6.43333733e-01 -1.89335108e-01 2.62538016e-01
7.36528456e-01 -1.79556739e+00 4.09233928e-01 1.15870965e+00
-2.71862522e-02 7.67888129e-01 1.09431124e+00 -9.02438104e-01
-1.13832784e+00 -9.04585779e-01 1.38079524e+00 -7.01223135e-01
5.67554653e-01 -3.46850812e-01 -8.38427484e-01 5.54694831e-01
6.75727725e-01 -6.50409400e-01 1.09613228e+00 2.73437798e-01
-1.52412131e-01 2.10773423e-01 -1.02089286e+00 7.00534940e-01
5.00414133e-01 -1.81313649e-01 -1.08506668e+00 4.13210653e-02
5.64119458e-01 -4.51798141e-01 -7.93198824e-01 2.70289660e-01
4.69198287e-01 -4.95181262e-01 4.09094751e-01 -6.53814971e-01
1.16249132e+00 -9.65984836e-02 2.34484047e-01 -1.34756005e+00
-1.26641989e-01 -7.86808252e-01 -4.33799237e-01 1.67645204e+00
9.59340394e-01 1.32333472e-01 9.23489451e-01 8.00040424e-01
-1.71432257e-01 -9.66646671e-01 -4.57439125e-01 -3.82740706e-01
2.88387358e-01 -1.90478072e-01 2.37055570e-01 4.91206408e-01
6.05039656e-01 1.07146692e+00 1.15335584e-01 -4.10053045e-01
5.51870525e-01 9.81463492e-02 7.29434490e-01 -1.20141804e+00
-4.04359140e-02 -6.90250993e-01 2.46447131e-01 -1.11075008e+00
-3.20294648e-02 -8.88264716e-01 3.06819677e-01 -2.13440466e+00
5.12216449e-01 -5.20991832e-02 1.08677156e-01 4.83905077e-01
-6.75633609e-01 -4.57850955e-02 3.55146646e-01 6.11215979e-02
-6.88896894e-01 3.38058233e-01 8.74950409e-01 -1.08755611e-01
-4.88545537e-01 -1.97878987e-01 -1.19218063e+00 6.91853762e-01
6.10164583e-01 -5.63325107e-01 -2.52109140e-01 -1.02431849e-01
3.60067159e-01 -1.82090834e-01 -1.93090364e-01 -8.96939099e-01
2.30791360e-01 2.62463186e-02 6.90151274e-01 -1.23121285e+00
1.28201112e-01 -3.76635551e-01 -5.89442194e-01 2.60879159e-01
-9.08483982e-01 8.84369910e-02 3.56559813e-01 2.24747404e-01
-1.34331599e-01 -5.83714128e-01 3.95657301e-01 -1.38666347e-01
-7.75436983e-02 -2.25316271e-01 -6.34938002e-01 4.95744735e-01
5.60797691e-01 -1.98516563e-01 -4.23966020e-01 -3.23026657e-01
-5.71302772e-01 4.64713991e-01 2.93408662e-01 1.38502553e-01
4.66210425e-01 -7.62308836e-01 -1.02046406e+00 -1.61422029e-01
-5.96450344e-02 1.71355903e-01 -4.04479913e-02 8.84052634e-01
-2.98539937e-01 2.16453418e-01 1.01904504e-01 -2.08970964e-01
-1.62047911e+00 1.98274508e-01 -3.88821632e-01 -7.36905277e-01
-7.23884642e-01 7.48604596e-01 -2.41305888e-01 -1.04587935e-01
3.40147018e-01 -3.36156577e-01 -5.85655510e-01 7.58222938e-01
6.49663270e-01 5.71076512e-01 2.19967470e-01 -3.40665221e-01
-2.94422120e-01 2.86935177e-02 -7.28901505e-01 -3.93321693e-01
1.61881959e+00 4.58683819e-02 -1.90727666e-01 4.15508568e-01
1.15977383e+00 7.66700447e-01 -9.82340097e-01 6.59448057e-02
4.55421120e-01 -2.45848112e-02 2.01538838e-02 -9.97751296e-01
-4.96217906e-01 7.35777974e-01 -3.69391203e-01 3.81139785e-01
8.51604283e-01 1.80721626e-01 7.96391249e-01 3.06872338e-01
-6.64720118e-01 -1.47028315e+00 -3.94424424e-02 7.09264755e-01
1.04064488e+00 -1.02668178e+00 6.48417652e-01 -6.30198896e-01
-6.48404598e-01 1.28482223e+00 6.40104473e-01 -1.34038895e-01
9.28381234e-02 6.66048110e-01 -4.88232598e-02 -3.01638246e-01
-9.09180403e-01 3.26539092e-02 4.49680716e-01 1.56967208e-01
7.17944324e-01 -2.63317674e-01 -6.99985683e-01 8.84907782e-01
-4.79155242e-01 -3.46396208e-01 9.50854123e-01 1.21282756e+00
-5.12312531e-01 -1.12551188e+00 -1.95790604e-01 9.48796511e-01
-9.54604089e-01 -2.11267084e-01 -1.10230434e+00 6.13437831e-01
-3.75773072e-01 9.69340920e-01 1.13701604e-01 -5.80396242e-02
5.44136167e-01 2.54021883e-01 2.32212603e-01 -1.03802383e+00
-1.24394000e+00 4.69895899e-01 6.03431702e-01 -1.06948204e-01
-2.03378387e-02 -8.48109365e-01 -1.54727769e+00 6.38542324e-02
-2.92972833e-01 5.19325495e-01 5.33741593e-01 7.98142314e-01
4.32967901e-01 9.16392744e-01 2.75133967e-01 -6.62078798e-01
-6.04068160e-01 -1.14124107e+00 -1.61166668e-01 3.89875829e-01
4.02025670e-01 2.64129676e-02 -4.03053761e-01 -2.15538638e-03] | [12.511263847351074, 9.501860618591309] |
f26b6286-aaab-4cf3-a9a4-267700580ecc | online-decision-based-visual-tracking-via | null | null | http://proceedings.neurips.cc/paper/2020/hash/885b2c7a6deb4fea10f319c4ce993e02-Abstract.html | http://proceedings.neurips.cc/paper/2020/file/885b2c7a6deb4fea10f319c4ce993e02-Paper.pdf | Online Decision Based Visual Tracking via Reinforcement Learning | A deep visual tracker is typically based on either object detection or template matching while each of them is only suitable for a particular group of scenes. It is straightforward to consider fusing them together to pursue more reliable tracking. However, this is not wise as they follow different tracking principles. Unlike previous fusion-based methods, we propose a novel ensemble framework, named DTNet, with an online decision mechanism for visual tracking based on hierarchical reinforcement learning. The decision mechanism substantiates an intelligent switching strategy where the detection and the template trackers have to compete with each other to conduct tracking within different scenes that they are adept in. Besides, we present a novel detection tracker which avoids the common issue of incorrect proposal. Extensive results show that our DTNet achieves state-of-the-art tracking performance as well as good balance between accuracy and efficiency. The project website is available at https://vsislab.github.io/DTNet/. | ['Yibin Li', 'Ran Song', 'Wei zhang', 'Ke Song'] | 2020-12-01 | null | null | null | neurips-2020-12 | ['template-matching'] | ['computer-vision'] | [-3.85617405e-01 -4.63914186e-01 -1.41007498e-01 -1.51800811e-02
-4.07058239e-01 -6.01987839e-01 5.43793023e-01 -2.55005509e-01
-5.19702256e-01 6.44302189e-01 -2.21764639e-01 -7.33895674e-02
1.35384306e-01 -5.70024848e-01 -6.08431816e-01 -8.22816670e-01
1.79121181e-01 2.99085259e-01 8.31711590e-01 1.09833121e-01
6.57564700e-02 5.46397507e-01 -1.57413900e+00 -1.47950754e-01
8.25125754e-01 1.03433359e+00 4.66354638e-01 5.00977278e-01
-1.93881504e-02 5.61137736e-01 -6.23016596e-01 -4.70817029e-01
5.73842406e-01 -2.08277360e-01 -2.42679000e-01 2.38916203e-02
6.99155509e-01 -3.59434932e-01 -3.84649485e-01 1.16698301e+00
6.29276931e-01 7.45684430e-02 3.47442865e-01 -1.44099081e+00
-3.11049819e-01 2.48079166e-01 -7.08953202e-01 4.41604853e-01
1.88063830e-01 4.88932490e-01 8.11976790e-01 -7.42136836e-01
4.63722765e-01 1.00901020e+00 7.21200287e-01 6.83178723e-01
-1.01508343e+00 -8.74727130e-01 5.17358541e-01 1.72260299e-01
-1.26918411e+00 -4.73621756e-01 5.18484712e-01 -5.57828844e-01
4.62083489e-01 -2.82145035e-03 8.82507801e-01 1.00555801e+00
3.90190095e-01 8.01998675e-01 9.26392555e-01 -1.92479476e-01
8.69979486e-02 6.02674484e-03 -1.06071413e-01 6.42049670e-01
7.17707396e-01 5.53226173e-01 -4.63555723e-01 7.78451338e-02
9.81753886e-01 6.16005175e-02 -1.50264904e-01 -7.27030635e-01
-1.26239324e+00 6.29985988e-01 5.22261024e-01 2.66059160e-01
-4.88289744e-01 3.23369920e-01 2.17563510e-01 1.83205470e-01
6.05692677e-02 7.45553151e-02 -2.10967466e-01 2.36909881e-01
-9.89876688e-01 3.39199662e-01 4.13449228e-01 9.07287419e-01
5.89980423e-01 2.80185193e-01 -4.38184053e-01 3.58020455e-01
7.42787957e-01 5.46861231e-01 2.06411615e-01 -9.43629801e-01
-2.69185193e-02 5.88655233e-01 4.98693794e-01 -8.61372173e-01
-3.37787867e-01 -7.60024309e-01 -4.84756708e-01 7.41885483e-01
6.70119822e-01 -3.28425348e-01 -8.42281222e-01 1.72260404e+00
7.04200506e-01 4.83477712e-01 -3.12313765e-01 1.17058527e+00
7.60599852e-01 1.87022403e-01 2.75495380e-01 -2.84275025e-01
1.23657656e+00 -1.14567208e+00 -7.55852163e-01 -2.26288036e-01
2.96009660e-01 -9.71415579e-01 1.76044226e-01 2.86876440e-01
-1.04299033e+00 -9.12199199e-01 -1.08560729e+00 4.26100463e-01
-1.59278184e-01 3.90340030e-01 4.98703003e-01 6.33165717e-01
-1.11460602e+00 4.55479860e-01 -9.13676322e-01 -5.64966083e-01
2.68570691e-01 5.06891310e-01 -1.06577389e-01 3.59979212e-01
-8.06033254e-01 8.54872882e-01 3.75253081e-01 1.96674645e-01
-1.06346631e+00 -3.11069489e-01 -3.81313354e-01 -2.02688336e-01
6.08253300e-01 -9.34780240e-01 1.41136622e+00 -8.15708816e-01
-1.40723395e+00 5.97301602e-01 -1.91776622e-02 -5.56068480e-01
9.74311173e-01 -3.33869487e-01 -4.26173031e-01 -2.18546182e-01
2.74188668e-01 8.19838941e-01 1.07666028e+00 -1.24270535e+00
-1.04626250e+00 -2.17488274e-01 4.96974327e-02 2.30066523e-01
-7.83518776e-02 9.37328562e-02 -6.41672790e-01 -6.37394488e-01
9.63528361e-03 -9.09474373e-01 -1.87895507e-01 4.88114774e-01
-1.25146359e-01 -4.31418389e-01 9.17504549e-01 -2.18856394e-01
1.21184134e+00 -1.98999369e+00 -4.03943807e-02 -1.21557854e-01
4.15641457e-01 6.06429040e-01 5.21507263e-02 2.05743238e-01
3.16052407e-01 -3.97698164e-01 3.77696782e-01 -3.53621721e-01
-1.51133107e-03 -1.15753617e-02 -1.34701625e-01 6.68840706e-01
4.51821880e-03 7.14339256e-01 -9.78385687e-01 -7.62814224e-01
4.65188652e-01 3.34449619e-01 -2.36689478e-01 2.15966746e-01
-1.73601151e-01 6.83452785e-01 -6.12241387e-01 7.77659476e-01
8.75775993e-01 -3.76848698e-01 2.86733210e-01 -3.34537029e-01
-5.09243906e-01 -4.73096594e-02 -1.51317263e+00 1.43589580e+00
1.35337472e-01 3.06181371e-01 2.36302584e-01 -6.42911673e-01
8.33812892e-01 2.10017845e-01 6.22158289e-01 -5.45271933e-01
3.20296913e-01 8.12184811e-02 2.72592634e-01 -3.89619693e-02
5.44575095e-01 1.77393064e-01 1.42248660e-01 2.61589378e-01
2.29987904e-01 4.71964568e-01 4.20399964e-01 9.31997523e-02
1.01694715e+00 6.41760707e-01 2.61822701e-01 -1.65219963e-01
5.89460552e-01 1.19986534e-01 1.04159141e+00 1.08661461e+00
-8.54689300e-01 2.62824684e-01 -2.54157037e-01 -5.03402412e-01
-7.37681746e-01 -1.17256725e+00 1.53305344e-02 1.01345289e+00
6.36921227e-01 -2.85576105e-01 -4.84774053e-01 -7.45600462e-01
1.42520174e-01 1.98840842e-01 -5.22539198e-01 -3.51650119e-02
-4.60313350e-01 -2.96325654e-01 3.55507225e-01 5.61640263e-01
5.98097324e-01 -1.01771629e+00 -9.26808953e-01 3.84235591e-01
1.42424069e-02 -1.09870851e+00 -6.22440517e-01 9.29315314e-02
-7.48348355e-01 -1.16964412e+00 -6.92741632e-01 -5.14338136e-01
5.34744918e-01 8.21478248e-01 8.51360619e-01 4.24873888e-01
-1.16978809e-01 4.65826362e-01 -3.97323072e-01 -5.16113758e-01
-2.21996367e-01 -1.39029294e-01 3.98013204e-01 3.02262641e-02
2.91910201e-01 -1.93438977e-01 -8.75658631e-01 6.71917617e-01
-3.82329494e-01 -1.48454502e-01 5.47989011e-01 6.76086724e-01
4.66314405e-01 -2.14399964e-01 2.67958850e-01 -1.64001822e-01
8.60954374e-02 -1.76561773e-01 -1.17779934e+00 4.03345495e-01
-4.28544849e-01 -1.99994802e-01 3.41223657e-01 -6.08202219e-01
-1.06995213e+00 3.86623353e-01 2.14509085e-01 -9.37249243e-01
-4.97264694e-03 -2.28592396e-01 9.31469351e-02 -5.13094723e-01
5.05782604e-01 1.97680444e-01 9.27009340e-03 -3.49126041e-01
1.07194625e-01 1.63589582e-01 4.31785911e-01 -3.17354977e-01
1.05191004e+00 6.02504671e-01 -3.12364370e-01 -2.82490045e-01
-6.99668825e-01 -5.93055069e-01 -6.76068485e-01 -9.18152392e-01
8.17845345e-01 -1.07029212e+00 -1.04761493e+00 5.03036857e-01
-1.05856705e+00 -5.53343669e-02 -7.01240376e-02 7.13235021e-01
-3.87117416e-01 3.40376705e-01 -3.43172342e-01 -9.64318335e-01
-2.56041974e-01 -1.28059375e+00 9.95079756e-01 7.47931361e-01
1.32705852e-01 -7.88237870e-01 2.46014982e-01 7.90973380e-02
3.82903039e-01 -2.90422216e-02 -2.65901476e-01 -5.28499007e-01
-1.00064838e+00 -9.92010459e-02 -2.98537195e-01 -3.45451050e-02
8.49218294e-02 2.50996739e-01 -7.84224927e-01 -7.45766163e-01
-2.55715370e-01 1.36001175e-02 9.17790353e-01 6.30003095e-01
6.65228009e-01 1.72365353e-01 -7.82573104e-01 3.53114426e-01
1.49825513e+00 3.62445563e-01 3.65623862e-01 4.98889327e-01
5.59150398e-01 1.49202272e-01 9.57718551e-01 4.99615699e-01
3.30165356e-01 1.03732193e+00 6.50160193e-01 3.69489528e-02
-3.75004709e-01 -2.96468902e-02 5.46017945e-01 4.79325801e-01
-1.12324059e-01 -2.19683021e-01 -7.77200282e-01 5.04131496e-01
-2.20483613e+00 -1.06149089e+00 -9.19092298e-02 2.38217783e+00
3.55176717e-01 2.91388512e-01 4.86180305e-01 -4.87612784e-01
1.13488925e+00 2.53359992e-02 -5.85334837e-01 2.12286219e-01
1.13216087e-01 -3.29747021e-01 6.22708261e-01 1.78776100e-01
-1.26592708e+00 1.06984484e+00 6.27699518e+00 8.51955295e-01
-1.13566804e+00 2.26062909e-01 6.83199093e-02 -2.53235530e-02
2.66482532e-01 5.57870977e-02 -1.30580449e+00 6.70365810e-01
5.03097832e-01 -1.22698449e-01 9.11746100e-02 7.55554676e-01
5.47826365e-02 -2.97079921e-01 -6.90004170e-01 8.97056997e-01
-3.19787681e-01 -1.24779224e+00 -2.01382548e-01 4.48890068e-02
5.06498039e-01 2.66802132e-01 -1.88582495e-01 3.22460026e-01
7.77785301e-01 -4.73703057e-01 9.01550114e-01 7.27097631e-01
1.74863458e-01 -3.61528933e-01 4.92016494e-01 3.09241295e-01
-1.65603328e+00 -2.79756542e-02 -3.46274704e-01 1.06975652e-01
1.80100277e-01 2.86015928e-01 -4.06792641e-01 8.47312450e-01
9.10388112e-01 6.52971745e-01 -5.68142056e-01 1.73685420e+00
-1.43102258e-01 1.53664976e-01 -3.18385035e-01 3.65630747e-03
2.06098452e-01 -1.06016137e-01 8.28730226e-01 8.27442110e-01
4.34997410e-01 -8.15675855e-02 7.15485573e-01 6.06462836e-01
1.65647686e-01 -2.10431367e-01 -4.94309187e-01 3.36120754e-01
7.60227621e-01 1.63873732e+00 -9.47315753e-01 -4.11932379e-01
-5.64078212e-01 5.94319224e-01 4.11333174e-01 1.29845783e-01
-1.24739373e+00 1.45566508e-01 6.85089588e-01 -9.62692127e-02
8.49782407e-01 -2.22301364e-01 6.06857985e-02 -1.19041395e+00
-8.40300322e-02 -5.90988874e-01 4.61422473e-01 -6.46405220e-01
-1.20057738e+00 5.60875475e-01 -1.86217085e-01 -1.82248604e+00
8.52692798e-02 -4.30053174e-01 -6.45215809e-01 4.89184260e-01
-1.40049481e+00 -1.09876907e+00 -3.97401154e-01 5.75535238e-01
3.91812325e-01 -1.34748161e-01 2.18061835e-01 5.59016407e-01
-6.78323507e-01 5.56870639e-01 -6.64136857e-02 1.04328185e-01
9.28583026e-01 -1.07329226e+00 1.31749749e-01 1.10772693e+00
1.75793853e-03 4.07506257e-01 8.12537968e-01 -8.69451404e-01
-1.30453646e+00 -1.11977148e+00 4.67061460e-01 -3.89333576e-01
6.63290620e-01 -7.16614500e-02 -7.54082382e-01 5.69449782e-01
4.51493561e-01 2.39263669e-01 3.33754003e-01 -9.76328701e-02
-1.57364175e-01 -2.38054380e-01 -9.77906048e-01 6.57260478e-01
1.22266912e+00 1.07948616e-01 -4.39367384e-01 3.07297856e-01
4.27351564e-01 -5.88515162e-01 -7.33467817e-01 5.91516137e-01
6.58291459e-01 -1.05673218e+00 8.96444738e-01 -1.31345376e-01
-4.14149016e-01 -1.09020019e+00 3.48627605e-02 -1.00754368e+00
-7.25309789e-01 -5.69649935e-01 -3.23610544e-01 1.07342184e+00
2.85050478e-02 -6.43391550e-01 8.26650262e-01 1.35949448e-01
-3.47119346e-02 -4.75996971e-01 -1.02632093e+00 -1.23005068e+00
-4.11096960e-01 -1.74797997e-01 3.80711675e-01 6.97613657e-01
-5.46137691e-01 1.07240669e-01 -5.19509554e-01 4.00709838e-01
1.07281196e+00 2.22262040e-01 1.02376759e+00 -1.39262581e+00
-2.50978768e-01 -5.93189597e-01 -3.91836107e-01 -1.06232297e+00
-1.05623402e-01 -6.14680648e-01 1.71888277e-01 -1.45675349e+00
1.19077109e-01 -4.47798908e-01 -5.44236779e-01 6.12824500e-01
-1.76765174e-01 2.73077011e-01 6.54345512e-01 4.71526951e-01
-1.28013766e+00 5.19553483e-01 1.28465080e+00 -5.77584505e-02
-1.80085018e-01 1.70053348e-01 -5.01219749e-01 5.06612659e-01
8.76049578e-01 -5.66760957e-01 5.64812198e-02 -3.07940513e-01
-2.78773248e-01 1.07273810e-01 5.90294778e-01 -1.46418965e+00
8.30072701e-01 7.27641582e-02 6.92852199e-01 -8.66850853e-01
2.18182757e-01 -8.74026895e-01 4.05983627e-01 7.55688369e-01
1.96061283e-01 1.74714729e-01 3.73801708e-01 7.09374666e-01
3.34752612e-02 -2.39701513e-02 9.07958031e-01 -8.21973532e-02
-9.93490219e-01 5.04252434e-01 -3.85168463e-01 -3.16868842e-01
1.39684510e+00 -5.45420945e-01 -3.75103980e-01 -1.83587641e-01
-7.54702687e-01 6.57917082e-01 5.06901145e-01 6.38432682e-01
3.46942157e-01 -1.41951084e+00 -5.99134862e-01 -1.10822268e-01
4.01630159e-03 -4.63437825e-01 2.54404038e-01 1.15285921e+00
-5.97920716e-02 3.28887463e-01 -4.59946215e-01 -9.68566179e-01
-1.43362832e+00 6.06617987e-01 5.70754349e-01 -3.36327493e-01
-5.44585347e-01 6.86215281e-01 3.14441890e-01 -2.77140677e-01
3.84573281e-01 9.60692540e-02 -2.68639535e-01 4.43453938e-02
5.65602601e-01 2.81165630e-01 -1.37194663e-01 -6.76092386e-01
-6.54610217e-01 6.25227690e-01 -3.16537708e-01 1.67006224e-01
8.95253778e-01 -2.45579198e-01 2.21991256e-01 1.22793689e-01
3.47064257e-01 -2.26113975e-01 -1.76412833e+00 -4.00700837e-01
-9.47441235e-02 -7.09636867e-01 1.22478187e-01 -8.00209939e-01
-1.24411654e+00 2.93832690e-01 9.73040640e-01 2.26811215e-01
1.02068341e+00 -1.74474031e-01 4.69329357e-01 9.24600437e-02
6.61990166e-01 -1.07780886e+00 1.00896537e-01 4.91548121e-01
5.79498291e-01 -1.35218608e+00 1.99607655e-01 -1.76898763e-01
-4.08905178e-01 1.00981188e+00 1.11602187e+00 -3.67941231e-01
4.96540576e-01 2.89199173e-01 1.26914605e-01 -8.26972723e-02
-7.87130773e-01 -7.86742687e-01 3.14788043e-01 5.26766539e-01
2.17369035e-01 -1.67826310e-01 -3.79709184e-01 5.22009023e-02
3.34827781e-01 9.60993990e-02 1.23150647e-01 9.92438197e-01
-6.41211331e-01 -1.22521925e+00 -6.84402287e-01 2.89948374e-01
-3.57430935e-01 3.03061604e-01 -2.09947407e-01 9.86792207e-01
2.21301422e-01 9.33503270e-01 4.52048667e-02 -3.86850178e-01
1.99376181e-01 -3.57008249e-01 7.51711249e-01 -2.94916421e-01
-7.85935223e-01 4.56677318e-01 -1.73069596e-01 -6.66328609e-01
-8.38054359e-01 -8.40387225e-01 -9.34146225e-01 -3.70943993e-01
-5.84769487e-01 -1.17519991e-02 4.47699398e-01 8.63100111e-01
3.84727329e-01 6.85061872e-01 5.09559631e-01 -1.10206783e+00
-3.83851737e-01 -6.69419289e-01 -3.77830267e-01 -2.28888225e-02
3.31794202e-01 -1.23717713e+00 -1.50598935e-03 -2.62341380e-01] | [6.392798900604248, -2.0803847312927246] |
4cadd82d-3145-40fd-9a97-3d35c8828551 | diagnosing-model-performance-under | 2303.02011 | null | https://arxiv.org/abs/2303.02011v4 | https://arxiv.org/pdf/2303.02011v4.pdf | Diagnosing Model Performance Under Distribution Shift | Prediction models can perform poorly when deployed to target distributions different from the training distribution. To understand these operational failure modes, we develop a method, called DIstribution Shift DEcomposition (DISDE), to attribute a drop in performance to different types of distribution shifts. Our approach decomposes the performance drop into terms for 1) an increase in harder but frequently seen examples from training, 2) changes in the relationship between features and outcomes, and 3) poor performance on examples infrequent or unseen during training. These terms are defined by fixing a distribution on $X$ while varying the conditional distribution of $Y \mid X$ between training and target, or by fixing the conditional distribution of $Y \mid X$ while varying the distribution on $X$. In order to do this, we define a hypothetical distribution on $X$ consisting of values common in both training and target, over which it is easy to compare $Y \mid X$ and thus predictive performance. We estimate performance on this hypothetical distribution via reweighting methods. Empirically, we show how our method can 1) inform potential modeling improvements across distribution shifts for employment prediction on tabular census data, and 2) help to explain why certain domain adaptation methods fail to improve model performance for satellite image classification. | ['Tiffany Tianhui Cai', 'Steve Yadlowsky', 'Hongseok Namkoong'] | 2023-03-03 | null | null | null | null | ['satellite-image-classification'] | ['computer-vision'] | [ 2.48691380e-01 -7.67888352e-02 -4.06400442e-01 -6.18805647e-01
-7.59653330e-01 -4.43918735e-01 5.75129807e-01 3.91090751e-01
-6.11770511e-01 8.84074152e-01 3.07657897e-01 -6.42153740e-01
-5.94223976e-01 -1.12706304e+00 -7.40340233e-01 -5.99474311e-01
-2.08712935e-01 5.97717822e-01 -4.31781299e-02 -8.71056989e-02
4.76335995e-02 3.23421896e-01 -1.53849638e+00 2.43424281e-01
8.44013572e-01 6.63628757e-01 1.09080419e-01 3.68362367e-01
6.99452385e-02 3.81291956e-01 -7.95001149e-01 -2.94399768e-01
3.70614171e-01 -2.93857783e-01 -5.32244802e-01 3.17980982e-02
3.99907678e-01 -2.53382295e-01 -1.07955679e-01 8.75301600e-01
3.81644666e-01 4.59999144e-01 1.12252450e+00 -1.24228644e+00
-5.78722656e-01 6.88529134e-01 -7.09895432e-01 5.98366976e-01
1.57107741e-01 2.77994543e-01 1.00635707e+00 -5.61012089e-01
5.36054850e-01 1.16438580e+00 9.81269479e-01 2.80239642e-01
-1.85234439e+00 -8.15604508e-01 4.46668565e-01 -8.00931901e-02
-1.39605892e+00 -4.52138871e-01 5.15955508e-01 -7.70311236e-01
8.30853403e-01 2.88149953e-01 3.04006875e-01 8.70499253e-01
-2.66439002e-03 4.56770837e-01 1.01747823e+00 -4.37748224e-01
2.40218133e-01 3.76921207e-01 3.41747195e-01 2.29435898e-02
3.18783045e-01 3.29590559e-01 -2.82697707e-01 -4.72992957e-01
2.70056993e-01 1.59517556e-01 -3.33109677e-01 -2.18185514e-01
-7.62354732e-01 9.49617386e-01 3.76885056e-01 3.72263081e-02
-3.47786844e-01 9.66903195e-02 2.18621954e-01 3.23980987e-01
6.77430511e-01 4.03464645e-01 -7.61911094e-01 -1.48004517e-01
-8.87462556e-01 6.55234396e-01 4.30132538e-01 4.88134295e-01
9.14141893e-01 -3.77858765e-02 -2.29074210e-01 1.21417308e+00
-8.88252482e-02 6.27842009e-01 4.92505312e-01 -7.55338550e-01
8.70749652e-01 5.47742128e-01 3.55041146e-01 -9.24957752e-01
-5.04171968e-01 -4.63531077e-01 -5.02948225e-01 9.66969412e-03
7.54098594e-01 -3.90966028e-01 -1.06016254e+00 2.15265989e+00
-1.45973593e-01 -1.08059369e-01 -1.71919227e-01 4.12308931e-01
2.14003071e-01 5.60035586e-01 4.25262451e-01 -2.53720731e-01
1.05601823e+00 -1.32779881e-01 -1.82198182e-01 -5.99481344e-01
9.51674044e-01 -3.66799414e-01 1.40550864e+00 -6.93338318e-03
-7.54475176e-01 -5.64527869e-01 -7.05826283e-01 5.58107734e-01
-3.88332099e-01 -4.91850413e-02 4.57576752e-01 8.65945041e-01
-6.99868083e-01 9.18757915e-01 -7.65922844e-01 -3.60563874e-01
4.11694676e-01 3.56758952e-01 -9.75608230e-02 -1.94350049e-01
-1.14967632e+00 9.06714201e-01 2.16372445e-01 -5.94158769e-01
-4.04046774e-01 -9.43028390e-01 -8.81776989e-01 5.52064657e-01
8.25785026e-02 -4.59574789e-01 1.02169573e+00 -1.05467176e+00
-4.81179029e-01 6.84625506e-01 -1.53286919e-01 -5.29291630e-01
1.57310352e-01 1.02051631e-01 -5.08068442e-01 -5.13399005e-01
3.41377079e-01 4.28120643e-01 6.88181460e-01 -1.21932340e+00
-8.58215988e-01 -6.64026082e-01 -2.36625597e-01 2.43001297e-01
-4.37561750e-01 -2.32480094e-01 1.15185864e-01 -6.79135084e-01
-8.51881057e-02 -9.72958922e-01 -1.63714364e-01 -6.48224652e-01
-7.82565922e-02 -3.14325690e-01 3.34697992e-01 -5.79117417e-01
1.59111536e+00 -2.20925450e+00 -2.31499612e-01 7.08976448e-01
3.75449173e-02 3.99302095e-02 -1.27181232e-01 2.96992064e-01
-4.45937753e-01 2.58048385e-01 -3.60822052e-01 -4.74118441e-03
4.68724184e-02 2.16825247e-01 -5.04329324e-01 1.59282461e-01
2.20230758e-01 3.69193137e-01 -7.06395209e-01 1.66850626e-01
-2.34390190e-03 -2.73634400e-02 -6.88973725e-01 -2.80626982e-01
-1.40508292e-02 -4.49624434e-02 -1.57436073e-01 4.61296678e-01
6.50170743e-01 -2.63622832e-02 3.19113642e-01 2.73507893e-01
-1.05797462e-01 3.96433264e-01 -1.13371551e+00 8.07694912e-01
-1.07929885e-01 3.88599157e-01 -2.00035632e-01 -1.33348668e+00
1.05778849e+00 -1.61403269e-01 4.67585117e-01 -5.92457116e-01
-2.72396713e-01 5.85102402e-02 2.67821223e-01 -1.43555611e-01
4.87147957e-01 -5.64924240e-01 -1.76961556e-01 5.61919749e-01
-2.85707861e-01 -2.03499198e-02 1.24001898e-01 -1.61133617e-01
1.29012656e+00 -2.31289715e-01 5.59637025e-02 -3.45681578e-01
-1.04022324e-01 1.15062624e-01 6.64813578e-01 9.83430743e-01
-2.13578716e-01 3.10476691e-01 7.99254298e-01 -3.66932660e-01
-1.02057707e+00 -1.31503057e+00 -6.18350089e-01 1.48925066e+00
-2.65037954e-01 -2.53835589e-01 -4.63382989e-01 -8.16969156e-01
6.32704854e-01 1.13512838e+00 -8.12282503e-01 -4.07495648e-01
-2.84431577e-01 -1.32211232e+00 4.17582750e-01 7.07406819e-01
2.70976633e-01 -8.13181460e-01 -4.27198231e-01 -4.98661064e-02
-1.17425196e-01 -3.36520642e-01 -3.27345312e-01 5.34136117e-01
-8.46264124e-01 -8.46266627e-01 -8.69455993e-01 -3.20381880e-01
6.21789277e-01 7.65829459e-02 1.48137045e+00 -5.72641089e-04
5.84231690e-02 3.43786091e-01 -2.22932220e-01 -6.61327302e-01
-2.48973593e-01 4.18672979e-01 1.83371723e-01 -4.53651309e-01
7.19932258e-01 -5.98374248e-01 -5.29180944e-01 3.37357849e-01
-8.48787367e-01 -5.32828510e-01 3.11768711e-01 8.84148359e-01
4.30551350e-01 2.50422120e-01 7.67463744e-01 -1.21704829e+00
6.88371241e-01 -9.92419541e-01 -3.04357946e-01 2.02685744e-01
-1.05056143e+00 1.71748966e-01 2.47701153e-01 -6.47271156e-01
-8.40924621e-01 -8.33478719e-02 -1.68447301e-01 -2.45363772e-01
-1.91413462e-01 8.72215807e-01 1.87796414e-01 5.70006132e-01
1.07642281e+00 3.17793489e-02 4.13255394e-02 -5.20396888e-01
2.93363091e-02 7.17649698e-01 3.27973723e-01 -8.00898612e-01
6.79654002e-01 1.28085718e-01 -3.85359347e-01 -4.23566848e-01
-6.26228273e-01 -1.99967459e-01 -2.59590745e-01 1.78592324e-01
4.00586694e-01 -8.80406022e-01 -3.28832179e-01 4.17516455e-02
-3.36240172e-01 -8.66156280e-01 -5.34369409e-01 5.65537810e-01
-4.14848149e-01 -8.59432854e-03 7.12119713e-02 -8.16403151e-01
1.23565912e-01 -8.69972646e-01 7.51645565e-01 3.69828790e-02
-5.38785934e-01 -1.04690516e+00 1.82439521e-01 4.28002886e-02
4.68206704e-01 1.67572930e-01 1.34549797e+00 -7.07508922e-01
1.02731094e-01 -2.25269079e-01 -2.22228900e-01 4.11527038e-01
3.76424462e-01 8.69662222e-03 -6.27916813e-01 -5.39706588e-01
-4.76209879e-01 -1.28638878e-01 1.09015822e+00 7.40578830e-01
1.32246220e+00 -1.14358135e-01 -6.60509884e-01 4.22793746e-01
1.11778104e+00 3.36241275e-01 6.68463051e-01 4.26297158e-01
9.32239443e-02 7.85351455e-01 3.75714511e-01 4.98003036e-01
3.81740332e-01 6.94974005e-01 -3.85259651e-03 -1.37252331e-01
1.97838008e-01 -2.32143492e-01 2.32193053e-01 -1.68242201e-01
5.01815677e-02 -2.68181935e-02 -1.36401689e+00 8.36700737e-01
-1.53965485e+00 -1.19157803e+00 3.06717277e-01 2.54804683e+00
7.84024954e-01 5.63969195e-01 4.46581662e-01 7.82329142e-02
7.07621336e-01 1.58992093e-02 -7.25816905e-01 -2.66770571e-01
2.22138673e-01 2.79505491e-01 6.86579287e-01 5.53485692e-01
-8.33961666e-01 4.78577793e-01 7.23100281e+00 9.34810519e-01
-9.92268026e-01 -1.38095915e-01 1.06172311e+00 -3.16960871e-01
-4.27771866e-01 -1.25770262e-02 -8.79631639e-01 8.64267409e-01
1.00363243e+00 -3.32683295e-01 2.76925832e-01 1.00362778e+00
5.42932041e-02 -2.38746285e-01 -1.10410094e+00 7.28591502e-01
-1.76202372e-01 -1.08793867e+00 3.22089866e-02 2.78399140e-01
6.02710247e-01 -1.66915916e-02 3.68788511e-01 7.58308649e-01
6.75246477e-01 -1.05557477e+00 5.18143654e-01 2.59912580e-01
9.93615091e-01 -9.79808509e-01 6.67030275e-01 5.99137604e-01
-8.81080389e-01 -5.06781757e-01 -3.53758752e-01 -4.45170671e-01
-3.05594891e-01 6.14700854e-01 -8.52474928e-01 1.33649424e-01
8.47597718e-01 4.83214021e-01 -5.95817626e-01 6.61605716e-01
4.25153345e-01 8.73309910e-01 -4.25080687e-01 3.61764312e-01
-3.53634953e-02 5.57235703e-02 1.90185577e-01 1.12568426e+00
5.19272149e-01 1.10621899e-02 6.40662983e-02 7.62234747e-01
2.63162721e-02 -2.56257087e-01 -8.81344676e-01 2.23311990e-01
7.06344545e-01 6.40462875e-01 -5.23172677e-01 -3.16820443e-01
-3.55747372e-01 2.56387681e-01 2.22798571e-01 6.57140076e-01
-7.55635500e-01 -4.17139858e-01 9.31458116e-01 6.40953183e-01
1.81410059e-01 8.54469389e-02 -4.10949677e-01 -9.63859558e-01
-1.69806167e-01 -7.21968889e-01 7.57292867e-01 -6.34327769e-01
-1.45322764e+00 2.87344515e-01 6.30669892e-01 -1.01148522e+00
-4.06159699e-01 -4.24259871e-01 -5.17902851e-01 1.17576540e+00
-1.24233639e+00 -4.41728801e-01 -2.06516702e-02 4.51194644e-01
2.97835529e-01 -1.91261798e-01 7.27445483e-01 2.90911913e-01
-5.92396975e-01 8.80736172e-01 3.79807293e-01 5.16034067e-02
7.29170561e-01 -1.27192485e+00 2.40885958e-01 4.97366548e-01
-9.12154466e-02 6.27982497e-01 6.60835922e-01 -5.69023311e-01
-2.39002585e-01 -1.13213563e+00 9.42762673e-01 -7.54728734e-01
1.74297869e-01 -4.34219744e-03 -9.21349585e-01 1.03662336e+00
-3.78782749e-01 -3.83749664e-01 9.14556086e-01 7.76076376e-01
-3.21839154e-01 -3.18593144e-01 -1.42977881e+00 4.70298409e-01
8.75375628e-01 -2.41212308e-01 -5.04531264e-01 2.14338943e-01
2.84724027e-01 -1.51091710e-01 -9.12227929e-01 5.07789135e-01
5.46141863e-01 -1.00360000e+00 1.00085580e+00 -8.07789683e-01
4.56797481e-01 -2.85145524e-03 -5.18068552e-01 -1.57142031e+00
-5.32065809e-01 -3.62114422e-02 2.79995888e-01 1.24890852e+00
7.53626466e-01 -8.75868797e-01 7.16726720e-01 1.02295768e+00
-1.43053737e-02 -8.12684715e-01 -9.59847867e-01 -7.25159705e-01
5.89979708e-01 -5.52369833e-01 9.28591847e-01 1.09319651e+00
-6.44424334e-02 1.68473169e-01 1.18400283e-01 -1.78345721e-02
2.22054526e-01 4.62239124e-02 6.35539234e-01 -1.25794935e+00
-3.65917414e-01 -6.15599155e-01 -1.48004830e-01 -8.38395298e-01
8.12050030e-02 -7.98192799e-01 -1.33866996e-01 -1.17361760e+00
4.50677812e-01 -9.76509631e-01 -3.87605548e-01 7.03948438e-01
-3.41528654e-01 1.22484090e-02 1.45514607e-01 4.35150027e-01
1.86255127e-02 2.93487042e-01 5.86090326e-01 -8.37006420e-02
-3.86496186e-01 1.86310381e-01 -9.84364331e-01 7.21093416e-01
7.15341985e-01 -6.52261853e-01 -5.75208426e-01 -4.72645253e-01
2.09596545e-01 6.97952975e-03 2.65328526e-01 -8.25690150e-01
-5.25666595e-01 -5.36757171e-01 7.89409459e-01 -4.32887286e-01
1.15975387e-01 -6.76938415e-01 1.55777872e-01 4.86450821e-01
-5.63916802e-01 1.63736537e-01 3.65843713e-01 4.34858441e-01
1.11406945e-01 -1.48842782e-01 7.90191054e-01 -3.47516611e-02
-2.78123498e-01 -1.48929983e-01 -3.61532897e-01 4.67467718e-02
9.12756264e-01 -3.02200288e-01 -2.29634315e-01 -3.95247281e-01
-9.80496168e-01 2.89214581e-01 5.49490213e-01 2.53273398e-01
9.50074047e-02 -1.34414375e+00 -7.10214674e-01 4.30400729e-01
1.77465990e-01 -2.78779507e-01 1.28915533e-01 5.83286881e-01
-3.26843723e-03 1.60364360e-01 -4.76626456e-02 -4.84146535e-01
-1.00480008e+00 2.78739035e-01 5.16852438e-01 -3.60623181e-01
-1.55729994e-01 9.00663614e-01 2.62079805e-01 -5.44107735e-01
1.01627372e-01 -4.88684326e-01 -2.35894099e-02 2.72497237e-01
3.21074247e-01 4.63503689e-01 -2.21046922e-03 -3.11041623e-01
-3.86561185e-01 2.78168380e-01 -9.43065509e-02 -1.82054028e-01
1.25313616e+00 -7.56164640e-02 3.68705451e-01 4.27631706e-01
7.86780477e-01 -4.90565300e-02 -1.33922505e+00 -1.63451180e-01
-2.42236014e-02 -7.24031091e-01 -1.37065381e-01 -9.33493316e-01
-7.65272379e-01 4.84399050e-01 8.04281890e-01 5.18475652e-01
1.15190876e+00 4.80171442e-02 3.89649212e-01 1.43597275e-01
6.35646805e-02 -1.11839342e+00 -1.91617072e-01 2.88743734e-01
7.25640178e-01 -1.21991396e+00 1.53475747e-01 -5.98931406e-03
-7.14096129e-01 6.06220305e-01 5.45918882e-01 -1.14352042e-02
7.75897384e-01 7.60512799e-02 -1.32465318e-01 -4.97899726e-02
-8.06186855e-01 -2.56636411e-01 1.93227500e-01 7.66787171e-01
4.44814056e-01 2.71326393e-01 -3.15661103e-01 8.80589664e-01
-3.28517795e-01 -9.47826430e-02 1.14796840e-01 6.73336148e-01
-3.94484222e-01 -1.03989971e+00 -5.98180115e-01 1.22374260e+00
-3.89777273e-01 -1.83511406e-01 -2.08200917e-01 1.03178108e+00
5.34471035e-01 7.19433069e-01 7.31864870e-01 -4.95154619e-01
5.94927549e-01 6.02203965e-01 1.67096183e-01 -8.47081482e-01
-3.65858316e-01 -1.44995704e-01 1.86312571e-01 -5.12655415e-02
-2.06755884e-02 -1.00316250e+00 -8.80820334e-01 -5.85071862e-01
-1.76553816e-01 6.04437962e-02 2.45060340e-01 7.55599320e-01
3.60677898e-01 3.43392521e-01 6.36414051e-01 -7.40765393e-01
-8.89238119e-01 -1.15907371e+00 -9.11457598e-01 7.41798699e-01
2.75757730e-01 -8.95851433e-01 -5.28113425e-01 -1.81626398e-02] | [8.61192798614502, 4.6730875968933105] |
6ea10ba5-fcc9-4b16-9a0f-e0a4b05642b1 | ac-band-a-combinatorial-bandit-based-approach | 2212.00333 | null | https://arxiv.org/abs/2212.00333v1 | https://arxiv.org/pdf/2212.00333v1.pdf | AC-Band: A Combinatorial Bandit-Based Approach to Algorithm Configuration | We study the algorithm configuration (AC) problem, in which one seeks to find an optimal parameter configuration of a given target algorithm in an automated way. Recently, there has been significant progress in designing AC approaches that satisfy strong theoretical guarantees. However, a significant gap still remains between the practical performance of these approaches and state-of-the-art heuristic methods. To this end, we introduce AC-Band, a general approach for the AC problem based on multi-armed bandits that provides theoretical guarantees while exhibiting strong practical performance. We show that AC-Band requires significantly less computation time than other AC approaches providing theoretical guarantees while still yielding high-quality configurations. | ['Kevin Tierney', 'Eyke Hüllermeier', 'Björn Haddenhorst', 'Viktor Bengs', 'Elias Schede', 'Jasmin Brandt'] | 2022-12-01 | null | null | null | null | ['multi-armed-bandits'] | ['miscellaneous'] | [-8.03978592e-02 -1.97660252e-01 -9.84190762e-01 -6.85165869e-03
-1.34662783e+00 -1.15214217e+00 2.04835325e-01 -5.06814010e-02
-1.55276284e-01 1.19575453e+00 -1.42514125e-01 -9.33311760e-01
-7.40278363e-01 -6.60753489e-01 -7.14811981e-01 -9.16887224e-01
-1.40670806e-01 9.77391720e-01 3.38491388e-02 2.51526479e-02
3.52895647e-01 6.61285400e-01 -1.02744091e+00 8.65546092e-02
6.78312004e-01 1.22161186e+00 -7.80638680e-02 6.66430593e-01
-1.17694370e-01 1.04069911e-01 -3.28764766e-01 -5.99270165e-01
5.04786611e-01 -6.33497775e-01 -1.13503718e+00 3.34097952e-01
2.29609907e-01 3.07954811e-02 -2.13808436e-02 1.06176531e+00
1.75083399e-01 2.36132089e-02 2.45335028e-01 -1.31609905e+00
-2.89912879e-01 1.15825868e+00 -1.01839626e+00 3.07623565e-01
1.67508304e-01 6.21475689e-02 1.57944024e+00 -1.91624925e-01
3.88209730e-01 1.18399978e+00 3.51852089e-01 1.81784406e-01
-1.71805012e+00 -3.62074465e-01 3.53906125e-01 1.75680384e-01
-1.28771806e+00 -3.46121341e-01 5.84859550e-01 1.40993744e-01
7.64901221e-01 7.51321912e-01 9.64976311e-01 5.90745151e-01
-3.93124431e-01 1.26132202e+00 1.04859746e+00 -7.52226472e-01
4.16694582e-01 6.84524840e-03 2.19345480e-01 4.48721468e-01
7.17013836e-01 2.01611906e-01 -4.14040089e-01 -7.25399792e-01
6.43319666e-01 -4.26789373e-01 -3.76509428e-01 -7.52178371e-01
-1.14149356e+00 9.39205468e-01 1.41109511e-01 2.34898463e-01
-4.19393867e-01 5.91764510e-01 1.26152247e-01 2.64689595e-01
3.88227880e-01 6.97620213e-01 -5.42021155e-01 -3.37422401e-01
-1.21485174e+00 5.95382214e-01 8.72763634e-01 1.15861034e+00
4.65033084e-01 -1.23842470e-01 -2.81759739e-01 5.95890105e-01
-7.75080323e-02 5.09698212e-01 -1.50431797e-01 -1.04291904e+00
5.44736385e-01 5.41729368e-02 7.49650419e-01 -6.51682019e-01
-2.65798301e-01 -1.01380289e+00 -3.91378045e-01 -3.60763103e-01
3.65875006e-01 -7.17586800e-02 -3.92377347e-01 1.77161074e+00
5.49256980e-01 -2.34239995e-02 -6.40690252e-02 9.09560323e-01
-4.42651361e-01 8.07917118e-01 -6.07220352e-01 -9.77418840e-01
9.42651331e-01 -1.50117564e+00 -7.20753789e-01 -9.67453271e-02
4.80221093e-01 -8.95582318e-01 9.20264006e-01 6.73541963e-01
-1.63278329e+00 3.19376320e-01 -9.25330818e-01 6.49465859e-01
1.50676519e-01 -1.72792062e-01 1.08912480e+00 1.06112468e+00
-1.09282172e+00 2.31791496e-01 -6.97710156e-01 -9.15040970e-02
3.91417772e-01 4.13223475e-01 1.82384208e-01 -1.94182783e-01
-3.28341335e-01 5.81844926e-01 6.04020655e-01 -9.85952094e-02
-5.41376710e-01 -7.84403682e-01 -2.84918249e-01 3.75838339e-01
1.42886305e+00 -7.13384926e-01 1.84114599e+00 -1.00675559e+00
-1.59302652e+00 3.38730991e-01 -7.34160468e-02 -6.79851294e-01
6.03021741e-01 -2.01018944e-01 -4.91078198e-02 1.74600542e-01
-1.63465708e-01 7.79306889e-02 4.51774538e-01 -1.41423416e+00
-1.07659709e+00 -1.04402177e-01 2.67299473e-01 3.20974402e-02
-4.25756395e-01 -1.79279253e-01 -7.08394051e-01 -4.88222599e-01
1.04046725e-01 -1.30678499e+00 -4.90124673e-01 -3.69167119e-01
-6.59482598e-01 -3.45789939e-01 2.60472149e-01 1.49576351e-01
1.66334248e+00 -1.88128281e+00 3.41263592e-01 7.18517065e-01
-1.96780428e-01 1.04082637e-01 -3.06875914e-01 6.11838698e-01
3.93109441e-01 3.68147939e-01 -7.46801272e-02 -8.16123262e-02
1.63945660e-01 4.14707959e-01 -4.27505940e-01 7.39156008e-01
-5.23028791e-01 7.49552488e-01 -9.24026847e-01 -4.73277748e-01
-2.14147374e-01 -4.94785249e-01 -1.02656257e+00 6.34378567e-02
-7.73272991e-01 -2.33871430e-01 -6.12890005e-01 8.69409859e-01
5.94826281e-01 -4.51575965e-01 6.81881130e-01 2.84541965e-01
-1.91810921e-01 5.10725453e-02 -1.20703876e+00 1.44230282e+00
-3.71800214e-01 1.95188075e-01 6.28334880e-01 -1.28358388e+00
2.92989910e-01 4.14371043e-02 6.72947109e-01 -3.09035122e-01
1.37379393e-01 5.77715278e-01 -9.14130583e-02 -6.99969307e-02
5.93007326e-01 1.43309399e-01 -9.83764604e-02 7.52474606e-01
-5.10854840e-01 -7.37619549e-02 5.41980445e-01 1.34009212e-01
8.98651004e-01 -1.48241922e-01 3.93725395e-01 -6.21153414e-01
7.68844709e-02 3.34043920e-01 5.43247283e-01 1.24023247e+00
3.76203470e-02 9.85328406e-02 5.10071456e-01 -4.34621960e-01
-1.13308024e+00 -8.49267960e-01 -1.05224745e-02 9.88478243e-01
3.60093713e-01 -3.60161901e-01 -8.22659969e-01 -4.15557027e-01
3.10559034e-01 7.56955266e-01 -4.75382358e-01 3.18968505e-01
-4.86821413e-01 -5.80789030e-01 2.39118889e-01 2.29529038e-01
4.28359121e-01 -3.10065895e-01 -7.51470089e-01 6.14016294e-01
-3.90514463e-01 -1.15469110e+00 -8.68750393e-01 1.15122266e-01
-9.37997341e-01 -1.06555927e+00 -3.04658562e-01 -3.09283465e-01
5.23680031e-01 8.86266053e-01 1.24362087e+00 3.20907474e-01
-7.80446008e-02 3.56304288e-01 -4.30326134e-01 -1.93389654e-01
-6.50218129e-02 5.70437729e-01 -3.04877870e-02 3.42614629e-04
-1.07348055e-01 -3.95963699e-01 -5.69859564e-01 5.29866517e-01
-9.64989960e-01 9.93423834e-02 7.34201133e-01 9.66804862e-01
9.40500498e-01 2.18301520e-01 4.79104400e-01 -9.36575770e-01
7.18311071e-01 -4.18495685e-01 -1.21743321e+00 7.28973746e-01
-9.83960032e-01 1.74290895e-01 4.62036610e-01 -3.33927602e-01
-7.68906832e-01 1.12154195e-02 1.43151104e-01 -3.15518647e-01
3.14385265e-01 7.95157552e-01 1.64697647e-01 -5.58786467e-02
4.72349077e-01 2.68168032e-01 -9.32121128e-02 -3.86624932e-01
4.67058122e-01 5.09381473e-01 5.55499196e-01 -1.44199395e+00
7.53166020e-01 3.13642263e-01 2.62094170e-01 -5.01060605e-01
-1.48865747e+00 -6.22353852e-01 1.63711071e-01 5.34534529e-02
-3.43631171e-02 -2.83033341e-01 -9.28462446e-01 -1.65593117e-01
-8.08813095e-01 -5.69013357e-01 -6.38543516e-02 2.66178191e-01
-1.19495702e+00 1.79340795e-01 -2.25795597e-01 -1.20501697e+00
-3.17559630e-01 -1.10274172e+00 6.74564302e-01 4.66112775e-04
-6.31182939e-02 -6.15139842e-01 6.71249926e-02 4.42013085e-01
4.23197716e-01 1.42875180e-01 9.57424819e-01 -5.54580629e-01
-1.01555443e+00 1.68147665e-02 -6.72903135e-02 -3.37423414e-01
-8.12711865e-02 4.55952063e-02 -8.72632638e-02 -8.05698156e-01
-4.47382599e-01 -2.25641131e-01 6.27001762e-01 6.75255716e-01
1.64962649e+00 -8.26885402e-01 -7.10657358e-01 6.52677476e-01
1.77704692e+00 1.72142535e-01 2.30456829e-01 8.01385939e-01
-1.86830401e-01 4.83362935e-02 9.87088561e-01 6.85444951e-01
-7.74666965e-02 9.56159472e-01 4.50864136e-01 4.45373356e-02
4.45517868e-01 -1.06000498e-01 -2.45665312e-01 1.75796866e-01
-3.54221351e-02 -5.18505633e-01 -8.52792025e-01 1.02414882e+00
-2.48728371e+00 -9.68800545e-01 2.02437297e-01 2.28038287e+00
1.05654681e+00 3.90076876e-01 7.07990468e-01 2.54650742e-01
6.99911118e-01 -7.50345662e-02 -6.66280925e-01 -6.51124954e-01
-1.03854932e-01 -1.29502147e-01 8.96589160e-01 4.38874036e-01
-7.72925079e-01 9.94341969e-01 7.51815605e+00 1.23702180e+00
-7.31854856e-01 -4.65006419e-02 7.61290491e-01 -6.74730539e-01
-5.02286553e-01 8.81931931e-02 -7.03433156e-01 3.11151654e-01
9.75196064e-01 -5.67843139e-01 1.17864442e+00 1.08271933e+00
2.39119008e-01 -1.38225123e-01 -9.82677281e-01 9.77487803e-01
-2.86848336e-01 -1.90826750e+00 -3.72732133e-02 5.08498549e-01
1.24717414e+00 -3.33572149e-01 2.79880494e-01 -1.33201718e-01
8.16720963e-01 -7.17906117e-01 8.97977710e-01 -4.39175032e-02
6.68870926e-01 -1.36472487e+00 4.77698773e-01 4.04616505e-01
-8.75149906e-01 -5.11293948e-01 -3.42406034e-01 8.53254050e-02
2.55440533e-01 8.02742779e-01 -6.36925399e-01 7.23942041e-01
7.02900112e-01 2.79755276e-02 1.78147107e-01 1.58369052e+00
3.91438827e-02 9.70439196e-01 -7.18752086e-01 -3.31833810e-01
8.67291212e-01 -4.48881745e-01 6.10671401e-01 1.10432255e+00
3.40896308e-01 2.72222042e-01 6.44858122e-01 5.53240061e-01
-1.27504006e-01 2.42159009e-01 -3.15454006e-01 -3.22522372e-01
8.78922403e-01 1.04415762e+00 -9.42475915e-01 -2.92172134e-01
-1.00720346e-01 4.30736572e-01 5.75759292e-01 4.17970121e-01
-1.08018708e+00 9.50302929e-02 7.87397563e-01 -2.36244556e-02
7.90543079e-01 -4.05930698e-01 -4.15583163e-01 -7.70946920e-01
3.41510884e-02 -9.88159835e-01 7.10901260e-01 -2.74890453e-01
-1.15875864e+00 2.61137784e-01 1.81629613e-01 -7.98410535e-01
-2.26363361e-01 -2.35078901e-01 -1.21342376e-01 1.38023317e-01
-1.49799836e+00 -8.83802414e-01 3.75864714e-01 4.98590201e-01
7.58079469e-01 -9.10890400e-02 3.90088856e-01 -1.55365169e-01
-5.97841978e-01 9.43346858e-01 7.50613987e-01 -6.51385546e-01
2.87902385e-01 -1.20243967e+00 -2.90823996e-01 9.02986288e-01
2.91761756e-01 5.90879440e-01 1.01127601e+00 -2.31917664e-01
-1.94531906e+00 -8.40034127e-01 2.50587970e-01 1.66434765e-01
8.11838508e-01 1.62254274e-01 -8.45387578e-02 7.41414547e-01
1.79062143e-01 -2.20052656e-02 8.54584336e-01 5.93714118e-01
-2.06683338e-01 -2.75496393e-01 -1.09958458e+00 8.31803441e-01
1.23905969e+00 1.57871753e-01 -1.65077671e-01 5.16762435e-01
6.56062007e-01 -5.39173663e-01 -5.54582179e-01 1.89081170e-02
5.36897957e-01 -9.22778785e-01 8.13944221e-01 -7.51612663e-01
1.22873105e-01 -8.12976733e-02 -3.39167297e-01 -1.39190924e+00
-5.59710026e-01 -1.33809698e+00 -5.18608093e-01 8.15205514e-01
6.67997897e-01 -7.64168203e-01 8.72258186e-01 4.05208915e-01
-4.87061143e-02 -1.06768715e+00 -1.11467981e+00 -1.27947378e+00
1.90823078e-01 -3.53180051e-01 1.11566174e+00 7.19348013e-01
1.90532818e-01 6.47691661e-04 -4.54958349e-01 7.22454637e-02
8.38762164e-01 1.01049554e+00 8.47120345e-01 -9.29998875e-01
-7.11817741e-01 -9.40581381e-01 3.51611316e-01 -1.46707857e+00
2.46005982e-01 -4.95366275e-01 -2.89386641e-02 -1.32880890e+00
5.30787349e-01 -8.03421795e-01 -3.27208966e-01 5.86385489e-01
3.40999477e-02 1.70511350e-01 1.32170916e-01 2.34599575e-01
-1.03322411e+00 2.37279728e-01 1.09974539e+00 -1.00154638e-01
-2.79811859e-01 3.59295845e-01 -1.08378887e+00 3.36443275e-01
9.70344663e-01 -5.05446017e-01 -3.44538420e-01 -3.90670657e-01
5.25241196e-01 4.98883843e-01 -2.89006293e-01 -4.44122702e-01
2.24342823e-01 -1.01248717e+00 -3.55716556e-01 -7.65444994e-01
5.49073853e-02 -1.08699250e+00 2.53312469e-01 5.94202161e-01
-4.77175236e-01 2.09612861e-01 1.93501458e-01 7.69750655e-01
8.75189602e-02 -2.84367263e-01 7.35508919e-01 -9.75570530e-02
4.29366902e-02 4.22894269e-01 -3.81832868e-01 -1.79524682e-02
1.34010899e+00 -6.09410144e-02 -4.13344771e-01 -4.43679065e-01
-3.30966502e-01 5.48151016e-01 4.52155113e-01 -1.99081585e-01
1.45723850e-01 -1.26017189e+00 -5.02496421e-01 -4.23700035e-01
-6.15942813e-02 -2.28431940e-01 -2.06804529e-01 8.55463743e-01
-4.75661993e-01 8.67357850e-01 1.00767471e-01 -5.16034007e-01
-1.24500120e+00 1.09668493e+00 1.66970506e-01 -6.56046152e-01
-3.21821123e-01 5.04818141e-01 -4.95390475e-01 2.31289521e-01
1.82722941e-01 1.68375000e-01 6.02272570e-01 -3.12122017e-01
4.82321560e-01 4.47633088e-01 -1.50515795e-01 3.74121446e-04
-1.01277754e-01 2.08347455e-01 -1.42940864e-01 -4.62256461e-01
1.55341983e+00 -1.49257720e-01 -1.68043539e-01 -1.04596376e-01
6.97004914e-01 3.22955906e-01 -9.97296214e-01 -5.79661250e-01
1.90583512e-01 -9.89072859e-01 4.11891311e-01 -8.49189401e-01
-1.25778246e+00 3.37968431e-02 -3.79360132e-02 8.03869545e-01
1.09983075e+00 2.32671127e-02 9.33338225e-01 7.50671268e-01
9.88184094e-01 -1.25453722e+00 -1.19726710e-01 1.65482294e-02
5.91760457e-01 -6.73936248e-01 3.57901037e-01 -5.53544998e-01
-2.48547003e-01 1.01579404e+00 2.91690707e-01 7.78094456e-02
2.68438160e-01 3.78933340e-01 -5.04558146e-01 1.75172389e-02
-8.96836579e-01 -3.57651204e-01 -1.75962299e-01 8.54642168e-02
-1.76662192e-01 3.92434835e-01 -8.55821967e-01 4.97550398e-01
-3.95905972e-01 -2.40165964e-01 5.49174249e-01 1.04382133e+00
-5.50623178e-01 -1.39071894e+00 -6.37737334e-01 3.61743629e-01
-6.74069285e-01 9.01251137e-02 -4.18602169e-01 7.26540208e-01
-7.15308487e-01 1.21900737e+00 -1.32114917e-01 1.63582474e-01
6.54886439e-02 -3.28775704e-01 7.94416189e-01 -1.87877491e-01
-3.88061017e-01 5.88476181e-01 5.80131590e-01 -6.35073602e-01
-4.21186298e-01 -6.75139606e-01 -7.14342773e-01 -8.17941904e-01
-7.19541430e-01 7.95460582e-01 4.88705993e-01 9.20011401e-01
3.53985161e-01 1.51638299e-01 9.02960122e-01 -5.21737576e-01
-1.14921308e+00 -2.82982379e-01 -6.53896272e-01 -9.11763459e-02
2.83732414e-01 -7.85316944e-01 -8.41281787e-02 -4.73281056e-01] | [4.537922382354736, 3.294719696044922] |
c4e5691f-94d3-4d0e-9b1a-fa40e3116c15 | state-representation-and-polyomino-placement | 2001.04233 | null | https://arxiv.org/abs/2001.04233v1 | https://arxiv.org/pdf/2001.04233v1.pdf | State Representation and Polyomino Placement for the Game Patchwork | Modern board games are a rich source of entertainment for many people, but also contain interesting and challenging structures for game playing research and implementing game playing agents. This paper studies the game Patchwork, a two player strategy game using polyomino tile drafting and placement. The core polyomino placement mechanic is implemented in a constraint model using regular constraints, extending and improving the model in (Lagerkvist, Pesant, 2008) with: explicit rotation handling; optional placements; and new constraints for resource usage. Crucial for implementing good game playing agents is to have great heuristics for guiding the search when faced with large branching factors. This paper divides placing tiles into two parts: a policy used for placing parts and an evaluation used to select among different placements. Policies are designed based on classical packing literature as well as common standard constraint programming heuristics. For evaluation, global propagation guided regret is introduced, choosing placements based on not ruling out later placements. Extensive evaluations are performed, showing the importance of using a good evaluation and that the proposed global propagation guided regret is a very effective guide. | ['Mikael Zayenz Lagerkvist'] | 2020-01-13 | null | null | null | null | ['board-games'] | ['playing-games'] | [ 2.25402210e-02 2.08092913e-01 -2.58258969e-01 2.29445174e-01
-3.22175533e-01 -6.95091903e-01 -9.41348001e-02 4.17759150e-01
-4.90169585e-01 1.05348146e+00 -1.37139931e-01 -4.12480652e-01
-6.93912387e-01 -9.52172399e-01 -5.78899205e-01 -8.05983961e-01
-5.74897110e-01 1.10277784e+00 6.99447155e-01 -5.09304762e-01
4.54415917e-01 3.13408583e-01 -1.34401715e+00 -7.64619112e-02
6.71184540e-01 9.76353049e-01 5.43196380e-01 8.04255068e-01
-2.44355574e-02 7.32151449e-01 -5.83206654e-01 -5.96277237e-01
8.15417230e-01 -3.28759313e-01 -8.12676370e-01 2.95417070e-01
-6.59557819e-01 -8.69776905e-02 4.25497055e-01 8.94929647e-01
5.63905895e-01 3.73978138e-01 2.84077018e-01 -1.57517421e+00
1.04425542e-01 9.37387943e-01 -7.84537435e-01 1.76823050e-01
4.08744097e-01 -1.08425260e-01 1.18345523e+00 -1.22741249e-03
7.93734789e-01 7.30225086e-01 3.77689570e-01 3.22933942e-01
-7.88645327e-01 -3.20809603e-01 3.85155469e-01 2.93292016e-01
-1.22973883e+00 2.75937408e-01 3.30143124e-01 -1.96396351e-01
9.55258489e-01 7.54675806e-01 1.12177455e+00 5.12293339e-01
2.97464758e-01 6.75709128e-01 1.11663854e+00 -7.14221358e-01
9.62156057e-01 3.95209305e-02 -1.95734516e-01 3.66545975e-01
6.94348693e-01 2.14727357e-01 -2.46520728e-01 -2.47162342e-01
6.29001498e-01 -2.53181368e-01 -8.16546753e-02 -7.92019010e-01
-3.65072817e-01 9.97657478e-01 2.40349978e-01 -1.04683436e-01
-6.53701961e-01 2.51732975e-01 4.85601664e-01 1.91208199e-01
3.34887803e-01 7.49772549e-01 -4.10543472e-01 -7.57415473e-01
-1.01862824e+00 8.22517037e-01 1.19683897e+00 1.28954148e+00
2.84824014e-01 -4.99271184e-01 -1.60250366e-01 6.64458871e-01
1.07473902e-01 1.24699488e-01 -9.26635489e-02 -8.28556716e-01
6.24186575e-01 4.34167147e-01 2.92210042e-01 -9.13289309e-01
-5.60337484e-01 -4.14927661e-01 -4.55469280e-01 6.35009348e-01
3.13234270e-01 -4.70612079e-01 -6.59403801e-01 1.54050660e+00
3.06238979e-01 -2.34750882e-01 -3.42799157e-01 1.05607688e+00
1.17103569e-02 5.18017530e-01 -2.38001898e-01 -2.94042140e-01
1.41425538e+00 -1.15040708e+00 -3.48801166e-01 -1.88267976e-01
4.46671635e-01 -7.64951706e-01 2.76553661e-01 9.13673818e-01
-1.66712499e+00 1.96460158e-01 -1.15957713e+00 3.75856340e-01
-3.66771281e-01 -2.28699818e-01 9.01727796e-01 1.02940536e+00
-1.28205192e+00 6.06158137e-01 -7.02881873e-01 -3.53396356e-01
1.65327922e-01 6.94161057e-01 7.71730999e-03 -2.70105481e-01
-8.22104752e-01 9.95378137e-01 4.27311808e-01 1.46089077e-01
-3.48403275e-01 -3.20067525e-01 -6.53878033e-01 2.44460553e-01
9.99699533e-01 -5.87210357e-01 1.31906664e+00 -6.69991434e-01
-1.59249067e+00 6.74439728e-01 2.66539097e-01 -6.06507182e-01
1.01805508e+00 2.31436551e-01 3.22327822e-01 -1.58302099e-01
2.25046292e-01 3.80697489e-01 3.10697168e-01 -1.23138046e+00
-1.13871825e+00 -5.37276343e-02 8.23341429e-01 5.56075692e-01
3.27258527e-01 3.39930803e-02 -5.45838416e-01 -2.70805538e-01
2.18692109e-01 -9.62008476e-01 -1.01365411e+00 -5.16098917e-01
-4.43837106e-01 2.42615908e-01 -1.92981005e-01 -3.37528855e-01
1.31589544e+00 -1.55520761e+00 3.86639625e-01 8.08184564e-01
9.62874740e-02 -9.09669325e-02 -8.45662784e-03 8.26614499e-01
1.58314575e-02 1.36963382e-01 1.16881505e-01 -1.70612216e-01
5.04122019e-01 3.96532297e-01 4.03390825e-02 3.49259257e-01
-2.20250681e-01 4.67049569e-01 -8.33794296e-01 4.36210632e-02
1.35505885e-01 -3.23397994e-01 -8.54422748e-01 4.07421635e-03
-4.31415021e-01 -2.99541652e-01 -3.14622283e-01 6.41896069e-01
9.56329048e-01 1.99037805e-01 4.00869817e-01 6.03668690e-01
-4.38584328e-01 1.47005007e-01 -1.74535024e+00 1.31518459e+00
-3.52152944e-01 -8.68294388e-02 6.35216892e-01 -8.25096726e-01
7.57363200e-01 -4.80990484e-02 5.46974421e-01 -7.35102296e-01
5.42841136e-01 3.37288767e-01 2.12027237e-01 -2.31533363e-01
9.21609879e-01 -7.36438334e-02 -2.39390388e-01 4.53667819e-01
-6.29745662e-01 -2.02814862e-01 8.82362008e-01 2.76235342e-01
1.48257565e+00 3.45592722e-02 3.50810736e-01 -6.13896668e-01
2.66360343e-01 3.46354783e-01 8.39874744e-01 7.81734526e-01
-6.53770790e-02 4.85208631e-01 1.06718802e+00 -4.36561406e-01
-1.05129981e+00 -5.31121731e-01 2.24472836e-01 1.21959627e+00
5.80618143e-01 -6.54846191e-01 -6.84492052e-01 -1.01573728e-01
-1.51271969e-01 3.98041010e-01 -8.37091744e-01 2.02366233e-01
-4.05723751e-01 -8.50933015e-01 -6.47158623e-02 4.71828431e-01
6.90939426e-02 -9.45620000e-01 -1.19785047e+00 5.86850822e-01
2.67703030e-02 -7.85800278e-01 -2.18933716e-01 9.81850803e-01
-4.70971346e-01 -1.31393778e+00 -7.68967152e-01 -6.59037232e-01
6.70691550e-01 4.13743228e-01 1.23377824e+00 4.16971743e-01
-3.48865420e-01 5.47002815e-02 -9.54528689e-01 -5.89789331e-01
7.29351640e-02 3.06227416e-01 -2.03524396e-01 -7.66132832e-01
6.51363358e-02 -5.74807107e-01 -3.61620516e-01 4.87093508e-01
-7.41690338e-01 2.98363239e-01 5.84941208e-01 8.53891432e-01
4.83159810e-01 4.36827600e-01 -1.07137114e-01 -1.11236596e+00
7.86561728e-01 -4.73102242e-01 -1.01951766e+00 1.84711128e-01
-3.06022137e-01 -5.87473474e-02 6.69160128e-01 1.45004421e-01
-4.77551550e-01 -3.26269925e-01 -1.47709772e-01 -1.70593131e-02
2.30648056e-01 3.45934808e-01 -2.28893399e-01 -2.51037508e-01
2.14903966e-01 -3.78573090e-01 -2.26540834e-01 -5.45050092e-02
9.22885910e-02 7.88753927e-02 -2.28924766e-01 -1.13814151e+00
6.11981928e-01 2.52962381e-01 2.79370129e-01 -3.17351550e-01
-6.16429597e-02 -6.55272484e-01 -3.17133635e-01 -2.29992852e-01
5.61124265e-01 -3.13605428e-01 -1.08952653e+00 2.49745813e-03
-9.34816599e-01 -6.22701585e-01 -5.16284823e-01 8.08279589e-02
-1.00885892e+00 -4.40920554e-02 -4.51852649e-01 -1.15263236e+00
2.15308875e-01 -1.49453211e+00 7.66063809e-01 3.72177213e-01
1.62561722e-02 -7.32672572e-01 3.90775055e-01 1.95710734e-01
4.89282638e-01 4.17976618e-01 5.97815394e-01 -3.44625741e-01
-6.93320096e-01 -7.25218058e-02 -7.20631778e-02 -3.07298601e-01
-4.00819927e-01 -2.89696101e-02 -2.40595371e-01 -3.93037468e-01
-3.72558802e-01 -1.34873509e-01 3.78918827e-01 6.35050833e-01
8.94740105e-01 -1.80547938e-01 -4.12598848e-01 6.21652603e-01
1.82536423e+00 5.43958724e-01 7.10015655e-01 1.00928640e+00
9.29232165e-02 5.71543336e-01 8.24531376e-01 7.56522596e-01
2.07291126e-01 8.44388545e-01 9.16698515e-01 3.56522799e-02
6.08725667e-01 6.50377423e-02 6.14379942e-02 4.41919684e-01
-5.26305139e-01 -6.99129224e-01 -6.72597587e-01 4.84090626e-01
-2.22634792e+00 -8.25294852e-01 -5.73369227e-02 2.21176767e+00
3.34845275e-01 4.65428233e-01 6.13761187e-01 4.43131149e-01
7.81991601e-01 -7.82881677e-02 -7.56457299e-02 -1.22194886e+00
-8.99909995e-03 7.22560167e-01 1.32265997e+00 7.38668144e-01
-8.93935084e-01 9.28690434e-01 5.69515038e+00 1.07291651e+00
-5.37769020e-01 9.51446779e-03 4.78851795e-01 -3.53682727e-01
-2.25873619e-01 3.10992867e-01 -6.24712765e-01 5.27871609e-01
3.92884254e-01 -3.14522982e-01 7.20127761e-01 9.03940916e-01
3.41406643e-01 -8.28322768e-01 -6.63565755e-01 7.09315717e-01
-1.76652506e-01 -1.36027837e+00 -6.70254230e-01 4.99365836e-01
7.36914039e-01 -3.61012965e-01 -2.63868093e-01 2.50402570e-01
9.11883712e-01 -9.60515022e-01 1.06426370e+00 -5.97648695e-02
2.07774624e-01 -1.21477985e+00 1.08735907e+00 3.32379937e-01
-1.19381917e+00 -8.79536420e-02 -6.27640128e-01 -7.37742841e-01
5.03099561e-01 3.41461241e-01 -5.58783233e-01 9.66831267e-01
4.83302861e-01 4.36200462e-02 -5.22335619e-02 1.81673920e+00
-3.51558119e-01 1.83451086e-01 -6.06935441e-01 -5.15273929e-01
8.06771755e-01 -6.31364286e-01 5.10535955e-01 1.04937041e+00
2.65133858e-01 3.58854681e-01 4.88206267e-01 4.74992394e-01
5.61062276e-01 2.69763350e-01 -2.11419389e-01 4.42604721e-01
4.31861222e-01 1.18755400e+00 -1.60704577e+00 3.24571818e-01
9.95164216e-02 7.76498139e-01 1.75270736e-01 3.92067172e-02
-1.01460779e+00 -2.35967889e-01 6.56243026e-01 2.29741186e-01
5.78168631e-01 -1.22304179e-01 -6.46296322e-01 -4.67011124e-01
1.48785919e-01 -7.06849456e-01 3.00810397e-01 -4.67290699e-01
-9.55798030e-01 5.37023604e-01 1.15270227e-01 -9.90961909e-01
-1.59734949e-01 -7.63841569e-01 -9.02130961e-01 8.10694039e-01
-1.29107022e+00 -6.94217026e-01 -5.51848635e-02 1.43422291e-01
4.03633624e-01 1.44462109e-01 4.48983550e-01 1.94404110e-01
-5.96469283e-01 5.46196699e-01 1.73875950e-02 -3.94901007e-01
-2.71605421e-02 -1.48935556e+00 1.66333184e-01 9.04878199e-01
-2.84197152e-01 2.40440711e-01 9.32585716e-01 -5.06723106e-01
-1.48606098e+00 -5.37406266e-01 4.99104351e-01 9.30524319e-02
5.30362964e-01 -5.51473081e-01 -5.40780090e-02 3.12136382e-01
2.83208281e-01 -6.98358655e-01 5.68606794e-01 3.13972384e-01
4.18553472e-01 9.77493078e-02 -1.21858966e+00 7.68365562e-01
1.10879123e+00 5.57675719e-01 -1.99863445e-02 4.79809910e-01
2.37692013e-01 -8.89246762e-01 -1.90261990e-01 -1.74867418e-02
4.19040978e-01 -1.27924442e+00 4.91639793e-01 -3.45375568e-01
4.23227429e-01 -2.39717960e-01 -1.60876557e-01 -1.38489211e+00
-6.97942376e-01 -1.06256008e+00 6.01735592e-01 8.67760718e-01
6.37775004e-01 -3.86909515e-01 1.33871150e+00 8.66524100e-01
-1.93626821e-01 -1.17543316e+00 -9.45218325e-01 -9.18497086e-01
-7.63656944e-03 -3.31752896e-01 5.18576860e-01 5.07316351e-01
5.15024483e-01 -4.40444015e-02 -3.54847282e-01 1.24102272e-01
3.90890896e-01 -9.48207825e-02 7.72570550e-01 -9.84168887e-01
-8.47323000e-01 -8.96070957e-01 -6.53228223e-01 -9.35001791e-01
-3.73167932e-01 -4.93762702e-01 2.57003307e-01 -1.82082701e+00
7.30145052e-02 -1.06809652e+00 1.97360739e-01 2.98660696e-01
2.21618697e-01 1.24174505e-01 6.04723215e-01 -2.48653710e-01
-8.75751555e-01 5.57703860e-02 1.16254532e+00 2.22257808e-01
-2.69230992e-01 3.60481799e-01 -7.79886007e-01 4.60126102e-01
7.59949148e-01 -3.51515502e-01 -4.56614196e-01 -2.50116706e-01
1.04755175e+00 3.07615489e-01 -2.37836793e-01 -1.06516576e+00
5.45129716e-01 -4.51599419e-01 -3.69897723e-01 -3.98344547e-01
3.26059461e-01 -1.07568598e+00 4.25196618e-01 6.52701139e-01
-3.27667631e-02 5.82967579e-01 3.73040110e-01 4.39765871e-01
1.13573641e-01 -8.16321015e-01 4.80253041e-01 -3.41078877e-01
-7.01958835e-01 2.13507891e-01 -4.37074751e-01 7.92829394e-02
1.57750738e+00 -6.14684761e-01 -1.34973347e-01 -4.35865939e-01
-8.65566313e-01 6.82723105e-01 7.18342721e-01 -3.00190505e-02
5.81239276e-02 -8.81514907e-01 -3.84131521e-01 -6.25243112e-02
-2.33572856e-01 1.36360973e-02 2.41270080e-01 8.14717114e-01
-1.36064529e+00 2.96835601e-01 -5.78929067e-01 -4.43035588e-02
-1.18542516e+00 5.16033709e-01 4.47954684e-02 -9.24176395e-01
-4.55597073e-01 1.22493029e+00 -3.50924194e-01 2.34251134e-02
3.75289381e-01 -5.18742979e-01 5.31621352e-02 4.79545593e-02
3.84863168e-01 5.75366735e-01 3.54006052e-01 4.02888283e-02
-4.45090055e-01 3.16556573e-01 -1.69039130e-01 -3.69409442e-01
1.40313971e+00 -5.70813753e-02 -1.87577561e-01 -1.24787048e-01
1.40581295e-01 1.96436033e-01 -9.91707563e-01 3.57004791e-01
-1.18324839e-01 -6.96083665e-01 -2.40546837e-01 -8.68785679e-01
-1.02657032e+00 3.19499820e-01 -3.56190093e-02 7.80188859e-01
1.16666758e+00 -4.80474770e-01 4.66547549e-01 -6.15577102e-02
1.30167925e+00 -1.34004533e+00 -5.20297468e-01 6.99511766e-01
4.30057496e-01 -5.88364542e-01 2.63817906e-01 -8.35677981e-01
-6.35531843e-01 9.94370401e-01 6.01586878e-01 -6.50324225e-01
4.32390302e-01 8.70068312e-01 -3.19264472e-01 -1.45950228e-01
-8.64387274e-01 -6.32718205e-01 -5.02406180e-01 7.01750338e-01
-9.42002535e-02 4.47829187e-01 -1.19059122e+00 7.54784286e-01
-5.89523673e-01 -1.92835376e-01 9.07451689e-01 1.18592966e+00
-5.54296136e-01 -1.59312880e+00 -5.94599366e-01 4.61852610e-01
-2.55412072e-01 -6.29101247e-02 -3.90467554e-01 9.94605839e-01
5.23666739e-01 8.88848364e-01 6.70801774e-02 -2.73773283e-01
2.91780412e-01 -4.98719633e-01 6.82690203e-01 -7.83814967e-01
-1.06202567e+00 3.62601988e-02 3.29885513e-01 -5.79508960e-01
-3.13101411e-01 -5.94013155e-01 -6.69568121e-01 -7.91774750e-01
-5.90631306e-01 7.12127686e-01 6.17319822e-01 9.69570398e-01
-9.11295488e-02 7.41982043e-01 4.22365844e-01 -9.28734064e-01
-2.96358228e-01 -4.58047986e-01 -1.17725360e+00 -2.45093390e-01
-5.19351184e-01 -1.00788999e+00 -7.90429562e-02 -7.17603445e-01] | [3.446017026901245, 1.5071183443069458] |
0f910413-3c3e-46ce-9523-a36e57d0e623 | echo-aware-adaptation-of-sound-event | 2202.09121 | null | https://arxiv.org/abs/2202.09121v1 | https://arxiv.org/pdf/2202.09121v1.pdf | Echo-aware Adaptation of Sound Event Localization and Detection in Unknown Environments | Our goal is to develop a sound event localization and detection (SELD) system that works robustly in unknown environments. A SELD system trained on known environment data is degraded in an unknown environment due to environmental effects such as reverberation and noise not contained in the training data. Previous studies on related tasks have shown that domain adaptation methods are effective when data on the environment in which the system will be used is available even without labels. However adaptation to unknown environments remains a difficult task. In this study, we propose echo-aware feature refinement (EAR) for SELD, which suppresses environmental effects at the feature level by using additional spatial cues of the unknown environment obtained through measuring acoustic echoes. FOA-MEIR, an impulse response dataset containing over 100 environments, was recorded to validate the proposed method. Experiments on FOA-MEIR show that the EAR effectively improves SELD performance in unknown environments. | ['Shoichiro Saito', 'Yasunori Ohishi', 'Masahiro Yasuda'] | 2022-02-18 | null | null | null | null | ['sound-event-localization-and-detection'] | ['audio'] | [ 1.64899990e-01 -8.14316988e-01 8.65996838e-01 -4.84824002e-01
-8.72358143e-01 -5.97219467e-01 -2.07267143e-02 1.41016707e-01
-5.92224717e-01 5.24508178e-01 3.26064020e-01 2.71823734e-01
-2.09998056e-01 -4.71668184e-01 -5.50465167e-01 -6.65992558e-01
-2.45747924e-01 -2.00492442e-01 5.01654208e-01 -2.10375994e-01
2.00992092e-01 3.93382639e-01 -1.82302201e+00 2.84950882e-01
5.55604041e-01 7.59652436e-01 7.09491193e-01 1.01882899e+00
3.02471638e-01 2.34424531e-01 -1.02717412e+00 4.33454514e-01
3.10233980e-01 -4.28062230e-01 -1.06482953e-01 -1.15660936e-01
4.19288248e-01 -1.47037879e-01 -1.83984563e-01 9.23973441e-01
1.02791393e+00 7.28972614e-01 3.82331043e-01 -9.82134938e-01
-3.07699412e-01 4.45673406e-01 1.05311908e-01 6.19963765e-01
2.62590200e-01 -3.04846227e-01 4.79978770e-01 -1.29734743e+00
7.40727857e-02 9.58950162e-01 9.45753396e-01 4.17477518e-01
-1.12716782e+00 -9.75564599e-01 1.12115508e-02 3.62632334e-01
-1.70071900e+00 -8.16755831e-01 9.29936528e-01 -2.90732712e-01
9.42821622e-01 2.39470154e-01 8.34560469e-02 1.15489125e+00
1.90716490e-01 1.97813570e-01 1.11332798e+00 -6.26464188e-01
5.36699951e-01 3.14724535e-01 2.47309938e-01 6.75321966e-02
1.65992796e-01 4.82053727e-01 -9.76231456e-01 -1.43347532e-01
5.16038239e-01 -4.75204587e-01 -4.09372330e-01 5.80970161e-02
-9.46751714e-01 3.45922530e-01 2.67265648e-01 5.01003623e-01
-5.22372365e-01 -1.60941482e-01 2.13729292e-01 5.16329706e-01
5.52816570e-01 5.40181100e-01 -6.42820001e-01 -1.05555981e-01
-7.13275790e-01 -1.67561900e-02 7.02630162e-01 8.02362740e-01
5.73732018e-01 5.38421273e-01 -6.01559095e-02 1.27821100e+00
2.20708653e-01 9.88649487e-01 4.83112723e-01 -6.51713431e-01
1.48048401e-01 -3.75479341e-01 3.89183968e-01 -1.01846635e+00
-6.31483316e-01 -1.04114068e+00 -5.26892543e-01 1.52190104e-01
1.92589611e-01 -2.67327458e-01 -9.33838546e-01 1.93419349e+00
3.75029325e-01 7.02011526e-01 3.04887056e-01 1.03973198e+00
7.18342662e-01 6.40423596e-01 8.08863193e-02 -2.73092479e-01
9.69024658e-01 -5.75475097e-01 -9.56130207e-01 -5.15143692e-01
1.00374952e-01 -1.02702141e+00 8.69323909e-01 7.55157828e-01
-3.71640027e-01 -1.10109138e+00 -9.01763439e-01 6.15878344e-01
-3.12200248e-01 3.67346704e-02 9.11680013e-02 8.53917420e-01
-8.01089823e-01 2.34613046e-01 -6.51883364e-01 -4.76455599e-01
-2.18150854e-01 7.47181401e-02 -4.05404270e-01 4.43156958e-02
-1.33781326e+00 6.53256416e-01 1.82075411e-01 3.40120256e-01
-1.19736314e+00 -4.15741891e-01 -9.40417886e-01 4.51847874e-02
2.62523264e-01 -1.90532636e-02 1.30131686e+00 -7.17934191e-01
-1.29614329e+00 8.50345939e-02 -3.34273398e-01 -3.69886100e-01
1.38853136e-02 -7.45863199e-01 -1.12471902e+00 -5.86517043e-02
1.42183870e-01 -1.91930145e-01 1.10096073e+00 -1.42563236e+00
-6.74276114e-01 -6.71042055e-02 -6.19129181e-01 2.47982427e-01
-4.64461178e-01 1.44824907e-01 -9.03781801e-02 -8.14714551e-01
3.83296579e-01 -7.54294395e-01 -1.10157669e-01 -4.72951055e-01
-1.05742998e-01 2.02361599e-01 8.26543629e-01 -9.13762033e-01
1.28099501e+00 -2.57672215e+00 -6.41111910e-01 4.13801372e-01
-3.45312685e-01 3.38193536e-01 -3.10286969e-01 4.68872398e-01
-1.48452878e-01 -4.39220101e-01 -7.65710473e-02 -1.83030054e-01
-1.34798646e-01 -2.91094016e-02 -3.58330488e-01 3.36179018e-01
-2.52498453e-03 -5.87636186e-03 -8.54127824e-01 -1.94657296e-01
2.44580597e-01 5.28509557e-01 -4.79056925e-01 4.79505926e-01
3.25452298e-01 7.74485707e-01 -3.45461547e-01 4.26037610e-01
7.48369992e-01 3.81851912e-01 -3.92065912e-01 -9.46883708e-02
-2.86761045e-01 8.18013325e-02 -1.77205288e+00 1.41864669e+00
-6.31495714e-01 7.74327815e-01 3.09506536e-01 -6.88797057e-01
1.28176892e+00 5.03354251e-01 2.16510832e-01 -7.99950540e-01
-3.56804393e-02 3.97495836e-01 1.57250911e-01 -8.13415051e-01
4.57384974e-01 -1.85005501e-01 -1.12558767e-01 4.48693633e-02
9.50556397e-02 -7.67663419e-02 -2.07391962e-01 -1.41540289e-01
1.41372073e+00 -3.10881878e-03 2.64247864e-01 5.47574274e-02
4.59117413e-01 -2.76026577e-01 8.77339900e-01 1.29712451e+00
-4.56939459e-01 7.23347545e-01 -6.84623182e-01 -1.66562125e-02
-7.19574451e-01 -1.35993421e+00 -6.13861442e-01 1.51659143e+00
1.31563351e-01 -1.35460779e-01 -5.28005183e-01 -1.97002769e-01
-2.74635345e-01 1.00850999e+00 -2.77848572e-01 -2.54532665e-01
-5.15341341e-01 -6.28721356e-01 5.61314702e-01 5.51589847e-01
5.00319958e-01 -1.13756561e+00 -5.94423294e-01 8.96102965e-01
-4.42752570e-01 -1.30136359e+00 -3.48856539e-01 5.77407718e-01
-3.87122482e-01 -5.94909728e-01 -4.19364333e-01 -6.37738109e-01
3.35136265e-01 4.63769227e-01 7.48856366e-01 -4.54800099e-01
-3.98357928e-01 6.16222203e-01 -6.62001729e-01 -8.28690052e-01
-3.85583729e-01 -5.24560094e-01 4.23656136e-01 2.52516717e-01
2.97709048e-01 -7.21676707e-01 -4.63182688e-01 6.83010817e-01
-6.82884157e-01 -5.54322124e-01 4.58254814e-01 9.20155823e-01
4.50778276e-01 5.91569722e-01 1.16930270e+00 -6.21550262e-01
7.31495500e-01 -4.73956794e-01 -2.79296994e-01 1.12984993e-01
-2.90504366e-01 -2.41275549e-01 5.12941122e-01 -7.25881159e-01
-1.61030471e+00 1.34045780e-01 -2.82737255e-01 -5.06779067e-02
-7.46112823e-01 4.40615535e-01 -2.38007441e-01 -7.52420649e-02
1.29475451e+00 2.37053365e-01 -8.17785442e-01 -7.44118869e-01
-2.36503333e-01 1.19925177e+00 8.98968101e-01 -5.00929832e-01
8.63586426e-01 2.19754040e-01 -1.92255437e-01 -1.20779669e+00
-7.62111545e-01 -9.00171936e-01 -4.98243511e-01 -2.43770331e-01
4.41972852e-01 -1.04178703e+00 2.10140888e-02 6.69895649e-01
-9.46738541e-01 -9.36358050e-02 -6.03569821e-02 1.13480437e+00
-1.74897119e-01 1.02722622e-01 -1.71381131e-01 -1.37666118e+00
-1.32925093e-01 -4.92345691e-01 7.65445292e-01 1.86185643e-01
-2.30267778e-01 -6.43552303e-01 5.06116569e-01 3.16031352e-02
7.49071836e-01 -4.90866136e-03 2.94468105e-01 -8.82011771e-01
-1.47317976e-01 -2.61187583e-01 2.83365756e-01 7.59159565e-01
4.47976083e-01 -3.82644266e-01 -1.64639044e+00 -2.87042975e-01
3.33471000e-01 1.60876557e-01 7.72603512e-01 3.43910635e-01
9.03568566e-01 1.59698069e-01 -1.88146353e-01 3.86869818e-01
1.16005135e+00 5.90658128e-01 4.65875328e-01 3.59101832e-01
2.08485633e-01 5.85627377e-01 9.72219706e-01 7.21777499e-01
-1.05242796e-01 6.07460499e-01 8.59453753e-02 -1.58659399e-01
-4.05706793e-01 -1.11566849e-01 3.65039378e-01 9.58617508e-01
2.08664745e-01 -4.67262387e-01 -8.25264454e-01 8.34907293e-01
-1.51567662e+00 -9.03115094e-01 -1.95809230e-01 2.35333657e+00
6.50631487e-01 3.24840136e-02 -3.88723344e-01 4.70971137e-01
7.65905797e-01 -2.54704654e-01 -4.95542914e-01 -1.80590451e-01
-1.92181543e-01 4.97335941e-01 1.83982834e-01 4.69781905e-01
-1.24864626e+00 6.02618039e-01 6.32360888e+00 4.71131057e-01
-1.22960544e+00 2.71052688e-01 -2.19825953e-01 2.79003322e-01
2.62778401e-01 -3.31156462e-01 -5.40222704e-01 1.47646368e-01
1.08745611e+00 -1.13969401e-01 2.88605869e-01 8.19544435e-01
5.53766012e-01 -2.71426082e-01 -9.31569874e-01 9.99449492e-01
1.87171876e-01 -3.49484652e-01 -7.07713604e-01 -5.57273746e-01
4.13020134e-01 7.44458884e-02 1.89130664e-01 4.81689364e-01
2.39919964e-02 -6.26298964e-01 7.62988687e-01 6.05775714e-01
4.81855452e-01 -6.05249226e-01 7.64535129e-01 4.89181429e-01
-1.24391937e+00 -2.40850031e-01 -3.21850866e-01 -7.43526965e-02
2.69308649e-02 6.29264593e-01 -1.40827310e+00 4.40107614e-01
1.10942781e+00 2.32648909e-01 -5.21757364e-01 1.68661714e+00
-1.78832516e-01 1.06916106e+00 -4.43223298e-01 2.48761863e-01
-2.78719693e-01 2.99572736e-01 1.07912910e+00 1.56637120e+00
7.83661306e-01 1.86004221e-01 3.68705913e-02 4.23292249e-01
3.62781584e-01 1.92756698e-01 -7.02314973e-01 2.62655914e-01
6.84477448e-01 1.02919567e+00 -4.06665117e-01 6.78545907e-02
-1.73289046e-01 8.84645939e-01 -2.37921447e-01 7.98344254e-01
-6.34385705e-01 -7.75653422e-01 4.20571625e-01 -1.58917606e-01
3.79172981e-01 -3.75638902e-01 8.00092146e-02 -7.75037050e-01
-2.16933782e-03 -7.23985851e-01 2.93368042e-01 -1.09758186e+00
-1.37160349e+00 9.21525002e-01 9.83917341e-03 -1.52144897e+00
-1.83198377e-01 -3.24954808e-01 -3.81022722e-01 9.69174623e-01
-1.41350710e+00 -6.14357471e-01 -5.24142087e-01 6.59527838e-01
6.76877499e-01 -2.06817582e-01 1.12076855e+00 6.50430083e-01
-4.02696848e-01 5.89384735e-01 4.58526939e-01 -3.37470882e-02
1.21993136e+00 -1.01148772e+00 1.05823413e-01 1.12546790e+00
4.56642389e-01 5.41081786e-01 1.24903882e+00 -7.40235806e-01
-8.62759471e-01 -1.25526500e+00 6.99703872e-01 -2.07416013e-01
4.08627301e-01 -6.40035152e-01 -1.34091520e+00 3.41109097e-01
-2.77827263e-01 2.86072910e-01 8.73196065e-01 5.68636470e-02
-3.74749631e-01 -3.66239101e-01 -1.21195567e+00 4.70609288e-04
9.50905383e-01 -6.80539846e-01 -8.13535929e-01 -4.51743975e-02
5.34057140e-01 -3.71775746e-01 -5.05261004e-01 5.29170990e-01
4.29618418e-01 -6.63467824e-01 9.05748308e-01 -3.00934732e-01
-5.56971908e-01 -6.52215958e-01 -6.82384491e-01 -1.78994656e+00
-3.96380275e-01 -4.42676425e-01 1.45344585e-01 1.47720385e+00
2.37506315e-01 -7.78908074e-01 9.12689418e-02 2.83264607e-01
-4.72097069e-01 4.53271180e-01 -8.69661272e-01 -1.14990079e+00
-4.57319826e-01 -1.00438488e+00 4.46724266e-01 7.22924829e-01
-4.80729640e-01 2.11076751e-01 -4.93614733e-01 1.02854049e+00
7.32405841e-01 -2.54925877e-01 7.45885193e-01 -1.27218008e+00
-3.82920057e-01 4.99918848e-01 -3.84045422e-01 -6.53265536e-01
-9.78826731e-03 -3.71552825e-01 9.65786099e-01 -1.20802903e+00
-2.24932030e-01 -6.22989655e-01 -9.43761230e-01 3.71623665e-01
-1.71666458e-01 2.03127861e-01 -1.91069290e-01 2.65802778e-02
-4.73055094e-01 5.78332663e-01 7.49478579e-01 1.24420881e-01
-4.00528699e-01 3.34649175e-01 -3.77235532e-01 8.47918749e-01
7.86221325e-01 -9.94438529e-01 -3.62111032e-01 -4.01501477e-01
-1.12685315e-01 5.07299714e-02 3.90611857e-01 -1.64836562e+00
4.37668353e-01 -7.14159310e-02 5.54110706e-01 -5.73691666e-01
5.52637756e-01 -1.10095561e+00 3.82524520e-01 5.02932332e-02
-3.63144159e-01 -4.11851227e-01 5.36071897e-01 8.26759636e-01
-4.44293588e-01 -3.63683790e-01 6.69301331e-01 2.52748191e-01
-1.09988654e+00 -1.92821562e-01 -6.27467036e-01 -8.53356197e-02
7.88460076e-01 -2.30275150e-02 -2.15933278e-01 -5.05943179e-01
-9.36282277e-01 -3.66088659e-01 -1.69805005e-01 4.97515172e-01
7.95433819e-01 -1.15220153e+00 -9.97707367e-01 4.97566700e-01
2.74880260e-01 -4.57103908e-01 4.95764792e-01 3.75565290e-01
1.15884811e-01 1.85187832e-01 -7.59018734e-02 -4.68196899e-01
-1.51917112e+00 2.79201686e-01 4.30874318e-01 3.67396146e-01
-5.62495708e-01 9.59535897e-01 3.46902281e-01 -3.79741043e-01
4.22761142e-01 -2.10624442e-01 -3.74063015e-01 -3.15922946e-01
7.72800684e-01 4.04779464e-01 4.43213701e-01 -7.40706921e-01
-4.64623988e-01 3.64384085e-01 2.79235095e-01 -4.67265338e-01
1.40753114e+00 -5.09121776e-01 4.03153658e-01 7.61657059e-01
8.93292665e-01 4.26297069e-01 -1.04372382e+00 -7.10673213e-01
3.33235395e-04 -7.25943744e-01 3.96098435e-01 -1.07937598e+00
-4.42197502e-01 8.61535013e-01 1.37757206e+00 2.10714430e-01
1.49009073e+00 -2.64381945e-01 5.45597434e-01 6.77466094e-01
5.85540950e-01 -1.28008413e+00 3.01660478e-01 6.63109541e-01
1.07722747e+00 -1.08686602e+00 -4.84687924e-01 -2.74402559e-01
-4.99097705e-01 9.89124775e-01 5.46053231e-01 1.14162296e-01
8.41685116e-01 5.65179944e-01 4.59978461e-01 1.64501250e-01
-5.41827202e-01 -3.76823127e-01 2.70346433e-01 1.06233227e+00
2.15921476e-01 -3.22062597e-02 1.88824609e-01 9.60331023e-01
-3.07181060e-01 -2.37875253e-01 3.42716962e-01 1.09323311e+00
-8.31855237e-01 -7.58331776e-01 -1.11684191e+00 1.47475779e-01
-5.79217792e-01 -1.11242473e-01 -1.16842836e-02 3.71730983e-01
3.15585405e-01 1.62397623e+00 -1.75633833e-01 -4.17753190e-01
8.38568270e-01 1.92553952e-01 1.13574497e-01 -6.92082584e-01
-5.55415511e-01 5.29955089e-01 1.18220821e-01 -3.67609143e-01
-2.17871442e-01 -7.38630354e-01 -1.34807098e+00 5.96946180e-01
-6.77432239e-01 3.98452252e-01 9.48475599e-01 6.47166491e-01
2.34301776e-01 1.03278410e+00 9.41643655e-01 -4.98464257e-01
-5.09908915e-01 -1.15132010e+00 -9.76734161e-01 2.74824291e-01
7.17233241e-01 -7.33938992e-01 -7.03828156e-01 3.22427064e-01] | [15.171523094177246, 5.328629016876221] |
4f1d404d-6360-4e5d-93b0-fa4024f4713c | a-peer-to-peer-federated-continual-learning | 2306.02037 | null | https://arxiv.org/abs/2306.02037v1 | https://arxiv.org/pdf/2306.02037v1.pdf | A Peer-to-peer Federated Continual Learning Network for Improving CT Imaging from Multiple Institutions | Deep learning techniques have been widely used in computed tomography (CT) but require large data sets to train networks. Moreover, data sharing among multiple institutions is limited due to data privacy constraints, which hinders the development of high-performance DL-based CT imaging models from multi-institutional collaborations. Federated learning (FL) strategy is an alternative way to train the models without centralizing data from multi-institutions. In this work, we propose a novel peer-to-peer federated continual learning strategy to improve low-dose CT imaging performance from multiple institutions. The newly proposed method is called peer-to-peer continual FL with intermediate controllers, i.e., icP2P-FL. Specifically, different from the conventional FL model, the proposed icP2P-FL does not require a central server that coordinates training information for a global model. In the proposed icP2P-FL method, the peer-to-peer federated continual learning is introduced wherein the DL-based model is continually trained one client after another via model transferring and inter institutional parameter sharing due to the common characteristics of CT data among the clients. Furthermore, an intermediate controller is developed to make the overall training more flexible. Numerous experiments were conducted on the AAPM low-dose CT Grand Challenge dataset and local datasets, and the experimental results showed that the proposed icP2P-FL method outperforms the other comparative methods both qualitatively and quantitatively, and reaches an accuracy similar to a model trained with pooling data from all the institutions. | ['Jianhua Ma', 'Dong Zeng', 'Zhaoying Bian', 'XiaoYu Zhang', 'Ruihong He', 'Hao Wang'] | 2023-06-03 | null | null | null | null | ['computed-tomography-ct'] | ['methodology'] | [-5.87760329e-01 -1.42460465e-01 -2.86494941e-01 -5.09512305e-01
-1.15947926e+00 -3.06751788e-01 1.79526612e-01 2.39508316e-01
-6.56369507e-01 9.11580563e-01 -1.37281045e-02 -3.28360677e-01
-5.89714706e-01 -7.03551471e-01 -7.19230056e-01 -1.13430274e+00
-1.13032117e-01 6.85857534e-01 1.87679857e-01 4.42594141e-01
-2.11643934e-01 7.86968291e-01 -6.68389738e-01 4.05624777e-01
7.67730892e-01 1.04199374e+00 2.72154480e-01 2.02547997e-01
-1.40758082e-01 1.12414885e+00 -2.55687714e-01 -2.59794444e-01
6.61093712e-01 -9.28139538e-02 -7.44559526e-01 -1.53770298e-01
1.47306337e-03 -5.91777563e-01 -3.41174692e-01 6.58249617e-01
7.35653877e-01 1.79954976e-01 1.93850115e-01 -1.45916450e+00
-1.46052152e-01 3.54033530e-01 -4.26023304e-01 1.01689197e-01
-2.10742682e-01 2.41254598e-01 3.77416790e-01 -5.87740898e-01
6.96748376e-01 6.82677031e-01 6.59686208e-01 6.24507964e-01
-9.11419570e-01 -1.06488740e+00 -6.28612414e-02 8.21985975e-02
-1.36388445e+00 4.52880887e-03 5.48420906e-01 -3.49410027e-01
4.44561392e-01 2.01659441e-01 5.28584123e-01 7.03387856e-01
3.62630546e-01 5.93362033e-01 1.29485190e+00 -1.42141566e-01
3.12107742e-01 3.21920961e-01 -1.20896652e-01 7.00904489e-01
1.30931482e-01 -9.72293168e-02 -3.71701479e-01 -7.25369930e-01
1.10739565e+00 3.66416156e-01 -5.32582760e-01 -6.47447288e-01
-1.05663145e+00 6.93790376e-01 8.03428829e-01 4.24677044e-01
-6.48996413e-01 -3.21283042e-02 7.28059232e-01 3.08619291e-01
2.90909290e-01 -1.94849089e-01 -4.92276430e-01 2.48177797e-01
-8.32723558e-01 -6.47007525e-02 6.59439743e-01 8.47725153e-01
6.77445412e-01 -4.17181164e-01 -3.62758450e-02 6.72863364e-01
1.33636147e-01 5.68251796e-02 7.20339894e-01 -9.17309105e-01
6.22034907e-01 5.37365735e-01 -1.30250514e-01 -6.88390195e-01
-3.84377450e-01 -4.07915652e-01 -1.16041446e+00 -5.21204509e-02
1.80708677e-01 -5.05197346e-01 -5.75817406e-01 1.58605385e+00
6.55174732e-01 3.33583713e-01 9.98005737e-03 8.42950761e-01
7.37862229e-01 4.90311801e-01 2.65469581e-01 -3.55667412e-01
8.94530833e-01 -1.05411255e+00 -5.67437410e-01 4.49961603e-01
7.06255198e-01 -3.90267372e-01 5.73512375e-01 4.54622507e-01
-7.75000930e-01 -2.09226087e-01 -7.03002393e-01 3.24826568e-01
-7.73609206e-02 -8.82479176e-02 6.52734935e-01 5.79119384e-01
-1.16383135e+00 4.46464300e-01 -1.13034809e+00 -2.47068822e-01
8.36186826e-01 7.24879086e-01 -7.78125226e-01 -4.58569109e-01
-8.98485124e-01 6.32963359e-01 2.52451450e-01 -4.38008383e-02
-1.13009548e+00 -1.06057823e+00 -3.26538295e-01 1.37722671e-01
3.12925756e-01 -7.21803844e-01 1.32174897e+00 -7.12253213e-01
-1.46830249e+00 3.95439208e-01 1.50466099e-01 -4.04352248e-01
9.59925830e-01 2.82921314e-01 -2.93166339e-01 3.64252388e-01
-1.09248441e-02 4.02209342e-01 3.90884101e-01 -1.28077543e+00
-5.25009871e-01 -3.25331748e-01 -2.58323431e-01 2.59737700e-01
-5.81794977e-01 1.45486385e-01 -4.34848994e-01 -2.37249896e-01
5.27710058e-02 -8.25578213e-01 -4.61120456e-01 4.54192370e-01
-5.78369498e-02 -2.54096597e-01 1.08467984e+00 -3.43606353e-01
9.17647839e-01 -2.01366448e+00 -3.87781113e-01 5.21243691e-01
4.99000311e-01 1.29219621e-01 1.05813466e-01 3.23125362e-01
-2.63550162e-01 6.87355250e-02 -8.34320262e-02 -4.70686316e-01
-3.41503918e-01 5.24705589e-01 4.90391523e-01 5.38019359e-01
-4.78730351e-01 5.48273265e-01 -7.12763906e-01 -1.07756293e+00
1.38820559e-02 3.55043322e-01 -6.05982006e-01 2.56059468e-01
1.57817692e-01 1.03239739e+00 -9.03768659e-01 6.12444699e-01
9.30511594e-01 -5.77639461e-01 3.41278881e-01 -1.42121449e-01
-1.94498658e-01 -3.27620178e-01 -1.01273787e+00 1.80519450e+00
-5.95596194e-01 -5.17329164e-02 6.19861841e-01 -9.31714177e-01
7.35187054e-01 1.06845188e+00 1.23925710e+00 -4.50678766e-01
3.45238507e-01 5.35700560e-01 -1.75482094e-01 -4.80268449e-01
-5.46193086e-02 -4.32356000e-01 1.63986892e-01 6.65864944e-01
1.56446800e-01 1.22702286e-01 -4.25099909e-01 2.19048321e-01
1.13716781e+00 -5.15430689e-01 -1.01490021e-02 -1.61742970e-01
7.05560267e-01 -2.87018437e-02 1.03885198e+00 7.73417890e-01
-6.71362102e-01 4.39736724e-01 1.53973907e-01 -6.00829661e-01
-6.64200366e-01 -8.24390709e-01 -2.78187037e-01 7.14069366e-01
-1.46240503e-01 -1.19683042e-01 -6.68438315e-01 -1.06821585e+00
-1.07119447e-02 2.46311396e-01 -3.46704841e-01 1.91529542e-01
-6.34738564e-01 -5.74245393e-01 4.79515553e-01 4.34612870e-01
9.15481627e-01 -8.37502658e-01 -3.39057863e-01 4.12354022e-01
-3.14102560e-01 -7.98699796e-01 -5.53007960e-01 1.92256793e-01
-1.10073662e+00 -1.15226305e+00 -8.56372774e-01 -6.62635505e-01
7.41369367e-01 1.89086199e-01 5.73023319e-01 1.28724217e-01
-1.81801170e-01 5.23235500e-01 -1.79175168e-01 -5.09999335e-01
-2.72192538e-01 1.08577259e-01 1.36574833e-02 3.88844401e-01
2.37392467e-02 -4.66572106e-01 -9.15249228e-01 5.55084407e-01
-1.03735685e+00 -1.18425883e-01 5.37421703e-01 8.18920195e-01
7.75224984e-01 5.31341247e-02 8.84079158e-01 -1.12160325e+00
4.79873121e-01 -8.36490214e-01 -5.88824034e-01 4.67546105e-01
-7.02910781e-01 -5.68860948e-01 9.41135585e-01 -2.28461549e-01
-1.36126685e+00 3.70068789e-01 1.65926561e-01 -8.06280077e-01
1.78573169e-02 5.61338663e-01 -6.29229993e-02 -6.51191235e-01
5.04154265e-01 1.17032737e-01 2.54331887e-01 -6.33665979e-01
-1.23467147e-01 8.12848985e-01 2.29578048e-01 -7.71590769e-01
4.81943399e-01 4.58750278e-01 1.91082619e-02 -1.48751542e-01
-2.57257134e-01 -3.45956862e-01 -4.90959346e-01 -3.50175798e-01
7.63943076e-01 -1.04229999e+00 -8.61465275e-01 6.48342669e-01
-1.10079825e+00 -1.54423118e-01 -1.79173976e-01 1.04837084e+00
-4.40576017e-01 7.91849494e-02 -8.50695670e-01 -2.94838369e-01
-7.55586326e-01 -1.37872994e+00 2.69182146e-01 3.87497187e-01
3.74643117e-01 -9.34317410e-01 9.42756981e-02 6.24285877e-01
9.00860906e-01 3.91182035e-01 8.86022627e-01 -7.98009932e-01
-8.59238505e-01 -3.87396336e-01 -1.83713049e-01 4.38592166e-01
2.81198949e-01 -4.18144256e-01 -7.19768465e-01 -6.34886324e-01
3.99665922e-01 -4.99149740e-01 -5.40429577e-02 3.10866505e-01
1.71257329e+00 -3.38670790e-01 -3.46784651e-01 7.19907105e-01
1.69717777e+00 3.32969427e-01 4.02942926e-01 4.03657943e-01
6.65772557e-01 2.77256459e-01 4.32689756e-01 6.33248270e-01
4.22452241e-01 8.54844078e-02 4.27857339e-01 -3.28183889e-01
2.41175294e-01 -3.64464894e-03 -3.18946242e-01 9.71354604e-01
-5.19965477e-02 1.24697655e-01 -1.06227851e+00 4.27263677e-01
-1.83687544e+00 -4.85700220e-01 7.86831230e-02 2.24352050e+00
6.76627398e-01 -3.89448136e-01 -1.90510392e-01 -4.93986756e-01
8.19647253e-01 -2.18894809e-01 -8.15720797e-01 -1.39042586e-01
1.70486301e-01 2.78598338e-01 7.29498625e-01 -1.30890161e-01
-8.11649323e-01 3.23999286e-01 5.34774303e+00 8.72457206e-01
-1.45839012e+00 7.96607196e-01 7.92609870e-01 -3.58852714e-01
1.78923890e-01 -8.29230696e-02 -2.53206044e-01 3.04886609e-01
8.43603492e-01 -5.95032930e-01 1.02469094e-01 1.03592324e+00
3.40540141e-01 -6.15676213e-03 -1.05416250e+00 9.60533500e-01
-4.00369167e-01 -1.59791684e+00 -7.81503320e-02 3.07987958e-01
8.62380624e-01 5.39169490e-01 -1.68191835e-01 1.67849258e-01
3.79692435e-01 -7.02016056e-01 7.47549683e-02 5.62911808e-01
7.36427367e-01 -9.18157518e-01 7.57042766e-01 5.92593014e-01
-1.16539347e+00 -2.53455311e-01 -3.01278174e-01 6.89702213e-01
-5.94198368e-02 2.88070381e-01 -7.31994510e-01 1.31711483e+00
9.07595575e-01 3.93931568e-01 -2.84760684e-01 1.45974457e+00
2.80102670e-01 5.30771136e-01 -5.03925502e-01 3.41034293e-01
3.70554388e-01 8.02685916e-02 1.05609164e-01 7.17566371e-01
3.70336413e-01 4.18414772e-01 3.95470589e-01 4.83934492e-01
-5.20366251e-01 5.21640182e-01 -2.84932226e-01 5.29604733e-01
6.12113774e-01 1.50680208e+00 -3.80985022e-01 -3.62886161e-01
-5.58517635e-01 4.45349365e-01 3.27634543e-01 -5.09243309e-02
-7.20315933e-01 -5.27619161e-02 2.40445688e-01 2.32681975e-01
1.44502386e-01 3.90560403e-02 1.48174569e-01 -1.07054055e+00
-1.21549226e-01 -9.48762298e-01 9.48977590e-01 -7.05958486e-01
-1.71066093e+00 7.57674813e-01 2.08750535e-02 -1.46694434e+00
1.13679148e-01 -1.14359416e-01 -8.54929686e-01 9.24333930e-01
-1.68184614e+00 -1.22749567e+00 -4.93518680e-01 1.57040620e+00
-4.85271476e-02 -1.89335138e-01 9.50544298e-01 7.09120214e-01
-7.24153101e-01 8.44606280e-01 3.97610664e-01 3.08727503e-01
9.02977347e-01 -6.77631795e-01 -7.35738873e-01 4.94159758e-01
-6.40384614e-01 6.81913912e-01 -1.90507591e-01 -5.28772831e-01
-1.35712135e+00 -1.38420999e+00 3.56223822e-01 2.01493248e-01
4.00679141e-01 7.70387100e-03 -1.10221982e+00 8.93018305e-01
2.67708808e-01 8.17153990e-01 1.04100823e+00 -2.72144139e-01
5.73062822e-02 -6.46206319e-01 -1.83329988e+00 5.08574769e-02
4.10345554e-01 -3.32392991e-01 -8.85191467e-03 7.47644007e-01
6.53026998e-01 -4.68066096e-01 -1.35849690e+00 2.00612366e-01
1.35686010e-01 -6.89014733e-01 6.02970898e-01 -3.58074635e-01
4.47109006e-02 -1.66305095e-01 -1.58017986e-02 -1.17852271e+00
-1.77583799e-01 -5.29668868e-01 3.70971322e-01 1.09762526e+00
9.84025747e-02 -1.26842880e+00 1.18660975e+00 1.20022011e+00
-1.97306484e-01 -8.17155659e-01 -1.58268297e+00 -8.43167901e-01
4.89927769e-01 -1.27686486e-01 9.75261092e-01 1.33252156e+00
-2.40576729e-01 -5.19959986e-01 -1.46459788e-01 2.91676164e-01
9.74740982e-01 -1.34095877e-01 7.03224301e-01 -9.55989897e-01
-5.24711668e-01 9.82975438e-02 -7.48643428e-02 -4.93576258e-01
-1.92169264e-01 -1.10711622e+00 -2.52978206e-01 -1.27509356e+00
3.43957126e-01 -1.22303832e+00 -8.80300105e-01 8.89247537e-01
2.45518535e-01 -1.65602401e-01 4.03843969e-01 6.78760648e-01
-5.59111893e-01 5.24241269e-01 1.62520087e+00 1.17322005e-01
4.67151478e-02 1.11512922e-01 -5.07646799e-01 4.16844338e-01
8.68692398e-01 -9.99006748e-01 -4.39935654e-01 -5.55233359e-01
-5.52324831e-01 7.29740620e-01 4.01537567e-01 -1.11983764e+00
9.95825231e-01 -1.88071638e-01 3.74447942e-01 -5.82772732e-01
1.26429707e-01 -1.40346289e+00 5.44286430e-01 5.96862674e-01
-9.42145381e-03 -5.21691740e-02 -8.11905693e-03 5.85188389e-01
-2.93578178e-01 -7.89918900e-02 8.75712752e-01 -5.78288496e-01
-6.14891052e-02 8.93462420e-01 -1.60828292e-01 -2.34180763e-01
1.32932281e+00 1.44492462e-01 -3.32683891e-01 -2.73316711e-01
-7.52060711e-01 6.67706788e-01 1.42287537e-01 -8.68834779e-02
4.18636709e-01 -1.24017787e+00 -7.11690784e-01 5.54512963e-02
-9.36971381e-02 5.06720304e-01 8.24306130e-01 1.15801430e+00
-5.83586931e-01 4.05687004e-01 -2.28072628e-01 -7.66634345e-01
-1.14712977e+00 3.72567892e-01 7.06219256e-01 -6.99233770e-01
-6.66123390e-01 6.37417376e-01 1.35689437e-01 -1.01040685e+00
2.60369807e-01 -6.68389781e-04 4.07623529e-01 -3.24981868e-01
3.02390337e-01 2.51234889e-01 5.52203119e-01 -2.28915855e-01
-4.97556120e-01 1.62766442e-01 -4.60400909e-01 1.05923019e-01
1.80554402e+00 5.32537140e-02 -3.21107686e-01 4.52958532e-02
1.53842688e+00 -4.02542025e-01 -1.15973532e+00 -5.81169009e-01
-4.85664904e-01 -6.69648170e-01 2.14346275e-01 -8.89448225e-01
-1.82256365e+00 4.67320830e-01 7.30313897e-01 -4.61227059e-01
1.27360427e+00 -3.13920647e-01 1.02528787e+00 2.91576207e-01
9.28822398e-01 -8.91725302e-01 -9.57122669e-02 -1.61843114e-02
6.97689116e-01 -1.30937207e+00 1.25525862e-01 -1.38223648e-01
-6.25585437e-01 1.13773882e+00 9.30588543e-01 6.24510795e-02
1.08562970e+00 1.68442935e-01 3.14810187e-01 -1.86242387e-01
-7.50641167e-01 7.71496773e-01 -2.95187563e-01 3.08724225e-01
2.57287957e-02 7.25460276e-02 -2.57761866e-01 7.33703792e-01
2.72430062e-01 5.94067514e-01 3.93760860e-01 1.44457495e+00
7.13705644e-02 -1.41456413e+00 -5.64429402e-01 5.10204494e-01
-4.37988967e-01 2.83486485e-01 1.83222592e-01 8.93140733e-01
2.99031466e-01 7.61798561e-01 -2.00153887e-01 -1.03327319e-01
3.33065182e-01 1.88454222e-02 9.37336013e-02 -4.19590801e-01
-1.32042050e+00 1.32669300e-01 -5.31756759e-01 -5.92299998e-01
-3.54323804e-01 -6.66719079e-01 -1.40579677e+00 -5.29008925e-01
-3.37978512e-01 3.43485087e-01 1.03248405e+00 6.17853761e-01
5.12006998e-01 4.62863237e-01 1.28134489e+00 -6.19274437e-01
-5.43859124e-01 -6.45746112e-01 -5.62841773e-01 1.44759461e-01
1.48982108e-01 -3.50886643e-01 -2.68801868e-01 -4.31984514e-01] | [6.04835319519043, 6.465025424957275] |
62032082-d6cf-4661-90fe-e7286859412f | epilnet-a-novel-approach-to-iot-based | 2111.03265 | null | https://arxiv.org/abs/2111.03265v1 | https://arxiv.org/pdf/2111.03265v1.pdf | EpilNet: A Novel Approach to IoT based Epileptic Seizure Prediction and Diagnosis System using Artificial Intelligence | Epilepsy is one of the most occurring neurological diseases. The main characteristic of this disease is a frequent seizure, which is an electrical imbalance in the brain. It is generally accompanied by shaking of body parts and even leads (fainting). In the past few years, many treatments have come up. These mainly involve the use of anti-seizure drugs for controlling seizures. But in 70% of cases, these drugs are not effective, and surgery is the only solution when the condition worsens. So patients need to take care of themselves while having a seizure and be safe. Wearable electroencephalogram (EEG) devices have come up with the development in medical science and technology. These devices help in the analysis of brain electrical activities. EEG helps in locating the affected cortical region. The most important is that it can predict any seizure in advance on-site. This has resulted in a sudden increase in demand for effective and efficient seizure prediction and diagnosis systems. A novel approach to epileptic seizure prediction and diagnosis system EpilNet is proposed in the present paper. It is a one-dimensional (1D) convolution neural network. EpilNet gives the testing accuracy of 79.13% for five classes, leading to a significant increase of about 6-7% compared to related works. The developed Web API helps in bringing EpilNet into practical use. Thus, it is an integrated system for both patients and doctors. The system will help patients prevent injury or accidents and increase the efficiency of the treatment process by doctors in the hospitals. | ['Priyansh Agrawal', 'Virender Ranga', 'Shivam Gupta'] | 2021-11-05 | null | null | null | null | ['seizure-prediction'] | ['medical'] | [-3.27293932e-01 -2.53768057e-01 2.79741913e-01 -1.51640132e-01
1.79187506e-01 -2.03910708e-01 4.73595336e-02 1.34685516e-01
-3.14652473e-01 9.11155760e-01 2.95483544e-02 -2.21283715e-02
-2.95543015e-01 -7.46482432e-01 -6.17900118e-02 -9.72682297e-01
-1.99464306e-01 2.88845360e-01 2.12495685e-01 -1.42671615e-01
2.91608274e-01 6.76906407e-01 -1.38589501e+00 7.76184648e-02
1.05403662e+00 1.04414618e+00 4.89880145e-01 9.85014811e-02
-1.81884676e-01 5.34392595e-01 -7.23446488e-01 5.15952520e-02
1.13530224e-02 -4.58954364e-01 -4.20706153e-01 -1.73103824e-01
-7.28643358e-01 -2.46650726e-01 -1.61341310e-01 1.12149823e+00
8.86350155e-01 -3.53623539e-01 4.29972768e-01 -1.05903339e+00
-2.17842579e-01 3.73886794e-01 -4.74845946e-01 3.61647308e-01
2.29133680e-01 -4.45184708e-02 -1.60459280e-01 -6.06021583e-01
-3.70641090e-02 5.56823313e-01 5.92215240e-01 6.01253688e-01
-5.95891953e-01 -9.96648371e-01 -4.88246232e-01 5.80074668e-01
-1.59284735e+00 -2.16186896e-01 6.40996158e-01 -4.57092524e-01
1.06933320e+00 2.36331120e-01 1.09304571e+00 7.35189497e-01
7.57138014e-01 2.97374576e-01 1.16995180e+00 -1.71898007e-01
2.09981829e-01 1.33836478e-01 9.11269709e-02 9.66348201e-02
4.13726568e-01 -1.72512338e-01 -5.34400046e-01 6.00568429e-02
5.69019973e-01 5.00147045e-01 -7.20722318e-01 2.57967383e-01
-9.76317346e-01 3.51237893e-01 3.31793487e-01 9.59483624e-01
-7.43553817e-01 -2.00571448e-01 4.02636647e-01 1.75953746e-01
3.06887895e-01 2.74325341e-01 -5.12257159e-01 -5.74797571e-01
-8.41011822e-01 2.43941676e-02 6.84340954e-01 3.55798274e-01
1.08268589e-01 7.77824968e-02 2.67022669e-01 6.05128467e-01
-9.76468995e-02 4.04391974e-01 1.11003494e+00 8.72017350e-03
1.33460894e-01 8.83986652e-01 -7.52132945e-03 -9.25178468e-01
-7.47757375e-01 -6.32960081e-01 -1.06827486e+00 1.98505580e-01
9.76948738e-02 -3.60341728e-01 -7.17247188e-01 1.37073314e+00
-3.21878493e-02 2.69062757e-01 1.32173896e-01 6.51183605e-01
8.74435723e-01 6.96797490e-01 2.87154149e-02 -4.23929542e-01
1.51564300e+00 -3.18260282e-01 -9.71443832e-01 -4.33416888e-02
4.34988886e-01 -7.73884654e-01 6.22542858e-01 8.69401991e-01
-7.74371088e-01 -1.52111053e-01 -9.67051566e-01 6.25031531e-01
-4.85044926e-01 7.42258281e-02 4.51899558e-01 4.70853508e-01
-9.48669434e-01 5.07858694e-01 -1.02672803e+00 -6.46067321e-01
2.89840788e-01 7.52903283e-01 -3.74249518e-01 1.56417668e-01
-1.13412058e+00 1.28845978e+00 4.44964498e-01 1.15715750e-01
-3.41655403e-01 -5.04871130e-01 -2.09860608e-01 6.45591989e-02
-1.11371346e-01 -2.72462696e-01 8.29669654e-01 -7.79265821e-01
-1.13301504e+00 5.53579390e-01 -4.99331057e-02 -4.41154391e-01
3.83835196e-01 -1.49443656e-01 -8.08886349e-01 -2.09167656e-02
-1.80150554e-01 2.55763292e-01 3.10963154e-01 -3.87134284e-01
-6.46154642e-01 -5.91300309e-01 -5.03955245e-01 2.00890630e-01
-4.47529405e-01 2.96370327e-01 6.75798282e-02 -4.62782025e-01
1.15915813e-01 -5.29921412e-01 1.63257375e-01 -5.24604619e-01
-1.82475090e-01 -3.90413225e-01 1.21233356e+00 -9.04405355e-01
1.34571385e+00 -2.04136038e+00 -1.43719897e-01 1.20084500e-02
2.34194264e-01 4.96784598e-01 5.58720410e-01 3.85235906e-01
-4.30767387e-01 -1.26658395e-01 -1.32771626e-01 3.05167288e-01
-4.89143610e-01 -6.66741002e-03 1.50179127e-02 3.83126855e-01
2.07758811e-03 5.02711952e-01 -4.18726504e-01 -1.02223895e-01
2.43009523e-01 8.53320181e-01 3.16781141e-02 2.34372750e-01
3.51336151e-01 7.75080860e-01 -5.46904206e-01 6.21607125e-01
6.14941001e-01 -3.19579653e-02 -2.62939751e-01 1.74653128e-01
-3.51815611e-01 1.73965752e-01 -1.02413583e+00 1.19473851e+00
-1.75391808e-01 5.69511235e-01 -2.92816401e-01 -1.21850491e+00
9.71484244e-01 9.70276475e-01 5.59937119e-01 -7.80619025e-01
4.48955774e-01 5.31031191e-01 3.49478543e-01 -1.11441803e+00
-5.43094695e-01 -4.76612337e-02 3.78623456e-01 2.58748978e-01
-4.43911821e-01 1.46659181e-01 9.18933004e-02 -2.11147383e-01
1.07500744e+00 -1.80880025e-01 6.31246626e-01 -3.86149794e-01
5.66184163e-01 -3.21933627e-01 7.15128183e-01 -1.15474440e-01
-2.90281437e-02 1.89157426e-02 2.16959506e-01 -7.26808250e-01
-6.17313623e-01 -6.10748887e-01 -5.81908107e-01 9.81368776e-03
1.19048201e-01 -5.55617772e-02 -9.70556140e-01 -7.76068866e-02
-3.15413922e-01 4.55431223e-01 -2.43635088e-01 -1.89949512e-01
-2.79336929e-01 -8.87864232e-01 3.26093882e-01 3.95518720e-01
1.14515769e+00 -1.43564439e+00 -1.15082526e+00 4.84019309e-01
-1.24916673e-01 -6.60054982e-01 1.05329491e-01 3.72098535e-01
-1.02041650e+00 -9.46705222e-01 -8.85701299e-01 -7.76394129e-01
6.36129737e-01 -1.80185348e-01 5.50357759e-01 2.98225045e-01
-5.17350554e-01 -3.26669246e-01 -3.19873184e-01 -7.57204652e-01
1.81894526e-02 -5.56498058e-02 4.54957157e-01 -1.27865627e-01
8.89304578e-01 -1.10274756e+00 -8.38419795e-01 1.43411979e-01
-6.49774015e-01 1.12614572e-01 5.99990964e-01 4.42459792e-01
2.05790654e-01 3.55740964e-01 9.35743034e-01 -2.99406052e-01
9.26963031e-01 -5.82994223e-01 -5.51216185e-01 1.29257575e-01
-6.85521364e-01 -1.51928216e-01 6.79213583e-01 -3.33268821e-01
-5.78281105e-01 -1.60543948e-01 -2.24519163e-01 -3.61981094e-02
-5.43345988e-01 3.95434022e-01 -1.86075628e-01 -1.77022815e-01
3.92840147e-01 4.92895275e-01 -3.00212473e-01 -5.22240341e-01
-7.91073561e-01 1.20270193e+00 4.13065046e-01 1.10869735e-01
1.74216181e-01 -7.89097026e-02 2.27361321e-02 -9.00408566e-01
-1.16834365e-01 -3.44538629e-01 -1.29194289e-01 -2.45747313e-01
9.92089987e-01 -7.23218143e-01 -9.04431820e-01 8.24854612e-01
-1.19470263e+00 1.04610279e-01 4.20599639e-01 8.62017572e-01
-6.32115752e-02 -1.57602709e-02 -4.36784327e-01 -9.22512650e-01
-7.57192016e-01 -1.31323159e+00 2.52000123e-01 6.82139754e-01
-3.25060070e-01 -6.76751852e-01 -9.87220854e-02 -2.94172317e-01
7.56253898e-01 3.22199941e-01 7.55413592e-01 -7.13521421e-01
-3.91993940e-01 -3.59011471e-01 -1.10262185e-02 2.92441666e-01
4.70001608e-01 -3.51382643e-01 -9.11757708e-01 -6.53315857e-02
4.93228197e-01 5.39091267e-02 2.64120817e-01 5.68094075e-01
1.25800812e+00 -9.79050994e-02 -5.15601933e-01 6.10251725e-01
1.52346849e+00 1.26083148e+00 8.60087752e-01 2.88316250e-01
1.95171908e-01 2.67500222e-01 2.00179711e-01 4.46849436e-01
-4.95972708e-02 4.33090121e-01 3.71264607e-01 -7.22687840e-02
1.25473559e-01 2.90678799e-01 1.77763775e-01 9.87429380e-01
-3.90627712e-01 -1.26618057e-01 -1.11015844e+00 5.07683456e-01
-1.47182596e+00 -9.96693134e-01 -3.58774036e-01 2.35353756e+00
8.20834100e-01 -4.22628969e-02 -9.33527946e-02 5.25406301e-01
6.76263332e-01 -6.90879583e-01 -5.16420305e-01 -4.11970854e-01
1.29150286e-01 6.15980983e-01 3.68618965e-01 1.19019439e-02
-6.94561422e-01 4.76080239e-01 5.69044971e+00 5.83499730e-01
-1.84411085e+00 1.24821596e-01 4.52878267e-01 6.00117305e-03
3.65031660e-01 -2.78226107e-01 -5.71208596e-01 1.08709300e+00
7.99545586e-01 -3.57500523e-01 4.95389491e-01 7.25695372e-01
4.81131911e-01 -3.88822526e-01 -7.40196705e-01 1.44931912e+00
-9.77228358e-02 -9.49510574e-01 -4.41750109e-01 3.58400680e-02
3.12456757e-01 2.87470780e-02 -2.31507123e-01 -1.33544937e-01
-4.88597751e-01 -1.16053092e+00 1.10689849e-01 7.04639435e-01
8.26094985e-01 -1.17141068e+00 1.16664636e+00 6.32341743e-01
-9.58661854e-01 -4.41640168e-02 -9.53713432e-02 -3.67889643e-01
9.45730135e-02 6.54150188e-01 -6.79721236e-01 1.33742720e-01
1.14684606e+00 5.42282522e-01 -1.57367706e-01 1.60855663e+00
-3.64561528e-01 5.11137247e-01 -4.14415807e-01 -3.35848182e-01
-1.15345649e-01 -2.60993749e-01 4.44800377e-01 7.08650351e-01
8.06911111e-01 5.75383306e-01 -3.00566673e-01 5.25733948e-01
1.49482429e-01 1.91097707e-01 -6.08431935e-01 1.05034150e-01
3.06005657e-01 1.20089161e+00 -8.93610954e-01 -1.68190330e-01
-2.13434786e-01 9.47092175e-01 -2.18210906e-01 8.18013847e-02
-6.52719557e-01 -9.56039310e-01 3.72809559e-01 2.23005757e-01
-2.54424810e-01 1.67737395e-01 -3.91855389e-01 -9.06165421e-01
1.77706152e-01 -8.00210297e-01 -1.54810145e-01 -8.17620814e-01
-7.69271076e-01 8.41114998e-01 -1.54313758e-01 -1.16012847e+00
-2.47465149e-01 -4.59658802e-01 -9.50191438e-01 1.08300543e+00
-1.16167164e+00 -5.56202710e-01 -4.71971691e-01 8.00410926e-01
5.10037184e-01 -2.04909250e-01 1.15356755e+00 5.49048603e-01
-6.10535085e-01 1.38157174e-01 -3.08818575e-02 1.67290922e-02
6.56347215e-01 -9.16766405e-01 -3.01918536e-01 8.42326522e-01
-4.63943183e-01 5.66977680e-01 8.00374031e-01 -6.42823815e-01
-8.45787108e-01 -5.29755235e-01 1.19694054e+00 1.46679118e-01
3.53897840e-01 -1.38198376e-01 -8.97898555e-01 2.64907300e-01
5.45018792e-01 -3.06321651e-01 6.45560145e-01 -4.25957263e-01
6.16599560e-01 -3.78291965e-01 -1.34772861e+00 4.14433002e-01
5.98690748e-01 -1.18330553e-01 -7.16152787e-01 2.89052874e-01
1.03992417e-01 -2.01316565e-01 -5.46454787e-01 4.08178329e-01
4.23206091e-01 -1.06802034e+00 3.41715217e-01 -1.54903412e-01
1.93513036e-01 -2.63597459e-01 4.20945972e-01 -1.51420391e+00
-3.22669856e-02 -4.83475238e-01 2.51560122e-01 8.52999747e-01
1.99722260e-01 -1.07404506e+00 5.23344159e-01 7.61363149e-01
-1.08000934e-01 -1.01811993e+00 -6.78923070e-01 -5.47150373e-01
-3.32720160e-01 -1.87376902e-01 7.89027452e-01 8.44316363e-01
5.72630107e-01 3.64721157e-02 -1.68829918e-01 1.01593688e-01
3.72898936e-01 -2.72116542e-01 9.39272270e-02 -1.49795377e+00
1.91966500e-02 -2.70067424e-01 -8.40177774e-01 -3.34203392e-01
-4.65189368e-01 -5.97617328e-01 -4.42542374e-01 -1.79077625e+00
1.45375773e-01 -1.80944204e-01 -3.78020108e-01 7.39694357e-01
3.08878273e-01 1.49347439e-01 -3.91706169e-01 1.12242483e-01
9.36538801e-02 1.42361760e-01 1.02099645e+00 1.92696095e-01
-4.56728905e-01 2.43405953e-01 -5.83148420e-01 8.85007858e-01
1.33659983e+00 -4.99869704e-01 -4.58447725e-01 -3.25947851e-01
2.33473241e-01 2.06253633e-01 4.02301699e-02 -1.43347383e+00
5.66514790e-01 1.12086803e-01 7.62944639e-01 -6.00633979e-01
1.27836466e-01 -1.13987637e+00 5.70460200e-01 9.61301744e-01
4.18865740e-01 3.79546732e-01 2.33578056e-01 -5.26365191e-02
-3.50369155e-01 -8.46443549e-02 7.64262080e-01 1.19388988e-02
-6.17477834e-01 2.33826309e-01 -5.95676422e-01 -4.69468296e-01
1.52644575e+00 -4.52571392e-01 -1.02721505e-01 -2.79445231e-01
-8.21346521e-01 7.50740021e-02 4.13198061e-02 1.43240586e-01
6.50415242e-01 -1.12730324e+00 -4.78681862e-01 3.78926784e-01
2.78883763e-02 -1.49834976e-01 3.29868346e-01 1.32049227e+00
-9.29346204e-01 5.01211345e-01 -6.27221882e-01 -2.98750341e-01
-1.43394482e+00 2.04361334e-01 5.17405391e-01 1.26893986e-02
-8.97517800e-01 7.15025783e-01 -6.21434934e-02 4.33413148e-01
4.25513864e-01 -3.72829109e-01 -6.31423056e-01 -2.59348840e-01
9.86360550e-01 3.08779657e-01 4.20260668e-01 -2.78137714e-01
-5.74912965e-01 5.68887889e-01 5.50318062e-02 1.69650912e-01
1.64622009e+00 2.84972608e-01 -4.88461852e-01 2.73850650e-01
7.50870347e-01 -4.10537086e-02 -5.81536710e-01 4.00245577e-01
-2.11957544e-01 -1.52764440e-01 2.00940892e-01 -1.28040349e+00
-1.21057165e+00 9.68995154e-01 1.00238431e+00 3.35699022e-01
1.61179900e+00 -2.41035312e-01 8.58590126e-01 2.94104636e-01
8.35202515e-01 -8.78616750e-01 -4.68695462e-01 1.72379553e-01
7.99771845e-01 -7.07600117e-01 -3.30036700e-01 -6.22650050e-02
-3.90826315e-01 1.30039334e+00 3.32814336e-01 -1.99590191e-01
8.91700685e-01 6.69679403e-01 -1.56462211e-02 -3.54459018e-01
-3.88911605e-01 1.63388178e-01 -2.63305306e-02 4.83306378e-01
6.65341616e-01 1.22310475e-01 -9.99691010e-01 9.75771844e-01
-3.05473477e-01 4.74525899e-01 1.95332080e-01 8.15075517e-01
-6.26353264e-01 -1.21364355e+00 -5.05258381e-01 6.44382060e-01
-9.55528736e-01 -8.50671530e-02 -2.13348791e-01 7.55802155e-01
6.38444364e-01 1.09041488e+00 4.45031933e-02 -3.68954092e-01
2.28296891e-01 2.55080193e-01 3.25451434e-01 -2.05956295e-01
-5.28017223e-01 1.11934100e-03 -3.30416173e-01 -4.20471698e-01
-2.56694466e-01 -4.09667581e-01 -1.63281310e+00 -2.97196358e-01
-2.81243056e-01 4.71877545e-01 1.03416228e+00 1.00276709e+00
3.63205791e-01 6.69410527e-01 4.54872280e-01 -2.98870265e-01
-1.03693984e-01 -1.17655480e+00 -1.02455616e+00 6.23217672e-02
2.74889395e-02 -7.85599530e-01 -3.09574038e-01 -1.25953659e-01] | [13.270492553710938, 3.469743013381958] |
3c34b03d-2597-4c70-a133-3381640c7830 | deep-unsupervised-key-frame-extraction-for | 2211.06742 | null | https://arxiv.org/abs/2211.06742v1 | https://arxiv.org/pdf/2211.06742v1.pdf | Deep Unsupervised Key Frame Extraction for Efficient Video Classification | Video processing and analysis have become an urgent task since a huge amount of videos (e.g., Youtube, Hulu) are uploaded online every day. The extraction of representative key frames from videos is very important in video processing and analysis since it greatly reduces computing resources and time. Although great progress has been made recently, large-scale video classification remains an open problem, as the existing methods have not well balanced the performance and efficiency simultaneously. To tackle this problem, this work presents an unsupervised method to retrieve the key frames, which combines Convolutional Neural Network (CNN) and Temporal Segment Density Peaks Clustering (TSDPC). The proposed TSDPC is a generic and powerful framework and it has two advantages compared with previous works, one is that it can calculate the number of key frames automatically. The other is that it can preserve the temporal information of the video. Thus it improves the efficiency of video classification. Furthermore, a Long Short-Term Memory network (LSTM) is added on the top of the CNN to further elevate the performance of classification. Moreover, a weight fusion strategy of different input networks is presented to boost the performance. By optimizing both video classification and key frame extraction simultaneously, we achieve better classification performance and higher efficiency. We evaluate our method on two popular datasets (i.e., HMDB51 and UCF101) and the experimental results consistently demonstrate that our strategy achieves competitive performance and efficiency compared with the state-of-the-art approaches. | ['Paolo Rota', 'Nicu Sebe', 'Bin Ren', 'Songsong Wu', 'Lei Ding', 'Hao Tang'] | 2022-11-12 | null | null | null | null | ['video-classification'] | ['computer-vision'] | [-1.51171163e-01 -8.23975861e-01 -3.24556828e-01 -5.63645884e-02
-5.21953106e-01 3.73628251e-02 2.36110255e-01 4.93515991e-02
-8.09447050e-01 4.91060227e-01 -2.48535462e-02 2.34024059e-02
-6.93214387e-02 -7.69868553e-01 -5.17604411e-01 -8.93670678e-01
-7.31296884e-03 -2.31744230e-01 7.55039990e-01 1.47707656e-01
2.15852037e-01 3.62172693e-01 -1.79468143e+00 2.33068198e-01
8.60433400e-01 1.56701589e+00 5.71403086e-01 3.29399019e-01
-1.66037709e-01 1.01370406e+00 -4.57092196e-01 -1.98622242e-01
6.91243857e-02 -2.64484227e-01 -5.01143098e-01 1.37255386e-01
-9.28636361e-03 -4.34329361e-01 -7.39864528e-01 1.12643397e+00
5.18775642e-01 3.51186246e-01 2.16334939e-01 -1.24167216e+00
-3.18619341e-01 2.36194029e-01 -7.34076142e-01 5.69744766e-01
6.60309121e-02 -1.29532978e-01 7.40346968e-01 -8.29966187e-01
4.20835316e-01 9.81820822e-01 4.54129159e-01 1.54213503e-01
-6.20261312e-01 -8.53306472e-01 1.87217295e-01 9.22548592e-01
-1.65671325e+00 -2.67286628e-01 8.20969522e-01 -3.27329040e-01
6.85714364e-01 5.38681857e-02 8.16717982e-01 7.29324758e-01
6.06582724e-02 1.10289955e+00 5.23976803e-01 -2.30766580e-01
1.36285275e-01 -2.45061919e-01 -8.72132555e-02 7.04482317e-01
3.82361077e-02 -2.73727983e-01 -3.43158752e-01 2.32932463e-01
8.37233663e-01 4.99499500e-01 -5.48757970e-01 -1.84449837e-01
-1.27360582e+00 6.19725287e-01 3.34664166e-01 6.23946905e-01
-5.32610893e-01 3.63409854e-02 7.79388845e-01 -1.00969568e-01
3.89927179e-01 -1.25284478e-01 -2.15892494e-01 -3.64158183e-01
-1.03270686e+00 1.26634449e-01 2.90889055e-01 7.39685893e-01
5.68088353e-01 5.21074533e-02 -1.94037899e-01 1.05991197e+00
9.34953019e-02 3.22294056e-01 7.62640834e-01 -9.35018957e-01
5.85679293e-01 6.94165647e-01 -6.32978305e-02 -1.49450314e+00
-3.37808341e-01 -2.06265077e-01 -1.16751158e+00 -1.83276176e-01
2.90971577e-01 -4.31750119e-02 -6.94280744e-01 1.33894014e+00
2.17681631e-01 4.87787902e-01 -8.62671807e-02 1.00374806e+00
9.05551314e-01 1.07010460e+00 8.65146071e-02 -6.09538674e-01
1.36811066e+00 -9.71080780e-01 -9.59068775e-01 2.27889061e-01
2.92186528e-01 -7.90255308e-01 6.97855890e-01 4.32672679e-01
-8.99206698e-01 -8.75697255e-01 -9.62740183e-01 9.20158178e-02
-1.81669831e-01 3.91422242e-01 4.68875319e-01 1.91311672e-01
-6.78340375e-01 4.58310664e-01 -1.00111461e+00 -2.41613507e-01
6.43382370e-01 2.54327804e-01 -3.45198721e-01 -1.69666439e-01
-1.21341622e+00 3.40914249e-01 8.48836124e-01 2.34099939e-01
-5.31488299e-01 -3.68455887e-01 -8.54078591e-01 2.71792740e-01
6.79515839e-01 -1.80227801e-01 1.06412172e+00 -9.48578060e-01
-1.36546266e+00 3.16096157e-01 -2.13861123e-01 -4.85326409e-01
3.87381911e-01 -2.02751711e-01 -5.33415139e-01 5.90858340e-01
-7.70125315e-02 6.58871591e-01 8.55899453e-01 -7.26333678e-01
-1.23094642e+00 -1.52895823e-01 -4.25102934e-02 1.87718272e-01
-9.32301521e-01 1.89201385e-01 -1.16754353e+00 -9.31980968e-01
4.73854579e-02 -6.74861789e-01 6.96507618e-02 -1.70293495e-01
-4.18635793e-02 -3.90715808e-01 1.21859837e+00 -7.65750170e-01
1.74451649e+00 -2.20194101e+00 2.13422254e-01 -8.15630797e-03
1.39700890e-01 7.77068615e-01 1.93167299e-01 1.91233724e-01
4.71970625e-02 -5.95472530e-02 -2.52671517e-03 -4.26882170e-02
-2.86478668e-01 1.10522114e-01 5.39382100e-02 4.33369726e-01
2.71762069e-02 7.31317580e-01 -6.81437016e-01 -8.53295863e-01
5.25758684e-01 6.82037830e-01 -4.18564737e-01 1.16167232e-01
5.56344166e-02 2.40527645e-01 -3.74242216e-01 6.45364761e-01
5.98541796e-01 -3.67637575e-01 7.72807524e-02 -4.24676210e-01
-2.76474893e-01 -2.29152754e-01 -1.24164379e+00 1.55287802e+00
-2.10077584e-01 6.12486780e-01 -1.75864935e-01 -1.28742826e+00
7.86478281e-01 4.56954747e-01 7.46169686e-01 -8.21332037e-01
4.09171283e-01 1.81433678e-01 -1.35088637e-01 -8.18815410e-01
4.14481610e-01 2.23197475e-01 1.60647035e-01 7.86721408e-02
2.76023224e-02 6.21243775e-01 6.81266606e-01 4.82972711e-02
7.30394065e-01 5.14165796e-02 3.16035748e-01 -1.01693258e-01
9.49279606e-01 -2.39976034e-01 1.05888033e+00 1.51149273e-01
-3.09418589e-01 4.35435712e-01 3.88596833e-01 -6.26114011e-01
-8.61392379e-01 -5.41564047e-01 -2.03424580e-02 7.94641674e-01
4.92361516e-01 -5.33546209e-01 -8.15093875e-01 -4.74915117e-01
-1.99795827e-01 1.03683747e-01 -3.75013024e-01 -9.08972770e-02
-7.74195611e-01 -8.03411007e-01 2.21611604e-01 6.78827882e-01
1.01256895e+00 -1.00504100e+00 -6.79735124e-01 2.83021092e-01
-5.39316952e-01 -1.45978069e+00 -4.07026589e-01 -5.51731229e-01
-8.40502799e-01 -1.12088776e+00 -9.51985657e-01 -1.00117481e+00
4.60999846e-01 7.17641175e-01 6.35034084e-01 2.54999578e-01
-2.32818171e-01 -5.21067381e-02 -6.22960091e-01 -1.04445487e-01
9.19235125e-02 1.03236049e-01 -2.27940921e-02 2.58378685e-01
5.19174814e-01 -3.90240610e-01 -6.83744133e-01 3.96499485e-01
-1.09154749e+00 2.15332612e-01 4.98541474e-01 6.76504374e-01
7.54472733e-01 6.52789593e-01 4.92185980e-01 -4.22412425e-01
3.46142352e-01 -3.44796300e-01 -7.21583784e-01 1.75503105e-01
-4.09122169e-01 -3.98095757e-01 9.12815332e-01 -3.79001081e-01
-8.90646935e-01 -8.12129751e-02 -2.22456977e-01 -7.50608444e-01
-7.97804669e-02 6.44345641e-01 -1.72496468e-01 2.33167801e-02
-9.83147174e-02 5.23762226e-01 5.07394150e-02 -5.68329871e-01
-1.76660165e-01 7.99249232e-01 5.32706797e-01 -2.20112324e-01
5.25113761e-01 3.58569682e-01 -1.64106324e-01 -9.46821630e-01
-5.15421450e-01 -5.64411163e-01 -5.46435714e-01 -4.83135611e-01
1.00721443e+00 -1.00454080e+00 -8.99752796e-01 8.15466344e-01
-1.01057482e+00 1.25153676e-01 2.37844020e-01 8.53357971e-01
-2.94447988e-01 5.91738343e-01 -7.09317446e-01 -6.75271213e-01
-3.50420117e-01 -1.41887844e+00 6.54594004e-01 5.15836895e-01
2.04860002e-01 -7.30808675e-01 -4.15222377e-01 2.59802550e-01
2.89142638e-01 3.13424557e-01 6.75476134e-01 -2.58597314e-01
-7.29397774e-01 -2.39709154e-01 -4.98547077e-01 3.44941676e-01
1.12378806e-01 1.77763402e-01 -6.15643084e-01 -2.84582734e-01
-1.33327961e-01 -1.05972745e-01 9.71819162e-01 5.64556599e-01
1.64851832e+00 -1.93598583e-01 -2.97345757e-01 5.25667429e-01
1.38530648e+00 6.29612207e-01 7.62143791e-01 5.40167153e-01
8.15643132e-01 3.39655250e-01 7.76063561e-01 5.23946285e-01
4.77665186e-01 8.13569188e-01 3.60056430e-01 -3.97835150e-02
2.58986484e-02 6.94064498e-02 2.69280553e-01 1.23906422e+00
-4.27488834e-01 -1.92829892e-01 -7.57466555e-01 4.85931963e-01
-2.10642362e+00 -1.27053022e+00 -2.92151459e-02 2.09381604e+00
5.15587926e-01 1.61549091e-01 2.48638436e-01 5.86655974e-01
9.33755219e-01 3.40199351e-01 -4.93715852e-01 2.26691082e-01
2.93840263e-02 -6.31409064e-02 2.10895717e-01 -4.31625657e-02
-1.43938231e+00 6.01473272e-01 4.88200712e+00 1.26129735e+00
-1.28554344e+00 1.48384631e-01 7.40162790e-01 -2.31549188e-01
5.42042255e-01 -5.98519623e-01 -7.07877755e-01 9.69794750e-01
7.12860644e-01 -1.09961197e-01 2.67958879e-01 8.44571412e-01
3.75707895e-01 -1.35614499e-01 -6.10704362e-01 1.50261068e+00
2.28750825e-01 -1.43563497e+00 3.81416418e-02 -5.71997166e-02
5.09900331e-01 -1.63330689e-01 -1.54387638e-01 2.57024586e-01
-3.75184208e-01 -6.89174950e-01 6.26588225e-01 4.73028839e-01
6.82086289e-01 -1.16619122e+00 1.14122903e+00 3.12956661e-01
-1.71199536e+00 -3.00424665e-01 -5.16113460e-01 1.08446680e-01
2.55428076e-01 5.96781671e-01 -5.06772995e-02 7.34319389e-01
1.11783791e+00 1.02387786e+00 -4.62090611e-01 1.41098094e+00
-4.93784472e-02 4.45930541e-01 -1.37733191e-01 4.93699498e-02
3.83973539e-01 -2.79409915e-01 2.82183528e-01 1.13178301e+00
5.42003095e-01 2.08113790e-01 4.61197913e-01 2.26644605e-01
-2.33599827e-01 3.39279324e-01 -1.47872195e-01 -1.02378659e-01
3.78971726e-01 1.39605749e+00 -1.05026269e+00 -5.47740221e-01
-7.03746080e-01 6.54370368e-01 1.26736552e-01 2.03828111e-01
-1.02772164e+00 -6.58442080e-01 4.90585983e-01 -1.00154944e-01
6.82985723e-01 -2.85437733e-01 3.43985587e-01 -1.31747985e+00
2.13077903e-01 -7.32140779e-01 5.51912844e-01 -4.89029646e-01
-8.96688819e-01 6.13503098e-01 -1.26087561e-01 -1.58236206e+00
-1.55030275e-02 -4.81753916e-01 -4.49131191e-01 3.93396080e-01
-1.66105163e+00 -7.52616167e-01 -6.53995872e-01 6.62907600e-01
8.25795770e-01 -2.41379142e-01 3.63433927e-01 9.52491224e-01
-1.11349165e+00 5.04694462e-01 2.66850591e-01 5.00226915e-01
4.77699637e-01 -7.07319856e-01 -5.60747944e-02 1.07245982e+00
-5.52382041e-03 3.95112425e-01 2.18665466e-01 -3.91667455e-01
-1.38352025e+00 -1.20805752e+00 4.93632734e-01 3.14923882e-01
5.26006937e-01 -1.01939090e-01 -1.03022563e+00 1.30686581e-01
1.03491008e-01 1.92753181e-01 5.92487156e-01 -3.98660064e-01
2.05298066e-02 -5.25071323e-01 -6.58645213e-01 4.27574515e-01
7.55118191e-01 -3.29083979e-01 -2.49280915e-01 1.22954011e-01
6.11084104e-01 -3.44534457e-01 -9.35147703e-01 6.15644217e-01
6.07100368e-01 -1.10491383e+00 9.11863804e-01 -9.90512297e-02
4.71240819e-01 -5.66472411e-01 -1.64458111e-01 -8.76645207e-01
-2.63740718e-01 -2.71530867e-01 -4.75973159e-01 1.29883802e+00
-1.91551238e-01 -4.15122837e-01 6.77004278e-01 3.33252698e-01
2.10460443e-02 -1.15438604e+00 -8.18320334e-01 -7.73801982e-01
-4.76897985e-01 -3.92080426e-01 5.12393653e-01 7.12962687e-01
-1.90244332e-01 1.80759564e-01 -5.05708277e-01 -6.47415742e-02
4.18007135e-01 2.81840891e-01 5.04168570e-01 -1.23338974e+00
1.05136670e-02 -5.15984118e-01 -7.00775206e-01 -1.07783484e+00
4.35447842e-02 -5.50504446e-01 -1.91415161e-01 -1.56451440e+00
3.78578603e-01 -2.09358260e-01 -6.63263261e-01 3.97948533e-01
-3.38921100e-01 3.89783144e-01 3.05527538e-01 5.38483679e-01
-1.02819550e+00 6.26271129e-01 1.15921271e+00 -4.62220144e-03
-1.83310911e-01 -7.27393702e-02 -3.86798382e-01 8.83400917e-01
7.56998301e-01 -2.19449133e-01 -3.26878935e-01 -3.53559732e-01
-1.25383973e-01 7.86910355e-02 1.54664144e-01 -1.36850417e+00
4.59221035e-01 1.32187098e-01 6.05807900e-01 -7.67583847e-01
2.78032064e-01 -8.52775335e-01 -2.34525162e-03 5.39667487e-01
1.88316703e-02 1.65413171e-01 1.32694304e-01 6.54079020e-01
-7.43817866e-01 -8.18686038e-02 8.84996355e-01 -3.74775566e-02
-1.18642330e+00 6.41633868e-01 -3.99325073e-01 -1.65428013e-01
1.25159538e+00 -1.35111630e-01 -2.48059332e-01 -2.69658476e-01
-2.47487396e-01 2.93569058e-01 3.24503541e-01 5.63992083e-01
7.66129673e-01 -1.59427810e+00 -4.03069347e-01 9.80720744e-02
-7.57808890e-03 -4.76698540e-02 4.87097949e-01 1.11242998e+00
-6.86076820e-01 4.22183096e-01 -2.75512040e-01 -6.36231124e-01
-1.46818590e+00 6.96756065e-01 -8.84721503e-02 -1.06213674e-01
-7.55833924e-01 5.31216741e-01 7.59369954e-02 2.56838053e-01
4.51476634e-01 -1.44702628e-01 -7.02697277e-01 3.24097604e-01
9.94350612e-01 4.66346890e-01 -2.63714753e-02 -9.50764716e-01
-2.79187977e-01 7.05997169e-01 -1.45798922e-01 4.34559703e-01
1.41303658e+00 -1.32671654e-01 -1.78840101e-01 3.00709844e-01
1.44807625e+00 -4.47713494e-01 -1.17381907e+00 -3.09533566e-01
-1.03368483e-01 -7.28306174e-01 2.61582822e-01 -9.37100872e-02
-1.55210865e+00 9.95793045e-01 5.97609341e-01 2.47230709e-01
1.44735420e+00 -4.03490961e-01 1.37873697e+00 3.40436101e-01
3.14524174e-01 -1.33159411e+00 8.41789395e-02 3.91895920e-01
3.92854929e-01 -1.25655603e+00 1.33693695e-01 -3.28666389e-01
-4.74725485e-01 1.24353826e+00 5.10555267e-01 -2.28656214e-02
4.56163734e-01 2.28041485e-02 -7.38905072e-02 7.96701908e-02
-4.72766608e-01 -2.01792702e-01 3.28869790e-01 8.73640031e-02
3.03140759e-01 -2.95630068e-01 -4.89162207e-01 4.95881647e-01
2.47341961e-01 2.53600329e-01 2.15534821e-01 9.11270857e-01
-5.60399652e-01 -8.61236751e-01 -2.83874452e-01 5.59323668e-01
-9.31602240e-01 9.34825987e-02 3.25369805e-01 7.70185888e-01
2.36994311e-01 8.99210632e-01 1.36610538e-01 -5.88810563e-01
1.02698110e-01 -1.36163935e-01 1.52796835e-01 -1.02287628e-01
-1.64342925e-01 3.58499467e-01 -2.50500202e-01 -6.56363606e-01
-8.78513694e-01 -4.41752195e-01 -1.23852766e+00 -2.79694945e-01
-3.78692120e-01 2.93426216e-01 4.88985568e-01 9.05830860e-01
3.94475996e-01 7.07978010e-01 6.16248131e-01 -9.29481566e-01
3.20372805e-02 -6.98563576e-01 -5.24321318e-01 4.97044861e-01
9.83181223e-02 -8.95806849e-01 -1.94948819e-02 3.26965243e-01] | [9.121851921081543, -0.07246740907430649] |
08be73b6-10b0-438d-bf24-9ac3874ddae1 | towards-speaker-age-estimation-with-label | 2202.11424 | null | https://arxiv.org/abs/2202.11424v1 | https://arxiv.org/pdf/2202.11424v1.pdf | Towards Speaker Age Estimation with Label Distribution Learning | Existing methods for speaker age estimation usually treat it as a multi-class classification or a regression problem. However, precise age identification remains a challenge due to label ambiguity, \emph{i.e.}, utterances from adjacent age of the same person are often indistinguishable. To address this, we utilize the ambiguous information among the age labels, convert each age label into a discrete label distribution and leverage the label distribution learning (LDL) method to fit the data. For each audio data sample, our method produces a age distribution of its speaker, and on top of the distribution we also perform two other tasks: age prediction and age uncertainty minimization. Therefore, our method naturally combines the age classification and regression approaches, which enhances the robustness of our method. We conduct experiments on the public NIST SRE08-10 dataset and a real-world dataset, which exhibit that our method outperforms baseline methods by a relatively large margin, yielding a 10\% reduction in terms of mean absolute error (MAE) on a real-world dataset. | ['Jing Xiao', 'Junqing Peng', 'Jianzong Wang', 'Shijing Si'] | 2022-02-23 | null | null | null | null | ['age-estimation', 'age-estimation'] | ['computer-vision', 'miscellaneous'] | [ 1.32582873e-01 -9.81695368e-04 -1.53832406e-01 -9.03366983e-01
-1.35958755e+00 -5.84122598e-01 3.78890693e-01 2.70482987e-01
-3.67531836e-01 6.45404816e-01 4.25005972e-01 5.43660834e-04
2.06936911e-01 -2.78332472e-01 -3.51705104e-01 -8.14039767e-01
1.54871359e-01 3.89045417e-01 -2.58255720e-01 5.54530501e-01
1.41615704e-01 -6.84506372e-02 -1.61345792e+00 -2.48427942e-01
9.41130698e-01 1.46406531e+00 -4.21288490e-01 4.32322651e-01
1.91838369e-01 5.91268301e-01 -7.18010247e-01 -5.74889541e-01
-2.68601011e-02 -6.71465546e-02 -7.38951027e-01 -2.07888335e-01
8.37770462e-01 -5.18772364e-01 -1.32101804e-01 9.24708962e-01
6.73654675e-01 7.68050998e-02 1.20155859e+00 -1.65674508e+00
-3.96035641e-01 9.24264014e-01 -9.36566114e-01 -2.73684770e-01
5.77867091e-01 -4.94456172e-01 1.04378331e+00 -7.32064903e-01
-1.97656304e-01 1.41574979e+00 9.00933921e-01 7.27014959e-01
-1.25603509e+00 -1.29882634e+00 3.24051201e-01 -6.72124773e-02
-1.64358926e+00 -7.87846804e-01 7.93998003e-01 -5.51266611e-01
1.19536549e-01 2.42569208e-01 2.17827156e-01 1.31420708e+00
-1.91316634e-01 7.14491904e-01 1.32243812e+00 -2.05342904e-01
4.05299693e-01 -3.73539291e-02 4.42792237e-01 3.67969215e-01
-3.36658284e-02 2.54962081e-03 -8.23802352e-01 -3.78161460e-01
1.54617786e-01 -1.93713635e-01 6.03796504e-02 3.04277372e-02
-9.74347174e-01 7.20408201e-01 -3.40620279e-02 -3.77939224e-01
-9.31121558e-02 1.44335568e-01 2.02379227e-01 2.03854159e-01
7.78617024e-01 -7.64202178e-02 -3.29809815e-01 -2.80358255e-01
-9.83116567e-01 4.79686409e-01 8.43028128e-01 6.90998495e-01
6.15836442e-01 -2.32833728e-01 3.61463875e-02 1.05968428e+00
5.90699017e-01 6.61800146e-01 2.74190187e-01 -1.12296116e+00
3.81906956e-01 2.51666486e-01 3.86450924e-02 -5.16999722e-01
-3.77163529e-01 -3.37427519e-02 -7.63754487e-01 6.70392513e-02
9.74734187e-01 -2.44088620e-01 -8.78170907e-01 2.39336061e+00
4.92348373e-01 2.12185651e-01 -3.44623365e-02 4.87450004e-01
6.00061834e-01 4.48064029e-01 4.05173659e-01 -6.35436714e-01
1.46024299e+00 -5.83147645e-01 -6.35712385e-01 -4.51725662e-01
3.98943961e-01 -6.29684150e-01 7.71471083e-01 5.60558319e-01
-9.55375373e-01 -3.24359298e-01 -1.02640319e+00 8.54348242e-02
1.00071467e-01 1.43915832e-01 4.75509137e-01 9.59214747e-01
-7.72403777e-01 2.31808722e-01 -7.53951430e-01 -1.12253889e-01
1.44612953e-01 5.90422928e-01 -2.98026413e-01 3.71164195e-02
-1.09824395e+00 3.94751608e-01 1.68380924e-02 -2.38023832e-01
-6.52260125e-01 -9.55593705e-01 -1.00645590e+00 -1.82221115e-01
1.52759001e-01 -4.82409447e-01 1.43673122e+00 -6.19427443e-01
-1.35866249e+00 1.04867995e+00 -3.45731318e-01 -3.55280250e-01
3.63421798e-01 -4.37620908e-01 -5.15765131e-01 -1.76923648e-01
9.46344212e-02 6.23713493e-01 1.18571329e+00 -1.13482165e+00
-7.38548934e-01 -6.33686900e-01 -1.79056779e-01 1.86084043e-02
-6.01590753e-01 3.44687164e-01 -7.99052045e-02 -7.85117686e-01
3.66354138e-01 -1.07812965e+00 2.33270615e-01 -1.33613452e-01
-2.58923382e-01 -4.66951489e-01 5.21165967e-01 -1.03491378e+00
1.68319106e+00 -2.42253351e+00 5.76605126e-02 2.29464844e-01
4.77332920e-01 -4.95126754e-01 2.30120033e-01 -8.11477825e-02
-8.15875903e-02 -9.25926398e-03 -1.11926980e-01 -8.96990657e-01
1.45068035e-01 -1.26860827e-01 -1.73490122e-01 5.50666690e-01
-2.52274483e-01 1.61800712e-01 -5.87074339e-01 -7.55402803e-01
-1.42926514e-01 2.22016230e-01 -4.69225883e-01 3.98753107e-01
1.37986913e-01 4.27428782e-01 -2.12893486e-01 7.38006592e-01
6.84059262e-01 2.32718214e-01 4.62479778e-02 -8.23425949e-02
8.71417969e-02 2.10347071e-01 -1.34857631e+00 1.60158980e+00
-4.98848259e-01 1.31908745e-01 1.92900926e-01 -6.68486238e-01
1.06645131e+00 3.76536369e-01 6.78648889e-01 -1.66540995e-01
1.72547311e-01 2.59932816e-01 -1.22868583e-01 3.47775035e-02
3.00665587e-01 -5.52782454e-02 -5.40517807e-01 8.39373648e-01
-7.56849423e-02 -3.02893836e-02 -1.16702490e-01 8.53988007e-02
8.54605019e-01 -3.15964699e-01 9.47163031e-02 -1.92376390e-01
4.85408992e-01 -1.00382543e+00 9.10093009e-01 6.45650566e-01
-5.92507780e-01 6.86655045e-01 4.10523683e-01 -6.33134171e-02
-8.25382590e-01 -1.46098387e+00 -2.67494082e-01 1.41382599e+00
-8.81216675e-02 -4.78239477e-01 -1.05040336e+00 -8.17005455e-01
5.72852716e-02 5.54050207e-01 -6.44030809e-01 -3.04111570e-01
-4.74385411e-01 -6.60346091e-01 6.57678902e-01 6.44657135e-01
1.65090308e-01 -6.07569456e-01 2.52889618e-02 -1.86280087e-02
-3.35538357e-01 -1.25784087e+00 -8.17935884e-01 -4.83223759e-02
-4.94372338e-01 -8.11779022e-01 -6.48377776e-01 -8.88524711e-01
4.79107708e-01 -2.72571355e-01 1.04997742e+00 -2.29278818e-01
4.31264378e-03 4.22121644e-01 -9.90524739e-02 -6.17607057e-01
-4.96237606e-01 2.29389295e-01 7.23479867e-01 1.49619445e-01
6.82525754e-01 -8.24712992e-01 -5.34243286e-01 3.34660739e-01
-1.93450749e-01 -3.06765705e-01 2.29590401e-01 7.09964335e-01
2.85952479e-01 -4.01036292e-02 1.12445509e+00 -4.90110368e-01
4.00393486e-01 -3.54365647e-01 -4.28656071e-01 1.88692942e-01
-9.16602671e-01 7.12604150e-02 1.07445680e-01 -7.32765138e-01
-1.06949127e+00 2.83577979e-01 -1.46160334e-01 -8.83751437e-02
-1.53305545e-01 2.52840549e-01 -4.24733818e-01 2.72289217e-01
3.88658196e-01 -5.67623340e-02 1.56595543e-01 -5.03590465e-01
2.59513974e-01 1.21579123e+00 9.17538881e-01 -8.60568762e-01
8.84891093e-01 1.03541300e-01 -2.08925888e-01 -5.08547902e-01
-1.16896546e+00 -3.46784770e-01 -3.91371697e-01 -3.13325047e-01
8.86253953e-01 -1.29049814e+00 -8.89767230e-01 1.07869864e+00
-8.53983939e-01 -7.79279992e-02 1.84825972e-01 6.22281671e-01
-4.80162919e-01 4.42780286e-01 -8.34965169e-01 -1.27515984e+00
-5.33068419e-01 -9.44066644e-01 1.27377701e+00 2.81667829e-01
-6.55957639e-01 -7.29513586e-01 -1.10510074e-01 6.79670334e-01
-1.20305419e-01 2.44802963e-02 9.64903414e-01 -7.56902933e-01
1.32038280e-01 -3.23385537e-01 -9.75709632e-02 3.18963975e-01
1.84390232e-01 2.21811719e-02 -1.40393770e+00 -4.21857417e-01
-2.42370367e-02 -6.95303857e-01 7.50093162e-01 4.48001981e-01
1.49934697e+00 -6.42643571e-02 -1.27566129e-01 3.11005533e-01
6.99986041e-01 -3.17807272e-02 2.61198074e-01 -3.01186871e-02
6.80276513e-01 8.05302024e-01 6.93668485e-01 6.96961462e-01
7.13946044e-01 6.98714614e-01 1.68993056e-01 2.75120139e-01
1.45382777e-01 -3.36153090e-01 5.01144290e-01 9.19207513e-01
9.42182988e-02 1.01921409e-01 -8.76099944e-01 3.15696746e-01
-1.54567194e+00 -7.65218318e-01 3.96311760e-01 2.67679095e+00
1.26445365e+00 1.40115157e-01 7.24879861e-01 5.62276065e-01
9.07643199e-01 1.41424254e-01 -4.93166715e-01 -1.80145815e-01
1.63581625e-01 -7.52223209e-02 1.97478950e-01 4.47389305e-01
-1.33796930e+00 3.69054139e-01 6.64952421e+00 7.21641123e-01
-9.56475377e-01 -9.43510532e-02 9.03455019e-01 -1.37749419e-01
-1.32654449e-02 -4.37831819e-01 -9.56735730e-01 6.70091510e-01
1.13859439e+00 -3.90972942e-01 5.37619352e-01 8.02704513e-01
-5.52195385e-02 -8.94350708e-02 -1.56445312e+00 1.43050015e+00
1.41268551e-01 -3.50760937e-01 -5.74567437e-01 2.35723421e-01
5.39089501e-01 -4.97694552e-01 4.60009843e-01 4.97600853e-01
3.29412460e-01 -1.04146838e+00 1.00191236e+00 1.73157141e-01
9.82336879e-01 -9.79520202e-01 4.85270083e-01 2.38604546e-01
-1.25961363e+00 -3.38962436e-01 1.67264864e-02 -8.28341320e-02
1.12709396e-01 7.60803938e-01 -3.42707276e-01 3.26993287e-01
7.54253149e-01 3.75177383e-01 -5.87502837e-01 7.21432090e-01
-5.61179183e-02 8.38881552e-01 -3.33186448e-01 3.38572711e-01
-5.11895359e-01 -9.22869369e-02 1.74397379e-01 7.85379291e-01
5.47948599e-01 -5.44276498e-02 2.15325937e-01 5.15171468e-01
-2.67643481e-01 3.79536115e-02 -1.38269201e-01 4.69378047e-02
1.12792325e+00 1.18670762e+00 -2.86475152e-01 -9.81106460e-02
-3.89286697e-01 7.55355597e-01 3.69605720e-01 1.69220820e-01
-8.96721780e-01 -2.11575016e-01 9.83102202e-01 -2.56828994e-01
-1.50209323e-01 -2.48008850e-03 -4.27364141e-01 -9.28569794e-01
9.41931680e-02 -9.10009086e-01 5.12661338e-01 -3.75903070e-01
-1.68394268e+00 2.07896933e-01 3.88623103e-02 -1.09959435e+00
-6.49499834e-01 -3.16371262e-01 -5.45282662e-01 7.49878943e-01
-8.53130460e-01 -1.21426809e+00 -2.38432333e-01 2.86348373e-01
4.08714950e-01 -8.60008672e-02 8.94952655e-01 5.73296964e-01
-7.50969946e-01 1.30222642e+00 -1.20006995e-02 1.50261551e-01
1.31686616e+00 -1.48745143e+00 2.07227722e-01 5.25587916e-01
-9.37270820e-02 5.02530277e-01 8.01294208e-01 -4.26789135e-01
-9.74655867e-01 -9.00546014e-01 9.81639445e-01 -6.57782793e-01
7.01232314e-01 -7.98722208e-01 -7.72625625e-01 6.05765998e-01
-2.20753670e-01 -3.13117921e-01 1.00728929e+00 6.91091835e-01
-8.94532561e-01 -2.99559444e-01 -1.20317066e+00 4.25196677e-01
1.03839731e+00 -6.96945548e-01 -4.53961313e-01 1.84010826e-02
8.49779665e-01 -3.78446132e-01 -1.30357039e+00 4.21555668e-01
1.01845622e+00 -8.13622713e-01 1.02173686e+00 -1.57314211e-01
3.08545947e-01 -2.51802839e-02 -2.34135300e-01 -1.20562994e+00
-5.83050586e-03 -4.71423775e-01 -5.31358004e-01 2.04543424e+00
3.70216817e-01 -5.31724453e-01 7.62840331e-01 1.11132991e+00
4.76138592e-02 -6.67652130e-01 -1.06039917e+00 -8.23986650e-01
2.88148373e-01 -6.31650209e-01 9.07625437e-01 6.99030936e-01
-2.14798283e-02 4.35712457e-01 -5.93522668e-01 3.42027843e-01
1.02152991e+00 -5.70140146e-02 5.92749476e-01 -1.61561418e+00
-2.26442724e-01 -4.17468518e-01 -3.44904482e-01 -8.63370299e-01
7.27516294e-01 -3.14888179e-01 3.35596234e-01 -8.48566949e-01
3.32502663e-01 -7.38535345e-01 -4.35940146e-01 3.53624165e-01
-5.62493205e-01 2.69248158e-01 -1.06663384e-01 6.13123551e-02
-5.04232168e-01 4.25117642e-01 4.05144989e-01 -3.62171680e-01
-1.24598093e-01 4.53224361e-01 -7.97762513e-01 8.48379254e-01
6.27414405e-01 -5.15698850e-01 -3.68210375e-01 -1.40299005e-02
7.31592719e-03 8.91377851e-02 -6.94681928e-02 -1.20305371e+00
4.70739231e-03 -1.21930689e-01 3.85454863e-01 -5.01732767e-01
5.39280355e-01 -5.58659554e-01 -1.90108344e-02 2.07633153e-01
-4.25932080e-01 -1.56535536e-01 -2.58933693e-01 5.58077753e-01
-1.55211076e-01 -7.55855292e-02 7.15022743e-01 6.23760521e-01
-2.40590274e-01 5.29286027e-01 -4.51791957e-02 1.99808910e-01
7.22537339e-01 3.16767730e-02 -3.62092227e-01 -7.14982629e-01
-6.16329193e-01 4.19800639e-01 4.83743846e-01 4.05030727e-01
1.82198673e-01 -1.53097975e+00 -6.61387980e-01 1.80136353e-01
4.01241452e-01 -1.59644578e-02 2.13069409e-01 8.38564336e-01
2.59578466e-01 -1.76917851e-01 1.91275701e-01 -5.53708553e-01
-1.65832877e+00 4.88579690e-01 5.47713675e-02 3.45403813e-02
1.51308458e-02 1.03625655e+00 4.52940047e-01 -4.99575406e-01
7.52963901e-01 -2.39804983e-02 -2.83956051e-01 3.11392814e-01
7.75950074e-01 5.16212344e-01 -1.73983648e-01 -7.04095542e-01
-4.70592976e-01 6.00203812e-01 -2.05991641e-01 -4.16482866e-01
9.15187538e-01 -4.59793150e-01 4.38445993e-03 7.98219800e-01
1.12048376e+00 2.39115238e-01 -1.27101517e+00 -3.69705647e-01
1.99709665e-02 -3.52726638e-01 -2.32462227e-01 -6.42558455e-01
-8.16847742e-01 6.83248878e-01 6.99440360e-01 1.50522888e-01
1.16808021e+00 2.43739933e-01 7.45455086e-01 -6.60170242e-02
2.52901286e-01 -1.26619697e+00 1.91331938e-01 2.93919712e-01
6.32345319e-01 -1.32067299e+00 9.29874554e-02 -5.89496851e-01
-5.91412663e-01 7.02361345e-01 7.69671082e-01 5.46067953e-01
6.66649282e-01 1.14575393e-01 4.36635822e-01 4.11365956e-01
-4.37378615e-01 2.40304500e-01 3.00633490e-01 7.67453134e-01
6.17842019e-01 2.31141493e-01 -2.88900107e-01 1.14886248e+00
-6.95614696e-01 -2.48414353e-01 1.42399862e-01 6.45934045e-01
-4.12754804e-01 -1.29564869e+00 -4.96640623e-01 6.43043935e-01
-7.17399836e-01 9.79292393e-02 -2.22315088e-01 1.72194988e-01
-1.09785482e-01 1.27818298e+00 1.85537264e-01 -6.78649545e-01
-4.26201187e-02 4.45183903e-01 5.04173994e-01 -4.25636888e-01
-2.20706940e-01 -1.60265919e-02 2.85740733e-01 -2.32477486e-01
-2.23829180e-01 -1.16471362e+00 -1.14714789e+00 -5.01516402e-01
-2.83027917e-01 1.67694971e-01 7.58996129e-01 8.74636769e-01
-1.02183156e-01 2.16562644e-01 1.12788761e+00 -5.98759472e-01
-1.04155731e+00 -1.11476696e+00 -9.70242560e-01 5.70562243e-01
2.11947665e-01 -8.22799683e-01 -7.70899057e-01 6.09950796e-02] | [14.13963508605957, 6.019794940948486] |
d733fb85-55c6-4617-8249-c6fb6b5eefcf | mixste-seq2seq-mixed-spatio-temporal-encoder | 2203.00859 | null | https://arxiv.org/abs/2203.00859v4 | https://arxiv.org/pdf/2203.00859v4.pdf | MixSTE: Seq2seq Mixed Spatio-Temporal Encoder for 3D Human Pose Estimation in Video | Recent transformer-based solutions have been introduced to estimate 3D human pose from 2D keypoint sequence by considering body joints among all frames globally to learn spatio-temporal correlation. We observe that the motions of different joints differ significantly. However, the previous methods cannot efficiently model the solid inter-frame correspondence of each joint, leading to insufficient learning of spatial-temporal correlation. We propose MixSTE (Mixed Spatio-Temporal Encoder), which has a temporal transformer block to separately model the temporal motion of each joint and a spatial transformer block to learn inter-joint spatial correlation. These two blocks are utilized alternately to obtain better spatio-temporal feature encoding. In addition, the network output is extended from the central frame to entire frames of the input video, thereby improving the coherence between the input and output sequences. Extensive experiments are conducted on three benchmarks (Human3.6M, MPI-INF-3DHP, and HumanEva). The results show that our model outperforms the state-of-the-art approach by 10.9% P-MPJPE and 7.6% MPJPE. The code is available at https://github.com/JinluZhang1126/MixSTE. | ['Junsong Yuan', 'Yujin Chen', 'Jianyu Yang', 'Zhigang Tu', 'Jinlu Zhang'] | 2022-03-02 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Zhang_MixSTE_Seq2seq_Mixed_Spatio-Temporal_Encoder_for_3D_Human_Pose_Estimation_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Zhang_MixSTE_Seq2seq_Mixed_Spatio-Temporal_Encoder_for_3D_Human_Pose_Estimation_CVPR_2022_paper.pdf | cvpr-2022-1 | ['monocular-3d-human-pose-estimation'] | ['computer-vision'] | [-4.25658315e-01 -3.14700603e-01 -1.17681675e-01 -2.02777497e-02
-4.22510862e-01 -1.15168750e-01 3.28051150e-01 -4.64734972e-01
-4.81575221e-01 4.02433157e-01 2.32642740e-01 2.95138657e-01
1.23275667e-01 -4.03676629e-01 -7.10573256e-01 -5.33999145e-01
-3.10128540e-01 2.75577903e-01 6.36227608e-01 -1.33237109e-01
-1.68463767e-01 2.12362275e-01 -1.19362140e+00 2.49665231e-01
3.34654689e-01 9.87217069e-01 2.86938429e-01 7.77909756e-01
2.92032540e-01 9.74891365e-01 -3.08371723e-01 -2.26748168e-01
2.95621097e-01 -5.12277365e-01 -6.30589247e-01 7.73120346e-03
3.76070380e-01 -5.34602880e-01 -7.86829472e-01 7.57516861e-01
6.79919660e-01 3.57668459e-01 3.18379283e-01 -1.29641962e+00
-1.89568639e-01 -1.31515060e-02 -1.07240427e+00 2.17766076e-01
5.24493992e-01 3.53120983e-01 7.17968047e-01 -9.19533551e-01
7.90128708e-01 1.46423399e+00 5.63871026e-01 3.87328982e-01
-7.41151273e-01 -7.30828166e-01 5.17209060e-04 5.84870756e-01
-1.45843315e+00 -1.87022582e-01 8.39346826e-01 -4.35450792e-01
9.43104923e-01 -5.02991490e-03 9.86287177e-01 1.00678861e+00
4.83700126e-01 1.03205645e+00 7.74333656e-01 -7.58559480e-02
-2.94809192e-01 -7.35543370e-01 -1.16720453e-01 9.10240710e-01
-2.51774698e-01 3.36108834e-01 -1.00501680e+00 -2.55329367e-02
1.34716356e+00 1.06307603e-01 -4.00401115e-01 -4.51003551e-01
-1.54560816e+00 4.15600717e-01 5.61881959e-01 9.69917923e-02
-4.22669917e-01 5.06374598e-01 5.03651917e-01 7.86533207e-02
3.61660033e-01 -2.71591157e-01 -4.36247766e-01 -4.02088940e-01
-8.27912331e-01 5.13870060e-01 4.69423503e-01 9.95476067e-01
4.98615175e-01 -1.53334692e-01 -1.41257375e-01 7.70299315e-01
4.34415787e-01 5.33347726e-01 2.56492555e-01 -1.33596659e+00
6.74888194e-01 2.95798779e-01 1.51626527e-01 -1.25479043e+00
-4.58777010e-01 -3.11826169e-01 -1.02885258e+00 8.28820169e-02
4.36347634e-01 -1.29475503e-03 -8.43012214e-01 1.52233315e+00
5.39414406e-01 6.45439088e-01 -3.36131722e-01 1.26014626e+00
7.15097785e-01 8.69653702e-01 -1.04261518e-01 5.68736866e-02
1.41540718e+00 -1.45225704e+00 -6.49168849e-01 -1.49226159e-01
4.04623479e-01 -8.52495432e-01 7.57172048e-01 2.89493442e-01
-1.28688622e+00 -9.08104777e-01 -9.17960584e-01 -2.81529129e-01
2.59763837e-01 1.69700190e-01 2.26012781e-01 -1.60737023e-01
-6.91452742e-01 5.18510461e-01 -1.36325729e+00 5.78588108e-03
1.11880004e-01 1.93389773e-01 -5.57607949e-01 -1.97841063e-01
-1.30323899e+00 6.58711076e-01 2.61744261e-01 2.67006278e-01
-6.45568490e-01 -5.75134337e-01 -9.62120950e-01 -3.07517499e-01
4.39923555e-01 -8.58936608e-01 1.13822246e+00 -4.04104650e-01
-1.50175631e+00 5.66852093e-01 -3.55617821e-01 -2.67200410e-01
8.42894614e-01 -7.15110183e-01 -1.39555410e-01 4.13628638e-01
6.33049086e-02 8.86189222e-01 7.14870393e-01 -9.41629767e-01
-6.87674403e-01 -2.53812045e-01 -1.27910346e-01 3.60715151e-01
9.43701640e-02 5.09256497e-02 -1.32646871e+00 -9.45602536e-01
1.36456892e-01 -1.23185575e+00 -2.67472893e-01 3.16895515e-01
-3.62630427e-01 -1.03243060e-01 8.17987204e-01 -8.97742867e-01
1.25568795e+00 -2.20591354e+00 5.81259847e-01 2.40516234e-02
2.20337704e-01 1.53826624e-01 -1.38840713e-02 2.57313550e-01
9.99866650e-02 -4.61835295e-01 -3.92587893e-02 -7.22572982e-01
-1.20460384e-01 2.84405529e-01 8.25525820e-02 6.60691619e-01
-1.20799318e-01 9.37176526e-01 -7.99116433e-01 -7.48162091e-01
4.20216978e-01 8.84674072e-01 -5.57620704e-01 3.31331015e-01
4.69326712e-02 7.44952738e-01 -3.81841689e-01 5.24386168e-01
5.86932540e-01 -3.14087093e-01 -2.25090832e-02 -3.52828741e-01
-2.11817119e-03 2.54323542e-01 -1.31066108e+00 2.09778023e+00
-1.29036367e-01 4.75437522e-01 -1.61470026e-01 -5.51407456e-01
5.80705941e-01 4.74664927e-01 7.76404917e-01 -7.32091665e-01
1.61330506e-01 6.57337755e-02 -1.58529907e-01 -4.63942885e-01
4.98501152e-01 3.56957257e-01 -2.13637017e-02 2.29562178e-01
-1.38454745e-02 3.13255131e-01 2.49045715e-01 5.61511144e-02
9.69307959e-01 5.26069641e-01 1.77819565e-01 4.20502983e-02
4.71714288e-01 -3.21249992e-01 8.99034441e-01 1.86885163e-01
-3.71324390e-01 9.06573713e-01 4.27244395e-01 -5.13386130e-01
-1.12629139e+00 -1.23552251e+00 4.20114905e-01 7.36433864e-01
4.79558676e-01 -7.55883276e-01 -5.51117301e-01 -4.49256778e-01
-1.72049761e-01 -9.88463871e-03 -4.57847953e-01 -5.09216897e-02
-1.09391356e+00 -2.30056494e-01 4.89523828e-01 8.02734077e-01
7.08275259e-01 -9.48435664e-01 -1.05876780e+00 2.78705925e-01
-6.96499407e-01 -1.34510815e+00 -9.70736921e-01 -2.43442014e-01
-8.29630375e-01 -1.16802120e+00 -1.16066480e+00 -6.99975491e-01
3.35107297e-01 2.74609804e-01 1.03555310e+00 1.35699317e-01
-2.82338172e-01 1.89149782e-01 -3.30894589e-01 3.43154669e-01
1.01151459e-01 -4.55727242e-02 6.65138066e-02 -2.95985639e-01
9.41831023e-02 -6.48722053e-01 -9.03405130e-01 7.55174160e-01
-6.34848118e-01 5.14183342e-01 4.22361851e-01 1.00441372e+00
6.50186718e-01 -1.87263191e-01 -1.40859351e-01 -2.21751556e-01
-1.32148166e-03 -2.55828351e-01 -3.47051501e-01 6.74168393e-02
1.54808104e-01 -2.85770409e-02 2.21521541e-01 -5.85293174e-01
-8.82433116e-01 1.17011964e-01 -2.29789183e-01 -9.57702637e-01
-4.24763784e-02 3.47309560e-01 -9.52237695e-02 2.75765836e-01
1.64866313e-01 2.30951086e-01 1.08954065e-01 -5.35483539e-01
1.08557247e-01 1.23651564e-01 9.23474729e-01 -6.51544333e-01
6.91783547e-01 4.82389838e-01 8.47805813e-02 -7.03384757e-01
-4.93308514e-01 -6.53214335e-01 -8.44484150e-01 -3.06804091e-01
1.05738056e+00 -1.20028245e+00 -7.14873672e-01 8.90190721e-01
-1.41066563e+00 -5.65870225e-01 8.19230974e-02 8.39714468e-01
-7.39242196e-01 6.18032575e-01 -1.08241391e+00 -3.37137341e-01
-2.83025265e-01 -1.33062840e+00 1.20488656e+00 4.60593812e-02
-3.88726056e-01 -6.74260795e-01 2.41376832e-01 2.86833555e-01
-8.86175632e-02 3.78723413e-01 4.02696908e-01 1.79605126e-01
-5.45254409e-01 1.85031351e-02 -6.45126179e-02 2.24089161e-01
2.55738571e-02 2.37211678e-02 -4.27631110e-01 -3.50231409e-01
-2.08683029e-01 -2.10786968e-01 8.26255381e-01 4.51315016e-01
8.48442972e-01 -9.30267349e-02 -2.55174279e-01 6.88959718e-01
9.17065680e-01 9.49183479e-02 8.25900316e-01 3.39887321e-01
9.69848454e-01 5.29233396e-01 8.93947065e-01 4.67069566e-01
6.30792499e-01 1.19304395e+00 2.69334078e-01 4.13799100e-02
-2.97063470e-01 -2.83981740e-01 6.27519608e-01 9.99547780e-01
-5.56003809e-01 -1.46712139e-01 -8.60817492e-01 4.69954908e-01
-2.14432383e+00 -9.59736466e-01 -1.63674802e-01 1.89913750e+00
6.90382719e-01 1.80896744e-01 4.05165464e-01 9.53239128e-02
6.51865244e-01 4.41433877e-01 -3.84685665e-01 2.29329526e-01
7.60573745e-02 6.34860992e-02 3.43095630e-01 4.93916094e-01
-1.08436680e+00 7.56226420e-01 4.99740267e+00 8.25272560e-01
-9.18210328e-01 1.75966680e-01 3.29190463e-01 -3.49154353e-01
1.94489822e-01 -2.27547392e-01 -6.24489009e-01 5.80782652e-01
5.12656391e-01 2.50591159e-01 1.14114610e-02 5.77097535e-01
2.30820596e-01 -1.96461767e-01 -1.02378035e+00 1.18806767e+00
-1.34969816e-01 -1.06211472e+00 -2.11752132e-01 6.16900660e-02
6.01133108e-01 1.69302020e-02 -3.66717437e-03 -1.11751102e-01
-6.20583743e-02 -7.23173201e-01 9.48664308e-01 7.07721770e-01
6.04194343e-01 -9.92312610e-01 5.84755003e-01 3.03662777e-01
-1.82449484e+00 3.45264882e-01 -1.16828747e-01 2.60983575e-02
5.12423277e-01 5.24249136e-01 -1.50939032e-01 6.91791654e-01
1.12259150e+00 1.06655455e+00 -3.29536974e-01 8.67161989e-01
-4.35990721e-01 2.99140722e-01 -4.72722709e-01 4.19229925e-01
1.75785229e-01 -1.71538778e-02 6.78269923e-01 1.06585276e+00
4.28293347e-01 1.30636856e-01 3.10617745e-01 4.54876631e-01
1.82144955e-01 -3.25787306e-01 -1.87717617e-01 3.42556804e-01
2.67342418e-01 8.52728665e-01 -4.96217459e-01 -3.56791317e-01
-5.61741829e-01 1.40167379e+00 9.07733291e-02 2.95215428e-01
-1.17411745e+00 -7.74846971e-02 9.10202324e-01 1.14163004e-01
4.55802798e-01 -7.39860237e-01 2.37131372e-01 -1.21422613e+00
3.77666801e-01 -9.01057899e-01 4.37714934e-01 -7.65031457e-01
-1.01284814e+00 4.23980355e-01 2.69199640e-01 -1.46478343e+00
-4.14662957e-01 -3.60914052e-01 -2.76831716e-01 7.82254994e-01
-1.13719165e+00 -1.12327397e+00 -3.62724692e-01 7.53097951e-01
6.03152335e-01 9.32068899e-02 4.20072109e-01 4.74294931e-01
-3.85713071e-01 5.50660014e-01 -2.64974415e-01 3.40274543e-01
9.55941081e-01 -9.68500137e-01 7.66368926e-01 7.67219782e-01
3.03414036e-02 4.39549834e-01 5.57170153e-01 -7.41912484e-01
-1.33338618e+00 -9.12808001e-01 9.22344744e-01 -3.28253418e-01
3.70729506e-01 -2.10808933e-01 -9.05171871e-01 5.94867527e-01
1.57453969e-01 4.46555734e-01 2.83590108e-01 -2.88252801e-01
-3.59225571e-01 7.56774098e-02 -6.26677990e-01 7.04205394e-01
1.22194993e+00 -4.49937254e-01 -5.02668917e-01 -1.20275907e-01
7.70741463e-01 -9.74910796e-01 -1.00759280e+00 5.13481319e-01
9.31195199e-01 -1.18120158e+00 1.29696846e+00 -8.46324041e-02
7.78623879e-01 -6.53910100e-01 -1.25583887e-01 -1.04436934e+00
-3.35814416e-01 -4.43288893e-01 -6.01257324e-01 7.80151963e-01
-2.23613441e-01 -2.23504663e-01 8.55658829e-01 1.76666886e-01
-3.06875370e-02 -1.05351055e+00 -1.05845225e+00 -8.83517444e-01
-1.91214278e-01 -5.13183534e-01 2.46650472e-01 6.40083253e-01
-2.27283418e-01 1.45420909e-01 -9.18267965e-01 9.57721099e-02
7.84894764e-01 -5.17819971e-02 1.09296167e+00 -6.53217793e-01
-6.70588255e-01 -2.02951759e-01 -5.74069321e-01 -1.66084671e+00
-6.49033710e-02 -3.01257551e-01 1.08176343e-01 -1.20968771e+00
2.22093254e-01 -6.13719113e-02 -1.90345675e-01 3.82178366e-01
-3.17348659e-01 3.92813951e-01 5.29622316e-01 2.93927014e-01
-5.81301272e-01 6.71302080e-01 1.59716427e+00 1.91863626e-02
-9.37012509e-02 -9.86015499e-02 2.32494473e-01 8.62294912e-01
6.99732661e-01 -4.87816662e-01 -3.74606371e-01 -5.30424953e-01
-1.03835694e-01 3.94376725e-01 8.80230963e-01 -1.31146097e+00
4.48260427e-01 -1.11185469e-01 6.06180489e-01 -1.04409575e+00
6.61057711e-01 -6.74240410e-01 6.00260973e-01 7.20375717e-01
-1.30538657e-01 4.13705349e-01 4.22800072e-02 5.59653938e-01
-3.57969940e-01 2.84910738e-01 7.27241576e-01 -7.25501403e-02
-8.79106820e-01 6.47325695e-01 -1.21339440e-01 9.29828212e-02
8.96424949e-01 -1.69822931e-01 1.78053100e-02 -3.75967711e-01
-7.38837659e-01 3.95633042e-01 5.60923040e-01 5.80060303e-01
8.01423073e-01 -1.55214381e+00 -6.32046044e-01 -5.12044840e-02
-7.78455138e-02 2.65036494e-01 6.80140138e-01 1.09636211e+00
-6.85055137e-01 2.86902428e-01 -4.05234516e-01 -9.01982784e-01
-1.46913123e+00 2.74305731e-01 3.62654835e-01 -5.48701227e-01
-9.86772060e-01 8.65268409e-01 3.43142331e-01 -1.01306394e-01
3.51195931e-01 -2.51833826e-01 1.84242859e-01 -2.70121366e-01
4.85442132e-01 6.39729321e-01 -3.31959188e-01 -9.97100234e-01
-5.50426900e-01 9.08855855e-01 9.76594836e-02 -2.49140963e-01
1.17951977e+00 -1.83388397e-01 -4.32272591e-02 3.89568657e-01
1.54982078e+00 -2.49701008e-01 -1.69630396e+00 -4.38312978e-01
-3.34510684e-01 -6.51341558e-01 -3.47496271e-01 -3.67215753e-01
-1.34720731e+00 9.26143050e-01 6.08801603e-01 -6.00194633e-01
1.15005171e+00 -6.64462298e-02 1.31966150e+00 1.37005940e-01
5.60166597e-01 -9.66128409e-01 5.35035312e-01 6.93334997e-01
9.25893068e-01 -9.80017960e-01 2.59706140e-01 -5.48305929e-01
-5.73881984e-01 1.01577997e+00 7.32513964e-01 -3.17545712e-01
6.73273802e-01 3.38557541e-01 -5.50832041e-02 -1.57869563e-01
-6.97475016e-01 2.76516192e-02 6.30684257e-01 3.69755507e-01
5.58328390e-01 -1.09314740e-01 -2.23828763e-01 3.30924690e-01
-7.20231012e-02 1.19722351e-01 1.92462765e-02 1.20887649e+00
-6.21500239e-02 -1.21498394e+00 -4.70381886e-01 -5.99272773e-02
-4.07675534e-01 1.95828825e-01 -1.18278582e-02 9.43855107e-01
1.94159999e-01 5.78831077e-01 4.94811088e-02 -6.15290284e-01
4.27037627e-01 -2.32991204e-01 6.78484142e-01 -1.77149475e-01
-3.46287727e-01 6.04178667e-01 -1.96332335e-02 -1.07055116e+00
-6.87652707e-01 -7.18295276e-01 -1.42379594e+00 -5.51557600e-01
1.26352280e-01 1.22581180e-02 5.70482165e-02 6.79179788e-01
2.83474684e-01 6.54994726e-01 3.18159640e-01 -1.20454276e+00
-5.04774824e-02 -8.25216055e-01 -3.59397799e-01 3.28259021e-01
4.07324672e-01 -9.85823393e-01 3.29433754e-02 2.55559474e-01] | [7.199881076812744, -0.632171630859375] |
20547f04-f4de-47cc-a5d4-4fed2dfaa2f6 | netflick-adversarial-flickering-attacks-on | 2304.01441 | null | https://arxiv.org/abs/2304.01441v1 | https://arxiv.org/pdf/2304.01441v1.pdf | NetFlick: Adversarial Flickering Attacks on Deep Learning Based Video Compression | Video compression plays a significant role in IoT devices for the efficient transport of visual data while satisfying all underlying bandwidth constraints. Deep learning-based video compression methods are rapidly replacing traditional algorithms and providing state-of-the-art results on edge devices. However, recently developed adversarial attacks demonstrate that digitally crafted perturbations can break the Rate-Distortion relationship of video compression. In this work, we present a real-world LED attack to target video compression frameworks. Our physically realizable attack, dubbed NetFlick, can degrade the spatio-temporal correlation between successive frames by injecting flickering temporal perturbations. In addition, we propose universal perturbations that can downgrade performance of incoming video without prior knowledge of the contents. Experimental results demonstrate that NetFlick can successfully deteriorate the performance of video compression frameworks in both digital- and physical-settings and can be further extended to attack downstream video classification networks. | ['Farinaz Koushanfar', 'Seira Hidano', 'Mojan Javaheripi', 'Shehzeen Samarah Hussain', 'Nojan Sheybani', 'Jung-Woo Chang'] | 2023-04-04 | null | null | null | null | ['video-classification'] | ['computer-vision'] | [ 4.75165129e-01 -6.09075315e-02 -3.91418636e-01 1.07618392e-01
-5.42344034e-01 -7.04016805e-01 3.09580475e-01 -3.96256834e-01
-1.09117053e-01 5.97313106e-01 1.16360351e-01 -7.80368984e-01
-4.16029058e-02 -6.48103297e-01 -1.23388410e+00 -7.47933090e-01
-6.85928285e-01 -3.69800150e-01 1.39167368e-01 7.52195567e-02
-1.41296983e-02 4.93548602e-01 -8.59417319e-01 3.76128942e-01
5.11737689e-02 1.53112328e+00 -2.69639045e-01 1.16137242e+00
7.06657588e-01 1.03270972e+00 -8.39488089e-01 -5.41019440e-01
6.80119038e-01 -2.79200405e-01 -2.58273125e-01 -1.61102101e-01
2.68098384e-01 -1.18945813e+00 -1.62904394e+00 1.14507830e+00
5.42478979e-01 -3.65301132e-01 3.21169138e-01 -1.59584844e+00
-6.30890369e-01 7.73505330e-01 -2.62660980e-01 5.44167459e-01
7.59276971e-02 4.79490101e-01 4.67267424e-01 -7.71616772e-02
1.54781535e-01 8.60570312e-01 7.49529302e-01 7.70585239e-01
-7.47017741e-01 -1.10198557e+00 -1.73797041e-01 7.27193058e-01
-1.28439116e+00 -8.49863052e-01 8.23849380e-01 1.30326658e-01
9.23790336e-01 3.85824382e-01 4.64475393e-01 1.63844252e+00
5.62868893e-01 5.96319675e-01 4.20830280e-01 -5.83395474e-02
3.57054591e-01 -4.96343702e-01 -7.53910899e-01 3.46545875e-01
4.44502205e-01 5.64094961e-01 -4.94108200e-01 -2.02157035e-01
8.33464503e-01 -6.25464842e-02 -6.39759302e-01 -1.05543055e-01
-8.39304388e-01 3.84654075e-01 3.45610082e-01 -2.10563883e-01
-1.97667792e-01 1.17851651e+00 7.25961626e-01 4.24932450e-01
1.60148829e-01 -2.15437692e-02 -3.28841925e-01 -5.53611040e-01
-8.50399852e-01 -2.42293358e-01 5.55997431e-01 1.22440135e+00
-2.00881258e-01 4.69906390e-01 -8.78984705e-02 -1.97244361e-01
2.38719776e-01 6.99296534e-01 2.76473939e-01 -1.35765743e+00
7.06874788e-01 -4.61627007e-01 1.32119611e-01 -1.16111231e+00
-1.36031523e-01 -3.09510410e-01 -1.00343251e+00 2.85615861e-01
2.25894719e-01 -5.69623351e-01 -5.40274858e-01 1.58551037e+00
-1.82751283e-01 9.76012826e-01 2.19204515e-01 8.90746415e-01
3.26094240e-01 7.12563157e-01 -1.92787588e-01 -3.72168779e-01
7.14915574e-01 -6.90221727e-01 -7.25046933e-01 -3.01852487e-02
1.60479277e-01 -5.10124147e-01 2.88841158e-01 6.26049221e-01
-1.47641754e+00 -3.56890976e-01 -1.57040882e+00 3.84796798e-01
1.87814295e-01 -4.91463125e-01 3.76557112e-01 1.27389050e+00
-9.86990750e-01 6.06742799e-01 -1.01726842e+00 3.01997274e-01
9.16674912e-01 5.47562242e-01 3.77751738e-02 -1.08010538e-01
-1.20545268e+00 1.89838693e-01 1.14846133e-01 1.47829596e-02
-1.48225749e+00 -9.20425236e-01 -5.28276622e-01 3.49480242e-01
3.36190999e-01 -6.15511298e-01 1.26920426e+00 -7.67495871e-01
-1.54809284e+00 2.41750225e-01 5.11429012e-01 -1.27111602e+00
8.89104843e-01 -4.49735969e-01 -9.30604339e-01 7.67218471e-01
-5.77092707e-01 4.03021097e-01 1.37474310e+00 -1.06586766e+00
-3.90709847e-01 6.33829981e-02 1.27184242e-01 -3.04149956e-01
-8.08500886e-01 -3.35209742e-02 -3.62031013e-01 -8.10611069e-01
-3.71840924e-01 -8.39400232e-01 -7.51478970e-02 3.89205277e-01
-4.76161122e-01 5.29307187e-01 1.52358675e+00 -4.56229836e-01
1.04844379e+00 -2.30252838e+00 -2.46702895e-01 6.46644756e-02
4.14738625e-01 7.39092290e-01 -1.09913453e-01 3.83880109e-01
1.38001934e-01 4.55308497e-01 1.55585721e-01 -1.46630466e-01
-2.01808568e-02 6.78937212e-02 -7.02689052e-01 7.79583156e-01
-2.08644137e-01 9.32379365e-01 -8.29378366e-01 3.83991599e-02
3.90262812e-01 8.74473393e-01 -7.30856538e-01 6.06277287e-02
-2.71357924e-01 3.83094966e-01 -3.16084117e-01 3.72812361e-01
8.93716395e-01 4.37052958e-02 2.35276341e-01 -4.39891487e-01
5.82522035e-01 1.62930533e-01 -5.51576436e-01 1.28324282e+00
-4.16831315e-01 1.22141802e+00 2.23357469e-01 -9.44527507e-01
2.57251710e-01 5.20942092e-01 6.54565334e-01 -7.75718331e-01
5.98537683e-01 -1.25750706e-01 -2.48149391e-02 -5.66110194e-01
1.77083269e-01 2.03591511e-01 -2.45488305e-02 2.73189247e-01
-2.91773260e-01 2.04021096e-01 -5.47428906e-01 2.63361841e-01
1.69770038e+00 -2.80537903e-01 -1.27990723e-01 7.31302127e-02
-3.25649045e-02 -6.53569162e-01 4.24304307e-01 1.06906688e+00
-6.23907149e-01 4.05832052e-01 3.81730378e-01 -2.46414885e-01
-1.40252995e+00 -1.39731014e+00 5.56971617e-02 4.03820902e-01
4.99235451e-01 -5.83756447e-01 -8.92617404e-01 -6.41525865e-01
-9.31169186e-03 5.73729932e-01 -9.52873677e-02 -7.39490330e-01
-5.30442834e-01 -2.87424654e-01 1.41803408e+00 7.11645544e-01
9.50343907e-01 -4.23573524e-01 -8.27970564e-01 1.56024680e-01
-3.43083665e-02 -1.81984615e+00 -5.63294232e-01 -1.56987235e-01
-4.68452662e-01 -1.04864979e+00 -3.70762944e-01 -3.66223007e-01
2.38714695e-01 4.83081430e-01 7.30949461e-01 1.02445439e-01
-1.91230237e-01 6.47015393e-01 -3.38966280e-01 -1.79844111e-01
-7.60590017e-01 -3.39592755e-01 2.85239041e-01 3.72713543e-02
8.46696645e-02 -1.03788698e+00 -8.06688905e-01 3.82401973e-01
-1.02135897e+00 -3.81230354e-01 1.89749092e-01 3.87431443e-01
1.90139309e-01 8.77289057e-01 3.68867874e-01 -1.87338039e-01
3.63382012e-01 -4.22871798e-01 -5.96110821e-01 -8.09024498e-02
-9.93324742e-02 -1.46737903e-01 1.05420077e+00 -6.85094357e-01
-6.02460682e-01 -4.57286507e-01 -1.58795848e-01 -1.08053136e+00
4.15386744e-02 6.83685094e-02 -4.45143163e-01 -5.36351681e-01
5.57715058e-01 6.79444969e-02 -2.20849529e-01 7.82204035e-04
2.70449966e-01 7.37469792e-01 1.04436529e+00 -2.79312760e-01
1.15382230e+00 8.80719543e-01 4.30967450e-01 -9.03204024e-01
-1.80744216e-01 2.55633354e-01 9.03515890e-02 -4.99084353e-01
7.07250774e-01 -1.10709250e+00 -1.11834431e+00 6.55868649e-01
-1.03970873e+00 -5.92175186e-01 -5.78053258e-02 4.98380572e-01
-7.27153957e-01 5.79399049e-01 -9.14246082e-01 -4.91028249e-01
-2.57488579e-01 -1.11921465e+00 8.00572991e-01 -1.47504602e-02
3.76581937e-01 -8.92389715e-01 -5.16353428e-01 2.69126326e-01
6.77085161e-01 4.64499176e-01 6.23764217e-01 -2.45058343e-01
-9.51688349e-01 -3.17840219e-01 -1.38430372e-01 5.08899212e-01
-5.31684421e-02 -5.70320673e-02 -1.00989640e+00 -7.16716528e-01
4.01314110e-01 -2.74559528e-01 6.87478304e-01 4.91225153e-01
1.62907875e+00 -7.65534103e-01 -3.00329119e-01 1.29066646e+00
1.26228464e+00 3.75871658e-01 1.16466522e+00 2.19329357e-01
6.59784138e-01 -3.56647998e-01 7.79452175e-02 7.61053860e-01
-3.76054436e-01 6.75802529e-01 9.63906765e-01 3.76413107e-01
-1.74098507e-01 -3.49761397e-01 7.37807393e-01 2.39201829e-01
2.62676269e-01 -9.75081921e-01 -4.18181121e-01 2.18188781e-02
-1.55208850e+00 -1.23817098e+00 5.35853691e-02 2.11718631e+00
2.65700519e-01 8.32929611e-01 -7.75501728e-02 3.79363775e-01
7.08979070e-01 2.65197188e-01 -8.96008551e-01 -2.33053640e-01
-8.97033289e-02 6.01752028e-02 1.32237554e+00 2.01611102e-01
-1.17308414e+00 5.80268502e-01 6.68790579e+00 8.82506013e-01
-1.26063490e+00 1.75966650e-01 5.76660931e-01 -4.95721370e-01
-9.19528008e-02 -4.60562736e-01 -2.79996067e-01 6.99699402e-01
1.38711941e+00 -3.73756826e-01 6.90009058e-01 7.75487840e-01
4.25461888e-01 4.49389040e-01 -1.19762635e+00 1.27894723e+00
2.48876780e-01 -1.64982831e+00 -2.37910263e-03 3.16983670e-01
5.01624465e-01 1.14133731e-01 6.12616599e-01 -1.73009112e-01
2.61406094e-01 -9.28716421e-01 7.71899343e-01 4.66507189e-02
1.24455547e+00 -9.28677738e-01 4.50995326e-01 3.28970514e-02
-1.04485965e+00 -4.80232060e-01 -3.63348246e-01 -4.10657972e-02
4.80219066e-01 4.68746215e-01 -3.55883807e-01 1.96205258e-01
7.11053789e-01 6.03583455e-01 -2.11293340e-01 1.08627772e+00
-2.30613768e-01 9.93029714e-01 -2.86904454e-01 3.74445260e-01
2.29873225e-01 4.29937690e-01 9.49676335e-01 9.47521031e-01
3.48118991e-01 1.11504577e-01 -4.63462956e-02 4.21520323e-01
-5.71393728e-01 -8.04233491e-01 -7.39965916e-01 4.10149321e-02
5.85982621e-01 4.55033064e-01 -4.15130705e-01 -1.16683647e-01
-2.35132709e-01 1.29727733e+00 -4.54341888e-01 5.02765179e-01
-1.52171290e+00 -1.91971824e-01 1.20550871e+00 4.47135530e-02
6.46981835e-01 -4.62174624e-01 -6.35743365e-02 -1.11125422e+00
1.31095141e-01 -9.20910537e-01 8.43509585e-02 -8.15343797e-01
-1.02005589e+00 2.05426842e-01 -1.65032819e-01 -1.36454701e+00
2.05913447e-02 -5.73571563e-01 -5.84411979e-01 3.00988294e-02
-1.34373617e+00 -7.01892436e-01 -2.36235574e-01 8.89633179e-01
4.45198566e-01 -4.08862501e-01 4.94118363e-01 6.27694905e-01
-4.79408145e-01 1.12688899e+00 4.90969628e-01 3.75696152e-01
3.11246514e-01 -4.63989973e-01 7.20012665e-01 1.33127749e+00
9.16751772e-02 4.71615512e-03 9.45509017e-01 -5.87742090e-01
-1.82198966e+00 -1.34811020e+00 -2.08828494e-01 1.61081001e-01
7.53233433e-01 -4.59216952e-01 -4.07229304e-01 6.79317594e-01
4.30547625e-01 4.53987896e-01 5.71215570e-01 -9.71180737e-01
-6.66030407e-01 -2.36022249e-01 -1.42060184e+00 7.76352346e-01
1.24609911e+00 -8.09010386e-01 1.23891212e-01 4.40138519e-01
1.04866564e+00 -4.38178062e-01 -5.68809152e-01 4.25401807e-01
6.19567156e-01 -1.00249922e+00 1.20846462e+00 -8.42754185e-01
3.01775932e-01 -1.32409036e-01 -5.42728484e-01 -1.03853977e+00
-7.85932094e-02 -1.58474267e+00 -7.75464356e-01 9.52444851e-01
3.04805357e-02 -3.33858043e-01 1.01123524e+00 3.57806593e-01
-1.48221493e-01 -1.44625545e-01 -1.18945730e+00 -1.38315535e+00
-1.62709635e-02 -9.31162298e-01 4.20831710e-01 5.71351409e-01
4.52099554e-02 -3.21196079e-01 -8.51030707e-01 8.49792302e-01
8.27424407e-01 -7.49964654e-01 6.34135842e-01 -2.99365163e-01
-7.18916833e-01 -3.63085777e-01 -1.08002949e+00 -1.33166385e+00
1.38339520e-01 -3.89191121e-01 -2.62397885e-01 -7.63973713e-01
-2.17329890e-01 -7.41650090e-02 -3.01688492e-01 -1.74501464e-01
3.40150416e-01 5.75737357e-01 3.72449309e-01 5.96752390e-02
-5.89193821e-01 4.49345052e-01 7.28295028e-01 -3.96139175e-01
2.03428060e-01 7.80201256e-02 -3.97264600e-01 4.65866178e-01
1.10376143e+00 -4.32253599e-01 -7.63306320e-01 -6.82348073e-01
1.94837287e-01 2.33245313e-01 6.31907523e-01 -1.70624125e+00
3.94775569e-01 -1.81598719e-02 2.44489610e-01 -9.47480742e-03
4.13716286e-01 -1.42220902e+00 2.07584336e-01 6.74764991e-01
-3.05397302e-01 -6.50821850e-02 3.45439792e-01 9.71639574e-01
3.16097170e-01 2.44206741e-01 8.03995371e-01 3.35060120e-01
-3.33991915e-01 4.47999656e-01 -6.94655538e-01 2.06421435e-01
1.36553514e+00 -7.89330602e-02 -7.21017420e-01 -9.72723424e-01
-4.83976036e-01 -1.16036601e-01 3.87256116e-01 3.56614083e-01
8.52859735e-01 -1.18454576e+00 -5.83578289e-01 2.24189982e-01
-3.70993614e-01 -6.46791697e-01 3.93052220e-01 4.15097624e-01
-8.37015569e-01 4.12242502e-01 -2.88761288e-01 -5.20495236e-01
-1.23764181e+00 1.15392292e+00 5.83389699e-01 1.83431089e-01
-9.16338146e-01 7.05044270e-01 -1.10944457e-01 7.49638081e-01
7.58089483e-01 -2.75426298e-01 5.59534192e-01 -9.01121378e-01
8.56198132e-01 4.36684579e-01 5.52063398e-02 -2.52926826e-01
-9.17723998e-02 8.18191022e-02 1.33136973e-01 -3.25488336e-02
9.08039033e-01 -3.75644177e-01 5.79418719e-01 -3.83569509e-01
1.43248844e+00 3.24622579e-02 -1.77726924e+00 1.39061451e-01
-7.44341135e-01 -7.04276681e-01 4.41808671e-01 -4.69193548e-01
-1.63017738e+00 7.32340276e-01 8.33870113e-01 4.71525282e-01
1.36673892e+00 -2.65166134e-01 1.52308190e+00 4.58839983e-01
5.34973085e-01 -7.19919264e-01 2.53496885e-01 -1.30263239e-01
2.32893854e-01 -6.56478763e-01 -1.92033723e-01 -4.84340250e-01
-4.45406176e-02 1.04266572e+00 2.01005816e-01 -1.68576494e-01
9.69213724e-01 9.07643199e-01 -2.32361525e-01 1.92869723e-01
-7.19358265e-01 4.55902845e-01 -3.33728135e-01 9.67794061e-01
-3.63629639e-01 3.74477245e-02 2.39753738e-01 2.52026021e-01
-1.68080218e-02 6.98313713e-02 9.54540074e-01 8.62151206e-01
-2.21371233e-01 -8.93981934e-01 -4.56856489e-01 3.66272211e-01
-8.49167526e-01 -8.68287608e-02 -9.41067375e-03 4.31942940e-01
3.58709805e-02 1.33670008e+00 1.95006162e-01 -9.23441589e-01
-4.00180444e-02 -5.31425953e-01 4.44720089e-01 3.06971341e-01
-1.14330955e-01 -1.40643001e-01 1.55806631e-01 -9.79905188e-01
-1.87723801e-01 -4.06007677e-01 -9.84002471e-01 -8.86615396e-01
1.09458350e-01 -2.39946201e-01 6.13881111e-01 9.87240016e-01
5.01775920e-01 7.16771245e-01 9.57655191e-01 -9.02073681e-01
-6.21535003e-01 -2.14614585e-01 -2.86095828e-01 3.56977701e-01
8.35200131e-01 -3.04373950e-01 -7.35678971e-01 1.56887650e-01] | [5.389522552490234, 7.9357757568359375] |
1c31579f-8073-4d12-80a0-af84604eb5fb | li-net-large-pose-identity-preserving-face | 2104.02850 | null | https://arxiv.org/abs/2104.02850v1 | https://arxiv.org/pdf/2104.02850v1.pdf | LI-Net: Large-Pose Identity-Preserving Face Reenactment Network | Face reenactment is a challenging task, as it is difficult to maintain accurate expression, pose and identity simultaneously. Most existing methods directly apply driving facial landmarks to reenact source faces and ignore the intrinsic gap between two identities, resulting in the identity mismatch issue. Besides, they neglect the entanglement of expression and pose features when encoding driving faces, leading to inaccurate expressions and visual artifacts on large-pose reenacted faces. To address these problems, we propose a Large-pose Identity-preserving face reenactment network, LI-Net. Specifically, the Landmark Transformer is adopted to adjust driving landmark images, which aims to narrow the identity gap between driving and source landmark images. Then the Face Rotation Module and the Expression Enhancing Generator decouple the transformed landmark image into pose and expression features, and reenact those attributes separately to generate identity-preserving faces with accurate expressions and poses. Both qualitative and quantitative experimental results demonstrate the superiority of our method. | ['Jizhong Han', 'Jiao Dai', 'Shuqiao Zou', 'Cai Yu', 'Zhaoxing Li', 'Tao Liang', 'Peng Chen', 'Jin Liu'] | 2021-04-07 | null | null | null | null | ['face-reenactment'] | ['computer-vision'] | [ 9.36804935e-02 4.18647602e-02 -3.37992385e-02 -7.91199207e-01
-3.58526260e-01 -7.10017741e-01 4.86994326e-01 -7.22461522e-01
-4.05577570e-02 4.34354961e-01 2.44761363e-01 4.84207869e-01
4.33079712e-02 -6.66404784e-01 -5.09627283e-01 -7.60848343e-01
4.64029580e-01 -3.47991823e-03 -4.42991972e-01 -3.40917200e-01
-1.02129512e-01 7.63478220e-01 -1.56096971e+00 -1.48492575e-01
7.18494058e-01 1.09147036e+00 -3.74098003e-01 -5.97246885e-02
-7.46315941e-02 4.14929032e-01 -4.16353703e-01 -6.54343069e-01
3.82499844e-01 -5.86802781e-01 -2.40638182e-01 2.38454342e-01
6.45946145e-01 -4.44673747e-01 -5.22523046e-01 1.31562221e+00
6.11755967e-01 -1.88282639e-01 2.53005326e-01 -1.76086366e+00
-1.09011436e+00 1.80343002e-01 -9.67100382e-01 -5.25989652e-01
4.37520802e-01 -4.28705104e-02 5.30550778e-01 -1.09739447e+00
6.60230994e-01 1.64684260e+00 7.28991628e-01 7.54640400e-01
-1.35832012e+00 -1.47220588e+00 2.74886429e-01 5.34090307e-03
-1.84604907e+00 -9.55511451e-01 1.35187399e+00 -1.26190200e-01
-1.28869504e-01 2.51199752e-01 7.50284433e-01 1.04980123e+00
-3.88703756e-02 4.38924968e-01 1.16050696e+00 -1.61647387e-02
-2.60229826e-01 1.90375298e-01 -5.23301840e-01 7.94794202e-01
1.13350168e-01 1.53422937e-01 -5.75154543e-01 2.38741934e-02
1.16630507e+00 2.07157940e-01 -3.46138149e-01 -5.44011950e-01
-1.12250841e+00 5.71107745e-01 5.99221230e-01 1.81943908e-01
-2.20431164e-01 -4.73540314e-02 2.77296275e-01 2.24678829e-01
2.79169649e-01 2.45780945e-01 -2.11005416e-02 1.74993530e-01
-7.69917488e-01 1.46872222e-01 2.55661547e-01 1.26236296e+00
1.24997795e+00 1.42249390e-01 -2.48208344e-01 9.78104591e-01
3.44501048e-01 5.89456320e-01 2.72650063e-01 -9.58005130e-01
-8.86707939e-03 7.18854308e-01 -1.77088112e-01 -1.65942621e+00
-2.86743879e-01 -3.98541093e-01 -1.01984572e+00 6.12231204e-03
-1.01351440e-01 -4.76953238e-02 -7.37802744e-01 2.29805565e+00
4.58485186e-01 2.19468579e-01 -1.16730630e-01 1.00552571e+00
1.08831203e+00 3.83447617e-01 -1.25459591e-02 -3.85813028e-01
1.41541409e+00 -6.96686864e-01 -1.01999056e+00 -1.80824086e-01
2.09460542e-01 -9.15286481e-01 9.75220680e-01 -1.16706021e-01
-9.32774961e-01 -5.67706645e-01 -1.17133987e+00 -3.60184669e-01
1.71365514e-02 3.60393494e-01 5.97892463e-01 5.82192004e-01
-9.93327022e-01 2.75968969e-01 -4.20980036e-01 2.40706336e-02
5.45149446e-01 5.90606034e-01 -9.05815661e-01 -2.37949640e-02
-1.22725642e+00 6.03719771e-01 4.31748666e-02 5.77777505e-01
-5.18725038e-01 -7.99121022e-01 -1.12678516e+00 -1.72629252e-01
1.56053647e-01 -4.20845628e-01 9.52008903e-01 -1.36514223e+00
-1.59257269e+00 1.17244732e+00 -2.53962994e-01 4.97752279e-01
6.49663925e-01 1.03292011e-01 -7.49882817e-01 -1.68241799e-01
2.55764782e-01 9.33976114e-01 1.11145568e+00 -1.51469553e+00
-3.32643777e-01 -7.03592420e-01 -8.13869610e-02 1.92703426e-01
-5.01064181e-01 -1.67175665e-01 -5.45041680e-01 -7.97001660e-01
5.75546980e-01 -8.26704144e-01 1.41976535e-01 4.70672101e-01
-3.27188671e-01 1.69645369e-01 1.12331402e+00 -6.13194406e-01
8.37040901e-01 -2.39334083e+00 1.11301273e-01 2.52381831e-01
3.54440659e-01 -1.31043836e-01 -5.55463254e-01 -7.36815929e-02
-4.73306447e-01 -2.48300079e-02 -3.15938541e-03 -3.13048929e-01
-1.48065612e-01 2.13975802e-01 -3.07057619e-01 6.68041825e-01
3.58518392e-01 1.06895459e+00 -7.98953354e-01 -6.76669002e-01
3.69626284e-02 8.84880543e-01 -5.09783506e-01 1.25942469e-01
1.69786006e-01 7.76897013e-01 -4.57086980e-01 6.99701667e-01
1.35197318e+00 3.65302175e-01 2.34376993e-02 -8.68952632e-01
-1.36893570e-01 -2.03573838e-01 -1.08258688e+00 1.84206235e+00
-5.09948313e-01 3.42586786e-01 4.03027892e-01 -4.22702551e-01
1.41491294e+00 1.19375601e-01 5.87492943e-01 -8.36886227e-01
4.38644946e-01 2.63309091e-01 -2.86179245e-01 -1.94247514e-01
2.12606758e-01 -4.57335114e-01 -4.61651850e-03 2.44075552e-01
-1.04868792e-01 -2.20051959e-01 -3.98957312e-01 -9.99873281e-02
2.02568188e-01 2.57140040e-01 -5.38600944e-02 -3.48431468e-01
8.40572417e-01 -6.10944688e-01 9.96402085e-01 -2.40156308e-01
-1.88378870e-01 7.36352026e-01 5.93381226e-01 -5.75395882e-01
-8.93322587e-01 -1.15426862e+00 -2.07515001e-01 8.70176196e-01
6.68573737e-01 -3.54175299e-01 -9.07608032e-01 -4.34438854e-01
-8.40070248e-02 2.98631847e-01 -7.97248304e-01 -6.75324976e-01
-6.67700827e-01 -5.39084315e-01 6.47858977e-01 2.87328631e-01
8.51129889e-01 -7.20179617e-01 1.08944587e-01 -6.96348771e-02
-2.89739847e-01 -1.00139606e+00 -1.04957175e+00 -7.06886828e-01
-3.41965795e-01 -9.67923880e-01 -6.45976365e-01 -1.11155283e+00
1.10457921e+00 1.83077857e-01 5.28021097e-01 2.27976397e-01
-1.40591234e-01 -1.45058766e-01 -2.70627253e-02 -1.75593600e-01
-2.58983433e-01 -1.87117428e-01 3.44802231e-01 6.58078134e-01
2.79532254e-01 -9.78526652e-01 -6.98418677e-01 5.41438222e-01
-8.32390249e-01 2.44962797e-01 4.87338156e-01 7.85118818e-01
5.12047112e-01 -1.59319773e-01 7.23480403e-01 -4.88015205e-01
3.93731058e-01 3.28958370e-02 -4.97330725e-01 2.20972821e-01
-4.15636480e-01 -1.01121470e-01 5.47104836e-01 -5.67826629e-01
-1.00193131e+00 2.86207587e-01 -2.01531067e-01 -7.18531668e-01
2.35858321e-01 -8.90660435e-02 -9.32285607e-01 -5.38361847e-01
3.35656703e-01 2.91893154e-01 4.08526927e-01 -2.45636865e-01
4.41627949e-01 4.64221209e-01 9.12753999e-01 -6.27198458e-01
1.19849288e+00 6.49459898e-01 9.92318541e-02 -5.00015318e-01
-6.98671818e-01 2.51122832e-01 -7.97086835e-01 -2.47751132e-01
5.41366518e-01 -1.06851077e+00 -9.70015824e-01 7.18535125e-01
-1.26156616e+00 4.70582694e-01 -1.35293022e-01 1.17691591e-01
-4.00487363e-01 2.54435569e-01 -4.61781561e-01 -3.01221490e-01
-2.88454860e-01 -1.10449314e+00 1.17319477e+00 6.01122499e-01
9.03806463e-03 -3.70984077e-01 -1.65278584e-01 -9.34683383e-02
4.14243877e-01 4.88994122e-01 7.77472556e-01 1.11634165e-01
-3.83098811e-01 -2.30082482e-01 -5.58808327e-01 1.89947605e-01
7.54602909e-01 1.76535904e-01 -1.07767308e+00 -4.78333622e-01
1.45965919e-01 -1.80582777e-01 3.56021881e-01 -7.80015215e-02
9.86241579e-01 -4.92736459e-01 -2.97122896e-01 1.09258139e+00
9.26948011e-01 2.74291217e-01 7.08166242e-01 -1.21320961e-02
9.69620705e-01 9.82866585e-01 4.95467067e-01 3.33501428e-01
3.50441873e-01 6.63863361e-01 2.16846198e-01 -3.75613958e-01
-1.02300175e-01 -9.17625844e-01 2.17208982e-01 8.12429190e-01
1.67183980e-01 3.99259180e-01 -2.99400747e-01 3.18736881e-01
-1.48098445e+00 -6.11602008e-01 4.39633727e-01 2.09286380e+00
1.05250347e+00 -4.20067102e-01 -1.35855615e-01 -5.01709618e-02
1.14199960e+00 2.15506017e-01 -7.27718890e-01 5.96578456e-02
-2.43234038e-01 6.19952008e-02 1.23727210e-01 3.85441482e-01
-8.78068447e-01 1.04493749e+00 5.71133280e+00 8.62939358e-01
-1.41533101e+00 1.15021177e-01 7.81700552e-01 -1.66909605e-01
-5.28979659e-01 1.24357557e-02 -6.15112007e-01 3.65362585e-01
-1.60046648e-02 -3.87086391e-01 4.88772839e-01 9.20649827e-01
9.15761963e-02 5.96496999e-01 -1.03701401e+00 1.49870145e+00
3.49931449e-01 -1.00553238e+00 3.57892096e-01 -1.41391248e-01
6.17927969e-01 -8.06826472e-01 2.07524329e-01 5.59590608e-02
-1.55609399e-01 -1.24089730e+00 8.87880147e-01 4.71174031e-01
1.39682698e+00 -1.08842254e+00 3.88495952e-01 -2.43039042e-01
-1.44523823e+00 8.73747766e-02 -3.63943934e-01 4.19999100e-02
2.21812263e-01 2.79403657e-01 -3.46582502e-01 4.84896302e-01
4.92527604e-01 8.62622917e-01 -5.34472525e-01 4.53728497e-01
-3.29122722e-01 -2.54417300e-01 -2.64566332e-01 2.70677805e-01
-1.48228854e-01 -6.15030587e-01 4.89275038e-01 5.90505838e-01
5.17609060e-01 3.49632129e-02 1.18156411e-01 1.32173216e+00
-4.60817873e-01 1.80496573e-01 -6.03107810e-01 1.05939798e-01
8.05799186e-01 1.58707118e+00 -4.27308619e-01 -2.74496824e-02
-1.49025768e-01 1.15171707e+00 6.46253768e-03 2.97237486e-01
-9.07922328e-01 -4.36963499e-01 1.19065869e+00 1.52834341e-01
-2.88206279e-01 -3.53621691e-03 -1.89583302e-01 -1.23753810e+00
2.26641387e-01 -8.66184175e-01 -1.31204560e-01 -8.25608850e-01
-1.16066265e+00 8.32388878e-01 -1.53653070e-01 -1.33271337e+00
1.40757799e-01 -2.72891343e-01 -5.65863669e-01 8.71826172e-01
-1.30922449e+00 -1.82420266e+00 -7.02368140e-01 6.67315900e-01
-5.47331981e-02 8.92842561e-02 5.63173354e-01 6.64961338e-01
-7.41376758e-01 1.19542968e+00 -3.05147499e-01 3.16504657e-01
9.68376160e-01 -4.31851685e-01 2.40234569e-01 4.75304425e-01
-1.28013879e-01 8.21528971e-01 3.92834783e-01 -4.66699272e-01
-1.39840853e+00 -1.24073756e+00 5.92415035e-01 -1.97869807e-01
2.12537393e-01 -7.54531324e-01 -8.57116520e-01 7.34755397e-01
-2.47351348e-01 1.66211426e-01 3.41455191e-01 -2.65809864e-01
-6.97800756e-01 -5.86293221e-01 -1.27612925e+00 9.73272085e-01
1.40696132e+00 -8.05583537e-01 -2.99942762e-01 -2.69744769e-02
7.81104028e-01 -3.94108951e-01 -7.39558101e-01 6.26193345e-01
9.39887941e-01 -8.57149243e-01 9.44419324e-01 -2.26780757e-01
3.12544137e-01 -6.52513087e-01 1.01431713e-01 -1.16933572e+00
-4.06461805e-01 -7.99483597e-01 6.92765057e-01 1.85319471e+00
4.06988822e-02 -9.47020054e-01 6.85097814e-01 5.31683683e-01
1.78139731e-01 -5.07518589e-01 -1.01354778e+00 -5.42999506e-01
-1.80253834e-02 1.79422200e-01 1.44322634e+00 1.34365463e+00
-1.95360810e-01 4.10565555e-01 -5.91843367e-01 6.96111172e-02
4.37262863e-01 3.30991775e-01 8.01060677e-01 -1.05685365e+00
4.07359987e-01 -3.72228444e-01 -5.77957094e-01 -8.77408683e-01
5.71553946e-01 -8.34598482e-01 -1.35738710e-02 -7.63018429e-01
9.87969935e-02 -6.08689547e-01 -8.21159407e-02 6.47883415e-01
8.13652128e-02 6.84909999e-01 5.55595793e-02 1.71384275e-01
1.05818443e-01 1.08964097e+00 1.73654532e+00 -2.02329606e-02
-9.90208760e-02 -4.48953062e-01 -1.04194355e+00 6.94422364e-01
5.17669737e-01 -2.68257082e-01 -5.57146192e-01 -3.90199095e-01
4.84366976e-02 -1.46454662e-01 5.29386222e-01 -7.95456409e-01
1.11220740e-01 -2.34009609e-01 7.50742674e-01 -2.60449052e-01
5.13062716e-01 -7.82629728e-01 5.33295095e-01 2.65611917e-01
-1.35993570e-01 2.50647068e-01 2.00796425e-01 1.46578163e-01
-3.27500015e-01 2.55819052e-01 1.14076674e+00 1.79905295e-01
-4.76643503e-01 8.62099648e-01 3.51496041e-01 -2.03257263e-01
1.11014986e+00 -3.82734209e-01 -1.55362234e-01 -5.07908940e-01
-3.88594002e-01 1.44010723e-01 1.01290393e+00 8.24368179e-01
6.25216365e-01 -2.05156207e+00 -7.95089722e-01 9.75122988e-01
2.24953070e-01 1.67204171e-01 4.84461606e-01 7.28653431e-01
-5.36149859e-01 -1.48612872e-01 -6.43387735e-01 -4.26960647e-01
-1.17525458e+00 6.33552253e-01 5.35598099e-01 5.78042507e-01
-3.64619136e-01 8.50659370e-01 6.94826722e-01 -7.20554173e-01
-2.26236477e-01 2.94600278e-01 -6.90849349e-02 1.84353322e-01
5.57014942e-01 1.32844329e-01 -2.20621467e-01 -1.30178797e+00
-4.97473031e-01 1.17615747e+00 -1.57916456e-01 2.85280608e-02
1.02381146e+00 -3.01390231e-01 -6.66072905e-01 -4.63822372e-02
1.57243609e+00 2.21657097e-01 -1.21088386e+00 -1.43971607e-01
-5.25663137e-01 -9.22776699e-01 -1.75260872e-01 -3.24792564e-01
-1.49720752e+00 8.93702626e-01 5.83474457e-01 -4.95425165e-01
1.33568549e+00 -2.57201016e-01 8.90507340e-01 -3.13803814e-02
3.73420238e-01 -8.30872655e-01 6.43001273e-02 1.34355620e-01
1.21747971e+00 -8.50857735e-01 -1.04954377e-01 -8.63924384e-01
-3.30710590e-01 8.85151982e-01 1.05427241e+00 -5.75281419e-02
5.69870591e-01 1.48209482e-01 3.20620656e-01 -6.73906058e-02
-8.04186687e-02 4.38029647e-01 2.41745442e-01 7.94851243e-01
2.61232227e-01 -5.19434586e-02 -1.10500865e-01 8.30719531e-01
-6.41044676e-01 -1.23761602e-01 1.53333440e-01 6.03760064e-01
1.14003234e-01 -1.11839485e+00 -4.26618129e-01 3.00505925e-02
-3.63993533e-02 2.54143891e-03 -5.91245055e-01 6.97522342e-01
3.78690183e-01 6.12311542e-01 1.94411755e-01 -7.67515361e-01
5.14972150e-01 -2.70195842e-01 5.86238921e-01 -2.37807170e-01
-1.48883343e-01 1.66745529e-01 -4.20752496e-01 -5.16267538e-01
-3.23934883e-01 -3.05507988e-01 -1.25919247e+00 -7.20411777e-01
-3.23780596e-01 5.74155673e-02 2.64196187e-01 5.16205013e-01
6.44196093e-01 1.67868137e-01 1.09742558e+00 -9.44096386e-01
-3.48810017e-01 -6.99731588e-01 -7.62189329e-01 6.77353084e-01
4.22808379e-01 -8.61075401e-01 -2.81198442e-01 2.17827968e-02] | [12.699849128723145, -0.04083140566945076] |
f8610014-7ae8-40d2-8e85-afd68fc3409e | ic3d-image-conditioned-3d-diffusion-for-shape | 2211.10865 | null | https://arxiv.org/abs/2211.10865v2 | https://arxiv.org/pdf/2211.10865v2.pdf | IC3D: Image-Conditioned 3D Diffusion for Shape Generation | In the last years, Denoising Diffusion Probabilistic Models (DDPMs) obtained state-of-the-art results in many generative tasks, outperforming GANs and other classes of generative models. In particular, they reached impressive results in various image generation sub-tasks, among which conditional generation tasks such as text-guided image synthesis. Given the success of DDPMs in 2D generation, they have more recently been applied to 3D shape generation, outperforming previous approaches and reaching state-of-the-art results. However, these existing 3D DDPM works make little or no use of guidance, mainly being unconditional or class-conditional. In this work, we present IC3D, an Image-Conditioned 3D Diffusion model that generates 3D shapes by image guidance. To guide our DDPM, we introduce CISP (Contrastive Image-Shape Pre-training), a model jointly embedding images and shapes by contrastive pre-training, inspired by the literature on text-to-image DDPMs. Our generative diffusion model outperforms the state-of-the-art in 3D generation quality and diversity. Furthermore, despite IC3D generative nature, we show that its generated shapes are preferred by human evaluators to a SoTA single-view 3D reconstruction model in terms of quality and coherence to the query image by running a side-by-side human evaluation. Ablation studies show the importance of CISP for learning structural integrity properties, crucial for realistic generation. Such biases yield a regular embedding space and allow for interpolation and conditioning on out-of-distribution images, while also making IC3D capable of generating coherent but diverse completions of occluded views and enabling its adoption in controlled real-life applications. | ['Matteo Matteucci', 'Matteo Frosi', 'Paolo Cudrano', 'Cristian Sbrolli'] | 2022-11-20 | null | null | null | null | ['single-view-3d-reconstruction', '3d-shape-generation'] | ['computer-vision', 'computer-vision'] | [ 4.11354095e-01 4.36161667e-01 1.48070827e-01 -6.71814457e-02
-9.34144974e-01 -6.69350445e-01 1.20699823e+00 -5.50268173e-01
5.95506802e-02 6.02308214e-01 4.24143940e-01 -1.68183729e-01
2.41216403e-02 -9.79125082e-01 -8.92249584e-01 -9.08916354e-01
2.41791159e-01 7.81713426e-01 1.60540566e-02 -8.85120705e-02
-2.65761595e-02 5.13637781e-01 -1.57407403e+00 3.44876170e-01
7.44310677e-01 8.27931643e-01 4.63637292e-01 7.74521768e-01
5.56515679e-02 3.99339020e-01 -5.45677066e-01 -5.60442388e-01
2.99518526e-01 -7.16879010e-01 -4.31284547e-01 4.05629486e-01
5.58458924e-01 -5.33909440e-01 -2.84828454e-01 6.19649351e-01
9.84823644e-01 -1.69206724e-01 1.21135819e+00 -1.13539052e+00
-1.29934216e+00 2.46689349e-01 -4.28158939e-01 -2.18480602e-01
5.12957215e-01 4.94240433e-01 9.13958490e-01 -1.11459446e+00
1.04525614e+00 1.31069052e+00 5.52157402e-01 8.79437983e-01
-1.68363929e+00 -2.20194623e-01 -4.14462984e-02 -3.30111861e-01
-1.13215995e+00 -4.49202329e-01 8.51875424e-01 -5.11924803e-01
8.09132099e-01 1.66838914e-01 6.03079617e-01 1.70612872e+00
1.22144632e-01 1.06621420e+00 1.34968138e+00 -4.12899524e-01
3.09356034e-01 1.45355314e-01 -8.38146389e-01 5.00862122e-01
1.56860352e-02 4.82001901e-01 -5.33498108e-01 -7.66771212e-02
1.20535934e+00 -2.98549831e-01 -4.52568293e-01 -5.62540531e-01
-1.28772581e+00 9.27704096e-01 3.02671939e-01 2.26034328e-01
-6.75848126e-01 3.43497664e-01 -1.51311487e-01 9.39874873e-02
7.56482661e-01 3.34090590e-01 1.98861398e-02 6.35197014e-02
-1.10955513e+00 5.66316187e-01 6.07044578e-01 1.14530337e+00
5.16020536e-01 4.30666327e-01 -7.24961638e-01 8.15466046e-01
3.12110513e-01 8.65617514e-01 3.04933161e-01 -9.75203633e-01
2.22720891e-01 2.53896266e-01 -9.85434745e-03 -7.72179425e-01
-9.33298934e-03 -6.07471406e-01 -1.00966477e+00 4.60273892e-01
1.80204272e-01 -9.36373994e-02 -1.23173678e+00 1.83954382e+00
2.71642655e-01 -1.94111183e-01 2.98511074e-03 8.17729533e-01
7.91403830e-01 7.85110533e-01 -8.48514810e-02 -4.78378348e-02
9.51944292e-01 -6.81497097e-01 -3.30506384e-01 -9.39716995e-02
1.46172434e-01 -9.23437893e-01 1.12740660e+00 5.07234812e-01
-1.39389062e+00 -7.03717113e-01 -6.81531012e-01 7.02795014e-02
-6.88773617e-02 2.76722591e-02 5.45405269e-01 8.70005786e-01
-1.53631747e+00 4.39776808e-01 -5.48425138e-01 -3.47441584e-01
7.66461670e-01 7.78795872e-03 -3.10539305e-01 -3.36608112e-01
-8.57462287e-01 8.21502626e-01 -1.69363664e-03 -1.45976931e-01
-1.57296085e+00 -8.78455281e-01 -8.42074037e-01 -3.21579725e-01
-2.21371017e-02 -1.45677221e+00 9.57915664e-01 -5.41928709e-01
-1.65560484e+00 1.15118635e+00 1.14561822e-02 -4.95610505e-01
8.88463795e-01 -6.11395650e-02 -5.98362014e-02 2.36125961e-01
1.15328491e-01 1.29608560e+00 1.30609059e+00 -1.72868884e+00
-1.29850030e-01 -1.57694310e-01 -6.14498705e-02 2.14828789e-01
2.15628147e-02 -5.60490727e-01 -5.48455119e-01 -9.38761294e-01
-1.64703708e-02 -1.01539040e+00 -3.09541404e-01 5.12978546e-02
-6.11271679e-01 2.29835720e-03 7.45915174e-01 -4.70354706e-01
7.21036017e-01 -1.99646521e+00 3.89683038e-01 2.24735383e-02
2.00342596e-01 5.83036281e-02 -4.36462045e-01 6.21532917e-01
-2.66813915e-02 3.22865248e-01 -4.02106911e-01 -8.04484427e-01
3.30795288e-01 2.45272651e-01 -4.88725752e-01 2.35486984e-01
4.81765926e-01 1.26675928e+00 -7.17519999e-01 -3.77070189e-01
4.40634519e-01 8.93500984e-01 -7.60789633e-01 4.51957673e-01
-4.85108823e-01 8.60175133e-01 -2.38326877e-01 5.34095943e-01
6.66279495e-01 -1.85792193e-01 -1.92819238e-01 -3.06804568e-01
2.85871834e-01 -1.86633199e-01 -8.18209469e-01 1.91955936e+00
-6.83262706e-01 4.24290508e-01 -1.95284694e-01 -6.18121803e-01
8.81551087e-01 3.53015423e-01 3.59992027e-01 -7.73080826e-01
-9.20029804e-02 3.24718297e-01 -3.47286999e-01 -2.19851196e-01
5.52424788e-01 -3.66344452e-01 1.33544832e-01 5.68405986e-01
3.54607373e-01 -9.93542075e-01 8.82399902e-02 4.52953935e-01
7.46778190e-01 5.94672918e-01 -1.47107005e-01 -7.29425997e-02
1.84056118e-01 -3.32623810e-01 -7.16983229e-02 8.94534230e-01
4.09608334e-01 1.55007803e+00 2.47510508e-01 -1.74292494e-02
-1.38097072e+00 -1.45564604e+00 -5.09394445e-02 4.53113884e-01
-1.17929727e-01 -1.20148197e-01 -9.01544213e-01 -6.58086300e-01
-2.56689876e-01 1.05508792e+00 -7.58687139e-01 4.38222429e-03
-2.97694892e-01 -7.58636653e-01 4.66098398e-01 3.55805606e-01
3.10853541e-01 -1.24270082e+00 -3.37955981e-01 3.36241603e-01
-1.32375553e-01 -1.17754066e+00 -5.23093283e-01 -8.19403529e-02
-8.75253260e-01 -6.99919581e-01 -1.60603523e+00 -5.82442522e-01
8.75004411e-01 6.71716854e-02 1.56154048e+00 -1.89622924e-01
-2.80023396e-01 9.00868118e-01 -3.18230689e-01 -2.39751458e-01
-8.85135233e-01 -4.87969443e-02 -1.81598932e-01 1.65592153e-02
-4.54677045e-01 -1.05400670e+00 -9.45688903e-01 3.61596465e-01
-1.27072453e+00 3.50353956e-01 8.45786750e-01 9.23089981e-01
8.13013494e-01 -1.52568191e-01 5.53979754e-01 -8.53944898e-01
8.12290967e-01 -1.95390135e-01 -2.82410085e-01 6.89957440e-02
-7.43396401e-01 1.20407321e-01 4.09909844e-01 -4.21353370e-01
-1.17907023e+00 -1.72705755e-01 -4.34911489e-01 -7.11109757e-01
-2.82536775e-01 1.28736451e-01 -1.62980631e-01 1.60172507e-01
8.50993276e-01 7.90032804e-01 9.51061398e-02 -3.14708114e-01
8.23857427e-01 1.96297303e-01 4.42540824e-01 -8.08486044e-01
9.09193277e-01 5.69645107e-01 5.81116192e-02 -7.06515908e-01
-5.08747041e-01 2.18273133e-01 -3.79280865e-01 -1.97831661e-01
1.00480795e+00 -8.87414992e-01 -2.37541661e-01 5.95452309e-01
-1.08657336e+00 -5.52811801e-01 -7.17781723e-01 4.18869033e-02
-1.00783706e+00 2.72634476e-01 -4.81210470e-01 -8.90325069e-01
-3.65766376e-01 -1.17316067e+00 1.54591811e+00 -5.39134741e-02
-1.92390025e-01 -1.11855900e+00 1.12943597e-01 2.34102935e-01
6.78361535e-01 4.48999047e-01 9.03712630e-01 -5.95485121e-02
-7.53094018e-01 -6.74548969e-02 -1.05359718e-01 5.91346443e-01
-2.17933804e-02 -1.57631904e-01 -1.09942293e+00 -2.07487270e-01
-1.40611783e-01 -3.27586651e-01 8.93034518e-01 7.08684444e-01
8.24969590e-01 -2.24557996e-01 -2.41932496e-01 6.44984126e-01
1.30305958e+00 -1.09722733e-01 9.60098147e-01 -1.11836553e-01
6.04676366e-01 4.92372304e-01 1.15464129e-01 3.51971328e-01
2.35079333e-01 6.61925375e-01 5.76477528e-01 -3.35239261e-01
-8.75818670e-01 -7.38220394e-01 3.60435247e-01 4.62420523e-01
-1.72850221e-01 -7.06645846e-01 -3.96589398e-01 6.84940279e-01
-1.39840782e+00 -9.48012054e-01 -6.34522364e-02 2.12816715e+00
7.84697115e-01 1.66221216e-01 1.07804462e-01 -1.30829677e-01
4.64432597e-01 2.56301105e-01 -4.67290908e-01 -1.63096920e-01
-5.61475515e-01 3.30664992e-01 1.03215091e-01 4.58395630e-01
-7.63552964e-01 7.68783152e-01 6.39867306e+00 1.21921921e+00
-8.90957892e-01 1.28240392e-01 1.13266385e+00 5.59631847e-02
-1.04765356e+00 -1.06362887e-01 -7.10951090e-01 4.19159472e-01
4.23076838e-01 2.20061123e-01 3.46514821e-01 7.09694684e-01
1.42651916e-01 -1.33782521e-01 -1.07949877e+00 1.12750876e+00
1.47062734e-01 -1.60893679e+00 5.80683887e-01 3.95517290e-01
1.21130753e+00 -1.03377581e-01 5.16863465e-01 1.59250915e-01
4.16928291e-01 -1.25603902e+00 1.01204479e+00 6.66353643e-01
1.15146768e+00 -4.51775789e-01 4.02098179e-01 3.33666623e-01
-6.54216826e-01 3.69077682e-01 -2.90310562e-01 5.99943876e-01
6.34118855e-01 1.01093209e+00 -7.49033332e-01 8.14711332e-01
4.08556372e-01 5.22980928e-01 -3.76158357e-01 6.55941963e-01
-4.63721544e-01 5.50602078e-01 -3.07230353e-01 2.38687932e-01
2.83397853e-01 -2.02985078e-01 6.70552671e-01 1.15207243e+00
8.01111817e-01 -5.93860075e-02 -1.49682939e-01 1.45311105e+00
5.45884995e-03 -2.13296875e-01 -8.81405473e-01 3.70591320e-03
1.44462407e-01 1.03253734e+00 -5.88565528e-01 -1.90825477e-01
5.98561764e-02 1.32743680e+00 -4.52546328e-02 5.15677273e-01
-7.53003001e-01 1.41376898e-01 2.23657504e-01 3.28278601e-01
6.59449935e-01 -2.37074330e-01 -1.98324919e-01 -8.79000604e-01
-5.56466989e-02 -8.76918018e-01 1.16947601e-02 -1.13455749e+00
-1.73415518e+00 9.70421255e-01 7.32933357e-02 -1.34805858e+00
-6.15843058e-01 -4.18544322e-01 -5.45570195e-01 1.06614029e+00
-1.34938991e+00 -1.58528161e+00 -2.21314102e-01 5.83082318e-01
5.34768105e-01 -1.59217775e-01 8.82389963e-01 1.22108713e-01
2.03268174e-02 4.43436503e-01 -1.90759301e-01 -3.30515862e-01
6.47994161e-01 -1.14903498e+00 7.12647617e-01 7.83011675e-01
3.94319445e-01 4.04484928e-01 7.55403101e-01 -6.66365385e-01
-1.31459498e+00 -9.75013256e-01 5.08824527e-01 -7.31071949e-01
4.64225672e-02 -3.96398902e-01 -4.21662241e-01 3.41217250e-01
4.62347537e-01 -1.01548396e-01 2.52883166e-01 -2.80521661e-01
-2.17583358e-01 1.61791325e-01 -1.23497796e+00 7.84185350e-01
1.45712948e+00 -3.97876024e-01 -1.33751705e-01 2.58389264e-01
6.27670169e-01 -4.61404353e-01 -7.42860436e-01 3.71539503e-01
4.04595703e-01 -1.38245773e+00 1.22017694e+00 -6.23664968e-02
8.63353670e-01 -2.42495239e-01 -3.32571775e-01 -1.57337570e+00
-3.13070834e-01 -9.18203235e-01 -1.50803998e-01 1.35661626e+00
5.14453173e-01 -3.43931019e-01 8.72502685e-01 3.04203540e-01
-3.58015001e-01 -8.41413021e-01 -7.34715104e-01 -8.13596547e-01
2.50265360e-01 -6.11482859e-01 3.82859230e-01 5.36892533e-01
-8.16541195e-01 3.29999089e-01 -6.15479231e-01 -2.02350348e-01
7.20360577e-01 2.11894393e-01 1.02513504e+00 -7.25518107e-01
-5.71660936e-01 -6.26140654e-01 -2.19472513e-01 -1.46001410e+00
-1.95493370e-01 -9.48050797e-01 -2.00803932e-02 -1.91744852e+00
-8.12734943e-03 -5.41636825e-01 4.15758520e-01 1.59331784e-01
-7.49956444e-02 6.37052178e-01 2.97160864e-01 2.45932296e-01
-1.31422907e-01 9.80275571e-01 1.86088765e+00 -2.16523007e-01
-4.51442003e-02 -4.57101800e-02 -7.66784430e-01 3.08087111e-01
3.46184999e-01 -3.88751686e-01 -8.13660562e-01 -4.81131911e-01
1.78019449e-01 -2.77962373e-03 6.53431296e-01 -9.95697975e-01
-1.62364185e-01 1.74919650e-01 6.33672476e-01 -6.96860731e-01
6.68313920e-01 -5.56621253e-01 5.69658399e-01 2.35095963e-01
-8.37132707e-02 -1.90176368e-01 6.34648800e-02 6.69516802e-01
-1.00677401e-01 -8.00518841e-02 6.34333134e-01 -3.13895792e-01
-4.32856560e-01 5.61663091e-01 -2.39700645e-01 2.39351839e-01
7.30225980e-01 -5.47773778e-01 -1.21335067e-01 -9.00581717e-01
-8.09179783e-01 -3.25023949e-01 6.42429173e-01 3.18237841e-01
8.00828218e-01 -1.42969799e+00 -1.15977693e+00 2.47334838e-01
4.82705273e-02 1.89191937e-01 5.27214468e-01 7.04434454e-01
-3.33212584e-01 2.52964556e-01 -3.71254720e-02 -9.57401097e-01
-7.12212443e-01 5.00902116e-01 1.31204411e-01 -4.35966045e-01
-7.87138224e-01 8.92729223e-01 7.82342732e-01 -5.08759797e-01
-9.65898484e-02 -5.95963411e-02 2.39778996e-01 -1.15120329e-01
2.35821828e-01 -1.22210674e-01 3.87790278e-02 -3.01723540e-01
-1.97267942e-02 6.94800377e-01 1.38439253e-01 -5.30413091e-01
1.42845440e+00 -9.18271393e-02 3.10091048e-01 -7.46084435e-04
9.30789292e-01 1.28346294e-01 -1.76724684e+00 7.60897845e-02
-6.94204867e-01 -4.33420956e-01 -1.25528380e-01 -1.07272422e+00
-1.24976242e+00 8.86107087e-01 5.18828928e-01 1.82931319e-01
1.11890471e+00 2.22519487e-01 7.32847571e-01 -2.28607714e-01
5.79600871e-01 -5.58201909e-01 4.79012847e-01 1.84051946e-01
1.39174342e+00 -9.70499694e-01 1.95991043e-02 -4.05124545e-01
-8.88097882e-01 8.15577209e-01 1.91114381e-01 -1.97294667e-01
5.68727136e-01 1.76469773e-01 -1.30884433e-02 -1.58638865e-01
-6.85848832e-01 -1.92142412e-01 5.39946914e-01 1.29794264e+00
3.34406704e-01 -1.66689917e-01 3.73032354e-02 2.84838229e-01
-2.65266448e-01 -1.16093598e-01 2.94937342e-01 5.18774033e-01
1.61995202e-01 -1.24997032e+00 -2.33293980e-01 1.72272190e-01
-1.62179947e-01 -3.50250304e-01 -3.19607496e-01 7.88523436e-01
3.45048122e-02 8.36248040e-01 -1.10290073e-01 -2.36153632e-01
3.52327108e-01 -1.64324448e-01 9.03866410e-01 -4.49945420e-01
-3.99138957e-01 3.02121401e-01 1.13029711e-01 -3.43255103e-01
-4.44772542e-01 -5.88320911e-01 -6.58193231e-01 -2.46582612e-01
-2.01566622e-01 -1.88264728e-01 6.52653456e-01 5.76282442e-01
7.03283906e-01 4.74364161e-01 5.79040527e-01 -1.36727917e+00
-4.17030871e-01 -9.48749244e-01 -4.90130663e-01 5.95150173e-01
-3.36961597e-02 -6.06662929e-01 -2.99849391e-01 2.12823063e-01] | [11.416590690612793, -0.48460322618484497] |
5952802c-6d1a-4ee2-91ff-493da36c3d4f | shape-of-you-precise-3d-shape-estimations-for | 2304.07389 | null | https://arxiv.org/abs/2304.07389v1 | https://arxiv.org/pdf/2304.07389v1.pdf | Shape of You: Precise 3D shape estimations for diverse body types | This paper presents Shape of You (SoY), an approach to improve the accuracy of 3D body shape estimation for vision-based clothing recommendation systems. While existing methods have successfully estimated 3D poses, there remains a lack of work in precise shape estimation, particularly for diverse human bodies. To address this gap, we propose two loss functions that can be readily integrated into parametric 3D human reconstruction pipelines. Additionally, we propose a test-time optimization routine that further improves quality. Our method improves over the recent SHAPY method by 17.7% on the challenging SSP-3D dataset. We consider our work to be a step towards a more accurate 3D shape estimation system that works reliably on diverse body types and holds promise for practical applications in the fashion industry. | ['Benjamin Biggs', 'Gerard Medioni', 'Achal Dave', 'Rohan Sarkar'] | 2023-04-14 | null | null | null | null | ['3d-human-reconstruction'] | ['computer-vision'] | [-7.10472018e-02 -1.28112286e-01 -4.23687920e-02 -4.18963701e-01
-6.79580748e-01 -7.55173385e-01 -2.76140925e-02 -1.62719548e-01
2.58560386e-02 3.22120517e-01 3.23126048e-01 -8.52895086e-04
2.65686989e-01 -3.71822715e-01 -7.93650270e-01 -1.18367195e-01
2.28128135e-01 9.21078324e-01 2.27287397e-01 -3.06558371e-01
-1.30320638e-01 5.36035120e-01 -9.85979438e-01 -1.48295090e-01
5.35124183e-01 9.44950223e-01 -2.58439720e-01 5.80871701e-01
4.58370507e-01 -6.94808662e-02 -1.91211671e-01 -9.58913684e-01
5.55219293e-01 -2.04573125e-01 -4.80505437e-01 4.31790769e-01
1.06423676e+00 -6.78505778e-01 1.40555769e-01 5.94853282e-01
9.43171561e-01 6.15136512e-02 5.09165883e-01 -8.60886395e-01
-5.04166007e-01 1.33839443e-01 -6.65347636e-01 -3.17374736e-01
6.73810244e-01 1.63636282e-01 7.27495551e-01 -8.87750506e-01
7.77877927e-01 1.20384479e+00 1.46692300e+00 6.28957093e-01
-1.33638573e+00 -3.17895561e-01 -9.49467197e-02 -3.93441677e-01
-1.26508260e+00 -5.68806231e-01 8.74801040e-01 -4.40818340e-01
6.47556961e-01 4.18550253e-01 1.28977537e+00 9.55631375e-01
-4.04174533e-03 9.57628608e-01 1.08468473e+00 -3.30186903e-01
-1.91690698e-01 -1.19802572e-01 -2.25452371e-02 9.42534447e-01
4.95644301e-01 5.60728349e-02 -4.66485351e-01 -2.55664080e-01
1.17749822e+00 -4.21337515e-01 2.02069432e-01 -8.51791978e-01
-1.18375206e+00 4.95035321e-01 2.79765904e-01 -2.90756464e-01
-3.12396675e-01 4.37142402e-01 2.95991153e-01 -5.84029406e-02
8.45879018e-01 4.33415622e-01 -3.76002371e-01 -2.07743749e-01
-1.19687283e+00 8.46620977e-01 1.03335774e+00 1.16555870e+00
-1.26300212e-02 4.81053255e-02 -6.46329746e-02 1.00286365e+00
6.90535784e-01 6.67813718e-01 -5.56659997e-01 -1.12395501e+00
6.03828430e-02 3.22912514e-01 3.85739803e-01 -1.20949328e+00
-5.02220333e-01 -4.25750971e-01 -2.10256532e-01 5.86348996e-02
8.34303141e-01 9.48985294e-02 -8.05998862e-01 1.18491268e+00
9.89583552e-01 -3.54671218e-02 -5.83976388e-01 1.55582619e+00
1.14912987e+00 4.87698279e-02 -8.59881043e-02 3.28593999e-01
1.30510890e+00 -1.03373265e+00 -3.11357528e-01 -1.05848886e-01
3.82527150e-02 -1.32298958e+00 9.06105995e-01 5.62346160e-01
-1.72030759e+00 -5.62755704e-01 -1.01505411e+00 -3.33393425e-01
2.41591573e-01 2.50109851e-01 8.87941599e-01 1.13057780e+00
-1.02072477e+00 5.01780570e-01 -8.78148794e-01 -6.71572387e-01
3.23216796e-01 4.97406155e-01 -1.90973654e-01 -1.44169360e-01
-5.69840074e-01 9.46096361e-01 -1.98706836e-01 2.52244800e-01
-6.08219326e-01 -7.06146061e-01 -8.59593391e-01 -6.81817055e-01
4.19325203e-01 -1.28412628e+00 1.23591709e+00 -4.32049036e-01
-1.57603645e+00 1.28042328e+00 -7.00239018e-02 -2.47802779e-01
6.91724360e-01 -4.84123290e-01 -1.59779146e-01 -1.73404247e-01
-1.97673321e-01 5.70725679e-01 9.69558537e-01 -1.32273531e+00
1.42596155e-01 -5.80851495e-01 -2.92694699e-02 2.91621357e-01
1.32353708e-01 2.24314228e-01 -1.07613349e+00 -7.10869610e-01
3.02427053e-01 -1.30757213e+00 -2.91520357e-01 5.78334451e-01
-4.00854409e-01 1.25612676e-01 1.24618761e-01 -1.30228603e+00
8.82772803e-01 -1.70738900e+00 3.37780386e-01 3.27942371e-01
3.10976177e-01 1.96065933e-01 1.67446718e-01 2.16411129e-01
7.16847658e-01 -1.42781720e-01 6.31945431e-02 -6.55903816e-01
1.35728702e-01 8.50720778e-02 2.36770332e-01 6.91965282e-01
6.91236630e-02 1.00327218e+00 -6.53394222e-01 -5.78591406e-01
3.30977499e-01 6.34885848e-01 -8.39959383e-01 -4.63449545e-02
-1.15232654e-01 5.25552690e-01 -3.29588801e-01 1.21005714e+00
7.52483428e-01 -2.29374766e-02 2.40954712e-01 -6.96918070e-01
1.65134400e-01 1.14860237e-02 -1.18937981e+00 1.93896246e+00
-9.62243453e-02 -5.85943721e-02 4.43767875e-01 -3.75728756e-01
1.18676269e+00 9.46542397e-02 8.18599105e-01 -1.76113784e-01
1.63608089e-01 2.00566530e-01 -2.46514514e-01 -2.38052562e-01
7.46615112e-01 -3.13254267e-01 -6.63097128e-02 8.47168788e-02
9.86493677e-02 -5.77703953e-01 -1.32466435e-01 -2.51724571e-01
4.75433826e-01 1.01013613e+00 1.33912504e-01 -2.31082082e-01
-4.47266698e-02 2.54917383e-01 4.90887970e-01 3.96519095e-01
-4.16276187e-01 9.47062552e-01 -1.05966277e-01 -6.09940469e-01
-1.46695328e+00 -1.25502098e+00 -1.50400013e-01 9.22816575e-01
2.51977861e-01 -6.06012404e-01 -5.21995127e-01 -5.81645668e-01
6.90726519e-01 2.39144102e-01 -3.61736655e-01 3.01961362e-01
-6.09486580e-01 -6.06215775e-01 4.67226654e-01 8.30778658e-01
1.33710906e-01 -3.60899001e-01 -1.78464904e-01 3.16718251e-01
-8.26285705e-02 -1.23380578e+00 -9.24814463e-01 -4.70682144e-01
-1.02916324e+00 -9.36787128e-01 -8.72138023e-01 -5.76974213e-01
5.54405093e-01 1.50010750e-01 1.58381629e+00 8.37755725e-02
-2.98208982e-01 8.77830505e-01 -2.93059289e-01 -5.19322574e-01
-3.73939663e-01 -2.48787269e-01 3.74214917e-01 -4.37796414e-01
2.70037234e-01 -2.80209810e-01 -9.13302124e-01 7.52406001e-01
-6.45225868e-02 2.42696419e-01 1.90643407e-02 5.73178172e-01
1.01580870e+00 -4.95060742e-01 3.02621543e-01 -5.47900736e-01
4.97879326e-01 -1.66071072e-01 -3.42802852e-01 -4.59871534e-03
-4.26753998e-01 -3.97444308e-01 8.57980698e-02 -3.86871576e-01
-7.89069831e-01 4.08489376e-01 -6.23067796e-01 -3.32549274e-01
7.15516284e-02 1.21506460e-01 3.56666371e-02 -6.86876357e-01
4.91959840e-01 -1.58060104e-01 3.49081635e-01 -9.39862669e-01
3.16658080e-01 3.22503835e-01 5.58236778e-01 -8.99698496e-01
7.12741256e-01 3.84211868e-01 2.42971152e-01 -7.88634777e-01
-8.08786333e-01 -6.20221138e-01 -8.11419725e-01 -5.04500866e-01
8.07302356e-01 -1.14890850e+00 -6.57438219e-01 3.32068443e-01
-6.86661720e-01 -3.20805907e-01 -1.13149166e-01 3.14379722e-01
-8.16885114e-01 6.42870367e-01 -6.98900104e-01 -8.75101507e-01
-7.66456485e-01 -9.70280170e-01 1.66860807e+00 -4.83052582e-02
-7.03376114e-01 -1.02922559e+00 2.43742745e-02 9.94679987e-01
1.83511049e-01 5.81692159e-01 1.96061432e-01 -2.33626235e-02
-2.89247841e-01 -2.71706372e-01 1.43858612e-01 2.94248730e-01
-1.81749314e-01 1.05352819e-01 -5.33548415e-01 -4.73431170e-01
-4.66664642e-01 -3.45430583e-01 4.99763340e-01 8.37165356e-01
8.36332679e-01 4.00127023e-02 -2.03073531e-01 8.20624888e-01
1.10483313e+00 -4.57114726e-01 3.21682274e-01 1.21813305e-02
8.79318058e-01 3.07456017e-01 8.56842339e-01 4.53981459e-01
8.87631655e-01 1.24421990e+00 2.32376248e-01 -3.09610218e-01
-7.17881143e-01 -5.53896070e-01 1.31503835e-01 1.09671807e+00
-8.16224456e-01 3.65272999e-01 -5.89216471e-01 4.49469239e-01
-1.60772836e+00 -6.58700824e-01 -3.66292685e-01 2.34635067e+00
8.07753563e-01 3.60932350e-02 9.63061690e-01 -2.32866928e-01
2.50406414e-01 -2.18646348e-01 -5.76136589e-01 -4.83307868e-01
1.52968243e-01 2.61897027e-01 6.07455790e-01 2.18023062e-01
-1.24888837e+00 8.75176668e-01 7.47848892e+00 3.60626876e-01
-6.17102921e-01 4.99393120e-02 2.50074923e-01 -2.90216148e-01
-2.19758198e-01 -2.79291451e-01 -9.76877868e-01 2.95019209e-01
3.12300026e-01 3.50262076e-01 4.93164361e-01 8.70615959e-01
2.24645421e-01 -2.02552625e-03 -9.96942282e-01 1.10396636e+00
3.08983922e-01 -1.04801369e+00 -4.22146738e-01 1.73303306e-01
6.55916631e-01 -2.62609929e-01 -6.13410585e-02 8.05407166e-02
2.36925364e-01 -7.81067967e-01 1.01323223e+00 7.45583296e-01
9.16608870e-01 -5.79497695e-01 3.23764473e-01 1.48250416e-01
-1.16002011e+00 4.89281386e-01 -5.23331881e-01 -1.23747410e-02
4.86771762e-01 5.32998383e-01 -1.16166055e+00 4.54976797e-01
7.45041072e-01 7.35732138e-01 -5.83672523e-01 1.39395964e+00
1.11965790e-01 8.69606972e-01 -5.78966677e-01 -1.11387104e-01
-4.32666570e-01 -2.59948105e-01 7.09693730e-01 1.03921640e+00
2.87268639e-01 -1.86965987e-02 4.93549496e-01 7.21324921e-01
-3.43191028e-02 1.16215058e-01 -5.10397077e-01 4.24225777e-02
2.34636009e-01 1.25822258e+00 -6.89770818e-01 1.55692339e-01
-1.78294703e-01 1.09994328e+00 9.08259824e-02 -6.46739677e-02
-8.45824122e-01 5.26005268e-01 6.30318642e-01 4.81399477e-01
3.23358774e-01 -7.54834056e-01 -5.59162021e-01 -1.19420862e+00
3.21713164e-02 -9.64184403e-01 9.71333310e-02 -8.60551715e-01
-1.54946685e+00 -1.23931810e-01 -4.76264991e-02 -1.09281194e+00
-1.26349047e-01 -5.78522325e-01 3.77346762e-02 6.77694023e-01
-8.80606234e-01 -1.94085026e+00 -2.30313465e-01 2.52529442e-01
4.97647226e-01 2.39575416e-01 8.96568894e-01 3.12029183e-01
-2.63056606e-01 9.12412763e-01 -1.56077832e-01 -2.82366157e-01
8.62003207e-01 -1.24121702e+00 9.71718013e-01 5.23096561e-01
1.82886496e-01 7.03960538e-01 8.78285527e-01 -9.14717853e-01
-2.21335649e+00 -7.79349625e-01 3.93006116e-01 -1.11087167e+00
1.93013921e-01 -4.19312000e-01 -2.06445649e-01 7.03074813e-01
-2.91639000e-01 -3.05039641e-02 9.29994941e-01 6.56702876e-01
-1.87314272e-01 1.71210244e-02 -1.44183016e+00 4.71330494e-01
1.47054517e+00 -5.83191179e-02 -5.43265343e-01 1.63745001e-01
2.68307328e-01 -1.04349983e+00 -1.59895074e+00 5.77181399e-01
1.39190698e+00 -6.46375656e-01 1.57550788e+00 -4.92698580e-01
1.43093735e-01 -3.45396787e-01 -2.46492505e-01 -1.18513823e+00
-5.89784682e-01 -5.57363153e-01 -6.74515247e-01 1.04592061e+00
-8.96403641e-02 -2.17124615e-02 9.44751263e-01 8.36579323e-01
-1.84102848e-01 -9.41172361e-01 -6.85052931e-01 -6.26773775e-01
-5.67746162e-02 -5.33311605e-01 3.93258989e-01 6.52126253e-01
-4.13994312e-01 7.09294677e-02 -9.87203360e-01 -1.38190269e-01
1.06033945e+00 3.28324080e-01 1.26761270e+00 -1.27950215e+00
-5.41700602e-01 -9.77956653e-02 -5.82484126e-01 -1.32605934e+00
-3.36533368e-01 -7.69695640e-01 -9.40831006e-02 -1.69350529e+00
3.63756418e-01 -5.43970168e-01 1.99188724e-01 4.30688560e-01
-4.94039543e-02 1.04583538e+00 3.77191991e-01 -1.02401637e-01
-6.14648283e-01 1.25951603e-01 1.74434638e+00 -2.09185202e-02
-1.75767601e-03 2.57221282e-01 -7.76647329e-01 7.97975659e-01
4.89257723e-01 1.38959646e-01 -7.72330537e-02 -5.21518588e-01
2.03683883e-01 -1.22578703e-01 6.79374397e-01 -8.87316465e-01
-2.12338299e-01 9.54383463e-02 9.72487926e-01 -8.90530884e-01
7.07716465e-01 -8.57872605e-01 5.17260075e-01 1.43904880e-01
1.74699485e-01 -2.55688936e-01 2.95970291e-01 3.40677798e-01
6.65565491e-01 3.82599770e-03 4.96227354e-01 -1.96746439e-01
-5.22738099e-01 3.67010802e-01 1.66085735e-01 -1.10962510e-01
7.29016483e-01 -5.58500707e-01 2.46115416e-01 -2.43486062e-01
-1.00596464e+00 2.19307151e-02 1.03537071e+00 3.76589209e-01
7.37885535e-01 -1.62682915e+00 -8.98928642e-01 1.78008556e-01
-3.20676379e-02 -2.41373405e-01 2.95494124e-02 9.46426392e-01
-6.50450230e-01 2.32108571e-02 -2.29143992e-01 -7.45032609e-01
-1.69021690e+00 2.31647730e-01 2.52058774e-01 3.30819041e-02
-6.07813597e-01 8.74354899e-01 -5.78502357e-01 -7.48338342e-01
-3.26778032e-02 -3.42170238e-01 2.35389903e-01 -3.72590780e-01
1.63478613e-01 6.69723570e-01 7.00467674e-04 -9.15750623e-01
-4.47778076e-01 9.24110830e-01 2.68129647e-01 1.88672855e-01
1.31180334e+00 -2.23337963e-01 1.58155933e-01 3.03200454e-01
7.36889780e-01 4.88988340e-01 -1.41920936e+00 5.07915318e-02
-3.94748390e-01 -6.66516662e-01 -3.08623072e-02 -1.10532868e+00
-8.03359389e-01 3.48469943e-01 5.70454419e-01 -6.52229488e-02
9.20543969e-01 1.87565908e-01 1.23743904e+00 -1.64465398e-01
8.90770912e-01 -9.25171196e-01 -2.71176428e-01 2.24577755e-01
9.88091767e-01 -1.25595701e+00 6.87805772e-01 -7.43529618e-01
-6.88832879e-01 9.19175684e-01 4.45362985e-01 -2.72209555e-01
5.11313140e-01 3.31933737e-01 -1.44753501e-01 -3.28975290e-01
-1.10608727e-01 -9.43410024e-02 1.12078834e+00 8.83229494e-01
8.88306141e-01 3.06117564e-01 -5.99456131e-01 7.79118955e-01
-2.75281698e-01 2.58321583e-01 -1.00530133e-01 6.26921654e-01
-1.11497119e-01 -1.25590515e+00 -5.88743627e-01 5.87603569e-01
-4.78390515e-01 1.51435539e-01 -4.38686162e-01 6.34884536e-01
1.12131752e-01 7.34198570e-01 -3.83047044e-01 -4.34908360e-01
6.51772976e-01 -8.34104270e-02 1.45696342e+00 -5.09035408e-01
-6.47884071e-01 4.75900412e-01 7.73914218e-01 -8.11449170e-01
-5.95980406e-01 -8.63677084e-01 -7.17660069e-01 -5.78465462e-01
-4.72366631e-01 -5.20147741e-01 8.23137820e-01 5.05780399e-01
3.56355757e-01 1.13364659e-01 1.70962825e-01 -1.31353295e+00
-6.15104020e-01 -6.13101423e-01 -5.59900224e-01 5.62780619e-01
-2.07270786e-01 -1.12407553e+00 4.08080250e-01 2.36613348e-01] | [7.013289928436279, -1.1194239854812622] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.